Dear Editor,

Bacteria are able to communicate with each other through small, diffusible molecules. In some species, this type of communication has been linked to virulence. Therefore, bacterial cell-cell communication pathways have recently been put forward as promising targets for new anti-microbial therapeutics 1. Several different communication systems have been identified in bacteria, both for intraspecies and for interspecies communication. The AI-2 signal is hypothesized to be involved in interspecies bacterial communication, since it is produced by an enzyme, LuxS, which is widely conserved among Gram-negatives and Gram-positives 2. Salmonella enterica serovar Typhimurium (S. Typhimurium), an important food pathogen that causes diseases ranging from self-limiting gastroenteritis to life-threatening systemic infections, also produces AI-2 through LuxS. In S. Typhimurium, however, the role of AI-2 as a signaling molecule is still controversial, including the possible role in its pathogenicity. Most studies unraveling the possible role of AI-2 have focused their efforts on studying the effects of a luxS mutation 3. However, LuxS also functions in the pathway for metabolism of S-adenosylmethionine (SAM), the major cellular methyl donor 4. As a result, the important metabolic role of LuxS makes interpretation of data reflecting either cell-cell signaling or metabolic effects very difficult 5. In this study, we therefore aimed at identifying the direct transcriptional targets of LsrR, the sole transcriptional regulator that has been associated with AI-2 in S. Typhimurium 6. A better knowledge of the Salmonella LsrR regulon may further clarify the role of AI-2. By using a combination of high-throughput technologies, we show that the LsrR regulator has a limited regulon in S. Typhimurium.

Till now, the only AI-2-dependent genes known in S. Typhimurium, as identified in a genetic screen for luxS-regulated genes 6, are in the lsr operon, which is repressed by LsrR when no AI-2 is present in the cell. The lsr operon consists of various genes for AI-2 transport and processing. lsrACD encode an ATP binding cassette (ABC) transporter, which is highly similar to the ribose transport system of E. coli and S. Typhimurium, that internalizes extracellular AI-2 6. LsrB is a periplasmic binding protein component that directly binds AI-2 and is required for AI-2 transport into the cell 7. LsrF and LsrG are involved in further processing of the internalized AI-2 8, 9. No function for the lsrE gene, which is homologous to genes encoding sugar epimerases, has yet been identified. Encoded in the opposite direction of the lsr operon and also shown to be repressed by LsrR in E. coli 10 are lsrR and lsrK. LsrK is a kinase that phosphorylates AI-2 upon its entry into the cell 8. Phospho-AI-2 then directly binds the repressor LsrR, thus relieving the repression of the lsr operon by LsrR 8, 10. In S. Typhimurium, AI-2 production is stimulated in conditions of low pH and high osmolarity, conditions that are prevalent in the intestine 11. Therefore, AI-2 has been hypothesized to play a role in host-associated phenotypes and even in the pathogenicity of S. Typhimurium, yet contradicting observations are reported 12.

Recently, the LsrR binding site in the promoter region of both lsrR and lsrA has been determined in vitro in E. coli 10. Additionally, the effect of an lsrR deletion on genome-wide expression has been studied in E. coli 13. However, no distinction between the direct targets of LsrR (constituting the LsrR regulon) and indirect targets – affected by an lsrR mutation modulating the rate at which AI-2 is imported and processed – has been made. Based on these observations, LsrR – being a specific AI-2-related regulator with a broad impact – has been put forward as a promising gateway towards further understanding AI-2-based communication and, possibly, towards establishing a foundation for antimicrobial therapies targeting LsrR 1. In this study, we therefore aimed at mapping the full LsrR regulon of S. Typhimurium, in order to better understand its role by identifying its direct target genes. To this end, we combined chromatin immunoprecipitation (ChIP) coupled with S. Typhimurium whole-genome tiling arrays (ChIP-chip) and transcriptomics comparing gene expression in an lsrR deletion mutant and wild-type S. Typhimurium.

ChIP-chip was performed as previously described (14 and Supplementary information, Data S1) on S. Typhimurium whole genome tiling arrays after 4 h of growth (1.5E9 CFU/ml) under aerobic conditions. We chose a time point at which the extracellular AI-2 level is higher for the wild-type than for the lsrR mutant (data not shown), indicating that at this time point LsrR is still repressing the lsr operon and, possibly, its other yet unidentified target genes. For LsrR, the intergenic region between lsrR and the lsr operon was the only in vivo bound target that could be identified (Figure 1A and 1B). Although this direct control has already been shown in vitro for E. coli 10, here we show that LsrR also binds the lsr operon promoter in vivo in S. Typhimurium. More importantly, however, no other targets bound by LsrR in vivo could be detected. To ensure that we did not fail to identify other LsrR target genes, bound by the regulator at another time point, ChIP-chip was repeated after 10 h of growth (5E9 CFU/ml) but no additional peaks could be observed (data not shown). ChIP-chip data were validated by performing ChIP and assessing the enrichment of the lsr promoter by quantitative PCR (Figure 1C).

Figure 1
figure 1

Genome-wide (A) and lsr region view (B) of LsrR ChIP-chip data. Each bar represents the log2 enrichment ratio, or difference between normalized ChIP and mock ChIP log2 ratios, for a particular probe. The peak indicating binding of LsrR in the lsr operon upstream intergenic region is indicated in (A) and a detailed view of the lsr region is given in (B). In the latter, the lower strip represents the annotated genome; genes encoded on the direct strand are represented by bars above the horizontal line, genes encoded on the complementary strand below. The negative peaks in (A) are an artefact, caused by the absence of Fels-1 and Fels-2 in S. Typhimurium SL1344 (used for the ChIP-chip analysis) as compared to S. Typhimurium LT2 (genome sequence used for the tiling array design, Supplementary information, Data S1). The validation of the ChIP-chip data by ChIP-qPCR is shown in (C); dnaG is a negative control (Supplementary information, Data S1). Values shown are log2 enrichment ratios of ChIP over mock ChIP samples and averages of triplicate qPCRs; standard deviations are indicated. Values are representative for three biological repeats. (D) Significantly differentially expressed genes in an lsrR mutant versus wild-type S. Typhimurium. aSTM numbers, gene names and gene functions are taken from NCBI Refseq NC_003197 16 and adapted according to recent literature. Results are sorted according to STM number. bDifferentially expressed genes were detected by t-test with multiple testing correction; see Supplementary information, Data S1.cDifferential expression of STM4071, lsrA and pyrI was validated by qRT-PCR (Supplementary information, Data S1). qRT-PCR log2 ratios of expression in the lsrR mutant versus the wild-type S. Typhimurium were 3.74 (± 0.59) for STM4071, 4.84 (± 0.23) for lsrA and −0.70 (± 0.13) for pyrI.

Full size image

To complement the ChIP-chip experiment – ChIP-chip data give information on the location of a regulator, not on its functioning – we performed a transcriptomics study comparing mRNA levels in the wild-type and the lsrR mutant strain CMPG5608 at the same time point (4 h) as for the ChIP-chip experiment (Supplementary information, Data S1). As expected, genes of the lsr operon were significantly upregulated in the lsrR mutant versus the wild-type, confirming the repressor function of LsrR (Figure 1D). Divergently transcribed from the lsr operon, lsrK and further downstream-located STM4071 (Figure 1B) were also upregulated in the lsrR mutant (Figure 1D). STM4071 has not been associated with the lsr genes before. STM4071 encodes a protein with similarity to polyketide synthesis domain proteins and was previously annotated as a putative mannose-6-phosphate isomerase; STM4071 has no homologue in E. coli. Significantly downregulated in the lsrR mutant versus the wild-type was pyrI, encoding the allosteric regulatory subunit of the aspartate carbamoyltransferase, involved in pyrimidine biosynthesis. Just below the analysis threshold were the differential expression values of pyrB, encoding the catalytic subunit of the aspartate carbamoyltransferase, and carAB, encoding the glutamine-hydrolyzing small and large subunits of the carbamoyl phosphate synthase; carbamoylphosphate is a precursor in pyrimidine biosynthesis. Differential expression of these genes may be an indirect effect of the lsrR mutation. Pyrimidine biosynthesis has been associated with AI-2 before 15. Through deleting lsrR, uptake and processing of AI-2 molecules is increased when compared to the wild-type. Possibly, the altered balance in AI-2-derived metabolites affects other (metabolic) pathways in S. Typhimurium, although suggesting a direct link between LsrR/AI-2 and pyrimidine biosynthesis is not evident. As a validation of the microarray results, we confirmed STM4071, lsrA and pyrI differential expression by quantitative real-time PCR (qRT-PCR; Figure 1D and Supplementary information, Data S1). These differential expression results agree with a transcriptomics study comparing gene expression in an lsrR mutant and that in wild-type E. coli 13: the lsr genes and a number of nucleotide biosynthesis genes were differentially expressed in both the E. coli and S. Typhimurium lsrR mutants versus the wild-type. The study in E. coli identified a much higher number of differentially expressed genes, but no distinction was made between the direct (LsrR-bound) and indirect target genes of LsrR 13. Furthermore, the study in E. coli sampled RNA at a different time point (early stationary phase at OD600 2.4 for E. coli, versus late exponential phase at OD600 1.3 for this study). However, our S. Typhimurium ChIP-chip experiment after 10 h of growth (stationary phase, OD600 2.8) did not show additional direct LsrR targets. Whereas the experimental conditions used in this study have been shown to be representative for studying AI-2 in S. Typhimurium 6, it cannot be excluded that LsrR might control additional genes under other experimental conditions. It is also plausible that AI-2 and LsrR can have different functions in E. coli versus S. Typhimurium. The lsr operons are different in E. coli and S. Typhimurium, for instance. As mentioned above, the STM4071 gene encoded downstream of lsrK and transcribed in the same direction has no homologue in E. coli, as is the case for lsrE located downstream of lsrG in S. Typhimurium.

In conclusion, by performing ChIP-chip and transcriptomics to delineate the AI-2-dependent LsrR regulon in S. Typhimurium, we show that LsrR controls lsr gene expression by binding the lsr promoter region in vivo. The absence of other LsrR-bound target genes shows that LsrR, under experimental conditions that have been shown to be relevant for the AI-2 phenotype, is a specialized regulator with a limited regulon in S. Typhimurium. This may limit the suitability of LsrR as a target to combat S. Typhimurium infections, although the function of LsrR as a regulator of AI-2 uptake may have broader indirect implications. Based on these results, we therefore suggest that research focusing on the AI-2 molecule itself – including its breakdown products – is highly relevant to unravel the role and impact of putative AI-2 signaling in S. Typhimurium.