Systems level analysis of two-component signal transduction systems in Erwinia amylovora: Role in virulence, regulation of amylovoran biosynthesis and swarming motility

Prokaryotes use their small size and metabolic diversity to dominate every conceivable niche on earth. A large part of this success comes from the evolution of elaborate sensory systems to monitor and respond to dramatic fluctuations in their environment. Primary means of signal transduction in bacteria involve two-component signal transduction systems (TCSTs) [1, 2]. The term "two-component" was coined in 1986 to describe a new class of regulatory systems found in bacteria; today, two-component systems represent major paradigms for signal transduction in prokaryotes and in lower eukaryotes [3, 4]. The prototypical two-component system consists of a histidine kinase protein (HK) containing a conserved kinase core and a response regulator protein (RR) containing a conserved regulatory domain [4, 5]. Extracellular stimuli are sensed by, and serve to modulate HK activities. The HK is autophosphorylated at a histidine residue, creating a high-energy phosphoryl group that is subsequently transferred to an aspartate residue in the RR via a reaction catalyzed by the RR itself. Phosphotransfer to the RR leads to activation of a downstream effector domain that elicits a specific response [1, 4].

Two-component systems are distributed at varying frequencies among organisms of all domains, including Eubacteria, Archaea, and Eukarya [1, 3]. The availability of complete genome sequences has allowed for a definitive assessment of the prevalence of two-component proteins. There are 30 HKs (five of which are hybrid kinases) and 32 RRs in Escherichia coli [6]. However, the number of two-component proteins differs greatly among bacteria. Often, parasitic bacteria encode fewer signaling proteins than relatively free-living bacteria [7]. Among sequenced plant pathogenic bacteria, the number of TCST genes is quite different. For example, the xylem-limited Xylella fastidiosa has the fewest TCST genes; whereas, the relatively adaptable Pseudomonas syringae and Xanthomonas spp. have the largest number of TCST genes [8, 9]. Perhaps, the most attractive reason for studying two-component systems in bacteria is that TCST systems are used by pathogenic bacteria to control expression of virulence factors required for infection. Many such TCST systems are identified in both human and animal pathogens and in plant pathogens [7, 10]. The Agrobacterium tumefaciens VirA/VirG system, the GacA/GacS of both Pseudomonas sp. and Pectobacterium carotovora, and the RpfCG of Xanthomonas spp. and X. fastidiosa are probably the most well-known and studied TCST systems involved in virulence gene expression in plant pathogens [10–12]. However, most of these studies focused on one or a few TCSTs. Only recently has genome-wide analysis of TCSTs in plant pathogenic bacteria become possible due to increased availability of whole genome sequences [8, 9, 13].

E. amylovora is the causal agent of fire blight, a devastating necrotic disease affecting apple, pear, and other rosaceous plants. Fire blight is one of the most important bacterial plant diseases worldwide that has a significant economic impact, resulting in crop losses of millions of dollars per year [14]. As a member of the Enterobacteriaceae, E. amylovora is related to many important human and animal pathogens such as Es. coli, Salmonella enterica, Shigella flexneri, Yersinia enterocolitica, and Y. pestis. Like many other Gram-negative plant pathogenic bacteria, E. amylovora utilizes both a type III secretion (T3SS) apparatus to deliver effector proteins into host plant cells, and the extracellular polysaccharide (EPS) amylovoran to cause disease [15–17]. In E. amylovora, structural components of T3SS encoded by the Hrp regulon are regulated by the two-component system HrpX and HrpY, which direct the expression of the NtrC family σ⁵⁴-dependent, enhancer-binding protein HrpS [18]. Both HrpY and HrpS function in the activation of expression of the alternate sigma factor HrpL, thereby regulating various genes and operons of the Hrp regulon. Expression of hrpX and hrpS is regulated by low pH, low nutrients, and low temperature conditions, mimicking the plant apoplast [18]. The biosynthesis of amylovoran is regulated by another TCST system, the RcsCDB phosphorelay system [17]. Recently, we identified several TCST genes including hrpX, grrS, and envZ that are induced during infection of host tissue in E. amylovora, indicating that TCSTs are key players in controlling the expression of virulence factors required for infection, and thus we hypothesize that there are likely networks controlling virulence gene expression [19]. However, other TCSTs in E. amylovora have either not been identified or have not yet been studied. This prompted us to use a systems approach to explore the function of TCSTs in E. amylovora. Understanding the genetics and molecular mechanisms of E. amylovora signaling will greatly enhance the likelihood of developing novel methods of controlling the disease.

The EnvZ-OmpR and GrrS-GrrA (also called GacSA, BarA-UvrY) are two widely-distributed and well-studied TCSTs in γ-proteobacteria. They represent those paradigms of signal transduction systems having pleiotropic effects, thus suggesting both systems are global regulators [2, 20]. EnvZ is a transmembrane sensor that predominantly responds to acidic pH conditions and changes in osmolarity, and subsequently phosphorylates OmpR. The EnvZ-OmpR system has been originally reported to govern the expression of ompC and ompF genes, encoding two major outer-membrane porins [21]. In addition to its role in porin osmoregulation, OmpR is involved in regulating various cellular components including flagellar gene expression, fatty acid transport, curli fibre formation, and cell division as a dual regulator; i.e., negative or positive [22]. In Salmonella spp., OmpR-EnvZ regulates another TCST system SsrA-SsrB, that in turn regulates the T3SS produced by Salmonella pathogenicity island 2 (SPI-2) [23–26]. OmpR also negatively regulates expression of invasin in Yersinia enterocolitica and the T3SS in P. syringae [27, 28].

GacS and GacA homologs have been identified in many Gram-negative bacterial genera, including Azotobacter, Erwinia, Escherichia, Legionella, Pectobacterium, Pseudomonas, Salmonella, Serratia, and Vibrio [12, 20, 29]. The sensor kinase GacS was initially discovered as a key regulator of virulence in the plant pathogen P. syringae [see review [12]]. The GacSA system has since been reported to regulate an array of phenotypes, including biofilm formation, alginate biosynthesis, production of toxins and extracellular enzymes [30, 31], proteases, siderophores, swarming motility [32], and type III secretion [10, 29, 33–35]. Similar to EnvZ-OmpR, GacSA also has a dual regulatory function; i.e., as a positive or negative regulator. The GacS/GacA (BarA/UvrY) system positively controls the expression of one to five genes specifying small RNAs (sRNAs), thus upregulating the productions of proteins that are otherwise repressed by small RNAs (RsmA/CsrA) [32, 36]. Most gacS/gacA mutants demonstrate reduced production of virulence factors and reduced virulence in a variety of host-pathogen systems [12, 20]. An alternative scenario is that the GacSA cascade downregulates the expression of flagellar genes in P. fluorescens or Es. coli [20].

In this study, our genome-wide analysis, using the recently closed genome sequence of E. amylovora Ea273 http://www.sanger.ac.uk/Projects/E_amylovora/ identified a total of 46 TCST genes in E. amylovora including 17 sensor kinases, 20 response regulators, three hybrid HKs, four RRs with enzymatic output domain (EAL-GGDEF proteins), and two kinases. A comparative genomic analysis of TCSTs was then conducted in related enterobacteria enabling classification of the TCSTs in E. amylovora. A total of 59 deletion mutants were generated, and their contribution to virulence, amylovoran biosynthesis, and swarming motility was characterized. Our findings suggested that TCSTs in E. amylovora played a major role in virulence and in the regulation of amylovoran biosynthesis and swarming motility. This indicated that networks of gene regulation existed in E. amylovora that respond to different environmental and host signals.

In this study, 46 TCSTs have been identified in E. amylovora, a pathogen of rosaceous plants which mainly resides in the plant xylem, but can also grow epiphytically on stigmas of flowers. Compared to other plant pathogenic enterobacteria such as D. dadantii and P. carotovora subsp. atroseptica, both of which can survive not only in plants, but also in soil, or the animal counterpart Es. coli, the number of TCSTs present in E. amylovora is relatively small, thus may reflecting the particular host niche for this pathogen. Furthermore, the genome size (3.9 Mbp) of E. amylovora is also the smallest among sequenced enterobacterial plant and mammalian pathogens [53]. However, E. amylovora still maintains both enterobacterial-specific TCSTs (such as the Rcs system) [54] and plant enterobacterial-specific ones (such as HrpXY) [39].

On the other hand, E. amylovora has almost the same number of TCSTs as that found in E. tasmaniensis, a saprophytic bacterium isolated from apple flowers in Tasmania, Australia [38]. Since both bacteria occupy the same ecological niche during colonization and can grow epiphytically on flowers, these bacteria might have developed and maintained similar TCSTs. This suggests that TCSTs may be evolutionarily maintained to cope with similar environmental and plant host signals in closely related bacteria. It is interesting to note that fire blight is endemic to North America, and it has subsequently spread to Europe and New Zealand in the 1950s and in 1917, respectively, but it has not yet been reported in Australia [14]. It is possible that these two bacteria may have not yet encountered each other, as the distribution of E. tasmaniensis outside of Australia remains unknown. Surprisingly, we found that the majority of TCSTs in these two bacteria share a high level of aa similarity and identity. These results suggest that similar TCSTs might play important roles for the survival and proliferation among closely-related bacteria in similar plant niches.

Currently, several thousands of TCSTs have been identified in sequenced bacterial genomes [55–57]. Although the basic biochemistry of TCSTs is well understood, some structural insights in phosphorylation-dependent changes of TCST domains are variable. Domain architecture has proven particularly informative for analyzing multidomain proteins involved in signal transduction and in predicting the functions of these signal transduction proteins [43, 58]. In RRs, structural characterization of DNA-binding domains has revealed several variations on the common helix-turn-helix (HTH) theme, such as NarL- and OmpR-types. Recently, many novel conserved domains have been described such as PAS, GAF, GGDEF, EAL, and HD-GYP, thereby affirming the complexity of bacterial signaling systems [43, 58]. In E. amylovora, most HKs belong to four common HK groups, and are periplasmic sensors. Moreover, most RRs in this bacterium are OmpR-, NarL-, NtrC-, and LytR-like proteins, the four most common families of DNA-binding RRs found in prokaryotes [43]. These results indicate that E. amylovora has maintained some basic signal transduction systems for the bacterium to survive. Phylogenetic and genomic analyses have revealed co-evolutionary relationships between cognate HKs and RRs. This seems obvious in E. amylovora as orphan HKs and RRs are rare.

It is well understood that TCSTs are involved in regulating virulence gene expression in plant bacterial pathogens [10]. Previous reports on TCSTs have demonstrated the importance of TCSTs in the virulence of bacterial plant pathogens. However, data on complete or global virulence regulation networks are lacking. A recent study by Qian et al. [13] provides a useful beginning towards a better understanding of the regulatory networks involved. A genome-wide mutagenesis of all 54 RRs in X. campestris pv. campestris has revealed that two novel RRs are involved in virulence, thus facilitating future studies on signaling networks in this bacterium [13]. In this study, we have utilized a reverse genetic approach and constructed 59 HK and RR mutants in E. amylovora, which will also provide valuable tools for future global gene expression assays using microarrays to deduce signaling networks in this bacterium.

Early studies have revealed that in E. amylovora, the Hrp T3SS, which delivers effector proteins into host plants, and the EPS amylovoran are two major virulence factors [52, 59, 60]. Previous reports have also indicated that the RcsCDB phosphorelay system regulates amylovoran biosynthesis, while the two-component system HrpXY regulates hrp-T3SS gene expression. Recently, we have further demonstrated that the Rcs system is essential for virulence in E. amylovora and may play a role in the survival of the pathogen [17]. Mutations in the Rcs system have rendered the organism non-pathogenic [17]. In this study, we have found that hrpX, hrpY and hrpXY mutants remain virulent, and could induce a spotty weak hypersensitive response (HR) on tobacco (Figure S2A, [see Additional file 2]); while, a hrpXYS triple mutant has a normal Hrp- phenotype. It is interesting to note that, in a previous report, Tn5-insertional mutants of hrpY have been reported to be non-pathogenic, and could not induce an HR on tobacco [18]. Two classes of hrpX insertional mutants have been identified, one similar to the hrpY mutant and the other that continue to cause disease and induce a spotty HR on tobacco [18]. Similar observations have been reported in Pantoea stewartii subsp. stewartii, causal agent of Stewart's wilt of corn [61]; wherein, Tn5-insertional hrpX and hrpY mutants exhibit a Hrp- phenotype and in-frame deletion hrpX mutants show reduced virulence. Since the hrpXY is transcribed as an operon, it is possible that Tn5 insertion could cause polar effects on the downstream genes such as hrpS (Figure S2B, [see Additional file 2]). Indeed, a recent study indicates that the hrp regulatory genes in P. stewartii subsp. stewartii participate in a novel regulatory loop that upregulates itself by readthrough transcription of hrpL into hrpXYS [62].

It has been proposed that in E. amylovora, both HrpY and HrpS regulate hrpL, encoding the master regulator of T3SS, and that the effects of HrpY and HrpS are additive [18]. Subsequent studies of P. stewartii subsp. stewartii, D. dadantii, and Pantoea herbicola pv. gyposophilae have demonstrated that HrpY initially activates hrpS by binding to its promoter, and then the HrpL is activated by HrpS as well as by other regulatory genes [39, 63]. Furthermore, it has been reported that HrpY in D. dadantii acts as both a positive and negative regulator [39]. In P. stewartii subsp. stewartii, besides the HrpY binding site, additional sequences further upstream of the hrpS promoter are also required for hrpS expression, suggesting that unknown regulatory proteins may act cooperatively with HrpY [63]. Microarray studies suggest that hrpL represents only one branch of the regulatory pathways downstream of hrpRS, and a large number of genes regulated by HrpRS are hrpL-independent in P. syringae [64]. In our study, HrpX, HrpY, and HrpS also act as negative regulators, as hrpX, hrpY, hrpXY, and hrpXYS mutants produce more amylovoran. Our virulence tests suggest that the T3SS remains functional in hrpX, hrpY, and hrpXY deletion mutants, but not in the hrpXYS mutant, whereby HrpS and HrpL may be activated by other unknown regulators except that production and/or translocation of HrpN (Harpin) in tobacco is severely attenuated as showed in an HR assay (Figure S2A, [see Additional file 2]). It is possible that host signals affecting gene expression may also be different in tobacco.

Various models have proposed that HrpX senses environmental signals in the plant apoplast or the Hrp-inducing medium to phosphorylate HrpY [18, 34, 35]. However, domain structure analysis has indicated that HrpX is a soluble cytoplasmic protein, and may sense intracellular signals. This suggests that other signaling pathways may also be involved in activating hrpXY, hrpS, or hrpL by sensing outside signals to regulate T3SS. Since the expression of HrpX, HrpS, and HrpL is regulated by low pH, also corresponding to conditions under which OmpR-EnvZ and GrrS-GrrA are activated, our results further suggest that both OmpR-EnvZ and GrrS-GrrA can regulate the hrpXY operon or hrpS either directly or indirectly as reported in other plant pathogenic bacteria [10, 28, 33, 34]. Further studies are needed to dissect the roles of OmpR-EnvZ and GrrS-GrrA in regulating T3SS.

Several regulatory genes have been previously reported to control amylovoran biosynthesis in E. amylovora, including the Rcs system, RcsA, Lon protease, and H-NS protein [17, 50, 65]. Here, we have further identified several groups of regulators, including both negative and positive regulators. These regulators may form a network that governs the production of amylovoran under different conditions to benefit pathogen survival or pathogenesis. Regulatory cascades are also likely to occur as global regulators such as OmpR-EnvZ and GrrS-GrrA may control expression of other regulatory genes or proteins, such as hrpXY and quorum sensing systems, as reflected in the amount of amylovoran produced in these mutants. However, we cannot rule out that cross-talk between different TCSTs may further complicate this scenario. Our study indicates that regulation of amylovoran biosynthesis is highly complex and further suggests that the pathogen has developed a system to control this major virulence factor.

Swarming is a flagella-driven form of motility for movement across solid surfaces as a group [66–70]. Swarmer cells are normally hyperflagellated and require extracellular components such as EPS and surfactants that enable mass migration [71, 72]. Previous studies have identified several global regulators in Es. coli and in other bacteria, including the Rcs system, OmpR and GrrSA, known to influence flagella biosynthesis, especially the master regulator flhDC [71, 73]. In this study, we have identified both negative (GrrSA) and positive (EnvZ/OmpR) regulators of swarming motility. It is easy to accept that the GacSA system may negatively regulate flagella biosynthesis, thus rendering grrSA mutants hypermotile. It is also obvious that flagella biosynthesis is not impaired in the envZ-ompR mutants as the swarming phenotype is different between the envZ-ompR mutants and the flhDC-fliA mutants. The obvious question that arises as to why envZ-ompR mutants are non-motile although they produce prolonged flagella. A recent study in Salmonella typhimurium has reported that mutations in chemotaxis pathways are impaired for swarming motility and further revealed a role of flagellum in sensing external wetness [72]. It has been proposed that swarming requires a fluid environment generated as bacteria extract water from the underlying agar gel [74]. Flagella are designed to work in this aqueous environment; that is swarming cells move in a thin layer of fluid over the surface of the agar. Further studies have revealed that the wetting agent that draws water out of the underlying agar is an osmotic agent [74]. It is likely that in our envZ-ompR mutants, this unknown osmotic agent provides a signal for EnvZ-OmpR system that may regulate chemotaxis response or creates high osmolarity so that water can be removed from the agar, thus affecting swarming motility. Further studies are needed to clarify this hypothesis.

In summary, we have identified and classified TCSTs, and have systematically generated TCST deletion mutants in E. amylovora. The mutants generated will serve as resources for future exploration of TCSTs in E. amylovora. In-depth characterization of deletion mutants and global gene expression will be our next goal. Our current data provide experimental evidence that TCSTs, especially those global regulators such as the Rcs system, EnvZ-OmpR and GrrA-GrrS, play important roles in virulence and in regulating virulence factors such as amylovoran production and swarming motility. These findings also suggest that multiple TCSTs may form complex and highly connected circuits and signaling networks in this important pathogen.