Role of Type III Effector Secretion during Bacterial Pathogenesis in Another Kingdom

Complex and sophisticated mechanisms of bacterial pathogenesis in plants, which have similarities to the mechanisms used by bacterial pathogens of mammals, such as Yersinia and Salmonella spp. (19, 76), have recently been identified. Mammalian and plant pathogens can use type III secretion systems (TTSSs) to colonize and parasitize susceptible hosts through the activity of translocated effectors. Because there are architectural and physiological limitations in plants that prevent an adaptive response, disease resistance in plants is usually associated with one or more forms of cellular immunity (more commonly referred to as induced resistance) that are somewhat analogous to the innate immunity observed in insects and mammals (38, 76). Cells in contact or close proximity to the pathogen usually respond defensively with an oxidative burst and a form of programmed cell death (PCD). All living cells of a plant appear to have the potential to recognize and respond locally to a pathogen. In some cases, the activation of these defense responses is dependent upon effectors that are translocated via the TTSS and is thus linked to the pathogenic activities of the pathogen.

In this paper, we will focus on the interactions of Pseudomonas syringae with its plant hosts and the activities of its TTSS. P. syringae is a collection of biochemically related strains that can have distinct host ranges. Recent studies have begun to establish biochemical activities for translocated effectors that have specific cellular targets in the host to aid in parasitism and to define the host range. Plants, in turn, express surveillance systems that appear to initiate defense responses by detecting alterations in these cellular targets. The goal here is to summarize recent advances in our understanding of the molecular events occurring during pathogenesis by P. syringae strains and how these events differ between susceptible and resistant plant hosts.

OVERVIEW OF P. SYRINGAE INTERACTIONS WITH PLANT HOSTS

P. syringae is a fluorescent pseudomonad in the γ-subgroup of proteobacteria that facultatively infects a wide range of economically important plant species. A typical symptom of P. syringae infection is an initial “water soaking” (darkening of the tissue) at the site of infection (indicative of altered membrane physiology) followed by slowly developing PCD, and in the case of exotoxin producers, a spreading chlorosis (yellowing of the tissue due to chlorophyll breakdown) (6, 56). The disease usually does not kill the plant but diminishes the yield and marketability of the product. Susceptible plant species can be found in divergent taxonomic groups of flowering plants. To put this statement in perspective, flowering plants arose in the fossil record at about the same time as the first mammals, suggesting that similar levels of diversity should be present for the two groups of organisms (although there are substantially more species of flowering plants). Most agriculturally important plant species are susceptible to at least one P. syringae strain. However, individual P. syringae strains usually have a very limited host range and only cause disease in a small subset of plant species. In fact, some strains can only infect a few varieties of a single plant species (37).

In many cases, asymptomatic epiphytic populations of the bacterium growing on exposed surfaces of plants provide the inoculum for infection (37, 56). A single bacterial cell is sufficient to initiate an infection. The invasion of the mesophyll tissue (loosely packed cells with air-filled intercellular spaces that are necessary for gas exchange) is usually aided by leaf wetting and/or tissue wounding and by bacterial motility. Once the bacterium is in the intercellular spaces of the tissue, it adsorbs to a host cell's surface. Unlike many mammalian pathogens, adsorption is likely to be passive, as normal physiological processes of the tissue remove water from intercellular spaces to restore gas exchange, thus bringing the bacterium in contact with a host cell. In addition, a specific adherence mechanism may not be necessary because fluid fluxes at the adhesion site are likely to be minimal. However, a candidate adhesin has been detected in some bacterial strains that are pathogenic to plants (68).

During the initial colonization of the tissue of a susceptible plant, the bacterium begins to multiply and produce virulence factors that contribute to symptom formation. Common virulence factors of P. syringae strains are extracellular polysaccharides (usually alginate, but levan can also be produced), derivatized peptide exotoxins, and several plant growth hormones, such as indoleacetic acid (associated with cell volume control, among other effects) and cytokinins (stimulate cell division but have other activities as well). The extracellular polysaccharide forms a hydroscopic glycocalyx that may protect the bacterium from oxidative stress and aids in tissue colonization (48). The exotoxins inhibit specific enzymes, such as glutamine synthetase or ornithine carbamoyl transferase, in surrounding host cells to suppress some defense responses, or they can directly alter the membrane physiology to favor parasitism (6, 31, 86). Over several days, bacterial populations in infected tissues can reach levels as high as 10⁹ cells/g of fresh weight of leaf tissue, but they are limited to the intercellular spaces. An important distinction from many mammalian pathogens is that P. syringae strains are noninvasive and remain external to the plant cell wall. This is a fundamental difference from many mammalian pathogens and impacts the nature of the interaction between the bacterium and its host and the strategy that plant pathogens use to parasitize the host.

The timing of the plant defense response to the initial colonization by the bacterium is a determining factor in the outcome of an interaction. In a susceptible plant, the host cells are usually slow to respond to the infection (39). As a result, the infecting bacterial population is able to spread into new tissues before the cellular defense responses of the initially colonized cells are activated. Thus, the pathogen is able to maintain a continuously expanding infection. Eventually, large areas of leaves and other organs or tissues become infected, and necrosis develops due to a slowly induced PCD (22, 84). In contrast, a resistant plant is able to rapidly initiate a cellular defense response during the initial colonization of the tissue that functions to prevent further spread of the infection. A single adsorbed bacterium elicits a cascade of cellular defense responses in the host cell of the resistant plant within 1 to 2 h. An oxidative burst is commonly observed by 3 to 6 h, with PCD typically developing within 12 h of infection. In a susceptible plant, similar responses are not detected until at least 48 to 96 h postinfection (19).

TYPE III SECRETION AND PATHOGENESIS

The ability to cause disease in susceptible plants and the induction of PCD in resistant plants have both been linked to the activities of a TTSS encoded by the central conserved region (CCR) of the P. syringae hrp pathogenicity island (PAI) (2, 26, 40). TTSSs are broadly conserved protein translocation systems that utilize a needle-like secreton and are found in bacterial pathogens of both plants and mammals (18). The hrp PAI CCR includes seven operons containing 26 genes that encode the TTSS and its dedicated regulatory system. Each of the nine universally conserved components of the TTSS secreton is present in the P. syringae system (8). Among TTSSs of mammalian pathogens, the Yersinia Ysc proteins are the closest homologs to the conserved Hrc products of the hrp PAI CCR, and 12 Ysc products share sequence similarity with counterparts of the P. syringae Hrp secreton (40).

The type III secretion process is similar to flagellar biogenesis. Secreted proteins (known as effectors) are loaded into a central channel of the TTSS secreton and exported from its distal end (18). The Hrp TTSS appears to be functionally interchangeable with its counterparts in other pathogenic bacteria. For example, the secretion of P. syringae-encoded effectors by Erwinia and Yersinia TTSSs has been reported (3, 34). In P. syringae, a pilus is assembled from the structural protein HrpA and extends from the secreton to the plant cell (46, 47, 54). This pilus can be more than 2 μm long and appears to be flexible. Immunolocalization was used to demonstrate that effectors are extruded from the tip of the HrpA pilus (46, 47, 54).

Like that of many TTSSs, the expression of the P. syringae TTSS is environmentally regulated during pathogenesis (40). A complex regulatory system that shares some similarities with that controlling flagellar biogenesis coordinates the assembly of the hrp-encoded TTSS. HrpR and HrpS, constitutively expressed from the hrpRS operon, are similar to the enhancer binding proteins that are typically part of two-component regulatory systems. HrpR and HrpS, however, lack the usual amino-terminal domain that modulates activity and do not require a cognate sensor kinase for their activities. HrpR and HrpS instead appear to form a heteromeric complex which then activates the RpoN-dependent hrpL promoter (41, 82). HrpL is an alternative sigma factor that is related to FliA, the alternative sigma factor involved in the expression of class III flagellar genes. HrpL directs the transcription of the operons encoding the components of the TTSS (82, 83). The regulated proteolysis of HrpR by the Lon protease provides one mechanism for the environmental regulation of hrpL expression (9) and therefore for the assembly of the secreton. Conditions that mimic, in part, the environment in plant tissue (amino acid starvation) suppress the Lon-mediated degradation of HrpR, thereby allowing the HrpR/HrpS complex to form. Another protein, HrpV, also negatively regulates hrp expression, albeit via an unknown mechanism (67). A GacS/GacA system has also been shown to modulate hrp expression, as is the case for several mammalian pathogens (16).

The primary function of the hrp-encoded TTSS appears to be the translocation of effectors into the plant cell cytosol to facilitate parasitism of the host (29, 66). Thus, the identification of the effectors expressed by a strain is critical for understanding the molecular interactions with the plant host. It has been relatively difficult, however, to identify the effectors produced by P. syringae strains. As in Salmonella and Yersinia, the secretion of P. syringae effectors through the hrp TTSS can occur without translocation into the host cell, but the level of secretion is several orders of magnitude lower than that in its mammalian counterparts. For example, the accumulation of Yop proteins in culture filtrates for Yersinia strains is obvious under inductive growth conditions (18), whereas only a few of the proteins secreted via the Hrp TTSS from just a couple of P. syringae strains can be easily detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis or immunoblotting under inductive conditions (80, 85). The proteins that can be detected, such as HrpW and HrpZ, do not appear to be translocated into plant cells, but instead may assist in early phases of the translocation process. HrpW has structural features that suggest that it may degrade components of the plant cell wall and could facilitate penetration by the HrpA pilus (15). HrpZ has been shown to form pores in lipid bilayers (52) and could aid in the breaching of the host cell plasma membrane by the Hrp TTSS.

Because physical methods for identifying effectors could not be easily applied, most of the initially identified effectors from P. syringae strains were discovered instead through phenotypic screens that attempted to isolate factors affecting the host range of a strain (51). At least 20 distinct effectors that required the hrp TTSS for phenotypic activity were identified by this process in a variety of P. syringae strains. Given that these phenotypic screens searched for genes that reduced virulence in an indicator plant (see below), an early name for effector genes was avirulence (avr) genes. Although mechanistically inaccurate, the avr name has been retained in some cases for historical reasons (25). More recently, identified effectors have been designated hop genes (encoding Hrp/Hrc outer proteins, analogous to yop genes) (80). Although their products are structurally divergent, all effector genes have been found to be transcriptionally dependent upon HrpL for expression.

Consistent with the conservation of the TTSS secreton, effectors of P. syringae strains share some characteristics with known effectors from mammalian pathogens. For instance, TTSS-secreted effectors are modular in nature, irrespective of the source. For the P. syringae effectors that have been analyzed for secretion signals, a cryptic secretion signal is located in the amino-terminal 50 amino acids (17, 30, 32, 65). The phenotypic and biochemical activities of effectors are usually associated with a carboxy-terminal domain in the proteins. Analogous to what has been observed for effectors of mammalian pathogens, some P. syringae effectors also have cognate chaperones, but most do not (18).

Analyses of the genomes of two P. syringae strains have revealed that individual strains encode a surprisingly large number of effectors. For example, by searches for HrpL-dependent promoters, together with similarity searches for homologs of known effectors, the genome of P. syringae DC3000 (a pathogen of tomato plants and Arabidopsis thaliana) was found to encode 58 known or likely effectors (17, 30). Similarly, the genome of the bean pathogen P. syringae B728a includes genes for 29 probable effectors, of which 25 are at least partially conserved in DC3000 (30). The genes for some of these effectors were located in effector loci associated with the hrp PAI (2, 14, 20), but the majority of the effectors were encoded by genes distributed elsewhere in the genome (11, 49). In DC3000, a cluster of nine candidate effector genes was found just downstream of the genes that encode the exotoxin coronatine (11), which is suggestive of a pathogenicity satellite. In other strains, clusters of effector genes have been identified on plasmids (26, 42), and several effector genes have been associated with transposable elements (49). Since most mammalian pathogens translocate considerably fewer effectors, the variety of effectors produced and apparently translocated into host cells by each P. syringae strain was unanticipated but could reflect the complexity of plant cells, which contain chloroplasts and other unique features (15, 30, 32, 36, 52) distinct from mammalian cells.

The variation in the host ranges of P. syringae strains may be due to differences in the effectors produced by the strains. Many effector genes appear to have been acquired via horizontal gene transfer. Plasmid-borne effector gene clusters and integron-like and transpositional redistribution mechanisms have been identified in P. syringae strains, as mentioned above (14, 43, 49). Once acquired, effectors continue to evolve through the accumulation of point mutations. For example, single point mutations were identified in alleles of an effector gene found in closely related P. syringae strains. These alleles affected which variety of the susceptible plant species (each of which carries a different set of R genes) responded to the expressing strain (77). Thus, the host range of a strain can be attributed, in part, to the set of effector genes that it expresses and to the activity of each individual effector.

VIRULENCE TARGETS OF SELECTED P. SYRINGAE EFFECTORS

The cellular activities of several translocated effectors from various P. syringae strains have been identified recently and were found to be similar to the activities of effectors produced by Yersinia and Salmonella spp. Like its counterparts, each P. syringae effector appears to have a specific cellular target in the host to facilitate parasitism either by direct manipulation of a physiological process in the host cell to aid in nutrient release or by the suppression of cellular defense systems. A common target of P. syringae effectors appears to be signal transduction pathways associated with the plasma membrane. In addition, some of these translocated effectors are processed in the host cell to become active.

Translocated cysteine proteases.

AvrPphB is an effector produced by some P. syringae strains that exhibits sequence similarity with the catalytic domains of YopT, Efa1, and other bacterial effectors that are members of this cysteine protease family (75). The cysteine protease effectors are essential virulence factors for mammalian pathogens. For example, the translocation of the Yersinia YopT effector into the cytosol of host cells causes the degradation of RhoA, a small membrane-bound GTPase found in the cytosol, and the disruption of the actin-based cytoskeleton (75). Efa1 is a related cysteine protease from enteropathogenic and enterohemorrhagic Escherichia coli that may also cause cytoskeletal rearrangements through a similar mechanism (62). Similar to these cysteine proteases, AvrPphB is required for virulence in susceptible hosts. After translocation into the host cell, AvrPphB is autocatalytically cleaved to expose an amino-terminal myristoylation motif that is necessary for its activity (Fig. 1A). Upon myristoylation, the mature protein localizes to the plasma membrane of the host cell, where AvrPphB has been shown to physically interact with and to specifically cleave PBS1, a Ser/Thr kinase (74, 78). The active-site Cys is required for this activity (74), consistent with its identification as a cysteine protease. Unfortunately, the normal cellular function of PBS1 is unknown, but it is presumably a component of a signal transduction system which may affect membrane function and it is essential to the defense response of resistant plants (see below).

FIG. 1.

(A) P. syringae effector interactions with susceptible host plants. P. syringae injects effectors into the cytosol of the host plant by using the Hrp TTSS. Once in the cytosol, the effectors target various host proteins to promote virulence. The cysteine proteases AvrPphB and AvrRpt2 are activated in the cytosol by autoproteolysis. AvrB, AvrRpm1, AvrPphB, and AvrPto are myristoylated and localize to the plasma membrane. AvrRpt2 also localizes to the plasma membrane. PBS1 and PTO autophosphorylate. AvrB and AvrRpm1 induce the hyperphosphorylation of RIN4. (B) P. syringae effector interactions with resistant plants. Resistant plants encode R proteins that monitor the cellular targets of the P. syringae effectors. Effector protein recognition initiates signal cascades that culminate in the PCD of the responding cell. P. syringae TTSS-secreted effectors are shown in orange; plant-encoded target proteins are shown in blue; plant proteins involved in signaling are shown in white; and R proteins are shown in yellow. Proteins thought to associate with the plasma membrane are represented by ovals, and cytosolic proteins are represented by rectangles. Zigzags represent protein myristoylation. To date, no single P. syringae strain has been identified that encodes or secretes all of the effectors mentioned above. Likewise, no single host plant is known to express all of the indicated R genes. See the text for descriptions of effector activities.

More recently, AvrRpt2 was cleverly identified as a cysteine protease. No local sequence alignments could be detected, but the predicted secondary structure of the active domain aligned with the secondary structure of staphopain, a cysteine protease from Staphylococcus epidermidis (4, 57). Like AvrPphB and the other related cysteine proteases, AvrRpt2 is autocatalytically processed (Fig. 1A). In the host plant, AvrRpt2 is cleaved after amino acid 71 to yield a stable C-terminal product (60). This posttranslocational processing requires at least one host factor (45). The cellular target of processed AvrRpt2 appears to be a host protein called RIN4, which is rapidly degraded in the presence of a catalytically active version of AvrRpt2. Inactivation of the predicted active-site Cys blocked the autocatalytic cleavage of AvrRpt2 and the AvrRpt2-mediated degradation of RIN4 (5, 57). Like that of PBS1, the biochemical function of RIN4 in the host cell under normal physiological conditions is not known, but it is required for the elicitation of defense responses in resistant plants, as described below.

Translocated PTPs.

HopPtoD2 is a chimeric effector whose amino-terminal 140 amino acids share sequence similarity with the corresponding region of AvrPphD, a broadly conserved translocated effector (10, 24). The carboxyl terminus contains a protein tyrosine phosphatase (PTP) active-site domain similar to those in TTSS-secreted effectors of Yersinia and Salmonella. In Yersinia, the PTP activity of YopH dephosphorylates proteins involved in cytoskeleton development, thus inhibiting phagocytosis by macrophages (21). SptP, a PTP encoded by Salmonella enterica serovar Typhimurium, also targets cytoskeletal components and inhibits the activation of a Salmonella-induced mitogen-activated protein kinase (MAPK) in host cells (21, 55). HopPtoD2 was shown to be an active PTP that is translocated into plant cells by the Hrp TTSS. A hopPtoD2 knockout mutant was substantially less virulent. Populations of the mutant grew at the same rate as the wild type during the initial 24 h of the interaction, but thereafter growth essentially stopped, consistent with the activation of a defense response. This suggested that HopPtoD2 might be a suppressor of one or more defense responses. Consistent with this hypothesis, the ectopic expression of a catalytically active version of HopPtoD2 in a P. syringae strain that lacked a hopPtoD2 allele delayed the onset of several defense responses, such as the reactive oxygen burst and PCD, in resistant plants. The transcription of a gene that is normally used as a marker for induced defense responses in surrounding cells was also suppressed (10).

Although a precise cellular target for this translocated PTP has not been identified, P. syringae HopPtoD2 likely functions similarly to SptP (Fig. 1A). There are several defense-associated MAPK cascades found in plants that appear to utilize tyrosine-phosphorylated proteins. The Arabidopsis genome includes more than 20 MAPK homologs that could be targets of HopPtoD2 PTP activity. A study by Espinosa et al. lent credence to this hypothesis by demonstrating that HopPtoD2 acts downstream of NtMEK2, a MAPK kinase that activates the MAPK's SIPK and WIPK, which are involved in the induction of defense responses (24). Whether these MAPKs are in fact the target of HopPtoD2 is currently under investigation.

Other effectors with known cellular targets: AvrPto and AvrPtoB.

Another effector, AvrPto, enhances symptom development and slightly increases the growth of populations of a P. syringae strain infecting susceptible tomato plants (13, 72). Upon translocation into host cells by the hrp TTSS, AvrPto is myristoylated or palmitoylated at the amino terminus to target AvrPto to the plasma membrane (Fig. 1A) (73). AvrPto was found to specifically interact with PTO, a Ser/Thr kinase similar to the IRAK-1 kinase of mammals (64). PTO, in turn, interacts with several PTO-interacting (PTI) proteins that seem to have a function in signaling (e.g., PTI1 is a kinase and PTI4, -5, and -6 are transcription factors) (64). These data suggest that PTO may function in a phosphorelay cascade which is necessary for a defense response.

A second effector, AvrPtoB, that also interacts with PTO was identified in the avrPto-expressing P. syringae strain by the use of PTO as bait in a yeast two-hybrid screen (Fig. 1A). AvrPtoB does not share any significant similarity with AvrPto and is much larger (50). Posttranslocational modifications of AvrPtoB have not been observed. However, AvrPtoB appears to promote virulence by acting as a general inhibitor of defense-associated PCD (1). AvrPtoB inhibited PCD induced by another P. syringae effector, a fungal elicitor of PCD, and even the mouse pro-apoptotic protein Bax. The ectopic expression of AvrPtoB inhibited the development of PCD in Saccharomyces cerevisiae as well. Recently, four other P. syringae-secreted effectors have been shown to suppress PCD in plants and yeast (44), indicating that PCD suppression may be a common pathogenic mechanism for P. syringae. The specific mechanism by which AvrPtoB and these other effectors inhibit PCD has not been established. Interestingly, AvrPto and AvrPtoB also enhance P. syringae virulence in plant species that do not carry a homolog of PTO. Therefore, it is likely that both proteins target at least one other plant protein.

Other P. syringae TTSS-secreted effectors.

For the majority of effectors that are translocated from P. syringae strains, the cellular fate, biochemical activity, and role in virulence have not been established. Some effectors have apparent chloroplast or mitochondrial localization signals, suggesting that their cellular targets are located in these organelles (30, 32). Like the above effectors, the majority seem to function in both pathogenicity in susceptible plants and the elicitation of defense responses in resistant plants, as described below. The insertional inactivation of most P. syringae effectors, however, only slightly affects virulence, suggesting that there is a redundancy in effector activities in host cells (26, 30).

HOST COMPONENTS AFFECTING EFFECTOR PHENOTYPES

Disease resistance in plants and the phenotype of an effector can be linked in many cases to the genotype of the host. Closely related plant varieties can differ in susceptibility to a particular pathogen and can differentially respond to a particular effector. For example, it is common for one variety of a plant species to be susceptible to a strain of P. syringae whereas another variety of the same plant species will be stably resistant to that same strain. A genetic analysis of these varieties usually reveals that the resistant variety carries at least one genetically dominant resistance (R) gene that confers the ability to respond defensively to that P. syringae strain. An analysis of the corresponding P. syringae strain reveals a phenotypically dominant effector gene (e.g., the aforementioned avr or hop genes) that is required for eliciting a defense response in the host variety expressing its cognate R gene (25).

Because of their role in disease resistance in plants, many R genes have been cloned and characterized (35). The most abundant and best studied class consists of cytoplasmic proteins composed of a central nucleotide binding (NB) motif and a leucine-rich repeat (LRR) at the carboxyl terminus. These are structurally similar to NOD1 and to portions of Apaf1, which are involved in caspase activation and the initiation of apoptosis (PCD) in humans (76). It is thought that the specificity of an R gene product resides in the LRR region in most, but not all, cases. The amino terminus of this group of R proteins contains either a Toll-like interleukin-1 receptor (TIR) or a coiled-coiled (CC) domain, which is thought to function in signal transduction. In the A. thaliana genome, there are approximately 150 members of the NB-LRR protein family that could be R genes (59). Other classes of potential R genes have also been identified. A few encode transmembrane domains, whereas others have extracellular LRR domains, as expected for a typical membrane-associated receptor (35).

PERCEPTION OF EFFECTORS BY RESISTANT HOSTS

For many bacterial pathogens, the activation of the defensive PCD response in resistant plants requires a functional TTSS and is linked to specific R genes. Many of the initially identified effectors were isolated due to their ability to elicit R-gene-dependent responses. Most early models had predicted that the translocated effectors would be ligands for receptors encoded by R genes (39). Direct interactions, however, could not be detected in many cases. Further analyses suggested that the recognition of an effector by an R gene product might be indirect. Instead of detecting the presence of the effector itself, the R gene product monitors the status of the particular cellular target of the effector (19, 38).

For example, a direct interaction of the cysteine protease AvrRpt2 with its cognate R gene product, RPS2, could not be detected by several sensitive methods. Instead, the effect of AvrRpt2 activity on its cellular target, RIN4, appeared to be the eliciting signal. RIN4 has been shown to physically interact with RPS2 at the plasma membrane (5, 53, 57) and thus could be an intermediate in the recognition process (Fig. 1B). The initiation of RPS2-dependent resistance was correlated with the proteolysis of RIN4 by AvrRpt2. Interestingly, RNA interference-mediated depletion of RIN4 leads to a constitutive PCD phenotype in RPS2-expressing plants, and the overexpression of RIN4 suppresses the AvrRpt2-mediated elicitation of PCD in RPS2-expressing plants (57). Thus, it appears that the AvrRpt2-directed elimination of RIN4 signals RPS2-mediated resistance. In this way, RPS2 senses RIN4 levels in cells and “guards” whatever role RIN4 plays in the host cell.

A slightly different process appears to occur during the induction of a defense response in plants expressing RPM1. RPM1 confers resistance to P. syringae strains expressing either AvrRpm1 or AvrB (7). Although the biochemical activities of these secreted effectors are not known, the mechanism by which they induce PCD in host cells is better understood. RPM1 is a peripheral plasma membrane protein that, like RPS2, interacts with RIN4 (58). In contrast to AvrRpt2, however, AvrRpm1 and AvrB appear to induce the phosphorylation of RIN4 via an unidentified kinase (Fig. 1B). This phosphorylation likely alters the activity of RIN4 to allow AvrRpm1 and AvrB to act as virulence factors in susceptible plants that lack the RPM1 allele (58). In resistant plants, however, RPM1 appears to detect or respond to the phosphorylated form of RIN4 to induce PCD. In this case, the R gene product appears to be sensing an effector-dependent modification of the cellular target of the effector.

In an analogous manner, the R gene product RPS5 appears to monitor PBS1, a Ser/Thr kinase that is the cellular target for the cysteine protease AvrPphB. The proteolysis of PBS1 by AvrPphB is required for the RPS5-mediated induction of defense responses (74, 81), but in this case, only partial cleavage of the cellular target is required to elicit a response (Fig. 1B). After cleavage by AvrPphB, the residual PBS1 derivative retains its kinase activity and is autophosphorylated. Inactivation of the kinase activity of the PBS1 derivative blocks the RPS5-dependent induction of a defense response (74). This suggests that a phosphorylated derivative or another phosphorylated substrate for PBS1 might interact with RPS5 to elicit the defense response.

Parallels also exist between the mechanism by which resistant plants detect the presence of AvrPto and AvrPtoB and that for the recognition of AvrPphB. The induction of PCD in resistant plants by AvrPto and AvrPtoB is dependent upon two host factors, namely PTO, the Ser/Thr kinase mentioned previously as the apparent cellular target of AvrPto, and PRF, a comparatively large (∼210 kDa) CC-NB-LRR protein (70). PRF is highly conserved in a wide variety of plant species, and its structural properties are similar to those of the largest class of R gene products. Plants that have a deletion in PRF are more susceptible to P. syringae strains that express AvrPto and/or AvrPtoB, suggesting that PRF may be the R gene product that responds to AvrPto and AvrPtoB (64). The kinase activity of PTO is required for AvrPto- and AvrPtoB-induced PCD as well as for signaling via the PTI proteins (64). This suggests that the kinase activity of PTO generates a phosphorylated intermediate that could be the actual ligand for PRF (Fig. 1B).

Irrespective of their mechanism of activation, NB-LRR proteins transduce signals through at least two different pathways to elicit a defense response. Mutations in either pathway render a plant susceptible to disease. CC-NB-LRR proteins appear to signal through one pathway, whereas TIR-NB-LRR proteins activate a separate pathway (35). The components of these pathways appear to be unique to plants. However, signals from both pathways converge on three proteins that are well characterized in mammalian cells. Homologues of RAR1 and SGT1 are implicated in cell cycle control and development in other eukaryotes. In plants, the corresponding homologs interact with each other and are required for disease resistance mediated by many, but not all, CC-NB-LRR and TIR-NB-LRR R proteins (35, 38). SGT1 is part of a ubiquitination complex in yeast that targets proteins for degradation. There is evidence that ubiquitination could be an important component of defense signaling in plants (38). A third protein, the chaperone HSP90, also interacts with RAR1 and SGT1 and is required for R-mediated defense responses (79). As a result, it appears that the RAR1/SGT1/HSP90 complex is a convergence point for signal transduction mediated through both classes of NB-LRR proteins. The precise mechanism by which RAR1, SGT1, and HSP90 interact to mediate signaling has not been established. Many of the defense responses have also been linked to MAPK cascades, as mentioned above (63).

The effector-mediated activation of any of these pathways leads to a wide range of responses in the adjacent cells of the plant. Changes in membrane permeability, an increased production of reactive oxygen species (ROS) and nitric oxide (NO), the activation of protein kinases, and transcriptional reprogramming are just some of the responses induced by pathogens in resistant plants (19, 35, 63). One of the earliest plant responses to an infection is a change in membrane permeability that leads to an influx of calcium and an efflux of potassium (71). Calcium then initiates a series of additional responses, possibly via calcium-dependent protein kinases. Both ROS production and PCD are dependent upon an increase in cytosolic calcium (63). An oxidative burst is typically observed 3 to 6 h after infection and has been attributed to an increase in the activity of a calcium-dependent NADPH oxidase. ROS and NO, in turn, stimulate salicylic acid (SA) production. The ROS, NO, and SA can act as secondary messengers to induce neighboring cells to modify their cell walls, produce antimicrobial products, and activate defense gene expression (19).

CONCLUDING REMARKS

What emerges from these studies is that the interactions of P. syringae strains with their plant hosts are highly complex, involving multiple pathogenicity and virulence factors to enable the successful colonization of the host. P. syringae strains express a variety of effectors that are translocated into the cytosol of host cells by their TTSS and are required for pathogenicity in the susceptible host. While most mammalian pathogens do not appear to secrete such a plethora of effectors, this appears to be a common feature of many plant pathogenic bacteria, such as Erwinia (26), Xanthomonas (12), and Ralstonia spp. (69). This diversity of effectors could be due to the complexity of the host cell itself and to the diversity of hosts with which a strain interacts.

While some P. syringae effectors are at least partially conserved among many strains, many are only found in a few strains. For example, of the effectors with known or predicted biochemical activities, orthologs of the cysteine protease AvrPphB are broadly conserved, whereas the PTP HopPtoD2 appears to have a limited distribution among P. syringae strains. This is consistent with the observation that each strain appears to be specialized to parasitize a specific plant host. Each effector is thought to have a specific cellular target to aid in the parasitism of a susceptible host. Because different hosts can be from divergent taxonomic families of plants, distinct effectors are likely to be necessary to facilitate parasitism. In addition, the deletion of individual effectors usually has little effect on virulence, yet the loss of the TTSS invariably abolishes pathogenicity. Therefore, P. syringae strains appear to express functionally redundant effectors. As an example, one P. syringae strain was found to translocate two effectors that interacted with PTO.

The plant hosts, in turn, are not passive recipients of these pathogens and have evolved complicated defense mechanisms that respond to effectors. Plants apparently express a large array of R genes in all living cells as part of a surveillance system to detect the presence of pathogens (be they viral, bacterial, fungal, protist, or animal). For those that respond to translocated bacterial effectors, the sensors do not appear to be receptors for effectors but rather seem to monitor the cellular targets of the P. syringae effectors. The loss or modification of the cellular target is the eliciting signal for the defense response in these cases. This implies that the cellular target in the host and the host's sensor likely coevolved, rather than the effectors and sensors, which could potentially evolve at very different rates. In addition, more than one sensor may monitor a single cellular target to detect different types of modifications. For example, RPS2 monitors the proteolytic degradation of RIN4, whereas RPM1 seems to detect the phosphorylation of RIN4.

It is also important to recognize that translocated effectors are not the only bacterial signals to which plant cells respond. In addition to translocated effectors, flagellin can elicit defense responses, and its cognate host cell surface receptor has been identified (Fig. 1B) (28). Acyl lactone elicitors of defense responses (syringolides) and lipopolysaccharides appear to elicit distinct sets of defense responses in some plants (23, 27, 33). Thus, plants have evolved multifaceted “immune” systems to prevent disease that must be evaded or suppressed for disease to occur.

Are host-pathogen interactions in plants and mammals examples of conservation or convergence? It appears that both processes probably occurred. TTSSs are well conserved among plant and mammalian pathogens, are mechanistically similar, and are important for pathogenesis. Nguyen et al. reported that all TTSSs likely have a common progenitor (61), and since many mammals consume plants there are many possibilities for genetic exchange between pathogens of these distinct hosts. Thus, studies of the TTSSs of plant pathogenic bacteria should be applicable to those of mammalian pathogens and vice versa. In addition to the translocation machinery, several translocated effectors are also at least partially conserved among pathogens. In one case, the cysteine proteases AvrPphB and YopT are likely orthologs of each other and thus represent conserved genes (75). In the case of the PTPs and the cysteine protease AvrRpt2, convergence appears to have occurred. Although P. syringae HopPtoD2 and Yersinia YopH are both PTPs, only the active site is conserved and there is little similarity in the other regions of these effectors (10, 24). AvrRpt2 is structurally similar to other cysteine proteases, but again, there is little primary sequence conservation (4). The cellular targets of these effectors, however, are very different. Whereas many mammalian pathogens alter the host cell morphology to either suppress or induce phagocytosis, for example, altered cell morphologies are rarely seen in plant hosts and are much more subtle when they are observed. Instead, the effectors from P. syringae strains seem to affect one or more signal transduction pathways to enhance nutrient release from cells or to suppress the induction of defense responses. This could explain why many of the effectors, particularly those that do not obviously suppress defense responses, associate with the plasma membrane. These signal transduction pathways and their roles in defense responses are just beginning to be elucidated for higher order plants.

On the host side, PCD and ROS production occur in both mammalian and plant hosts, and there are some similarities in the biosynthetic pathways for ROSs and signaling during the initiation of PCD (35, 63). Superficially, several components of signal transduction cascades leading up to the above responses are conserved, but the complete pathways are not. The overall mechanism by which a resistant plant host recognizes a pathogen appears to be very different from what occurs in a mammalian host. The absence of an adaptive response in plants is, of course, a stark difference. Until the function of each effector produced by a pathogen and the mechanisms for the induction of defense responses are fully established, we are left with intriguing similarities and open-ended questions. Understanding the interactions of bacteria with plant hosts, however, may offer important clues to basic features of innate immunity of eukaryotes in general.