In Gram-negative bacteria, quorum sensing control of gene expression is mediated by transcription factors of the LuxR family, whose DNA-binding affinity is modulated by diffusible N-acyl homoserine lactone (AHL) signalling molecules. In Serratia sp. ATCC 39006 and the plant pathogen Erwinia carotovora ssp. carotovora (Ecc), the biosynthesis of the β-lactam antibiotic 1-carbapen-2-em-3-carboxylic acid (Car) is under quorum sensing control. This study has revealed that, uniquely, the LuxR family transcriptional activator CarR39006 from Serratia 39006 has no detectable affinity for cognate AHL molecules. Furthermore, CarR39006 was shown to be naturally competent to bind to its target promoter with high affinity, activate transcription and resist cellular proteolysis, and was unaffected by AHL signals. Experiments with chimeric proteins suggest that the C-terminal DNA-binding domain of CarR39006 may be responsible for conferring AHL independence. In contrast, we show that the homologous CarREcc protein binds to its 3O-C6-HSL ligand with high affinity, and that the highly conserved Trp-44 residue is critical for this interaction. Unlike TraR from Agrobacterium tumefaciens, CarREcc is not directly protected from cellular proteolysis by AHL binding, but via AHL-induced DNA binding. At physiological protein concentrations, AHL binding induces CarREcc to bind to its target promoter with higher affinity and activate transcription.
A diverse array of bacterial species regulate gene expression in response to population cell density, in a process termed quorum sensing (QS). Gram-negative proteobacteria sense population density using diffusible N-acyl homoserine lactone (AHL) molecules. AHLs act as ligands to modulate the DNA-binding ability of cytoplasmic transcription factors of the LuxR family, thereby controlling expression of key gene sets (Whitehead et al., 2001). LuxR family proteins are highly divergent in primary sequence, but all share conserved secondary structure and a common two-domain architecture (Fig. S1). The N-terminal domains of LuxR proteins contain an amphipathic AHL-binding pocket (Vannini et al., 2002; Zhang et al., 2002). In TraR from Agrobacterium tumefaciens, the cognate 3O-C8-HSL inducer is completely buried within this pocket, and forms four key hydrogen bonds with the highly conserved TraR residues Tyr-53, Trp-57, Asp-70 and Thr-129 (Vannini et al., 2002; Zhang et al., 2002) (Fig. S1). In the LuxR family transcriptional activators TraR, LuxR from Vibrio fischeri and LasR from Pseudomonas aeruginosa, binding of the AHL inducer drives the formation of homodimers (Choi and Greenberg, 1992; Qin et al., 2000; Zhu and Winans, 2001; Kiratisin et al., 2002; Schuster et al., 2004; Bottomley et al., 2007). Dimerization facilitates binding of a helix–turn–helix motif in the C-terminal domain of the LuxR protomers to palindromic DNA sequences upstream of target promoters, thereby recruiting RNA polymerase for transcription (Vannini et al., 2002; Zhang et al., 2002; Schuster et al., 2004; Urbanowski et al., 2004). AHL binding was also shown to protect TraR from degradation by the cellular proteases ClpP and Lon (Zhu and Winans, 2001).
The LuxR family transcriptional activator CarR controls production of the β-lactam antibiotic carbapenem (1-carbapen-2-em-3-carboxylic acid; Car) in the plant pathogen Erwinia carotovora ssp. carotovora (Ecc) and in Serratia ATCC 39006 (S39006) (Parker et al., 1982; Bycroft et al., 1988; McGowan et al., 1995; Cox et al., 1998). In Ecc, the 3O-C6-HSL inducer synthesized by the AHL synthase CarI interacts with the CarREcc protein, allowing it to bind to the promoter of the Car biosynthetic operon (carA-H) and activate transcription (McGowan et al., 1995; 1996; 1997; 2005; Welch et al., 2000). Unlike other LuxR proteins, data from in vitro experiments suggested that CarREcc may exist as a pre-formed dimer, which forms a higher-order multimer upon the addition of 3O-C6-HSL (Welch et al., 2000). In addition to its role as an AHL-dependent regulator, CarREcc was also reported to activate Car biosynthesis in the absence of AHL molecules, when overexpressed in trans (McGowan et al., 1995). This phenomenon, which is indicative of a mass-action effect caused by high concentrations of apo-CarREcc, has not been reported for any other LuxR protein. Even when strongly overexpressed, TraR, LuxR from V. fischeri and RhlR and LasR from P. aeruginosa all required their cognate AHL molecule to function (Zhu and Winans, 1999; Lamb et al., 2003; Schuster et al., 2004; Urbanowski et al., 2004).
Serratia ATCC 39006 possesses a homologous carA-H biosynthetic cluster under the transcriptional control of the CarR39006 protein (Cox et al., 1998; Thomson et al., 2000). CarR39006 and CarREcc share strong overall amino acid identity (59.3%) and similarity (75.0%) (Fig. S1). Carbapenem production in Serratia 39006 is completely dependent upon production of the short-chain C4-HSL signal by the SmaI synthase (Thomson et al., 2000). However, in an AHL-deficient smaI mutant, a subsequent mutation in the gene encoding a second LuxR homologue (the SmaR repressor) restored wild-type carbapenem production (Slater et al., 2003), suggesting that CarR39006 can activate carA-H transcription in the absence of AHLs. According to this model, QS control of Car production is imposed by the AHL-responsive SmaR protein, via transcriptional repression of the carR39006 gene (and the carA promoter itself) in the absence of AHLs (Slater et al., 2003; Fineran et al., 2005a). In support of this model, expression of carR39006in trans complemented the carbapenem production defect in an Ecc carREcc mutant, in the absence of AHLs (Cox et al., 1998). However, the mass-action effect observed with overexpressed CarREcc made the significance of this result unclear. It was also recently reasoned that CarR39006 might be inherently transcriptionally active, but could become deactivated by binding to cognate AHLs (Tsai and Winans, 2010), as has been proposed for EsaR from Pantoea stewartii (Schu et al., 2009).
The aim of this study was therefore to confirm and investigate the phenomenon of AHL independence in CarR39006, in comparison with its more extensively studied homologue CarREcc. We show that, uniquely, the Serratia CarR39006 protein has no detectable affinity for AHLs, but can still bind to its target promoter with high affinity and activate transcription in the absence of AHLs. We further show that, unlike TraR, protection of CarREcc and CarR39006 from cellular proteolysis relies on DNA binding. Finally, our data suggest that the C-terminal domain of CarR39006 may be responsible for conferring AHL independence.
CarR39006 has no detectable affinity for Erwinia or Serratia AHLs
First, we analysed the binding of CarREcc and CarR39006 to cognate AHL signals, using isothermal titration calorimetry (ITC). The TraR protein from A. tumefaciens forms a hydrogen bond between its Trp-57 residue and the carbonyl group of the AHL lactone ring (Vannini et al., 2002; Zhang et al., 2002). This H-bond is also formed between the equivalent Trp-60 residue of LasR from P. aeruginosa and its 3O-C12-HSL ligand (Bottomley et al., 2007; Zou and Nair, 2009). This functional tryptophan residue is very highly conserved among LuxR homologues, including CarREcc (Cox et al., 1998) (Fig. S1). In contrast, CarR39006 contains a cysteine at this position (Cys-44), which forms much weaker H-bonds than tryptophan (Fig. S1). It was previously hypothesized that this W44C substitution might confer AHL independence to CarR39006 (Slater et al., 2003). The remaining three H-bonding residues are conserved in CarR39006 (Fig. S1).
To test the importance of the strongly conserved tryptophan residue for AHL binding, the highly soluble His6-tagged AHL-binding domain of CarREcc (residues 1 to 167) was purified by Ni-NTA affinity chromatography, and ITC analysis demonstrated its strong affinity for the Ecc inducer 3O-C6-HSL (Fig. 1A). Based on the observed stoichiometry of the interaction, binding of 3O-C6-HSL to CarREcc1-167 was fitted to a two binding site model, probably reflecting one molecule of 3O-C6-HSL binding to each protomer in a CarREcc dimer (Welch et al., 2000) (Fig. 1A). The values for the calculated dissociation constants (Kd = 0.72 µM and 3.53 µM) are consistent with previous fluorescence quenching data (Welch et al., 2000). Site-directed mutagenesis of the Trp-44 residue to cysteine completely abolished any detectable interaction between CarREcc and 3O-C6-HSL (Fig. 1A). This strongly suggests that H-bonding by the Trp-44 residue is essential for AHL binding.
The AHL-binding ability of the N-terminal domain of CarR39006 was then compared with that of His6-CarREcc1-167. No detectable interaction was observed between the His6-CarR390061-167 protein and either 3O-C6-HSL or the Serratia C4-HSL ligand (Fig. 1B), suggesting that CarR39006 is unresponsive to the cognate Erwinia and Serratia AHL signals.
At physiological concentrations, CarREcc requires AHLs to activate transcription
To directly isolate the activity of CarREcc and CarR39006 in the absence of any endogenous host regulation, these proteins were expressed in the heterologous host Escherichia coli DH5α and assessed for their ability to activate transcription of their target carbapenem operon promoters in the presence of AHL inducers.
When expressed in E. coli DH5α from the pQE80L-derived plasmid pSP78 induced with 1 µM IPTG, hexahistidine-tagged CarREcc required 1 µM 3O-C6-HSL to activate transcription from the Ecc PcarA::lacZ transcriptional reporter plasmid pSP14 (Fig. 2A). No β-galactosidase activity was detected with either the empty pQE80L vector (Fig. 2A), or 3O-C6-HSL in the absence of His6-CarREcc (data not shown). His6-tagged CarREcc retained ∼ 85 % of the transcriptional activity of the untagged protein (data not shown). Non-cognate AHLs of varying acyl chain length and oxidation state were unable to induce His6-CarREcc to activate Ecc carA transcription (Fig. 2A), consistent with the observation that CarREcc has a low affinity for these signals (Welch et al., 2000). As previously reported, the non-cognate 3O-C8-HSL signal also induced His6-CarREcc to activate carA transcription, to a higher level than the endogenous 3O-C6-HSL (Welch et al., 2000) (Fig. 2A).
Binding to AHLs protects CarREcc from cellular degradation
Previous studies demonstrated that binding to AHLs protected the A. tumefaciens TraR protein from cellular proteolysis (Zhu and Winans, 2001). In contrast, MrtR from Mesorhizobium tianshanense and RhlR from P. aeruginosa were stably expressed, but unable to activate transcription, in the absence of AHLs (Lamb et al., 2003; Yang et al., 2009). To assess whether binding to cognate AHLs was having any stabilizing effect on the CarREcc protein in vivo, duplicate samples from the PcarA transcription experiment were separated by SDS-PAGE and probed with anti-His antibodies to detect His6-CarREcc. Consistent with the model for TraR, His6-CarREcc was only detected in E. coli DH5α cultures grown in the presence of 3O-C6-HSL and 3O-C8-HSL (Fig. 2A). Based on the experiments with TraR, it was hypothesized that binding to cognate AHLs protected His6-CarREcc from degradation by cellular proteases.
To test this hypothesis, a full-length version of the ‘signal-blind’ His6-CarREcc W44C mutant protein was expressed from pSP90 in E. coli DH5α carrying pSP14 (Fig. 2B). The signal-blind W44C variant was unable to resist cellular proteolysis or activate carA transcription in the presence of 3O-C6-HSL (Fig. 2B), suggesting that the ability to bind to AHLs is essential for CarREcc to function and resist degradation at physiologically relevant concentrations. When the wild type and W44C variants were expressed with higher IPTG concentrations, CarREcc protein levels were comparable via Western blotting, suggesting that the W44C mutation was not intrinsically attenuating CarREcc production or stability (Fig. S2). These data also do not support the previous hypothesis that the W44C substitution in CarR39006 might be responsible for conferring AHL independence (Slater et al., 2003).
When overexpressed, CarREcc accumulates and activates carA transcription in the absence of AHLs
Based on the AHL-dependent nature of carA transcription (Fig. 2A) it was predicted that induction of pSP78 with 1 µM IPTG resulted in approximately wild-type His6-CarREcc protein levels. We next investigated the previously reported phenomenon of AHL-independent carA transcriptional activation, by overexpressing His6-CarREcc. Consistent with previous studies in Ecc and with the data in Fig. 2A, in E. coli DH5α cultures induced with lower IPTG concentrations (0 µM and 1 µM) PcarA transcription was dependent upon 3O-C6-HSL (Fig. 3A). However, as IPTG induction increased to 2 µM, carA transcription was activated in the absence of AHL, to ∼ 60% of the level induced by 3O-C6-HSL. With 5 µM IPTG, Ecc carA transcription was identical in the presence or absence of 3O-C6-HSL. In each case, His6-CarREcc protein levels (assessed by Western blot) correlated with the increase in AHL-independent carA transcription. At higher IPTG induction levels, His6-CarREcc appeared to somehow overcome degradation by cellular proteases, and activate carA transcription.
AHL-independent Ecc carA transcription and protein accumulation at higher IPTG induction levels was also observed with the signal-blind His6-CarREcc W44C variant (Fig. 3B). This supports the hypothesis that AHL-independent CarREcc function is caused by a mass-action effect that is unrelated to the ability to bind to AHLs.
CarR39006 activates transcription and resists proteolysis in the absence of AHLs
We next tested the effect of AHLs on the transcriptional activity of the Serratia CarR39006 protein. In previous studies, it was unclear whether the apparent AHL-independent carA transcription in Serratia 39006 was the result of some inherent property of the CarR39006 protein, or due to other Serratia-specific regulatory factors or features of the Serratia carA promoter. To investigate AHL independence in the absence of potential complications from Serratia-specific regulation, CarR39006 was expressed in a heterologous host, E. coli DH5α. In this host background, transcription from the Serratia PcarA::lacZ transcriptional reporter plasmid pTA16 was activated by expression of His6-CarR39006 from pSP79 (Fig. 4A). However, unlike CarREcc, His6-CarR39006 activated transcription in the absence of AHLs at all IPTG concentrations tested, consistent with previous genetic studies in Ecc and Serratia (Cox et al., 1998; Fineran et al., 2005a). Moreover, the addition of 1 µM C4-HSL or 3O-C6-HSL to cultures did not significantly alter carA transcription (Fig. 4A). Western blotting also showed that the accumulation of the His6-CarR39006 protein was not altered by the addition of either C4-HSL or 3O-C6-HSL (Fig. 4A). His6-CarR39006 was also unresponsive to a broader range of non-cognate AHLs (Fig. 4B). Taken together, these data are consistent with a transcriptional activator that functions independently of AHLs.
Site-directed mutagenesis to convert the Cys-44 residue back to the highly conserved tryptophan slightly reduced the stability of His6-CarR39006, but did not restore AHL dependence (Fig. S3), strongly suggesting that the W44C substitution is not the sole cause of AHL independence in CarR39006.
CarR39006 has a naturally strong affinity for its target promoter
It has been demonstrated for other LuxR family proteins that binding to their cognate AHL increases their affinity for target DNA (Qin et al., 2000; Schuster et al., 2004; Urbanowski et al., 2004). Unlike other LuxR family proteins, the full-length His6-CarREcc and His6-CarR39006 proteins were estimated to be 60–70% soluble in the absence of AHLs, when expressed under low-temperature (16°C) conditions (data not shown). We were therefore able to assess the DNA-binding affinity of Ni-NTA purified His6-CarR proteins by in vitro electrophoretic mobility shift assays (EMSA), using digoxygenin (DIG)-labelled carA promoter fragments (Fig. 5).
In the absence of 3O-C6-HSL, His6-CarREcc displayed an intrinsic specific affinity for the Ecc carA promoter, with 100 nM His6-CarREcc forming a DNA–protein complex (Complex 1, Fig. 5A). Binding was assessed to be relatively weak, as only part of the DIG-labelled DNA was shifted. The weak affinity of apo-CarREcc for its promoter DNA is consistent with AHL-independent carA transcription when the protein was overexpressed (Fig. 3A). When 3O-C6-HSL was added, His6-CarREcc formed DNA–protein complexes at lower protein concentrations (20 nM), indicating enhanced affinity of CarREcc for the carA promoter (Fig. 5B). Addition of 3O-C6-HSL to the reaction mixture containing 40 nM and 100 nM His6-CarREcc also caused the formation of a more highly retarded DNA–protein complex (Complex 2, Fig. 5B). This complex may be compatible with carA DNA binding to an AHL-induced higher-order CarR multimer, as previously suggested (Welch et al., 2000). However, this EMSA method is not appropriate for accurate assessment of multimeric state. Alternatively, this DNA–protein complex could represent CarR dimers bound to two distinct sites in the same carA promoter fragment, perhaps facilitating transcription via DNA looping (De Carlo et al., 2006). Regardless of the CarREcc multimeric state, the more highly retarded DNA–protein complex completely shifted the carA DNA, indicating that AHL-bound CarREcc protein had a higher DNA affinity than apo-His6-CarREcc. Enhanced binding of CarREcc to carA promoter DNA correlates with activation of carA transcription in the presence of 3O-C6-HSL (Fig. 3A).
Like the wild-type protein, His6-CarREcc W44C had an intrinsic weak affinity for the carA promoter (Fig. 5C), but this was not altered by 3O-C6-HSL (Fig. 5D). This is consistent with a signal-blind CarREcc variant that is unresponsive to AHLs, and therefore locked in a state with low DNA-binding affinity.
In contrast, His6-CarR39006 displayed a strong intrinsic affinity for the Serratia carA promoter in the absence of AHLs, with 20 nM His6-CarR39006 forming a DNA–protein complex (Fig. 5E). In fact, the DNA-binding ability of His6-CarR39006 in the absence of AHLs closely matched that of His6-CarREcc in the presence of 3O-C6-HSL (Fig. 5B) consistent with the hypothesis that CarR39006 is a ‘locked-on’ version of CarREcc. Reactions containing 100 nM His6-CarR39006 formed the more highly retarded DNA–protein complex (Complex 2), formation of which required 3O-C6-HSL with His6-CarREcc. The affinity of His6-CarR39006 for the Serratia carA promoter was not altered by either C4-HSL or 3O-C6-HSL (Fig. 5F and G).
CarR protease resistance requires target promoter DNA
The ability of His6-CarREcc to bind to 3O-C6-HSL correlated with enhanced DNA affinity (Fig. 5B) but it was unclear how this was linked with protection from cellular proteolysis (Fig. 2). In TraR, AHL-mediated protease protection occurred in the absence of target promoter DNA, suggesting that AHL binding was protecting the protein directly (Zhu and Winans, 2001). Like the TraR model, 3O-C6-HSL could be directly protecting His6-CarREcc, thereby allowing it to accumulate and activate carA transcription. Alternatively, CarREcc-3O-C6-HSL complexes could be resisting proteolysis via enhanced binding to target DNA. In this second model, protection from proteolysis would only occur in the presence of promoter DNA.
To test if either of the two models was correct, His6-CarREcc accumulation was assessed in the absence of the carA promoter, by expressing pSP78 (encoding His6-CarREcc) in E. coli DH5α carrying the promoterless pRW50 progenitor plasmid instead of the PcarA::lacZ reporter pSP14. In the presence of 3O-C6-HSL, His6-CarREcc was only detectable by Western blot in E. coli cells containing the carA promoter (pSP14) but not in its absence (pRW50) (Fig. 6A). Similarly, His6-CarR39006 expressed from pSP79 was only detectable in the presence of the Serratia carA promoter (pTA16) (Fig. 6B). In fact, His6-CarREcc and His6-CarR39006 expressed in the presence of pRW50 required approximately 50- to 200-fold higher IPTG induction to accumulate to a similar level as when expressed with pSP14 or pTA16 (Fig 6C and D). This suggests that carA promoter DNA is involved in stabilizing the His6-CarR proteins.
More importantly, addition of 3O-C6-HSL did not significantly protect His6-CarREcc from proteolysis in the absence of target promoter DNA (Fig. 6C). This result disputes the model that CarREcc is directly protected from cellular proteolysis by AHL binding, and instead suggests that the protective action of 3O-C6-HSL may be mediated indirectly, via binding of His6-CarR proteins to the carA promoter. This model of CarR protease protection is strikingly different from that currently proposed for TraR (Zhu and Winans, 1999; 2001).
The C-terminal domain of CarR39006 may confer AHL independence
The C-terminal DNA-binding domains of CarREcc and CarR39006 share an amino acid identity of 67.4% (86.0% similarity). The two proteins are especially well-conserved in the predicted recognition helix (α9) of the helix–turn–helix DNA-binding motif (Fig. S4). CarR39006 expressed from a multicopy plasmid can also complement a carREcc mutation in Ecc 39048 (Cox et al., 1998). By expressing pSP78 in the presence of pTA16, and pSP79 with pSP14, we next showed that CarREcc and CarR39006 were functionally interchangeable, activating maximal transcription from the heterologous carA promoter (Fig. S5). Moreover, the response of each CarR protein to AHLs was retained while activating either carA promoter, suggesting that the ability of CarR39006 to activate carA transcription in the absence of AHLs was due to an inherent property of the protein itself (Fig. S5).
However, these data gave no clues as to the functional differences between the CarREcc and CarR39006 proteins that caused them to behave so differently. Given the strong sequence identity between the C-terminal DNA-binding domains of the two CarR proteins, it was previously hypothesized that AHL independence might be conferred by the N-terminal domain of CarR39006 (Slater et al., 2003). To test this hypothesis, the CarR39006 N-terminal domain (residues 1 to 158) was fused to the C-terminal domain of the CarREcc protein (residues 159 to 244) to form a CarR39006-Ecc domain chimera, and vice versa to create a CarREcc-39006 chimera.
To identify which domain of the CarR39006 protein was responsible for AHL independence, these CarR domain chimeras were expressed in E. coli DH5α carrying pSP14. The His6-CarREcc-39006 chimera expressed from pSP89 accumulated and activated carA transcription at low IPTG induction levels, and was not significantly affected by the addition of 3O-C6-HSL (Fig. 7A). This result was very similar to that induced by wild-type His6-CarR39006, suggesting that the C-terminal domain of CarR39006 may be conferring ligand independence, contrary to our previous hypothesis. Control experiments showed that the CarR39006 C-terminal domain alone could not activate carA transcription (data not shown). In contrast, the His6-CarR39006-Ecc chimera expressed from pSP88 was far less able to accumulate, or activate transcription (Fig. 7B). In fact, the His6-CarR39006-Ecc protein acted in a similar fashion to that of wild-type His6-CarREcc grown in the absence of 3O-C6-HSL, or His6-CarREcc W44C grown in the presence of 3O-C6-HSL (Fig. 3).
This study has furthered our understanding of the mode of action of the CarREcc and CarR39006 proteins. Our data confirm that the CarREcc protein binds to its 3O-C6-HSL ligand in vitro with high affinity, and that the hydrogen bond formed between 3O-C6-HSL and the highly conserved Trp-44 residue is critical for this interaction. The complete abolition of detectable AHL binding in the W44C mutant protein suggests that this key hydrogen bond may be required for the formation or stabilization of additional H-bonds between CarREcc residues and the AHL ligand. In vitro, AHL binding induced the CarREcc protein to bind to the carA promoter with higher affinity. In the heterologous host E. coli DH5α, enhanced DNA binding had the twin effect of preventing proteolytic turnover and activating carA transcription. The W44C variant of His6-CarREcc cannot bind to 3O-C6-HSL, and so remains in a low-affinity state, which cannot resist proteolysis or activate carA transcription at physiologically relevant concentrations. The mechanism by which DNA-bound CarREcc resists cellular proteolysis is unclear, but must differ significantly from the direct protection afforded to the TraR protein from A. tumefaciens by 3O-C8-HSL (Zhu and Winans, 1999; 2001).
This study has also confirmed the ability of CarREcc to bind to target DNA and activate transcription in the absence of AHLs, when overexpressed. This is in stark contrast to other activators of the LuxR family, in which the apo-form had no detectable affinity for target DNA (Qin et al., 2000; Schuster et al., 2004). The window between AHL-dependent and AHL-independent carA transcription was narrow, with mild overexpression of CarREcc sufficient to activate carA transcription. The AHL-independent behaviour of CarREcc when overexpressed also demonstrates that care must be taken when attempting to interpret the physiological mechanism of proteins based purely on in vitro data and the expression of physiologically artefactual protein concentrations from multicopy plasmid systems.
This study has also revealed that, uniquely, the previously uncharacterized CarR39006 protein from Serratia 39006 has no detectable affinity in vitro for AHL molecules, and that its ability to bind to its target carA promoter with high affinity, activate transcription and successfully resist cellular proteolysis is unaffected by AHLs. The results of in vivo experiments in the heterologous E. coli host closely match those obtained from several genetic studies in Serratia, implying that AHL independence is conferred by the CarR39006 protein itself, and not by other Serratia-specific regulatory factors. High-affinity binding of purified His6-CarR39006 to purified target DNA, in the absence of any other molecule, suggests that CarR39006 does not require a ligand to function. Before this study, it had been demonstrated that some QS systems can respond to non-AHL molecules, such as Rhodopseudomonas palustris, which responds to a p-coumaroyl-homoserine lactone derived from lignin (Schaefer et al., 2008). Similarly, OryR from Xanthomonas oryzae pv. oryzae is activated by undefined plant signals from rice (Ferluga and Venturi, 2009). Other LuxR family proteins have been shown to be inherently transcriptionally active, but are inactivated by cognate AHLs (Schu et al., 2009). In contrast, binding of CarR39006 to DNA is neither improved nor attenuated by AHLs, thereby representing the first truly ligand-independent LuxR family protein.
The binding of apo-CarREcc and apo-CarR39006 to DNA is reminiscent of the ‘repressor’ class of LuxR proteins, such as EsaR from P. stewartii (Von Bodman et al., 1998). Members of this class are more accurately described as ‘apo-functional’, as in some cases their competence to bind DNA in the absence of AHLs can result in transcriptional activation (Von Bodman et al., 2003; Schu et al., 2009). Members of this class are inactivated by AHL signals, via a poorly defined mechanism, resulting in derepression (or deactivation) of transcription. A very recent review of LuxR family proteins proposed that these apo-functional LuxR proteins fall into a single phylogenetic clade (Tsai and Winans, 2010). Interestingly, the CarREcc and CarR39006 activators were identified as the closest relatives of this apo-functional class (Fig. S6). Thus, CarREcc and CarR39006 may represent evolutionary intermediates between AHL-activated and AHL-inactivated LuxR proteins. Our data show that CarR39006 represents a third class of LuxR protein, which is neither activated nor inactivated by AHLs.
Experiments with domain chimeras suggest that the C-terminal DNA-binding domain of CarR39006 may be responsible for conferring AHL independence. Attaching this C-terminal domain onto the CarREcc N-terminus resulted in high-affinity DNA binding and transcriptional activation, abolishing the need for an AHL signal. The N-terminal domain residues involved in AHL binding are probably no longer selectively advantageous in the AHL-independent CarR39006, so the loss of the critical Trp-44 residue may have occurred by neutral drift. In contrast, attachment of the signal-blind CarR39006 N-terminal domain onto the CarREcc C-terminal domain (which is normally activated by 3O-C6-HSL binding to the N-terminus) resulted in low affinity binding to the carA promoter, protease sensitivity and poor transcriptional activation.
The C-terminal DNA-binding domains of CarREcc and CarR39006 are extremely similar, containing only three amino acid differences within alpha helices α7, α8, α9 and α10 (Fig. S4). This strong amino acid identity is consistent with the ability of these two proteins to activate transcription from the two carA promoters interchangeably (Fig. S5). Interestingly, secondary structure prediction suggests that the extreme C-terminus of CarR39006 may contain an additional short alpha helix, which is not predicted for CarREcc or any other LuxR family protein analysed, and which may be functionally significant (Fig. S4). Further studies are required to identify the C-terminal regions that are responsible for the unique mechanism by which the signal-blind CarR39006 protein has evolved to overcome the requirement for an AHL ligand.
Bacterial strains and culture conditions
Escherichia coli DH5α was grown at 37°C in Luria–broth (LB) at 300 r.p.m. or on LB agar containing 1.5% (w/v) agar. Media were supplemented with ampicillin (Ap, 100 µg ml−1) and tetracycline (Tc, 35 µg ml−1), and growth (OD600) was measured as described previously (Poulter et al., 2010). When required, an appropriate amount of isopropyl-β-d-thiogalactopyranoside (IPTG) was added (see Results). Synthetic N-acyl homoserine lactones were synthesized as previously described (Glansdorp et al., 2004). All experiments were performed in triplicate and plotted as mean ± SD.
DNA manipulation and sequence analysis
Molecular biology techniques and sequencing were performed as described previously (Poulter et al., 2010). Oligonucleotide primers are shown in Table S1. All plasmids were verified by DNA sequencing. Sequence data were analysed using GCG (Genetics Computer Group, University of Wisconsin) and compared with GenBank DNA or non-redundant protein sequence databases using BLAST (Altschul et al., 1997). Multiple sequence alignments were performed using MultAlin (Corpet, 1988), and images were generated with ESPript (Gouet et al., 1999). Secondary structure prediction analysis was performed using PSIpred (McGuffin et al., 2000).
Construction of CarR expression vectors
All proteins expressed in this study contained N-terminal hexahistidine (His6) tags, and all plasmids are shown in Table 1. CarR expression vectors pSP78 (His6-CarREcc) and pSP79 (His6-CarR39006) were created by PCR cloning of the carREcc and carR39006 genes into pQE80L (Qiagen) BamHI and PstI sites, using primers SP277 and SP257, and SP278 and PF67 respectively. For expression of the CarR AHL-binding domains (amino acids 1 to 167), the primers SP277 and SP279 (carREcc) and SP278 and SP280 (carR39006) were used to PCR clone the first 501 bp of the carR genes into pQE80L BamHI and PstI sites, creating plasmids pSP76 (His6-CarREcc1-167) and pSP77 (His6-CarR390061-167).
Table 1. Bacterial strains and plasmids used in this study.
Substitution of residue 44 of CarREcc and CarR39006 was performed by overlap extension PCR (Warrens et al., 1997). A CarREcc W44C variant was created by PCR amplification of the 5′ 148 bp of carREcc using SP277 and the mutagenic primer SP273, and the 3′ 618 bp of CarREcc using mutagenic primer SP274 and SP257. Using the resulting products as template, PCR amplification with SP277 and SP257 resulted in a spliced 735 bp product, which was cloned into pQE80L BamHI and PstI sites to generate plasmid pSP90. An identical strategy was used for the CarR39006 C44W variant plasmid pSP91, using primers SP278 and SP95 (5′ end) and SP96 and PF67 (3′ end). The CarREcc W44C ligand-binding domain (His6-CarREcc1-167 W44C) was expressed from plasmid pSP111, created by PCR cloning the 5′ 501 bp of carREcc W44C from pSP90 into pQE80L BamHI and PstI sites, using primers SP277 and SP279.
Construction of CarR domain chimeras
Chimera proteins were constructed, in which the CarR39006 N-terminal AHL-binding domain (residues 1 to 158) was fused to the CarREcc C-terminal DNA-binding domain (residues 159 to 244), and vice versa. The 5′ 474 bp of carR39006 was PCR amplified using primers SP278 and SP299, and the 3′ 261 bp of carREcc amplified using SP300 and SP257. Using the resulting products as template, PCR amplification with SP278 and SP257 resulted in a final 735 bp spliced product in which the gene fragment encoding the CarR39006 N-terminal domain was fused to the fragment encoding the CarREcc C-terminal domain. An identical strategy was used using primers SP277 and SP307, and primers SP308 and PF67, resulting in a spliced carREcc-39006 PCR product. These products were cloned into pQE80L BamHI and PstI sites to create plasmid pSP88 and pSP89 respectively.
Purification of His6-tagged proteins
CarR proteins were purified from E. coli BL21/DE3 carrying plasmids pSP76, pSP77, pSP78, pSP79, pSP90 or pSP111. 5 ml overnight cultures were grown in LB with Ap at 37°C, and used to inoculate 1 l of pre-warmed LB and Ap. Expression cultures were grown at 37°C with shaking at 200 r.p.m. to an OD600 of 0.5, then rapidly cooled to 20°C, induced with 1 mM IPTG and incubated overnight at 16°C. His6-tagged proteins were purified from harvested cells using Ni-NTA resin (Qiagen) as described previously (Gristwood et al., 2008). Elution fractions containing protein were pooled and dialysed overnight against 1 l Dilution Buffer (50 mM NaH2PO4/300 mM NaCl/pH 8.0) at 4°C.
carA promoter lacZ fusion experiments
The Ecc carA promoter (nucleotides −210 to +127, relative to the transcriptional start site) was PCR cloned with primers SP152 and SP153 into the BamHI and HindIII sites of pRW50 (Lodge et al., 1992) to give plasmid pSP14. The Serratia carA promoter (-199 to +50) was previously cloned into pRW50, to create plasmid pTA16 (Gristwood et al., 2008). E. coli DH5α cells carrying pRW50, pSP14 or pTA16, plus pSP78, pSP79, pSP88, pSP89, pSP90 or pSP91, were grown in 5 ml of LB, plus Ap, Tc, IPTG and synthetic AHLs to an OD600 of 0.5. Promoter expression (measured as Miller Units) was determined as described previously (Fineran et al., 2005b).
Western blot analysis for CarR proteins
Western blot analysis was performed against bacterial cell samples taken from promoter lacZ fusion experiments (normalized to OD600 and separated by SDS-PAGE), using a primary mouse monoclonal anti-His antibody (Novagen) and a goat anti-mouse polyclonal HRP conjugated secondary antibody (Sigma).
Electrophoretic mobility shift assays experiments were performed with 3′ DIG-labelled carA promoter fragments from pSP14 and pTA16, using a DIG Gel Shift Kit (Roche). EMSA reaction mixtures (20 µl) contained the indicated amount of His6-CarR protein and 1.5 nM of DIG-labelled DNA, in binding buffer [20 mM HEPES, pH 7.6, 1 mM EDTA, 10 mM (NH4)2SO4, 1 mM DTT, 0.02% Tween and 30 mM KCl], 5 µg ml−1 poly-L-lysine and 50 µg ml−1 poly[d(A-T)]. Specific competition reactions contained 50-fold excess (75 nM) unlabelled carA promoter DNA. An rsmA promoter fragment from Erwinia carotovora ssp. atroseptica SCRI1043 (Bell et al., 2004) was used as a non-specific competitor.
Isothermal titration calorimetry was performed on a VP-ITC micro-calorimeter (MicroCal) at 30°C. Proteins were dialysed against ITC Buffer (50 mM NaH2PO4/300 mM NaCl/1% DMSO/pH 8.0). Synthetic AHLs (50 mM in DMSO) were diluted into Dilution Buffer (50 mM NaH2PO4/300 mM NaCl/pH 8.0) to 600 µM. Standard settings were: 27 injections (2 µl, 5 µl, then 25 × 10 µl), 120 s initial delay, 30°C, reference power 17, stirring speed 310 r.p.m. Based on the observed stoichiometry of binding, raw data were processed and plotted using a two binding site model (Origin software). Dissociation constants (Kd) were derived from Ka values (Kd = 1/Ka) calculated from the slope of the isotherm curve.
We thank members of the Salmond group, Dr M. Welch and Dr L. Evans for useful discussions, P. Sledz for ITC assistance, Dr D. Marsden for ITC conditions, Dr P. Fineran for providing plasmid pTA16 and I. Foulds and A. Rawlinson for technical assistance. This work was supported by the Biotechnology and Biological Sciences Research Council.