The rhizosphere- and plant-associated biotopes are densely populated by numerous microbial species. The ability of a pathogen to survive under varying and competitive conditions as well as the ability to succeed in the interaction with its host are important elements of a plant pathogen's ecological fitness. In the competition for nutrients, bacteria employ numerous strategies. One widespread strategy is the production of antimicrobial compounds including compounds targeting closely related bacteria in the same nutritional niche. Such antimicrobial compounds are of great interest because they affect bacterial population dynamics and, consequently, survival and virulence (Riley & Wertz, 2002; Gardner et al., 2004). Furthermore, some antimicrobial compounds may have additional regulatory functions (e.g. Hauge et al., 1998; Eijsink et al., 2002; Kodani et al., 2005; Linares et al., 2006). Finally, the exploitation of narrow-spectrum antimicrobial compounds is an attractive strategy for the targeted combat of bacterial infections, e.g. in plant disease control (Montesinos, 2007).
The two main categories of antimicrobials from bacteria are the bacteriocins (ribosomally synthesized proteinaceous substances; usually narrow spectrum) and the antibiotics (secondary metabolites; usually broader spectrum). The important ecological roles as well as the great application potential of bacteriocins are well recognized in some fields such as the food fermentation industry (Cotter et al., 2005). While these aspects in principal are equally important in plant pathology, very little is known about bacteriocin production by bacterial plant pathogens and their close relatives. In this review, we summarize current knowledge in the field.
Several attempts have been made to form a useful classification scheme for bacteriocins from Gram-positive bacteria (Klaenhammer 1993; Eijsink et al., 2002; Cotter et al., 2005; de Jong et al., 2006; Heng & Tagg, 2006). Class I bacteriocins are lantibiotics, i.e. peptides that contain thioether bridges and unusual, posttranslationally modified amino acids. Currently, lantibiotics are divided into types A (relatively linear and flexible, cationic peptides), B (more globular, rigid peptides with no or negative net charge) and C (two-component bacteriocins) (Chatterjee et al., 2005). Class I bacteriocins have a range of activities generally resulting in membrane destabilization, pore formation and/or inhibition of cell-wall synthesis through binding to specific lipids (Chatterjee et al., 2005; Hasper et al., 2006). Class II bacteriocins form a large and highly diverse group of nonmodified peptides (Eijsink et al., 2002). They affect target cells in similar ways as class I bacteriocins, but their receptors seem to be proteins rather than lipids (Diep et al., 2007). Class III bacteriocins are not peptides but heat-labile proteins that are categorized according to their ability to lyse cells. Other classes of bacteriocins recognized in Gram-positive bacteria comprise bacteriocins that carry essential lipid or carbohydrate moieties and the cyclic peptides (De Jong et al., 2006; Heng & Tagg, 2006).
The best-known bacteriocins produced by Gram-negative bacteria are the colicins produced by Escherichia coli. Colicins are a family of antimicrobial proteins with narrow host ranges, acting primarily on other strains of E. coli and its close enteric relatives (Riley & Gordon, 1999). The colicins show large variation, in terms of both structure and their effects on target cells, effects that include pore-formation, inhibition of cell-wall synthesis, DNAse activity and RNAse activity (Braun et al., 2002). Another group of protein bacteriocins from Gram-negative bacteria comprises proteins that affect the target cell membrane by self-assembly into particles that resemble tails of bacteriophages (e.g. some subclasses of the pyocins produced by Pseudomonas aeruginosa; Michel-Briand & Baysse, 2002). Generally, protein bacteriocins from both Gram-negative and Gram-positive bacteria show large variation, as illustrated by some of the examples described below. Many members of the Gram-negative Enterobacteriaceae also produce antimicrobial peptides called microcins. This group of antimicrobial peptides has its own subclassification and includes compounds with and without posttranslational modifications (Duquesne et al., 2007). Another group of peptide bacteriocins from Gram-negative bacteria comprises the trifolitoxins that are discussed below. It should be noted that genome analyses indicate that Gram-negative bacteria also may produce ‘Gram-positive-like’ peptide bacteriocins (Dirix et al., 2004).
Bacteriocins from plant pathogenic bacteria
Most plant pathogenic bacteria are Gram-negative bacteria, and almost all known bacteriocins produced by these bacteria are proteins. Among bacteria residing in the soil and the rhizosphere as well as among saprophytic bacteria, Gram-positive species are more frequent. In such bacteria, many peptide bacteriocins, especially from Class I (lantibiotics), have been identified and characterized.
Genome sequencing of the plant pathogenic Pseudomonas syringae pv. syringae revealed the presence of S-type pyocins (Feil et al., 2005), which are also found in strains of the opportunistic human pathogen Pseudomonas aeruginosa. These are high-molecular-weight bacteriocins (65–80 kDa) that are related to colicins. S-type pyocins are complexes of two components: a large component with killing activity and a smaller immunity protein (Michel-Briand & Baysse, 2002). They act mainly on other species of Pseudomonas (Sano et al., 1993). Very recently, Pectobacterium carotovorum (previously known as Erwinia carotovora ssp. carotovora) was shown to produce a smaller colicin- and pyocin-like antimicrobial protein of about 55 kDa, called carocin S1. This protein was shown to inhibit other strains of the same species probably by exerting DNAse activity (Chuang et al., 2007; see also Kyeremeh et al., 2000). Colicin-like proteins are also found in the genomes of both the sequenced strains of Xylella fastidiosa (Simpson et al., 2000; van Sluys et al., 2003) and in Xanthomonas oryzae pv. oryzae (Ochiai et al., 2005).
As judged by genome sequences, many plant pathogenic bacteria appear to contain genes encoding so-called RTX-toxins, in particular haemolysins, which are usually considered virulence factors of Gram-negative bacteria (Lally et al., 1999; van Sluys et al., 2002). Interestingly, Oresnik et al. (1999) showed that Rhizobium leguminosarum produces such an RTX-type toxin of about 100 kDa. More importantly, they also showed that this protein indeed provides a competitive advantage (in terms of nodule occupancy) in the competition with some (not all) closely related strains. Hert et al. (2005) showed that production of (noncharacterized) protein bacteriocins enables Xanthomonas perforans to exert an antagonistic effect on the tomato pathogen Xanthomonas euvesicatoria, both in the laboratory and in field trials.
Bacteriocins that exert their antimicrobial action by self-assembling into cytotoxic phage tail-like fibers have also been observed in Gram-negative plant pathogens. One of the best-studied cases is that of the carotovoricins produced by Pectobacterium carotovorum. It has been shown that carotovoricins produced by one Pectobacterium carotovorum strain kill other Pectobacterium carotovorum strains and that the killing spectra of these bacteriocins are determined by the structure of the tail fibers (Nguyen et al., 2001; Yamada et al., 2006). Phage-tail-like bacteriocins have also been found in Ralstonia solanacearum (previously Pseudomonas solanacearum) (Arwiyanto et al., 1993) and it has been shown that avirulent bacteriocin producers reduce the development of bacterial wilt on tobacco (Chen & Echandi, 1984). A phage-tail-like bacteriocin produced by Serratia plymithicum has been found to inhibit growth of Erwinia amylovora (Jabrane et al., 2002). Phage-tail-like bacteriocins have also been found in Rhizobium (Lotz & Mayer, 1972). It is conceivable that the particle-like bacteriocins found during early work on Pseudomonas syringae by Haag & Vidaver (1974) also belong to this type.
There are a several examples of plant pathogens producing other protein bacteriocins that do not resemble the ‘standard’ types of antimicrobial proteins mentioned above and/or that have not been sufficiently characterized to allow some sort of classification. Xanthomonas campestris pv. glycines, the causative agent of bacterial pustule on soybean, produces a heterodimeric bacteriocin called glycinecin, which is encoded by two separate genes, glyA and glyB, that give rise to a 39- and a 14-kDa subunit, respectively (Heu et al., 2001). The antagonistic range of this bacteriocin has been found to mainly include other pathovars of X. campestris, and also X. oryzae pv. oryzae, which causes bacterial blight of rice (Heu et al., 2001). Glycinecin shares no significant similarity with any protein sequences known to date and acts by permeabilizing the membranes of target cells (Pham et al., 2004). Another case is the 30-kDa lectin-like putidacin (LlpA), produced by a rhizosphere isolate of Pseudomonas sp. BW11M1. This bacteriocin contains regions that resemble mannose-binding domains of lectins in monocotyledonous plants (Parret et al., 2003). Putidacin has inhibitory activity against strains of a number of Pseudomonas species, including pathovars of Pseudomonas syringae. Interestingly, two lectin-like bacteriocins with similar inhibitory spectra were recently identified in the well-known biocontrol strain Pseudomonas fluorescens Pf-5 (Parret et al., 2005). Finally, Lavermicocca et al. (1999) purified a bacteriocin from Pseudomonas syringae pv. ciccaronei, which potentially consists of three proteins (45–76 kDa). These authors showed that this bacteriocin effectively inhibits the causative agent of olive knot disease, Pseudomonas syringae ssp. savastanoi, in field trials.
Very little is known about protein bacteriocins from Gram-positive plant pathogenic bacteria. The best-studied case concerns ipomicin, a heat-sensitive 10-kDa protein produced by the sweet-potato pathogen Streptomyces ipomoea (Zhang et al., 2003). The antagonistic activity seems limited to closely related strains, i.e. primarily other strains of S. ipomoea. Holtsmark et al. (2007) recently showed that the Gram-positive tomato pathogen Clavibacter michiganensis ssp. michiganensis secretes a 14-kDa protein that inhibits growth of the closely related potato pathogen Clavibacter michiganensis ssp. sepedonicus.
Whereas peptide bacteriocins are abundant in nature, including bacterial microbial soil ecosystems, and despite the fact that genome searches reveal the presence of potential bacteriocin genes in most bacterial genomes (e.g. Dirix et al., 2004; Nes & Johnsborg, 2004; de Jong et al., 2006), very little is known about peptide bacteriocins produced by plant pathogens. The potential use of bacteriocins from other sources to control plant pathogens has recently been reviewed by Montesinos (2007).
Trifolitoxins are peptide bacteriocins produced by Gram-negative species such as Agrobacterium tumefaciens and R. leguminosarum. These bacteriocins consist of eleven amino acids carrying several (partly unknown) posttranslational modifications (Scupham & Triplett, 2006). Several studies confirm the role of these compounds in competition with closely related species (Robleto et al., 1998; Herlache & Triplett, 2002). These trifolitoxins are very interesting peptide bacteriocins, but more work is needed before they can be compared with other peptide bacteriocins and classified.
Holtsmark et al. (2006) described the first example of a plant pathogen producing a known type of peptide bacteriocin, namely a lantibiotic belonging to class I, type B. This bacteriocin, named michiganin A, is produced by the Gram-positive tomato pathogen C. michiganensis ssp. michiganensis and inhibits growth of the closely related potato pathogen C. michiganensis ssp. sepedonicus (Holtsmark et al., 2006, 2007; Fig. 1). Michiganin A resembles other type B lantibiotics such as actagardine (Zimmermann & Jung, 1997) and mersacidin (Prasch et al., 1997), which are produced by an Actinoplanes sp. and a Bacillus sp., respectively. These lantibiotics are thought to exert their antimicrobial action through interfering with the incorporation of lipid II into peptidoglycan (Bauer & Dicks, 2005). Michiganin A was found to inhibit C. michiganensis ssp. sepedonicus at concentrations in the 10–100 nM range (20–200 ng mL−1), which are typical inhibitory concentrations for peptide bacteriocins.
It is widely assumed that antimicrobial substances secreted by bacteria, be it bacteriocins or antibiotics, mainly have a role in competition. However, there are now several examples in the literature that link bacteriocins and antibiotics to other functions, such as signalling, virulence and sporulation (e.g. Kuipers et al., 1995; Hauge et al., 1998; Whitehead et al., 2002; Kodani et al., 2005; Linares et al., 2006). For example, the potential regulatory role of classical antibiotics was recently discussed by Linares et al. (2006) in a study on the regulation of virulence in Pseudomonas aeruginosa. Several class I and class II peptide bacteriocins produced by Gram-positive bacteria are known to have an additional function in that they act as an auto-inducing peptide pheromone that stimulates bacteriocin production (Kuipers et al., 1995; Brurberg et al., 1997; Hauge et al., 1998; Eijsink et al., 2002). This enables concerted bacteriocin production in bacterial subpopulations of the same species, thus increasing competitive power. Kodani et al. (2005) showed that a lanthionine produced by Streptomyces tendae is involved in aerial hyphae formation and has (limited) antimicrobial activity, providing another example of a bacteriocin-like peptide with a regulatory function. Most interestingly from a plant pathology point of view, Lee et al. (2006) recently showed that the avirulence gene product AvrXa21 produced by a number of X. oryzae pv. oryzae strains seems to be a secreted peptide dependent on a specific type of ATP-binding cassette transporter (RaxB; Burdman et al., 2004) that is characteristic for the production of many peptide bacteriocins and cognate regulatory pheromones. Whether the hypothesized peptide is a peptide pheromone or a bacteriocin, or perhaps is multifunctional, remains to be seen. Interestingly, both sequenced strains of X. oryzae pv. oryzae contain several candidate peptide bacteriocins/pheromones, as do the genomes of other Gram-negative bacteria (Dirix et al., 2004; Table 1).