Elements in the LftR Repressor Operator Interface Contributing to Regulation of Aurantimycin Resistance in Listeria monocytogenes

Listeria monocytogenes is an environmental bacterium that lives in the soil but can infect humans upon ingestion, and this can lead to severe invasive disease. Adaptation to these entirely different habitats involves massive reprogramming of transcription.

ABSTRACT

The bacterium Listeria monocytogenes ubiquitously occurs in the environment but can cause severe invasive disease in susceptible individuals when ingested. We recently identified the L. monocytogenes genes lieAB and lftRS, encoding a multidrug resistance ABC transporter and a regulatory module, respectively. These genes jointly mediate resistance against aurantimycin, an antibiotic produced by the soil-dwelling species Streptomyces aurantiacus, and thus contribute to the survival of L. monocytogenes in its natural habitat, the soil. Repression of lieAB and lftRS is exceptionally tight but strongly induced in the presence of aurantimycin. Repression depends on LftR, which belongs to subfamily 2 of the PadR-like transcriptional repressors. To better understand this interesting class of transcriptional repressors, we here deduce the LftR operator sequence from a systematic truncation and mutation analysis of the P_lieAB promoter. The sequence identified is also present in the P_lftRS promoter but not found elsewhere in the chromosome. Mutational analysis of the putative operator in the P_lftRS promoter confirmed its relevance for LftR-dependent repression. The proposed operator sequence was sufficient for DNA binding by LftR in vitro, and a mutation in this sequence affected aurantimycin resistance. Our results provide further insights into the transcriptional adaptation of an important human pathogen to survive the conditions in its natural reservoir.

IMPORTANCE Listeria monocytogenes is an environmental bacterium that lives in the soil but can infect humans upon ingestion, and this can lead to severe invasive disease. Adaptation to these entirely different habitats involves massive reprogramming of transcription. Among the differentially expressed genes is the lieAB operon, which encodes a transporter for the detoxification of aurantimycin, an antimicrobial compound produced by soil-dwelling competitors. While lieAB is important for survival in the environment, its expression is detrimental during infection. We here identify critical elements in the lieAB promoter and its transcriptional regulator LftR that contribute to habitat-specific expression of the lieAB genes. These results further clarify the molecular mechanisms underlying the aurantimycin resistance of L. monocytogenes.

INTRODUCTION

The Gram-positive bacterium Listeria monocytogenes prospers in different ecological niches. It is well adapted for life in the environment, where it is present in the soil and on plant surfaces or decaying plant material, as well as in surface water (1, 2). However, L. monocytogenes is also frequently found in fecal samples of birds, mammals, and humans either asymptomatically carrying the bacterium or suffering from an infectious condition called listeriosis (2). When ingested, the bacterium can reach the gastrointestinal tracts of humans and animals. In susceptible groups, it can even cross the gut-blood barrier (3). This transition into the bloodstream involves a cascade of steps, including (i) invasion of epithelial cells in a phagocytosis-like process, (ii) escape from the phagosome, (ii) multiplication within the cytosol, and (iv) successive spreading to neighboring cells (3–5). Different sets of genes become activated under each condition to ensure the survival of L. monocytogenes during environmental stages or during infection (3). The most important virulence genes are controlled by the master regulator PrfA, which activates their expression inside the host (6, 7) and, together with posttranscriptional and posttranslational mechanisms, ensures that their expression is adjusted to temperature, nutrient availability, and redox conditions (7, 8). Outside a mammalian host, L. monocytogenes experiences different stressful conditions, including UV radiation, heat, cold, desiccation, predation by protists, antibiotic pressure, or nutrient limitation. Most of these stresses are mitigated by the induction of specific stress responses (9).

Recently, we identified in L. monocytogenes the regulatory protein LftR (10), which belongs to the class of PadR-like transcriptional repressors. These proteins repress transcription by binding to operator sites in the promoter regions of their target genes (11–13) and are grouped into two subfamilies according to their molecular size and mechanism of dimerization (12). Repression in subfamily 1 proteins is relieved through allosteric domain rearrangements that occur when the repressor binds its cognate ligand, usually small toxic compounds, such as phenolic acids (12). In subfamily 2 proteins, such as Lactococcus lactis LmrR, compound binding has been shown to affect the dynamic equilibrium of repressor conformations, but how this controls promoter recognition is less clear (14). L. monocytogenes LftR belongs to the latter family. LftR tightly represses the transcription of the lftRS operon and that of the lieAB genes, encoding a multidrug resistance ABC transporter (10). Virtually no LieAB is formed under normal growth conditions, but the lieAB genes are massively overexpressed in cells lacking LftR (10, 15). LftR-dependent repression of the P_lieAB promoter is relieved through aurantimycin A (15), a cyclic hexadepsipeptide produced by the soil dweller Streptomyces aurantiacus (16, 17). Aurantimycin A is exported out of the cell by LieAB, thus mediating aurantimycin A resistance. It has been identified as a specific inducer of the LftR response in a screen using a natural compound library with ∼700 different compounds from myxobacteria, streptomycetes, and fungi (15). Presumably, this resistance mechanism contributes to the survival of L. monocytogenes in the soil. Whether aurantimycin directly binds to LftR is not known. Genetic data indicate that activation of LftR-dependent promoters requires lftS, located in the same operon (15). Possibly, LftS recognizes aurantimycin and transmits this signal to LftR.

Characteristically, operator sequences of PadR-type regulators are palindromes composed of two 4-bp inverted repeats separated by an 8- to 10-bp spacer. For PadR proteins of different species, the ATGT-N_8–10-ACAT consensus motif has been described as the operator sequence (11, 12, 18). Interestingly, such sequence motifs cannot be found in the promoters of LftR-regulated genes, but a putative operator in the P_lftRS promoter has been predicted by in silico sequence analysis (19). The high dynamic range of the P_lieAB promoter, i.e., the difference in promoter activity between its off and on states, makes it an interesting candidate for the development of tunable gene expression systems to be used in bacterial genetics and synthetic biology. In order to better understand the P_lieAB promoter and its regulation, we performed experiments that allowed the identification of critical elements in the LftR repressor operator interface that promote aurantimycin resistance.

RESULTS

The lieAB genes are sufficient to confer aurantimycin resistance.

We previously demonstrated that the lieAB genes are important determinants of the aurantimycin resistance in L. monocytogenes. Our experiments suggested that LieAB exports aurantimycin out of the cell (15). However, it remained unclear whether the lieAB genes were the sole aurantimycin resistance determinants or whether other gene products cooperate with LieAB to detoxify aurantimycin. In order to test whether lieAB alone confers aurantimycin resistance to another bacterial species that is sensitive to this antibiotic, we expressed the L. monocytogenes lieAB genes driven by the xylose-inducible P_xyl promoter in Bacillus subtilis as a heterologous host. The MIC of aurantimycin A was reported to be 0.02 μg/ml for B. subtilis ATCC 6633 (16). In good agreement, we measured an aurantimycin MIC of 0.08 ± 0 μg/ml for B. subtilis strain 168 (Table 1). The lieAB-expressing B. subtilis strain BSAH1 (amyE::P_xyl-lieAB) had a slightly elevated MIC of 0.13 ± 0.05 μg/ml during growth in the absence of xylose, likely due to background lieAB expression from the leaky P_xyl promoter. However, when lieAB expression was induced by xylose, the MIC of aurantimycin increased to 1.28 ± 0 μg/ml (Table 1), a value that is close to the reported MIC value for L. monocytogenes lacking lftR (15). Aurantimycin is the only known LieAB substrate and was identified in a screen with approximately 700 natural compounds (15). However, besides the expression of lieAB induced by aurantimycin, the expression of lieAB can also be induced by rhodamine 6G, which is of synthetic origin, even though LieAB did not confer resistance against this compound to L. monocytogenes (15). In good agreement, the expression of lieAB also did not affect the rhodamine 6G resistance of B. subtilis (Table 1). Taken together, these findings demonstrate that the expression of lieAB is sufficient to confer the aurantimycin resistance level of an intrinsically aurantimycin-resistant (Aur^r) bacterium (L. monocytogenes) to a normally aurantimycin-sensitive species (B. subtilis).

TABLE 1.

Aurantimycin resistance of B. subtilis expressing lieAB

Strain	Genotype	Presence of xylose	MIC (μg/ml) of:
			Rhodamine 6G	Aurantimycin
168	wt	–	2 ± 0	0.08 ± 0
168	wt	+	2 ± 0	0.08 ± 0
BSAH1	amyE::Pxyl-lieAB	–	2 ± 0	0.13 ± 0.05
BSAH1	amyE::Pxyl-lieAB	+	2 ± 0	1.28 ± 0

Spontaneous aurantimycin-resistant suppressors inactivate LftR.

We previously noticed that the growth of L. monocytogenes in the presence of toxic aurantimycin concentrations forced the selection of Aur^r suppressors in wild-type (wt) and ΔlftS mutant cells lacking LftS, the coregulator of LftR (15). For isolation of Aur^r suppressor mutants, the wild type and the ΔlftS mutant LMKK26 were grown overnight in the presence of 750 ng/ml aurantimycin A. This aurantimycin concentration is toxic to L. monocytogenes wild-type and ΔlftS cells but tolerated by ΔlftR cells overexpressing lieAB (15). Aur^r clones from several independent rounds of suppressor selection were isolated, and their lftR genes were sequenced. All tested suppressors had acquired mutations in the open reading frame (ORF), the promoter, or the ribosomal binding site of lftR. While most mutations in the lftR ORF were frameshifts or premature stop codons inactivating LftR, two LftR amino acid exchanges (G27S and T46M) were identified in suppressors isolated from the wild type, and three other amino acid differences were identified in the ΔlftS mutant (T5I, R53C, G70D) (Fig. 1A). Determination of the MIC of aurantimycin A showed that all of the strains were as resistant to aurantimycin A as mutants lacking lftR or lftRS (Fig. 1B). Next, we purified the wild type and the G27S and T46M mutants of LftR-strep (Fig. S1) and analyzed their ability to retard a P_lieAB promoter fragment in an electrophoretic mobility shift assay (EMSA). This shows that LftR binds to the P_lieAB promoter, as three individual LftR-DNA nucleoprotein complexes were detected. In contrast, the G27S and T46M proteins have lost this binding activity (Fig. 1C). The T46M and R53C exchanges caused amino acid exchanges in the DNA-binding helix, whereas the remaining three mutations affected residues in the α0 (T5I) and α2 (G27S) helices as well as the loop between β-sheets β1 and β2 (G70D) (Fig. 1A). Like the T46M and R53C exchanges in the DNA-binding helix, the last three mutations also affected amino acid positions which are located on the DNA-binding interface of the LftR dimer (Fig. 1D). This suggests that the remaining spontaneously occurring Aur^r suppressors also acquire aurantimycin resistance by derepression of LftR-dependent promoters via the inactivation of LftR promoter binding. Remarkably, promoter binding of LftR was insensitive to aurantimycin (Fig. S2). This is congruent with the observation that induction of P_lieAB by aurantimycin requires LftS (15). Possibly, LftS senses aurantimycin.

FIG 1.

Aurantimycin-resistant suppressor mutants inactivate LftR. (A) Alignment of PadR proteins belonging to subfamily 2 and including L. monocytogenes LftR, Bacillus cereus Bce3449, L. monocytogenes LltR, B. cereus BC_4206, and Lactococcus lactis LmrR. Identical amino acids are shaded in black, and similar amino acids are shaded in gray. Residues mutated in aurantimycin-resistant suppressors are shown in red above the alignment and the secondary-structure elements of LftR below the alignment (19). (B) LftR suppressor mutations leading to increased aurantimycin A resistance. MICs of aurantimycin A for L. monocytogenes strains EGD-e (wt), LMSH26 (ΔlftR), LMSH124 (lftR G27S), LMSH132 (lftR T46M), LMKK26 (ΔlftS), LMKK31 (ΔlftRS), LMSH141 (lftR G70D ΔlftS), LMSH142 (lftR R53C ΔlftS), and LMSH145 (lftR T5I ΔlftS). Average values and standard deviations are calculated from three independent experiments. The standard deviation is 0 where no error bars are shown. Significance is indicated by an asterisk (P < 0.05, t test). Please note that MICs were determined already after 10 h of growth to prevent selection of suppressors that would falsify the result at later time points (15). (C) Binding of LftR and LftR variants from Aur^r suppressors to the P_lieAB promoter. Electrophoretic mobility shift assay showing the interaction of wild-type LftR and the G27S and T46M variants with a 165-bp P_lieAB promoter fragment. Molar ratios are indicated. Numbers indicate the positions of the three different LftR-DNA complexes. (D) Structural model of DNA binding by the LftR dimer (PDB accession no. 6ABQ). The two LftR monomers are shown in yellow and blue, and amino acid residues mutated in the lftR suppressors are shown in red.

Identification of the minimal P_lieAB promoter.

The LftR operator sequence is presently unknown. To obtain better insight into LftR-dependent repression, stepwise truncations of a previously studied P_lieAB promoter fragment (15) were constructed and fused to lacZ (Fig. 2A). The resulting promoter-lacZ fusions were inserted into the chromosome of wild-type L. monocytogenes strain EGD-e, and their induction by aurantimycin A was studied in agar diffusion assays in which the reporter strains were poured into X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside)-containing agar plates and aurantimycin was spotted on top. Induction of P_lieAB-lacZ expression by aurantimycin A resulted in a clear blue ring of β-galactosidase activity surrounding the zone of aurantimycin application (Fig. 2B). Aurantimycin prevents growth in the direct vicinity of the application site but induces the promoter-lacZ fusions within its diffusion zone. In contrast, background promoter activity becomes evident outside this zone. The initially characterized promoter fragment (315 bp) contained the 3′ end of the lmo0978 gene located upstream of lieAB, a potential hairpin structure probably representing the transcriptional terminator of lmo0978 as well as the putative −10 and −35 boxes of the P_lieAB promoter. The P_lieAB promoter was still induced by aurantimycin A when the 3′ end of lmo0978 and its terminator were removed. The shortest promoter fragment, which still showed induction by aurantimycin A, covered the 122 bp lying upstream of lieAB (Fig. 2A and B). In contrast, induction was not observed when the fragment was further truncated so that the −35 box was no longer present (Fig. 2B). This was confirmed using densitometric intensity measurements along vertical lines through the center of the aurantimycin spots (Fig. S3). The activity of this set of promoters in the absence and the presence of aurantimycin was also measured by determination of β-galactosidase activity. In this experiment, all promoter variants down to the nucleotide (nt) 122-to-1 version were activated by aurantimycin 19- to 24-fold. In contrast, the nt 97-to-1 truncation was induced only 4-fold by aurantimycin (Fig. 2C). This suggests that nucleotides 1 to 122 upstream of the lieA start codon must contain critical sequence motifs required for control of the P_lieAB promoter through LftR and for maximal promoter activity under derepressed conditions.

FIG 2.

Truncation analysis of the P_lieAB promoter. (A) Scheme illustrating the tested P_lieAB promoter truncations. The arrows show the position of a putative terminator hairpin of the lmo0978 gene, and the black boxes indicate the position of the −35 and −10 boxes in the P_lieAB promoter. (B) Agar plates showing the induction of the P_lieAB promoter and truncated versions thereof by aurantimycin A. Strains LMSH5 [P_lieAB(1-315)-lacZ], LMSH30 [P_lieAB(1-193)-lacZ], LMSH31 [P_lieAB(1-152)-lacZ], LMSH32 [P_lieAB(1-122)-lacZ], and LMSH33 [P_lieAB(1-97)-lacZ] were poured into X-Gal containing BHI agar, and 2-μl droplets of a 2.5-mg/ml aurantimycin A solution were spotted on top of the solidified agar. Images were taken after overnight incubation at 37°C. The same image acquisition and modification parameters were applied to all panels. A densitometric evaluation is shown in Fig. S3. (C) Activities of promoter truncations and their induction by aurantimycin measured by determination of β-galactosidase activity. The same strains as in panel B were grown in BHI broth to mid-logarithmic growth phase in the absence or the presence of 100 ng/ml aurantimycin. The experiment was repeated three times, and average values, standard deviations, and induction ratios are shown. The asterisk marks a statistically significant difference from the wild type (P < 0.01, t test).

Identification of the LftR operator in the P_lieAB promoter.

PadR-type regulators recognize inverted repeats that consist of 4- to 9-nt-long half-sides separated by a 4- to 8-bp-long spacer (11, 18, 20), and inspection of the P_lieAB sequence identified such a motif (TTACTA-N₅-TAGTAA) (Fig. 3A, shown in red). The first repeat unit of this motif (nt 89 to 84 upstream of lieA) even overlapped the −10 box of the P_lieAB promoter (nt 89 to 83) and thus seemed well suited to serve as the LftR operator. We tested this conjecture by altering the second repeat unit (nt 78 to 73), but this had no effect on the inducibility of the P_lieAB-lacZ reporter in strain LMSH38 (data not shown). Other candidate sequences could not be deduced from the sequence; thus, we systematically replaced nucleotides 69 to 61, 60 to 51, 50 to 41, 40 to 31, and 30 to 21 in blocks with their complementary nucleotides (i.e., A↔T and G↔C) (Fig. 3A). With this strategy, most nucleotides between the −10 box and the ribosomal binding site (nt 19 to 9) were included in the analysis. Replacement of the blocks from nt 69 to 61 and 50 to 51 resulted in promoter variants with 12- and 31-fold-stronger background activities, respectively, in the absence of aurantimycin, indicating relief of LftR repression. However, their derepression was not complete, as aurantimycin induced both promoters still 2- to 4-fold (Fig. 3B and C). On the other hand, the nt 69-to-61 (1,182 ± 142 Miller units [MU]) and nt 60-to-51 (1,755 ± 88 MU) promoters were almost as active in the presence of aurantimycin as the fully derepressed wild-type promoter in the ΔlftR mutant (1,938 ± 218 MU). Thus, residual promoter repression in the presence of aurantimycin, as it occurs with the wild-type promoter due to a negative-feedback loop (15), is nearly abrogated in these promoter variants. The negative-feedback loop results from aurantimycin-induced lftRS expression, leading to increased LftR repressor levels that limit P_lieAB induction (15). In contrast, replacements of blocks from nt 50 to 41, 40 to 31, and 30 to 21 resulted in promoter variants with background activities similar to that of the wild-type promoter (Fig. 3C), but inducibility of the promoter with the nt 30-to-21 transversion by aurantimycin was reduced (Fig. 3). This suggests that residues 69 to 51 upstream of the lieAB start codon are important for repression of P_lieAB through LftR, whereas residues in the nt 30-to-21 region are important for promoter activation. However, inverted repeats that may represent LftR binding sites could not be identified in these regions.

FIG 3.

Mutational analysis of the minimal P_lieAB promoter. (A) Positions of 9- to 10-mer mutated sequence stretches with A↔T and G↔C transversions in the P_{lieAB(1–122)} promoter fragment. 9- to 10-mers that lead to promoter derepression are shown in blue, and all others are yellow. The −35 and the −10 boxes of the P_lieAB promoter are indicated by a black background. An inverted repeat overlapping the −10 box is shown in red. See the text for details. (B) Induction of the P_{lieAB(1–122)}-lacZ reporter variants by aurantimycin A. Strains LMSH32 [P_{lieAB(1–122)}-lacZ], LMSH44 [P_{lieAB(1–122) transversion 69–61}-lacZ], LMSH45 [P_{lieAB(1-122) transversion 60–51}-lacZ], LMSH46/[P_{lieAB(1–122) transversion 50–41}-lacZ], LMSH47 [P_{lieAB(1–122) transversion 40–31}-lacZ], and LMSH48 [P_{lieAB(1-122) transversion 30–21}-lacZ] were tested as described for Fig. 2B. The same acquisition and modification parameters were applied to all images. (C) β-Galactosidase activity of the same strains grown in BHI broth (±100 ng/ml aurantimycin) at 37°C. The experiment was carried out three times, and average values and standard variations are shown. Fold changes with and without aurantimycin are indicated. The asterisks mark statistically significant differences from the wild type (P < 0.05, t test). Strain LMSH16 with the promoterless lacZ gene (lacZ) and strain LMSH98 carrying the P_{lieAB(1–122)}-lacZ reporter but lacking lftR (ΔlftR) were included as controls.

Next, we changed two bases at a time between residues 69 and 51 upstream of the lieA start codon. Exchange of bases 69 to 68, 65 to 64, 61 to 60, and 57 to 56 into their complementary bases (Fig. 4A) caused relief of P_lieAB repression (Fig. 4B), and this effect was confirmed by quantification of β-galactosidase activity in the absence of aurantimycin (Fig. 4C). The promoter with an exchange of the GT dinucleotide at positions 69 and 68 (strain LMSH52) showed the strongest degree of derepression. β-Galactosidase activity was 314 ± 28 MU in this strain, indicating a 38.7-fold increase of promoter activity in the absence of aurantimycin compared to that of the wild-type P_lieAB-lacZ fusion present in strain LMSH32 (8.1 ± 2 MU). However, this activity level did not reach the level of the wild-type P_lieAB-lacZ fusion when fully derepressed in the ΔlftR background (2,594 ± 249 MU). To decide whether the nt 69-to-68 exchange lowered the intrinsic activity of the P_lieAB promoter or whether the promoter with the nt 69-to-68 exchange is still LftR repressed, we tested its activity in the ΔlftR background (strain LMSH101). The activity of the P_lieAB promoter with the nt 69-to-68 exchange increased to 1,937 ± 82 MU in this strain, indicating a residual LftR-dependent repression of this mutant promoter (Fig. 4B). Taken together, the main LftR operator must be located between residues 69 and 56.

FIG 4.

Identification of the LftR operator sequence in P_lieAB. (A) Sections of the P_{lieAB(1–122)} promoter fragment and positions of twin nucleotide exchanges tested for their effect on LftR repression (blue and yellow highlighting, as described in the legend for panel C. Numbers correspond to nucleotides upstream of the lieA start codon. (B) Induction of the P_{lieAB(1–122)}-lacZ reporter variants carrying twin nucleotide exchanges by aurantimycin A (aur.) in an agar diffusion assay. Strains LMSH32 [P_{lieAB(1–122)}-lacZ, wt], LMSH52 (nt 69 and 68), LMSH53 (nt 67 and 66), LMSH54 (nt 65 and 64), LMSH55 (nt 63 and 62), LMSH56 (nt 61 and 60), LMSH57 (nt 59 and 58), LMSH58 (nt 57 and 56), LMSH59 (nt 55 and 54), LMSH60 (nt 53 and 52), and LMSH61 (nt 51) were tested as described in the legend to Fig. 2B for induction by aurantimycin A. The same acquisition and modification parameters were applied to all images. Strains LMSH16 (lacZ, wt) and LMSH98 (P_lieAB-lacZ ΔlftR) were used as controls. (C) Quantification of β-galactosidase activity in the same set of strains. Average values and standard deviations are calculated from three independent experiments. Blue, P_lieAB variants with >10-fold derepression of background promoter activity; yellow, those with <10-fold derepression; gray, controls. Asterisks mark statistically significant differences from the activity of the wild-type P_lieAB promoter in the wild-type background (P < 0.01, t test).

Confirmation of the LftR operator in the P_lftRS promoter.

Interestingly, a 14-bp sequence similar to P_lieAB nucleotides 69 to 56 was found within the P_lftRS promoter region (P_lftRS nucleotides 51 to 38) (Fig. 5A). Unlike the LftR operator motif in P_lieAB, the motif present in P_lftRS contains a perfect inverted repeat (GTAT-N₆-ATAC) and simultaneously overlaps the putative −10 box of P_lftRS. In good agreement with the hypothesis that this sequence serves as the LftR operator, the consensus operator sequence (GTAWTAC-N₃-ATAC) is present only twice in the L. monocytogenes EGD-e genome, i.e., upstream of lieAB and in front of the lftRS operon. This is in good agreement with our previous transcriptome sequencing (RNA-Seq) analysis, which showed that lieAB and lftRS are the only targets of LftR (15).

FIG 5.

LftR operator in the P_lftRS promoter. (A) LftR operator sequences in the P_lieAB and P_lftRS promoters (dark-blue background). The single-nucleotide difference in the LftR boxes is highlighted by a light-blue background. The −10 box of P_lftRS is marked with a dashed underline, the inverted repeat in the LftR operator of P_lftRS is indicated with arrows, and the putative secondary LftR operator in P_lieAB is marked by a red underline. (B) Mutant variants of the P_lftRS promoter fragment constructed in this study. Numbers correspond to nucleotides upstream of the lftR start codon. The putative LftR box is colored as in panel A. Mutated nucleotides are shown in red. (C) Induction of the P_lftRS-lacZ reporter variants by aurantimycin A. Strains LMSH7 (P_lftRS-lacZ, wt), LMSH69 (nt 51 to 48), LMSH70 (nt 41 to 38), and LMSH71 (nt 47 to 45) were tested as described in the legend to Fig. 2B. The same acquisition and modification parameters were applied to all images. Strains LMSH16 (lacZ, wt), LMSH100 (P_lftRS-lacZ ΔlftR), and LMSH102 (P_lftRS-41-38-lacZ ΔlftR) were used as further controls. (D) Quantification of β-galactosidase activity in the same set of strains. Average values and standard deviations are calculated from three experiments. Blue, P_lftRS variants with >10-fold derepression of background promoter activity; yellow, those with <10-fold derepression; gray, controls. Asterisks mark statistically significant differences from the activity of the wild-type P_lftRS promoter in the wild-type background (P < 0.01, t test).

In order to verify the putative LftR operator sequence with a second LftR-dependent promoter, we replaced the two halves of the inverted repeat in P_lftRS en bloc with their complementary nucleotides (Fig. 5B) and tested induction of the resulting promoters by aurantimycin A (Fig. 5C). Replacement of the first half (51-GTAT-48) inactivated the P_lftRS promoter completely, probably because this half overlaps the −10 region. In contrast, replacement of the second half (41-ATAC-38) led to promoter derepression (Fig. 5C). However, as for the lieAB promoter, derepression was only partial, as shown by quantification of β-galactosidase activities. Here, the P_lftRS promoter with the exchange of the second inverted repeat generated 732 ± 105 MU in the wild-type background of strain LMSH70 (20.1-fold derepression) and 3,491 ± 58 MU in the ΔlftR background of strain LMSH102 (Fig. 5D). The TAC trinucleotide downstream of the first repeat is also conserved in the LftR operator sequences of both promoters (Fig. 5A) and contributes to the repression of P_lieAB. As it overlaps the −10 box of the lftRS promoter, its contribution to LftR-dependent repression cannot be studied without completely inactivating P_lftRS (Fig. 5B to D). These data agree with the idea that GTAWTAC-N₃-ATAC acts as the LftR operator sequence.

To confirm that this sequence is sufficient for LftR binding, the LftR operator was transplanted into the P_divIVA promoter (Fig. 6A), which is not subject to LftR-mediated repression (15). The interaction of LftR with the native P_divIVA promoter and the mutant version containing the LftR binding site was then studied by electrophoretic mobility shift assay. The native P_divIVA promoter did not interact with LftR (Fig. 6B). However, when the LftR operator is inserted into the P_divIVA promoter, a specific interaction can be observed (Fig. 6B). This proves that the LftR operator motif identified here is sufficient for LftR binding.

FIG 6.

The LftR operator is sufficient for LftR binding. (A) DNA sequences of the P_divIVA promoter fragment and the artificial P_divIVA-Op_LftR promoter fragment containing the LftR operator, which were tested for LftR binding. The LftR operator motif is highlighted by a blue background. (B) EMSA showing the interaction of LftR with the P_divIVA and P_divIVA-Op_LftR promoters.

Interaction of LftR with its operator.

To show a direct interaction of LftR with its operator, we tested the binding of LftR to P_lieAB-69-68, in which the first nucleotides of the operator motif (nt 69 and 68; GT) were mutated and which represents the P_lieAB promoter mutation with the highest degree of derepression in vivo (Fig. 4C). While formation of LftR promoter complex no. 3 (likely representing an unspecific nucleoprotein complex formed in the presence of large LftR amounts) was unaffected by this mutation, the formation of LftR nucleoprotein complex no. 2 was strongly reduced and the formation of complex no. 1 strongly increased (Fig. 7A). This shows that a mutation in the identified LftR operator alters the binding of LftR to the P_lieAB promoter in vitro and thus confirms the relevance of this motif for aurantimycin A-dependent control of lieAB expression. However, it also indicates that LftR must have one or more secondary so-far-unidentified binding site in the promoter of the lieAB genes.

FIG 7.

Effect of LftR operator mutations on repressor binding and aurantimycin resistance. (A) EMSA showing the interaction of LftR with different promoter fragments. Different concentrations of LftR were mixed with the P_lieAB promoter and a mutant variant thereof, in which the first nucleobases, GT, of the LftR operator were replaced by CA (P_lieAB-69-68) and analyzed. A fragment encompassing the P_gpsB promoter (15) was used to control the binding specificity of LftR. The different LftR promoter complexes are indicated on the right. (B) Effect of the P_lieAB-69-68 mutation on aurantimycin resistance. Aurantimycin MICs for strains EGD-e (wt), LMTE1 (P_lieAB-69-68), and LMSH26 (ΔlftR). The experiment was repeated three times, and average values and standard deviations are shown. Asterisks mark statistically significant differences from the activity of the wild type (P < 0.01, t test).

Next, we tried to replace the native P_lieAB promoter by the P_lieAB-69-68 variant in the EGD-e chromosome but failed despite multiple attempts. However, this replacement was possible in the ΔlftR mutant, probably because the promoter region is not LftR occupied in this strain and accessible to the recombination machinery. The lftR gene was then reintroduced at its native site in the chromosome of this strain by allelic replacement (yielding strain LMTE1, P_lieAB-69-68) to measure the effect of the P_lieAB-69-68 promoter mutation on aurantimycin resistance. The MIC of aurantimycin for this strain was 1.25 ± 0 μg/ml (n = 3), which is close to that of the ΔlftR mutant (2.5 ± 0 μg/ml) (Fig. 7B). This shows that the transversion of the nt 69-68 dinucleotide in the P_lieAB promoter relieves repression of lieAB expression through LftR and confirms that LftR cannot repress the P_lieAB promoter effectively when the LftR operator site is mutated at these two critical positions.

DISCUSSION

We here have determined the conserved sequence motif GTAWTAC-N₃-ATAC as the operator sequence of the PadR-type transcriptional repressor LftR. This motif is exclusively found in the two LftR-regulated promoters of L. monocytogenes, P_lieAB and P_lftRS (15), and is composed of an imperfect 4-bp inverted repeat sequence, which is separated by a spacer of 6 nucleotides. Imperfect inverted repeats are also found in the operator sites of other genes controlled by PadR-type repressors, such as the lmrR gene of L. lactis (18). Mutation of the two repeats in the LftR operator either en bloc or by introduction of dinucleotide exchanges causes relief of repression in reporter gene assays with both LftR-dependent promoters. The first three nucleotides of the spacer region (TAC) are also important for LftR-dependent repression, since this sequence is conserved in both promoters and mutation of this motif in the P_lieAB promoter relieved its repression. Moreover, the GTAWTAC-N₃-ATAC motif (including the TAC trinucleotide) is unique to the P_lieAB and P_lftRS promoters, whereas 192 copies of the GTAW-N₆-ATAC motif (i.e., the same motif without the TAC trinucleotide) are present in the L. monocytogenes EGD-e genome (21). Thus, the extension of the first repeat unit by the TAC motif ensures the required sequence specificity of the LftR operator that is needed for specific repression of two promoters only. Our previous RNA-Seq analysis of a ΔlftR mutant has shown that the transcriptional start point of the lieAB operon is located at position 63 ± 1 bp upstream of the lieA start codon and 19 to 22 bp downstream of the −10 box (15). The relevance of the unusually long distance between the −10 box and the experimentally defined transcriptional start site is currently not clear, but binding to the site of transcription initiation would explain how LftR blocks lieAB transcription.

The P_lftRS promoter had a stronger background activity than the P_lieAB promoter in previously published experiments (15). This pattern of activity likely ensures the production of sufficient background LftR levels for full repression of P_lieAB even under noninducing conditions. However, it contradicts the observation that the more strongly repressed P_lieAB promoter contains only an imperfect inverted repeat in its LftR operator, whereas the ideal inverted repeat is present in the LftR operator of P_lftRS. A solution to this contradiction might come from the observation that a less conserved secondary LftR binding site is present in the P_lieAB promoter. This secondary motif (GTAATAA-N₃-ATAC) (underlined in Fig. 5A) overlaps the consensus LftR operator and might help to further tighten repression. However, when we added mutations in this potential secondary operator site in the P_lieAB promoter with the 69-68 exchange, which is currently the P_lieAB promoter variant with the strongest degree of derepression, no further increase of derepression could be observed (data not shown). Possibly, mutations in the secondary operator site affect the background promoter activity only under normally repressed conditions and do not affect the degree of derepression.

Another remaining question concerns the fact that we do not observe the complete derepression of the P_lieAB and P_lftRS promoters when they are mutated in their consensus LftR operator motifs. The mutant variants of both promoters with the highest degree of derepression are still repressed 5- to 6-fold by LftR, indicating residual LftR binding. Possibly, binding of LftR to lesser-conserved secondary binding sites is still possible in these promoter mutants. In fact, L. lactis LmrR protects two sites in the promoter of its target gene lmrCD, and only one of them contains the inverted repeat motif (22). Likewise, LmrR binds to multiple sites in the promoter of its own gene, containing only one operator sequence (18). It seems that LmrR also binds to DNA sequences not containing the typical operator. Agustiandari et al. proposed that the operator may act as a strong nucleation site for binding of repressor molecules and facilitates the formation of bigger nucleoprotein complexes that also cover adjacent sequence stretches (18). In good agreement, the most recent LmrR-dependent transcriptional regulation model assumes that LmrR exists in equilibrium between nonspecific DNA absorption and specific operator interaction. Effector binding to LmrR is believed to shift this equilibrium toward the nonspecifically DNA-bound state, leading to relief of repression (14). Like LmrR, LftR belongs to subfamily 2 of the PadR-type repressors (12) and might thus regulate its target promoters in a similar way. Such binding of LftR to secondary sites in the proximity of the main operator would explain residual LftR-dependent regulation in our operator mutants and also is in good agreement with the residual binding of LftR to promoter variants mutated in the LftR operator in vitro. More detailed structural and biochemical experiments would be useful to address these unanswered questions in the future.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Strains are listed in Table S1 in the supplemental material. L. monocytogenes was routinely cultivated in brain heart infusion (BHI) broth or on BHI agar plates at 37°C. B. subtilis strains were grown in LB or on LB agar plates. Erythromycin (5 μg ml⁻¹), kanamycin (50 μg ml⁻¹), spectinomycin (50 μg ml⁻¹), and X-Gal (100 μg ml⁻¹) were added when required. Escherichia coli TOP10 was used as the standard cloning host (23).

General methods, manipulation of DNA, and oligonucleotide primers.

The transformation of E. coli and isolation of plasmid DNA were performed using standard methods (23). The transformation of B. subtilis was performed as described elsewhere (24). The preparation of electro-competent L. monocytogenes cells and transformation of L. monocytogenes were carried out as described previously (25). Restriction and ligation of DNA was performed by following the manufacturer´s instructions. All primer sequences are listed in Table S2.

Construction of bacterial plasmids and strains.

Plasmid pSAH43 was constructed for overproduction of LftR. To this end, the lftR gene was amplified using the primer pair SAH208/SAH209 and introduced by restriction-free cloning into pSH304. Plasmid pSH304 had been constructed in several steps. First, the gfp gene of plasmid pSG1164 was amplified using the primer pair SV142/SV143 (the reverse primer fuses a C-terminal Strep-tag to gfp), and cloned into pET11a using NheI/BamHI, yielding plasmid pSH105. Next, divIVA-gfp from plasmid pDG7 was amplified with the oligonucleotides SV128/SV143 and cloned into pSH105 using NdeI. This results in plasmid pSH145, which carries divIVA-gfp-strep. The gfp part of this fusion was then removed by QuikChange mutagenesis using primers SHW318/SHW319, and this yielded plasmid pSH304, expressing divIVA-strep. The G27S and T46M mutations were introduced into pSAH43 by QuikChange mutagenesis using the primer pairs SHW848/SHW849 and SHW850/SHW851, respectively.

For construction of promoter-lacZ fusions carrying truncated versions of P_lieAB, fragments were amplified using the primer pairs SAH170/SAH99 (nt 1 to 193), SAH171/SAH99 (nt 1 to 152), SAH172/SAH99 (nt 1 to 122), or SAH173/SAH99 (nt 1 to 97) and cloned into pBP117 using BamHI/EcoRI, yielding plasmids pSAH26, pSAH27, pSAH28, and pSAH29, respectively.

Plasmid pSAH31 was constructed by amplification of the P_lieAB promoter using the primers SAH187/SAH99 and cloning of the resulting fragment into pBP117 using BamHI/EcoRI.

Plasmids pSAH38 to pSAH42 were constructed to fuse P_{lieAB(1–122)} promoter variants carrying 9-mer or 10-mer stretches with continuous A↔T and G↔C transversions to lacZ. To this end, regions immediately up- and downstream of the stretch to be mutated were amplified from plasmid pSAH28 with SHW306/SAH188 and SAH195/SAH91 (pSAH38), SHW306/SAH189 and SAH196/SAH91 (pSAH39), SHW306/SAH190 and SAH197/SAH91 (pSAH40), and SHW306/SAH191 and SAH198/SAH91 (pSAH41), as well as SHW306/SAH192 and SAH199/SAH91 (pSAH42). The two up- and downstream fragments were fused together in a splicing by overlap extension PCR (SOE-PCR), and an internal fragment was amplified with SAH99/SAH172 and cloned into pBP117 using BamHI/EcoRI.

Likewise, plasmids pSAH47 to pSAH56 were constructed to fuse P_{lieAB(1–122)} promoter variants containing 2-mer stretches with A↔T and G↔C transversions to lacZ. Regions immediately up- and downstream of the stretch to be mutated were amplified from plasmid pSAH28 with primers SHW306/SAH227 and SAH228/SAH91 (pSAH47), SHW306/SAH229 and SAH230/SAH91 (pSAH48), SHW306/SAH231 and SAH232/SAH91 (pSAH49), SHW306/SAH233 and SAH234/SAH91 (pSAH50), SHW306/SAH235 and SAH236/SAH91 (pSAH51), SHW306/SAH237 and SAH238/SAH91 (pSAH52), SHW306/SAH239 and SAH240/SAH91 (pSAH53), SHW306/SAH241 and SAH242/SAH91 (pSAH54), SHW306/SAH243 and SAH244/SAH91 (pSAH55), and SHW306/SAH245 and SAH246/SAH91 (pSAH56). The two up- and downstream fragments were fused together in a SOE-PCR, and the desired fragment was amplified with SHW306/SAH91 and cloned into pBP117 using BamHI/EcoRI.

Plasmids pSAH63 to -65 were generated to confirm LftR binding in the lftRS operator. A fragment upstream of the lftRS operator was amplified using primers SAH79 and SAH265. Primer SAH266 (pSAH63), SAH267 (pSAH64), or SAH268 (pSAH65) carrying the respective mutation were used together with primer SAH108 to amplify the region downstream of the mutation to be introduced. Both fragments were extended on each other by PCR using the outer primers SAH79 and SAH108. The fragment generated in the latter reaction was then inserted into pBP117 by SOE-PCR.

Plasmid pSAH89 was constructed to insert the LftR operator sequence GTAATACAGAATAC into the P_divIVA promoter. To this end, a QuikChange mutagenesis was performed using the oligonucleotides SAH317/SAH318 and plasmid pSAH19 as the template.

Plasmid derivatives of pBP117 were transformed into L. monocytogenes by electroporation, and kanamycin-resistant clones were selected. Plasmid integration was verified by PCR.

Plasmid pSAH90 was constructed to introduce the two sequence transversions at positions 69 and 68 upstream of the lieAB start codon into the chromosome. To this end, a synthetic double-stranded DNA (dsDNA) fragment that covers the P_lieAB promoter region from 314 bp upstream of the lieAB start codon onwards to the 76th nucleotide of the lieA gene and already included the desired nucleotide exchanges was commercially produced (Integrated DNA Technologies, Belgium). This fragment was amplified in a PCR using the primers SAH332/SAH333 and inserted into pMAD using EcoRI/NcoI. For reintroduction of lftR into the ΔlftR mutant, plasmid pTE3 was constructed by removing the transposon (Tn) insertion from lftS that is present on plasmid pKK64 in a PCR with SHW893/SHW894 as the primers. pMAD plasmids were used to modify chromosomal loci using an integration/excision protocol described elsewhere (26). Allelic replacement was confirmed by PCR and DNA sequencing.

For the construction of plasmid pSAH91, lieAB was amplified from EGD-e chromosomal DNA with the primers SAH334/SAH335. The obtained PCR product was inserted into pSG1154 by SOE-PCR. B. subtilis transformants were selected on LB agar plates containing spectinomycin (50 μg/ml). Integration of the plasmid into the amyE locus was confirmed by the absence of amylase activity on starch plates and verified by PCR.

β-Galactosidase reporter assays.

Aurantimycin A was tested for induction of promoter-lacZ reporters in diffusion assays. To this end, a previously published protocol was used (15). For quantitative β-galactosidase measurements, strains were cultivated until an optical density at 600 nm (OD₆₀₀) of 0.5 to 0.6 was reached and collected by centrifugation. The cell pellet was washed with 600 μl H₂O and then resuspended in 1.2 ml Z buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 20 mM 2-mercaptoethanol). Cells were sonicated, and debris was removed by centrifugation (12,000 × g, 2 min). The protein content of the resulting supernatant was determined using Roti-Nanoquant (Roth, Germany). For β-galactosidase activity measurements, the samples were appropriately diluted in Z buffer to a final volume of 1,000 μl and incubated at 30°C for 10 min. The reaction was started by adding 200 μl ONPG (o-nitrophenyl-β-d-galactopyranoside) in 4 mg/ml Z buffer and stopped by adding 500 μl 1 M Na₂CO₃ as soon as the first sample turned clear yellow. Absorption was measured at 420 nm against Z buffer incubated with ONPG as the blank, and Miller units (MU) were calculated.

Determination of MICs.

B. subtilis strains were grown on LB agar or in LB broth with the addition of 1% xylose when required. Initially, cells from LB plates were grown overnight at 37°C with shaking. The overnight culture was diluted in fresh medium to an OD₆₀₀ of 0.1 before transference of 100 μl of the diluted culture to a 96-well plate containing 100 μl medium with a geometric dilution series of twice the final aurantimycin A or rhodamine 6G concentration that was to be tested. The plate was incubated at 37°C with shaking for 19 to 24 h. MIC determination for L. monocytogenes was similarly performed in 96-well plates in a total volume of 200 μl. Overnight cultures were used to inoculate 200-μl BHI aliquots with various concentrations of aurantimycin A (geometric dilution series) at an OD₆₀₀ of 0.05. The microtiter plates were incubated at 37°C and examined for growth after 10 h (due to suppressor formation at later time points) (15). The MIC was defined as the lowest concentration of antibiotic at which no growth could be observed.

Purification of LftR proteins.

LftR-strep proteins were overexpressed in E. coli BL21, which was cultivated in LB broth containing ampicillin (100 μg/ml) at 37°C. Gene expression was induced when the culture had reached an OD₆₀₀ of 0.5 by isopropyl-β-d-thiogalactopyranoside (IPTG) addition (final concentration, 1 mM). After three more hours of continued cultivation, the cells were harvested, washed once with ZAP buffer (10 mM Tris-HCl, pH 7.5, 200 mM NaCl), and disrupted in ZAP buffer containing 100 mM phenylmethyl sulfoxide (PMSF) using an Emulsiflex homogenizer (Avestin, Germany). Cell debris was removed by centrifugation (6,000 × g, 5 min, 4°C), and the resulting supernatant was ultracentrifuged (100,000 × g, 30 min at 4°C) or filtered through a Minisart filter with a pore size of 0.45 μm (Sartorius). LftR-strep proteins were purified using affinity chromatography and Strep-Tactin Sepharose (IBA Lifesciences, Germany) according to the manufacturer’s instructions. Fractions containing purified proteins were pooled, aliquoted, and stored at –20°C. Aliquots of these samples were separated by standard SDS-polyacrylamide gel electrophoresis.

Electrophoretic mobility shift assay.

To show the interaction of LftR with specific DNA sequences, 165-bp-long DNA fragments were amplified using PCR. The wild-type lieAB promoter was amplified by PCR using primers SAH99 and SAH172 from EGD-e genomic DNA, modified P_lieAB-69-68 with the G69C and T68A exchanges was amplified using the same primers but pSAH47 as the template, and P_gpsB was amplified using primers SAH324 and SAH325 and EGD-e as the template.

The P_divIVA promoter was amplified from genomic DNA using primers SAH326 and SAH327. The P_divIVA promoter containing the LftR operator was amplified using the same primers but plasmid pSAH89 as the template.

One hundred to 150 ng of the purified DNA was added to EMSA buffer (end concentrations, 120 mM HEPES, 300 mM KCl, 30 mM MgCl₂, 0.3 mg/ml bovine serum albumin [BSA], 30% glycerol, 0.3 mM EDTA, pH 8), mixed with different amounts of purified LftR, and incubated for 15 min at room temperature before being loaded on a 12% acrylamide gel (0.6× Tris-borate-EDTA [TBE], 12% acrylamide/bisacrylamide). The gel was run at 120 V for 90 min, and the DNA was visualized after being stained with ethidium bromide.