The bacterial strains and plasmids used in this study are listed in Supplementary Table 1. Bacteria were routinely cultured in Lysogeny Broth (LB) or in Dulbecco’s modified Eagle’s medium (DMEM) containing glucose [0.45% (wt/vol)] and L-glutamine (584 mg/l), but not sodium pyruvate (Gibco-BRL Life Technologies, Waltham, MA, USA), supplemented with 1% LB. When necessary, antibiotics were added at the following concentrations: kanamycin (Km) 30 μg ml^–1, ampicillin (Ap) 100 μg ml^–1, tetracycline (Tc) 12 μg ml^–1, streptomycin (Sm) 100 μg ml^–1 and chloramphenicol (Cm) 25 μg ml^–1. To induce the expression of the LEE virulence genes, bacteria were grown either in 50 ml DMEM under shaken or static + 5% CO₂ conditions at 37°C, while shaken LB broth was used as the non-inducing or repressing condition (Martinez-Laguna et al., 1999; Bustamante et al., 2001, 2011).

Standard genetic and molecular techniques were applied as described previously (Sambrook and Russell, 2001). Restriction enzymes were obtained from Thermo Scientific, Waltham, MA, USA and used according to the manufacturer’s instructions. PCR reactions were performed in 50 μl using Platinum Taq DNA polymerase (Invitrogen, Waltham, MA, USA). The oligonucleotides used for PCR amplification were synthesized at the Oligonucleotide Synthesis Facility of the Instituto de Biotecnología/UNAM, Cuernavaca, México, and are listed in Supplementary Table 2.

To construct the pT3GrlR plasmid, a DNA sequence including the RBS and the coding region of grlR was amplified by PCR using the GREPKE-F/GREPX-R oligonucleotide pair. The product obtained was digested with KpnI-XhoI enzymes and ligated into vector pMPM-T3 (Mayer, 1995), digested with the same enzymes. To generate plasmids pT3GrlRA, pT3GrlRA/I44A, and pT3GrlRA/R54A fragments were amplified by PCR with the oligonucleotides XHINTERGRLAF/HIGRLAR (Supplementary Table 2), using plasmids pTEPGrlA1, pTEPGrlA1/I44A, and pTEPGrlA1/R54A as templates (Jimenez et al., 2010), respectively. The resulting products were digested with XhoI-HindIII and cloned into the pT3GrlR plasmid digested with the same enzymes.

The pDnaK-CAT plasmid was constructed by PCR amplifying a fragment comprising from position −394 to + 127 of the transcription start site of the dnaK gene, using the DnaKF/DnaKR oligonucleotides. The product was digested with BamHI-HindIII enzymes and cloned into the pKK232-8 vector (Pharmacia Biotech) (Brosius, 1984), previously digested with the same enzymes.

Chromosomal DNA from the EPEC E2348/69 strain was used as the PCR template. All plasmids generated were verified by DNA sequencing and evaluated for their ability to complement the corresponding mutants through a profile of secreted proteins or western blotting against virulence proteins.

The grlR and grlA mutants were produced by generating chromosomal in-frame deletions of grlR codons 6 to 118 or grlA codons 6 to 132 by the sacB gene-based allelic exchange method as described previously (Edwards et al., 1998). Deletion of the grlRA operon spans from codon 6 of grlR to codon 132 of grlA. Suicide plasmids were generated by cloning PCR-amplified fragments containing the described deletions flanked by approximately 800 to 1000 bp on each side into XbaI/SacI-digested plasmid pRE112. The resulting suicide plasmids pRE112DgrlREP, pRE112DgrlAEP, and pRE112DgrlRAEP, respectively (Supplementary Table 1), were conjugated into strains WT EPEC E2348/69, Δhns (JPEP36) and Δler to generate single and double mutants (Supplementary Table 1).

To generate the chromosomally 3xFLAG-tagged strains, a modification of the λRed recombinase system was used as described previously (Datsenko and Wanner, 2000; Uzzau et al., 2001). The PCR fragment to tag the native grlR gene was generated using oligonucleotides grlR-FLAGH1P1 and grlR-FLAGH2P2 and the grlA gene using oligonucleotides grlA-FLAGH1P1 and grlA-FLAGH2P2, and plasmid pSUB11 DNA as template. The resulting PCR products were electroporated into WT EPEC or its ΔgrlR or ΔgrlA mutants to generate the EPEC grlR:3xFLAG, grlA:3xFLAG, grlR:3xFLAG ΔgrlA and ΔgrlR grlA:3xFLAG (Supplementary Table 1).

Single and double mutants were verified by PCR amplification and DNA sequencing. The tagged strains contain the tag in their native gene chromosomal locations; thus, are expressed from the grlRA operon promoter.

Chloramphenicol acetyltransferase (CAT) activity, derived from the expression of the transcriptional fusions to the cat reporter gene, was determined as follows (Martinez-Laguna et al., 1999). The strains containing the transcriptional fusions were grown in 5 ml of LB supplemented with antibiotics and incubated overnight at 37°C. The next day, culture pellets were adjusted to an OD₆₀₀ = 1.0 with 1X phosphate buffered saline (PBS) (10 mM Na2HPO4, 2 mM KH2PO4, 137 mM NaCl and 2.7 mM KCl) and used to inoculate 50 ml of LB or DMEM supplemented with antibiotics with one ml of each suspension. Cultures were incubated at 37°C under shaking or static plus 5% CO₂ growth conditions, and 1 ml samples were collected when the cultures reached an OD₆₀₀ = 0.8 and 1.0. The cell pellet was obtained by centrifugation at 14,000 rpm for 2 minutes and washed with 1 ml of TDTT buffer (Tris-Hcl 50 mM pH 7.8 and dithiothreitol 30 μM), centrifuged again and resuspended in 0.5 ml of the same buffer. Lysis was achieved by sonication for 5 min with pulses of 10 s per minute and 5 s rest. Soluble extracts were separated from cell debris by centrifugation at 12,000 rpm for 20 min at 4°C, and 5 μl aliquots of each extract were added in duplicate to a 96-well microtiter plate followed by the addition of 200 μl of the reaction mix containing 1 mM DTNB [5,58-dithio-bis(2-nitrobenzoic acid)] (Research Organics), 0.1 mM Tris-Hcl pH 7.8, 0.1 mM acetyl-CoA (Pharmacia LKB Biotech Inc., Alameda, CA, USA), and 0.1 mM chloramphenicol (Sigma Chemical Co., Saint Louis, MO, USA). Changes in absorbance at 410 nm were recorded every 5 s for 5 min with a CERES 900C automatic microplate reader and the KC3 program set to kinetic mode. A CAT standard curve (from 0 to 2500 U/ml) was used to interpolate the activities of each sample. The protein concentration of each extract was determined by adding 10 μl of each extract in duplicate and 200 μl of the reaction mixture (25 ml of solution A + 500 μl of solution B) of the “BCA Protein Assay Kit” (Pierce) to a 96-well plate. Then incubated at 37°C for 30 min; the absorbance was read at a wavelength of 562 nm using an automatic microplate reader CERES 900 C and the KC3 program. The CAT-specific activity was determined by dividing the CAT activity by the protein concentration of each extract. The data are expressed as μmol/min/mg of protein and are the results from at least three independent assays done in duplicate. The empty vector pKK232-8 did not show measurable levels of CAT activity in any of the strains used in the study (data not shown).

EPEC secreted proteins were analyzed as described previously (Deng et al., 2004). Briefly, bacterial cultures in LB and DMEM were incubated at 37°C under shaking or static conditions. Triplicate 1.5 ml samples were collected per culture when the cultures reached an OD₆₀₀ of 1.0 and were subjected to centrifugation at 17,900 x g for 5 min in Eppendorf tubes. Bacterial pellets were saved when convenient, and 1.3 ml of each supernatant was separated into fresh tubes and 160 μl of 100% trichloroacetic acid (TCA) was added per tube. Proteins were allowed to precipitate at 4°C overnight. Subsequently, the proteins were concentrated by centrifugation at 20,000 × g for 30 min and resuspended in 10 μl of 1× Laemmli loading buffer and the triplicate samples were mixed in the same tube, boiled for 5 min and resolved by 12% SDS-polyacrylamide gel electrophoresis (PAGE).

Bacteria were grown in DMEM or LB under shaking or static conditions until they reached an OD₆₀₀ of 1. Total extracts were obtained from 3 ml samples of the bacterial cultures. Cells were resuspended in 500 μl of urea solution (8 M) and lysed by sonication. Aliquots of total extracts were mixed with Laemmli buffer, boiled for 5 min and analyzed by SDS-PAGE (12% acrylamide) and, subsequently, transferred to 0.45-μm-pores-size PVDF membrane (Millipore) using a semi-dry transfer chamber (BioRad, Hercules, CA, USA).

The membrane was blocked in 10% non-fat milk and incubated with one of the following primary antibodies: a 1:10,000 dilution of monoclonal anti-Tir, a 1:7,500 dilution of polyclonal anti-intimin (kindly provided by J.A. Giron), a 1:20,000 dilution of polyclonal anti-EspA (kindly provided by J. Kaper), a 1:10,000 dilution of polyclonal anti-EscJ (kindly provided by Dr. Bertha González-Pedrajo), a 1:5,000 dilution of monoclonal anti-FLAG M2 (Sigma-Aldrich, Saint Louis, MO, USA), a 1:15,000 dilution of polyclonal anti-maltose binding protein (MBP) (New England Biolabs, Ipswich, MA, USA), a 1:10,000 dilution of polyclonal anti-DnaK (Invitrogen, Waltham, MA, USA) and a 1:50,000 dilution of monoclonal anti-GroEL (Sigma-Aldrich, Saint Louis, MO, USA). The membrane was washed with 1x phosphate buffer saline (PBS) containing 0.05% Tween 20 and incubated with a 1:10,000 dilution of secondary anti-rabbit or anti-mouse antibodies coupled to horseradish peroxidase (Thermo Scientific, Waltham, MA, USA). The membrane was developed with the commercial kit “Western Lightning Chemiluminescence Reagent Plus” (Perkin Elmer, Waltham, MA, USA) according to the manufacturer’s instructions.

EPEC E2348/69 grlR:3xFLAG, grlA:3xFLAG, ΔgrlR grlA:3xFLAG, and grlR:3xFLAG ΔgrlA strains were transformed with pT3GrlA1/I44A or pT3GrlA1/R54A and grown overnight in 5 ml of LB medium supplemented with antibiotics. The next day, 0.5 ml of these cultures were inoculated into 20 ml DMEM or LB medium supplemented with antibiotics and allowed to grow at 37°C with shaking. When the cultures reached an OD₆₀₀ of 1.0, 5 ml samples were taken and the cells were concentrated by centrifugation at 18,000 xg at 4°C, resuspended in 500 μl PBS and lysed by sonication. Samples were taken at an OD₆₀₀ of 0.3, 0.6, 0.8, and 1.0 from 50 ml cultures when needed.

Total extracts were resuspended in a non-denaturing buffer (50% glycerol, 25 mM Tris-Hcl pH 6.8 and 0.05% bromophenol Blue) and separated by 12% native PAGE at 4°C. The gel was transferred to a 0.45-μm-pore-size PVDF membrane (Millipore) and processed for western blotting.

Simultaneously, aliquots obtained from the same assays were processed for 12% SDS-PAGE and western blotting.

Data analyses were performed by two-way ANOVA followed by Tukey’s multiple comparisons posttest using GraphPad Prism version 8.4.3 (471) for Mac OS. P-values < 0.05 were considered statistically significant.

GrlR represses LEE gene expression mainly under non-permissive conditions such as growth in LB (Deng et al., 2004; Lio and Syu, 2004; Iyoda and Watanabe, 2005). To further explore this feature, we evaluated the expression of the LEE1, LEE2, LEE4, LEE5, and dnaK promoters fused to the cat reporter gene in WT EPEC E2348/69 and its ΔgrlR non-polar deletion mutant cultured under LEE inducing (DMEM/37°C) and repressing (LB/37°C) growth conditions. Under inducing conditions, the CAT reporter activity of all LEE fusions increased between 1.5- to 4-fold in EPEC ΔgrlR compared to the WT strain (Figure 1A); however, under repression conditions, the fold increase was higher, between 5 and 13.5 (Figure 1A). In contrast, dnaK-cat expression, used as a LEE unrelated control, was similar in both strains and growth conditions (Figure 1A).

The analysis of the T3SS-dependent protein secretion profile of WT, ΔgrlR and Δler EPEC strains showed that, under repressing conditions (LB), the ΔgrlR mutant abundantly secreted proteins into the medium, while the WT strain did not secrete detectable levels of LEE-encoded proteins (Figure 1B). This phenotype reversed by complementing the mutant strain with a plasmid containing the grlR gene (pT3GrlR) (Figure 1B). Growth under inducing conditions (DMEM) allowed secretion of LEE-encoded proteins by the WT strain and higher secretion levels by the ΔgrlR strain. Moreover, complementing the ΔgrlR mutant with the pT3GrlR plasmid suppressed protein secretion, even under inducing conditions (DMEM), indicating that GrlR overexpression can override the presence of GrlA under T3S permissive conditions (Figure 1B). As expected, the Δler mutant did not secrete under both growth conditions.

These results highlight the negative regulatory effect that GrlR exerts on the LEE genes in agreement with previous reports (Deng et al., 2004; Lio and Syu, 2004; Iyoda and Watanabe, 2005).

H-NS globally represses LEE-gene expression, while Ler acts as an antagonist to overcome this repression; thus, in the absence of H-NS, the role of Ler is dispensable (Friedberg et al., 1999; Leh et al., 2017; Shin, 2017). Moreover, GrlR expression from a plasmid strongly represses the expression of the LEE genes in WT EPEC even under conditions permissive to LEE expression (Deng et al., 2004; Lio and Syu, 2004; Jobichen et al., 2007). Evidence showing that GrlR and GrlA establish protein-protein interactions led to propose that GrlR negatively regulates LEE gene expression by preventing GrlA from binding to and activating the ler promoter (Creasey et al., 2003; Jobichen et al., 2007; Jimenez et al., 2010; Padavannil et al., 2013).

Based on such evidence, we hypothesized that overexpression of GrlR in EPEC lacking H-NS, which expresses LEE genes constitutively even under non-permissive growth conditions in a Ler- and GrlA-independent manner, would not affect LEE gene expression. Therefore, we analyzed the transcriptional activity of the LEE1-, LEE2-, LEE4-, and LEE5-cat fusions and the expression of LEE-encoded proteins in WT and Δhns EPEC strains carrying or not plasmid pT3GrlR (Supplementary Table 1). Consistent with previous reports for other A/E pathogens, overexpression of GrlR in WT EPEC repressed the transcriptional activity of all LEE fusions tested (Figure 2A), as well as the expression of Tir, intimin and EspA, as shown by western blot (Figure 2B), and of EspC (Supplementary Figure 1A), an autotransporter encoded outside the LEE and whose expression is also regulated by Ler (Elliott et al., 2000; Li et al., 2004; Vidal and Navarro-Garcia, 2008). Unexpectedly, GrlR also significantly repressed the transcriptional activity of the LEE promoters in the EPEC Δhns mutant (Figures 2A, B). Under these conditions, all strains similarly expressed proteins DnaK and MBP, used as loading controls, indicating that repression did not result from an unspecific pleiotropic effect in EPEC physiology due to GrlR overexpression.

To address this concept, we evaluated the effect of GrlR expressed from a plasmid under growth conditions where GrlA is not the primary activator of Ler expression. When EPEC grows in static DMEM cultures under 5% CO₂ atmosphere at 37°C, the EAF plasmid-encoded regulator PerC is the main activator of LEE1 promoter expression and therefore of ler (Bustamante et al., 2011). Thus, we analyzed the secreted protein profile and intimin and EspA expression in WT EPEC and the Δhns mutant harboring plasmid pT3GrlR or the empty vector pMPM-T3, grown in shaken and static + 5% CO₂ DMEM. As shown, GrlR expressed from pT3GrlR repressed protein secretion (Supplementary Figure 1A) and expression of EspA and Intimin in both strains and conditions, but not that of DnaK (Supplementary Figure 1B), confirming the results shown in Figure 1B.

These data suggested that GrlR can repress LEE gene expression through an alternative pathway, probably independent of its interaction with GrlA.

Ler and GrlA establish a positive regulatory feedback loop that counteracts the repression exerted by H-NS (Barba et al., 2005; Jimenez et al., 2010). According to the results described above, GrlR and H-NS act cooperatively to repress the expression of LEE genes. To further explore this observation, we evaluated the expression of the LEE1-cat, LEE2-cat, and LEE5-cat transcriptional fusions in EPEC WT and Δhns grown under inducing and repressing conditions and also analyzed the secretion profile of these strains from samples of their culture supernatants. Under inducing conditions, the activity of the LEE2-cat and LEE5-cat fusions increased about 2-fold in the absence of either of the two repressors, whereas that of the LEE1-cat fusion was 4.7-fold and 3.7-fold in the ΔgrlR and Δhns mutants, respectively (Figure 3A). Under repressing conditions, the derepression observed for the LEE1, LEE2, and LEE5 promoters operons was about 13-, 7-, and 14-fold, respectively, in the absence of GrlR; whereas in the Δhns mutant it was 2-, 3-, and 5-fold, respectively, relative to the activity displayed in the WT strain (Figure 3A). In agreement with these results, while the WT strain poorly secreted LEE-encoded proteins under repressing conditions (LB), the ΔgrlR and Δhns mutants showed a clear increase in protein secretion (Figure 3B).

To delve deeper into the role of both H-NS and GrlR in the negative regulation of the LEE genes, we analyzed the transcriptional activity of the LEE2-cat fusion and the protein secretion profile in WT EPEC and its Δler, Δhns, ΔgrlR, ΔgrlA, ΔgrlRA, ΔlerΔgrlR, and ΔgrlRΔhns isogenic mutants grown under inducing and repressing conditions (Figure 4). As expected, compared to the WT strain, the Δler and ΔgrlA strains significantly downregulated LEE2-cat fusion expression and protein secretion, while the ΔgrlR mutant expressed higher levels of LEE2-cat activity (Figures 4A, B). Interestingly, the ΔgrlRA double mutant displayed a phenotype similar to the ΔgrlA single mutant, confirming the important role of GrlA in ler activation and, indirectly, in derepression of Ler-dependent promoters that in the absence of GrlR remained repressed by H-NS (Figures 4A, B). In the ΔgrlRΔhns double mutant, the transcriptional activity of the LEE2-cat fusion showed expression levels similar to those observed in the ΔgrlR and Δhns single mutants under inducing conditions, while under repressing conditions, its expression was closer to the activity observed in the ΔgrlR strain (Figure 4A). These results coincided with the secreted protein profile since, in DMEM and LB, the ΔgrlR and Δhns single mutants and the double ΔgrlRΔhns mutant showed increased secretion in contrast to the WT strain (Figure 4B). Finally, the phenotype of the ΔgrlRΔler mutant confirmed that Ler is essential to antagonizing the repression exerted by H-NS regardless of the growth condition (Figures 4A, B).

The inclusion of the ΔlerΔhns double and the ΔlerΔhnsΔgrlRA quadruple mutant in this assay was cumbersome due to the poor growth shown by these strains. To overcome this limitation, we expressed ectopically H-NS^G113D, a dominant negative version of H-NS that carries a mutation affecting its DNA binding capacity without altering the oligomerization domain (Ueguchi et al., 1996), to inhibit H-NS activity. WT EPEC and its mutant derivatives Δler, Δhns, ΔgrlR, ΔgrlA, ΔgrlRA, ΔlerΔgrlR,ΔlerΔgrlA, ΔlerΔgrlRA and ΔescN carrying plasmid pT6-HNS/G113D, expressing H-NS^G113D under the control of an arabinose inducible promoter (Bustamante et al., 2008; Jimenez et al., 2010), were grown under repressing conditions with and without arabinose. We collected samples from these cultures to analyze the secreted proteins by PAGE and the expression of Tir and EscJ in total cell extracts by western blotting. In the absence of arabinose, we did not observed T3S of EspA, EspB and EspD nor Tir and EscJ expression in the WT or any strain carrying the Δler or ΔgrlA deletions (Figures 5A, B). In contrast, upon induction of H-NS^G113D, de-repression of T3S was evident even in the absence of Ler and/or GrlA, while the ΔgrlR mutant showed protein secretion and Tir and EscJ expression both without or with H-NS^G113D induction (Figures 5A, B). The ΔescN strain, a mutant defective in T3S, was used as a control that did not secrete virulence proteins even upon H-NS^G113D induction, except for EspC, a non-T3 secreted protein, whose gene is also repressed by H-NS and derepressed by Ler (Figure 5A). In contrast, Tir and EscJ were observed in total cell extracts of this strain when H-NS^G113D was induced (Figure 5B).

Taken together, these results indicate that GrlR and H-NS negatively regulate the expression of LEE operons at two independent levels that are indirectly cooperative (see Section “Discussion”).

We previously observed that under inducing conditions, the negative effect of GrlR is counteracted when co-expressed with GrlA in C. rodentium (Deng et al., 2004). A possible explanation of this phenotype is that the co-expression of both proteins prevents the titration effect that overexpressing only GrlR from a plasmid has on the native levels of GrlA. However, there is also the possibility that under these conditions, the interaction between GrlR and GrlA also plays a role in preventing the GrlR-mediated independent repression described above (Figure 2 and Supplementary Figure 1).

To evaluate this possibility, we transformed plasmids pT3GrlRA, pT3GrlR, pTEPGrlA1 and the empty vector pMPM-T3 (Supplementary Table 1) into WT EPEC and grew the resulting strains under inducing and repressing conditions to analyze the secretion profile. As shown in Figure 2, GrlR overexpression repressed the LEE genes and inhibited T3S even under inducing conditions, while overexpression of GrlA overcame the repressing effect of growth in LB (Figure 6A). In turn, as observed in C. rodentium (Deng et al., 2004), GrlR-GrlA co-expression prevented the negative effect of GrlR, even under repressing conditions, but to a lesser extent (Figure 6A).

To further explore this notion, we took advantage of the phenotype of two GrlA mutants, GrlA/I44A and GrlA/R54A, that cannot activate the LEE1 operon because they no longer bind to the ler regulatory region due to a single amino acid change in the HTH domain. Additionally, the GrlA/I44A mutant cannot interact with GrlR (Jimenez et al., 2010; Padavannil et al., 2013). The genes encoding these mutants were cloned into pT3GrlR to co-express them with WT grlR. WT EPEC containing the resulting plasmids pT3GrlRA/I44A or pT3GrlRA/R54A were then grown under inducing and repressing conditions to examine their T3S profile. Under inducing conditions, T3S in the presence of pT3GrlRA/I44A was similar to the secretion profile seen with pT3GrlR, as in both cases protein secretion was significantly reduced (Figure 6A). In contrast, in the presence of GrlR-GrlA/R54A, T3S was unaltered (Figure 6A), suggesting that GrlA/R54A probably prevented GrlR-mediated repression by interacting with it and not because it was over inducing Ler expression. Moreover, in LB medium, where the WT strain does not secrete T3S substrates, we observed a clear de-repression of virulence protein secretion in the presence of GrlR-GrlA/R54A, but not with GrlR-GrlA/I44A, suggesting that GrlA/R54A titrated away both the endogenous GrlR expressed from the chromosomal gene and that expressed from the plasmid (Figure 6A). To confirm the above observations, we performed a similar experiment with the EPEC ΔgrlR strain but grown under repressing conditions (LB medium). Co-expression of GrlR with GrlA/I44A did not prevent GrlR from complementing the repression phenotype and the effect was similar to that observed for the strain with pT3GrlR (Figure 6B); however, this was not the case when GrlR was co-expressed with GrlA/R54A, as protein secretion was still observed (Figure 6B). Western blotting to detect intimin, Tir and EspA in total cell extracts of the same strains, showed similar results (Figure 6C).

These results are consistent with the notion that GrlA, in addition to its activating effect on the ler promoter as a DNA binding protein, could also counteract the alternative GrlR repression effect by interacting with it.

The results described above further indicated that GrlR has a critical role as a repressor under non-permissive growth conditions for LEE expression, thus implying that EPEC expresses it in both LB and DMEM. To analyze GrlR and GrlA expression and their interaction in more detail, we grew the EPEC grlR:3xFLAG, grlA:3xFLAG, ΔgrlA-grlR:3xFLAG and ΔgrlR-grlA:3xFLAG strains, as well as the ΔgrlA-grlR:3xFLAG strain carrying the plasmids encoding the GrlA/I44A and GrlA/R54A mutants (Supplementary Table 1), under inducing and repressing conditions and analyzed total cell extracts by native PAGE and western blotting from samples taken from each culture. As shown in Figure 7, the formation of the GrlR-GrlA complex occurred under both growth conditions (Figure 7A, lanes 1, 2, 7, and 8, upper band). The EPEC grlR:3xFLAG strain also showed the GrlR dimer and monomer (Figure 7A, lanes 1 and 7), while the banding pattern of the ΔgrlA-grlR:3xFLAG strain, where the upper band is not present due to the absence of GrlA (Figure 7A, lines 3 and 9), confirmed the nature of the second and third bands. The EPEC grlA:3xFLAG strain only showed a band corresponding to the GrlR-GrlA complex (Figure 7A, lines 2 and 8), while in the absence of GrlR, GrlA-3xFLAG did not enter the gel (Figure 7A, lines 4 and 10), likely due to its isoelectric point (theoretical pI: 9.71), which changes when in complex with GrlR. Moreover, the expression of the plasmid-encoded GrlA/I44A and GrlA/R54A mutants in the ΔgrlA-grlR:3xFLAG strain further confirmed these observations as the GrlA/I44A mutant, which does not interact with GrlR, did not form the GrlR₂-GrlA complex (Figure 7A, lines 5 and 11) seen with the GrlA/R54A mutant (Figure 7A, lines 6 and 12). Interestingly, the GrlA/R54A mutant drove GrlR to mainly form the complex with GrlA since the dimeric and monomeric forms were not visible. In part, this may be why overexpression of GrlA in the EPEC WT strain overrode the repression in LB (Figure 6A). SDS-PAGE and anti-FLAG western blot with the same cell extracts showed the presence of the tagged proteins (Figure 7B).

Native PAGE analysis also suggested, based on the banding pattern shown by GrlR and GrlA (as discussed above), that GrlR and GrlA are expressed and interact, although not to the same extent, independently of the growth conditions. To analyze this observation in more detail, we took samples from LB and DMEM cultures of the EPEC grlR:3xFLAG and grlA:3xFLAG strains at different growth stages and analyzed them by SDS-PAGE and native PAGE followed by western blot. The qualitative analysis showed that GrlR and GrlA were expressed at moderately higher levels in DMEM than in LB; however, although enriched in DMEM, heterotrimers formed under both growth conditions independently of the expression levels of these proteins (Figures 8A, B, upper panels).

These data allowed us to hypothesize that heterotrimer formation plays a bidirectional role, on the one hand, preventing the activation of ler by reducing the levels of free GrlA and, on the other, antagonizing the independent repressor function of GrlR. A future more quantitative analysis of the relative concentrations of GrlA under both growth conditions, as well as the effect of these conditions on the ability of GrlA to activate the ler promoter and on the kinetics of GrlR-GrlA interaction, will allow a better understanding of the mechanism that regulates the expression of LEE genes in response to environmental cues.

Together, these results further supported the notion that GrlA counteracts the repressor effect of GrlR by interacting with it and showed that the GrlR₂-GrlA heterotrimeric complex, the GrlR-GrlR dimer and the GrlR monomer are present in EPEC under both growth conditions.

To further explore the notion that GrlR can repress the constitutive expression of LEE promoters independently of its interaction with GrlA, we assessed the activity of the LEE1 and LEE2 promoters in a background lacking not only H-NS, but also GrlA and Ler. We then transformed the empty vector pMPM-T3 or plasmids pT3GrlR, pTEPGrlA1 o pT3GrlRA into the Δhns mutant of the LEE-negative lab strain E. coli MC4100 (JPMC1) containing the LEE1-cat or LEE2-cat fusion. The WT strain was the control, where the LEE1-cat fusion, but not the LEE2-cat fusion, was active in the presence of pTEPGrlA1 (Figure 9A), as previously shown (Bustamante et al., 2011). Interestingly, GrlR also repressed the constitutive expression of the LEE1-cat and LEE2-cat fusions in the Δhns background but not when co-expressed with GrlA (Figure 9B). This result supported the proposal that GrlR also negatively regulates LEE gene expression independently of its interaction with GrlA. Also, GrlA had a dual role as a positive regulator by, in addition to its function as a ler activator, preventing GrlR’s negative effect, most likely by trapping it in the heterotrimer. Defining the underlying mechanism by which GrlR represses the constitutive expression of the LEE promoters will require further investigation.