Correspondence: Stephen Busby, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. Tel.: 44 121 414 5439; fax: 44 121 414 5925; e-mail: firstname.lastname@example.org
The Escherichia coli melR gene encodes the MelR transcription factor that controls melibiose utilization. Expression of melR is autoregulated by MelR, which represses the melR promoter by binding to a target that overlaps the transcript start. Here, we show that MelR-dependent repression of the melR promoter can be enhanced by the presence of a second single DNA site for MelR located up to 250 base pairs upstream. Parallels with AraC-dependent repression at the araC–araBAD regulatory region and the possibility of the MelR-dependent repression loop formation are discussed. The results show that MelR bound at two distal loci can cooperate together in transcriptional repression.
The activity of many bacterial promoters is controlled by transcription repressors, and many cases have now been described where efficient repression requires interaction between repressors bound at two separated DNA targets, resulting in looping of the intervening DNA (Browning & Busby, 2004). One of the first cases to be described was repression by the Escherichia coli AraC protein at the araC-araBAD intergenic regulatory region, which requires AraC binding to two target sites, I1 and O2, separated by 210 base pairs (reviewed by Schleif, 2010).
In previous work, we have studied the interactions of MelR at the E. coli melibiose operon regulatory region (Wade et al., 2000, 2001). MelR is a member of the AraC family of transcription factors and is essential for melibiose-dependent triggering of the melAB operon that encodes products needed for melibiose catabolism and transport. The melR gene is located upstream of the melAB operon, and the melR and melAB promoters are divergent, with the transcript start sites separated by 256 base pairs (Webster et al., 1987). We showed that transcription activation at the melAB promoter requires binding of MelR to four target sites (denoted 1′, 1, 2 and 2′) and binding of the global regulator, cyclic AMP receptor protein (CRP) to a target site located between MelR sites 1 and 2 (Fig. 1a). Transcription initiation at the melR promoter is dependent on activation by CRP and is repressed by MelR binding to a single target site (denoted R) overlapping the melR transcript start. Wade et al. (2000) reported that efficient MelR-dependent repression of the melR promoter requires upstream sequences that covered the melAB promoter and that the most important element in repression is MelR binding at target site 2. Further detailed analysis by Samarasinghe et al. (2008) showed that MelR bound at sites 1 and 1′ plays a role in repression, and images from atomic force microscopy suggested that repression is due to a nucleoprotein complex consisting of four MelR subunits and ~170 base pairs of DNA between MelR-binding target site 2 and target site R.
Most members of the AraC family of transcription regulators function as homodimers of two subunits with the N-terminal domain of each subunit involved in ligand binding and dimerization, and the C-terminal domain responsible for DNA binding (Gallegos et al., 1997). C-terminal domains of AraC family members are highly conserved, carry two helix-turn-helix motifs and bind to asymmetric ~18 base pair target operator sequences. As it is well established that effective transcriptional repression can result from the two subunits of a single AraC dimer binding to two separated target sites (Schleif, 2010), and as MelR has been shown to dimerise (Bourgerie et al., 1997; Kahramanoglou et al., 2006), we revisited the E. coli melibiose operon regulatory region to investigate whether two DNA sites for MelR could be manipulated to produce efficient MelR-dependent repression of the melR promoter.
Materials and methods
In this work, we exploited the low-copy-number lac expression vector plasmid, pRW50, encoding resistance to tetracycline (Lodge et al., 1992). The starting points of the work were pRW50 derivatives carrying the TB22 and TB23 EcoRI-HindIII fragments (Fig. 1b) containing the E. coli melR promoter, as described by Samarasinghe et al. (2008). These recombinant pRW50 derivatives each carry a melR promoter::lacZ fusion, and they were propagated in the WAM1321 E. coli K-12 Δlac Δmel strain to measure melR promoter activity. Cells were grown in minimal medium with fructose, as a carbon source, and 35 μg mL−1 tetracycline, as in the study by Samarasinghe et al. (2008), and the Miller (1972) method was used to quantify β-galactosidase expression. For the different melR promoter fusions studied here in our conditions in the absence of MelR, β-galactosidase activity levels range from 360 to 400 standard Miller units. To quantify repression by MelR, cells also carried pJW15, encoding melR or empty vector, pJW15ΔmelR, and 80 μg mL−1 ampicillin was included in the media, as described by Kahramanoglou et al. (2006). In experiments to measure effects due to MalI, cells also carried pACYC–malI, encoding malI or empty vector, pACYC-ΔHN (Lloyd et al., 2010), and 10 μg mL−1 chloramphenicol was included in the media.
Derivatives of the TB22 and TB23 EcoRI-HindIII fragments, illustrated in Figs 1-4, were constructed by standard recombinant DNA technology using synthetic oligos purchased from Alta Bioscience (http://www.altabioscience.com/) and cloned into pRW50. The complete annotated base sequence of each fragment is listed in the Data S1 (Supporting information), and the DNA sequences were checked by the functional genomics facility of the University of Birmingham College of Life and Environmental Sciences (http://www.genomics.bham.ac.uk/).
Results and discussion
Optimal MelR-dependent repression due to two DNA sites for MelR in the same orientation
To investigate MelR-dependent repression at the melR promoter, we exploited different melR promoter::lac fusions carried by derivatives of the pRW50 low-copy-number lac expression plasmid, and β-galactosidase expression was measured in the WAM1321 E. coli K-12 Δlac Δmel host strain, containing either plasmid pJW15, encoding melR or empty vector. The starting experiment compared MelR-dependent repression of the melR promoter carried on the TB22 and the TB23 fragments, illustrated in Fig. 1b. The 251 base pair TB22 EcoRI-HindIII fragment carries DNA sequence from 192 base pairs upstream of the melR promoter transcript start (position −192) to 59 base pairs downstream (+59) and includes MelR target site 2, whilst in the 227 base pair TB23 fragment, MelR target site 2 is deleted. Results illustrated in Fig. 1c show that, as expected, the deletion of site 2 in the TB23 fragment causes a clear reduction in MelR-dependent repression of the melR promoter and confirms previous observations (Wade et al., 2000).
Previously, we identified the DNA target site for MelR subunits as an 18 base pair asymmetric sequence (Webster et al., 1987; Wade et al., 2001). By convention, we denote the location of each site by its centre with respect to the target promoter. Hence, at the melR promoter, MelR-binding site R is located at position +2.5 (i.e. between base pairs 2 and 3 downstream from the melR promoter transcript start) and MelR-binding site 2 is located at position −174.5 (i.e. between base pairs 174 and 175 upstream from the melR promoter transcript start). To investigate whether the binding of two MelR subunits could be sufficient to repress the melR promoter efficiently, we constructed the TB31, TB28 and TB33 fragments, illustrated in Fig. 1b. TB31 carries the core melR promoter sequences exactly as in TB22 and TB23, but DNA sequence upstream of position −80 is replaced by unrelated sequence. TB28 and TB33 are derivatives of TB31 carrying a single consensus 18 base pair site for MelR at position −174.5. In the TB28 fragment, this site has the same orientation as site 2 in the starting TB22 fragment, whilst, in TB33, this site has the opposite orientation, which is the same as for site R. Results illustrated in Fig. 1c show that MelR-dependent repression of the melR promoter in the TB31 and TB28 fragments is weak, but is increased to ~90% with the TB33 fragment.
The weak repression of the melR promoter carried on the TB31 fragment must be due to MelR binding to the target site R alone, and this result is consistent with the study by Wade et al. (2000). The ~90% repression found with the TB33 fragment must be due to MelR binding to the targets at both positions −174.5 and +2.5 and interaction between MelR bound at the two loci. Strikingly, repression is greatly reduced with the TB28 fragment (Fig. 1b and c), and this was expected from our previous work in which we replaced MelR target sites 1 and 1′ and the adjacent DNA site for CRP (Samarasinghe et al., 2008). Hence, as for AraC-dependent repression at the araC-araBAD intergenic region, efficient repression with just two bound regulator molecules depends on both target sequences being in the same orientation (Carra & Schleif, 1993).
MelR-dependent repression at the TB33 melR promoter is insensitive to spacing but reduced by a decoy DNA site for MelR
The centre-to-centre distance between the two DNA sites for MelR in the TB33 fragment is 176 base pairs. To investigate the relation between spacing and repression, we constructed a series of derivative fragments with the upstream MelR target at different locations, ranging from position −254.5 to position –−83.5. This is illustrated in Fig. 2, which also lists the percentage MelR-dependent repression for each case. The data show that repression is largely unaffected as the upstream DNA site for MelR is moved through ~170 base pairs, including translocation by five base pairs to the opposite face of the DNA helix (compare repression with TB33, TB332 and TB333).
A simple explanation for our observations is that repression of the melR promoter in the TB33 fragment and its derivatives is due to a bridging interaction between MelR bound at the upstream and downstream DNA sites and subsequent loop formation, and this interaction must be sufficiently flexible to accommodate different distances and different face of the DNA helix juxtapositions between the sites. We suppose that the lack of efficient repression with the TB23 fragment (Fig. 1c) must be due to interactions between MelR bound at site 1 and site 1′ that preclude interaction with site R (Fig. 1b). To investigate this, we constructed the TB33P and TB33R derivatives illustrated in Fig. 3a. These fragments are derivatives of TB33 that contain a supplementary upstream DNA site for MelR organised in either the same orientation (TB33P) or opposite orientation (TB33R). Results illustrated in Fig. 3b show that the presence of the supplementary DNA site for MelR significantly reduces MelR-dependent repression of the melR promoter, presumably because the supplementary site acts as a decoy for MelR–MelR interactions.
Effects of MalI binding between MelR targets on MelR-dependent repression
The flexibility in the spacing of the two DNA sites for MelR observed in the experiment illustrated in Fig. 2 suggested that it would be interesting to insert an intervening site for another DNA-binding protein. In recent work, Lloyd et al. (2010) identified the DNA site for the E. coli MalI repressor (that is a member of the LacI family) as a symmetric 16 base pair sequence element. Figure 4a illustrates an experiment where one or two of these elements were inserted between the two DNA sites for MelR in the TB334 fragment. Results in Fig. 4b show that in the absence of a plasmid encoding MalI, as expected, these insertions have but small effects on MelR-dependent repression of the melR promoter. However, with plasmid pACYC-malI, which encodes MalI, there is a clear small significant relief of repression with the TB334I-1 and TB334I-2 fragments carrying one or two MalI operator elements, but no relief with the control TB31, TB33 or TB334 fragments.
The expression of many transcription repressors is autoregulated by repression (Browning & Busby, 2004). Kahramanoglou et al. (2006) proposed a two-state model for MelR in which, in the absence of its ligand, melibiose, MelR acts as an autorepressor of its own production by repressing the melR promoter. Samarasinghe et al. (2008) showed that this repression was due to the formation of a nucleoprotein complex involving four MelR subunits. Here, we report that it is possible to construct simpler derivatives of the melR promoter where only two MelR targets are needed for efficient repression (Fig. 1), and there are clear parallels between this and AraC-dependent repression at the araC–araBAD intergenic region, where repression is dependent on interaction between two AraC subunits bound to targets separated by 210 base pairs (Schleif, 2010). An explanation for the observed repression with the TB33 fragment is that MelR subunits bound at the upstream and downstream DNA targets interact and result in loop formation, as for AraC. However, there appears to be more flexibility in how the two DNA sites for MelR can be juxtaposed, compared to AraC. Hence, AraC-dependent repression is disrupted by +5 base pair insertions (Lee & Schleif, 1989), whilst MelR-dependent repression is not (Fig. 2). The simplest explanation for this would be that the linker joining the N- and C-terminal domains is more flexible in MelR than in AraC. This flexibility is underscored by the experiment in Fig. 4 where MalI binding failed to completely disrupt repression. This experiment also argues that the mechanism of MelR-dependent repression with TB33 is different to the mechanism operating at the more complex wild type melibiose operon regulatory region in TB22 (Fig. 1), where repression depends on the formation of a nucleoprotein complex.
In the new constructs described here, efficient repression of the melR promoter by MelR requires interaction between MelR bound immediately adjacent to the transcript start and upstream-bound MelR, and this can be subverted by the insertion of a supplementary DNA site for MelR (Fig. 3). Hence, efficient repression results from two, but not from three, DNA sites for MelR. Our experiments underline the diversity of protein–DNA architectures that can be responsible for transcription repression.
This work was supported by the UK BBSRC with a project grant to S.J.W.B. and a summer studentship to D.D.