The evolution of transcriptional regulators through the recruitment of DNA-binding domains by enzymes is a widely held notion. However, few experimental approaches have directly addressed this hypothesis. Here we report the reconstruction of a plausible pathway for the evolution of an enzyme into a transcriptional regulator. The BzdR protein is the prototype of a subfamily of prokaryotic transcriptional regulators that controls the expression of genes involved in the anaerobic degradation of benzoate. We have shown that BzdR consists of an N-terminal DNA-binding domain connected through a linker to a C-terminal effector-binding domain that shows significant identity to the shikimate kinase (SK). The construction of active synthetic BzdR-like regulators by fusing the DNA-binding domain of BzdR to the Escherichia coli SKI protein strongly supports the notion that an ancestral SK domain could have been involved in the evolutionary origin of BzdR. The loss of the enzymatic activity of the ancestral SK domain was essential for it to evolve as a regulatory domain in the current BzdR protein. This work also supports the view that enzymes precede the emergence of the regulatory systems that may control their expression.
Citation: Durante-Rodríguez G, Mancheño JM, Rivas G, Alfonso C, García JL, Díaz E, et al. (2013) Identification of a Missing Link in the Evolution of an Enzyme into a Transcriptional Regulator. PLoS ONE 8(3): e57518. https://doi.org/10.1371/journal.pone.0057518
Editor: Frances M. Sladek, Univeristy of California Riverside, United States of America
Received: September 10, 2012; Accepted: January 22, 2013; Published: March 19, 2013
Copyright: © 2013 Durante-Rodríguez et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the Ministry of Economy and Competitiveness of Spain [BIO2009-10438, CSD2007-00005, BIO2008-04478-C03-03, BFU2010-17929, CSD-2006-00015]. GD-R was the recipient of a FPI fellowship from the Ministry of Economy and Competitiveness of Spain. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Regulation of transcription through the action of small molecules that directly bind to a transcription factor is widespread in all life forms. A large number of transcriptional regulators contain a DNA-binding domain fused to an effector-binding domain. Binding of the effector results in a conformational change, which influences the properties of the transcription factor and, accordingly, results in activation or repression of transcription .
The effector-binding protein domains of transcriptional regulators appear to have evolved by distinct selective forces. In some cases, the effector-binding protein domains appear to derive from catalytic proteins, which may or may not retain the active site residues in their binding pockets during evolution and thus could possibly behave as bifunctional proteins . There are also a few examples of transcriptional regulators that might have evolved from enzymes that have lost their catalytic activity. In eukaryotes, we could mention the Gal80 regulator involved in the catabolism of galactose in Saccharomyces cerevisiae and Kluyveromyces lactis , and TAFII150 protein in Drosophila melanogaster which controls transcription mediated by the RNA polymerase (RNAP)-II . In prokaryotes, the HutC regulator of Pseudomonas putida, involved in histidine utilization , and the FarR regulator of Escherichia coli, which controls the expression of Krebs cycle genes and responds to fatty acids , contain effector-binding domains similar to chorimaste lyases . However, the evolutionary pathways giving rise to these regulators have not yet been reproduced in the laboratory.
Anaerobic pathways provide reliable model scenarios to study protein evolution since they may preserve relics of ancient events occurring on early anaerobic Earth. In an effort to understand the evolution of regulators, we have reconstructed the most probable evolutionary pathway followed by the BzdR regulator. The BzdR protein of the facultative anaerobe β-Proteobacterium Azoarcus sp. CIB is the prototype of a new subfamily of transcriptional regulators. This regulator controls the expression of genes involved in the aerobic or anaerobic degradation of benzoate .
The predicted domain organization of BzdR consists of an N-terminal region (NBzdR, residues 1–90), homologous to the DNA-binding domain of members of the helix-turn-helix (HTH)-XRE transcriptional regulator, connected through a linker sequence to a C-terminal region (CBzdR, residues 131–298), This C-terminal region shows 23% sequence identity with the E. coli shikimate kinase I (SKI) (aroK gene product). In addition CBzdR conserves the SKI P-loop-containing nucleoside triphosphate hydrolase fold, the Walker-A motif and the Gly present in the Walker B-motif of purine nucleotide-binding proteins, and recognizes the inducer molecule benzoyl-CoA (Figure 1) , . In this work, we have experimentally reproduced the most likely evolutionary pathway followed by the BzdR protein. First, we show that NBzdR and CBzdR are real functional domains able to bind DNA and the effector molecule, respectively. We then constructed functionally active synthetic BzdR-like regulators by fusing the DNA-binding domain of BzdR to E. coli SKI protein as the effector-binding domain. The observed functionality of these synthetic regulators provides strong support for a role of an ancestral SK enzyme in the evolutionary origin of the BzdR protein.
Diagram showing the modular architecture of the BzdR protein, its NBzdR, NBzdRL and CBzdR domains, and the Q1, Q2 and Q1ΔL chimeras. The N-terminal domain (NBzdR), C-terminal domain (CBzdR) and linker region of BzdR are indicated in orange, blue and green, respectively. The E. coli SKI enzyme is shown in violet, and the Asp168Ala substitution in the Q2 chimera is indicated by an asterisk in red.
Materials and Methods
Strains, Plasmids, Growth Conditions, and Molecular Biology Procedures
The E. coli strains and plasmids used in this work are listed in Table 1. The construction of recombinant plasmids featuring the BzdR modules and the chimeras Q1, Q1ΔL and Q2 is detailed in the supplementary materials. E. coli cells were grown on Luria-Bertani (LB)  or M63  medium supplemented with the appropriate carbon source at 37°C. Anaerobic growth in LB medium using 10 mM nitrate as the final electron acceptor was achieved as previously reported . Standard molecular biology techniques were performed as described previously .
β-Galactosidase activities were measured using permeabilized cells as described by Miller . Shikimate kinase assays were performed according to previously established procedures  with the modifications detailed below. Plasmid pJCD-PN was used as as supercoiled PN template for in vitro transcription assays conducted as previously described .
Gel Retardation and DNase I Footprinting Assays
The PN DNA probe was obtained and mixed with the purified proteins at the concentrations indicated in each assay according to a previously described procedure .
Recombinant Plasmid Constructions
To clone the NBzdR, NBzdRL and CBzdR domains, NbzdR, NBzdRL and CbzdR fragments were first PCR amplified from pECOR7 plasmid using the oligonucleotide pairs 5′HisReg (5′-CGGGATCCTTTCCAACGATGAGAACTCATCAC-3′; engineered BamHI site underlined)/N1BzdR (5′-GGGAAGCTTTCACTCCGCCTCCTCGCGCACG-3′; engineered HindIII site underlined), 5′HisReg (see above)/N2BzdR (5′-GGGAAGCTTTCACCTGCGCGCGCTTCGCCCC-3′; engineered HindIII site underlined) and 5′C2 (5′-CGGGATCCTCAGCGTGCCAGGACTTCGAGG-3′; an engineered BamHI is underlined)/3′HisReg (5′-GGGAAGCTTTCAGCGTGCCAGGACTTCGAGG-3′; engineered HindIII site underlined). Next, the PCR amplified products were cloned into the pQE32 vector by BamHI/HindIII double digestion, giving rise to plasmids pQE32-His6NBzdR, pQE32-His6NBzdRL and pQE32-His6CBzdR, respectively. The NbzdR and NBzdRL fragments were also subcloned under the control of the Plac promoter into the pKC01 vector by EcoRI/HindIII double digestionyielding plasmids pCK01NBzdR and pCK01NBzdRL, respectively. The aroK gene that encodes the SKI enzyme was PCR amplified from the E. coli CC118 genome using 5′His-aroK oligonucleotides (5′-CGGGATCCTTGCAGAGAAACGCAATATCTTT-3′; engineered BamHI site underlined)/5′aroKquim (5′-AACTGCAGTTAGTTGCTTTCCAGCATGTGAATA-3′; engineered PstI underlined) and then cloned into the pQE32 vector by BamHI/PstI double digestion, giving rise to plasmid pQE32- His6SKI.
To construct the Q1 and Q1ΔL chimeras, NBzdR, NBzdRL and aroK were first PCR amplified from pECOR7 plasmid and aroK from the E. coli CC118 genome using the oligonucleotide pairs 5′HisReg (see above)/newN1quim (5′-GGGGTACCCTCCGCCTCCTCGCGCACG-3′; engineered KpnI site underlined) or 5′HisReg (see above)/5′N2qim (5′-GGGGTACCCCTGCGCGCGCTTCG-3`; engineered KpnI site underlined) and 3′aroKqim* (5′-GGGGTACCATGGCAGAGAAACGCAATATCTTT-3′; engineered KpnI site underlined)/5′aroKquim (see above), respectively. The NBzdR and NBzdRL amplified products were cloned into the pGEM-Teasy vector by BamHI/KpnI double digestion, giving rise to plasmids pGEMT-NBzdR and pGEMT-NBzdRL, respectively. The aroK amplified product was then cloned into the pGEMT-NBzdR and pGEMT-NBzdRL vectors by KpnI/PstI double digestion, giving rise to plasmids pGEMT-Q1ΔL and pGEMT-Q1. These constructions were digested using BamHI and PstI enzymes to subclone the Q1ΔL and Q1 genes with N-terminal His6-tag into the pQE32 plasmid, generating plasmid pQE32-His6Q1ΔL and pQE32-His6Q1.
The Q2 chimera containing the Asp168Ala substitution was constructed by directed PCR mutagenesis using pQE32-His6Q1 plasmid as template. Fragment A spannings from the start codon of aroK to the codon that encodes the Asp36 residue was PCR-amplified with oligonucleotides 3′aroKquim (anneals at the 5′-end of the aroK start codon) and 3′D36A (5′-AATCTCTTGAGCGGAATCGTAAAATTC-3′; anneals at the region spanning codon Asp36 of the aroK gene and replaces it with an Ala codon). Fragment B, spanning from the codon that encodes the Asp36 residue to the 3′-end of aroK, was PCR-amplified with oligonucleotides 5'D36A (5′-GAATTTTACGATTCCGCTCAAGAGATT-3′; anneals at the region spanning codon Asp36 of the aroK gene and replaces it with an Ala codon) and 5'aroKquim (anneals at the 3′-end of aroK). A second round of PCR amplification uses as templates the A and B fragments, and as primers the 3'aroKquim and 5'aroKquim oligonucleotides, which anneal at the start and stop codon of the aroK gene, respectively. The resulting aroK mutant gene that contains the Asp36Ala substitution was KpnI/PstI-double digested, replacing the wild-type aroK gene cloned in pQE32-His6Q1 to give rise to plasmid pQE32-His6Q2.
All constructions were confirmed by DNA sequencing.
Shikimate Kinase Activity Assays
Shikimate kinase activity was determined at 25°C by coupling the release of ADP to pyruvate kinase (EC 184.108.40.206) and lactate dehydrogenase (EC 220.127.116.11) reactions. Shikimate-dependent oxidation of NADH was monitored at 340 nm (ε = 6180 M−1 cm−1). The standard assay mixture contained (final concentration) 50 mM triethanolamine hydrochloride/KOH buffer, pH 7.0, 50 mM KCl, 5 mM MgCl2, 1.6 mM shikimate, 2.5 mM ATP, 1 mM phosphoenolpyruvate, 0.1 mM NADH, 3 units of pyruvate kinase (Sigma)/ml, 2.5 units of lactate dehydrogenase (Sigma)/ml, and the target protein to be assayed. Km values for shikimate were obtained from the Lineweaver-Burk plot . Although it has been described that E. coli SKI has a high Km (5–20 mM) for shikimate, this value was obtained using an SKI isolated as a heterologous mix of oligomers . Our results show that the His6-SKI has a Km for shikimate (0.4 mM) similar to that reported previously for the E. coli SKII . In the competition assays, benzoyl-CoA or AMP were added at a 10 mM concentration to the standard assay conditions, and shikimate kinase activity was compared to that obtained in the absence of these inhibitors.
Binding of benzoyl-CoA to CBzdR was determined by monitoring the change in protein fluorescence upon ligand addition. Measurements were obtained in an SML-Amicon 8000 spectrophotometer using quartz cuvettes and 0.2-cm excitation and 1-cm emission path lengths (λexcitation = 275 nm and λemission 312 nm; slit widths = 5 nm). Titrations were performed at 25°C by adding a benzoyl-CoA stock solution to 0.4 ml of 20 mM Tris-HCl, pH 8.0, 100 mM KCl containing CBzdR (1 µM initial concentration). The final ligand concentration ranged from 0 to 300 µM. Corrections were made for protein dilutions, background signal, and for the instrument response as indicated by the manufacturer. Control titrations with buffer alone produced no change in the emission signal after correlation. The Kd value for benzoyl-CoA in CBzdR was calculated by non-linear regression using the equation for a one-site saturation model ([L]b/[P]t = ([L]free/Kd+[L]free) implemented in the GraphPad Prism software (GraphPad software, San Diego); where [L]b, is the concentration of bound ligand; [P]t, the total protein concentration, and [L]free, the concentration of free ligand. The fraction of bound ligand was calculated as (Io−I)/(Io−Imin); where Io is the maximum CBzdR fluorescence intensity (no ligand added); I is the fluorescence intensity after addition of an aliquot of ligand, and Imin is the fluorescence intensity at saturating concentration of ligand (this value was estimated by nonlinear regression to a single hyperbolic decay curve). The data obtained in at least three experimental trials were averaged to calculate the dissociation constant.
Results and Discussion
NBzdR Is a Functional DNA-Binding Domain
To confirm our hypothesis that the molecular architecture of BzdR arises from the fusion of two individual functional domains, the NBzdR and CBzdR domains were expressed, purified and characterized as independent proteins (Figure 1). Purified NBzdR protein was able to bind to the operator region of the PN promoter, as illustrated by gel shift (data not shown) and DNase I footprinting experiments (Figure 2A). The expression of NBzdR in E. coli cells with a chromosomal PN::lacZ fusion elicited the same repression of the PN promoter observed in cells expressing the entire BzdR regulator (Figure 2B). In vitro transcription experiments also revealed the ability of NBzdR to inhibit the PN promoter (Figure 2C). The predicted five α helices of NBzdR constitute the minimal structural domain since deletion of helix 5 (residues 77 to 86) did not allow for the production in E. coli of a stable protein that could be detected by SDS-PAGE (not shown).
A. DNase I footprinting experiments were performed using the PN probe and the purified regulators His6-BzdR (control), His6-NBzdR, and His6-NBzdRL. The figure shows the results of footprinting assays conducted in the absence of the regulators (lane C), or the presence of 50, 100, or 200 nM of His6-NBzdR (lanes 1 to 3, respectively) or His6-NBzdRL (lanes 4 to 6, respectively). Lanes 7 and 8 are footprinting assays containing 50 and 100 nM of purified His6-BzdR. Lane AG shows the A+G Maxam and Gilbert sequencing reaction. Protected regions (I, II, and III) are indicated with brackets. The −10 box and the transcription initiation site (+1) of the PN promoter are also shown. Phosphodiester bonds hypersensitive to DNase I cleavage are indicated by asterisks. B. In vivo effect of the N-terminal domain of BzdR on the repression of the PN promoter. E. coli AFMCPN cells (containing a PN::lacZ fusion chromosome insertion of the) harboring plasmid pCK01BzdR (BzdR), pCK01NBzdR (NBzdR) or pCK01NBzdRL (NBzdRL) or the control plasmid pCK01 (-), were grown anaerobically in LB medium until the mid-exponential culture phase. β-galactosidase activity is expressed in Miller units. Results from three independent experiments (n = 3) and errors bars are shown. C. Effect of BzdR, NBzdR and NBzdRL on in vitro transcription from PN. Multiple-round in vitro transcription reactions were performed using the pJCD-PN plasmid template, which produces a 184-nt mRNA from PN (arrow), 50 nM E. coli RNAP, and 20 nM Fnr* activator. Transcription reactions were carried out in the absence of repressor (lane -) or presence of 40 nM purified His6-BzdR (lanes BzdR), His6-NBzdR (lanes NBzdR) or His6-NBzdRL (lanes NBzdRL) and in the absence (-) or presence (+) of 2 mM benzoyl-CoA.
We observed that when NBzdR was fused to the linker region, designated NBzdRL (residues 1 to 130; Figure 1), it was also able to repress the activity of PN both in vivo and in vitro (Figure 2B and Figure 2C). Moreover, NBzdRL showed greater protection to this promoter (operator regions I and II) than NBzdR and similarly to that observed for complete BzdR protein (Figure 2A). Gel shift experiments also revealed that NBzdRL was able to retard the migration of the PN promoter at lower concentrations than those required for NBzdR (data not shown). These results suggest that the BzdR linker sequence, whose main function is to facilitate interdomain communication , might also play a role in stabilization and binding of the N-terminal domain to the target DNA.
BzdR is a dimer in solution . To test whether this property is conferred by the N-terminal domain, we examined the association state of the NBzdR and NBzdRL proteins. Sedimentation velocity experiments revealed that both proteins sedimented as single species with s-values of 1.8 and 2.3 S, respectively. Also, sedimentation equilibrium experiments confirmed that these species were dimers (Figure S1), demonstrating that BzdR dimerization is an intrinsic property of the N-terminal domain.
We should mention that benzoyl-CoA was unable to rescue the repression caused by NBzdR or NBzdRL over the PN promoter both in vitro (Figure 2C) and in vivo (data not shown), confirming that NBzdR is not the target of the inducer molecule and, therefore, behaves as a super-repressor. All of these results suggest that NBzdR is a true functional domain that retains the ability of native BzdR protein to bind the PN promoter and repress its activity.
CBzdR Is a Functional Benzoyl-CoA-Binding Domain
The modelled 3D structure of CBzdR suggests the presence of a deep groove that could interact with the benzoyl-CoA inducer molecule , triggering a conformational change in BzdR protein . To establish if effector binding was specific of the C-terminal domain of BzdR, we investigated whether CBzdR would interact with benzoyl-CoA. As expected, fluorescence emission by CBzdR was dramatically modified upon incubation with benzoyl-CoA (Figure S2). Based on the structural model for CBzdR , fluorimetry data could be fitted to a single binding site, indicating the formation of a weakly bound CBzdR-benzoyl-CoA complex (Kd 203±35 µM) as previously observed for the parent BzdR regulator (Kd 157±58 µM) . This suggests that benzoyl-CoA induces a conformational change in CBzdR and that it therefore, retains its effector-binding ability in the absence of the NBzdR domain, thus behaving as an independent functional domain.
Analytical ultracentrifugation experiments revealed that CBzdR behaves as a single species with an s-value of 1.8 S, compatible with a globular monomer (Figure S3). The observation that CBzdR is a monomer that undergoes conformational changes in the presence of its cognate ligand (benzoyl-CoA), resembles the behaviour of SKs which undergo extensive conformational changes in the presence of their substrates, i.e., ATP and shikimate . However, despite this conserved similarity with SKs, CBzdR showed no detectable shikimate kinase activity (Figure S4A).
The Q1 Chimera: a Bifunctional Regulator that Controls the PN Promoter in vitro
The results described above demonstrate that the organization of BzdR is modular and shaped by two functional domains, i.e., NBzdR, which binds to the target promoter and CBzdR, which recognizes benzoyl-CoA as its inducer molecule. This molecular architecture suggests that BzdR may have arisen from the fusion of an HTH-XRE DNA-binding domain to a protein structurally resembling SK or an ancestor of this enzyme. The structural similarity between ATP and shikimate, the two substrates of SK, and the two ends of the benzoyl-CoA molecule, i.e., 3P-ADP and the benzoyl group, suggest that the latter could also be recognized by the SK enzyme. In agreement with this hypothesis, we previously observed that benzoyl-CoA inhibits (by 44%) E. coli SKI activity in a similar manner to AMP (34% inhibition), a well-known SK inhibitor , suggesting that benzoyl-CoA can be considered a newly identified ligand of the well-studied SK enzyme.
To unravel the evolutionary pathway that could have given rise to the current BzdR regulator, we used a protein engineering approach and constructed the chimeric protein Q1 by the fusing NBzdRL to E. coli SKI protein (Figure 1). Interestingly, Q1 showed SK activity yielding a Km for shikimate of 0.6 mM, similar to that exhibited by E. coli SKI purified at our laboratory (Km 0.4 mM) (Figure S4A). The enzyme activity of the SK domain in the Q1 chimera was also demonstrated in vivo in an E. coli aroK aroL double mutant strain lacking SK activity that is unable to grow in minimal medium since it cannot synthesize aromatic amino acids . E. coli aroK aroL cells transformed with a plasmid that expressed the Q1 protein were observed to grow in glycerol-containing minimal medium (Figure S4B), indicating that Q1 was able to rescue the lack of SK activity in the mutant strain.
In analytical ultracentrifugation experiments, Q1 protein behaved as a single species with an s-value of 3.7 S consistent with a protein dimer (Figure S3). This finding is in agreement with the fact that this chimeric protein contains the NBzdR homodimeric domain, and reveals that dimerization does not prevent the catalytic activity of the monomeric SKI enzyme.
The Q1 chimera was able to bind to the three operator regions of the PN promoter conferring similar protection to that shown by the BzdR regulator (Figure 3A). In addition, it was capable of inhibiting PN promoter activity in vitro (Figure 3B, lanes 2-3). Accordingly, we can conclude that both domains are fully and separately active in the Q1 chimera.
(A) DNase I footprinting experiments performed out using the PN probe and the purified regulators His6-BzdR (control) and His6-Q1. The figure shows the results of footprinting assays conducted in the absence of the regulators (lane C), or presence of 25, 50, 100, and 200 nM of His6-BzdR (lanes 1 to 4, respectively) or His6-Q1 (lanes 5 to 8, respectively). Lane AG shows the A+G Maxam and Gilbert sequencing reaction. Protected regions (I, II, and III) are indicated by brackets. The -10 box and the transcription initiation site (+1) of the PN promoter are also shown. Phosphodiester bonds hypersensitive to DNase I cleavage are indicated by asterisks. (B) In vitro effect of the Q1 and Q1ΔL chimeras on the activity of PN. Multiple-round in vitro transcription reactions were performed using the pJCD-PN plasmid template, which produces a 184-nt mRNA from PN (arrow), 50 nM E. coli RNA polymerase (RNAP), and 20 nM Fnr* activator . Transcription reactions were conducted in the absence of chimeric regulator (lane 1) or presence of 25 nM (lane 2) or 50 nM (lanes 3–12) of purified His6-Q1 or His6-Q1ΔL proteins. Shikimate (S) or shikimate plus ATP (S+ATP) were added at 1 mM (lanes 4 and 7), 2 mM (lanes 5 and 8) or 4 mM (lanes 6, 9, 11 and 12). (C) In vivo effect of BzdR (control) and the Q1 and Q2 chimeras on the activity of the PN promoter. β-galactosidase activity (in Miller units) of E. coli MC4100 cells harboring plasmid pSJ3PN (PN::lacZ) and the plasmids pCK01BzdR (BzdR), pCK01Q1 (Q1), pCK01Q2 (Q2), or the control plasmid pCK01 (-). Cells were grown anaerobically until mid-exponential phase in LB medium supplemented, when indicated, with 5 mM shikimate (S). Results from three independent experiments (n = 3) and errors bars are shown. (D). In vitro effect of the Q2 chimera on the activity of PN. In vitro transcription reactions were performed as in panel B, in the absence of Q2 (lane 1) or presence of 25 nM (lane 2) or 50 nM (lanes 3–9) of purified His6-Q2 protein. Shikimate (S) or shikimate plus ATP (S+ATP) were added at 1 mM (lanes 4 and 7), 2 mM (lanes 5 and 8) or 4 mM (lanes 6 and 9).
It is well known that SK enzymes undergo profound conformation changes when they interact with their substrates, ATP-Mg2+ and shikimate. The binding of shikimate leads to a structural state (loaded conformation) that, in the presence of ATP-Mg2+, is maintained for a short period of time sufficient to trigger the catalytic reaction to produce 3P-shikimate. The enzyme subsequently recovers its starting structural state (unloaded conformation) (Figure S5) , . Since the Q1 chimera is able to bind and inhibit the activity of the PN promoter in the absence of shikimate and ATP (Figures 3A and 3B), we anticipated that the presence of shikimate alone could provoke a conformational change in the SK domain, i.e., to a loaded-like conformation, that would lead to the release of the Q1 protein from the PN promoter. When increasing concentrations of shikimate were added to the in vitro transcription assay mix, significant PN promoter de-repression was observed (Figure 3B, lanes 4–6). Moreover, this de-repression was not observed when the Q1 chimera was replaced with Q1ΔL protein (Figure 3B, lanes 10–12), a new chimera lacking the linker region between the NBzdR and SK domains (Figure 1). These data suggest that the Q1 chimera is a bifunctional regulator that controls PN promoter activity and responds to shikimate as an inducer molecule in vitro by generating a conformational change transmitted to the DNA-binding domain through the linker region. Interestingly, benzoyl-CoA was also shown to offset the Q1-dependent repression of PN (Figure 4, lane 5) with an efficiency similar to that shown by shikimate (Figure 4, lane 4) but lower than that observed for the native BzdR (Figure 4, lane 11). As previously noted for BzdR , benzoate could not mimic the inducer effect of its CoA derivative (benzoyl-CoA) in the Q1 protein (Figure 4, lane 6), emphasizing the importance of the CoA moiety in effector recognition.
In vitro transcription reactions were run as in Figure 2B, in the absence of repressor proteins (lane 2) or presence of 40 nM purified His6-Q1 (lanes 3 to 6), His6-Q2 (lanes 7 to 10) or His6-BzdR (lanes 11 to 14). The ligands shikimate (S) (lanes 4, 8, 12), benzoyl-CoA (BCoA) (lanes 5, 9, 11), and benzoate (Bz) (lanes 6, 10) were added at 1 mM; ATP was added at 4 mM (lane 13). Lanes -, no ligand added. Lane 1, control assay without RNAP.
The observation that CBzdR lacks SK activity (Figure S4A) raises the question of whether the existence of such activity in the Q1 chimera may interfere in vivo with effector-dependent de-repression of the PN promoter. This possibility was confirmed by monitoring PN promoter activity in E. coli cells expressing the PN::lacZ reporter fusion and the bifunctional Q1 chimera. It was observed that Q1 protein inhibited the activity of PN as efficiently as the control BzdR repressor even after the addition of shikimate to the culture medium (Figure 3C). These data strongly suggest that the presence in the cell of both substrates of SK, i.e., shikimate and ATP-Mg2+, prevents de-repression of the PN promoter. The lack of PN induction by shikimate in vivo was also confirmed in vitro when ATP-Mg2+ and shikimate were added together to the transcription assay (Figure 3B, lanes 7–9). These results are in agreement with the rapid conformational changes observed in the SK enzyme after binding of shikimate and ATP-Mg2+ during catalysis , , and suggest the SK domain of the Q1 protein does not remain in the loaded-like conformation (Figure S5) for sufficient time to trigger the release of the repressor from the target promoter.
Taken together, these results confirm that the SK activity of the Q1 regulator prevents effector-specific de-repression of the target promoter by shikimate under physiological conditions.
Evolving the Q1 Chimera towards a BzdR-like Regulator: the Q2 Chimera
The results reported above suggest that CBzdR may have evolved from an SK-like enzyme that lost its enzyme activity while maintaining its benzoyl-CoA binding ability to efficiently control the PN promoter. To simulate the predicted molecular history of BzdR in the laboratory, we engineered a Q1 derived protein containing a single mutation that abolished its enzyme activity.
Since SK activity is strictly dependent on Mg2+, and binding of this metal requires one Thr and two Asp conserved residues , we engineered a new synthetic protein, a so-called Q2 chimera (Figure 1), by replacing the Asp168 residue of the Q1 protein (corresponding to Asp36 of SKI) with an Ala residue. This substitution avoids Mg2+ binding while maintaining the overall 3D structure of the SK domain. As expected, the purified Q2 protein showed no SK activity (Figure S4A), but was still able to efficiently inhibit the in vitro activity of the PN promoter (Figure 3D, lanes 2–3). The addition of shikimate or benzoyl-CoA (but not benzoate) to the in vitro transcription reaction released the repression of the PN promoter (Figure 3D, lanes 4–6; Figure 4, lanes 8–10) suggesting that the SK domain of Q2 retained its ability to bind both ligands and transfer the corresponding conformational changes to the NBzdR domain. Interestingly, shikimate also attenuated the repression exerted by Q2 even in the presence of ATP-Mg2+ (Figure 3D, lanes 7–9). Moreover, in vivo experiments revealed that the addition of shikimate to the culture medium of E. coli cells expressing the PN::lacZ reporter fusion and the Q2 chimera reduced the repression exerted by Q2 (Figure 3C), a behaviour not observed for the Q1 chimera. These results indicate that when the PN promoter is repressed by the Q2 chimera it can be de-repressed in vivo by shikimate.This observation is consistent with the fact that the catalytically inactive SK domain of the synthetic Q2 regulator can adopt a permanent loaded-like conformation under physiological conditions when shikimate is present, thus triggering the release of the repressor from the target promoter.
All these results point to a role of the Q2 chimera as a synthetic transcriptional regulator that can control the PN promoter by recognizing shikimate and benzoyl-CoA as inducer molecules. Shikimate and ATP were not effective CBzdR ligands since they were unable to significantly impair the BzdR-dependent inhibition of PN (Figure 4, lanes 12–13). Thus, the Q2 chimeric protein represents a BzdR-like regulator with a broader effector-binding range and is a likely intermediate in the evolutionary history of the current BzdR protein.
Several lines of research have suggested that early enzyme forms exhibited broad substrate specificity, or “substrate ambiguity”, and later diverged to give rise to the specific enzymes we know today , , , . Proteins acquire new functions without losing their original function, and gene duplication may follow the emergence of a new function, rather than initiate it . In effect, it is known that the SK domain can occur as an independent enzyme, e.g., AroK and AroL proteins in bacteria , or as a domain of multifunctional enzymes, e.g., AROM protein in Ascomycetes . Moreover, the duplication of genes coding for SK-like domains is very common in plants and some of these proteins have acquired novel functions, e.g., SKL1, which seems essential for chloroplast biogenesis and SKL2, which encodes a key protein for species-adaptive molecular evolution in Arabidopsis thaliana . Although these duplications are less frequent in prokaryotes, E. coli also has two SKs, and one of these (SKI) could play a role in an as yet unknown cell function since aroK mutants are resistant to mecillinam . In contrast with examples proposed previously whereby an ancestral SK evolved into a novel enzyme, our work shows that the SK domain could also have evolved to give rise to a regulatory protein. We suggest that SK behaves as a “stem” protein domain, able to interact with different substrates and thus facilitating the acquisition of new functions over time. In effect, a similar SK domain has also been detected in a putative regulator (plpp0115 gene product) in the γ-proteobacterium Legionella pneumophila .
We report the reconstruction of an analogue of an ancestral protein that could have generated one of the domains of the current BzdR regulator, whose modular architecture comprises two independent functional domains, i.e., a DNA-binding domain (NBzdR) and an effector-binding domain (CBzdR) (Figure 5). The fact that we were able to construct and characterize a bifunctional Q1 regulator strongly suggests that fusion of an HTH-XRE DNA-binding domain to an SK domain was a major step in the evolution of BzdR. In addition, this SK domain was found to behaves as a stem protein domain able to interact with different substrates. The finding that SK recognizes benzoyl-CoA provides additional evidence for our hypothesis. However, the loss of enzyme activity of the SK domain may have been a critical event to improve the efficiency of the ancestral BzdR regulatory protein. Indeed, this was indicated by the catalytically-inactive Q2 chimera which, unlike Q1, was able to respond in vivo to the inducer molecules. This, synthetic Q2 protein constitutes the first shikimate-dependent transcriptional regulator described so far. Our finding that it is possible to construct a transcriptional regulator by fusing an appropriate DNA-binding domain to an enzyme that recognizes the target compound, paves the way for a novel rational approach to create new à la carte regulatory proteins. During the course of evolution of the BzdR regulator, the Q2-like ancestral protein proposed likely suffered additional mutations preventing its recognition of shikimate while increasing its binding affinity for benzoyl-CoA, the inducer and first intermediate of the catabolic pathway controlled by BzdR.
The red arrows represent fusion between a shikimate kinase ancestor and a DNA-binding domain of the HTH-XRE family. This fusion rendered a protein equivalent to the Q1 chimera. A series of mutations led to the loss of enzyme activity (chimera Q1) and finally to increase in the affinity of the protein for the inducer benzoyl-CoA. Sk: shikimate; 3-P-Sk: 3-phosphoshikimate; Bz-CoA: benzoyl-CoA.
Besides recreating a plausible pathway for the evolution of an enzyme into a transcriptional regulator, our findings also support the notion that enzymes preceded the emergence of the regulatory systems that control their expression.
Study of the oligomerization state of NBzdR (A) and NBzdRL (B) proteins in solution. Sedimentation equilibrium data (grey dots) and best fit analysis assuming a protein dimer (black line) and monomer (red line) species. The lower panels show the difference between estimated values and experimental data for protein dimers (residuals). The data indicate that NBzdR and NBzdRL proteins are dimers, demonstrating that BzdR dimerization is an intrinsic property of the N-terminal domain.
Conformational changes in CBzdR induced by benzoyl-CoA binding. Intrinsic fluorescence of His6-CBzdR as a function of benzoyl-CoA (Bz-CoA) concentration. Data points represent the decrease in the His6-CBzdR (7.5 µM) fluorescence emission maximum (312 nm) expressed in arbitrary units (a.u.) upon excitation at 275 nm in the presence of increasing concentrations of Bz-CoA. Insets, fitting Bz-CoA binding to His6-CBzdR [P] to a single site model. This result suggests that benzoyl-CoA induces a conformational change in CBzdR and, therefore, that CBzdR retains its effector-binding ability in the absence of the NBzdR domain, thus behaving as an independent functional domain.
Study of the oligomerization state of CBzdR and Q1 proteins in solution. Sedimentation coefficient distribution c(s) corresponding to the sedimentation velocity of purified His6-CBzdR (broken blue line) and His6-Q1 (solid blue line) proteins. The protein concentration distribution pattern (c(s)) and sedimentation coefficient (S) are represented in the graph. The standard s-value of the protein did not change significantly with protein concentration over the range examined (1-30 µM). The Q1 protein behaved as a single species with an s-value 3.7 S consistent with a protein dimer, whereas CBzdR behaved as a single species with an s-value of 1.8 S compatible with a globular monomer.
Shikimate kinase activity of E. coli SKI and CBzdR, Q1 and Q2 proteins. A. SK activities are shown for the purified His6-SKI (red line), His6-CBzdR (orange line), His6-Q1 (black line) and His6-Q2 (green line). Despite its conserved similarity with SKs, CBzdR showed no detectable shikimate kinase activity. B. Growth curves for E. coli ALO807strain defective in the aroL and aroK genes harboring plasmid pQE32-His6Q1 (expresses the His6-Q1 protein) (red lines) or the control pQE32 plasmid (green lines). Cells were grown in M63 minimal medium supplemented with 30 mM glycerol in the presence (continuous lines) or absence (discontinuous lines) of 0.4% (w/v) casamino acids. The results of a single representative experiment are shown, and values were reproducible in three separate experiments with standard deviation values of <10%. These results demonstrate that the Q1 protein retains SK activity in the SK defective mutant strain.
Diagram showing the conformational changes the SK enzyme undergoes during a catalytic cycle. The SK enzyme shows the LID region (white), the NB region (green) responsible for ATP-recognition, the ESB region (blue) involved in shikimate recognition, and the RC region (red), which constitutes the rest of the SK structure. A. “Unloaded conformation” of the SK enzyme. B. “Loaded conformation” of SK in the presence of shikimate. c. Structural conformation of the SK in the presence of ATP-Mg2+ and shikimate. Thick black arrow represents the release of the two products of the reaction, 3P-shikimate and ADP, and the recovery of the initial “unloaded conformation” of the SK enzyme. Adapted from Hatmann et al. 
The authors thank to V. de Lorenzo for critical reading of the manuscript and inspiring discussions, A. Valencia for the technical assistance and Secugen S.L. for help with the gene sequencing procedures.
Conceived and designed the experiments: MC GD-R ED. Performed the experiments: GD-R JMM CA. Analyzed the data: MC ED GR. Contributed reagents/materials/analysis tools: JMM GR CA. Wrote the paper: MC ED JLG.
- 1. McAdams HH, Srinivasan B, Arkin AP (2004) The evolution of genetic regulatory system in bacteria. Nat Rev Gene 5: 169–178.
- 2. Commichau FM, Stülke J (2008) Trigger enzymes: bifunctional proteins active in metabolism and in controlling gene expression. Mol Microbiol 67: 692–702.
- 3. Lohr D, López J (1995) GAL4/GAL80-dependent nucleosome disruption/deposition on the upstream region of the yeast GAL1-10 and GAL80 genes. J Biol Chem 270: 27671–27678.
- 4. Aravind L, Koonin EV (1998) Eukaryotic transcription regulators derive from ancient enzymatic domains. Curr Biol 8: R111–R113.
- 5. Allison SL, Phillips AT (1990) Nucleotide sequence of the gene encoding the repressor for the histidine utilization genes of Pseudomonas putida. J Bacteriol 172: 5470–5476.
- 6. Quail MA, Dempsey CE, Guest JR (1994) Identification of a fatty acyl responsive regulator (FarR) in Escherichia coli. FEBS Lett 356: 183–187.
- 7. Barragán MJ, Blázquez B, Zamarro MT, Mancheño JM, García JL, et al. (2005) BzdR, a repressor that controls the anaerobic catabolism of benzoate in Azoarcus sp. CIB, is the first member of a new subfamily of transcriptional regulators. J Biol Chem 280: 10683–10694.
- 8. Durante-Rodríguez G, Valderrama JA, Mancheño JM, Rivas G, Alfonso C, et al. (2010) Biochemical characterization of the transcriptional regulator BzdR from Azoarcus sp. CIB. J Biol Chem 285: 35694–35705.
- 9. Sambrook J, Russell DW (2001) Molecular Cloning: A Laboratory Manual., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY
- 10. Miller JH (1972) Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
- 11. Millar G, Lewendon A, Hunter MG, Coggins JR (1986) The cloning and expression of the aroL gene from Escherichia coli K12. Biochem J 237: 427–437.
- 12. Durante-Rodríguez G, Zamarro MT, García JL, Díaz E, Carmona M (2006) Oxygen-dependent regulation of the central pathway for the anaerobic catabolism of aromatic compounds in Azoarcus sp. CIB. J Bacteriol 188: 2343–2354.
- 13. Lineweaver H, Burk D (1934) The determination of enzyme dissociation constants. J Am Chem Soc 56: 658–666.
- 14. De Feyter R (1987) Shikimate kinase from Escherichia coli K12. Methods Enzymol 142: 355–361.
- 15. Whipp MJ, Pittard AJ (1995) A reassessment of the relationship between aroK-aroL-encoded shikimate kinase enzymes of Escherichia coli. J Bacteriol 177: 1627–1629.
- 16. Vonhein C, Schlaiderer GJ, Schulz GE (1995) Movie of the structural changes during a catalytic cycle of nucleoside monophosphate kinases. Structure 3: 483–490.
- 17. Hartmann DM, Bourenkov GP, Oberschall A, Strizhov N, Bartunik HD (2006) Mechanism of phosphoryl transfer catalyzed by shikimate kinase from Mycobacterium tuberculosis. L Mol Biol 364: 411–423.
- 18. Løbner-Olensen A, Marinus MG (1992) Identification of the gene (aroK) encoding shikimic acid kinase I of Escherichia coli.. J Bacteriol 174: 525–529.
- 19. Krell T, Coggins JR, Lapthorn AJ (1998) The three-dimensional structure of shikimate kinase. J Mol Biol 278: 983–997.
- 20. Jensen RA (1974) Enzyme recruitment in evolution of new function. Annu Rev Microbiol 30: 409–425.
- 21. O'Brien PJ, Herschlag D (1999) Catalytic promiscuity and the evolution of new enzymatic activities. Chem Biol 6: R91–R105.
- 22. Peisajaovich S, Tawfik D (2007) Protein engineers turned evolutionists. Nat Meth 4: 991–994.
- 23. Galvão TC, Mencía M, de Lorenzo V (2007) Emergence of novel functions in transcriptional regulators by regression to stem protein types. Mol Microbiol 65: 907–919.
- 24. Aharoni A, Gaidukov L, Khersonsky O, Gould SM, Roodveldt C, et al. (2005) The ‘evolvability’ of promiscuous protein functions. Nat Genet 37: 73–76.
- 25. Coggins JR, Boocock MR, Chaudhuri S, Lambert JM, Lumsden J, et al. (1987) The arom multifunctional enzyme from Neurospora crassa. Methods Enzymol 142: 325–341.
- 26. Fucile G, Falconer S, Christendat D (2008) Evolutionary diversification of plant shikimate kinase gene duplicates. PLoS Genetics 4: e1000292.
- 27. Vinella D, Gagny B, Joseleau-Petit D, D'Ari R, Cashel M (1996) Mecillinam resistance in Escherichia coli is conferred by loss of a second activity of the AroK protein. J Bacteriol 178: 3818–3828.
- 28. Cazalet C, Rusniok C, Brüggemann H, Zidane N, Magnier A, et al. (2004) Evidence in the Legionella pneumophila genome for exploitation of host cell functions and high genome plasticity. Nat Genet 36: 1165–1173.
- 29. Silhavy TJ, Berman ML, Enquist LW (1984) Experiments with gene fusions. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
- 30. Ferrández A, García JL, Díaz E (2000) Transcriptional regulation of the divergent paa catabolic operons for phenylacetic acid in Escherichia coli. Characterization of a new aerobic hybrid patway. J Biol Chem 275: 12214–12222.
- 31. López-Barragán MJ, Carmona M, Zamarro MT, Thiele B, Boll M, et al. (2004) The bzd gene cluster, coding for anaerobic benzoate catabolism, in Azoarcus sp. strain CIB. J Bacteriol 186: 5762–5774.
- 32. Manoil C, Beckwith J (1985) TnphoA: a transposon probe for protein export signals. Proc Natl Acad Sci U S A 82: 8129–8133.
- 33. Ferrández A, Miñambres B, García B, Olivera ER, Luengo JM, et al. (1998) Catabolism of phenylacetic acid in Escherichia coli. Characterization of a new aerobic hybrid pathway. J Biol Chem 273: 25974–25986.