Phytotoxic secondary metabolites and peptides produced by plant pathogenic Dothideomycete fungi


  • Ioannis Stergiopoulos,

    1. Department of Plant Pathology, University of California Davis, Davis, CA, USA
    Search for more papers by this author
  • Jérôme Collemare,

    1. Laboratory of Phytopathology, Wageningen University and Research Centrum, Wageningen, The Netherlands
    2. Centre of Biosystems Genomics, Wageningen, The Netherlands
    Search for more papers by this author
  • Rahim Mehrabi,

    1. Laboratory of Phytopathology, Wageningen University and Research Centrum, Wageningen, The Netherlands
    2. Department of Wheat Breeding, Seed and Plant Improvement Institute, Karaj, Iran
    Search for more papers by this author
  • Pierre J.G.M. De Wit

    Corresponding author
    1. Laboratory of Phytopathology, Wageningen University and Research Centrum, Wageningen, The Netherlands
    2. Centre of Biosystems Genomics, Wageningen, The Netherlands
    • Department of Plant Pathology, University of California Davis, Davis, CA, USA
    Search for more papers by this author

Correspondence: Pierre J.G.M. De Wit, Radix Bulding 107, Droevendaalsesteeg 1, 6708 PB Wageningen, The Netherlands. Tel.: +31 317 483130; fax: +31 317 483412; e-mail:


Many necrotrophic plant pathogenic fungi belonging to the class of Dothideomycetes produce phytotoxic metabolites and peptides that are usually required for pathogenicity. Phytotoxins that affect a broad range of plant species are known as non-host-specific toxins (non-HSTs), whereas HSTs affect only a particular plant species or more often genotypes of that species. For pathogens producing HSTs, pathogenicity and host specificity are largely defined by the ability to produce the toxin, while plant susceptibility is dependent on the presence of the toxin target. Non-HSTs are not the main determinants of pathogenicity but contribute to virulence of the producing pathogen. Dothideomycetes are remarkable for the production of toxins, particularly HSTs because they are the only fungal species known so far to produce them. The synthesis, regulation, and mechanisms of action of the most important HSTs and non-HSTs will be discussed. Studies on the mode of action of HSTs have highlighted the induction of programed cell death (PCD) as an important mechanism. We discuss HST-induced PCD and the plant hypersensitive response upon recognition of avirulence factors that share common pathways. In this respect, although nucleotide-binding-site-leucine-rich repeat types of resistance proteins mediate resistance against biotrophs, they can also contribute to susceptibility toward necrotrophs.


Plant pathogenic fungi are commonly divided into biotrophic, hemi-biotrophic, and necrotrophic pathogens. Biotrophic pathogens thrive on living host tissues and often secrete effectors to suppress the host immune system, thus enabling them to cause disease (Stergiopoulos & de Wit, 2009). In contrast, necrotrophic fungal pathogens thrive on dead host tissue that has become necrotized by secreted cell wall degrading enzymes, phytotoxic secondary metabolites (SMs), peptides, and reactive oxygen species (ROS; Horbach et al., 2011). Finally, hemi-biotrophic pathogens initially establish a biotrophic relationship with their host and produce toxins only at later stages of the infection to kill the host cells and enable the pathogen to complete its life cycle on the dead tissue (Horbach et al., 2011). Secreted phytotoxins contribute to virulence or pathogenicity by disrupting host cells and inducing the release of nutrients to facilitate colonization of host tissues (Wolpert et al., 2002; Berestetskiy, 2008). Although the division of pathogenic fungi in these three lifestyles provides a general impression about their infection and interaction with their host, the distinction between biotrophs and hemi-biotrophs is often less clear.

Classification of toxins has been traditionally based on the spectrum of their biological activity rather than on their chemical properties or structural characteristics. In this way, two major classes are distinguished as follows: host-specific toxins (HSTs) and non-HSTs (Wolpert et al., 2002; Berestetskiy, 2008).

HSTs are biologically active only against a particular plant species and essentially determine the host range of the producing pathogen (Wolpert et al., 2002). Often, not all genotypes of a host plant species are sensitive to the toxin and similarly not all isolates of a pathogen species produce the toxin. Thus, the genetic basis of interaction between HSTs and plant genotypes resembles that of race-specific effectors produced by (hemi-)biotrophic plant pathogens and cognate resistant genes. The difference is that, unlike resistant genes, the presence of a HST target allele in a specific plant genotype confers susceptibility to the pathogen rather than resistance. This concept is generally known as the inverse gene-for-gene model, as opposed to the classical gene-for-gene model that holds for (hemi-)biotrophic pathogen-host interactions (Oliver & Solomon, 2010). Nearly all HSTs identified so far are produced by necrotrophic pathogens of the order of Pleosporales within the class of Dothideomycetes. Treatment of the plant genotype with the HST alone invokes the same disease symptoms as the pathogen itself, while disruption of the pathogen's ability to produce the toxin results in complete or severe loss of pathogenicity on the plant genotype (Wolpert et al., 2002).

Unlike HSTs, non-HSTs are biologically active against a broad spectrum of plant species (Berestetskiy, 2008) and are neither the sole determinants of pathogenicity nor define host specificity. Instead, they mainly contribute to virulence of the producing pathogen. A wide variety of non-HSTs implicated in disease development have been characterized from different Dothideomycete plant pathogenic fungi and they act on different physiological processes common in plants, including energy production (tentoxin: Alternaria spp.), lipid biosynthesis (cyperin: Ascochyta cypericola), actin polymerization (cytochalasins: many fungal species), production of ROS (cercosporins: Cercospora spp.) and various others (Berestetskiy, 2008).

This review focuses on HSTs and non-HSTs produced by species of Dothideomycetes, the largest and taxonomically most diverse class of Ascomycete fungi (Hane et al., 2011). The vast majority of HSTs and non-HSTs are low molecular weight SMs (Collemare & Lebrun, 2011) and different species of Dothideomycetes produce different profiles of SMs that are dispensable for growth. Their production is highly dependent on the physiological or morphological stage of the fungus and the conditions under which it is grown. SM non-HSTs and HSTs produced by fungi during infection of the host can be important determinants of pathogenicity or virulence. Based on their chemical structure, SMs are mostly classified as polyketides, nonribosomal peptides, alkaloids, terpenes, or metabolites of mixed biosynthetic origin (Collemare & Lebrun, 2011). Some HSTs are ribosome-produced peptides, as reported for the fungi Pyrenophora tritici-repentis (Ptr) and Stagonospora nodorum (Friesen et al.,2008ab).

SM HSTs produced by Dothideomycetes

SM HSTs are mainly produced by species of Cochliobolus and Alternaria. Those produced by Alternaria species are covered in another review to be published in this journal (Tsuge, et al. 2012; this issue). Several pathogenic species and races of Cochliobolus (anamorph state Bipolaris or Curvularia) are known to produce an array of structurally diverse HSTs that are primary virulence factors (Table 1; Fig. 1). HST producing species of Cochliobolus are so far exclusively pathogens of grasses. For most of these species the ability to produce the HST as well as sensitivity of host plants to the toxin are controlled by single Mendelian genes or genetic loci. However, the mode of action and the way HSTs manipulate the host to enable infections differ significantly from one producing species to another. On the other hand resistance toward the HST is conferred either by the lack of a target site of the toxin or by the ability to detoxify the HST. Here, we summarize what is known about the remarkable complexity and role during plant–microbe interactions of the major HSTs produced by species of Cochliobolus.

Figure 1.

Structures of SM HSTs produced by fungal species of Dothideomycetes.

Table 1. Biological activities of SM and proteinaceous HSTs produced by species of Dothideomycetes
ToxinSpeciesChemical structureMode of actionPlant target
HC-toxin Cochliobolus carbonum Cyclic tetrapeptideSuppression of host defensesHDACs
T-toxin Cochliobolus heterostrophus Linear polyketideDisruption of mitochondrial activityURF13: 13-kDa mitochondrial protein
PM-toxin Mycosphaerella zeae-maydis Linear polyketideDisruption of mitochondrial activityURF13: 13-kDa mitochondrial protein
Victorin Cochliobolus victoriae Cyclic chlorinated pentapeptideInduction of PCDLOV1: CC-NB-LRR disease resistance protein
HS-toxin Bipolaris sacchari Sesquiterpene glucosideDepolarization of plasma membraneUnknown
Phomalide Leptosphaeria maculans Cyclic pentadepsipeptideUnknownUnknown
DepsilairdinLeptosphaeria maculans Laird 2 and Mayfair 2Sesquiterpenic depsipeptideUnknownUnknown
Maculansins A and B Leptosphaeria maculans d-mannitol derived SMUnknownUnknown
PC-toxin Periconia circinata Peptidyl chlorinated polyketideInduction of PCDUnknown
SV-toxins I and II Stemphylium vesicarium UnknownInduction of PCDUnknown
PtrToxA Pyrenophora tritici-repentis 13.2-kDa proteinTsn1-mediated induction of PCDChloroplasts, ToxABP1
PtrToxB Pyrenophora tritici-repentis 6.5-kDa proteinInduction of PCDProbably chloroplasts
SnToxA Stagonospora nodorum 13.2-kDa proteinTsn1-mediated induction of PCDChloroplasts, ToxABP1
SnTox1 Stagonospora nodorum 10.3-kDa proteinSnn1-mediated induction of PCDProbably chloroplasts
SnTox2 Stagonospora nodorum 7–10-kDa proteinSnn2-mediated induction of PCDProbably chloroplasts
SnTox3 Stagonospora nodorum 25.8-kDa proteinSnn3-mediated induction of PCDUnknown
SnTox4 Stagonospora nodorum 10.3-kDa proteinSnn4-mediated induction of PCDProbably chloroplasts

HC-toxin is a cyclic tetrapeptide produced by C. carborum, a fungus causing Northern corn leaf spot and ear rot of maize (Walton, 2006). The disease emerged in the first half of the 20th century and HC-toxin isolated from Cochliobolus carbonum was identified as a major determinant of host specificity and virulence in the 1960s. HC-toxin is only produced by race 1 isolates of the fungus, which generally cause larger lesions on susceptible maize cultivars compared with the HC-toxin-deficient races 2 and 3 (Walton et al., 1997). HC-toxin is an important virulence factor and a major determinant of the disease on maize. Consistent with the emergence of the disease, C. carbonum race 1 isolates were only reported after the introduction of new corn varieties in the late 1930s that contained a susceptibility locus to the HC-toxin (Ullstrup & Brunson, 1947). The structure of HC-toxin is cyclo (d-Pro-l-Ala-d-Ala-l-Aeo), where Aeo stands for 2aminoepoxyoxo decanoic acid, a side chain with a carbonyl function that is indispensable for toxicity on susceptible hosts (Fig. 1; Walton et al., 1982). The mode of action of HC-toxin is based on inhibition of histone-deacetylases (HDACs) that are generally associated with repression of gene expression and consequently, their inhibition leads to H3 and H4 histone-hyperacetylation in maize and changes in transcript levels (Brosch et al., 1995). Thus, epigenetic transcriptional modification in presumably host defense genes is thought to be the primary role of HC-toxin in disease (Brosch et al., 1995). Host susceptibility to the toxin on the other hand is conferred by the lack of production of a carbonyl HC-toxin reductase (HCTR), an enzyme that detoxifies HC-toxin through chemical reduction of a side carbonyl group (Meeley & Walton, 1991; Walton, 2006). The NADPH-dependent HCTR in maize is encoded by the dominant Hm1 nuclear gene, the first plant resistance gene ever to be cloned (Johal & Briggs, 1992). In the absence of Hm1, a second dominant allele at an unlinked locus, Hm2, can also confer resistance to the fungus, albeit at intermediate levels as compared to Hm1 and only in adult plants. Molecular cloning and sequencing of both genes has indicated that Hm2 is a duplication of Hm1 (Johal & Briggs, 1992; Multani et al., 1998). Hm homologs appear to be evolutionary conserved in all monocots, suggesting that Hm-mediated resistance to C. carborum is of ancient origin and probably wide-spread in grasses (Han et al., 1997). However, two molecular events in maize, a transposon insertion in Hm1 and a deletion in Hm2, and subsequent inbreeding for hybrid seed production led to the breakdown of the natural mechanism of resistance in this crop and the emergence of disease (Multani et al., 1998). On the other hand, C. carbonum seems to employ two different mechanisms of self-protection against the HC-toxin. The first one is presumably based on active export of the toxin by the TOXA transporter, a member of the major facilitator superfamily of transporters (Pitkin et al., 1996). Additionally, C. carbonum displays two HDAC activities, one of them being toxin-resistant (Brosch et al., 2001).

T-toxin, formerly known as HMT-toxin or BMT-toxin, is synthesized by the necrotrophic fungus Cochliobolus heterostrophus, the causal agent of southern corn leaf blight (SCLB) in maize. Production of T-toxin is associated with high aggressiveness of the pathogen on maize cultivars that carry the ‘Texas male sterile cytoplasm’ (T-cms). Indeed, race T isolates that produce T-toxin are significantly more virulent on T-cms genotypes than the nontoxin-producing race O isolates (Tegtmeier et al., 1982). However, although it enhances virulence, T-toxin is not required for pathogenicity of C. heterostrophus race T strains on T-cms maize genotypes as toxin-deficient disruption mutants of the fungus are still able to cause disease symptoms (Walton, 1996). The existence of C. heterostrophus race T isolates became widely acknowledged in the early 1970s, when an SCLB epidemic ravaged cornfields across the US, causing what is considered by many as one of the most economically devastating diseases of the 20th century (Ullstrup, 1972). This epidemic was triggered by the vast monoculture of susceptible maize lines, after the introduction in the 1950s and 1960s of T-cms parental lines for the production of hybrid lines that did not necessitate hand or mechanical emasculation (Ullstrup, 1972). T-toxin is a linear polyketide containing between 35 and 49 carbons, although found predominately as a C41 derivative (Fig. 1; Kono et al., 1985). Its mode of action is based on conformational changes induced upon binding of the toxin on the gene product of T-urf13, the Texas cytoplasmic male sterility gene. T-urf13 is located on the mitochondrial chromosome of T-cytoplasm genotypes and encodes a 13 kDa oligomeric protein (URF13) that is assembled in the inner mitochondrial plasma membrane (Wise et al., 1987). URF13 causes premature degeneration of the mitochondria-rich tapetal cell layer of anthers at an early microspore stage, resulting in pollen abortion. Direct binding of T-toxin to URF13 induces pore formation and rapid permeabilization of the inner mitochondrial membrane that disrupts regular function and elicits a number of responses, including uncoupling of oxidative phosphorylation, leakage of small molecules, breakdown of electrochemical gradient and ATP-production, and membrane swelling (Levings et al., 1995). These T-toxin-dependent effects are also observed when URF13 is expressed in Escherichia coli, yeast, tobacco, and insects). Although the mode of action of T-toxin and related polyketides is solved, little is known about the function of URF13 in mitochondria, which originates from a series of recombination events involving other mitochondrial genes (Dewey et al., 1986). This origin raises the question of URF13 occurrence in nature, and consequently of the function of T-toxin for C. heterostrophus. Indeed, distribution of SM biosynthetic pathways is discontinuous among fungi, suggesting that maintenance of SM production in a given species is owing to evolutionary constraints. It is possible that T-toxin may play another role in the biology of C. heterostrophus rather than solely being a HST on maize. PM-toxin is an analog of T-toxin that is produced by the yellow leaf blight pathogen of corn Mycosphaerella zeae-maydis (recently renamed Pyronellaea zeae-maydis (Aveskamp et al., 2010). This fungus is pathogenic on T-cms corn mainly owing to the production PM-toxin. PM-toxin is a polyketide with a structure similar and its role in the pathogenicity of only M. zeae-maydis seems to be more important than T-toxin for C. heterostrophus as mutants deficient for PM-toxin production are no longer pathogenic on T-cms maize genotypes (Yun et al., 1998).

Victorin is a cyclic pentapeptide (Fig. 1) produced by the fungus Cochliobolus victoriae, the causal agent of the Victoria blight of oats. During the 1940s an epidemic swept across oat fields in the US that were planted with lines containing the ‘Victoria-type’ of resistance toward the crown rust fungus Puccinia coronata f. sp. avenae (Meehan & Murphy, 1946; Meehan, 1947). Resistance to crown rust is conferred by the dominant Pc-2 gene, a desirable trait that was extensively bred in oats and lead to the Victoria blight epidemic, as Pc-2 containing oats turned out to be susceptible to the new disease. Conversely, pathogenicity of C. victoriae on Pc-2 lines depended largely on the production of victorin (Macko et al., 1985). Victorin is a HST required for pathogenicity because only oat genotypes that are sensitive to victorin are susceptible to C. victoriae, treatment with victorin alone is able to reproduce disease symptoms on sensitive plants (Navarre & Wolpert, 1999; Lorang et al., 2004). Sensitivity of oat to victorin is controlled by a single dominant allele at the Vb locus. However, extensive genetic and mutational analysis failed to separate this locus from Pc-2 that confers resistance to rust, suggesting that Vb and Pc-2 were the same or closely linked loci (Mayama et al., 1995). This implied an unexpected relation between disease resistance genes that at the same time were conferring susceptibility to a necrotrophic pathogen. Nevertheless, physiological studies on the mode of action of victorin supported this connection. Indeed, in sensitive plants victorin triggers both the induction of host defense responses as well as responses typical of a programed cell death (PCD; Navarre & Wolpert, 1999). The mechanism underlying the induction of PCD by victorin is unclear owing to contradictory reports that suggest the target of victorin to be either in mitochondria, where it would inhibit two components of the glycine decarboxylase complex (GDC; Navarre & Wolpert, 1995), or at the plasma membrane, as the PCD occurs before victorin enters the cell and is independent of GDC binding (Tada et al., 2005). Understanding the mechanism of C. victoriae susceptibility was considerably advanced by the molecular dissection of victorin susceptibility in Arabidopsis thaliana, which led to the identification on chromosome 1 of the Locus Orchestrating Victorin effects 1, or LOV1 (Lorang et al., 2007). LOV1 encodes a coiled-coil-nucleotide-binding-site-leucine-rich repeat (CC-NBS-LRR) protein that is highly similar to intracellular resistance proteins of the RPP8 family. Although perception of victorin by LOV1 induces several disease resistance-associated responses, including rapid induction of the pathogenicity-related gene PR-1 and production of the phytoalexin camalexin, LOV1-mediated susceptibility does not require salicylic acid, jasmonic acid, or ethylene-mediated signaling pathways (Lorang et al., 2007). In a similar way, RAR1, SGT1b and HSP90, which are required by many resistance genes for the expression of complete resistance, do not contribute to LOV1-mediated susceptibility (Lorang et al., 2007). Interestingly, a thioredoxin is required for victorin sensitivity (Sweat & Wolpert, 2007), which is reminiscent of the involvement of the CITRX thioredoxin in tomato Cf-dependent resistance responses (Rivas et al., 2004). These results suggest that victorin may actually target a typical resistance protein that is normally involved in the recognition of a naturally occurring pathogen of A. thaliana (Sweat et al., 2008). Such a target may also exist in oat, the natural host of C. victoriae, and could be the resistance gene located at the Pc-2 locus. In that case, the Vb/Pc-2 locus might be the first example of a gene involved in resistance against a biotroph (P. coronata f. sp. avenae)and susceptibility toward a necrotroph (C. victoriae).

HS-toxin, also known as helminthosporoside, is produced by Bipolaris sacchari (also known as H. saccahri or Drechslera sacchari), the causal agent of the eyespot disease of sugarcane. HS-toxin consists of four β-1,5-linked galactofuranose units linked with the asymmetric sesquiterpene core (C15H24O2) at carbon positions 2 and 13 (Fig. 1; Macko et al., 1983). HS-toxin causes lesions and electrolyte leakage only on sugarcane cultivars susceptible to B. sacchari and thus can be considered as a HST (Schroter et al., 1985). Membrane depolarization is also observed as a result of its activity, an effect that is reversible by increasing temperature at 30 °C or higher (Schroter et al., 1985). However, although a good correlation between sensitivity to the toxin and susceptibility to the pathogen is in general observed, other factors might be contributing to the disease as well, because in some cases, disease severity did not correlate with sensitivity to the toxin (Bournival et al., 1994). In addition to HS-toxin, B. sacchari can produce lower molecular weight toxin-analogs (toxoids) that lack one or more of the galactose residues and act antagonistically to the HS-toxin, thus protecting sensitive tissues against the toxin (Livingston & Scheffer, 1984). The biological relevance of the production of these analogs by the fungus is not understood yet, but their production could suggest that they interact or compete for the yet to be identified HS-toxin receptor.

The plant pathogenic fungus Leptosphaeria maculans (anamorph Phoma lingam) causes blackleg disease, one of the most devastating diseases of the oilseed crops canola and rapeseed (Howlett et al., 2001). The fungus is known to produce an array of SMs under diverse conditions and there is some evidence that a few of them might act as HSTs (Table 1; Fig. 1; Pedras & Yu, 2009). Phomalide was originally isolated from liquid minimal-medium cultures of L. maculans that were grown for 24–60 h but the toxin is known to be produced by the fungus during infection of susceptible hosts as well, suggesting a role during pathogenesis (Elliott et al., 2007). The toxin causes reddish, chlorotic and necrotic lesions on canola leaves that closely resemble those induced by the pathogen. These symptoms are only present on susceptible canola and rapeseed and no damage is observed on resistant Brassica species, such as brown mustard and white mustard, even when the toxin is applied at high concentrations (10−4 M; Ward et al., 1999). The strong correlation between susceptibility to the toxin and pathogenic range of the fungus suggests that phomalide is an important determinant of host specificity and virulence. The structure of phomalide has been recently resolved as a rare cyclic depsipeptide with three α-amino acids and two α-hydroxy acids (Fig. 1; Siodlak et al., 2011). Interestingly, the structural analog isophomalide that contains (Z)-dehydrobutyrine shows no phytotoxic activity against Brassica species, while the saturated dihydrophomalide causes chlorotic symptoms on the resistant brown mustard but not on canola, indicating that both the presence and configuration of the (E)-double bond is crucial for selective phytotoxicity (Siodlak et al., 2011). Another depsipeptide produced by L. maculans is depsilairdin. The toxin is isolated from Laird 2 and Mayfair 2, two L. maculans isolates that are avirulent on canola but virulent on brown mustard, thus exhibiting reverse pathogenicity as compared to the rest of the isolates of the fungus (Pedras et al., 2004). Treatment of brown mustard leaves with the toxin triggers the formation of chlorotic and necrotic spots, symptoms that resemble those caused by infections with the pathogen (Pedras et al., 2004). The chemical structure of depsilairdin has been resolved and it is shown to be a sesquiterpenic depsipeptide that contains a (2S,3S,4S)-3,4-dihydroxy-3-methylprolyl residue (Dhmp), an amino acid that has never been described before (Fig. 1; Pedras et al., 2004). Recently, the complete synthesis of depsilairdin and three depsilairdin analogs, in which the Dhmp residue was replaced with l-proline and cis- and trans-4-hydroxy-l-proline has been achieved. Analysis of phytotoxicities showed that the three analogs were no longer toxic to both brown mustard and canola, suggesting that the presence of Dhmp in the chemical structure of depsilairdin is required for its biological activity and host-selectivity to brown mustard (Ward & Pardeshi, 2010). A new set of selective phytotoxins, named maculansins A and B have also been recently isolated from virulent isolates of L. maculans (Pedras & Chumala, 2011). Notably, both maculansin A and B show higher toxicity on leaves of resistant brownmustard than on leaves of susceptible canola. Given the fact that Laird 2 and Mayfair 2 are pathogenic on brown mustard, it will be interesting to examine whether these two isolates can also produce maculansins A and B and whether these two metabolites contribute to virulence on brown mustard or relate to a host jump. Structurally, these toxins are derivatives of mannitol containing the unusual chromophore 2-isocyano-3-methyl-2-butenoyl (Fig. 1). Together with brassicicolin, a host selective phytotoxic metabolite produced by Alternaria brassicicola, these two metabolites seem to constitute a unique group of d-mannitol derived fungal metabolites with phytotoxic activities (Pedras & Chumala, 2011).

Less common pathogenic fungi also produce toxic SMs that might be determinants of a host range. PC-toxin, also known as peritoxin, is produced by the necrotrophic soil-borne fungus Periconia circinata, which causes milo disease, a root and crown rot of sorghum (Macko et al., 1992; Dunkle & Macko, 1995). Peritoxins (Fig. 1) are hybrid molecules consisting of a peptide and a chlorinated polyketide with equal toxicities and which are known as peritoxins A and B, respectively (Macko et al., 1992). Treatment of sorghum seedlings with peritoxin induces visual symptoms of disease only in susceptible genotypes (Dunkle & Macko, 1995). In addition, it causes a number of cytological and developmental alterations, including inhibition of mitosis, electrolyte leakage, vacuolar expansion, defects in secretory activity and endomembrane flow, dispersion of heterochromatin in the nuclei, autolysis and other hallmarks of apoptosis (Dunkle & Macko, 1995). Taken together, these observations suggest that the mode of action of peritoxin may be controlled by a disease resistance gene. Indeed, susceptibility to the pathogen and sensitivity to peritoxin in sorghum are both controlled by the single gene Pc-B that encodes a NBS–LRR type of resistance protein (Nagy & Bennetzen, 2008). To confirm the dual role of this type of proteins in resistance and susceptibility, it would be interesting to identify the biotrophic pathogen that is recognized by Pc-B. SV-toxins I and II are produced by Stemphylium vesicarium, the causal agent of brown spot of European pear, a disease of economic importance in Mediterranean pear-producing areas (Singh et al., 1999). Despite the broad host range of S. vesicarium, biological studies have shown that gain of pathogenicity on pears is owing to the ability of the fungus to produce SV-toxins, suggesting a parasitic differentiation in the species owing to the acquisition of HSTs. Indeed, sensitivity of pear cultivars to the pathogen is directly correlated to their susceptibility toward the two toxins. SV-toxins selectively induce vein necrosis on susceptible pear cultivars, with SV-toxin I exhibiting almost 10 × higher toxicity than SV-toxin II (Singh et al., 1999). On the cellular level, SV toxins induce electrolyte leakage from susceptible tissues and cause plasmalemmal invagination at both ends of plasmodesmata and other plasma membrane disorders that resemble those induced by the Alternaria AK, AF and ACT HSTs (Singh et al., 2000). These effects suggest the target site of SV-toxins to be located on the plasma membrane near plasmodesmata. However, as this target still remains elusive, it is not yet known whether the observed plasma membrane disorders are caused by direct binding of the SV-toxin to the membrane, or indirectly (Singh et al., 2000).

Biosynthesis and regulation of SM HSTs

The biosynthesis of SMs involves many enzymatic steps. In fungi, the genes encoding proteins that catalyze these different steps are frequently organized in clusters and thus are located at the same locus and are often co-regulated (Keller & Hohn, 1997). Such SM gene clusters usually contain a key enzyme [e.g. a polyketide synthase (PKS), a nonribosomal peptide synthetase (NRPS), a terpene cyclase (TC), or a dimethylallyl tryptophan synthase] responsible for the biosynthesis of the first intermediate. In addition, they may also contain genes encoding additional enzymes, transporters and transcription factors (Keller et al., 2005). Most studies to understand the biosynthesis of SM HSTs have so far focused on identifying the corresponding gene clusters. These efforts however, have only been successful for T-toxin and HC-toxin. There is no doubt that the availability of more fungal genomes will allow the characterization of additional SM biosynthetic pathways.

The biosynthesis of T-toxin in C. heterostrophus is controlled by the two unlinked loci TOX1A and TOX1B. These two loci behave like a single genetic locus because they are located on chromosomes that form a four-armed linkage group owing to a reciprocal translocation (Kodama et al., 1999). The sequence complexity of these loci generated by extensive genetic rearrangements has so far hampered the identification of genes involved in T-toxin production. However, the availability of the genome sequence of the fungus accelerated their delineation, and a total of nine genes (PKS1, PKS2, LAM1, OXI1 and TOX9 at the TOX1A locus; DEC1, RED1, RED2 and RED3, at the TOX1B locus) are now implicated in the production of T-toxin (Inderbitzin et al., 2010). All genes at the TOX1A and TOX1B loci are embedded in A/T rich regions and are flanked by multiple repeats (Inderbitzin et al., 2010). PKS1 and PKS2 encode fully reducing iterative type I PKSs that retain all the typical domains of this enzymatic family (Yang et al., 1996; Baker et al., 2006). However, the contribution of each PKS to the synthesis of the T-toxin polyketide backbone is still unknown. It is also unclear how the two polyketides are linked with each other, as they involve a unique carbon-carbon bond (Baker et al., 2006). Further on, DEC1 encodes an acetoacetate decarboxylase and its disruption completely abolishes the production of T-toxin (Rose et al., 2002). This enzyme is likely responsible for the decarboxylation of the penultimate precursors, which explains that all polyketides of the T-toxin family terminate with no carboxyl function. Single mutants of the remaining six genes show decreased T-toxin production, while double and triple mutants completely lose their ability to produce the toxin (Inderbitzin et al., 2010). These genes encode five dehydrogenases and one protein with unknown function (TOX9). The contribution of these accessory enzymes in T-toxin biosynthesis is still unknown. The transcript regulation of the TOX1 genes has not been studied yet, but production of T-toxin in vitro and during maize infection indicates that they are all expressed under such conditions. Yet, a transcription factor that could regulate the expression of these genes has not been identified at none of the two TOX1 loci. Thus, it is possible that genes located on other scaffolds are involved in T-toxin production and regulation. Recently, T-toxin production was shown to be under the epigenetic control of LaeA and VeA orthologs in C. heterostrophus, which regulate sporulation and secondary metabolism in Aspergillus nidulans (Wu et al., 2012). Indeed, a positive correlation between expression of TOX1 genes and consequently production of T-toxin is observed in the dark, in mutants that overexpress or are depleted of LaeA and VeA activity, respectively (Wu et al., 2012). These are the only regulators of T-toxin identified so far.

PM-toxin produced by M. zeae-maydis is structurally related to T-toxin and involves a gene cluster that contains homologs of TOX1 genes (PKS1, RED1 and RED2; Inderbitzin et al., 2010). Although the three genes are required for PM-toxin biosynthesis in M. zeae-maydis, this gene cluster does not provide any additional information on the biosynthesis of T-toxin.

Biosynthesis of HC-toxin is governed by the TOX2 locus in C. carbonum. The locus spreads over a 600-kb genomic region and contains at least seven genes in multiple copies (HTS1, TOXA, TOXC, TOXD, TOXE, TOXF and TOXG; Ahn et al., 2002). Like the TOX1 locus, the TOX2 locus appears to be subject to rearrangements, likely caused by the activity of transposable elements (Panaccione et al., 1996). Apart of TOXD, all other genes present in this locus are shown to be involved in the biosynthesis, export or regulation of HC-toxin production, and consequently are all required for pathogenicity on maize. HTS1 is considered the key gene in the biosynthesis of the HC-toxin. The gene encodes a 570-kDa NRPS, composed of four modules and each responsible for the addition of a single aminoacid in the peptide product (d-Pro, l-Ala, d-Ala and l-Aeo, respectively; Panaccione et al., 1992). The modular order in HTS1 fully corresponds to the amino acid order in the final peptide product. The epimerization of l-Pro to d-Pro is catalyzed by an epimerase domain present within the first module of HTS1. The second d-amino acid is epimerized by another enzyme, an alanine racemase, encoded by TOXG (Cheng & Walton, 2000). However, this amino acid is not required for HC-toxin activity as Δ to xg mutants are still able to produce a minor form of HC-toxin, in which alanine is replaced by glycine (Cheng & Walton, 2000). This is consistent with the finding that HC-toxin activity is owing to both the terminal epoxide and the vicinal carbonyl group of Aeo (Brosch et al., 1995). The biosynthesis of the Aeo chain is an important step in HC-toxin production and requires several other TOX2 genes. ToxF is a putative branched-chain amino acid transaminase that possibly aminates a precursor of Aeo (Cheng et al., 1999), while ToxC is a fatty acid synthase that is involved in the production of the decanoic acid backbone of Aeo (Cheng et al., 1999; Ahn et al., 2002). ToxA is an MFS transporter that is likely involved in the export and self-protection of the fungus against HC-toxin (Pitkin et al., 1996). All TOX2 genes are expressed in vitro and are induced during spore germination (Jones & Dunkle, 1995). Induction of transcription likely involves ToxE, as this bZIP transcription factor can bind to a 10-base motif found in promoters of all the ToxA genes (Pedley & Walton, 2001). However, regulation of this gene cluster is certainly more complex, as TOXE is not required for HTS1 expression. A gene cluster in Fusarium semitectum was characterized for the production of the cyclic tetrapeptide HDAC inhibitor apicidin (Jin et al., 2010). This gene cluster consists of homologs of HTS1, TOXA, TOXE and TOXF, for which the corresponding mutants display phenotypes similar to those of C. carbonum mutants in terms of toxin production. The apicidin gene cluster also contains a α-subunit fatty acid synthase, while ToxC is a β-subunit fatty acid synthase. Comparison of the TOX2 locus with the apicidin gene cluster provides new clues for HC-toxin biosynthesis. The TOX2 locus especially, may contain additional genes required for HC-toxin production, such as cytochrome P450 monooxygenases. Sequencing of the C. carbonum genome would be needed to fully understand HC-toxin biosynthesis and the evolution of the TOX2 locus/apicidin gene cluster that led to the production of two different HDAC inhibitors.

Proteinaceous HSTs produced by Dothideomycetes

In addition to SM HSTs, some necrotrophic plant pathogens secrete small toxic peptides that induce host necrosis or chlorosis, and can act as key determinants of fungal virulence or pathogenicity (Oliver & Solomon, 2010). These ribosome-synthesized HSTs act as effectors of necrotrophy (Friesen et al.,2008ab). Much of what is known today on the mode of action of proteinaceous HSTs and their interaction with host susceptibility genes comes from detailed studies on two important Dothideomycete necrotrophic wheat pathogens, Ptr (Ciuffetti et al., 2010) and S. nodorum (Table 2; Friesen & Faris, 2010).

Table 2. Biological activities of SM non-HSTs produced by species of Dothideomycetes
ToxinSpeciesChemical structureMode of actionPlant target
ToxinSpeciesChemical structureMode of actionPlant target
CercosporinCercospora spp.PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
ElsinochromesElsinoë spp.PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
Phleichrome Cladosporium phlei PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
Cladochrome Cladosporium cladosporioides PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
Calphostin Cladosporium cladosporioides PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
Shiraiachrome Shiraia bambusicola PerylenequinoneLight-dependent production of ROS, lipid peroxidationMembrane lipids
Rubellin and uredinorubellin Ramularia collo-cygni Anthrachinone derivativesLight-dependent production of ROS, fatty acid peroxidationMembrane lipids
Sirodesmin PL Leptosphaeria maculans ETPProduction of ROS through redox cyclingMultiple cellular targets
Phomalirazine Leptosphaeria maculans ETPProduction of ROS through redox cyclingMultiple cellular targets
Ophiobolin Cochliobolus heterostrophus SesquiterpeneInhibition of calmodulinCalmodulin
Bipolaroxin Bipolaris cynodontis SesquiterpeneUnknownUnknown
Solanapyrone Ascochyta rabiei DecalinNonspecific binding to proteins (likely membrane proteins)Multiple (membrane) proteins
Papyracillic acid Ascochyta agropyrina 1,6-Dioxaspiro[4,4]noneneUnknownUnknown
Phyllostictine Phyllosticta cirsii OxazatricycloalkenoneUnknownUnknown
Phyllostoxin Phyllosticta cirsii Pentasubstituted bicyclo-octatrienyl acetic acid esterUnknownUnknown
1-O-β-D (14-hydroxy-4, 12-eicosadienoyl)-glucoside Curvularia andropogonis  UnknownUnknown
CytochalasinSeveral DothiodeomycetesCytochalasinsBlocking of polymerization and elongation of actinActin
Herbarumin Phoma herbarum Nonelide lactoneInhibition of calmodulinCalmodulin
Pinolidoxin Ascochyta pinoides Nonelide lactoneInhibitor of PALPAL
Putaminoxin Phoma putaminum Nonelide lactoneUnknownUnknown
Stagonolide Stagonospora cirsii Nonelide lactoneLikely inhibition of photosynthesisUnknown
SphaeropsidinSphaeropsis sapinea f. sp. cupressi; Diplodia mutilaPimarane diterpeneUnknownUnknown
Sphaeropsidone and episphaeropsidoneSphaeropsis sapinea f. sp. cupressiDimedone methyl ethersUnknownUnknown

Ptr is the causal agent of tan spot of wheat. The pathogen is known to produce two well characterized proteinaceous HSTs, namely PtrToxA and PtrToxB, which interact in a highly specific manner with the host plant and are important determinants of pathogenicity on wheat (Ciuffetti et al., 2010). PtrToxA and PtrToxB induce necrotic and chlorotic symptoms, respectively, on leaves of susceptible wheat genotypes. Sensitivity to one of these toxins renders the host susceptible to the pathogen, while transfer of each individual toxin into a nonpathogenic isolate of the fungus can restore pathogenicity.

PtrToxA was the first proteinaceous HST ever described from a filamentous fungus (Ballance et al., 1989). The toxin induces extensive necrotic symptoms on the leaves of susceptible wheat genotypes, in a light-dependent manner. Although the protein was originally described in Ptr, several lines of evidence support that the gene was horizontally transferred to this fungus from S. nodorum, a necrotrophic pathogen of wheat that is frequently found in mixed infections with Ptr (Friesen et al., 2006). Indeed, while S. nodorum has been long recognized as one of the major and regularly occurring pathogens in this crop, infections by Ptr have only been reported since 1941. A highly homologous genomic region of 11 kb, containing the ToxA gene and a transposase is shared between the two fungi. ToxA from Ptr and S. nodorum share 99.7% similarity and a similar exon-intron structure, while the ITS sequences of the two fungi are only 83% similar. This high similarity and the absence of the gene in related species are consistent with an interspecific gene transfer (Friesen et al., 2006). Moreover, population analysis revealed larger heterogeneous distribution and higher allelic variation in the ToxA gene from natural populations of S. nodorum than from populations of Ptr. This suggests a founder effect in Ptr, supporting the hypothesis that the direction of transfer was most likely from S. nodorum to Ptr. Population analysis also revealed the presence of PtrToxA in c. 80% of a worldwide collection of isolates of the fungus (Friesen et al., 2006). However, not all races of Ptr can produce this toxin as, so far, the presence of PtrToxA has only been detected in races 1, 2, 7 and 8 (Ciuffetti et al., 2010). While virulence in the fungus is conferred by secretion of PtrToxA, sensitivity of wheat to the toxin is conferred by the presence of the dominant Tsn1 allele (Anderson et al., 1999). Wheat genotypes that lack Tsn1 are significantly less susceptible to races of Ptr that produce PtrToxA. Tsn1 has been recently cloned and it is shown to be a member of the NBS-LRR class of resistance genes. The encoded Tsn1 disease resistance-like protein contains a NBS, a LRR and a serine/threonine protein kinase domain, all of which are necessary for sensitivity to PtrToxA (Faris et al., 2010).

PtrToxA is the product of a single copy gene. The premature protein consists of a 22 amino acid signal peptide, followed by a 4.3 kDa pro-domain (amino acids 23–60) that is required for proper folding of the mature protein (Tuori et al., 2000). Both domains are cleaved prior to secretion of the 13.2 kDa mature ToxA protein. Crystallization of native ToxA showed that it adopts a single-domain structural conformation, with a β-barrel fold that presents a solvent-exposed loop containing an arginyl-glycyl-aspartic acid (RGD) cell attachment motif (Sarma et al., 2005). The RGD motif is embedded within a stretch of 10 amino acids (amino acids 137–146) that shares similarity with the mammalian protein vitronectin, a protein known to interact with integrin receptors present on the plasma membranes through its RGD motif (Manning et al., 2004). Mutational analysis showed that the RGD motif of PtrToxA as well as other solvent-exposed amino acid residues are required for the necrosis-inducing activity of the toxin (Meinhardt et al., 2002). Cytological studies revealed that PtrToxA is rapidly internalized into mesophyll cells of sensitive wheat cultivars, a process that is facilitated by the RGD motif (Manning et al., 2008). In addition, conversion of RGD motif to RGE completely abolishes PtrToxA internalization, toxin activity and development of symptoms (Meinhardt et al., 2002; Manning et al., 2004). These data support the hypothesis that the RGD motif of PtrToxA interacts with a high-affinity binding site present on wheat mesophyll cells to facilitate toxin internalization and eventually cell death. However, the identity of the receptor(s) at the cell membrane as well as the precise mechanism underlying PtrToxA internalization remains to be discovered. In this respect, transcriptional profiling has shown that treatment of sensitive wheat leaves with PtrToxA results in increased transcript levels of several receptors, including lectin-like receptors (Pandelova et al., 2009). Similar receptors have been previously identified to interact with RGD-containing peptides and proteins (Gouget et al., 2006). Consequently, Tsn1 has been suggested to be the receptor for PtrToxA, albeit the Tsn1 protein contains no obvious transmembrane domains and it is thus most likely located intracellularly. This might also explain why yeast two-hybrid experiments have failed to show a direct interaction between Tsn1 and PtrToxA. Instead, it is possible that Tsn1 monitors the condition of a yet to be identified receptor of PtrToxA and mediates PCD after interaction with PtrToxA (Faris et al., 2010), similar to what has been proposed as the guard hypothesis in biotrophic pathogen-host interactions (Jones & Dangl, 2006). Finally, based on cytological experiments, endocytosis has also been suggested as a mechanism for internalization as treatment with endocytosis inhibitors prevented PtrToxA internalization (Manning et al., 2008).

After internalization, PtrToxA localizes inside the chloroplasts where it interacts with a protein designated as ToxA binding protein 1 (ToxABP1; Manning et al., 2007). The interaction between PtrToxA and ToxABP1 is mediated through the vitronectin-like loop of PtrToxA but does not require the RGD motif present in this loop. Both transcript and protein levels of ToxABP1 are light-regulated, as is PtrToxA-induced PCD (Manning et al., 2007). Database searches revealed that ToxABP1 is conserved in several cyanobacteria and plants, including rice, potato and A. thaliana. It is shown that the ToxABP1 homolog in A. thaliana, Thf1, is localized in both chloroplasts and stromules (Wang et al., 2004). Stromules are extrusions of plastids connecting them to other membrane systems, including the plasma membrane. Thus, it is possible that after internalization, ToxABP1 provides a route for PtrToxA from the plasma membrane to the chloroplasts, where it likely functions. Functional analysis of Thf1 and Psb29, the ToxABP1 homolog in the blue-green alga Synechocystis (Keren et al., 2005), revealed that knock-out mutants show a reduced growth rate, loss of chlorophyll, reduced photosystem II (PSII) activity with increasing light intensities, impaired chloroplast development, and a lack of organized thylakoids (Wang et al., 2004; Keren et al., 2005). Interestingly, treatment of sensitive wheat cells with PtrToxA also leads to the disruption of the photosynthetic machinery in a light-dependent manner, loss of chlorophyll and the disorganization of thylakoids (Manning et al., 2007). These similarities suggest that the PtrToxA mode of action is associated with disruption of the photosynthetic machinery by interfering with chloroplast function. Thus, it seems that PtrToxA induces alterations in photosystems I and II (PSI and PSII) by interacting with ToxABP1, leading to light-dependent accumulation of ROS in chloroplasts that disrupts their photosynthetic capacity and triggers PCD (Manning et al., 2009; Pandelova et al., 2009). This hypothesis is supported by the light-dependence of PCD activity of PtrToxA and the light-regulated transcription of ToxABP1 and Tsn1. Moreover, inhibition of ROS accumulation by the antioxidant N-acetyl cysteine abolishes the onset of PCD, suggesting that the observed light-depended accumulation of ROS in the chloroplasts is required for PtrToxA-induced PCD (Manning et al., 2009). However, silencing of ToxABP1 does not completely abolish the development of PtrToxA-induced PCD and it is likely that additional plant proteins interact with PtrToxA or are necessary for full toxin activity (Ciuffetti et al., 2010).

PtrToxB is the second proteinaceous HST from Ptr also known as chlorosis-inducing toxin. The toxin is predominantly produced by race 5 isolates of the fungus (Lamari et al., 2003). Unlike the single-copy PtrToxA, PtrToxB is present in multiple copies, ranging from two to up to ten in different races of the fungus (Martinez et al., 2004; Strelkov et al., 2006). Higher PtrToxB-copy number causes higher transcript and toxin production levels, which relates to increased symptom development and virulence on susceptible wheat plants (Amaike et al., 2008). For instance, high virulence of race 5 Ptr isolates positively correlates with high levels of PtrToxB transcripts produced by eight to ten copies of PtrToxB present in these isolates, whereas weakly pathogenic isolates of race 5 possess only two copies of PtrToxB (Strelkov et al., 2006). High transcript levels of PtrToxB also result in higher numbers and more rapid development of appressoria, suggesting that apart from its capacity to induce chlorosis in toxin-sensitive wheat, PtrToxB may also play a role in the basic pathogenicity of the fungus (Aboukhaddour et al., 2012). Interestingly, races 3 and 4 of the fungus that are nonpathogenic on PtrToxB-sensitive wheat genotypes appear to possess a single-copy of a nonfunctional homolog of PtrToxB, termed toxb. Although toxb is transcriptionally active, its expression levels are considerably lower as compared to wild-type isolates and the produced peptide is not biologically active and does not induce chlorosis in PtrToxB-sensitive wheat (Martinez et al., 2004; Amaike et al., 2008).

PtrToxB encodes a small secreted pre-protein of 87 amino acids, consisting of a 23 amino acid signal peptide that is cleaved to yield a mature peptide of 64 amino acids with a predicted molecular mass of 6.5 kDa. Four cysteine residues are present in the mature PtrToxB, which are predicted to form two disulfide-bridges and likely generate a tightly folded tertiary structure, rendering the peptide resistant to heat, organic solvents and proteases (Ciuffetti et al., 2010). Such characteristics are frequently found in secreted fungal apoplastic effectors (Stergiopoulos & de Wit, 2009), and it is hypothesized that PtrToxB may also be active in the apoplast (Ciuffetti et al., 2010). Apart from the four cysteine residues, there are no other structural features present in PtrToxB that could point to a particular mode of action. Domain-swaps between the N-terminal, C-terminal and central parts of the wild-type PtrToxB and the inactive toxb protein from race 4 were performed to determine the active domains involved in the induction of chlorosis (Betts et al., 2011). The two homologs share c. 86% and 81% similarity and identity, respectively, with most of the variation present in the mature part of the two proteins. The analysis showed that, although all of the PtrToxB protein domains are required for full biological activity, the N-terminal region has the greatest impact on PtrToxB chlorosis-inducing activity (Betts et al., 2011). Site-directed mutagenesis of the two amino acids in the N-terminal region that differ between PtrToxB and toxb showed both of them to be essential for full PtrToxB activity and the sequence surrounding the first cysteine residue necessary for proper folding of the protein (Betts et al., 2011). Thus, PtrToxB does not contain a single contiguous motif required to trigger chlorosis in sensitive wheat genotypes. PtrToxB-induced chlorosis is light-dependent and likely involves the production of ROS, in this way resembling the mode of action of PtrToxA (Strelkov et al., 1998). Similar to PtrToxA, it is proposed that PtrToxB causes a rapid disruption of the photosynthetic processes in sensitive wheat, leading to the generation of ROS and oxidative stress, chlorophyll photo-oxidation and the development of chlorosis (Kim et al., 2010). The early chlorosis developed by PtrToxB coincides with induced degradation of chlorophyll (Strelkov et al., 1998). Gene expression profiling and biochemical studies also support the role of ROS production and chloroplast involvement in PtrToxB-triggered chlorosis (Ciuffetti et al., 2010). Additionally, comparative microarray analysis between PtrToxA- and PtrToxB-induced responses illustrated that the two different toxins evoke similar host defense responses (Ciuffetti et al., 2010), suggesting that the cellular targets of PtrToxA and PtrToxB and their mode of actions are fairly similar.

Unlike other proteinaceous HSTs, PtrToxB homologs are present across a broad range of plant pathogenic Ascomycete fungi, including species of Cochliobolus, Alternaria, and sister-species of Pyrenophora like P. bromi (Andrie et al., 2008). Surprisingly, the taxonomic distribution of PtrToxB extends even outside the order of Pleosporales, to fungi such as Magnaporthe grisea, a member of the Sordariomycetes. This suggests that PtrToxB has been inherited from a common ancestral species that predates the split between Dothideomycetes and Sordariomycetes. Pyrenophora bromi is the causal agent of brown leaf spot of smooth bromegrass (Bromus inermis) and as in Ptr, different isolates of the fungus possess multiple copies of ToxB (Andrie et al., 2008). However, in contrast to PtrToxB, the ToxB alleles within and between isolates of P. bromi are not necessarily identical. Furthermore, P. bromi ToxB isoforms do not induce chlorosis on the pathogen's host smooth bromegrass but they are able to develop chlorosis in ToxB-sensitive wheat cultivars (Andrie et al., 2008).

The necrotrophic pathogen S. nodorum (anamorph Phaeosphaeria nodorum) causes S. nodorum glume blotch (SNB) on wheat, one of the major and economically important diseases in wheat growing areas of the world. To date, five HSTs have been identified from S. nodorum, including SnToxA, SnTox1, SnTox2, SnTox3 and SnTox4. The corresponding host susceptibility genes have been identified as Tsn1, Snn1, Snn2, Snn3 and Snn4, respectively (Friesen et al.,2008ab). Both additive and epistastatic effects have been reported when multiple compatible HST sensitivity gene interactions segregate in the same population (Friesen et al.,2008ab). In this respect, the ToxA-Tsn1, SnTox1-Snn1, SnTox2-Snn2, SnTox3-Snn3 and SnTox4-Snn4 interactions were shown to account for as much as 95%, 58%, 47%, 17% and 41% of the variation in SNB development, respectively (Friesen et al.,2008ab). Thus, each of the five HSTs characterized so far is contributing to the development of SNB. The SnToxA gene will not be discussed here as it is nearly identical to PtrToxA. As discussed previously the latter has been gained by Ptr, by horizontal gene transfer from S. nodorum (Friesen et al., 2006) and overall, SnToxA and PtrToxA have comparable modes of action.

SnTox1 was the first HST to be described from S. nodorum (Liu et al., 2004). The toxin induces necrosis on wheat genotypes carrying Snn1, a single dominant gene (Liu et al., 2004). Snn1 has not been cloned yet, although high-resolution mapping identified two EST-based markers as candidate genes that co-segregate with Snn1 and encode an NBS-LRR type of resistance protein (Reddy et al., 2008). SnTox1 encodes a protein of 117 amino acids, consisting of a predicted 17 amino acid signal peptide and a 100 amino acid cysteine-rich mature peptide with an estimated molecular mass of 10 kDa (Liu et al., 2012). Sixteen cysteine residues are present in the mature protein, a feature commonly found in apoplastic fungal effectors, suggesting that SnTox1 may also function in the apoplast (Liu et al., 2012). Interestingly, SnTox1 possesses a chitin-binding domain (ChtBD) with a conserved C-terminal chitin-binding motif that is mostly present in plant chitin-binding proteins (ChtBD1 or CBM18 superfamily). This domain is different from the ChtBD present in the apoplastic Avr4 effector protein (ChtBD2 or CBM14 superfamily) from the tomato pathogen Cladosporium fulvum (Stergiopoulos & de Wit, 2009). It is unknown whether the ChtBD1 present in SnTox1 is functional in binding chitin (Liu et al., 2012). A global population analysis involving 777 isolates of S. nodorum showed that SnTox1 is the predominant toxin gene present in these isolates (c. 85%) as compared to other S. nodorum toxin genes, such as SnToxA (c. 36%) and SnTox3 (c. 60%). However, SnTox1 was absent from all S. nodorum isolates that were collected from wild grasses and other taxonomically related species such as Ptr, P. teres, P. bromi and M. graminicola that were nonpathogenic on wheat (Liu et al., 2012). Analysis of sequence variation in SnTox1 identified at least 12 different nucleotide haplotypes, 11 of which encode different protein isoforms indicating that SnTox1 is under diversifying selection. However, none of the characterized nonsynonymous substitutions were present in the putative ChtBD as was reported for C. fulvum Avr4 (Stergiopoulos et al., 2007). Deletion of SnTox1 from virulent isolates of the fungus renders them nonpathogenic on Snn1 wheat lines while conversely, expression of the SnTox1 gene in nonpathogenic isolates renders them pathogenic on Snn1 wheat lines (Liu et al., 2012). Studies on the mode of action of SnTox1 suggest that it is similar to SnToxA. Indeed, SnTox1, SnToxA, as well as SnTox3, all have similar expression patterns with transcript abundance peaking at 3 days post inoculation (dpi), which correlates with the onset of necrosis development (Liu et al., 20092012; Faris et al., 2011). SnTox1-induced PCD and disease susceptibility in Snn1 lines are also light-dependent as is the case for the SnToxA-Tsn1 interaction (Friesen et al., 2007). Comparative analysis of the infection process between nonpathogenic and pathogenic isolates of S. nodorum has shown that SnTox1 plays a significant role during the initial penetration process. Most likely, SnTox1 is required to trigger Snn1-mediated PCD of epidermal cells thus providing the fungus with nutrients for further invasive growth (Liu et al., 2012).

SnTox2 is the third proteinaceous HST identified from S. nodorum. The toxin causes necrosis on wheat genotypes carrying the Snn2 sensitivity gene (Friesen et al., 2007). SnTox2 has only partially been characterized and its molecular mass has been estimated to be between 7 and 10 kDa. The toxin is protease-sensitive and similar to SnToxA and SnTox1 its ability to trigger PCD in susceptible wheat genotypes is light-dependent (Friesen et al., 2007). This suggests that all three toxins may recruit similar host components to induce PCD. A wheat recombinant inbred population segregating for Snn2 and Tsn1 (ToxA sensitivity) showed that the SnTox2–Snn2 interaction accounts for 47%, while the SnToxA–Tsn1 interaction accounts for 20% of the disease variation (Friesen & Faris, 2010). The two interactions were almost completely additive and together accounted for 65% of the disease variation. The additive effects conferred by multiple interactions of toxins with cognate sensitivity genes might explain why necrotrophic pathogens like S. nodorum produce so many different toxins (Friesen et al., 2007; Friesen & Faris, 2010).

SnTox3 induces PCD on wheat genotypes carrying the Snn3 toxin-sensitivity locus. SnTox3 is a single copy, intronless gene encoding a pre-protein of 230 amino acids with an estimated molecular mass of 26 kDa (Liu et al., 2009). After removal of a 20 amino acid signal peptide and a predicted pro-sequence, the mature SnTox3 contains six cysteine residues that most likely are involved in the formation of three disulfide bonds, as treatment of Pichia pastoris-produced SnTox3 with the reducing agent dithiothreitol (DTT) destroyed PCD activity. No homologs of SnTox3 have been found yet in other microbes, while the distribution of SnTox3 in natural populations of S. nodorum shows variable deletion frequencies in different populations from around the world and with a global presence frequency of c. 60.1%. In total, 11 nucleotide substitutions have been identified in SnTox3 from global populations of the fungus, which however, give rise to only four haplotypes of the produced protein (Liu et al., 2009). Expression profiling of SnTox3 showed that similar to SnToxA, transcription of this gene peaks at 3 dpi and decreases dramatically from 5 dpi onwards. This time frame coincides with hyphal proliferation at 3 dpi and the onset of symptom development, and at 5 dpi with transition from hyphal growth to asexual sporulation (Liu et al., 2009; Friesen & Faris, 2010). SnTox3-disruption caused decrease in virulence of the fungus on Snn3 wheat lines, whereas heterologous expression of SnTox3 in a nonpathogenic strain of S. nodorum was sufficient to render this strain virulent on Snn3 lines (Liu et al., 2009; Friesen & Faris, 2010). Together these results demonstrate that SnTox3 is an important virulence factor for the fungus on wheat lines carrying the Snn3 gene. Almost nothing is known about the mode of action of SnTox3 but the nature of its interaction with Snn3 seems to be different from the previously discussed SnToxA–Tsn1, SnTox1–Snn1, and SnTox2–Snn2 interactions. Indeed, SnTox3 activity on Snn3 wheat is not light-dependent and the toxin is less important for disease development, as it accounts for a maximum of only 17% of the disease variation (Friesen et al.,2008ab; Friesen & Faris, 2010). The relatively low impact of the SnTox3–Snn3 interaction can easily be masked by other key interactions, like the SnTox2–Snn2 and SnToxA–Tsn1 interactions. In fact, in the presence of the SnToxA–Tsn1 interaction the SnTox3–Snn3 interaction is practically undetectable (Friesen et al.,2008ab; Friesen & Faris, 2010).

SnTox4 is the most recently identified proteinaceous HST from S. nodorum. The toxin has only partially been characterized and its molecular mass is estimated between 10 and 30 kDa (Abeysekara et al., 2009). Sensitivity of wheat to SnTox4 is governed by Snn4, a single dominant gene (Abeysekara et al., 2009). Just like the previously described interactions between ToxA–Tsn1, SnTox1–Snn1, and SnTox2–Snn2, the SnTox4–Snn4 interaction is also light-dependent. However, this interaction results in the development of only mottled necrotic symptoms as compared to the severe and extensive necrosis caused by the other proteinaceous S. nodorum HSTs. The Snn4 locus is responsible for 41% of the phenotypic variation associated with susceptibility to SNB indicating that the SnTox4–Snn4 interaction is a significant factor in the development of the disease.

Non-HSTs produced by plant pathogenic Dothideomycetes

Dothideomycetes are well known for the production of HSTs, but they also produce many non-host-specific and highly toxic SMs (Table 2; Fig. 2). Perylenequinones, are major a class of photo-activated polyketide non-HSTs that include the well-studied cercosporin and elsinochrome toxins. These dark red pigments contain a highly oxidized pentacyclic core and harbor an axial chirality owing to their helical structure (Mulrooney et al., 2010). Cercosporin, produced by species of Cercospora was the first perylenequinone (Fig. 2) isolated from fungi (Daub, 1982). Since then, more than 20 perylenequinones have been characterized from various fungi, most of them belonging to the Dothideomycetes (Daub et al., 2005). In addition to being phytotoxic, some of these compounds have other interesting properties as well, including inhibition of protein kinase C (calphostins; Barak et al., 1991), antiviral activity (calphostin C; Hudson et al., 1997) and anti-proliferative effects on tumor cell lines (cercosporin and calphostin C; Vandenbogaerde et al., 1998). Owing to their photo-activation, perylenequinones are potent photosensitizing agents that could be used in photodynamic therapy (Olivo & Chin, 2006).

Figure 2.

Structures of SM non-HSTs produced by fungal species of Dothideomycetes.

Cercosporin was first isolated from the fungus Cercospora kikuchii, the causal agent of the ‘purple speck’ disease of soybean. Since then, the toxin has been isolated from many Cercospora species growing in vitro or in planta, indicating that plant pathogenic strains produce this compound during infection of the host. Cercosporin is phytotoxic on diverse plant species, including Nicotiana tabacum and Phaseolus vulgaris, implying that this is a non-HST (Fajola, 1978). In addition, cercosporin has antimicrobial properties against bacteria (Fajola, 1978). An important role of cercosporin in pathogenicity was demonstrated for C. kikuchii as mutants of the fungus impaired in cerscosporin production could not trigger lesions on soybean (Upchurch et al., 1991). Similarly, in Cercospora nicotianae, cercosporin deficient knock-out mutants trigger fewer necrotic lesions on tobacco as compared to the wild type (Choquer et al., 2005).

Elsinochromes (A, B, C and D) are perylenequinone (Fig. 2) pigments synthesized by many Elsinoë species, including Elsinoë fawcettii and Elsinoë australis that are important pathogens of various citrus species (Chung, 2011). Elsinochrome is toxic to citrus and tobacco cell suspensions (Liao & Chung,2008ab) and can be isolated from necrotic lesions of infected citrus leaves (Wang et al., 2009), suggesting an important role for pathogenicity. Although no correlation could be found between in vitro production levels of elsinochrome and pathogenicity on the host (Wang et al., 2009), this toxin is required for full virulence. Indeed, E. fawcettii knock-out mutants unable to produce elsinochrome trigger fewer necrotic lesions as compared to the wild type (Chung & Liao, 2008; Liao & Chung,2008ab). Finally, the bindweed bio-control fungus Stagonospora convolvuli are also reported to produce elsinochrome (Chung, 2011).

Several Dothideomycetes are known to produce photoactive compounds that are similar to cercosporin and elsinochrome, including phleichrome, calphostin, cladochrome and shiraiachrome (Arnone et al., 1988; Mulrooney et al., 2010). Cladosporium phlei is the causal agent of the purple spot disease of timothy and a producer of phleichrome, a natural product with both phytotoxic and antimicrobial properties (Fig. 2; Araki & Shimanuki, 1983). While the role of phleichrome in pathogenesis is not yet proven, it is shown that phleichrome production is induced by diketopiperadine compounds produced by Epichloe typhina, an endophytic fungus also found on timothy (Seto et al., 2005). Phleichrome is toxic to E. typhina in the presence of light, suggesting a role for this toxin in antagonistic interactions between the two fungi. The mild pathogen of wheat Cladosporium cladosporioides is known for its in vitro production of the perylenequinone compounds cladochromes and calphostins (Fig. 2; Williams et al., 2008). Previously, only cladochromes A and B were reported from Cladosporium cucumerinum during infection of cucumber (Arnone et al., 1988). These, and other cladochrome derivatives, can be isolated from in vitro growing cultures of the fungus only with the combined use of chemical inducers that modify chromatin conformation, suggesting that they play a role during pathogenesis. The chemical synthesis of calphostins and their cytotoxicity on animal cells have been extensively studied owing to their protein kinase C inhibitory and antitumor activities (Barak et al., 1991; Mulrooney et al., 2010). Finally, shiraiachrome A and B are produced by the bamboo pathogen Shiraia bambusicola (Fig. 2; Wang et al., 1992). Like other perylenequinones derivatives, these compounds show antitumor, antiviral and protein kinase C inhibitory activities (Wang et al., 1992). Shiraiachrome A also exhibits anti-angiogenic properties, which explains the use of S. bambusicola in the treatment of rheumatoid arthritis in Chinese folk medicine (Tong et al., 2004).

The mode of action of perylenequinones has been extensively studied, mainly owing to their light activated properties that could be exploited as photosensitizers in photodynamic therapy. All fungal perylenequinones share a basic 3,10-dihydroxy-4,9-perylenequinone chromophore that can absorb light-energy (Daub et al., 2005). This results in conversion of the compound to an electron-active triplet state that can be subsequently reduced to react with cellular molecules or oxygen to generate singlet oxygen (e.g. production of ROS, such as superoxide anion, hydrogen peroxide and hydroxyl radicals; Daub et al., 2005). The mode of action and cell damage invoked by perylenequinones have been extensively studied using mainly cercosporin as a model. This compound is not specifically toxic to any particular cell type, as it relies on the generation of toxic ROS products that are harmful to bacterial, fungal, plant and animal cells. The studies have shown that it induces lipid peroxidation of liposomes, microsomes, rat liver, membrane lipids of tobacco cells and pea internode mitochondria (Cavallini et al., 1979; Daub, 1982). In particular, the toxin destroys unsaturated acyl chains of lipids that consequently trigger extensive structural changes in membranes (Daub & Briggs, 1983). Lipid peroxidation occurs rapidly after application of the toxin, with electrolyte leakage observed within 2 min and cell damage within 45 min upon exposure to light. These effects however, can be inhibited by singlet oxygen quenchers, suggesting that lipid peroxidation involves the production of singlet oxygen (Cavallini et al., 1979; Daub & Briggs, 1983). Similarly to cercosporin, elsinochrome induces electrolyte leakage in a light-dependent manner (Liao & Chung,2008ab). In this case, the use of singlet oxygen quenchers confirmed the production of singlet oxygens upon exposure to light (Liao & Chung,2008ab). Finally, both cercosporin and elsinochrome are known to induce the production of superoxide in a light-dependent manner (Daub & Briggs, 1983; Liao & Chung,2008ab).

The basis of cell resistance to perylenequinones was studied using cercosporin as a model, and diverse mechanisms have been identified that play a role in resistance against this compound and consequently against Cercospora species as well. One mechanism seems to be reductive detoxification of cercosporin by cercosporin-resistant fungi (Daub et al., 2005). Indeed, these fungi show a higher reduction activity of diverse dyes with different redox potentials than cercosporin-sensitive fungi, such as Aspergillus flavus and Neurospora crassa (Daub et al., 1992). Accordingly, over-expression of the flavin adenine dinucleotide (FAD)-dependent pyridine nucleotide reductase Cpd1p in Saccharomyces cerevisiae resulted in resistance to cercosporin (Ververidis et al., 2001). This mechanism of detoxification that partially contributes to resistance to singlet oxygen-generating compounds is active locally around the hyphae and reversible, allowing the toxin to spontaneously re-oxidize into a form that is toxic for the plant cell (Daub et al., 2005). Transgenic tobacco plants expressing CDP1 are resistant to cercosporin, indicating that this gene could be used as a source of resistance against cercosporin-producing pathogens (Panagiotis et al., 2007). Also, several bacteria species have the ability to degrade cercosporin by removing one methoxyl group and adding one oxygen atom that generates an unstable green metabolite (xanosporic acid) with no activity against cercosporin-sensitive fungi and plant cells (Mitchell et al., 2002). Mutant screening and expression profiling in the presence of cercosporin suggested that an oxidoreductase is involved in cercosporin degradation by the bacterial species Xanthomonas campestris pv. zinniae (Taylor et al., 2006). However, expression of this oxidoreductase in noncercosporin-degrading bacteria did not result in cercosporin degradation, suggesting that other factors are required as well. Identification of the complete set of bacterial genes that degrade cercosporin is of interest for the genetic engineering of plants resistant to cercosporin-producing pathogens.

An additional mechanism of resistance to cercosporin that is commonly found in fungi is based on active efflux of the toxin. At least three membrane transporters, namely cercosporin facilitator protein (CFP), Ctb4 and Atr1 are implicated in transport of the toxin in various Cercospora species (Callahan et al., 1999; Choquer et al., 2007; Amnuaykanjanasin & Daub, 2009). The MFS transporter CFP is involved in secretion and self-protection against cercosporin in C. kikuchii (Callahan et al., 1999; Upchurch et al., 2001) and likely in C. nicotianae as well (Amnuaykanjanasin & Daub, 2009). Ctb4 is also an MFS transporter that is encoded by a gene located within the cercosporin biosynthesis gene cluster (Chen et al.,2007ab). However, although secretion of cercosporin and consequently virulence are impaired in Δctb4 deletion mutants, these mutants are still tolerant to the toxin (Choquer et al., 2007). Finally, Atr1 is an ABC transporter involved in cercosporin export and partial self-protection against it, as disruption and overexpression mutants of the fungus are slightly more sensitive or tolerant to the toxin, respectively (Amnuaykanjanasin & Daub, 2009). Further analyses of these three transporters are needed to better understand their roles in cercosporin export and resistance. CRG1 is a Zn(2)Cys(6) transcription factor that among others regulates the expression CFP, which explains its role in resistance against cercosporin. However, over-expression of CFP in a Δcrg1 mutant only results in a slight increase in tolerance to cercosporin (Amnuaykanjanasin & Daub, 2009), suggesting that several transporters, and other mechanisms as well, are involved in resistance to cercosporin.

A third mechanism used by C. nicotianae for self-protection against cercosporin is based on pyridoxine production (vitamin B6), a quencher of singlet oxygen. Pyridoxine biosynthesis is controlled by the SOR1 gene (Ehrenshaft et al., 1998), which is found in various perylenequinone producing fungi (Ehrenshaft et al., 1999). Pyridoxine quenches singlet oxygen at similar rates to other efficient antioxidants, as for example vitamins C and E (Ehrenshaft et al., 1999). In addition, pyridoxine is able to quench superoxide (Jain & Lim, 2001).

Plants also developed antioxidant defense mechanisms, including catalases and superoxide dismutases (SODs). Transcriptome analysis in maize after treatment with cercosporin showed significant increase in SOD gene expression (Williamson & Scandalios, 1992). These genes are certainly good candidates for engineering of plants with resistance to cercosporin. Indeed, transgenic sugar beets expressing SOD genes from tomato show increased resistance to cercosporin, and consequently to infection by Cercospora beticola (Tertivanidis et al., 2004).

Rubellins and uredinorubellins are also photoactivated compounds produced by the plant pathogenic fungi Ramularia collo-cygni and the hyper-parasite Ramularia uredinicola, respectively (Miethbauer et al., 2008). Although these molecules have a different structure as compared to perylenequinones (Fig. 2), they show a similar type of photo-activation, suggesting a similar mode of action. However, a role in virulence for both compounds is not yet known.

Another class of non-HSTs are the epipolythiodioxopiperazine (ETP) compounds. These SMs are characterized by a bridged disulfide piperazinedione six-membered ring, derived from a cyclic dipeptide, which seems to be required for their biological activity (Mullbacher et al., 1986). Their potential use as chemotherapeutic agents provoked investigations on their mode of action, and studies were mainly performed on gliotoxin, a toxin produced by the human opportunistic pathogen Aspergillus fumigatus. Nevertheless, a few ETP toxins have been isolated from Dothideomycete fungi as well. In this respect, sirodesmin PL produced by L. maculans is one of the most well-studied ETP phytotoxins (Fig. 2; Rouxel et al., 1988). Phytoxicity of sirodesmin PL suggests a role for this toxin in blackleg disease on Brassica napus (Rouxel et al., 1988). Knock-out mutants unable to produce sirodesmin do not completely abolish pathogenicity but show decreased virulence (Gardiner et al., 2004; Elliott et al., 2007). In addition to phytotoxicity, sirodesmin PL also shows antimicrobial activity toward gram-positive bacteria, which relies on the disulfide bridge structure of the toxin (Boudart, 1989). An analog of sirodesmin PL that likely originates from the same biosynthetic pathway is phomalirazine (Pedras et al., 1989), produced by the fungus P. lingam, showing that several bioactive compounds can originate from a single pathway.

The mode of action of ETPs like sirodesmin PL has been mainly studied using gliotoxin as a model. Toxicity of ETPs likely involves the sulfur atoms in the central diketothiopiperazine moiety that either inactivate proteins by conjugation with thiol residues or generate ROS (Gardiner et al., 2005). Several examples of covalent interaction with proteins have been reported for gliotoxin, but so far, none for the ETPs produced by plant pathogenic fungi. These toxins do not exhibit target specificity and thus can disrupt many different functions in the cell. For example, gliotoxin and other ETPs can act as immunosuppressors as the transcription factor NF-κB that is involved in immune responses is a target of these compounds (Pahl et al., 1996). The immunosuppressor activity may also be owing to inhibition of the chymotrypsin-like activity of the 20S proteasome (Kroll et al., 1999). ETPs can also trigger PCD (Waring et al., 1988), which appears dependent on caspase-3-like activity and production of ROS (Zhou et al., 2000). Mitochondrial proteins are also targeted by ETPs, as shown by the decreased rate of ATP synthesis owing to inhibition of ATP-synthases, and the induced alterations in their function is coupled to the apoptosis triggered by gliotoxin (Gardiner et al., 2005). However, induction of ROS by gliotoxin and other ETPs is not linked with apoptosis because quencher molecules do not inhibit PCD (Waring et al., 1995). Overall, stimulation of ROS appears to be of little importance to explain the PCD activity of ETPs, especially because the low concentration of gliotoxin required to induce apoptosis is not sufficient to induce an oxidative stress (Gardiner et al., 2005). Instead, PCD activity of ETPs could be explained by their effect on calcium homeostasis as gliotoxin interacts with thiol residues of a redox sensitive plasma membrane calcium channel (Hurne et al., 2002). This interaction in thymocytes results in an increase in calcium influx that can cause oxidative damage and subsequent necrosis. However, the role of all these processes in toxicity remains uncertain.

Many other toxic compounds are produced by Dothideomycete fungi, but their role in pathogenicity has not been proven or has not been studied in great detail. In addition to T-toxin for example, C. heterostrophus also produces ophiobolins that are sesquiterpene phytotoxins (Fig. 2) with potential herbicidal and antinematocidal properties (Tsipouras et al., 1996; Evidente et al., 2006). A mechanism for their antimicrobial and antitumor activity was proposed based on competition for binding to the membrane ivermectin receptor, which modulates an invertebrate specific glutamate-gated chloride channel (Tsipouras et al., 1996). Ophiobolins are also active against certain Zygomycetes as they inhibit sporangiospore germination of Mucor circinelloides and in addition cause morphological defects (Krizsan et al., 2010). The Bermuda grass pathogen Bipolaris cynodontis produces two phytotoxic sesquiterpenes named bipolaroxins (Fig. 2). The fatty acid ester of bipolaroxin is an inhibitor of the steroidal plant growth regulator brassinolide, but it is not yet known whether this activity is responsible for toxicity (Kim et al., 1998). Furthermore, the chickpea pathogen Ascochyta rabiei is known to produce the phytotoxic polyketides, solanapyrones (Fig. 2), which are suspected to play a role in microbial pathogenesis (Kasahara et al., 2010). Papyracillic acid (Fig. 2) produced by Ascochyta agropyrina is toxic to plants, including its host quack grass, which implies that it could be used as a control agent against this noxious perennial weed (Evidente et al., 2009). Phyllosticta cirsii is a pathogen of the noxious weed Cirsium arvense that produces four oxazatricycloalkenones, collectively known as phyllostictines. These are highly toxic on the host plant and show antimicrobial activity against Gram-positive bacteria as well (Evidente et al., 2008). In addition, P. cirsii produces phyllostoxin, a penta substituted bicyclo-octatrienyl acetic acid ester (Fig. 2) that is also highly toxic on its host plant (Evidente et al., 2008). Curvularia andropogonis is a pathogen of Java citronella and is known to produce a highly toxic 1-O-β-d (14-hydroxy-4, 12-eicosadienoyl)-glucoside in vitro and in planta (Alam et al., 1997). The purified toxin induces symptoms of the disease on the host plant but it is also toxic to other plants. Finally, many Dothiodeomycete fungi produce cytochalasins (Fig. 2), compounds that inhibit actin and could thus be important for plant infection, as plant cells reorganize actin filaments upon contact with pathogens (Scherlach et al., 2010).

Phoma herbarum is a ubiquitous saprophyte that can also be pathogenic on plants and animals. Its virulence is based on the production of diverse natural products, including the phytohormone gibberellin and the phytotoxic herbarumins (Rivero-Cruz et al., 2003). These SMs are nonelide lactones (Fig. 2) that inhibit germination and growth of Amaranthus hypochondriacus (Furstner et al., 2002). The compounds are analogs of the phytotoxins pinolidoxin and putaminoxin, produced by the Dothideomycetes, Ascochyta pinoides and Phoma putaminum, respectively (Evidente et al.,1998a ,b). Pinolidoxin and putaminoxin (Fig. 2) are toxic on diverse plant species, but show little or no antifungal activity and animal toxicity. Their phytotoxicity is correlated with the integrity of the nonelide ring, and the presence of hydoxyl groups and a propyl side chain (Evidente et al.,1998a ,b). Pinolidoxin is a potent inhibitor of phenylalanine ammonia lyase (PAL) activity, which plays a key role in plant defense responses (Furstner et al., 2002). Stagonolide is also a nonelide compound (Fig. 2) produced by Stagonospora cirsii, a pathogen of the weed C. arvense (Yuzikhin et al., 2007). This toxin shows selectivity as its host plant is more sensitive than the non-hosts tomato, pepper, and cucumber (Yuzikhin et al., 2007). In addition to causing necrosis on leaves, stagonolide is a strong inhibitor of root growth in seedlings of Asteraceae species, including C. arvense. All these nonelide compounds are potent herbicidal agents that could be used to control noxious weeds such as C. arvense.

Sphaeropsidins are phytotoxic diterpenes (Fig. 2) that are produced by Sphaeropsis sapinea f. sp. cupressi and Diplodia mutila, which cause canker diseases of Italian cypress. Tests on the biological activity of these toxins against several species of cypress as well as non-host species such as tomato, mung bean and oak, indicated that these are non-HSTs, able to cause symptoms both on host and non-host plants (Sparapano et al., 2004). In addition to being phytotoxic, sphaeropsidins are also toxic against diverse bacteria and fungi, including known pathogens of cypress such as Seiridium cardinale, Seiridium cupressi and Seiridium unicorpe (Evidente et al., 2011). The antifungal activity of sphaeropsidins could suggest that these toxins have a role in antagonistic activities against other fungal pathogens during infection of the host and thus could be used as potential biological control agents. Next to sphaeropsidins, S. sapinea f. sp. cupressi is also known to produce two phytotoxic dimedone methyl ethers, named sphaeropsidone and episphaeropsidone (Evidente et al.,1998a ,b). Different species of cypress exhibit variable levels of sensitivity to these two metabolites, although in general sphaeropsidone appears to be less phytotoxic than episphaeropsidone. Both metabolites are also able to induce symptoms on non-host plants such as tomato. Symptoms induced by sphaeropsidone and episphaeropsidone on plants are in general less severe than those induced by sphaeropsidins. Finally, both sphaeropsidone and episphaeropsidone show antimicrobial activity against a range of differed fungi, including the cypress pathogens S. cardinale and S. cupressi although, in all cases sphaeropsidone appeared to be less active than episphaeropsidone (Evidente et al.,1998a ,b).

Biosynthesis and regulation of SM non-HSTs

A gene cluster for the biosynthesis of cercosporin was identified in C. nicotianae that consists of eight genes, CTB1-to-CTB8 (Chen et al.,2007ab). CTB1 encodes a nonreducing type I PKS that is the key enzyme responsible for cercosporin biosynthesis. Δctb1 knock-out mutants are unable to produce cercosporin and induce fewer and smaller necrotic spots compared with the wild type (Choquer et al., 2005). CTB2 encodes an O-methyltransferase that is required for the biosynthesis of cercosporin and virulence of the fungus (Chen et al.,2007ab). However, Δctb2 deletion mutants appear to accumulate another yellow pigment that could correspond to a yet structurally uncharacterized precursor. CTB3 encodes an O-methyltransferase (N-terminus)/FAD-dependent monooxygenase (C-terminus) that is required for cercosporin production and full virulence and similarly to Δctb2, the Δctb3 deletion mutants also accumulate a yellow pigment that is not phytotoxic (Dekkers et al., 2007). Deletion of CTB5, CTB6 and CTB7 that encode putative FAD/FMN- or NADPH-dependent oxidoreductases also resulted in accumulation of a yellow pigment and absence of cercosporin production (Chen et al.,2007ab). Accordingly, the mutants triggered fewer lesions compared with the wild-type. These functional analyses determined genes involved in the production of cercosporin, while a biosynthetic pathway that involves oxidation, hydration and methylation steps has been proposed (Chen et al.,2007ab). However, none of the precursors accumulated in the different mutants has been characterized yet and experimental proof for the proposed route is still unavailable. CTB4 encodes an MFS transporter that is involved in the secretion of the toxin (Choquer et al., 2007). Indeed, deletion mutants are defective in both cercosporin biosynthesis and secretion, and accumulate a brown pigment. The lower amount of cercosporin suggests the existence of a feedback inhibition. Remarkably, CTB4 is not involved in self-protection because the mutants retained normal tolerance to cercosporin. The last gene found in the gene cluster was CTB8, which encodes a Zn(2)Cys(6) transcription factor (Chen et al.,2007ab). Δctb8 deletion mutants are not able to produce cercosporin and are less virulent than the wild type, mainly owing to the absence in expression of the other seven genes present in the CTB cluster. CTB8 appears to act downstream of the previously identified regulator CRG1 (Chung, 2003), as CTB8 (and obviously other CTB genes) is down-regulated in a crg1 disrupted mutant, while CRG1 expression is not affected in a Δctb8 mutant (Chen et al.,2007ab). The suggested feedback inhibition upon accumulation of cercosporin or its precursors has been confined in Δctb1, Δctb2 and Δctb3 mutant backgrounds that show down-regulation of genes in the cluster (Chen et al.,2007ab). However, this regulation appears more complex as transcription of CTB genes is not modified in the Δctb4 mutant (Choquer et al., 2007). Induction of CTB genes and production of cercosporin is affected by many environmental stimuli including light, carbon, nitrogen, pH, salts, ions and media composition (Choquer et al., 2005; Chen et al.,2007ab; Dekkers et al., 2007; You et al., 2008). Signaling in response to these diverse environmental stimuli likely involves global regulators such as VeA, but also a calcium/calmodulin and a MAP kinase pathway (Chung, 2003; Shim & Dunkle, 2003).

Elsinochrome toxin produced by Elsinoë species is also synthesized through a polyketide pathway. In E. fawcettii, EfPKS1 encodes a nonreducing PKS that is involved in the biosynthesis of elsinochrome, conidiation, and virulence on rough lemon (Liao & Chung,2008ab). Other genes adjacent to EfPKS1 likely play a role in elsinochrome biosynthesis. RDT1 encodes a reductase with strong similarity to the 1,3,8-trihydroxynaphthalene reductase that is involved in melanin biosynthesis (Chung & Liao, 2008). ECT1 encodes a putative membrane transporter that could be responsible for secretion and self-protection against the toxin. A role in elsinochrome production for PRF1, a gene encoding a putative prefoldin protein, is unclear. At this locus, other genes encoding hypothetical proteins (OXR1, EfHP1 to EfHP4) could be involved in elsinochrome biosynthesis but no function could be predicted (Chung & Liao, 2008). The last gene found at this locus is TSF1, which encodes a putative transcription factor with DNA binding motifs that are characteristic of C2H2 and Zn(2)Cys(6) zinc finger transcription factors (Chung & Liao, 2008). Similarly to Δefpks1 mutants, Δtsf1 deletion mutants were unable to produce elsinochrome and were altered in conidiation and pathogenicity on rough lemon. Accordingly, EfPKS1, RDT1, PRF1 and EfHP1 are down-regulated in these mutants, while expression of ECT1 and the other genes at this locus is unaltered. This suggests that only four genes are involved in the elsinochrome biosynthetic pathway. However, the regulation of this gene cluster seems to be more complex. Indeed, expression of all these genes is not completely correlated with elsinochrome production under nitrogen limitation, high pH or high glucose concentration (Chung & Liao, 2008). Like cercosporin, EfPKS1 and elsinochrome production seems also to be regulated by light (Liao & Chung,2008ab). Similarly in Δefpks1 mutants, RDT1, TSF1, PRF1 and ECT1 expression was nearly abolished, suggesting a feedback inhibition regulatory network (Chung & Liao, 2008). In addition to Tsf1, EfPKS1 (but not TSF1) is directly regulated by the EfSte12 transcription factor (Yang and Chung, 2010). EfSTE12 expression is not affected in both Δefpks1 and Δtsf1 deletion mutants. Although EfPKS1 is down-regulated in a Δefste12 deletion mutant, this strain is still able to produce a smaller amount of elsinochrome and is not affected in pathogenicity or conidiation (Yang & Chung, 2010). Overall, it seems that production of elsinochrome or intermediates may be involved in conidium differentiation, but the underlying genetic mechanism is still unclear.

The biosynthesis pathway of sirodemin PL in L. maculans is determined by a gene cluster that consists of 18 genes, including sirP a two-module NRPS-encoding gene (Gardiner et al., 2004). The proposed first step in the sirodemin biosynthetic pathway is the transfer of a dimethylallyl group to either the dipeptide cyclo-l-tyrosyl-l-serine or free l-tyrosine, which leads to the formation of phomamide (Gardiner et al., 2004). O-prenylation of l-tyrosine in the presence of dimethylallyl diphosphate is performed by the prenyltransferase enzyme encoded by SirD (Gardiner et al., 2004; Kremer & Li, 2010). Consistent with the predicted biosynthetic pathway, the gene cluster contains genes that encode a thioredoxin reductase (sirT), an acetyl transferase (sirH), two methyl transferases (sirM and sirN), three cytochrome P450 monooxygenases (sirB, sirC and sirE) and one oxidoreductase (sirO; Gardiner et al., 2004). Several genes encode proteins with no obvious function in sirodesmin biosynthesis. The glutathione S-transferase sirG and dipeptidase sirJ could be involved in self-protection during biosynthesis. Finally, sirI encodes an aminocyclopropane-1-carboxylic acid synthase and sirQ, sirR and sirS encode putative progesterone 5-b-reductases (Gardiner et al., 2004). Apart from sirP that whose disruption results in loss of sirodesmin production (Gardiner et al., 2004) and reduced pathogenicity (Elliott et al., 2007), all other genes from this gene cluster await functional analysis. The sirodesmin gene cluster also contains a gene encoding an ABC transporter (sirA). Disruption of this gene results in induction of sirP expression and increased accumulation of sirodesmin in the medium (Gardiner et al., 2005). However, this mutant is more sensitive to sirodesmin, suggesting that sirA is involved in self-protection against the toxin. SirZ encodes a Zn(2)Cys(6) transcription factor that regulates sirodesmin production (Fox et al., 2008). Indeed, sirZ-silenced mutants show down-regulation of genes from the gene cluster (sirD, sirP and sirH). Recently, the expression of the NRPS (sirP) and transcription factor encoding genes (sirZ) was shown to be controlled in response to amino acid availability by cpcA (Elliott et al., 2011). A sirodesmin-related compound is gliotoxin, produced by the human pathogen A. fumigatus. Remarkably, the gene clusters for ETP biosynthesis in both fungi share similarities, suggesting a common origin (Gardiner et al., 2004). An ETP-like gene cluster was also identified in Penicillium lilacinoechinulatum that share similarities with gene clusters in L. maculans and A. fumigatus (Fox & Howlett, 2008). However, although these gene clusters share similarities, the transcription factor PlgliZ is not able to complement a gliZ mutant in A. fumigatus (Fox & Howlett, 2008) and gliA provides resistance to gliotoxin only and not to sirodesmin (Gardiner et al., 2005).

Biosynthesis of the other fungal toxins described previously likely involves gene clusters. For example, a gene cluster involved in solanapyrone production was identified in Alternaria solani (Kasahara et al., 2010). This gene cluster consists of six genes (SOL1 to SOL6) that encode a PKS, an O-methyltransferase, a dehydrogenase, a transcription factor, a flavin-dependent oxidase, and a cytochrome P450 monooxygenase, respectively. All these activities are required in the proposed biosynthetic pathway leading to solanapyrone (Kasahara et al., 2010), although functional analysis remains to be done. Expressing SOL1 in Aspergillus oryzae and producing Sol5 in P. pastoris experimentally confirmed the activities of the PKS (SOL1) and oxidase (SOL5) enzymes. The PKS was shown to produce desmethylprosolanapyrone, but cannot catalyze any Diels–Alder cyclization. This activity was demonstrated for the purified oxidase that catalyzed the formation of solanapyrone A from prosolanapyrone II (Kasahara et al., 2010). Therefore, SOL5 encodes a diels-alderase enzyme involved in oxidation and cycloaddition of prosolanapyrone. As A. rabiei also produces solanapyrones, it is expected that a similar gene cluster is present in its genome.

Conclusions and future prospects

Fungal plant pathogens are divided into necrotrophs and biotrophs according to their feeding behavior. Biotrophs exhibit an intimate interaction with living host cells for feeding, while necrotrophs kill host cells and feed on the nutrients released from them. These different lifestyles require different molecular weaponry. Biotrophs secrete effector proteins that exploit host cells without killing them (Stergiopoulos & de Wit, 2009), whereas necrotrophs produce SM or proteinaceous HSTs that kill their host plant (Wolpert et al., 2002; Friesen et al.,2008ab). HSTs induce PCD in specific genotypes of the host that is mediated by dominant HST sensitivity genes. As a necrotrophic pathogen produces a HST to promote disease and the host contains a receptor that is required for HST sensitivity, the relation between necrotrophic pathogens and their host plants is often considered as the inverse of the classical gene-for-gene system, the hallmark of biotrophic plant–pathogen interactions (Stergiopoulos & de Wit, 2009). In a gene-for-gene system, recognition of a biotroph-derived effector protein by a host resistance receptor protein leads to a resistance response known as effector-triggered immunity (ETI) that results in PCD (Jones & Dangl, 2006). Thus, PCD provides resistance to biotrophs but susceptibility to necrotrophs (Fig. 3). It has been suggested that necrotrophic pathogens might have subverted resistance mechanisms acquired by plants to combat biotrophic pathogens (Friesen et al.,2008ab) and have recruited weaponry to survive the harsh conditions prevailing in dead or dying host tissue. This notion is supported by the fact that resistance genes active against biotrophs on one hand and susceptibility genes active against necrotrophs on the other, both encode similar types of resistance receptor proteins. Plant disease resistance against biotrophs is conferred by genes that encode extra- or intracellular NBS-LRR types of resistance receptor proteins (Maekawa et al., 2011). However, in recent years host genes encoding NBS-LRR proteins with a similar structure as disease resistance proteins have been reported to also mediate sensitivity to SM and proteinaceous HSTs. The A. thaliana LOV1 gene for example encodes aCC-NBS-LRR protein required for sensitivity to the HST victorin produced by C. victoriae (Lorang et al., 2007). Similarly, sensitivity to the HST PC-toxin produced by P. circinata is dependent on an NBS-LRR protein (Pc-B)in the host plant sorghum (Nagy & Bennetzen, 2008). Susceptibility of oat to C. victoriae producing victorin and resistance to the obligate biotroph P. coronata are correlated and it is suggested that the same CC-NBS-LRR protein may mediate PCD leading to susceptibility to C. victoriae and resistance to P. coronata, respectively. However, this hypothesis still needs additional experimental support. Whereas, LOV1 and Pc-B mediate recognition of a SM HST, Tsn1 mediates recognition of the proteinaceous HST ToxA produced by the necrotrophic pathogens Ptr (Sarma et al., 2005) and S. nodorum (Friesen et al.,2008ab). Also Tsn1 encodes an NBS-LRR receptor protein containing an S/T protein kinase that is required for ToxA sensitivity and disease susceptibility (Faris et al., 2010). This NBS-LRR protein is very similar to the RPG5 stem rust resistance protein active against P. graminis (Brueggeman et al., 2008). Thus, NBS-LRRs can be involved in HST-triggered susceptibility (ETS) to necrotrophs and ETI to biotrophs. Many NBS-LRR resistance proteins active against biotrophs do not interact directly with cognate fungal effectors but guard he host targets of these effectors. When effectors of biotrophic pathogens interact with host targets, this is sensed by the guarding resistance proteins that subsequently mediate PCD (Jones & Dangl, 2006). A similar guard model may also hold for necrotrophic plant–pathogen interactions. Tsn1 is an NBS-LRR-like host protein that does not interact directly with ToxA but might guard the host target of ToxA (Faris et al., 2010).

Figure 3.

(CC)-NBS-LRR proteins mediate resistance to (hemi-)biotrophs and susceptibility to necrotrophs. Some plant resistance proteins (R) and proteins that mediate sensitivity to HSTs, both belong to the (CC)-NBS-LRR class of proteins. Avirulence proteins (Avr) secreted by (hemi-)biotrophic pathogens as well as proteinaceous (e.g. ToxA) and secondary metabolite HSTs (e.g. victorin) that are secreted by necrotrophic pathogens, target specific plant proteins. According to the guard model, these targets are guarded by (CC)-NBS-LRR proteins. Binding and concomitant modulation of a host target by an Avr or a HST is sensed by the (CC)-NBS-LRR proteins that mediate PCD, thus providing resistance to a (hemi-)biotroph (green arrows) and susceptibility to a necrotroph (brown arrows). Both cases comply with a gene-for-gene type of interaction. Tsn1 likely senses the binding of the proteinaceous HST ToxA produced by Pyrenophora tritici-repentis, to its plant target and mediates PCD resulting in in plant susceptibility. The SM HST victorin produced by Cochliobolus victoriae likely targets host proteins that are monitored by LOV1 in Arabidopsis thaliana. In oat, the victorin susceptibility Vb locus is linked to the Pc-2 locus that confers resistance to the obligate biotrophic rust pathogen Puccinia coronata f. sp. avenae. Likely, this rust fungus secretes an Avr protein that is directly or indirectly recognized by Pc-2 that mediates PCD.

Thus, both proteinaceous effectors of biotrophic fungal pathogens and SMs or proteinaceous HSTs of necrotrophic fungal pathogens interact with host targets that are guarded by (CC)-NBS-LRR receptor proteins that mediate PCD. Necrotrophic, in contrast to (hemi-)biotrophic, pathogens are tolerant to harsh conditions prevalent in necrotic tissues (Lorang et al., 2007). Is (CC)-NBS-LRR-mediated PCD induced by HSTs different from PCD induced by biotrophic proteinaceous effectors and how is resistance then achieved against necrotrophs? Some plants resistant to necrotrophs might lack a target for a HST. Indeed such plants are resistant to infections by a necrotroph, as is for example the case for C. heterostrophus where absence of URF13 in the host leads to insensitivity to T-toxin. But for (hemi-)biotrophic and necrotrophic pathogens producing proteinaceous HSTs like ToxA and the SnToxs (Friesen et al.,2008ab; Ciuffetti et al., 2010) the situation might be different. Several common signaling pathways are associated with both ETS against necrotrophs and ETI against (hemi-)biotrophs. Thus, a classic gene-for-gene relationship proposed for biotrophs might also hold true for some (hemi-)biotrophs and some necrotrophs producing proteinaceous toxins, whereas the inverse gene-for-gene relationship might only hold true for a subgroup of necrotrops. Biotrophs and hemi-biotrophs cannot thrive on dead host tissue likely because they are not equipped with the appropriate weaponry to survive in necrotic tissues, whereas necrotrophs are equipped with the suited weaponry to colonize necrotic tissues. It is expected that future research will identify important biochemical components of this latter weaponry as more and more genomes of Dothideomycete fungi become available.