Repression of the Escherichia coli melR promoter by MelR: evidence that efficient repression requires the formation of a repression loop

Transcriptome data are insufficient to control false discoveries in regulatory network inference

Inference of causal transcriptional regulatory networks (TRNs) from transcriptomic data suffers notoriously from false positives. Approaches to control the false discovery rate (FDR), for example, via permutation, bootstrapping, or multivariate Gaussian distributions, suffer from several complications: difficulty in distinguishing direct from indirect regulation, nonlinear effects, and causal structure inference requiring "causal sufficiency," meaning experiments that are free of any unmeasured, confounding variables. Here, we use a recently developed statistical framework, model-X knockoffs, to control the FDR while accounting for indirect effects, nonlinear dose-response, and user-provided covariates. We adjust the procedure to estimate the FDR correctly even when measured against incomplete gold standards. However, benchmarking against chromatin immunoprecipitation (ChIP) and other gold standards reveals higher observed than reported FDR. This indicates that unmeasured confounding is a major driver of FDR in TRN inference. A record of this paper's transparent peer review process is included in the supplemental information.

Characterisation of a putative AraC transcriptional regulator from Mycobacterium smegmatis

MSMEG_0307 is annotated as a transcriptional regulator belonging to the AraC protein family and is located adjacent to the arylamine N-acetyltransferase (nat) gene in Mycobacterium smegmatis, in a gene cluster, conserved in most environmental mycobacterial species. In order to elucidate the function of the AraC protein from the nat operon in M. smegmatis, two conserved palindromic DNA motifs were identified using bioinformatics and tested for protein binding using electrophoretic mobility shift assays with a recombinant form of the AraC protein. We identified the formation of a DNA:AraC protein complex with one of the motifs as well as the presence of this motif in 20 loci across the whole genome of M. smegmatis, supporting the existence of an AraC controlled regulon. To characterise the effects of AraC in the regulation of the nat operon genes, as well as to gain further insight into its function, we generated a ΔaraC mutant strain where the araC gene was replaced by a hygromycin resistance marker. The level of expression of the nat and MSMEG_0308 genes was down-regulated in the ΔaraC strain when compared to the wild type strain indicating an activator effect of the AraC protein on the expression of the nat operon genes.

Microarray analysis of the Ler regulon in enteropathogenic and enterohaemorrhagic Escherichia coli strains

The type III protein secretion system is an important pathogenicity factor of enteropathogenic and enterohaemorrhagic Escherichia coli pathotypes. The genes encoding this apparatus are located on a pathogenicity island (the locus of enterocyte effacement) and are transcriptionally activated by the master regulator Ler. In each pathotype Ler is also known to regulate genes located elsewhere on the chromosome, but the full extent of the Ler regulon is unclear, especially for enteropathogenic E. coli. The Ler regulon was defined for two strains of E. coli: E2348/69 (enteropathogenic) and EDL933 (enterohaemorrhagic) in mid and late log phases of growth by DNA microarray analysis of the transcriptomes of wild-type and ler mutant versions of each strain. In both strains the Ler regulon is focused on the locus of enterocyte effacement - all major transcriptional units of which are activated by Ler, with the sole exception of the LEE1 operon during mid-log phase growth in E2348/69. However, the Ler regulon does extend more widely and also includes unlinked pathogenicity genes: in E2348/69 more than 50 genes outside of this locus were regulated, including a number of known or potential pathogenicity determinants; in EDL933 only 4 extra-LEE genes, again including known pathogenicity factors, were activated. In E2348/69, where the Ler regulon is clearly growth phase dependent, a number of genes including the plasmid-encoded regulator operon perABC, were found to be negatively regulated by Ler. Negative regulation by Ler of PerC, itself a positive regulator of the ler promoter, suggests a negative feedback loop involving these proteins.

Activation of sarX by Rbf is required for biofilm formation and icaADBC expression in Staphylococcus aureus

A major constituent of many Staphylococcus aureus biofilms is a polysaccharide known as the polysaccharide intercellular adhesin, or poly N-acetylglucosamine (PIA/PNAG). PIA/PNAG is synthesized by the 4 gene products of the icaADBC operon, which is negatively regulated by the divergently transcribed icaR gene. We previously reported the identification of a gene, rbf, involved in the positive transcriptional regulation of icaADBC transcription by repressing icaR in S. aureus strain 8325-4. However, we were unable to show binding of Rbf to DNA upstream of icaR or icaA, suggesting that Rbf may control expression of an unknown factor(s) that, in turn, regulates ica expression. Here we report that the unknown factor is SarX protein. Results from epistasis assays and genetic complementation analyses suggest that Rbf upregulates SarX, which then downregulates IcaR, thereby activating icaADBC. Electrophoretic mobility shift assays revealed that SarX protein bound to a sequence upstream of icaR within the icaA coding region. Cross-linking and immunoprecipitation experiments further suggested that Rbf binds to the sarX promoter in S. aureus. These results demonstrate that Rbf and SarX represent a regulatory cascade that promotes PIA-dependent biofilm formation in S. aureus.

Two DNA sites for MelR in the same orientation are sufficient for optimal MelR-dependent repression at the Escherichia coli melR promoter

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.

Transcriptional activation of the Azotobacter vinelandii polyhydroxybutyrate biosynthetic genes phbBAC by PhbR and RpoS

We previously showed that in Azotobacter vinelandii, accumulation of polyhydroxybutyrate (PHB) occurs mainly during the stationary phase, and that a mutation in phbR, encoding a transcriptional regulator of the AraC family, reduces PHB accumulation. In this study, we characterized the roles of PhbR and RpoS, a central regulator during stationary phase in bacteria, in the regulation of expression of the PHB biosynthetic operon phbBAC and phbR. We showed that inactivation of rpoS reduced PHB accumulation, similar to the phbR mutation, and inactivation of both rpoS and phbR resulted in an inability to produce PHB. We carried out expression studies with the wild-type, and the rpoS, phbR and double rpoS-phbR mutant strains, using quantitative RT-PCR, as well as phbB : : gusA and phbR : : gusA gene fusions. These studies showed that both PhbR and RpoS act as activators of phbB and phbR, and revealed a role for PhbR as an autoactivator. We also demonstrated that PhbR binds specifically to two almost identical 18 bp sites, TGTCACCAA-N(4)-CACTA and TGTCACCAA-N(4)-CAGTA, present in the phbB promoter region. The activation of phbB and phbR transcription by RpoS reported here is in agreement with the observation that accumulation of PHB in A. vinelandii occurs mainly during the stationary phase.

Rbf promotes biofilm formation by Staphylococcus aureus via repression of icaR, a negative regulator of icaADBC

We previously reported the identification of a gene, rbf, involved in the regulation of biofilm formation by Staphylococcus aureus 8325-4. In an effort to study the mechanism of regulation, microarrays were used to compare the transcription profiles of the wild-type strain with an rbf mutant and an rbf overexpression strain of the clinical isolate UAMS-1. Among the genes affected by rbf overexpression are those of the intercellular adhesion (ica) locus; however, expression of these genes was not affected by an rbf deletion in the chromosome. The icaADBC genes are responsible for production of poly-N-acetylglucosamine (PNAG), a major constituent of biofilm. The icaR gene encodes a negative regulator of icaADBC. In UAMS-1 carrying an Rbf-encoding plasmid, Rbf was found to repress icaR transcription with a concomitant increase in icaADBC expression and increased PNAG and biofilm production relative to isogenic strains lacking the plasmid. Sequencing of the rbf gene from UAMS-1 showed that there was a 2-bp insertion affecting the 50th codon of the rbf open reading frame, suggesting that rbf is a pseudogene in UAMS-1. This finding explains why deletion of rbf had no effect on biofilm formation in UAMS-1. To further characterize the Rbf regulation on biofilm we compared biofilm formation, icaA and icaR transcription, and PNAG production in 8325-4 and its isogenic rbf and icaR single mutants and an rbf icaR double mutant. Our results are consistent with a model wherein rbf represses synthesis of icaR, which in turn results in derepression of icaADBC and increased PNAG production. Furthermore, purified rbf did not bind to the icaR or icaA promoter region, suggesting that rbf controls expression of an unknown factor(s) that represses icaR. The role of rbf in controlling the S. aureus biofilm phenotype was further demonstrated in a clinical strain, MW2.

Differences in the mechanism of the allosteric l-rhamnose responses of the AraC/XylS family transcription activators RhaS and RhaR

Proteins in the largest subset of AraC/XylS family transcription activators, including RhaS and RhaR, have C-terminal domains (CTDs) that mediate DNA-binding and transcription activation, and N-terminal domains (NTDs) that mediate dimerization and effector binding. The mechanism of the allosteric effector response in this family has been identified only for AraC. Here, we investigated the mechanism by which RhaS and RhaR respond to their effector, l-rhamnose. Unlike AraC, N-terminal truncations suggested that RhaS and RhaR do not use an N-terminal arm to inhibit activity in the absence of effector. We used random mutagenesis to isolate RhaS and RhaR variants with enhanced activation in the absence of l-rhamnose. NTD substitutions largely clustered around the predicted l-rhamnose-binding pockets, suggesting that they mimic the structural outcome of effector binding to the wild-type proteins. RhaS-CTD substitutions clustered in the first HTH motif, and suggested that l-rhamnose induces improved DNA binding. In contrast, RhaR-CTD substitutions clustered at a single residue in the second HTH motif, at a position consistent with improved RNAP contacts. We propose separate allosteric mechanisms for the two proteins: Without l-rhamnose, RhaS does not effectively bind DNA while RhaR does not effectively contact RNAP. Upon l-rhamnose binding, both proteins undergo structural changes that enable transcription activation.

Autoregulation of the Escherichia coli melR promoter: repression involves four molecules of MelR

The Escherichia coli MelR protein is a transcription activator that autoregulates its own promoter by repressing transcription initiation. Optimal repression requires MelR binding to a site that overlaps the melR transcription start point and to upstream sites. In this work, we have investigated the different determinants needed for optimal repression and their spatial requirements. We show that repression requires a complex involving four DNA-bound MelR molecules, and that the global CRP regulator plays little or no role.

Roles of effectors in XylS-dependent transcription activation: intramolecular domain derepression and DNA binding

XylS, an AraC family protein, activates transcription from the benzoate degradation pathway Pm promoter in the presence of a substrate effector such as 3-methylbenzoate (3MB). We developed a procedure to obtain XylS-enriched preparations which proved suitable to analyze its activation mechanism. XylS showed specific 3MB-independent binding to its target operator, which became strictly 3MB dependent in a dimerization-defective mutant. We demonstrated that the N-terminal domain of the protein can make linker-independent interactions with the C-terminal domain and inhibit its capacity to bind DNA. Interactions are hampered in the presence of 3MB effector. We propose two independent roles for 3MB in XylS activation: in addition to its known influence favoring protein dimerization, the effector is able to modify XylS conformation to trigger N-terminal domain intramolecular derepression. We also show that activation by XylS involves RNA polymerase recruitment to the Pm promoter as demonstrated by chromatin immunoprecipitation assays. RNA polymerase switching in Pm transcription was reproduced in in vitro transcription assays. All sigma(32)-, sigma(38)-, and sigma(70)-dependent RNA polymerases were able to carry out Pm transcription in a rigorous XylS-dependent manner, as demonstrated by the formation of open complexes only in the presence of the regulator.

Transcription activation by the DNA-binding domain of the AraC family protein RhaS in the absence of its effector-binding domain

The Escherichia coli L-rhamnose-responsive transcription activators RhaS and RhaR both consist of two domains, a C-terminal DNA-binding domain and an N-terminal dimerization domain. Both function as dimers and only activate transcription in the presence of L-rhamnose. Here, we examined the ability of the DNA-binding domains of RhaS (RhaS-CTD) and RhaR (RhaR-CTD) to bind to DNA and activate transcription. RhaS-CTD and RhaR-CTD were both shown by DNase I footprinting to be capable of binding specifically to the appropriate DNA sites. In vivo as well as in vitro transcription assays showed that RhaS-CTD could activate transcription to high levels, whereas RhaR-CTD was capable of only very low levels of transcription activation. As expected, RhaS-CTD did not require the presence of L-rhamnose to activate transcription. The upstream half-site at rhaBAD and the downstream half-site at rhaT were found to be the strongest of the known RhaS half-sites, and a new putative RhaS half-site with comparable strength to known sites was identified. Given that cyclic AMP receptor protein (CRP), the second activator required for full rhaBAD expression, cannot activate rhaBAD expression in a DeltarhaS strain, it was of interest to test whether CRP could activate transcription in combination with RhaS-CTD. We found that RhaS-CTD allowed significant activation by CRP, both in vivo and in vitro, although full-length RhaS allowed somewhat greater CRP activation. We conclude that RhaS-CTD contains all of the determinants necessary for transcription activation by RhaS.

Mutational analysis of the Escherichia coli melR gene suggests a two-state concerted model to explain transcriptional activation and repression in the melibiose operon

Transcription of the Escherichia coli melAB operon is regulated by the MelR protein, an AraC family member whose activity is modulated by the binding of melibiose. In the absence of melibiose, MelR is unable to activate the melAB promoter but autoregulates its own expression by repressing the melR promoter. Melibiose triggers MelR-dependent activation of the melAB promoter and relieves MelR-dependent repression of the melR promoter. Twenty-nine single amino acid substitutions in MelR that result in partial melibiose-independent activation of the melAB promoter have been identified. Combinations of different substitutions result in almost complete melibiose-independent activation of the melAB promoter. MelR carrying each of the single substitutions is less able to repress the melR promoter, while MelR carrying some combinations of substitutions is completely unable to repress the melR promoter. These results argue that different conformational states of MelR are responsible for activation of the melAB promoter and repression of the melR promoter. Supporting evidence for this is provided by the isolation of substitutions in MelR that block melibiose-dependent activation of the melAB promoter while not changing melibiose-independent repression of the melR promoter. Additional experiments with a bacterial two-hybrid system suggest that interactions between MelR subunits differ according to the two conformational states.

Characterization of functional domains of the Vibrio cholerae virulence regulator ToxT

The toxT gene encodes an AraC family transcriptional activator that is responsible for regulating virulence gene expression in Vibrio cholerae. Analysis of ToxT by dominant/negative assays and a LexA-based reporter system demonstrated that the N-terminus of the protein contains dimerization determinants, indicating that ToxT likely functions as a dimer. Additionally, a natural variant of ToxT with only 60% identity in the N-terminus, as well as a mutant form of ToxT with an altered amino acid in the N-terminus (L107F), exhibited altered transcriptional responses to bile, suggesting that the N-terminus is involved in environmental sensing. The C-terminus of ToxT functions to bind DNA and requires dimerization for stable binding in vitro, as demonstrated by gel shift analysis. Interestingly, a dimerized form of the ToxT C-terminus is able to bind DNA in vitro but is transcriptionally inactive in vivo, indicating that the N-terminus contains determinants that are required for transcriptional activation. These results provide a model for a two-domain structure for ToxT, with an N-terminal dimerization and environmental sensing domain and a C-terminal DNA-binding domain; unlike other well-studied AraC family proteins, both domains of ToxT appear to be required for transcriptional activation.

PchR-box recognition by the AraC-type regulator PchR of Pseudomonas aeruginosa requires the siderophore pyochelin as an effector

Under iron limitation, the opportunistic human pathogen Pseudomonas aeruginosa produces the siderophore pyochelin. When secreted into the extracellular environment, pyochelin complexes ferric ions and delivers them, via the outer membrane receptor FptA, to the bacterial cytoplasm. Extracellular pyochelin also acts as a signalling molecule, inducing the expression of pyochelin biosynthesis and uptake genes by a mechanism involving the AraC-type regulator PchR. We have identified a 32 bp conserved sequence element (PchR-box) in promoter regions of pyochelin-controlled genes and we show that the PchR-box in the pchR-pchDCBA intergenic region is essential for the induction of the pyochelin biosynthetic operon pchDCBA and the repression of the divergently transcribed pchR gene. PchR was purified as a fusion with maltose-binding protein (MBP). Mobility shift assays demonstrated specific binding of MBP-PchR to the PchR-box in the presence, but not in the absence of pyochelin and iron. PchR-box mutations that interfered with pyochelin-dependent regulation in vivo, also affected pyochelin-dependent PchR-box recognition in vitro. We conclude that pyochelin, probably in its iron-loaded state, is the intracellular effector required for PchR-mediated regulation. The fact that extracellular pyochelin triggers this regulation suggests that the siderophore can enter the cytoplasm.

The AraC-type regulator RipA represses aconitase and other iron proteins from Corynebacterium under iron limitation and is itself repressed by DtxR

The mRNA level of the aconitase gene acn of Corynebacterium glutamicum is reduced under iron limitation. Here we show that an AraC-type regulator, termed RipA for "regulator of iron proteins A," is involved in this type of regulation. A C. glutamicum DeltaripA mutant has a 2-fold higher aconitase activity than the wild type under iron limitation, but not under iron excess. Comparison of the mRNA profiles of the DeltaripA mutant and the wild type revealed that the acn mRNA level was increased in the DeltaripA mutant under iron limitation, but not under iron excess, indicating a repressor function of RipA. Besides acn, some other genes showed increased mRNA levels in the DeltaripA mutant under iron starvation (i.e. those encoding succinate dehydrogenase (sdhCAB), nitrate/nitrite transporter and nitrate reductase (narKGHJI), isopropylmalate dehydratase (leuCD), catechol 1,2-dioxygenase (catA), and phosphotransacetylase (pta)). Most of these proteins contain iron. Purified RipA binds to the upstream regions of all operons mentioned above and in addition to that of the catalase gene (katA). From 13 identified binding sites, the RipA consensus binding motif RRGCGN(4)RYGAC was deduced. Expression of ripA itself is repressed under iron excess by DtxR, since purified DtxR binds to a well conserved binding site upstream of ripA. Thus, repression of acn and the other target genes indicated above under iron limitation involves a regulatory cascade of two repressors, DtxR and its target RipA. The modulation of the intracellular iron usage by RipA supplements mechanisms for iron acquisition that are directly regulated by DtxR.

Transcription activation at the Escherichia coli melAB promoter: interactions of MelR with the C-terminal domain of the RNA polymerase alpha subunit

We have investigated the role of the RNA polymerase alpha subunit during MelR-dependent activation of transcription at the Escherichia coli melAB promoter. To do this, we used a simplified melAB promoter derivative that is dependent on MelR binding at two 18 bp sites, located from position -34 to -51 and from position -54 to -71, upstream of the transcription start site. Results from experiments with hydroxyl radical footprinting, and with RNA polymerase, carrying alpha subunits that were tagged with a chemical nuclease, show that the C-terminal domains of the RNA polymerase alpha subunits are located near position -52 and near position -72 during transcription activation. We demonstrate that the C-terminal domain of the RNA polymerase alpha subunit is needed for open complex formation, and we describe two experiments showing that the RNA polymerase alpha subunit can interact with MelR. Finally, we used alanine scanning to identify determinants in the C-terminal domain of the RNA polymerase alpha subunit that are important for MelR-dependent activation of the melAB promoter.

Expression of the Azotobacter vinelandii poly-beta-hydroxybutyrate biosynthetic phbBAC operon is driven by two overlapping promoters and is dependent on the transcriptional activator PhbR

The Azotobacter vinelandii phbBAC genes encode the enzymes for poly-beta-hydroxybutyrate (PHB) synthesis. The phbR gene, which is located upstream of and in the opposite direction of phbBAC, encodes PhbR, a transcriptional activator which is a member of the AraC family of activators. Here we report that a mutation in phbR reduced PHB accumulation and transcription of a phbB-lacZ fusion. We also report that phbB is transcribed from two overlapping promoters, p(B)1 and p(B)2. The region corresponding to the -35 region of p(B)1 overlaps the p(B)2 -10 region. In the phbR mutant, expression of phbB from the p(B)1 promoter is significantly reduced, whereas expression from the p(B)2 promoter is slightly increased. Two phbR promoters, p(R)1 and p(R)2, were also identified. Transcription from p(R)2 was shown to be dependent on sigma(S). Six conserved 18-bp sites, designated R1 to R6, are present within the phbR-phbB intergenic region and are proposed to be putative binding targets for PhbR. R1 overlaps the -35 region of the p(B)1 promoter. A model for the regulation of phbB transcription by PhbR is proposed.

The central domain of Escherichia coli TyrR is responsible for hexamerization associated with tyrosine-mediated repression of gene expression

TyrR from Escherichia coli regulates the expression of genes for aromatic amino acid uptake and biosynthesis. Its central ATP-hydrolyzing domain is similar to conserved domains of bacterial regulatory proteins that interact with RNA polymerase holoenzyme associated with the alternative sigma factor, sigma(54). It is also related to the common module of the AAA+ superfamily of proteins that is involved in a wide range of cellular activities. We expressed and purified two TyrR central domain polypeptides. The fragment comprising residues 188-467, called TyrR-(188-467), was soluble and stable, in contrast to that corresponding to the conserved core from residues 193 to 433. TyrR-(188-467) bound ATP and rhodamine-ATP with association constants 2- to 5-fold lower than TyrR and hydrolyzed ATP at five times the rate of TyrR. In contrast to TyrR, which is predominantly dimeric at protein concentrations less than 10 microm in the absence of ligands, or in the presence of ATP or tyrosine alone, TyrR-(188-467) is a monomer, even at high protein concentrations. Tyrosine in the presence of ATP or ATPgammaS promotes the oligomerization of TyrR-(188-467) to a hexamer. Tyrosine-dependent repression of gene transcription by TyrR therefore depends on ligand binding and hexamerization determinants located in the central domain polypeptide TyrR-(188-467).

DNA binding of the transcription activator protein MelR from Escherichia coli and its C-terminal domain

MelR is an Escherichia coli transcription factor belonging to the AraC family. It activates expression of the melAB operon in response to melibiose. Full-length MelR (MelR303) binds to two pairs of sites upstream of the melAB transcription start site, denoted sites 1' and 1 and sites 2 and 2', and to a fifth site, R, which overlaps the divergent melR promoter. The C-terminal domain of MelR (MelR173) does not activate transcription. Here we show that, like MelR303, when MelR173 binds to sites 1 and 2 it recruits CRP to bind between these sites. Hence, the C-terminal domain is involved in heterologous interactions. MelR173 binds to the R site, which has 11 of 18 bp identical to sites 1 and 2 but, surprisingly, does not bind to site 1', which has 12 of 18 bp identical, nor to site 2'. Using electrophoretic mobility shift assays, we show that the binding of MelR303 to sites 1' and 2' is due to cooperative binding with the adjacent site. This homologous cooperativity requires the N-terminal domain of the protein. Activation of the melAB promoter requires MelR to occupy site 2', which overlaps the -35 hexamer. Hence, both domains of MelR are required for transcription activation.

The AraC transcriptional activators

The AraC family of bacterial transcriptional activators regulate diverse genetic systems. Recent X-ray diffraction studies show that the monomeric MarA and Rob activators bind to their asymmetric degenerate DNA sites via two different helix-turn-helix elements. Activation by MarA, SoxS or Rob requires a particular orientation of the asymmetric binding sequence (and hence the activator), depending on its distance from the -10 RNAP signal. Genetic studies are beginning to clarify how the activators interact with RNAP. Growing evidence suggests that for the sugar metabolism activators, multiple binding sites upstream of the promoter anchor the activator in a repressing or nonactivating configuration. By interaction with the sugar and/or CRP, the activator is allosterically altered so it can bind a new set of sites that enable it to activate the promoter. Surprisingly, the virulence activator, Rns, must bind to both upstream and downstream sites in order to activate the rns promoter.

Transcription activation at the Escherichia coli melAB promoter: the role of MelR and the cyclic AMP receptor protein

MelR is a melibiose-triggered transcription activator that belongs to the AraC family of transcription factors. Using purified Escherichia coli RNA polymerase and a cloned DNA fragment carrying the entire melibiose operon intergenic region, we have demonstrated in vitro open complex formation and activation of transcription initiation at the melAB promoter. This activation is dependent on MelR and melibiose. These studies also show that the cyclic AMP receptor protein (CRP) interacts with the melAB promoter and increases MelR-dependent transcription activation. DNAase I footprinting has been exploited to investigate the location of MelR-and CRP-binding sites at the melAB promoter. We showed previously that MelR binds to two identical 18 bp target sequences centred at position -100.5 (Site 1) and position -62.5 (Site 2). In this work, we show that MelR additionally binds to two other related 18 bp sequences: Site 1', centred at position -120.5, located immediately upstream of Site 1, and Site R, at position -238.5, which overlaps the transcription start site of the divergent melR promoter. MelR can bind to Site 1', Site 1, Site 2 and Site R, in both the absence and the presence of melibiose. However, in the presence of melibiose, MelR also binds to a fifth site (Site 2', centred at position -42.5) located immediately downstream of Site 2, and overlapping the -35 region of the melAB promoter. Additionally, although CRP is unable to bind to the melAB promoter in the absence of MelR, in the presence of MelR, it binds to a site located between MelR binding Site 1 and Site 2. Thus, tandem-bound MelR recruits CRP to the MelR. We propose that expression from the melAB promoter has an absolute requirement for MelR binding to Site 2'. Optimal expression of the melAB promoter requires Sites 1', Site 1, Site 2 and Site 2'; CRP acts as a 'bridge' between MelR bound at Sites 1' and 1 and at Sites 2 and 2', increasing expression from the melAB promoter. In support of this model, we show that improvement of the base sequence of Site 2' removes the requirement for Site 1' and Site 1, and short circuits the effects of CRP.