Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Activation of transcription initiation at the Escherichia coli melAB promoter is dependent on MelR, a transcription factor belonging to the AraC family. MelR binds to 18 bp target sites using two helix–turn–helix (HTH) motifs that are both located in its C-terminal domain. The melAB promoter contains four target sites for MelR. Previously, we showed that occupation of two of these sites, centred at positions −42.5 and −62.5 upstream of the melAB transcription start point, is sufficient for activation. We showed that MelR binds as a direct repeat to these sites, and we proposed a model to describe how the two HTH motifs are positioned. Here, we have used suppression genetics to confirm this model and to show that MelR residue 273, which is in HTH 2, interacts with basepair 13 of each target site. As our model for DNA-bound MelR suggests that HTH 2 must be adjacent to the melAB promoter −35 element, we searched this part of MelR for amino acid side-chains that might be able to interact with σ. We describe genetic evidence to show that MelR residue 261 is close to residues 596 and 599 of the RNA polymerase σ70 subunit, and that they can interact. Similarly, MelR residue 265 is shown to be able to interact with residue 596 of σ70. In the final part of the work, we describe experiments in which the MelR binding site at position −42.5 was improved. We show that this is detrimental to MelR-dependent transcription activation because bound MelR is mispositioned so that it is unable to make ‘correct’ interactions with σ.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The Escherichia coli melibiose (mel) operon encodes proteins that facilitate the metabolism of melibiose. Expression of the mel operon is controlled by a single promoter (the melAB promoter), the expression of which is co-dependent on two transcription activators, MelR and the cyclic AMP receptor protein (CRP) (Webster et al., 1987; Belyaeva et al., 2000). MelR is a member of the AraC family of transcription regulators (reviewed by Gallegos et al., 1997; Martin and Rosner, 2001; Egan, 2002), and its activity is controlled by melibiose. CRP is a global regulator controlled by the second messenger, cyclic AMP, the level of which is controlled by the availability of glucose (reviewed by Kolb et al., 1993). Thus, the co-dependence of the melAB promoter on two activators couples induction to two physiological signals, the presence of melibiose and the absence of glucose.
In previous studies of the melAB promoter, we identified four DNA sites for MelR binding (organised as two pairs: sites 1′ and 1 and sites 2 and 2′) and one site for CRP binding (Fig. 1A; Belyaeva et al., 2000). Each DNA site for MelR covers 18 bp and is occupied by a single MelR subunit. MelR subunits consist of two domains, an N-terminal melibiose-binding domain and a C-terminal DNA-binding domain that carries two helix–turn–helix motifs (HTH), which is typical of AraC family members (Michán et al., 1995; Howard et al., 2002). Each MelR subunit binds with the two HTH motifs penetrating the DNA major groove on the same face of the target DNA duplex (Fig. 1C). We showed previously that the two MelR subunits bound to site 2 and site 2′ are organized as a direct repeat, with HTH 2 (residues 259–282, located towards the C-terminus of MelR) contacting the downstream part of both sites, and with HTH 1 (residues 211–232) contacting the upstream part (Fig. 1C; Grainger et al., 2003). We also showed that the site for CRP binding is located between MelR binding site 1 and site 2, centred between basepairs 81 and 82 upstream of the melAB transcript start site, i.e. position −81.5 (Fig. 1A; Belyaeva et al., 2000).
In vitro studies with the melAB promoter have revealed that its co-dependence on MelR and CRP results from an unusual mechanism (Wade et al., 2001). In the absence of CRP, MelR can occupy sites 1′ and 1 and site 2, but is unable to occupy site 2′, which overlaps the promoter −35 hexamer element. This appears to be because site 2′ is the weakest of the four DNA sites for MelR. Occupation of site 2′, and concomitant activation of the melAB promoter, requires the presence of melibiose and the binding of CRP at its target. Interestingly, CRP is unable to bind at its target site in the absence of MelR, and it is recruited to the melAB promoter by a direct interaction with prebound MelR. Thus, we have proposed that MelR and CRP bind co-operatively to form a large nucleoprotein complex at the melAB promoter. The activity of this complex is triggered by melibiose, which induces MelR to occupy site 2′. We have shown that it is the occupation of site 2′, which overlaps the melAB promoter −35 element, that is most important for transcription activation. We have proposed that this is because MelR activates transcription by a class II-type mechanism that involves direct contact with domain 4 of the RNA polymerase σ subunit bound at the target promoter −35 element (Grainger et al., 2003).
In this study, we have used genetic analysis to investigate interactions between MelR and domain 4 of the RNA polymerase σ70 subunit. Because of the complexity of the wild-type melAB promoter, we worked with two simplified derivatives, JK16 and JK19, in which the base sequence of MelR binding site 2′ was altered such that MelR-dependent activation became independent of CRP and of MelR binding to sites 1 and 1′ (Fig. 1B). In our previous study, we argued that the downstream parts of MelR binding sites 2 and 2′ must be occupied by HTH 2 of MelR. Thus, in the first part of this work, we used suppression genetics to confirm that this is the case, and that this part of MelR is well placed to contact domain 4 of the RNA polymerase σ70 subunit. In the second part of the study, we show that MelR residues D261 and T265 can interact with amino acid residues located near the C-terminus of the σ70 subunit. Finally, we investigated transcription activation by MelR at melAB promoter derivatives in which the DNA sequence of MelR binding sites 2 and 2′ are identical. We show that activation at these promoters is inefficient because MelR is incorrectly positioned.
Results and discussion
MelR binding to the melAB promoter and to a simplified derivative
In previous papers, we argued that, although MelR binds to four different sites at the melAB promoter (illustrated in Fig. 1A), it is the occupation of site 2′, centred at position −42.5 and overlapping the melAB promoter −35 hexamer element, that is essential for transcription activation (Belyaeva et al., 2000; Wade et al., 2001). MelR bound at site 2′ appears to act as a typical ‘class II’ transcription activator, probably by interacting with the RNA polymerase σ subunit. Of the four DNA sites for MelR at the melAB promoter, site 2′ is the weakest and, thus, occupation of this site requires prebinding of MelR to the upstream sites, and of CRP. In order to study the molecular details of transcription activation by MelR more easily, we have worked with the JK16 and JK19 melAB promoter derivatives (illustrated in Fig. 1A and B). These derivatives carry site 2′ sequences that are altered such that their DNA sequence approaches that of site 1′. Expression from the JK16 and JK19 melAB promoter derivatives remains completely dependent on MelR, but is independent of CRP and independent of MelR binding to the upstream site 1 and site 1′ (Belyaeva et al., 2000; C. L. Webster, unpublished data).
In our most recent study, we showed that MelR binds as a direct repeat to site 2 and site 2′ at the JK19 promoter, and that each MelR subunit is oriented such that HTH 2 binds towards the downstream end of each site (Grainger et al., 2003). Genetic epistasis experiments were used to derive a model that predicted that residue R273, in the recognition helix of HTH 2, contacts the basepair at position 13 of each site. Thus, the first aim of this study was to use suppression genetics to confirm this model. To do this, we constructed the p39T p59T derivative of the JK19 promoter in which the G:C basepair at position 13 of both site 2 and site 2′ was substituted by a T:A basepair. Both the starting JK19 and the JK19 p39T p59T fragments were cloned into the broad-host-range lac expression vector, pRW50, to create derivatives carrying pmelAB::lac fusions. These plasmids were transformed into the WAM132 ΔlacΔmelR host strain. Colonies give a white Lac– phenotype on MacConkey lactose indicator plates, because the melAB promoter cannot be activated in the absence of melR. The introduction of the melR gene, cloned into the pJW15 plasmid, resulted in a Lac+ phenotype for WAM132 cells carrying the JK19 pmelAB::lac fusion. In contrast, WAM132 cells carrying the JK19 p39T p59T derivative::lac fusion remained Lac–, because the p39T and p59T substitutions (at position 13 of site 2′ and site 2) prevent MelR binding to its target sites and, hence, prevent MelR-dependent activation. In parallel experiments, we investigated the effects of the p39T and p59T substitutions alone, and found that they result in intermediate levels of MelR-dependent activation of the melAB promoter (C. L. Webster, unpublished results).
To perform the suppression genetics experiment, DNA fragments from pJW15, carrying melR, were resynthesized using error-prone polymerase chain reaction (PCR) and recloned into pJW15 to give a library of random melR mutants. DNA from this library was used to transform WAM132 ΔmelRΔlac cells containing pRW50 carrying the JK19 p39T p59T derivative::lac fusion. After screening 20 000 transformants from three different libraries, we isolated seven independent colonies that displayed a strong Lac+ phenotype. The pJW15 DNA was isolated from each colony, and the base sequence of the insert carrying melR was determined. We found that the melR genes in each of the seven pJW15 derivatives encoded a substitution of residue R273. Thus, although the entire melR gene had been mutagenized, these screens yielded only melR encoding the RC273, RG273 and RS273 derivatives. Assays of promoter activities, described in Fig. 2, showed that these substitutions restored levels of transcription activation from the p39T p59T derivative of the melAB JK19 promoter to between 20% and 60% of that found with the starting JK19 promoter. In each case, activity was dependent on the addition of melibiose to the growth medium, just as with the starting pJW15 plasmid (data not shown).
The error-prone PCR mutagenesis method was set up to make single base substitutions and, hence, could only create certain amino acid changes at codon 273 of MelR. Thus, in a complementary experiment, a library of pJW15 derivatives encoding melR with random sequences at codon 273 was made. From this library, we selected four further MelR derivatives, RQ273, RN273, RT273, RV273, that suppress the effects of the p39T p59T substitutions in the JK19 promoter. Two of these derivatives, RN273 and RV273, restored activation to a level nearly equal to that observed with the starting JK19 promoter fragment (Fig. 2). Further experiments showed that none of the seven different substitutions of MelR residue R273 causes an increase in MelR-dependent activation of the JK19 promoter (most cause a small decrease; Fig. 2). Thus, our observation, that only substitutions of MelR residue 273 permit MelR-dependent activation of the p39T p59T derivative of the melAB JK19, argues strongly that MelR residue R273 must be in close proximity to position 13 of MelR binding sites 2 and 2′. This is in agreement with the model presented in Fig. 1C that was generated from the findings of Grainger et al. (2003).
Alanine substitutions in MelR that give positive control phenotypes
Figure 1C illustrates our conclusion that, during transcription activation by MelR, HTH 2 binds adjacent to the promoter −35 hexamer and is well placed to contact domain 4 of the RNA polymerase σ70 subunit. This accords with models deduced from studies with two other AraC family members, the E. coli RhaS protein and the E. coli SoxS protein. For RhaS, Bhende and Egan (2000) used genetic epistasis to argue that RhaS residue D241 (equivalent to MelR D261) contacts residue R599 in the RNA polymerase σ70 subunit during activation at the rhaBAD promoter. For SoxS, Griffith and Wolf (2002) used alanine scanning to identify a surface-exposed patch that contributes to activation at class II SoxS-dependent promoters. This patch includes SoxS residue D75, which aligns with RhaS residue D241 and hence MelR residue D261 (Gallegos et al., 1997). Thus, we used site-directed mutagenesis to create melR derivatives encoding MelR with alanine substitutions at positions equivalent and adjacent to the class II-specific residues identified by Griffith and Wolf (2002). These were transferred to pLG314 (a kanR plasmid that carries melR) and transformed into WAM132 ΔmelRΔlac cells containing pRW50, carrying the KK43 (wild type), JK16 or JK19 melAB promoter fragments. Melibiose-dependent β-galactosidase expression was then measured. Some MelR derivatives (LA260 and LA264) were totally defective in their ability to activate transcription at these promoters, whereas another (DA256) exhibited similar activity to the wild-type protein (data not shown). These versions of MelR were not studied further. However, MelR with the KA257, SA258, DA261 or TA265 substitutions was significantly less active than the wild-type protein at the different promoters (Fig. 3A shows data with JK16).
To test that the effects were not the result of a defect in DNA binding, we constructed the DCG1 promoter. This is a derivative of the activator-independent extended ‘−10’ TB-TC promoter (Burr, 2000), which is derived from the galP1 promoter. DCG1 carries the 18 bp MelR binding site 1 centred at position −61.5 of the TB-TC promoter. WAM132 ΔmelRΔlac cells containing pRW50 carrying the DCG1 promoter score as Lac+ because of the DCG1::lac fusion. However, lac expression is sharply repressed by MelR encoded by pLG314, as a result of MelR binding to its target at position −61.5. Results illustrated in Fig. 3B show that this repression is unaffected by the KA257, SA258, DA261 or TA265 substitutions. Thus, we conclude that MelR with the KA257, SA258, DA261 or TA265 substitution is defective in its ability to activate transcription but not in its ability to bind to DNA.
Interaction between MelR residue 261 and σ70 studied by suppression genetics
In the next experiment, we used suppression genetics to investigate whether the putative MelR activation determinant, defined by the different alanine substitutions, can interact with domain 4 of the RNA poIymerase σ70 subunit. As many class II activators interact with σ70 using negatively charged amino acid side-chains (Dove et al., 2003), we focused on MelR residue 261, and constructed a derivative of pLG314 encoding MelR with the DK261 substitution that introduces a positive charge at this position. We checked that the DK261 substitution reduced the ability of MelR to activate transcription at the KK43, JK16 and JK19 promoters, while not affecting its DNA binding ability, as judged by repression of the DCG1 promoter (data not shown). A reduction in activation was observed with each of the three melAB promoter derivatives tested, but a clear phenotype change on MacConkey indicator plates was found only with the JK16 promoter. Thus, we used the JK16 promoter to search for σ70 mutants that could suppress the activation defect due to DK261. To do this, the rpoD gene (which encodes σ70) in plasmid pVRσ was mutagenized by error-prone PCR from codon 536 to codon 613. This resulted in a library of mutant pVRσ plasmids encoding σ70 with random substitutions in domain 4. Library DNA was transformed into WAM132 ΔmelRΔlac cells containing pRW50 with the JK16 melAB promoter and pLG314 encoding MelR DK261. As expected, most transformants scored as Lac–, but a small number of Lac+ colonies, due to mutated pVRσ, were found. Three Lac+ colonies were found to contain pVRσ encoding σ70 with the RG599, RC599 and RS599 substitutions. Experiments illustrated in Fig. 4 show that each of these substitutions increases activation of the JK16 melAB promoter by MelR DK261 but not by wild-type MelR. A further Lac+ colony carried pVRσ encoding σ70 RH596 and KN297. When separated from each other, neither of these two substitutions could specifically suppress the effects of MelR DK261 (data not shown). However, we tested the effects of other changes in R596 (previously isolated by V. Rhodius) and found that pVRσ encoding σ70 RC596 and RS596 gave a small but reproducible specific suppression of the effect of the DK261 substitution in MelR (Fig. 4). It is likely that we had failed to find these derivatives in our screen of random mutants because of their small effects.
Study of the MelR–σ70 domain 4 interface using genetic epistasis
The above results show that MelR residues K257, S258, D261 and T265 are important for transcription activation, and suggest that MelR residue 261 can interact with residues 599 and 596 in domain 4 of σ70. To investigate interactions between MelR and domain 4 of σ70 further, we measured the effects of the KA257, SA258, DA261 and TA265 substitutions in MelR in the presence of σ70 with alanine substitutions at residues 590, 591, 593, 595, 596, 597, 598, 599 or 600. To do this, WAM132 ΔmelRΔlac cells containing pRW50 with the JK16 melAB promoter::lac fusion, and pVRσ encoding σ70 with the different alanine substitutions, was transformed with pLG314 derivatives, encoding wild-type MelR or the KA257, SA258, DA261 and TA265 derivatives, and β-galactosidase activities were measured. Our aim was to identify alanine substitutions in σ70 that resulted in differential activation by the different MelR mutants.
The results, summarized in Fig. 5, show that alanine substitutions at positions 590, 591, 595, 598 and 600 in σ70 have little or no effect on activation of the JK16 melAB promoter by wild-type MelR or by the KA257, SA258, DA261 and TA265 mutant proteins. In contrast, with σ70 carrying the alanine substitution at position 596, the reduction in activation by MelR resulting from the DA261 and TA265 substitutions (but not the KA257 and SA258 substitutions) is greatly attenuated. With σ70 carrying the alanine substitution at position 599, the deleterious effect on activation of the DA261 substitution in MelR (but not the KA257, SA258 and TA265 substitutions) is greatly reduced. Thus, in agreement with the results from the suppression genetics experiments, residue 261 in MelR appears to interact with residues 596 and 599 in σ70. Similarly, MelR residue 265 may interact with σ70 residue 596. Interestingly, with σ70 carrying alanine substitutions at positions 593 and 597, the reduction in activation by all four different alanine substitutions in MelR is attenuated. To explain this, we suggest that the KA593 and KA597 substitutions in σ70 remove unfavourable interactions with the mutant MelR derivatives.
Construction of melAB promoter derivatives with a consensus 2′ site
The JK16 and JK19 melAB promoter derivatives (illustrated in Fig. 1A and B) carry site 2′ sequences that were altered such that the base sequences approach that of site 1′. Prompted by our finding that MelR binds as a direct repeat at these promoters, and that it is occupation of site 2′ that is crucial for transcription activation, we made derivatives in which the site 2′ sequence was replaced by the consensus site 2 sequence. To facilitate these constructions, we started with the JK22 melAB promoter fragment, a shortened derivative of JK19 in which upstream DNA sequences had been removed without affecting promoter activity (Grainger et al., 2003). Figure 6A illustrates the resulting JK30 melAB promoter derivative, carrying a direct repeat of the site 2 sequence, such that the base sequence of site 2′ is identical to that of site 2. To measure the activity of this promoter, the JK30 fragment was cloned into pRW50 to create a pmelAB::lac fusion, and the resulting plasmid was transformed into WAM132 ΔmelRΔlac cells containing pLG314, and β-galactosidase expression was measured. Unexpectedly, we found that MelR was completely unable to activate transcription at JK30 in any of the conditions tested (Fig. 6B). As this negative result might have been due to alterations affecting the −35 hexamer element, we constructed JK31, a derivative of JK30 in which the base sequence at the downstream end of site 2′ corresponds to the base sequence of the starting JK22 promoter (Fig. 6A). Measurements of the activity of the JK31 promoter, illustrated in Fig. 6B, showed that, although it is more active than JK30, it is substantially less active than the starting JK22 promoter.
To understand the reasons for the low activity of the JK30 and JK31 promoters, we investigated the binding of MelR. In our first experiment, electrophoretic mobility shift assays (EMSAs) were used to monitor the binding of MelR to DNA fragments carrying the JK22 or JK31 promoters. Our previous work had shown that MelR can bind to a single consensus DNA target site as a monomer and that, once bound to the DNA, this MelR molecule can recruit a further MelR subunit to an adjacent low-affinity site by dimerization (Howard et al., 2002). Results illustrated in Fig. 7A show that two MelR–DNA complexes, *1 and *2, are formed with both fragments. The *1 complex has a higher mobility and is formed by a single MelR molecule binding to the DNA. The low-mobility complex, *2, is formed by two MelR molecules binding to the DNA. In the case of JK22, in which only one of the two MelR binding sites is consensus, it appears that MelR first binds to the high-affinity site 2, creating the *1 complex, before binding to the lower affinity site 2′ to create the *2 complex. JK31 contains two high-affinity MelR binding sites and thus, as expected, the *2 complex is formed more readily. We conclude that MelR binds more co-operatively to JK31 than to JK22.
As tighter co-operative binding of MelR to site 2′ at the JK31 melAB promoter might be expected to improve rather than hinder transcription activation, we examined the positioning of bound MelR subunits. In our previous work, we described a derivative of MelR containing a single cysteine residue at position 269, Cys269 MelR (Grainger et al., 2003; Fig. 1C). This derivative could be purified and conjugated at Cys-269 with the inorganic nuclease, p-bromoacetamidobenzyl-EDTA-Fe (FeBABE). FeBABE-tagged Cys269 MelR was then bound to labelled melAB promoter fragments, and the patterns of DNA cleavage generated after activation of the FeBABE reagent gave information about the location and orientation of bound MelR. In our previous study, we showed that the FeBABE reagent attached to a single bound MelR subunit generated cleavage of the DNA minor groove at two locations on the same face of the DNA helix, separated by 10 bp. Results illustrated in Fig. 7B show that FeBABE-tagged Cys-269 MelR generates an almost identical pattern of DNA cleavage at the JK22 and JK31 promoters (note that the JK31 fragment is cleaved much more readily than JK22, because of tighter co-operative binding of MelR). Four discrete regions of DNA cleavage, occurring near positions −60, −50, −40 and −30, are observed, with the two upstream and two downstream sets of cleavages resulting from MelR binding to site 2 and site 2′ respectively. Closer examination of the data, however, shows that, although the two upstream sets of cleavage at the JK22 and JK31 promoters are identical, the two downstream sets of cleavage at JK31 are shifted upstream by 1–2 bp. From this, we conclude that MelR bound at site 2′ at the JK31 promoter is mispositioned and that this accounts for the low activity of the JK31 melAB promoter. It is likely that mispositioning is a consequence of alternative MelR–DNA contacts being formed or a difference in promoter DNA bending.
To test the possibility that mispositioning of MelR bound at site 2′ at the JK31 melAB promoter has corrupted the MelR–σ70 domain 4 interface, we performed an epistasis experiment. To do this, we compared the activity of the KA257, SA258, DA261 and TA265 MelR mutants at the JK19 and JK31 promoter fragments (Fig. 8). As in the previous experiment, the KA257, SA258, DA261 and TA265 substitutions cause similar reductions in the activity of the JK16 melAB promoter (Fig. 5). Our results (Fig. 8) show that the KA257 and SA258 substitutions cause even larger defects in activation at the JK31 melAB promoter. In contrast, the DA261 and TA265 substitutions cause smaller defects. An explanation for this is that MelR bound at site 2′ at the JK31 promoter, and hence the D261 and T265 side-chains, are misplaced. Thus, alanine substitutions at D261 and T265 have lesser effects on MelR-dependent transcription activation.
Many transcription activator proteins bind to DNA sites that overlap the −35 hexamer at target promoters. It is thought that many such activators function by making a contact with domain 4 of the RNA polymerase σ70 subunit, which recruits the transcriptional machinery to the promoter DNA (reviewed by Dove et al., 2003). In some cases, the molecular details of this interaction have been determined. For example, for bacteriophage λ cI protein at the PRM promoter, genetics and structural modelling have been used to show that a negatively charged surface of cI, close to its DNA-binding domain, contacts specific residues in the C-terminal domain 4 of the RNA polymerase σ70 subunit. Structural modelling suggests that residue E34 of cI protein interacts with σ70 domain 4 residue R588, and that residue D38 of cI contacts σ70 domain 4 residue R596 (Nickels et al., 2002). Similarly, for the AraC family protein RhaS, a genetic analysis argues that negatively charged residue D241 makes contact with residue R599 in domain 4 of σ70 (Bhende and Egan, 2000).
In this work, we used genetic analysis to characterize the MelR–σ70 domain 4 interface. In a preliminary experiment, we confirmed the orientation of MelR when bound to site 2′. This was particularly important in order to identify the parts of bound MelR that would be well-placed to contact domain 4 of σ70 bound to its −35 element during transcription activation at the melAB promoter. We then showed that alanine substitutions at MelR residues K257, S258, D261 or T265 lead to a severe activation defect in MelR, and we used genetic suppression and epistasis experiments to explore interactions between MelR and σ70. The interpretation of all suppression and epistasis experiments is complicated. In this instance, the problems are compounded because activation of the melAB promoter probably involves many different MelR–RNA polymerase interactions, because alternative interactions (either favourable or unfavourable) may be created in mutants and because the effects of σ mutants were assayed in a diploid strain. Nevertheless, the present data show clearly that the deleterious effects of substitutions at residue 261 in MelR can be attenuated by changes in R599 and R596 in σ70. Similarly, the deleterious effect of the TA265 substitution in MelR is reversed by the RA596 substitution in σ70. The simplest explanation for these results is that, during MelR-dependent activation of the melAB promoter, residue 261 in MelR is close to residues 599 and 596 in σ70, and that residue 265 in MelR is close to residue 596 in σ70.
To assess the feasibility of these conclusions, we constructed a simple structure-based model of the MelR–σ70 domain 4 interface (Fig. 9). Our starting point was the MarA–marbox structure of Rhee et al. (1998) and the structure of Thermus aquaticusσA bound to −35 element DNA (Campbell et al., 2002). Two basepairs of the MarA–marbox structure DNA were aligned with the equivalent basepairs of the T. aquaticusσA domain 4/−35 element DNA structure using weblab viewer, version 3.1 (Molecular Simulations). We then superimposed the amino acid sequence of MelR onto MarA, and the amino acid sequence of σ70 onto σA. Our model, illustrated in Fig. 9, shows that residue R596 of σ70 domain 4 is positioned so that it can interact with both residues D261 and T265 of MelR, and that σ70 residue R599 is proximal to MelR residue D261, but not to T265. Our interpretation of the genetic results is that, with wild-type MelR and σ70, these interactions do occur. However, as the activation defects due to the DA261 or TA265 substitutions in MelR can be almost fully reversed by particular alanine substitutions in σ70, and as it is unlikely that the alanine residues are close enough to interact, we conclude that any single interaction between MelR and σ70 cannot be essential. We suggest that the surface of MelR near D261 and T265 has evolved to accommodate the surface of σ70 near R599 and R596, and that the most likely consequences of the alanine substitutions in MelR are to create unfavourable interactions (for example, resulting from clashes between charged and uncharged side-chains). Other genetic approaches will be needed to identify whether there are MelR side-chains that make contacts with RNA polymerase that are essential and cannot be replaced.
As far as we know, our study represents the most detailed analysis of interactions between an AraC family protein and the RNA polymerase σ70 subunit to date. The available evidence suggests that many AraC family proteins bind to DNA sites overlapping the −35 hexamer at target promoters, and that the surface of the protein including HTH 2 binds adjacent to domain 4 of the σ70 subunit (Martin and Rosner, 2001). In some cases, strong interactions will have evolved that are essential for transcription activation. In other cases, such as described here, the activator may have evolved so that it can form a number of complementary interactions with σ70. In either case, the precise position of the AraC family protein, bound to its target overlapping the −35 hexamer, is likely to be important. An interesting, and unexpected, conclusion from this work is that correct positioning depends on the precise nature of the binding sequence, and that optimal positioning depends on a binding sequence that is far from the consensus.
Strains, plasmids and promoter fragments
Bacterial strains, plasmids and the melAB promoter fragments used in this work are listed in Table 1. Standard techniques for recombinant DNA manipulations were used throughout. Automated DNA sequencing was done at the University of Birmingham genomics suite (funded by BBSRC grant 6/JIF13209). Table 2 lists the primers used for manipulation of the different melAB promoter fragments. EcoRI–HindIII fragments carrying the starting melAB promoter and derivatives (JK16, JK19, JK22, JK30 and JK31) were cloned in the vector pUC18. Note that these derivatives contain a single PstI site located in the melAB promoter inserts, 5 bp downstream of MelR binding site 2′. By convention, promoter sequences are numbered with the transcription start point as +1 and with upstream and downstream locations denoted by ‘–’ and ‘+’ prefixes respectively. Promoter mutations are referred to as pNB, where B is the mutant base on the non-template strand and N is its location.
Table 1. . Bacterial strains, plasmids and promoter fragments.
Derivative of JK22 where the sequence of MelR binding site 2′ has been replaced by the sequence of MelR binding site 2
Derivative of JK30 where positions 16 and 18 of MelR binding site 2′ (located at −34 and −36 with respect to the melAB transcription start site) have been altered to resemble the sequence of the wild-type promoter
Derivative of the galP1 promoter repressed by MelR
Introduces MelR binding site at position −61.5 of galP1
The JK19 p39T p59T fragment was generated by sequential PCRs. The first round of PCR was done using the primers D25322 and D14314 with pUC18 carrying the JK19 fragment (pUC18/JK19) as template. The resulting PCR product was digested with EcoRI and HindIII and cloned into purified EcoRI–HindIII vector. The resulting plasmid, pUC18, carrying the p59T derivative of JK19, was used as a template in a second round of PCR, using the primers D3407 and D34121. This PCR product was digested with EcoRI and PstI and cloned into pUC18/JK19 vector, creating a pUC18 derivative carrying the JK19 p39T p59T promoter on an EcoRI–HindIII fragment that was subcloned into pRW50 as required.
The JK30 and JK31 fragments were generated by PCR, using primer D34119 for JK30 or D34120 for JK31, with the flanking primer D3407. pUC18 carrying the JK22 melAB promoter fragment (pUC18/JK22) was used as a template. The resulting PCR products were digested with EcoRI and PstI and cloned into EcoRI–PstI vector isolated from pUC18/JK22. The resulting EcoRI–HindIII JK30 and JK31 fragments were subcloned into pRW50 as required.
The DCG1 promoter was generated by PCR using the primers D31250 and D4600 and the galP1 promoter derivative TB-TC (Burr, 2000) cloned in the vector pAA121 as a template. The PCR product was digested with EcoRI and HindIII and cloned into EcoRI–HindIII pAA121 vector. The promoter was subcloned into pRW50 on an EcoRI–HindIII fragment as required.
Mutagenesis of melR
Primers used for random and site-directed mutagenesis of melR are listed in Table 2. A library of random mutations in the segment of melR encoding residues 100–302 was prepared by error-prone PCR (Zhou et al., 1991) using the primers D39890 and D4600 with pJW15 as a template. PCR products were digested with NsiI and HindIII and cloned into pJW15 NsiI–HindIII vector to generate a library of pJW15 mutant derivatives. A second library carrying random mutations throughout the entire melR gene was generated using primers D3407 and D4600. PCR products were digested with EcoRI and HindIII and cloned into pJW15 EcoRI–HindIII vector. Libraries were transformed into WAM132 cells containing pRW50 carrying the p39T p59T derivative of the JK19 fragment. A further library of melR derivatives, in which codon 273 was completely randomized, was constructed in the vector pJW15 by megaprimer PCR. The first round of PCR used a mutagenic primer, D38405, that had a degenerate sequence at codon 273, in conjunction with the flanking primer D4600 and the template pJW15. This created a DNA fragment that was used as a primer in a second PCR with the flanking primer D39890. The final PCR product was cloned into pJW15 and transformed into the tester strain described above. Strong Lac+ colonies were selected, and the melR insert in pJW15 isolated from these derivatives was sequenced using primers D3407 and D39890. In these experiments, we also isolated colonies with weak Lac+ phenotypes and found that these carried pJW15 derivatives that give MelR-dependent activation of the melAB promoter in the absence of melibiose (C. L. Webster, unpublished data).
Cloned melR in plasmids pLG314 or pJW15 was used as templates for site-directed mutagenesis of residues 257–265. Mutagenic primers were used in conjunction with the flanking primer D38105 or D4600 to generate PCR products that were purified and used as a megaprimer in a second round of PCR with the appropriate flanking primer (D39890 or D38424). The resulting DNA fragments were restricted and cloned into pLG314 or pJW15 as appropriate.
Mutagenesis of rpoD
Error-prone PCR, using primers D38796 and D29200, was used to prepare a library of random mutations in the segment of the rpoD gene encoding residues 536–613, as in our previous work (Rhodius and Busby, 2000). The library was transformed into WAM132 cells carrying JK16 cloned into pRW50 and pLG314 encoding DK261 MelR.
Measurement of promoter activities in vivo
DNA fragments containing the melAB promoter were cloned into pRW50, a low-copy-number lac expression vector, to generate pmelAB::lac fusions. β-Galactosidase levels in cells carrying these recombinants were measured by the Miller (1972) method. Activities, shown in Miller units, are the average of at least three independent experiments, and error bars show one standard deviation either side of the mean. Cells were grown in media either with or without melibiose exactly as in our previous work (Webster et al., 1987): data shown are those obtained from cultures with melibiose unless stated otherwise. Assays were performed in a ΔmelRΔlac (WAM132) background carrying different pJW15 or pLG314 derivatives, encoding wild-type and mutant MelR, and different pVRσ derivatives encoding wild-type or mutant σ70.
Preparation of proteins
For the EMSA experiments shown in Fig. 7A, MelR was overexpressed in BL21 (λDE3) cells carrying plasmids pLysS and pCM117-303. Crude cell extracts containing MelR were prepared and used as described by Michán et al. (1995). For FeBABE experiments, we used a derivative of pCM117-303, encoding MelR with a single cysteine at position 269, to overexpress the protein (Grainger et al., 2003). MelR used for FeBABE experiments was overexpressed, purified and conjugated with FeBABE as in our previous work. Protein concentrations were determined according to the protocol of Bradford (1976).
DNA templates were derived from caesium chloride-purified preparations of plasmid pAA121 carrying the JK19 or JK31 fragments. Purified AatII–HindIII fragments were purified and labelled at the HindIII end using [γ-32P]-ATP and polynucleotide kinase. FeBABE footprinting experiments were performed as in our previous work (Grainger et al., 2003). DNA fragments were used at a final concentration of 10 nM in buffer containing 60 mM Hepes, pH 8.0, 1.25 µM potassium glutamate and 40 µg ml−1 herring sperm DNA. Incubations contained 10 mM melibiose and 0.75 µM MelR as indicated.
This work was supported by the UK BBSRC with a project grant to S.J.W.B. and E.I.H., and by the UK MRC with a PhD studentship award to D.C.G. We are grateful to Robert Martin, Rick Wolf, Robert Schleif and Susan Egan for many helpful discussions. We would also like to thank Jen-Min Huang for help in constructing derivatives of pLG314 encoding MelR DA261 and MelR DK261.