Host-selective toxins (HSTs) are effectors produced by some necrotrophic pathogenic fungi that typically confer the ability to cause disease. Often, diseases caused by pathogens that produce HSTs follow an inverse gene-for-gene model where toxin production is required for the ability to cause disease and a single locus in the host is responsible for toxin sensitivity and disease susceptibility. Pyrenophora tritici-repentis represents an ideal pathogen for studying the biological significance of such inverse gene-for-gene interactions, because it displays a complex race structure based on its production of multiple HSTs. Ptr ToxA and Ptr ToxB are two proteinaceous HSTs produced by P. tritici-repentis that are structurally unrelated and appear to evoke different host responses, yet each toxin confers the ability to cause disease. This review will summarize the current knowledge of how these two dissimilar HSTs display distinct modes of action, yet each confers pathogenicity to P. tritici-repentis.
Host–pathogen combinations displaying classical gene-for-gene characteristics have provided an essential genetic model for extensive molecular evaluation of disease. Due to these evaluations, considerable progress has been and is being made towards the understanding of the contribution of pathogen effectors to virulence in the absence of recognition (compatibility) or avirulence through dominant interactions with a corresponding host resistance gene product (incompatibility). However, far less progress has been made in identifying effectors that play a dominant role in pathogen virulence and in conditioning host susceptibility (compatibility). Yet, the identification and characterization of these factors are essential not only for a comprehensive understanding of plant host–pathogen interactions but also for providing insights into the evolution of the ability of a microorganism to cause disease. Host selective toxins (HSTs) produced by some fungal plant pathogens, including Pyrenophora tritici-repentis (Ptr), are ideal effectors with which to address these topics because their production is, in most cases, essential for compatibility of the pathogen with its host.
Ptr is the causal agent of tan spot of wheat (Triticum aestivum). Tan spot of wheat has been identified in major wheat-growing areas throughout the world and is a disease of significant economic importance. The Ptr–wheat interaction is the inverse of the ‘classical’ gene-for-gene interaction given that virulence on the part of the pathogen and disease susceptibility on the part of the host are the genetically dominant factors (Wolpert et al., 2002; Strelkov & Lamari, 2003). Ptr displays a complex race structure and at least eight races have been described and designated races 1 through 8 (Lamari et al., 2003). Currently, these race distinctions are attributed to the production of three HSTs (Ptr ToxA, B, and C); that is, each race is differentiated by the expression of one or a combination of these toxins. Sensitivity of wheat to a single HST appears to be sufficient for Ptr to cause disease, and in those cases tested, the expression of a HST in a nonpathogenic isolate renders the isolate pathogenic. Thus, each of these toxins can confer pathogenicity. This raises the question of whether these toxins all evoke the same or a similar host response and, if not, what is the common element in the host response that confers virulence to the pathogen? This review presents the progress made towards understanding how two proteinaceous HSTs produced by Ptr, Ptr ToxA (ToxA) and Ptr ToxB (ToxB), contribute to pathogen virulence and host susceptibility.
Ptr ToxA: a model protein in an ‘inverse’ gene-for-gene system
Characteristics of the ToxA gene and its protein
ToxA, which induces necrosis on ToxA-sensitive wheat cultivars, was the first proteinaceous HST to be described (Ballance et al., 1989; Tomas et al., 1990; Tuori et al., 1995). Isolation and transformation of the ToxA gene into a nonpathogenic race 4 isolate of Ptr was sufficient to render that isolate pathogenic on ToxA-sensitive wheat, confirming the importance of ToxA as a pathogenicity factor (Ciuffetti et al., 1997). ToxA is produced by races 1, 2, 7 and 8, either alone (race 2) or in combination with ToxC (race 1), ToxB (race 7) or both ToxB and ToxC (race 8) (Strelkov & Lamari, 2003). In all Ptr isolates thus far tested, ToxA is single copy (Ballance et al., 1996; Ciuffetti et al., 1997; Lamari et al., 2003). A highly similar 11-kb region of genomic DNA containing ToxA has been described in another wheat pathogenic fungus, Stagonospora nodorum (Friesen et al., 2006). Because of a larger number of polymorphisms in ToxA in S. nodorum (SnToxA, 13 haplotypes) as compared with Ptr (PtrToxA, three haplotypes (Ballance et al., 1996; Ciuffetti et al., 1997; Friesen et al., 2006)), and a fairly recent emergence of severe outbreaks of tan spot symptoms, it was suggested that S. nodorum donated this region of genomic DNA to Ptr in c.1941. However, the number and geographic distribution of ToxA-containing S. nodorum isolates tested (124 isolates from 8 geographic regions) was more representative than that of the Ptr isolates tested (49 N. American; 6 S. American; 1 European), in terms of the global spread of these pathogens. Also, more complex Ptr races were not represented, as North American isolates (88% of those tested) are predominantly race 1 (Ali & Francl, 2003). Additionally, the haplotypes of PtrToxA in regions where SnToxA shows the greatest diversity, that is Central Asia and South Africa (Friesen et al., 2006), were not determined. If horizontal gene transfer explains the presence of ToxA in both fungi, a more complete library of PtrToxA haplotypes would be justified in order to determine the direction of transfer. That ToxA might have existed in an as yet unidentified donor (Friesen et al., 2006) or progenitor is also possible. It will be important to study more isolates of both pathogens, as well as related species and other wheat pathogens, to gain a deeper understanding of how this gene came to be present in these two economically important fungi.
ToxA encodes a 23 amino acid (aa) signal peptide (Ballance et al., 1996; Ciuffetti et al., 1997) and a 4.3-kDa pro-domain required for proper folding, both of which are cleaved before secretion of the mature ToxA protein (13.2 kDa) (Fig. 1) (Tuori et al., 2000). The three-dimensional structure of ToxA has been determined to be a single domain protein with a β-barrel fold that presents a solvent-exposed loop containing an arginyl-glycyl-aspartic acid (RGD) motif (Fig. 1) (Sarma et al., 2005). ToxA resembles the fibronectin type III domain, through which fibronectin utilizes an RGD motif to interact with plasma membrane proteins. The RGD motif of ToxA, as well as other aa present in the solvent-exposed loop, is required for activity (Meinhardt et al., 2002; Manning et al., 2008).
The RGD motif of ToxA suggests an internalization mechanism
Several methods were used to show that ToxA can be detected within ToxA-sensitive but not ToxA-insensitive mesophyll cells (Manning & Ciuffetti, 2005), and receptor binding is likely to be the first step required for internalization (Fig. 2). The presence of an RGD motif suggests that ToxA interacts with an extracellular receptor, and the observation that mesophyll cells act as an affinity matrix suggests that these cells contain a high-affinity receptor that recognizes this motif. In fact, we can estimate the number of ToxA-binding receptors by treating ToxA-sensitive tissue and measuring the affected area. Using several concentrations of ToxA, the ToxA-binding receptor in each case was estimated to be relatively abundant at c. 3 × 106 per cell with a dissociation constant (Kd) for the interaction between ToxA and these sites of c. 1 nM. These data support high-affinity binding between ToxA and the ToxA receptor (Manning et al., 2008).
Co-treatment with ToxA and RGD-containing peptides results in a decrease in symptom development and a reduction of internalized ToxA (Manning et al., 2008). Additionally, the rate and amount of internalized toxin are adversely affected when certain aa within the RGD-containing, solvent-exposed loop are mutated. Mutagenesis of the RGD motif to arginyl-glycyl-glutamate (RGE) completely abolishes protein internalization and toxic activity. Taken together, these data provide strong evidence that the RGD-containing loop of ToxA is important for receptor binding and internalization into sensitive wheat cells. Interestingly, ToxA induces an increase in transcript levels of receptors, including lectin-like receptors (Pandelova et al., 2009), previously identified as a site of interaction for RGD-containing peptides and proteins (Gouget et al., 2006). Current efforts to identify the ToxA receptor take into account the importance of the RGD motif in protein uptake and symptom development.
Internalization and intracellular expression of ToxA result in cell death
How internalization of ToxA occurs is unknown, although microscopic (Manning & Ciuffetti, 2005) and transcriptional evidence (Pandelova et al., 2009), and experiments with endocytosis inhibitors (V. A. Manning & L. M. Ciuffetti, unpublished), point to endocytosis as the likely mechanism. In crystals used to determine ToxA monomeric structure, ToxA was present as a trimer with the shape of a three-blade pinwheel (Sarma et al., 2005). The significance of this structure is not known; however, in planta expression studies indicate that ToxA interacts with itself, suggesting that the trimeric conformation may be required for toxin function once the protein is internalized (V. A. Manning et al., unpublished). Additionally, if ToxA is expressed internally in ToxA-insensitive wheat, barley (Hordeum vulgare), Arabidopsis thaliana or tobacco (Nicotiana benthamiana), cell death occurs (Manning & Ciuffetti, 2005; Tai & Bragg, 2007; Manning et al., 2010). The most logical explanation of these data is that the Tsn1 locus, which governs sensitivity to ToxA (Gamba et al., 1998; Anderson et al., 1999), is involved in toxin uptake. A recent report suggests that Tsn1 encodes a protein containing ‘R gene-related protein kinase, nucleotide binding, and leucine-rich repeat domains’ (J. D. Faris et al., pers. comm.). Given this proposed structure of Tsn1, it is unclear if this protein has an extracellular domain that participates in the recognition and uptake of ToxA or if it functions cytoplasmically. In the latter scenario, Tsn1 could play some role in the endocytic process, allowing ToxA to accumulate within sensitive cells. Clearly, these findings present interesting questions that, once answered, are likely to reveal novel roles for the products of resistance-like genes in plant processes.
Chloroplasts as the site-of-action of ToxA
Microscopy studies indicate chloroplast localization of ToxA in sensitive wheat mesophyll cells (Fig. 2) (Manning & Ciuffetti, 2005) and several responses to ToxA treatment suggest that the interaction of ToxA with chloroplasts contributes to symptom development. ToxA-induced cell death is light-dependent (Manning & Ciuffetti, 2005) and results in reactive oxygen species (ROS) accumulation in the chloroplasts (Manning et al., 2009); inhibiting ROS accumulation prevents necrosis. There is a global decrease in photosystem (PS) I and II protein complexes if ToxA-induced ROS accumulation occurs. Low transcript levels of PSI- and PSII-related genes in response to ToxA treatment suggests that a reduction in protein production contributes to this global decrease (Pandelova et al., 2009). The increase of transcript levels of ROS-scavenging enzymes suggests that detoxification mechanisms are engaged, but that these are insufficient to prevent ROS accumulation and cell death (Fig. 2). If ToxA-induced ROS production is inhibited, a decrease in supercomplexes and light harvesting complex a/b protein 4 (Lhcb4) indicates that the chloroplasts are attempting to reduce photosynthetic capacity and an increase in intermediate complexes suggests that PS homeostasis is perturbed (Manning et al., 2009).
ToxA has been shown to interact with a chloroplast-localized protein, ToxABinding Protein 1 (ToxABP1) (Manning et al., 2007). Thylakoid formation 1 (Thf1), the well-conserved homolog of ToxABP1 in A. thaliana, has been implicated in PSII biogenesis/degradation (Keren et al., 2005), possibly through modulation of FtsH, a protease required for removal of photodamaged D1 protein (Zhang et al., 2009). Loss of function of Thf1 is therefore likely to result in oxidative stress and inhibition of chloroplast development. Some of the characteristics of ToxA-induced symptom development, that is, light-dependent ROS accumulation and perturbation of PSII homeostasis, are consistent with ToxA altering ToxABP1 function. In fact, silencing of ToxABP1 in wheat partially recapitulates the phenotype of internal expression of ToxA and also reduces the extent of necrosis induced by ToxA (Manning et al., 2010). However, because ToxABP1 silencing does not lead to a complete abrogation of ToxA-induced symptom development, it is likely that other interactions between ToxA and plant proteins are necessary for full toxin activity. We are currently investigating the possibility that ToxA might be directly involved in altering the function of a ROS-detoxifying enzyme in the chloroplasts.
The ‘inverse’ gene-for-gene system and effector-triggered immunity
A growing body of evidence suggests that ‘inverse’ gene-for-gene-induced susceptibility responses are similar to ‘classical’ gene-for-gene-induced defense-associated responses, commonly referred to as effector-triggered immunity (ETI) (Wolpert et al., 2002; Strelkov & Lamari, 2003; Jones & Dangl, 2006). This is consistent with the involvement of resistance (R) genes in the susceptibility response (Lorang et al., 2007). In support of this, microarray analysis of ToxA- vs mock-infiltrated ToxA-sensitive leaves provided strong evidence of an early up-regulation of defense-related genes including those encoding pathogenesis-related (PR) proteins, and enzymes involved in the phenylpropanoid, jasmonic acid and ethylene biosynthesis pathways (Fig. 2) (Pandelova et al., 2009). Early up-regulation of PR proteins and phenylalanine ammonia lyase (PAL), as demonstrated by quantitative polymerase chain reaction (qPCR), was reported for another ToxA-sensitive cultivar in response to ToxA (Adhikari et al., 2009). To probe the contribution of the Tsn1 locus in the ToxA response, Adhikari et al. (2009) used microarray analysis to compare the ToxA-induced responses between a ToxA-sensitive line and a ToxA-insensitive ethylmethane sulfonate (EMS) mutant of that same line. In this comparison, the majority of genes were regulated when cell death was far advanced and only a few defense response genes were up-regulated at early time-points. This is surprising as defense response genes are typically associated with early response to pathogen effectors (Abramavitch et al., 2006), and not with necrotic tissue. The interpretation of this study is hindered by the limited microarray data set provided and the lack of the availability of a complete data set in a public database. Application of improved annotations (Pandelova et al., 2009) to a complete data set might provide greater insight into the role of ETI in the Tsn1–ToxA interaction.
Ptr ToxB reveals the complexity of the ‘inverse’ gene-for-gene system
The ToxB open reading frame (ORF) present in isolates of races 5 and 6 is 261 bp in length and translates into an 87-aa pre-protein (Fig. 3) with a 23-aa signal peptide (Martinez et al., 2001; Strelkov & Lamari, 2003). The predicted molecular mass of the protein is 6.5 kDa (Martinez et al., 2001) and the protein contains four cysteine residues (Fig. 3, asterisk) that, based on mass spectrometry analysis, are predicted to be involved in two disulfide bonds (M. Figueroa Betts, unpublished). The ToxB homolog present in race 4, toxb, is 86% similar to ToxB (Martinez et al., 2004) and has been shown to be transcribed at low levels (Amaike et al., 2008). The two proteins are 81% identical, with toxb encoding a putative 23-aa signal peptide and mature protein that contains an additional amino acid compared to ToxB (Fig. 3). Despite the high degree of sequence similarity, heterologously expressed toxb appears to be inactive (Kim & Strelkov, 2007), providing evidence that these few sequence differences are responsible for its lack of activity. The ToxB homolog present in race 3 differs in the flanking upstream sequence, altering the putative start codon and increasing the putative secretory signal to 41 aa (Fig. 3). However, transcription does not appear to be affected and, if proteolytic processing occurs, it would produce a mature protein identical to ToxB (Strelkov et al., 2006). There are at least three possible explanations for the lack of chlorosis-inducing activity by race 3 isolates that contain this ToxB homolog: failure to be targeted to the secretory pathway as a result of the altered signal peptide; improper processing within the secretory pathway resulting in altered protein folding; and/or insufficient copy number.
Races 5 and 6 contain genes that encode identical ToxB proteins; however, the exact aa sequence of ToxB produced by races 7 and 8 is not known. Regardless, ToxB produced by all of these races induces chlorosis when infiltrated into ToxB-sensitive cultivars (Lamari et al., 2003). Chlorosis symptoms induced by native and heterologously expressed ToxB are typically detectible 48 h post infiltration and the severity of symptoms correlates with the amount of toxin used for infiltration (Strelkov et al., 1999; Kim & Strelkov, 2007). Purified ToxB is hydrophilic and stable when exposed to organics and heat. The stability of the protein is consistent with resistance to proteolytic enzymes and probably attributable to a tightly folded conformation (L. M. Ciuffetti et al., unpublished). We hypothesize that ToxB might act in the apoplast because of common attributes shared with apoplastic effectors, including small size, high cysteine content, and resistance to proteases.
The comparison of ToxB with other protein sequences and a search for putative functional motifs have not revealed obvious sequences that might contribute to toxic activity. However, studies involving chimeric proteins that contain combinations of coding regions from ToxB and the inactive homolog toxb are revealing important sequence features for biological activity (M. Figueroa Betts et al., unpublished). These data, in combination with ongoing studies to determine the tertiary structure of ToxB and toxb, will greatly enhance our understanding of the structural components necessary for activity of this proteinaceous toxin.
The onset of symptoms triggered by ToxB coincides with a decrease in chlorophyll a (Chla) and Chlb. Although the mode-of-action is not well understood, it has been shown that chlorosis is light-dependent and could involve the production of ROS (Strelkov et al., 1998). Recent microarray and biochemical studies support the role of ROS production and chloroplast involvement in ToxB-triggered responses (I. Pandelova et al., unpublished). Additionally, comparative microarray analysis between ToxA- and ToxB-induced responses illustrates that two distinctly different toxins evoke similar host defense responses. Furthermore, this study supports the idea that common defense responses are activated in both ‘classical’ and ‘inverse’ gene-for-gene interactions. Given that cell death is the ultimate outcome of both interactions and necrotrophic pathogens, including Ptr, exploit cell death for survival, it will be important to differentiate which of these responses are essential for the induction of cell death.
Ptr ToxB-containing isolates display virulence proportional to copy number
Unlike ToxA-containing isolates, where a single copy of the gene is sufficient to induce symptom development on ToxA-sensitive cultivars, ToxB-containing pathogenic races appear to require more than one copy to induce significant symptoms (Lamari et al., 2003; Martinez et al., 2004). Several ToxB loci have been cloned from two race 5 isolates collected in two distinct geographic regions, DW7 from North Dakota (Ali et al., 1999) and Alg3-24 from Eastern Algeria (Strelkov & Lamari, 2003). Alignments of six (out of an estimated nine) ToxB loci from DW7 showed high levels of conservation over > 900 nt, including identical ToxB ORFs (Martinez et al., 2004). Flanking these conserved regions are retrotransposon-like sequences. Additionally, some loci contain inversions, inverted repeats and conserved insertions. These observations suggest that unequal crossing over with similar sequences in the genome has led to amplification of ToxB-containing loci. Three ToxB loci from Alg3-24 display up to 100% identity with loci from DW7 over large regions, and include the same retrotransposon-like sequences, inverted repeats and insertions (Strelkov et al., 2006). Interestingly, ToxB loci present in a low-virulence race 5 isolate, 92-171R5 (Strelkov et al., 2002), appear to have significant differences in regions upstream of the ToxB ORF when compared to sequenced DW7 and Alg3-24 loci (Strelkov et al., 2006).
As indicated, copy number variation of ToxB is linked to virulence, with more virulent isolates having a greater estimated copy number than their less virulent counterparts (Strelkov et al., 2002, 2006; Martinez et al., 2004). Correlation between amounts of ToxB transcript produced and the number of ToxB copies (Strelkov et al., 2006; Amaike et al., 2008) suggests that increased production of ToxB could be responsible for the observed increase in virulence. To confirm that copy number contributes to virulence/pathogenicity, nonpathogenic race 4 isolates were transformed with ToxB, driven by the ToxA promoter (Ciuffetti et al., 1997), with methods effective in obtaining transformants with a wide range of copy numbers (Fig. 4). Copy number and virulence were compared to those of virulent race 5 and nonpathogenic race 4 isolates by Southern blot analysis and inoculation of ToxB-sensitive wheat, respectively. As previously demonstrated (Martinez et al., 2004), the race 5 isolate DW7 contains multiple copies of ToxB (Fig. 4a, right panel) and is highly virulent when inoculated onto the ToxB-sensitive cv Katepwa (Fig. 4a, left panel); whereas the race 4 isolate, SD20, contains one copy of the toxb gene that encodes an inactive protein (Fig. 4a, right panel) and is not virulent (Fig. 4a, left panel). Because toxb is present in race 4 as a single copy, it can be used as an internal control for comparison and classification of the transformed isolates into categories (H, high; M, medium; L, low) according to ToxB copy number (Fig. 4b, right panel; compare band intensities between toxb and ToxB). Inoculation of ToxB-containing transformants on a ToxB-sensitive cultivar indicated that the higher the copy number, the greater the virulence of the transformant (Fig. 4b, left panel). These findings further support that ToxB confers virulence in a copy number-dependent manner.
Another function for ToxB homologs?
ToxB transcript is present at early stages of infection in mycelia and conidia in both resistant and susceptible interactions. In addition, higher levels of ToxB transcript correlate with more rapid development of appressoria (Amaike et al., 2008). These observations have raised the question of whether ToxB plays an alternate role during early stages of infection.
Southern blot analysis has identified potential homologs of ToxB in Cochliobolus, Alternaria, and other members of the genus Pyrenophora (Andrie et al., 2008). In Pyrenophora bromi, the causal agent of brown leaf spot of smooth bromegrass (Bromus inermis), homologs of ToxB (PbToxB) are present in single or multiple copies depending on the isolate. Several PbToxB loci have been cloned from different P. bromi isolates and, unlike ToxB, the ORF is not necessarily identical within a single isolate or between isolates. The proteins are c. 80% similar to ToxB and toxb (Fig. 3) and the genes are expressed at low levels (R. M. Andrie & L. M. Ciuffetti, unpublished data). Although there is no obvious symptom development in leaves of bromegrass treated with heterologously expressed PbToxB proteins, these proteins induce chlorosis on ToxB-sensitive wheat cultivars. This suggests that an increase in copy number and/or gene expression could potentiate the pathogenicity of P. bromi on wheat. An additional role for ToxB and related proteins could explain the occurrence of various homologs of ToxB in other Dothideomycete plant pathogens and the maintenance of a homolog in nonpathogenic race 4 isolates (Andrie et al., 2008).
ToxA and ToxB are two structurally unrelated, necrotrophic effectors produced by Ptr that each condition compatibility in its interaction with wheat. ToxA displays a solvent-exposed loop, which is essential for function and mediates internalization into plant cells. ToxB has no discernable functional domains; however, it is resistant to proteases, probably as a consequence of a tightly folded conformation, suggesting that it could function as an apoplastic effector. ToxA and ToxB are not only structurally distinct, but also appear to promote host cell death through very different mechanisms. Host responses induced by ToxA are rapid and lead to necrosis, whereas changes induced by ToxB are slower and result in chlorosis. Data suggest that ToxA binds to a high-affinity receptor, is rapidly internalized, and alters photosystem homeostasis, which leads to an accumulation of ROS. Throughout this process, major transcriptional reprogramming occurs. In contrast, cell death induced by ToxB does not appear to involve a high-affinity receptor and studies thus far indicate that ToxB probably has an extracellular site-of-action. However, current results do support the possibility that, ultimately, both ToxA and ToxB impact chloroplast function and evoke a similar transcriptional response in the host. While highly speculative at this point, these data suggest an important role for altered chloroplast function in mediating pathogenicity of Ptr. Future studies clarifying the mechanism-of-action of all the HST effectors produced by this pathogen will define their primary role in contributing to the pathogenicity/virulence of this important pathogen.
This review is dedicated to the memory of Dr Lakhdar Lamari, University of Manitoba, Canada. Dr Lamari’s love for and dedication to the field of plant pathology and especially the Pyrenophora tritici-repentis/wheat interaction have had and will continue to have a significant and long-lasting impact; we are grateful for all of Dr Lamari’s outstanding contributions.
As a result of restrictions on the number of references allowed in this review, the authors would like to express their apologies in cases where references related to the primary data were omitted. In these cases a later publication of these authors that contained primary data citations was included. The authors would like to thank Dr Tom Wolpert for helpful discussion and manuscript review. Work performed in the Ciuffetti laboratory was funded by grants from the United States Department of Agriculture, CSREES and the National Science Foundation.