ABSTRACT Title: VIEWING QUORUM SENSING IN A GLOBAL NETWORK: STUDIES ON SDIA-DEPENDENT SYSTEM 1AND AUTOINDUCER-2 MEDIATED SYSTEM-2 OF NON-PATHOGENIC ESCHERICHIA COLI K-12 Songhee Kim, Doctor of Philosophy, 2005 Directed By: Professor William E. Bentley, Chemical Engineering & Professor Sam W. Joseph, Cell Biology and Molecular Genetics Quorum sensing is a process by which bacteria send out and perceive signal molecules that enumerate their population density. When a threshold signal level is achieved, these bacteria coordinate their behavior and exhibit multicellularity. This can be viewed as a sort of sociological behavior, wherein bacteria communicate with each other resulting in altered societal structure. Studies on the quorum sensing of common laboratory strains of E. coli K-12 have been independently directed on two mainly recognized systems, each associated with an autoinducer family (AI-1 and AI-2). This study has specifically addressed a question of the functionality of a sensor associated with system 1 (a LuxR type sensor, SdiA). Interestingly, this sensor, which is also a transcriptional regulator, was found to positively regulate the uptake of the quorum system 2 signaling molecule (AI-2). An sdiA mutant showed altered mRNA levels for motility genes, purine de novo synthesis and rRNAs, which are likely to link to phenotypic changes such as decreased motility and increased aggregation and stress resistance. This study also describes the regulatory effects of major molecular chaperones, DnaJKE and GroESL, on AI-2 level found in conditioned medium, suggesting a linkage between the well-characterized heat shock stress response and quorum sensing. The finding that there is cross-regulation between E. coli quorum system 1 and 2, and the interplay with global regulatory networks such as the heat shock response, will contribute to the development of ?engineered? quorum systems in E. coli that serve industrial purposes and also to the potential control of pathogenic bacteria. VIEWING QUORUM SENSING IN A GLOBAL NETWORK: STUDIES ON SDIA-DEPENDENT SYSTEM 1 AND AUTOINDUCER-2 MEDIATED SYSTEM-2 OF NON-PATHOGENIC ESCHERICHIA COLI K-12. By Songhee Kim. Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2005 Advisory Committee: Sam W. Joseph, Chair William E. Bentley James Culver Zhongchi Liu Sergei Sukharev (Dean?s Representative) ii Dedication To my loved ones, to whom I will always be indebted: Hyun Kyung Kim, my mother-in- law, Kyung Ja Kim and Jang Ok Kim, my mother and my father. iii Acknowledgements I feel like I have arrived at one stop after a long journey. I appreciate the support and direction of my advisors Dr. Bentley and Dr. Joseph. Their enthusiasm toward the research and their compassion and humor are so precious to me. I thank Dr. Culver, Dr. Liu, and Dr. Sukharev for the valuable advices and guidance they provided during the course of my study. I also extend my sincerest thanks to the members of the Bentley Lab, who share the memory and friendship during the voyage, especially to John. During the last 6 years, I have been so blessed with the births of my two sons, Jaeyoon and Jaeyoung. Most of all, I give my best thanks to my loved husband, Hojeong Park, whose trust never wavered encouraging me even in my restless moments, and who has provided the shoulder to lean on. Thank you. iv Table of Contents Dedication........................................................................................................................... ii Acknowledgements............................................................................................................iii Table of Contents............................................................................................................... iv List of Figures..................................................................................................................... v List of Tables ..................................................................................................................... vi Chapter1: Introduction........................................................................................................ 1 Overview......................................................................................................................... 1 Literature review............................................................................................................. 4 1-1. Discovery of quorum sensing.............................................................................. 4 1-2. Molecular biology of quorum sensing: qualifying features for quorum sensing 6 1-3. Quorum signal activation................................................................................... 22 1-4. Quorum sensing in the global regulatory circuits.............................................. 29 1-5. Quorum sensing in enteric bacteria (problem recognition)............................... 32 Purpose of study and approach ..................................................................................... 36 Chapter 2: Quorum-sensing under stress-effects of heat shock on Escherichia coli AI-2 signaling............................................................................................................................ 38 Abstract......................................................................................................................... 38 Introduction................................................................................................................... 39 Material and Methods ................................................................................................... 44 Results and Discussion ................................................................................................. 48 Chapter 3: Transcriptome analysis of SdiA mutation; potential for AI-1?AI-2 crosstalk in Escherichia coli ................................................................................................................ 69 Abstract......................................................................................................................... 69 Introduction................................................................................................................... 70 Material and Methods ................................................................................................... 74 Results and Discussion ................................................................................................. 80 Chapter 4: Conclusions................................................................................................... 125 Multiple quorum signalings in one organism: testing the regulatory effect from system 2 to system 1 of Escherichia coli. ............................................................................... 125 Functionality of SdiA examined by deletion (chapter 3)............................................ 126 Autoinducer 2 (AI-2) level is actively regulated by molecular chaperones under stress (Chapter2) ................................................................................................................... 130 Significance of the study for potential applications.................................................... 133 Perspectives................................................................................................................. 133 Bibliography ................................................................................................................... 135 v List of Figures Figure 1. The conversion of S-adenosyl methionine (SAM) to AI-2. .............................. 14 Figure 2. The multiple sequence alignment of six members of the LuxR family. ........... 17 Figure 3. Currently suggested Vibrio harveyi and V. fischeri quorum signal pathways. 24 Figure 4. The activity of quorum signals on Vibrio harveyi luminescence...................... 27 Figure 5. The genetic components of E. coli and Salmonella quorum sensing................ 33 Figure 6. The comparison of tube culture and flask culture (30?C).................................. 50 Figure 7. AI-2 from flask cultures at constant 42 ?C (with glucose)................................ 53 Figure 9. AI-2 activity before and after 1 hr heat shock to stationary phase cells............ 56 Figure 10. AI-2 activity before and after 1hr heat shock to exponential phase cells........ 60 Figure 11. The effects of glucose and temperature on AI-2. ............................................ 64 Figure 12. Macroscale phenotypic differences. ................................................................ 83 Figure 13. The reduced swimming motility of W2X2 (W3110 sdiA - ). ............................ 84 Figure 14. Lethal shock assays on the late exponential phase cells.................................. 87 Figure 15. Differential transcriptome expression. ............................................................ 90 Figure 16. Motility-related genes around sdiA in the chromosome.................................. 96 Figure 17. Operon-wide changes with putative functions (metabolism associated). ..... 100 Figure 18. The positive regulation of SdiA on AI-2 uptake machinery. ........................ 105 Figure 19. Reduced rRNA of W2X2 (W3110 sdiA - )...................................................... 111 Figure 20. Searching for SdiA binding motifs................................................................ 121 Figure 21. Significantly changed IG regions with congruency. ..................................... 122 Figure 22. Significantly changed IG regions with incongruency-opposite. ................... 123 Figure 23. Significantly changed IG regions with incongruency-indifference............... 124 Figure 24. Suggested convergence of E. coli quorum system 1 and 2. .......................... 129 Figure 25. Metabolic flux for AI-2 synthesis.................................................................. 132 vi List of Tables Table 1. The examples of acylated homoserine lactones and corresponding sensors ........ 9 Table 2. Bacterial strains used for studying involvement of molecular chaperones. ....... 47 Table 3. Changes of AI-2 activity by 1 hr heat shock at stationary phase........................ 57 Table 4. Changes of AI-2 levels by 1 hr heat shock at exponential phase. ...................... 61 Table 5. Bacterial strains and a plasmid used for the transcriptome analysis................... 79 Table 6. Significant changes in ORF ................................................................................ 94 Table 7. Examination on growth phase transition. ......................................................... 110 Table 8. Literature-based comparative analysis.............................................................. 116 Table 9. Significant changes in IGs ................................................................................ 120 1 Chapter1: Introduction Overview Unicellular microorganisms have evolved to dynamically adjust their physiological characteristics to changing environments for their own survival and propagation. Accordingly, bacterial sensory machineries have been developed that enable cells to perceive signals emanating from both abiotic and biotic sources. The idea that bacteria may sense their own population density and that of other bacterial species? and then transform this knowledge into a coordinated phenotype dates from studies in the ?70?s with a variety of microorganisms including Cyanobacteria, Firmicutes, Actinobacteria, and Proteobacteria. The first discovery of bacterial mass-action was centered on the luminescence of Vibrio fischeri, a symbiotic marine microorganism (Eberhard, 1972; Nealson et al., 1970). This concept was coined ?Quorum Sensing? in the early 1990s (Fuqua et al., 1994). Now, quorum sensing has become a widespread subject of research, as bacterial interactions with eukaryotic systems, i.e. plant and animal kingdoms, have been shown to be quorum-sensing associated. Topics including pathogenesis, symbiotic associations, and even bacterial consortia in biofilm communities are linked to quorum sensing. Advantages of gauging cell density within certain physical ranges were inferred by researchers using ecological or teleological arguments, such as in competitions among microbes faced with limited nutrients, or from the necessity to avoid impending anti- bacterial immune system for human pathogens. The advantages in quorum sensing may have placed some part of selective pressure on bacterial speciation (McFall-Ngai, 2002). 2 It is not surprising that quorum regulation on non-virulent genes has thus far received only minimal attention unlike that of virulent ones. The original intent of my research was to test the possibility of engineering bacterial metabolism by disturbing or stimulating the quorum sensing mechanism for the purpose of increasing the yield of heterologous protein overexpression. Measurement of the Escherichia coli produced quorum signal molecule, AI-2, was an important element in early experiments. It became clear early on, that very little was known about the AI-2 synthesis and uptake mechanisms in E. coli, and importantly, what genes were regulated by this autoinducer. Besides the architecture of E. coli quorum system was incompletely composed of two systems (or more) and several conflicting experimental results were reported on the functionality and interaction between quorum systems. Drawing linkages between quorum sensing and other signal transduction mechanisms turned out to be more difficult than we had imagined, so my study was refocused on determining the functionality of a quorum sensing transcriptional regulator. In short, opposing the ideas that non-pathogenic E. coli strains do not have potential for behaving group-wise, this study was motivated to examine E. coli quorum sensing. We expect that this basic questioning on the quorum regulation can be applied through engineering and/or interfering cell-cell communication in the future; controlling bacterial virulence and recombinant protein expression. Unlike antibiotics-based treatment of problematic bacteria (from human-interest standpoint) dauntingly challenged by occurrence of resistant ones, this new approach is envisioned not to place unwanted selective pressure on the bacteria. Also the idea associated with the signal-mediated gene transcription can be engineered develop tightly controlled bacterial signal mediated 3 transcription system, to control recombinant gene expression [Reviewed in (March and Bentley, 2004)]. In the rest of this Chapter, the diverse features of Gram negative bacterial quorum sensing are highlightened by the required components for successful quorum sensing autoregulation. Further, I introduce fundamental questions we have on quorum sensing in E. coli. In Chapter 2, I present evidence of AI-2 signaling that links global signal networks through molecular chaperones. In particular, heat shock experiments are used to probe the role of molecular chaperones on AI-2 synthesis. Chapter 3 mainly describes results from transcriptome analyses of sdiA deletion and attempts to provide an association with phenotype. Finally, conclusions drawn out from my research are presented along with the remaining questions suggested for future research. 4 Literature review 1-1. Discovery of quorum sensing Euprymna scolopes and Vibrio fischeri interaction During nocturnal feeding periods, Hawaiian squid Euprymna scolopes emits luminescent light matching the intensity of moonlight, downward from their bodies. The luminescence is made possible by a symbiotic marine bacterium, Vibrio fischeri. Living in the center of the squid body cavity, V. fischeri exerts about 20 % of its metabolic potential for light emission when found at high cell density (e.g., ~ 10 10-11 cells/ml) (Makemson, 1986). What they gain by spending this much energy has intrigued many scientists. Importantly, the luminescence from E. scolopes exhibits periodicity that is controlled by the internalized V. fischeri cell density. Each morning, more than 95 % of the symbionts are expulsed from the animal host into the surrounding seawater thereby dissipating the luminescence. The squids stay in the sand, hiding during daytime hours. Meanwhile unexpelled 5 % bacterial cells grow and repopulate the light organ as nightfall approaches (Lee and Ruby, 1994; McFall-Ngai, 1990). In this way, a periodic cycle of luminescence can be synchronized with that of bacterial mass population. Initiated from the discovery by ecologists, the squid-bacterial symbiosis became the prototypic model of quorum sensing and has intrigued many other research groups in the field of development and evolution who noted the importance of its dynamicity. For instance, reciprocal interactions and competition between host and bacteria, or genes between bacterial species, have been studied (Lee and Ruby, 1994). Also, mechanisms 5 governing the essential bacterial communication including its biochemical features have been actively studied at the molecular level (Graf and Ruby, 1998; Lupp et al., 2003; Miyamoto et al., 2003). Since the discovery of quorum sensing in V. fischeri, a diverse array of bacterial organisms have been studied for possible density dependent phenotypic changes and have often been reported as quorum sensing organisms. From these pools of proposed quorum-sensing bacteria, discerning real quorum-dependent bacteria has become an issue. 6 1-2. Molecular biology of quorum sensing: qualifying features for quorum sensing We define a quorum sensing bacterium as one that detects a small molecule that is either self-secreted or is otherwise present in its immediate growth environment, and through the detection of this signal molecule, it coordinates its own behavior as a function of its own cell density. Traditionally, qualified quorum sensing bacteria are equipped with at least three components: signal molecules produced by autoinducer synthases, their cognate receptors or transcriptional regulators, and genes that are placed under the control of the signals. Note that non-self secreted signals from other bacteria can function as effectors, which are specifically referred as alloinduction. Hence, the genetic requirement that a signal synthase encoding component be present is not as critical. In general, researchers pursue fundamental insight on quorum sensing systems in both directions; by starting with phenotypes that appear to be cell density dependent and unraveling their molecular componentry or by starting with known molecular components and searching for phenotypic differences. Searching for genes involved with bacterial luminescence is one example of the first case, while screening genes affected by exogenous autoinducers or by mutation in regulators (receptors) is exemplary of the second case. I have chosen the latter and have focused on SdiA, an E. coli homologue to V. fischeri quorum sensing transcriptional regulator, LuxR. 7 1-2-1. Signals: diversity versus universality Signal-molecules are usually produced as byproducts of pivotal enzymatic reactions. S-adenosylmethionine (SAM) is the common precursor for both autoinducer families, AI-1 and AI-2. In plants, SAM has also been known as the precursor of the plant signal, ethylene, a gaseous plant hormone that induces the ripening of fruits (White et al., 1994). The apparent uniqueness of SAM in its utility to serve as a precursor to signals among various biological systems is reviewed in (Fontecave et al., 2004). Using SAM as precursors, either species-specific signals by diversifying signal molecules and/or use species-non specific signals by using seemingly uniform molecules are produced by many Gram negative bacteria. Diversity Diverse forms of Gram-negative bacteria commonly in the form of Acylated Homoserine Lactone signals, whose differences are crucial in the signal specificity (AHL or hereafter AI-1). Over ten different AI-1s have been identified from different bacterial species, often through the use of a combination of high performance liquid chromatography (HPLC) purification, nuclear magnetic resonance (NMR) spectroscopy, and mass spectroscopy (Schaefer et al., 2000). AI-1s have a common five-membered homoserine lactone ring and can be diversified by conjugated acyl side chains differing in length, backbone saturation, or the side chain substitution (Table 1). The acyl chains identified thus far vary from 4 to 14 carbons (all even-numbered) in length. The length of the acyl chain proportionately increases the hydrophobicity of the signals and determines each signal?s membrane permeability. Fundamentally, AI-1 is able to freely move across cell membranes. For example, titrated V. fischeri 3-oxo-C6-HSL was used to characterize free 8 diffusion across cell membranes (Kaplan and Greenberg, 1985). However, when the acyl chain is too long, it is hard to rely on the simple diffusion thus active secretion is needed. The length of the acyl chain also affects signal?s turnover rate (see below). In the synthesis of AI-1 molecules, amide bond formation between the acyl- group and the amino group of S-adenosyl methionine (SAM) is critical (Hanzelka and Greenberg, 1996). Three types of AHL synthases, LuxI, AinS (LuxM), and HdtS types, have been identified (Laue et al., 2000; Swift et al., 2001). Amongst them, LuxI family members are evolutionarily conserved. In vitro catalysis of AI-1 synthesis was reported with V. fischeri LuxI (Schaefer et al., 1996), A. tumefaciens TraI protein (More et al., 1996) and P. aeruginosa RhlI protein (Parsek et al., 1999). The second type of AI-1 synthase (AinS family) is non-homologous to the typical LuxI proteins. In this group, products of luxM (V. harveyi, 25-kDa) and ainS (V. fischeri, 46 -kDa) genes are included. LuxM has 34 % identity with the C-terminal 218 residues of AinS (Fuqua et al., 2001). The putative third group of AI-1 synthase, HdtS (homolog of PlsC from E. coli), was identified from P. fluorescens (Laue et al., 2000). It is closely related to lysophosphatidic acid acyltransferase, which is needed for cell membrane biosynthesis. 9 AI-1 Bacteria Regulators Chain length ? R-group a Target function LuxR-LuxI 6 =O Bioluminescence Vibrio fischeri AinR-AinS b 8 -H Bioluminescence Vibrio harveyi LuxM/LuxN- LuxO-LuxRvh e 4 -OH Luminescence,proteases, polyhydroxybutyrate synthesis, and others Vibrio anguillarum VanR-VanI 10 =O Proteases Agrobacterium tumefaciens TraR-TraI 8 6 6 =O =O Conjugal transfer * Pantoea stewartii EsaR-EsaI 6 =O Exopolysaccharide Rhodobacter sphaeroides CerR-CerI 14 d -H Aggregation Yersenia enterocolitica YenR-YenI 6 -H. =O Target functions unknown LasR-LasI 12 =O Host interaction RhlR-RhlI 4 -H Rhamnolipids Pseudomonas aeruginosa QscR C ? ? Host interaction Erwinia stewartii EsaR-EsaI 6 =O virulence, exopolysaccharide Escherichia coli SdiA-? ? ? Cell division,? Table sources are from reviews of (Fuqua et al., 1996; Fuqua et al., 2001) and * from (Zhang et al., 1993) a. Moiety at ? position (R in diagram; fully reduced, H; hydroxyl, OH; carbonyl, O) b. AinR is homologous to two component sensor kinases; AinS is not homologous to LuxI c. QscR is a LuxR homologoue that inhibits the activity of LasR d. Has an unsaturated bond between position 7-8 on the acyl chain e. LuxM/LuxN-LuxO-LuxR e Phosphorelay, LuxRvh is not a LuxR family protein Table 1. The examples of acylated homoserine lactones and corresponding sensors (denoted as AI-1 in this dissertation) 10 Although the three family proteins apparently differ in structure, they catalyze a common reaction recognizing SAM as an amino acid substrate, not as a methyl-donor, unlike most other SAM-utilizing enzymes. Acyl-acyl carrier protein (acyl-ACP) is known to be most efficient as an acyl donor for LuxI proteins, while acyl-ACP and acyl-CoA conjugates provide acyl side chains for AinS family proteins (Hanzelka and Greenberg, 1996; Hanzelka et al., 1999). Acyl-ACP conjugates are intermediates in fatty acid synthesis, while acyl-CoA compounds are typically intermediates of fatty acid ? -oxidation. Sequential reactions initiated by SAM binding followed by butyryl-ACP have been suggested from a study using purified soluble P. aeruginosa RhlI (Parsek et al., 1999). Studies on HdtS are incomplete, but based on the activity of E. coli PlsC, both acyl-coA and acyl-ACP could be used as substrates for autoinducer synthesis (Laue et al., 2000). Several reports show that single AI-1 synthases can produce multiple types of AI-1 molecules (Zhu et al., 1998), implying a range of Acyl-substrate specificity for the enzymes. Also many bacteria are reported to have multiple AI-1 synthases. Production of a range of AI-1s with different length of acyl side chains may provide bacteria with greater versatility in adaptation to various ecosystems (Yates et al., 2002). How they differentiate self- or non-self-derived signals is a key in the signal specificity. In natural environments, AI-1 is not accumulated to high concentrations. This is partly explained by natural lactone ring hydrolysis by hydroxide ions in alkaline conditions. Such laconolysis increases with increased temperature and decreases as the acyl side chains become more elongated (C12 and beyond). Opened lactone rings can be closed reversibly by lowering the pH. Enzymatic lactonolysis also occurs in the natural environments. Bacterial species, such as Variovorax paradoxus and Bacillus subspecies., can degrade AI- 11 1s as a source of organic compounds, energy, and nitrogen (Dong et al., 2000; Leadbetter and Greenberg, 2000). Quorum signals are often detected with bioassays using luminescence (lux) or ? - galactosidase (lacZ) genes or by the production of purple compound, violacein, in response to the presence of signal molecules. Currently used AI-1 monitoring systems are CviR- based Chromobacterium violaceum CV026 (Atkinson et al., 1999), LuxR based E. coli biosensor strains E. coli (pSB536) for C4-HSL and E. coli (pSB1705) for 3-oxo-12-HSL (Winson et al., 1998), and TraR-based A. tumefaciens NT1 (pZLR4). Using multiple biosensors in combination enables one to detect the entire range of known AI-1s (Ravn et al., 2001). Universality AI-2 is the most recently identified signal molecule and it is considered to be ?universal? since about half of the fully sequence bacterial genomes, both Gram-positives and Gram-negatives, have genetic potential to produce this molecule, as they have a V. harveyi luxS homologue in their genome (Waters and Bassler, 2005). The AI-2 biosynthetic pathway is now elucidated in which SAM is demethylated to S-adenosylhomocysteine (SAH), and then sequentially degraded to S-ribosylhomocysteine (SRH), and (4S)-4,5-dihydroxy-2,3-pentanedion (DPD) as depicted (Fig. 1). Most prokaryotes use this two step conversion of SAH to homocysteine, by enzymes Pfs (5?- Methylthioadenosine/S-adenosylhomocystein (MTA/AdoHcy) nucleosidase) and LuxS. While LuxS is essential for AI-2 synthesis, Pfs appears not as essential and works also in the SAM decarboxylation pathway producing polyamines (Lee et al., 2003). In contrast, 12 Archea, Eukarya, and Alphaproteobacteria prefer the one-step conversion of SAH to adenosine and homocysteine (Sun et al. 2004). The resulting homocysteine can be degraded further to ? -ketobutyrate and then to succinyl CoA or can be recycled back to methionine and SAM (Berg et al.2001). The final structure of V. harveyi AI-2 was first reported to be a borated furanosyl diester as co-crystallized with LuxP protein, which specifically binds AI-2 in the periplasmic space. However, AI-2 from S. typhimurium is not borated (Miller et al., 2004). Considering borate is abundant in the ocean (~0.4 mM) but not in the terrestrial environment (Waters and Bassler, 2005), another level of species-dependent variation must exist even in AI-2 reflecting the condition of natural habitats (Miller et al., 2004). However, AI-2 is still considered to be a universal quorum signal molecule. While Vibrio species use the periplasmic protein, LuxP, for AI-2 binding and then subsequently initiate signaling via a two-component phosphotransfer from the inner- membrane, both E. coli and Salmonella uptake AI-2 through an ATP-dependent cassette transporter into the cytoplasm (Taga et al., 2003; Wang et al., 2005). Although there remains speculation as to the reasons for ATP consumption in AI-2 uptake, none have been experimentally verified. Especially noteworthy is that in the gut, high concentrations of AI- 2 are expected since AI-2 synthesis is associated with many enteric bacteria. Indeed, from the human stool specimens and from in vitro artificial intestine model, AI-2 activities were detected (Sperandio et al., 2003). Presence of AI-2 was also found from stored foods (Cloak et al., 2002). So far, the most commonly used AI-2 biosensor strain is V. harveyi derivative BB170, which was developed by null-mutation of V. harveyi quorum receptor of AI-1, but keeping 13 the sensory of AI-2 intact (Bassler et al., 1997). Recently, a third sensory of V. harveyi was identified (Henke and Bassler, 2004). Discovery of the third parallel sensory has not raised concerns on BB170 reporter activity (Fig. 4.A). 14 Figure 1. The conversion of S-adenosyl methionine (SAM) to AI-2. Modified from (Miller et al., 2004; Zhao et al., 2003). The activated methyl group of SAM is used for methylating RNA, DNA, certain metabolites and some proteins leading to the formation of the toxic metabolite S-adenoylhomocysteine (SAH). Bacteria commonly remove the SAH using two enzymatic steps using Pfs and LuxS, which simultaneously generates 4,5-dihydoxy-2,3-pentandione (DPD). DPD is an extremely unstable molecule and cyclized as a furanone. AI-2 molecule co-crystallized as a ligand for LsrB of S. typhimurium is (2R,4S)-2-methyl-2,3,3,4-tetrahydorxytetrahydrofuanon (R-THMF), while AI-2 molecule as a ligand of V. harveyi is reported as a borated S- THMF. LuxS reaction also produces homocysteine, which can be either recycled to methionine synthesis (see Fig. 23) or be degraded. 15 1-2-2. Signal dependent transcriptional regulator Structure: signal binding and DNA binding Most well known regulatory elements for the autoinducer AI-1 quorum circuit, are an evolutionarily conserved group of proteins known as the LuxR family. Working as transcriptional regulators, this group of proteins has two functional domains, an N-terminal (about two-thirds) domain for binding autoinducer and a C-terminal (one third) domain for DNA binding and transcriptional activation. They also interact with RNAP ? -subunit C- terminal domain (? CTD) to recruit RNA polymerase to target promoters (e.g., V. fischeri LuxR system (Stevens et al., 1999)). Sequence similarities in C-terminus helix-turn-helix- containing regions define the group [Reviewed in (Choi and Greenberg, 1991)], while sequence variations in N-terminus signal binding domains with a hydrophobic pouch indicates signal-specificity driven protein evolution (Fig. 2). Coevolution between the LuxI and LuxR may have enabled bacterial cells to differentiate subtle structural features of signals from the diverse pool of quorum signals present in the environments. Structural features of the regulatory proteins are as well understood as those of signal molecules, partly due to difficulties in protein purification and the time involved with structural determination. Generally LuxR-type proteins aggregate and become insoluble upon overexpression in E. coli. This problem has at times been solved by the addition of signal molecules at the same time as overexpression, which implies a critical requirement of AI-1 for effective translation and folding of the LuxR polypeptide. This was noted in studies with TraR (co-crystallized as a complex with its cognate signal, 3OC8-HSL) and V. fischeri LuxR, where cognate AI-1 addition increased the protein-yield by stabilizing LuxR conformation and increasing its DNA binding affinity after multimerzation (Stevens and 16 Greenberg, 1999)(Urbanowski et al., 2004). In this way, full length of V. fischeri LuxR was purified for the first time. Sometimes, an autoinducer-independent truncated form of LuxR has been used to study its transcriptional activity (Sitnikov et al., 1996). Additional LuxR family proteins that have been successfully crystallized are: Pantoea stewartii ssp. stewartii EsaR (Minogue et al., 2002), Erwinia chrysanthemi ExpR (Nasser et al., 1998), Erwinia CarR (Welch et al., 2000), and A. tumefaciens TraR (Zhu and Winans, 1999). These proteins are sometimes cocrystalized with their cognate signal or DNA showing the strength of the interaction. Conserved DNA sequences for V. fischeri LuxR protein binding have been reported as a ?lux box?: a 20-bp inverted repeat (ACCTGTAGGATCGTACAGGT). This sequence is centered at position 42.5 b away from the transcriptional start site. The lux box and its location with respect to other promoter elements is critical for LuxR-dependent activation of the lux genes (Egland and Greenberg, 2000)). A ?Las box? has also been defined by the presence of a conserved sequence identified from all the genes known to be activated by the LasR/RhlR/QscR regulators. The las box, NNCT (N) 12 AGAA, is centered at -40 region (Egland and Greenberg, 2000; Whiteley and Greenberg, 2001). Suggested A. tumefaciens ?Tra boxes? are ATGTGCAGATCTGCACAT, ATGTGCAGATCTGCACGT, and ATGTGCAAATCTGCACTT. Each of these is positioned differently relative to the transcription start site (Fuqua and Winans, 1996). The consensus sequences with dyad structures observed from the three LuxR proteins support bimolecular binding of LuxR on the promoter and indicate varied target motifs in LuxR family proteins among species. 17 SdiA-Salmonella MQENDFFTWRRAMLLRFQEMAAAEDVYTELQYQTQRLEFDYYALCVRHPVPFTRPKISLR 60 SdiA-Escherichia MQDKDFFSWRRTMLLRFQRMETAEEVYHEIELQAQQLEYDYYSLCVRHPVPFTRPKVAFY 60 LasR-Pseudomonas -------MALVDGFLELERSSGKLEWSAILQKMASDLGFSKILFGLLPKDSQDYENAFIV 53 LuxR-Vibrio -MKNINADDTYRIINKIKACRSNNDINQCLSDMTKMVHCEYYLLAIIYPHSMVKSDISIL 59 ExpR-Erwinia --------MSQLFYNNETISRIIKSQFDMALSHYGDIKYAYMVLNKKKP-----TEILII 47 TraR-Agrobacterium ---------MQHWLDKLTDLAAIEGDECILKTGLADIADHFGFTG-YAYLHIQHRHITAV 50 SdiA-Salmonella TTYPPAWVTHYQSENYFAIDPVLKPENFRQGHLHWD-DVLFH----EAKAMWDAAQRFGL 115 SdiA-Escherichia TNYPEAWVSYYQAKNFLAIDPVLNPENFSQGHLMWN-DDLFS----EAQPLWEAARAHGL 115 LasR-Pseudomonas GNYPAAWREHYDRAGYARVDPTVSHCTQSVLPIFWE-PSIYQTR--KQHEFFEEASAAGL 110 LuxR-Vibrio DNYPKKWRQYYDDANLIKYDPIVDYSNSNHSPINWN-IFENNAVNKKSPNVIKEAKTSGL 118 ExpR-Erwinia SNHHDEWREIYQANNYQHIDPVVIAALNKITPFPWD-EDLLVSTQLKMSKIFNLSREHNI 106 TraR-Agrobacterium TNYHRQWQSTYFDKKFEALDPVVKRARSRKHIFTWSGEHERPTLSKDERAFYDHASDFGI 110 SdiA-Salmonella RRGVTQCVMLPNRALGFLSFSRSSLR--CSSFTYDEVELRLQLLARESLSALTRFEDDMV 173 SdiA-Escherichia RRGVTQYLMLPNRALGFLSFSRCSAR--EIPILSDELQLKMQLLVRESLMALMRLNDEIV 173 LasR-Pseudomonas VYGLTMPLHGARGELGALSLSVEAENRAEANRFMESVLPTLWMLKDYALQSGAGLAFEHP 170 LuxR-Vibrio ITGFSFPIHTANNGFGMLSFAHSEKDN-YIDSLFLHACMNIPLIVPSLVDNYRKINIANN 177 ExpR-Erwinia TNGYTFVLHDHSNNLVMLSIMIDESNVSNIDDVIESNKDKLQMTLMTIHAETISLYREMI 166 TraR-Agrobacterium RSGITIPIKTAN---GFMSMFTMASDKPVIDLDREIDAVAAAATIGQIHARISFLRTTPT 167 SdiA-Salmonella MAP-------EMRFSKREKEILKWTAEGKTSSEIAIILSISENTVNFHQKNMQKKFNAPN 226 SdiA-Escherichia MTP-------EMNFSKREKEILRWTAEGKTSAEIAMILSISENTVNFHQKNMQKKINAPN 226 LasR-Pseudomonas VSK-------PVVLTSREKEVLQWCAIGKTSWEISVICNCSEANVNFHMGNIRRKFGVTS 223 LuxR-Vibrio KS--------NNDLTKREKECLAWACEGKSSWDISKILGCSERTVTFHLTNAQMKLNTTN 229 ExpR-Erwinia RNKEDERSNDKDIFSQRENEILYWASMGKTYQEIALILDIKTGTVKFHIGNVVKKLGVLN 226 TraR-Agrobacterium AED-------AAWLDPKEATYLRWIAVGKTMEEIADVEGVKYNSVRVKLREAMKRFDVRS 220 SdiA-Salmonella KTQIACYAAATGLI------- 240 SdiA-Escherichia KTQVACYAAATGLI------- 240 LasR-Pseudomonas RRVAAIMAVNLGLITL----- 239 LuxR-Vibrio RCQSISKAILTGAIDCPYFKN 250 ExpR-Erwinia AKHAIRLGIELQLIRPVQS-- 245 TraR-Agrobacterium KAHLTALAIRRKLI------- 234 Figure 2. The multiple sequence alignment of six members of the LuxR family. E. coli SdiA, S. typhimurium SdiA, P. aeruginosa LasR, V. fischeri LuxR, E. carotovora ExpR, and A. tumefaciens TraR protein sequences were aligned by ClustralW algorithm (EMBL-EBI). Identical residues are distinguished by black bold letters and conserved residues are done by blue (less conserved) or red (more conserved) ones. Secondary structures were adopted from the TraR structure suggested from (Vannini et al., 2004). 18 Transcription activation or repression Members of the LuxR family proteins have different mechanisms for transcriptional activation. P. aeruginosa LasR, A. tumefaciens TraR, V. fischeri LuxR, and Erwinia carotovora CarR, the proteins bind to the DNA target sequences only when complexed with their corresponding AI-1. On the other hand, E. chrysanthemi ExpR can bind to its DNA target sequence even in the absence of its corresponding AI-1(Reverchon et al., 1998). ExpR can also function as a transcriptional repressor of its own synthesis. For this bimodular regulation, it represses DNA transcription in the absence of AI-1, while the presence of AI causes a conformational change making it dissociate from the target DNA. Similar examples can be found from Pantotea stewartii EsaR and Serratia marcescens SpnR. Most of the studied LuxR family proteins are not able to bind to the target sequences when complexed with AI-1 and when not complexed with AI-1 (Medina et al., 2003). Another exception is RhlR, which can bind on rhlAB promoter in the presence and absence of AI-1. . 19 1-2-3. Target genes (function) By far, the most dominant targets for quorum sensing regulated genes are in association with pathogenesis (Ahmer, 2004; Newton and Fray, 2004; Suga and Smith, 2003). In particular, those that are reported to influence or control virulence are the most widely reported and studied, although there is an exception in that of luciferase encoding genes from Vibrio. AI-1 controlling functions Genes regulated by AI-1/LuxR family can be grouped into (exo) enzymes, antibiotics and siderophore production, and surface motility for gaining nutritional and spatial advantages in natural niches, capsular polysaccharide for virulence and aggregation, and producing gene transfer and conjugation agent for propagation. Certain AI-1s possess pharmacological and immunomodulatory activity such that their direct role as virulence determinants was implied (Yates et al., 2002). AI-2 controlling functions The luxS and AI-2 are reported to be involved in global regulation of physiologic functions and virulence. However, due to the fact that AI-2 synthesis is associated with activated methyl cycle, the actual mechanisms of signal transduction and gene regulation by the autoinducer are difficult to determine. Still, an increasing number of reports show phenotypic changes by in conjunction with the luxS gene (e.g., a Streptococcus pyogenes mutant showed enhanced haemolytic activity and a reduction in secreted proteolytic activity (sagA and speB)(Lyon et al., 2001). LuxS has also been shown to induce a hemin 20 acquisition protein and arginine-specific protein (RgpA), but repressed a hemin-regulated protein and an exonuclease (Chung et al., 2001). LuxS was shown to be associated with acid susceptibility (Wen and Burne, 2004) and biofilm development in mixed cultures of Streptococcus gordonii and Porphyromonas gingivalis cells (Davies et al., 1998; Li et al., 2002; Merritt et al., 2003). Also, in the gram positive Clostridium perfringens, the importance of LuxS and AI-2 was reported in the production of the three toxins, alpha-, kappa-, and theta- toxins (Ohtani et al., 2002). Independent microarray-based transcriptome analyses were done to show the consequence of luxS mutation in EHEC 0157::H7 (Sperandio et al., 2001) and the effects of conditioned media with or without AI-2 on W3110 (DeLisa et al., 2001). Both studies showed global effects of AI-2/LuxS, in varied functions including DNA processing, carbohydrate metabolism, surface- and outer membrane-associated function, and virulence. To date, the only confirmed AI-2 dependent gene transcription identified in E. coli was also shown in Salmonella, is limited to AI-2 uptake machinery, lsr (luxS regulated) operon and the methionine synthase gene, metE. 21 1-2-4. Methodology for identifying quorum signaling components Identification of quorum sensing components has been accomplished by various molecular biological techniques. Commonly, E. coli is used as a surrogate host, as its relatively well-annotated genome and ease of genetic manipulation for harboring reporters or regulatory genes on multicopy plasmids, make it amenable to a variety of techniques. Interestingly, so far ignored E. coli quorum signaling (as a host) affects the activity of heterogeneous quorum system (Lindsay and Ahmer, 2005). Advanced from the non-quantitative Northern analysis techniques, recently- developed microarray technologies have been used as systemic tools for identifying the architecture and dynamics of gene networks (Tegner et al., 2003). In vivo expression technology using promoterless reporter genes, are also used for positive selection of promoters that are specifically activated under certain conditions. Although microarrays and promoter-fusions are carried out at similarly large scales, they differ in that one represents the real quantity of RNA levels while the other is inevitably associated with protein expression. Screens of putative quorum components require further confirmation with in vitro assays, which provide information on the mechanisms of gene expression regulation. Importantly, with the increasing number of fully sequenced bacterial genomes, homologous sequence searches have proven quite effective and efficient in identifying conserved quorum components. 22 1-3. Quorum signal activation 1-3-1. Well-known model organisms, Vibrio species Direct signal activation Perhaps the simplest quorum sensing mechanism is based on the direct binding of AI-1? LuxR to activate transcription, as shown in Vibrio fischeri (Fig. 3). For this, AI-1 diffuses out and then comes back into the cell as the cell density increases. A second, more complex, type of signal transfer was also identified from bacteria of the genus, Vibrio (see below). Phosphorelay V. harveyi quorum sensing uses two component sensors initiating phosphorelay cascade for density dependent activation of the lux operon. In this model (Lenz et al., 2004), at low cell density when the autoinducers are present at low concentrations, phosphate flows toward LuxO, which belongs to a large and highly conserved family of ? 54 -dependent transcriptional regulators. LuxO has three conserved domains: the response regulator domain, the ? 54 activation domain, and a HTH motif for direct DNA binding (Lilley and Bassler, 2000). Together with ? 54 , phospho-LuxO can activate the expression of small regulatory RNAs (sRNAs). The complexes of the sRNAs and sRNA chaperone protein Hfq destabilize the mRNA of the quorum-sensing master regulator LuxR VH , resulting in the indirect repression of lux operon transcription. LuxR VH is not homologous to V. fischeri LuxR. 23 At high cell density, the binding of autoinducers to membrane receptors makes the phosphate flow away from LuxO. In this condition, LuxO is dephosphorelyated and the regulatory sRNAs are not expressed. As a consequence, LuxR translation and its expression are not free from the repression of destabilizing sRNA-Hfq complex. The result is the transcriptional activation of the lux operon. In the suggested V. harveyi models, autoinducers do not need to reenter the cells and can initiate the phosphorelay from the innermembrane bound sensory proteins, LuxN and LuxPQ (Freeman et al., 2000). Multiple components for quorum signaling were identified from V. harveyi, but LuxI-LuxR VF homologues have not been found to date. Bacterial phosphorelay is close to eukaryotic phosphotransfer featured with a conversion of response regulator from kinase to phosphatase. However it is distinguished from eukaryotes in that aspartate and histidine residues are used for phosphotransfer rather than threonine and serine. Interestingly, a second quorum component pair (synthase/regulator) was identified in V. fischeri. AinS, and its corresponding regulator, LitR, are suggested to be operated at low cell density in laboratory cultures. Along with LuxS, the AinS/LitR pair, is homologous to quorum components of V. harveyi and is therefore, considered to operate in an analogous manner to the LuxO system of V. harveyi. It is not known whether sRNAs are also involved in bioluminescence control in V. fischeri (Lupp and Ruby, 2005). 24 V. harveyi V. fischeri Figure 3. Currently suggested Vibrio harveyi and V. fischeri quorum signal pathways. Currently identified three parallel quorum signaling pathways of V. harveyi and V. fischeri are depicted with signal synthase (ovals), transcriptional regulators (colored rectangle), and the target genes (yellow arrows). V. fischeri LitR is homologous to V. harveyi LuxR VH , while V. fischeri LuxR does not share homology to it. Note that LuxI-originated signal molecule binds (light blue circles) the V. fischeri LuxR to activated the lux operon while other quorum signal molecules initiate the signaling from the inner membrane (Visick, 2005). LuxM LuxS LuxP LuxQ LuxU LuxO LuxR VH luxCDABE sRNA CqsA LuxN CqsS AinS LuxS LuxP LuxQ LuxU LuxO LitR CDABE sRNA CqsA LuxN luxI LuxR luxR fla LuxI AI-2 AI-2 AI-1 AI-1 AI-1 25 1-3-2. Quantitative analysis signal activation (dose-responsive) The physiologically relevant AI-1 concentrations for target gene activation have been calculated from dose-response activation curves (Blosser and Gray, 2000; Michael et al., 2001; Zhu et al., 1998). For this purpose, in vitro synthesized signal molecules are preferred over conditioned media (CM) due to their directed action and the potential for directly assigning activity to the presence of the specific molecule (as opposed to an array of molecules present in CM). A study with C. violaceum reporter strains tested with a synthetic AI-1(C6) showed the maximum production of pigment at a concentration of 20 nM, while a minimum detectable response occurred with 1 nM of the signal (Blosser and Gray, 2000). TraR activity on TraI promoter by additional cognate AI-1 (3-oxo-C8-HSL) showed that c.a. 1 nM of AI-1 can induce the half maximal response, while c.a. 5 nM can saturate the activity (Zhu et al., 1998). When transcriptional activity of S. typhimurium SdiA on the rck promoter was tested with 10 AI-1 type analogues, it showed a half-maximum response concentration of 1-5 nM, using the most potent signal (oxoC10). rck is the gene which was found from the promoterless lacZ genetic screen searching for sdiA-dependent gene expression (Michael et al., 2001). From the three studies presented here, approximately 5 nM of AI-1 seems to be sufficient to activate target gene transcription at levels up to half the maxima. At present, direct quantification of AI-2 is not available due to its extreme polarity. So the concentration is measured rather indirectly from in vitro synthesized AI-2 (Ellman?s test for free sulfhydryl groups, calculated assuming a 1:1 ratio between homocysteine and AI-2)(Schauder et al., 2001) or by measuring luminescence of BB170 cultures after adding 26 serially diluted conditioned media. About 100 nM of in vitro-synthesized AI-2 accomplished the half-maximal level of light production from BB170 reporter (Fig. 4.B) (Schauder et al., 2001). 27 A. B Figure 4. The activity of quorum signals on Vibrio harveyi luminescence. (A) Density dependent luminescence of V. harveyi single, double, and triple autoinducer sensor. Light units are reported as counts min ?1 ml ?1 x 1,000/CFU ml ?1 . Arrow is to show the increase luminescence by exogenous quorum signal. Figure was adopted and modified from (Henke and Bassler, 2004). (B) Activities of in vitro synthesized AI-2 and its structure analogues (MHF, DMHF, and HF) were measured using V. harveyi reporter BB170. Figure was adopted from (Schauder et al., 2001). 10 5 10 6 10 7 10 8 10 9 10 6 10 5 10 4 10 3 10 2 10 1 Colony forming units Relative light units w/t luxO - sensor1- (BB170) sensor2 - sensor3 - Li g ht Production Compound concentration (? M) AI-2 28 1- 3-3. Signal interception and eavesdropping in environments The diversity or universality of signals imply intra- or inter bacterial species communication as well as cross kingdom gene regulation. Situations, such as the presence of quorum sensing-dependent target genes in the absence of the endogenous signal synthases, have intrigued many research groups. In the environment, bacteria such as Pseudomonas aeruginosa may modulate their behavior by monitoring environmental conditions and probably by eavesdropping on the other bacterial AI-2 and/or other signals. P. aeruginosa does not have the luxS genes, but has genes responding to the AI-2. Examples of inter-kingdom signal events can be found from studies using halogenated furanones from marine algae Delisea pulchra inhibiting infection by plant pathogen Serratia liquefaciens. Structurally similar to the AI-1, halogenated furanones can competitively inhibit the bacterium?s quorum sensing system (Rasmussen et al., 2000). Also mammalian hormone, such as epinephrine, is reported to be analogue of autoinducers of enterohemorragic E. coli autoinducer (Sperandio et al., 2003). 29 1-4. Quorum sensing in the global regulatory circuits 1-4-1. Catabolite repression and autoinduction There are more factors controlling the luminescence of V. fischeri besides the commonly reported autoinduction system. It has long been appreciated that autoinduction of bioluminescence in many marine bacteria is also subject to catabolite repression of AI-1 synthesis. That is, the presence of glucose was reported to delay bioluminescence of Vibrio (Ruby and Nealson, 1976). In contrast, the addition of glucose to the enteric bacteria, E. coli and Salmonella cultures increases and prolongs the duration of AI-2 activity (Surette et al., 1999), implying a different physiological role of glucose in the metabolism-linked Quorum signal regulation. 1-4-2. Global regulators Quorum sensing, as a global gene regulation system, was first shown in P. aeruginosa (ChaponHerve et al., 1997) . In this study, the roles of stationary phase sigma factor, rpoS, and genes coding for components of the general secretory pathway (xcp) suggested that cells may anticipate stationary phase by sensing the crowdedness. Induction of RhlR by RpoS was reported, but this is controversial, since opposite observation of induction of RpoS by RhlR and C4-HSL mediated quorum sensing was also reported from P. aeruginosa (Latifi et al., 1996; Whiteley et al., 2000). It is not conclusive which one is hierarchically higher, or take precedence, but it is highly likely that RpoS and quorum regulators are closely linked and reciprocally influencing the other. 30 1-4-3. Molecular chaperones and proteases Molecular chaperones and proteases control the quality of proteins newly emerging from ribosomes (about 10 to 20 % of the total) by binding to exposed hydrophobic patches on proteins. Chaperones prevent partially unfolded from being aggregated to functionally inactive states. They also facilitate the destruction of structurally damaged and/or misfolded proteins occurring from various stresses, such as oxidative stress and heat shock stress. This process consumes a substantial amount of energy in the form of ATP and requires a balance between the levels of chaperones and proteases (Wickner et al., 1999). The first introduction of molecular chaperones as components within quorums sensing systems was demonstrated by the overexpression of a cloned V. fischeri luxR gene in E. coli (Dolan and Greenberg, 1992). For proper expression of LuxR, GroES? GroEL complex also was overexpressed. Another group has focused on the role of proteases in controlling the life-span of the transcriptional regulator (Zhu and Winans, 2001). An alarmed state resulting from ?crowdedness? needs to be tuned so that when the cell density level decreases, the activated system whould be down-regulated. Proteolysis of the transcriptional regulator has been suggested as one way to terminate the activated LuxR-AI system in A. tumefaciens. 1-4-4. Small RNA Small regulatory RNA first appeared in quorum circuitry in the V. harveyi model, as demonstrated by (Lenz et al., 2004). The discovery of regulatory sRNA among the prokaryotes is very recent (last 5 years). Prokaryotic sRNA activity can be found in the transition to stationary phase, iron homeostasis, and polarity in some operons. A major 31 class of sRNAs works by binding to RNA chaperone, Hfq, followed by base-paring with target mRNAs. This base pairing affects the translation or mRNA decay either positively or negatively, depending on the prevailing conditions and targets (Gottesman, 2004). Although it is highly likely there are many sRNA lurking in the genome, their small size and the requirements of suitable conditions for their expression make it difficult to confirm them. Various approaches using microarrays, bioinformatics, and immunoprecipitation with Hfq have been employed to hunt down and characterize novel orphan sRNAs. Sensitivity to environmental conditions endows small RNA-based regulators the advantage of a shortened adjustment period, which make them appropriate for sensing changes of population density. 1-4-5. Others Some environmental conditions, such as iron and oxygen limitation, promote V. fischeri luminescence at low cell density in monocentrid fish, while also limiting the growth of V. fischeri cells (Dunlap and Kuo, 1992). Also, the production of AI-2 in E. coli was regulated by environmental parameters, such as the availability of preferred carbon sources (DeLisa et al., 2001; Surette et al., 1999) and acid stresses (Kirkpatrick et al., 2001; Stancik et al., 2002) suggesting that the AI-2 mediated quorum sensing is linked to nutritional limitation and stress responses. Other control elements, such as gacA, rsaL, and Vfr were reported along with the specific quinolones produced by P. aeruginosa (Parsek and Greenberg, 2000) 32 1-5. Quorum sensing in enteric bacteria (problem recognition) 1-5-1. Gene phylogenies of the quorum sensing components Despite having a highly annotated genome, understanding of E. coli quorum sensing is not as advanced as it is in less-well characterized species. Currently identified quorum sensing components are; a luxR homolog, sdiA, and the only signal molecule with a known synthetic pathway (AI-2, luxS). It is noteworthy that although hormone-like signal molecule AI-3 was reported from E. coli serotype O157:H7 (EHEC), its synthesis is still luxS dependent. There are no luxI homologs for AI-1 synthesis (Fig. 5). Regarding its functionality (as mentioned above), we now know that AI-2 activates its own up-take machinery encoding genes, lsr operon. The functionality of sdiA is associated with cell-division. However, its expression and functionality are not fully understood and agreed on. Details regarding SdiA are described further in Chapter 3. Three molecular phylogeny studies regarding the genetic components for quorum sensing were reported. From them, we could find information relating to E. coli quorum sensory components. From the two studies regarding the phylogenetic trees of LuxR homologues, E. coli SdiA is reported to be closer to that of Pseudomonas spp., which in turn, are not as closely related to other enterobacterial species as to E. coli in the SSU rRNA phylogenetic tree (Gray and Garey, 2001; Lerat and Moran, 2004; Sun et al., 2004). While cognate luxI and luxR gene pairs seem to have become linked physically and/or functionally early in history, still, many bacterial species contain multiple LuxI and /or LuxR homologues and product or detect distinct signals. 33 Figure 5. The genetic components of E. coli and Salmonella quorum sensing. Currently identified quorum components are depicted. In the quorum system 1, LuxR type transcriptional regulator, SdiA, is identified with its target operon, ftsQAZ. In the quorum system 2, signal synthase LuxS produces AI-2 molecules, which are secreted outside the cells and reenter the cell through ATP binding cassette type-transporter, Lsr. Once entering the cells, AI-2 is phosphorelated by LsrK and binds the LsrR. WhileAI-2- free LsrR represses the lsr structural gene transcript, phospho-AI-2 bound LsrR cannot bind the lsr promoter anymore, derepressing the transcription of the lsr structural genes. The lsrR and lsrK are divergently transcribed from the lsrACDBFGE. Note that E. coli does not have the most distal gene, lsrE, which is present in Salmonella lsr operon. Model figure was adopted and modified from (Ahmer, 2004). LuxS P SdiA LsrR Without phospho-AI-2 LsrR P With phospho-AI-2 LsrB LsrD LsrA LsrC LsrA LsrK lsrACDBFGE lsrKR ftsQAZ and ? 34 Multiple LuxR homologues can also permit the independent activation of different gene functions in response to a single autoinducer signal (Gray and Garey, 2001). The authors? view is that the E. coli luxI component was lost in evolution. As previously mentioned, the absence of signals could not be pivotal for their functionality, especially in mixed cell culture environments. Theoretically, without endogenous signal production, bacteria can still detect the presence of coexisting or competing bacterial species by sensing the environmental concentration of signals. Supporting this notion, Pseudomonas aeruginosa is reported to sense the AI-2 from the other bacterial species (Duan et al., 2003). Also, Ahmer and colleagues have shown that AI-1 originated from other species can modulate Salmonella SdiA activity (Michael et al., 2001; Smith and Ahmer, 2003). However genes are robustly changing their functions in the long run, it is also possible that SdiA may have gained new functions and can recognize new types of signal molecules as effectors present in pure cell culture environments. Two independent phylogenetic trees for LuxS (Lerat and Moran, 2004; Sun et al., 2004), agree with each other showing the ancestral presence of luxS in ? -Proteobacteria. Some cases of incongruence with 16S rRNA based microbial phylogeny can be found from Bifidobacterium longum, Helicobacter pylori, Clostridium acetobutylicum and Borrelia burgdorferi. The orthologs of the S. typhimurium lsr genes were found in most Enterobacteriales as well as in Sinorhizobium meliloti and some Bacillus species (Sun et al., 2004). Unlike the broad distribution of LuxS, AI-2 signal cascades components (Fig. 3.) are present in only Vibrio strains tested by a reciprocal best hit strategy (the periplasmic AI-2 binding protein LuxP, hybrid sensor kinase LuxQ, and the regulator protein LuxU, and the 35 terminal two component response regulator LuxO orthologs). However, the unidirectional best hit search expanded the LuxP homologues search to 28 species with D-ribose binding proteins and 104 organisms both with the hybrid sensor kinase LuxQ, and the two- component regulator LuxO. These studies show the uniqueness of Vibrio species quorum sensing different from that of E. coli and Salmonella. 1-5-2. Enteric environmental conditions Gastrointestinal tracts of the stomach, small intestine, and large intestine are the site of food digestion and the major habitat of E. coli. Both commensal and pathogenic E. coli should pass the acid barriers of the stomach (pH 2) and enter the small intestine, in which pH becomes gradually more alkaline. In the intestinal cavity of the small intestine lower ileum, bacterial cell numbers are commonly about 10 5 -10 7 per gram. In the large intestine, there commonly exist 10 7 facultative anaerobes and 10 10 -10 11 obligate anaerobe cells per gram of intestinal content. Facultative E. coli are present in the intestines but at numbers smaller than obligate anaerobes, such as Clostridium and Bacteroides. The growth rate of bacteria in the lumen is reportedly one to two doublings per day. Intestinal flora has a profound influence on the animal, carrying out a wide variety of metabolic reactions. Of special note are the roles of the intestinal flora in modifying compounds of the bile. Bile acids are steroids produced in the liver and excreted into the intestine via the gall bladder. Their role is to promote emulsification of fats in the diet so they can be effectively digested. Intestinal microbes transform these bile acids (Brock and Madigan, 1991). 36 Purpose of study and approach Quorum sensing is now a common terminology describing sociological group-wise bacterial adaptation to changes in its own or others? cell density. In this chapter, I contrasted quorum systems by categorizing them under the criteria of diversity and universality of the signal molecules. Uses of diverse or universal forms of signals or both may give cells different kinds of advantages, which are not yet understood and is needed to be investigated. Challenges properly describing quorum dependent features from non-pathogenic E. coli were from the absence of ?diversified? signal, AI-1 and its synthase, although they are genetically competent for receiving AI-1 type signals by possessing LuxR homologue encoding gene, sdiA. Does it mean that they are neither to reveal their presence to others nor to count themselves; but rather, only interested in counting the other species? This basic question on the functionality of System 1 was complicated or overwhelmed System 2, which actively produces and takes up its signal molecule, AI-2. System 2 is easily affected by environmental factors, but it is not quite clear ?how?. Looking at this puzzling architecture of the E. coli quorum sensing, comprised of the seemingly incomplete system 1 and the environmentally sensitive system 2, has intrigued me and inspired me to investigate the functionality of system 1 and find, at the molecular level responsible regulators effecting the changes in AI-2 upon stress exposure. Experimental approach to solve the above questions follows below. 1. The functionality of E. coli quorum system 1 was examined by deleting sdiA and performing microarray analysis. It was expected when combined with the microarray 37 results from the overexpression of sdiA, these data could help to identify sdiA- dependent genes, which may begin to explain SdiA function. 2. Different kinds of molecular chaperones and their wild type parental strains were examined for the AI-2 activity under different culturing conditions. By varying environmental stimuli, reproducible effects of chaperone activity on system 2 were expected. 38 Chapter 2: Quorum-sensing under stress-effects of heat shock on Escherichia coli AI-2 signaling Abstract Sensing cell-density and subsequent physiological adjustments are tuned to a global signaling network that receives various cues from the environment. In an effort to identify components linking quorum-sensing to the global sensory system, we examined whether defects in the molecular chaperones, DnaJKE and GroESL, influence AI-2 profiling. We tested the conditioned media from 20 different E. coli strains grown in LB media in order to characterize the kinetics of AI-2 accumulation and depletion, and to link these phenomena to specific molecular chaperones reportedly involved in the folding of quorum sensing transcriptional regulators. The examined E. coli strains include those with null and point mutations in the major chaperones, DnaJKE or GroESL and their isogenic parental strains. Imposing ?high temperature? as an environmental stress to chaperone mutated E. coli cells growing in Luria-Bertani media with glucose supplementation showed uncontrolled AI-2 production beyond the stationary phase, while wildtype cells in the same conditions showed the commonly reported increase and decline exemplified by the growth of cells at 30?C. We present AI-2 profiles in various genetic backgrounds as evidence of active AI-2 decrease towards the stationary phase that is controlled by the molecular chaperone complexes, DnaJKE and GroESL. Key words: AI-2 profile, molecular chaperones, heat shock, glucose 39 Introduction The dynamic physico-chemical conditions of natural habitats place evolutionary pressures on bacteria for the development of sensory signal mediated stress-adjustment processes. Blocking or stimulating such signaling pathways, interwoven as a complex web of interacting genes and proteins, often leads to changes in other signaling pathways. Quorum sensing, signal mediated bacterial assessment of population size in confined environments and subsequently adjusted phenotypic changes in response, is not an exception and is open to cues of not only biotic conditions but also many abiotic conditions. Thus ?cell density? is more likely to be a relative term reflecting the number of cells within a particular environment. A most recent review on quorum sensing can be found in (Waters and Bassler, 2005). Physico-chemical conditions can influence the onset of quorum activation. Examples include the luminescence of the Vibrio species, which was noted to change with the chemical composition of the culture medium. The presence of glucose delayed luciferase gene expression (Dunlap and Greenberg, 1988), while iron restricted conditions caused the earlier induction of luminescence, as compared to fast-growing iron sufficient cultures (Dunlap, 1992). An artificially induced stringent response, a nutrient stress response in Pseudomonas aeruginsa, was shown to induce premature expression of secreted virulence factors such as elastase (Van Delden et al., 2001). One exemplary signal AI-2, produced both from Gram+ and Gram- bacteria, has been suggested as a universal quorum signal molecule and has distinct characteristics in response to the growth conditions. Not alone to the specificity of AI-2 as a quorum 40 signal, its production level and pattern to environmental conditions have been mainly paid attention to. Related with its culminating timing, interesting observations were noted wherein AI-2 was produced early from Campylobacter jejuni, C. coli, Salmonella typhimurium, and E. coli grown in brucella broth at 4 ?C. This observation was media dependent, since the same strains in different food or broth conditions at the similar temperatures did not produce AI-2. The reason for the early appearance of AI-2 activity at 4 ?C is not known (Cloak et al., 2002). However, most of growth condition responsive AI-2 has been reported in a quantitative manner. Surrette and Bassler reported that the presence of glucose, high osmolarity (0.4 M), and low pH (pH5) activated AI-2 production in Salmonella typhimurium LT2. In contrast, low osmolarity (0.1%) and heat shock (43?C) affected AI- 2 production negatively (Surette and Bassler, 1999). Low pH was also shown to induce AI-2 synthase, LuxS, expression in E. coli (Kirkpatrick et al., 2001) and also in Lactococcus lactis (Frees et al., 2003). Our laboratory, using chemostat cultures, has also reported changes of AI-2 activity by various environmental stimuli from E. coli K-12. Namely, AI-2 levels increased with pulsed addition of glucose, Fe(III), NaCl, and dithiothreitol, and decreased with aerobiosis, amino acid starvation, and overexpression of cloned human interleukin-2 (DeLisa et al., 2001). To explain the quorum sensing responding to varied environmental perturbations as mentioned above, many investigators have turned to experimentally examining cross regulation between quorum regulators and other global transcriptional regulators. For example, Cyclic AMP receptor protein (CRP) frequently activates the expression of 41 LuxR family proteins, i.e. V. fischeri luxR (Dunlap and Greenberg, 1988) and P. aeruginosa lasR (Albus et al., 1997). The cAMP-CRP associated repression of quorum sensing genes was reported, less frequently. Repression of expI in Erwinia chrysanthemi by CRP was suggested as a reason for declined signal concentration in the stationary phase (Reverchon et al., 1998). AI-2 profiling of E. coli and Salmonella typhimurium is such that its activity increases along with growth until reaching a maximum in the late exponential phase. When cells enter the stationary phase, the AI-2 levels drop. Now it is known that the disappearance of AI-2 is actively mediated by an ATP-binding cassette type transporter (Taga et al., 2003; Taga et al., 2001). The decreaseo of E. coli AI-2 upon stationary phase can be explained by cAMP-CRP; it not only downregulates the AI-2 synthesis (indirect) but also activates AI-2 uptake machinery expression (lsr operon) (Wang et al., 2005). In Pseudomonas putida, cross-regulation of global response regulator GacA and stationary-phase sigma factor RpoS on the regulation of AHL system was shown (McDougald et al., 2003). Related with this, Affymetrix-based transcriptome analysis of P. aeruginosa further testified that more than 40 % of previously suggested quorum-controlled genes were also RpoS-dependent (Schuster et al., 2003) . Here, we are especially interested in the possibility of molecular chaperones as another type of quorum regulator. Molecular chaperones constitute a group of proteins that stabilize otherwise unstable, non-native states through controlled binding and release thus preventing aggregation [reviewed in (Wickner et al., 1999)]. This group of proteins has multi-functionality for facilitating the correct folding, oligomeric assembly and transport of proteins (and even RNAs) to a particular subcellular compartment, or to a 42 location for disposal by degradation (Aoki et al., 2000; Hartl, 1997; Reid and Flynn, 1996). Most folding of newly synthesized proteins in the cytosol is aided by two different sets of molecular chaperones; a 70-kDa chaperone (also called as heat shock proteins, hsp70s, DnaK) and a 60kDa chaperone family (hsp60s, GroEL). They fulfill distinct roles in protein folding so that Hsp70s and their partner proteins can stabilize translating and/or newly synthesized polypeptides until all segments of the chain necessary for folding are available (co-translational binding), whereas chaperonins (GroESL) promote protein folding in the sequestered environment of their central cavity (Hartl, 1997). The chaperonin, GroEL, along with its cofactor GroES, is the only known chaperone system in E. coli that is essential under all growth conditions (Fayet et al., 1989; Horwich et al., 1993). Various structurally unrelated proteins can be the targets of GroESL. Especially related with stress-induced changes, chaperones are thought to control both intercellular and extracytoplasmid stress activated ? factors, although there is an opposing observation that the control of GroESL was not important for Bacillus subtilis, ? B , a general stress response transcription factor (Scott et al., 1999). A central role of DnaJKE is to down regulate the ? 32 ?dependent heat shock response. ? 32 is one of the principal ? factors regulating heat shock gene expression in E. coli [reviewed in (Bukau, 1993)]. While expression of DnaJKE itself is dependent on heat and ? 32 , a negative feed-back loop exists such that the ? 32 -dependent heat shock response is eliminated by DnaJKE. The function of the mechanism apparently is to maintain optimal levels of the heat shock proteins needed for normal abundance of unfolded proteins (Craig and Gross, 1991) 43 Molecular chaperones have received attention from researchers working in quorum sensing especially those who tried to overexpress LuxR proteins for in vitro experiments or who tried to construct heterogeneous quorum systems using E. coli as host cells. While the former group emphasizes the importance of protease activity of molecular chaperones in non-quorum conditions (Zhu and Winans, 1999), the latter group emphasizes the requirement of GroESL in the conformational activation of LuxR proteins available for quorum activation on the lux operon (Dolan and Greenberg, 1992). In this study, hypothesis was made that molecular chaperones would be a hub interconnecting quorum sensing to other global signaling networks, as shown from its pleiotropic functionality. We examined AI-2 signaling profiles affected by mutations in molecular chaperones. A high temperature condition was imposed to various chaperone mutants and their otherwise isogenic parental strains. Cells were cultured in Luria-Bertani media (LB) or LB supplemented with non-limiting concentrations of glucose. By combining the three factors, chaperones, high temperature, and glucose, we aimed to clarify a general effect of heat shock on AI-2 activity and to see the functional manifestation of the major chaperone as a stress modulator or as a cell process maintainer. Finally, we wanted to see if the experimental method (heat shock or high temperature) could be used as a parameter for discerning the steps on which the major chaperones are exerting their regulatory influence: AI-2 synthesis or uptake/degradation. 44 Material and Methods Strains E. coli chaperone mutants were a gift of Drs. Costa Georgopolous, Debbie Ang, Pierre Genevaux (D?partement de Biochimie M?dicale, Centre M?dical Universitaire, Universit? de Gen?ve, Geneve, Switzerland). Vibrio harveyi reporter strain BB170 was provided by Dr. Bonnie Bassler (Department of Molecular Biology, Princeton University, Princeton, USA). The genotypes of each strain are shown in Table 2 . Media Cells were cultured in Luria-Bertani media (pH 7.4). Glucose supplemented LB media was prepared by adding sterile 20 % (w/v) glucose stock at a 1/25 fold dilution (final concentration of 0.8 % (v/v), 8.88 mM). Culture condition and heat shock Overnight seed cultures were prepared in LB media at 30 ?C. Then cells were freshly cultured either as tube cultures or flask cultures. Tube culture experiments were initiated by 1/50 fold dilution of overnight seeds in new LB media . Polytest tube: 17x100 mm polystyrene test tubes were used (Fisher Scientific C., Inc., Pittsburgh, PA). Each of 20 E. coli strains was dispensed to four 15 ml test tubes with the volume of 2 ml. Two replicate sets of cultures were used to obtain growth curves at either constant 30 ?C or 42 ?C. For flask culturing of E. coli cells, overnight cells were 100-fold diluted to 25 ml LB (250 ml flask culture). Cell culturing conditions were identical to those of tube cultures. Especially for 1 hr heat shock treatment, three replicate sets of tube cultures were prepared (1/50 fold dilution of overnight seeds) and were initially cultured at 30 ?C until temperature upshift at either 3 hours (HS-case study 2) or 7 hours (HS-case study 1) after 45 initial inoculation. The three replicate tube cultures were used for obtaining AI-2 activities of initial cultures without stress, the cultures grown the additional 1 hour at 30 ?C or cultured for the 1hour at 42 ?C, respectively. Temperature upshift was done by transferring each set of E. coli cultures for heat shock into a 42 ?C incubator (air chamber). Cells were placed without shaking during the period (1 hr) and culture tubes were placed on ice for 2 minute to stop the reaction. In this report, three independent experiments were performed for constant temperature cell cultures (30 ?C or 42 ?C) to obtain the AI-2 levels, while 1 hr heat shock experiments were performed just once, but in duplicate. Conditioned media Conditioned media were prepared by centrifugation of E. coli cultures at 16,000xg for 10 min 4 ?C. Supernatants were filtered with 0.2 ?m HT Tuffryn filters (Pall Corp., Ann Arbor, MI). AI-2 assay and reported values The AI-2 activity in the E. coli conditioned media was measured following the method in (Bassler et al., 1997). BB170 was grown for 16 h with shaking at 30?C in AB medium with 10 ? g/ml of kanamycin, diluted 1:5000 in fresh AB medium and aliquoted to sterile 12x75mm tubes (Fisher Scientific C., Inc., Pittsburgh, PA). Cell?free culture fluids were added to a final concentration of 10 % (v/v) to these tubes. Negative controls contained 10 % (v/v) sterile LB or LB plus 0.8 % glucose. Tubes were shaken at 175 rpm and 30 ?C in a shaker (New Brunswick Scientific). Luminescence was measured by quantifying light production as a function of V. harveyi cell density (EG&G Berthold, Gaithersburg, MD). Here, AI-2 levels are reported either as AI-2 fold-induction (normalized luminescence by negative control) or as relative AI-2 fold-induction (normalized luminescence by isogenic parent strains). 46 Especially for 1 hr heat shock experiments, AI-2 levels are reported after two-step normalization to discern the portion attributable to natural changes over time. Calculation was done as follows. When the initial AI-2 activity at time 0 is set as AI2 0 , AI-2 activity after addition 1 hr incubation at 30 ?C as AI2 1, and AI-2 activity after 1 hr incubation at 42 ?C AI2 1 ' , changes by heat and time is AI2 1 ' / AI2 0 , changes by time is AI2 1 / AI2 0 . Net changes by heat shock (relative) is defined to be (AI2 1 ' / AI2 0 )/ (AI2 1 / AI2 0 ) = AI2 1 ' /AI2 1 . 47 Strains Relevant characteristics or Genotypes E. coli C600 F - thi1 thr1 leuB6 lacY1 tonA21 supE44? - MF746 C600 dnaK756 MF634 C600 dnaJ259 CG938 C600 dnaJ259 mini Kan R W3110 F - ? - IN(rrnD-rrnE) rph-1 B178 W3110 galE relA CG712 B178 zjd::Tn10 groES30 CG714 B178 groEL140 CG2245 B178 groES + L + Tn10 CG2244 B178 groES619 Tn10 CG2241 B178 groEL44 Tn10 DA15 B178 phoA::Tn10 DA16 DA15 grpE280 MC4100 (argF-lac)U169 araD139 rpsL150 deoC1 relA1 ptsF25 flbB5501 rbsR F CU247 MC4100 cbpA::Kan R dnaJ::Tn10-42(Tet R ) WKG15 MC4100 djlA::Spc R GP108 MC4100 dnaJ::Tn10-42(Tet R ) GP109 MC4100 cbpA::Kan R GP110 MC4100 djlA::Spc R dnaJ::Tn10-42(Tet R ) GP111 MC4100 djlA::Spc R cbpA::Kan R GP112 MC4100 dnaJ::Tn10-42(Tet R ) cbpA::Kan R GP113 MC4100 djlA::Spc R dnaJ::Tn10-42(Tet R ) cbpA::Kan R V. harveyi BB170 BB120 luxN::Tn5 sensor1 - , sensor 2 + Tet R, tetracycline resistance gene; Kan R , kanamycin resistance; Spc R ,spectinomycin Mutated gene of interest is marked with bold letters Mutations are designated by single letter amino acid code and residue number. Table 2. Bacterial strains used for studying involvement of molecular chaperones. 48 Results and Discussion Chaperones are likely to be involved in nearly every cellular function. We examined whether this involvement could be manifested in AI-2 signaling by examining the effects of mutations in major chaperone, DnaJKE and GroELS, on AI-2 levels. Strains surveyed in these experiments are listed in Table 2. E. coli chaperone mutants have either substituted amino acid residues in the GroESL or DnaJKE complexes or have null-mutations in one or more of the hsp40 homologues, dnaJ, cbpA, and djlA. Two different experimental scales were adopted, test tube cultures (holding capacity of 15 ml) and flask cultures (holding capacity of 250 ml). Conditioned media from the tube cultures were collected at 4 hr and 8 hr after inoculation, while conditioned media from the flask cultures were collected every 2 hr for a comprehensive time course study. AI-2 activity of chaperone mutants w/out heat shock (LB, 30 ?C) Thermally permissive conditions were expected to minimize the role of heat-shock proteins DnaJKE and GroESL on AI-2 signaling. All mutants tested in our study did not show noticeable changes in growth rate when cultured at 30 ?C (Fig. 6). It was noteworthy (and perhaps fortunate) that the mutations examined had minimal impact on cell growth (Fig. 6). This is particularly important as deciphering effects of these chaperone mutants on AI-2 mediated signaling could be difficult, as results might be due to secondary effects of delayed or otherwise altered cell density. When normalized by the AI-2 level of parental strain, both DnaJKE and GroESL mutants showed higher AI-2 49 levels (values >1) in both scales of experiments in the late exponential phase, which corresponded to 4 hrs of tube cultures or 6 hrs of flask cultures. In this permissive condition, mutations in groESL (Fig. 1.C) exhibited significantly increased AI-2 levels, especially at 4 and 6 hr of flask cultures. Similar results were obtained at 6 hr with the dnaJKE point mutants but to a slightly lesser extent (Fig. 6.B). Examining the different combinations of hsp40 knockout mutants, it appeared as though there was no synergism between the effects of ? dnaJ, ? dnaK, and ? grpE, as double mutants and the triple mutant were not distinguished from the single mutants at any of the sampled time points. The lone exception to this was found with the triple mutant, ? dnaJ?cbpA?djlA, at the 4 hr sample of the tube experiments (Fig. 6.A). It is therefore noteworthy that in almost all cases, deletion or impairment of the chaperones investigated resulted in increased AI-2 levels at the times (entr?e to stationary phase) when AI-2 is typically highest. Exceptions to this increase were tube cultures of ? dnaJ at 4 hr, grpE280 at 4 and 8 hr, dnaK756 at 8 hr, groES30 at 8hr and the flask cultures of grpE280 at 6hr. These showed slightly lower levels of AI-2 compared to the controls, however, but these observations did not appear to be systematically consistent. The apparent AI-2 decrease by grpE280 from both tube and flask cultures remains intriguing. 50 Figure 6. The comparison of tube culture and flask culture (30?C). The 20 strains were cultured either in culture tube (upper panels) or flasks (lower panels). Reported data are normalized AI-2 levels by parental control strains. The control strains are represented with black bars in graphs. (A)MC4100 and its 8 derivatives with deleted hsp40s, (B) point mutants of dnaJ, dnaK and grpE. Control strains for dnaJ and dnaK mutants are C600 derivative, while control of grpE is a B178 derivative. (C) groESL point mutants are B178 derivatives (groE+ Tn10 nearby). From the two independent experiments with tube culture and three independent experiments with flask cultures, representative experiment is presented. X-axis represents the time after initiating culturing. Left-y axis represents relative fold induction of AI-2, while right-y axis represents OD600nm. Errors associated with AI-2 measurement were less than 10% of the value. A B C Tube cultures Flask cultures 51 A continuous high-temperature condition affected both growth rate and AI-2 profiling Raising the temperature increases the cellular requirement of chaperone function. This is partly due to the increased rates of growth and certain types of protein occurring at higher temperature or due to the increase of temperature sensitive folding- intermediates of many proteins [reviews in (King et al., 1996)]. Constant or prolonged high temperature condition is considered different from traditional heat shock (typically short time period). In this condition, the majority of DnaJKE would be occupied by structure-damaged proteins rather exerting negative regulatory effects on ? 32 regulon expected immediately following transient heat shock Further, heat-inducible but mutated chaperone expression are assumed to lower the activity on structurally liable proteins, due to the decrease either in substrate-turnover rates or in the interaction with other chaperones or proteases. When the 20 E. coli strains were cultured at higher temperature (42 ?C), growth defects were found from most of the chaperone mutants, which can be ascribed to accumulated toxic protein aggregates in the cells. As shown in Fig. 7, deletion of dnaJ (Fig. 7.A), grpE280 and point mutations in groESL (Fig. 7.C) showed significant growth defects. The triple hsp40 mutant showed decreased turbidity all along the growth profile, particularly after 3 hrs. Increasingly affected growth could be found in the sequence of groES30, groEL140, groEL44, and groES619. Unlike the growth defects by grpE280, the dnaK756 or dnaJ259 mutants were resistant to 42 ?C (Fig. 8.B). It is noteworthy that high temperature cultivation at 42 ?C, of resistant cells increased their growth rates in the exponential phase so that they reached the stationary phase only 4 hr after inoculation 52 (and only 3 hr for the tube cultures, data not shown). This is about 1 hr earlier than the cultures at 30 ?C or without heat shock conditions. Decreases in the growth rate and yield of chaperone mutants were coincident with alterations in AI-2 production. As seen in Fig. 7.A, the deletion in dnaJ had a stronger impact on cell growth than deletion in cbpA or djlA. The lowest growth yield was found from the ? dnaJ?cbpA?djlA strain indicating apparent synergistic effects on growth and yield (Fig. 8.A). Significantly increased AI-2 levels were found from ? dnaJ?djlA, and elevated AI-2 levels were also detected from dnaJ mutants (? dnaJ?djlA >? dnaJ?cbpA> ? dnaJ?cbpA?djlA> ? dnaJ in series). The ? cbpA or ? djlA or the double mutants did not display altered growth and AI-2 production characteristics relative to their controls. Notable growth decreases were observed for grpE280, again yielding profound changes in AI-2 levels. This was in contrast to dnaK756 and dnaJ259. The influence of growth rate was clear with groESL mutant strains, which increased AI-2 levels sequentially groES30, groEL140, groES619, and groEL44. Thus, the increase of AI-2 was apparently inversely-proportionate to the growth rate changes. Growth defects observed from grpE and groESL mutants occurred during the exponential phase (2-4 hrs). Intriguingly, its influence on AI-2 signaling lasted longer and was progressive over the stationary phase (limited by our observation period of 10 hrs). 53 Figure 7. AI-2 from flask cultures at constant 42 ?C (with glucose). Growth defects of the strains were pronounced at 42 ?C (right panels) contrasting to 30 ?C (left panels), which were almost identical to those of flask cultures (data not shown). (A) Growth-defected dnaJ knockout strains showed elevated AI-2 activity from the MC4100 and its 8 hsp40 knockout strains at 42 ?C. (B) Amongst point mutants of dnaJ, dnaK and grpE, only grpE280 showed growth defect and elevated AI-2 level. (C) Growth defects and concomitant AI-2 increases were apparent in groESL point mutants. From three independent experiments with flask cultures, a representative experiment is presented. Errors associated with AI-2 measurement were less than 10% of the value. 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 0 200 400 600 800 1000 1200 0246810 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 0 200 400 600 800 1000 1200 0246810 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 time (hr) AI-2 fold in duction lnOD600nm 30 ?C 42 ?C A B C A B C 54 Application of heat shock to tube cultures (LB) 1 hr duration for heat shock was chosen as a relatively short but effective duration to enunciate shock effects. Surette and Bassler used 2 hr at 43 ?C to cultures previously grown for 6 hr at 30 ?C (Surette and Bassler, 1999). Delisa et al., (DeLisa et al., 2001) upshifted the culture temperature to 42 ?C within a steady-state culture (? =0.75 h -1 ) and continuously monitored AI-2 for the ensuing 4 hr. In both studies, lowered levels of AI-2 were attributed to the applied heat shock. Here, we report that a culture?s growth state, during which the shock is applied, affects the subsequent AI-2 profiling. We present two case studies, performed at the exponential phase or at the stationary phase. Case study of heat shock 1: heat shock in the stationary phase increases AI-2 activity (slowed decrease in AI-2 over time) To distinguish changes due to heat shock from ?normal? changes along typical batch growth, AI-2 activities of a strain at three conditions were measured and compared: base case culture without stress of cells cultivated for 3 hr (denoted AI2 0 ), the same culture grown an additional 1 hr at 30 ?C (denoted AI2 1 ) or the same cells cultured for an additional 1 hr at 42 ?C (denoted AI2 1 ' ). Heat shock of the stationary phase cultures (at 7 hr post inoculation in tubes) resulted in cells maintaining a higher level of AI-2 activity than the cells kept at 30 ?C for the same duration (expressed as AI2 1 ' /AI2 1 ). During this time, control cells take up AI-2 and the levels naturally decline over time (expressed as AI2 1 / AI2 0 ). We observed that heat shock increased the AI-2 levels both from the wild type and chaperone mutants. Changes in AI-2 among the parental strains are depicted in Fig. 8. The higher AI-2 levels observed after heat shock agreed with the general increase shown in cultures kept at 42 55 ?C for the entire culture (shown previously in Fig. 6 and 7). Heat shock induced AI-2 increases were observed during the stationary phase from point mutated chaperones (B178 and C600 and their derivatives). MC4100 and its derivative hsp40 knockouts showed no differences due to the heat shock, which is surprising, considering previous observations of increased AI-2 with ? dnaJ mutants (both at 42?C or 30 ?C regardless of presence of glucose). There was no hierarchical dominance of dnaJ over cbpA and djlA in this decrease. The relative insensitivity of dnaJ mutants from the control is intriguing. As an explanatory scenario, presuming that AI-2 synthesis is negatively targeted by chaperones, the chosen time point of temperature upshift (at 7 hr) may have been too late to see a ?chaperone effect?. That is, if the chaperones have already worked on the AI-2 synthesis machinery, then a subsequent heat shock would be of no consequence to the AI-2 profile. This appears to have been the scenario, hence we believe that the involvement of the chaperones is on the synthesis pathway as opposed to the depletion pathway. We note, however, ? dnaJ?cbpA?djlA was resistant to the decrease over the 1 hour (Table 3), which implies that when combined, Hsp40s have synergistically affected AI-2 uptake to a certain degree. B178 pheA::Tn10 ? C600 thr::Tn10 and their derivatives also showed decreases in AI-2 over time (AI2 1 / AI2 0 ), but had higher levels of AI-2 after the heat shock (AI2 1 ' /AI2 1, AI2 1 ' /AI2 0 ). From the C600 series, a decrease in AI-2 over time was not shown from the two dnaJ259 mutants. Higher AI-2 activity after heat shock is clear from dnaK756, and grpE280. The grpE280 increased more than 56 % than that of its parental cells. The groESL mutants showed increase of AI-2 from 6 to 60 %. 56 0 50 100 150 200 250 MC4100 C600 thr::Tn10 B178 pheA::Tn10 B178 B178 +Tn10 A I - 2 f o l d i ndu ct i o without HS with HS Figure 8. AI-2 activity before and after 1 hr heat shock to stationary phase cells. AI-2 activity of the stationary phase parental cells (precultured 7 hrs) was compared before and after the shock. Over the 1hr of temperature-shift, parental control cells showed increased AI-2 levels (single experiment based data). 57 Genotypes Change by time (AI2 1 / AI2 0 ) Change by HS over time (AI2 1 '/ AI2 0 ) Net Change by HS (AI2 1 '/ AI2 1 ) w/t (MC4110) 0.59 0.67 1.14 ? dnaJ 0.73 0.83 1.15 ? dnaJ? cbpA 0.63 0.58 0.92 ? dnaJ? djlA 0.80 0.81 1.02 ? dnaJ ? cbpA ? djlA 0.85 1.01 1.18 ? cbpA 0.62 0.81 1.30 ? djlA 0.59 0.61 1.03 ? cbpA ? djlA 0.68 0.70 1.03 w/t (B178 pheA::Tn10) 0.58 1.07 1.84 grpE280 0.69 1.97 2.88 w/t (C600 thr::Tn10) 0.53 0.77 1.45 dnaK756 0.53 1.13 2.11 dnaJ259 0.99 0.70 0.70 dnaJ259 mini-kanR nearby 1.05 1.23 1.16 w/t (B178) 0.68 1.53 2.26 w/t (B178 +Tn10) 0.98 1.14 1.16 B178 groES30 Tn10 nearby 0.59 1.13 1.91 GroES619 0.80 1.00 1.24 GroEL140 0.69 1.10 1.59 B178groEL44 Tn10 nearby 0.88 1.28 1.46 Table 3. Changes of AI-2 activity by 1 hr heat shock at stationary phase. Relative AI-2 changes were reported as normalized values. Wild type parental strains are marked with boldic letters. Net change ratio by heat (AI2 1 ' /AI2 1 ) is marked with bold letters. AI2 0, initial AI-2 activity at 30 ?C (at time 0), AI2 1,: AI-2 activity after addition 1 hr incubation at 30 ?C (at time 1hr), AI2 1 ' : AI-2 activity after 1 hr 42 ?C heat shock (at time 1hr) (single experiment based data). 58 Case studies of heat shock 2: heat shock in the exponential state Heat shock to exponentially growing cells either decreased AI-2 or showed no effects (Table 4). The change patterns in AI-2 level of the parental strains are depicted in Fig. 9. Examples of chaperone mutants following the pattern of parental controls could be found in MC4110 and its hsp40 knockouts and C600 and its dnaJK point mutants. In contrast, grpE280 showed resistance to the decreases otherwise shown in its parental control. Looking closely at the AI-2 decreases of MC4100 and its derivative strain cells, the ? dnaJ?cbp?djlA mutant did not decrease AI-2 levels (only 2 % decrease) as it didn?t in the stationary HS experiment (Table 3). This again implies a synergistic effect on uptake or degradation machinery delaying the normally-hastened AI-2 uptake by heat shock at this growth state. Combined with ? dnaJ, deletion of cbpA (? dnaJ?cbp, 10 % decrease) seems to have contributed the higher AI-2 levels after the HS than djlA (? dnaJ ? djlA 30 % decrease), however, alone as ? cbp and ? djlA, they did not differ (c.a. 50 % decrease). Point mutations in groESL showed clear decreases more by heat shock than by time, while the parental strain did not change its AI-2 level by either time or heat shock. The AI-2 decrease by temperature upshift (exponential phase) described here was later found to be a general phenomenon of heat shock effects in LB glucose cultures (below). Various patterns in AI-2 levels were observed among the chaperone mutants. No differences could be attributed to dnaJK point mutations, groESL point mutations, or to ? dnaJ?cbp?djlA. Of course, this set of experiments only tests a short time period and at one condition. Thus, differences here might not be uniquely ascribed to the mutations tested. That is, there are other conditions wherein the mutations might be more relevant. In this 59 context, the clearer differences seen in stationary phase HS appear to be associated with the required chaperone functionality. Stationary phase induced chaperon activities were reported (Dukan and Nystrom, 1998). 60 0 50 100 150 200 250 300 350 400 450 MC4100 C600 thr::Tn10 B178 pheA::Tn10 B178 B178 +Tn10 A I - 2 f o ld in d u c t i o without HS with HS Figure 9. AI-2 activity before and after 1hr heat shock to exponential phase cells. AI-2 activity of the exponential phase parental cells (precultured 3 hrs) was compared before and after the shock. Over the 1hr of temperature-shift, parental control cells showed increased AI-2 levels (single experiment based data). 61 Genotypes Change by time (AI2 1 / AI2 0 ) Change by HS over time (AI2 1 '/ AI2 0 ) Net Change by HS (AI2 1 '/ AI2 1 ) w/t (MC4100) 1.02 0.57 0.56 ? dnaJ 1.02 0.68 0.67 ? dnaJ? cbpA 0.93 0.84 0.90 ? dnaJ? djlA 1.16 0.81 0.70 ? dnaJ ? cbpA ? djlA 0.97 0.95 0.98 ? cbpA 1.34 0.70 0.52 ? djlA 1.03 0.52 0.51 ? cbpA ? djlA 0.91 0.63 0.70 w/t (B178 pheA::Tn10) 0.89 0.66 0.74 grpE280 0.87 0.87 1.00 C600 thr::Tn10 0.82 0.84 1.03 dnaK756 1.23 1.12 0.91 dnaJ259 0.79 0.91 1.15 dnaJ259 mini-kanR nearby 0.88 0.93 1.05 w/t (B178) 0.94 0.96 1.02 w/t (B178 +Tn10) 0.78 0.71 0.91 B178 groES30 Tn10 nearby 0.98 0.95 0.97 GroES619 0.78 0.58 0.75 GroEL140 0.98 0.54 0.55 B178groEL44 Tn10 nearby 0.80 0.50 0.63 Table 4. Changes of AI-2 levels by 1 hr heat shock at exponential phase. Relative AI-2 change values were reported. Wild type parental strains are marked with bold letters. Net change ratio by heat (AI21' /AI21) is marked with bold letters. AI20: initial AI-2 activity at 30 ?C (at time 0), AI21: AI-2 activity after addition 1 hr incubation at 30 ?C (at time 1hr), and AI21? : AI-2 activity after 1 hr 42 ?C heat shock (at time 1hr) (single experiment based data). 62 Glucose effects: Presence of glucose at high temperature pronounced the chaperone activity Glucose has been shown to increase AI-2 level and its period of sustained increase in conditioned media (Surette and Bassler, 1998). Our lab has also have noted the global impact of glucose in the transcriptome [microarray analysis (Wang et al. in preparation)], aspiring to reveal possible glucose-specific-effects. At the same time, we tried to discern glucose effects from AI-2 profile changes among chaperone mutations. For this, cells were grown in 4 different conditions: in the presence or absence of glucose combined with different culturing temperature of 37 ?C or 42 ?C. Conditioned media from the cultures were collected every 2 hours for AI-2 measurement (Fig. 10). Of the 4 conditions, the strongest effects of chaperone mutants were found at 42 ?C with glucose. Growth rate affected chaperone mutants (dnaJ knockouts, grpE280 and groESL mutants) progressively increased AI-2 toward the stationary phase (Fig. 10. A.and C). The progressive increase in AI-2 is not clear at the other three conditions; when glucose is not supplemented or when the temperature is low or when those two conditions are combined. At 42?C, without glucose, complete depletion of AI-2 was observed within the time range studied from the wild type strains. Sustained AI-2 activity of hsp40 mutants and groESL point mutants was clear (Fig. 10.C), which is not seen in the dnaJKE point mutants (Fig. 10.B). Despite the affected growth rate of dnaJ259, its AI-2 activity did not differ from that of the parental strain. We have marked this as an exceptional case of the general idea that the increase of AI-2 is inversely-proportional to the growth rate. At 30?C, without glucose supplementation, no significant growth rate changes were found. Only dnaJ knockouts showed higher level of AI-2 and its slow disappearance, 63 while point mutants of dnaJKE did not differ much from the parental strains in the AI-2 profiles. The most noticeable difference was found at 10 hr and at this time, again, the dnaJ259 mutant showed the biggest effects. The groESL mutants showed generally 1-3 fold higher AI-2 levels along growth, especially at 8 hr, where the biggest difference was seen (~10 to 20 fold). 64 A. Figure 10. The effects of glucose and temperature on AI-2. (A) MC4100 and its 8 derivatives with deletions in three hsp40s. (B) Point mutants of dnaJ, dnaK and grpE. Control strains for dnaJ and dnaK mutants are C600 derivative, while control of grpE is a B178 derivative. (C) The groESL point mutants are B178 derivatives (groE + Tn10 nearby). Culture conditions are indicated. From three independent experiments with flask cultures, a representative experiment is presented. Errors associated with AI-2 measurement were than 10% of the value. * Figures are presented in the other section. 30 C w/o glc 0 20 40 60 80 100 120 140 160 180 200 0246810 -4 -3 -2 -1 0 1 2 30 C with glc * 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 42C w/o glc 0 20 40 60 80 100 120 140 160 180 200 0246810 -4 -3 -2 -1 0 1 2 42 C with glc * 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 time (hr) AI-2 fo ld ind u ctio n ln OD 600 nm 30 C w/o glc 30 C with glc * 42 C w/o glc 42 C with glc* 65 B. Figure 10. The effects of glucose and temperature on AI-2 (contd.). B) Point mutants of dnaJ, dnaK and grpE. Control strains for dnaJ and dnaK mutants are C600 derivative, while control of grpE is a B178 derivative. time (hr) AI-2 fo ld ind u ctio n 30 C w/o glc 0 20 40 60 80 100 120 140 160 180 200 024 6810 -4 -3 -2 -1 0 1 2 30 C with glucose * 0 200 400 600 800 1000 1200 0 2 4 6 8 10 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 42 C w/o glc 0 20 40 60 80 100 120 140 160 180 200 0246 8 10 -4 -3 -2 -1 0 1 2 42C with glc * 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 ln OD 600 n 66 C . Figure 10. The effects of glucose and temperature on AI-2 (contd.). C) The groESL point mutants are B178 derivatives (groE + Tn10 nearby). 30 C w/o glc 0 20 40 60 80 100 120 140 160 180 200 0246810 -4 -3 -2 -1 0 1 2 42 C w/o glc 0 20 40 60 80 100 120 140 160 180 200 0246810 -4 -3 -2 -1 0 1 2 42 C with glc * 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 30 C with glc * 0 200 400 600 800 1000 1200 0246810 -4 -3 -2 -1 0 1 2 time (hr) AI-2 fo ld ind u ctio n ln OD 600 nm 30 C w/o glc 30 C with glc* 42 C w/o glc 42 C with glc* 67 Presented data for the changed AI-2 profiles elucidate the importance of chaperone in the control of AI-2 toward the stationary phase. Especially when contrasted with parental cells with wildtype molecular chaperones in the high temperature glucose culture, chapareone mutants showed a surprising and preserved pattern of AI-2, progressively increasing AI-2, while wild type cells reduce AI-2 levels under the same condition The results presented in this study provide clear evidence that the decrease of AI-2 shown in the stationary phase requires the activity of molecular chaperones, DnaJKE and GroESL complex (Fig. 10). Interestingly, we view this decrease in AI2 concentration (fold activity) as a function of altered synthesis rate. When the involved chaperones are mutated, the AI-2 level seems to increase unabated well into the late-exponential and stationary phases. Glucose drives the metabolic flow towards AI-2 synthesis. We cannot pinpoint exact major target proteins under the regulatory effects of the chaperone from the methodology chosen here (screening AI-2 profiles). The observed phenomena could be attributed to one or a multiple number of proteins that are influenced by the chaperones. When we couple AI-2 profiles of wildtype cells and mutants grown under standard conditions to the same cultivated under heat shock, we can clearly conclude that the chaperones are negatively regulating the AI-2 synthesis so that AI-2 is not accumulated at stationary phase culture media. Since the concentration of glucose (0.8%) used in this study is high enough to totally inhibit the AI-2 uptake machinery (lsr operon), the chances of defects in AI-2 uptake machinery by the chaperone mutants are rare. 68 It is obvious from the results obtained here, that the level of AI-2 and the direction of AI-2 accumulation depend on chaperones and on the prevailing metabolic state of the cells. At stationary phase, heat shock increased AI-2 levels in the conditioned media and especially did from some of the chaperone mutants. During exponential phase, heat shock reduced AI-2 levels; however clear differences from the chaperone mutants were limited. Hence, the changes in AI-2 were different at the different states. This is entirely consistent with a recent microarray study by Wang et al, wherein they found discordant sets of genes that were affected by LuxS mutation when tested in the presence/absence of glucose at different ODs. Thus, perhaps it is now understandable, that the cells use quorum sensing signaling in a finely tuned and regulated manner, wherein the genes regulated and the signals perceived vary, but the prevailing environmental conditions, which themselves are ever changing. 69 Chapter 3: Transcriptome analysis of SdiA mutation; potential for AI-1?AI-2 crosstalk in Escherichia coli Abstract Enteric bacteria Salmonella and Escherichia coli has genetic potential to deploy a cell density dependent response with common tools, LuxI type transcriptional regulator, SdiA (system 1), and seemingly independent AI-2 signal synthase (system 2). While signal synthesis and uptake machinery of the system 2 has been widely studied, absence of identified signal and its synthase for the system 1 has suspended a consensus on its functionality. In this work, comparative analysis of transcriptome was performed between E. coli W3110 and its otherwise isotype sdiA mutant collected at late exponential phase in Luria-Bertani media. Using E. coli K-12 Affymetrix microarray, significant changes (cutoff of 2 fold) were found from small numbers of genes and IGs by the mutation, with notable reduction in rRNAs levels counterbalanced by induced purine de novo synthesis genes and a reduction in AI-2 uptake machinery transcripts (lsr operon). Additionally reporting increased resistance to general stresses and decrease in motility of the mutant, we also presented consensus sequences as possible binding motifs and predicted small RNA expression based on the information of sub-transcript level of changes in the intergenic regions (IGs). Contrasting with the most of the previous studies focused on SdiA overexpression, this work investigated the effects of sdiA deletion systemically for the first time demonstrating the the possibility of its functionality in the pure culture condition especially on the quorum system 2. Key words: SdiA, lsr operon, rRNA, purine de novo synthesis. 70 Introduction Many prokaryotic species sense neighboring populations, adapt their physiological states to environmental cues, and trigger group-wide or coordinated physiological processes in a process termed ?quorum sensing?. Quorum sensing involves an array of external signaling molecules (from peptides to furanones), several internal regulatory molecules (from proteins to small RNA), and an assortment of regulatory motifs (from direct signal binding of transcriptional regulators to hierarchical phosphorelay cascades). Among the most well known quorum regulator proteins are the LuxR-type proteins that have specificity toward acylated homoserine lactones (AHL), referred to as autoinducer-1 (AI-1). AI-1s consist of diverse acyl-side chains that are linked to S-Adenosylmethionine (SAM). There are currently three known protein families catalyzing AI-1 synthesis, including the well-studied LuxI-family, AinS, and HdtS. The signal synthase/regulator pair, LuxI-LuxR, are conserved among many bacterial species and are considered to have co-evolved (Lerat and Moran, 2004). Numerous examples of AI-1 synthase/regulator pairs can be found in organisms such as Vibrio fischeri (LuxI-LuxR, AinS-AinR), Pseudomonas aeruginosa (LasI-LasRI, RhlI-RhlR), Agrobacterium tumefaciens (TraI- TraR), Pantoea stewartii (EsaI-EsaR), Rhodobacter sphaeroides (CerI-CerR), Vibrio anguillarum (VanI-VanR), Yersinia enterocolitica (YenI-YenR) among others [Review in (Fuqua et al., 2001)]. The enteric bacterium Escherichia coli also has a LuxR homologue named sdiA (suppressor of division inhibition, 67 % aa similarity with 27 % identical aa compared with V. fischeri LuxR). SdiA was initially cloned and characterized as a factor restoring a cell division defect of ftsZ84 (fts: filamentation temperature sensitive) upon 71 overexpression (Wang et al., 1991). Escherichia coli and Salmonella appear to be distinct from the above group of bacteria, because they do not have a luxI homologue (Lerat and Moran, 2004). Experiments with several different sensor cells were indeed unsuccessful in detecting AI-1 derivatives from E. coli. Instead, E. coli has components associated with an alternative pathway proposed as a cross-species quorum sensing system. This signal molecule, autoinducer AI-2, is synthesized by three enzymatic reactions involving a set of methyltransferases, a nucleosidase (Pfs) and AI-2 synthase (LuxS). Consistent with many QS signal molecules, the AI-2 concentration in the media of E. coli cultures increases with growth until the late exponential phase. Surprisingly, E. coli and Salmonella actively uptake AI-2 and decrease its level in the culturing media. Conserved among Enterobacteriales, lsr operon (lsr: luxS regulated) encodes a ABC-type transporter and subsequent modification enzymes controlling the consumption of AI-2 (Chen et al., 2002; Sun et al., 2004; Taga et al., 2003; Taga et al., 2001; Wang et al., 2005). Despite the absence of AI-1 signal molecules in CM from E. coli, unknown signals to control SdiA activity have been speculated for over a decade. For example, unidentified signals present in stationary phase enterohaemorrhagic E. coli (EHEC) 0157:H7 DMEM (serum-free Dulbecco?s modified Eagle medium) CM were shown to bind SdiA and negatively regulate the expression of virulence factors (Kanamaru et al., 2000). Chemically synthesized AHLs were shown to activate SdiA mediated ftsQ 2 p transcription (ftsQAZ cluster, fts: filamentation temperature sensitive) (Sitnikov et al., 1996). Also, SdiA expression itself was affected by exogenously added quorum signals. Depending on the experimental conditions, increased sdiA expression was reported from 72 AI-2 synthesizing E. coli O157:H7 DMEM cultures (11.6 fold compared with AI-2 non- producers) (Sperandio et al., 2001) and E. coli W3110 exposed to CM with an 800-fold differential in AI-2 concentration. (2 fold) (DeLisa et al., 2001). Conflicting observation of decreased sdiA transcript by CM addition was reported from a study with stationary phase CM of S. enterica serovar. Typhimurium (30 %) (Volf et al., 2002). These studies have clearly failed to delineate the functional role of SdiA in quorum sensing systems and, while not intended, also did not find SdiA-regulated genes beyond ftsQAZ. Ahmer and colleagues (Ahmer et al., 1998) adopted a genetic screen for SdiA dependent genes in Salmonella typhimurium (using MudJ transposon mutations) where they found three chromosomally located genes that were differentially controlled by plasmid-encoded SdiA (Michael et al., 2001). None had sequence similarity to E. coli genes. Interestingly, none of these genes were differentially regulated by SdiA when sdiA was expressed from its natural position in the chromosome. The remaining of seven fusions were located within an operon on a virulence plasmid (Ahmer et al., 1998; Michael et al., 2001; Smith and Ahmer, 2003). Also, Wei et al. performed a transcriptome analysis using PCR-based spotted microarray slides to determine the consequence of overexpressed sdiA in exponentially growing cells (OD 600nm ? 0.45) (Wei et al., 2001). Their study was preceded by their discovery of increased resistance to DNA damaging mitomycin C (MCM) of E. coli by sdiA amplification (Wei et al., 2001) and thus focused on changes in acridine efflux pumps that relate to MCM resistance. Herewith, we have attempted to unveil the functionality of sdiA when expressed from the natural chromosomal position instead of using plasmid vectors. That is, since overexpressed SdiA is aggregate-prone(Kanamaru et al., 2000; Yamamoto et al., 2001), 73 our approach which employs gene knockout, was expected to reduce possible interference from the stress response associated with protein overexpression (Bentley et al., 1990). Indeed, while culture conditions and assay formats were different from Wei et al, distinctive changes in transcriptome profile were observed and reported here, with an notable evidence for the cross-talk between the AI-1 and AI-2 quorum sensing circuits in E. coli. Phenotypic changes associated with the sdiA deletion were presented ranging from increased tolerance toward general stresses to decreased motility. Finally, utilizing the antisense probe format spanning intergenic regions and partly trying to link the transcriptome and phenotypes, expression of novel small RNA or possible promoter region was predicted. 74 Material and Methods Strains and growth conditions. The strains used in this study are derivatives of E. coli K-12, W3110 (Table 5). P1 vir transduction (Sternberg and Maurer) was done to transfer sdiA::kan R from WX2 (Wang et al. 1991) to W3110 and ZK126 (W3110 lac - ) chromosomes, and the new mutant strains were named W2X2 and ZKWX2, respectively. Successful transduction was checked by antibiotic resistance (sdiA::kan R ) and the change in genomic DNA surrounding the sdiA using PCR and the absence of sdiA mRNA expression (RT-PCR). Overnight cultures were prepared in Luria-Bertani media (pH adjusted to 7.4 with NaOH?HCl) with 10 ? g/ml kanamycin and/or 100 ? g/ml ampicillin according to Sambrook (Sambook and Russell 2001). The overnight cultures were diluted 1/100 fold to initiate fresh cultures until OD 600nm reaches to be 1.8 at 37 ?C at 250 rpm (250 ml flask, 50 ml vol of cell culture). Note that kanamycin was only used for maintaining W2X2 (in O/N culture), and was not used in experiments. LB agar plates were used for E. coli colony counting in shock experiments and L- marine (LM) agar plates were used for V. harveyi BB170 colony counting in AI-2 assay. Media composition: RNA extraction and quantification of mRNA transcript 200 ? l of cell cultures samples were immediately treated with RNA stabilizer (400 ? l, Qiagen, Valencia CA) and stored at -80 ?C before the RNA extraction. RNeasy kit was used with DNase treatment to eliminate chromosomal DNA contamination (Qiagen, Valencia CA). Isolated RNA was resuspended in diethylpyrocarbonated (DEPC)-treated water and the 75 RNA concentration was measured by UV absorption (260nm). RNA quality and the rRNA quantity of the total RNA were checked by electrophosis in a 1.2 % (w/v) of agarose gel with formaldehyde (220 mM). The initial quantification of mRNA of sdiA along the time course was done by using 0.05 ?g of total RNA as a template for reverse transcription using 40 units of SuperScript III TM RT and 50 ng of random-primer. Samples for quantitative real-time PCR was prepared with qPCR SYBR?Green mix (Invitrogen, Calsbad CA). Fluorescence intensity was recorded in time using GeneAmp 5700 (Applied Biosystems, Foster City, CA). Time course data was normalized by making a ribosomal protein expression a constant (internal control, rpsK). The primers used for sdiA expressions are; 5?-GAATTTCTTTTTCGCGCTTG-3? and 5?-TTTTTGTCCTTTCCCGTTG-3?. Affymetrix microarray analysis and data mining Antisense GeneChip? arrays were used as specified by the manufacturer. Two independent experiments were performed for each strain, W3110 and W2X2. Total RNA (10?g) was used for synthesis of cDNA. Subsequent steps (cDNA converstion, RNaseI treatment, biotinylation, hybridization and scanning) were performed at the CBR Microarray facility. Image files were then analyzed using Gene Chip Operation System (GCOS version 1.12) (Affymetrix, Inc., Santa Clara, CA). (http://www.umbi.umd.edu/~cbr/macore/macorestart.htm) For detection call analysis, each of the two replicates needed to get present called. Detection calls in Affymetrix chips are assessed by probe pair saturation and calculation of a detection p-value. Default values of signal threshold (Tau 0.015) for a discrimination 76 score, R= (PM-MM)/(PM+MM) and cut-offs of significance p-values (reflecting the confidence of the detection call, ? 1 =0.04 and ? 2 =0.06) were used in this study. For differential call analysis, 4 pairwise batch analyses were performed for W3110 (baseline) and W2X2 (experiment). The results were averaged and reported as log 2 IR (Induction Ratio). From the genes with differential calls, absolute Log 2 IR values of >1.0 was used for ?significantly? changed genes and values between 0.5 and 1 for ?moderate? transcript changes. Genes with absoulte log 2 IR < 0.5 were denotated to be with marginally altered. Description of probe set design: Attempted to lower apparent signal loss by averaging over large regions, individual probe intensities of selected genes and intergenic region were interrogated. For each probe, corrected intensity (CI) was calculated by subtracting the value of mismatch (MM) probe from the value of the corresponding perfect match (PM) probe (PM-MM values). Bad probes are defined when the signal of MM is bigger than that of PM (CI<0) and were given a value of 0. Results from the biological duplicate experiments were averaged and then relative induction ratio (IR: CI W2X2 / CI W3110 ) was reported in log 2 scale (log 2 IR). High resolution graphs of expression- and change patterns across ORFs and IGs were used to determine the expression of small RNA and promoter. In silico consensus motif search To equally represent 5?UTR of each promoter region, each coding sequence of IG was chosen to cover 200 bp upstream of translation start sites (except ftsQ 2p ) as inputs in an in silico study using BioProspector. Two block model was adopted with a spacer from 8 to 10 bp between the blocks. 77 Analytic assays ? -galactosidase assays were performed following the method of Miller (Miller, 1972). Transcriptional fusion reporter for lsrA is a derivative of pFYZ1 (Wang et al., 2005) (Table 1). Miller Units =1000 x [(OD 420 -1.75xOD 550 ]/ (TxVxOD 600 ). AI-2 levels were determined following Bassler et al. (Bassler et al., 1993). Briefly, E. coli conditioned media were collected by removing cells with centrifugation (for 10 min of 16000xg at 4?C) and subsequent filtration (Miller pore, 0.2 ? m). 20 ? l of E. coli CM was added to 180 ? l of freshly diluted overnight cultured BB170 (D=1/5000, AI-2 reporter strain) to initiate culturing at 30 ?C with aeration (175 rpm). Light production was measured with a luminometer (EG&G Berthold, Gaithersburg, MD) and the reported AI- 2 levels were normalized by counts of Vibrio cells. Phenotypic characterization Shock resistance Late exponential state cells were exposed to different shock conditions, oxidative, acidic, and thermal shock. For oxidative and acidic shock, 10 ? l of cells were placed to 1ml of PBS with H 2 O 2 (from 5mM to 40 mM) and LB (pH 3.0, adjusted with HCl), respectively. At each sample point, 100 ? l of shocked cells were withdrawn, immediately serially diluted by PBS (pH 7.5), and spread on LB agar plates. For heat shock, cells (40? l) were loaded into 1.5 ml Eppendorf tubes and placed into a 55 ?C water bath. At indicated time points, aliquots were withdrawn and immediately mixed with room temperature PBS (pH 7.5) and then serially diluted before spreading onto LB agar plates. All the viable colony counting was done after 24 hr incubation of the plates at 37 ?C. 78 Autoaggregation To see bacterial auto-aggregation, late-exponentially growing cells were transferred to 15 ml falcon tubes and stored statically at 4?C. Cells were stored for 12 hrs and 1 week before taking the image pictures. Motility Swimming motility was examined by swimming zone on soft agar plates (0.3 %). Late exponential phase broth cultures were placed on top of the plates and then stored at 37 ?C chamber for 12 hrs. SEM (scanning electron microscopy) Images for late exponential phase broth culture cells were taken. For this, cells were collected and gently washed three times with PBS buffer (pH 7.4, centrifugation by 2000g for 10 min) and then fixed with 2 % glutaldehyde (1hr at RT and 9 hr at 4 ?C). Cells were collected with 0.2 ? l filters and the residual glutaldehyde was washed out with PBS buffer three times before the filters were dehydrated with ethanol (70 %, 95 % and 100 %). The filters were fully dehydrated with Denton Vacuum freezer (Denton DCP-1 critical point dryer) and coated with Ag-Pd (Denton DV 502/503 vacuum evaporator). Coated filters were examined under EM (Model). This process was performed in the Biological Ultrastructure Laboratory in UMCP. 79 Table 5. Bacterial strains and a plasmid used for the transcriptome analysis. Strains or plasmid Characteristics Source E. coli W3110 WX2 W2X2 ZK126 ZKWX2 V.harveyi BB170 Plasmid F - ? - rph1 ? lac-pro met pro zzz::Tn10 thy supD r K - m K sdiA::kanR W3110 sdiA::kanR W3110 ? lacU169 tna-2 ZK126 sdiA::KanT BB120 luxN::Tn5 (sensor 1 - , sensonr 2 + ) pLW11 (lsrA fusion, pFYZ1 derivative) Lab stock Rothfield This study This study B. Bassler Lab stock 80 Results and Discussion Phenotypic changes from sdiA null mutation Growth and colony morphology As previous researchers linked SdiA to a cell division defect, we examined W3110 sdiA - for growth defects and other growth-related phenotypic changes. E. coli W3110 and W3110sdiA - (W2X2) cells were grown identically in LB medium (Fig. 11.A). When we examined the wild type sdiA expression level over the course of a typical culture using quantitative RT-PCR, an apparent mid-exponential peak was observed (OD 600 1.8, Fig. 1.A); expression remained high thereafter. Based on this observation, we performed phenotypic characterization of the sdiA mutant at the same growth phase, late-exponential state. Colony shape and cell morphology under a light microscope did not differ between E. coli W3110 and W3110 sdiA - (W2X2) cells (Fig. 11.B). Swimming and SEM Swimming motility, however, was significantly attenuated in W2X2 cells innoculated onto soft agar plates (0.3 %) (Fig. 12.C) and in the LB broth (Fig. 12.B). Also, while cells showed no apparent sign of defects in division under the microscope, SEM images revealed clear differences between the two strains such that W3110 had significantly more flagellar or fimbrial appendages as well as a more developed extracellular matrix (ECM) (Fig. 12.D.E). Autoaggregation We also noticed that LB cultured W2X2 cells immediately settled to the bottom of centrifuge tubes within an hour, while W3110 cells remained suspended in the broth. After long-term static storage, W3110 cells eventually settled down to the bottom along with a slimy layer compressed on top of cell pellet, which is not shown in W2X2 culture (Fig. 11.F, right panel). It is highly likely that the extracellular matrix 81 delayed direct cell-cell contact, which is mediated by short appendage protein such as Agn43. E. coli autoaggregation was reported to be only possible in the minimal media but not in LB media. 82 2 ? m 2 ? m C W3110 W2X2 W3110 W2X2 12 hr 12 hr 2 week 2 week B W2X2 E W3110 W2X2 W3110 A 0 2 4 6 8 10 12 14 16 18 0123456 time (hr) AU -4 -3 -2 -1 0 1 2 l n O D 600 n m 83 Figure 11. Macroscale phenotypic differences. (A) Cell growth and expression of the sdiA gene. Growth curve for W3110 and W2X2 in LB media (semi-log graph: X-axis culture indicates sampling time and y-axis indicates logOD600nm). Quantitative Real time PCR was performed to see the expression pattern of sdiA along the time course. One of ribosomal protein-encoding gene (rpsK) was used as an internal control and used for normalizing the sdiA expression at different time points. Y-axis represents relative abundance (arbitrary units). (B) Colony morphology W3110 and W2X2 were cultured on LB (upper panel) and McConkey (lower panel) agar plates for 3 days at 37?C. (C) Autoaggregation of W2X2. Late exponential phase W3110 and W2X2 cultures were statically stored at 4 ?C for 12 hrs (left panel). Same tubes after prolonged incubation (1week, right panel) showed the slimy top on the pellet. Autoaggregation of W2X2 is visible within an hour of static storage. 84 Figure 12. The reduced swimming motility of W2X2 (W3110 sdiA - ). Swimming ability of sdiA deletion mutant, W2X2, was examined. (A)Overnight cultured W2X2 and W3110 in LB broth were placed on the LB agar containing 0.3 % Bacto agar (Difo) and incubated at regular 37 ?C for 11 hr and then scanned. (B) Swimming speed of the wild type and mutant cells were compared by motility tracking (Dr. Losert, W. Dept of Physics, UMCP). 300 E. coli cells of each strain were individually tracked and distributed by probability. 3 time points of broth cultures were compared (1 hr, 2 hr and 4 hr after initiation of culture, each representing an early-, or mid- , or late exponential state). 0.4 0.3 0.2 0.1 0.0 15 1050 1 hr 2 hr 4 hr W2X2 (( sdiA - ) p 0.2 0.1 0.0 151050 ? m per second W31100 1 hr 2 hr 4 hr p 0.4 0.3 ? m per second A B 85 C. W3110 D. W2X2 Figure 12. (contd) (C) and (D) are scanning electron microscopy images of W3110 and W2X2 at late exponential stage. 86 Shock resistance assa When examined on the resistance toward severe insults, W2X2 survived better than wild type W3110 in general (heat shock at 55?C, acid shock at pH 3, and oxidative stress with 5 and 15 mM H 2 O 2 ). For heat shock tests, cells were aliquoted in 1.5 ml Eppendorf tubes (40 ? l) and set in a hot water bath (55?C). Every 1 min, each of the tubes was withdrawn and immediately serially diluted in PBS buffer (RT) and was plated on nutrient agar plates for counting the viable colonies (Fig. 13.A). In both cases, there was an apparent lag in cell death, but the lag was sustained in the case of W2X2. Ultimately the first order specific cell death rate was indistinguishable between the two, suggesting the death mechanism was consistent. Increased physical aggregation of W2X2 is speculated to have provided chances of the deep-core-centered cells from being slowly heated up about 1 min in our experiment set-up. For acid shock tests (Fig. 13.B), cells were centrifuged, resuspended in pH 3.0 LB for the indicated times, and rapidly serially diluted in PBS (pH 7.4). Both wild type and sdiA null mutant W2X2 were sensitive to acid shock, however, the specific cell death rate for W2X2 was significantly lower than that in wild type W3110. Hydrogen peroxide tests were performed identically, with cell resuspension in PBS (pH 7.4) containing 15mM H 2 O 2 . Again, W2X2 cells were more tolerate to the applied stress (Fig. 13.C). . 87 Figure 13. Lethal shock assays on the late exponential phase cells. W2X2 showed stronger resistance to lethal condition than wild type W3110. (A) 55 ?C thermal shock. Lines along the data points are predicted as follows. W3110:y=5x10 9 X -7.0 , b=-7.0+0.63; W2X2: y=10 11 X -7.8 , b=-7.8+0.67.(B) pH3 acid shock responses are reported as the number of colony forming units of cells (CFU/ml) after the stress. (C) 15 mM H 2 O 2 oxidative stress response is reported as the ratio of the survived cells after the stress. Error bars of A and B are based on 3 independent experiments. Closed squares stand for W3110 and open squares stand for W2X2. 88 Comparative analysis using DNA microarrays Having established significant phenotypic variations between W3110 and W2X2 cultivated in LB media and/or during late exponential phase (OD 600 1.8), a transcriptome analysis was performed using Affymetrix E. coli antisense microarrays. Biological replicates were prepared on separate occasions and total RNA mixtures were extracted for the repeated chips. The RNA mixtures were converted to cDNA for further analysis using the chips (see Material and Methods). The antisense arrays are made of DNA oligo probes designed to hybridize all cDNAs from all open reading frames (ORFs) and intergenic regions (IGs). It is noteworthy that IG probes are designed to reflect the orientation of neighboring ORFs. Initially limiting our analysis to detectable ORFs and IGs by signal intensity and reliability, our microarray data were initially grouped based on the present?absent calls, resulting in ~70 % of W3110 and 64 % of W2X2 of the total 4347 ORFs and 60 % of W3110 and 56 % of W2X2 of the total 2886 IGs assessed as present (Fig. 14). Beyond differential responses that are discussed in detail below, spots only expressed in W3110 were presumably attributed to a positive role of sdiA. Vice verse a putative negative role of SdiA can be found from spots expressed only in W2X2. SdiA is more likely to function as an activator than a repressor because the expression of 165 more ORFs and 92 more IGs were reduced with sdiA deletion. This is further supported by the fact that correlation with the overexpression study was limited to genes induced with the increase of sdiA transcripts (discussed below). A complete list of log 2 IR for the ORFs and IGs is available on our group website (http://www.umbi.umd.edu/~cbr/lab_web/home.htm). Next, ORFs and IGs both absent called from W3110 and W2X2 (absent in both from Fig. 3, black pies) were omitted and a total of 100 ORF (2.2 % of the total ORFs) 89 and 33 IGs were identified outside a 2-fold induction ratio (log 2 2=1) cutoff (Table 6 and 9). Judging from the signal intensity in the microarray, the expression level of sdiA in wild type W3110 was quite low compared to the other global regulators such as RpoS. The low differential value of log 2 IR W2X2/W3110 (-2.4) is considered due to the low expression or short life-span of sdiA transcript. Whether this caused the generally mild differential IR between the strains is not known. 90 Figure 14. Differential transcriptome expression. ORFs and IGs are grouped by detection calls. Absent called ORFs and IGs both from W3110 and W2X2 are colored black and were not further included in Table 6 and 9. Remaining data were further discussed in the text which include; the spots present only in W3110 or W2X2 (dotted or lined in the pie graph) and spots both present in W3110 and W2X2 (the white portion). OR I 91 ORF The 102 ORFs were grouped functionally based on the Affymetrix database; metabolism/nucleotide synthesis/motility related (42), regulatory proteins (13), transport /binding proteins (6), other functions (3) and uncharacterized or hypothesized functions (38). Interestingly, many of regulon- or operon-wild changes were noted. Metabolism and cell processes The most highly affected ORF by sdiA deletion was hdhA, which encodes a bile acid modifying 7-alpha-hydroxysteroid dehydrogenase (hdhA, -2.4) (Yoshimoto et al., 1991). Produced from liver, bile acids are secreted to digestive systems to emulsify dietary fats. Although active modification and thus diversification of the bile acid by enteric bacteria are known, specific purposes of the activity are not clear. If the altered transcription level resulted in expressed protein levels, it would not be too much exaggerated presumption that SdiA would function in the bacterial association in the host digestive system. Meanwhile the biggest increase by the sdiA deletion occurred with glcB (2.2), a gene associated with a glycol utilization pathway. Its product, inducible malate synthase G, converts glyoxylate to malate. In the same glc operon, unknown functional glcG was also significantly induced with a marginal increase of a divergently transcribed transcriptional activator gene, glcC. Three intriguing regulon-wide changes were found; genes for sulfate assimilation, purine de novo synthesis, and motility. Reduction of SO 4 2- to SO 3 2- and H 2 S is needed for cysteine and methionine synthesis catalyzed by proteins encoded as cysDNC and cysJIH operons. Although it is generally difficult to link specific amino acid pathways with the 92 mutation of an indirectly associated gene (Hua et al., 2004), our finding of potentially increased expression of genes for sulfur reduction may have potential to be linked to the synthesis of SAM, the common substrate of AI-1 and AI-2. Importantly supporting reduced swimming motility of W2X2, 28 genes involved in flagella, chemotaxis, and motility were down-regulated. Of these genes, only the flg operon is located at min 24 of the chromosome, while the rest of them are clustered at min 43 of the chromosome, wherein sdiA is located (Fig. 15.B). The three levels of hierarchical regulation are as indicated in the figure. The two clusters of genes are separated by 11 kbp apart. Genes for purine (AMP and GMP) biosynthesis are scattered across the genome and all of the 12 genes were induced by the mutation. Exceptionally, the expression of PurR repressor (GalR/LacI family) controlling the regulon did not change in this study (Meng et al., 1990). Why in the LB rich media, W2X2 induced more genes for the nucleic acid, the building block of the nucleic acid, is quite odd. The rest of three genes (speC, cvpA, and sodA) in the group of metabolism and cell process could be related with the purine de novo synthesis. First, a constitutive form of ornithine decarboxylase (biodegradative, producing putrescine) encoding speC (1.18) is known to be positively regulated by GTP in its expression (HOLTTA et al., 1972) Thus, observed induction of purine de novo synthesis pathway may have preceded the speC induction. Also known to raise pH in the acidic environment, possible increase putrescine may have contributed to the increased acid resistance or other shock resistance. Secondly, cvpA (1.0) is located within the purF operon and commonly regulated by PurR (Fath et al., 1989). It is needed for the production of plasmid-borne small extracellular toxin protein, colicin V. Lastly, sodA is one of two distinct superoxide dismutases (1.2). Its 93 production, manganese-SOD, is only present under aerobic conditions (Touati, 2000). Also MnSOD is a part of ribonucleoside-diphosphate reductase activating system (Coves et al., 1995), which is required to convert nucleotide to deoxynucleotide. Thus it is likely that the increased purine synthesis demanded the activation of ADP reductase and GDP reductase. 94 Metabolism and cell processes log 2 IR cysC adenosine 5'-phosphosulfate kinase 1.6 cysD ATP:sulfurylase , subunit 2 1.8 cysH 3'-phosphoadenosine 5'-phosphosulfate reductase 1.3 cysI sulfite reductase, alpha subunit 1.2 cysJ sulfite reductase (NADPH), beta subunit 1.4 cysN ATP-sulfurylase, subunit 1, probable GTPase 1.3 guaA GMP synthetase (glutamine-hydrolyzing) 1.5 guaB IMP dehydrogenase 2.0 purC SAICAR synthetase 1.4 purD GAR synthetase 1.6 purE AIR carboxylase, catalytic subunit 1.2 purF PRPP amidotransferase 1.2 purH AICAR formyltransferase 1.7 purK AIR carboxylase, CO(2)-fixing subunit 1.3 purL FGAM synthetase 1.2 purM AIR synthetase 1.6 purT formyltetrahydrofolate deformylase 1.5 pyrB aspartate carbamoyltransferase 1.0 glcB malate synthase G 2.2 speC ornithine decarboxylase isozyme 1.2 cvpA membrane protein for colicin V production 1.0 sodA superoxide dismutase, manganese 1.2 Transport and binding proteins cysP thiosulfate binding protein 1.2 sbp periplasmic sulfate-binding protein 1.1 melB melibiose permease II 1.0 Others prfC peptide chain release factor RF-3 1.4 flu (agn43) outer membrane fluffing protein, similar to adhesin 1.0 Uncharacterized hypothetical proteins glcG a gene within the glycolate utilization operon 1.4 b2001 hypothetical protein 1.1 b4188 conserved hypothetical protein 1.2 b1287 putative oxidoreductase 1.5 b2789 YgcZ MFS transporter 1.3 b3337 bacterioferritin-associated ferredoxin 1.8 b3646 putative membrane protein 1.5 b3654 YicE NCS2 tranporter 1.6 b1388 putative phenylacetate-CoA oxygenase 1.4 b1389 putative phenylacetate-CoA oxygenase 1.1 b1390 putative phenylacetate-CoA oxygenase 1.6 b1391 putative phenylacetate-CoA oxygenase 1.2 b1392 putative oxygenase reductase 1.7 b1393 putative enoyl-CoA hydratase 1.6 b1394 putative enoyl-CoA hydratase 1.5 b1395 putative 3-hydroxy-acyl-CoA dehydrogenase 2.0 b1396 with some similarity to thioesterases 1.3 b3114 2-ketobutyrate formate-lyase 1.3 b3115 propionate kinase / acetate kinase C 1.6 b3125 tartronate semialdehyde reductase 1.0 b3127 YhaU MFS transporter 1.6 Table 6. Significant changes in ORF (upregulated). 95 Metabolism and cell processes log 2 IR flgB cell-proximal portion of basal-body rod -1.3 flgD initiation of hook assembly -1.1 flgF cell-proximal portion of basal-body rod -1.0 flgG cell-distal portion of basal-body rod -1.0 flgH basal-body outer-membrane L ring -1.2 flgI homolog of Salmonella P-ring -1.1 flgK hook-filament junction protein 1 -1.4 flgL hook-filament junction protein -1.5 flgM anti-FliA (anti-sigma) factor -1.2 flgN protein of flagellar biosynthesis -1.2 fliD filament capping protein -1.1 fliE basal-body component -1.5 fliF basal-body MS-ring and collar protein -1.2 fliG for motor switching and energizing -1.2 fliH flagellar biosynthesis -1.1 fliJ flagellar fliJ protein -1.1 fliL flagellar biosynthesis -1.1 cheA proton conductor component of motor -1.3 hdhA 7alpha-hydroxysteroid dehydrogenase -2.4 Regulatory proteins sdiA transcriptional regulator of ftsQAZ -2.3 soxS regulation of superoxide response -1.4 rpoE heat shock oxidative stress -1.2 rseA negative regulatory protein for RpoE -1.1 fliA alternative sigma factor 28 -1.5 fliT repressor of class 3a and 3b operons -1.4 fliS repressor of class 3a and 3b operons -1.4 flgM anti-FliA (anti-sigma) factor -1.2 flhD regulator of flagellar biosynthesis -1.2 cheW positive regulator of CheA protein -1.3 trg ribose sensor receptor -1.2 cheZ antagonist of CheY as switch regulator -1.0 tar aspartate sensor receptor -1.1 Transport and binding proteins srlE glucitol/sorbitol-specific IIB component -1.1 araH high-affinity L-arabinose transport -1.1 fruA PTS system, fructose-specific transport -1.8 Others amyA cytoplasmic alpha-amylase -1.2 Uncharacterized hypothetical proteins b0250 hypothetical protein -1.2 b1760 conserved hypothetical protein -1.5 b1904 conserved hypothetical protein -1.7 b1935 hypothetical protein -1.3 b4109 hypothetical protein -1.3 b1194 protein involved in flagellar function -1.3 b1566 hypothetical protein; gene is regulated by FliA -1.8 fliZ possible cell-density responsive regulator of sigmaF -1.5 b3525 protein involved in flagellar function -1.2 b2147 putative dihydrothymine dehydrogenase -1.2 b1967 heat shock protein (Hsp) 31 -1.7 b3570 putative ATP-binding protein -1.2 b1513 ATP-binding component of ABC transporter -1.3 b1515 membrane component of ABC transporter -1.5 b1516 periplasmic binding protein of ABC transporter -1.1 b1517 putative aldolase -1.1 b1518 conserved hypothetical protein -1.3 Table 6. (contd.)Significant changes in ORF (down regulated) 96 A. B. Figure 15. Motility-related genes around sdiA in the chromosome. (A) Transcript changes in sdiA ?IG-yecC after sdiA deletion. Subtranscript level changes were shown as described in Material and Methods. 42 probes are representing the region (14 of each). Open circle indicates the relative induction ratio by sdiA deletion (log 2 RI W2X2/W3110 , right Y-axis) while bars indicate the signal intensity (PM-MM, corrected, left Y-axis). The sdiA deletion induced the upstream IG expression. However, the signal in the IG is not strong (absent called). Upstream gene yecC showed no changes by the sdiA mutation. Gel figure shows the absence of sdiA mRNA from W2X2 using RT-PCR (with 40 PCR cycles). (B) Genomic context of sdiA relating with the motility genes. The sdiA gene is indicated (red triangle) in the middle of flagella, motility and chemotaxis regulon (region IIa IIb, around 1960000-2028000). The hierarchy of the regulon, (Class I, II and III) is shown along with the specific microarray log 2 IR values for each gene. flhDC (-1.2 -1.0) fliA (-15) fliFGHIJK (-1.2 -1.1 -1.1 -0.8 -1.1 -0.6) motAB cheWA (-1.5 -1.6 -1.3 -1.3) fliMNOPQR (-0.7 -0.9 -1.1 -0.7 -0.4 -0.7 -0.7) fliE ( -1.5) Class II Class I Class III tar tap cheRBYZ (-1.1 -0.8 -0.9 -0.8 -1.0) fliC fliDST (-0.9 -1.2 -1.4 -1.2) sdiA -2000 -1500 -1000 -500 0 L14 -6 -4 -2 0 2 L o g 2 (R ela t iv e I n t e n s ity ) -Correcte d Si gnal Intensi t y sdiA(1994133?1994855) yecC(1995085?1995834) IG 1278 ?sdiA transcript (RT-PCT) W3110 W2X2 97 Uncharacterized hypothetical proteins (metabolism) Among the 39 genes that were annotated as uncharacterized in the Affymetrix database, three induced groups of genes appeared as parts of operons (Fig. 16). Since the functionality of each of them have been reported in metabolism, it is appropriate to introduce them here. These are proposed to function in L-threonine degradation (tdc), phenylacetate degradation (paa), and glycolate/galactarate/glucarate degradation (gar). First, the tdc operon (min 68.3, 7 genes of b3112-b3118 with 3 IGs, 4 significant and 6 moderate increases, anaerobic)) is composed of genes encoding threonine dehydratase (tdcB), permease (tdcC), propionate kinase (tdcD), 2-ketobutyrate formate- lyase (tdcE), hypothetical protein (tdcF), and serine deaminase (tdcG). It has two genes encoding operon-specific transcriptional activators TdcA (LysR type) and TdcR. The changes of tdcABCDEFG expression were evenly around 2 fold-levels, while changes in tdcR and downstream gene yhaO (may function as amino acid/proton symporter) were much smaller. Besides TdcA and TdcR, multiple transcriptional factors including CRP, IHF (integration host factor), and HU (histone like protein units) are directly involved in its promoter expression (Hagewood et al., 1994). Secondly, genes for putative phenylacetate degradation pathway (min 31.2-31.5, 11 genes of b1388-b1398, 9 significant and 2 moderate increases) showed induction with the last two distal genes with moderate changes. Following the operon exists its transcriptional regulator encoding gene paaX as a transcriptional unit, however paaX expression did not change by the mutation. Also seen in the PurR of purine de novo pathway, the absence of changes in the regulator in contrast to the changes in the structural genes in the operon is likely to have resulted from difference in the expression 98 timing and controlled level of expression (transcription for the structural vs. translation for the regulatory genes) between the regulators and structural genes or from the difference in mRNA degradation speed (i.e. more rapid degradation of the transcriptional regulator mRNA). Finally, induced galactarate/glucarate degradation pathway operon (5 genes of b3124-b3128 with 1 IG, 2 significant, 1 moderate, and 3 marginal increases) encodes two transcriptional units of garD and garPLRK encode galactarate dehydratase (garD), MFS transporter (garP), aldolase (garL), reductase (garR), and glycerate kinase (garK). As indicated in the Fig. 5, the changes in garP for MFS transporter and garR for tartonate semialdehyde reductase were greater than the others?. A common regulatory protein, SdaR, was identified for the both directional transcription toward garD and garP (Monterrubio et al., 2000). Importantly, its third transcriptional unit, gudPD (b2789?b2787), was also induced in W2X2 [MFS transporter (1.3) and glucarate dehydratase (0.8)], proving it as another example of regulon-wide changes. Particularly, the apparent increase of the two MFS transporters (garP and gudP) is intriguing. Increased glycolate (glc operon, above) and galactarate utilization commonly ends in the production of 3-phosphoglycerate for TCA cycle. Combining the some of the changes mentioned above, an interesting sequence on glyoxylate metabolism can be state. Glycoxylate is generated from both purine degradation and glycolate. If purine degradation were prior to the increased purine de novo pathway, glyoxylate level increased naturally in W2X2 (combined with gar) and could draw precursors from the TCA cycle (glyoxylate bypass) to glyoxylate cycle for 99 making building blocks, such as nucleic acids. It is noteworthy that the glyoxylate cycle is often associated with growth on acetate, which, in turn, accumulates during the late exponential phase during growth in LB. Indeed, the induction of the phenylacetate degradation may have increased acetate levels in W2X2 cytosol (paa operon, above). This reasoning, consistently, results in increased purine synthesis. 100 Microarray Relative Signal Ratio (log 2 W2X2/W3110) 0.88 0.83 0.98 1.25 1.55 0.93 0.78 0.88 0.48 0.95 1.38 1.13 1.60 1.20 1.68 1.63 1.48 2.02 1.25 0.38 0.53 -0.35 0.38 1.03 0.88 1.63 0.33 Figure 16. Operon-wide changes with putative functions (metabolism associated). Three induced operones suggested to function (still annotated as putative in Affymetrix databse) are depicted and marked for the differential expression (log 2 IR). These are suggested to involve in L-threonine degradation (tdc operon), phenylacetate degradation (paa operon), and glycolate degradation (gar operon). Notably, positive regulator of paaX did not change. yhaO tdcG F tdcE D t tdcC tdcB tdcA tdcR rnpB garK garR garL garP garD paaZ paaA B C D E F G paa H I J K paaX 101 Transport and binding proteins Among the group of transport and substrate binding, three induced (cysP, sbp, and melB) and three reduced (srlE, araH and fruA) gene expressions were found. Induction of thiosulfate/sulfuate binding proteins (cysP and sbp) can be regarded as the precedence of the increased sulfate assimilation. Their products, CysP and Sbp, are both localized in the periplasm and functionally identical interacting commonly with the components of ABC-type transporter CysATW (Sirko et al., 1995). Related with sugar transport, melibiose permease and low affinity glucitol/sorbitol transporter showed operon-wide changes. Among the genes in melibiose (? -D-galactopyranosyl-(1? 6)-D-glucopyanose) permease operon (melRAB), the increase of permease gene, melB, was most apparent than those of a transcriptional activator and ? -galactosidase encoding genes (melR and melA). Among the decreases of genes in glucitol/sorbitol transporter (phosphoenolpyruvate dependent phosphotransferase system, srlAEBD), the change of srlE (-1.1) was most clear and listed in Table 2. The operon has a transcriptional activator (srlM) and repressor (srlR) nearby (Postma et al., 1993). In our study, the srlM was also down-regulated, however, srlR expression did not change. This suggests that W2X2 may have increased utilization of cytosolic glycolate (glc operon, mentioned above) and rather decreased its uptake, which is reasonable. Both mel and srl operons are known to be catabolically repressed by cAMP-CRP complex. We also noted interesting reduction of another transporter encoding gene clusters (b1511-b1518), which is discussed below. 102 AI-2 transporter, lsr operon Most interestingly, we noted that the ATP binding cassette-type transporter at 34.5 min of the E. coli chromosome, also identified as the lsr operon, was significantly reduced in the W2X2 cells (Fig. 17). Originally predicted as the low-affinity transporter of the two ribose transporter systems in E. coli, the lsr operon was recently characterized for AI-2 uptake and modification both in E. coli and Salmonella (Wang et al., 2005; Xavier and Bassler, 2005). Transcribed opposite direction from the ABC transporter structural genes (b1513-b1515, lsrACDB) and putative AI-2 processing enzyme encoding genes (b1517-1518, lsrFG), genes encoding a negative regulator (lsrR, b1512, SorC family) and a kinase (lsrK, b1511) exist in the upstream as another transcription unit (Fig. 17.A). The lsr operon of E. coli does not have the last operonic gene, the lsrE counterpart of Salmonella, but is followed with a trans-aconitate methyltransferase encoding gene, tam. Unlike the previous proposition that lsr operon is homologous to rbs operon for a high affinity ribose transporter (Taga et al., 2001), only one component (rbsD) was moderately affected (-0.6, Fig. 6.A). RbsD converts ribose from pyran to furan for accelerated utilization (Ryu et al., 2004). To confirm the changes in the lsr operon, transcriptional fusion assays for lsrA- lacZ fusion were performed in a lac - background (E. coli ZK126, Fig. 17.B). Beginning with the growth of the cells, unlike the W3110 background, ZK126 cells with an sdiA deletion (ZKWX2) had a lower growth yield and growth rate in the LB cultures compared to the parental control ZK126. Interestingly, adding consumable amount of glucose at the beginning of the cultures (1.11 mM, 0.1 %) reduced the difference and increase the promoter activity from both of the strains. AI-2 levels in CM from these cultures were typically higher in CM from the parental cells than the sdiA - cells (Fig. 103 17.C). Regardless of the presence of glucose in the media, sdiA - resulted in lower lsrA transcription (~2-3 fold as measured by ? -galactosidase expression), which is consistent with microarray results (Fig. 17.B). The differences were particularly noteworthy during the late exponential phase but not in the early exponential phase (2-3 hr), which may explain why this operon was not detected in the experiment for overexpression by microarray. Catabolite repression is the major driving force of lsr structural gene transcription finely tuned by AI-2 responsive regulator (LsrR) (Wang et al., 2005). The reduced lsr operon expression in our study showed that presence of SdiA somehow assists CRP- driven lsr transcription. Following the reasoning of Bassler and colleagues, increased AI- 2 must have activated LsrK kinase for derepressing LsrR, which in turn induced the transcription toward lsrA and the following genes of the operon. Non-consumable amounts of glucose (0.8%) abolished the expression of the operon both from W3110 and W2X2. These data demonstrate that the systems 1 and 2 of E. coli quorum sensing merge on the ATP-binding cassette type transporter characterized as the AI-2 specific uptake machinery. Although, in general AI-2 and AI-1 molecules share the amino acid precursor, S-adenosyl methionine (SAM), genes involved in AI-2 synthesis (pfs and luxS) were not affected. Whether there is a physical interaction with SdiA and the lsr promoter was not tested. Convergence of different quorum sensing pathways has been suggested in other species when there are multiple pathways present in an organism, as in the case of Pseudomonas and Vibrio species. Most recently, two different quorum regulators in 104 Streptococcus pneumoniae were shown to positively regulate the same ABC transporter for transporting peptide pheromones (Knutsen et al., 2004) 105 A. Relative Signal Ratio (log 2 W2X2/W3110) -0.90 -0.85 -1.27 -0.52 -1.50 -1.12 -1.05 -1.27 -0.62 -0.60 -0.05 0.20 -0.15 0.05 0.48 B. Transcriptional fusion assay C. AI-2 assay Figure 17. The positive regulation of SdiA on AI-2 uptake machinery. (A)Divergent transcriptions from b1511-b1518 (lsr operon) are depicted with the differential expression value (log 2 IR) obtained from our microarrays. Suggested homologues rbs operon is also depicted. (B) Low copy vector based lsrA-transcriptional fusion assay was performed (Wang et al. 2005) in the presence and absence of glucose (0.1%). X-axis represents time and y-axis represents miller units. Error bars are obtained from two independent experiments. (C) Corresponding AI-2 levels in the conditioned media were indicated with the growth curves. lsrK lsrR lsrA lsrC lsrD lsrB lsrF lsrG tam rbsD rbsA rbsC rbsB rbsK rbsR 0 200 400 600 800 1000 1200 1400 1600 1800 2000 12345678 time (hr) A I - 2 f o ld in d u c t io n 0.00001 0.0001 0.001 0.01 0.1 1 10 O D 600 n m 0 50 100 150 200 250 300 350 400 2345678 time (hr) M ill e r U n it s ZK126 ZK126 0.1% glc ZKWX2 ZKWX2 0.1 %glc 106 Ag43 and biofilm formation Importantly, a gene coding antigen 43 (Ag43), the major reversible phase-variable protein in the outer membrane, was induced (flu, agn43, 1.0). As a species-specific antigen, Ag43 is well-known to make cells aggregate (Kjaergaard et al., 2000), which is in good agreement with the observable W2X2 phenotype seen in Fig. 11.F. There were no significant differences in the other possible adhesin, such as curli proteins or fimbriae. The transcription of the Ag43 is known to be regulated by differential methylation of Dam and OxyR interaction. OxyR binding on the promoter blocks the methylation thus inhibits flu expression. Locked-off flu expression by oxyR mutant was reported (Henderson and Owen, 1999). It is noteworthy that quorum-sensing often controls cell motility, biofilm formation, and siderophore production. There are three papers using DNA microarrays for E. coli biofilm formation (Beloin et al., 2004; Ren et al., 2004; Schembri et al., 2003). Interestingly, sdiA, lsr operon, flu and many other genes were detected at least once to be differentially expressed in biofilm compared to the planktonic cells. 107 Transition to stationary phase (Table 7) The transition period to stationary phase is typically characterized by extensive reprogramming on translational- and transcriptional- machinery, nucleoid, and cell process (Ishihama, 1999). By the sdiA mutation, translation process seemed most affected caused by the changes in ribosomal RNA (rRNA) components as apparently shown in Table 3. The rRNAs are encoded in 7 operons, each of which makes a precursor of 16S, 23S, and 5S rRNA and at least one tRNA genes. Designated as 18 b-numbered ORFs in E. coli Affymterix chips (8 for 5S, 6 for 23S, and 4 for 18S rRNA), almost all of the rRNAs transcriptionally decreased significantly or moderately keeping the tRNA expression levels equal. When small effector molecules that regulate rRNA transcription were considered as possible explanation of the differences, relA and spoT for the stringent response effector, ppGpp, did not differ between the two strains. However, as mentioned above, altered purine nucleotide synthesis is likely to have been realized as altered NTP concentration for initiating rRNA transcription (iNTP) (Paul et al., 2004).The differences in rRNA expression, were confirmed by a RNA gel with equal amounts of total RNA in loaded samples (Fig. 18). Unlike rRNA, genes encoding entire 55 ribosome protein components (22 for 30S ribosome subunit and 23 for 50S subunit) did not change much; only three moderate decreases (rpsU, rpmB, rpmG: -0.60, -0.80, and -0.58) and one moderate increase (rpsQ, 0.85) were found. Other genes related with translation were also examined. Of three translation termination factors, only RF-1 (peptide chain releasing factor, prfC, 1.35) was induced while keeping the other two RF-1 and RF-2 unvaried. Additionally, translation repressor operon relBEF slightly decreased (-0.72, -0.47, and -0.10, respectively). Interestingly, the ribosomal modulation factor (rmf) for dimerizing ribosomes in the 108 stationary phase was also reduced in W2X2. Physiological outcome of the changes in the translational machinery found from the genomic analysis needs to be defined in the future. One of the limits in current microarray studies is the inability to distinguish the rates of transcription initiation and transcription turnover. However, considering the rRNAs and tRNAs are processed from same precursors, the reduction in 5S, 16S, and 23S rRNA content keeping the equivalent expression of tRNA and even the entire ribosomal protein component levels is a convincing evidence suggestive of rRNA destabilizing force. Impacts on translation machinery by quorum sensing are quite undefined. Previously significant changes of ribosomal protein component transcription by the luxS-mutation in EHEC 0157:H7 was reported (Sperandio et al., 2001), and there is no report on LuxR-mediated possible translation machinery modulation to our knowledge. The decreased rRNA quantity reminds us of common growth limits, however, there were no noticeable differences in growth rate in W2X2. And arrested growth nucleotide synthesis, especially of pyrimidine, tends to decrease. However, observed induction in the purine de novo pathway confirmed that this phenomenon differs from the common growth arrest. Unlike rRNA levels associated with translation machinery, the expression of transcription machinery did not show dramatic changes. Of the 9 RNA polymerase components (rpoABCDEHNSZ), the core RNA polymerase and constitutive sigma factor transcript levels were conserved. Exceptionally alternative sigma factor rpoE and rpoS expression decreased and are discussed below. 109 Finally, 12 known nucleoid-associated DNA binding proteins were examined and only stationary-phase protein IHF was found to change, with a moderate decrease. Functioning in general cell process and stress response, 25 chaperone genes (including putative genes) were also examined and marginal decreases in hslV and hslU and moderate decreases in mopB, mopA, and htpG were shown as seen in Table 7. 110 Gene Description log 2 IR Ribosomal RNA with a change call rrlA 23S rRNA of rrnA operon -0.9 rrlC 23S rRNA of rrnC operon -0.6 rrlD 23S rRNA of rrnD operon -0.5 rrlE 23S rRNA of rrnE operon -1.1 rrlG 23S rRNA of rrnG operon -1.1 rrlH 23S rRNA of rrnH operon -1.1 rrsA 16S RNA of rrnA operon -0.9 rrsC 16S RNA of rrnC operon -0.8 rrsG 16S RNA of rrnG operon -0.9 rrsH 16S RNA of rrnH operon -0.7 Chaperone with a change call hslU heat shock protein hslVU, ATPase subunit, -0.5 hslV heat shock protein hslVU, proteasome-related -0.5 mopB GroES, 10 Kd chaperone binds to Hsp60 in pres. -0.6 mopA GroEL, chaperone Hsp60, ATPase -0.5 htpG chaperone Hsp90, heat shock protein C 62.5 -0.8 All RNA polymerase components rpoA RNA polymerase, alpha subunit -0.2 rpoB RNA polymerase, beta subunit -0.2 rpoC RNA polymerase, beta prime subunit -0.2 rpoD RNA polymerase, sigma(70) factor -0.2 rpoE RNA polymerase, sigma-E factor -1.2 rpoH RNA polymerase, sigma(32) factor -0.0 rpoN RNA polymerase, sigma(54 or 60) factor -0.3 rpoS RNA polymerase, sigma S (sigma38) factor -0.4 rpoZ RNA polymerase, omega subunit 0.1 Table 7. Examination on growth phase transition. Differential expression of components in translation and translation machinery and cell process (by chaperones) were looked into and the genes with change calls were selected and listed. Note that significant rrl and rrs changes (bold letters) are not included in the Table 6. 111 Figure 18. Reduced rRNA of W2X2 (W3110 sdiA - ). Equal amounts of total RNA (both 1.0 ? g and 0.5 ? g RNA) of W3110 and W2X2 were loaded onto formaldehyde (220 mM) RNA gel with 1.2 % agarose for visual comparison. Samples from W2X2 showed apparently reduced rRNA level compared to those of W3110 in the specific growth stage (late exponential). W3110 W2X2 W3110 W2X2 Total RNA 1? g 1 ? g 0.5? g 0.5 ? g 23S rRNA 16S rRNA 112 Comparative analysis with other regulators A relatively high ratio of regulatory proteins was affected by the sdiA mutation, possibly due to an intrinsic complicacy of transition in growth phase (as mentioned above). When we looked into known global regulators with changes, reduced extracytoplasmic stress and oxidative stress associated factors (rpoE, -1.23 ? soxS, -1.35, respectively) were found, along with factors involved in starvation/anaerobic environmental adjustment (rpoS,-0.43 ? acrA,-0.80) to a less extent. Intrigued with the possibility of their regulatory effects on the genes catalogued in Table2, we attempted to compile the target genes of each regulator and examined their differential ratios from expectation. Sources of the compared literature and database are listed in Table 4. [rpoS regulon (Weber et al., 2005); soxS and its homologues rob and marA regulon (Barbosa and Levy, 2000; Martin and Rosner, 2002; Pomposiello et al., 2001); acrA regulon EcoCyc; rpoE regulon (Dartigalongue et al., 2001; Rezuchova et al., 2003)]. Details of comparative studies are as follows. First, among the 140 genes expressed at the RpoS inducing conditions of stationary phase, osmotic upshift and acid stress (Weber et al., 2005), 7 genes got change calls, especially with 2 significant changes of amyA and hdhA. Oddly, major stationary phase response gene, osmY, did not change in our study. Secondly, many genes of soxS/rob/marA regulon showed changes in our study, especially with three significant changes of srlE, pyrB, and guaB. Still, several genes showed opposite change-pattern from expectation, e.g. significant induction of sodA and 4 moderate reductions of b1452, fumC, and pqiA. Thirdly, dual regulator of ArcA, composing a two-component system with a sensor kinase ArcB, represses aerobic metabolism genes when respiration is not possible (such a condition of the arrested 113 growth during stationary phase). We found 22 genes with change calls, and most of them showed expected changes in agreement with the decrease of soxS expression. Finally, functioning in polypeptides folding in the envelope and lipopolysaccharide biosynthesis/transport, ? E is known to control at least 31 promoters for 73 genes (Dartigalongue et al., 2001; Rezuchova et al., 2003). None of them changed significantly, however, the expression of three ORFs showed moderate changes. Unexpectedly, transcription of its downstream neighbor encoding its own anti-RpoE factor (rseA, -1.05) (Missiakas et al., 1997) and another gene under the suppression of RpoE decreased (a growth inhibiting gene ssnA, -0.8). In summary, nonetheless the inconclusiveness originated from different experimental conditions independently adopted in different studies, several changes ascribable to changes in the global regulators were identified: six RpoS regulated genes were affected, three SoxS regulated genes, and three ArcA regulated genes (Table 8). In an analogous manner, our results were compared with that of sdiA overexpression study using PCR-based microarrays (Wei et al., 2001). Surprisingly, only two genes, yedU and yhiW, showed correlated expression between the two studies i.e. increased expression with increased level of sdiA transcripts. Despite the low expression level of yhiW in our study (not listed in Table 2 due to being absent called), it has same a HTH domain homologous to that of Salmonella srgC, which was shown to be sdiA dependent (AraC/XylS family). If it were indeed directly SdiA-dependent, the increased acid resistance of the W2X2 strain would be glutamate-dependent (Ma et al., 2002). The yedU encodes a heat shock protein Hsc31(Mujacic et al., 2004). Interestingly, both of 114 them were also identified as members of RpoS regulon (Table 8), thus, overlapped regulation of SdiA and RpoS is highly likely to exist on these genes. Also there were 6 moderate changes showing correlation, which includes ftsQ and Z, the products of the only confirmed sdiA- dependent transcript, ftsQ2p, supporting the null mutation of sdiA. However, uvrY, the gene directly downstream of sdiA and suggested to be SdiA- dependent from another overexpression study (Suzuki et al., 2002; Wei et al., 2001), did not change in our study. Similarly, genes for acridine efflux pump shown to be the reason of increased antibiotics resistance of sdiA overexpressing strain (Wei et al., 2001) showed only marginal change only in acrA (-0.38). This supports and explains the absence of hypersensitivity to MMC by a sdiA knockout strain (Wei et al., 2001). More than that, strikingly, 46 genes even showed the opposite pattern deviated from the expectation (25 significance- and 21 moderate decreases). With the exceptions of cysteine degradation and nucleotide conversion, tnaA and aphA, all of them are motility-related. The decreases of the regulatory genes for chemotaxis ? motility and subsequent changes in downstream genes were depicted in Fig. 15.B. This change in transcription turned out as reduced swimming motility of W2X2 and was visually further proven as less attached flagella on W2X2 (Fig. 12). Indeed, motility must be one common target sensitive to quorum signaling [e.g. LuxS-dependent motility and attachment defects were reported (Sperandio et al., 1999)]. Possible explanation for the apparent discrepancy between the two studies can be easily found from the dramatically different growth phases chosen to be compared (OD 600 0.45 vs 1.8), or due to the general stress response resulting from protein overexpression (Bentley et al., 1990). However, from more radical standpoints, SdiA may adopt 115 bimodular DNA binding conformation and subsequent selections of target pools depending on stoicheometry of its own concentration and the cognate signals (or even presence and absence). Still it needs to be kept in mind that the transcript level changes in the transcriptional regulator could be minor compared to changes in the post- transcriptional levels as one caveat of this comparative approach. 116 Global regulator Literature or public database used Target genes changed in our study gene log 2 I R Ref. Target s # Significant Moderate Marginal rpoS 1 -0.43 Weber et al. 140 (-)amyA,hdhA, yhiW, gadA, otsA, yedU (-)lrp, b1519, b3100 (+)pdhR, b1806 soxS 2 -1.35 Barbosa et al., Pomposiello et al., 264 (-) srlA_2 (+)pyrB, guaB (-) lldP, tnaA, mtlR, gatD, yeaQ, hupB, marA, gatC, zwf, gatA, tnaL, malE (+) ndh, prmA, rpsQ, and yfhO (-)mglB, ytfK, wrbA, tpx, mdaA, madB, yljA, and fur acrA 3 -0.80 EcoCyc 73 sodA, glcB, glcG b2997, caiF, fadAB, fumC, hybABCDE, lldP, ndh aceABK, dctA, sucABC, and tpx rpoE 4 -1.23 Dartigalongue Rezuchova et al. 73 htrA, ygiM, b1806 sdiA -2.35 Wei et al. 137 yedU, yhiW ftsQ, ftsZ, glnH, pntA, pntB, and ybgI Table 8. Literature-based comparative analysis. Examination of cross-regulation with other well-known global regulators and comparison with the transcriptome alteration by plasmid born sdiA (overexpression). Absent called genes in our study are indicated by underlines. 117 IG: SdiA consensus sequences and sRNA expression Affymetrix probe set formats enable users to examine the changes in the intergenic regions (IGs), which can be useful for screening putative SdiA binding motifs (in promoters) and searching the expression of novel and small RNAs. We had 319 IGs (of the 2886 IGs) with change calls and 33 of them showed more than 2 fold changes by the mutation. The induction ratio and signal intensities of 33 IGs were interrogated to a sub- transcript-level and manually grouped to be either showing congruence (20) or incongruence (12) by the relation with those of the flanking ORFs (or IGs) (Fig 20-22, Table 9). Since most of IGs in the congruent group are transcribed in a same direction with the neighbors (Fig. 20), they are likely to be in operon units or be up- or downstream transcribed but untranslated regions of neighboring genes (UTR). Here, 4 IGs could clearly fit as 5?-UTRs of proximal genes, thus they were used as inputs for searching SdiA consensus motifs below. The IGs without congruency were further divided into two groups; opposite (6, Fig. 21) or indifferent (6) to the neighbors in changes (Fig. 22). Details of our data analysis are as follows. Consensus motif Firstly, previous biochemical analysis on ftsQ p1p2 promoters suggested ?SdiA box? with inverted sequences, 5-AAAAG(N) 8 GAAAA-3 (Yamamoto et al., 2001). It was reported that this sequence also exists in the promoter of AI-2 regulated genes eae and espA of E. coli 0157:H7 and in the sdiA regulated gene srgC of S. typhimurium from the genome database search. Limits in single gene based model sequence are unavoidable to covere many possible SdiA dependent genes suggested either from the study of Wei et al. (Wei et al., 2001), or from our study. In order to search 118 consensus sequences in promoter regions, 5? ?UTR positioning IGs from three sources were examined; genes showing congruency with significant changed IG (Table 9 ycfB, modA, purA, and cvp, mentioned above), significantly changed genes with congruent IG (Table 6. yjfN and hdhA), and genes-of-interest (yhiW and yedU commonly detected from sdiA overexpression- and deletion, b1513 of AI-2 uptake machinery, and the ftsQ 2p as an internal control, respectively). Details of the analysis are described in Material and Methods. The most highly scored sequences are presented in Fig. 19. and importantly, the sequence proposed inverted sequence of ?SdiA box? is one of the screened (Yamamoto et al., 2001). Using our results on changes of IGs in the context of proximal gene transcript changes, a search for SdiA binding motifs more in general term was performed in silico. Since LuxR type proteins generally bind as a dimer, we tested the two block models, giving each block of 5 bps and intervals of 8-12 bps. One of the motifs agreed with the ?SdiA box? (Fig. 19), although there are other possible motifs with more weight. Small RNA expression Secondly, extensive presence of structural or regulatory sRNAs are now confirmed (so far about 60 sRNAs are confirmed experimentally) or predicted through genome-wide searching efforts (Carter et al., 2001; Chen et al., 2002; Tjaden et al., 2002; Wassarman et al., 2001). Of the 33 IGs, 9 IGs contain a either confirmed or predicted sRNA. Studying for the IG expression, we have a notion that unexpected expression and changes in IG expression can be ascribed to small RNA if not resulted from the technical errors. Orphan IGs, being not associated with the flanking genes in the expression pattern (i.e. positioned neither in the 5?-UTR as a promoter, nor in the middle of operon, nor at 3?-UTR by incomplete termination of the precedent gene) 119 were selected as 13 IGs of incongruence in our grouping scheme. Notably, IGs neighboring 6 of 5S rRNA (out of 8) genes were included, which may be linked to decrease in rRNA transcripts. It is interesting that IGs neighboring the rest of two remaining 5S rRNAs were already predicted as small RNAs, although they were not significantly changed in our study (Carter et al., 2001; Tjaden et al., 2002). Besides, three predicted IGs from the previous studies in the incongruent group (IS094, tp18, and tk11) further supports our approach of sRNA prediction to be reasonable. However, the information on IG expression using E. coli antisense Affymetrix chips is limited due to probe set design biased to the neighboring ORF transcription direction. Judging from the orientation of the probes, cDNA (see Material and Methods) of IS079 (downstream of hdhA) is only detectable by the probes, while cDNA from C0465 and tp2 cDNA cannot hybridize with the IG probes and the orientation of tp18, tp30 and tk11 is unknown. Regarding the signal intensity, many of IG showed lower signals than neighbors even though they were in congruent group by the induction ratio (e.g. IG_1073 of IS079?IG_1259 in operon? IG_2719 in promoter region). Same cases were found in the incongruent group (lower signal from IG_2430?IG_2599 (lower level). However, there were exceptional cases of stronger signals (IG_1512, IG_2280). Whether the signal difference across the IG and neighbors reflects real transcript level ? stability of mRNA or merely resulted from different efficiency of reverse transcription remains as a question. 120 IG ID log 2 IR start IG length flanking Affy. Characteristics? sRNA(orientation) Congruency ref IG_68 1.3 122857 160 pdhR/aceE >>> incomplete termination tp2 (<)* Rivas et al IG_507 -1.2 794146 166 b0762/modA >>> promoter IG_715 -1 1134735 52 flgG/flgH >>> operon IG_743 1.3 1191855 35 ycfC/ycfB <>> promoter IG_1073 -1.8 1695077 220 gusR/hdhA <<< incomplete termination IS079 (<) Chen et al IG_1258 -1.3 1969009 45 tap/tar <<< opreon IG_1259 -1.4 1970716 144 tar/cheW <<< operon C0465(>)* Chen et al. IG_1260 -2.5 1975164 126 motA/flhC <<< operon tp30(unknown) Rivas et al IG_1265 -1.3 1983093 69 araG/araF <<< operon IG_1283 -1.3 2004102 77 fliT/amyA >>> (stem and loop) promoter IG_1291 -1.2 2010803 448 fliE/IG_1292 <>> operon, promoter IG_1294 -1.1 2019524 62 fliN/fliO >>> operon IG_1335 1.1 2072681 114 flu/yeeR >>> incomplete termination, operon? C0486(<) Tjaden et al IG_1511 1.2 2426647 94 ubiX/purF <<< incomplete termination IG_1512 1.5 2428259 36 purF/cvpA <<< operon, strong signal IG_1621 1.4 2595639 212 purC/nlpB <<< stem and loop, terminater IG_1636 1.4 2630556 68 guaA/guaB <<< operon IG_1799 1 2886336 74 cysH/cysI <<< operon C0673(>) Tjaden et al. IG_2719 1.1 4402162 103 yjeT/purA >>> promoter mapping IG_2767 1.5 4468527 33 yjgF/pyrI <<< incomplete termination Incongruency 1-Opposite IG_132 1.1 228663 93 rrlH/rrfH >>> small intensity compared to the flanks IG_2133 2.1 3421180 37 rrfF/thrV <<< small intensity compared to the flanks IG_2429 1.2 3944232 92 rrlC/rrfC >>> small intensity compared to the flanks IG_2430 1 3944444 52 rrfC/aspT >>> small intensity compared to the flanks IG_2572 1.1 4169124 92 IG_2571/rrfB >>> IG_2599 1.4 4210526 93 rrlE/rrfE >>> small intensity compared to the flanks IG_1279 1.2 1994855 231 sdiA/yecC <<< IS094 (>) Chen et al. Incongruency 2-Indifference IG_1915 1.5 3084422 303 yqgD/metK <>> IG_2280 1.4 3693591 35 yhjQ/yhjR <<< IG_287 1.4 445897 142 yajR/cyoE <<< Small intensity IG_1222 1.1 1910601 191 htpX/prc <<< Small intensity tp18 (unknown) Rivas et al IG_1629 1.7 2618181 85 uraA/upp <<< Small intensity tk11 (unknown) Rivas et al IG_2172 1 3487818 85 argD/pabA <<< Small intensity Affy. Probe set orientation designed in antisense microarray ? Based on information from high-resolution probe analysis * Experimentally confirmed sRNA Table 9. Significant changes in IGs 121 Spacer Two Block Motifs Score Sites 8 CATTT (N)8 TCAAC 2.01 11 GACAA (N)8 GCAGA 2 7 TTGTG (N)8 AAAAG 1.96 11 CAGTT (N)8 TCAAG 1.96 12 GCAAT (N)8 ATGAT 1.95 11 AAAAG (N)8 GAAAA 1.95 7 9 TGCCA (N)9 ATGAA 2.14 12 TGACA (N)9 ATGAA 2.1 9 GCATT (N)9 TCAAG 2.09 9 GCATT (N)9 TCAAG 2.06 9 GCATT (N)9 TCAAG 2.06 9 TTTTG (N)9 ACAGT 2.05 9 10 TGCCC (N)10 ATGAA 2.25 13 TGTCA (N)10 ATGAA 2.24 13 11 TGCCA (N)11 TGAAG 2.26 15 TGCCC (N)11 TCAAG 2.26 15 TGTCA (N)11 ATGAA 2.21 12 12 AATGT (N)12 ATGAA 2.28 13 TGCCA (N)12 TGAAG 2.25 9 TGTCA (N)12 GAAAA 2.24 13 TGCCA (N)12 GAAGA 2.21 12 Weblog, Berkley Figure 19. Searching for SdiA binding motifs. 10 chosen IGs were used as inputs for analysis with BioProspector (Liu et al., 2002). Two block model (each block 5bp and the gap of 8 to 12 bp) was tested and most highly scored motifs are listed in the box along with the scores and the number of sites with the sequences. The most closely matching sequence to the ?SdiA box? is illustrated using WebLog (Berkely.). 1 2 3 4 5 (6 7 8 9 10 11 12 13) 14 15 16 17 18 122 Congruency Figure 20. Significantly changed IG regions with congruency. Probe intensities are shown as corrected probe intensity (PM-MM) for significantly changed IGs and the flanking ORFs. Individual probes are numbered as in Afffymetrix probe library but aligned on the map of clockwise strand (+) (x-axis), but not distributed in physical map scale. Upward bars indicate transcripts of clockwise direction and downward bars indicate the counterclockwise directional transcripts (left y-axis). Circles show the relative expression ratio of each probe (log 2 IR W2X2/W3110 , right y-axis). Represented IG expressions are considered to have resulted from incomplete transcription of upstream genes. 0 1000 2000 3000 4000 5000 L1 L6 L1 1 IG5 IG1 0 IG1 5 R4 R9 R1 4 -3 -2 -1 0 1 2 3 4 5 IG 68 pdhR ace -5000 -4000 -3000 -2000 -1000 0 L1 4 L9 L4 IG 1 4 IG 9 IG 4 R1 5 R1 0 R5 -8 -7 -6 -5 -4 -3 -2 -1 0 IG 1073 uid hdhA 0 500 1000 1500 2000 2500 3000 3500 4000 L1 L6 L1 1 IG 5 IG 1 0 IG 1 5 R4 R9 R1 4 0 0.5 1 1.5 2 2.5 3 3.5 4 IG 1335 flu b200 -5000 -4000 -3000 -2000 -1000 0 L1 4 L9 L4 IG 1 1 IG 6 IG 1 R1 1 R6 R1 0 0.5 1 1.5 2 2.5 3 3.5 4 IG 1512 purF cvpA -2000 -1500 -1000 -500 0 L1 5 L1 0 L5 IG 1 0 IG 5 R1 0 R5 0 0.5 1 1.5 2 2.5 3 3.5 4 IG 1511 ubi pur -10000 -8000 -6000 -4000 -2000 0 L1 4 L9 L4 IG 6 IG 1 R1 1 R6 R1 0 0.5 1 1.5 2 2.5 3 3.5 4 IG 2767 yjg pyrI 123 Incongruency-opposite Figure 21. Significantly changed IG regions with incongruency-opposite. Probe intensities are shown as corrected probe intensity (PM-MM) for significantly changed IGs and the flanking ORFs.Note that the five of the represented IGs are located upstream of rrf with the exception of IG_2430, located downstream of rrfC. Upward bars indicate transcripts of clockwise direction and downward bars indicate the counterclockwise directional transcripts (left y-axis). Circles show the relative expression ratio of each probe (log 2 IR W2X2/W3110 , right y-axis). Represented IG expressions are considered to have resulted from incomplete transcription of upstream genes. -25000 -20000 -15000 -10000 -5000 0 L1 L6 L1 1 L1 6 IG 5 -6 -4 -2 0 2 4 IG 2133 0 5000 10000 15000 20000 25000 L1 IG 5 R1 R6 -2 -1 0 1 2 3 4 IG 132 0 5000 10000 15000 20000 25000 L1 L6 L1 1 L1 6 IG 5 R1 R6 R1 1 -2 -1 0 1 2 3 4 IG 2429 0 5000 10000 15000 20000 25000 L1 L6 IG 4 IG 9 R3 -2 -1 0 1 2 3 4 IG 2599 0 5000 10000 15000 20000 25000 L1 L6 IG 3 IG 8 R4 R9 -6 -4 -2 0 2 4 IG 2572 0 5000 10000 15000 20000 25000 L1 L6 L1 1 IG 4 R4 -2 -1 0 1 2 3 4 IG 2430 rrlH rrfH rrfF thrV rrlC rrfC rrfC asp IG25 rrfB rrlE rrfE 124 Incongruency-indifference Figure 22. Significantly changed IG regions with incongruency-indifference. Probe intensities are shown as corrected probe intensity (PM-MM) for significantly changed IGs and the flanking ORFs.Notably, IG_2280 has strong signals not observed from the flanking genes.Upward bars indicate transcripts of clockwise direction and downward bars indicate the counterclockwise directional transcripts (left y-axis). Circles show the relative expression ratio of each probe (log 2 IR W2X2/W3110 , right y-axis). Represented IG expressions are considered to have resulted from incomplete transcription of upstream genes. -500 0 500 1000 1500 2000 2500 L1 4 L9 L4 IG 1 IG 6 IG 1 1 R5 R1 0 -1 0 1 2 3 4 IG 1915 yggD metK -2500 -2000 -1500 -1000 -500 0 L1 3 L8 L3 IG 9 IG 4 R1 5 R1 0 R5 0 0.5 1 1.5 2 2.5 3 3.5 4 IG 2280 yhjQ yhj -6000 -5000 -4000 -3000 -2000 -1000 0 L1 4 L9 L4 IG 1 5 IG 1 0 IG 5 R1 0 R5 -1 0 1 2 3 4 IG 287 yajR cyoE -3000 -2500 -2000 -1500 -1000 -500 0 L1 5 L1 0 L5 IG 1 1 IG 6 IG 1 R1 1 R6 R1 -1 0 1 2 3 4 IG 1222 prc htpX -3000 -2500 -2000 -1500 -1000 -500 0 L1 5 L1 0 L5 IG 1 1 IG 6 IG 1 R1 1 R6 R1 -3 -2 -1 0 1 2 3 4 IG 1629 upp ura -3000 -2500 -2000 -1500 -1000 -500 0 L1 L6 L1 1 IG 1 0 IG 5 R1 0 R5 -2 -1 0 1 2 3 4 5 IG 2172 argD pabA 125 Chapter 4: Conclusions Multiple quorum signalings in one organism: testing the regulatory effect from system 2 to system 1 of Escherichia coli. As introduced in Chapter 1, quorum sensing is mediated by cell signals emitted to the environment that are either diverse (species specific) and/or conserved (universal signal) and are then perceived by cells modulating activity of specific regulatory receptor components. Bacterial quorum sensing, in many cases, is composed of plural numbers of quorum sensory, which then integrate into a global network either in a parallel or in hierarchical manner for timely and effective gene expression. Escherichia coli K-12, the experimental organism investigated in this study, also has indications of multiple-quorum signaling. We denoted one of the quorum systems in E. coli as system 1 represented by LuxR type protein SdiA and the other quorum system as system 2 by universal signal molecule AI-2. Importantly, we have noticed that the co- presence of sdiA and pfs/luxS in an organism is unique to ? -proteobacteriacae of E. coli, Salmonella subspecies, Y. pseudotuberculosis, V. parahaemolyticus, and V. cholerae (NCBI BLASTX search, unidirectional homology search). Unlike Y. pseudotuberculosis and Vibrio species producing AI-1 signals, E. coli and Salmonella do not have the LuxI type proteins nor have the detectable AI-1 activity making the system 1 found in these organisms seemingly incomplete. The unique architecture of E. coli quorum sensing raised questions about the functionality of E. coli quorum components and the relationship between the systems and their association with established global regulatory networks. I tried to answer some of these questions in this dissertation. Chapter 2 highlighted the impacts of environmental 126 factors on the quorum sensing system 2. In particular, cultivation temperature and heat shock were examined to delineate the apparent crosswiring of quorum system 2 with quorum system 1, and to attempt to understand how the quorum system 2 signal, AI2, is synthesized and degraded in response to environmental stresses. To date, system 1 and system 2 have been considered independent of one another. Using a confirmed SdiA target promoter, ftsQ 2p of E. coli, possible AI-2 modulation (using conditioned media) of SdiA activity in E. coli was determined to be negative by two research groups (Michael et al., 2001; Surette and Bassler, 1999). Although not described in this study as preliminary studies, we also confirmed that regulatory effects of AI-2 may not directly affect SdiA in flask cultures by using a luxS knockout mutant (data not shown). Also we were unable to demonstrate direct stable binding of AI-2 to SdiA when tested by mixing synthetic AI-2 with purified His 6 -tagged SdiA protein (data not shown). Theoretically, the quantity of unbound and bound AI-2 could be measured by separating the appropriate receptor protein fraction from the solution. However, aggregation of SdiA and the limits of the quantitative assay for AI-2 activity were too difficult to overcome, leaving room for improvement in the future before making a conclusion. Functionality of SdiA examined by deletion (chapter 3) Naturally, we questioned the appropriateness of relying on the single identified target promoter (ftsQ 2p, cell division associated) as a probe of SdiA functionality. Especially since its signal responsiveness is complicated by being associated with cell division (cell divisions are affected by numerous factors). To screen more of the SdiA- 127 dependent genes, we performed a comparative microarray analysis between wild type W3110 and the sdiA - mutant, W2X2. At the same time, we looked for phenotypic differences resulting from the sdiA deletion, such as changes in stress resistance and reduced motility. Using the high throughput technology of microarrays, we obtained valuable information on the genomewide transcriptome changes, including all open reading frames (ORFs) and intergenic regions (IGs). We noted two intriguing classes of results signifying marked changes in expression. Namely, the sdiA - strain had downregulated a group of genes assigned to AI-2 uptake and modification (lsr operon, b1511-b1518) and secondarily, an ? -hydroxysteroid dehydrogenase for modifying bile acids (hdh) was also observed to be reduced in expression. Changes associated with the lsr operon (microarray data, transcriptional fusion assay, and AI-2 activity) are presented as conclusive evidence of cross-communication between system 1 and system 2, in which system 1 positively affects system 2 in the regulatory network. As bile acids can function as physiological ligands for the farnesoid X receptor controlling cholesterol and bile acid homeostasis (Makishima et al., 1999; Wang et al., 1999), the change in E. coli hdh expression is highly likely to associate the sdiA-dependent E. coli signaling to a mammalian signaling in the intestine. Along with the discovery of a mammalian hormone influencing enterohemorrhagic E. coli virulence genes (Sperandio et al., 2003), symbiotic- or pathogenic interaction of E. coli by interkingdom signaling is more prevalent than we expected. Studies testing this idea are expected in the future. From the altered transcriptome in the sdiA mutant, we noted changes in other transcriptional regulators of arcA, soxS, rpoE and rpoS. Selecting a specific target of each 128 regulator, which were also changed by sdiA deletion, left the questions as to whether the changes are solely subsequent to the changes of global regulatory proteins (ArcA, SoxS, RpoE and RpoS) or is SdiA directly regulating them in a competitive or cooperative manner with other regulators. The same question arose from the unexpected orphan expression of IGs with significant changes, which we suggest as possible sRNA. We expect that cross- or synergistic- regulatory cascades involving the other regulatory proteins or sRNA could be, when combined, a clue for explaining differences in rRNA and stress resistance developed by the mutation. Of practical use, we performed an in silico study, wherein 10 IGs were searched based on several criteria, for conserved sequences with a topology of ?LuxR family protein? binding sites. Experimental confirmation of the searched motifs will expand the list of SdiA-dependent targets and also provide ?sensors? for quantifying SdiA activity. While our study on SdiA aimed to find its functionality in the endogenous system of pure cultures, continuous efforts towards understanding interactions between other bacteria have been reported through ?eavesdropping? signals originated from other bacterial species or through competitive binding on exogenous AI-1 sensor plasmids (Smith and Ahmer, 2003)(Lindsay and Ahmer, 2005). Thus, we can say more about SdiA function, but mechanisms of action are yet to be elucidated. 129 Figure 23. Suggested convergence of E. coli quorum system 1 and 2. Previously divided E. coli quorum system 1 and 2 are now illustrated to be converged on the AI-2 uptake machinery encoding genes, lsr operon. Whether the positive activation of SdiA on the lsr operon is direct or indirect has not yet determined. The SdiA-mediated transcriptional activation on ftsQ2p was not affected by AI-2 in the flask culture condition (data not shown here). LuxS P SdiA LsrR Without phospho-AI-2 LsrR P With phospho-AI-2 LsrB LsrD LsrA LsrC LsrA LsrK lsrACDBFGE lsrKR ftsQAZ P2 OM IM cAMP-CRP LuxR-type ? AI-2 130 Autoinducer 2 (AI-2) level is actively regulated by molecular chaperones under stress (Chapter2) The second quorum system has the functional signal, AI-2. AI-2 production is closely linked to detoxifying SAH and recycling homocysteine. However, by using the two enzymatic conversion of SAH to homocysteine, most eubacteria can produce AI-2, which is not produced in eukaryotes. Apart from the active studies on functionality of AI- 2, our work in Chapter 2 was focused on the impacts of environmental stimuli, high temperature, and the regulatory effects of molecular chaperones, DnaJKE and GroESL. When cells have wild type GroESL and DnaJKE, high temperature (42 ?C) incubation reduced AI-2 levels in glucose supplemented cultures. Interestingly, cultures of groESL point mutants, dnaJ knockouts, and grpE280 mutants at 42 ?C in the presence of glucose did not decrease AI-2 levels even after the cells entered the stationary phase, possibly due to defects in growth. From these results, we concluded that system 2 is connected to global signaling, mediated by molecular chaperones. There are several reports on the regulatory effects of environmental factors on AI-2 production (DeLisa et al., 2001, 2001; Hardie et al., 2003; Surette and Bassler, 1999). None of them reported the ?negative? regulatory aspects of these chaperones. The negative regulatory effects of chaperones can be described as follows: glucose drives metabolic carbon flux continuously towards AI-2 synthesis. At 42 ?C, non- functional chaperones (either by point mutations or by deletion) do not exert their negative regulatory roles on their normal targets in the AI-2 production pathway, letting them produce AI-2 (possibly through the interaction with proteases). 131 Observations from high temperature experiments give insight into the activity of essential AI-2 synthase LuxS, which should be structurally stable and functionally active at 42 ?C. Indeed, stable maintenance of active LuxS in the stationary phase cells has been reported (Hardie et al., 2003; Xavier and Bassler, 2005), however its temperature dependent activity has not yet been reported. If LuxS were not the main targets of these chaperones, what protein could interact with the chaperones, directly or indirectly, to and directly govern AI-2 synthesis? One attractive candidate could be found from methionine de novo synthesis supplying the SAM for AI-2 synthesis. Thermolabile homoserine O- succinyltransferase (MetA, see Fig. 24), the first enzyme in the methionine synthesis is known to be structurally unstable, tending to aggregate at the temperature higher than 44 ?C. Its interaction with DnaJ and protease ClpB has been reported (Biran et al., 2000; Gur et al., 2002)(Mogk et al., 1999). Further biochemical analyses on protein-protein interaction and observing half-life of possible substrates of the chaperones are required to test this idea. 132 Figure 24. Metabolic flux for AI-2 synthesis. Sulfur is assimilated to amino acids, cysteine and methionine. Homocysteine is generated from cysteine and oxaloacetate in the de novo synthesis of methionine. MetK converts methionine to S-adenosyl methionine (SAM) using ATP. Further transform of SAM to AI-2 is depicted in Figure1.While most bacteria use the two-step conversion of SAH to homocysteine (Pfs and LuxS mediated), one-step conversion also occurs in eukaryotes and some archae bacteria as depicted (SAH hydrolase, SahH*). Major enzyme encoding genes are indicated with gray ovals. The figure was modified from (Vendeville et al., 2005). SO 4 2 S 2- Cysteine cysK O-Succinylhomoserine Oxaloacetate Aspartate Homoserine metA metB Cystathionine metC Pfs LuxS X X-CH 3 DPD SahH* metK metE metH O CH 3 OH HO HO HO AI-2: R-THMF (LsrB ligand) SRH SAM SAH Methionine Homocysteine 133 Significance of the study for potential applications The key element in controlling pathogens by manipulating quorum signaling resides the specificity of the signal action not to place unwanted selected pressure on the organism. Although the chosen E. coli K-12 model system is not equipped with the virulence genes for direct assessment of the manipulated quorum system 1, still, relatively small numbers of changed genes by the absence of the sensor protein expression suffices the first requirement. Remaining issues on how to augment the strength of the quorum sensory can be accomplished by searching effectors and screening the target genes, which will automatically lead to develop as ?inducible system? for recombinant protein expression. The in silico study, performed based on the microarray data and literature-based comparative analysis, is going to be tested in vitro and in vivo as an important reference. Besides, associated with the intriguing phenotypic characteristics, such as the accelerated autoaggregation, screened SdiA-dependent promoter is expected to be developed as a biosensor to study further bacterial social behavior under various environmental conditions (both natural habitats and harsh conditions) and to screen inhibitory chemical substances with highly specificity on the quorum sensory (as mentioned above). This needs to be distinguished from the global regulatory networks initiated from abiotic environmental cues, which was well exemplified in Chapter 2. Perspectives Taken together quorum sensing of E. coli is in a finely tuned and regulated state are globally interconnected with other sensory systems. Unraveling this sensing mechanism is difficult in the flask cultures where various quorum inhibitory materials may accumulate over time. However difficult, study of E. coli quorum signaling is 134 progressively being understood and is likely to provide very real insight into mechanisms of bacterial virulence and pathogenicity and also commensalism. 135 Bibliography Ahmer, B.M.M., van Reeuwijk, J., Timmers, C.D., Valentine, P.J., and Heffron, F. (1998) Salmonella typhimurium encodes an SdiA homolog, a putative quorum sensor of the LuxR family, that regulates genes on the virulence plasmid. Journal of Bacteriology 180: 1185-1193. Ahmer, B.M.M. (2004) Cell-to-cell signalling in Escherichia coli and Salmonella enterica. Molecular Microbiology 52: 933-945. Albus, A., Pesci, E., RunyenJanecky, L., West, S., and Iglewski, B. (1997) Vfr controls quorum sensing in Pseudomonas aeruginosa. Journal of Bacteriology 179: 3928- 3935. Aoki, K., Motojima, F., Taguchi, H., Yomo, T., and Yoshida, M. (2000) GroEL binds artificial proteins with random sequences. Journal of Biological Chemistry 275: 13755-13758. Atkinson, S., Throup, J., Stewart, G., and Williams, P. (1999) A hierarchical quorum- sensing system in Yersinia pseudotuberculosis is involved in the regulation of motility and clumping. Molecular Microbiology 33: 1267-1277. Barbosa, T., and Levy, S. (2000) Differential expression of over 60 chromosomal genes in Escherichia coli by constitutive expression of MarA. Journal of Bacteriology 182: 3467-3474. Bassler, B., Greenberg, E., and Stevens, A. (1997) Cross-species induction of luminescence in the quorum-sensing bacterium Vibrio harveyi. Journal of Bacteriology 179: 4043-4045. Bassler, B.L., Wright, M., Showalter, R.E., and Silverman, M.R. (1993) Intercellular Signaling in Vibrio harveyi - Sequence and Function of Genes Regulating Expression of Luminescence. Molecular Microbiology 9: 773-786. Beloin, C., Valle, J., Latour-Lambert, P., Faure, P., Kzreminski, M., Balestrino, D., Haagensen, J., Molin, S., Prensier, G., Arbeille, B., and Ghigo, J. (2004) Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression. Molecular Microbiology 51: 659-674. Bentley, W.E., Mirjalili, N., Andersen, D., Davis, R., and Kompala, D. (1990) Plasmid- encoded protein-the principal factor in the metabolic burden associated with recombinant bacteria. Biotechnology and Bioengineering 35: 668-681. Biran, D., Gur, E., Gollan, L., and Ron, E. (2000) Control of methionine biosynthesis in Escherichia coli by proteolysis. Molecular Microbiology 37: 1436-1443. Blosser, R., and Gray, K. (2000) Extraction of violacein from Chromobacterium violaceum provides a new quantitative bioassay for N-acyl homoserine lactone autoinducers. Journal of Microbiological Methods 40: 47-55. Bukau, B. (1993) Regulation of the Escherichia coli heat shock response. Molecular Microbiology 9: 671-680. Carter, R., Dubchak, I., and Holbrook, S. (2001) A computational approach to identify genes for functional RNAs in genomic sequences. Nucleic Acids Research 29: 3928-3938. ChaponHerve, V., Akrim, M., Latifi, A., Williams, P., Lazdunski, A., and Bally, M. (1997) Regulation of the xcp secretion pathway by multiple quorum-sensing modulons in Pseudomonas aeruginosa. Molecular Microbiology 24: 1169-1178. 136 Chen, S., Lesnik, E., Hall, T., Sampath, R., Griffey, R., Ecker, D., and Blyn, L. (2002) A bioinformatics based approach to discover small RNA genes in the Escherichia coli genome. Biosystems 65: 157-177. Chen, X., Schauder, S., Potier, N., Van Dorsselaer, A., Pelczer, I., Bassler, B.L., and Hughson, F.M. (2002) Structural identification of a bacterial quorum-sensing signal containing boron. Nature 415: 545-549. Choi, S.H., and Greenberg, E.P. (1991) The C-terminal region of the Vibrio fischeri LuxR protein contains an inducer-independent lux gene activating domain. PNAS 88: 11115-11119. Chung, W., Park, Y., Lamont, R., McNab, R., Barbieri, B., and Demuth, D. (2001) Signaling system in Porphyromonas gingivalis based on a LuxS protein. Journal of Bacteriology 183: 3903-3909. Cloak, O., Solow, B., Briggs, C., Chen, C., and Fratamico, P. (2002) Quorum sensing and production of autoinducer-2 in Campylobacter spp., Escherichia coli O157: H7, and Salmonella enterica serovar typhimurium in foods. Applied and Environmental Microbiology 68: 4666-4671. Cloak, O., Solow, B., Briggs, C., Chen, C., and Fratamico, P. (2002) Quorum sensing and production of autoinducer-2 in Campylobacter spp., Escherichia coli O157: H7, and Salmonella enterzca serovar typhimurium in foods. Applied and Environmental Microbiology 68: 4666-4671. Coves, J., Delon, B., Climent, I., Sjoberg, B., and Fontecave, M. (1995) Enzymatic and chemical-reduction of the iron center of the Escherichia coli ribonucleotide reductase protein R2 the role of the C-terminus. European Journal of Biochemistry 233: 357-363. Craig, E., and Gross, C. (1991) Is hsp70 the cellular thermometer? Trends in Biochemical Sciences 16: 135-140. Dartigalongue, C., Missiakas, D., and Raina, S. (2001) Characterization of the Escherichia coli sigma(E) regulon. Journal of Biological Chemistry 276: 20866- 20875. Davies, D., Parsek, M., Pearson, J., Iglewski, B., Costerton, J., and Greenberg, E. (1998) The involvement of cell-to-cell signals in the development of a bacterial biofilm. Science 280: 295-298. DeLisa, M., Valdes, J., and Bentley, W. (2001) Mapping stress-induced changes in autoinducer AI-2 production in chemostat-cultivated Escherichia coli K-12. Journal of Bacteriology 183: 2918-2928. DeLisa, M., Valdes, J., and Bentley, W. (2001) Quorum signaling via Al-2 communicates the "metabolic burden" associated with heterologous protein production in Escherichia coli. Biotechnology and Bioengineering 75: 439-450. DeLisa, M., Wu, C., Wang, L., Valdes, J., and Bentley, W. (2001) DNA microarray- based identification of genes controlled by autoinducer 2-stimulated quorum sensing in Escherichia coli. Journal of Bacteriology 183: 5239-5247. Dolan, K., and Greenberg, E. (1992) Evidence that GroEL, not Sigm-32, is involved in transcriptional regulation of the Vibrio fischeri luminescence genes in Escherichia coli. Journal of Bacteriology 174: 5132-5135. 137 Dong, Y., Xu, J., Li, X., and Zhang, L. (2000) AiiA, an enzyme that inactivates the acylhomoserine lactone quorum-sensing signal and attenuates the virulence of Erwinia carotovora. PNAS 97: 3526-3531. Duan, K., Dammel, C., Stein, J., Rabin, H., and Surette, M. (2003) Modulation of Pseudomonas aeruginosa gene expression by host microflora through interspecies communication. Molecular Microbiology 50: 1477-1491. Dukan, S., and Nystrom, T. (1998) Bacterial senescence: stasis results in increased and differential oxidation of cytoplasmic proteins leading to developmental induction of the heat shock regulon. Genes & Development 12: 3431-3441. Dunlap, P., and Greenberg, E. (1988) Control of Vibrio fischeri lux gene transcription by a cyclic-AMP receptor protein LuxR protein regulatory circuit. Journal of Bacteriology 170: 4040-4046. Dunlap, P. (1992) Iron control of the Vibrio fischeri luminescence system in Escherichia coli. Archives of Microbiology 157: 235-241. Dunlap, P., and Kuo, A. (1992) Cell density-dependenet modulation of the Vibrio fischeri luminescence system in the absence of autoinducer and LuxR protein. Journal of Bacteriology 174: 2440-2448. Eberhard, A. (1972) Inhibition and activation of bacterial luciferase synthesis. Journal of Bacteriology 109: 1101-&. Egland, K., and Greenberg, E. (2000) Conversion of the Vibrio fischeri transcriptional activator, LuxR, to a repressor. Journal of Bacteriology 182: 805-811. Fath, M., Mahanty, H., and Kolter, R. (1989) Characterization of a purF operon mutation which affects colicin-V production. Journal of Bacteriology 171: 3158-3161. Fayet, O., Ziegelhoffer, T., and Georgopoulos, C. (1989) The GroES and GroEL heat- shock gene-products of Eshcerichia coli are essensital for bacterial-growth at all temperatures. Journal of Bacteriology 171: 1379-1385. Fontecave, M., Atta, M., and Mulliez, E. (2004) S-adenosylmethionine: nothing goes to waste. Trends in Biochemical Sciences 29: 243-249. Freeman, J., Lilley, B., and Bassler, B. (2000) A genetic analysis of the functions of LuxN: a two-component hybrid sensor kinase that regulates quorum sensing in Vibrio harveyi. Molecular Microbiology 35: 139-149. Frees, D., Vogensen, F., and Ingmer, H. (2003) Identification of proteins induced at low pH in Lactococcus lactis. International Journal of Food Microbiology 87: 293- 300. Fuqua, C., and Winans, S. (1996) Conserved cis-acting promoter elements are required for density-dependent transcription of Agrobacterium tumefaciens conjugal transfer genes. Journal of Bacteriology 178: 435-440. Fuqua, C., Winans, S., and Greenberg, E. (1996) Census and consensus in bacterial ecosystems: The LuxR-LuxI family of quorum-sensing transcriptional regulators. Annual Review of Microbiology 50: 727-751. Fuqua, C., Parsek, M., and Greenberg, E. (2001) Regulation of gene expression by cell- to-cell communication: Acyl-homoserine lactone quorum sensing. Annual Review of Genetics 35: 439-468. Fuqua, W., Winans, S., and Greenberg, E. (1994) Quorum sensing in bacteria:the LuxR- LuxI family of cell density-responsive transcriptional reuglators. Journal of Bacteriology 176: 269-275. 138 Gottesman, S. (2004) The small RNA regulators of Escherichia coli: Roles and mechanisms. Annual Review of Microbiology 58: 303-328. Graf, J., and Ruby, E. (1998) Host-derived amino acids support the proliferation of symbiotic bacteria. Proceedings of the National Academy of Sciences of the Unites States of America 95: 1818-1822. Gray, K., and Garey, J. (2001) The evolution of bacterial LuxI and LuxR quorum sensing regulators. Microbiology-SGM 147: 2379-2387. Gur, E., Biran, D., Gazit, E., and Ron, E. (2002) In vivo aggregation of a single enzyme limits growth of Escherichia coli at elevated temperatures. Molecular Microbiology 46: 1391-1397. Hagewood, B., Ganduri, Y., and Datta, P. (1994) Functional-analysis of the tdcABC promoter of Eschecrichia coli: Roles of TdcA and TdcR. Journal of Bacteriology 176: 6214-6220. Hanzelka, B., and Greenberg, E. (1996) Quorum sensing in Vibrio fischeri: Evidence that S-adenosylmethionine is the amino acid substrate for autoinducer synthesis. Journal of Bacteriology 178: 5291-5294. Hanzelka, B., Parsek, M., Val, D., Dunlap, P., Cronan, J., and Greenberg, E. (1999) Acylhomoserine lactone synthase activity of the Vibrio fischeri AinS protein. Journal of Bacteriology 181: 5766-5770. Hardie, K., Cooksley, C., Green, A., and Winzer, K. (2003) Autoinducer 2 activity in Escherichia coli culture supernatants can be actively reduced despite maintenance of an active synthase, LuxS. Microbiology-SGM 149: 715-728. Hartl, F. (1997) Molecular chaperone action in protein folding. FASEB Journal 11: A1006-A1006. Henderson, I., and Owen, P. (1999) The major phase-variable outer membrane protein of Escherichia coli structurally resembles the immunoglobulin A1 protease class of exported protein and is regulated by a novel mechanism involving Dam and OxyR. Journal of Bacteriology 181: 2132-2141. Henke, J., and Bassler, B. (2004) Three parallel quorum-sensing systems regulate gene expression in Vibrio harveyi. Journal of Bacteriology 186: 6902-6914. Hotta, E., Janne, J., and Pispa, J. (1972) Ornithine decarboxylase from Escherichia coli: stimulation of the enzyme activity by nucleotides. Biochemical and Biophysical Research Communications 47: 1165-&. Horwich, A., Low, K., Fenton, W., Hirshfield, I., and Furtak, K. (1993) Folding in vivo of Bacterial cytoplasic proteins-role of GroEL. Cell 74: 909-917. Hua, Q., Yang, C., Oshima, T., Mori, H., and Shimizu, K. (2004) Analysis of gene expression in Escherichia coli in response to changes of growth-limiting nutrient in chemostat cultures. Applied and Environmental Microbiology 70: 2354-2366. Ishihama, A. (1999) Modulation of the nucleoid, the transcription apparatus, and the translation machinery in bacteria for stationary phase survival. Genes to Cells 4: 135-143. Kanamaru, K., Kanamaru, K., Tatsuno, I., Tobe, T., and Sasakawa, C. (2000) SdiA, an Escherichia coli homologue of quorum-sensing regulators, controls the expression of virulence factors in enterohaemorrhagic Escherichia coli O157: H7. Molecular Microbiology 38: 805-816. 139 Kaplan, H., and Greenberg, E. (1985) Diffusion of autoinducer is involved in regulation of the Vibrio fischeri luminescence system. Journal of Bacteriology 163: 1210- 1214. King, J., HaasePettingell, C., Robinson, A., Speed, M., and Mitraki, A. (1996) Thermolabile folding intermediates: Inclusion body precursors and chaperonin substrates. FASEB Journal 10: 57-66. Kirkpatrick, C., Maurer, L., Oyelakin, N., Yoncheva, Y., Maurer, R., and Slonczewski, J. (2001) Acetate and formate stress: Opposite responses in the proteome of Escherichia coli. Journal of Bacteriology 183: 6466-6477. Kjaergaard, K., Schembri, M., Ramos, C., Molin, S., and Klemm, P. (2000) Antigen 43 facilitates formation of multispecies biofilms. Environmental Microbiology 2: 695-702. Knutsen, E., Ween, O., and Havarstein, L.S. (2004) Two separate quorum-sensing systems upregulate transcription of the same ABC transporter in Streptococcus pneumoniae. Journal of Bacteriology 186: 3078-3085. Latifi, A., Foglino, M., Tanaka, K., Williams, P., and Lazdunski, A. (1996) A hierarchical quorum-sensing cascade in Pseudomonas aeruginosa links the transcriptional activators LasR and RhIR (VsmR) to expression of the stationary- phase sigma factor RpoS. Molecular Microbiology 21: 1137-1146. Laue, R., Jiang, Y., Chhabra, S., Jacob, S., Stewart, G., Hardman, A., Downie, J., O'Gara, F., and Williams, P. (2000) The biocontrol strain Pseudomonas fluorescens F113 produces the Rhizobium small bacteriocin, N-(3-hydroxy-7-cis- tetradecenoyl)homoserine lactone, via HdtS, a putative novel N-acylhomoserine lactone synthase. Microbiology-UK 146: 2469-2480. Leadbetter, J., and Greenberg, E. (2000) Metabolism of acyl-homoserine lactone quorum- sensing signals by Variovorax paradoxus. Journal of Bacteriology 182: 6921- 6926. Lee, J., Cornell, K., Riscoe, M., and Howell, P. (2003) Structure of Escherichia coli 5 '- methylthioadenosine/S-adenosylhomocysteine nucleosidase inhibitor complexes provide insight into the conformational changes required for substrate binding and catalysis. Journal of Biological Chemistry 278: 8761-8770. Lee, K., and Ruby, E. (1994) Effect of the squid host on the abundance and distrubution of symbiotic Vibrio fischeri in nature. Applied and Environmental Microbiology 60: 1565-1571. Lee, K., and Ruby, E. (1994) Competition between Vibrio fischeri strains during initiation and maintenance of a light organ symbiosis. Journal of Bacteriology 176: 1985-1991. Lenz, D., Mok, K., Lilley, B., Kulkarni, R., Wingreen, N., and Bassler, B. (2004) The small RNA chaperone Hfq and multiple small RNAs control quorum sensing in Vibrio harveyi and Vibrio cholerae. Cell 118: 69-82. Lerat, E., and Moran, N. (2004) Evolutionary history of quorum-sensing systems in bacteria. Molecular Biology and Evolution 21: 903-913. Lerat, E., and Moran, N.A. (2004) The evolutionary history of quorum-sensing systems in bacteria (Vol 21, pg 903, 2004). Molecular Biology and Evolution 21: 1612- 1612. 140 Li, Y., Lau, P., Tang, N., Svensater, G., Ellen, R., and Cvitkovitch, D. (2002) Novel two- component regulatory system involved in biofilm formation and acid resistance in Streptococcus mutans. Journal of Bacteriology 184: 6333-6342. Lilley, B., and Bassler, B. (2000) Regulation of quorum sensing in Vibrio harveyi by LuxO and Sigma-54. Molecular Microbiology 36: 940-954. Liu, X.S., Brutlag, D.L., and Liu, J.S. (2002) An algorithm for finding protein-DNA binding sites with applications to chromatin-immunoprecipitation microarray experiments. Nature Biotechnology 20: 835-839. Lupp, C., Urbanowski, M., Greenberg, E., and Ruby, E. (2003) The Vibrio fischeri quorum-sensing systems ain and lux sequentially induce luminescence gene expression and are important for persistence in the squid host. Molecular Microbiology 50: 319-331. Lupp, C., and Ruby, E. (2005) Vibrio fischeri uses two quorum-sensing systems for the regulation of early and late colonization factors. Journal of Bacteriology 187: 3620-3629. Lyon, W., Madden, J., Levin, J., Stein, J., and Caparon, M. (2001) Mutation of luxS affects growth and virulence factor expression in Streptococcus pyogenes. Molecular Microbiology 42: 145-157. Ma, Z., Richard, H., Tucker, D., Conway, T., and Foster, J. (2002) Collaborative regulation of Escherichia coli glutamate-dependent acid resistance by two AraC- like regulators, GadX and GadW (YhiW). Journal of Bacteriology 184: 7001- 7012. Makemson, J. (1986) Luciferase-dependent oxygen-consumption by bioluminescent Vibrios. Journal of Bacteriology 165: 461-466. Makishima, M., Okamoto, A., Repa, J., Tu, H., Learned, R., Luk, A., Hull, M., Lustig, K., Mangelsdorf, D., and Shan, B. (1999) Identification of a nuclear receptor for bile acids. Science 284: 1362-1365. March, J., and Bentley, W. (2004) Quorum sensing and bacterial cross-talk in biotechnology. Current Opinion in Biotechnology 15: 495-502. Martin, R.G., and Rosner, J.L. (2002) Genomics of the marA/soxS/rob regulon of Escherichia coli: identification of directly activated promoters by application of molecular genetics and informatics to microarray data. Molecular Microbiology 44: 1611-1624. McDougald, D., Srinivasan, S., Rice, S., and Kjelleberg, S. (2003) Signal-mediated cross- talk regulates stress adaptation in Vibrio species. Microbiology-SGM 149: 1923- 1933. McFall-Ngai, M. (1990) Crypsis in the pelagic environment. American zoologist 30: 175- 188. McFall-Ngai, M. (2002) Unseen forces: The influence of bacteria on animal development. Developmental Biology 242: 1-14. Medina, G., Juarez, K., Valderrama, B., and Soberon-Chavez, G. (2003) Mechanism of Pseudomonas aeruginosa RhlR transcriptional regulation of the rhlAB promoter. Journal of Bacteriology 185: 5976-5983. Meng, L., Kilstrup, M., and Nygaard, P. (1990) Autoregulation of PurR repressor synthesis and involvement of PurR in the reuglation of purB,purC,purL, purMN 141 and guaBA expression in Escherichia coli. European Journal of Biochemistry 187: 373-379. Merritt, J., Qi, F., Goodman, S., Anderson, M., and Shi, W. (2003) Mutation of luxS affects biofilm formation in Streptococcus mutans. Infection and Immunity 71: 1972-1979. Michael, B., Smith, J., Swift, S., Heffron, F., and Ahmer, B. (2001) SdiA of Salmonella enterica is a LuxR homolog that detects mixed microbial communities. Journal of Bacteriology 183: 5733-5742. Miller, S., Xavier, K., Campagna, S., Taga, M., Semmelhack, M., Bassler, B., and Hughson, F. (2004) Salmonella typhimurium recognizes a chemically distinct form of the bacterial quorum-sensing signal Al-2. Molecular Cell 15: 677-687. Minogue, T., Wehland-von Trebra, M., Bernhard, F., and von Bodman, S. (2002) The autoregulatory role of EsaR, a quorum-sensing regulator in Pantoea stewartii ssp stewartii: evidence for a repressor function. Molecular Microbiology 44: 1625- 1635. Missiakas, D., Mayer, M., Lemaire, M., Georgopoulos, C., and Raina, S. (1997) Modulation of the Escherichia coli sigma(E) (RpoE) heat-shock transcription- factor activity by the RseA, RseB and RseC proteins. Molecular Microbiology 24: 355-371. Miyamoto, C., Dunlap, P., Ruby, E., and Meighen, E. (2003) LuxO controls LuxR expression in Vibrio harveyi: evidence for a common regulatory mechanism in Vibrio. Molecular Microbiology 48: 537-548. Mogk, A., Tomoyasu, T., Goloubinoff, P., Rudiger, S., Roder, D., Langen, H., and Bukau, B. (1999) Identification of thermolabile Escherichia coli proteins: prevention and reversion of aggregation by DnaK and ClpB. EMBO Journal 18: 6934-6949. Monterrubio, R., Baldoma, L., Obradors, N., Aguilar, J., and Badia, J. (2000) A common regulator for the operons encoding the enzymes involved in D-galactarate, D- glucarate, and D-glycerate utilization in Escherichia coli. Journal of Bacteriology 182: 2672-2674. More, M., Finger, L., Stryker, J., Fuqua, C., Eberhard, A., and Winans, S. (1996) Enzymatic synthesis of a quorum-sensing autoinducer through use of defined substrates. Science 272: 1655-1658. Mujacic, M., Bader, M., and Baneyx, F. (2004) Escherichia coli Hsp31 functions as a holding chaperone that cooperates with the DnaK-DnaJ-GrpE system in the management of protein misfolding under severe stress conditions. Molecular Microbiology 51: 849-859. Nasser, W., Bouillant, M., Salmond, G., and Reverchon, S. (1998) Characterization of the Erwinia chrysanthemi expl-expR locus directing the synthesis of two N-acyl- homoserine lactone signal molecules. Molecular Microbiology 29: 1391-1405. Nealson, K., Platt, T., and Hastings, J. (1970) Cellular control of syntehsis and activity of bacterial luminescent system. Journal of bacteriology 104: 313-&. Newton, J., and Fray, R. (2004) Integration of environmental and host-derived signals with quorum sensing during plant-microbe interactions. Cellular Microbiology 6: 213-224. 142 Ohtani, K., Hayashi, H., and Shimizu, T. (2002) The luxS gene is involved in cell-cell signalling for toxin production in Clostridium perfringens. Molecular Microbiology 44: 171-179. Parsek, M., Val, D., Hanzelka, B., Cronan, J., and Greenberg, E. (1999) Acyl homoserine-lactone quorum-sensing signal generation. PNAS 96: 4360-4365. Parsek, M., and Greenberg, E. (2000) Acyl-homoserine lactone quorum sensing in Gram- negative bacteria: A signaling mechanism involved in associations with higher organisms. PNAS 97: 8789-8793. Paul, B., Ross, W., Gaal, T., and Gourse, R. (2004) rRNA transcription in Escherichia coli. Annual Review of Genetics 38: 749-770. Pomposiello, P., Bennik, M., and Demple, B. (2001) Genome-wide transcriptional profiling of the Escherichia coli responses to superoxide stress and sodium salicylate. Journal of Bacteriology 183: 3890-3902. Postma, P., Lengeler, J., and Jacobson, G. (1993) Phosphoenolpyruvate - carbohydrate phosphotransferase systems of bacteria. Microbiological reviews 57: 543-594. Rasmussen, T., Manefield, M., Andersen, J., Eberl, L., Anthoni, U., Christophersen, C., Steinberg, P., Kjelleberg, S., and Givskov, M. (2000) How Delisea pulchra furanones affect quorum sensing and swarming motility in Serratia liquefaciens MG1. Microbiology-UK 146: 3237-3244. Ravn, L., Christensen, A., Molin, S., Givskov, M., and Gram, L. (2001) Methods for detecting acylated homoserine lactones produced by Gram-negative bacteria and their application in studies of AHL-production kinetics. Journal of Microbiological Methods 44: 239-251. Reid, B., and Flynn, G. (1996) GroEL binds to and unfolds rhodanese posttranslationally. Journal of Biological Chemistry 271: 7212-7217. Ren, D., Bedzyk, L., Thomas, S., Ye, R., and Wood, T. (2004) Gene expression in Escherichia coli biofilms. Applied Microbiology and Biotechnology 64: 515-524. Reverchon, S., Bouillant, M., Salmond, G., and Nasser, W. (1998) Integration of the quorum-sensing system in the regulatory networks controlling virulence factor synthesis in Erwinia chrysanthemi. Molecular Microbiology 29: 1407-1418. Rezuchova, B., Miticka, H., Homerova, D., Roberts, M., and Kormanec, J. (2003) New members of the Escherichia coli sigma(E) regulon identified by a two-plasmid system. FEMS Microbiology Letters 225: 1-7. Ruby, E., and Nealson, K. (1976) Symbiotic association of photobacterium fischeri with marine luminous fish monocetris-japonica-model of symbiosis based on bacterial studies. Biological Bulletin 151: 574-586. Ryu, K., Kim, C., Kim, I., Yoo, S., Choi, B., and Park, C. (2004) NMR application probes a novel and ubiquitous family of enzymes that alter monosaccharide configuration. Journal of Biological Chemistry 279: 25544-25548. Schaefer, A., Val, D., Hanzelka, B., Cronan, J., and Greenberg, E. (1996) Generation of cell-to-cell signals in quorum sensing: Acyl homoserine lactone synthase activity of a purified Vibrio fescheri LuxI protein. Proceedings of the National Academy of Sciences of the Unites States of America 93: 9505-9509. Schaefer, A., Hanzelka, B., Parsek, M., and Greenberg, E. (2000) Detection, purification, and structural elucidation of the acylhomoserine lactone inducer of Vibrio fischeri 143 luminescence and other related molecules. Bioluminescence and Chemiluminescence,PT C 305: 288-301. Schauder, S., Shokat, K., Surette, M., and Bassler, B. (2001) The LuxS family of bacterial autoinducers: biosynthesis of a novel quorum-sensing signal molecule. Molecular Microbiology 41: 463-476. Schembri, M., Kjaergaard, K., and Klemm, P. (2003) Global gene expression in Escherichia coli biofilms. Molecular Microbiology 48: 253-267. Schuster, M., Lostroh, C., Ogi, T., and Greenberg, E. (2003) Identification, timing, and signal specificity of Pseudomonas aeruginosa quorum-controlled genes: a transcriptome analysis. Journal of Bacteriology 185: 2066-2079. Scott, J., Smirnova, N., and Haldenwang, W. (1999) A Bacillus-specific factor is needed to trigger the stress-activated phosphatase/kinase cascade of sigma(B) induction. Biochemical and Biophysical Research Communications 257: 106-110. Sirko, A., Zatyka, M., Sadowy, E., and Hulanicka, D. (1995) Sulfate and thiosulfate transport in Escherichia coli K-12: evidence for a functional overlapping of sulfate-binding and thiosulfate-binding proteins. Journal of Bacteriology 177: 4134-4136. Sitnikov, D., Shadel, G., and Baldwin, T. (1996) Autoinducer-independent mutants of the LuxR transcriptional activator exhibit differential effects on the two lux promoters of Vibrio fischeri. Molecular & General Genetics 252: 622-625. Sitnikov, D.M., Schineller, J.B., and Baldwin, T.O. (1996) Control of cell division in Escherichia coli: Regulation of transcription of ftsQA involves both RpoS and SdiA-mediated autoinduction. Proceedings of the National Academy of Sciences of the United States of America 93: 336-341. Smith, J., and Ahmer, B. (2003) Detection of other microbial species by Salmonella: Expression of the SdiA regulon. Journal of Bacteriology 185: 1357-1366. Sperandio, V., Mellies, J.L., Nguyen, W., Shin, S., and Kaper, J.B. (1999) Quorum sensing controls expression of the type III secretion gene transcription and protein secretion in enterohemorrhagic and enteropathogenic Escherichia coli. Proceedings of the National Academy of Sciences of the United States of America 96: 15196-15201. Sperandio, V., Torres, A., Giron, J., and Kaper, J. (2001) Quorum sensing is a global regulatory mechanism in enterohemorrhagic Escherichia coli O157: H7. Journal of Bacteriology 183: 5187-5197. Sperandio, V., Torres, A., Jarvis, B., Nataro, J., and Kaper, J. (2003) Bacteria-host communication: The language of hormones. PNAS 100: 8951-8956. Stancik, L., Stancik, D., Schmidt, B., Barnhart, D., Yoncheva, Y., and Slonczewski, J. (2002) pH-dependent expression of periplasmic proteins and amino acid catabolism in Escherichia coli. Journal of Bacteriology 184: 4246-4258. Stevens, A., Fujita, N., Ishihama, A., and Greenberg, E. (1999) Involvement of the RNA polymerase alpha-subunit C-terminal domain in LuxR-dependent activation of the Vibrio fischeri luminescence genes. Journal of Bacteriology 181: 4704-4707. Suga, H., and Smith, K. (2003) Molecular mechanisms of bacterial quorum sensing as a new drug target. Current Opinion in Chemical Biology 7: 586-591. Sun, J., Daniel, R., Wagner-Dobler, I., and Zeng, A. (2004) Is autoinducer-2 a universal signal for interspecies communication: a comparative genomic and phylogenetic 144 analysis of the synthesis and signal transduction pathways. BMC Evolutionary Biology 4: -. Surette, M., and Bassler, B. (1998) Quorum sensing in Escherichia coli and Salmonella typhimurium. PNAS 95: 7046-7050. Surette, M., and Bassler, B. (1999) Regulation of autoinducer production in Salmonella typhimurium. Molecular Microbiology 31: 585-595. Surette, M., Miller, M., and Bassler, B. (1999) Quorum sensing in Escherichia coli, Salmonella typhimurium, and Vibrio harveyi: A new family of genes responsible for autoinducer production. PNAS 96: 1639-1644. Suzuki, K., Wang, X., Weilbacher, T., Pernestig, A., Melefors, O., Georgellis, D., Babitzke, P., and Romeo, T. (2002) Regulatory circuitry of the CsrA/CsrB and BarA/UvrY systems of Escherichia coli. Journal of Bacteriology 184: 5130-5140. Swift, S., Downie, J., Whitehead, N., Barnard, A., Salmond, G., and Williams, P. (2001) Quorum sensing as a population-density-dependent determinant of bacterial physiology. Advances in Microbial Physiology, 45: 199-270. Taga, M., Semmelhack, J., and Bassler, B. (2001) The LuxS-dependent autoinducer Al-2 controls the expression of an ABC transporter that functions in Al-2 uptake in Salmonella typhimurium. Molecular Microbiologu 42: 777-793. Taga, M., Miller, S., and Bassler, B. (2003) Lsr-mediated transport and processing of Al- 2 in Salmonella typhimurium. Molecular Microbiology 50: 1411-1427. Taga, M.E., Semmelhack, J.L., and Bassler, B.L. (2001) The LuxS-dependent autoinducer Al-2 controls the expression of an ABC transporter that functions in Al-2 uptake in Salmonella typhimurium. Molecular Microbiology 42: 777-793. Tegner, J., Yeung, M., Hasty, J., and Collins, J. (2003) Reverse engineering gene networks: Integrating genetic perturbations with dynamical modeling. PNAS 100: 5944-5949. Tjaden, B., Saxena, R., Stolyar, S., Haynor, D., Kolker, E., and Rosenow, C. (2002) Transcriptome analysis of Escherichia coli using high-density oligonucleotide probe arrays. Nucleic Acids Research 30: 3732-3738. Touati, D. (2000) Iron and oxidative stress in bacteria. Archives of Biochemistry and Biophysics 373: 1-6. Urbanowski, A., Lostroh, C., and Greenberg, E. (2004) Reversible acyl-homoserine lactone binding to purified Vibrio fischeri LuxR protein. Journal of Bacteriology 186: 631-637. Van Delden, C., Comte, R., and Bally, M. (2001) Stringent response activates quorum sensing and modulates cell density-dependent gene expression in Pseudomonas aeruginosa. Journal of Bacteriology 183: 5376-5384. Vannini, A., Volpari, C., and Di Marco, S. (2004) Crystallization and preliminary X-ray diffraction studies of the quorum-sensing regulator TraM from Agrobacterium tumefaciens. Acta Crystallographica Section D-Biological Crystallography 60: 146-148. Vendeville, A., Winzer, K., Heurlier, K., Tang, C., and Hardie, K. (2005) Making 'sense' of metabolism: Autoinducer-2, LuxS and pathogenic bacteria. Nature Reviews Microbiology 3: 383-396. Visick, K. (2005) Layers of signaling in a Bacterium-Host association. Journal of Bacteriology 187: 3603-3606. 145 Volf, J., Sevcik, M., Havlickova, H., Sisak, F., Damborsky, J., and Rychlik, I. (2002) Role of SdiA in Salmonella enterica serovar Typhimurium physiology and virulence. Archives of Microbiology 178: 94-101. Wang, H., Chen, J., Hollister, K., Sowers, L., and Forman, B. (1999) Endogenous bile acids are ligands for the nuclear receptor FXR BAR. Molecular Cell 3: 543-553. Wang, L., Hashimoto, Y., Tsao, C., Valdes, J., and Bentley, W. (2005) Cyclic AMP (cAMP) and cAMP receptor protein influence both synthesis and uptake of extracellular autoinducer 2 in Escherichia coli. Journal of Bacteriology 187: 2066-2076. Wang, X.D., Deboer, P.A.J., and Rothfield, L.I. (1991) A Factor That Positively Regulates Cell-Division by Activating Transcription of the Major Cluster of Essential Cell-Division Genes of Escherichia coli. EMBO Journal 10: 3363-3372. Wassarman, K., Repoila, F., Rosenow, C., Storz, G., and Gottesman, S. (2001) Identification of novel small RNAs using comparative genomics and microarrays. Genes & Development 15: 1637-1651. Weber, H., Polen, T., Heuveling, J., Wendisch, V., and Hengge, R. (2005) Genome-wide analysis of the general stress response network in Escherichia coli: sigma(S)- dependent genes, promoters, and sigma factor selectivity. Journal of Bacteriology 187: 1591-1603. Wei, Y., Lee, J.M., Smulski, D.R., and LaRossa, R.A. (2001) Global impact of sdiA amplification revealed by comprehensive gene expression profiling of Escherichia coli. Journal of Bacteriology 183: 2265-2272. Wei, Y., Vollmer, A., and LaRossa, R. (2001) In vivo titration of mitomycin C action by four Escherichia coli genomic regions on multicopy plasmids. Journal of Bacteriology 183: 2259-2264. Welch, M., Todd, D., Whitehead, N., McGowan, S., Bycroft, B., and Salmond, G. (2000) N-acyl homoserine lactone binding to the CarR receptor determines quorum- sensing specificity in Erwinia. EMBO Journal 19: 631-641. Wen, Z., and Burne, R. (2004) LuxS-mediated signaling in Streptococcus mutans is involved in regulation of acid and oxidative stress tolerance and biofilm formation. Journal of Bacteriology 186: 2682-2691. White, M., Vasquez, J., Yang, S., and Kirsch, J. (1994) Expression of Apple 1- aminocyclopropane-1-carbosylates synthease in Escherichia coli: kinetic characterization of wild-type and active-stie mutant forms. PNAS 91: 12428- 12432. Whiteley, M., Parsek, M., and Greenberg, E. (2000) Regulation of quorum sensing by RpoS in Pseudomonas aeruginosa. Journal of Bacteriology 182: 4356-4360. Whiteley, M., and Greenberg, E. (2001) Promoter specificity elements in Pseudomonas aeruginosa quorum-sensing-controlled genes. Journal of Bacteriology 183: 5529- 5534. Wickner, S., Maurizi, M., and Gottesman, S. (1999) Posttranslational quality control: Folding, refolding, and degrading proteins. Science 286: 1888-1893. Winson, M., Swift, S., Fish, L., Throup, J., Jorgensen, F., Chhabra, S., Bycroft, B., Williams, P., and Stewart, G. (1998) Construction and analysis of luxCDABE- based plasmid sensors for investigating N-acyl homoserine lactone-mediated quorum sensing. FEMS Microbiology Letters 163: 185-192. 146 Xavier, K.B., and Bassler, B.L. (2005) Regulation of uptake and processing of the quorum-sensing autoinducer AI-2 in Escherichia coli. Journal of Bacteriology 187: 238-248. Yamamoto, K., Yata, K., Fujita, N., and Ishihama, A. (2001) Novel mode of transcription regulation by SdiA, an Escherichia coli homologue of the quorum-sensing regulator. Molecular Microbiology 41: 1187-1198. Yates, E., Philipp, B., Buckley, C., Atkinson, S., Chhabra, S., Sockett, R., Goldner, M., Dessaux, Y., Camara, M., Smith, H., and Williams, P. (2002) N-acylhomoserine Lactones undergo lactonolysis in a pH-, temperature-, and acyl chain length- dependent manner during growth of Yersinia pseudotuberculosis and Pseudomonas aeruginosa. Infection and Immunity 70: 5635-5646. Yoshimoto, T., Higashi, H., Kanatani, A., Lin, X., Nagai, H., Oyama, H., Kurazono, K., and Tsuru, D. (1991) Clonging ans sequencing of the 7-alpha-hydorxysteroid dehydrogenase gene from Escherichia coli HB101 and characterization of the expressed enzyme. Journal of bacteriology 173: 2173-2179. Zhang, L., Murphy, P., Kerr, A., and Tate, M. (1993) Agrobacterium conjugation and gene-regulaiton by N-acyl-L-homoserine lactones. Nature 362: 446-448. Zhao, G., Wan, W., Mansouri, S., Alfaro, J., Bassler, B., Cornell, K., and Zhou, Z. (2003) Chemical synthesis of S-ribosyl-L-homocysteine and activity assay as a LuxS substrate. Bioorganic & Medical Chemistry Letters 13: 3897-3900. Zhu, J., Beaber, J., More, M., Fuqua, C., Eberhard, A., and Winans, S. (1998) Analogs of the autoinducer 3-oxooctanoyl-homoserine lactone strongly inhibit activity of the TraR protein of Agrobacterium tumefaciens. Journal of Bacteriology 180: 5398- 5405. Zhu, J., and Winans, S. (1999) Autoinducer binding by the quorum-sensing regulator TraR increases affinity for target promoters in vitro and decreases TraR turnover rates in whole cells. PNAS 96: 4832-4837.