Tansley review. Microbiological control of soil-borne phytopathogenic fungi with special emphasis on wilt-inducing Fusarium oxysporum.

Transcription

1 New Review Tansley review Microbiological control of soil-borne phytopathogenic fungi with special emphasis on wilt-inducing Fusarium oxysporum Author for correspondence: Claude Alabouvette Tel: Received: 17 April 2009 Accepted: 11 July 2009 Claude Alabouvette 1, Chantal Olivain 1, Quirico Migheli 2 and Christian Steinberg 1 1 UMR 1229, INRA Université de Bourgogne, Microbiologie du Sol et de l Environnement, 17 rue Sully, BP 86510, F Dijon Cedex, France; 2 Dipartimento di Protezione delle Piante and Istituto Nazionale di Biostrutture e Biosistemi, Università degli Studi di Sassari,Via Enrico De Nicola 9, I Sassari, Italy Contents Summary 529 I. Biological control of plant diseases: state of the art 530 II. Main modes of action of biological control agents 530 III. The protective strains of F. oxysporum: an unexplored model 532 IV. Future directions for the study of the protective capacity of strains of F. oxysporum 539 V. How to make biological control successful in the field? 540 References 541 Summary doi: /j x Key words: biocontrol, biotrophy, competition, ecological fitness, induced resistance, plant defense reactions, priming, root colonization. Plant diseases induced by soil-borne plant pathogens are among the most difficult to control. In the absence of effective chemical control methods, there is renewed interest in biological control based on application of populations of antagonistic micro-organisms. In addition to Pseudomonas spp. and Trichoderma spp., which are the two most widely studied groups of biological control agents, the protective strains of Fusarium oxysporum represent an original model. These protective strains of F. oxysporum can be used to control wilt induced by pathogenic strains of the same species. Exploring the mechanisms involved in the protective capability of these strains is not only necessary for their development as commercial biocontrol agents but raises many basic questions related to the determinism of pathogenicity versus biocontrol capacity in the F. oxysporum species complex. In this paper, current knowledge regarding the interaction between the plant and the protective strains is reviewed in comparison with interactions between the plant and pathogenic strains. The success of biological control depends not only on plant microbial interactions but also on the ecological fitness of the biological control agents. Ó 529

2 530 Review Tansley review New I. Biological control of plant diseases: state of the art At the beginning of the 21st century, humanity is facing challenges regarding food safety in relation to climatic changes and the energy crisis (Muller et al., 2008; Toth et al., 2008). Contrary to earlier predictions, the prices of agricultural commodities are increasing dramatically because of a shortage of production. The world s population is continuously growing and agricultural production must increase to ensure global access to safe food in sufficient amounts. At the same time, the use of fertilizers and chemical pesticides must be reduced to minimize the deleterious environmental impact of agriculture. In this context, there is a renewed interest in alternative approaches to pest, disease and weed control. The expression alternative control methods, meaning alternative to synthetic chemical pesticides, covers a broad array of different approaches based on agricultural practices and the application of natural products and beneficial (micro)organisms known as biological control agents (BCAs). It is important to clearly define what is meant by biological control, as there is a tendency to consider all methods using natural products to promote growth, by stimulating plant defense responses to stresses of abiotic or biotic origin, as part of the biological control approach. Although useful, most of these methods do not fit the definition of biological control. According to Eilenberg (2006), biological control or biocontrol is the use of living organisms to suppress the population density or impact of a specific pest organism, making it less abundant or less damaging than it would otherwise be. The main difference between biological control and other control methods is the use of living populations of beneficial organisms, which have several modes of action and thus avoid the risk of rapid appearance of resistance in the target population. The different components of biological control of plant diseases were reviewed by Cook & Baker (1983). Organisms that can be used to achieve biological control include avirulent or hypovirulent individuals or populations within the pathogenic species, antagonistic microorganisms, and the host plant itself, manipulated to have greater or more effective resistance to the pathogen. In this review we will mainly focus on microbiological control of plant diseases, which is based on the use of avirulent, nonpathogenic or antagonistic microorganisms; and we will limit our review to biological control of soil-borne pathogens, as root diseases are much more difficult to control than aerial diseases (Alabouvette et al., 2005). Biological control of soil-borne diseases was proposed more than 40 yr ago. A symposium held in Berkley in 1965 was titled Ecology of soil-borne plant pathogens; prelude to biological control (Baker & Snyder, 1965). The two main approaches to biological control of soil-borne pathogens were already proposed during this symposium: enhancement of the naturally occurring populations of antagonists and introduction of a selected BCA. As the approach based on enhancing natural biological control processes has been recently reviewed (Alabouvette & Steinberg, 2006) we will focus on the use of BCAs. Despite the increasing number of scientific papers dealing with biological control, there are still a very limited number of products on the market, the situation varying according to country. In the European Union, only two dozen microorganisms have been included in Annex I of the directive , which lists the active substances authorized for use in plant protection. Together, these biological products represent only a very limited share of the market in plant protection products. It is therefore interesting to review progress and failures in biological control research and to identify the bottlenecks that prevent more rapid success in the application of microbial control. After summarizing the main modes of action of BCAs, we will review knowledge of the interactions between the plant and protective strains of Fusarium oxysporum able to control wilt induced by pathogenic strains of F. oxysporum. II. Main modes of action of biological control agents Several modes of action are involved in the protective capabilities of BCAs, and generally a given strain acts through several mechanisms expressed successively, simultaneously or synergistically. 1. Microbial antagonism Microbial antagonism results from direct interactions between two microorganisms sharing the same ecological niche. Three main types of direct interaction may be characterized: parasitism, competition for nutrients or plant tissues, and antibiosis. Parasitism Parasitism of a plant pathogen by other microorganisms, including viruses, is a widely distributed phenomenon. The parasitic activity of strains of Trichoderma spp. towards pathogens such as Rhizoctonia solani has been extensively studied (Chet & Baker, 1981). It involves specific recognition between the antagonist and its target pathogen and several types of cell wall-degrading enzymes (CWDEs) that enable the parasite to penetrate the hyphae of the pathogen. Other mycoparasites such as Coniothyrium minitans (Jones et al., 2004) and Sporidesmium sclerotivorum (Adams & Fravel, 1993) are effective in controlling diseases caused by Sclerotinia spp. and other sclerotia-forming fungi. This type of antagonism, which causes death of the target organism, mainly results in a decrease in the inoculum density. Parasitism of fungal

3 New Tansley review Review 531 pathogens by viruses or virus-like particles can also induce hypovirulence. For example, hypovirulent strains of Cryphonectria parasitica are used to control chestnut blight. Hypovirulence is contagious; dsrnas can be transmitted from a hypovirulent strain to a virulent compatible strain and, under favorable conditions, hypovirulence can spread naturally in diseased forests (Milgroom & Cortesi, 2004). Competition for nutrients Competition for nutrients is a general phenomenon regulating the population dynamics of microorganisms sharing the same ecological niche and having the same physiological requirements when resources are limited. Competition for nutrients, especially for carbon, is common in an oligotrophic milieu such as soil, and is considered to be responsible for the well-know phenomenon of fungistasis (Lockwood, 1977; De Boer et al., 2003), which is the inhibition of fungal spore germination in soil. Although difficult to demonstrate experimentally, competition for nutrients in soil is certainly one of the modes of action of many BCAs such as Trichoderma spp. For example, the reduction in the germination rate of chlamydospores of F. oxysporum in the rhizosphere of cotton Gossypium sp. and melon Cucumis melo in the presence of Trichoderma harzianum T35 was attributed to competition for nutrients (Sivan & Chet, 1989). Competition for carbon between pathogenic and nonpathogenic F. oxysporum is one of the main modes of action of biocontrol strains of F. oxysporum. By studying the quantitative interactions in soil between a pathogenic strain and several nonpathogenic strains of F. oxysporum, Couteaudier & Alabouvette (1990) established that some nonpathogenic strains were more competitive for a carbon source than others and should be selected for biological control. However, the intrinsic capacity of a strain to use the carbon source efficiently corresponds to a competitive advantage only when the two populations are of the same size. Larkin & Fravel (1999) studied the dose response relationships between pathogenic and biocontrol strains of F. oxysporum governing biological control and defined an effective biocontrol dose for each protective strain. These authors attributed to differences in the mode of action of the strains the differences in dose response relationships and concluded that the strain Fo47 exerted its biological control effect mainly through competition for nutrients, while the strain CS-20 acted primarily by inducing resistance. In practice the BCA Fo47 needs to be introduced at an inoculum density much higher than that of the pathogen to be controlled. Competition for minor elements also frequently occurs in soil. For example, competition for iron is one of the modes of action by which siderophore-producing Pseudomonas spp. limit the growth of pathogenic fungi and reduce disease incidence or severity (Schippers et al., 1987; Bakker et al., 1991; Loper & Henkels, 1997). Moreover, iron competition from the siderophore-producing Pseudomonas spp. was shown to enhance the antagonistic effect of the protective strain Fo47 by making the pathogen more susceptible to competition for carbon (Lemanceau et al., 1992; Lemanceau & Alabouvette, 1993). Competition for colonization of the plant tissues Competition might also occur for colonization of the root surface and plant tissues. It has been proposed that intense colonization of the root surface by an antagonist would prevent access of the pathogen to the infection sites (see section III, 6: Competition for surface colonization ). Using several approaches to quantify root colonization by a nonpathogenic and a pathogenic strain of F. oxysporum, Eparvier & Alabouvette (1994) observed that the glucuronidase activity of the GUS-transformed pathogen was reduced in the presence of the protective strain and concluded that these strains were competing for root colonization. However, another interpretation of these results can be proposed: the presence of the protective strain might directly or indirectly inhibit the metabolic activity of the pathogen without reducing its biomass. Postma & Luttikholt (1996) considered the hypothesis of direct competition between two strains of F. oxysporum within the vessels of the host plant. They showed that some nonpathogenic strains were able to reduce the colonization of the carnation Dianthus caryophyllus stem by the pathogen, resulting in a decrease in disease severity. This hypothesis has never been supported by further data obtained by artificial inoculation of the plant by wounding. Antibiosis Antibiosis is the antagonism resulting from the production by one microorganism of secondary metabolites toxic for other microorganisms. Antibiosis is a very common phenomenon responsible for the biocontrol activity of many BCAs such as fluorescent Pseudomonas spp., Bacillus spp., Streptomyces spp. and Trichoderma spp. Very diverse molecules have been described and their role in the suppression of several plant pathogens has been documented (Fravel, 1988; Weller & Thomashow, 1993). They include not only antibiotics sensu stricto, but also bacteriocines, enzymes such as CWDEs, and volatile compounds with an antifungal activity. A given strain of BCA may produce several types of secondary metabolite, having different functions and effective against different species of fungal pathogen. For example, the strain CHAO of Pseudomonas fluorescens produces siderophores, phenazines, 2,4-diacetylphloroglucinol and cyanide, different combinations of these metabolites being responsible for the antagonism expressed against Gaeumannomyces graminis var. tritici and Chalara elegans (Défago & Haas, 1990). Strains of Trichoderma spp. produce many types of secondary metabolite (Sivasithamparam & Ghisalberti, 1998) including antibiotics (Howel, 1998) and CWDEs (Kubicek & Penttilä, 1998; Lorito, 1998), the role of which has been clearly established in biocontrol activity (Vinale et al., 2008). It is important to emphasize

4 532 Review Tansley review New that a single antifungal metabolite does not account for all the antagonistic activity of a BCA. Different secondary metabolites produced by a given strain of BCA might be responsible for antagonistic activities toward different pathogens. Therefore, one must avoid any generalization from one patho-system to another patho-system, and from one BCA to another BCA. The best example comes from the work of Woo & Lorito (2007), who demonstrated that a strain of T. harzianum produced different secondary metabolites depending not only on the plant to which it was applied but also on the target pathogen infecting that plant. 2. Induced resistance of the plant At present, most of the research dealing with biocontrol is focused on the plant microbial interactions leading to enhanced disease resistance. Any plant reacts to physical stresses such as heat, frost, drought, salt, inoculation with pathogenic or nonpathogenic microorganisms and chemical molecules of natural or synthetic origin by expressing defense reactions. The first evidence of systemic protection induced by a microorganism was reported by Kuć (1987), who found that cucumber Cucumis sativus protection against Colletotricum orbiculare after pre-inoculation of the cotyledons with this same pathogen. Most of the research dealing with induced resistance (IR) in plants has focused on signaling pathways at the biochemical and molecular levels. On the basis of the data obtained, a distinction between induced systemic resistance (ISR) and systemic acquired resistance (SAR) has been proposed. SAR is activated by necrotizing pathogens, resulting in enhanced resistance to a broad spectrum of pathogens in organs distant from the site of infection (Ryals et al., 1996; Hammerschmidt, 1999). The expression of SAR depends on the accumulation of salicylic acid (SA) and is associated with the induction of pathogenesis-related (PR) proteins. ISR corresponds to the resistance induced by plant growth-promoting rhizobacteria inoculated onto the roots and involving the jasmonic acid (JA) and ethylene (ET) pathways (Pieterse et al., 1998). These two main pathways are not independent and there are some common nodes; for example, both SAR and ISR are controlled by the regulatory protein non-expressor of pathogenesis-related genes1 (NPR1). Cross-communication between defense pathways provides a regulatory potential that allows the plant to finetune its defense responses (Pieterse & van Loon, 2004). It was assumed that protection conferred by IR was based on direct activation of defense reactions. However, based on the most recent research on ISR, the phenomenon of priming appears to be a common feature of the plant s immune system that offers protection (Van Hulten et al., 2006). According to Goellner & Conrath (2008), priming is a component of IR in plants and corresponds to a physiological state in which an induced plant shows a faster or greater activation of defense responses after infection with a challenging pathogen. Most of our knowledge of plant microbial interactions has been obtained from studies on plant pathogen interactions, which might not be representative of the interactions between plants and BCAs. Recently, studies of the signaling pathways induced in the plant by BCAs have been carried out, Trichoderma spp. and fluorescent Pseudomonas spp. being the two groups of BCAs on which most attention has been focused in this field. Research on ISR-inducing Pseudomonas showed that some of the microbe-associated molecular patterns (MAMPs), such as flagellin and lipopolysaccharides (LPSs), that are known to play a role in the interactions between pathogenic Pseudomonas spp. and the plant, are involved in the protection conferred by beneficial Pseudomonas spp. (Van Wees et al., 2008). Purified flagellin and LPS of the nonpathogenic resistance-inducing strains of P. fluorescens or Pseudomonas putida had the capability to induce resistance in the plant. However, this capacity varied depending on the plant species to which these molecules were applied. Moreover, flagellin- and LPS-defective mutants of these beneficial bacteria were often as effective in controlling the disease as the wild-type strains, suggesting that multiple MAMPs are involved in the activation of the plant s immune response (Bakker et al., 2007). Some of the antibiotics produced by the beneficial bacteria could also function as inducers of the immune response. For example, the 2,4-diacetyl-phloroglucinol produced by many fluorescent Pseudomonas spp. was demonstrated to induce resistance (Weller et al., 2002). Similarly, many of the secondary metabolites involved in hyperparasitism or antibiosis of Trichoderma spp. are also able to trigger plant resistance. Inoculation of Trichoderma spp. onto the roots resulted in the up-regulation of different endogenous defense-related proteins and enzyme activities in the plant, a situation typically occurring when the plant faces pathogen attack (Woo & Lorito, 2007). III. The protective strains of F. oxysporum:an unexplored model 1. Protective capacity of strains of F. oxysporum The idea of using nonpathogenic strains of F. oxysporum to control Fusarium diseases came from studies of soils naturally suppressive to Fusarium wilts, such as the soil from Châteaurenard (Louvet et al., 1976). These soils harbor high populations of nonpathogenic F. oxysporum and Fusarium solani whose involvement in the mechanism of soil suppressiveness was confirmed experimentally using a variant of Koch s postulates (Rouxel et al., 1979). Strains of F. oxysporum were much more efficient in establishing suppressiveness in soil than other species of Fusarium (Tamietti & Alabouvette, 1986). There is a great diversity among

5 New Tansley review Review 533 soil-borne nonpathogenic strains of F. oxysporum in their capacity to protect plants against their specific pathogens (Alabouvette et al., 1987; Forsyth et al., 2006), and some effective strains have not been isolated from soil but from the stems of healthy plants (Ogawa & Komada, 1984; Postma & Rattink, 1992). Moreover, it is well established that a pathogenic strain applied to a nonhost plant is able to protect it against further infection by its specific forma specialis. This phenomenon was first described by Matta (1971) and defined as cross-protection or premunition. Recently, Minerdi et al. (2008) showed that a protective strain of F. oxysporum was in fact a pathogenic strain belonging to the forma specialis lactucae, associated with several bacteria. The cured strain was pathogenic on the plant, suggesting that the protective effect was not a fungal trait but was dependent on the interaction with the ectosymbiotic bacteria. A fundamental question remains: how to characterize a nonpathogenic strain of F. oxysporum? Nonpathogenic is a negative definition; it means that the strain is lacking pathogenicity on the plant species on which it has been inoculated. Ideally, identification of pathogenic versus nonpathogenic strains should be based on molecular criteria. However, other than inoculation of the plant, there is no universal tool enabling characterization of an F. oxysporum strain s pathogenicity. Methods have been recently developed to identify formae speciales, such as albedinis, ciceris, cucumerinum, dianthi and lycopersici, among pathogenic strains, but these methods do not enable one to distinguish between pathogenic and nonpathogenic strains (Lievens et al., 2008). Considering the narrow host specificity of pathogenic strains, inoculation of a very large collection of different plant species would be needed to determine that a soil-borne strain is probably nonpathogenic. For this reason it seems preferable to focus on protective rather than on nonpathogenic strains of F. oxysporum, these protective strains being either true nonpathogenic strains or pathogenic strains applied to a nonhost plant. However, as there is great diversity among protective strains of F. oxysporum, one should be very cautious when generalizing to other strains the mode of action involved in the protection conferred on the plant by a given nonpathogenic strain or a pathogenic strain applied to a nonhost plant. Whatever their origin, protective strains of F. oxysporum have been shown to control Fusarium wilt in many crops, including asparagus (Asparagus officinalis; Elmer, 2004), banana (Musa sp.; Forsyth et al., 2006), basil (Ocimum basilicum; Larkin & Fravel, 1999), carnation (Dianthus caryophyllus; Garibaldi et al., 1986), chickpea (Cicer arietinum; Hervas et al., 1997), cucumber (Cucumis sativus; Mandeel & Baker, 1991), cyclamen (Cyclamen persicum; Minuto et al., 1995), gladiolus (Gladiolus sp.; Magie, 1980), melon (Cucumis melo; Rouxel et al., 1979), tomato (Solanum lycopersicum; Olivain et al., 1995; Fuchs et al., 1997) spinach (Spinacia oleracea; Katzube et al., 1994) and watermelon (Citrullus lanatus; Larkin et al., 1996). In contrast to other BCAs such as Pseudomonas spp. or Trichoderma spp., the protective strains of F. oxysporum are effective mostly against pathogenic F. oxysporum. Only a few papers have reported efficacy against Pythium ultimum (Benhamou et al., 2002), Phytophthora capsici (Silvar et al., 2009) and Verticillium dahliae (Pantelides et al., 2009). Moreover, some endophytic strains of nonpathogenic F. oxysporum have been shown to reduce damage caused by Meloidogyne incognita in tomato roots (Dababat & Sikora, 2007). Plant protection depends on the population density of the BCA or more precisely on the ratio of pathogen versus protective strain (Larkin & Fravel, 1999). The protective strains are usually more effective when they are applied a few days before inoculation of the pathogen; and the protection is often improved when the strains are associated with rhizobacteria, especially fluorescent pseudomonads (Lemanceau & Alabouvette, 1993; Saman, 2009). The efficacy of biological control conferred by protective strains of F. oxysporum depends on environmental conditions, especially soil type (Larkin & Fravel, 2002). Despite the great number of scientific papers dealing with the efficacy of protective strains of F. oxysporum, the modes of action of theses strains have not been fully elucidated. The first studies focused on direct microbial antagonism and, as summarized above, later experiments demonstrated the role of competition for nutrients in soil and the rhizosphere (see section II, 1: Competition for nutrients ). The hypothesis of antibiosis was not examined, because that there was no obvious inhibition of growth of pathogenic strains confronted, in vitro, with protective strains. Today, the few teams interested in the protective activity of F. oxysporum are focusing on plant fungal interactions. In this article, we will review knowledge related to this field of research, focusing on similarities and differences between pathogenic and protective strains. 2. Virulence factors and host specificity determinants Pathogenic strains of F. oxysporum exhibit a complex pattern of host specificity, which is the basis of their classification into formae speciales and physiological races. The genetic factors regulating host specificity remain largely unknown, and the discovery of isolates infecting the same host whilst having independent evolutionary origins supports the hypothesis that host specificity may have arisen convergently (O Donnell et al., 1998). Attempts to elucidate conserved virulence factors and determinants for host specificity deserve attention, not only for our comprehension of pathogenesis in F. oxysporum, but also for the identification of the molecular traits governing the protective capability of F. oxysporum strains.

6 534 Review Tansley review New Several CWDEs, such as endopolygalacturonase, pectate lyase, xylanase, and subtilisin-like protease, are expressed during the infection process, but their role in pathogenesis is still a matter of debate. Transformation-mediated knockout of genes encoding CWDEs often has no effect on virulence, as multiple genes encoding similar and functionally redundant enzyme activities are likely to coexist in F. oxysporum, and may compensate for the loss of the disrupted genes. Many transcription factors regulate the expression of genes encoding secreted enzymes such as CWDEs, and understanding the involvement of these enzymes in virulence requires analysis of the role of transcriptional regulators (Di Pietro et al., 2003; Michielse & Rep, 2009). Similarly, it would be interesting to identify enzymes required for plant protection and to study the role of the genetic factors regulating their expression under different environmental conditions. Recently, new transcription factor genes were found to be implicated in virulence in F. oxysporum f. sp. melonis (Imazaki et al., 2007), f. sp. lycopersici (Michielse et al., 2009), and f. sp. phaseoli (Ramos et al., 2007). Dissecting the role of these transcriptional regulators would help to elucidate how pathogenic as well as protective strains of F. oxysporum interact with the plant. For example, sequencing the promoter of an ABC transporter characterized in F. oxysporum showed that the sequences differed between eight phytopathogenic and 11 biocontrol strains, suggesting that the ATP Binding Cassette (ABC) transporter Fo ABC1 may be regulated differentially between pathogenic and biocontrol strains of the fungus (Fravel et al., 2008). Signal transduction processes, and particularly cyclic adenosine monophosphate protein kinase A (camp-pka) and mitogen-activated protein kinase (MAPK) cascades, are implicated in the physiological, morphological, and metabolic adaptation of F. oxysporum to the plant environment. MAPK (fmk1) and G-protein subunits a and b are required for full pathogenicity, as the fungal infection of the plant is blocked by inactivating either the MAPK or the camp cascade (Di Pietro et al., 2003). One may wonder whether the soil-borne protective strains possess the same traits as pathogens in relation to plant colonization (see section III, 5: Colonization of the root surface and penetration into the root ). Proteomic analysis of the interaction between plants and F. oxysporum led to the identification of the first avirulence gene of F. oxysporum in xylem sap during colonization of tomato by F. oxysporum f. sp. lycopersici. Its product Secreted in Xylem 1 (Six1) is required for full virulence as well as for avirulence on tomato plants carrying the resistance gene I-3 (Rep et al., 2004). New avirulence genes, all encoding small proteins secreted in xylem during infection, share some common characteristics in the promoter region (Houterman et al., 2007) and are located at a virulence locus that is absent from strains belonging to other formae speciales as well as from the protective strain Fo47 (Van der Does et al., 2008). The genome of F. oxysporum f. sp. lycopersici strain 4287, which has been well characterized over the last 10 yr, was recently sequenced, along with the genomes of F. verticillioides, F. solani, and F. graminearum. The sequence is now available at the following web site institute.org/annotation/genome/fusarium_graminearum/multi Home.html. While only c. 50% of the F. solani genome can be aligned with the genomes of other sequenced Fusarium species, the vast majority (> 79%) of the F. graminearum, F. oxysporum and F. verticillioides genomes are syntenic regions with high sequence identity. The relatedness of these three genomes provides us with an unprecedented ability to predict genes, to determine orthologs, and to define regulatory sequences in these fungi. This powerful tool, coupled to the availability of genome-wide microarrays, is now ready to be fully exploited in order to develop the functional genomics of F. oxysporum, and to improve our understanding not only of its pathogenicity and host specificity traits, but also of its ability to compete in complex environments and to act as a biocontrol agent. 3. Role of fungal secondary metabolites in the protective capacity of F. oxysporum Strains of F. oxysporum, like strains belonging to other Fusarium species, produce many different secondary metabolites, whose role in relation to biological control has not been studied. Application of a culture filtrate of F. oxysporum has been shown to trigger the plant defense response in Arabidopsis thaliana cell cultures, but the chemical nature of the elicitors has not been characterized (Davies et al., 2006). Among the secondary metabolites, fusaric acid is the main toxin found in culture filtrates. It is produced by both pathogenic and nonpathogenic F. oxysporum and its role in pathogenicity is still controversial. Depending on the concentration, fusaric acid either participates in fungal pathogenicity by decreasing plant cell viability or acts as an elicitor of plant defense reactions (Bouizgarne et al., 2006). At nontoxic concentrations, fusaric acid was able to induce the synthesis of phytoalexins and also to induce rapid responses such as the production of reactive oxygen species and an increase in cytosolic calcium in A. thaliana cell suspensions. Necrosis-and-ethylene inducing protein (Nep1), isolated from the culture filtrate of F. oxysporum f. sp erythroxyli, was shown to induce ethylene biosynthesis and necrosis in a wide variety of Eudicots (Bailey, 1995). Nep1-like proteins play a dual role in plant pathogen interactions: Nep1 functions like a toxin by facilitating cell death as a component of diseases caused by necrotrophic plant pathogens, and when triggering the plant immune responses it resembles MAMPs such as bacterial flagellins (Bae et al., 2006; Qutob et al., 2006).

7 New Tansley review Review 535 Tomatinase is the name given to a family of extracellular enzymes produced by fungal pathogens attacking tomato. This enzyme detoxifies the tomato phytoanticipin a tomatine, which triggers programmed cell death in F. oxysporum f. sp lycopersici (Ito et al., 2007), and also suppresses the induced responses of the host (Ito et al., 2004). The ability to produce tomatinase is associated with the capacity of some nonpathogenic strains to colonize the stem of tomato (Ito et al., 2005). It appears that, depending on the concentration, secondary metabolites, like other pathogenicity factors that are produced by strains of F. oxysporum, are involved in both virulence and elicitation of plant defense reactions. However, the results described above do not provide a clear demonstration of the role of these molecules in the protective capacity of F. oxysporum. Knowing that MAMPs of fluorescent pseudomonads are involved in suppression of disease, it would be very interesting to consider the potential role of fusaric acid, Nep1 and tomatinase in relation to the biocontrol potential of F. oxysporum. 4. Recognition of strains of F. oxysporum by the plant Ion fluxes and the production of active oxygen species are the first physiological events that occur in the plant in response to infection by a microorganism. In many plant microbial models, these early physiological events enable us to distinguish a compatible interaction from an incompatible one. A few papers report the production of an oxidative burst by plant cells in response to inoculation with pathogenic strains of F. oxysporum (Davies et al., 2006; Mandal et al., 2008). There is a single study comparing the physiological events in flax (Linum usitatissimum) cells inoculated with either a protective (Fo47) or a pathogenic (Foln3) strain of F. oxysporum (Olivain et al., 2003). Immediately after inoculation, flax cells responded almost identically to the two strains, with a Ca 2+ influx, a first phase of H 2 O 2 production and an increase in the extracellular ph being detected. Two hours post-inoculation, there was a clear difference in the responses: a second phase of Ca 2+ influx and of H 2 O 2 production was detected only in the presence of the protective strain. Similarly, alkalinization of the extracellular medium was higher in the case of Fo47 compared with Foln3. Differential responses were also observed upon inoculation of tomato and muskmelon (Cucumis melo) cells with Fo47 or their respective pathogen. Moreover, when a pathogenic strain of F. oxysporum f. sp. melonis (Fom24) was applied to flax (a nonhost plant) cells, it induced the same responses as the protective strain Fo47. Taken together, these results allowed us to conclude that plant cells respond differently to inoculation with a plant pathogen and inoculation with a nonpathogenic strain. However, as the nonpathogenic strains used were also protective, it was not possible to ascertain whether this differential reaction Chemiluminescence (relative units) Time (mpi) post-inoculation Fig. 1 Kinetics of H 2 O 2 production induced in a flax cell suspension after inoculation with germinated microconidia per gram fresh weight of cells of the protective strain of Fusarium oxysporum, Fo47 (closed circles), or its mutants 94 (open circles) and 505 (open triangles), which are defective in their protective capacity. H 2 O 2 production was measured using the chemiluminescence of luminol and was monitored at 10-min intervals for 5 h. The uninoculated control is represented by closed triangles. mpi, minutes post inoculation. was correlated with their protective capacity. To answer this question, transposon-mediated mutagenesis was carried out to generate mutants impaired in their biocontrol capacity from strains Fo47 (Trouvelot, 2002) and Fom24 (L haridon et al., 2007). Results showed that it was not possible to differentiate the protective strain from the nonprotective one based on H 2 O 2 production as this varies independently of the characteristics of the strains (Fig. 1). 5. Colonization of the root surface and penetration into the root Early studies (Nash & Snyder,1967; Huisman, 1988; Gordon et al., 1989) using isolation techniques showed that pathogenic strains as well as native saprophytic strains of F. oxysporum are present at the root surface of many plant species growing in field soils. Indeed, the protective strains of F. oxysporum have the capacity to actively colonize the root surface. The use of transformed strains expressing different reporter genes allowed direct observation of fungal growth at the root surface. Most of the studies were performed in vitro, and the pattern of root colonization was dependent on the mode of inoculation. When the seedling root was dipped in a conidial suspension before being transferred to a plant nutrient solution, growing hyphae were observed everywhere on the root surface and, a few days post-inoculation, the whole root surface was covered by a dense network of hyphae (Olivain & Alabouvette, 1997). In some cases, hyphae were observed growing preferentially along cellular junctions at the root surface (Bolwerk et al., 2005). The pattern of root colonization differs between hydroponics and soil conditions: the root apices, which are heavily colonized in hydroponics, are not colonized in soil (Olivain et al., 2006; Nahalkova et al., 2008). In hydro-

8 536 Review Tansley review New 100 Disease incidence Time (dpi) post-inoculation Fig. 2 Protective capacities of two strains of Fusarium oxysporum f. sp. lycopersici (Fol 4287 and Fol007) and of their respective mutants fmk1 and N40, which are defective in their capacity to colonize the plant root. Flax cv. Viking was inoculated with the pathogenic strain (Foln3) and the other strains at respective concentrations of and microconidia per ml of substrate. The Fusarium wilt incidence is expressed as per cent of wilted plants: control (open squares), pathogenic strain Foln3 (closed squares), strain Fol 4287 (open circles), its mutant Dfmk1 (closed circles), strain Fol007 (open triangles) and its mutant N40 (closed triangles). The characteristics of these mutants have been described by Di Pietro et al. (2001) and Duyvesteijn et al. (2005), respectively. dpi, days post-inoculation. ponics, conidia are uniformly distributed on the root surface and the colonization is intensive at the apex where root exudates are abundant. In infested soil, the root grows faster than the hyphae arising from germinating conidia or chlamydospores attached to the soil particles. Germ tubes attached to soil particles grow toward other soil particles; they do not show any tropism toward the root surface. Penetration of F. oxysporum into the root is a crucial step in the biological cycle of this fungus. By using mutants of protective strains affected in their ability to penetrate the roots we demonstrated that colonization of the superficial cell layers is a necessary condition for the expression of the protective capacity of the strains (Fig. 2). However, the capacity to colonize is not sufficient to induce resistance in the plant. Mutants of Fo47 and Fom24, which colonize the root tissues to the same extent as the wild-type strains, have lost the capacity to protect the plant. Growing hyphae were observed penetrating root hairs and the epidermis either inter- or intracellularly. None of the studies investigating protective strains have reported appressoria, but some images recently acquired using fluorescent microscopy (Fig. 3) might be interpreted as showing specialized penetration structures. The penetration of the fungus into the root takes place during the 48 h following inoculation. Hyphae are observed penetrating the mature as well as the younger zones of the root. The penetration points are few in number, despite the intensive surface colonization observed a few Fig. 3 Hyphal swelling formed at the site of penetration of the protective strain Fo47 into the epidermis of a tomato root. Observations were made by conventional epifluorescent microscopy, 16 h after tomato root inoculation with germinated microconidia of Fo47. Hyphae were stained with the fluorescent probe WGA Alexa fluor 594. Image: C. Humbert and S. Aimé. days post-inoculation. In fact, surface colonization and internal colonization of the root are probably independent phenomena, one being related to the saprophytic growth and the other to the parasitic growth of the fungus inside the tissues. 6. Competition for surface colonization Both pathogenic and protective strains are able to colonize the root surface and to penetrate the root, following the same pattern. It was therefore proposed that competition for infection sites could occur. Mandeel & Baker (1991) postulated that the root surface has a finite number of infection sites which could be protected by increasing the inoculum density of the nonpathogenic strain. Unfortunately, the term infection sites is rather ill defined: do they correspond to sites of attachment, colonization, or penetration? Results obtained by Recorbet & Alabouvette (1997) showed that conidia rapidly bind to the root surface, the attachment being accurately described in terms of the Langmuir adsorption isotherm, indicating the existence of a single class of specific high-affinity adherence sites on the root surface. As marked strains are available, several studies have reexamined this hypothesis of competition for infection sites, with contrasting results. Bolwerk et al. (2005), working in a gnotobiotic sand system, concluded that the protective strain Fo47 uses the mechanism of competition for niches and nutrients as a biocontrol strategy against F. oxysporum f. sp. radicis-lycopersici. According to these authors, competition takes place for two types of niche: the attachment sites on root hairs and the cellular junctions of the root, where the fungi preferentially grow. Competition results in a

9 New Tansley review Review 537 reduction in the number of attachment and colonization sites available to the pathogen. By contrast, the use of probability models to calculate the relative efficiency of infection and competition for nutrients and infection sites by nonpathogenic and pathogenic strains of F. oxysporum led to the conclusion that competition for nutrients and or infection sites is an insignificant factor in biocontrol of Fusarium wilt diseases by non-pathogenic fusaria (Mandeel, 2007). In our opinion, it is not possible to refer to competition for infection sites, as evidence of the existence of specific infection sites is lacking. Competition for root colonization is more likely to be a result of competition for nutrients, as exudation, which provides nutrients for microbial development, is not uniformly distributed along the root surface, but tends to occur at root junctions and root tips. 7. Colonization of the root tissues The pattern of root colonization by the well-characterized protective strain Fo47 has been described in many plant species: eucalyptus (Eucalyptus sp.; Salerno et al., 2000), flax (Linum usitatissimum; Olivain et al., 2003), muskmelon (C. Olivain, unpublished), pea (Pisum sativum; Benhamou & Garand 2001), tomato (Bolwerk et al., 2005; Olivain et al., 2006; Nahalkova et al., 2008), and pepper (Capsicum annuum; J. Veloso, pers. comm.). The capacity of other protective strains such as Fo5a4 (Olivain & Alabouvette, 1997), CS20 (Fravel et al., 2003) and 70T01 (Bao & Lazarovitz, 2001) and of pathogenic strains applied to nonhost plant (Beswetherick & Bishop, 1993) to colonize plant roots was also described. In contrast to the intensive colonization of the root surface, fungal growth in the inner root was mainly restricted to the outer cell layers, including the epidermis, the hypodermis and rarely some layers of cortical cells. In a few cases, hyphae were observed colonizing the deep layers of the cortex. This intensive colonization could result from the high inoculum pressure used in these artificial in vitro experimental procedures. The host root cells reacted by forming structural barriers created by thickening and coiling of the cell walls which entrapped the fungus. Typical defense reactions such as wall appositions, intercellular plugging and intracellular osmiophilic deposits were frequently observed. In these reactive areas, dead cells with an aggregated cytoplasm could be observed. A coating material along the secondary wall accumulated in noninvaded xylem vessels. The chemical nature of these deposits was determined to be callose in the wall appositions (Benhamou & Garrand, 2001) and lignin inside the hypodermic cells (C. Olivain, unpublished). Moreover, the invading hyphae of Fo47 suffered from substantial alterations such as breakdown of the plasma membrane and disorganization of the cytoplasm (Benhamou & Garrand, 2001; Olivain et al., 2003). These plant responses to fungal invasion contributed to stopping the progression of the hyphae of protective strains which did not reach the stele. These observations led to the conclusion that the protective strains of F. oxysporum act as elicitors of plant defense reactions. By contrast, there is no unique pattern of root colonization by pathogenic strains of F. oxysporum. Pathogenic strains are responsible for different types of disease, such as crown and root rots and vascular wilt, and in some cases root rot may initiate or facilitate vascular colonization (Kroes et al., 1998). These different types of pathogenic F. oxysporum are quite dissimilar in the mechanisms by which they colonize their hosts and induce symptoms. The first steps of root colonization are similar to those described for the protective strains. The germ tubes colonize the root surface producing a dense net-like mycelium and, 24 h after inoculation, surface hyphae produce branches, which penetrate the root inter- or intracellularly. At this point, observations regarding the penetration and colonization processes diverge. Specialized penetration structures have sometimes been reported (Parry & Pegg, 1985; Farquhar & Peterson, 1989). Following penetration, fungal hyphae are observed both inter- and intracellularly, and either there is no apparent disorganization of the cortex (Bishop & Cooper, 1983; Olivain & Alabouvette, 1999) or there is evident destruction of the cell structure (Smith & Peterson, 1983; Benhamou et al., 1994; Baayen & Rijkenberg, 1999). In the first case, F. oxysporum grows within the host cells without inducing damage (Baayen & Förch, 2001) such as plasmolysis. Rodriguez-Galvez & Mendgen (1995) observed that the plasma membrane followed the finger-like projections of papillae without any evidence of degradation, demonstrating that the fungus infects living cells. In the second case, the fungus secretes phytotoxic compounds that facilitate penetration and subsequent colonization. These phytotoxins affected cells even when they were not directly in contact with the invading hyphae (Czymmek et al., 2007). In response to infection, the host cells show similar defense reactions to that described above for the protective strains. However, these defense reactions, being less intense and numerous, are not sufficient to stop the progression of the pathogen toward the stele and the xylem vessels. 8. Is cell death related to the biocontrol capacity? The fact that dead cells are observed in areas where the plant reacts to invasion by F. oxysporum strains led us to suggest an analogy between resistance induced by the protective F. oxysporum strains and the hypersensitive reaction, which is characterized by rapid death of the infected cells. The capacity of pathogenic and protective strains of F. oxysporum to induce cell death was studied in vitro using cell cultures of different plant species (flax, tomato and muskmelon). Starting 14 h post-inoculation, there was a greater percentage of dead cells in the presence of the biocontrol strains than in the presence of the pathogenic ones.

10 538 Review Tansley review New Cell death (%) Time (hpi) post-inoculation Fig. 4 Time course of cell death in cultured tomato cells inoculated with germinated microconidia per gram fresh weight of cells from either the protective strain Fom24 (closed circles) or the mutant 157 (open circles), which is defective in its protective capacity. The percentage of dead cells was estimated at the time of inoculation and then every 2 h from 14 to 24 h by staining with neutral red. Uninoculated cells are represented by closed triangles. hpi, hours post-inoculation. To determine whether this difference in the kinetics of cell death could be attributed to the protective capacity of the strains, the same types of experiment were conducted using mutants of biocontrol strains affected in their protective capacity. Whatever the strains and the plants considered, the percentage of dead cells increased more rapidly in the presence of the biocontrol strains than in the presence of their mutants affected in their protective capacity (Fig. 4). These results, which must be confirmed with a broader array of strains, allow one to distinguish a protective from a nonprotective strain of F. oxysporum, and lead to a fundamental question: is the cell death induced by different strains of F. oxysporum of the same nature? As we proposed the hypothesis that the protection induced by Fo47 could be compared to the HR induced by biotrophic fungi, and knowing that HR induces programmed cell death, it would be interesting to characterize the type of death induced by Fo47 and to compare it to the cell death induced by the pathogen in the host plant. 9. Production of plant antimicrobial compounds in response to F. oxysporum invasion The plant reacts to microbial infection by producing antimicrobial compounds of a diverse nature, the purpose of which is to suppress fungal growth or activity. Several studies have reported the susceptibility of strains of F. oxysporum to defense molecules produced by the plant. In vitro, phytoalexins express a fungistatic activity against pathogenic strains of F. oxysporum (Zhang et al., 1993; Stevenson et al., 1997). Moreover, phenolic compounds were shown to inhibit the production of cell walldegrading enzymes by the pathogen (Mandavia et al., 2003). In planta, Matta et al. (1969), indicated that phenolic compounds are involved in the resistance of tomato to Fusarium wilt. Since this early report, only two studies have considered the production of phenolic compounds in plants inoculated with protective strains of F. oxysporum. Applied to asparagus, biocontrol strains of F. oxysporum increased peroxidase (POX) and phenylalanine ammonialyase(pal) activities and lignin deposition (He et al., 2002). Similarly, biocontrol strains applied to tomato usually increase the content of ferulic, caffeic and vanillic acids (Panina et al., 2007). Based on these results it would be interesting to enlarge this approach to other strains and plant species and to consider the relative susceptibility of pathogenic and protective strains of F. oxysporum to phenolic compounds. Many studies have compared the accumulation or activity of PR proteins during the interaction between (i) a susceptible or a resistant plant cultivar and the same pathogenic strain of F. oxysporum, and (ii) a virulent or an avirulent race of a pathogenic F. oxysporum strain and the same plant cultivar. Only a few studies have dealt with the plant response to inoculation with a protective strain of F. oxysporum. In comparison to the noninoculated control, the protective strain Fo47 induced increased activity of chitinases, b-1,3-glucanase, b-1,4-glucosidase, peroxidase and PR1 in tomato (Fuchs et al.,1997; Duijff et al., 1998; Cachinero et al., 2002). In contrast, the activities of chitinase and b-1,3-glucanase isoforms appeared to be stimulated in tomato roots infected by F. oxysporum f. sp. lycopersici as compared with roots colonized with the protective strain (Recorbet et al., 1998). In a recent study, Aimé et al. (2008) compared the accumulation of five PR protein transcripts in tomato cell cultures and in roots and leaves of tomato plants challenged with either the protective strain Fo47 or the pathogenic strain Fol8. Results showed a lower expression of PR protein-encoding genes in the plants or cells inoculated with the protective strain. These results suggest that priming of the plant defense reaction could be one of the modes of action of strain Fo47. This hypothesis is consistent with the observation that peroxidase and PAL contents were higher in asparagus inoculated with protective than with pathogenic strains and increased more rapidly after inoculation with the pathogen (He et al., 2002). Similarly, in banana, several defense-related genes were first down-regulated upon inoculation with nonpathogenic F. oxysporum endophytes and then up-regulated after challenge with the nematode Radopholus similis (Paparu et al., 2007). These few studies dealing with the expression of plant defense reactions in response to inoculation with a protective strain of F. oxysporum do not provide any information about the metabolic pathways triggered by the protective strains. In contrast, most of the recent studies dealing with plant defense induced by pathogenic strains