Introduction

Bacteria have evolved complex genetic circuits to regulate their physiological activities and behaviors in response to a variety of environmental cues. Quorum sensing (QS) is a bacterial cell-cell communication process by which bacterial cells produce, secrete, and detect small-molecule signals named autoinducers, whose concentration corresponds to population density. These extracellular signals include N-acylhomoserine lactones (autoinducer-1, AI-1 or AHL) and 4,5-dihydroxy-2,3-pentanedione derivatives (generally called autoinducer-2, AI-2), which serve as signal molecules of QS system 1 and system 2, respectively 1, 2. When they reach a threshold concentration in the local environment, a signal transduction cascade is initiated, which ultimately results in a coordinated and population-wide alteration in bacterial behaviors such as bioluminescence, exopolysaccharide secretion, biofilm formation, and virulence 2, 3, 4.

QS system 1 includes protein pairs similar to LuxR and LuxI. The system 1 autoinducer named AI-1 is produced by LuxI and is detected by LuxR 3. Two QS systems have been identified in Escherichia coli K-12 5. E. coli encodes a LuxR homologue, SdiA (suppressor of cell division inhibitor), but it does not encode a LuxI homolog or synthesize any AHL molecule detected by SdiA 5. SdiA of E. coli and Salmonella enterica serovar Typhimurium, however, responds to several AHLs generated by other microbial species 6, 7, 8, 9. Recently, a report indicated that E. coli uses SdiA to monitor indole and AHLs to control biofilms; however, it was not ascertained whether indole itself binds to SdiA 10. A previous study showed that SdiA could support or inhibit RNA polymerase binding to the promoters, thereby affecting the transcription of the target genes 11. Overexpression of SdiA affects the expression of a battery of genes, including those that control cell division, motility, chemotaxis, and multidrug efflux pump genes 5, 6, 12, 13.

E. coli has an intact QS system 2 that includes nine genes (e.g. luxS, lsrR, lsrK, and lsrACDBFG). The cyclic AMP (cAMP)-cAMP receptor protein (CRP) complex stimulates the expression of lsrR and the lsr operon, which includes lsrACDBFG, while LsrR represses their expression and is located adjacent to, while being transcribed divergently from, the lsr operon 14, 15, 16. The system 2 autoinducer named AI-2 is synthesized by LuxS and accumulates extracellularly. Following internalization by the Lsr transporter encoded by genes in the lsr operon, AI-2 is phosphorylated by LsrK, and phospho-AI-2 binds specifically to LsrR and antagonizes it. LsrF and LsrG are required for further processing of phospho-AI-2 17. It has been recently reported that the mean biofilm thickness and biomass of lsrR or lsrK mutants are lower than those of the wild type, while the global small RNA (sRNA) regulator DsrA and the sRNA cell division inhibitor DicF are induced 2- to 4.4-fold in both lsrR and lsrK mutants 18.

A class of enzymes containing GGDEF domains synthesize the second messenger 3′,5′-cyclic diguanylic acid (c-di-GMP) that is later hydrolyzed by EAL or HD-GYP domain proteins 19, 20, 21, 22, 23. In some cases, the GGDEF or EAL domains can play a regulatory role in the absence of a catalytic role; e.g., a catalytically inactive GGDEF domain of the CC3396 protein has retained the ability to bind GTP and, in response, can activate the neighboring EAL domain 24. As a second messenger in a variety of bacterial species, c-di-GMP influences various aspects of physiology and behavior, such as motility, sessility, virulence, biofilm, and cell-cell communications 25. The protein encoded by ydiV in E. coli, which is 52% identical at the amino-acid level to its S. typhimurium homolog CdgR, consists solely of an EAL domain 26. In a cdgR mutant of S. typhimurium, the c-di-GMP concentration was about 7-fold higher than that in a wild-type (WT) strain, but CdgR did not exhibit c-di-GMP phosphodiesterase activity in vitro, probably due to lack of suitable trial conditions or indirect regulation of c-di-GMP by CdgR, such as through induction of a phosphodiesterase 27.

Three QS circuits of Vibrio harveyi converge to control one set of genes in a process that involves sRNA species 28, 29. In E. coli, the cAMP-CRP complex stimulates the expression of both lsrR and the lsr operon by binding to the promoter regions, while LsrR represses their expression 14, 15. The relationship between QS system 1 and QS system 2 in E. coli, however, remains obscure. Here, we show that the addition of glucose to the culture inhibited the expression of sdiA and ydiV. The expression of ydiV was stimulated by SdiA, and extracellular AI-1 from Vibrio fischeri further up-regulated ydiV expression in an sdiA-dependent manner. Furthermore, we discovered that ydiV and cAMP contribute to the interaction between the two QS systems in E. coli. On the basis of these results, we propose a working model that describes the signaling pathway linkage in E. coli that includes the two QS systems, cAMP signaling and YdiV.

Results

Effects of extracellular AI-1 and glucose on ydiV expression

The sequence 5′-AAA AGN NNN NNN NGA AAA-3′ was identified as the SdiA binding site within the ftsQ promoter region, which is hereafter designated as the SdiA box 11. Bioinformatics analysis suggested the presence of an SdiA box in the ydiV promoter: 5′-AAA AGg acc cct GAA AA-3′ (−128 to −112 relative to the translation start site of ydiV, spacer shown in lowercase letters). A microarray analysis was used in E. coli to identify genes that respond to plasmid-based expression of sdiA, and it revealed that the expression of ydiV is elevated with increased sdiA dosage 12. All of the above imply that extracellular AI-1 might up-regulate ydiV expression in an sdiA-dependent manner.

To investigate whether ydiV is regulated by chromosomal sdiA in vivo, we constructed an sdiA mutant and the ydiV-lacZ promoter fusion plasmid pPydiV and tested ydiV-lacZ expression. As indicated by β-galactosidase activity assays, sdiA deletion lowered the level of ydiV expression by 30%, leaving significant expression of ydiV even in the absence of SdiA (Figure 1A). Several AHLs generated by other microbial species activate SdiA of E. coli and S. typhimurium by binding and changing its conformation 6, 7, 8, 9; thus, if chromosomal sdiA up-regulates the expression of ydiV, the addition of AHL would elevate ydiV expression. Our results showed that the addition of AHL (referring to oxoC6, a V. fischeri AHL signal molecule) resulted in ydiV up-regulation in WT but not in the sdiA mutant, as expected (Figure 1B and 1C). Under aerobic conditions, ydiV expression in WT increased 1.5-fold (in Luria-Bertani (LB)) or 2.2-fold (in LB plus AHL) more than that in the sdiA mutant in 12-h-old cultures (Figure 1B). In 24-h-old cultures, the fold-difference for ydiV expression in WT versus the sdiA mutant was not greater than that in 12-h-old cultures under aerobic conditions (data not shown). Under microaerobic conditions, the ydiV expression in WT increased 4-fold (in LB) or 6-fold (in LB plus AHL) more than that in the sdiA mutant in 24-h-old cultures with OD600 values around 0.57 (Figure 1C). These data demonstrate that AHL from V. fischeri further stimulates ydiV expression in WT but not in the sdiA mutant.

Figure 1
figure 1

Effects of SdiA and AHL on the expression of ydiV. (A) SdiA stimulates ydiV expression. Conditions for cell growth and β-galactosidase activity assays are described in Materials and Methods. E. coli ZK126 (WT) and isogenic sdiA mutant carrying plasmid pPydiV (ydiV-lacZ) were grown in LB. At different time points during cell growth, aliquots were collected for the measurement of the OD600 (squares and triangles) and β-galactosidase activity (bars). (B and C) Effects of AHL on the transcription of ydiV. β-Galactosidase activity under aerobic conditions (B) and under microaerobic conditions (C) in E. coli ZK126 (WT) and isogenic sdiA mutant carrying plasmid pPydiV (ydiV-lacZ) was measured after growth in LB or LB containing AHL.

Full size image

To further evaluate whether SdiA can bind to the ydiV promoter region and regulate the expression of ydiV, a series of DNA fragments were synthesized, and a gel shift assay was performed with purified SdiA or AHL-SdiA complex. As shown in Figure 2A and 2B, AHL-SdiA complex and SdiA bound the ydiV promoter fragments (S-ydiV) in a dose-dependent manner, and the addition of 10 μM AHL to SdiA-binding buffer caused a more intense gel shift (about 20-fold) when using SdiA but not AHL-SdiA complex. The gel shift with AHL-SdiA complex was more intense (about 20-fold) than that with SdiA (Figure 2B). To assess nonspecific binding, we used both the mutated ydiV promoter fragments (M-ydiV) and a control DNA (S-con) that is devoid of an SdiA box. The gel shift assay shows that the AHL-SdiA complex bound to M-ydiV, S-ftsQ, and S-con as well (Figure 2A), among which S-ftsQ is the positive control. To exclude any protein contaminant, the purity of SdiA and AHL-SdiA complex preparation was ascertained by SDS-PAGE (Figure 2C). In summary, extracellular AI-1 causes more intense binding of SdiA to the ydiV promoter region, and it up-regulates ydiV expression in an sdiA-dependent manner.

Figure 2
figure 2

Binding of SdiA to the ydiV promoter. Gel shift assays were performed as described in Materials and Methods. The symbol (+) represents the addition of 10 μM AHL to SdiA-binding buffer. The arrow denotes the SdiA-DNA complexes. (A) Gel shift assay was conducted using DNA fragment S-ydiV (lanes 1-4), M-ydiV (lane 5), S-ftsQ (lane 6) and S-con (lane 7) as probes. AHL-SdiA complex was used: lane 1, 0 pmol; lane 2, 7.5 pmol; lanes 3-7, 15 pmol. (B) Gel shift assay was conducted using DNA fragment S-ydiV as probe, SdiA (lane 1, 0 pmol; lanes 2 and 4, 7.5 pmol; lane 3, 15 pmol), and AHL-SdiA complex (lane 5, 7.5 pmol). (C) The purity of SdiA and AHL-SdiA complex was analyzed by SDS-PAGE. Lane 1, protein standard; lane 2, SdiA; lane 3, AHL-SdiA complex.

Full size image

To check whether glucose plays an important role in sdiA and ydiV expression, we used the ydiV-lacZ promoter fusion plasmid pPydiV described above and constructed the sdiA-lacZ promoter fusion plasmid pPsdiA. The addition of 0.8% glucose to the growth medium decreased the β-galactosidase activity from the sdiA promoter (2.4-fold) and ydiV promoter (3.6-fold) in WT. ydiV expression in the sdiA mutant was also down-regulated by the addition of glucose (3.5-fold), while the addition of 10 mM cAMP did not restore β-galactosidase activity to the level observed when the cells were grown in LB alone. These results suggest that glucose inhibits sdiA and ydiV expression and that the effect of glucose on ydiV expression is not mediated by SdiA.

The sdiA-ydiV double mutation results in QS system 2 inhibition

In V. harveyi, three QS circuits converge to control one set of genes, as mentioned above 28, 29. However, the interaction between QS systems 1 and 2 in E. coli remains unclear. Together, previous studies and our result show that ydiV, sdiA, and QS system 2 are all repressed by glucose 14, 15. Some complex processes, namely virulence and biofilm, are regulated by QS system 1, QS system 2, and EAL domain proteins 10, 25, 27, 30, 31, 32. It stands to reason that QS system 1 and system 2 in E. coli might be linked and that ydiV might be involved.

β-Galactosidase activity assays show that lsrR and lsrACDBFG expression was not affected by the deletion of sdiA or ydiV, while sdiA-ydiV double mutation resulted in lsrR repression (>2.3-fold) and lsrACDBFG repression (>4.8-fold) (Figure 3A and 3B). As repression of lsrACDBFG can further affect AI-2 internalization and the subsequent cleavage of phospho-AI-2, we speculated that transport of AI-2 might also be affected in the sdiA-ydiV double mutant. We assayed the extracellular AI-2 profiles of WT and the mutants. Consistent with our results described above, the sdiA or ydiV mutant was indistinguishable from WT in its ability to produce and import AI-2, while the sdiA-ydiV double mutant showed much slower removal of AI-2 from extracellular fluids compared with that of the other three genotypes when grown in LB at 30 °C (Figure 3C). These results suggest that ydiV plays an important role in the interaction between QS system 1 and system 2, and QS system 2 is repressed by sdiA-ydiV double mutation.

Figure 3
figure 3

QS system 2 is inhibited in the sdiA-ydiV double mutant. (A and B) The expression of lsrR and the lsr operon is repressed in the sdiA-ydiV double mutant. E. coli ZK126 (WT), sdiA, ydiV, and sdiA-ydiV double mutant carrying plasmid pPlsrR (lsrR-lacZ) (A) and pPlsrA (lsr-lacZ) (B) were grown in LB. At different time points during cell growth, aliquots were collected for measurement of the OD600 (squares, circles, and triangles) and β-galactosidase activity (bars). (C) Extracellular AI-2 activity profiles. Overnight cultures of E. coli ZK126 (WT), sdiA, ydiV, and sdiA-ydiV double mutant were diluted 100-fold into fresh LB medium. At different time points during cell growth, aliquots were collected for measurement of the OD600 (open symbols) and AI-2 activity (solid symbols) as described in Materials and Methods.

Full size image

QS system 2 inhibition is caused by decreased intracellular cAMP concentration in the sdiA-ydiV double mutant

In E. coli, AI-2 is synthesized by LuxS and released into the extracellular environment. AI-2 is bound by the periplasmic protein LsrB and internalized by the Lsr transporter (LsrACDB). Following internalization, AI-2 is phosphorylated by LsrK; then, phospho-AI-2 binds specifically to LsrR and antagonizes it. LsrF and LsrG are required for further processing of phospho-AI-2 14, 17. The sdiA-ydiV double mutation resulted in QS system 2 inhibition, raising the question of whether this is due to the modulation of a specific factor involved in regulating both lsrR and the lsr operon. The cAMP-CRP complex stimulates the expression of lsrR and the lsr operon, while LsrR represses their expression 14, 15, suggesting that cAMP, the CRP, or both may be affected by sdiA-ydiV double mutation.

To test this hypothesis, β-galactosidase activity in WT, sdiA mutant, ydiV mutant and sdiA-ydiV double mutant carrying plasmid pPcrp (crp-lacZ) was determined. The expression of crp was not affected (data not shown). Intracellular cAMP concentration in these strains with different genotypes was also determined. A decreased intracellular cAMP concentration (about 2-fold, P < 0.05) was observed in the sdiA-ydiV double mutant, while the intracellular cAMP concentration in the sdiA mutant or the ydiV mutant was not significantly different from that in the WT strain (Figure 4A). If the decreased intracellular cAMP concentration in the sdiA-ydiV double mutant caused inhibition of QS system 2, the addition of cAMP should offset the sdiA-ydiV double mutation effects on lsrR and lsrACDBFG expression. To further corroborate our findings, the expression of lsrR-lacZ and lsr-lacZ in WT and the mutants was measured after 12 h of growth in LB or LB plus cAMP as described for the experiments shown in Figure 3A and 3B. As shown in Figure 4B, the addition of 10 mM cAMP offset the sdiA-ydiV double mutation effects on lsrR and lsrACDBFG expression. These data allowed us to conclude that sdiA and ydiV play a synergistic role in the regulation of intracellular cAMP, and a decrease in the intracellular cAMP concentration caused by the sdiA-ydiV double mutation leads to down-regulation of lsrR and the lsr operon. Hence, cAMP and ydiV are crucial in connecting QS system 1 and QS system 2 in E. coli.

Figure 4
figure 4

QS system 2 inhibition is caused by a decreased intracellular cAMP concentration in the sdiA-ydiV double mutant. (A) Intracellular cAMP concentration determination. Intracellular cAMP concentrations were measured by the cAMP enzyme immunoassay system and expressed in femtomoles per milligram (wet weight) of cells. The P value for the sdiA-ydiV double mutation was < 0.05. (B) The sdiA-ydiV double mutation effects on lsrR and lsr operon expression are offset by the addition of cAMP. Strains were grown in LB or LB plus 10 mM cAMP. β-Galactosidase activity in strains carrying plasmid pPlsrA (lsr-lacZ) or pPlsrR (lsrR-lacZ) was measured after 12 h of growth.

Full size image

Discussion

Previous studies have shown that an unidentified extracellular factor in E. coli conditioned medium down-regulated expression of the sdiA gene that normally functions to stimulate ftsQP2 expression 33. The addition of indole up-regulated expression of sdiA (2.9-fold) and led to a decrease in ftsQP2 expression 10, so further investigation is needed to determine whether the unidentified extracellular factor, which was shown to regulate sdiA and to regulate the ftsQ2p cell division promoter via SdiA, is indole. The sdiA gene was not affected by luxS deletion in LB with or without glucose 15, and our study showed that deletion of lsrR did not affect sdiA expression in LB (data not shown). Thus, the extracellular factor is unlikely to be AI-2, and AI-2 does not signal to SdiA in these experimental conditions.

A previous study showed that S. typhimurium SdiA, which is 69% identical at the amino-acid level to its E. coli homolog, could partially complement the E. coli sdiA gene for activation of ftsQAZ at promoter 2 and for suppression of filamentation caused by an ftsZ(Ts) allele 34. In S. typhimurium, overexpressed plasmid-based sdiA, but not chromosomal sdiA, can activate four genes, srgA, srgB, rck, and srgC 34. Microarray analysis revealed that more than 100 genes are regulated by sdiA overexpression, and the expression of ydiV was elevated with increased sdiA dosage 12. However, the genes identified with the microarray were not tested for a response to chromosomal sdiA or AHL. In the present study, we found that sdiA deletion lowers the level of ydiV expression, and AHL from V. fischeri further stimulates ydiV expression in WT but not in the sdiA mutant. Our results also show that, under microaerobic conditions, the fold-difference for ydiV expression in WT versus sdiA mutant is higher than that under aerobic conditions. This result suggests that there might be an interaction between respiratory control and ydiV expression. E. coli sdiA plays a more important role under microaerobic conditions and may be used to detect the transition from a free-living state to the intestinal environment 34.

As shown by the gel shift assay, we found that the addition of AHL to SdiA-binding buffer caused more intense binding of SdiA to the ydiV promoter region. The addition of AHL to SdiA-binding buffer did not cause more intense binding of the AHL-SdiA complex to the ydiV promoter region because the SdiA binding pocket had been occupied by AHL during induction. The gel shift with the AHL-SdiA complex was more intense than that with SdiA. These results suggest that AHL supports SdiA binding to the ydiV promoter. Our results further corroborate the previously stated hypothesis that SdiA has very low levels of activity in the absence of AHL and that the addition of AHL leads to a more stable folded protein 5, 8, 35. So far, the binding specificity of SdiA is still controversial. The SdiA binding site within the ftsQP2 promoter, which is designated as the SdiA box, was identified previously 11. However, this SdiA box is not present in the promoters of any SdiA-regulated genes of Salmonella, including the Salmonella srgE-lux fusion that was placed into E. coli K-12 and was activated in an sdiA-dependent and AHL-dependent manner 5. A DNA fragment (73 bp from +112 to +184 with respect to the ftsQP2 initiation site) formed complexes with SdiA as shown by the gel shift assay, but clear protection was not observed in this region as shown by DNase I footprinting, and there was no homolog of SdiA box present in this DNA fragment 11. To assess nonspecific binding, we used both the mutated ydiV promoter fragment (M-ydiV) and a control DNA (S-con) that is devoid of an SdiA box. Our results show that the AHL-SdiA complex, as well as SdiA (data not shown), can bind to the ydiV promoter fragments (S-ydiV), the ydiV promoter fragments containing base-substitutions in the SdiA box (M-ydiV), the ftsQP2 fragments (S-ftsQ) as a positive control, and a 44 bp DNA fragment of the multicloning site of pBluescript II KS+ (S-con) (Figure 2A). The negative control DNA (S-con) displayed reduced binding as compared to ydiV promoter fragments (Figure 2A). The results from the SDS-PAGE analysis indicate that there were no protein contaminants in the SdiA and AHL-SdiA complex preparations (Figure 2C). As shown in Figure 2A and 2B, the gel shift is very weak even with 7.5 pmol of SdiA, a result that is consistent with a previous study 11. There is no apparent difference between the specific ydiV promoter and the mutated ydiV promoter when the gel shift assay was conducted with 7–28 pmol of SdiA (data not shown). Our results show that, whether or not AHL is present, SdiA binds DNA nonspecifically, at least in vitro in our experimental conditions. Because the action of SdiA is pleiotropic, one would expect that other physicochemical factors or proteins might be involved in the control of binding specificity of SdiA in vivo.

Although the evidence for indirect interplay between QS and c-di-GMP signaling is accumulating, a direct connection, such as the activation of a c-di-GMP synthetase or phosphodiesterase by an autoinducer, has yet to be reported 36, 37, 38. Besides YdiV (EAL), a GGDEF domain protein YedQ was also up-regulated with increased sdiA dosage 12. These results indicate that SdiA might play an important role in GGDEF/EAL-related cellular processes. Lee et al. 10 recently reported that AHLs inhibited E. coli biofilm formation through SdiA in a dose-dependent manner and that sdiA mutation caused a 51-fold increase in biofilm formation. Since biofilm formation is suppressed upon overproduction of EAL domain proteins 25, 39, the SdiA-dependent expression of ydiV (EAL) might also contribute to the SdiA-mediated inhibition of biofilm formation in E. coli.

Interestingly, E. coli has 19 GGDEF and 17 EAL domain proteins, with an overlap of seven proteins containing both domains. Temporal or spatial regulatory mechanisms might be employed by a single GGDEF or EAL protein to significantly affect a certain function 40. Although the function of these 29 GGDEF/EAL domain proteins in E. coli is poorly understood, it is obvious that they play a key role in c-di-GMP signaling 40, 41. So far, it is not clear whether YdiV in E. coli acts as a phosphodiesterase or just plays a regulatory role. Here, we have shown that YdiV is tightly connected with QS systems and cAMP signaling in E. coli.

The cAMP-CRP complex stimulates expression of both lsrR and the lsr operon of E. coli, which are divergently transcribed, by directly binding to the two different CRP binding sites in the intergenic region 14, 15. CRP adopts an active conformation only when it binds cAMP, and the concentration of the active cAMP-CRP complex is influenced by the intracellular concentration of cAMP 42. In the present study, we found that the intracellular cAMP concentration, but not the expression of crp, was decreased by about 2-fold in the sdiA-ydiV mutant, and the decreased cAMP concentration subsequently led to inhibition of expression of lsrR and the lsr operon. In a later trial, the addition of cAMP offset the sdiA-ydiV double mutation effects on lsrR and lsrACDBFG expression. These results raise the question of how the decreased intracellular concentration of cAMP could be caused by the sdiA-ydiV double mutation.

E. coli strains employ a complex mechanism to control intracellular cAMP concentration, and, so far, this process has not been completely elucidated. The intracellular cAMP concentration in E. coli can be controlled by its own synthesis and hydrolysis 42. cAMP is synthesized by adenylate cyclase (AC) which is encoded by the cyaA gene, and the activity of AC is regulated transcriptionally and post-translationally 42. The transcription of cyaA is negatively regulated by the cAMP-CRP complex; thus, a large increase in cAMP synthesis occurs in a crp mutant 43. However, the cellular levels of cAMP are regulated mainly by post-translationally modified IIAGlc 43, 44. The phosphorylated form of IIAGlc, of the phosphoenolpyruvate:sugar phosphotransferase system (PTS), binds to AC and stimulates its activity only in the presence of additional unknown factor(s) from E. coli extracts 44. The hydrolysis of cAMP by endogenous cAMP phosphodiesterases provides another way to affect the intracellular cAMP concentration. In E. coli, cAMP is hydrolyzed by CpdA 45. In the present study, we have discovered that the EAL domain protein encoded by ydiV is involved in the regulation of cAMP production and that it plays an important role in the interaction between QS system 1 and QS system 2 in E. coli. As shown in Figure 4A, the intracellular concentration of cAMP is decreased in the sdiA-ydiV mutant than that in the ydiV mutant, indicating that SdiA positively regulates cAMP production in the mutants with an ydiV deletion, as does YdiV in the mutants with an sdiA deletion. It is possible that SdiA and YdiV synergistically regulate the intracellular concentration of cAMP to control QS system 2 in E. coli. The factors involved in cAMP production, such as cyaA, IIAGlc, or CpdA, might be affected by sdiA-ydiV double mutation. Identification of the exact mechanism of decreased intracellular cAMP concentration in the sdiA-ydiV double mutant will be the subject of further studies.

In the present study, we investigated the environmental cues and regulatory pathway interactions that influence the E. coli expression profile and physiological activities. As shown in Figure 5, QS system 1 and QS system 2 in E. coli are linked via the EAL domain protein encoded by ydiV and cAMP, and in this process, unidentified X factor(s), possibly cyaA, IIAGlc, or CpdA, might be involved. When glucose is present in the growth medium, the expression of lsrR and the lsr operon is apparently repressed, and the level of luxS transcription increases 14, 15. Here, we found that SdiA, a LuxR homolog, and the EAL domain protein encoded by ydiV were also repressed at the transcriptional level by the addition of glucose and that the effect of glucose on ydiV expression was not mediated by SdiA. These results suggest that QS systems and YdiV in E. coli mainly function under specific growth conditions, especially during a shortage of glucose or other PTS sugars. The expression of ydiV is up-regulated by QS system 1, and it contributes to the control of intracellular cAMP concentration, which leads to the inhibition of QS system 2 in the sdiA-ydiV double mutant. Interestingly, lsrR and lsrB of E. coli are repressed (4- to 5-fold) under high levels of c-di-GMP in cells overexpressing yddV, a diguanylate cyclase 46. Thus, it is clear that E. coli cells have complex hierarchical regulatory systems and that these important signaling pathways, including QS systems, cAMP, and GGDEF/EAL domain proteins, are linked together for their control. Identification of the mechanisms employed by the sdiA-ydiV double mutant to control intracellular cAMP concentration will enable us to address the different roles of sdiA and ydiV during this process and to clarify further intracellular and intercellular signaling in E. coli.

Figure 5
figure 5

Conceptual model of signaling pathway interactions in E. coli. AI-2 is internalized by the Lsr transporter encoded by lsrACDB and phosphorylated by LsrK. Phospho-AI-2 binds specifically to LsrR and antagonizes it. Expression of lsrR and the lsr operon is inhibited by LsrR and stimulated by the cAMP-CRP complex. LsrF and LsrG are required for further processing of phospho-AI-2. Besides the inhibition effect on QS system 2, glucose represses the expression of sdiA and ydiV. The expression of ydiV is stimulated by the AHL-SdiA complex, and it contributes to the control of intracellular cAMP concentration, which leads to the inhibition of QS system 2. Regulatory effects are shown by solid (direct) or dashed lines (indirect or unclear mechanisms). Positive and negative regulatory effects are indicated by lines with triangular arrowheads and by lines with blunt ends, respectively.

Full size image

Materials and Methods

Bacterial strains, plasmids and media

The bacterial strains and plasmids used in this work are listed in Table 1. E. coli ZK126 was used as WT strain 47. Bacteria were grown in LB medium or on LB plates containing 1.5% agar under aerobic conditions unless otherwise indicated. The autoinducer bioassay (AB) and Luria-marine (LM) media are described in detail elsewhere 48. Unless otherwise stated, glucose, cAMP, and N-(3-oxohexanoyl)-homoserine lactone (oxoC6, which serves as a V. fischeri AHL signal molecule) were added at 0.8%, 10 mM, and 10 μM, respectively. AHL was added to media as dilutions from a 10 mM stock solution in ethyl acetate. The final concentration of ethyl acetate was 0.1% and had no effect on the growth or AHL response of the reporter. Ampicillin (Ap) and kanamycin (Kan) were used respectively at 100 and 50 μg/ml when necessary.

Table 1 Bacterial strains and plasmids used in this study
Full size table

Plasmid construction

The plasmids used in this study are listed in Table 1 and standard procedures were employed 49. Genes and promoters were prepared by PCR using E. coli ZK126 chromosomal DNA as template. Extractions of DNA from agarose gels were performed using Wizard® PCR Preps DNA Purification System (Promega, Madison, WI). Primers used in this study are listed in Table 2. All constructs made by PCR were sequenced to verify their integrity.

Table 2 Oligonucleotides used in this study
Full size table

Plasmid pFZY1, which was used to construct promoter fusion vectors in this study, is a low-copy-number vector (one to two copies per cell) with the multiple restriction site linker of M13mp18 inserted upstream from a promoterless galK′-lacZYA reporter segment 50. To create pPydiV (ydiV-lacZ), pPsdiA (sdiA-lacZ), and pPcrp (crp-lacZ), the promoter regions were amplified by PCR using the primers P-ydiV-F/P-ydiV-R, P-sdiA-F/P-sdiA-R, and P-crp-F/P-crp-R, respectively. The purified PCR products were digested with appropriate restriction enzymes, as shown in Table 2, and ligated into digested pFZY1. Plasmids pPlsrR (lsrR-lacZ) and pPlsrA (lsr-lacZ) were constructed by amplification of the same promoter region using primers P-lsr-R and P-lsr-A followed by purification, digestion with EcoRI and insertion into EcoRI-digested and dephosphorylated pFZY1, while the transcriptional direction was determined by PCR using primers P-lsr-R (or P-lsr-A) and Mp18-R that anneals to pFZY1.

To express a His-tagged version of SdiA, pGsdiA was constructed. DNA fragments containing the sdiA coding region were amplified by PCR using primers G-sdiA-F/G-sdiA-R. After digestion with the appropriate restriction enzymes as shown in Table 2, the amplified DNA fragments were cloned into the corresponding sites of pET-28a(+). pGsdiA expresses SdiA with a N-terminal hexahistidine tag.

Gene disruption

Chromosomal deletions of sdiA, ydiV, and lsrR were constructed by methods described previously 51. We used the phage l Red recombination system to replace sdiA with an sdiA::Kan fragment, which was amplified by PCR from the plasmid pKD4 using primers D-sdiA-F/D-sdiA-R (Table 2). The resulting PCR product was treated with DpnI, purified and introduced by electroporation into E. coli ZK126 carrying the Red recombinase expression plasmid pKD46, which was cured later by growth at 42 °C. Recombinants were selected on LB supplemented with kanamycin. Deletions of ydiV and lsrR were constructed similarly by PCR amplification of pKD4 with primers D-ydiV-F/D-ydiV-R and D-lsrR-F/D-lsrR-R, respectively. To construct strain SCH604S, the KanR gene of strain SCH603 was eliminated using the previously described method, and pCP20 was employed 51. Strain SCH604SB was constructed similarly by introducing an ydiV::Kan fragment into E. coli SCH604S containing plasmid pKD46. All of the gene deletions were verified by PCR tests.

Protein purification

His-tagged SdiA was prepared as described previously with some modifications 11, 35. Transformants of E. coli BL21(DE3) containing pGsdiA were grown in LB to an optical density of 0.6 at 600 nm, induced with 0.8 mM IPTG, or induced with 0.8 mM IPTG as well as supplied with 100 μM AHL to form AHL-SdiA complex during induction with IPTG and then incubated for an additional 3 h. All subsequent procedures were performed at 4 °C. The cultures were harvested and resuspended in 40 ml of buffer A (20 mM Tris-HCl, pH 8.0, at 4 °C; 500 mM NaCl; 5 mM 2-mercaptoethanol and 5% glycerol). After the addition of PMSF (1 mM) and lysozyme (0.1 mg/ml), the cells were incubated on ice for 30 min and then lysed by sonication. After centrifugation at 16 000 rpm for 30 min, the supernatant was applied to a nickel-NTA column and eluted with a gradient of 20-250 mM imidazole in buffer A. Eluted fractions were subjected to SDS-PAGE analysis. Fractions containing the overexpressed His-tagged SdiA were pooled and dialyzed against SdiA-dialysis buffer (20 mM Tris-HCl, pH 7.6, at 4 °C; 200 mM KCl; 10 mM MgCl2; 0.1 mM EDTA; 1 mM DTT and 50% glycerol) and stored at −80 °C until use. The purity of SdiA and AHL-SdiA complex was analyzed by SDS-PAGE.

Gel shift assay

Double-stranded DNA fragments including the ydiV promoter fragments (S-ydiV, S-ydiV-F/S-ydiV-R), the ydiV promoter fragments containing base-substitutions in the SdiA box (M-ydiV, M-ydiV-F/M-ydiV-R), the ftsQ promoter fragments (S-ftsQ, S-ftsQ-F/S-ftsQ-R), and a 44 bp DNA fragment of the multicloning site of pBluescript II KS+ (S-con, S-con-F/S-con-R) were produced by boiling and slowly cooling synthetic DNA oligonucleotides (Table 2). The DIG gel shift kit (Roche Ltd., Mannheim, Germany) was used for the labeling of DNA fragments and the detection of signals according to the manufacturer's instructions. Band intensities were analyzed with Stratagene EagleSight software.

The labeled DNA fragments (75 fmol) were incubated with various amounts of purified SdiA or AHL-SdiA complex at 4 ºC for 30 min in 10 μl of SdiA-binding buffer (20 mM Tris-HCl, pH 8.0, at 4 °C; 50 mM NaCl; 3 mM magnesium acetate; 0.1 mM EDTA; 0.1 mM DTT and 25 μg ml−1 BSA) 11. The DNA-protein complexes that formed were separated by 8% PAGE in 0.5× TBE buffer.

Measurement of intracellular cAMP concentrations

Overnight cultures of E. coli were diluted 100-fold into fresh LB medium. Cells were incubated at 30 °C with shaking at 200 rpm for 21 h. Two milliliters of cultures were removed and centrifuged. Cells were suspended in 200 μl of 0.1 N HCl and heated at 100 °C for 10 min. Each sample was then centrifuged at 12 000 rpm for 10 min, and the supernatant fluid was adjusted to pH 7.0 with 0.1 N NaOH. The supernatant fluid was again centrifuged to remove precipitated material. Intracellular cAMP was then measured using the cAMP EIA kit (Assay designs, Ann Arbor, MI). The average intracellular cAMP concentration was expressed in femtomoles per milligram (wet weight) of cells. All assays were performed in duplicate cultures.

Extracellular AI-2 activity assays

Overnight cultures of E. coli were diluted 100-fold into fresh LB medium. Cells were incubated at 30 °C with shaking at 200 rpm in flasks. Two milliliters of cultures were removed at regular intervals for the determination of the OD600 and the preparation of cell-free fluids. Cell-free culture fluids were prepared by centrifugation of the cultures at 12 000 rpm for 2 min, followed by filtering cleared supernatants through sterilized 0.22-μm-pore filters, and the resulting cell-free fluids were stored at −20 °C.

The AI-2 activity in cell-free E. coli fluids was measured using the V. harveyi strain BB170 bioluminescence reporter assay, as described previously 14, 16, 48. Strain BB170 was grown in AB medium overnight and diluted 1:5 000 into fresh AB medium. Parallel 96-well microtiter plates containing 180 μl of diluted BB170 cultures and 20 μl of cell-free fluids were incubated at 30 °C with shaking. Measurements of luminescence were conducted hourly with a luminometer, while strain BB170 cell density was determined by spreading dilutions onto solid LM medium and counting colonies after overnight growth at 30 °C. Relative light units (RLU) were defined as light emission [counts min−1 ml−1 × 103/CFU ml−1]. AI-2 activity is reported as fold induction of RLU over the background obtained when LB medium instead of cell-free culture was added.

β-Galactosidase assays

Overnight cultures of E. coli were diluted 1:100 into fresh LB medium. The cultures were incubated at 30 °C with shaking at 200 rpm. The microaerobic environment used for bacterial propagation was created by using 40 ml volumes of LB broth in 60-ml serum bottles capped with butyl rubber stoppers, and the cultures were incubated statically at 30 °C. At different time points during cell growth, aliquots were removed for the determination of the OD600, and β-galactosidase assays were performed as described previously 52. β-Galactosidase activity is expressed in Miller units. All assays were performed in triplicate. The error bars in the graphs indicate the standard deviations.