Potato diseases caused by soft rot erwinias: an overview of pathogenesis



Three soft rot erwinias, Erwinia carotovora ssp. carotovora, E. carotovora ssp. atroseptica and E. chrysanthemi are associated with potatoes causing tuber soft rot and blackleg (stem rot). Latent infection of tubers and stems is widespread. As opportunistic pathogens, the bacteria tend to cause disease when potato resistance is impaired. Pathogenesis or disease development in potato tubers and stems is discussed in terms of the interaction between pathogen, host and environment, microbial competition and recent findings on the molecular basis of pathogenicity. Emphasis is placed on the role of free water and anaerobiosis in weakening tuber resistance and in providing nutrient for erwinias to multiply. Blackleg symptoms are expressed when erwinias predominate in rotting mother tubers, invade the stems and multiply in xylem vessels under favourable weather conditions. Soft rot erwinias tend to out-compete other bacteria in tuber rots because of their ability to produce larger quantities of a wider range of cell wall-degrading enzymes. However, despite extensive studies on their induction, regulation and secretion, little is known about the precise role of the different enzymes in pathogenesis. The putative role of quorum-sensing regulation of these enzymes in disease development is evaluated. The role certain pathogenicity-related characters, including motility, adhesion, siderophores, detoxifying systems and the hrp gene complex, common to most bacteria including symbionts and saprophytes, could play in latent and active infections is also discussed.


Three soft rot coliforms, Erwinia carotovora ssp. carotovora (Ecc), E. carotovora ssp. atroseptica (Eca) and E. chrysanthemi (Ech), are pathogenic to potato (Pérombelon & Kelman, 1980). They have recently been classified in the resuscitated genus Pectobacterium as P. ssp. carotovorum, P. carotovorum ssp. atrosepticum and P. chrysanthemi (Hauben et al., 1998), but the nomenclature has not yet been widely accepted by plant pathologists. The bacteria are Gram-negative, nonsporing, facultative anaerobes, characterized by the production of large quantities of extracellular pectic enzymes. They rely mainly on the production of these enzymes together with a wide range of other plant cell wall-degrading enzymes to cause disease (Collmer & Keen, 1986). Additional pathogenicity-associated characters have been identified, which could also be involved in the establishment of the bacteria in plant tissues and in a free-living or saprophytic life phase. They are the main cause of tuber decay in store and blackleg or stem rot in the field (Pérombelon, 1992).

Of the three soft rot erwinias, Ecc has the broadest host range worldwide. The restriction of Eca to potato could be attributed as much to genuine host specificity as to ecological factors favouring its survival from one season to the next on a vegetatively reproduced crop in cool temperate regions. In contrast, Ech is pathogenic to many plants in tropical and subtropical regions, but can also affect certain crops (maize, dahlia) in temperate regions. It can be divided into several loosely related biovars, serovars and pathovars. However, as many strains of the groupings tend to exhibit overlapping host ranges, the species is not easily subdivided at the infraspecific level (Young et al., 1992). Latent infection of potato tubers by the soft rot erwinias is widespread and, because disease tends to develop only when host resistance is impaired, they have been described as opportunistic pathogens (Pérombelon & Kelman, 1980). Of the three bacteria, only Eca and Ech appear to cause blackleg symptoms, but all three can cause tuber soft rot (Pérombelon & Kelman, 1987). Their pathogenesis is temperature-dependent: Eca tends to cause blackleg at temperatures < 25°C, and Ech, regardless of biovar, at higher temperatures (Pérombelon et al., 1987). However, what are known as cold strains of Ech have recently been found that can also cause blackleg in cool temperate countries. The apparent nonpathogenicity of Ecc could be attributed to its inability to compete successfully in rotting mother tubers with Eca or Ech or even with other saprophytic bacteria also commonly present on tubers (Pseudomonas spp., Bacillus spp., Clostridium spp.) (Pérombelon et al., 1979). However, when occasionally conditions allow Ecc to predominate in rotting mother tubers, it can cause blackleg (Molina & Harrison, 1980). Ecc is also the main causal agent of aerial stem rot where competition is less severe, especially when high temperatures occur, frequent aerial irrigation is practised and water is contaminated (Powelson, 1980).

In the past, attention has focused mostly on ecology of the soft rot erwinias and epidemiology of the diseases they cause (Kikumoto, 1980; Pérombelon & Kelman, 1980; Pérombelon, 1982, 1992; Stanghellini, 1982) and, recently, on their pathogenicity and host resistance (Collmer & Keen, 1986; Kotoujansky, 1987; Lyon, 1989; Barras et al., 1994; Salmond, 1994; Pérombelon & Salmond, 1995; Hugouvieux-Cotte-Pattat et al., 1996).

A comprehensive assessment of pathogenesis, which encompasses the interactions between host, pathogen and the environment leading to latent and active infections, by focusing exclusively on soft rot erwinia pathogenesis or disease development in potato tubers and stems is the subject of this review. Disease development, with emphasis placed on the transition from latent to active infection, is examined in two steps: (i) the role of environmental conditions and host reaction, and (ii) the implications of recent molecular studies on erwinia pathogenicity.

Pathogenesis of tuber soft rot

Latent infection of tubers by soft rot erwinias is widespread in most commercial seed grade stocks, ranging from undetectable (< 101 cells/g peel) to ~106 cells/g peel (Pérombelon, 1992). The bacteria are sited intercellularly, in lenticels and in wounds, usually beyond the phelloderm layer, and possibly, to a lesser extent, in the vascular system (xylem). Although essential for bacterial flagellar activity and bacterial multiplication, it is unlikely that there is a water film on the cell walls and still less free water in the intercellular spaces. Plant cells usually suffer from water deficit (high osmotic potential), ranging commonly from –5 to –8 bar in tubers to –5 to –20 bar in leaf tissue during the day, depending on the balance between water loss and uptake, which affects turgidity (Billing, 1987).

Cells of erwinias in tuber tissue appear to remain quiescent or dormant, even when relatively large numbers (up to 106 cells/g peel) are present. For example, this occurs when tubers are inoculated by vacuum infiltration in a bacterial suspension. After free water has been absorbed, the bacteria are in direct contact with the hydrated walls of live cortical cells. Bacterial numbers can fluctuate depending on tuber storage conditions, increasing under moist and decreasing under dry conditions. However, they usually do not exceed the critical level at which tissue maceration occurs, ~107 cells/g diseased tissue (Pérombelon et al., 1979), unless the tubers remain wet long enough to allow uninterrupted bacterial multiplication. The nature of this dormancy is still not clear. It should be emphasized that tubers also commonly harbour other pectolytic saprophytic bacteria (Bacillus spp., Clostridium spp., Flavobacterium spp. and Pseudomonas spp.) which, if given the opportunity, can also cause rotting (Lund, 1979). A possible explanation for this dormancy is low nutrient levels and/or, more likely, unavailability of free water (aw), which restricts bacterial growth. It is implied that the bacteria are unable to extract sufficient nutrient for growth, possibly as a result of host–pathogen interaction, which could be of a passive or active nature (see below). The bacteria, however, can survive for several months, long enough to bridge the gap between one cropping season and the next (Pérombelon, 1992). What is not known is whether there is a slow turnover in the erwinia cell population or whether the bacterial cells remain truly dormant throughout latency.

For multiplication of erwinias to proceed in tuber tissue, the inhibiting factors (water and nutrient availability and host resistance) have to be overcome. The main environmental factor for a shift from latency to disease development is the presence of water on tubers, which triggers a cascade of events leading to the onset of rotting (Pérombelon & Lowe, 1975). A water film on tubers leads rapidly, depending on prevailing temperature, to anaerobiosis within the tubers (Burton & Wigginton, 1970). Oxygen within the tubers, once depleted by tissue respiration, is not renewed by diffusion from the air because of the water film. Anaerobiosis impairs oxygen-dependent host resistance systems (phytoalexins, phenolics, free radicals, etc.). It also inhibits cell wall lignification and suberization, which offer protection from pectic enzyme degradation. Thus, injecting tuber tissue with cell-free erwinia culture supernatants containing large quantities of pectic enzymes results in tissue maceration under anaerobic but not under aerobic conditions (Maher & Kelman, 1983). It has been shown that, following wounding, cells bordering the wound site respond, under aerobic but not under anaerobic conditions, with deposition of callose at the primary pit fields. This is followed by intussusception of lignin in the middle lamellae and primary cell walls and suberization along the inner surfaces of the primary walls (Thomson et al., 1995). Failure of soft rot erwinias to initiate a rot under natural conditions in wounded tuber tissue under aerobic conditions is more related to inability of the bacteria to degrade the lignified middle lamellae than to inhibition of bacterial growth by toxic substances, such as phenolics and phytoalexins (Maher & Kelman, 1984). However, pre-formed, oxygen-independent resistance can affect bacterial growth; for example, resistance to tuber rot has been associated with high calcium level and pectin methylation of the cell walls (McGuire & Kelman, 1984; McMillan et al., 1993).

Anaerobiosis on its own will not result in breakdown of latency and rotting. Presence of free water is essential (Pérombelon & Lowe, 1975). It has been shown that availability of water also results in swelling of cortical cells and breaching of the phelloderm layer in lenticels, which tend to open (proliferate). In addition, anaerobiosis increases cell membrane permeability, resulting in leakage of cell contents. As a result, the bacteria can penetrate deeper into the tissue, multiply and eventually produce cell wall-degrading enzymes. Subsequent loosening of cells following enzymatic degradation and weakening of cell walls (Grimault et al., 1997) could cause cells to burst as osmotic pressure is released, providing additional nutrient (Hall & Wood, 1970). This is less evident at incipient plasmolysis (Tribe, 1955), and the extent of tuber tissue maceration by pectic enzymes is related to the water potential of the tissue (Alberghina et al., 1973). It is also possible that certain recently identified pathogenicity determinants (hrp and avr genes) could be involved and could contribute to host cell death (see below). A further effect of anaerobiosis could be to increase the virulence of the bacteria. Although their growth rate in vitro in a nutrient-rich pectate medium under anaerobic conditions is two to three times slower than under aerobic conditions, pectic enzyme production, in particular pectate lyase (Pel), per bacterial cell is twice that under aerobic conditions (M. C. M. Pérombelon, unpublished data).

Once these conditions have been satisfied, bacterial growth can proceed relatively without hindrance, even from a low inoculum level, resulting in rotting when the critical population level (107 cells/g diseased tissue) is reached. Soft rot erwinias tend to grow more rapidly and produce more pectic enzymes under most field conditions than the many soil-borne pectolytic bacteria also present –Clostridium spp., Bacillus spp., Pseudomonas spp. – thereby allowing them to predominate at the advancing rot lesion (Pérombelon et al., 1979). However, these nonerwinia bacteria and also soil saprophytes may grow on near equal terms behind the rotting front on nutrients made available by the dominant bacterium (I. K. Toth, S.C.R.I., Dundee, personal communication, 1999).

Inoculation of tubers under aerobic conditions by wounding instead of vacuum infiltration entails release of cellular contents, which stimulate bacterial growth but also the induction of host resistance mechanisms and wound-healing responses. What follows is a competition between bacterial multiplication and development of host resistance. The inoculum load needs to be sufficiently high for the bacterial population to reach a critical size for rotting to develop before inhibition by host resistance or resorption of liquid from damaged cells in the intercellular space. Otherwise, bacterial multiplication is arrested and the infection becomes latent. Rotting can develop under apparently aerobic conditions from a low inoculum when temperatures are high (> 35°C), probably because tuber tissue respiration rate depletes oxygen faster than can be replaced by diffusion, resulting in localized anaerobiosis. At the same time, tuber tissue metabolism is adversely affected and leakage of cell contents can occur to promote bacterial growth, which occurs more rapidly with increased tuber turgidity. This situation can occur in hot tropical conditions in tuber soft rots triggered by tuber moth larvae, which not only carry the bacteria deep within the tuber tissue, but also their feeding activity provides additional nutrients from damaged cells for bacterial growth (Elphinstone & Wiersema, 1988).

Pathogenesis of blackleg and aerial stem rot

In contrast to tuber rot, stem diseases generally develop under aerobic conditions. Blackleg develops when large numbers of the pathogen invade the stems after multiplication in the rotting mother tubers. The disease does not develop in plants grown without mother tubers (microplants) or in plants grown from erwinia-free seed, even in heavily contaminated soil. Factors that favour rotting of mother tubers by erwinias will also favour blackleg development. The more important factor is soil water level (rainfall/irrigation) which, if prolonged, induces development of the anaerobic conditions in mother tubers, favouring bacterial multiplication and initiation of rotting (Pérombelon et al., 1989).

The relative resistance of potato cultivars to tuber soft rot and blackleg can be different: high or low tuber resistance can be associated with either high or low blackleg resistance. However, cultivar rating based on incidence of blackleg appears to be more related to stem than to tuber resistance, but the rate at which mother tubers rot may affect blackleg incidence (Allefs et al., 1995). The bacteria are translocated passively in the transpiration stream from the vascular system of the rotting mother tuber to the stem xylem vessels. The nature of stem resistance could be directly related to anatomical features of stem tissues. It has been suggested that the presence of a lignified barrier at the junction between the mother tuber and emerging sprouts/stems could restrict translocation of the bacteria, and hence stem invasion, more effectively in resistant than in susceptible cultivars (Weber, 1990). Since the early histopathological studies of blackleg by Artschwager (1920), showing extensive lignification of vascular tissues and development of sclereids in stem cortex and pith and formation of protein crystals in leaf cells as a result of infection, little more has been done. These changes, however, tend to increase plant resistance to infection, as lignification is an effective resistance mechanism by restricting cell wall degradation by pectic enzymes (Dean & Kuc, 1987). Moreover, it develops sooner and more extensively in resistant than in susceptible cultivars (Weber, 1990). Blackleg symptoms usually develop on young shoots before lignification or higher up on older stems above the woody base.

Stem invasion by soft rot erwinias soon after emergence can result in blanking (rotting and death of the whole plant) (Pérombelon, 1992). When Eca blackleg develops early in the season, stem rot is essentially an extension of mother tuber rotting. Stunting, chlorosis and wilting symptoms, caused by restriction of water flow in the xylem vessels following infection, tend to develop at that stage under dry conditions. The situation is probably different later in the season when plants have become more independent of their mother tubers and underground parts of stems more lignified, and hence more resistant to maceration. Stem invasion, even by large numbers of bacteria, does not necessarily result in blackleg symptom development. This depends mainly on prevailing environmental conditions, especially rainfall. Vascular tissue of apparently healthy plants often has high populations of the pathogen. For example, especially in cool conditions, it is not unusual to find that only one of the three or four stems emerging from a mother tuber is diseased and the others, while heavily infected, remain symptomless (Pérombelon et al., 1987).

A high proportion of stems of apparently healthy plants grown from contaminated seed tubers usually harbour low numbers of the bacteria in a quiescent form, often for a long time (Helias et al., 2000). Latency can be reversed when weather conditions are favourable (wet conditions) to induce bacterial multiplication, leading eventually to blackleg. However, it is not clear why the bacteria remain quiescent and how wet weather conditions can induce the bacteria to multiply to cause disease under presumably essentially aerobic conditions. It is possible that under wet conditions when the transpiration rate is low, there is little translocation of erwinia cells from the mother tuber or stem base high up the stems via the xylem vessels. Although xylem vessels are nutrient-poor, especially in carbon and nitrogen sources, nutrient could seep and accumulate in the vessels from adjacent living cells when there is little water translocation. Under these conditions, erwinias that are already present, especially at the base of stems, would be able to multiply and obstruct water flow. It is known that blackleg symptoms on young plants develop first at the stem base and not at the top of stems or in leaves. However, how precisely the bacteria induce rotting of the stem is still not clear. It is possible that, following multiplication in xylem vessels to the critical population level when symptoms develop, cell wall-degrading enzyme activity allows the bacteria to invade the cortex in large numbers to cause rotting. This process is probably repeated as the bacteria move up the stem in the xylem vessels or in the cortical tissue. It is notable in the case of aerial stem rot, when stems are infected naturally (insect and wind damage and irrigation water) or artificially (inoculation), that the bacteria must gain access to the vascular system in order for rotting to spread along the stem. This implies that wounds are sufficiently deep to breach the xylem vessels (Hellmers & Dowson, 1953). Otherwise, rotting tends to be localized, possibly arrested by resistance mechanisms active under aerobic conditions. In contrast, under dry atmospheric conditions but with adequate soil moisture, the bacteria are likely to be translocated rapidly in the transpiration stream from rotting mother tubers to the top of the plant or into leaves before they have an opportunity to multiply and cause a rot. Although bacterial numbers remain low, discoloration of the vascular system and wilting followed by desiccation of mainly the top leaves can develop, especially when caused by Ech, indicating some restriction of water flow (Lumb et al., 1986).

The reason why blackleg is not more prevalent, even in field experiments in which all seed tubers have been inoculated and despite the fact that latent infection of stems is generally widespread, is probably because the pathogen has to be the dominant erwinia present and has a population large enough to initiate disease. Bacterial numbers tend to vary during the growing season in both rotting mother tubers and stems, increasing and decreasing depending on fluctuations in general weather conditions and variation in microclimatic conditions at plant level and within the field (Pérombelon et al., 1987). Symptoms are expressed only in a few instances, when the pathogen predominates firstly in the rotting mother tubers and subsequently, following invasion, in stems. The higher the initial seed contamination level, the more likely it is that large numbers of the pathogen will develop in the rotting mother tubers, and hence the greater are the risks of stem invasion and blackleg incidence (Bain et al., 1990). Competition in rotting mother tubers, modulated by environmental conditions, especially temperature, determines whether erwinias will predominate and which one, if more than one is present.

Aerial stem rots, which develop after wounding from small inoculum loads (air- or water-borne erwinias or bacteria in latent stem infection), tend to occur only under wet conditions (frequent irrigation or prolonged rainfall) and when senescence has set in. Under these conditions, water and nutrients are readily available for rapid growth of erwinias, usually the more widely distributed Ecc.

Implications of recent molecular studies on erwinia pathogenicity

Role of cell wall-degrading enzymes

Soft rot erwinias produce a wider range of enzymes able to degrade plant cell wall components compared with most pectolytic saprophytic bacteria. These enzymes include pectinases, cellulases, proteases and xylanases, which have different properties (acidic or basic pI, high and low optimum pH, periplasmic or extracellular and exo- or endo-mode of action) (Collmer & Keen, 1986). The ability to produce a wider range of enzymes/isoenzymes more rapidly and in larger quantities than pectolytic saprophytic microorganisms enables erwinias to invade living plants more readily and cause disease (Lapwood, 1957; Zucker & Hankin, 1970; Liao, 1989). In contrast to pathogens specialized to infect a narrow host range, the mechanisms that have evolved in the soft rot erwinias apparently allow a high degree of plasticity in their metabolism. This enables the bacteria to exploit a wide range of substrates, living or dead plant tissues.

Of the different enzymes, the pectinases are believed to be the most important in pathogenesis, being responsible for tissue maceration and, indirectly, cell death. This is achieved by degradation of pectic substances in the middle lamella between cells. Four main types of enzymes are produced, three with a high optimum pH (~8), namely pectate lyase (Pel), pectin lyase (Pnl) and pectin methyl esterase (Pme), and one, polygalacturonase (Peh), with a low optimum pH (~6) (Collmer & Keen, 1986). Some attack preferentially pectic substances with a low degree methylation and others attack preferentially those with a high degree of methylation. Characteristically, they are usually present in multiforms or isoenzymes encoded by independent genes. For example, Ech produces five major Pel grouped into two families (PelA, D, E and PelB, C) and at least three minor Pel (I, L, Z) induced preferentially in plant tissues, belonging to three other families. In contrast, E. carotovora produces three major Pel belonging to the PelB, C family, an intracellular Pel homologous to a Pel from Yersinia pseudotuberculosis and several minor plant-induced Pel (Hinton et al., 1989; Jafra et al., 1999). The role of these enzymes at molecular level in pathogenesis has been reviewed fully by Barras et al. (1994). Only aspects relevant to potato infection will be highlighted here. Considerable information is now available on the molecular basis of enzyme production in vitro, namely the different types, their induction, synthesis, regulation, secretion and export, and, to a lesser extent, the effect of environmental and host factors on their expression. Although E. carotovora and Ech produce similar symptoms on potato, the pectic enzymes produced and their regulation and secretion systems are somewhat different. Moreover, most of the relevant work has been done on Saintpaulia ionantha (African violet), tobacco and chicory, and only to a lesser extent on potato. Not surprisingly, our understanding of the role of these enzymes in potato pathogenesis is still fragmented, notably the relative importance of the different enzymes and isoenzymes when produced in planta.

As with fungal pathogens (Cooper, 1983), pectic enzyme and isoenzyme expression in erwinias is sequential in planta (Yang et al., 1992; Masclaux et al., 1996). This would suggest that the genes concerned are also regulated separately as well as by global regulatory systems, such as the quorum-sensing system (see below). A possible biological function for the sequential enzyme production may be to maximize the activity of the main enzyme responsible for tissue maceration, Pel (pH optimum ~8), in plant tissue where the apoplastic tissue fluid pH is ~6. Recent studies with Ech on chicory leaves indicate a pH-dependent, sequential regulation of pelA, D, E genes (Nachin & Barras, 2000). PelA, D are expressed early during infection under acidic conditions, while PelE is mainly transcribed once the conditions have become basic. PelA action is known to result in predominantly pectic oligomers with a high degree of polymerization (10–15 residues). These residues are able to activate plant defence systems and to induce a rapid influx of H+ in plant cells, resulting in a rise in the pH of the apoplastic fluid (Baker et al., 1990). Since E. carotovora does not produce PelA, D, E (Barras et al., 1994), it is possible that the combined action of Peh, which has a pH optimum near to that of tissue apoplastic fluid, and of exo-Pel, could have the same effect as Ech PelA in raising the pH (Yang et al., 1992).

Using potato tubers as a model, rotting was still obtained, although to different degrees, when inoculated with Ech Pel A, B, C, D, E mutants. This was probably because of complementation by the other isoenzymes and the production of enzymes induced mainly in planta (Beaulieu et al., 1993). PelE was found to be more important than PelB, C in causing disease (Payne et al., 1987). Similar results were obtained with Saintpaulia, PelB, C being less important than PelA, D, E (Boccara et al., 1988). However, PelB, C are probably responsible for tissue maceration by E. carotovora, as the other Pel are absent. This difference could be attributed to the fact that regulation of extracellular enzymes in E. carotovora and Ech involves different systems (Barras et al., 1994). Ech minor Pel could play a more important role in potato tuber tissue maceration than hitherto thought. Inactivation of any one of Ech minor pel genes results in less maceration, associated with lower induction of major Pel synthesis and bacterial multiplication, than mutation of genes coding for the major Pel (Jafra et al., 1999). The minor Pel may produce pectic enzyme inducers and more suitable substrates for enhanced major Pel activity. Of the other cell wall-degrading enzymes, only Pme has been shown to play an important role in Ech pathogenesis (Boccara & Chatain, 1989), which tends to support the apparent relationship between increased potato resistance to soft rot and higher cell wall pectin methylation (McMillan et al., 1993). In contrast, Peh, Pnl and cellulase appear to contribute to the pathogenicity of E. carotovora (Barras et al., 1994). As the differential effect of physical factors on isoenzyme production has been obtained usually using nonpotato host models, extrapolation to potato is not advisable because the relative importance of each enzyme is host-dependent (Hugouvieux-Cotte-Pattat et al., 1992). However, a differential effect of temperature on Pel and Peh production by erwinias in vitro could explain the different ability of Eca and Ech to cause blackleg in cool and hot climates: enzyme production by the former but not by the latter is depressed at temperatures > 30°C (Pérombelon et al., 1987; Lanham et al., 1991). The molecular basis for this effect is unknown.

The development of maceration symptoms when a soft rot erwinia population has reached a threshold level can be explained in terms of a cell-density-dependent regulatory or quorum-sensing system for extracellular enzymes (Swift et al., 1996). Enzyme production is switched on when both cell numbers and the inducer, homoserine lactone (HSL), which is secreted by the bacteria, has reached a critical level. The loss of virulence in Ecc expI mutants unable to produce HSL (Jones et al., 1993; Pirhonen et al., 1993) and in an Ecc strain cloned with a Bacillus gene (aiiA) coding for an enzyme that breaks down HSL (Dong et al., 2000) can be construed to support a role for the system in pathogenesis. It has been suggested that quorum sensing allows the bacteria to multiply within host tissue without triggering host resistance. Otherwise early enzymatic degradation of pectic substances could generate phytoalexin elicitors (Swift et al., 1996). However, as rotting is initiated under anaerobic conditions, which entail loss of host resistance, the rationale for such a regulation system is questionable. A quorum-sensing regulation system is present in a wide range of bacteria, pathogenic as well as saprophytic, and not surprisingly their genes are relatively highly conserved (Pierson et al., 1998). Its primary function in soft rot erwinias may be related to situations that hold good whether inside and outside plants. It is possible that its role in relation to exoenzyme regulation in pathogenesis is secondary and may be an extension of its function in a saprophytic growth situation related to cellular energy economy, e.g. the lux gene system in Vibrio fischeri (Meighen & Dunlap, 1993). After bacterial growth has exhausted available nutrient, enzyme production is useful to exploit further nutritionally the present niche and promote tissue invasion. Similarly, the co-regulation of antibiotics by the erwinias would help to control competing bacteria attracted by the released nutrient. However, quorum-sensing systems could have a more specific role in pathogenesis as they are also involved in the regulation of hrpN-encoded harpin in E. carotovora (Mukherjee et al., 1997 (see below).

Despite the considerable progress made recently, an integrated picture of the mode of action of the different cell wall-degrading enzymes in potato tissue maceration has not yet emerged. Further information is needed on the coordinated and sequential expression of the enzymes/isoenzymes, especially those expressed mainly in planta, e.g. minor Pel, Pnl, in potato tuber and stem tissues under disease-inducing environmental conditions. When assessing their role in potato pathogenesis, especially under anaerobic conditions, it should be borne in mind that the enzymes are present not only to facilitate penetration and tissue colonization, but also to promote exploitation of the environment for nutrient. This includes degraded pectic polymers and cellular contents released from cells killed as a result of enzymatic activity. This interpretation would fit with the concept of soft rot erwinias as opportunistic potato pathogens.

Role of other pathogenicity determinants

As a generalization, cell wall-degrading enzymes can be said to be virulence determinants responsible for symptom development once growth of the bacteria has been initiated. For this to happen, favourable environmental conditions must prevail. In addition, it is likely that a set of what can loosely be described as pathogenicity characters have to be activated possibly to overcome the state of latency and to allow bacterial multiplication. Several potential candidates have been identified and probably more will be in the future. Some of the factors are involved in the establishment and growth of the bacteria in potato tissue and others in actively overcoming host resistance. However, it is sometimes difficult to differentiate their role as pathogenicity determinants from their function in general cellular housekeeping, which also applies in a saprophytic growth phase. The definition of pathogenicity genes is unavoidably subjective. In its widest sense, they are responsible for pathogenicity but are also likely to have other functions. In a narrower definition, they would be related solely to pathogenesis and, if inactivated, the mutants would be expected to be nonpathogenic while retaining the fitness of the wild-type strain, including ability to survive and grow outside living plants. As discussed below, however, few if any pathogenicity genes are likely to fall into the latter category. Furthermore, there is the problem of how to determine bacterial fitness, i.e. what characters are to be evaluated and under what conditions? The same difficulty applies to pathogenicity testing, which usually involves inoculation with artificially large numbers of the bacteria under inappropriate environmental conditions, thus bypassing the critical transition from latent infection by low bacterial numbers to active infection at higher populations.

Motility of Erwinia spp. has been described as a pathogenicity determinant, as it appears to be necessary for the successful invasion and infection of potato plants. Direct screening of Tn5-generated Eca mutants in potato stems has shown that a pleiotropic reduced-virulence mutant, defective in flagella assembly proteins, was not motile (Mulholland et al., 1993). Another Eca mutant also showed multiple cell-surface defects, including alterations in synthesis of outer membrane proteins, lipopolysaccharide (LPS), enterobacterial common antigen (ECA) and flagella as well as reduced synthesis of Pel and cellulase. Not surprisingly, it also exhibited reduced virulence on potato plants (Toth et al., 1999). The mutant phenotype was due to a defect in a single gene (rffG), which in other bacteria is involved in the production of ECA and LPS. Disruption of the cell surface could also have an effect on the flagellar protein assembly and enzyme secretory apparatus. Reduced virulence in this instance is probably more attributable to lower enzyme production than to the loss of motility or alterations in ECA and LPS. Motility is likely to be not very important when infecting tuber tissue from an inoculation site, since the bacteria tend to diffuse out from the soft rot lesion. However, it may have an ecological role in the contamination of tuber lenticels by erwinia cells in wet soils.

Adhesion, usually attributed to fimbriae or pili, is a feature common to most saprophytic and pathogenic plant and animal bacteria. It is considered to be an important pathogenicity character in many enteric pathogenic bacteria. In the case of Erwinia spp., only some erwinia strains are fimbriate, usually with different fimbriae types (Wallace & Pérombelon, 1992). Although E. carotovora cells have been shown to adhere, with varying degree, to different potato tissues, it appears to be independent of the presence or absence of known adhesins (Wallace & Pérombelon, 1993). Other undetected adhesive factors could be involved, e.g. flagella in the case of Pseudomonas fluorescens (De Flaun et al., 1990). Recently, an outer membrane protein immunologically similar to the intimin protein of pathogenic E. coli strains was found also to be produced by Ech (Duartéet al., 2000). In E. coli, adhesion to intestinal mucosa involves the binding of intimin to the Tir protein secreted by a type III secretion system into the host cells before they are killed. Although Ech can also induce mammalian cell death (Duartéet al., 2000), the question remains regarding the role of intimin-like proteins in plant pathogenesis.

Another putative pathogenicity-related factor is the outer cell wall surface component of Gram-negative bacteria, LPS, which has long been associated with plant and animal diseases. In bacteria–animal interactions, LPS on the one hand helps in protecting the bacteria against antibacterial substances, such as bile salts, and on the other triggers changes in the host resulting in increased resistance. Its role in plant pathogenesis has been examined especially in relation to virulence and hypersensitive reaction (HR) in different bacteria–plant interactions. Thus, LPS mutants of Ech exhibit a reduced virulence phenotype on Saintpaulia as well as phage resistance (Schoonejans et al., 1987). Protein–LPS complexes from Ech and several other plant pathogenic bacteria inhibit the hypersensitive reaction (HR) when infiltrated in host leaves prior to challenge by the pathogen (Mazzucchi et al., 1979; Sequeira, 1983). This effect, termed localized induced reaction (LIR), probably reflects the LPS-induced antibacterial environment, attributed mostly to antimicrobial phenolics conjugates, at the inoculation site. As bacterial numbers tend to be higher eventually than in an HR situation, LIR is believed to increase plant tolerance to bacteria (Mazzucchi et al., 1979). In Xanthomonas campestris pv. campestris, LIR is transitory, developing after 10–18 h and is lost after 30 h, possibly as a result of hrp gene activity (Newman et al., 2000). The fact that LPS preparations from E. coli and Salmonella spp. can also prevent HR expression (Newman et al., 2000) suggests that LPS is unlikely to play an important specific role in pathogenesis. In the case of the soft rot erwinias, the requirement for anaerobiosis for disease development would inhibit LIR, but LPS could still provide some protection against pre-formed antimicrobial substances in tubers.

Soft rot erwinias can produce catechol and hydroxamate siderophores (Persmark et al., 1989; Ishimaru & Loper, 1992). Latterly, attention has been focused on the role of the iron uptake system (siderophores) as a virulence determinant of Ech infecting Saintpaulia (Enard et al., 1988). A similar role for such systems has been described for several animal-pathogenic bacteria (Payne, 1993). Soluble forms of iron, essential for all forms of life, are not readily available in plant or animal tissues. Under iron-limiting conditions, growth of bacteria able to express high-affinity iron acquisition systems is favoured and, if they are also pathogenic, their virulence is apparently increased. However, siderophores probably also have a role to play in the saprophytic free-living phase of these bacteria, e.g. at iron-limiting oxygen levels and pH ranges present in many soils (Bossier et al., 1988). It is relevant to note that siderophores are also produced by many rhizobacteria such as fluorescent Pseudomonas spp., sometimes used as biological control agents acting by competing with erwinias on potato (Kloepper, 1983). Recently, an additional role has been attributed to the E. amylovora siderophore, desferrioxamine, which has a more direct bearing on pathogenicity. Depending on its concentration, the siderophore interacts with H2O2 and peroxidases present in the affected tissue either to enhance oxidative stress induced by harpin, coded by a hrpN gene (see below), or to protect bacterial cells by inhibiting generation of reactive oxygen species (Dellagi et al., 1998). It remains to be seen whether this scenario would apply equally to the soft rot erwinias, bearing in mind that rotting develops under anaerobic conditions.

In an attempt to widen the range of known pathogenicity-related genes, genetic strategies have been applied to identify genes that are activated only when the bacteria multiply in plant tissues. Novel pathogenicity genes are identified when infection of host plants by mutated strains is greatly reduced whilst retaining their known virulence factors (Beaulieu & Van Gijsegem, 1990). The problem with this approach is that some of these genes could be involved primarily in cellular housekeeping, especially under stress conditions in and outside plants, and any effect on pathogenicity is an unavoidable sequel. The distinction between the two functions of the genes is not always readily made and may depend on prevailing circumstances. One such plant-inducible gene of Ech infecting Saintpaulia is msrA coding for peptide methionine sulphoxide reductase, which protects and repairs proteins against active oxygen damage (El Hassouni et al., 1999). The gene is present in most living organisms and has been implicated in the pathogenicity of mammalian bacterial pathogens. It is essential for full virulence expression on Saintpaulia and chicory leaves. The reductase activity in compatible host–pathogen interactions is probably high enough to protect the bacteria against basic tissue-reactive oxygen species. A similar role has been attributed to the sap gene of Ech, which provides a detoxification mechanism that enables the bacteria to resist antimicrobial peptides from their hosts (Lopez-Solanilla et al., 2001). Not unexpectedly, this system has also been reported in Salmonella typhimurium and is probably also present in other bacteria. Interestingly, mutation in the sap gene decreased Ech virulence on potato tubers and chicory leaves to a greater extent than mutations in pelL or hrp genes. The implication is that plant tissues contain toxic peptides, which can be neutralized by the Sap system of Ech.

Both the MsrA and Sap systems could have a role in the infection of potato tubers by soft rot erwinias. Latent infection could be the result of a state of equilibrium between MsrA activity and reactive oxygen level or Sap activity level and antimicrobial peptide level, in addition to lignified cell walls and absence of free water inhibiting bacterial growth and activity as discussed above. However, when free water is available and anaerobic conditions prevail, the equilibrium could be upset with a reduction in the activity of the toxic substances, thus favouring bacterial multiplication and eventually tissue maceration. These detoxifying genes could play a greater role in stem infection, which occurs under aerobic conditions. Finally, as a reaction to the presence of erwinia cells, LIR could develop in tuber tissue under aerobic conditions, contributing to the development of a latent state. This would be reversed once anaerobic conditions developed.

Two sets of genes, hrp and avr, have been closely associated with the expression of pathogenicity and host specificity in most plant pathogenic bacteria (Collmer, 1998). The hrp cluster encodes a type III protein-secretion system that is highly conserved in both plant and animal bacteria. It is used to secrete certain hrp-encoded proteins such as harpins and pilins and to deliver Avr proteins across the walls and plasma membrane of living plant cells. Avr proteins and, to a lesser extent, harpins induce rapid cell death leading to HR. As a result, bacterial invasion is arrested in incompatible interactions with host plants. Avr proteins also appear to play a role in compatible host–pathogen interaction. Host specificity at intra- or interpathovar and at species levels is usually defined by avr genes (Leach & White, 1997). The hrp gene complex is present in all subspecies of E. carotovora as well as in Ech (Mukherjee et al., 1996). Some workers have noted HR caused by these bacteria when the masking effect of pectic enzyme activity has been reduced through targeted mutagenesis (Bauer et al., 1994; Mukherjee et al., 1997). The gene complex is transcribed in plants and minimal growth media but not in rich media. Expression of the genes is affected by environmental signals, such as carbon and nitrogen sources, pH, osmolarity, temperature and possibly plant-derived substances, depending on the bacterium (Lindgren, 1997). In E. carotovora, the hrp cluster is probably activated by HrpL itself responding to as yet unknown environmental or plant signal(s) through the action of other Hrp proteins. Production of hrpN-encoded harpin is under the control of two global regulatory systems, which are up- and downregulated by the quorum-sensing signal and RsmA, respectively. The role of hrp genes in the pathogenesis of the soft rot erwinias remains to be clarified. For example, infection of chicory leaves by a Ech hrp-2, HR mutant is reduced, but not prevented, although the mutant retained the ability to produce the full complement of extracellular enzymes (Bauer et al., 1994). In contrast, an Ecc hrpN, HR strain was as virulent as the wild type (Mukherjee et al., 1997). It is unlikely that hrp genes in these bacteria have the same importance in pathogenesis as in nonopportunistic pathogens, such as Pseudomonas syringae pathovars or E. amylovora. It has been suggested that, being tightly regulated in Ecc, hrp genes are expressed only under certain conditions, e.g. in an early stage in infection before production of large amounts of extracellular enzymes (Mukherjee et al., 1997). Thus, they could be involved in the transition from latent to active infection. It is relevant to note that the type III secretion system and related hrp genes have been found in nonpathogenic bacteria, including P. fluorescens and Rhizobium spp. (Preston et al., 1998). Similarly, the apparent poor development of host specificity in the soft rot erwinias with their promiscuous host range (Collmer, 1998) would suggest that avr genes are unlikely to be present or, if present, do not play an important role in pathogenesis. However, the secretion of Avr proteins of P. syringae by an Ech Hrp system functioning heterologously in E. coli cells carrying the Ech hrp genes indicates that avr genes could be present in Ech (Ham et al., 1998).

In the context of potato pathogenesis, soft rot erwinia hrp and avr genes could be involved in a subtle way to increase virulence, e.g. by inducing leakage of cellular fluids to provide nutrient for bacterial growth, as in E. stewartii infecting corn (Coplin et al., 1992). Assuming that the genes are functional under anaerobic conditions, a low level of expression may help in providing water and nutrient for the bacteria to grow until the critical cell number is reached for pectic enzyme production. This would be in addition to host cell contents leakage attributed to low oxygen level, as discussed above. However, the genes would not be expressed during the state of latency of the bacteria in potato tissues. Other factors, such as those discussed above, could override hrp and avr gene expression to maintain latency until favourable environmental conditions develop. It is possible that induction and regulation of hrp gene expression in these bacteria respond to certain environmental or plant signals as described earlier. Alternatively or additionally, avr genes could be involved in defining ecological fitness of the bacteria, a role which has also been attributed to them in other bacteria (Leach & White, 1997). Finally, these genes could have a role to play in allowing the erwinias to infect preferentially certain hosts, for example in Eca and Ech, which have a certain degree of host specificity, but less likely Ecc with its wide host range. In short, much remains to be done in analysing the role of these genes in erwinia diseases of potato.

Conclusions and future outlook

Although soft rot erwinias can be said to be highly successful pathogens when infecting potato, they can also be viewed as opportunistic pathogens. Latent infection is widespread, and bacterial multiplication is initiated and disease develops when host resistance is impaired. The role of erwinia pathogenicity characters in latency, tissue colonization, disease development and their response to environmental conditions remains to be clarified. Most of the characters are also present in many saprophytic bacteria and are probably also involved in general cellular housekeeping. Genetic systems responsible for disease development could be derived from those involved in a free-living or saprophytic-life phase in order to exploit nutrient-rich living plants. It is becoming increasingly clear that there is little difference at the genetic level in what are believed to be pathogenicity-associated characters between saprophytes, endophytes, symbionts and plant and animal pathogens (Preston et al., 1998). For example, some strains of Pseudomonas aeruginosa are pathogenic to both animals and plants (Rahme et al., 1995). These bacterial groups tend to share many characters implicated in pathogenesis, including cell wall-degrading enzymes, protein secretion types II and III systems, detoxifying enzymes and siderophores. Differences in their behaviour are probably due mainly to variation arising from evolutionary adaptation of the relevant genes, such as gene duplication and differences in their regulation, in order to exploit new environments. In addition, lateral transfer of certain gene clusters (e.g. hrp), sometimes known as ‘pathogenicity islands’, between different bacterial species could also have an important evolutionary role; it could be responsible for the transformation of a saprophyte into a plant pathogen (Preston et al., 1998). Progressive adaptation could lead to greater specialization in the infection system and host range and specificity; for example, although sharing many pathogenicity genes, Ech and, to a lesser extent, Eca are more specialized regarding host specificity than Ecc. Therefore, when analysing the role of pathogenicity determinants of soft rot erwinias, it is important to bear in mind that the bacteria are capable of behaving both as saprophytes and pathogens depending on external circumstances. Our understanding of pathogenesis would gain from parallel studies on pathogenic and saprophytic growth phases of the bacteria as modulated by environmental conditions.

Advances in the development of molecular techniques, such as microarray analytical systems for gene expression, would facilitate an evaluation of the role of soft rot erwinia genes expressed in pathogenesis, in particular in the transition from latent to active infection, and in saprophytic situations. Similarly, genomics studies, including sequencing the whole genome, would greatly facilitate the identification of hitherto unknown genes involved in potential pathogenicity mechanisms, as in the case of Xylella fastidiosa (Simpson et al., 2000). Finally, proteomics could be particularly useful for a better understanding of survival strategies of erwinias in response to stress conditions in planta and in the environment. The combined effect would lead to a rational analysis of bacterium–host–environment interaction, which would include defining the relative contribution of each parameter. The overall aim would be the definition of a model integrating anatomical, biochemical, physiological and environmental aspects of erwinia–potato interaction. On one hand, this would encompass latent infection and, on the other, disease development from its inception to symptom expression.


I thank A. K. Chatterjee, A. Kelman and I. K. Toth for useful comments on the manuscript.