Large‐scale molecular genetic analysis in plant‐pathogenic fungi: a decade of genome‐wide functional analysis

Summary

Plant‐pathogenic fungi cause diseases to all major crop plants world‐wide and threaten global food security. Underpinning fungal diseases are virulence genes facilitating plant host colonization that often marks pathogenesis and crop failures, as well as an increase in staple food prices. Fungal molecular genetics is therefore the cornerstone to the sustainable prevention of disease outbreaks. Pathogenicity studies using mutant collections provide immense function‐based information regarding virulence genes of economically relevant fungi. These collections are rich in potential targets for existing and new biological control agents. They contribute to host resistance breeding against fungal pathogens and are instrumental in searching for novel resistance genes through the identification of fungal effectors. Therefore, functional analyses of mutant collections propel gene discovery and characterization, and may be incorporated into disease management strategies. In the light of these attributes, mutant collections enhance the development of practical solutions to confront modern agricultural constraints. Here, a critical review of mutant collections constructed by various laboratories during the past decade is provided. We used Magnaporthe oryzae and Fusarium graminearum studies to show how mutant screens contribute to bridge existing knowledge gaps in pathogenicity and fungal–host interactions.

Introduction

Fungal pathogens impact negatively on the global economy, food, human and animal welfare. This is translated into considerable price increases in staple foods, such as barley, maize, rice and wheat. As a result, numerous studies have been conducted to highlight the plight and probable cause of action in dealing with associated effects of fungal diseases. One review article, based on a survey of 495 votes from the international community of fungal pathologists, lists 10 fungal species that are highly destructive to major crop plants world‐wide (Dean et al., 2012). Most fungi listed in this review have been the subject of genetic analysis in order to understand their pathogenic mechanisms. However, collections of analysed and published mutants are only available for certain pathogenic species, including Botrytis cinerea, Colletotrichum higginsianum, Fusarium graminearum, Fusarium oxysporum and Magnaporthe oryzae. An extensive number of analysed mutant collections are available for F. graminearum and M. oryzae, more than for any other fungi. Therefore, genome‐wide functional screens of these two fungi can serve as roadmaps for the assessment of relevant areas of fungal pathogenicity and disease management.

Fusarium graminearum is a predominant pathogen of head scabs or Fusarium head blight (FHB) in wheat and barley (Turkington et al., 2014). This pathogen is also a cause of stalk and ear rot infections in maize (Turkington et al., 2014). Fusarium graminearum produces several important mycotoxins, including commonly studied toxins, such as deoxynivalenol (DON), nivalenol (NIV) and zearalenone (ZEA). These mycotoxins can accumulate post‐harvest in the presence of high moisture levels (Magan et al., 2010). As a result, they have the potential to contaminate food and feed, which, in turn, present a health hazard to animals and humans (Darwish et al., 2014; Zain, 2011). Essentially, the establishment of the genetic basis of F. graminearum mycotoxin production will aid in the design of effective control strategies against FHB epidemics and associated health impacts. Fusarium graminearum mutant collections are therefore a promising molecular genetics tool for the high‐throughput discovery of genes associated with mycotoxin production and regulation.

Magnaporthe oryzae is an ascomycete filamentous fungal pathogen of rice, particularly in Asia, which relies on this crop as a major food source (Toriyama et al., 2005). Moreover, a recent outbreak of wheat blast in Bangladesh destroyed 15 000 hectares of wheat, establishing M. oryzae as an emerging pathogen of this crop (Callaway, 2016). Much of what is known about M. oryzae virulence is mostly derived from rice infection studies. Magnaporthe oryzae infects rice through an asymptomatic biotrophic phase that allows growth and feeding on living tissues (Marcel et al., 2010; Oliveira‐Garcia and Valent, 2015; Talbot and Wilson, 2009). This fungus can switch to become a necrotrophic pathogen. Although effectors are known to facilitate pathogen colonization by manipulation of host immunity during biotrophy, they have been shown to trigger host cell death in Nicotiana benthamiana during late infection stages of M. oryzae (Chen et al., 2013). Therefore, effectors that are expressed following the biotrophic phase seem to play an important role in the necrotrophic lifestyle of this pathogenic fungus (Chen et al., 2013). The ability of M. oryzae to switch from the biotrophic to the necrotrophic phase allows feeding on living and dead plant tissues. Hence, M. oryzae is classified as a hemibiotroph. However, this fungus is one of the few hemibiotrophs that are experimentally tractable, with its genome and that of its main host fully sequenced (Dean et al., 2005; Ebbole, 2007; Goff et al., 2002). This makes M. oryzae a practical choice for immediate use of molecular genetics tools, such as mutant collections.

For the past 10 years, successful genetic screens of mutant collections have been conducted in various background strains of F. graminearum and M. oryzae (Table 1). Such screens have produced vast amounts of data, much of which has been integrated to define the parasitic mechanisms used by these fungi to infect their host plants. Owing to the importance of these collections for agriculture, a comprehensive and updated assessment of their functional data is required. The current review provides such an assessment by showing how F. graminearum and M. oryzae mutant screens advance certain areas of fungal biology, including virulence gene discovery, modes of pathogenicity and regulation, and host–pathogen interactions.

Table 1.

Mutant collections in Magnaporthe oryzae and Fusarium graminearum.

Mutant collection	Strain	Host(s) tested	Isolates	Isolates with altered phenotypes (%)	Genes identified/characterized	Reference
Agrobacterium tumefaciens‐mediated	M. oryzae KJ201	Rice	21 070	2.6	203	Jeon et al. ( 2008)
	M. oryzae Guy11	Barley cv. Golden Promise and rice cv. CO‐39	5248	0.1	1	Li et al. ( 2010)
Bidirectional genetics	M. oryzae KJ201	Rice cv. Nakdongbyeo	1139	11.2	3	Park and Lee ( 2013)
Targeted gene deletion	M. oryzae Ina72	Barley cv. Nigrate and rice cv. Shin No. 2	78	1.2	1	Saitoh et al. ( 2012)
	F. graminearum Z3643	Not tested on host plant	127	32	40	Kim et al., 2015
Plasmid‐mediated integration	F. graminearum PH‐1	Wheat cv. Bobwhite	1170	0.6	146*	Baldwin et al. ( 2010)
	F. graminearum PH‐1	Wheat cv. Bobwhite	650	1.2	456*	Urban et al. ( 2015)
Transcription factor	F. graminearum 3639	Not tested on host plant	657	25.8	170	Son et al. ( 2016)
	M. oryzae 70‐15	Barley and rice cv. CO39	104	58.6	61	Lu et al. ( 2014)
	M. oryzae 70‐15	Barley and rice cv. CO39	47	97	46	Cao et al., 2016

*These genes were predicted in the chromosomal regions lost as a result of insertions, but were not formally characterized individually.

Mutant Collections

Agrobacterium tumefaciens‐mediated (ATM) collections

Jeon et al. (2007) constructed 21 070 hygromycin‐resistant strains from M. oryzae KJ201 using ATM transformation and transfer DNA (T‐DNA) tagged insertion. ATM transformation uses the ability of A. tumefaciens to randomly integrate T‐DNA molecules from its plasmid into the host genome. Pathogenicity screening of T‐DNA insertion strains on rice leaves identified 2.6% (559/21 070) mutants (Table 1) with defects in infection and asexual reproduction (Jeon et al., 2007). Approximately 181 of these insertion strains are exclusively defective in pathogenicity, showing abnormal conidial shape and an inability to form an appressorium, a specialized cell that penetrates the rice leaf cuticle using turgor pressure. These pathogenicity defects are derived from disruptions in 203 independent genes, only a small number of which have been well characterized (Chi et al., 2009; Goh et al., 2011a, 2011b; Jeon et al., 2007, 2008). Jeon et al. (2007) also provided a platform to analyse T‐DNA insertion mutants using the ATM and high‐throughput phenotype databases available at http://atmt.snu.ac.kr and http://www.phi-base.org/ (Winnenburg et al., 2007), respectively. The use of these databases with the ATM mutant collection allows researchers to link target genes with mutant phenotypes, and this approach will facilitate the direct discovery of gene function in M. oryzae.

Li et al. (2010) also used ATM T‐DNA mutagenesis to generate 5248 hygromycin‐resistant insertion strains that were screened on susceptible barley cultivar (cv.) Golden Promise using a barley cut‐leaf assay (Table 1). As a result, seven phenotypically altered strains were recovered and later verified on rice cv. CO‐39. Among these strains, LY‐130 was attenuated for pathogenicity (Table 1) on both hosts. T‐DNA genomic insertion in LY‐30 was mapped at the third exon of RIC8. This gene encodes a cytoplasmic guanine‐nucleotide exchange factor (GEF) which triggers cell development through interaction with the α‐subunit of the heterotrimeric G protein (Gα, Gβ/Gγ) in Neurospora crassa (Wright et al., 2011). Targeted gene disruption (TGD) mutants of RIC8 display colony morphology that is typical of LY‐30, but atypical of the wild‐type M. oryzae strain. These mutants further display vegetative growth, conidiation, appressorium formation, fertility and penetration defects, which explain the accompanying defects in pathogenicity on abraded barley and rice leaves. In mammals, RIC8 homologues are fold chaperones and GEFs, as well as therapeutic targets for the inhibition of Gα subunit‐induced cancers (Papasergi et al., 2015). Therefore, RIC8 proteins seem to firmly influence G protein signalling across eukaryotic species, suggestive of their conserved status (Afshar et al., 2005; Papasergi et al., 2015; Peters and Rogers, 2013; Tall et al., 2003). The M. oryzae ATM collection of Li et al. (2010) is therefore a compelling target for the development of broad‐spectrum fungicidal agents.

The bidirectional genetics (BiG) collection

Recycled ectopic transformants were generated in a BiG platform in M. oryzae KJ201 (Park and Lee, 2013). About 11% (128/1139) of mutants in this collection showed defects in vegetative growth, melanization and asexual reproduction (Park and Lee, 2013; Table 1). These BiG mutants were constructed by TGD‐based forward genetics, and reverse genetics involving the recycling of ectopic mutant isolates generated by random integration in MGG_09225 and MGG_14496 genes. MGG_09225 codes for a putative acetate transmembrane transporter with a GPR1/FUN34/YaaH motif that is conserved in many species, including Aspergillus nidulans, Saccharomyces cerevisiae and Yarrowia lipolytica (Gentsch et al., 2007; Robellet et al., 2008). MGG_14496 codes for a Forkhead transcription regulator that is implicated in cellular development, differentiation and immunity (Coffer and Burgering, 2004; Friedman and Kaestner, 2006; Park et al., 2014). The application of the BiG platform was verified by the assessment of disruptions in the putative histidinol‐phosphate aminotransferase gene, the promoter region of a homologue of S. cerevisiae vacuolar sorting‐associated protein 74 gene and a region homologous to the N. crassa frequency protein encoding gene. Park and Lee (2013) provided a detailed technical evaluation of this collection for use in fungal genetic research. More mutant collections are yet to be constructed using this platform. The BiG platform will nonetheless be useful for mutant generation in filamentous fungi where TGD strategies often result in the generation of ectopic transformants, in addition to the desired mutant strains.

Plasmid‐mediated integration (PMI) collections

Baldwin et al. (2010) reported 1170 F. graminearum insertion mutants that were created using PMI. These insertion mutants were obtained from an initial screen of 5000 primary transformants. Eight disease‐attenuated F. graminearum or daf mutants were identified by the assessment of pathogenicity on susceptible wheat cv. Bobwhite (Baldwin et al., 2010; Table 1). These daf mutants are defective in causing disease symptoms in grains and young seedlings. They are also unable to hinder root development and invade various plant tissue layers. Based on microscopic experiments, defects in mycelial formation caused the inability of daf mutants to negatively affect wheat development. DON production in some of these mutants (daf10, daf15, daf26 and daf38) was significantly lower relative to that of the wild‐type strain, with daf10 producing almost undetectable DON levels. This PMI collection was subjected to genetics‐ and culture‐based analysis using genome‐wide sequencing and DON‐inducing conditions, respectively. In a genetics‐based analysis, a deletion of a 350‐kb segment in daf10 containing 146 predicted genes was identified. A culture‐based analysis suggested that this segment may contribute to the observed lower DON levels in daf10. Presumably, this is a result of the fact that some of the genes in this region, including TRI1, potentially contribute to DON production and pathogenesis. Interestingly, the loss of a 350‐kb segment also had an impact on other genomic parts of daf10 (Baldwin et al., 2010). These included 93 genes located outside the deleted segment that were expressed more strongly in daf10 than in the wild‐type strain under DON‐inducing conditions. Some of these genes, including TRI15 and TRI101, are key transcription factors (TFs) in DON production. These findings indicate that F. graminearum DON production is a complex process that is controlled by a number of genes, some of which are still unknown. Although the combinatorial effect of these predicted genes is established in DON production, mutants of each gene should be investigated individually for their role in DON production.

In a separate PMI‐derived collection of 650 insertion mutants, eight F. graminearum strains with significant virulence reduction were identified by Urban et al. (2015) (Table 1). Tantamount to what Baldwin et al. (2010) had observed using whole‐genome sequencing, some insertions in the F. graminearum genome showed broad deletions in segments of several chromosomes (Urban et al., 2015). These included chromosomes 1 and 4, where single insertion sites were accompanied by a 5′‐end deletion of 503.6‐kb and 182.4‐kb segments, respectively. As mutants in this collection show highly reduced virulence, it may be possible that some predicted genes in these deleted chromosomal segments take part in DON production and regulation.

The F. graminearum PMI insertion collections highlight the role of whole‐genome sequencing and bioinformatics tools, such as GeneChips (Baldwin et al., 2010) and FindInsertSeq (Urban et al., 2015), in the deep analysis of mutants with complex phenotypes. Noteworthy, these collections also highlight how insertion mutant generation can cause unexpected genomic alterations that are often associated with these complex mutant phenotypes.

TGD collection of differentially expressed genes

Kim et al. (2015) constructed 127 knock‐out strains in F. graminearum strain Z3643 from 127 genes that were differentially expressed during sexual development (Table 1). These genes were selected from 1245 genes based on their expression profiles and putative functional roles (Kim et al., 2015). Forty genes in this collection control sexual and hyphal development, as well as virulence and a number of pleiotropic phenotypes. Mutants of genes controlling sexual development formed barren perithecia, a smaller number of perithecia or no perithecia. Other mutants showed larger perithecia, delayed production of perithecia and perithecia with stage‐specific defects (Kim et al., 2015). Mutants of genes controlling pleiotropic phenotypes represented 12 genes. Among these genes was MYT2, a gene coding for a regulator of perithecium size, and ICL1, coding for the isocitrate lyase required for perithecium development, but not for virulence, in F. graminearum (Lee et al., 2009). One pleiotropic mutant disrupted for FGSG_07368 was identified from a group of 50 differentially expressed TFs. This mutant produced three times more ZEA and a smaller number of fertile perithecia than did the wild‐type strain (Kim et al., 2015). This suggests that FGSG_07368 down‐regulates genes related to ZEA production and up‐regulates those controlling perithecium fertility. Thus, ZEA toxin production and sexual development are possibly regulated at different infection stages by FGSG_07368. Other genes from the F. graminearum collection of Kim et al. (2015), involved in hyphal development and virulence were also characterized. These included some genes previously uncharacterized, and FGL1, which plays a role in wheat and maize cob infections (Voigt et al., 2005). Overall, genes in this TGD collection were selected based on functional similarity with genes required for sexual development, hyphal growth, virulence and toxin production. Therefore, Kim et al. (2015) provided an assessment of how these processes are coordinated to allow the adaptation of F. graminearum to dynamic environmental conditions.

TGD collection of secreted protein genes

A TGD method was used to generate an M. oryzae collection of 78 mutants representing genes coding for putative secreted proteins (Saitoh et al., 2012). A disruption in one of these genes, named MC69, produced a TGD mutant that was defective in virulence on susceptible barley cotyledons and rice cv. Shin No. 2 (Saitoh et al., 2012; Table 1). Plant infection assays showed that mc69 cannot induce appressorium‐mediated penetration or form invasive hyphae against barley and rice plants. The MC69 gene is constitutively expressed and encodes a secreted protein required for rice infections in the wild‐type M. oryzae strain. The amino acid sequence of the MC69 protein is conserved and is presumed to be essential for virulence in other fungi (Saitoh et al., 2012). Indeed, Saitoh et al. (2012) found that this protein plays a crucial role in the pathogenicity of the cucumber anthracnose fungus, Colletotrichum orbiculare, as deletion of its homologous gene results in reduced lesions on host cucumber and N. benthamiana leaves. In view of these findings, MC69 appears to be both a candidate effector gene and a probable target for broad‐spectrum fungicides that may be developed in future. As shown later in this review, the mutant collection of Saitoh et al. (2012) is valuable for searching for candidate fungal effectors.

Identification of TFs

Identification of Zn₂Cys₆ zinc TFs

Sequence‐specific DNA‐binding proteins include TFs that regulate a myriad of transcriptional changes triggered in response to environmental stimuli. The Zn₂Cys₆ zinc binuclear cluster is a predominant class of TFs in fungi (Todd et al., 2014). Lu et al. (2014) constructed 104 Zn₂Cys₆ zinc TF mutants using a high‐throughput shuttle vector containing functional elements that allow replication in S. cerevisiae, Escherichia coli and A. tumefaciens. Fifty‐eight per cent of these mutants represent TF genes controlling mycelial growth, conidiation and appressorium development, as well as M. oryzae barley and rice pathogenicity (Lu et al., 2014; Table 1). Some of the TFs represented by these mutants regulate multiple processes, arguing for the presence of interconnected networks in filamentous fungi. These include CNF1 (conidial production negative regulatory factor 1) and GPF1 (growth and pathogenicity regulatory factor 1), whose mutants are severely affected in pathogenicity. Indeed, gene expression profiles of these mutants indicate that networks controlled by CNF1 and GPF1 are highly interconnected in M. oryzae; CNF1 and GPF1 co‐regulate 2030 genes, including other TFs. Strikingly, the expression of 2021 of these co‐regulated genes is unidirectional, with 1105 and 916 genes down‐regulated and up‐regulated by CNF1 and GPF1, respectively. It will be interesting to determine whether these TFs are co‐expressed in planta or complement each other through the binding of downstream targets, and how this affects overall M. oryzae virulence. Therefore, Lu et al. (2014) have provided fresh insights into the controlled fundamental processes in M. oryzae. This study has the potential to guide the development of fungicides that target TFs responsible for ‘switching on’ M. oryzae development and virulence genes.

Identification of Cys₂‐His₂ zinc TFs

The Cys₂‐His₂ zinc finger TF cluster is the second largest class of TFs in ascomycetes, but is predominant in basidiomycete fungi (Todd et al., 2014). Cao et al. (2016) constructed 47 null mutants disrupted for these TFs in M. oryzae 70‐15 using a previously reported method (Lu et al., 2014; Table 1). These TF mutants were obtained from an initial screening of 5414 primary transformants. Forty‐four of the 47 mutants represent genes regulating M. oryzae development and pathogenicity. With the exception of a single disruptant in MGG_15508, 46 of the 47 mutants in this TF collection contain at least two independently derived isolates. Among Cys₂‐His₂ zinc finger TFs identified to control development were eight TFs (CREA, CON7, GCP1, GCF2, GPF2, GPF3, GCF3 and REI1) involved in the negative regulation of M. oryzae mycelial growth on complete agar medium. Several mutants, including gcf3, vrf2, gcf2 and creA, were altered in mycelial growth and displayed differences in growth inhibition relative to the wild‐type strain towards abiotic stress‐inducing cues, such as ionic and nutrient stress, as well as hypertonic pressure (Cao et al., 2016). Another key observation was that olive oil promoted the growth of creA, whereas sorbitol, the reduced form of glucose, suppressed growth in this mutant. Consistently, creA displayed high expression of lipid catabolism genes in the presence of glucose relative to the wild‐type strain. These observations suggest that glucose is essential for the activation of M. oryzae CREA, which ultimately represses lipid catabolism genes.

Cao et al. (2016) also identified 28 genes involved in M. oryzae conidiation. Among these genes, disruption of virulence regulatory factor 1, or VRF1, caused the abnormal growth of appressoria and subsequent reduction in melanin. This gene was also highly expressed during appressorial formation in the wild‐type strain, which supports its importance for melanin accumulation and appressorial development in M. oryzae. These findings link VRF1 with M. oryzae pathogenicity. Consistently, VRF1, together with VRF2, was among the genes required for pathogenicity on rice and barley. VRF1 and VRF2 mutants produced almost no blast symptoms on rice leaves, and displayed severe penetration and invasion defects on barley leaves, compared with other mutants and the wild‐type strain. Genome‐wide transcriptional analysis of vrf1 and vrf2 indicated stage‐specific regulation of genes by VRF1 and VRF2. An early infection stage is controlled by VRF1 through the regulation of 2514 genes, some of which are associated with appressorium development and penetration of the plant cuticle. The late infection stage, however, is controlled by VRF2, which regulates 1045 genes required for mycelial development and plant cell invasion. The fact that genes regulated by VRF1 more than double those regulated by VRF2 may be a result of M. oryzae's requirement to recognize, attach and adjust to the plant's surface, and to overcome physical barriers. As M. oryzae also occupies the plant's surface with other microbes and invaders, some of these VRF1 regulated genes may have roles in fitness for competition of space and resources. Therefore, the study of Cao et al. (2016) spotlights new regulatory networks that depend on Cys₂‐His₂ zinc finger TFs directed to fine tune M. oryzae developmental stages, survival and virulence progress.

Identification of TFs from various families

Son et al. (2011) constructed an F. graminearum mutant collection of 657 mutants deleted for putative TFs. This collection was derived from an initial screening of 8741 primary transformants. Represented TFs come from various families, including Zn₂Cys₆ and Cys₂‐His₂, winged helix repressor DNA‐binding, nucleic acid‐binding, OB‐fold, Myb and b‐ZIP families. TF mutant strains in this collection correspond to over 11 000 phenotypes derived from 17 phenotypic assays, which include mycelial growth, sexual development, virulence, toxin production and stress adaptation. This is by far the largest TF collection in filamentous fungi, and reports 170 mutant phenotypically altered strains constructed in the background of F. graminearum strain GZ3639 (Table 1).

During screening of this TF collection, 124 mutants displayed multiple phenotypes, whereas 46 displayed individual phenotypes (Son et al., 2011). Of these mutants, 31 were affected for perithecium development, virulence, growth or DON production, with 14 TF mutants from this group producing no perithecia and DON, and showing reduced virulence. Fusarium graminearum TF strains with multiple phenotypes showed phenotypic correlation between sexual development, virulence, growth and toxin production, as well as correlation between osmotic and reactive oxygen species (ROS) stress (Son et al., 2011). A similar trend of TF phenotypic correlation of virulence, growth and stress was observed in the encapsulated human pathogenic yeast, Cryptococcus neoformans (Jung et al., 2015). Screening the F. graminearum TF collection has therefore identified highly interlocked and possibly conserved fungal networks that underlie phenotypic correlations, which may allow F. graminearum to bypass fitness barriers exerted by plant hosts.

Identification of Fungal Effectors

Symbiotic microbes, such as rhizobacteria, interact harmlessly with plants, and this interaction often benefits both the resident microbe and the plant. Mutations in genes coding for surface molecules on these microbes enable symbionts to evade the initial plant immune response, called pathogen‐triggered immunity (PTI) (Pel and Pieterse, 2012; Trdá et al., 2014). However, pathogenic microbes that subvert PTI can impair plant growth and damage leaf or root surface tissues (Dodds and Rathjen, 2010). This occurs after these pathogens have obtained access to the plant's interior through wounds, stomatal openings or gas exchange pores. PTI‐adapted pathogens can also release avirulence (Avr) effectors and target them to specific plant locales (Stergiopoulos and de Wit, 2009). Apoplastic effectors are translocated in the space outside the plant cell membrane, whereas cytoplasmic effectors are translocated inside the plant. In resistant host plants, effectors can interact with the matching plant's resistance (R) gene and activate effector‐triggered immunity (ETI) (Dodds and Rathjen, 2010; Lo Presti et al., 2015). This Avr–R interaction prevents pathogen colonization by activating programmed cell death, called the hypersensitive response (HR), which deprives the pathogen of plant cell‐derived nutrients (Dodds and Rathjen, 2010; Lo Presti et al., 2015). However, in plants that do not harbour R gene(s) to the corresponding Avr gene(s), pathogen effectors can manipulate downstream HR‐activating genes and other ETI‐dependent response pathways, such as hormone signalling, resulting in effector‐triggered susceptibility (Kazan and Lyons, 2014). Therefore, Avr effectors can alter host cell structure and function, and make the host environment conducive to pathogen colonization (Dodds and Rathjen, 2010; Kazan and Lyons, 2014; Lo Presti et al., 2015). For this reason, the identification of effectors and an understanding of their mechanisms of action are important for plant resistance breeding.

Effectors facilitate feeding schemes, such as biotrophy, where feeding occurs on living plant tissues. This is exemplified by pathogens such as the flax rust, bean rust and wheat stem rust fungi (Ellis et al., 2007, 2009; Kemen et al., 2005; Upadhyaya et al., 2014). A large number of effectors have been identified in M. oryzae which facilitate feeding on both living and dead plant tissues. These effectors include AvrPi54, Avr‐Pia, Avr‐Pii, Avr‐Pik/km/kp, Avr‐Pita, Pwl1, Pwl2, Bas2, Avr‐Piz‐t, Avr‐Pi9, Avr‐Pib, Iug6, Iug9 and Slp1 (Dong et al., 2015; Khang et al., 2010; Mentlak et al., 2012; Oliveira‐Garcia and Valent, 2015; Ray et al., 2016; Wu et al., 2015; Yoshida et al., 2009; Zhang and Xu, 2014; Zhang et al., 2015). Other studies have started to characterize plant cell death‐inducing effectors in M. oryzae (Chen et al., 2013) and the rice false smut fungus, Ustilaginoidea virens (Fang et al., 2016). These types of effectors are highly expressed during necrotrophy, a stage that allows the growth and feeding of fungal pathogens on dead plant tissues. Therefore, effectors that facilitate growth and feeding on living and dead plant tissues sustain the hemibiotrophic lifestyle of pathogens, such as M. oryzae. Putative effectors, such as lipolytic enzymes (e.g. LIP1) and endoxylanases (e.g. XYLA), have been demonstrated to breach plant defences and elicit immune reactions in plants in the hemibiotrophic fungus F. graminearum (Feng et al., 2005; Sella et al., 2013). Recently, Sperschneider et al. (2016) have predicted 292 candidate effectors in this fungus, 42.5% of which are putative cytoplasmic effectors as their expression occurs in planta in wheat and barley. Other studies have identified 63 secreted proteins, many of which are hydrolytic enzymes, in F. graminearum (Ji et al., 2013). Apoplastic effectors belong to this group of enzymes as they can mediate plant cell breakdown at the apoplastic space. Recently, 34 cysteine‐rich F. graminearum candidate effectors have been identified by Lu and Edwards (2016). These were expressed in infected wheat heads, with expression in half of them correlating with FHB development. Functional analyses will determine whether they interact with the plant's immune responses.

The use of mutant screens in fungal effector identification is limited. To date, only M. oryzae mutant screens have been useful in the identification of candidate effectors. These screens include that of the ATM collection, screened on susceptible rice cv. Nakdongbyeo (Chi et al., 2009). The Deltades1 mutant, disrupted for a pathogenicity gene required for plant innate defence suppression (DES1) through scavenging of apoplastic host‐derived ROS, was identified in this screen (Chi et al., 2009). The authors observed that the DES1 disruption mutant is unable to prevent ROS accumulation and displays hypersensitivity towards H₂O₂ concentrations that cause oxidative damage to fungal cells. Therefore, the DES1‐mediated ROS scavenging pathway may counteract PTI during initial plant immunity. PTI is responsible for rapid and transient accumulation of apoplastic ROS at the pathogen attack site, and can prevent pathogen colonization (Lamb and Dixon, 1997). Although DES1 interacts indirectly with PTI, sequence comparison with homologues in other fungi indicates that the DES1 protein lacks a signal peptide, a characteristic feature of effector proteins (Chi et al., 2009). Screening of the fungal TGD collection identified the MC69 gene in M. oryzae that is conserved in C. orbiculare (Saitoh et al., 2012). This gene affects pathogenicity in these fungal pathogens towards their inherent hosts (Saitoh et al., 2012). Although the MC69 protein contains a secretory signal peptide, it is not clear whether this protein interferes directly or indirectly with PTI or ETI responses in plants (Saitoh et al., 2012). Recently, Mogga et al. (2016) have identified the M. oryzae hypothetical effector gene 13 (HEG13) with virulence function on barley cv. Vada. The resultant mutant was originally constructed in the M. oryzae TGD collection and was named mc33 (Saitoh et al., 2012). In this study, the mc33 mutant did not affect the pathogenicity of M. oryzae when it was inoculated on the immune‐compromised barley cv. Nigrate, as well as on susceptible rice cv. Shin No. 2. Taken together, these data suggest that the secreted effector candidate gene disruptant may show different phenotypes if the host cultivar is changed (Mogga et al., 2016; Saitoh et al., 2012).

Citations of Mutant Collections

Citations reflect the seminal stance and relevance of collections in science. We conducted a scientific literature search using Google Scholar for reviews, theses, research articles, books and book chapters citing F. graminearum and M. oryzae collections discussed in this review. The F. graminearum TGD collection of differentially expressed genes (Kim et al., 2015) and the M. oryzae TF collection (Cao et al., 2016) were excluded from this analysis. These collections were not cited by any article by the time (31 August 2016) the literature search was finalized. We also excluded 20 scientific papers citing mutant collections that were either not translatable to the English language or appeared more than once in our search results. Therefore, our analysis is based on 116 articles citing three F. graminearum mutant collections (referred to as Fg PMI yr2010, Fg PMI yr2015 and Fg TF) and 281 articles citing five M. oryzae mutant collections (referred to as Mo ATM yr2007, Mo ATM yr2010, Mo BiG, Mo TGD and Mo TF yr2014) (Fig. 1). These names are used for discussions in this section. To conduct a more thorough analysis, citations were split into 12 different scientific aspects. We found that more citations of mutant collections came from research articles and theses than from reviews and books or book chapters (Fig. 1A). Combined, these represent the actual number of scientific citations. However, our discussions of key observations cover citations in terms of specific aspects. Also taken into account are mutant collections that appear in more than one category. These include the BiG collection which appears in the ‘Development of functional genomics tools’ and ‘Mutant collections’ categories.

Figure 1.

Citations of Fusarium graminearum (Fg) and Magnaporthe oryzae (Mo) mutant collections. (A) Citations based on scientific aspects. (B) Number of citations recorded each year since Mo ATM yr2007 (a) and Fg TF (b) collections were made public. ATM, Agrobacterium tumefaciens‐mediated; BiG, bidirectional genetics; PMI, plasmid‐mediated integration; TGD, targeted gene disruption; TF, transcription factor.

Mo ATM yr2007, reported by Jeon et al. (2007), is one of the oldest collections constructed using ATM transformation. This mutant collection has been cited by 52 articles describing the development of functional genomics tools (Fig. 1A), with most articles describing the use of ATM transformation. The presumption is that the recognition of the Mo ATM yr2007 collection is a result of the fact that ATM transformation was an emerging tool at the time it was released. Mo ATM yr2007 has also been cited by 47 articles describing the identification and characterization of genes other than effectors and TFs (Fig. 1A). Mutants of genes characterized in these articles were derived using the ATM method. Mo TGD is currently the only M. oryzae collection describing the role of secreted proteins in fungal pathogenicity (Saitoh et al., 2012). This collection was subsequently cited by 32 articles reporting on effectors (Fig. 1A), arguing for its importance in effector research. Mo ATM yr2010 of Li et al. (2010) contains the largest number of citations in the G protein signalling group (Fig. 1A). Screening of this collection shares important insights into G protein signalling in M. oryzae (Li et al. 2010). This is an aspect of interest in upstream signalling for many eukaryotic species, including fungal pathogens. The Fg TF collection describes a large number of TFs belonging to different families using an array of mutant phenotypes (Son et al., 2011). Accordingly, many TF articles have cited this collection rather than the Mo TF collection, which focuses only on Zn₂Cys₆ zinc TFs (Lu et al., 2014). The latter collection has been cited by several articles analysing autophagy (degradation of cytoplasmic components), some of which characterized TFs, such as CCA1, associated with components of this pathway.

Although mutant collections are cited by fewer articles in the comparative analyses category, the numbers are two‐fold higher in the non‐fungal category (Fig. 1A). In this category, we have included organisms simpler than filamentous fungi, such as bacteria and yeasts, and those with a more complex biology, such as insects and plants. These comparative studies permit research on commonalities between plant‐, animal‐ and human‐pathogenic species. For instance, a study reporting a TF collection of a human‐pathogenic yeast, Cr. neoformans, cited most collections in this non‐fungal group (Jung et al., 2015). Some similarities between pleiotropic TF mutants of Cr. neoformans and F. graminearum can be noted. These include correlations between virulence, growth and stress in these fungi. Therefore, these comparisons may point to regulatory networks with evolutionarily conserved frameworks and functions across pathogens with unparalleled hosts. Likewise, comparative analysis between non‐fungal and fungal pathogens may underscore unique processes that influence host or niche preferences.

The numbers of citations generated per year following the release of the Mo ATM yr2007 and Fg TF collections were also surveyed (Fig. 1Ba,b). These mutant collections show the most literature citations which can help to predict citation trends of emerging collections (Fig. 1A). For instance, we can predict from these data that some mutant collections may receive little (e.g. Mo ATM yr2007) or no (e.g. Fg TF) recognition in the year in which they are released, but become more recognized in the years that follow (Fig. 1Ba,b). Collections not yet cited, such as the Mo TGD (Kim et al., 2015) and Mo TF (Cao et al., 2016) collections, also display this trend. Another factor contributing to missing citations in the year of release is the effect of some collections having to spill over to the following year. This implies that collections may still be recognized and cited in the same year in which they are released. However, this will potentially be delayed by factors such as long manuscript developments and journal review processes. Citations of the Fg PMI collection show this trend (Urban et al., 2015). This collection was released in April 2015, but was cited by an article that was probably submitted towards the end of 2015, but published the following year in February (Lu and Edwards, 2016). The yearly citation data also reflect the impact of scientific developments on citations of fungal mutant collections. For instance, Mo ATM yr2007 was cited by more articles between 2008 and 2009 (Fig. 1Ba). These were a combination of articles on gene identification, development of functional genomics and bioinformatics tools. The Fg TF collection was cited from 2012 to 2013, with a larger number of articles in 2014 compared with previous years (Fig. 1Bb). Interestingly, of the 65 articles published during this time, more than one‐half (38) analysed regulatory processes, particularly involving the role of TFs in these processes.

Limitations

Limitations in gene discovery

The number of transformants produced during mutant collection construction decreases substantially after initial screening. This results in fewer genes being discovered (Table 1). For instance, screening of 21 070 and 5248 M. oryzae ATM transformants, as well as 78 TGD transformants, resulted in only 2.6%, 0.1% and 1.2% transformants, respectively, that were phenotypically altered (Table 1). These transformants represented 203 pathogenicity genes and one pathogenicity gene, respectively, in M. oryzae (Table 1). This warrants advanced screening technologies that can detect subtle changes in mutants that show wild‐type properties towards the conditions used during initial screening. Perhaps, the use of phenotypic microarrays (PMs) can meet this requirement. PMs measure respiration signals instead of growth‐related phenotypes, suggesting that even small phenotypic alterations may be detected (Gardiner et al., 2009; Homann et al., 2005; Lei and Bochner, 2013). As explained below, mutant collections with less phenotypically altered transformants should also be tested against different sets of conditions by other research groups.

Lack of additional screening

The applicability of the fungal mutant collections remains at arm's length as only a few have been additionally screened. This increases the possibility that the contribution of the majority of collections will go unnoticed based on limited evidence. Some of the additionally screened collections include the ATM collection, screened on susceptible rice cv. Nakdongbyeo; this screening identified DES1, a gene involved in the scavenging of host‐derived ROS (Chi et al., 2009). The M. oryzae mutant, mc33, originally characterized by screening the TGD collection on susceptible barley cv. Nigrate and rice cv. Shin No. 2 (Saitoh et al., 2012), was recently characterized on barley cv. Vada (Mogga et al., 2016). This screening showed that mc33 is attenuated on this cultivar, but not on barley cv. Nigrate and rice cv. Shin No. 2, used by Saitoh et al. (2012). Therefore, MC33 shows a cultivar‐specific function in virulence. This cultivar‐specific function could have been missed if no additional screening had been conducted. The F. graminearum TF collection has undergone additional phenotypic screening (Son et al., 2011). This screening identified TFs involved in the regulation of responses to osmotic stress and DNA damaging chemicals (Son et al., 2015, 2016). These studies indicate that additional screens may uncover previously under‐represented roles of already discovered genes.

Limitations in vetting of mutant collections

Deletion or insertion strains may harbour unexpected mutations which often complicate phenotypic analysis and lead to erroneous conclusions. However, several vetting strategies have been implemented to rule out phenotypes resulting from these mutations in the generated mutant strains. These strategies include complementation, which involves integration of the deleted gene into its original locus to restore wild‐type properties. Examples include the PMI strains constructed by Baldwin et al. (2010) and Urban et al. (2015). These collections display unexpected mutations in the form of chromosomal deletions. In these examples, whole‐genome sequencing was used to vet insertion mutants and to provide the genetic basis of the observed phenotypes. Although this strategy seems fitting for the vetting of mutants in these PMI collections, it will be challenging to implement strategies, such as complementation, as deleted chromosomal segments are often large. In addition, Baldwin et al. (2010) and Urban et al. (2015) verified insertions only in some daf mutants. Vetting will be required for the remainder of the mutants to formally use them in fungal genetic studies. Jeon et al. (2007) vetted the M. oryzae collection using TGD and the characterization of T‐DNA insertion sites from ATM mutant strains. Complementation of these TGD mutants and other vetting strategies will be required to resolve potential phenotypic inconsistencies which may occur in different culture conditions.

Conclusions and Future Work

The role of mutant collections in areas such as effector biology remains inadequate. Only the M. oryzae TGD collection of secreted proteins has set the groundwork for candidate effector search and the understanding of R gene‐mediated host resistance. The broadest meaning of the term ‘effector’, inclusive of any pathogen‐derived molecule that interferes with the normal functioning of host defences, has been proposed (Kazan and Lyons, 2014). Based on this broad definition, the contribution of mutant collections in effector discovery may be substantial. An exciting opportunity exists to use available TF collections to guide the development of alternative and effective fungicides. Such fungicides are expected to target conserved TF proteins, instead of their countless downstream targets. This will substantially cut the associated developmental costs of fungicides. Considering effector and TF discovery, a unified approach with potential significance in fungal disease control programmes is conceivable. This approach will coalesce effector‐assisted breeding and biological control with TFs as potential targets.

Affordable genome sequencing platforms and ongoing improvements in bioinformatics tools will increase the number of sequenced and annotated fungal genomes. As a result, the past decade of fungal research has experienced a surge in the phenotypic information derived from mutant collection screening. A better way in which this amount of data can be processed effectively is to develop a dedicated and accessible web‐based resource that is easy to use. A resource like this could be linked to other web‐based resources, including those designed to support sequencing projects. An intuitive way in which molecular plant pathologists could conveniently query data from these collections using this proposed web‐based resource would be an integrated view of such data. Pathologists would also be allowed to update this database with information from newly generated collections coming from screens performed from different laboratories. Therefore, this web‐based resource would encourage molecular plant pathologists to readily preserve information from mutant collections.