On the trail of a cereal killer: recent advances in Fusarium graminearum pathogenomics and host resistance

SUMMARY

The ascomycete fungal pathogen Fusarium graminearum (sexual stage: Gibberella zeae) causes the devastating head blight or scab disease on wheat and barley, and cob or ear rot disease on maize. Fusarium graminearum infection causes significant crop and quality losses. In addition to roles as virulence factors during pathogenesis, trichothecene mycotoxins (e.g. deoxynivalenol) produced by this pathogen constitute a significant threat to human and animal health if consumed in respective food or feed products. In the last few years, significant progress has been made towards a better understanding of the processes involved in F. graminearum pathogenesis, toxin biosynthesis and host resistance mechanisms through the use of high‐throughput genomic and phenomic technologies. In this article, we briefly review these new advances and also discuss how future research can contribute to the development of sustainable plant protection strategies against this important plant pathogen.

INTRODUCTION

The fungal pathogen Fusarium graminearum sensu stricto (O'Donnell et al., 2004) is the causative agent of Fusarium head blight (FHB) disease on wheat and barley and ear rot of maize. Major FHB epidemics that have occurred in the last two decades have established F. graminearum as a major global pathogen of cereals. The F. graminearum infection cycle starts with the macroconidia of the fungus overwintering in the soil or on plant debris. The widespread adoption of no‐till practices and stubble retention in recent years most probably accounts for the increased occurrence of FHB. The fungus produces fruiting bodies, called perithecia, and ascospores released from these perithecia infect cereal heads during flowering, especially under conducive climatic conditions (e.g. high humidity during flowering). Infection by the pathogen results in reduced grain yield as well as shrivelled and mycotoxin‐contaminated grains that are not suitable for human or animal consumption. Interestingly, F. graminearum is not just a specialized pathogen of the cereal head, but can also cause crown rot (Stephens et al., 2008) and root rot (Henkes et al., 2011; Lanoue et al., 2010), diseases which may be important in some environments.

Given the economic importance of this pathogen, F. graminearum and related Fusarium species are currently under intense investigation. This review aims to provide an overview of recent developments, particularly those derived from new genomics‐based technological approaches, and is intended to complement previously published reviews (Goswami and Kistler, 2004; Trail, 2009; Walter et al., 2010; Xu and Nicholson, 2009) covering earlier studies of F. graminearum infection biology.

NEW INSIGHTS INTO THE F. GRAMINEARUM COLONIZATION PROCESS

Biotrophy versus necrotrophy

Over the years, one contentious issue has been whether F. graminearum exhibits a biotrophic lifestyle during the initial stages of infection of floral tissues (Brown et al., 2010; Trail, 2009). A recent detailed microscopic study of the F. graminearum infection process in wheat heads (Brown et al., 2010) found no indication of necrotrophy at the initial stages of infection, as the advancing F. graminearum hyphae remained in the intercellular spaces of wheat rachis cells before subsequent intracellular growth, which presumably leads to subsequent cell death and necrosis (Fig. 1). Therefore, F. graminearum may be classified as a hemibiotrophic pathogen, although Brown et al. (2010) concluded that, during floral infection, F. graminearum probably displays a unique lifestyle different from previously described fungal lifestyles. Similarly, crown infection by F. graminearum also seems to follow a hemibiotrophic lifestyle, characterized by a relatively long symptomless period, followed by massive tissue necrosis and a rapid increase in fungal biomass (Stephens et al., 2008). To our knowledge, no microscopic investigation of root infection by F. graminearum has been reported (Henkes et al., 2011; Lanoue et al., 2010), but recent microscopic analysis of wheat root infection by the related pathogen Fusarium culmorum has shown the development of rapid necrosis that was visible within a few days of inoculation (Beccari et al., 2011).

Figure 1.

A simplified model of Fusarium graminearum–wheat interaction. Fusarium graminearum produces effectors, cell wall‐degrading enzymes (see Kikot et al., 2009 for a review) and toxins to colonize wheat. The plant responds to infection by producing defence‐related hormones, pathogenesis‐related (PR) proteins, reactive oxygen and proteins involved in cellular detoxification. Although some of the plant's responses help to restrict infection, others can potentially be exploited by the pathogen to aid pathogenicity. See text for further details. Fg, F. graminearum.

New penetration mechanisms

Recent research utilizing detached wheat glumes in an infection assay has revealed new penetration mechanisms by F. graminearum. Rittenour and Harris (2010) observed two different hyphal structures, called ‘subcuticular hyphae’ and ‘bulbous infection hyphae’, on detached and infected wheat glumes. Of these two structures, the bulbous infection hyphae were thought to be specific to infection, partly because the development of these hyphae was dependent on the fungal GPMK1 gene (Rittenour and Harris, 2010) encoding a mitogen‐activated protein kinase (MAPK). GPMK1 has been shown previously to be involved in F. graminearum pathogenicity (Jenczmionka et al., 2003; Urban et al., 2003). Fusarium graminearum can also form lobate appressoria and infection cushions during the infection of florets through caryopses and husks (Boenisch and Schäfer, 2011). However, the formation of these structures was not affected in a tri5 mutant deficient in trichothecene production (Boenisch and Schäfer, 2011), consistent with the notion that these toxins are not necessary for the initial infection of wheat heads. Specialized infection structures were not observed during crown infection by F. graminearum (Stephens et al., 2008), and Beccari et al. (2011) did not observe any bulbous hyphae during the infection of wheat roots by F. culmorum. This suggests that different penetration strategies may be used by these pathogens when infecting different tissues.

Tissue‐specific invasion strategies

The use of F. graminearum strains expressing reporter genes, combined with confocal microscopy, has been instrumental in expanding our understanding of the F. graminearum infection processes. For instance, using F. graminearum transformed with a TRI5 promoter‐GFP gene construct, Ilgen et al. (2009) showed that TRI5 expression, an indicator of the toxin deoxynivalenol (DON) biosynthesis, was activated in a tissue‐specific manner in inoculated wheat heads. The reporter gene driven by the TRI5 promoter was not expressed on anthers, which are the initial targets of the pathogen during floral infection. However, during the spread of the pathogen from inoculated to uninoculated spikelets (4–7 days post‐inoculation), extensive reporter gene expression was detected in the rachis node, suggesting that the rachis tissue elicits DON biosynthesis in the pathogen (Ilgen et al., 2009). This finding is consistent with earlier reports that, at least in wheat, the rachis constitutes a formidable barrier to the spread of F. graminearum, and that DON biosynthesis in the pathogen is required to overcome this major obstacle (Jansen et al., 2005; Maier et al., 2006).

Recent genetic data also support the importance of the rachis tissue in F. graminearum infection processes. The introduction of the long arm of chromosome 7E (7EL) from Thinopyrum elongatum, a wild relative of wheat, to hexaploid Chinese Spring (CS) provided a significant increase in resistance to F. graminearum (Miller et al., 2011). Comparative analysis of the infection process using a green fluorescent protein (GFP)‐transformed F. graminearum strain showed that the advance of infectious hyphae from inoculated spikelets to adjacent spikelets was blocked in the rachis node of CS‐7EL, and this blockage was associated with the deposition of a brown substance, the nature of which is currently unknown (Miller et al., 2011). The authors have also noted the possibility that the differences observed in rachis internode lengths between CS and CS‐7EL might have contributed to the 7EL‐mediated F. graminearum resistance in CS (Miller et al., 2011).

The analysis of crown infection by F. graminearum identified at least three distinct stages of infection (spore germination, infection and initial colonization of leaf epidermis and extensive colonization of crown), and microarray analysis revealed that different F. graminearum genes were expressed in these distinct stages of crown rot infection, indicating the complexity of the processes involved in initial penetration and spread within the infected tissue (Stephens et al., 2008). A comparative analysis of independent microarray data revealed overlapping fungal gene expression patterns between early stages of crown infection and head infection by F. graminearum (Stephens et al., 2008). Some fungal pathogens, such as Magnaporthe oryzae, the causative agent of blast disease in rice, are known to exhibit a tissue‐adapted invasion strategy (Marcel et al., 2010). It would therefore be interesting to functionally test whether F. graminearum uses the same pathogenicity genes required for FHB to infect vegetative tissues, such as crowns and roots, or similar strategies to infect head, root and crown tissue.

Trichothecenes do not appear to be required to cause visible disease symptoms during crown infection in bread wheat, but do contribute to full fungal biomass production and stem colonization (Desmond et al., 2008a; Mudge et al., 2006). The involvement of trichothecene as a virulence factor in F. culmorum‐mediated crown rot disease on durum wheat has also been demonstrated recently (Scherm et al., 2011). The silencing of TRI6, a transcription factor known to regulate the expression of toxin biosynthesis genes in F. culmorum, led to reduced DON production and crown rot symptom development. In contrast, some of the transformants that showed unexpectedly high DON and TRI6 transcript levels, despite being transformed with a TRI6 silencing construct, showed increased symptom development, suggesting that TRI6 and DON play a role in the infection of the crown/stem tissue during crown rot disease development (Scherm et al., 2011).

Host‐specific invasion strategies

Differences in the F. graminearum pathogenicity processes required for FHB on different host species are known to exist. For instance, F. graminearum‐produced DON does not seem to act as a virulence factor on barley (Maier et al., 2006). To understand this host‐specific adaptation, the F. graminearum genes expressed during the infection of wheat and barley have been compared (Lysøe et al., 2011a) and, surprisingly, found to be substantially different. Recently, FHB disease caused by F. graminearum in the model cereal Brachypodium distachyon has been demonstrated (Peraldi et al., 2011). Interestingly, although F. graminearum does not directly infect leaves of wheat and Arabidopsis without wounding and the addition of exogenous DON (Chen et al., 2006), it is able to directly penetrate and infect leaf tissues of Brachypodium (Peraldi et al., 2011), again suggesting that host‐specific infection strategies may be employed by these pathogens.

FORWARD AND REVERSE GENETIC ANALYSES OF F. GRAMINEARUM GENES

The availability of a genome sequence (Cuomo et al., 2007) and an efficient genetic transformation system for F. graminearum has facilitated forward genetic analyses of gene function, and transposon‐, plasmid‐ and restriction enzyme‐mediated integrations (REMI) have been employed to generate F. graminearum insertion mutant libraries and to identify pathogenicity mutants (Dufresne et al., 2008; Seong et al., 2005). Recently, Baldwin et al. (2010a) generated 5000 F. graminearum transformants by random plasmid insertion, and screened 1000 mutants from this collection for altered pathogenicity. This screen identified eight mutants, named ‘disease‐attenuated F. graminearum’ or ‘daf’, with reduced virulence towards wheat heads. One of the mutants further characterized was daf10 (disease‐attenuated F. graminearum10), which contained a large deletion of 350 kb from the end of chromosome 1. This deletion is predicted to harbour 146 genes, some of which may be involved in toxin production. Consistent with this, the daf10 deletion mutant did not produce DON and, as expected, showed reduced virulence towards wheat in inoculation assays (Baldwin et al., 2010a). Another mutant generated through random plasmid insertion was top1, affected in a gene encoding a topoisomerase 1 enzyme (Table 1). This enzyme is involved in uncoiling DNA during transcription and DNA replication, and seems to be required for fungal sporulation and virulence assessed in wheat ear infection assays (Baldwin et al., 2010b).

Table 1.

Examples of recently characterized Fusarium graminearum genes and their roles in fungal pathogenesis.

Gene	Function	Mutant phenotype	Effect on pathogenicity	Reference
Npc1	Involved in sterol trafficking	Contain more ergosterol than wild‐type	Reduced pathogenicity	Breakspear et al. (2011)
Mid1	Stretch‐activated calcium‐permeable ion channel	Reduced ascospore discharge, reduced growth and abnormal ascospores	No effect	Cavinder et al. (2011)
FgStuA	Transcriptional regulator	Impaired spore production and reduced production of secondary metabolites	Reduced pathogenicity	Lysøe et al. (2011b)
HDF1	Histone deacteylase	Reduced toxin production	Reduced pathogenicity	Li et al. (2011a)
FgERG24B	Sterol C‐14 reductase	Increased sensitivity to amine‐containing fungicides	No effect	Liu et al. (2011a)
CYP51A, ‐B and ‐C	14‐α demethylases	CYP51A and CYP51C deletion mutants show increased sensitivity to fungicides containing sterol demethylation inhibitor	Not determined	Liu et al. (2011b)
FgATG15	Autophagy‐like lipase	Aberrant fungal growth and aerial hyphae	Reduced pathogenicity	Nguyen et al. (2011)
FgPac1	Transcription factor	Reduced development under neutral and alkaline pH, and an earlier Tri gene induction and toxin accumulation at acidic pH	Unknown	Merhej et al. (2011a)
FgVe1 (Velvet)	Fungal development, secondary metabolism and virulence	Aberrant growth and conidiation, reduced aurofusarin biosynthesis, aberrant Tri gene expression and reduced toxin production	Reduced virulence	Merhej et al. (2011c)
aurT, aurZ and aurS	Biosynthesis of the red polyketide pigment aurofusarin	Aberrant pigment phenotypes	Not determined	Frandsen et al. (2011)
ZIF1	b‐ZIP transcription factor unique to filamentous ascomycetes	Reduced toxin production	Reduced pathogenicity	Wang et al. (2011)
FgRrg‐1	Response regulator implicated in the high osmolarity in glycerol (HOG) pathway	Increased sensitivity to osmotic stress mediated by NaCl, KCl, sorbitol or glucose, and to metal cations Li+, Ca2+ and Mg2+	Reduced pathogenicity	Jiang et al. (2011)
FgTOP1	Topoisomerase I required to relax supercoiled DNA	Defects in sporulation	Reduced pathogenicity	Baldwin et al. (2010b)
FgTEP1b	Tensin‐like phosphatase 1 involved in phosphatidylinositol pathway	Increased sensitivity to lithium; reduced conidia production	Reduced pathogenicity	Zhang et al. (2010)
FgPtc1	Type 2C protein phosphatase	Mycelium growth in response to lithium	Reduced pathogenicity	Jiang et al. (2010)
ROA	Encodes a novel gene required for normal ascospore morphogenesis and discharge	Abnormal ascospores	Not determined	Min et al. (2010)
FTL1	Encodes a transducin‐β‐like gene acting as a component of the Set3 complex involved in late stages of ascospore formation	Significantly reduced in conidiation	Reduced pathogenicity	Ding et al. (2009)
Ssk1 and Hog1	Histidine kinase two‐component response regulator and mitogen‐activated protein kinase, respectively	Aberrant reproductive development	Reduced pathogenicity	Oide et al. (2010)
CCH1	L‐type calcium ion channel	Ascospore discharge and mycelial growth	No effect	Hallen and Trail (2008)
TRI6 and TRI10	Transcriptional regulators of trichothecene biosynthesis	Reduced toxin production	Reduced pathogenicity	Seong et al. (2009)
ZRA1	ABC transporter involved in zearalenone biosynthesis	Reduced zearalenone production	Not determined	Lee et al. (2011)
EBR1	A novel Zn2Cys6 transcription factor	Reduced growth	Reduced pathogenesis	Zhao et al. (2011)

Similarly, an earlier screen of 6500 antibiotic‐resistant mutants generated by REMI identified several pathogenicity mutants (Seong et al., 2005), including the ftl1 mutant defective in the transducin‐β‐like protein FTL1 (Table 1). The ftl1 mutant has been characterized in detail recently and displays defects in its ability to move from infected anthers to flowers. Although no alterations in DON biosynthesis were evident in the mutant, the ftl1 mutant showed an increased sensitivity to plant defensins (Ding et al., 2009), which might explain the reduced virulence phenotype of the mutant.

Recent gene knock‐out mutants of F. graminearum from both forward and reverse genetic experiments (Table 1) can be placed into broad categories. Some of the mutations affect fungal development without having any effect on virulence. For instance, knock‐out mutants of the Mid1 gene, encoding a calcium ion channel, caused reduced ascospore discharge, reduced growth and abnormal ascospores, but had no effect on virulence (Cavinder et al., 2011), although in the field this mutation would probably reduce inoculum fecundity. In the great majority of knock‐out mutants described, fungal development, toxin biosynthesis and virulence were all affected. Recent examples include mutants of regulatory genes FgStuAp, encoding a transcription factor (Lysøe et al., 2011b), and FgATG15, encoding an autophagy‐like lipase involved in growth and the production of aerial hyphae (Table 1; Nguyen et al., 2011).

Gene knock‐out mutant analysis has also identified genes that affect different steps in F. graminearum pathogenicity processes. For instance, in the hdf1 mutant deficient for the HDF1 gene encoding a histone deacetylase, the initial penetration process was not altered, but the mutant failed to spread within wheat heads (Table 1; Li et al., 2011b). The hdf1 mutant also showed reduced DON levels, supporting the proposed role of DON in pathogen spread. A correlation between reduced DON levels and compromised pathogen spread within infected heads has also been observed in other F. graminearum mutants. For instance, daf10 (Baldwin et al., 2010a), zif1 (Wang et al., 2011) and Fgatg15 (Nguyen et al., 2011) all show reduced DON levels, as well as reduced virulence towards wheat heads (Table 1).

Although most gene knock‐outs described so far have reduced toxin biosynthesis and attenuated virulence, mutants that show increased toxin biosynthesis and virulence have also been identified. For instance, knock‐out mutants of the TRI6‐regulated genes FGSG_00007 and FGSG_10397, predicted to encode a putative cytochrome P450 monooxygenase and a protein with a partial terpene cyclase domain, respectively, showed elevated DON levels, as well as increased virulence (Gardiner et al., 2009a), suggesting that these two loci act as negative regulators of toxin biosynthesis. The product of the FGSG_10397 locus was later found to be a longiborneol synthase required for the production of culmorin, a tricyclic sesquiterpene known to be produced by F. graminearum and designated as CLM1 (McCormick et al., 2010). Disruption of CLM1 in F. graminearum resulted in not only complete loss of culmorin production, but also in an increase in 15‐acetyldeoxynivalenol (15ADON) levels (Gardiner et al., 2009a; McCormick et al., 2010). Increased DON levels and virulence have also been found in the recently described pac1 mutant compromised in the sensing of extracellular pH (Table 1; Merhej et al., 2011a).

FUSARIUM GRAMINEARUM TRANSCRIPTOME, PROTEOME AND METABOLOME ANALYSES

The advent of genome‐scale microarrays (Becher et al., 2011; Güldener et al., 2006) has enabled the analysis of the F. graminearum transcriptome to better understand the fungal infection programmes on diverse tissues and hosts (Hallen‐Adams et al., 2011; Lysøe et al., 2011a; Stephens et al., 2008), as well as in different F. graminearum mutants (Baldwin et al., 2010a; Gardiner et al., 2009a; Hallen and Trail, 2008; Lee et al., 2011; Seong et al., 2009). Microarray analysis of global gene expression in Tri6 and Tri10 deletion mutants affected in toxin biosynthesis and fungal virulence revealed hundreds of genes potentially regulated by these two transcription factors (Seong et al., 2009).

Proteome studies conducted in F. graminearum have identified a number of secreted proteins that may act as effectors during fungal growth in culture and infection of wheat heads (Paper et al., 2007). More than 50% of F. graminearum proteins identified during infection contained putative secretion signals (Paper et al., 2007), suggesting that these proteins may have effector functions, although this remains to be tested experimentally. Another proteomics study has observed changes in the accumulation of 135 of the 435 detected F. graminearum proteins when F. graminearum was grown on a medium that promotes trichothecene biosynthesis (Taylor et al., 2008). Although most of these proteins were functionally unknown, their sheer numbers are an indication of the complex processes associated with mycotoxin biosynthesis in F. graminearum. The broad metabolic impact of DON production has also been further demonstrated via a comparative analysis of the DON‐deficient tri5 mutant with wild‐type F. graminearum using both transcriptome and metabolome approaches, where widespread alterations in a number of carbon and nitrogen metabolism pathways were observed (Chen et al., 2011).

IN SILICO PREDICTION OF F. GRAMINEARUM PATHOGENICITY GENES

Large‐scale analysis of protein–protein interaction networks (interactome) has the potential to reveal new insights into F. graminearum pathogenesis. Towards this aim, Zhao et al. (2009) used computational methods to predict protein–protein interactions in F. graminearum based on the assumption that F. graminearum proteins that are evolutionarily conserved will also show conserved interaction patterns similar to those known for well‐studied species, such as Saccharomyces cerevisiae, Drosophila melanogaster, Caenorhabditis elegans, Homo sapiens, Musculus musculus, Schizosaccharomyces pombe and Escherichia coli. The predicted F. graminearum interactome database currently contains about 52% of the F. graminearum proteome, consisting of 3745 proteins and 27 102 predicted interactions. Of the three predicted protein‐protein interactions, one of the interactions between a phosphoprotein phosphatase 3‐R catalytic chain and a calcineurin B was verified experimentally using yeast two‐hybrid experiments (Zhao et al., 2009).

The availability of both microarray and predicted protein interaction data has led to a systems biology approach to computationally predict F. graminearum pathogenicity genes (Liu et al., 2010). In this work, genes that are predicted to interact with previously described F. graminearum pathogenicity genes (Zhao et al., 2009) and are differentially expressed during infection were identified. Genes involved in the G‐protein‐coupled receptor and MAPK signalling pathways that are known to affect the pathogenesis process (Oide et al., 2010; Zhang et al., 2011) were represented among the in silico‐identified F. graminearum genes. Also identified were F. graminearum genes with homology to those known to affect pathogenesis in other fungi. This approach shows promise to narrow down the number of candidate genes that can be further functionally tested.

FUSARIUM GRAMINEARUM EFFECTORS AND VIRULENCE FACTORS

Pathogen effectors usually contain amino‐terminal secretion signals and, after secretion by pathogens, either remain in the extracellular spaces of the host plant or enter host cells and have the ability to modify (i.e. suppress) host defences. Analysis of the F. graminearum genome has indicated that 1250 genes encode secreted proteins in this pathogen, and it is possible that a substantial number of these proteins are effectors (reviewed by Schmidt and Panstruga, 2011). In addition, recent microarray analyses have revealed that 43% of the genes expressed in the pathogen during infection contain putative secretion signals, when compared with only 12.8% of the genes containing such sequences in the whole genome, suggesting that large numbers of F. graminearum effectors may be deployed during infection (Lysøe et al., 2011a). However, to date, no report has described the virulence‐associated roles for any effector from F. graminearum.

Some secondary metabolites, such as the phytotoxin coronatine from the phytopathogenic bacterium Pseudomonas syringae, act as effectors in modulating host defences (Schulze‐Lefert and Panstruga, 2011). Similarly, DON acts as a virulence factor promoting F. graminearum aggressiveness on wheat. Although DON is known to act as a eukaryote protein synthesis inhibitor in vitro (Pestka, 2010), how DON acts on plant cells to affect virulence is currently unknown. Exogenous application of DON to wheat and Arabidopsis triggers both programmed cell death and strong defence gene expression (Desmond et al., 2008a; Nishiuchi et al., 2006) In animal cells, DON also triggers programmed cell death and an immune reaction via a ribotoxic shock process (Pestka, 2010). The DON‐mediated activation of immune responses in both animals and plants requires MAPK signalling pathways (Nishiuchi et al., 2006; Pestka, 2010) that are known to act downstream from pattern recognition receptors (Tena et al., 2011). Therefore, DON may be considered as an F. graminearum‘effector’ with a cross‐kingdom action.

WHAT INDUCES TOXIN BIOSYNTHESIS IN F. GRAMINEARUM?

Given the importance of DON as a virulence factor, the regulation of its biosynthesis has received considerable attention (Merhej et al., 2011a). Importantly, DON biosynthesis in F. graminearum appears to be much higher during infection than during growth in culture, suggesting that signals generated in planta may play important roles in the triggering of DON biosynthesis (Mudge et al., 2006; Voigt et al., 2005). There is evidently a complex spectrum of possible regulatory signals for toxin production in infected plants, but an over‐arching observation from this body of work is that F. graminearum appears to stimulate host stress and defence responses and to recognize these as cues for the activation of mycotoxin biosynthesis. Table 2 contains examples of compounds that regulate TRI gene expression and DON biosynthesis in F. graminearum in vitro. Some of the DON‐inducing compounds identified are naturally present in the plant or induced during infection (see below). However, the evidence that DON‐inducing compounds are present at the appropriate concentration in a given tissue during infection is mostly unavailable.

Table 2.

Examples of molecules acting as inducers or repressors of trichothecene mycotoxin (e.g. deoxynivalenol, DON) biosynthesis in Fusarium graminearum.

Inducer molecule	Mode of action	Reference
Polyamines	Activates Tri gene expression and toxin biosynthesis	Gardiner et al. (2009b)
Sugars (sucrose, 1‐kestose and nystose)	Activates Tri4 and Tri5 expression and DON and 3ADON biosynthesis	Jiao et al. (2008)
Precocenes and piperitone from the essential oils of Matricaria recutita and Eucalyptus dives	Inhibits Tri4, Tri5, Tri6 and Tri10 expression and DON biosynthesis	Yaguchi et al. (2009)
Magnesium ions	Inhibits Tri5, Tri6 and Tri12 expression and toxin biosynthesis	Pinson‐Gadais et al. (2008)
Reactive oxygen (e.g. H2O2)	Activates Tri gene expression and toxin biosynthesis	2006, 2007); Audenaert et al. (2010)
pH	Low pH activates Tri gene expression and toxin biosynthesis	Gardiner et al. (2009c); Merhej et al. (2010)
Ferulic acid	Inhibits Tri5 expression and DON biosynthesis	Boutigny et al. (2009)
Cinnamic‐derived acids and phenolic acids	Ferulic acid and coumaric acid activates DON biosynthesis	Ponts et al. (2011)
Cobalt chloride	Induces TRI4 and TRI6 expression and DON biosynthesis	Tsuyuki et al. (2011)
Warm temperature (e.g. 25 °C)	Induces trichothecene production	Kokkonen et al. (2010)
Salt (NaCl) in culture	Inhibits Tri5 expression and DON biosynthesis	Ochiai et al. (2007)

3ADON, 3‐acetyldeoxynivalenol.

Amine inducers

To identify nutritional inducers of DON biosynthesis in F. graminearum, Gardiner et al. (2009b) used an indicator F. graminearum strain expressing the GFP reporter gene under the control of the TRI5 gene promoter. In this work, commercially available ‘phenotype array’ plates containing different nitrogen sources were initially used to grow the TRI5‐GFP F. graminearum strain (Fig. 2). The effects of selected GFP‐inducing nitrogen sources on DON biosynthesis were further verified in larger scale experiments. Interestingly, the nitrogen sources that most effectively induced DON biosynthesis were ornithine, agmatine and putrescine, all being part of polyamine biosynthesis, a well‐established stress response pathway in plants. In the follow‐up work, it was further shown that polyamine gene expression and polyamine levels are stimulated by F. graminearum infection (Gardiner et al., 2010b). This has led to the hypothesis that F. graminearum exploits a stress‐induced polyamine pathway in the host plant to activate toxin production and enhance virulence.

Figure 2.

Examples of strategies aimed at the high‐throughput identification of novel compounds that inhibit fungal growth (A) (Schreiber et al., 2011) or induce mycotoxin biosynthesis (B) (Gardiner et al., 2009b). See text for additional details. GFP, green fluorescent protein; Fg, Fusarium graminearum.

pH

The acidity of the culture medium is another factor that promotes DON biosynthesis in the pathogen (Gardiner et al., 2009c; Merhej et al., 2010). It has been proposed that the fungal pH sensing system may be involved in pH‐mediated alteration of DON biosynthesis (Gardiner et al., 2009c). Indeed, the PacC transcription factor, a key component of the fungal regulatory system involved in sensing extracellular pH, negatively regulates TRI gene expression and toxin production. The FgDPac1 deletion mutant showed an earlier induction of TRI gene expression and toxin accumulation under low pH (Table 1; Merhej et al., 2011b).

Reactive oxygen

Reactive oxygen species (ROS), such as hydrogen peroxide (H₂O₂), have been identified as one of the signals that induce DON biosynthesis in the pathogen. Interestingly, the addition of H₂O₂ to F. graminearum cultures leads to relatively late induction of TRI gene expression (2–4 days after treatment) and toxin accumulation (Ponts et al., 2006). It seems likely that the induction of TRI gene expression by H₂O₂ requires several downstream metabolic or signalling steps for DON induction. The ROS‐degrading enzyme catalase suppresses TRI gene expression and toxin accumulation (Ponts et al., 2007), and several H₂O₂‐generating fungicides have been reported to induce TRI gene expression and DON biosynthesis (Audenaert et al., 2010). H₂O₂ is also generated in the host by exogenously applied DON (Desmond et al., 2008a), and it is possible that a feedback loop exists in which increased cellular ROS in the host further stimulates DON biosynthesis.

Phenolic acids

Ponts et al. (2011) have investigated the effect of five phenolic acids (ferulic, p‐coumaric, caffeic, syringic and p‐hydroxybenzoic acid) on DON biosynthesis. Of the phenolic acids tested, ferulic acid and coumaric acid enhanced toxin production, whereas p‐hydroxybenzoic acid treatment showed a reduction (Table 2). Interestingly, 2005, 2008) have identified higher levels of cinnamic acid and ferulic acid in resistant wheats compared with susceptible genotypes. These compounds induce toxin biosynthesis, but also show antifungal activity on F. graminearum (Bollina et al., 2010; Ponts et al., 2011). These defensive metabolites may act as cues for the pathogen to boost toxin production in an attempt to stimulate host cell death, which allows infection to continue.

HOST GENES EXPRESSED DURING PATHOGENESIS

Transcriptome studies

So far, there have been several microarray studies in both wheat and barley, investigating which host genes are induced during infection (reviewed by Bischof et al., 2011). Overall, the classes of host genes induced by F. graminearum overlap substantially with those found in many other plant–microbe interactions. In these microarray studies, gene expression patterns, either in the same genotype with and without infection or in genetically diverse genotypes that display differences in FHB resistance, have been analysed (e.g. Foroud et al., 2011). The use of isogenic lines that differ in quantitative trait loci (QTLs) conditioning FHB would be a better alternative than the use of genetically diverse lines to pinpoint possible resistance functions.

One of the QTLs associated with type‐II FHB resistance (i.e. resistance to spread within the spike) is the Fhb1 locus located on the short arm of the 3B chromosome in wheat. The mechanism by which Fhb1‐mediated resistance occurs is currently unknown and, to identify potential mechanisms of Fhb1‐mediated resistance, F. graminearum‐responsive gene expression in isogenic wheat lines that differ by this QTL have been compared using microarray analyses (Golkari et al., 2009; Jia et al., 2009). Although large numbers of wheat genes responded to F. graminearum infection in both susceptible and resistant near‐isogenic lines (NILs), only a small subset of 14 genes showed differential expression between NILs that differed by the presence and absence of the Fhb1 locus (Jia et al., 2009). Future analyses of these candidate genes may reveal new insights into the potential mechanism(s) of resistance conferred by the Fhb1 locus. Remarkably, in another recent microarray study that compared isogenic pairs of three QTLs known to confer quantitative resistance to FHB in barley, different sets of differentially expressed genes were found, suggesting that the mechanisms of resistance conferred by different QTLs might be distinct (Jia et al., 2011). Ding et al. (2011) analysed a disease‐susceptible wheat mutant with its partially resistant wild‐type using proteome and transcriptome analyses, as well as measurements of the defence regulators salicylic acid (SA) and jasmonic acid (JA). They postulated that early induction of the SA defence pathway was critical for resistance in wheat.

The mycotoxin DON is one of the best‐known F. graminearum virulence factors and causes programmed cell death and has defence‐inducing activities (Desmond et al., 2008a; Masuda et al., 2007; Nishiuchi et al., 2006). Therefore, recent work has also focused on the transcriptome responses triggered by DON in the host (Walter et al., 2008). Using a custom‐made wheat cDNA array, it was found that, among other things, DON modulates JA biosynthesis and signalling as the wheat homologues of the Arabidopsis MYC2 and 12‐oxophytodienoic acid (OPDA) were up‐regulated (Walter and Doohan, 2011). Perhaps, because of the toxicity of DON to plant cells, DON exposure also triggers strong expression of genes potentially involved in DON detoxification. For instance, treatment of barley heads with DON induced the expression of genes encoding ABC transporters, uridine diphosphate (UDP)‐glucosyltransferases, cytochrome P450s and glutathione‐S‐transferases, plant enzymes that are potentially involved in DON detoxification (Gardiner et al., 2010a; Li et al., 2010).

Proteome studies

So far, several proteome studies have been conducted to identify differentially expressed proteins in infected grains of wheat (Dornez et al., 2010; Wang et al., 2005; 2005, 2006), barley (Geddes et al., 2008; Yang et al., 2011; Zantinge et al., 2010), maize (Mohammadi et al., 2011) and wild relatives of wheat (Eggert et al., 2011; 2010a, 2010b) and barley (Eggert and Pawelzik, 2011). However, very little overlap, other than the accumulation of some well‐described pathogenesis‐related (PR) proteins (e.g. chitinases and peroxidases, etc.), has been observed for the F. graminearum‐induced proteins detected in these studies. It was proposed that differences observed in different studies could be a result of the use of different inoculation methods and growth conditions (Yang et al., 2010a). In addition, proteome studies appear to suffer from the lack of sensitivity in detecting changes in low‐abundance proteins. This may explain why many potential F. graminearum effectors were not identified in fungal proteomic studies, although microarray experiments often revealed that mRNAs encoding secreted proteins were induced during pathogenesis in F. graminearum.

Metabolome studies

A number of metabolome studies have also been conducted in F. graminearum‐infected hosts. In a recent study, Kumaraswamy et al. (2011a) identified a total of 626 metabolites whose levels were altered by F. graminearum in two barley genotypes that showed different levels of quantitative resistance to FHB. The majority of these metabolites could be grouped into fatty acid, phenylpropanoid and flavonoid pathways. Among the metabolites belonging to the fatty acid pathway, methyl jasmonate (MeJA), as well as the JA precursors linolenic and linolenic acids, were particularly notable, and increased accumulation of MeJA following F. graminearum challenge in the resistant barley genotype relative to the susceptible genotype has led to the proposal that the JA pathway is the predominant defence signalling pathway operating in barley against F. graminearum (Kumaraswamy et al., 2011a).

NEW INSIGHTS INTO HOST RESISTANCE FROM MODEL PATHOSYSTEMS

Earlier research has established that F. graminearum can infect Arabidopsis (Urban et al., 2002), and this model pathosystem has been exploited to dissect the potential roles of signalling pathways in host resistance to F. graminearum. In Arabidopsis, the SA pathway appears to be required for FHB resistance. The Arabidopsis SA signalling mutants, npr1 and eds11, as well as the SA‐deficient mutant sid1 and transgenic nahG plants display increased susceptibility when F. graminearum or F. culmorum are inoculated on leaves or floral organs, respectively (Cuzick et al., 2008; Makandar et al., 2010). In contrast, the JA pathway appears to mediate disease susceptibility based on the increased disease resistance observed in the receptor mutant coi1 (Makandar et al., 2010; Schreiber et al., 2011). This is consistent with recent observations on defence pathway induction in resistant and susceptible wheats (Ding et al., 2011)

Similar to JA signalling, the ethylene (ET) signalling pathway is also required for susceptibility to F. graminearum in both Arabidopsis and wheat. The Arabidopsis ET signalling mutants ein2, ein3 and etr1 show smaller necrotic lesions than do wild‐type plants when their leaves are inoculated with F. graminearum. In contrast, ET‐overproducing mutants, eto1 and eto2, as well as the ctr1 mutant with constitutively activated ET signalling display enhanced disease susceptibility. Furthermore, transgenic wheat heads silenced for the expression of the wheat EIN2 (ETHYLENE INSENSITIVE2) gene were more resistant to FHB. This suggests that ET signalling is exploited by F. graminearum to cause disease in both monocotyledonous and dicotyledonous plants (Chen et al., 2009). Although it is not known how JA and ET pathways provide susceptibility to F. graminearum, one reason may be the potential involvement of these pathways in pathogen‐induced senescence responses (Thatcher et al., 2009).

PLANT PROTECTION AND TOXIN REDUCTION STRATEGIES

Antifungal transgenic approaches

Recent studies have shown that transgenic wheat independently expressing a pectin methyl‐esterase inhibitor (Volpi et al., 2011), an antifungal plant defensin (AFP) (Li et al., 2011b), a class II chitinase from barley (Shin et al., 2008) and polygalacturonase‐inhibiting protein (Ferrari et al., 2011) provides quantitative FHB resistance under glasshouse conditions. Of these proteins, chitinase, AFP2 and polygalacturonase directly inhibit pathogen growth, whereas pectin methyl‐esterase inhibitor‐mediated resistance is associated with an inability of F. graminearum to grow on methyl‐esterified pectin. The pyramiding of multiple transgenes with different modes of action may be a way to achieve stronger and more durable resistance.

More recently, a new class of sesquiterpenoid phytoalexins, called zealexin, with inhibitory activities against F. graminearum has been identified in maize, responding to attack by F. graminearum as well as synergistically in jasmonate‐treated plants (Huffaker et al., 2011). Purified zealexin inhibited F. graminearum growth in physiologically active concentrations, indicating that transgenic enhancement of the levels of this phytoalexin can be beneficial to reduce cob rot in maize.

Mycotoxin reduction strategies

As discussed above, various polyamines and phenolic acids have been shown to be strong inducers of DON biosynthesis in F. graminearum, and the inhibition of the biosynthesis of these host metabolites in the host plant may lead to reduced toxin levels. However, these host metabolites are involved in other functions, such as resistance to other stresses, and therefore their inhibition would need to be spatially and temporally restricted (e.g. in the developing grain tissue and only during infection) to reduce possible collateral effects.

Another way of reducing DON levels is to express genes that are responsible for DON detoxification (Karlovsky, 2011). Previous research has identified an Arabidopsis UDP‐glucosyltransferase (UGT) involved in DON detoxification (Poppenberger et al., 2003). More recently, several DON‐inducible UGTs have been identified in wheat (Desmond et al., 2008b) and barley (Gardiner et al., 2010a), and a barley gene product was able to detoxify DON. In a separate study, Schweiger et al. (2010) expressed DON‐inducible barley UGT cDNAs in yeast and tested the resistance of recombinant yeast strains to DON. In these assays, one of the barley UGTs, called HvUGT13248, conferred DON resistance in yeast. However, the effects of both barley genes on disease and DON deposition in transgenic cereals have yet to be demonstrated. The formation of DON–glutathione conjugates may also help to reduce the effect of DON on plant tissue (Gardiner et al., 2010a).

The identification of the cellular targets of DON would also potentially increase resistance if such targets could be modified to reduce toxin sensitivity. A recent example of this strategy is the simultaneous reduction of DON accumulation and the enhancement of F. graminearum resistance under field conditions in wheat transgenically expressing a truncated form of the yeast ribosomal protein L3 (Di et al., 2010). The authors have speculated that the expression of yeast protein possibly interferes with the targeting of the wheat L3 protein by DON, and hence reduces the cellular sensitivity to DON. Indeed, DON is known to bind to eukaryotic ribosomes and to interfere with the protein synthesis apparatus (Pestka, 2010). To identify other potential targets of trichothecene action in an unbiased manner, McLaughlin et al. (2009) employed a large‐scale screen to identify individual yeast mutants showing increased mycotoxin resistance. This screen identified several mutants deficient for genes involved in mitochondrial translation and mitochondrial membrane function, suggesting that these functions confer toxin sensitivity (McLaughlin et al., 2009). It remains to be determined whether similar processes provide sensitivity to trichothecene mycotoxins in plant cells.

Novel defence compounds

The external application of novel compounds with effects on pathogen growth is also a promising approach to reduce disease incidence. A recent example of this strategy is the complete prevention of disease development in both wheat and maize by the external application of CNI‐1493, a compound that inhibits fungal deoxyhypusine synthase (DHS) activity without affecting grain development (Woriedh et al., 2011). The F. graminearum DHS gene, which is strongly induced during the early stages of infection, is involved in hypusine biosynthesis required for the activity of protein eukaryotic initiation factor‐5A (eIF‐5A), and eIF‐5A is thought to transport fungal mRNAs required for pathogenesis out of the nucleus to facilitate their translation by ribosomes (Woriedh et al., 2011). Despite its promise in disease prevention, the application of CNI‐1493 under field conditions seems to be unlikely because of possible nondiscriminatory toxic effects on other organisms.

Using a high‐throughput screening system in which Arabidopsis leaves were incubated with suspected defence‐activating small molecules (Fig. 2), Schreiber et al. (2011) identified sulphamethoxazole and the indole alkaloid gramine as providing protection against F. graminearum in the assay. These two compounds were also effective in providing protection in wheat when sprayed onto heads prior to inoculation. Although possible protection mechanisms afforded by these compounds are not known, this finding certainly suggests the relevance of the Arabidopsis pathosystem to F. graminearum–wheat interaction. Recent research in our laboratory has shown that gramine can be highly toxic to F. graminearum if the tests are conducted on agarose medium, but not on agar (D. M. Gardiner, unpublished work), as conducted by Schreiber et al. (2011). This is probably because sulphated polysaccharides in agar may bind amines such as gramine.

Biological control

Inhibition of FHB through the use of biocontrol organisms may offer environmentally friendly disease control compared with chemical treatments. So far, some success in disease reduction has been reported with bacterial strains Bacillus spp. and Pseudomonas spp. under controlled conditions (He et al., 2009). Recent research has provided new insights into a Pseudomonas‐mediated disease protection system for F. graminearum. Using a split root inoculation system and carbon tracer elements, Henkes et al. (2011) demonstrated that the infection of roots by F. graminearum results in the distribution of more and less carbon to uninfected and infected roots, respectively, perhaps in an attempt to reduce fungal growth before antifungal defences can kick in. In contrast with plants inoculated directly with F. graminearum through the roots, barley plants primed with Pseudomonas fluorescens before inoculation did not show the same patterns of distorted distribution of carbon allocation and were also less affected developmentally in response to infection (Henkes et al., 2011). This study provides a possible explanation of the nature of processes operating in the host plant in response to root infection by F. graminearum, but it is not known whether nutrient supply pathways are affected in other plant parts.

Another study compared the F. culmorum‐triggered global gene expression patterns in barley heads that were either primed or nonprimed with the concurrent application of FHB biocontrol strain P. fluorescens and F. culmorum. As suspected, the application of biocontrol bacteria altered the expression patterns of large numbers of genes, including genes encoding lipid transfer proteins and protease inhibitors. This study also implicated the JA pathway as a modulator of P. fluorescens‐mediated priming against F. culmorum infection in barley (Petti et al., 2010). Together, these studies suggest that biological control agents for F. graminearum may act through the stimulation of intrinsic resistance in the host rather than direct inhibitory effects on the pathogen.

Resistance breeding

Although this review has focused mainly on the molecular aspects of the host–pathogen interaction, significant progress has also been made in recent years in the genetic dissection of F. graminearum resistance in the host. For example, large numbers of QTLs controlling resistance to F. graminearum have been identified in both wheat and barley (reviewed by Buerstmayr et al., 2009; Handa et al., 2008), and isogenic lines that differ by the presence and/or absence of some QTLs have been produced. The mechanisms of resistance conditioned by these QTLs are currently unclear and, as progress is made in wheat genome sequencing, the genomic regions harbouring these QTLs can certainly be better defined and their effects on disease resistance studied.

Breeding selections made on the basis of resistance‐associated host metabolites (both induced during F. graminearum infection and having in vitro inhibitory activity against F. graminearum) could be a way to speed up the selection process if reliable associations between these metabolites as biomarkers and host resistance can be established (Bollina et al., 2011). For example, in a relatively small‐scale metabolomic study, phenylalanine, p‐coumaric acid, jasmonate, linolenic acid and DON levels have been associated with resistance in a small subset of barley genotypes (Kumaraswamy et al., 2011b).

FUTURE PERSPECTIVES

New F. graminearum genomes?

The availability of a sequenced F. graminearum genome (Cuomo et al., 2007) has undoubtedly enhanced significantly our ability to uncover new insights into this pathogen. The sequencing of additional genomes from F. graminearum strains that differ by virulence and toxin production, etc., as well as from related species such as F. culmorum and F. pseudograminearum, would provide useful information about the evolution of pathogenicity and host specificity in these pathogens. Indeed, recent research has shown that the differential activity of TRI8, resulting from DNA sequence variations in the coding region of this gene, is a critical determinant of whether an F. graminearum isolate produces 3‐acetyldeoxynivalenol (3ADON) or 15ADON. It appears that TRI8 has a C‐15 esterase or C‐3 esterase activity in 3ADON‐ and 15ADON‐producing F. graminearum isolates, respectively (Alexander et al., 2011). Recent technical developments (e.g. next‐generation sequencing technologies) would make such sequencing efforts a highly feasible exercise.

Although additional F. graminearum genomes are currently not available, existing genome sequences from related Fusarium species have the potential to provide new insights into the evolution of F. graminearum. Indeed, comparative analysis of the F. graminearum genome with the recently completed genomes of F. verticillioides, F. oxysporum f.sp. tomato and F. solani (Nectria haematococca) has revealed substantial sequence similarity and conservation of the order of nearly 9000 genes in these three pathogens (Ma et al., 2010). Interestingly, genes that are unique to each species are found in telomeric chromosomal regions that also show high sequence diversity and recombination rates and contain transposons. This has led to the suggestion that these regions of high variability have evolved through telomeric fusion of ancestral chromosomes (reviewed by Rep and Kistler, 2010).

Functional analyses of host and pathogen genes

So far, only a few F. graminearum mutants have been analysed in detail. The availability of an F. graminearum gene‐indexed mutation resource base, such as those for yeast and Arabidopsis, where knockout strains are systematically identified, deposited and distributed to the scientific community, would significantly speed up the analysis of gene function in F. graminearum.

The scale and throughput of stable transformation of wheat are limiting factors for the functional analysis of candidate host genes for resistance. An alternative to forward genetic analyses of gene function in wheat would be to employ reverse genetic studies. For instance, high‐throughput methods for the identification of deletions in candidate wheat genes of interest have recently been developed (Fitzgerald et al., 2010). To overcome the difficulty of host analysis, at least initially, model systems, such as yeast, Arabidopsis and Brachypodium, are currently being exploited. Brachypodium, being a close relative of wheat, offers new advantages over the Arabidopsis system. In addition to being a susceptible host to F. graminearum (Peraldi et al., 2011), Brachypodium is transformable, has a sequenced genome and contains diverse germplasm and emerging functional genomic resources, such as T‐DNA insertion lines, that can facilitate host gene function analyses.

New challenges

How farming practises and environmental shifts, such as climate change, may affect the evolution of virulence and toxin biosynthesis in F. graminearum is another area of future research. Recent research has shown that larger numbers of F. graminearum isolates with the 3ADON chemotype and with higher aggressiveness and greater levels of grain DON content can be found in current F. graminearum isolates than those collected between 1980 and 2000 in North Dakota (Puri and Zhong, 2010). Although the exact reason for this adaptation is unknown, it has been postulated that the use of resistant host genotypes and fungicides may have led to the development of more aggressive isolates (Puri and Zhong, 2010). Another study examining F. graminearum isolates throughout the USA has found that nivalenol (NIV)‐producing F. graminearum isolates have already established in small‐grain‐growing regions of southern Louisiana (Gale et al., 2011), despite earlier reports showing that such chemotypes were rare (Desjardins et al., 2008). Although the potential toxicity of NIV versus DON toxin‐producing F. graminearum isolates is still controversial (Gale et al., 2011; Puri and Zhong, 2010), it will be important to maintain regular surveillance to monitor possible changes in toxin chemotypes.

FHB: STILL A LOOMING DISASTER?

In their nearly decade‐old pathogen profile, Goswami and Kistler (2004) rightfully warned that F. graminearum and related pathogens could produce disastrous outcomes unless concerted efforts were devoted to a better understanding of the complex nature of these particular plant–pathogen interactions. As reviewed in this article, significant progress has been made towards this goal, although we still have a long way to go towards the development of a good understanding of F. graminearum–cereal interactions and sustainable plant protection strategies.

NOTE ADDED IN PROOF

Recently, an Fg deletion mutant library comprising mutant strains for 657 putative Fg transcription factors has been constructed and the analysis of these mutants under various conditions revealed diverse mutant phenotypes, including alterations in mycelial growth, sexual development, conidia production, virulence, toxin production and stress responses (Son et al., 2011). This library provides a useful resource for functional analysis of Fg transcription factor genes.