Integration of Cell-to-Cell Signals in Soil Bacterial Communities

Abstract

Soil bacteria employ multitiered signaling mechanisms to structure multicellular communities, coordinate behaviors within these communities, and to interact with their eukaryotic hosts. Bacteria deploy signals to estimate numbers of bacterial cells within diffusion-limited environments and then modulate gene expression in response to changes in population densities. This behavior is known as “quorum sensing” (QS). Many Gram-negative bacteria use N-acyl homoserine lactones (AHLs) as QS signals. In γ-proteobacteria, AHL-mediated signaling itself is regulated by the orthologs of the GacS/GacA two-component global regulatory system. In response to an unknown signal, GacS/GacA orthologs control genes involved in quorum sensing, virulence, stress survival, motility, and the production of secondary metabolites and exoenzymes. GacS is an unorthodox sensor kinase predicted to autophosphorylate in the presence of a currently unknown endogenous signal. In Pseudomonas spp., GacS interacts with the orphan sensor kinases RetS and LadS, which modulate activity of GacS.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

In complex and uncertain environments, such as soils, unicellular organisms depend on various cues to coordinately regulate genes and behaviors important for survival of their populations. How microbes perceive environmental and metabolic signals and how they integrate this information to modulate gene expression is a central question in microbial ecology. Unlike obligate parasites and symbionts of eukaryotes, genomes of which have undergone considerable reductions, soil bacteria, as well as opportunistic pathogens contain a large number of genes involved in environmental sensing. For example, of the approximately half of the Pseudomonas aeruginosa genes (for which functions could be assigned at least tentatively), 2% are two-component regulatory systems and 7.2% are transcriptional regulators (Stover et al. 2000). Regulatory proteins make up 8.7% of the genome of S. meliloti 1021 (including 36 pairs of sensor kinases and response regulators) (Galibert et al. 2001). These genome-wide comparisons indicate that the ability to accurately sense the environment and to change gene expression accordingly is crucial to the bacteria that spend at least a part of their life cycle in soils.

Quorum sensing (QS) is one of the mechanisms by which microbes change global patterns of gene expression in response to increases in their population densities within a diffusion-limited environment [rev. Waters and Bassler (2005) and Hense et al. (2007)]. At least a dozen kinds of small molecules and short peptides have been designated as signals for QS-like regulatory cascades in proteobacteria, firmicutes, and yeast. Even though the ability to sense the presence of other bacteria (including siblings) appears to be evolutionarily conserved in different clades of unicellular organisms, the signals used in QS and mechanisms by which they are perceived are distinct [rev. Waters and Bassler (2005), Diggle et al. (2007a), and Dobretsov et al. (2009)]. Since the original discoveries of QS in Streptomyces and Vibrio fischeri (Khokhlov et al. 1967; Eberhard et al. 1981), QS research has exploded with over 2,400 peer-reviewed publications on the subject cataloged in Pubmed (as of October 2009). Therefore, instead of surveying behaviors controlled by QS in various soil bacteria, this chapter will focus on the mechanisms by which various regulatory inputs (including QS signals) are integrated in well-characterized soil and plant-associated bacteria.

Our goals for this Chapter are to highlight signal integration mechanisms that are common to many proteobacteria and then to identify regulatory mechanisms that are unique to specific systems. Due to space limitations, this review will not be comprehensive. Rather, we will briefly survey what is known about QS in soil microbial communities, then focus on QS in two model bacteria (Sinorhizobium meliloti and Pseudomonas aeruginosa) and then review the role of orthologs of GacS/GacA in the integration of environmental inputs and QS in γ-proteobacteria.

2 Examples of QS in Soil Bacterial Communities

Proteobacteria rely on several types of QS systems to regulate the expression of genes important for structuring of microbial communities, for their interactions with other organisms and eukaryotic hosts, and for production of secondary metabolites. The best-characterized QS signals in Gram-negative bacteria are N-acyl-L-homoserine lactones (AHLs). Soil and plant-associated rhizobia, pseudomonads, serratia, chromobacteria, pectobacteria, pantoeae are among those that produce AHL QS signals (Elasri et al. 2001; Hirsch et al. 2003; Von Bodman et al. 2003; Sanchez-Contreras et al. 2007; Van Houdt et al. 2007). AHLs produced by different bacterial species differ in the length of their side chains, degree of saturation, and the decorations at the third carbon of the side chain.

Synthesis of AHLs from acyl-acyl carrier proteins (acyl-ACP) and S-adenosyl methionine (SAM) is typically catalyzed by the homologs of LuxI (Fuqua and Eberhard 1999). Once produced, AHLs with shorter acyl side chains diffuse freely across cell membranes. However, molecules with longer acyl side chains (e.g., 3-oxo-C₁₂-HSL of P. aeruginosa) can be actively transported (Pearson et al. 1999). Inside bacterial cells, AHLs are typically detected by the homologs of LuxR proteins (Zhang et al. 2002; Koch et al. 2005). Homologs of LuxR have an AHL-binding N-terminal domain and a DNA-binding C-terminal binding domain. Binding of an AHL to the cognate receptor protein typically leads to the oligomerization of the protein and allows the LuxR-AHL complex to bind its target DNA sequences, “lux boxes” (Stevens et al. 1994; Zhang et al. 2002; Bartels et al. 2007). This is the most common mechanism of AHL signal perception in bacteria, although some variations on this common theme have been reported.

Mechanisms of AHL synthesis and perception differ in some bacteria. For example, some species of Vibrio can synthesize AHLs via AinS and LuxM enzymes (Gilson et al. 1995; Ng and Bassler 2009). In V. harveyi and V. cholerae, AHLs are perceived by a hybrid histidine kinase LuxN, which upon perception of an AHL signal contributes to phosphorylation of a phosphorelay protein LuxU, and this eventually sets off the QS regulatory cascade in V. harveyi and V. cholerae (Ng and Bassler 2009). The AHL receptor, ExpR in Pantoeae stewartii functions as a repressor, which upon binding of a cognate AHL is released from DNA sequences within regulated promoters and this allows transcription of the downstream genes (Von Bodman et al. 2003). Salmonella, Klebsiella and E. coli do not synthesize their own AHL signals, but have a functional AHL receptor, SdiA (Ahmer 2004; Kim and Surette 2006; Smith et al. 2008). Even though the function of SdiA during interactions of Salmonella with its animal hosts is not yet completely understood, it is clear that the ability to perceive AHLs is dispensable during plant-associated growth (Noel et al. 2010). Collectively, these results indicate that even though the mechanisms of AHL synthesis and perception differ in some bacteria, AHL-mediated QS is widespread in Gram-negative bacteria.

Not all Gram-negative bacteria have AHL-mediated QS, and some use other signals as well as AHLs for QS regulation. Even members of the same family or closely related bacteria may completely differ in regard to their QS machinery (Cha et al. 1998; Elasri et al. 2001; Charkowski 2009). Interestingly, the ability to synthesize AHLs is more common in plant-associated bacteria recovered from the rhizosphere than among bacteria isolated from bulk soil (Elasri et al. 2001). Such heterogeneity in the distribution of QS genes within the genome in closely related bacteria could be explained – at least in part – by the observation that AHL synthases and AHL receptors are often carried on self-transmissible plasmids or integrative and conjugative elements (Marketon and Gonzalez 2002; Cho et al. 2009; Ramsay et al. 2009), or are sometimes found flanked by phage or transposon sequences (Williamson et al. 2005; Wei et al. 2006; Hao et al. 2010), suggestive of horizontal transfer. Curiously, genomes of many bacteria (including S. meliloti and P. aeruginosa, discussed below in detail) contain “orphan” AHL receptors, which are not encoded next to an AHL synthase (Stover et al. 2000; Pellock et al. 2002; Bartels et al. 2007). Unlike synthase-receptor QS gene clusters that are often carried on plasmids or contained within variable regions of genomes of soil bacteria, “orphan” AHL receptors are typically orthologous in closely related bacteria. These observations raise interesting questions regarding the evolutionary origin and ecological functions of AHL-mediated QS (Lerat and Moran 2004; Diggle et al. 2007a; Hense et al. 2007), questions for which no satisfactory explanations currently exist.

Screens of metagenomic libraries using AHL reporter bioassays have identified AHL synthases encoded next to luxR AHL receptors (Williamson et al. 2005; Hao et al. 2010). In addition to a canonical luxI AHL synthase, Williamson et al. (2005) also identified a clone homologous to either 2-isopropylmalate synthase or a homocitrate synthase, encoded next to a transcriptional activator. Overexpression of this synthase gene in E. coli cells carrying an AHL-responsive gfp reporter triggered activation of the AHL GFP reporter, indicating that it likely encoded a novel non-luxI AHL synthase or a synthase of an AHL-mimic compound (Williamson et al. 2005). A functional screen for QS inhibitors also identified a clone encoding a cytochrome-like molecule, capable of inhibiting responses of an AHL reporter, although the mechanism of this inhibition is currently unknown (Williamson et al. 2005).

The role of QS regulation in natural environments, where there are no sharp limits to diffusion of QS signals, is much less well characterized. AHL synthase and receptor mutants generally have reduced virulence in both plant and animal pathogens, indicating that such regulation has important functions in the natural environments of these hosts. Likewise there is evidence that AHL-mediated signaling takes places in situ in natural aquatic biofilms and soil microcosms (Gantner et al. 2006; Gao and Teplitski 2008). In the root zone (rhizosphere) of tomato and wheat, Pseudomonas putida produced AHLs (as measured by activation of an AHL-responsive GFP reporter). AHLs produced by P. putida formed a quorum gradient with highest responses within 5μm but extending as far as 40–80μm over the root surface (Gantner et al. 2006). Recombinant-based in vivo expression technology (RIVET) demonstrated that the AHL synthase gene sinI in S. meliloti was expressed during the colonization of rhizosphere of M. truncatula (Gao and Teplitski 2008). Expression of sinI was variable at early stages of the rhizosphere colonization, but stabilized once bacteria reached the stage where they are predicted to form microcolonies on the root surfaces (Gao and Teplitski 2008). Collectively, these observations indicate that AHL-mediated QS is common in such natural environments.

2.1 The Role of Sinorhizobium meliloti Quorum Sensing in the Interaction with Its Plant Hosts

2.1.1 Signal Exchange Leading to the Establishment of the S. meliloti–Medicago Symbiosis

Most commonly, S. meliloti is found in the symbiosis with alfalfa and related legumes from genera Medicago and Trigonella (Jones et al. 2007). It is also a common soil bacterium, capable of living saprophytically outside of plant hosts. The establishment and maintenance of the symbiosis between S. meliloti and its legume hosts is complex and requires precisely timed signal exchange between the partners (Gage 2004; Mitra et al. 2004; Jones et al. 2007). The symbiosis begins with initial contacts via extracellular signals produced by each partner (specific flavonoids from the plant hosts and chitin lipo-oligosaccharide Nod factors from the bacterium) (Mulligan and Long 1985; Peters et al. 1986; Caetano-Anolles et al. 1992). This is followed by the initiation of new root nodule meristems and the formation of infection threads (Long et al. 1982; Jones et al. 2007). Infection threads are tubular cell wall invaginations of host root hairs that carry multiplying Sinorhizobium cells into the differentiating nodule tissue. Bacterial exopolysaccharides (EPS) and host lectins play important signaling roles in the infection process (Bauer 1977; Gage 2004; Mitra et al. 2004; Skorupska et al. 2006). Infecting bacteria are released from the threads into envelopes of host plasma membrane, where Sinorhizobium differentiates into “bacteroids” that multiply to fill most of the infected host cell. Bacteroid formation and the differentiation of tissues within the nodule generate a zone of symbiotic nitrogen fixation (Timmers et al. 2000; Barnett and Fisher 2006; Capela et al. 2006). Sinorhizobium provides the host with fixed nitrogen in the form of NH₄, in exchange for the photosynthetically fixed C, mostly in the form of C4-dicarboxylic acids (Poole and Allaway 2000; Djordjevic 2004). Many of the N-fixing host and bacterial cells subsequently senesce, with renewed multiplication of bacteria in older nodules (Timmers et al. 2000). The establishment and maintenance of this symbiosis provides many opportunities for QS regulation and for host manipulation of QS.

QS regulation is common in rhizobia (Gonzalez and Marketon 2003; Sanchez-Contreras et al. 2007). Mutants in various QS-regulated genes have proven to be defective in nodule initiation (Gao et al. 2005). In addition, cell division, symbiotic plasmid transfer, gene expression in the rhizosphere, symbiosome development and N-fixation, and nodule number are QS-regulated in Rhizobium species [rev. Gonzalez and Marketon (2003) and Sanchez-Contreras et al. (2007)]. In general, QS mutants of rhizobia can still infect and form nodules on their legume hosts. This suggests that the primary role of QS in rhizobia is to optimize many aspects of host interaction rather than determine any single essential step.

2.1.2 QS Signal Generation During S. meliloti–Medicago Interactions

QS in Sinorhizobium is complex and consists of multiple AHL signals and receptors. In laboratory shake cultures, the genomically sequenced strain 1021 and its derivative 8530 produce a diversity of AHLs with long acyl chains: C₁₂-HSL, C₁₄-HSL, 3-oxo-C₁₄-HSL, C₁₆-HSL, C_16:1-HSL, 3-oxo-C_16:1-HSL, and C₁₈-HSL (Marketon et al. 2002; Teplitski et al. 2003a; Gao et al. 2005). S. meliloti strains also produce QS-active substances that are chemically distinct from AHLs in that they are more hydrophyllic than known AHLs and are resistant to hydrolysis by AiiA, an AHL lactonase (Gao et al. 2007). In strains 1021 and 8530, all AHLs are produced in vitro by the sole AHL synthase, SinI, which is encoded on the chromosome (Gao et al. 2005), although the presence of a second cryptic AHL synthase has not been completely ruled out (Marketon et al. 2002). In addition to sinI, S. meliloti AK631 also contains another AHL synthase encoded on a plasmid (Marketon and Gonzalez 2002), which directs synthesis of AHLs with short side chains (Marketon and Gonzalez 2002; Teplitski et al. 2003a).

To find out whether QS signal synthesis is activated in S. meliloti during symbiosis with M. truncatula, regulation of merodiploid reporters (sinI+ sinI::gus and sinI+ sinI::tnpR-gus) was tested during different stages of the symbiosis (Gao and Teplitski 2008). When the expression of sinI was quantified using the RIVET reporter, it was determined that the expression of sinI was variable during the early stages of rhizosphere colonization, then reached a plateau within a week (Gao and Teplitski 2008), a time period sufficient to form steady-state population densities in the rhizosphere (Caetano-Anolles et al. 1992). The sinI promoter was induced strongly in older parts of 3-week-old nodules, indicating that SinI AHLs are actively synthesized during symbiosis, but not at detectable levels in the tip region where the nodule meristem is located (Gao and Teplitski 2008). Thus, the synthesis of AHL QS signals may be limited in younger nodule tissues and in the rhizosphere near the growing tip where infections take place. This could be biologically relevant. The observation that sinI is expressed in M. truncatula nodules is consistent with the earlier report that QS signals are produced by R. etli bacteroids in bean nodules (Daniels et al. 2002). Transcriptomic analyses indicate that sinI was expressed in the bacteroids formed by the expR- strain S. meliloti 1021 (Barnett et al. 2004). Interestingly, despite the fact that the sinI transcriptional fusions were strongly expressed in the nodules formed by the wild type S. meliloti (Gao and Teplitski 2008), qualitative RT-PCR assays indicate that the accumulation of the sinI mRNA was downregulated inside the nodules (Gurich and Gonzalez 2009). Technical limitations of the two assays aside, collectively these results could suggest posttranscriptional regulation of the sinI mRNA stability in expR-dependent manner. Evidence that levels of sinI mRNA (as indicated by both microarray and quantitative RT-PCR analyses) decrease in the hfq mutant in vitro (M. Gao and M. Barnett, manuscript in preparation) supports the hypothesis that posttranscriptional regulation of sinI by RNA binding proteins like Hfq may affect QS during normal symbiotic interactions between the bacterium and its plant hosts, adding another level of regulatory complexity to QS regulation in the host.

2.1.3 Responses of S. meliloti to AHL QS Signals

InS. meliloti strain 8530, there are two known AHL receptors (SinR and ExpR) and perhaps several additional, but untested AHL receptors capable of recognizing AHLs (Pellock et al. 2002; Bartels et al. 2007). The SinR receptor modulates expression of ~30 genes, including sinI, the AHL synthase (Marketon et al. 2002; Hoang et al. 2004; Gao et al. 2005). The ExpR receptor is a functional“orphan” AHL receptor (Bartels et al. 2007). Orthologs of expR are present in the genomes of S. medicae, Rhizobium leguminosarum, and Agrobacterium tumefaciens. In S. meliloti 1021, expR is interrupted by a native insertion sequence, but this sequence excises spontaneously to provide the functional AHL receptor in the strain 8530 (Pellock et al. 2002). ExpR regulates surface swarming and activates synthesis of a symbiotically important EPSII (Becker et al. 1997; Pellock et al. 2002; Gao et al. 2005). Transcriptome studies indicate that ExpR is involved in regulating the expression of about 140 genes under laboratory shake culture conditions (Hoang et al. 2004; Gurich and Gonzalez 2009). Parallel proteome studies identified additional 50 proteins subject to ExpR-dependent QS regulation (Gao et al. 2005). In addition to the ~200 genes/proteins that are QS-regulated in laboratory cultures via the ExpR receptor, another 200 proteins have been identified as QS-regulated by AHL receptors other than ExpR (Chen et al. 2003; Hoang et al. 2004; Teplitski et al. 2004; Gao et al. 2005).

One of the most surprising results from the global studies of QS in S. meliloti is the lack of overlap between the sets of QS-regulated genes/proteins. Only about 5% of the genes/proteins identified as QS-regulated were seen in more than one study. Different culture methods, different receptor backgrounds, and different means of distinguishing QS-regulated genes/proteins probably account much of this lack of overlap. But the lack of overlap between these several studies teaches us two important things: first, that the we are far from identifying the full range of QS-regulated functions in Sinorhizobium and the contribution of posttranscriptional regulatory systems (like those mediated by an RNA binding protein Hfq); and second that further analysis of QS in laboratory cultures would not be very useful in helping us understand the roles of QS in the real biology of the bacterium.

2.1.4 Plant Hosts Detect Rhizobial AHLs and Manipulate Bacterial Signaling

Since bacterial pathogens and symbionts rely on QS regulation to optimize their infection of host organisms, it is not surprising that eukaryotic hosts have evolved the means to take advantage of this dependence. Eukaryotes have different mechanisms to deal with QS in the bacteria they encounter (Teplitski et al. 2010). Both plants and animals can inactivate AHLs (please see a chapter by Dessaux et al. in this book). Plant hosts, including M. truncatula, can detect bacterial AHLs and activate potent defense responses (Mathesius et al. 2003; Schuhegger et al. 2006). These extensive and specific responses of the host to the AHL signals from bacteria are likely an integral part of the QS-related interplay between Sinorhizobium and M. truncatula.

Plants, algae, and fungi also produce a variety of compounds that disrupt or manipulate bacterial QS circuitry and affect QS-regulated behaviors in situ. In pioneering studies, the marine red seaweed Delisea pulchra was found to produce a set of 20–30 different halogenated furanones. These algal furanones inhibit QS in Gram-negative bacteria (Givskov et al. 1996). Seedlings of many plants, including pea, rice, tomato, soybean, and M. truncatula, secrete compounds that stimulate or inhibit AHL receptor-dependent responses in QS reporter bacteria (Teplitski et al. 2000; Gao et al. 2003, 2007; Mathesius et al. 2003; Karamanoli and Lindow 2006). The plant compounds are therefore able to mimic bacterial AHL QS signals, and they appear to be secreted at levels that would affect QS regulation in associated bacteria during natural encounters (Teplitski et al. 2000), although not all bacterial AHL receptors are susceptible to manipulation by plant AHL signal-mimics in situ (Gantner et al. 2006). Pea and M. truncatula were found to produce 10–20 chromatographically separable AHL mimic compounds. Most of these AHL mimic compounds were preferentially extracted into methanol rather than ethyl acetate, indicating that they are chemically different from bacterial AHLs (Teplitski et al. 2000; Gao et al. 2003).

In addition to AHL-mimics, L-canavanine, a compound commonly found in exudates of alfalfa, was shown to inhibit QS-mediated behaviors in reporter strains and ExpR-responsive genes in S. meliloti (Keshavan et al. 2005). L-canavanine is an analog of arginine, and a strong inhibitor of bacterial arginine deaminase (Lu et al. 2005; Li et al. 2008). It is likely that down-regulation of QS responses may result from a general toxicity or could be an indirect consequence of the production of O-ureido-L-homoserine, a known intermediary produced from L-canavanine via a reactive thiouronium intermediate (Li et al. 2008). Because the addition of L-arginine at least partially reversed QS inhibitory effects of L-canavanine (Keshavan et al. 2005), the indirect effect of L-canavanine on metabolism and/or protein synthesis (Keshavan et al. 2005) is one explanation of its biological activity in QS assays.

The discoveries that rhizobia produce AHLs during normal interactions with the legume hosts, and that the plant hosts detect bacterial AHLs, respond to them and also produce compounds capable of modulating bacterial QS behaviors indicate that the bi-directional exchange of AHL signals may function as another regulatory layer in the establishment of this tightly coevolved symbiosis. In contrast to the red algae, which only produce inhibitory halogenated furanones, the ability of legumes to produce stimulatory AHL mimics, AHL-degrading enzymes, and compounds that directly or indirectly inhibit QS suggests that these plants may use their battery of QS-related signals and proteins to manipulate QS in the bacteria they encounter rather than just broadly suppressing QS.

2.1.5 S. meliloti Responds to Non-AHL QS Signals from Other Microbes

In soils, sinorhizobia are exposed to signals from other bacteria and eukaryotes. Recent evidence suggests that even though S. meliloti does not produce some of the known chemical classes of QS signals, it recognizes and responds to non-AHL QS signals from other soil microorganisms. For example, in addition to AHL QS signals, most Gram-negative bacteria and firmicutes produce furanone signals collectively known as autoinducer-2 (AI-2) [rev. Waters and Bassler (2005)]. Because chemical structures of AI-2 produced by different bacteria are closely related and because these signals appear to be wide spread in different clades of bacteria, their function in interspecies bacterial communication has been intensively studied [e.g., rev. Waters and Bassler (2005)].

Even though S. meliloti does not produce AI-2 signals, nor do published genomes of sinorhizobia contain homologues of the AI-2 synthase luxS, it detects and responds to AI-2 signals from other bacteria (Pereira et al. 2008). The symbiotic megaplasmid B (pSymB) of S. meliloti 1021 contains an ait operon (a uto i nducer-2 transporter, Smb21023-Smb21016), which is organizationally similar and functionally homologous to the AI-2 transporter operon lsr from Salmonella enterica (Pereira et al. 2008). The protein encoded by Smb21016 binds to a synthetic AI-2 ((2R, 4S)-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran), as demonstrated by crystallography and by the aitK- and aitA-dependent removal of exogenously supplied AI-2 to the growth medium with the associated changes in the transcription of the wild-type ait operon (Pereira et al. 2008). Consistent with the lack of AI-2 synthesis by the wild-type S. meliloti, coinoculation of the wild type and AI-2-insensitive strain onto Medicago seedlings demonstrated no effect of the ait mutations on seedling colonization (Pereira et al. 2008). In its ability to detect and respond to AI-2 from other bacteria, S. meliloti is similar to P. aeruginosa, which also does not produce but responds to AI-2 signals from other bacteria (Duan et al. 2003). Now that it is clear that S. meliloti responds to the synthetic and bacterially produced AI-2 signals (Pereira et al. 2008), it will be interesting to delineate the role of this behavior during the interactions of S. meliloti with AI-2-producing soil microbiota.

Within soil microbial communities, S. meliloti likely encounters QS-related signals produced by eukaryotic microbes. Proteomic and gene expression studies using synthetic signal analogs and HPLC-purified fractions containing eukaryotic signals, preliminary screens using 10mM farnesol, a compound known to effect QS-like responses in dimorphic yeast (Hornby et al. 2001), identified promoters for ndvA and bhbA as those up-regulated by farnesol in the sinI-strain of S. meliloti (M. Gao and K. Jogelekar, unpublished). In another study, treatment of S. meliloti 1021 cultures with an HPLC-purified fraction containing an active compound produced by the soil alga Chlamydomonas reinhardtii affected accumulation of 34 polypeptides. This fraction contained an activity capable of stimulating the LasR AHL receptor (Teplitski et al. 2004). Many of the polypeptides accumulated in response to an activity from C. reinhardtii were also accumulated in response to AHL signals produced by the wild-type cultures of S. meliloti, indicating that the purified AHL mimic compound produced by C. reinhardtii affects QS in the wild-type soil bacteria (Teplitski et al. 2004). The chemical structure of this algal signal is currently unknown (Teplitski et al. 2004). However, similar studies identified lumichrome as a stimulatory signal produced by C. reinhardtii (Rajamani et al. 2008). The role of lumichrome in the interactions between soil algae and bacteria is discussed in the accompanying Chap. 16

2.2 QS in Pseudomonas aeruginosa, a Model Environmental Bacterium and Opportunistic Pathogen

P. aeruginosa is a model environmental bacterium and an important opportunistic pathogen of animals, humans, and plants. The adaptation of this microorganism to such variety of environmental niches is reflected in the large size of its genome and by the large proportion of genes devoted to transcriptional regulation and signal transduction (Stover et al. 2000). These include functions encoding complex QS multisignaling systems. The current understanding is that in this microorganism QS is mediated by AHLs synthesized by LasI (a 3-oxo-C₁₂-HSL synthase) and RhlI (a C₄-HSL synthase) and by 2-alkyl-4-quinolone (AQ) signal molecules like the Pseudomonas quinolone signal, PQS (2-hepty-2-hydroxy-4-quinolone), and HHQ (2-heptyl-4-quinolone) (Xiao et al. 2006a, b; Williams and Camara 2009). This last class of signals includes more than 50 AQS produced by P. aeruginosa upon expression of the pqsABCDE operon. Figure14.1 summarizes the complex epistatic interactions between these QS systems. AHL signaling depends on the lasRI and rhlRI genes, where LasR and RhlR are LuxR-type transcriptional regulators binding 3-oxo-C₁₂-HSL and C₄-HSL, respectively. The lasIR and rhlIR systems are hierarchically arranged, and lasI is autoregulated by LasR, and LasR activates rhlRI, which is in turn autoregulated. A negative autoregulatory loop is also triggered by LasR binding of the RsaL repressor (Rampioni et al. 2007).

Fig.14.1

Signal integration in pseudomonads. Under noninducing conditions, the GacS–GacA regulatory cascade is blocked by binding of RetS to GacS. Under these conditions, rsm sRNAs are not produced. Under inducing conditions, GacS autophosphorylates and then transphosphorylates GacA (Asp and His residues involved in phosphotransfer are indicated as D and H, respectively; ~P indicates phosphate group). Phosphorylated GacA induces transcription of rsm sRNAs, which titrate RsmA. Under these conditions, quorum sensing genes are induced and production of biofilm is favored. Las QS system is mediated by 3-oxo-C₁₂-HSL, Rhl QS system is mediated by C₄-HSL and the PQS system requires quinolone signals. In addition to Gac-Rsm system (and not shown on the diagram), Las QS system is positively regulated by an orphan AHL receptor VqsR, cAMP-binding protein Vfr, polyphosphate kinase Ppk and DksA, a transcriptional factor that interacts with RNA polymerase through a secondary channel. Negative regulation of Las QS system is effected by a global transcriptional regulator AlgR2, HN-S like regulator MvaT, and an orphan AHL receptor QscR

In P. aeruginosa, 10% of the genome transcriptionally responds to AHL signals (Schuster et al. 2003; Wagner et al. 2003). Transcriptome and genetic studies have identified many direct and indirect targets for the LasR and RhlR regulators (Whiteley and Greenberg 2001). LasR activates virulence factors such as elastase (LasB), proteases (LasA), exotoxin A (toxA), and type II secretion (XcpR, XcpP). RhlR activates the production of rhamnolipids, but it is also necessary for the optimal production of LasB elastase, LasA protease, pyocyanin, and cyanide in vitro [reviewed in Williams and Camara (2009)]. The differential regulation of QS target genes depends on the recognition of conserved las, rhl boxes acting as binding sites for either or both regulators. The LasIR system has been initially shown to be involved in the differentiation of P. aeruginosa biofilms. For instance, a mutant defective in 3-oxo-C₁₂-HSL production forms an abnormal, flat, and undeveloped biofilm that is sensitive to low concentrations of the detergent sodium dodecyl sulfate (SDS) (Davies et al. 1998). The Las system, and to some extent the Rhl system, was later shown to contribute to the expression of the pel polysaccharide, one of the components of the P. aeruginosa pellicle in model biofilms (Sakuragi and Kolter 2007). More controversial is the role of rhlIR in model biofilm production, as rhamnolipids trigger Pseudomonas detachment from biofilms (Boles et al. 2005), but rhamnolipids have been shown to play a role in various aspects of biofilms development, including microcolony formation, maintenance of open channels, mushroom cap formation and dispersion from the biofilm (Davey et al. 2003; Espinosa-Urgel 2003).

The AHL-based QS is further complicated by the presence of two orphan LuxR-type regulators, QscR and VqsR. QscR has its own regulon of more than 400 genes. QcsR can function as either a repressor or an activator, but it has not been shown to bind AHLs (Dong et al. 2005; Juhas et al. 2005). VsqR is required for AHL production, it has been shown to bind 3-oxo-C₁₂-HSL and to regulate lasI (Fig.14.1).

The other branch of QS in Pseudomonas involves AQS molecules produced by the pqsABCDE operon under the positive control of PqsR (MvfR), a LysR regulator activated by LasR. PSQ and HHQ (a precursor of PQS) signals both act as coinducers of PqsR (Diggle et al. 2006, 2007b; Xiao et al. 2006a; Coleman et al. 2008; Farrow et al. 2008, and Fig.14.1). Exported HHQ is taken up by adjacent bacterial cells and converted into PQS by PqsH, a putative mono-oxygenase (Diggle et al. 2006; Xiao et al. 2006a; Bera et al. 2009). PqsR positively feeds back on rhlR expression. A direct negative regulatory feedback involves RhlR, which represses both pqsR and pqsABCDE. Mutations in either QS systems reciprocally reduce expression and therefore reduce virulence, as PQS signaling also affects elastase, lecA, pyocianin, biofilm formation. Transcriptome studies have revealed that PQS induces an iron starvation response upregulating genes involved in iron scavenging and siderophore biosynthesis; PQS can directly chelate ferric iron at high affinity, probably trapping the iron complexes outside the cell envelope (Bredenbruch et al. 2006; Fletcher et al. 2007). PQS also interacts with P. aeruginosa envelope causing the release of membrane vesicles, which plays a role in cell autolysis in biofilms (Baysse et al. 2005; Diggle et al. 2007b).

2.3 Integration of QS into Global Regulatory Networks

The expression of QS signal synthases and receptors is itself tightly controlled and at the crossroad of a complex networks of regulators (Fig.14.1). For example, in Pseudomonas, AHL-mediated QS circuits are controlled at the level of lasR transcription by Vfr, a catabolite regulator and cyclic-AMP binding protein (Albus et al. 1997). RsaL, a regulatory gene divergently transcribed upstream of the lasI promoter, directly represses lasI and is itself controlled by LasR (Rampioni et al. 2007). The stationary phase sigma factor RpoS negatively regulates rhlI expression (Whiteley et al. 2000), while an alternate sigma factor RpoN can up- or down-regulate rhlI and functions as a negative regulatory of lasI (Heurlier et al. 2003; Thompson et al. 2003). The nucleoid-like regulator MvaT is involved in growth phase-dependent expression of QS genes and behaviors affecting LecA, elastase, pyocyanin, swarming, biofilms, efflux pumps expression (Diggle et al. 2003). Under phosphate-limiting conditions, PhoB stimulates rhlR (Jensen et al. 2006). The transcriptional regulators DksA, AlgR, PtxR negatively controls rhlI, while PtxR or VqsM positively control lasI (Dong et al. 2005). Finally, positive regulation of lasR, rhlR is exerted by GacS/GacA, a two component signal transduction system acting as global regulator in γ-proteobacteria. Among other effects, mutations in this two-component system reduce production of the AHL signals and accumulation of the mRNAs encoding the AHL synthase and AHL receptor (Parkins et al. 2001; Quinones et al. 2004). The contribution of the GacS/GacA to the regulation of QS appears to be evolutionarily conserved in γ-proteobacteria and is further discussed below.

3 GacS/GacA Is a Two-Component System Controlling Environmental Adaptation, Biofilm Formation, and Motility in γ-Proteobacteria

3.1 Discovery of GacA and GacS in γ-Proteobacteria

gacA (global antibiotic and cyanide control) was first discovered in biocontrol pseudomonads in screens for mutants defective in the production of fungicidal secondary metabolites (Laville et al. 1992). The regulatory role for gacA orthologs in controlling production of toxins and antibiotic compounds is conserved in many soil and host-associated bacteria from the genus Pseudomonas [rev. Haas et al. (2002) and Teplitski and Ahmer (2004)]. In P. aeruginosa, gacA mutants were identified as less virulent in nematodes, plants, and mice (Tan et al. 1999). In the tomato pathogen Pseudomonas syringae pv tomato DC3000 (Pst DC3000), disruption of gacA reduced the virulence of the bacterium and reduced its ability to elicit a hypersensitive (apoptosis-like) response in plants. In several strains of P. syringae, gacA orthologs control the production of N-acyl homoserine lactone signals and other traits associated with fitness (Chatterjee et al. 2003; Quinones et al. 2004). Therefore, in pseudomonads gacA orthologs have a global regulatory role in controlling behaviors related to environmental fitness and adaption.

In various pathogenic bacteria, gacA orthologs were identified in screens of mutants defective in aspects of virulence. Salmonella sirA, for example, was identified in a screen for mutants with reduced expression of genes encoded within the Salmonella Pathogenicity Island 1 (SPI-1) (Johnston et al. 1996). Orthologs of gacA are also present in symbiotic and saprophytic γ-proteobacteria. In the soil bacterium, Azotobacter vinelandii gacA controls production of alginate (Castaneda et al. 2001). In Vibrio fischeri, gacA is required for colonization of the squid host (Euprymna scolopes) and regulates genes involved with chemotaxis, substratum-dependent motility, and production of N-acyl homoserine lactone signals (Whistler et al. 1998; Whistler et al. 2007). Of all the sequenced genomes of γ-proteobacteria (as of October 2009), gacA orthologues are only absent from genomes of some insect endosymbionts: genomes of these bacteria have undergone considerable reduction (Gil et al. 2004). This suggests the central role for GacA in all γ-proteobacteria in controlling behaviors that are possibly related to environmental sensing and adaptation.

GacA orthologs are “orphan” response regulators, that is, unlike many two-component response regulators, they are not encoded near a sensor kinase. Of the published genome sequences, the gacA stop codon either overlaps the uvrC start codon or lies just within the uvrC coding sequence in Salmonella, Photobacterium, Vibrio, Yersinia, Serratia, and Pectobacterium. The gacA-uvrC tandem gene arrangement is conserved throughout the γ-proteobacteria, with the exception of Xanthomonas spp. and Xylella spp. These genetic arrangements suggest that uvrC and gacA are often translationally coupled. Indeed, E. coli gacA ortholog (uvrY) was identified as an ORF containing regulatory elements for uvrC, a gene involved in nucleotide excision repair (Moolenaar et al. 1987). Similarly, in P. fluorescens, P. aeruginosa, E. coli, and Pectobacterium carotovorum, polar mutations in gacA result in altered UV tolerance, because a portion of the gacA transcript extends through uvrC (Moolenaar et al. 1987; Laville et al. 1992; Reimmann et al. 1997; Eriksson et al. 1998). However, gacA orthologs themselves do not function in UV resistance. The genomic regions upstream of gacA orthologs are much more variable than the downstream regions [rev. Teplitski and Ahmer (2004)]. Among the genera Salmonella, Klebsiella, and Escherichia, the gacA orthologs are located downstream of sdiA (but within a separate transcriptional unit). SdiA is a LuxR-type AHL receptor used by these bacteria to detect the quorum sensing signals of other bacterial species (Ahmer 2004). However, this arrangement is not conserved in other genera.

Because gacA is not encoded next to a sensor kinase, genetic (and later, biochemical) evidence was used to establish a link between gacA and a sensor kinase encoded by gacS. The first identified gacS ortholog (“lemA” for “non lesion forming”) was detected in P. syringae pv syringae based on the inability of the mutant to cause disease lesions on bean leaves (Hrabak and Willis 1992; Kitten et al. 1998). The gacS ortholog of a free-living soil nitrogen fixing bacterium A. vinelandii was identified in a screen of clones that complemented a nonmucoid (alginate-deficient) phenotype (Castaneda et al. 2000). Consistent with the postulated GacS–GacA interaction, various mutant screens identified gacS mutants as having phenotypes similar to those of the gacA mutants (Chancey et al. 1999).

3.2 GacS–GacA Signal Transduction

3.2.1 Structure/Function Analysis of GacS Orthologs

GacS orthologs are highly conserved among the γ-proteobacteria [Fig.14.2 and Heeb and Haas (2001)]. The gacS gene encodes an unorthodox sensor kinase with a HAMP phosphatase domain (residues 188–240), histidine kinase A (dimerization/phosphoacceptor) domain (residues 281–344), HATPase_c domain (Histidine kinase-like ATPase, residues 400–507); REC (signal receiver domain, residues 668–784) and a histidine phosphotransfer (HPT) domain (residues 860–910). Similar to other two-component sensor kinases, these three domains have conserved H295, D718, and H861 residues, predicted to have a role in autophosphorylation (H295) and phosphotransfer to GacA (Zuber et al. 2003). Corresponding single amino acid replacements in Pseudomonas GacS (H294, D717, H863) resulted in a gacS null phenotype, suggesting that these residues play an important role in affecting downstream gene regulation (Zuber et al. 2003). GacS orthologs have a predicted periplasmic loop (M33 to L162 of GacS) and the deletion of the predicted periplasmic domain did not affect the in vivo function of gacS (Zuber et al. 2003). Similarly, the ability to auto- and trans-phosphorylate in vitro was not affected by a deletion of the first 198 residues of BarA (an ortholog of GacS in Salmonella and E. coli) (Pernestig et al. 2001; Teplitski et al. 2003b, 2006).

Fig.14.2

CLUSTALW alignment of GacS orthologs. GacS is an unorthodox sensor kinase with a HAMP phosphatase domain (residues 188–240), histidine kinase A (dimerization/phosphoacceptor) domain (residues 281–344, with a conserved H295), HATPase_c domain (Histidine kinase-like ATPase, residues 400–507); REC (signal receiver domain, residues 668–784, with a conserved D718) and a histidine phosphotransfer (HPT) domain (residues 860–910, with a conserved H861). GacS orthologs from P. aeruginosa PAO1 (GenBank accession # NP_249619), P. fluorescens Pf-5 (#YP_261539), P. syringae pv tomato DC3000 (#NP_791516), A. vinelandii (#AAF04854), E. coli K12 (#YP_003035194), S. enterica sv Typhimurium LT2 (#NP_461879), Pectobacterium carotovorum (#CAA74020), and K. pneumoniae KP342 (YP_002236863) are shown

In P. fluorescens, the histidine kinase (HisKA), the ATPase, and the receiver REC domains of GacS were all needed for the interactions with GacA, based on the results of a bacterial two-hybrid assay (Workentine et al. 2009). The 60 amino acid residues immediately upstream of the REC domain were also required for the interactions of GacS with GacA (Workentine et al. 2009). The HPT domain was dispensable in the interactions of GacS with GacA (Workentine et al. 2009). Only HAMP domain was required for the interactions of two GacS proteins (Workentine et al. 2009). Deletion of the M190-E270 residues of HAMP domain and their replacement with Glu-Ala-Phe resulted in a signal-independent ON gacS (Zuber et al. 2003).

Important functional differences exist between GacS orthologs of pseudomonads and enteric bacteria. In pseudomonads, GacS interacts with two other sensor kinases, LadS and RetS (see below). This interaction leads to either inhibition or activation of GacS autokinase activity, presumably upon perception of unknown signals. Direct interaction between GacS and LadS or RetS has been shown (Goodman et al. 2009; Workentine et al. 2009). Enteric bacteria, such as E. coli, Salmonella, and their plant-associated relatives (Klebsiella, Erwinia, Pantoea, and Pectobacterium) lack orthologs of RetS and LadS. Even though the possibility that functionally similar sensor kinases may be present in enteric bacteria has not been ruled out.

3.2.2 Sensor Kinases RetS and LadS Modulate Function of GacS

In pseudomonads, RetS and LadS exert opposing effects on the GacS/GacA system (Ventre et al. 2006; Goodman et al. 2009; Workentine et al. 2009). The gene encoding the unorthodox hybrid sensor kinase RetS was isolated from a mutant producing an increased biofilm and was subsequently shown to be required for the activation of a type 3 secretion system, concomitant with the repression of biofilm formation, and colonization/dissemination in murine acute infection models (Goodman et al. 2004; Zolfaghar et al. 2005). gacA, gacS, rsmA and the diguanylate cyclase sadC were isolated as suppressor mutants of the RetS phenotype, implying a genetic link between them and retS (Goodman et al. 2004). It was later shown that genes under RetS control are inversely regulated by two other sensor kinases, GacS and LadS (Goodman et al. 2004; Ventre et al. 2006). Consistent with this model, microarray analyses revealed that the regulatory activity of RetS is channeled through GacS and GacA (Brencic et al. 2009).

RetS and LadS contain putative periplasm-facing signal receiver domains, located between a single N-terminal trans-membrane segment on one side and seven trans-membrane segments on the other. This class of periplasmic receptors is hypothesized to assume a jelly roll folding conformation as seen in lectins, implying that the signal may be a carbohydrate (Goodman et al. 2009). Neither LadS nor RetS interacts with GacA, but both LadS and RetS interacted with GacS, although interactions of GacS with LadS were weaker than GacS–GacS or GacS– RetS interactions. Complexes formed by GacS and RetS in cell membranes were very stable and resistant to treatment with mild detergents (Goodman et al. 2009). These observations indicate that sensor kinases LadS and RetS do not affect the phosphorylation status of the response regulator GacA, rather they modulate activity of the sensor kinase GacS in response to an unknown signal. One working hypothesis is that under ground signal conditions, RetS could form a homodimer, sequestered from GacS. Binding of an external signal may dissociate the homodimer and make RetS monomers available for inhibitory interactions with GacS (Fig.14.1 and (Goodman et al. 2009).

The RetS/GacS system constitutes a novel paradigm for bacterial signal transduction systems. Contrary to what other known sensor kinase do, RetS directly regulates the activity of GacS via formation of heteromeric membrane complexes (Goodman et al. 2009). This activity is consistent with the absolute dispensability of its conserved phosphorelay residues of RetS (Goodman et al. 2009). Sensor kinases involved in multistep phosphorelay reactions, such as GacS, share a conserved modular architecture and a common mechanism of phosphate transfer. They function as homodimers in which the g-phosphate from ATP bound by one monomer is transferred intermolecularly to the conserved histidine on the histidine kinase domain of the opposing monomer. The experiments described by Goodman et al. (2009) are consistent with a model in which RetS blocks GacS phosphorylation at this very early step of the signal transduction pathway, by disrupting the formation of GacS homodimers and instead forming a nonproductive GacS:RetS heterodimer or a higher-order structure of GacS and RetS homodimer (Fig.14.1 and Goodman et al. 2009).

RetS and GacS are constitutively coexpressed under most conditions [Goodman et al. (2009) and references therein]. Therefore, the effect of RetS on GacS must come at the posttranscriptional or posttranslation level, likely in response to a yet unknown signal. For example, in the pathogenesis model for P. aeruginosa PAO1, RetS is hypothesized to perceive a cue important for the switch between acute and chronic infections (Goodman et al. 2009). In fluorescent soil pseudomonads, RetS was responsible for the temperature-dependent regulation of sRNAs and the downstream genes involved in the production of secondary products, while effects of LadS were independent of the temperature (Humair et al. 2009).

It appears, therefore, that the Gac signaling network in pseudomonads consists of RetS, LadS, and GacS/GacA and controls the expression of a significant number of genes at the level of mRNA translation and/or stability. Taken together, these genetic approaches identified multiple components of a molecular switch controlling adaptation to various niches relevant to its pathogenesis and adaptation to different environmental niches.

3.3 Structure/Function Analysis of GacA Orthologs

Similarly to other FixJ/LuxR-type transcriptional regulators, GacA functions as a dimer (as demonstrated by the bacterial two-hybrid experiments (Workentine et al. 2009). Unlike LuxR AHL receptor proteins (multimerization of which is facilitated by binding of a ligand directly to the N-terminus of the protein), orthologs of GacA become active once phosphorylated by a sensor kinase GacS (as discussed above). The crystal structure for NarL, a member of the FixJ family of regulators, has been determined (Baikalov et al. 1996). By comparison with NarL, GacA orthologs are predicted to have an N-terminal receiver domain and a C-terminal DNA binding domain (Baikalov et al. 1996). The two domains are connected by an a-helix (a6, residues S127–S136). The N-terminal receiver domain (M1-S127) is comprised of five parallel b-sheets surrounded by five a-helices (Baikalov et al. 1996). D54 is the predicted phosphorylation site. In some two-component response regulators a D54E mutation leads to a constitutive ON (kinase-independent) phenotype by mimicking the phosphorylated state of the aspartate residue. However, in most cases D54E mutants are unable to accept the phosphate group and have a null phenotype.

Several amino acid residues are conserved in the N-terminus of all FixJ-type regulators. The following regions of GacA are highly conserved and are predicted to interact with the phosphorylation site: D8-D9, S34-C35, P58-I61, T82-E86, S103-A107 (Fig.14.3). Consistent with the prediction that these conserved loops interact with the phosphorylation site, C86T, C84T, and G105R yielded a gacS-independent, constitutively ON, GacA in P. fluorescens (Gaffney and Lam 1999). In several systems, mutations in the residues corresponding to D8-D9 and V83 alter the function of the protein and result in constitutively active response regulators (Smith et al. 2004). A V88A mutation in NarL (and also in the corresponding residue of CheY) results in a ligand or cofactor-independent, but phosphorylation-dependent regulator (Egan and Stewart 1991). Curiously, while V83 (corresponding to V88 of NarL) is conserved in the GacA orthologs of E. coli, Salmonella, P. aeruginosa, Xylella fastidiosa, and Azotobacter, natural V83A variations are found in GacA orthologs of Yersinia pestis, P. syringae DC3000, and L. pneumophila (isoleucine is found at position 83 of P. carotovorum, Photobacteirum profundum, and V. fischeri). The consequences, if any, of the V83A and V83I variations are unknown. It will be of interest to explore whether, like in CheY and NarL, V83 functions as a site for interaction of GacA orthologs with a ligand, a cofactor or docking with another protein to provide a hydrophobic pocket for V83, and whether A83 or I83 found in some bacteria may result in a different set of regulatory requirements for a particular GacA ortholog. Additionally, point mutations in P. fluorescens gacA, yielding M20I (in α1), and Q132R (in α6) resulted in kinase-independent regulators (Gaffney and Lam 1999). M20 is conserved only in GacA orthologs of Pseudomonas spp., Azotobacter, and Yersinia; isoleucine (or leucine) residues in the corresponding position are found in the GacA orthologs of E. coli, Salmonella, Legionella, Xylella, Photobacterium, Pectobacterium, and Vibrio and also in the related response regulators NarL, UhpA, FimZ, and RcsB. Q132 is present in most orthologs of GacA, but not in the GacA ortholog of Xylella or the related two-component response regulators, NarL, UhpA, FimZ, and RcsB. Because these two residues (M20 and Q132) are found in a limited number of proteins, it is difficult to speculate about the biochemical basis of the constitutive ON phenotype observed in the corresponding P. fluorescens gacA mutants.

Fig.14.3

CLUSTALW alignment of GacA orthologs and NarL. The predicted protein sequences of GacA orthologs show a high degree of conservation. For example, S. Typhimurium SirA and E. coli UvrY share 96% amino acid identity, well above the average of 90% for these two genera. GacA of P. aeruginosa and UvrY are 60% identical. Conserved amino acid residues are shaded. The N-terminus (residues 5–119) contains a signal receiver domain, the C-terminus (149–205) contains DNA-binding domain typical of all LuxR-type transcriptional regulators. GacA orthologs from S. enterica (GenBank Accession # AAC08300), plant-associated K. pneumoniae 342 (#ACI09250), E. coli K-12 (#AP_002529), soft rot pathogen P. carotovorum (#YP_003018235), V. fischeri (#AAQ83589), Photobacterium profundum (YP_130426), L. pneumophila (#AAL79360), A. vinelandii (#AAK97431), P. syringae pv. syringae DC3000 (#NP_792819), P. fluorescens Pf-5 (#YP_260665), P. aeruginosa PAO1 (#AAG05974), X. fastidiosa (#NP_780164), and NarL from E. coli (#CAA48935) are shown. Conserved Asp54 residue, the predicted site of phosphorylation, is indicated by an asterisk

The C-terminal DNA binding domain of the GacA protein consists of four α-helices (α7-10). The central helices α8 and α9 are predicted to form a helix-turn-helix motif, supported by the flanking hydrophobic core of helices α7 and α10. Based on the comparison with the solved crystal structure for NarL, residues I157, I170, and V181 are predicted to build a hydrophobic cluster that fixes α7–α9 in their positions. L175 is predicted to anchor the α8–α9 loop to the cluster, while G164 assures a proper angle between α7 and α8 (Baikalov et al. 1996). Spontaneous mutations predicted to affect the structure of the DNA-binding helices in P. fluorescens GacA resulted in a null phenotype (L173P in α8 of the HTH and A203V in α10 of the HTH) (Whistler et al. 1998; Bull et al. 2001). This is consistent with the reduction in activity of the corresponding mutants in the HTH-domain of LuxR from V. fischeri (Egland and Greenberg 2001; Trott and Stevens 2001). Though not yet demonstrated biochemically, these mutants are likely to be defective in their ability to interact with target DNA sequences. As shown in Fig.14.3, these residues are highly conserved in the GacA orthologs of different γ-proteobacteria.

In addition to the transphosphorylation by GacS, GacA orthologs appear to be capable of being phosphorylated from the cellular pool of acetyl phosphate (Ac-P) independently of the sensor kinases (Lawhon et al. 2002). At pH 6.7, acetate (or Acetyl-CoA) is phosphorylated to Ac-P via ackA (or pta) and at least partially compensates for the barA mutation in S. enterica (Lawhon et al. 2002). This effect of Ac-P may explain why, in general, the consequences of the barA mutation are not as severe as the consequences of the sirA mutation. The ability of GacA orthologs to be phosphorylated from alternate P-donors under some conditions may also explain why the phenotypes and transcriptional units under control of gacS orthologs overlap only partially (Lawhon et al. 2002).

3.4 GacA Regulons in Soil γ-Proteobacteria

3.4.1 Evolutionarily Conserved Targets of the GacS/GacA Orthologs: The csr RNA

Since their original discovery, systematic efforts have been made to identify GacA regulons in various bacteria. In Salmonella, for example, the initial studies identified only the horizontal acquisitions Salmonella Pathogenicity Island (SPI) 1, SPI4, and SPI5, which are controlled by the sirA (gacA)-dependent expression of SPI1 regulators hilA and/or hilC (Johnston et al. 1996; Ahmer et al. 1999). A more ancient function was identified when it was determined that SirA represses flagellar genes via controlling the Csr regulatory system (Pernestig et al. 2001; Teplitski et al. 2003b). In fact, recent studies have demonstrated that the effects of GacA orthologs on motility and horizontally acquired virulence genes are mediated via the posttranscriptional regulatory system Csr (Rsm) (Kay et al. 2005; Lapouge et al. 2008; Brencic et al. 2009).

The Csr (Rsm) system depends on the activity of the RNA-binding protein CsrA (known as RsmA in pseudomonads and erwinias). Binding of CsrA (RsmA) to mRNA either stabilizes or destabilizes transcripts (Romeo et al. 1993; Romeo 1998b). Published genomes of pseudomonads and erwinias also contain a second CsrA homologue, RsmE, with apparently redundant functions (Reimmann et al. 2005). Small noncoding regulatory RNA antagonizes effects of CsrA proteins. Mechanistically, this antagonism is based on the folding of csr/rsm sRNAs into secondary structures which expose loops and bulges, which bind CsrA and reduce the intracellular concentration of the unbound CsrA (Romeo et al. 1993; Romeo 1998b).

The number and sequence of csr (rsm) sRNAs are species-specific. For example, all pseudomonads contain rsmY and rsmZ sRNA; however, P. fluorescens also contains rsmX (Kay et al. 2005; Brencic et al. 2009). Genome-wide searches identified at least one additional gacA-dependent sRNA in pseudomonads, although its role in the GacS/GacA-Rsm signal transduction is not yet fully elucidated (González et al. 2008). Functional homologs of rsm sRNAs in E. coli and S. enterica are csrB and csrC, and rsmB and rsmC in erwinias. There is little sequence conservation among these sRNAs; however, they all form stem-loop secondary structures with exposed GGA bulges (Romeo 1998a, b; Valverde et al. 2004; Kay et al. 2005; Lapouge et al. 2008). These features appear to be sufficient for their function, as demonstrated by the ability of over-expressed rsmY or rsmZ from P. fluorescens CHA0 to mimic a csrA mutation in E. coli (Valverde et al. 2004) and further supported by the experiments in which synthetic sRNA containing exposed GGA bulges functionally complemented the triple rsmX rsmY rsmZ mutant of P. fluorescens (Valverde 2009).

The link between orthologs of gacA and the Csr system was first demonstrated by showing that the effects of Pseudomonas gacA on the downstream genes were posttranscriptional and required the Rsm system (Blumer et al. 1999; Kay 2005 #3014). Consistent with the predicted role of the Csr system in the GacA regulatory cascade, the overexpression of RsmA mimics a gacA mutation in Pseudomonas (Blumer et al. 1999), and a double rsmA rsmE mutant overcame the gacS mutation (Reimmann et al. 2005). Conversely, overexpression of the genes encoding regulatory sRNAs can complement mutations in gacS or gacA orthologs and a deletion of all csr RNAs mimics a gacA mutation in different bacterial species (Aarons et al. 2000; Altier et al. 2000a, b; Heeb et al. 2002; Valverde et al. 2003; Weilbacher et al. 2003). A more rigorous support for this hypothesis was provided by proteomic and transcriptomic profiling of P. aeruginosa gacA and rsmY rsmZ mutants (Kay et al. 2006; Brencic et al. 2009), which demonstrated that in P. aeruginosa gacA-regulated proteins and transcripts are controlled via the rsmY and/or rsmZ. It is of note, that while the effects of gacA in P. aeruginosa are due to the regulatory effects of GacA on the Csr system, CsrA has additional targets: CsrA regulates expression of the ribosomal protein S1 independently of the csr sRNAs (Kay et al. 2006).

The effects of GacA on the Csr (Rsm) posttranscriptional regulatory system are due to the control of the expression of csr/rsm sRNA by GacA orthologs. Phosphorylated GacA of P. aeruginosa most likely binds to the conserved sequences (TGTAAGCATTAACTTACA or TGTAAGCCAAGGCTTACA) within the promoter regions of the target sRNA genes (Heeb et al. 2002; Kay et al. 2006). A similar sequence (TGTAAGACAAGGTGAAAC) was identified within the predicted promoter region for the S. enterica csrB, which was shifted by phosphorylated SirA (Teplitski et al. 2003b). These observations suggest that the binding site for GacA orthologs may be conserved in γ-proteobacteria.

Interestingly, in vitro gel mobility shift assays demonstrated that a fragment upstream of hilA (the horizontally acquired regulator of SPI1) was shifted by phosphorylated SirA (Teplitski et al. 2003b). This shifted fragment contained the sequence TTAAAGCACAGGCATAAG, which is similar to the hypothetical conserved GacA DNA binding site. Furthermore, this shifted fragment is distinct from the hilA promoter controlled by other regulators (Ellermeier et al. 2005). It remains to be established whether/how this and other horizontally acquired genes became integrated into the SirA (GacA) regulon of Salmonella both at the level of direct transcriptional regulation by SirA and by posttranscriptional regulation via the Csr system.

3.4.2 Orthologs of gacS/gacA Are Central to Biofilm Formation in γ-Proteobacteria

Almost invariably, mutants in gacS, gacA are defective in biofilm formation. Biofilms are multicellular highly structured, sessile microbial communities that form on submerged or wet surfaces, including surfaces of plant leaves, roots, and plant-associated fungi [rev. Davey and O’toole (2000), Hall-Stoodley et al. (2004), and Danhorn and Fuqua (2007)]. Formation of a biofilm is a complex, multistep process that requires surface conditioning, reversible attachment, adhesion, cell aggregation, biofilm maturation, and eventual dispersal (Hall-Stoodley et al. 2004). It has been hypothesized that one of the early steps preceding biofilm formation requires that the bacterial cells “make a decision” as to whether to continue active colonization of surfaces via surfactant-mediated flagella-dependent swarming motility or to form sessile biofilms (Verstraeten et al. 2008). Consistent with this hypothesis, swarming motility and biofilm formation are often divergently regulated [rev. Verstraeten et al. (2008)]. Furthermore, consistent with this hypothesis, nonmotile nonflagellated flhDC S. enterica mutants formed more biofilm compared with the wild type (Teplitski et al. 2006). Mutant analysis revealed that flagellated but chemotaxis-defective mutants or those with paralyzed flagella formed reduced biofilms, while mutants in structural flagellar genes increased biofilm formation (Teplitski et al. 2006). It should be noted, however, that the effects of flhDC on biofilm formation even by the same bacteria are not always the same in different studies, especially when different media or surfaces are used [rev. Davey and O’Toole (2000)].

Orthologs of gacA downregulate swarming or swimming motility in pseudomonads and vibrios (Goodier and Ahmer 2001) and also positively regulate biofilm formation in the same bacteria (Parkins et al. 2001). Both of these effects are due to the effects of the GacS/GacA cascade on the Csr (Rsm) posttranscriptional regulatory system. As described above, GacA up-regulates transcription of the csr (rsm) regulatory sRNAs, which antagonize activity of the CsrA (RsmA) RNA-binding protein and this promotes synthesis of polymers required for biofilm formation (Suzuki et al. 2002; Wang et al. 2005; Brencic et al. 2009). It is not yet known whether the GacS/GacA-Csr regulatory cascade contributes to this swarming vs. biofilm “decision making” or whether it is one of the regulatory cascades, which sets off as the consequence of this decision.

4 The Elusive GacS Signal

Currently, there is little information regarding the signal(s) detected by GacS. Gene activation mediated by GacS/GacA orthologs has been observed during growth on common laboratory media and within plant and animal tissues. It is not yet clear whether the signal that ultimately triggers the function of GacA is the same in all bacteria. As discussed above, in pseudomonads (but not in Enterobacteriacea), phosphorylation status of GacS depends on the input of two other sensor kinases RetS and LadS. Even though homologues of retS, ladS are absent from the genomes of Enterobacteriacea, it may be premature to rule out the possibility that RetS or LadS-like sensor kinases alter the function of GacS in enterobacteria. On another extreme, in Xyllela a functional GacA is present, while a GacS ortholog is missing (Shi et al. 2009). As shown in Fig.14.3, the N-terminus of GacA _X.f. contains the conserved amino acid residues involved in phosphorylation. Therefore, it is likely that GacA _X.f. does not directly bind a ligand, rather is phosphorylated by a yet-unknown sensor kinase. As described below, several research groups are focusing their efforts on identifying a signal for the GacS/GacA regulatory system.

4.1 Evidence for the Self-Produced GacS Signal

During batch culture, GacS of P. fluorescens responds to a compound in its own culture supernatant by increasing the expression of rsmZ fivefold as the cultures entered stationary phase (Heeb et al. 2002). Follow-up studies demonstrated that hcnA (another gene under gacA control) was similarly activated by the culture supernatant extract. The ability of the signal to affect expression of hcnA::lacZ depended on the presence of the sensor kinase gacS. Production of the signal is controlled by the GacS/GacA-Rsm system (Kay et al. 2005). The signal perceived by GacS was extractable from the stationary phase cultures with dichloromethane (Heeb et al. 2002; Zuber et al. 2003). Even though the GacS signal is present in stationary phase cultures and is hydrophobic, it is not an AHL (Heeb et al. 2002; Heurlier et al. 2004).

4.2 Evidence for the Eukaryotic Contribution to the GacS/GacA-Mediated Signaling

A compound from seed exudates was reported to stimulate the expression of a luxAB fusion in a gene under GacS/GacA control in a plant-associated Pseudomonas spp. DSS73 (Koch et al. 2002). The effect of the plant compound required functional gacS, suggesting that the signal (or signal-mimic) for GacS may be produced by eukaryotes as well. Expression of rsmB from Dickeya dadantii was modestly (~1.5-fold) up-regulated in the presence of 100mM o-coumaric acid, a common plant phenolic acid (Yang et al. 2008). Expression of gacA was not affected by the compound; however, the effect of o-coumaric acid on rsmB required a functional gacS, suggesting the o-coumaric acid may be affecting the cascade at the level of signal perception or transduction (Yang et al. 2008).

5 Conclusions and Future Directions

One could argue that soils in temperate regions truly represent the most extreme environments on Earth. After all, environmental conditions in hot springs or permafrost are reasonably predictable. In temperate soils, on the other hand, microorganisms deal with cycles of flooding and drought, oscillations of temperature, predation, intra- and interspecies competition, as well as uncertain availability of nutrients. To sense environmental conditions and to change patterns of gene regulation accordingly, soil bacteria have evolved sophisticated signal transduction mechanisms.

In γ-proteobacteria, the GacS/GacA two-component regulatory system is one of the mechanisms for posttranscriptional regulation of gene expression in response to cues that reflect environmental conditions. Orthologs of gacA are found in the genomes of all γ-proteobacteria, with the exception of those that have undergone extreme reductions. The structure and function of GacA orthologs is highly conserved in γ-proteobacteria: upon phosphorylation of a conserved aspartate residue, these FixJ-type regulators bind within promoters of sRNA genes and thus set off the regulatory cascade, which is mediated by a posttranscriptional regulatory protein CsrA (RsmA). The regulatory inputs leading to the phosphorylation of GacA are variable. GacA orthologs of Pseudomonas spp., S. enterica, and E. coli are phosphorylated by GacS orthologs. The same mechanism of GacA activation is highly likely in other γ-proteobacteria based on the wide distribution of gacS orthologs in genomes of most other γ-proteobacteria. Interestingly, genomes of Xylella and Xanthomonas lack identifiable orthologues of gacS. However, gacA _X.f. was shown to be functional and complemented gacA mutations in heterologous hosts. This discovery may provide important clues for the understanding of the function and the evolution of the GacS–GacA signal transduction. In pseudomonads, the phosphorylation status of GacS depends on two other sensor kinases, RetS and LadS; orthologs of retS and ladS are not present in the genomes of enteric bacteria. Signals required for the autophosphorylation of GacS and activation of RetS are not known. However, the available evidence indicates that GacS responds to a self-produced hydrophobic molecule, while activity of RetS is dependent on the temperature. Several research groups are working on identifying the signal(s) that set off the GacS/GacA regulatory cascade as well as inhibitors that would block the function of this regulatory system.

While the GacS/GacA system is present only in γ-proteobacteria, quorum sensing mediated by N-acyl homoserine lactone (AHL) signals has been characterized in all subgroups of proteobacteria. Even though AHL-mediated signaling is wide spread, it is not universal: closely related bacteria may differ substantially in the repertoire of the AHLs they produce and perceive. At least in part, this is due to the fact that the genes responsible for the synthesis of AHLs and their perception are borne on plasmids, transposons, or other mobile elements. Interestingly, those AHL receptor genes that are evolutionarily conserved among distantly related bacteria are often encoded on chromosome and are unlinked to AHL synthase genes in most soil bacteria. In light of these observations, one may wonder whether the AHL-mediated QS is a “social behavior” or whether it is more akin to a “social disease,” which is transmitted by mobile genetic elements. Anthropomorphic comparisons aside, these observations highlight the need for an unbiased evaluation of the evolutionary origins of AHL-mediated quorum sensing and its ecological consequences.