The transcriptional regulator TamR from Streptomyces coelicolor controls a key step in central metabolism during oxidative stress

Authors


For correspondence. E-mail agrove@lsu.edu; Tel. (+1) 225 578 5148; Fax (+1) 225 578 8790.

Summary

Multiple antibiotic resistance regulator (MarR) family transcriptional regulators usually regulate gene activity by responding to specific ligands. Here we show that TamR (trans-aconitate methyltransferase regulator), a MarR homologue from Streptomyces coelicolor, functions in oxidative stress responses to regulate a key step in central metabolism. The gene encoding TamR is oriented divergently from the tam gene, which encodes trans-aconitate methyltransferase. Trans-aconitate methyltransferase methylates trans-aconitate, which is formed when cis-aconitate is released during aconitase-mediated isomerization of citrate to isocitrate; trans-aconitate, but not its methyl ester, is a potent inhibitor of aconitase. We show that TamR binds with high affinity to the intergenic region between the tamR and tam genes. Notably, trans-aconitate attenuates DNA-binding by TamR, as do citrate, cis-aconitate and isocitrate, which are the substrate, intermediate and product of aconitase respectively. In vivo, hydrogen peroxide and citrate induce significant upregulation of the tam (SCO3132), tamR (SCO3133) and aconitase (SCO5999) genes. Since oxidative stress leads to disassembly of the [4Fe-4S] cluster that is essential for aconitase activity, resulting in accumulation of citrate and release of cis-aconitate and its subsequent conversion to trans-aconitate, we propose that TamR mediates a novel regulatory function in which the inhibitory effects of trans-aconitate and accumulated citrate are alleviated.

Introduction

Members of the multiple antibiotic resistance regulator (MarR) family of transcriptional regulators are involved in a variety of important biological processes, including aromatic compound catabolism, virulence factor biogenesis and stress responses (for review, see Wilkinson and Grove, 2006; Perera and Grove, 2010a). Responding to specific ligands is a characteristic of many MarR family proteins, which often repress the genes under their control until ligand is bound (Cohen et al., 1993; Ariza et al., 1994; Buchmeier et al., 1997; Wilkinson and Grove, 2004; Davis and Sello, 2010; Perera and Grove, 2010b). When the MarR homologue controls transcription of a gene encoding a metabolic enzyme, the ligand may be the substrate for this enzyme as exemplified by MobR from Comamonas testosteroni. MobR regulates transcription of the mobA gene, which encodes 3-hydroxybenzoate 4-hydroxylase, and repression of gene activity by MobR is relieved on binding 3-hydroxybenzoate (Hiromoto et al., 2006). Another example is HucR from Deinococcus radiodurans, which belongs to a subfamily of MarR homologues that respond to the ligand urate by attenuated DNA binding in vitro and increased gene activity in vivo; since HucR functions as a repressor by binding a cognate site that overlaps core promoter elements, attenuated DNA binding by HucR results in more efficient recruitment of RNA polymerase and therefore increased transcription. HucR regulates the transcription of a gene encoding uricase, an enzyme that participates in purine degradation by converting urate into 5-hydroxyisourate (Wilkinson and Grove, 2004; 2005; Perera et al., 2009).

HucR differs from canonical MarR homologues by having an N-terminal extension, which adopts a helical conformation (Bordelon et al., 2006). Biochemical mapping of the urate-binding site in HucR identified four residues required for urate-mediated attenuation of DNA binding, with one of these residues deriving from the unique N-terminal extension (Perera et al., 2009). Using the sequence of HucR as a query to Blast bacterial genomes revealed the existence of potential urate-responsive transcriptional regulators (UrtR) in other bacterial species, two of which were experimentally confirmed; Agrobacterium tumefaciens PecS and Burkholderia thailandensis MftR respond to the ligand urate, but they do not regulate a gene encoding uricase. Instead, these transcription factors have been speculated to participate in responses to oxidative stress by regulating transcription of a divergently oriented gene encoding an efflux pump involved in secretion of antioxidants (Grove, 2010; Perera and Grove, 2010b). The latter inference was based on experimental reports that Dickeya dadantii (Erwinia chrysanthemi) PecS regulates expression of a gene encoding the efflux pump PecM, which transports the antioxidant indigoidine (Rouanet and Nasser, 2001).

Another potential urate-responsive MarR homologue was predicted in the genome of Streptomyces coelicolor based on bioinformatics analyses (Perera and Grove, 2011). S. coelicolor is a soil-dwelling bacterium and the model organism of Streptomyces species, characterized by a complex lifecycle and production of an extensive set of secondary metabolites (Hopwood, 1999; Bentley et al., 2002; Challis and Hopwood, 2003; Gehring et al., 2004). Notably, a gene encoding a putative homologue of HucR was identified that is not oriented divergently from a gene encoding an efflux pump, but instead a gene annotated as a putative trans-aconitate methyltransferase (Fig. 1A). A genomic organization involving divergently encoded genes is common for loci encoding MarR homologues and predicts regulation of the divergently oriented gene by the transcription factor, a prediction based on the frequent reports of such mode of regulation (e.g. Ariza et al., 1994; Davis and Sello, 2010; Perera and Grove, 2010b).

Figure 1.

Genetic locus organization of tamtamR and related metabolic reactions.

The enzyme trans-aconitate methyltransferase functions to regulate key metabolic pathways. Citrate isomerization is an important reaction in both the citric acid and glyoxylate cycles, in which citrate is converted to isocitrate by aconitate hydratase (aconitase) via the intermediate cis-aconitate (Fig. 1B). If cis-aconitate is released from the enzyme-substrate complex, it can spontaneously be converted to the more stable isomer trans-aconitate (Ambler and Roberts, 1948). Trans-aconitate is an efficient inhibitor of aconitase and if this toxic by-product accumulates, it inhibits the citrate isomerization step (Saffran and Prado, 1949; Lauble et al., 1994; Cai and Clarke, 1999; Cai et al., 2001). One circumstance under which trans-aconitate may accumulate is oxidative stress; aconitase contains a [4Fe-4S] cluster that is essential for catalytic activity, and this cofactor is disassembled under oxidative stress conditions rendering the enzyme non-functional (Verniquet et al., 1991; Gardner and Fridovich, 1992). The enzyme trans-aconitate methyltransferase functions to prevent accumulation of trans-aconitate by using it as a substrate to catalyse a methyl group transfer from S-adenosyl-methionine (SAM), resulting in formation of the trans-aconitate methylester. Esterification of trans-aconitate significantly attenuates its ability to inhibit aconitase and to interfere with key steps in central metabolism (Cai and Clarke, 1999; Cai et al., 2001). The fate of the trans-aconitate methyl ester is unclear.

In this paper, we report that the predicted HucR homologue (named TamR for trans-aconitate methyltransferase regulator) functions to control expression of the genes encoding trans-aconitate methyltransferase (tam) and TamR. Consistent with attenuated DNA binding in vitro, gene activity is increased under oxidative stress conditions under which accumulation of citrate and trans-aconitate occurs. Thus, TamR functions in oxidative stress responses to alleviate the consequences of aconitase inactivation.

Results

The genomic locus containing tamR and tam genes is conserved among Actinomycetales

S. coelicolor gene SCO3133 encodes a predicted MarR homologue that is divergently oriented from gene SCO3132 annotated as a putative trans-aconitate methyltransferase (tam; Fig. 1A). The intergenic region would be predicted to contain −10 and −35 promoter elements of both genes. Based on the function of this MarR homologue described below, we propose the name TamR (trans-aconitate methyltransferase regulator). In addition to what is seen in the S. coelicolor genome, the tamtamR locus organization consisting of divergently oriented tam and tamR genes is conserved in the genomes of several other bacteria that belong to the order Actinomycetales.

Comparing the amino acid sequence of TamR proteins with that of canonical MarR homologues revealed that all TamR homologues contain the N-terminal extension that is characteristic of urate-responsive MarR homologues and that they harbour the four residues shown to be involved in urate binding (Fig. 2A). TamR homologues were found to be highly conserved among the species in which they are encoded, including a wide region (boxed in Fig. 2A), which covers the turn between α4 and α5 and the N-terminal half of α5. Since α5 corresponds to the DNA recognition helix identified from structures of MarR homologues complexed with DNA (e.g. Hong et al., 2005), conservation of DNA binding sites for TamR would be predicted.

Figure 2.

Conservation of TamR.

A. Sequence alignment of TamR proteins and other MarR homologues. Secondary structure elements are from the structure of HucR and aligned with the sequence of HucR (Bordelon et al., 2006). The N-terminal helix α1 is absent in E. coli MarR and Methanobacterium thermoautotrophicum MTH313. In addition to the TamR homolog from S. coelicolor (Q9K3T1), Streptomyces avermitilis (Q82HD8), Kribbella flavida (D2PPC7) and Thermomonospora curvata (D1A7I8), the alignment includes D. radiodurans HucR (Q9RV71), E. coli MarR (P27245), M. thermoautotrophicum MTH313 (O26413) and A. tumefaciens PecS (Q7D1T4). TamR proteins, D. radiodurans HucR and A. tumefaciens PecS all belong to the urate-responsive UrtR subfamily of MarR homologues. Red frame highlights conservation of residues near and within the DNA recognition helix (α5).

B. Phylogenetic analysis of selected MarR homologues from the information of amino acid sequences. Phylogenetic tree was generated with MEGA4 using neighbour-joining method with 500 bootstrap replicates. The tree is drawn to scale. The evolutionary distances are in units of the number of amino acid substitutions per position. Grey background shading denotes TamR proteins. The scale bar represents an evolutionary distance of 0.05.

TamR homologues were found to be encoded by bacterial species that belong to the order Actinomycetales, particularly the genus Streptomyces. Phylogenetic analysis of TamR and other MarR homologues revealed clustering of TamR homologues, with the most closely related homologues being the identified urate-responsive proteins (e.g. A. tumefaciens PecS and D. radiodurans HucR) while homologues such as E. coli MarR are more distantly related (Fig. 2B). Thus, bacterial species that encode TamR homologues are evolutionarily closely related, raising the possibility that a common ancestor may have adopted the genomic locus containing the tam and tamR genes, perhaps by a gene duplication of an existing marR homologue followed by selection for the novel function (i.e. responding to ligands associated with aconitase function).

Consistent with sequence conservation of the DNA-binding helices of the analysed TamR homologues, all corresponding genomic loci encoding the tam–tamR gene pair were seen to feature an 18 bp conserved palindromic site in the tamtamR intergenic region that is predicted to constitute the TamR binding site (Fig. 3A). Sequence comparison and WebLogo revealed the consensus sequence of TamR binding sites (Fig. 3 and B). In particular, residues at the centre of each half-site of the binding motif (positions 3–8 and 11–16) were found to be highly conserved. In analogy with other MarR homologues, the layout of the genomic locus and the presence of conserved palindromic sequences in the intergenic region also leads to the prediction that TamR autoregulates its own gene transcription. Examination of the S. coelicolor tam–tamR intergenic region did not reveal other sequence repeats likely to function as binding sites for transcription factors.

Figure 3.

Sequence consensus of TamR binding site.

A. Predicted TamR binding sequences (18 bp inverted repeat sequences) from the intergenic region of tamtamR loci from 12 different bacteria. The respective tamR genes are named at the right, preceded by the distance (in bp) between the start codon and the nearest identified binding motif.

B. WebLogo (http://weblogo.berkeley.edu/logo.cgi) representing the consensus TamR binding site. The relative frequency of base pairs at each position is represented by the height of each nucleotide.

C. Predicted TamR binding site (grey shading) in the promoter region of aconitase encoding genes. Asterisks indicate positions where nucleotides are identical in all sequences.

All analysed tam–tamR intergenic regions were found to contain a palindromic site that matches the consensus sequence (Fig. 3A). However, the tamtamR intergenic region in the chromosome of S. coelicolor has some special characteristics. In addition to the conserved TamR binding site (site 1), five additional predicted TamR binding sites were identified (sites 2–6; Fig. 4). These five putative TamR binding sites are similar to each other, but more divergent from the identified consensus sequence compared to site 1. All six sites are adjacent to each other, overlapping by three base pairs.

Figure 4.

S. coelicolor tamtamR intergenic region. The tam and tamR genes are oriented divergently. The sequence of the intergenic region is shown with predicted binding sites. The conserved TamR binding site is site 1. Five additional TamR binding sites are located adjacent to the conserved binding site (sites 2–6). Each TamR binding site overlaps the preceding site by three base pairs. Bottom panel shows alignment of all six cognate sites. Asterisks indicate positions where nucleotides are identical in all sequences.

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 PATSER (http://rsat.ulb.ac.be/genome-scale-patser_form.cgi) to find other genes potentially regulated by TamR. The promoter regions of genes identified by this method were examined, revealing putative TamR binding sites in the promoters of two genes that encode enzymes of the citric acid cycle, aconitase (SCO5999) and malate synthase (SCO6243). Notably, the TamR binding site was conserved in the promoters of aconitase genes in all 12 bacterial species seen to feature the tam–tamR genomic locus (some species encode two aconitase genes, only one of which containing the conserved TamR binding site) (Fig. 3C). The presence of predicted TamR binding sites in the promoters of genes encoding aconitase in all these bacterial species further points to a link between TamR and citrate isomerization.

TamR binds to the tamtamR intergenic region

The tamR gene was cloned from S. coelicolor genomic DNA and the His6-tagged protein expressed in E. coli and purified to apparent homogeneity (Fig. 5A). Since MarR proteins are expected to exist as dimers, size-exclusion chromatography was used to determinate the oligomeric state of purified TamR, which was found to exist as single species by native gel electrophoresis (data not shown). TamR eluted from the gel filtration column at 41.4 kDa (Fig. 5B), which is consistent with the molecular mass of the expected dimer (theoretical molecular mass of recombinant TamR dimer is 41.7 kDa). Dimeric TamR could also be detected in SDS-PAGE gels following cross-linking with glutaraldehyde (data not shown).

Figure 5.

Characterization of TamR.

A. Purified TamR in 15% SDS-PAGE gel. Lane 1, TamR; lane 2, molecular mass marker (Bio-Rad; Mw indicated at right).

B. Gel-filtration analysis of TamR. The standard curve was generated from the Kav of molecular weight standards (grey squares) versus Log(10)(Mw) of these standards. The Kav of TamR is indicated as the black circle.

C. Far-UV CD spectrum of TamR. Ellipticity measurements are represented in units of millidegrees (mdeg; machine units).

D. Thermal stability of TamR. Fluorescence emission resulting from the binding of SYPRO Orange to denatured protein was measured as a function of temperature.

Far-UV circular dichroism spectroscopy showed that the secondary structure composition of TamR is about 57% α-helix, 10% β-sheet and 33% random coil (Fig. 5C), with the secondary structure composition estimated based on the CD spectrum using DichroWeb (Whitmore and Wallace, 2004; 2008). This is similar to the composition of HucR, which contains about 55% α-helix and 5% β-sheet (Bordelon et al., 2006). TamR was quite stable with a melting temperature (Tm) of 59.9 ± 0.3°C (Fig. 5D). This is comparable to other MarR homologues, which also have relatively high melting temperatures (Wilkinson and Grove, 2004; Andresen et al., 2010; Perera and Grove, 2010b). Thus, TamR exists as a stable dimer at physiologically relevant temperatures and features the secondary structure content expected for a MarR homologue.

To determine whether TamR binds the tamtamR intergenic region, a 247 bp sequence named tamO representing this intergenic region was used in electrophoretic mobility shift assays (EMSA). TamR bound to tamO forming two clearly distinguishable complexes (Fig. 6A). TamR bound with high affinity as evidenced by an apparent dissociation constant (Kd) of 16.5 ± 1.2 pM (Fig. 6B). This Kd likely represents an upper limit, as conditions for formation of complex 1 are nearly stoichiometric ([DNA] ∼ Kd). A TamR–tamO complex (identified as C1 in Fig. 6A) was formed at relatively low TamR concentrations (Kd ∼ 16 pM) and with increasing concentrations, another TamR–tamO complex (C2) appeared; consistent with the observation that DNA is effectively saturated to form complex 1 before additional sites are filled, formation of complex 1 and 2 is not cooperative, as evidenced by a Hill coefficient of 1.0 ± 0.0.

Figure 6.

Binding of TamR to tamtamR intergenic region tamO.

A. EMSA showing binding of TamR. DNA (0.015 nM) was titrated with TamR (lanes 1–15 representing reactions with 0, 1.4 pM, 2.9 pM, 5.7 pM, 11.4 pM, 22.9 pM, 45.7 pM, 91.4 pM, 0.18 nM, 0.37 nM, 0.73 nM, 1.46 nM, 2.92 nM, 5.85 nM and 11.7 nM TamR respectively). Two relatively stable complexes (C1 and C2) were detectable. Complexes (C1 and C2) and free DNA (F) are identified by arrows.

B. Fractional complex formation was plotted as a function of TamR concentration. Data represented by filled circles considers the sum of complex 1 and complex 2 as complex (Kd = 16.5 ± 1.2 pM; nH = 1.0 ± 0.0), while data represented by filled squares considers only complex 2 as complex and the sum of free DNA and complex 1 as ‘free DNA’ (Kd = 4.5 ± 1.0 nM; nH = 1.8 ± 0.1; note that Kd does not represent the affinity for a single site). Error bars represent standard deviation from three independent repeats. Inset shows example of electrophoresis at room temperature, revealing six TamR–tamO complexes (lane 2; other experimental conditions are unaltered). Bands corresponding to C1 and C2 in panel (A) are indicated, with bands corresponding to additional TamR-DNA complexes visible at intermediate electrophoretic mobilities. Lane 1 contains DNA only.

C. Binding of 0.015 nM labelled tamO to 0.18 nM TamR was challenged with increasing concentration of unlabelled 247 bp tamO DNA (lanes 3–6: 0.038, 0.075, 0.15, 0.225 nM) or the same concentration of 3000 bp plasmid pGEM5 (lanes 7–10). Reaction in lane 1 contained labelled DNA only. Reaction in lane 2 contained no competitor DNA.

D. Challenge of complex 2 formation by competition assay. All conditions are the same as (C) except 0.73 nM TamR is used. Note that reactions in (C) and (D) were carried out at higher ionic strength (0.5 M Tris) compared to reactions in (A) (50 mM Tris).

At sufficiently high TamR concentrations (∼ 1.5 nM), complex 1 disappeared and all DNA was bound as complex 2, with 50% conversion of complex 1 to complex 2 at a > 10-fold higher TamR concentration than that required for half-maximal conversion of free DNA to complex 1. The difference in migration between complex 1 and complex 2 indicates that there are more than two TamR dimers bound to tamO in complex 2 (Fig. 6A). This is consistent with the prediction that there are six TamR binding sites in the intergenic region, one (conserved TamR binding site 1) with relatively high TamR binding affinity and five sites with lower affinity for TamR (Fig. 4). Indeed, six bands corresponding to TamR–tamO complexes were detected when electrophoresis was performed under different conditions (inset to Fig. 6B). Quantification of complex 2 formation, considering complex 1 as ‘free DNA’, yielded a Hill coefficient of 1.8 ± 0.1 (Fig. 6B), suggesting cooperativity of binding to sites 2–6 (Fig. 4).

Specificity of TamR binding to tamO was assessed by EMSA experiments in which unlabelled specific DNA (tamO) or nonspecific DNA (plasmid pGEM5) were combined with labelled tamO DNA and TamR. Two different concentrations of TamR were used to examine specificity of complex 1 and complex 2 separately. Using a concentration of TamR where only complex 1 is seen (0.18 nM TamR; Fig. 6C), only addition of unlabelled tamO could compete with labelled tamO for binding to TamR. That not all labelled DNA was unbound at the highest concentration of competitor reflects that total tamO DNA (0.24 nM) was only in modest excess over TamR. By contrast, complex formation between TamR and tamO was not significantly affected by addition of non-specific pGEM5 DNA (up to 15-fold molar excess of the 3000 bp pGEM5 compared to the concentration of 247 bp tamO). Using a concentration of TamR where complex 2 is formed (0.73 nM TamR; Fig. 6D), unlabelled tamO DNA efficiently competed for formation of complex 2 despite TamR being in excess over unlabelled tamO, with all DNA converted to complex 1; this is consistent with the inference that the affinity of TamR for site 1 (Fig. 4) is higher than the affinity for adjacent sites. As for complex 1 formation, addition of nonspecific pGEM5 DNA had no effect on formation of complex 2. These experiments show that TamR binds sequence-specifically to tamO.

TamR responds to trans-aconitate, cis-aconitate, citrate and isocitrate by attenuated DNA binding

As trans-aconitate is the substrate of trans-aconitate methyltransferase, its effect on TamR–tamO binding was measured. In addition, effects of three related compounds, citrate, isocitrate and cis-aconitate, which are the substrate, product and intermediate of the citrate isomerization reaction, respectively, were determined. All these four compounds have close structural and metabolic relationships (Fig. 1B).

With increasing concentration of trans-aconitate, the binding of TamR to tamO was significantly attenuated. Formation of both TamR–tamO complex 1 (with an IC50 of 70.5 ± 1.9 mM) and TamR–tamO complex 2 (with an IC50 of 35.8 ± 0.4 mM) was attenuated by trans-aconitate (Fig. 7 and E). Cis-aconitate, citrate and isocitrate also attenuated the binding of TamR to tamO (Fig. 7B–D, F–H). For complex 1, the binding of TamR to tamO was attenuated by cis-aconitate, citrate and isocitrate with IC50 values of 62.6 ± 1.4, 88.2 ± 3.1 and 99.1 ± 2.5 mM respectively (Fig. 7F–H). For complex 2, the IC50 values were 53.0 ± 1.2, 71.5 ± 2.0 and 61.3 ± 1.5 mM for cis-aconitate, citrate and isocitrate respectively (Fig. 7–H). Evidently, TamR exhibits little discrimination between these structurally related compounds, which more efficiently attenuated formation of complex 2 than complex 1. To confirm that attenuation of DNA binding was not due to compromised structural integrity of TamR on ligand binding, CD spectra and thermal stabilities of TamR were measured in presence of ligand. The CD spectrum of TamR in presence of 100 mM trans-aconitate did not indicate significant changes in secondary structure content (data not shown). Addition of ligand had no effect on protein stability, even at the highest concentration of 50 mM (Table 1); this could perhaps reflect changes in protein flexibility that compensate for stabilizing effects of ligand binding or that ligand binding perturbs the energy landscape of the native state ensemble by preferred binding to a less stable substate.

Figure 7.

Effect of different ligands on the binding of TamR to tamO.

A–D. Reactions in lanes 1, 8 and 9 contained labelled tamO DNA only. Ligand concentrations in lanes 2–7 and 10–15 are 0, 10, 20, 50, 75, 100 mM respectively. DNA (0.015 nM) was incubated with 0.73 nM TamR (lanes 2–7) or 0.09 nM TamR (lanes 10–15). Complexes (C1 and C2) and free DNA (F) are identified.

E–H. Normalized complex 1 and complex 2 formation as a function of ligand concentration. Error bars represent the standard deviation of three independent repeats.

Table 1. Thermal stability of TamR in presence of ligands
LigandTm (°C)
None59.9 ± 0.3
Trans-aconitate59.7 ± 0.4
Cis-aconitate59.9 ± 0.2
Citrate59.6 ± 0.2
Isocitrate60.2 ± 0.5

Since TamR was identified based on similarity to urate-responsive MarR homologues, the effect of urate on DNA binding of TamR was also investigated. Other intermediates in the purine degradation pathway (including xanthine, hypoxanthine and allantoin), which are structurally similar and metabolically related to urate were also examined. EMSA results revealed that urate and related ligands have little or no effect on DNA binding by TamR (data not shown), consistent with TamR responding to distinct ligands.

In vivo effect of hydrogen peroxide and citrate on TamR-mediated gene regulation

To investigate gene regulation by TamR in vivo, transcript levels of S. coelicolor tam and tamR were measured under conditions in which intracellular citrate concentrations are increased, either by uptake of citrate or by inactivating aconitase with hydrogen peroxide, an event associated with accumulation of both citrate and trans-aconitate. In S. coelicolor and other Gram-positive bacteria, citrate can be transported across cell membranes by the CitMHS family of transporters when citrate forms complexes with specific metal ions (Lensbouer et al., 2008; 2010). In S. coelicolor, the function of one member of the CitMHS family of transporters (SCO1710; CitSc) has been experimentally documented, and the protein was found to transport citrate efficiently when it forms complexes with Fe3+ or Ca2+, but not with Mg2+, Ni2+ or Co2+. Exogenous citrate cannot be transported by the CitMHS family of citrate transporters unless it is complexed with metal; consistent with this observation, exposing S. coelicolor cultures to exogenous citrate did not have any effect on transcript levels of tam (relative expression level 1.1 ± 0.4 for cultures grown in presence of citrate relative to cultures to which no citrate was added) or tamR (relative level 1.3 ± 0.1) (Fig. 8A). However, if the media was supplemented with 100 mM citrate and 5 mM Fe3+, qRT-PCR results showed that the transcript level of tam is elevated 11.3 ± 1.9 fold. In contrast, the transcript level of tamR was only modestly increased (1.7 ± 0.3 fold). Consistent with the uptake of only citrate complexed with appropriate metals, supplementing the media with citrate and Ca2+ lead to similarly increased transcript levels (14.5 ± 3.2 fold for tam and 2.7 ± 0.4 for tamR; Fig. 8A). These data suggest that citrate functions as a TamR ligand in vivo, and that expression of the tam gene is more sensitive to citrate compared to that of tamR.

Figure 8.

In vivo gene regulation.

A. Relative abundance of tam and tamR transcripts after exposure to 100 mM citrate, 100 mM citrate + 5 mM Fe3+, 100 mM citrate + 5 mM Ca2+ or 10 mM H2O2. Quantitative RT-PCR was used to measure the relative mRNA levels of tam and tamR genes and reference control gene (rpoA).

B. Relative abundance of sacA transcript after exposure to 100 mM citrate+10 mM Ca2+ or 10 mM H2O2. Note that y-axis is expanded compared to graph in (A). Error bars represent standard deviation of three repeats.

In S. coelicolor, the activity of aconitase is closely linked to intracellular citrate concentrations; inactivating the gene encoding aconitase dramatically increases intracellular citrate concentrations from nearly undetectable to > 14 mM (Viollier et al., 2001). Aconitase contains a [4Fe-4S] cluster, which plays a critical role in catalysis, and its conversion to a [3Fe-4S] cluster results in loss of enzymatic activity (Beinert et al., 1983; Robbins and Stout, 1989; Verniquet et al., 1991; Gardner and Fridovich, 1992; Flint et al., 1993). The in vivo effect of hydrogen peroxide on the transcript level of tam and tamR was therefore investigated, as it can cause accumulation of intracellular citrate by inactivating aconitase as well as cause release of cis-aconitate from inactivated enzyme (Viollier et al., 2001). Quantitative RT-PCR results demonstrated that expression of tam and tamR genes is significantly elevated in response to hydrogen peroxide, with transcript levels of tam and tamR increasing by 27.3 ± 3.9 fold and 3.0 ± 0.9 fold respectively (Fig. 8A).

Because a TamR binding site is predicted in the promoter region of the aconitase gene sacA, transcription of sacA was also investigated. Incubation of S. coelicolor cultures with citrate (100 mM) plus Ca2+ (10 mM) or with H2O2 (10 mM) results in increased expression of sacA by 1.9 ± 0.1 fold and 2.5 ± 0.1 fold respectively (Fig. 8B). To examine further the role of TamR in mediating regulation of the sacA gene, EMSA experiments were performed using a DNA fragment representing the sacA promoter. As shown in Fig. 9A, TamR bound this DNA forming one distinct complex, consistent with the presence of a single conserved palindromic binding motif in the sacA promoter. The affinity of TamR for this DNA was 1.9 ± 0.2 nM. These data also corroborate the validity of the consensus TamR binding motif. As for TamR binding to the tam–tamR intergenic region, binding to the sacA promoter was attenuated in the presence of trans-aconitate, cis-aconitate, citrate and isocitrate (Fig. 9B-C and data not shown). Collectively, these data suggest that TamR responds to intracellular accumulation of citrate by differentially upregulating tam and tamR genes, and they are consistent with TamR contributing to regulation of sacA gene activity.

Figure 9.

Binding of TamR to sacA promoter DNA.

A. EMSA showing binding of TamR. Labelled sacA promoter DNA (0.5 nM) was titrated with TamR; the TamR concentrations in lanes 1–9 are 0.04, 0.08, 0.16, 0.31, 0.63, 1.25, 2.50, 5.0 and 10.0 nM respectively. Reaction in lane 10 contained labelled DNA only. Complexes (C) and free DNA (F) are identified.

B and C. Effect of trans-aconitate (B) and citrate (C) on binding of TamR to the promoter region of sacA. Ligand concentrations in lanes 2–6 are 0, 25, 50, 75, 100 mM respectively. Reactions contained 0.63 nM TamR and 0.5 nM DNA. Reaction in lanes 1 contained DNA only. Reactions in all panels were performed at high ionic strength (0.5 M Tris).

Discussion

Aconitase function is coupled to oxidative stress

Both eukaryotic and prokaryotic aconitases are bifunctional proteins (Haile et al., 1992; Alén and Sonenshein, 1999). In the presence of iron, their [4Fe-4S] clusters are assembled, and the proteins function as aconitases. When iron is limiting and [4Fe-4S] cluster assembly is compromised, the proteins lose aconitase function and serve instead as RNA-binding proteins to regulate expression of genes associated with iron metabolism by binding iron response elements (IREs) in the mRNA. This dual function permits the cells to sense and respond to changes in cellular iron concentration. When aconitase catalytic function is compromised and citrate cannot be metabolized, secreted citrate can in turn function as an iron-chelator, facilitating uptake of iron to reassemble [4Fe-4S] clusters and restore catalytic activity.

The [4Fe-4S] cluster of aconitase is disassembled by peroxide stress. Consistent with this observation, the mRNA-binding property of aconitase has also been associated with regulation of genes involved in oxidative stress responses (e.g. Tang et al., 2002). Such function is also likely for aconitase encoded by Streptomycetes (Michta et al., 2012). Under these conditions, TamR may contribute to restoring the catalytic function of aconitase by participating in regulating its gene activity and by preventing inhibition of functional enzyme by removing the inhibitor trans-aconitate.

TamR promotes metabolic flux through the citric acid and glyoxylate cycles

Reactive oxygen species cause inactivation of aconitase due to disruption of its [4Fe-4S] cluster, resulting in elevated citrate and trans-aconitate concentrations. Under these conditions, tam, tamR and aconitase genes are upregulated (Fig. 8). Citrate has been shown to accumulate to > 14 mM concentration on inactivation of aconitase (SCO5999), which was experimentally shown to be the primary vegetative aconitase in S. coelicolor (Viollier et al., 2001). We observe increased gene activity not only on inactivation of aconitase by H2O2, but also after uptake of iron–citrate or calcium-citrate complexes. For iron–citrate complexes, the most relevant species has been shown to be a monoiron–dicitrate complex (Silva et al., 2009). Using 5 mM Fe3+, and considering a monoiron–dicitrate complex as the most relevant species, the extracellular concentration of transportable citrate would be 10 mM. Under these conditions, an ∼ 11-fold upregulation of tam was observed, while an increase of ∼ 27-fold was seen on exposure to H2O2. Given that intracellular citrate concentrations may rise to > 14 mM on inactivation of aconitase, this suggests that we observe increased gene activity under physiologically relevant concentrations of citrate. That exposure to H2O2 results in a greater increase in gene expression may be a consequence of a higher intracellular citrate concentration compared to that obtained after uptake of exogenous citrate, or it may be due to contributing effects of trans-aconitate formed as a result of cis-aconitate released from inactivated aconitase. TamR contains no cysteines (oxidation of which could be associated with conformational changes in the protein and possible modification of DNA-binding activity) and would not be expected to respond directly to oxidative stress. Thus, our data suggest that TamR serves an important function in ensuring metabolic flux through the citric acid and glyoxylate cycles by responding to the increased cellular levels of citrate and trans-aconitate that occur under oxidative stress conditions to effect upregulation of key metabolic enzymes.

Ligand-binding by TamR

TamR was identified based on homology to UrtR, specifically conservation of the N-terminal helix and residues shown to be involved in urate-mediated attenuation of DNA binding (Perera and Grove, 2011). Yet, TamR does not bind urate or other intermediates of purine metabolism, but compounds associated with aconitase function. TamR exhibits little preference for either of these ligands, all of which share significant negative charge. The IC50 is also comparable for disruption of complex 1 and for conversion of complex 2 to complex 1, despite the binding of five additional TamR dimers in complex 2 (Figs 6 and 7). If each additional TamR dimer in complex 2 must bind ligand for complex 2 to be converted to complex 1, a higher ligand concentration would be required compared to conversion of complex 1 to free DNA. This is not observed. It is possible that differences in the DNA sites may impose distinct conformational changes in the ligand binding pockets on DNA binding, causing high-affinity bound TamR to be less sensitive to ligand, or that ligand binding to a single TamR dimer that is part of complex 2 leads to cooperative disassembly of this complex. That saturation of all six TamR sites requires > 10-fold higher TamR concentration compared to that required to saturate site one is consistent with lower affinity binding to the adjacent sites. We also note that the arrangement of cognate sites with a 3 bp overlap would place the centres of each palindrome about 15 bp apart, predicting that adjacent TamR dimers bind on opposite faces of the DNA duplex. In vivo, however, tam gene expression is more sensitive to ligand than that of tamR (Fig. 8). This may reflect more efficient repression of tam gene expression by TamR, perhaps due to TamR protein concentrations that are insufficient to saturate the adjacent sites, resulting in a greater net increase in tam gene expression when ligand is present compared to the increase in tamR gene expression. In addition, differential promoter strength or the participation of other regulatory proteins may contribute to the observed differences.

The TamR ligand-binding pocket is distinct from that of urate-responsive MarR homologues. Sequence alignments reveal that TamR proteins share residues not otherwise conserved among urate-responsive transcriptional regulators, particularly in helix two, which lines the predicted ligand-binding pocket (Fig. 2A). A model of S. coelicolor TamR based on the structure of HucR is shown in Fig. 10. The residues previously shown to participate in urate binding to HucR occupy equivalent positions in TamR (identified in cyan). One of these residues is a tryptophan from helix one, which is absent from other MarR proteins for which a structure has been reported. Since this helix would be expected to block access to the ligand-binding pocket from this direction, a more likely access route for ligands is from the underside of the protein near the dimer interface. Notably, two arginine residues from helix two, which are conserved only among TamR proteins are seen to face the ligand-binding pocket (shown in magenta), perhaps affording selectivity for highly negatively charged ligands. In addition, a tryptophan at the beginning of helix three is seen to form a lid at the bottom of the binding pocket. While the predicted role of these residues in ligand binding to TamR awaits confirmation, their presence only in TamR proteins suggests a mechanism for discrimination between urate and ligands associated with aconitase function.

Figure 10.

Model of TamR. Model based on structure of HucR (2fbk), using SwissModel. Arginine 41 and tryptophan 53 from one subunit and arginine 35 from the other are identified in magenta. Residues previously shown to participate in urate binding to HucR are shown in cyan. Secondary structure elements α1 and α2 forming one edge of the ligand-binding pocket are identified, along with the DNA recognition helix α5. Figure generated with PyMOL.

Model for expression of the TamR regulon

A possible model for gene regulation by TamR is shown in Fig. 11. Conserved TamR sites are located in the promoters of tam and sacA, with more divergent sites near the start of the tamR gene. At low concentrations of TamR, the high-affinity site in the tam gene promoter is filled, and tam gene expression is repressed. Occupancy of the site in the sacA promoter would attenuate sacA gene expression; however, sacA regulation is likely complex and predicted to include the ferric uptake regulator Fur, as a site for this transcription factor is predicted to overlap the TamR site (Muschko et al., 2002). Based on this model, a tamR mutant would be predicted to exhibit constitutive tam gene expression. With no TamR bound to divergent sites in the tamR gene promoter, tamR is expressed. As TamR accumulates, it can fill the cognate sites in the tamR gene promoter, resulting in repression of gene expression. We note that the observed mode of TamR binding to the tam–tamR intergenic region involving cooperative binding to low-affinity sites near the tamR gene, but no cooperativity of binding to the high affinity site near the tam gene and the adjacent low-affinity sites is consistent with differential TamR-mediated regulation of the tam and tamR genes. The tam–tamR intergenic regions contains six adjoining DNA sites of which cooperative binding is only observed for the five low-affinity sites; this may imply that occupancy of high- and low-affinity sites leads to differential structural changes in DNA or protein and that only occupancy of low-affinity sites is compatible with cooperative binding. Based on this model of repression, the negative autoregulation of tamR prevents excessive accumulation of TamR, yet ensures adequate levels of TamR for repression of tam gene expression and a more sensitive response to ligands. Thus, even if physiological conditions do not change, cellular concentrations of TamR would be predicted to fluctuate within a range that ensures sensitive control of the tam gene.

Figure 11.

Proposed model for regulation of tam, tamR and sacA by TamR. Elliptical symbols represent TamR, small filled circles represent ligand. Genes are denoted by large arrows. Ligand concentrations may be increased by uptake of citrate or by inactivation of aconitase, resulting in accumulation of both citrate and trans-aconitate.

If aconitase is inactivated, intracellular citrate concentrations will increase. In addition, if the intermediate cis-aconitate is released, trans-aconitate concentrations will increase because cis-aconitate will be converted to the more stable isomer. These events are intertwined and may happen at the same time. Both citrate and trans-aconitate will attenuate the binding of TamR to its cognate sites, resulting in increased gene transcription; upregulation of sacA ensures production of functional aconitase, and production of trans-aconitate methyltransferase results in esterification of trans-aconitate, preventing its inhibition of aconitase. The concomitant upregulation of tamR ensures that sufficient TamR is available to restore the repressed state of tam gene expression once citrate and trans-aconitate concentrations return to normal levels.

Taken together, we propose that TamR constitutes a divergent member of the urate-responsive transcriptional regulator (UrtR) family that mediates a novel regulatory function that ensures metabolic flux through the citric acid and glyoxylate cycles during oxidative stress. While TamR is unique in binding ligands associated with aconitase function and in regulating a key step in central metabolism, it shares with other characterized members of the UrtR family a primary function in oxidative stress responses.

Experimental procedures

Sequence alignment and phylogenetic analysis

Amino acid sequences of selected MarR homologues were aligned using the MUSCLE sequence alignment server (Edgar, 2004). Residues were shaded according to their identity and similarity using BOXSHADE v3.21. The sequence of HucR (D. radiodurans) was used to identify secondary structure elements (Bordelon et al., 2006). Phylogenetic tree was generated with MEGA4 using neighbour-joining method and pre-aligned sequences (Saitou and Nei, 1987; Tamura et al., 2007). Five hundred bootstrap replicates were analysed to generate bootstrap consensus tree and to estimate the statistical confidence values. Positions that contain gaps were eliminated during calculation. The tree was drawn to scale. The evolutionary distances are in units of number of amino acid substitutions per site. The sequence logo, which shows the consensus sequence of the TamR binding site was generated using WebLogo (Crooks et al., 2004) at http://weblogo.berkeley.edu/.

Cloning and purification of TamR

S. coelicolor A3(2) M145 strain was kindly provided by Gregg Pettis, LSU. After S. coelicolor was grown in tryptone yeast extract broth (ISP medium 1), the genomic DNA was isolated using the salting out method (Kieser et al., 2000). Forward primer 5′-CACTACACTGATCCATATGGAGGAC-3′ and reverse primer 5′-GACCTGGACGGGAATTCAGCC-3′ were used to amplify the gene encoding TamR (SCO3133; restriction sites underlined). The PCR product was cloned into the NdeI–EcoRI sites of pET28b (Novagen), which introduces an N-terminal His6-tag, and the recombinant plasmid transformed into E. coli TOP10 (Invitrogen). After the correct construct was confirmed by sequencing, it was transformed into E. coli BL21(DE3)pLysS for protein expression. A single colony was used to inoculate an overnight culture. For overexpression, the overnight culture, which was grown at 37°C (250 rpm) in Luria–Bertani media (with 50 μg ml−1 kanamycin) was diluted 1:500 with LB media containing 50 μg ml−1 kanamycin. Cultures were grown at 37°C (250 rpm). When the OD600 reached about 0.6, overexpression of protein was induced with 0.2 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG) for 3 h. The induced cultures were chilled on ice. Cells were harvested by centrifugation and stored at −80°C.

After cell pellets were thawed on ice, cells were resuspended in ice-cold lysis buffer (50 mM potassium phosphate buffer (pH 7.0), 400 mM NaCl, 10% glycerol, 10 mM imidazole, 0.15 mM phenylmethylsulphonyl fluoride (PMSF), 1 mM 2-mercaptoethanol). Sonication was used to disrupt cells. DNase I was added to digest nucleic acids. This solution was centrifuged at 28 000 g for 60 min. The supernatant was filtered through filter paper and loaded onto a HIS-Select Nickel Affinity column (Sigma), previously equilibrated with lysis buffer. The column was washed by gravity flow with 10 volumes of wash buffer (50 mM potassium phosphate buffer (pH 7.0), 400 mM NaCl, 10% glycerol, 20 mM imidazole, 0.15 mM PMSF, 1 mM 2-mercaptoethanol). Elution buffer [50 mM potassium phosphate buffer (pH 7.0), 400 mM NaCl, 10% glycerol, 250 mM imidazole, 0.15 mM PMSF, 1 mM 2-mercaptoethanol] was used to elute proteins. Peak fractions were pooled and dialysed against 2 l dialysis buffer [50 mM potassium phosphate buffer (pH 7.0), 400 mM NaCl, 10% glycerol, 0.15 mM PMSF, 1 mM 2-mercaptoethanol]. The purity of protein preparations was ascertained using SDS-PAGE, followed by staining of gels with Coomassie brilliant-blue. TamR concentration was determined based on its absorbance at 280 nm using the calculated extinction coefficient (12 490 M−1 cm−1). All experiments were performed with His6-tagged TamR.

DNA binding assays

The intergenic segment between S. coelicolor tamR (SCO1133) and tam (SCO1132) genes was amplified using primers tamO-Fw (5′-TCCGGCGTGGCGCAGGTACT-3′) and tamO-Rv (5′-GCGACCAGCCGATCGACCT-3′). This 247 bp tamO DNA contains the entire intergenic region and extends 54 bp into the coding region of SCO1132 and 29 bp into the coding region of SCO1133. The 247 bp DNA was 32P-labelled at the 5′-ends using T4-polynucleotide kinase (T4-PNK). 32P-labelled tamO (0.015 nM) was incubated with TamR in binding buffer [20 mM Tris (pH 8.0), 50 mM NaCl, 0.06% detergent BRIJ58 (Pierce), 20 μg ml−1 bovine serum albumin (BSA), 2% glycerol] at 25°C for 30 min. Complex and free DNA were resolved using 6% non-denaturing polyacrylamide gels [39:1(w/w) acrylamide : bisacrylamide]. After the gel was pre-run for 20 min in 0.5× Tris–borate–EDTA (TBE) buffer at 4°C, the samples were loaded and run at 10 V cm−1 for 2 h in 0.5× TBE buffer. The gel was dried and exposed to phosphor screens. Results were visualized using a Storm 840 phosphorimager (GE Healthcare), and analysed using ImageQuant 5.1. Data were fitted to the Hill equation: f = fmax[TamR]n/(Kd + [TamR]n), where [TamR] is the protein concentration, f is fractional saturation, Kd is the apparent equilibrium dissociation constant, and n is the Hill coefficient. Specificity of interaction between TamR and tamO was measured using competition assay in which unlabelled non-specific plasmid DNA (pGEM5) or unlabelled 247 bp tamO were used as competitor, with competitor DNA added prior to labelled tamO DNA. The binding buffer for these assays contained 0.5 M Tris. Binding of TamR to DNA representing the sacA promoter was performed as described for binding to tamO, except that the binding buffer contained 0.5 M Tris and that 0.5 nM labelled sacA DNA was used. The 179 bp sacA promoter DNA (comprising four base pairs of the coding region and 175 bp upstream) was amplified from S. coelicolor genomic DNA using primers 5′-CCCCATGTACTAGAGTTATCT-3′ and 5′-ACACGACAGTCTCCTTCA-3′.

To determine the effect of ligand, the binding buffer used was 0.5 M Tris (pH 8.0), 50 mM NaCl, 0.06% BRIJ58, 20 μg ml−1 BSA, 2% glycerol. Note that the higher buffer concentration was necessitated to prevent pH changes on addition of urate, xanthine, and hypoxanthine, which were dissolved in 0.4 M NaOH, and that this higher ionic strength reduces the affinity of TamR for both DNA and ligands. Other ligands were dissolved in distilled water and added before addition of TamR to DNA. The concentrations of TamR were 0.09 nM and 0.73 nM, respectively, when measuring the effect of a ligand on complex 1 and complex 2. After 30 min incubation at 25°C, samples were analysed using EMSA under the conditions described above. For quantification, the gel region considered as complex 1 included C1 and the region between C1 and free DNA, while the gel region considered as complex 2 included C2 and the gel region between C2 and C1 to account for complex dissociation during electrophoresis. Data were analysed by fitting to exponential decay equation: f = AekL, where f is fractional saturation, L is the ligand concentration, A is the saturation plateau, and k represents the exponential decay constant. Quantification results derive from at least three independent experiments.

Gel filtration

A Bio-Gel P-100 (GE Healthcare) column (0.7 × 100 cm) was pre-equilibrated and eluted with mobile phase buffer [50 mM potassium phosphate buffer (pH 7.0), 150 mM NaCl]. Markers used to create the standard curve include bovine serum albumin (66.0 kDa), ovalbumin (44.0 kDa), myoglobin (17.0 kDa) and vitamin B12 (1350 Da). The equation Kav = (VE − VO)/(VT − VO) was used to calculated the Kaverage (Kav) of a protein. In this equation, VE, VO and VT represent the retention volume of the protein, void volume of the column and the geometric bed volume of the column respectively.

Circular dichroism spectroscopy

A Jasco J-815 circular dichroism spectrometer (Jasco, Inc.) was used to measure the far UV circular dichroism spectrum of 10 μM TamR in CD buffer [12.5 mM potassium phosphate buffer (pH 7.0), 100 mM NaCl, 2.5% glycerol, 0.25 mM 2-mercaptoethanol] at 20°C. Where added, trans-aconitate was included at a final concentration of 100 mM. Measurements were conducted at 1 nm steps in triplicate. A quartz cuvette with 0.1 cm path length was used. Secondary structure composition was calculated using the analysis K2d Program from the website DichroWeb (Whitmore and Wallace, 2004; 2008). The maximum error for this K2d analysis was 0.097 and the goodness of fit was determined from the NRMSD value of 0.105.

Thermal stability

TamR (8 μM) in a measurement buffer [200 μM Tris (pH 8.0), 200 mM NaCl] was mixed with reference fluorescent dye 5× SYPRO Orange (Invitrogen). When added, ligands were included at a final concentration of 50 mM after adjusting the pH of ligand solutions to 8.0 with NaOH. Fluorescence emission was measured over a temperature range of 5−94°C in 1°C increments for 45 s using an Applied Biosystems 7500 Real-Time PCR System. SYBR green filter was used for detection. Total fluorescence yield was corrected using a measured result from a reaction without protein. The sigmoidal part of the melting curve was fit to a four-parameter sigmoidal equation using Sigma Plot 9. At least three independent experiments were performed.

In vivo regulation of gene activity

S. coelicolor cultures, which were germinated from spores in yeast extract-malt extract media (YEME media, with 10.3% sucrose) were grown for 22 h and then treated with either citrate (100 mM), citrate (100 mM) combined with Fe3+ or Ca2+ (5 mM), or H2O2 (10 mM) for 2 h before cells were harvested by centrifugation. For determination of sacA transcript level, S. coelicolor cultures were grown in ISP medium 1 (without glucose or sucrose to avoid potential catabolite repression) before incubation with citrate (100 mM) and Ca2+ (10 mM) or H2O2 (10 mM) for 2 h. The total RNA was isolated using illustra RNAspin Mini Isolation Kit (GE Healthcare) after the pellet was quickly washed with 50 mM sodium phosphate buffer (pH 6.4) twice. AMV reverse transcriptase (New England BioLabs) was used to generate cDNA for quantitative PCR (qPCR). Applied Biosystems 7500 Real-Time PCR system was used to carry out qPCR with SYBR green I as fluorescent dye and gene rpoA (house-keeping gene encoding RNA polymerase alpha subunit whose expression is not expected to vary under the experimental conditions) as internal control. Comparative CT (2−ΔΔCT) method was used for data analysis after data validations (Schmittgen and Livak, 2008).

Acknowledgements

We thank Gregg Pettis for providing the Streptomyces coelicolor A3(2) M145 strain. Supported in part by the National Science Foundation (MCB-1051610 to A.G.). The authors have no conflict of interest to declare.

Ancillary