Equilibrium binding and kinetic characterization of putative tetracycline repressor family transcription regulator Fad35R from Mycobacterium tuberculosis

Authors


S. Kumaran, Council of Scientific and Industrial Research, India, Institute of Microbial Technology, Sector 39-A, Chandigarh 160036, India
Fax: +91 172 269632
Tel: +91 172 6665474
E-mail: skumaran@imtech.res.in

Abstract

Fatty acids play critical role in the survival and virulence of Mycobacterium tuberculosis (Mtb). Activation of fatty acids by acyl-CoA synthetases (Fad) into fatty acyl-CoA is the first and one of the crucial steps in fatty acid metabolism. Mtb possesses 36 fatty acyl-CoA synthetases, unlike Escherichia coli, which has single enzyme. However, the mechanisms by which the expression of these multiple Fad genes is regulated remain uncharacterized. We characterized the DNA- and ligand-binding properties of a putative tetracycline repressor family regulator, named Fad35R, located upstream of the Fad35 gene and ScoA-citE operon. We identified a palindromic regulatory motif upstream of Fad35 and characterized the binding of Fad35R to this motif. Equilibrium binding studies show that Fad35R binds to this motif with high affinity (Kd∼ 0.033 μm) and the specificity of binding was confirmed by an electromobility gel shift assay. Kinetic studies indicate that faster association (ka,avg∼ 5.4 × 104 m−1·s−1) and slower dissociation rates (kd,avg∼ 5.84 × 10−4 s−1) confer higher affinity. The affinity for the promoter is maximum at 300 mm NaCl but decreases rapidly beyond this range. Ligand-binding studies indicate that Fad35R binds specifically to tetracycline and also binds to fatty acid derivatives. The promoter-binding affinity is decreased significantly in the presence of palmityl-CoA, suggesting that Fad35R can sense the levels of activated fatty acids and alter its DNA-binding activity. Our results suggest that Fad35R may be the functional homologue of FadR and controls the expression of genes in a metabolite-dependent manner.

Structured digital abstract 

Abbreviations
EMSA

electromobility gel shift assay

FA

fatty acid

HTH

helix-turn-helix

ITC

isothermal titration calorimetry

Mtb

Mycobacterium tuberculosis

PCA

palmitoyl-CoA

RNAP

RNA polymerase

SPR

surface plasmon resonance

TetR

tetracycline repressor

Introduction

Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis, encodes hundreds of genes (∼18%) whose functions remain to be investigated [1,2]. Many of these hypothetical or untested putative genes are essential for the virulence and survival of this bacterium [3,4]. Constantly changing environmental milieu and subsequent changes in the levels of intracellular metabolites pose many challenges to the bacterium [5]. Most of the fundamental responses occur via the deployment of relevant specific transcription factors to bring about an alteration in gene expression [6]. Analysis of Mtb, as well as Mycobacterium avium, reveals that Mtb possesses a large number of transcriptional regulators (∼200) and almost 25% of these are predicted to be tetracycline repressor (TetR) family regulators [7]. TetR family proteins are naturally designed molecular sensors of small ligands and are known to interact with a variety of structurally unrelated molecules [8,9,44]. Recently, the TetR family has attracted wide spread scrutiny because regulators belonging to this family control the expression of genes involved in development, fatty acid (FA) metabolism, drug resistance and many other cellular processes [10–12].

Among the many components that are considered to be vital for the survival and pathogenicity of Mtb under hostile conditions, FAs were shown to assist the Mtb in the survival and maintaining their virulence [13]. They impart functional diversity to the dynamic lipid profile responsible for effective establishment of Mtb in a wide variety of environments [14]. Mtb has approximately 250 genes for FA metabolism compared to E. coli, which has only approximately 50 genes for FA metabolism [15]. The expression of genes involved in FA metabolism pathways was shown to be regulated by TetR family members and some of these genes have been proposed as potential targets for drug design [16,17]. Mainly, TetR/GntR family transcriptional regulators control genes of FA metabolism, although only very few of them from Mtb have been characterized so far [18,19]. FadR, a bonafide member of the TetR family was shown to regulate the transcriptional control of FA metabolism in E. coli, Bacillus subtilis and Thermus thermophilus [20–22]. FadR from E. coli has been characterized as a member of GntR family, whereas FadRs from B. subtilis and T. thermophilus belong to the TetR family [21,22]. The functional homologue of FadR, however, has not been characterized in Mtb, although gene annotation studies have identified several putative TetR family transcription factors as potential regulators of FA metabolism [23].

Although it is well established that FAs play vital role in reshaping the virulence of Mtb in host, details of the regulatory circuits of FA-mediated gene regulation are not yet known [13]. E. coli possesses a single fatty acyl-CoA synthetase (Fad) of 62 kDa, whereas Mtb has approximately 36 fatty acyl-CoA synthetases (Fad1–36) [1,24]. The presence of such large numbers of Fads increases the complexity of FA degradation and presents a challenge with respect to deciphering the regulatory mechanisms of FA homeostasis. Many long- and medium-chain FAs, which are products of Fad operons, are ligands of TetR family regulators with an ability to alter the DNA-binding properties of regulators upon binding to them [21,22]. Both structural scaffolds and intracellular concentrations of FAs play critical roles in selectively altering the DNA-binding properties of regulators [19–22]. The exquisite feedback circuit built to switch genes on and off through metabolite-mediated control needs to be deciphered in detail to allow an understanding of the transcriptional wiring of FA metabolism. When scanning the Mtb genome, we found that, out of 36 Fad homologues, three of them (Fad5, Fad8 and Fad35) have an adjacent GntR/TetR family regulator gene placed immediately next to their coding region. The coding region of TetR family regulator, in general, is found to be upstream of their target genes and, hence, it is likely that the adjacent regulator may exert transcriptional control over its downstream genes/operons [25,26]. Fad35 codes for an acetyl-coA synthetase that plays a crucial role in the β-oxidation and degradation of FA [1]. The identified Fad35R present upstream of Fad35 gene has a unique 23 amino acid N-terminal extension that is not present in known TetR family members. We hypothesized that Fad35R may bind to promoter regions of the Fad35 gene and that its DNA-binding activity may be sensitive to levels of activated long-chain FAs, which are products of the Fad35 gene.

To test whether Fad35R can be a regulator of the Fad35 gene, we cloned, expressed, purified and characterized the solution properties, as well as DNA-binding and ligand-binding properties, of Fad35R from Mtb. Purified Fad35R is homodimer in solution, specifically recognizing tetracycline family antibiotics, as well as binding citrate and long-chain fatty acyl-CoA. We deduced a Fad35R recognition motif upstream of the Fad35 gene using bioinformatics analysis and employed biophysical studies to confirm the specific recognition of this regulatory motif by Fad35R with high affinity. Kinetics of the Fad35R–DNA interaction revealed that the off-rate is very slow and the interaction is not sensitive to salts at lower salt concentrations. In addition, the results show that long-chain fatty acyl-CoAs, which are products of Fad35, can bind to Fad35R and reduce its affinity for the promoter DNA recognition. In summary, the findings of the present study suggest that a TetR family regulator, Fad35R, may be functional homologue of FadR and that its DNA-binding activity is sensitive to levels of FA metabolites.

Results

Cloning, expression and biophysical characterization of Fad35R from Mtb

The Fad35R gene was cloned into pET28a (+) vector, and expressed and purified as described in the Materials and methods. Fad35R was purified to homogeneity by nickel affinity and gel filtration chromatography. The purified Fad35R migrated as an approximately 23-kDa protein band (Fig. 1A) by 12% SDS/PAGE analysis. We determined the assembly state of Fad35R using size-exclusion chromatography and dynamic light scattering (Details provided in Doc. S2). The size-exclusion profile shows an elution profile consistent with a dimeric molecule of approximately 46 kDa (Fig. 1A). The dialyzed protein was concentrated to approximately 2.9 mg·mL−1 (122.0 μm) and light scattering profiles were recorded at 25 °C. A single exponential model can be used to adequately describe the normalized autocorrelation function plotted as a function of delay time (Fig. 1B). The Stokes radius of the molecule was calculated from the estimated translational diffusion coefficient, D20,w (3.5 × 10−7 cm2·s−1) from Eqn (2). An estimation of the mean hydrodynamic radius from D20,w, using Eqn (2) yields a value of approximately 7.0 nm, as expected of a protein with a molecular size of approximately 53 kDa, which is similar to the value of 46 kDa expected for the Fad35R homodimer. We examined the secondary structural content using CD spectroscopy, which indicates that Fad35R consists of more than 70%α-helix structures (Fig. 1C), as expected for TetR family transcriptional regulators [27–29]. To better understand the structural features of CD spectrum, we employed a homology modelling approach using the crystal structure of a transcriptional regulator of FA degradation (TetR/AcrR family) from B. subtilis (Protein Data Bank code: 1VI0) [30]. Overall, the modelled structure indicated predominantly helical structures with more than 70% residues in helical conformation, which is consistent with the results of the CD studies (Fig. 1C and Fig. S1). Superimposition of our model with the EthR structure shows that the DNA-binding helix-turn-helix (HTH) domain structures are very similar, although the ligand-binding C-terminal domains exhibit significant deviation, as expected for a TetR family regulator [28]. In summary, the purified Fad35R protein is a dimer in solution, properly folded with a high helical content, and the model structure displays a N-terminal HTH DNA-binding domain and a variable C-terminal ligand-binding domain, as expected for a TetR family regulator.

Figure 1.

 Analytical and spectroscopic characterization of Fad35R. (A) Size-exclusion chromatography analysis of the purified Fad35R protein in buffer B (20 mm Tris pH 8.0, 150 mm NaCl and 10% glycerol). Size-exclusion profiles monitored at two wavelengths; 260 nm (dash) and 280 nm (thick). The loading concentration for the proteins was 203.0 μm (4.6 mg·mL−1). The observed elution volume of 160 mL is expected for the Fad35R homodimer under our experimental conditions. Inset: 12% SDS/PAGE analysis of affinity and gel filtration purified Fad35R. Lane 1, marker; lane 2, supernatant; lane 3, flow through; lane 4, size-exclusion purified; lane 5, His-tag cleaved and purified Fad35R. (B) Dynamic light scattering studies performed on the purified Fad35R at higher protein concentration (122.0 μm). The plot of normalized autocorrelation function as a function of delay time (tau) for the scattering intensity obtained from Fad35R solution. Estimation of hydrodynamic radius from the autocorrelation function yields a value of approximately 7.0 nm, which is expected for approximately 53.0 kDa assuming a spherical shape. (C) CD spectrum of Fad35R in buffer in the presence (dashed line) and absence of tetracycline (solid line); Fad35R concentration is 2.5 × 10−6 m. CD signature of Fad35R indicates that the purified protein is folded.

Identification of the Fad35R-binding site and analyses of Fad35R–promoter DNA interactions

TetR family regulators have been found to interact with promoter sequences containing different signatures. TetR from E. coli and other organisms binds preferably to palindromic sequences, with a few TetR family members that bind as tandem dimers (two or four dimers may bind motifs with inverted repeats with variable linker lengths inbetween) [31–34]. We searched for binding motifs within the 114-bp intergenic region of the Fad35R and Fad35 genes (Complete intergenic sequence provided in Doc. S1). Initial inspection did not show any obvious palindromic motifs; instead, two direct repeat sequences of 8 bp in length separated by 8-bp linker sequences were identified (Fig. 2A, boxes). However, most of TetR family regulators recognize palindromic motifs as a result of the homodimeric assembly. We compared the sequences of intergenic regions of 40 putative TetR family regulators aiming to deduce any consensus motif. Details of sequence alignment and sequences of intergenic regions are provided in Doc. S3. We deduced two palindromic motifs using these approaches. A consensus palindrome was deduced (from extensive comparative analysis of 40 putative intergenic regions) which overlaps -10 consensus sequence of RNAP. In general, the binding site for TetR regulator is found to be downstream of the −10 box. Therefore, we employed an alternative approach in which we aligned the upstream region of the Fad35 gene against palindromic motifs reported for FadR regulators from other bacteria (Detailed sequences provided in Table S1) [35]. Using this method, we found a diffused palindrome that shows similarity to the binding motif of DhaS (Fig. 2A, blue) [34]. This diffused palindrome is 24 bp in length and is present downstream of the −10 box, as expected for the binding site of TetR family repressors. Interestingly, this diffused palindrome also overlaps with two imperfect direct repeats found within this region (Fig. 2A, boxes). We designed a 41-bp stretch that includes the predicted palindrome (24 bp) and an additional 6–11 bp on either side. Binding of Fad35R to its site is expected to block the RNA polymerase (RNAP) binding site by occluding the −10 box of the RNAP binding site. In addition to altering the structural properties of the RNAP binding site, binding of Fad35R to its site may also physically obstruct RNAP binding. It is known that actual contact site (physical interaction) of a protein on the DNA is smaller than the total site size (occluded) estimated by conventional equilibrium methods [36]. Therefore, the addition of extra nucleotides on binding sites may be necessary to form the correct protein–DNA complex. Furthermore, it was observed that EthR oligomerizes with 1 : 8 stoichiometry (DNA to protein) and cooperative oligomerization is favoured toward the transcription start codon [32]. Therefore, we added additional nucleotides on the right to ensure that we covered the entire binding region, as well as to test whether Fad35R can bind with a > 1 : 2 ratio (1 : 4 as well as 1 : 8). Any cooperative binding of tandem dimers can be observed more clearly with this longer template.

Figure 2.

 Bioinformatics analyses of Fad35R and binding sites and the protein–DNA interaction. (A) Top: architecture of the Fad35R regulon and the predicted intergenic regions (IG1, IG2 and IG3) and the identity of genes. Middle: a closer view of IG1 and the promoter architecture showing the position of the −10 boxes of the RNAP binding site. Bottom: relative position and sequence of the 41-bp motif containing the predicted palindromes, diffused (blue) and consensus (red) motif overlapping with -10 box; DhaS binding motif is aligned to the predicted diffused palindrome; motifs highlighted within boxes are possible direct repeats; a two-headed arrow under a sequence shows the Fad35R-binding region. (B) Scan of fluorescence emission at different concentrations of 41-bp dsDNA; all binding experiments were performed in buffer B. The addition of 41-bp dsDNA decreases fluorescence intensity. The protein concentration is 5.0 μmexcit = 292 nm and (λexcit = 345 nm) and a 25–40 μm dsDNA stock was used. (C) Fluorescence quenching titration of the Fad35R with 41-bp dsDNA. Two titrations were performed with 41-bp dsDNA; (•) and (○). The data were fitted to a two identical sites binding model (Eqn 1) and the solid line represents the best fit to the data, yielding Kobs = 3.0 (0.1) × 107 m−1; ▪, PhoP-Box dsDNA used as nonspecific DNA. (D) EMSA for the binding of indicated concentrations of Fad35R to 5′-end-labelled 10 nm oligonucleotide-based DNA (lanes 1 to 8) (reaction conditions were as described in the Materials and methods). Protein–DNA complexes were detected by autoradiography. Open arrowheads indicate unbound DNA and filled arrowheads indicate a shifted protein–DNA complex. (E) ITC analysis of the interaction between Fad35R (40.0 μm; syringe) and 41-bp dsDNA (2.0 μm; cell). Lower panel: integrated heat responses per injection from (A) plotted as normalized heat per mole of injectant. The solid line represents the best fit of the data to a single-site binding model with parameters Kobs = 1.4 (0.1) × 107 m−1.

The extent of tryptophan fluorescence quenching upon DNA binding was monitored (Fig. 2B, C). Analyses of binding isotherm indicates that the Fad35R homodimer binds to DNA with a 1 : 1 molar ratio with Kobs∼ 3.0 ± 0.2 × 107 (0.3) m−1. The affinity of Fad35R for the promoter DNA is similar to that estimated for the TetR–TetO interaction [30]. The specificity of promoter binding by Fad35R was tested by examining the interaction of Fad35R to nonspecific DNA, a canonical PhoP-box motif. Fad35R binds to PhoP-box very weakly and does not show saturation, even at a very high protein concentration (Fig. 2C). To confirm direct physical interaction in the presence of competing nonspecific DNA, electromobility gel shift assay (EMSA) experiments were performed. As shown in Fig. 2D, the addition of increasing amounts of purified Fad35R to the [32P]-end labelled 41-bp DNA fragment in the presence of competing nonspecific DNA resulted in the retardation of DNA fragment migration in native polyacrylamide gel.

Determination of the biochemical properties, including the energetics of interactions, is important for understanding the mechanism of promoter recognition. To examine the energetics of protein–DNA interactions, we performed isothermal titration calorimetry (ITC) experiments at 25 °C (Fig. 2E). The binding of Fad35R comprises an exothermic reaction and follows a simple 1 : 1 binding isotherm, consistent with fluorescence quenching experiments. This indicates that Fad35R may not oligomerize as was observed for EthR. Analyses of the binding reaction yield the thermodynamic parameters; Kobs∼ 1.4 ± (0.1) × 107m−1Gobs ∼ −40 ± 0.7 kJ·m−1); ΔHobs ∼ −106 ± 12 kJ·m−1; TΔS∼ 64 kJ·m−1. The molar enthalpy observed for the binding of Fad35R dimer to promoter DNA is almost two-fold higher than the molar enthalpy observed for the binding of QacR to IR1 (ΔHobs ∼ −60 kJ·m−1) [37, 38].

Surface plasmon resonance (SPR) studies and the kinetics of Fad35R–DNA interactions

Real-time binding kinetics using the SPR technique has provided mechanistic information on protein–DNA interactions [39]. To further understand Fad35R binding to its promoter region, we immobilized 5′-biotin labelled DNA on a streptavidin-coated chip and monitored the difference signal after subtraction of any nonspecific binding of Fad35R to the chip. Time-dependent binding and dissociation of Fad35R to promoter DNA at varied protein concentrations are shown in Fig. 3A. The linear increase and quick saturation of the resonance signal indicates that the association rate constant for Fad35R binding to promoter DNA is very high and that Fad35R dissociates very slowly. Slow dissociation was clearly evident from the dissociation phase (Fig. 3A) and, even after 10 min of dissociation in the buffer, a significant amount of Fad35R is still bound to DNA. Analyses of association kinetics at higher Fad35R concentrations (30–100 nm) shows that Fad35R binds rapidly, and the estimated binding parameters are obtained (Fig. 3B,C and Table 1). Equilibrium constants estimated from real-time kinetics experiments are similar to the equilibrium constants determined from steady-state fluorescence experiments.

Figure 3.

 SPR analysis of the Fad35R interaction with its cognate dsDNA, 41-bp dsDNA. All binding experiments were performed at 25 °C, buffer B. (A) The plot of the SPR difference resonance signal of Fad35R binding to 5′-biotin-41-bp dsDNA monitored at a series of Fad35R concentrations. (B) Kinetics of Fad35R binding at 25 °C. The association data were fit to single-site model for the binding of one Fad35R homodimer. (C) The dissociation phase of SPR data in which Fad35R dissociates from the 41-bp dsDNA is plotted versus time.

Table 1.   Kinetic and equilibrium parameters for the binding of Fad35R to 41-bp dsDNA determined by SPR at 25 °C. Values of Ka were calculated using Ka = ka /kd.
[Fad35R] (nm) k a,1 (m−1·s−1) k d,1 (s−1) K a,1 (m−1)
 102.9 × 104 (0.1)4.2 × 10−4 (0.1)6.9 × 107 (0.1)
 202.7 × 104 (0.1)4.3 × 10−4 (0.1)6.3 × 108 (0.1)
 252.6 × 104 (0.1)4.0 × 10−4 (0.1)6.5 × 107 (0.1)
 303.9 × 104 (0.1)4.9 × 10−4 (0.1)7.9 × 107 (0.1)
 404.1 × 104 (0.1)4.7 × 10−4 (0.1)8.7 × 107 (0.1)
 506.0 × 104 (0.1)5.3 × 10−4 (0.1)1.1 × 108 (0.1)
 755.2 × 104 (0.1)8.5 × 10−4 (0.2)6.1 × 107 (0.1)
1007.8 × 104 (0.2)5.0 × 10−4 (0.1)1.6 × 108 (0.1)
2509.0 × 104 (0.1)6.3 × 10−4 (0.2)1.4 × 108 (0.1)

The kinetics of the formation of protein–DNA complexes is studied as a function of temperature and salt. The overall association constants increase with temperature, yielding temperature-dependent activation energy. From the temperature dependence of the association rate constants of protein–DNA interactions (Table 2), an Arrhenius plot of ln(kon) versus 1/T was obtained and the activation energy, Ea, was evaluated from the plot. Analysis of the Arrhenius plot yielded an Ea for the Fad35R–DNA interaction of approximatey 8 kJ (Fig. 4B). Furthermore, we studied the salt (NaCl) dependence of the interaction, aiming to understand the nature of the forces that drive the interaction. All experiments were performed at a fixed Fad35R concentration (30 nm) at 25 °C (Fig. 4A). The interesting feature of the NaCl-dependent kinetics is that both amplitude and rates of association increase as the NaCl concentration is raised from 50 to 300 mm, although they decrease as the NaCl concentration is increased further (Fig. 4C). Analysis of the kinetic data shows that the association rate constant attains a maximum value at 300 mm, and thereafter it falls off rapidly (Table 3).

Table 2.   Effect of temperature on kinetic constants (ka,1 and kd,1) and estimated equilibrium association constant (Ka,1).
  k a,1 (m−1·s−1) k d,1 (s−1) K a,1 (m−1)
10 °C
 30.0 nm1.6 × 104 (0.1)2.2 × 10−4 (0.1)7.3 × 107 (0.1)
15 °C
 30.0 nm2.2 × 104 (0.1)3.4 × 10−4 (0.1)6.5 × 107 (0.1)
20 °C
 30.0 nm4.6 × 104 (0.1)3.2 × 10−4 (0.1)2.7 × 108 (0.1)
25 °C
 30.0 nm6.1 × 104 (0.1)2.3 × 10−4 (0.1)2.6 × 108 (0.1)
30 °C
 30.0 nm9.2 × 104 (0.1)2.7 × 10−4(0.1)3.4 × 108 (0.1)
Figure 4.

 Temperature and salt dependence of Fad35R–41-bp dsDNA interactions. (A) Kinetics study of Fad35R (30 nm) at different temperatures. (B) Arrhenius plot of apparent rate constant as a function of 1/T. Activation energy estimated from the slop of the liner fit is 8 kJ·mol−1 (30 nm). (C) Association kinetics of Fad35R–41-bp dsDNA interaction as a function of NaCl concentrations: (a) 50 mm, (b) 150 mm, (c) 300 mm, (d) 450 mm and (e) 600 mm.

Table 3.   Effect of salt concentration on kinetic constants (ka,1 and kd,1) and estimated equilibrium association constant (Ka,1). NA, not available.
[NaCl] (mm) k a,1 (m−1·s−1) k d,1 (s−1) K a,1 (m−1)
 506.9 × 103 (0.1)3.3 × 10−4 (0.1)2.0 × 107 (0.1)
1008.8 × 103 (0.1)2.3 × 10−4 (0.1)3.8 × 107 (0.1)
3001.4 × 105 (0.1)1.0 × 10−4 (0.5)1.4 × 109 (0.2)
4007.0 × 103 (0.1)2.0 × 10−4 (0.1)3.5 × 107 (0.1)
600NANANA

Fad35R binds tetracycline and FA derivatives

To test whether Fad35R can recognize tetracycline antibiotics and other ligands of FA metabolism, we examined the interaction of Fad35R with multiple ligands. First, a variety of antibiotics, including tetracycline were used. In all the experiments, the tryptophan fluorescence signal of Fad35R was used to monitor the protein–ligands interaction. The addition of tetracycline quenched tryptophan fluorescence and the extent of quenching was used to construct the binding isotherm (Fig. 5A). To understand the specificity of Fad35R for tetracycline, titrations were performed with nontetracycline family antibiotics, such as ampicillin and streptomycin. The results obtained demonstrate that both ampicillin and streptomycin do not show any significant binding even at millimolar concentrations, suggesting that Fad35R does not recognize these antibiotics (Fig. 5B). To understand the nature of FadR35–tetracycline interactions, binding studies were performed at different NaCl concentrations (Fig. 5C). Analysis of the binding isotherm shows that, as a result of increasing the NaCl concentration from 50 to 300 mm, the affinity for tetracycline is not altered significantly (Kd∼ 0.6 μm at 50 mm and Kd∼ 0.3 μm at 300 mm) for tetracycline.

Figure 5.

 Characterization of ligand binding properties. (A) Scan of fluorescence emission at different concentrations of tetracycline. The addition of tetracycline decreases fluorescence intensity. The protein concentration is 5.0 μmexcit = 292 nm and λexcit = 345 nm) and 50–150 μm tetracycline stock was used. (B) Fluorescence quenching titration of the Fad35R with tetracycline, streptomycin and ampicillin: (•) tetracycline (0.22 mm stock); (○) ampicillin (5 mm); and (□) streptomycin (5 mm). (C) Fluorescence quenching titration of the Fad35R with tetracycline in buffer A. Titrations were performed in the same buffer with different NaCl concentrations; (⋄) 50 mm; (•) 100 mm; (○) 150 mm; and (×) 300 mm. The data were fitted to a two identical sites binding model (Eqn 3) and the solid line represents the best fit to data. (D) Fluorescence quenching titration of citrate to Fad35R. The solid line represents the best fit of the data to a two identical sites binding model (Eqn 1) with parameters Kobs = 0.33 (0.1) × 103 m−1. (E) Binding of palmityl-CoA to Fad35R and associated fluorescence quenching. The binding constant was estimated to be Kobs = 1.6 (0.1) × 104 m−1 (Kd∼ 62.3 μm). (F) Effect of palmitoyl-CoA on DNA-binding properties of Fad35R. The reactions contained a radiolabelled 41-bp DNA (20 nm), purified His6-tagged Fad35R protein (1.5 μm) and palmitoyl-CoA (concentrations as indicated) as a potential regulator. Filled arrowheads indicate band shifts produced by the Fad35R–DNA complex and open arrowheads indicate the release of free probe back as a function of palmitoyl-CoA concentration, suggesting the derepression of Fad35R by fatty acyl-CoA.

In addition to Fad35, a continuous array of genes (ScoA-citE operon) involved in FA metabolism is found upstream of Fad35R (Fig. 1). Unlike Fad35, which has its own promoter region, the other seven genes shown in Fig. 1 are under the control of single promoter region, with the last gene coding for citrate lyase (CitE), which is necessary for the regeneration of acetyl-CoA. Activated FAs catalyzed by Fad35-like genes are known to bind FadR with a high affinity and to regulate gene expression [19,21]. In addition, long-chain FAs such as palmitic acid negatively regulate (whereas citrate positively regulates) the activity of AccA1 and AccA2 [40]. Therefore, we tested the binding of citrate and activated FAs to Fad35R. The results obtained indicate that both citrate and palmitoyl-CoA (PCA) bind with binding affinities of ∼3.0 ± 0.2 mm and 62.3 ± 5.0 μm, respectively (Fig. 5D, E). Although the binding data for PCA cannot be described by a simple 1 : 1 binding, binding was clearly noticeable. To test whether the binding of PCA would have any effect on the DNA-binding properties of Fad35R, we performed EMSA studies in the presence of PCA. We used sufficiently high concentration of Fad35R where a maximum shift was observed in the absence of PCA (Fig. 5F). Interestingly, as the concentration of PCA is increased, the binding of Fad35R to DNA was abolished, indicating that activated long-chain FAs such as PCA can weaken the affinity of Fad35R for its promoter DNA (Fig. 5F). The results obtained in these studies confirm that Fad35R can bind multiple ligands and that the DNA-binding properties of Fad35R are sensitive to levels of the products of FA metabolism.

Discussion

Although a general response mechanism exists for all characterized TetR family members, differences in the DNA-binding mechanisms have been observed from both structural and biochemical studies [7,32,41]. Different DNA recognition mechanisms adopted by structurally homologous proteins of the same family suggest the existence of a functional disparity in both ligand and DNA recognition by these family members, and such differences would be critical for the regulatory actions of these proteins. It was observed that both TetR and QacR exhibit the same level of specificity in binding to their respective promoter elements, although they achieve this specificity via different mechanisms [7,37,41]. Limited biochemical and structural studies on a few members of this widely distributed family are insufficient to enable a deciphering of the detailed features of the regulatory mechanism(s) adapted by TetR family members.

In the present study, we purified the previously uncharacterized TetR family regulator, Fad35R, from Mtb and characterized its ligand- and DNA-binding properties. The identified Fad35R gene is located upstream of Fad35 and consists of an independent intergenic region, IG1 (Fig. 2A). The Fad35R-binding region encompasses canonical −10 sequences and the binding of Fad35R to this region would block RNAP binding. The present study shows that Fad35R binds to this promoter DNA containing a 24-bp palindromic motif with high affinity and the specificity of binding is further confirmed by EMSA studies. The binding constant determined from kinetic studies for Fad35R and promoter DNA interactions [Kobs,avg∼9.2 × 107 m−1 (Kd∼ 11.1 nm)] is similar to the affinity estimated for the TetR–tetO1 interaction (Kobs∼ 7.6 × 107 m−1) and is also comparable to the binding affinity of QacR, a TetR family repressor from Staphylococcus aureus, to its operator DNA (Kobs∼ 2.0 × 107 m−1) [30,31]. Our model structure indicates that Fad35R has an HTH DNA-binding amino terminal domain and a variable C-terminal ligand-binding domain and thus is classified as a typical member of TetR family. Residues Pro37 and Tyr40 are predicted to bind DNA in alignment with their counterpart residues of EthR and may serve as specificity determining residues [28].

The results of our ligand-binding studies suggest that Fad35R can bind tetracycline, activated FAs and small molecules such as citrate. The binding of these molecules may have physiological relevance, as shown in the present study. Binding of activated FA decreases the affinity of Fad35R for DNA. Fatty-acyl derivatives are known ligands for FadR, the transcriptional regulator that controls the expression of genes involved in FA metabolism [19–21]. Therefore, it is possible that the identified Fad35R may be the homologue of FadR in Mtb, and it may work as a transcriptional switch by responding to intracellular levels of activated FAs such as palmityl-CoA. The predicted Fad35R regulon, as shown in Fig. 1, consists of three intergenic elements (IG1, IG2 and IG3) and the genes encoded by Fad35R regulon include enzymes involved in FA synthesis, degradation and the tricaroboxylic acid cycle. Products of FadE19 (acetyl-CoA dehydrogenase) and AccA1/D1 (acetyl-CoA carboxylases) control the steps of β-oxidation and biosynthesis of FA, respectively [1,26]. The citrate lyase, citE, which catalyzes the formation of acetyl-CoA from citrate and acetate, plays a main role in controlling the flux of acetyl-CoA into tricaroboxylic acid [1,40]. The functional coordination of the Fad35R regulon and the relationship between the components are shown in Fig. S2). Activation of FA by enzymes present in the Fad35R regulon is the first committed step in FA oxidation. We demonstrate that Fad35R binds to metabolites synthesized by genes in the regulon and also show that palmityl-CoA, a potential substrate of Fad35, can alter the DNA-binding activity of Fad35R (Fig. 6). Both QacR and ActR bind different ligands in different configurations with a significant flexibility in the ligand-binding pocket [8,34,37]. It is possible that Fad35R may also have such extended plasticity for recognizing chemically different molecules. A systematic understanding of ligand-binding specificities of Fad35R and their effect on promoter binding may be necessary to dissect out the transcriptional wiring of FA metabolism.

Figure 6.

 A schematic model depicting the activated fatty acid mediated regulation of the Fad35 gene by Fad35R. (A) Under low levels of activated FA, Fad35R remains unliganded. Unliganded Fad35R binds DNA with high affinity and represses gene expression. (B) Under elevated FA levels, activated FA binds to the ligand binding site of Fad35R and dissociates Fad35R from the DNA by reducing DNA binding affinity. FA binding traps the Fad35R in a binding incompetent conformation, thus allowing the expression of Fad35.

Thermodynamic data obtained as a function of salt concentration are used to understand the physicochemical mechanism of binding. The increase in binding affinity as the salt concentration is raised to 300 mm is different from that expected for regular protein–DNA interactions. The increase in salt concentrations is known to reduce the affinity for DNA as a result of the release of ions from the protein–DNA interface [29,41]. By contrast, in a few systems for which binding constants are increased at higher salt concentrations, it was shown that binding was accompanied by the removal of water molecules from the buried hydrophobic surface area [41,42]. The crystal structure of TetR in complex with cognate DNA shows that the TetR binding interface is hydrophobic and very few water molecules were found [42]. Ordered water molecules bound to a pre-bound state are released into bulk solvent, increasing the net entropy of the system, thus favouring binding. A previous study explained the effect of salt concentration on the activity of water in the bulk solvent [42]. Water activity in the bulk solvent is reduced at higher salt concentrations, although binding of Fad35R to its promoter DNA leads to the release of buried water molecules, which in turn increase the water activity in the bulk solvent. However, a further increase in salt concentration results in a significant loss of binding affinity. The modelled structure of Fad35R and the crystal structures of other TetR family regulators show few electrostatic interactions between the side chains of protein (Lys68 and Arg58) and phosphates of the DNA backbone. Kinetic data indicate that association rate is decreased to 100-fold at 400 mm and a sudden decrease in SPR signal amplitude, as well as binding affinity, is observed at a concentration of NaCl > 450 mm. Analysis of kinetic data > 400 mm NaCl is difficult because of the very low amplitude and low affinity. The cooperative decrease of the DNA-binding affinity of Fad35R at higher salt concentrations (but increased affinity at lower salt concentrations) indicates that the specificity of interaction is achieved via two different types of forces, with their respective contributions being dependent on the ionic strength of the medium. Such dual specificity availed by Fad35R may also imply that a sudden increase in salt concentrations as a result of stochastic fluctuations or temporal high flux may up-regulate the expression of FadD35 and other genes.

Although the interaction between TetR and TetO has been studied in detail, much variability exists in terms of their response to signal and DNA-binding modes in multiple TetR systems [19,27,32]. In the absence of information on the FadR homologue in Mtb, the present study provides the first biochemical characterization of a potential FadR homologue, Fad35R, which binds to promoter region of Fad35 and alters its DNA binding in the presence of activated FAs. To uncover the role of the Fad35R-mediated signalling cascade(s), further studies employing structural, biochemical and genetic methods are required.

Materials and methods

Materials

All chemicals and reagents were of analytical reagent grade and were obtained from Sigma (St Louis, MO, USA) and Invitrogen BioServices India Pvt. Ltd (Whitefield, Bangalore, India). All oligonucleotides used in the present study were of analytical quality and were obtained from Midland Certified Reagent Company (Midland, TX, USA) or Sigma (St Louis, MO, USA).

Protein expression and purification

Fad35R gene was amplified from Mtb (H37Rv) genome using the primers (forward, 5′-TATTGCTAGCGTGACAGCGTCCGCCC-3′; reverse, 5′-TATTGAATTCCTATAGACAACGATCCGCG-3′). The nucleotide sequence of Fad35R is 100% identical with respect to both H37Ra (MRA_2532) and H37Rv (Rv2506). The amplified gene product was digested with NheI and EcoRI and ligated into pET28a(+) vector. For the expression of Fad35R, E. coli, BL21 (DE3) was used as expression host. E. coli strains were grown at 37 °C on LB medium containing kanamycin (50 μg·mL−1). Protein expression was induced by 1 mm isopropyl thio-β-d-galactoside and the induction was carried out at 37 °C for 3 h at 220 r.p.m. Cells were harvested, resuspended in buffer A (50 mm Tris, 150 mm NaCl, 10% glycerol, 0.1 mm dithiothreitol) and lysed by sonication at pulse rate of 10 s ‘on’; 20 s ‘off’ for 5–6 min, repeated five times. The lysate was centrifuged at 11 400 g. for 30 min and supernatant was passed through Ni-nitrilotriacetic acid affinity chromatography. N-terminally His-tagged Fad35R was eluted from the Ni-nitrilotriacetic acid column using elution buffer (50 mm Tris, 150 mm NaCl, 10% glycerol, 250 mm imidazole, pH 8.0). The His-tag was removed by thrombin cleavage and Fad35R was purified by passing the cleavage mixture through Ni-nitrilotriacetic acid/benzamidine sepharose columns. The purified FadR was dialyzed against buffer A and applied on to Hiprep 16/60 Sephacryl S-200 column (GE Healthcare, Little Chalfont, UK) equilibrated with the same buffer. The size-exclusion column was calibrated as described (Doc. S2). The size-exclusion purified Fad35R protein was monitored by 12% SDS/PAGE followed by Coomassie Brilliant Blue R-250 staining. The purity of protein was found to be > 98%.

Measurement of protein and DNA concentration

The oligonucleotide containing Fad35R promoter region (41 bp) was synthesized and purified using denaturing gel and reverse phase HPLC. Both top and bottom strands were dissolved in the buffer A and dialyzed, and the concentrations of each strand were determined spectrophotometrically using the molar extinction coefficient calculated from their sequence. The oligonucleotides were mixed in a 1 : 1 ratio to a final concentration of 10.0 μm and heated to 95 °C for 5 min followed by slow cooling to room temperature, allowing them to reanneal gradually. The concentration of gel filtration purified Fad35R was determined using the molar extinction coefficient (ε = 8480 m−1) calculated from the sequence. Both protein and DNA were dialyzed versus respective buffers before the experiments.

Determination of secondary structure by CD

CD measurements were carried out with a JASCO-810 spectropolorimeter (Jasco, Tokyo, Japan) equipped with a Peltier type temperature controller (PTC-348W). Far-UV spectra were obtained in a quartz cuvette with a 1-mm light path length and each spectrum obtained was an average of 10 scans. The ellipticity of protein CD spectra is reported as the mean residue ellipticity (deg·cm−2·dmol−1). In the case of Fad35R–tetracyline complexes, the spectrum for protein is subtracted. Deconvolution of far UV-CD spectra was performed using jwsse32 software (Jasco).

Dynamic light scattering studies

Dynamic light scattering studies were performed using a Delsa Nano instrument (Beckman Coulter, Beckman Coulter, Fullerton, CA, USA) equipped with 512 channels. An He-Ne laser was used with 45 mW of power and the angle of scattering was 165°. Measurements were performed with a 100-μm aperture using quartz cuvette containing 1.8 mL of sample in a thermostated cell holder. The autocorrelation function of the scattering data was analyzed to obtain the translation diffusion coefficient of the protein using:

image(1)

Where τ is the auto correlation time and is defined as:

image(2)

Where D is the diffusion coefficient and q is the scattering vector.

Fluorescence titration measurements

Titrations of Fad35R with dsDNA were examined by monitoring the intrinsic tryptophan fluorescence of Fad35R using a Varian spectrofluorimeter (Varian Inc., Palo Alto, CA, USA). Experiments were performed in buffer B (20 mm Tris, 20 mm NaCl, 10% glycerol). The excitation wavelength and emission wavelengths for monitoring Fad35R-DNA interaction were 292 and 345 nm, respectively. Slit widths were set to 5 nm for all experiments and the photomultiplier tube voltage was adjusted to achieve the maximum signal for a given protein concentration. All experiments were performed at 23.0 ± 1 °C. Initial readings of both the protein, Fprotein,0 and buffer, Fbuff,0 were taken, with F0 = Fprotein,0 − Fbuff,0 defined as the initial fluorescence of the sample. The sample cuvette was then titrated with aliquots of dsDNA and mixed, equilibrated for 2–3 min before measurement. Data points from five such measurements were averaged to obtain Fobs,i. All measurements were corrected for dilution (Eqn 3), and inner filter effects (Eqns 4, 5) as described previously [43,44]:

image(3)

where Fi,cor is inner filter corrected fluorescence, Vo is the initial volume and Vi is the volume after each addition, fo is the initial fluorescence and fi is the fluorescence after each addition. The inner filter correction, C, was empirically determined as described previously [44]. For samples with low absorbance values as used in the experiments, C is simplified to:

image(4)
image(5)

where εp and εd are extinction coefficients of protein and DNA, and PT,i and DT,i are the total protein and DNA concentrations after each addition. The relative fluorescence quenching upon dsDNA binding is defined as:

image(6)

Binding of DNA to Fad35R was analyzed using a binding model that describes the binding of dimeric Fad35R to the DNA.

image(7)

where n is the number of binding sites, Qobs is the fluorescence quenching corresponding DNA bound Fad35R, D is the concentration of free DNA in solution and Kobs is the association constant for Fad35R-DNA interaction. Experimental data were fit to appropriate models using a nonlinear least squares method.

ITC

ITC experiments were performed in the indicated buffer A using a nano-ITC (TA Instruments, Newcastle, DE, USA). Both the Fad35R and dsDNA(s) were dialyzed extensively versus reaction buffer. All samples and buffer solutions were degassed at room temperature before use. The cell volume was 1.0 mL and the syringe volume was 250 μL. The volumes of injections used for the titration experiments were 10.0 μL and continuous stirring at 300 r.p.m. was used for mixing the reactants. Experiments were carried out by either titrating Fad35R (15–40 μm) (in the syringe) to the dsDNA (−3 μm) or vice versa. Control experiments were performed to determine the heat of dilution for each injection by injecting the same volumes of Fad35R. The data were analyzed using software provided by the manufacturer (TA Instruments). We used the binding model that considers equal affinities for two monomers (n = 2) of the homodimer and the heat released upon interaction is described by Eqn (8).

image(8)

where Qitot is total heat after the ith injection, V0 is the volume of calorimetric cell, Kobs is the observed equilibrium constant; n represents the number of binding sites and ΔH is the molar enthalpy. Estimates of Kobs and ΔH were obtained by fitting the experimental data to the model and the best-fit parameters were selected based on the lowest chi-squared values.

EMSA

PAGE-purified oligonucleotides were purchased from Sigma. The oligonucleotide concentrations were determined from absorbance at 260 nm, using calculated extinction coefficients. The 41-bp DNA fragments used to assess sequence-specific DNA binding by Fad35R were 5′-TACGTTAGTGACGATTAACCGAAGTGTCCAGCATGAGTCGT-3′ and its complement 5′-ACGACTCATGCTGGACACTTCGGTTAATCGTCACTAACGTA-3′ (the direct repeat sites are underlined). dsDNA probes were prepared as described above. DNA probes for the EMSA were generated by 5′-32P labelling of oligonucleotides using T4 polynucleotide kinase (New England Biolabs, Beverly, MA, USA) and [γ-32P] ATP (BRIT, Mumbai, India). Unincorporated nucleotides and the labelled oligonucleotide were separated using a Sephadex-G50 quick-spin column (GE Healthcare). Binding reactions (10 μL) contained DNA (20 nm) and His-Fad35R in 20 mm Hepes-Na+ (pH 7.5), 50 mm NaCl, 1 mm EDTA, 200 μg of BSA·mL−1, 10% glycerol, 1 mm dithiothreitol, 10 mm MgCl2 and 200 ng of sheared herring sperm DNA (20 μg·mL−1). Incubation was performed for 20 min at 20 °C. The reactions were analyzed by electrophoresis using nondenaturing 7% polyacrylamide gel [acrylamide-bisacrylamide (40 : 1.1) in 1× Tris-borate-EDTA buffer]. The gels were run at 70 V for 4–5 h at 4 °C, dried and visualized by autoradiography using a phosphorimager.

Modelling the tertiary structure of Fad35R

The nucleotide derived amino acid sequence of Fad35R was compared with the nonredundant database using blastp available at NCBI [45]. The multiple sequence alignment of the retrieved sequences was carried out using clustalx [46] with default values for gap opening and extension penalties. The crystal structure of TetR family regulatory protein RHA5900 (Protein Data Bank code: 2IBD) served as the template for the modelling of Fad35R using modeller 9v7 [47]. Out of 10 models generated, the best model was selected on the basis of DOPE score. The soundness and stereochemical validation of the model were executed using procheck [48].

SPR

We monitored the protein–DNA interaction in real time with a BIAcore3000 (BIAcore AB, Uppsala, Sweden) using a streptavidin-linked sensor chip. All experiments were performed in buffer B. Biotin-labelled dsDNA (5′-biotin-TACGTTAGTGACGATTAACCGAAGTGTCCAGCATGAGTCGT-3′) was immobilized on the streptavidin-coated sensor chip channel 2 or 4, which were used for monitoring the binding, and flow cell 1 or 3 lacking the immobilized DNAs were references. Control experiments were performed by testing the nonspecific binding of Fad35R to channels 1 and 3 and showed a < 3% change in signal. 5′-biotin-41-bp dsDNA was immobilized in channels 2 or 4 by constant injection of 20 μL·min−1 over 100–150 s. The DNA bound chip was washed with buffer and concentration-dependent experiments were initiated by the injection of purified thrombin-cleaved Fad35R into channels 2 or 4 (10–500 nm) in the running buffer (20 μL·min−1). The reference channel (1 or 2) subtracted difference signal was used for the analysis. After the difference signal was constant, buffer lacking Fad35R was used to initiate the dissociation phase. The experimental data were analyzed as described below using origin 5.0 (OriginLab Corporation, One Roundhouse Plaza, Suite 303, Northampton, MA 01060, USA). We used the following single-site binding model to fit the experimental data:

image(9)

where C is the concentration of Fad35R dimer, t is time (s), R is maximum response (response units), ka is the on-rate and kd is the off-rate.

Acknowledgements

This work was supported by Council of Scientific and Industrial Research, CSIR, India. S.A., V.S. and G.B. received salary support from Department of Biotechnology, DBT, India. A.K.S., M.M. and M.D. are recipients of research fellowship from CSIR. We thank Dr Alok Mondal for reading and editing the manuscript.

Ancillary

Advertisement