SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

In Gram-negative bacteria, N-acylhomoserine lactone (HSL) is used as a signal in cell–cell communication and quorum sensing (QS). The model prokaryote Escherichia coli lacks the system of HSL synthesis, but is capable of monitoring HSL signals in environment. Transcription factor SdiA for cell division control is believed to play a role as a HSL sensor. Using a collection of 477 species of chemically synthesized HSL analogues, we identified three synthetic signal molecules (SSMs) that bind in vitro to purified SdiA. After SELEX-chip screening of SdiA-binding DNA sequences, a striking difference was found between these SSMs in the pattern of regulation target genes on the E. coli genome. Based on Northern blot analysis in vivo, a set of target genes were found to be repressed by SdiA in the absence of effectors and derepressed by the addition of SSMs. Another set of genes were, however, expressed in the absence of effector ligands but repressed by the addition of SSMs. Taken together, we propose that the spectrum of taget gene selection by SdiA is modulated in multiple modes depending on the interacting HSL-like signal molecules.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

Quorum sensing (QS) is a bacterial cell–cell communication process by which bacteria communicate using extracellular signals called autoinducers (AIs). In the Gram-negative bacteria, three QS systems have been identified, which differ in the molecular species of QS signals. The best characterized QS signal is N-acylhomoserine lactone (HSL), referred to AI-1, which is involved as the language for intraspecies communication and in monitoring the own population density (Fuqua et al. 1996). The LuxI-family proteins synthesize these HSL signal molecules, whereas the LuxR-family proteins play a role in monitoring HSL signals (Waters & Bassler 2005). A variety of HSLs have been identified, each differing in length, oxidation, and saturation of the acyl chain. Each bacterium carries a pair of HSL synthetase and HSL sensor. For population density-dependent expression of luminescence, Vibrio fischeri LuxI produces AI-1 with the structure of N-3-oxohexanoy HSL (oxoC6-HSL) (Eberhard et al. 1981), whereas Vibrio harveyi produces AI-1 with the structure of N-butyryl HSL (C4 HSL) by LuxLM (Cao & Meighen 1989). Pseudomonas aeruginosa contains two pairs of HSL synthetase and HSL sensor, LasI/LasR, and RhlI/RhlR. LasI synthesizes N-(3-oxo-dodecanoyl)-L-HSL (oxoC12-HSL) (Pearson et al. 1994), whereas RhlI synthesizes N-butyryl-L-HSL (C4-HSL) (Pearson et al. 1995, 1997). The HSL sensors are considered to play roles in regulation of a wide range of environmental signal-response genes in a cell density-dependent manner.

Later another group of QS signals, referred to AI-2, was identified as the interspecies communication signal in a broad range of bacterial species (Bassler 2002). AI-2 is a furanosyl borate diester derived from cyclization of 4,5-dihydroxy-2,3-pentanedione (DPD), which is made from S-ribosylhomocysteine, an intermediate in the breakdown of S-adenosylhomocysteine (Schauder et al. 2001). Enterohemorrhagic E. coli (EHEC) is the agent responsible for outbreaks of bloody diarrhea. EHEC uses a QS regulatory system to sense the external condition within the host intestine and to activate the genes essential for intestinal colonization. EHEC produces another ill-characterized QS signal AI-3, which is used, together with the host-derived epinephrine, for cross-communication between E. coli and host animals. AI-3 is involved in the interkingdom communication with host animals (Sperandio et al. 2003). Indole was also predicted to be involved in cell–cell communication. Indole production is one of the unique characteristics of E. coli and is used as a diagnostic maker for distinguishing it from other enteric bacteria. Indole generated by E. coli inhibits QS system in P. aeruginosa, and E. coli mutants defective in indole production are not competitive in coculture with P. aeruginosa (Chu et al. 2012).

Bacterial species of the genera Escherichia and Salmonella are unique in that they have a LuxR homologue, SdiA, but they do not contain a LuxI homologue or any other enzyme family that can synthesize HSLs (Ahmer 2004). Because of the lack of LuxI-type enzyme for HSL synthesis, E. coli is considered to sense an as yet unidentified HSL-like signal produced by other bacteria in environment using SdiA sensor (Michael et al. 2001). SdiA was identified as the ‘suppressor of the cell division inhibitor’ that controls the transcription of the ftsQAZ operon involved in cell division (Wang et al. 1991; Garcia-Lara et al. 1996; Sitnikov et al. 1996; Weiss 2004). SdiA is composed of two domains, N-terminal effector-binding domain and C-terminal DNA-binding domain, and is considered to play as a transcription factor. Its crystal structure was solved using crystals that had a hexagonal space (Wu et al. 2008), and its direct interaction with HSL was proposed from modeling studies (Yao et al. 2006). Using the newly developed PS-TF (promoter-specific transcription factor) screening system, we have identified as many as 15 regulators, each monitoring a specific environmental factor or condition, are involved in regulation of the sdiA promoter (Shimada et al. 2013). This finding agrees with the observations that SdiA is involved in the variety of phenotypes including biofilm formation, motility, antibiotic sensitivity, and virulence expression (Eberl 1999; Wei et al. 2001b; Dyszel et al. 2010). Up to the present time, however, the only known regulation target of SdiA transcription factor is the ftsQAZ operon (Sitnikov et al. 1996; Yamamoto et al. 2001), which encodes the division assembly protein FtsQ, the protein FtsA for recruitment of FtsZ to Z-ring, and the tubulin-like cell division protein FtsZ, respectively (Hale & de Boer 1997).

In this study, we attempted to identify the full range of regulation targets of SdiA in E. coli. The most common signal-transduction mechanisms in E. coli are the one-component systems, in which a single polypeptide contains both an effector-binding sensory domain and a DNA-binding domain (Ishihama 2010, 2012). As the natural effector ligand for SdiA has not been identified, we first tried to identify an HSL-like effector ligand(s). For this purpose, we searched for effectors, which are recognized in vitro by purified SdiA, from a collection of 477 species of the chemically synthesized HSL analogues (Smith et al. 2003a; Igarashi & Suga 2011). This library was used for the selection of effectors that influence cell–cell communications in vivo of P. aeruginosa (Smith et al. 2003b; Chung et al. 2011). Here, we identify three ‘synthetic signal molecules’ (SSMs) that affect the binding in vitro of purified SdiA to the ftsQ promoter, the only known regulation target of SdiA. The influence of these three SSMs on the recognition pattern of SdiA on the E. coli genome was then analyzed using the improved genomic SELEX screening system (Shimada et al. 2005). As the recognition target genes of SdiA were found to be different depending on the effector species, transcription in vivo of the regulation target genes was examined in the presence and absence of the SSMs. Taken all the results together, we propose that SdiA is a unique transcription factor in E. coli, because SdiA interacts with multiple species of effector ligands and its promoter recognition specificity is modulated to various directions depending on the interacting effectors.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

Search for effector molecules affecting SdiA functions

Laboratory strains of E. coli do not synthesize HSL-type signals, the widely used QS signals in Gram-negative bacteria, but are capable of monitoring QS signals in environment supposedly produced by other bacteria. To identity HSL-type chemicals recognized by SdiA, we used a library of HSL analogues, which were chemically synthesized with the procedure of Igarashi and Suga (2011) (see Fig. S1 in supporting information for details of the library). In this study, we examined a total of 477 species from three libraries of HSL analogues: C4 library (96 different amines acylated with the C4 chain) (Fig. S1A in supporting information); C12 library (96 different amines acylated with the 3-oxoC12 HSL or C12 chain) (Fig. S1B in supporting information); and a total of 255 species of 5KH-13 library (52 lactone ring variants, each joined with the acyl chains of 4, 6, 8, 10, and 12 in length (Fig. S1C in supporting information).

The DNA-binding affinity and specificity of transcription factors often change after interaction with specific effector molecules, so-called inducers or corepressors. As an attempt for screening chemicals that affect SdiA functions, we tested the DNA-binding activity in vitro of purified SdiA to the only known target, ftsQAZ promoter. When 0.5 pmol of 300-bp-long fluorescent-labeled ftsQ promoter fragment was mixed with the increasing amounts (1–100 pmol) of SdiA for 15 min at 37 °C and subjected to PAGE, the fluorescent probe shifted to slowly migrating a smear band at the SdiA concentration above 15 pmol, indicating the formation of SdiA-ftsQ promoter complexes (Table 1A). We then examined possible influence of all the chemically synthesized HSL analogues on the gel-shift pattern of SdiA-promoter complexes. For this purpose, 5 pmol each of HSL analogues was added into the reaction mixture containing 0.5 pmol ftsQ promoter probe and a fixed amount (10 pmol) of SdiA protein. After 15-min incubation at 37 °C, the mixtures were directly subjected to gel-shift assay. Among a total of 477 species of HSL analogues tested, a significant difference in the gel-shift pattern was reproducibly observed for three HSL analogues, A4-30C12 (afterward referred to A), F12S-C12 (F), and K12-C12 (K) (Fig. 1). These HSL analogues with the binding activity to ftsQ promoter were hereafter referred to ‘synthetic signal molecules’ (SSMs). The smear gel pattern of SdiA-ftsQ promoter complexes implies either a weak affinity of SdiA to the ftsQ DNA probe or a conformational heterogeneity of SdiA-probe complexes. The SSM dose-depending increase in SdiA binding to the ftsQ DNA probe was observed for all three SSMs (Table 1B).

Table 1. Levels of SdiA-ftsQ promoter complex formation
SdiA (pmol)ftsQ promoter (pmol)SdiA-ftsQ promoter complex (%)
(A) In the absence of effectors
100.5<5
200.515
500.545
1000.5100
SdiA (pmol)ftsQ promoter (pmol)HSL analogues (pmol)SdiA-ftsQ promoter complex
HSLSSM-ASSM-FSSM-K
(B) In the presence of effectors
100.50.0110   
100.50.0220   
100.50.0550   
100.50.01 30  
100.50.02 35  
100.50.05 85  
100.50.01  40 
100.50.02  55 
100.50.05  90 
100.50.01   35
100.50.02   45
100.50.05   75
SdiA (pmol)ftsQ promoter (pmol)SSM-A SSM-F SSM-K (pmol)SdiA-ftsQ promoter complex
SSM-ASSM-FSSM-K
  1. The binding of SdiA protein to the fluorescent-labeled ftsQ promoter probe was examined by PAGE as described in Materials and Methods. The level of SdiA-promoter complex formation was determined by measuring both the complex peaks and the unbound free probe. The fluctuation level of each measurement was approximately 10%.

(C) In the presence of two effectors
100.50.1001000
100.50.100.05805
100.50.100.106010
100.50.100.203030
100.50.100.401060
100.50.1001000
100.50.100.0585<5
100.50.100.10705
100.50.100.205010
100.50.100.4010035
image

Figure 1. HSL-like chemicals that interact SdiA. Screening of SdiA-binding chemicals was carried out by gel-shift assay using a collection of 477 species of chemically synthesized HSL-like molecules. Three synthetic signal molecules (SSMs) were identified to bind to SdiA: (A) A4-3OC12 (referred to A); (B) F12S-C12 (referred to F); and (C) K12-C12 (referred to K).

Download figure to PowerPoint

In the presence of A, the SdiA-ftsQ promoter complex (SdiA-DNA complex-1) was observed apparently at the same position with that formed in the absence of SSMs (Fig. 2A), but the fluorescent intensity increased markedly (Table 1B), indicating that the affinity to the target promoter increased in the presence of A. Moreover, the set of DNA sequences recognized by SdiA-A complex was completely different with that by SdiA alone (see below). In the presence of authentic HSL (C4-HSL produced by Vibrio harveyi and Pseudomonas aeruginosa), the SdiA-HSL complex also moved to the same position with complex-1, and the affinity of SdiA to the ftsQ promoter probe increased significantly as in the case of effector A (Table 1B), suggesting that the influence of A on the structure of SdiA is similar to that of the authentic C4-HSL.

image

Figure 2. Influence of HSL chemicals on the DNA-binding activity of SdiA. The binding activity of SdiA to a promoter fragment of ftsQ, the only known target of SdiA, was examined by gel-shift assay under the conditions described in 'Experimental procedures' in the presence of 0.5 pmol fluorescent-labeled ftsQ promoter fragment and 10 pmol SdiA. In the first three panels, the binding of SdiA on to the ftsQ probe was examined in the presence of signal species of SSMs: (A) A4-30C12 (referred to A); (B) F12S-C12 (F); and (C) K12-C12 (K). The arrows indicate C1, C2, and C3 complexes formed in the presence of SSM-A, SSM-F and SSM-K, respectively. The gel-shift assay was repeated five times using different batches of purified SdiA with or without His-tag, but the SdiA-SSM complexes always formed smear bands. In gel-shift assays, the influence of sequential addition of two species of SSM on the formation of SdiA-ftsQ promoter complex was examined: (D) influence of SSM-F addition on C1 (SdiA-A) complex or influence of SSM-A addition on C2 (SdiA-F) complex and (E) influence of SSM-K addition on C1 (SdiA-A) complex or influence of SSM-A addition on C3 (SdiA-K) complex. Details are described in Table 1.

Download figure to PowerPoint

In contrast, the SdiA-ftsQ promoter complex formed in the presence of F migrated on PAGE to a different position (SdiA-DNA complex-2) (Fig. 2B). Likewise, the SdiA-ftsQ promoter complex formed in the presence of K moved to a different position (SdiA-DNA complex-3) (Fig. 2C). These findings altogether indicate the effector ligand-induced alteration in the form of SdiA-ftsQ promoter complex: the complex-1 (C1) in the presence of HSL and A; the complex-2 (C2) in the presence of F; and the complex-3 (C3) in the presence of K. The difference in mobility on PAGE might be due to the difference in the affinity of SdiA with different HSL chemicals or the difference in the structure of HSL-associated SdiA-DNA complexes.

After formation of C1 complex in the presence of A, it was converted into C2 by adding excess amounts of F (Fig. 2D). In contrast, the C2 complex was converted into C1 complex after subsequent addition of A. Likewise, the C1 complex was converted into C3 by adding excess amount of K, whereas the preformed C3 complex was converted into C1 by adding effector A (Fig. 2E). This finding suggests that all three SSMs (A, F, and K) bind to the same or over-lapping sites on the effector-binding N-terminal domain of SdiA protein. As an attempt to estimate the affinity difference between three SSMs, we next determined the SSM concentration-dependent conversion of SdiA-ftsQ promoter complexes. The conversion of C1 complex to C2 increased concomitant with the increase in SSM-F addition (Table 1C). Likewise, the conversion of C1 to C3 complex increased with the increase in SSM-K level (Table 1C). The concentration needed for conversion of the C1 complex was lower for SSM-F than SSM-K, indicating that the higher affinity to SdiA is higher for SSM-F than SSM-K.

The mobility shift of SdiA-ftsQ promoter complex was not observed in the presence of some well-known low-molecular-weight signal molecules such as cAMP, c-di-GMP, indole, and AI-2 (data not shown), furthermore supporting the conclusion that the binding of SSMs to SdiA is specific. Next possible influence of these SSMs on the DNA-binding specificity of SdiA was then examined by the genomic SELEX screening.

Genomic SELEX screening of SdiA-binding sequences

For the identification of genome DNA sequences that are recognized by E. coli SdiA, we employed the genomic SELEX screening system (Shimada et al. 2005), in which a library of E. coli genome DNA fragments of 200–300 bp in length was used for screening of DNA sequences with SdiA-binding activity instead of synthetic oligonucleotides with all possible sequences used in the original SELEX method (Ellington & Szostak 1990; Tuerk & Gold 1990; Singer et al. 1997). Into the mixture of E. coli genomic DNA fragments, fourfold molar excess of the purified His-tagged SdiA protein was added and the SdiA-DNA complexes were affinity purified. In the early stage of this genomic SELEX cycle, the SdiA-bound DNA fragments gave smear bands on PAGE as did the original genome fragment mixture. SdiA-associated DNA fragments were recovered and PCR amplified and subjected to next cycle of SELEX. After three SELEX cycles, DNA fragments with high affinity to SdiA were enriched, forming several discrete bands on PAGE.

For the identification of the whole set of SdiA-binding sequences, we then subjected the mixture of SELEX fragments to the SELEX-chip analysis. In this study, SELEX fragments were labeled with Cy5, whereas the original DNA library was labeled with Cy3. The mixture of fluorescent-labeled samples was hybridized to a DNA-tilling microarray (Oxford Gene Technology, Oxford, UK) (Shimada et al. 2008, 2011b; Teramoto et al. 2010). For elimination of the bias of library DNA, the ratio of fluorescence intensity bound to each probe between the test sample (Cy5) and the original library DNA (Cy3) was measured and plotted against the corresponding position on the E. coli genome (Fig. 3). On the DNA-tilling array used, the 60-bp-long probes are aligned along the E. coli genome at 105-bp intervals, and therefore approximately 300-bp-long SELEX fragments should bind to two or more consecutive probes. The height of SdiA-binding correlates with the affinity to the SdiA protein.

image

Figure 3. Genomic SELEX screening of SdiA-binding DNA sequences. Genomic SELEX screening of SdiA-binding sequences was carried out in the absence (A) or presence of HSL (B), SSM-A (C), SSM-F (D), and SSM-K (E). After the genomic SELEX, a collection of DNA fragments was subjected to SELEX-chip analysis using the tilling array of E. coli K-12 genome. The Y-axis represents the relative level of SdiA-bound DNA fragments, whereas the X-axis represents the position on the E. coli genome. The regulation targets were predicted based on the location of SdiA-binding sites. When SdiA-binding sites are located within spacer regions, only the flanking genes are indicated under light green background (for type-A spacers, the genes of both sides of bidirectional transcription units are indicated, whereas for type-B spacers, the genes to be transcribed are indicated). The genes carrying the SdiA-binding site inside the open-reading frames are indicated under light orange background. The set of major regulation targets of SdiA is described in Table S1 in Supporting information.

Download figure to PowerPoint

First, we carried out SELEX-chip analysis for SdiA alone in the absence of effector ligands. The location of SdiA-binding sites thus identified was classified into four groups: (i) type-A, intergenic spacers between bidirectional transcription units; (ii) type-B, intergenic spacers upstream of one transcription unit but downstream of another transcription unit; (iii) type-C, intergenic spacers located downstream of both transcription units; and (iv) type-D, inside open-reading frames. High-level peaks of SdiA binding are shown in Fig. 3A. The highest peak of SdiA binding was detected within type-A spacer between divergently transcribed yqjH and ygjI genes (69.27–69.29 min). YqjH is a FAD-containing NADPH-dependent ferric reductase that plays a role in the adaptation to iron starvation, whereas YqjI regulates expression of the yqjH gene together with the ferric uptake regulator Fur (Miethke et al. 2011; Wang et al. 2011). The binding of SdiA to this site was always detected in the absence and presence of SSMs albeit at different levels (see below). Besides the yqjH-yqjI operon, another type-A spacer between the yjiC and yjiD genes (98.14–98.18 min) was always detected in the absence and presence of SSMs. IraD, renamed from YjiD, interacts with RssB sRNA and inhibits proteolysis of RpoS sigma factor by ClpXP protease (Bougdour et al. 2008). In E. coli, transcription factors generally bind near the promoter region, and thus, we focused detailed analysis on the SdiA-binding sites located upstream of genes within 8 type-A and 10 type-B spacers.

By setting the cut-off level higher than 20% the level of the highest yqjH-yqjI peak, a total of 31 peaks were detected (Table 1A in supporting information), including the high-level peaks shown in Fig. 3A, of which 20 peaks were detected only in the presence of SdiA alone without HSL ligands. Among the total of 17 candidate genes (7 within type-A spacers and 10 within type-B spacers) under the control of SdiA alone, more than half encode the membrane-integrated transporters including yagG (sugar transporter), ydcS (polyamine transporter), gatA (galactitol PTS), mglC (methyl-galactoside transporter), ypjA (adhesion-like autotransporter), ascF (cellobiose/arbutin/salicin transporter), ygcS (transporter for unidentified substrate), yrbG (calcium/sodium:proton antiporter), and alsC (D-allose transporter). The genes coding for five transcription factors (YbeO, YgaV, AscG, YqjI, and GadW) are also included in this list, implying that a number of genes are indirectly controlled by SdiA. The ftsQAZ region formed a low-level peak (approximately 15% the highest peak at yqjH/yqjI) under the experimental conditions employed. Transcription of the ftsQAZ operon is also initiated together with the upstream mraZ operon consisting of 12 genes including the genes for cell division proteins, FtsL and FtsI. A low level of SdiA binding was also identified within this mraZ promoter region.

Influence of HSL analogues on the DNA recognition specificity of SdiA: in vitro SELEX-chip analyses

We next analyzed possible influence of the newly identified three SSMs: A, F, and K, as effectors on the DNA-binding activity and/or specificity of SdiA. For identification of the recognition targets of SdiA in the presence of each SSM, we again carried out the SELEX-chip analysis of SdiA-binding sequences in the presence of individual SSMs. In parallel, we also carried out SELEX-chip analysis of SdiA in the presence of authentic HSL. The set of high-level peaks of SdiA binding are shown in Fig. 3B–E (see also Table S1 in Supporting information). These SELEX-chip patterns are significantly different from that in the presence of SdiA alone (Fig. 3A), and moreover, a striking difference was observed between four SELEX-chip patterns in the presence of HSL and SSMs, indicating that each SSM modulates the DNA recognition properties of SdiA in a different mode.

In the presence of SSM-A, a total of 44 SdiA-binding sites were identified, of which 10 and 22 were located within type-A and type-B spacers, respectively (Table 1B in supporting information). In the presence of SSM-A, the pattern of SdiA-binding sites on the genome (Fig. 3B) was completely different from the SdiA pattern without the effector ligands (Fig. 3A). The list of 32 genes predicted to be under the control of SdiA-SSM-A complex includes a number of stress-response genes such as tig (peptidyl-prolyl cis/trans isomerase), rsfA (ribosome silencing factor), hchA (Hsp31 chaperone), yfcV (stress-response fimbriae), ypjA (stress-response adhesin), and sspA (stringent starvation protein). A number of the genes encoding cell surface components for cell–cell interaction are also included, such as hlyE (hemolysin E), yfcV (fimbrial-like adhesin), yqjH (siderophore interaction protein), ypfA (adhesion-like autotransporter), and fimA (major type-I fimbrin). Interestingly a high level of SdiA binding was detected in the promoter region of zipA encoding the Z-ring assembly factor, which initially interacts with FtsZ leading to assemble more then ten Z-ring components. The recognition pattern of SdiA-SSM-A complex (Fig. 3B) is similar to that in the presence of authentic HSL (Fig. 3E). Within the total of 44 peaks in the presence of SSM-A (Table S1B in supporting information) and of 39 peaks in the presence of authentic HSL (Table S1E in supporting information), overlapping was observed for 20 peaks. In good agreement with the similarity in DNA recognition patterns, the structure of SSM-A is similar to that of authentic HSL.

In the presence of SSM-F, a total of 26 SdiA-binding sites (5 sites within type-A spacers and 12 within type-B spacers [described in Table S1C in Supporting information]) with the peak level higher than 20% of the highest ygjH-ygjI peak were identified (Table S1C in supporting information), of which 7 were identified only in the presence of SSM-F. Noteworthy is that besides the set of stress-response genes, the genes related to translation were identified, including rsfA (ribosome silencing factor), glyW (tRNA-Gly), gltX (Glu-tRNA synthetase), and rrsG (16S rRNA). In the presence of SSM-K, a total of 37 SdiA-binding sites (14 sites within type-A spacers and 11 type-B spacers) were identified (Table S1D in Supporting information), of which 14 were detected only in the presence of SSM-K. As in the case of SSM-F, some genes related to translation appear to be under the control of SdiA, including thrS (Thr-tRNA synthetase), rrsG (16S rRNA), ileY (tRNA-Ile), rimP (ribosome maturation protein), rrsC (16S rRNA), and thrU (tRNA-Thr). These findings together suggest that upon expose to SSM-F and SSM-K, the DNA recognition properties of SdiA are modulated so as to control the genes for translation apparatus.

Taken together, we concluded that the target selection pattern of SdiA changes depending on the molecular species of HSL-like analogue. This finding indicates that SdiA is one of the unique transcription factors, because the activity and specificity of its regulatory function are under the control of multiple effector ligands. On the ftsQ promoter, we identified the SdiA-binding sequence of 29 bp in length, which contains a 18-bp-long core sequence, referred to SdiA box, with AAAA sequence at both ends, each being separated by a 10-bp-long spacer (Yamamoto et al. 2001). The SdiA-binding regions in the presence of HSL analogues contained SdiA box-like sequences with AT-rich segments, but the identification of SdiA-binding sites and of possible difference in the effector-dependent recognition sequences awaits foot-printing analyses.

Influence of HSL analogues on the DNA recognition specificity of SdiA: in vivo Northern blot analyses

To examine the influence of SSMs on the target selectivity of SdiA in vivo, we next carried out Northern blot analysis of mRNA levels for some representative target genes. Total RNA was extracted from E. coli grown in the absence and presence of each of three newly identified SSMs as well as the authentic HSL. On the basis of Northern blot pattern, the genes under the control of SdiA can be classified into two groups. The first group genes such as fimA (type-I bimbriae pilin), ypjA (adhesin-like autotransporer), zipA (cell division protein), and hchA (Hsp31 molecular chaperone) are transcribed in the absence of effector addition, but their mRNA levels decreased to various extent by the addition of SSMs (Fig. 4A). This group of genes must be expressed under the steady state of growing phase but repressed upon exposure to SSMs. In contrast, mRNA level of the second group genes including ybaO (uncharacterized DNA-binding transcription factor), yqiH (periplasmic pilin chaperone), acpP (acyl carrier protein for fatty acid biosynthesis), and rsfS (ribosomal silencing factor) was low in the absence of effectors but increased in the presence of HSL-like chemicals (Fig. 4B). This group of genes must be repressed under the normal growth conditions, but derepressed upon exposure to QS signals.

image

Figure 4. Influence of SdiA-binding SSMs on transcription of newly identified SdiA target genes. E. coli was grown in LB medium at 37 °C in the absence or presence of 0.01 mm each of three SSMs (A, F, and K) and authentic HSL. In the middle of exponential growth phase, total RNA was prepared by the hot ohenol method (Aiba et al. 1981) and subjected to Northern blot analysis using the indicated DIG-labeled probes. Group A indicates mRNAs that were detected in the absence of effector ligands, but decreased by the addition of SSMs and HSL, whereas group B indicates mRNA that were not detected in absence of ligands but increased by the addition of SSMs and HSL. Lane 1, no ligand addition; lane 2, SSM-A; lane 3, SSM-F; lane 4, SSM-K; lane 5, authentic HSL.

Download figure to PowerPoint

Two different modes of the regulation are generally in good agreement of the binding mode of SdiA as detected by SELEX-chip analysis. The level of SdiA binding in the absence of effector ligands is low for the first group targets that are expressed in the absence of SSMs, but high for the second group targets that are not expressed in the absence of SSMs (see Fig. 3). Taken together, we conclude that SdiA is a repressor for a set of target genes in the absence of effectors, but the expression of this group of genes is derepressed by the addition of SSMs; however, SdiA also plays as an activator for another set of genes but the expression of this group of genes becomes repressed after interaction with SSMs.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

Regulatory signals of SdiA function

Bacterial species of the genera Escherichia and Salmonella are unique in that they have a LuxR homologue, SdiA, but they do not contain a LuxI homologue or any other enzyme family that can synthesize HSLs (Ahmer 2004). SdiA was originally identified as a regulator of the ftsQAZ operon encoding the cell division apparatus, but several lines of evidence indicate the involvement of SdiA in control of various phenotypes such as biofilm formation and virulence expression (Eberl 1999; Wei et al. 2001b; Dyszel et al. 2010). The involvement of effector for control of the SdiA function was first indicated by the influence of a conditioned culture medium on the regulatory activity of SdiA (Garcia-Lara et al. 1996). Later, SdiA in Salmonella enterica serovar typhimurium was found to monitor AHLs produced by other bacterial genera (Michael et al. 2001; Smith & Ahmer 2003), and as a result, to activate the rck (resistance to complement killing) operon on a virulence plasmid and srg (SdiA-regulated gene) with unknown function (Ahmer et al. 1998; Smith & Ahmer 2003). These observations raised the hypothesis that E. coli also monitors an HSL-like signal(s) produced by other bacteria (Michael et al. 2001). Besides HSL, SdiA has been proposed to monitor other as yet unidentified external signals including indole (Lee et al. 2007). In this study, however, we failed to confirm the binding of a set of well-known signal molecules with SdiA protein.

One effective approach for the identification of signal molecules that are monitored by a LuxR-family protein is to examine the influence of synthetic HSL analogues on the expression of the genes under the control of test LuxR-family protein. Such approaches have been useful in other bacteria. For instance, Schaefer et al. (1996) examined a total of 25 species of chemically synthesized N-(3-oxohexanoyl)-HSL derivatives with alterations in the acyl side chain in luminescence induction in Vibrio fischeri and found that none showed a higher activity than the authentic AI-1, N-3-oxododecanoyl-HSL (3OC12-HSL). Passador et al. (1996) tested a total of 21 species of the acylated derivatives of L-homoserine lactone, L-homoserine thiolactone, and lactam analogues for their ability to act as autoinducers in Pseudomonas aeruginosa and concluded that the length of acyl chain is crucial in both in vivo activation of LasR-regulated target genes and in vitro binding to LasR.

Using a promoter trap library to screen for E. coli promoters whose expression is affected by synthetic N-hexanoyl-L-homoserine lactone (C6-HSL), six upregulated and nine downregulated promoters were identified, which also responded to synthetic 3-oxo-N-hexanoyl-L-homoserine lactone (3-oxo-C6-HSL) (Van Houdt et al. 2006). The HSL-dependent regulation of these promoters is eliminated by knockout of sdiA, supporting the functional role of different HSL signals on the regulation function of SdiA. By making a set of sdiA mutants in E. coli, Lee et al. (2009) identified that different sites on SdiA protein are involved in sensing HSL signals from other bacteria and its own signal indole. The three-dimensional structure of a complex between the N-terminal domain of E. coli SdiA and a candidate N-octanoyl-L-homoserine lactone (C8-HSL) has been calculated in solution from NMR data (Yao et al. 2006, 2007). The structural study indicated the association of SdiA with not only C8-HSL but also additional effectors such as xylose. Other HSL derivatives are also capable of acting as folding-switch effectors for SdiA. The observed structural heterogeneity of the bound ligand in the complex, together with the variety of HSL-type molecules that can apparently act as folding switches in the SdiA system, is consistent with the postulated biologic function of the SdiA protein as a detector of the presence of other species of bacteria.

One short-cut approach for the identification of the signal molecules that are recognized by E. coli SdiA is to test possible influence of HSL analogues on the DNA-binding activity in vitro of purified SdiA. In this study, we examined, for the first time, a systematic analysis of as many as 477 species of HSL analogues on the recognition and binding activity in vitro of DNA sequences by SdiA. Interestingly, two of three SSMs, F and K, were previously identified as QS inhibitors for P. aeruginosa (Smith et al. 2003a; J. Igarashi and H. Suga, unpublished data). In contrast, SSM-A was a novel compound, which was previously unidentified as neither QS agonist nor antagonist. In this study, these SSMs were examined for their influence on the gene selectivity by SdiA. The genomic SELEX screening herein described indicates that SdiA responds to various HSL analogues and alters its recognition targets depending on the signal species. This finding raises the possibility that in nature such as in host intestine, SdiA recognizes various species of the signal molecule released from co-existing bacteria, resulting in different mode of the genome expression for adaptation. Transcription in vivo of the candidate genes under the direct control of SdiA was then examined in the presence and absence of SdiA and in the presence and absence of effector ligands.

Functional response of SdiA to different signals

Microarray analysis in the presence of SdiA over-expression indicated more than threefold transcription elevation for as many as 75 genes including sdiA itself and the ftsQAZ genes and reduction for 62 genes including the motility and chemotaxis genes (Wei et al. 2001a). Over-expression of sdiA causes mytomycin resistance (Wei et al. 2001b). Likewise, SdiA over-expression triggers the resistance to some drugs such as fluoroquinolone and ceftazidime (Tavio et al. 2010). Over-expression of SdiA confers multidrug resistance and increased levels of AcrAB drug efflux pump (Rahmati et al. 2002). The change in genome expression pattern examined using plasmid-encoded over-expressed SdiA was, however, reported to be different from that observed with the genome-coded SdiA (Dyszel et al. 2010). The SdiA encoded by a single copy activates the genes (gadW, gadE, yhiD, and hdeA) for glutamate-dependent acid-resistance system and represses the genes encoding flagella and curli fimbriae. Accordingly, the bacterial motility and increased adherence to epithelial cells are enhanced in the sdiA mutant (Sharma et al. 2010). Several lines of evidence indicate that in response to HSLs, the genome expression pattern and phenotypical properties of E. coli are modulated by means of the HSL responsive transcriptional regulator SdiA. SdiA mutants with altered sensitivity to HSLs and indole reduced or increased biofilm formation (Lee et al. 2009). SdiA is necessary for enterohemorrhagic E. coli (EHEC) colonization of cattle (Hughes et al. 2010). Metagenome analysis indicated that HSLs within the mammalian gastrointestinal tract are produced by Bacteroidetes. One of the regulation targets of SdiA is the ydiV gene encoding an EAL domain-containing protein that is involved in degradation of c-di-GMP, a signal of cell–cell communication and growth switch between biofilm formation and planktonic growth (Jenal & Malone 2006). Expression of YdiV is induced by the addition of HSL (Zhou et al. 2008). These observations altogether indicate that SdiA regulates not only the ftsQAZ operon for cell division but also a number of genes for adaptation to the variety of stressful environmental conditions in nature and survival in host animals.

Transcription factors under the control of multiple effectors

Here, we provided both in vitro and in vivo evidence indicating the alteration of gene selectivity of SdiA by interaction with HSL analogues. Moreover, the spectrum of target gene selection by SdiA is modulated in multiple modes depending on the effector molecules. One unique finding of this study is that a single and the same transcription factor responds to multiple effector ligands.

Escherichia coli contains as many as 300 species of transcription factors, each monitoring a specific factor or condition in environments (Ishihama 2010, 2012). The majority of E. coli transcription factors belong to the one-component signal-transduction system, in which a single polypeptide contains both an effector-binding sensory domain and a DNA-binding domain. The activity of this group's transcription factors has been believed to be controlled by a single species of the effector ligand, that is, inducer or corepressor. Recently, however, the involvement of two effectors have been identified in a number of cases such as allantoin and glyoxalate for AllR (Hasegawa et al. 2008); arginine and lysine for ArgP (or IciA) (Marbaniang & Gowrishankar 2011); glyoxylate and pyruvate for IclR (Lorca et al. 2007); hypoxanthine and guanine for PurR (Houlberg & Jensen 1983); ATP and maltotriose for MalT (Schreiber & Richet 1999); and uracil and thymine for RutR (Shimada et al. 2007). One unique exception is TyrR that regulates a set of genes for synthesis and transport of aromatic amino acids. The activity and specificity of TyrR are regulated by tyrosine, phenylalanine, and tryptophan (Pittard 1996). Here, SdiA was identified as a novel type of transcription factors, of which the activity and specificity are controlled by multiple species of the effector ligand.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

Collection of HSL analogues

A library of AHL analogues was prepared by chemical synthesis according to the published procedure (Igarashi & Suga 2011). From this collection, we used in this study a total of 477 HSL analogues as shown in Supporting Information Fig. S1, which include AHL analogues with different chain length and different acyl conformation and AHL analogues with different lactone rings. The influence of these SSMs on the growth and pathogenicity of P. aeruginosa has been analyzed in detail (Smith et al. 2003b; Chung et al. 2011).

Bacterial strains and culture conditions

Escherichia coli BL21(DE3) [F- ompT hsd (rB- mB-) dcm gal λ(DE3)] (Studier & Moffatt 1986) was used for expression and purification of SdiA. E. coli K-12 BW25113 [lacIq rrnB lacZ hsdR araBAD rhabAD] and its otherwise isogenic mutant strain JW1901 lacking the sdiA gene were products of the Keio collection (Baba et al. 2006) and obtained from National Bio-Resource Center (National Institute of Genetics, Japan). Cells were cultured in LB medium or YESCA medium (Pratt & Silhavy 1998). When necessary, 100 μg/ml ampicillin and 50 μg/ml kanamycin were added in to the medium.

Expression and purification of the SdiA protein

The sdiA-coding sequence was amplified by PCR using E. coli K-12 W3350 genome DNA as a template, and after digestion with NdeI and NotI within the primer sequences, cloned into an expression vector pET21a(+) (Novagen) at the corresponding sites. His-tagged SdiA was expressed in pSdiA-transformed E. coli BL21(DE3) and affinity purified according to the standard purification procedure (Shimada et al. 2005; Yamamoto et al. 2005). The purity of SdiA was higher than 95% as judged by SDS-PAGE.

Gel-shift assay of SdiA-DNA complexes

For estimation of SdiA binding to the ftsQ promoter, the gel-shift assay was used. A fluorescent-labeled promoter fragment of the ftsQAZ operon, the only known target of SdiA, was generated by PCR amplification of using the ftsQ promoter assay vector as the template, a pair of primers (5′-FITC-labeled vector sequence and 3′-proximal vector sequence at the vector-fstQ junction), and Ex Taq DNA polymerase. The PCR product with FITC at 5′ termini was purified by PAGE, and then used for gel-shift assay under the standard conditions (Shimada et al. 2005). For estimation of the amount of SdiA protein for binding to the ftsQ promoter, a fixed amount of ftsQ promoter probe (0.5 pmol) and various amounts (1–100 pmol) of purified SdiA in 0.01 ml were incubated at 37OC for 15 min, and then directly subjected to PAGE. Next possible influence of HSL analogues on the binding of SdiA protein to the ftsQ promoter was examined in the presence of 0.5 pmol ftsQ promoter, 10 pmol SdiA protein, and 0.05 mm each of HSL analogues.

Genomic SELEX screening for SdiA-binding sequences

The improved genomic SELEX system was as described previously (Shimada et al. 2005). A mixture of DNA fragments of the E. coli K-12 W3110 genome was prepared after sonication of purified genome DNA and cloned into a multicopy plasmid pBR322. In each SELEX screening, the DNA mixture was regenerated by PCR. For SELEX screening, 5 pmol of the mixture of DNA fragments and 10 pmol His-tagged SdiA were mixed in a binding buffer (10 mm Tris-HCl, pH 7.8 at 4 °C, 3 mm magnesium acetate, 150 mm NaCl, and 1.25 mg/ml bovine serum albumin) and incubated for 30 min at 37 °C. The DNA-SdiA mixture was applied to a Ni-NTA column, and after washing out unbound DNA with the binding buffer containing 10 mm imidazole, DNA-SdiA complexes were eluted with an elution buffer containing 200 mm imidazole.

For SELEX-chip analysis, PCR-amplified SELEX-DNA fragments and the original DNA library were labeled with Cy3 and Cy5, respectively, and then combined. The fluorescent-labeled DNA mixtures were hybridized to a DNA-tilling array consisting of 43,450 species of 60-bp-long DNA probe, which are designed to cover the entire E. coli genome at 105-bp interval (Oxford Gene Technology, Oxford, UK) (Shimada et al. 2008, 2011a; Teramoto et al. 2010). The fluorescent intensity of test sample at each probe was normalized with that of the corresponding peak of original library. After normalization of each pattern, the Cy5/Cy3 ratio was measured and plotted along the E. coli genome.

Northern blot analysis

Total RNAs were extracted from E. coli cells by the hot phenol method (Aiba et al. 1981). RNA purity was checked by electrophoresis on 0.8% agarose gel in the presence of formaldehyde followed by staining with methylene blue. The amounts of RNAs used were measured by staining rRNAs with ethidium bromide. Northern blot analysis was carried out essentially as described previously (Shimada et al. 2011a,b). DIG-labeled probes were prepared by PCR amplification using E. coli AI-0053 (W3110) genomic DNA as template, DIG-11-dUTP (Roche), and dNTP as substrates, gene-specific forward and reverse primers, and Ex Taq DNA polymerase (Takara). Total RNAs were incubated in formaldehyde-MOPS (morpholinepropanesulfonic acid) gel-loading buffer for 10 min at 65° C for denaturation, subjected to electrophoresis on formaldehyde-containing 1.5% agarose gel, and then transferred to nylon membrane (Roche). Hybridization was carried out with DIG easy Hyb system (Roche) at 50 °C overnight with a DIG-labeled probe. For detecting the DIG-labeled probe, the membrane was treated with anti-DIG-AP Fab fragments and CDP-Star (Roche), and the image was scanned with LAS-4000 IR multicolor (Fuji Film).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

This work was supported by Grant-in-Aid for Scientific Research (A) (21241047), (B) (18310133) and (C) (25430173) to AI and Grant-in-Aid for Grants-in-Aid for Young Scientists (B) (24710214) to TS from MEXT (Ministry of Education, Culture, Sports, Science and Technology of Japan) and MEXT-Supported Program for the Strategic Research Foundation at Private Universities 2008-2012 (S0801037).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
gtc12139-sup-0001-FigS1.pptxapplication/mspowerpoint1272K

Figure S1 List of HSL analogs.

gtc12139-sup-0002-TableS1.pdfapplication/PDF62K

Table S1 SdiA-binding sites on the E. coli genome

gtc12139-sup-0003-DataS1.docxWord document42K 

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.