INTRODUCTION

Top Previous Next References

Pseudomonas spp. produce a wide spectrum of phytotoxic compounds (Table 1). Among the most well-characterized bacterial phytotoxins are those produced by the plant pathogen Pseudomonas syringae. This review summarizes our current understanding of the mechanism of action, biosynthesis, and regulation of four distinct classes of phytotoxins, including the lipodepsipeptides (syringomycins, syringopeptins), coronatines, phaseolotoxin, and tabtoxin.

P. syringae is reported to induce a wide variety of symptoms on plants, including blights (rapid death of tissue), leaf spots, and galls. The species is divided into pathogenic variants (pathovars), which vary in host range. Two distinct reactions are possible when P. syringae cells are infiltrated into plant tissue. One potential outcome is a compatible, susceptible interaction which is characterized by a symptom called water soaking, a reaction which is followed by pathogen proliferation and advanced symptom development. In contrast, resistant host cells undergo a reaction known as the hypersensitive response and become necrotic 12 to 24 h after inoculation. A cluster of genes termed the hrp region (for "hypersensitive response and pathogenicity") is conserved in phytopathogenic prokaryotes and affects the ability of a bacterium to induce a hypersensitive response in nonhost plants, pathogenicity in host plants, and the ability to grow within or on the surface of plants (83). It is important to note that the hrp gene cluster is required for pathogenicity of P. syringae on plant hosts. The hrp genes are known to encode genes for the regulation and biosynthesis of a type III secretion pathway that is similar in both plant and animal pathogens and is used to secrete virulence proteins (228). It is becoming increasingly evident that mechanisms which function in clinical pathogens of animals, such as the type III secretion systems in Salmonella, Shigella, and Yersinia, are similar to those in phytopathogenic species (206, 221). However, in addition to the hrp genes, phytopathogenic pseudomonads encode gene products that significantly enhance pathogen virulence, including extracellular polysaccharides, phytotoxins, cell wall-degrading enzymes, and phytohormones (3, 49, 56, 92).

Although phytotoxins are not required for pathogenicity in P. syringae, they generally function as virulence factors for this pathogen, and their production results in increased disease severity. For example, P. syringae phytotoxins can contribute to systemic movement of bacteria in planta (198), lesion size (24, 287), and multiplication of the pathogen in the host (24, 75, 172). The phytotoxins produced by P. syringae can substantially enhance the virulence of producing pathogens, even though some disease can occur in their absence.

The toxins produced by P. syringae are varied in origin and include monocyclic -lactam (tabtoxin), sulfodiaminophosphinyl peptide (phaseolotoxin), lipodepsinonapeptide (syringomycin), and polyketide (coronatine) structures (163). Knowledge of phytotoxin structure is extremely important since structural information may provide important clues about the biosynthetic processes involved. Fortunately, several P. syringae phytotoxins have structural analogies to antibiotics that are produced via nonribosomal mechanisms in Streptomyces and Bacillus spp. These pathways have served as predictive models for the synthesis of selected phytotoxins.

It has been difficult to obtain information on intermediates in the biosynthetic pathways to various phytotoxins. One reason for the lack of characterized intermediates in these diverse pathways is that nonribosomal synthesis is generally catalyzed by multifunctional proteins or polypeptide complexes and intermediates are transferred between enzymatic domains and not released into the cytoplasm. Furthermore, conversion of intermediates to the final product may occur very rapidly and impede detection and characterization.

The biosynthesis of nonribosomal peptides has been intensively investigated for a number of years, and there are several excellent reviews on this subject (126, 147, 247, 278). According to the current model, these peptides are synthesized via a thiotemplate mechanism by large multisubunit enzymes ranging from 100 to 1,600 kDa (247). All thiotemplate multienzymatic systems are composed of amino acid-activating domains that catalyze the adenylation of the constituent amino acids and the formation of thioesters. The general sequence of reactions includes (i) carboxyl activation of the substrate amino acid by adenylate formation, (ii) acylation of enzyme-attached pantothenoyl-thiols, and (iii) directed transfer to the next acyl intermediate with condensation (Fig. 1A). The completed peptide is released from the enzyme complex by cyclization, amidation, or hydrolysis (126).

View larger version (30K): [in this window] [in a new window]   FIG. 1.   (A) Reaction sequence catalyzed by multifunctional peptide synthetases. (a) Carboxyl activation of the first amino acid (A1) and formation of the aminoacyl adenylate; (b) activation of the second amino acid (A2) and formation of the aminoacyl adenylate; and (c) the condensation reaction. Abbreviations: Enz, enzyme; A, amino acid, M, divalent metal ion (Mg2+, Mn2+, Ca2+). Numbering indicates specific domains within an individual multienzyme; for example, Enz1 and Enz2 are two distinct domains within the same enzyme. Amino acids (A1 and A2) are two individual amino acids. For additional information, see reference 278. (B) Domain structure of the amino acid-activating module SyrE1. The SyrE1 module contains approximately 1,500 amino acids organized into four domains as defined by Stein and Vater (252). The relative positions of conserved core sequences are shown for each domain. The two elongation domains contain a characteristic HHxxxDG motif (E) (54). The five core sequences described by Stachelhaus and Marahiel (247) are located within the amino acid-adenylating domain. Core 2 has a sequence (SGTTGxPKGV) resembling the Walker type A motif involved in ATP binding, and the core 4 sequence (TGD) carries a motif associated with ATPase activity. A role in catalyzing aminoacyl adenylate formation is suggested for cores 3 and 5 (247). A region between cores 2 and 3 is associated with substrate recognition (S) (47), and SyrE1 exhibits substrate specificity for L-Ser (95). Core 6, with a characteristic LGGHSL motif, is located in the 4'-phosphopantetheine carrier domain. The motif contains a conserved serine to which the 4'-phosphopantetheine cofactor is covalently attached, which in turn is the site of thioester formation. The next module, SyrE2, carrying three domains (i.e., amino acid adenylating, 4'-phosphopantetheine carrier, and elongation domains), follows the SyrE1 module.

The isolation, sequencing, and characterization of genes encoding multifunctional peptide synthetases has indicated a multidomain arrangement in which an adenylation domain consisting of 600 amino acid residues is highly conserved and repeated. Biochemical studies with specific domains have confirmed the multidomain structure predicted by sequence data (147). An elongation domain of about 500 amino acids separates successive adenylation domains (54). The organization of the multifunctional peptide synthetase is colinear to the amino acid sequence of the corresponding peptide product (247). Turgay et al. (266) described a superfamily of adenylate-forming enzymes which includes all peptide synthetases and several adenylating enzymes. A typical adenylation-thiolation module of a peptide synthetase contains a series of conserved sequences with the same order and spacing (252). Five regions originally designated as core motifs (266) are conserved in the adenylation domain of peptide synthetases (Fig. 1B). Although the function of core 1 remains unknown, cores 2 to 5 are presumed to be involved in ATP binding and hydrolysis (Fig. 1B). Core 2 has a glycine-rich sequence that contains a potential phosphate-binding loop, whereas core 4 shows relatedness to ATPases. Conti et al. (47) identified a substrate-binding pocket between cores 2 and 3 of a gramicidin S synthetase module by analysis of the crystal structure of the adenylation domain. Substrate specificity in various peptide synthetases is presumably mediated by the amino acids lining the substrate-binding pocket. Core 6, which is located in the 4-phosphopantetheine carrier domain adjacent to the adenylation module (Fig. 1B), is associated with covalent binding of the substrate amino acid and contains a serine residue to which the cofactor 4'-phosphopantetheine is covalently attached (147).

The involvement of peptide synthetases and adenylate-forming enzymes in the biosynthesis of coronatine and syringomycin has been established and is discussed in further detail below.

Polyketides constitute a huge family of structurally diverse natural products including those with antibiotic, chemotherapeutic, and antiparasitic activities. Most of the research on polyketide synthesis in bacteria has focused on compounds synthesized by Streptomyces or other actinomycetes, and several excellent reviews have been recently published (99, 116, 123). However, in addition to coronatine, it is important to note that Pseudomonas produces a variety of antimicrobial compounds from the polyketide pathway, including mupirocin (pseudomonic acid) (73), pyoluteorin (52), and 2,4-diacetylphloroglucinol (14).

Polyketide synthesis is related to fatty acid biosynthesis; the latter begins with the condensation of acetyl coenzyme A (acetyl-CoA) (starter unit) and malonyl-CoA (chain extender unit) and then proceeds with a cycle of modifications on the carbonyl group of malonyl-CoA (i.e., reduction to a hydroxyl, dehydration to produce a double bond, and further reduction to form a saturated fatty acid) (Fig. 2). Unlike fatty acid synthases, a polyketide synthase (PKS) can accept additional substrates in the starter and extender groups and the products vary in the extent of reduction. PKSs are generally classified as type I or II systems and consist of protein complexes that act on covalently bound substrates that are attached as thioesters to an acyl carrier protein (ACP) (117). Polyketides synthesized by type I PKSs usually have a fairly reduced structure and are synthesized by large multifunctional proteins that consist of individual domains which catalyze specific and discrete reactions in a nonreiterative fashion (Fig. 3) (102). The functional activities catalyzed by domains within the type I PKS are often apparent in the structure of the growing polyketide chain (Fig. 3C), and nucleotide sequencing has become an important tool in predicting the biosynthetic route to polyketides synthesized by a type I PKS. Conversely, type II PKSs are most often associated with synthesis of aromatic polyketides, and biosynthesis occurs on monofunctional proteins that associate in a complex. Unlike the type I system, the type II PKS may utilize one or more enzymes in a reiterative fashion.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Reaction sequence in the synthesis of fatty acids. The starting units for the fatty acid synthase are acetyl-CoA and malonyl-CoA; these are converted into acetyl-ACP and malonyl-ACP by acetyl and malonyl transacylase, respectively. The fatty acid synthase proceeds with the condensation of these two precursors and then continues with a cycle of reduction, dehydration, and further reduction of the keto group (asterisk). Two key differences between polyketide synthases (PKS) and fatty acid synthase include the choice of starter units and the extent of reduction.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   Biosynthesis of erythromycin A. (A) The nucleotide sequence of the eryA region contains three ORFs designated eryAI, eryAII, and eryAIII (29, 60). (B) These genes encode three proteins which constitute erythromycin B synthetase (DEBS); these are designated DEBS 1, DEBS 2, and DEBS 3 (39). (C) Each DEBS protein contains two modules, each with domains for acetyltransferase (AT), ACP, and -keto synthase (KS) activity. Some modules contain additional domains for dehydratase (DH), enoyl reductase (ER), ketoreductase (KR), and TE activity. (D) Cyclization and release of the DEBS 3 product results in the formation of 6-deoxyerythronolide B (DEB), and additional tailoring steps result in the production of erythromycin A (E). Modified from reference 99.

The structure of coronatine (COR) (Fig. 4a) is unusual and has two distinct components: the polyketide coronafacic acid (CFA) (Fig. 4f) and coronamic acid (CMA), an ethylcyclopropyl amino acid derived from isoleucine (108, 160, 196). COR is generally the predominant coronafacoyl compound synthesized by COR producers and also the most toxic; however, other coronafacoyl compounds which contain various amino acid substituents conjugated to CFA via an amide linkage may be synthesized (Fig. 4b to e) (161, 165, 166, 169).

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Structures of COR and coronafacoyl compounds.

Precursor feeding studies with ¹³C-labeled substrates demonstrated that CFA is a novel polyketide synthesized from one unit of pyruvate, one unit of butyrate, and three acetate residues (196) (Fig. 5). Recent studies have suggested that the pyruvate used for CFA biosynthesis is converted into -ketoglutarate before incorporation to CFA and that -ketoglutarate may serve as the starter unit for CFA assembly (194). Little information is available about potential intermediates in the biosynthetic route to CFA, probably because such intermediates remain enzyme bound. However, Mitchell et al. (171) have identified a cyclopentenone compound, 2-(1-oxo-2-cyclopenten-2-ylmethyl)butanoic acid, which may function as an intermediate or shunt product of the CFA biosynthetic pathway.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Biochemical pathways involved in the synthesis of COR and coronafacoyl compounds in P. syringae pv. glycinea PG4180. COR consists of a polyketide component, CFA, coupled (CPL) via amide-bond formation to an amino acid component, CMA. CFA is synthesized as a branched polyketide from three acetate units, one pyruvate unit, and one butyrate unit via an unknown sequence of events (196). CMA is derived from isoleucine via alloisoleucine and cyclized by an unknown mechanism (160, 195). CMA functions as an intermediate in the COR biosynthetic pathway, which indicates that cyclization of L-alloisoleucine to form CMA occurs before CFA and CMA are coupled (170). The coronafacoyl analogues, CFA-Ile and CFA-aIle result from amide bond formation between CFA and isoleucine and alloisoleucine, respectively, and are not utilized further in the synthesis of COR.

Parry et al. (195) provided an important clue about the route to CMA by demonstrating that L-alloisoleucine was a more immediate precursor to CMA than was isoleucine. Initially, two possible pathways to COR from CFA were proposed; one route involved the direct coupling of isoleucine (or alloisoleucine) to CFA to form the coronafacoyl conjugates coronafacoylisoleucine (CFA-Ile), and coronafacoylalloisoleucine (CFA-aIle), followed by an oxidative cyclization on the amino acid moiety of the conjugate to form COR (289). This scheme was proposed based on the natural occurrence of CFA-Ile and CFA-aIle in a variety of COR producers (169). An alternative route involved the isomerization of isoleucine to form alloisoleucine, cyclization of alloisoleucine to form CMA, and conjugation of CFA and CMA via amide bond formation (Fig. 5). Support for the latter route developed from the demonstration of CMA as a defined intermediate in the COR pathway (170). The biosynthetic block to COR in several mutants was eliminated when CMA was exogenously supplied, and other mutants were found to excrete CMA when CFA synthesis was blocked (22, 170). Furthermore, CFA-negative mutants could produce COR when supplied with exogenous CFA but not with CFA-Ile or CFA-aIle, indicating that the latter compounds were not operative in coronatine synthesis (170). Our current understanding of the COR biosynthetic pathway is summarized in Fig. 5.

The final step in the pathway to COR is presumed to be the ligation or coupling of CFA and CMA by an amide linkage. The enzyme(s) catalyzing this reaction is thought to lack rigid specificity for the amino acid substrate since a variety of coronafacoyl-amino acid conjugates have been isolated, including CFA-Ile, CFA-aIle, coronafacoylvaline, norcoronatine, and CFA conjugated to serine and threonine (161, 165, 166, 169).

The primary symptom elicited by COR is a diffuse chlorosis that can be induced on a wide variety of plant species (82). Interestingly, the reaction of Arabidopsis thaliana to exogenously applied COR is atypical; instead of chlorosis, anthocyanins accumulate at the site of inoculation and the tissue develops a strong purple hue (27). COR is also known to induce hypertrophy, inhibit root elongation, and stimulate ethylene production (74, 122, 226, 277). Several research groups have noted the remarkable structural and functional homologies between COR and methyl jasmonate (MeJA), a plant growth regulator derived from the octadecanoid signaling pathway which is elicited by biological stress (241, 280). COR and MeJA induce analogous biological responses in Arabidopsis seedlings, Eschscholtzia californica cell cultures, and potato tissue; these results have led researchers to suggest that COR functions as a molecular mimic of the octadecanoid signaling molecules produced by higher plants (75, 85, 129, 276, 283). Furthermore, Feys et al. (75) generated a coronatine-insensitive (coi1) mutant of Arabidopsis that was insensitive to the effects of both COR and MeJA, suggesting a similar mode of action.

Light microscopy was used to compare the effects of COR, CFA, and MeJA on tomato tissue (190). Several changes were induced in tomato tissue exposed to the phytotoxin; for example, the epidermal wall was significantly thicker in COR-treated tissue and the chloroplasts stained more intensively and were smaller (190). One of the most pronounced differences was the appearance of spherical and cubical proteinaceous structures in the vacuole of COR-treated tomato tissue. These structures were markedly similar to the proteinase inhibitors which had been previously found in plant tissues exposed to various biological stresses (2, 243, 244). The presence of proteinase inhibitors in the COR-treated tissue was confirmed by demonstrating that this tissue significantly inhibited the activity of both chymopapain and chymotrypsin (190). Recently, polyclonal antibodies to both chymopapain (34) and chymotrypsin (225) inhibitors were used to confirm the identity of the proteinaceous structures in COR-treated tomato tissue. When COR-treated tissue was incubated with antisera to chymopapain inhibitor and then with a secondary antibody conjugated to gold, the cubical crystals were densely labeled with gold particles, indicating that these structures were chymopapain inhibitor (188). A similar experiment indicated that the spherical crystals were chymotrypsin inhibitor (188); thus, we concluded that both chymopapain and chymotrypsin inhibitors are specifically induced in response to COR in tomato tissue.

Although COR, CFA, and MeJA induced the production of proteinase inhibitors, only COR caused cell wall thickening, changes in chloroplast structure, and chlorosis; CFA and MeJA did not induce these changes in tomato tissue (190). Consequently, the CMA moiety, or perhaps the amide linkage between CFA and CMA, may impart additional biological activities to COR in tomato. Further differentiation of COR and MeJA was demonstrated by Krumm et al. (131), who showed that jasmonic acid and COR induce the production of distinctly different volatile compounds in Phaseolus lunatus. Therefore, COR does not function solely as a molecular mimic of MeJA in some plant species, and the mechanism of action of COR may remain unclear until putative receptors for the toxin are localized in various plant species.

Production of the phytotoxin COR has been demonstrated in five pathovars of P. syringae, i.e., pv. atropurpurea, glycinea, maculicola, morsprunorum, and tomato, which infect ryegrass, soybean, crucifers, Prunus spp., and tomato, respectively (159, 168, 282). Although production of COR outside the species P. syringae is thought to be rare, Xanthomonas campestris pv. phormiicola, a pathogen of New Zealand flax, also produces several coronafacoyl compounds (162, 261).

Tn5 mutagenesis has been used to obtain COR-defective (COR) mutants of P. syringae pv. atropurpurea, glycinea, morsprunorum, and tomato (23, 25, 176, 289). In several studies, COR was shown to play a distinct role in virulence (24, 172, 229); however, it is important to note that strains of P. syringae pv. glycinea, maculicola, morsprunorum, and tomato that do not produce COR have been isolated (159, 168, 271). Several reports have shown that the COR biosynthetic cluster occurs on indigenous plasmids (23, 25, 138, 229, 297); consequently, the potential instability of plasmid-located COR genes might explain the variability in COR production among strains of P. syringae (51, 271). Although the COR gene cluster has been frequently associated with large (80- to 110-kb) plasmids, these genes can also be chromosomal (51).

COR biosynthesis has been intensively studied in P. syringae pv. glycinea PG4180 because this strain is easy to manipulate genetically, consistently synthesizes large amounts of COR in vitro (20 to 40 mg/liter), and infects soybean, a host which is easy to cultivate (26). Transposon mutagenesis indicated that the COR biosynthesis genes in P. syringae pv. glycinea PG4180 are located on a 90-kb plasmid designated p4180A (25). The involvement of p4180A in COR production was demonstrated by transforming this plasmid into two nonproducers of COR, P. syringae pv. syringae PS51 and PS61 (22). Organic acids were then extracted from PS51 and PS61 transformants containing p4180A and analyzed by high-pressure liquid chromatography (HPLC) and combined gas chromatography-mass spectrometry. PS51 and PS61 transformants containing p4180A produced both CFA and COR, indicating that p4180A encodes all genes necessary for the biosynthesis of coronafacoyl compounds in P. syringae (22).

A variety of approaches have been used to characterize the COR biosynthetic cluster encoded by p4180A: (i) saturation Tn5 mutagenesis, (ii) feeding studies using exogenously supplied CFA and CMA, (iii) complementation of selected mutants with cloned DNA, (iv) expression of selected regions of the COR gene cluster in COR nonproducers, and (v) nucleotide sequence analysis (22, 143, 203, 213, 270, 272, 273, 289). Saturation Tn5 mutagenesis indicated that a 32-kb contiguous region was absolutely required for COR biosynthesis, and a physical map was developed by using the restriction enzymes BamHI and SstI (Fig. 6C) (22, 272, 289). Two regions in the COR biosynthetic cluster contained structural genes for CMA and CFA biosynthesis; these were separated by a 3.4-kb regulatory region (REG; Fig. 6B). Transcripts in the COR biosynthetic gene cluster were identified by a combination of the following approaches: (i) complementation of selected mutants with subcloned DNA from the COR biosynthetic cluster, (ii) expression of functional regions of the COR gene cluster in COR nonproducers, (iii) nucleotide sequence and primer extension analyses, and (iv) transcriptional fusions to a promoterless glucuronidase gene (142, 213, 270, 272, 273). These approaches indicated that the COR gene cluster in PG4180 consists of six transcripts (Fig. 6A).

View larger version (15K): [in this window] [in a new window]   FIG. 6.   Functional and physical map of the COR biosynthetic gene cluster. (A) Horizontal lines with arrowheads indicate the transcriptional organization of the COR gene cluster. (B) Functional regions of the COR biosynthetic cluster: CMA, CMA biosynthetic gene cluster; REG, regulatory region; CFA, CFA biosynthetic gene cluster. (C) Physical map of the COR gene cluster; the enzymes used for restriction mapping were BamHI and SstI. (D) Expanded view of SstI fragments 4 to 8, which contain the CFA biosynthetic gene cluster. Abbreviations: 1, cfa1; 2, cfa2; 3, cfa3; 4, cfa4; 9, cfa9.

The nucleotide sequence of the 6.9-kb region containing the CMA biosynthetic gene cluster revealed the presence of four genes designated cmaA, cmaB, cmaT, and cmaU (37, 197, 270) (Fig. 6A). The CMA biosynthetic gene cluster was shown to encode two transcripts; one transcript was monocistronic and contained cmaU, and the second was polycistronic and contained three cotranscribed genes designated cmaA, cmaB, and cmaT (Fig. 6A). Start sites for both transcripts were determined by primer extension (270). The deduced amino acid sequence of cmaA indicated that the enzyme contains an amino acid-activating domain, whereas cmaB showed extensive homology to syrB2, a gene encoding an enzyme required for syringomycin synthesis (94, 95, 292). The deduced amino acid sequence of cmaT suggested that it functions as a thioesterase (TE), providing further support to the role of a thiotemplate mechanism for CMA biosynthesis (270). CmaT has now been overproduced in E. coli, and assays with a variety of esters and thiolesters indicated that the overproduced protein was a functional esterase in vitro (197). However, sequence analysis of cmaU has failed to reveal anything useful about its potential function in CMA biosynthesis.

A region required for the coupling of CFA and CMA via amide bond formation was sequenced, and a 1.4-kb gene designated cfl (coronafacate ligase) was identified (Fig. 6D) (22, 143). Cfl is most closely related to enzymes that activate carboxylic acids by adenylation; consequently, this enzyme may catalyze the adenylation of CFA and the ligation of the CFA-adenylate to CMA. Coronafacate ligase has been overproduced in Escherichia coli and P. syringae in soluble form (214). Although the precise function of the enzyme remains unclear, construction of a nonpolar mutation in cfl suggested that the enzyme may also function in CFA biosynthesis (214).

Complementation experiments with CFA-defective mutants and an extensive series of subclones demonstrated that the cfl-CFA region was contained in a single 19-kb transcriptional unit (Fig. 6A) (142). Transcriptional fusions to a promoterless glucuronidase gene and primer extension analysis indicated that transcription initiated upstream of cfl in SstI fragment 8 and proceeded through SstI fragment 4 (142, 143) (Fig. 6). The 5' end of the transcript contains six discrete ORFs including cfl and cfa1 to cfa5 (143, 203) (Fig. 6D). Sequence analysis of cfa1, cfa2, and cfa3 revealed relatedness to ACP, fatty acid dehydratase, and -ketoacyl synthetase, respectively (203). The function of cfa4 could not be predicted from database searches, whereas the translation product of cfa5 showed relatedness to acyl-CoA ligases (203). Both cfa1 and cfa3 were overexpressed in E. coli, and protein products close to the predicted size were visualized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (203).

The continued sequencing of SstI fragments 5 and 6 indicated the presence of two large open reading frames (ORFs) that were designated cfa6 (8.0 kb) and cfa7 (6.2 kb) (Fig. 6D) (212). Both proteins exhibit a high degree of similarity to 6-deoxyerythronolide B synthase (Fig. 3), suggesting that a type I PKS participates in CFA synthesis. Both Cfa6 and Cfa7 were overproduced in E. coli and shown to encode multifunctional PKSs with antigenic similarity to DEBS 2 (212). Two additional genes, cfa8 and cfa9, mapped downstream of cfa7 (Fig. 6D); cfa8 was required for the biosynthesis of both CFA and COR, and the predicted translational product showed similarity to crotonyl-CoA reductases from Streptomyces spp. (213). Crotonyl-CoA reductase catalyzes one step in the conversion of acetoacetyl-CoA to butyryl-CoA, and the latter product is used as a 4C extender in polyketide synthesis (80). Consequently, the recruitment of a ccr gene into the CFA gene cluster may reflect the requirement of butyryl-CoA as a precursor for CFA synthesis. cfa9 showed relatedness to thioesterases (TEs) involved in the synthesis of gramicidin, tyrocidine, and tylosin (130, 155, 178). Furthermore, Cfa9 contained the GxSxG and GxH motifs characteristic of diisopropyl fluorophosphate-sensitive animal and avian TEs (43). Analysis of a cfa9 mutant indicated that this gene was dispensable for CFA and COR production but may increase the release of enzyme-bound products from the COR pathway (213). The complete nucleotide sequence of the CFA biosynthetic gene cluster has facilitated the development of a model which incorporates the activities of both the mono- and multifunctional proteins (212).

A variety of nutritional and environmental factors have been examined for their effect on COR production in P. syringae pv. glycinea PG4180 (189). Temperature had a highly significant effect on COR biosynthesis in P. syringae pv. glycinea PG4180, with maximal production at 18°C and negligible yields at 30°C (189). Interestingly, growth of PG4180 was relatively unaffected over a range of temperatures tested (14 to 30°C). This response to temperature is consistent with symptom development in the field, since P. syringae pv. glycinea is predominantly a cool-weather pathogen. CFA and CMA were also subject to the same pattern of temperature control, with optimal production at 18°C (187, 270). Recently, Rohde et al. (220) showed that COR production was thermoregulated in selected strains of P. syringae pv. atropurpurea, maculicola, morsprunorum, and tomato, which may indicate that temperature is a common regulatory control for COR biosynthesis in other pathovars of P. syringae.

The production of both CFA and CMA in PG4180 is regulated at the transcriptional level by temperature. Transcriptional fusions of the CFA and CMA promoter regions to a promoterless glucuronidase gene indicated that transcriptional activity in both biosynthetic gene clusters was maximal at 18°C and significantly lower at 28°C (142, 191, 214, 270). The higher level of transcriptional activity for the CMA and CFA biosynthetic promoters at 18°C helps explain why COR production is optimal at this temperature.

A regulatory region was isolated which controls both CFA and CMA production; the nucleotide sequence of this region revealed the presence of three genes, corP, corS, and corR (Fig. 6A) (273). The deduced amino acid sequences of corP and corR indicated relatedness to response regulators that function as members of two-component regulatory systems, and the translational product of corS showed sequence similarity to histidine protein kinases which function as environmental sensors (273). Response regulators control the adaptive response in two-component regulatory systems and are characterized by an N-terminal receiver domain which functions as the phosphorylation site and a C-terminal effector domain with a DNA-binding, helix-turn-helix (H-T-H) motif (192, 193). Both domains are strongly conserved in CorR; CorP, however, contains the highly conserved receiver domain (at least two aspartate residues and a conserved lysine) but lacks the H-T-H motif. The N-terminal receiver domains of CorR and CorP are almost identical when aligned, suggesting a shared specificity for the same phosphodonor protein(s). The COR regulatory system is modified from the two-component paradigm since it contains two response regulator proteins together with a single sensor protein.

Both CorR and CorP showed relatedness to response regulators in the RO_III group, which includes NarL, BvgA, and FixJ (193). Several of these response regulators function as positive activators of transcription and bind to specific target sequences upstream of the promoters they regulate (1, 36). Complementation analysis of a corR mutant, PG4180.P2, and transcriptional fusions to a promoterless glucuronidase gene (uidA) indicated that CorR functions as a positive regulator of COR gene expression (202). Deletion analysis of the cfl upstream region was used to define the minimal amount of DNA required for full transcriptional activity of a cfl::uidA fusion (202). A fragment located upstream from the cfl transcriptional start site was used in gel retardation and DNase I footprinting assays to define the specific bases bound by CorR. This region was also conserved in the promoter region for cmaA, suggesting that both the CFA and CMA structural genes are controlled by CorR, a positive activator of COR gene expression (202).

Gene fusions indicated that expression of corR and corP was not significantly different at 18 and 28°C. In contrast, expression of corS was regulated by temperature, and a corS::uidA fusion showed maximal transcriptional activity at 18°C and 15-fold less activity at 28°C (273). The use of transcriptional fusions and complementation analyses indicated that each regulatory gene was independently transcribed (Fig. 6A). Furthermore, experiments with the corS::uidA fusion indicated that transcription of corS was autoregulated and required functional copies of corR, corS, and corP (273).

COR production in PG4180 was significantly affected by the carbon source, glucose levels, amino acid supplements, complex carbon and nitrogen sources, and osmolarity (189). Transcriptional fusions were used to determine if any of these factors impacted transcriptional activity in the COR gene cluster (191). Glucose levels and selected carbon and amino acid sources significantly affected the expression of the cfl and cmaA operons. In general, gene expression increased with increasing amounts of glucose but was strongly repressed when selected carbon sources (xylose and fructose) and amino acids (isoleucine and valine) were added to the medium. Interestingly, changes in osmolarity and the addition of complex C and N sources did not significantly affect COR gene expression. In contrast, several researchers have shown that transcription of the hrp and avirulence genes (avr) genes in P. syringae is repressed by complex C and N sources and by increased osmolarity (106, 109, 144, 210, 227). Furthermore, the hrp and avr genes were shown to be transcriptionally activated in response to fructose (106, 144, 227). Obviously, some of the signals for activation of the hrp and COR gene clusters are different.

Several approaches have been used to investigate the potential stimulation of COR synthesis by host plants. Palmer and Bender amended the growth medium for PG4180 with extracts from soybean tissue or with plant-derived secondary metabolites but found no evidence that these substances substantially increased COR production in vitro (189). In a subsequent study, the activities of cmaA::uidA and cfl::uidA transcriptional fusions were compared in vitro and in soybean leaves; however, there was no evidence that COR gene expression in PG4180 was higher in plant tissue (188). In contrast, Ma et al. (145) showed that COR biosynthesis in P. syringae pv. tomato DC3000 is plant inducible. Gene fusions indicated that a single transcriptional unit designated CorII was expressed at a higher level in planta than in vitro. Other results indicate that shikimic and quinic acids may be signals for COR gene induction in DC3000 (137). These observations suggest that the signals for induction of COR synthesis differ in PG4180 and DC3000.

Syringomycin is representative of the cyclic lipodepsinonapeptide class of phytotoxins, which are composed of a polar peptide head and a hydrophobic 3-hydroxy fatty acid tail (77, 240) (Fig. 7). Characteristically, three forms of syringomycin are produced, differing only by the length of the 3-hydroxy fatty acid moiety, which is either decanoic (SRA₁), dodecanoic (SRE), or tetradecanoic (SRG) acid. An amide bond attaches the 3-hydroxy fatty acid to an N-terminal serine residue, which in turn is linked to 4-chlorothreonine at the C terminus by an ester linkage to form a macrocyclic lactone ring. Other distinctive structural features are a trio of uncommon amino acids (2,3-dehydroaminobutyric acid, 3-hydroxyaspartic acid, and 4-chlorothreonine) at the C terminus and the presence of D-isomers of serine and 2,4-diaminobutyric acid (231). Chlorination of the syringomycin molecule is important for biological activity (86). Syringomycin is produced by most strains of P. syringae pv. syringae isolated from a wide range of host plants. Syringotoxin and syringostatin are related lipodepsinonapeptides produced by strains isolated from citrus and lilac hosts, respectively (13, 76). The saprophytic strain MSU 16H from barley produces a fourth type of lipodepsinonapeptide called pseudomycin (12). All the lipodepsinonapeptides differ in amino acid sequence between positions 2 and 6, as depicted in Fig. 7. Despite these structural differences, the various types of lipodepsinonapeptides exhibit similar degrees of biological activity (245).

View larger version (14K): [in this window] [in a new window]   FIG. 7.   Structures of syringomycin, syringostatin, syringotoxin, and pseudomycin. The four lipodepsinonapeptides differ in the amino acid sequence between positions 2 and 6. The 3-hydroxy fatty acyl group is a derivative of either decanoic acid (syringomycin), dodecanoic acid (syringomycin and syringostatin), tetradecanoic acid (all four lipodepsinonapeptides), or hexadecanoic acid (pseudomycin); some forms of pseudomycin are acylated by 3,4-dihydroxytetradecanoate or 3,4-dihydroxyhexadecanoate. Abbreviations of nonstandard amino acids: Asp(3-OH), 3-hydroxyaspartic acid; Dab, 2,4-diaminobutyric acid; Dhb, 2,3-dehydroaminobutyric acid; Hse, homoserine; Orn, ornithine; Thr(4-Chl), 4-chlorothreonine; aThr, allothreonine.

Syringomycin induces necrosis in plant tissues, and early studies of its mode of action established that the plasma membrane of host cells is the primary target (8, 199). The amphipathic lipopeptide structure of syringomycin promotes its insertion into the lipid bilayers of membranes to form pores that are freely permeable to cations (105). The toxin causes an increase in transmembrane fluxes of K⁺, H⁺, and Ca²⁺ that are deadly to cells (32, 182, 258). Pore formation in lipid bilayers is a highly efficient process based on evidence that only nanomolar amounts of syringomycin are required for measurable activity. This is especially apparent in assays of tobacco protoplasts, where ⁴⁵Ca²⁺ influx and membrane lysis occur at a threshold syringomycin concentration of 50 ng/ml (103). Furthermore, Hutchison et al. (105) demonstrated pore-forming activity for SRE in a pure black-lipid membrane, thereby showing that the activity does not arise from opening of native ion channels found in membranes patched from cells. Many medically important bacteria produce pore-forming proteins or peptides that cause cytolysis as a result of massive ion fluxes (31). Syringomycin represents the first example of a virulence factor from a plant-pathogenic bacterium that targets host plasma membranes to form ion channels in lipid bilayers and causes cytolysis (103, 105).

From biophysical analysis of channel formation in planar lipid bilayer membranes, a picture is emerging of how syringomycin pores function and are formed (118). Shortly after insertion into the lipid bilayer, monomers of syringomycin aggregate into pore complexes. Based on the voltage-dependent behavior of ion channels, Feigin et al. (71) concluded that the channel was formed by at least six molecules of syringomycin. The lipophilic portion of each toxin subunit resides in the core of the bilayer, and the hydrophilic peptide head resides close to the surface of the membrane. Individual channels can become aggregated into clusters that exhibit synchronous opening and closing (118). The channel radius is approximately 1 nm for a syringomycin pore (105, 118). This is comparable in size to the channel dimensions of transmembrane pores formed by tolaasin, a lipodepsipeptide produced by the mushroom pathogen Pseudomonas tolaasii (211), and alpha-hemolysin, a cytolytic protein produced by E. coli (30).

It appears that sterols influence channel formation by syringomycin but are not components of the channel structure (70). Julmanop et al. (112) and Taguchi et al. (256) reported that sterols, particularly ergosterol, promoted the binding of syringomycin to cells. Because sterols can play a significant role in channel formation and are known to be essential for the cytotoxic activity of many pore-forming cytotoxins, such as streptolysin O (62), and lipopeptides, such as iturin A (132), an analogous role for sterols was proposed in channel formation by syringomycin. However, Feigin et al. (71) showed that the toxin readily formed ion channels in artificial membrane bilayers that lacked sterols. Subsequently, Feigin et al. (70) demonstrated that the addition of 50 mol% of ergosterol, stigmasterol, or cholesterol (sterols abundant in fungal, plant, and animal cells, respectively) to bilayers failed to alter the channel conductance properties of syringomycin.

The syringomycin pores are freely permeable to a series of monovalent and divalent cations (105). Based on studies by Takemoto and associates (215, 216, 293), an influx of H⁺ appears to be accompanied by an efflux of K⁺ across the syringomycin channel. This K⁺-H⁺ exchange generates an electrochemical gradient and collapses the pH gradient of the plasma membrane, resulting in acidification of the cytoplasm. The most conspicuous effect of channel formation by syringomycin is a rapid and sustained influx of Ca²⁺ ions that activates a cascade of events associated with cellular signaling in plants (120, 259, 260). For example, cytoplasmic influxes of Ca²⁺ caused by low concentrations of syringomycin lead to induction of kinase-mediated phosphorylation of membrane proteins (32) and the incorporation of 1,3--callose into plant cell walls (120). Consequently, the phosphorylation of a proton pump ATPase, as observed by Bidwai and Takemoto (32), appears to result from activation of protein kinases as modulated by a free Ca²⁺ signal (119). A short-term stimulation of the plasma membrane ATPase occurs (33). The resultant increase in ATP hydrolysis promotes the pumping of H⁺ and Ca²⁺ back out into the extracellular space, which in the long-term is ineffective against the collapse of the cation gradients of the cell. The ultimate benefit to the bacterium from pore formation is the systematic release of nutrients into the intercellular spaces of host tissues (105) and the alkalization of intercellular fluids, resulting in a more favorable environment for bacterial growth (42).

Sphingolipids, which are major lipid components of eukaryotic plasma membranes, have been associated with cell sensitivity to syringomycin (45, 90). A SYR2 mutant of the fungus Saccharomyces cerevisiae is highly resistant to syringomycin due to a failure to produce 4-hydroxylated sphingolipids such as phytoceramide (90). The total sphingolipid content of a SYR2 mutant is unchanged from the wild-type strain, and the mutant exhibited an abundance of ceramide moieties containing only dihydrosphingosine but not phytosphingosine. The significance of these observations on the biological activity of syringomycin is unclear, but sphingolipids play important roles in signal transduction as mediators of growth suppression and programmed cell death (185).

Another feature of the amphipathic syringomycin molecule is that it exhibits potent biosurfactant activity capable of lowering the interfacial tension of water to 31 mN/m, compared to a value of 73 mN/m for HPLC-grade water (105). Although pure preparations of syringomycin have a critical micelle concentration of 1.2 mg/ml, the surfactant properties are apparent at much lower concentrations. The surface-active properties of syringomycin are similar to those of other biosurfactants produced by fluorescent pseudomonads, including viscosin (183) and tolaasin (104). Biosurfactant activity appears to play an important role in spread of the bacterium across plant surfaces by reducing the surface tension of water and by concentrating relatively sparse nutrients at solvent interfaces (105).

Besides being phytotoxic, syringomycin and related lipodepsinonapeptides exhibit fungicidal activity toward a broad spectrum of filamentous fungi, such as Geotrichum candidum, and yeasts, such as Rhodotorula pilimanae (133, 245). Accordingly, assays for antifungal activity are conveniently used in bioassays of the toxins, which are active at concentrations as low as 0.8 µg/ml against yeasts (245). The related lipodepsinonapeptides syringotoxin and syringostatin have nearly equivalent broad-spectrum antifungal activity. Syringomycin also lyses erythrocytes, but 10-fold-higher concentrations of the toxin (i.e., 0.75 µg/ml) are required for comparable activity, as observed in assays of tobacco protoplasts (103, 105). Recently, efforts have been made to capitalize on the potent antifungal activities of syringomycin to control clinically important fungi, such as Candida spp. (245), and postharvest fungal pathogens of citrus, such as Penicillium digitatum (38). However, the strong hemolytic activity of syringomycin and related lipodepsinonapeptides remains a serious obstacle to commercial development.

Syringomycin biosynthesis occurs on a multifunctional complex of enzymes by a thiotemplate mechanism as originally described for peptide antibiotics produced by Bacillus, Streptomyces, and filamentous fungi (126). The first evidence for a nonribosomal mechanism of toxin synthesis originated with the association of large proteins, 470 kDa or larger, with lipodepsinonapeptide production (179, 288). Nontoxigenic (Tox) mutants were identified that were altered in the formation of these large proteins, which were speculated to function as peptide synthetases. Another important milestone was the resolution of the cyclic lipodepsipeptide structure of syringomycin containing nonproteinogenic amino acids and D-amino acids (77, 240). The occurrence of these unusual amino acids in the syringomycin peptide chain is indicative of nonribosomal multifunctional synthetases that catalyze the formation of peptides that contain modified amino acids (126, 252). Subsequently, Grgurina and Mariotti (89) used ¹⁴C-labeled amino acids to determine that L-threonine is the precursor of both 2,3-dehydroaminobutyric acid and 4-chlorothreonine and that aspartic acid was incorporated into 2,4-diaminobutyric and 3-hydroxyaspartic acids. The 3-hydroxy fatty acid tail of syringomycin appears to be derived from 3-hydroxyalkanoates that are accumulated in fluorescent pseudomonads and utilized as carbon and energy reserves (67).

Molecular genetic evidence for a thiotemplate mechanism of syringomycin synthesis came from analysis of the syrB1 gene that encodes an amino acid activation module characteristic of peptide synthetases (291). The SyrB1 amino acid sequence revealed six core sequences typical of the amino acid-activating modules described previously (247) (Fig. 1B). The core sequences of SyrB1 exhibited the characteristic order and spacing of thioester-forming modules involved in the synthesis of gramicidin S and other peptide antibiotics (291). Recent biochemical analysis demonstrated that the amino acid-activating module of SyrB1 catalyzes the recognition and activation of L-threonine (95). Thus, the syringomycin multienzyme system represents one of the first described for Pseudomonas, although it appears that all fluorescent pseudomonads synthesize peptide-containing metabolites, such as pyoverdin (154), by the thiotemplate mechanism.

The syringomycin (syr) gene cluster encompasses a DNA region of approximately 37 kb on the chromosome of P. syringae pv. syringae B301D (Fig. 8). Six ORFs in the syr gene cluster are predicted to encode proteins involved in the synthesis (syrB1, syrB2, syrC, and syrE) (95, 291), secretion (syrD) (209), and regulation (syrP) (292) of syringomycin. The most striking feature of the syr gene cluster is the presence of an enormous (28.4-kb) ORF, designated syrE (95). Sequence analysis indicated that syrE encodes a 1,039-kDa synthetase containing eight amino acid activation modules (95). Thus, SyrE represents the largest protein reported for a prokaryote to date. Precedence for the formation of large synthetases exists in the fungus Tolypocladium niveum, where the simA gene encodes a 1,689-kDa synthetase involved in cyclosporin A synthesis (281).

View larger version (17K): [in this window] [in a new window]   FIG. 8.   Physical map of the syringomycin gene cluster of P. syringae pv. syringae compared to that of the surfactin gene cluster of B. subtilis. The amino acid-adenylating domains and 4'-phosphopantetheine carrier domains are indicated by cross-hatched and stippled regions, respectively, for each module. The small circles attached above the proteins indicate the regions carrying motifs characteristic of thioesterases. The 37-kb syr gene cluster encodes four proteins (SyrB1, SyrB2, SyrC, and SyrE) involved in biosynthesis. The syrD and syrP genes encode proteins predicted to be involved in secretion and regulation, respectively. The 31-kb srf gene cluster encodes five proteins (SrfA-A, SrfA-B, SrfA-C, SrfA-TE, and Sfp) involved in surfactin synthesis. The sfp gene product has been proposed to function as a 4'-phosphopantetheine transferase (147). The functions of the products of orf5 through orf8 are unknown.

The domain organization of the predicted SyrE synthetase is illustrated in Fig. 8; each amino acid activation module contains domains for elongation, adenylation, and thiolation. The various SyrE adenylation domains exhibit a high degree of identity ranging from 45 to over 90%. Because the adenylation domain harbors the regions responsible for amino acid recognition and activation (247), the E1 adenylation domain of syrE was amplified, cloned, overexpressed, and used in biochemical assays (95). The E1 adenylation domain recognized L-serine as a substrate in ATP-pyrophosphate (PP_i) exchange reactions, indicating that syringomycin synthesis is initiated by activating the N-terminal amino acid, L-serine. The adenylation domain of E1 is preceded by an elongation domain, which is presumed to catalyze the acyl transfer of the 3-hydroxy fatty acid from SyrC to L-Ser bound to E1 as a thioester (54, 252). Elongation domains were previously shown to precede the first amino acid activation domain in SrfA-A which initiates the synthesis of the lipopeptide surfactin (252).

A distinctive structural feature of the SyrE synthetase is the fusion of a TE domain to the last amino acid activation module, E8 (Fig. 8) (95). This domain contains the GxSxG motif characteristic of TEs involved in the synthesis of fatty acids, polyketides, and peptide antibiotics (252). All bacterial peptide synthetases exhibit a TE domain at the C-terminal end of the synthetase that carries the last amino acid activation module; the TE domain presumably catalyzes a termination reaction in which the thioesterified peptide is released by hydrolytic cleavage from the synthetase. The importance of the TE domain in peptide antibiotic production in bacteria was demonstrated by Schneider and Marahiel (237); these researchers deleted the TE region of srfA to srfC, which encodes the last synthetase in surfactin synthesis, and reported a large reduction in surfactin yield. Unique to SyrE is the positioning of the TE domain relative to the E8 module, which are separated by intervening elongation and thiolation domains (95). In effect, the C terminus of SyrE contains elements of a ninth module that lacks an adenylation domain.

The ninth adenylation domain appears to be encoded by syrB (291). It is now known that syrB represents an operon that expresses two proteins (290): the 68-kDa SyrB1 protein carrying adenylation and thiolation domains, and the 35-kDa SyrB2 protein homologous to the CmaB protein in the COR gene cluster (37). The substrate specificity of the overproduced SyrB1 adenylation domain was analyzed in ATP-PP_i exchange reactions (95). SyrB1 activated L-threonine in exchange reactions, whereas a series of amino acids including L-serine and 4-chlorothreonine were not recognized as substrates. Therefore, SyrB1 presumably activates and binds L-threonine as a thioester, which may be subsequently modified to yield 4-chlorothreonine prior to transfer to the SyrE synthetase. Although the function of SyrB2 in syringomycin synthesis is unknown, one can speculate that it functions in the modification of L-threonine bound to the SyrB1 thiolation domain or in the cyclization of the mature peptide.

The syrC gene is located between the syrB operon and syrE and encodes an enzyme with TE activity (Fig. 8) (291). SyrC is similar to several proteins containing TE motifs, including CmaT, which is presumed to function as a coronamic acid thioesterase (270). The 48-kDa SyrC protein contains a GxCxG motif with a Cys substituted for Ser as the active site, a change that does not significantly affect the catalytic activity of TEs (285). The TE activity of SyrC was demonstrated by Grgurina et al. (87), who overproduced SyrC as an N-terminal His-Tag fusion protein in E. coli. SyrC catalyzed the hydrolysis of CoA from a 3-hydroxydodecanoyl-CoA substrate, supporting the hypothesis that SyrC functions as a thioesterase in syringomycin biosynthesis with a potential role in acyltransfer of a 3-hydroxy fatty acid. SyrC recognized a series of linear long chain acyl-CoA derivatives as substrates but did not utilize medium- or short-chain fatty acid derivatives. Site-directed mutagenesis of syrC was used to generate Cys-to-Gly mutations within the TE motif of SyrC, resulting in a mutant SyrC which lacked TE activity (290). Thus, it appears that SyrC catalyzes the transfer of the 3-hydroxydodecanoyl moiety from the corresponding CoA derivative to the amino group of serine bound to the E1 module of SyrE (Fig. 8).

In summary, syringomycin synthesis is catalyzed by four proteins, namely, SyrB1, SyrB2, SyrC, and SyrE (Fig. 8). In comparison, the surfactin gene cluster contains synthetases carrying seven amino acid activation modules (Fig. 8) (247). The adenylation of the N-terminal serine and covalent attachment of the amino acid to the synthetase by a carboxyl thioester initiate the biosynthetic process on the E1 module of SyrE (Fig. 1B) (95). Within the adenylation domain of E1, a serine-specific pocket located between core 2 and core 4 (47, 291) recognizes L-serine. ATP is bound to core 2 and subsequently cleaved by an ATPase located at core 4, leading to the formation of serine adenylate. The activated serine is then transferred to the thiolation domain and covalently linked by a thioester bond to the cofactor, 4'-phosphopantetheine, at a conserved serine residue within core 6 (247, 252). The TE activity of SyrC catalyzes the hydrolysis of 3-hydroxydodecanoyl-CoA and may transfer the 3-hydroxy fatty acid to the amino group of serine bound to the E1 module, thus forming a 3-hydroxydodecanoyl-L-serine conjugate. The intermediate is transferred to the E2 module, where the second amino acid, D-serine, is covalently bound as a thioester to 4'-phosphopantetheine. Each SyrE module carries a molecule of 4'-phosphopantetheine covalently bound to the thiolation domain, and this mediates the sequential transfer of carboxyl thioester-activated amino acids between aligned modules (253). Synthesis continues with a series of elongation cycles until an octapeptide is synthesized and bound at the thiolation domain of the E8 module of SyrE. At this point, the adenylation domain of SyrB1 interacts with the E8 module to incorporate the last amino acid, L-threonine, which is ultimately modified as 4-chlorothreonine. The TE domain at the C terminus of SyrE is presumed to release the mature syringomycin product from the synthetase. SyrB2 is speculated to function either in modifying L-threonine bound to SyrB2 or in cyclization of the mature peptide to form a lactone ring. Once cyclized, syringomycin is exported across the cytoplasmic membrane by the ATP-binding cassette (ABC) transporter protein, SyrD (209).

The regulation of syringomycin production is complex, based on evidence that both nutritional factors and plant signal molecules modulate toxin production by P. syringae pv. syringae (92). Iron exerts a positive regulatory effect on syringomycin production based on evidence that Fe³⁺ concentrations of 2 µM or higher are required for expression of a syrB-lacZ transcriptional fusion (174) and for maximum yields of toxin by strain B301D (91). In contrast, syringomycin production is repressed by inorganic phosphate concentrations of 1 mM or higher (91), and this resembles the phosphate-mediated down-regulation of antibiotic biosynthesis genes in many bacteria (139). The discovery that specific plant signal molecules also play a significant role in activating syrB gene expression and syringomycin production (discussed below) demonstrates that diverse environmental factors control the expression of genes dedicated to toxigenesis in P. syringae pv. syringae. Unfortunately, little is known about the complex genetic network responsible for the perception and transduction of these signals to the syr transcriptional apparatus.

Toxin gene clusters encoding a multienzyme system of peptide synthesis commonly carry regulatory elements that directly control the expression of biosynthesis genes (126). Surprisingly, the 37-kb syr cluster contains only one gene, syrP, that exerts regulatory effects on syringomycin production (292). A syrP mutant has an unusual pleiotropic phenotype with respect to syringomycin production and is substantially reduced in virulence on immature cherry fruits. In particular, syringomycin production by a syrP mutant is relatively insensitive to high inorganic phosphate concentrations in agar media. The syrP gene is located between the syrB and syrD genes (Fig. 8) and encodes a 40-kDa protein that may function in a phosphorelay signal transduction pathway. The SyrP protein exhibits similarity to the phosphoacceptor/transfer regions of histidine kinases such as CheA (296) and KinA (204), which are regulatory elements in phosphorelay pathways (7). Thus, SyrP may function in a phosphorelay system as an intermediate phosphate transmitter between a sensor protein and a response regulator (292). Phosphorelays govern major developmental commitments in microorganisms, and an important advantage of phosphorelays is that multiple signals can be integrated at intermediate steps in the regulatory network (7). Nevertheless, a role for SyrP in a phosphorelay mechanism of toxin production remains speculative until other members of such a regulatory system are identified and characterized. It is unclear whether additional regulatory genes are located within the right border region of the syr cluster (i.e., downstream of syrE) or are linked to the syringopeptin gene cluster (Fig. 8).

Global regulators of syringomycin production which are not physically linked to the syr cluster have been identified (101, 219). The gacS (lemA) and gacA genes encode members of a two-component sensory transduction system in P. syringae pv. syringae that regulates toxigenesis and the ability to cause necrotic lesions in plants. Sequence analysis indicated that GacS is a transmembrane protein which presumably functions as a histidine protein kinase that undergoes phosphorylation in response to environmental stimuli (101). GacA is a response regulator protein that is presumably phosphorylated by GacS (219). Like other members of the FixJ subclass of response regulators (41), GacA carries the phosphorylation site at the N terminus and a H-T-H motif at the C terminus (219). Nevertheless, it has not been demonstrated that GacS interacts directly with GacA to facilitate phosphate transfer in a two-component signal transduction system. An unusual feature of the gacS-gacA two-component system is that the two genes are not physically linked. Furthermore, the GacS-GacA homologs are conserved in fluorescent pseudomonads and control the expression of a diversity of cellular functions including production of protease, pectate lyase, pigments, and various antifungal metabolites (78, 125).

The GacS-GacA protein pair appears to be at the top of the regulatory hierarchy controlling syringomycin production, and little is known about the intermediary regulators. Kitten et al. (125) identified salA as a member of the gacS-gacA regulon that can restore syringomycin production to a gacS mutant if salA is overexpressed. A salA mutant is phenotypically distinguished from gacS or gacA mutants by a lack of suppression of protease production. Correspondingly, expression of a syrB-lacZ reporter was reduced to less than 3% in a salA mutant. The predicted SalA protein sequence exhibits an H-T-H DNA-binding motif with similarity to response regulators such as FixJ (5). It remains to be determined if SalA, as a putative response regulator, binds directly to the promoter region of the syrB operon or, rather, activates the syrB operon indirectly through one or more intermediary regulators such as SyrP. Thus, it appears that salA, as an element in the gacS-gacA regulon, controls the expression of the pathway that leads to syringomycin production and formation of necrotic lesions in plants (125). The gacS-gacA regulon also controls the expression of the ahlI_Pss gene specifying the synthesis of N-acylhomoserine lactone by P. syringae pv. syringae (61). However, syringomycin production is not controlled by the N-acylhomoserine lactone quorum-sensing signal. The environmental signals that activate the gacS-gacA regulon have not been identified, although they do not appear to be phenolic plant signal molecules such as arbutin (175, 219).

The syrA gene identified by Xu and Gross (288) appears to encode a regulatory protein required for syringomycin production and pathogenicity. The syrA gene, which lies outside the syr gene cluster, has not been sequenced, and its position in the regulatory network controlling toxigenesis remains to be determined.

Phytotoxin production by P. syringae pv. syringae is modulated by the perception of signals in the plant environment. The primary signals are specific phenolic glycosides that are abundant in the leaves, bark, and flowers of many plant species parasitized by P. syringae pv. syringae (175). For example, arbutin (Fig. 9) is a phenolic -glucoside distributed to more than 10 dicot families; in pear (Pyrus communis), arbutin constitutes 3 to 5% of the leaf material (156). In cherry (Prunus avium) leaves, two flavonol glycosides (quercetin 3-rutinosyl-4'-glucoside and kaempferol 3-rutinosyl-4'-glucoside) and one flavanone (dihydrowogonin 7-glucoside) glycoside (Fig. 9) have been identified as signal molecules based on the induction of a syrB-lacZ fusion as a reporter of gene activity (173). The plant signal molecules that activate toxin production by P. syringae pv. syringae are chemically distinct from those that activate the vir genes of Agrobacterium tumefaciens and the nod genes of rhizobia (152, 205). An intact glucosidic linkage is a structural feature of all syrB-inducing phenolic signal molecules; plants accumulate and store phenolic compounds as glycosides, which are more water soluble and less chemically reactive (100). The flavonoid signal molecules from cherry are equivalent in syrB-inducing activity and are abundant, as evidenced by the recovery of more than 11 mg (dry weight) of dihydrowogonin 7-glucoside per g from phloem (264).

View larger version (12K): [in this window] [in a new window]   FIG. 9.   Structures of phenolic plant signal molecules known to activate the syrB gene involved in syringomycin biosynthesis by P. syringae pv. syringae. Arbutin is found in several plant species including pear (Pyrus communis L.). The flavonol glycosides (quercetin 3-rutinosyl-4'-glucoside and kaempferol 3-rutinosyl-4'-glucoside) and the flavanone glucoside (dihydrowogonin 7-glucoside) are abundant in the leaves of sweet cherry (Prunus avium L.).

Phenolic signal activity is markedly enhanced in the presence of sugars that occur in large quantities in leaf tissue (173, 175). Sucrose and D-fructose are the most active sugars, causing a 10-fold stimulation of signal activity when phenolic signals occur at low concentrations. These two sugars also exhibit intrinsic low-level syrB-inducing signal activity in the absence of the phenolic inducer. Cherry contains an abundance of sucrose and fructose based on estimates of more than 3% of the dry weight of various tissues (121, 236). Consequently, the concentrations of both sugars exceed the threshold of 10 ppm required for significant induction of the syrB operon (175). The mechanism by which sugars augment the sensitivity of P. syringae pv. syringae to the phenolic signal is unknown. In A. tumefaciens, the ChvE sensory system mediates enhanced induction of vir genes (40). However, the two plant-microbe systems differ in sugar specificity, with the most conspicuous distinction being the inability of sucrose and D-fructose to enhance vir gene induction in the presence of the phenolic signal acetosyringone (4, 242).

In addition to the specific activation of syrB, the entire syringomycin biosynthetic apparatus is stimulated by plant signals in nearly all strains of the bacterium. The stimulatory effect of plant signal molecules on syringomycin was quite evident for some strains of P. syringae pv. syringae based on the recovery of up to 10-fold-higher toxin yields in a defined medium supplemented with arbutin and D-fructose (208). Furthermore, some strains required plant signal molecules for the production of syringomycin. In strains producing syringotoxin or syringostatin instead of syringomycin, plant signal molecules also stimulated toxigenesis. Such a network of communication between the plant and the bacterium that controls toxigenesis reflects the ability of P. syringae pv. syringae to adapt to a dynamic plant environment. The sensory mechanism favors the bacterium by detecting specific phenolic glycosides that signal the bacterium to rapidly activate virulence genes. P. syringae pv. syringae aggressively attacks a wide range of plants, and it would not be surprising to find that all host plants contain phenolics with the fundamental chemical structures responsible for signal activity.

Syringopeptins represent a second class of lipodepsipeptide phytotoxins produced by strains of P. syringae pv. syringae (9). In contrast to lipodepsinonapeptides, syringopeptins contain either 22 or 25 amino acids depending on the specific bacterial strain (Fig. 10). The N-terminal amino acid, 2,3-dehydro-2-aminobutyric acid, is acylated by either 3-hydroxydecanoic or 3-hydroxydodecanoic acid. An ester bond between allothreonine and the C-terminal tyrosine residue forms a lactone ring. A high percentage of hydrophobic amino acids are found in syringopeptin, and Ballio et al. (