My goal is to understand quorum sensing, the process of cell-cell communication in bacteria. Until recently, the ability of bacteria to communicate was considered an anomaly that occurred only in a few marine Vibrio species. It is now clear that cell-cell communication is ubiquitous in the bacterial world and that understanding this process is fundamental to all of microbiology, including industrial and clinical microbiology, and ultimately to understanding the development of higher organisms. To pursue my goal of understanding bacterial communication, I have combined genetics, biochemistry, structural biology, chemistry, microarray studies, bioinformatics, and modeling.
Quorum sensing, which involves the production, release, and subsequent detection of chemical signaling molecules called autoinducers, allows bacteria to regulate gene expression in response to changes in cell-population density. As a population of bacteria grows, the extracellular concentration of autoinducer increases. When a threshold is reached, the group responds with a population-wide alteration in gene expression. Processes controlled by quorum sensing are usually ones that are unproductive when undertaken by an individual bacterium but become effective when undertaken by the group. For example, quorum sensing controls bioluminescence, secretion of virulence factors, biofilm formation, sporulation, and the exchange of DNA. Thus, quorum sensing is a mechanism that allows bacteria to function as multicellular organisms. Our recent studies show that bacteria make, detect, and integrate information from multiple autoinducers, some of which exclusively facilitate intraspecies communication, while others enable communication between species.
Our study of quorum sensing is providing insight into intra- and interspecies communication, population-level cooperation, and the design principles underlying signal transduction and information processing at the cellular level. These investigations are leading to synthetic strategies for controlling quorum sensing. Objectives include development of antimicrobial drugs aimed at bacteria that use quorum sensing to control virulence, and improved industrial production of natural products such as antibiotics.
When I began studying quorum sensing, only two bacteria were known to make autoinducers: Vibrio fischeri and Vibrio harveyi. Both are bioluminescent marine bacteria; V. fischeri is a symbiont, whereas V. harveyi is free-living. The V. fischeri communication circuit had been shown to consist of the LuxI protein, which synthesizes an acyl-homoserine lactone autoinducer, and the LuxR protein, responsible for autoinducer binding and subsequent activation of transcription of the luciferase operon (LuxI/R homologs have since been found in more than 70 species of Gram-negative bacteria). At the time, I reasoned that, because V. harveyi is free-living, it must be able to adapt to a complex and changing environment. Therefore, its cell-cell communication circuit might be more sophisticated than that of V. fischeri. For this reason, I chose to examine V. harveyi. Indeed, we found that the V. harveyi quorum-sensing components are not similar to LuxI and LuxR.
Our genetic analyses revealed that V. harveyi possesses two quorum-sensing systems, and signal transduction occurs via a phosphorylation/dephosphorylation cascade (Figure 1). System 1 is composed of LuxN, which binds autoinducer-1 (AI-1; N-3-hydroxybutanoyl-L-homoserine lactone), and system 2 is composed of LuxPQ, which binds autoinducer-2 (AI-2; a furanosyl borate diester). Sensory signals from both LuxN and LuxPQ are channeled to an integrator protein called LuxU, and from LuxU to LuxO. LuxO indirectly represses the expression of many targets, including the luciferase operon luxCDABE.
We initially focused on LuxN, a nine-transmembrane domain protein from Vibrio harveyi, and the founding example of membrane-bound receptors for acyl-homoserine lactone (AHL) autoinducers. We used mutagenesis and suppressor analyses to identify the AHL-binding domain of LuxN and discovered LuxN mutants that confer decreased and increased AHL sensitivity. Our analysis of dose-response curves of multiple LuxN mutants pins these inverse phenotypes on quantifiable opposing shifts in the free-energy bias of LuxN for its kinase and phosphatase states. To extract signaling parameters, we exploited a strong LuxN antagonist, 1 of 15 small-molecule antagonists we identified. We found that quorum-sensing-mediated communication can be manipulated positively and negatively to control bacterial behavior and that signaling parameters can be deduced from in vivo data.
We extended our studies to a nonluminescent relative of V. harveyi, Vibrio cholerae, a major human pathogen and the cause of the endemic disease cholera. Using a classical genetics approach, we identified and characterized the genes encoding two autoinducer-sensor systems as well as a growth-phase-regulated sensory system. We used microarray analysis combined with in vitro and in vivo virulence assays to show that this complex communication system controls the entire V. cholerae virulence regulon (>70 genes), and consistent with this, we found that quorum-sensing-defective mutants are completely avirulent.
Our examination of autoinducer detection and response patterns in different bacteria showed that hundreds of species of bacteria produce AI-2, while only V. harveyi produces AI-1, leading us to hypothesize that V. harveyi uses AI-1 for intraspecies communication and AI-2 for interspecies communication. We cloned the genes responsible for AI-2 production from V. harveyi, Escherichia coli, and Salmonella typhimurium and showed that in each case it was the same gene, which we named luxS. Conserved luxS homologs exist in half of all sequenced Gram-negative and Gram-positive bacteria. We proposed that these genes define a new family of proteins involved in autoinducer production. To support this, we demonstrated that all bacterial species possessing luxS produce AI-2. Moreover, luxS mutants have been constructed in nearly every known luxS-containing bacterium; in each case, disruption of luxS eliminated AI-2 production. To date, AI-2 has been shown to control gene expression in dozens of bacterial species. For example, it controls growth in Bacillus anthracis and virulence in V. cholerae, E. coli O157, and many other clinically important pathogens.
We determined the biosynthetic pathway for AI-2 and showed that LuxS is the AI-2 synthase (Figure 2). LuxS hydrolyzes S-ribosylhomocysteine (SRH) to homocysteine and 4,5-dihydroxy-2,3-pentanedione (DPD). DPD undergoes spontaneous rearrangements, making identification of AI-2 difficult. With crystallographer Frederick Hughson (Princeton University), we crystallized the sensor protein LuxP in complex with the AI-2 ligand and solved their structures. DPD cyclizes and is subsequently converted to the final V. harveyi AI-2–signaling molecule by the addition of borate. We also used crystallography to demonstrate that the S. typhimurium AI-2–binding protein LsrB binds a chemically distinct form of AI-2 lacking boron (Figure 2). The V. harveyi and S. typhimurium AI-2 molecules interconvert upon release from their respective binding proteins, revealing new sophistication in the chemical lexicon used for bacterial interspecies signaling. These structural studies facilitated our development of enzymatic and chemical synthesis procedures to make the AI-2s (with chemist Martin Semmelhack, Princeton University).
For in-depth examination of AI-2 signaling in V. harveyi, we combined x-ray crystallographic and functional studies to show that AI-2 binding causes a major conformational change within LuxP (see Figure 1), which in turn stabilizes a quaternary arrangement in which two LuxPQ monomers are asymmetrically associated. We proposed that formation of this asymmetric quaternary structure, rather than conformational changes within single LuxQ subunits, is responsible for repressing the kinase activity of both LuxQ subunits and triggering the transition of V.harveyi into quorum-sensing mode. We also examined signal processing and integration in the V. harveyi quorum-sensing circuit with theoretical physicist Ned Wingreen (Princeton University). We observed that the V. harveyi quorum-sensing system can discriminate between no autoinducer, AI-1 only, AI-2 only, and AI-1 + AI-2, demonstrating that the binary information encoded in the presence or absence of one or both autoinducers is preserved by V. harveyi.
A genetic screen for additional components of the V. harveyi and V. cholerae quorum-sensing circuits revealed the protein Hfq (Figure 1). Hfq mediates interactions between small, regulatory RNAs (sRNAs) and specific mRNA targets that alter the stability of the target transcripts. We showed that Hfq mediates the destabilization of the mRNA encoding the quorum-sensing master regulators LuxR (V. harveyi) and HapR (V. cholerae), implicating an sRNA in the circuit. Bioinformatics identified four candidate sRNAs. Surprisingly, any one of the sRNAs proved sufficient for full control of quorum sensing. It is interesting that sRNAs are important in quorum sensing because sRNAs are involved in other quasi-developmental bacterial processes (e.g., entry into stationary phase, iron starvation, and stress response). Small (micro) RNA molecules (miRNAs) are also important in eukaryotic gene regulation and development.
Using fluorescence-activated cell sorting, we analyze the molecular mechanism underlying the sRNA-mediated quorum-sensing transition both in cell populations and also in individual bacterial cells. We had shown that quorum-sensing target genes exhibit differential responses to the individual autoinducers. This new study demonstrated that a differential response to each autoinducer input state also occurs at the level of the sRNAs, luxR gene expression, and LuxR protein production (see Figure 1). Individual cell analyses revealed that, in each case, all the bacteria in the population respond in unison to the various autoinducer inputs. We propose that the V. harveyi quorum-sensing transition is not switch-like but rather operates in a graded manner, and that this signaling arrangement, which employs shared regulatory proteins, nonetheless provides V. harveyi a mechanism to uniquely respond to different autoinducer input states.
In new work, we showed that the combined action of two feedback loops, one involving the sRNA-activator LuxO and one involving the sRNA-target HapR (V. cholerae) and LuxR (V. harveyi), promotes gene dosage compensation between the four sRNA genes. Gene dosage compensation adjusts the total sRNA pool and provides the molecular mechanism underlying sRNA redundancy. The dosage-compensation mechanism is exquisitely sensitive to small perturbations in sRNA levels. Precisely maintained sRNA levels are required to direct the proper timing and correct patterns of expression of quorum-sensing regulated target genes.