|Home | About | Journals | Submit | Contact Us | Français|
Bacterial autoinducer 2 (AI-2) is proposed to be an interspecies mediator of cell-cell communication that enables cells to operate at the multicellular level. Many environmental stimuli have been shown to affect the extracellular AI-2 levels, carbon sources being among the most important. In this report, we show that both AI-2 synthesis and uptake in Escherichia coli are subject to catabolite repression through the cyclic AMP (cAMP)-CRP complex, which directly stimulates transcription of the lsr (for “luxS regulated”) operon and indirectly represses luxS expression. Specifically, cAMP-CRP is shown to bind to a CRP binding site located in the upstream region of the lsr promoter and works with the LsrR repressor to regulate AI-2 uptake. The functions of the lsr operon and its regulators, LsrR and LsrK, previously reported in Salmonella enterica serovar Typhimurium, are confirmed here for E. coli. The elucidation of cAMP-CRP involvement in E. coli autoinduction impacts many areas, including the growth of E. coli in fermentation processes.
Bacteria have evolved complex genetic circuits to modulate their physiological states and behaviors in response to a variety of extracellular signals. In a process termed quorum sensing, or density-dependent gene regulation, bacteria produce, release, and respond to signaling molecules (autoinducers), which accumulate as a function of cell density. Quorum sensing allows bacteria to communicate with each other and coordinate their activities at a multicellular level. The autoinducers of many gram-positive bacteria are secreted peptides (30, 42), while gram-negative bacteria use small chemical molecules (60). Among gram-negative bacteria, the LuxI/LuxR signal synthase-signal receptor system is the most studied at the molecular level, with the signaling species being a family of N-acylhomoserine lactones. However, the cross-species autoinducer, autoinducer 2 (AI-2), has received intense interest recently because the gene for its terminal synthase, luxS, is present in over 55 bacteria and its activity can be readily assayed biologically (61). It is known that quorum sensing regulates diverse cellular processes, including bioluminescence (19, 34), spore formation (33), motility (18, 22), competence (35), conjugation (20), antibiotic synthesis (2, 17), virulence (38, 44, 50), and biofilm maturation (13, 45).
Our laboratory is interested in understanding and controlling microbial behavior in bioreactors in order to enhance recombinant protein synthesis and yield. Since quorum sensing is emerging as a global regulator of many intracellular processes, including those that influence protein synthesis, efforts to understand this “tunable” controller are essential. In our previous work using chemostat cultures (14), many stimuli were found to affect the level of AI-2. Among these, the pulsed addition of glucose, a common carbon source for recombinant Escherichia coli fermentations, resulted in increased AI-2 levels, but with the dynamic response dependent on the steady-state growth rate (e.g., dilution rate) of the culture. Indeed, AI-2 production on a per-cell basis was linearly proportional to the growth rate of the cells (14). Also, the level of AI-2 in extracellular fluids was reduced relative to that in controls in direct response to protein overexpression. This was attributed to the metabolic burden commonly associated with the requisite redirection of metabolites and resources (15).
The level and possible role of AI-2 as a mediator of intra- and intercellular coordination have been the subject of varied interpretations, as recently reviewed by Ahmer (1). Outside of the Vibrio genus, only the lsr operon of Salmonella has been found to directly respond to AI-2, although mutation of its synthase, LuxS, has far-reaching effects (16, 51), but this evidence is muddled by the coincident effects on cell growth (1) and the potential accumulation of S-ribosylhomocysteine (SRH). Hence, it is important to delineate pleiotropic effects of luxS gene knockout and of varying the glucose level and growth rate from molecular events directly attributed to AI-2. In the present work, we investigated the mechanistic effects of glucose on the synthesis and uptake of AI-2 in E. coli W3110.
Schauder et al. demonstrated that AI-2 is produced from S-adenosylmethionine (SAM) in three enzymatic steps and that LuxS is an AI-2 synthase (48) (Fig. (Fig.1).1). SAM serves as methyl donor in a variety of methylation reactions and as a propylamino donor in polyamine biosynthesis (24). The methyl group of SAM is transferred by several methyltransferases to its acceptors, resulting in production of S-adenosylhomocysteine (SAH). Accumulation of SAH is toxic to the cell, and it is rapidly degraded by a nucleosidase, Pfs, into adenine and SRH. LuxS acts on SRH to produce homocysteine, which can be recycled to methionine (24) and 4,5-dihydroxy-2,3-pentanedione, which likely undergoes cyclization and further rearrangement to yield AI-2 (10, 39, 48). Interestingly, the same nucleosidase, Pfs, participates in the SpeD-directed SAM decarboxylation pathway that ultimately produces methylthioribose during synthesis of polyamines. In some bacteria, Pfs is also involved in the pathway responsible for N-acylhomoserine lactone production (40, 43, 56).
The uptake of AI-2 was recently elucidated by Taga et al. in Salmonella enterica serovar Typhimurium (55). A genetic screen for luxS-regulated genes in serovar Typhimurium resulted in identification of the lsr operon (55). The lsrACDBFGEoperon encodes an AI-2 uptake and modification system that has similarity to the ribose transport system of E. coli and serovar Typhimurium. The AI-2 transport apparatus is encoded by the first four genes (lsrA, lsrC, lsrD, and lsrB). Upon entry into cells, AI-2 is phosphorylated by a kinase encoded by lsrK (54). The phospho-AI-2 appears to be the inducer that releases LsrR-mediated repression of lsr transcription (54). LsrR is homologous to SorC, a DNA-binding transcriptional regulator involved in sorbose metabolism in Klebsiella pneumoniae (55). At this time, LsrR has not been shown to bind to the promoter region of the lsr operon, and it is not clear whether the effect of phospho-AI-2 on LsrR is direct, although these are likely, based on the available data. LsrF and LsrG are responsible for further processing of phospho-AI-2 (54). In E. coli, there exists a similar set of genes, including an lsr operon (b1513 operon) and homologues to lsrK (ydeV) and lsrR (ydeW); there is no lsrE homologue. To date, there have been no studies regarding the function and/or regulation of the E. coli lsr operon.
It has long been known that the presence of glucose (or other phosphotransferase system [PTS] sugars) in the growth medium of E. coli and serovar Typhimurium cultures affects the level of extracellular AI-2 (25, 52). Moreover, maximal AI-2 activity is typically observed during mid- to late exponential phase, and this extracellular activity is removed when glucose becomes depleted (52). These findings suggest a linkage between catabolite repression and AI-2 production and transport. In this study, we show that catabolite repression influences AI-2 accumulation through the cyclic AMP (cAMP)-cAMP receptor protein (CRP) complex, which directly stimulates transcription of the lsr operon and indirectly represses luxS expression. cAMP-CRP is shown to bind to the upstream region of the promoter of the lsr operon and works with the LsrR repressor to regulate AI-2 uptake. A working model describing the appearance and disappearance of AI-2 in E. coli cultures is presented, along with our interpretation of AI-2 regulation.
The bacterial strains used in this study are listed in Table Table1.1. Luria-Bertani broth (LB) contained 5 g of yeast extract (Sigma) liter−1, 10 g of Bacto tryptone (Difco) liter−1, and 10 g of NaCl liter−1. Glucose and cAMP, when present, were added at 0.8% and 10 mM, respectively. The autoinducer bioassay (AB) and Luria-marine (LM) media are described in detail elsewhere (6, 23). When necessary, media were supplemented with antibiotics at the following concentrations: ampicillin, 60 μg ml−1; kanamycin, 50 μg ml−1.
The plasmids used in this study are listed in Table Table11 and were generated using standard procedures (47). Restriction enzymes, T4 DNA ligase, and Vent DNA polymerase were used as specified by the manufacturer (New England Biolabs, Beverly, Mass.). An E. coli W3110 chromosomal DNA preparation was performed using the QIAGEN DNeasy tissue kit (QIAGEN, Valencia, Calif.). Extractions of DNA from agarose gels were performed using the QIAGEN Qiaquick gel extraction kit. Oligonucleotides were from Gene Probe Technologies (Gaithersburg, Md.). DNA sequencing was performed at the DNA Core Facility of the Center of Biosystems Research (University of Maryland Biotechnology Institute). All constructs made by PCR were sequenced to verify their integrity.
Plasmid pFZY1 is a mini-F derivative (average copy number, one to two per cell) with a polycloning site upstream of a promoterless galK′-lacZYA reporter segment (32). To create pLW10, the luxS promoter region was amplified by PCR using primers luxSpF and luxSpR (Table (Table2).2). This DNA fragment was cloned into the pCR-Blunt vector (Invitrogen). The resulting plasmid, pLW9, was digested with EcoRI; the fragment containing the luxS promoter was inserted into the EcoRI site of pFZY1, producing plasmid pLW10. The luxS promoter region (−104 to +36 relative to the luxS start codon) contains the native sequence up to the transcriptional terminator of the upstream gene gshA. To create pYH10, the pfs promoter region (−337 to +113 relative to the pfs start codon) was amplified by PCR using the primers pfspF and pfspR (Table (Table2).2). The purified PCR product was digested with EcoRI-BamHI and was inserted into EcoRI-BamHI-digested pFZY1. To create pLW11, the promoter region of the lsrACDBFG operon [−307 to +92 relative to the start codon of lsrA (b1513)] was amplified by PCR using primers lsrpF and lsrpR (Table (Table2).2). The purified PCR product was digested with EcoRI-BamHI and was inserted into EcoRI-BamHI-digested pFZY1.
To create pLW12, the method of site-directed mutagenesis by PCR (21, 28) was used and modified as follows. Two subsequent PCR steps were carried out. In the first round of PCR, lsrPM and lsrPR were used as primers with pLW11 as template. The PCR products were purified with a QIAGEN MinElute PCR purification kit and used as a megaprimer in the second round of PCR with pLW11 as the template and lsrPF as another primer. The products in the second round of PCR were purified using the QIAGEN Qiaquick gel extraction kit and cloned into pCR-Blunt II TOPO vector (Invitrogen), which was transformed into Top10 competent cells (Invitrogen). The plasmids were prepared from the transformants and sequenced to confirm the mutation of the lsr promoter region. The plasmid with correct insertion was digested with EcoRI-BamHI, and the DNA fragment containing a mutated lsr promoter was inserted into EcoRI-BamHI-digested pFZY1.
The one-step replacement method described by Datsenko and Wanner (12) was used to construct a crp deletion in E. coli W3110 and ZK126. We used the phage λ Red recombination system to replace crp with a crp::kan PCR fragment. PCR was done using pKD4 as template and the primers crpHP1 and crpHP2 (Table (Table2).2). The PCR products were treated with DpnI and introduced by electroporation into E. coli W3110 or ZK126 containing plasmid pKD46, which expresses the Red recombinase and was cured later by growth at 37°C. Recombinants were selected on LB supplemented with kanamycin. Deletions of luxS, lsrR, and the lsrACDBFG operon were constructed similarly by PCR amplification of pKD4 with primers luxSHP1 and luxSHP2, lsrRHP1 and lsrRHP2, and lsrHP1 and lsrHP2, respectively, except that in making the deletion of lsrK, pKD13 was used as template with primers lsrKHP1 and lsrKHP2 for PCR (Table (Table2).2). The deletion of genes was verified by PCR tests.
Unless otherwise stated, an overnight culture grown in LB or LB plus 0.8% glucose was diluted 100-fold to an optical density at 600 nm (OD600) below 0.03 in LB or LB plus 0.8% glucose. Cells were incubated at 30°C with shaking at 250 rpm in Erlenmeyer flasks. Samples were removed at regular intervals. Cell-free culture fluids were prepared by centrifugation of the E. coli culture at 12,000 rpm for 10 min in a microcentrifuge. Cleared supernatants were filtered (0.2-μm-pore-size HT Tuffryn filters; Pall Corp., Ann Arbor, Mich.) and stored at −20°C.
E. coli cell-free culture fluids were tested for the presence of AI-2 by inducing luminescence in Vibrio harveyi reporter strain BB170. The assays were performed as outlined by Surette and Bassler (52). Briefly, BB170 was grown for 16 h with shaking at 30°C in AB medium, diluted 1:5,000 in fresh AB medium, and aliquoted to sterile 12- by 75-mm tubes (Fisher Scientific Co., Inc., Pittsburgh, Pa.). Cell-free culture fluids were added to a final concentration of 10% (vol/vol) to these tubes. Positive controls contained 10% (vol/vol) cell-free culture fluid from BB152, while negative controls contained 10% (vol/vol) sterile LB or LB plus 0.8% glucose. Tubes were shaken at 175 rpm and 30°C in an air shaker (New Brunswick Scientific), and hourly measurements of luminescence were taken. Luminescence was measured as a function of V. harveyi cell density by quantitating light production with a luminometer (EG&G Berthold, Gaithersburg, Md.). V. harveyi cell density was determined by spreading identical dilutions used for luminescence measurements onto solid LM medium and counting colonies after overnight growth. Relative light units (RLU) were defined as follows: [(counts per minute per milliliter) (103)]/(CFU per milliliter). AI-2 activities were obtained by dividing the RLU produced by the reporter after addition of E. coli culture fluid by the RLU of the reporter when growth medium alone was added (52, 53). The obtained values were in a linear range.
Cultures of E. coli were grown overnight in LB, diluted 100-fold into fresh LB, grown to mid-exponential phase, and then diluted into different medium with the OD600 below 0.03. The cultures were incubated at 30°C with shaking at 250 rpm in flasks. Samples were removed at intervals for determination of the OD600 and β-galactosidase activity. Specific activity of β-galactosidase is expressed in Miller units (37).
The 140-bp EcoRI fragment containing the promoter region of the luxS gene was prepared from pLW9. The 42-, 89-, and 120-bp DNA fragments containing the wild-type or mutated promoter regions of the lsr operon were synthesized by Integrated DNA Technologies (Coralville, Iowa). A digoxigenin gel shift kit (Boehringer Mannheim) was used for labeling of DNA fragments and detection of signals according to the manufacturer's instructions. Binding reactions were performed by incubating the labeled DNA fragments with various amounts of purified CRP (generously provided by Fred Schwarz, University of Maryland Biotechnology Institute) in 20 μl of binding buffer (10 mM Tris-HCl [pH 8.0], 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 50 μg of bovine serum albumin/ml, 15 μg of sonicated salmon sperm DNA/ml, 100 μM cAMP). Following incubation at 37°C for 10 min, 5 μl of gel loading buffer (0.25× TBE, 60%; glycerol, 40%; bromphenol, 0.2% [wt/vol]) was added, and mixtures were electrophoresed in a 6% native polyacrylamide gel in 0.5× TBE buffer (45 mM Tris-borate, 1 mM EDTA; pH 8.0) containing 100 μM cAMP. DNA bands were detected according to the manufacturer's instructions.
Extracellular AI-2 activity is produced when E. coli is grown with glucose or other PTS saccharides (52). The levels of extracellular AI-2 in the cell-free culture fluids of E. coli W3110 grown in LB and LB containing 0.8% glucose are shown in Fig. Fig.2.2. AI-2 was produced under both growth conditions but at different levels. In the presence of glucose, AI-2 activity increased during the exponential phase, reached the maximum at 4 h, and then declined slowly. This result is consistent with a previous report (52). However, we also found that when grown in LB in the absence of glucose, the bacteria still produced AI-2 activity during exponential phase, but at a much lower level. Furthermore, no AI-2 activity in cell-free culture fluids was detected after the cells entered stationary phase.
Glucose is known to affect gene expression through the cAMP-CRP complex, and the presence of glucose in the growth medium results in decreased levels of cAMP and CRP (26, 36). To check whether cAMP and CRP are involved in the production of extracellular AI-2 activity, we compared the extracellular AI-2 activity produced by W3110 with that of isogenic crp and cyaA mutants grown in LB (Fig. (Fig.3).3). Deletion of the crp gene or the cya gene in W3110 resulted in much higher AI-2 activity throughout the cell cultures (peak levels increased ~4-fold). In addition, the AI-2 activity remained high during stationary phase, suggesting the involvement of cAMP and CRP in regulating biosynthesis and/or removal of AI-2 from the medium. The differences in AI-2 activity of cell-free fluids from the wild type and crp and cya mutants grown with glucose (data not shown) were not as marked as the differences observed when cells were grown without glucose, probably due to the already very low level of cAMP and CRP in the cells in the presence of glucose.
The introduction of plasmid-borne crp or cya reduced the AI-2 activity to levels closer to that of W3110. In fact, the Δcrp mutant LW1 transformed with crp+ plasmid pHA7E resulted in even lower AI-2 activity than that produced by the wild type. This may be due to the use of multicopy pHA7E, which increased the concentration of CRP above that in W3110.
Since deletion of either crp or cya greatly enhanced extracellular AI-2 activity, we tested whether this was due to modulation of some genes involved in AI-2 biosynthesis. lacZ fusions were constructed to check the regulation of luxS and pfs, which encode enzymes responsible for AI-2 synthesis. We found that addition of 0.8% glucose to the growth medium increased the β-galactosidase activity from the luxS promoter (<2-fold), while addition of 10 mM cAMP partly offset this glucose effect (Fig. (Fig.4A).4A). Moreover, deletion of the crp gene increased the expression of luxS (Fig. (Fig.4A).4A). These results suggest that the expression of the luxS gene is negatively regulated by cAMP and CRP.
In contrast to the effect of glucose on luxS transcription, the presence of glucose lowered the level of β-galactosidase activity expressed from the pfs promoter, principally in the stationary phase (Fig. (Fig.4B).4B). The addition of cAMP to the cells grown in LB with glucose did not restore β-galactosidase activity to the level when the cells were grown in LB alone. Furthermore, deletion of crp had no effect on β-galactosidase expression from the pfs promoter relative to that in the wild-type control. While these results suggest that glucose plays a role in pfs expression, the control is likely to be through a mechanism other than the cAMP-CRP complex
We tested whether the action of CRP and cAMP on luxS transcription was mediated by cAMP-CRP binding near the promoter of luxS. CRP binding sites contain a palindromic sequence in which two conserved motifs, TGTGA and TCACA, are separated by a spacer (31). The length of the spacer is usually 6 bp, but spacers of 7 or 8 bp, and possibly 9 bp, have been observed (3, 4, 31, 46). Although a potential CRP binding site, 5′-TGGGAagaaagagtTCAGA-3′, was present (spacer shown in lowercase letters), a gel mobility shift assay showed no binding of the cAMP-CRP complex to the luxS promoter region (data not shown), suggesting that the effect of CRP and cAMP on luxS transcription was probably indirect.
Although deletion of the crp gene resulted in an increase of luxS expression (<2-fold) in the absence of glucose (Fig. (Fig.4A),4A), this seemed insufficient to explain the dramatic differences in extracellular AI-2 activities in the wild type and crp mutant, particularly during the stationary phase (Fig. (Fig.3).3). Thus, we investigated the role of cAMP and CRP in AI-2 uptake. First, to confirm the role of the lsr operon in E. coli, we made a deletion of the entire lsrACDBFGE.c. operon (see Materials and Methods). The Δlsr(ACDBFG)E.c mutant showed much slower removal of AI-2 from extracellular fluids relative to the wild type when grown in the absence of glucose (Fig. (Fig.5A),5A), although the removal of AI-2 was not completely blocked. It is likely that there is an alternative mechanism for AI-2 removal from the E. coli extracellular medium. To explain a similar observation in serovar Typhimurium, Taga et al. suggested that there may exist another low-affinity transporter for AI-2 uptake (54). Deletion of lsrRE.c. resulted in accelerated removal of AI-2 from extracellular fluids (Fig. (Fig.5A)5A) relative to the wild type when grown in LB, similar to a serovar Typhimurium lsrRS.t. mutant (54). Finally, deletion of lsrKE.c caused a severe defect in AI-2 removal (Fig. (Fig.5A).5A). This is also consistent with the serovar Typhimurium lsrKS.t. mutant, where it was suggested that the absence of LsrK prevents sequestration of AI-2 in cytoplasm in the form of phospho-AI-2 (54). In summary, the function and control of the lsrE.c.operon seems to operate similarly as in serovar Typhimurium in its role as an AI-2 autoregulated transporter and processing system.
Although the ΔlsrR, ΔlsrK, and Δlsr(ACDBFG) mutants and the wild-type cells displayed different rates of AI-2 removal when grown in LB without glucose, all of the mutants and the wild type showed very similar extracellular AI-2 levels when glucose was present in the growth medium. The levels of AI-2 were relatively high throughout the stationary phase (Fig. (Fig.5B).5B). These results suggest that the presence of glucose may affect the regulation of AI-2 uptake.
To investigate involvement of glucose-mediated catabolite repression in lsr regulation, we constructed a lacZ fusion under control of the promoter region of the lsrACDBFG operon. When the wild-type cells (ZK126) were grown in LB medium, lsr expression remained very low until the cells entered the stationary phase, consistent with the accumulation of AI-2 as observed earlier (Fig. (Fig.33 and and66 and Table Table3).3). The addition of 0.8% glucose to the growth medium strongly decreased transcription from the lsr promoter in the wild type and in all of the lsr mutants (Table (Table3).3). These results are consistent with the significantly higher extracellular AI-2 activities of these cells (Fig. (Fig.5B).5B). Deletion of the crp gene decreased lsr expression (Table (Table3),3), indicating that CRP is needed to activate transcription from the lsr promoter. Deletion of lsrR caused a significant increase in lsr expression (Fig. (Fig.66 and Table Table3),3), confirming the role of the LsrR protein as a repressor protein. These transcription results are consistent with the rapid AI-2 removal observed in the ΔlsrR mutant (Fig. (Fig.5A).5A). The much lower lsr expression levels in the exponential phase in the ΔlsrR mutant and other strains were likely caused by the low levels of glucose and other PTS sugars initially present in the LB medium. It is also noteworthy that deletion of either luxS or lsrK resulted in a much lower level of lsr expression during late-exponential and stationary phases (Fig. (Fig.66 and Table Table3),3), supporting the AI-2/phospho-AI-2-dependent regulation, as shown in serovar Typhimurium (54). It was further shown here that LsrR-mediated repression of lsr expression is not complete, as noted by significant expression levels in the ΔluxS or ΔlsrK mutants (Fig. (Fig.6).6). Finally, deletion of the entire lsrACDBFG operon resulted in a significant increase in transcription from the lsr promoter (Fig. (Fig.66 and Table Table3).3). This was, at first, unexpected since the absence of the Lsr transporter decreases uptake of AI-2, which is recruited to enhance lsr transcription. Thus, in the absence of AI-2 we would expect effective LsrR-mediated repression. We suggest that LsrK phosphorylates endogenous AI-2 and/or AI-2 imported by the alternative transporter, which then derepresses transcription from the lsr promoter through LsrR (see Discussion). Moreover, since LsrF and LsrG are not present, which are reported to promote AI-2 degradation (54), the inducer phospho-AI-2 may persist.
To evaluate whether cAMP-CRP directly modulates lsr transcription, we performed gel mobility shift assays. The promoter region of the lsr operon contains two potential CRP binding sites (Fig. (Fig.7A).7A). Site 1 is located 19 to 34 bp upstream of the transcription start site (predicted via the website http://www.fruitfly.org/seq_tools/promoter.html), while site 2 is located at residues −60 to −77. Our results revealed that cAMP-CRP binds to the DNA fragment containing both site 1 and site 2, but not the fragment containing only site 2 (Fig. (Fig.7B).7B). A shorter DNA fragment containing only site 1 was bound by cAMP-CRP (Fig. (Fig.7C),7C), and there was no binding of cAMP-CRP to a DNA fragment that contained only site 1 with substitutions in 4 bp in one of the CRP binding motifs (Fig. (Fig.7C).7C). These data confirm that cAMP-CRP specifically recognizes and binds to the CRP binding site located in site 1, and not site 2. Interestingly, the spacer for this binding site is 8 bp, while most of the known CRP binding sites contain a conventional 6-bp spacer (3, 4, 31). It has been demonstrated that cAMP-CRP binds to many sites with an 8-bp spacer in vitro (3, 4, 9, 49), although we have found no reports showing such CRP binding sites are functional in vivo. In this study, in vivo experiments were performed with wild-type ZK126 cells carrying pLW12, which contains the identical lsr promoter region as pLW11 but with the four base substitutions in the CRP binding site as used in the gel shift assay (Fig. (Fig.7A,7A, seq4). Significantly decreased transcription from the mutated lsr promoter was found (Fig. (Fig.66 and Table Table3),3), indicating a requirement for this site in cAMP-CRP-mediated activation. These results demonstrate that cAMP-CRP stimulates transcription of the lsrACDBFG operon by directly binding to the upstream region of the lsr promoter and that mutation in the CRP binding site abolishes this stimulation.
AI-2 was first discovered in V. harveyi as a quorum signal to regulate bioluminescence (5, 6). In addition to communicating cell density, it is suggested that AI-2 may relay information pertaining to the growth phase and metabolic potential of the bacterial cells (14, 62). In this study, we investigated the regulation of AI-2 production and uptake. We showed that cAMP and CRP negatively regulate luxS transcription. However, the action of CRP and cAMP on luxS transcription is not mediated by cAMP-CRP binding to the promoter region of luxS, as indicated by the gel shift assay (data not shown). It is possible that cAMP-CRP may control another transcriptional regulator(s) that modulates luxS expression. Interestingly, we found that the expression of pfs was reduced by the presence of glucose, but not through the cAMP-CRP complex. Considering that Pfs is involved in both AI-2 synthesis and polyamine formation pathways (Fig. (Fig.1),1), regulation of AI-2 production through LuxS appears more practical and efficient, although other factors, such as the concentrations of precursors, may also affect AI-2 synthesis. Beeston and Surette found that the extracellular AI-2 activities were correlated with the transcription profile of pfs in serovar Typhimurium (7), which is similar to our result in the presence of glucose (Fig. (Fig.22 and and4).4). Despite the correlation, the earlier (54, 55) and present works suggest that the extracellular activities of AI-2 are controlled by a dynamic regulatory process involving many factors that affect its synthesis and transport. In the stationary phase the AI-2 levels in culture medium appear to be decided by the rate of its import rather than the rate of synthesis.
Since E. coli possesses almost all genes homologous to those in the lsr operon of serovar Typhimurium, we were curious whether the function and regulation of the E. coli Lsr transporter were similar to those of serovar Typhimurium and whether cAMP and CRP were involved in regulation of AI-2 transport. Our data are consistent with both, although some differences were noted. For example, lsrE is apparently absent in E. coli. This gene is homologous to rpe, which encodes the ribulose phosphate epimerase, suggesting the possibility that an additional step(s) is required for AI-2 modification in serovar Typhimurium. Examination of the upstream region of the serovar Typhimurium lsr operon reveals one potential CRP binding site (spacer shown in lowercase letters): 5′-TGAGAgtttttTGACC-3′ (−32 to −47 relative to the predicted transcriptional start site of the lsr operon). This site has a 6-bp spacer, and its function has yet to be confirmed. While the molecular basis for the cAMP-CRP-mediated regulation may be slightly different, the function and control of the Lsr transporter are similar. In a genetic screen for the regulator of the lsr operon in serovar Typhimurium, mutation of either cya or ptsI caused reduced transcription of the lsrC-lacZ reporter (54). Since Cya is directly involved in the production of cAMP, and since PtsI is enzyme I of the PTS and is required for the activation of Cya, it is not surprising that mutation of cya or ptsI results in lower expression of the lsr operon.
We noted that deletion of the whole lsr operon in E. coli did not completely block AI-2 import (Fig. (Fig.5A),5A), which was also the case for the Lsr transport mutants of serovar Typhimurium (54, 55). Taga et al. proposed the existence of another low-affinity Lsr-independent transport mechanism(s) (54). Interestingly, while the transport of AI-2 is slower in this mutant, there is much higher transcription from the lsr promoter relative to that in both the wild type and the ΔluxS mutant (Fig. (Fig.6)6) (all in LB medium). In other words, transcription from the lsr promoter can still be induced even without AI-2 import by the Lsr transporter. Perhaps endogenous AI-2 and/or AI-2 imported from an alternative transporter can serve as the substrate for the LsrK kinase to produce the inducer, phospho-AI-2. Consistent with this, lsr expression in a serovar Typhimurium ΔlsrB mutant with a defective Lsr transporter is higher than that in a ΔluxS mutant, although lower than that in the wild type (54). The reduced lsr expression in the ΔlsrB mutant relative to that in the wild type probably results from a lower rate of AI-2 phosphorylation due to the absence of AI-2 uptake from a functional Lsr transporter. The higher induction from the lsr promoter in the E. coli ΔlsrACDBFG mutant is probably also influenced by the loss of the phospho-AI-2 degradation by LsrF and LsrG. Hence, LsrF and LsrG may function as a signal terminator in the wild-type E. coli cells, as shown in serovar Typhimurium (54).
There are, therefore, several regulators that influence the expression level of the Lsr transporter. Since σs is very important in sensing stress during the transition to stationary phase, we investigated the possibility that σs plays an additional role in controlling AI-2 transport. Our results showed that mutation of rpoS (encoding σs) causes an eightfold increase in the transcription of the E. coli lsr operon during the late exponential growth phase when cells are grown in the absence of glucose (Table (Table3),3), and extracellular AI-2 activities in ΔrpoS mutant are significantly decreased (data not shown). Notley-McRobb et al. reported that the mutation of rpoS induces expression of certain transporter genes under glucose limitation, probably due to loss of competition between σs and σ70 for core RNA polymerase (41). Also, it may be possible that σs controls other regulators that affect lsr expression. In addition to σs, Taga et al. reported that transcription of the lsrC-lacZ reporter in serovar Typhimurium is lowered eightfold by mutation of the lon gene, which encodes the Lon protease (54). How Lon is related to the regulation of lsr expression is unclear.
Our working model for AI-2 synthesis and transport in E. coli (Fig. (Fig.8)8) is that cAMP-CRP acts as a global controller, while LsrR functions as a specific controller. When glucose or other PTS sugars are present in the growth medium, low intracellular levels of cAMP and CRP result, and there is almost no transcription of the lsr operon. Instead, the level of luxS transcription increases, and it is likely that the metabolic flux of precursors (SAM, SAH, etc.) also increases concomitant with an increased demand for biosynthesis. It is also likely that other cellular activities requiring methylation (through SAM) will impact the level of the substrate for AI-2 synthesis. Thus, with sufficient glucose, increased carbon and energy storage may result and the net effect is an acceleration in AI-2 synthesis and export. This is consistent with the exponential-phase results shown in Fig. Fig.55 and and6,6, where there was minimal lsr transcription and the AI-2 levels were significantly higher (~3 times) in both the wild type and the ΔlsrACDBFG mutant (having drastically impaired AI-2 uptake) in the presence of glucose.
However, when glucose or other PTS sugars are absent, the cAMP-CRP complex binds to the upstream region of the lsr promoter and stimulates its transcription. It is under this condition that LsrR, LsrK, LsrF, and LsrG play a greater role in regulating and “tuning” AI-2 uptake. If there is no inducer (phospho-AI-2), LsrR represses transcription of the lsr operon. During the late exponential phase, a basal level of lsr expression likely results in uptake and phosphorylation of a small amount of AI-2. Endogenous AI-2 and AI-2 imported by a low-affinity transporter may also provide substrate for LsrK. The phospho-AI-2 inactivates the repressor LsrR and increases transcription of the lsr operon. Then, more AI-2 is transported into the cells and the cycle continues until LsrF and LsrG provide feedback control.
In summary, we have shown that E. coli cells synthesize and secrete AI-2 in the early growth phase and take it up during the stationary phase under glucose limitation. The simplest interpretation is that the bacterial cells use AI-2 as a carbon source in the absence of the preferred glucose, just like they use lactose, arabinose, and galactose, which are similarly regulated. This interpretation does not address its role as a signal molecule, however. Taga et al. reported that serovar Typhimurium cannot grow in minimal medium containing AI-2 as the sole carbon source (55). Winzer et al. suggested that utilization of AI-2 as the sole carbon source might require additional conditions (59). Yet, they suggested further that AI-2 may be a toxic by-product of SAM metabolism, which is excluded during early growth and taken up and metabolized at a later stage (for detoxifying and recycling the energetically expensive “ribose equivalent” unit) (58, 59). We found that the ΔluxS mutant of E. coli ZK126 grows as well as the wild type in LB medium (Fig. (Fig.6),6), but Sperandio and coworkers found that mutation of luxS in enterohemorrhagic E. coli O157:H7 resulted in a faster growth of the mutant when grown in Dulbecco's modified Eagle's medium and that there was a global effect on gene expression (51). It is therefore possible that the specific growth condition and/or specific genotype might affect or determine the role of AI-2 as a signal molecule. Indeed, there are many other reports suggesting that AI-2 may participate in the control of certain genes and physiological activities (reviewed in reference 61). Thus, while the role of AI-2 as a signal molecule remains unclear in E. coli W3110, it is clear that these cells have a complex hierarchical regulatory system for its control that is in part based on AI-2, suggesting potential involvement of AI-2 in the regulation of additional cellular processes.
We thank F. P. Schwarz for the gift of CRP and B. L. Bassler, H. Aiba, R. Kolter, A. Hochschild, L. I. Rothfield, and M. Berlyn for generously providing strains and plasmids used in the study. We thank J. Li for some help in making the lsrK deletion. We are grateful to S. W. Hutcheson and R. C. Stewart for helpful discussions.
This work was supported by the U.S. Army, SBCCOM, Aberdeen Proving Ground, Md. (DAAD 13-01-C-0036).