Host‐induced gene silencing: a tool for understanding fungal host interaction and for developing novel disease control strategies

SUMMARY

Recent discoveries regarding small RNAs and the mechanisms of gene silencing are providing new opportunities to explore fungal pathogen–host interactions and potential strategies for novel disease control. Plant pathogenic fungi are a constant and major threat to global food security; they represent the largest group of disease‐causing agents on crop plants on the planet. An initial understanding of RNA silencing mechanisms and small RNAs was derived from model fungi. Now, new knowledge with practical implications for RNA silencing is beginning to emerge from the study of plant–fungus interactions. Recent studies have shown that the expression of silencing constructs in plants designed on fungal genes can specifically silence their targets in invading pathogenic fungi, such as Fusarium verticillioides, Blumeria graminis and Puccinia striiformis f.sp. tritici. Here, we highlight the important general aspects of RNA silencing mechanisms and emphasize recent findings from plant pathogenic fungi. Strategies to employ RNA silencing to investigate the basis of fungal pathogenesis are discussed. Finally, we address important aspects for the development of fungal‐derived resistance through the expression of silencing constructs in host plants as a powerful strategy to control fungal disease.

INTRODUCTION

More than 25 years ago, Sanford and Johnston (1985) proposed the concept of parasite‐derived resistance (PDR) as a radical strategy to control plant disease. In PDR, the expression of a parasite gene in the plant host interferes with the parasite's ability to grow and develop, resulting in disease resistance. Perhaps the most successful application of PDR is the use of transgenic papaya plants expressing Papaya ringspot virus (PSRV) coat protein to protect against virus infection (Gonsalves, 1998). With the exception of viruses, it was generally believed that the lack of ready macromolecule exchange between host and parasites would be a major impediment to more widespread development of PDR.

Over the past decade or so, the discovery of small RNAs and the elucidation of RNA silencing mechanisms have rekindled an interest in PDR. Today, it is strongly suggested, on the basis of reporter gene silencing, that RNA molecules traffic between host and parasites (Tinoco et al., 2010; Tomilov et al., 2008; Westwood et al., 2009). Growing evidence suggests that RNA silencing constitutes a crucial component of many plant–parasite interactions in addition to viruses, including responses to insects, bacteria, nematodes and fungal infections (Huang et al., 2006; Katiyar‐Agarwal and Jin, 2010; Mao et al., 2007; Ruiz‐Ferrer and Voinnet, 2009; Turner et al., 2006). Several very recent published reports on the use of host‐induced gene silencing (HIGS) to control invading fungi are probable forerunners of many more applications to come.

In this article, we describe: (i) the central components of RNA silencing, emphasizing recent discoveries in plant pathogenic fungi (PPF); (ii) the fundamental principles of the use of RNA silencing to study PPF; (iii) applications in HIGS using fungus‐derived genes to confer disease resistance; and (iv) future perspectives of HIGS for the control of fungal disease.

GENERAL MECHANISM OF EUKARYOTIC RNA SILENCING

Approximately 20 years ago, anomalies observed during efforts to functionally characterize genes led to the discovery of post‐transcriptional gene silencing (PTGS). This phenomenon is known as co‐suppression in plants, quelling in Neurospora crassa and RNA interference (RNAi) in Caenorhabditis elegans and other animals (Lee et al., 1993; Napoli et al., 1990; Romano and Macino, 1992). Subsequent further independent investigations revealed that much of the underlying machinery and mechanisms is conserved across organisms.

PTGS can be generalized as a sequence‐specific regulatory mechanism for the silencing of gene expression triggered by long double‐stranded (dsRNA) or hairpin‐structured RNAs. These molecules are processed into 21–35‐nucleotide‐long single‐stranded RNAs that associate with and direct the RNA‐induced silencing complex (RISC) to the target transcript. This results in mRNA degradation or translational repression (Fig. 1) (Esquela‐Kerscher and Slack, 2006). There are distinct classes of small RNA molecules, such as short interfering RNAs (siRNAs), microRNAs (miRNAs), tRNA‐derived RNA fragments (tRFs) and Piwi‐interacting small RNAs (piRNAs). Currently, siRNAs and miRNAs are the best elucidated classes of small RNAs in plants and animals, and are discussed in more detail. Although it is clear that siRNA and miRNA processing pathways are distinct, in plants they typically begin in the nucleus, where Dicer and members of the ribonuclease III (RNase III) family bind to long dsRNA or hairpin RNAs and process them into siRNAs or miRNAs, respectively. In animals, Drosha is responsible for the slicing off of the long stem‐loop dsRNA structure, which is then processed into siRNAs or miRNAs by Dicer (for more details, see elsewhere: Baulcombe, 2004; Carthew and Sontheimer, 2009; Miyoshi et al., 2010; Sato et al., 2011). To our knowledge, it is unclear how siRNA is exported from the nucleus. In contrast, endogenously derived short hairpin RNA molecules (miRNA) are transported to the cytoplasm by a nuclear pore protein named Exportin‐5 (Katahira and Yoneda, 2011; Kim, 2004; Lund et al., 2004; Yi et al., 2003). In other instances, short dsRNA molecules may be derived from exogenous sources, such as viruses replicating in plant cells. Once in the cytoplasm, endogenously or exogenously derived short dsRNA molecules are processed into small dsRNA, and loaded into RISC. Here the ‘passage’ strand is eliminated, and the ‘guide’ RNA strand directs Argonaute to specifically degrade the target RNA transcript through the slicing ribonuclease activity of Argonaute or to interfere with translation (Esquela‐Kerscher and Slack, 2006). These small RNAs can also be used as primers by RNA dependent‐RNA polymerase (RdRp) to resynthesize dsRNAs (Melnyk et al., 2011). This amplification step ensures that several rounds of RNA silencing are triggered. Several members of the Argonaute family have been described. For example, AGO associates with 21–25‐nucleotide‐long small RNAs (siRNAs and miRNAs), whereas PIWI binds to 23–30‐nucleotide‐long RNA (piRNAs) molecules to silence repetitive element transcripts (Brennecke et al., 2007; Hartig et al., 2007). Here, we refer to the principle RNA silencing proteins as Dicer, Argonaute and RdRp.

Figure 1.

General mechanism and potential application of RNA silencing for the development of plant disease control strategies. In plant cells (A), endogenous RNA silencing is triggered by transcription of double‐stranded RNA (dsRNA) (or hairpin) molecules in the nucleus (A1). These precursor molecules are processed by ribonuclease III (RNase III) enzymes, such as Dicer (or Drosha in animals), producing RNA molecules, which are loaded and transported from the nucleus by Exportin‐5. In the cytoplasm, 21‐nucleotide dsRNAs are produced by Dicer activity (A3). On elimination of the passage strand, the guide single‐stranded RNA (ssRNA) associates with Argonaute (A4) and directs the RNA‐induced silencing complex (RISC) to bind to target endogenous transcripts (A5), resulting in the degradation of transcripts (A6). Unbound siRNA binds to mRNAs (A7) through the action of RNA‐dependent RNA polymerase (RdRp) and resynthesizes dsRNA which is cut into 21‐nt dsRNA molecules by Dicer (A8), leading to RNA silencing amplification. In pathogen‐infected plant cells (B), integration (B1) and transcription (B2) of the host‐induced gene silencing (HIGS) construct results in HIGS dsRNA molecules (or hairpins) that are processed into 21‐nt dsRNA molecules (B3 and B4; as explained in A). Alternatively, virus‐induced gene silencing (VIGS) (B5) is used to produce VIGS dsRNA precursors that are processed into 21‐nt dsRNA molecules by Dicer (B6). VIGS dsRNA molecules lead directly to the silencing of viral target transcripts in the plant cytoplasm (B7). The action of plant RdRp also leads to viral RNA silencing amplification (B8). HIGS dsRNA molecules are exported from plant cells and lead to the silencing of nematode targets on nematode feeding (B9), or the silencing of fungal transcripts on fungal infection (B10).

In fungi, a diverse number of small RNA classes have been described over the last few years. Recently, microRNA‐like RNAs (milRNAs) have been discovered in N. crassa (Lee et al., 2010). The molecular machinery involved in the biogenesis of milRNAs appears to be significantly different from that reported for miRNAs in plants and animals. Thus, it is still a matter of debate whether genuine miRNAs are present in filamentous fungi. Other small RNAs in fungi include quelling‐deficient 2 (QDE‐2)‐interacting small RNAs (qiRNAs) and Dicer‐independent small interfering RNAs (disiRNAs) in N. crassa (Lee et al., 2009, 2010). Endogenous short RNAs (esRNAs) have been discovered in Mucor circinelloides, and LTR retrotransposon‐siRNAs (LTR‐siRNAs) and tRFs in Magnaporthe oryzae and Aspergillus fumigatus (Jöchl et al., 2008; Nicolas et al., 2010; Nunes et al., 2011).

FUNDAMENTAL PRINCIPLES OF RNA SILENCING IN PPF

A knowledge of RNA silencing protein machinery in fungi was initially derived from studies of N. crassa and referred to as ‘quelling’ (Catalanotto et al., 2002, 2004; Cogoni and Macino, 1997; Fulci and Macino, 2007; Lee et al., 2009; Li et al., 2010b; Maiti et al., 2007; Nakayashiki, 2005; Nolan et al., 2008; Romano and Macino, 1992; Schumann et al., 2010; Xiao et al., 2010). Although quelling was first described in 1992, it was several years later that Cogoni and Macino discovered three distinct loci responsible for PTGS through the characterization of quelling‐deficient (qde) mutants (Cogoni and Macino, 1997). The qde‐1, qde‐2 and qde‐3 loci encode proteins that correspond to RdRp, Argonaute and RecQ DNA helicase, respectively. Small RNAs produced by Dicer associate with qde‐2 and require qde‐1 and qde‐3 for biogenesis (Catalanotto et al., 2002, 2004). Over the past few years, there have been several examples of RNA silencing in PPF, including members of the Ascomycota, Basidiomycota and Zygomycota, as well as in Phytophthora spp. (fungus‐like organisms; Oomycetes within the Kingdom Stramenopila) (Table 1).

Table 1.

Summary of RNA silencing in plant pathogenic fungi.

Phylum	Species	Targeted gene	Promoter used	RNA precursor	Comments	Reference
Ascomycota	Alternaria alternata	Putative hydrolase (ACTT2), a host‐selective ACT‐toxin	trpC *	Hairpin†	Silenced transformants were nonpathogenic	Miyamoto et al. (2008)
		Green fluorescent protein (GFP)	trpC *	Hairpin	Silencing of GFP suggests RNA silencing functionality	Miyamoto et al. (2008)
		Enoyl‐reductase (ACTTS2), a host‐selective ACT‐toxin	trpC *	Hairpin	Silencing of ACCTS2 transcripts resulted in lack of ACT‐toxin production and complete loss of pathogenicity	Ajiro et al. (2010)
	Aspergillus flavus and A. parasiticus	Transcription factor (aflR)‡	gpd *	Hairpin	The atoxigenic silenced strain was stable in corn and wheat	McDonald et al. (2005)
	Bipolaris oryzae	Polyketide synthase gene (PKS1)	trpC *	Hairpin	The silencing construct was stable during asexual propagation	Moriwaki et al. (2007)
	Botrytis cinerea	Superoxide dismutase (BCSOD1)	oliC *	Sense and antisense constructs	Both constructs reduced transcript levels but had no effect on virulence	Patel et al. (2008)
	Cladosporium fulvum	Hydrophobin gene (HCf‐1)	n/a	Antisense RNA	Silenced transformants reverted to the wild‐type	Hamada and Spanu (1998)
		First exons of six hydrophobin coding genes	gpd *	Chimeric hairpin	This work highlights that changes in gene expression of nontargeted genes pose a challenge for the silencing of multiple genes	Lacroix and Spanu (2009)
	Cochliobolus sativus	GFP, a host‐selective toxin (ToxA) and a polyketide synthase (CsPKS1)		Hairpin	Development of pSGate1 silencing vector	Leng et al. (2011)
					CsPKS1‐silenced strains resulted in albino phenotypes
	Colletotrichum gloeosporioides	Transcription factor (PAC1)	trpC *	Hairpin	pTroya, a Gateway RNA silencing vector, was developed from pSilent‐1 backbone for high‐throughput application in fungi	Shafran et al. (2008)
	Colletotrichum lagenarium	eGFP	trpC *	Hairpin	Development and application of pSilent‐1 to silence fungal transcripts	Nakayashiki et al. (2005)
	Fusarium graminearum	Transcription factor (Tri6)	gpd *	Hairpin	Silencing transformants were less virulent on wheat	McDonald et al. (2005)
	Fusarium solani f.sp. pisi	β(1,3)‐d‐glucan synthase (FsFKS1)	alcA	Hairpin	Imperfect cell wall formation led to reduced spore viability and lysis of spore and mycelia in silenced transformants	Ha et al. (2006)
	Fusarium solani	Chitosanase (CSN1)	gpd *	Hairpin	Silenced transformants were more virulent than wild‐type	Liu et al. (2010)
	Magnaporthe oryzae	eGFP	trpC *	Sense, antisense and hairpin RNA constructs	Hairpin construct more efficiently triggered eGFP silencing	Kadotani et al. (2003)
		MPG1	trpC *	Hairpin	Development of pSilent‐1 vector as tool for gene silencing for Ascomycete fungi	Nakayashiki et al. (2005)
		PKS‐like gene	trpC *	Hairpin	Extends application of the pSilent‐1 vector	Nakayashiki et al. (2005)
		eGFP	trpC *	dsRNA	Development of a vector with convergent opposing promoters vector [pSilent‐Dual1 (pSD1)];	Nguyen et al. (2008)
			gpd *
					pSD1 triggers RNA silencing slightly less efficiently than pSilent‐1
		37 genes involved in calcium signalling	trpC *	Chimeric dsRNA	This co‐silencing approach facilitates screening of transformants based on GFP fluorescence	Nguyen et al. (2008)
			gpd *
					Work identifies many calcium signalling proteins involved in hyphal growth, sporulation and pathogenicity
	Ophiostoma novo‐ulmi	Endopolygalacturonase (Epg1)	gdp	Hairpin	A longer stem‐loop RNA fragment (409 bp) resulted in greater reduction of mRNA transcripts compared with a 200‐bp fragment	Carneiro et al. (2010)
	Sclerotinia sclerotiorum	B regulatory subunit (rgb1) of 2A phosphoprotein phosphatase (PP2A)	trpC *	Hairpin	pSilent‐1 vector triggered high levels of silencing in two transformants which exhibited reduced hyphal radial growth, impaired sclerotial maturation and were unable to cause disease on tomato and Arabidopsis thaliana	Erental et al. (2007)
	Verticillium longisporum	Chorismate synthase (Vlaro2)	trpC *	Hairpin	pSilent‐1 vector was used and silenced 80% of Vlaro2 transcripts	Singh et al. (2010)
					Silenced transformants were able to infect Arabidopsis thaliana and Brassica napus
	Venturia inaequalis	Trihydroxynaphthalene reductase (THN) and GFP	gpdA §	Chimeric hairpin	Silenced transformants had a light brown phenotype and were able to infect apple (Royal Gala)	Fitzgerald et al. (2004)
Basidiomycota
	Melampsora lini	Effector protein (AvrL567)	AvrL567 native promoter	Hairpin	Silenced rust lines were virulent on flax plants carrying L6 immune receptor	Lawrence et al. (2010)
	Moniliophthora perniciosa	GFP, hydrophobin (MpHYD3) and 1‐cys peroxiredoxin (MpPRX1)	n/a	In vitro synthesized dsRNA	dsRNA shown to be feasible for transforming this fungus	Caribé dos Santos et al. (2009)
	Ustilago hordei	GUS and mating‐type gene (bW)	Hsp70 native promoter	Hairpin	U. hordei transformants (MAT1) mated with MAT‐2 exhibited reduced production of mating hyphae	Laurie et al. (2008)
Zygomycota
	Mucor circinelloides ¶	Carotenogenic gene (carB)	Native promoter		3% of silenced transformants showed albino phenotype	Nicolas et al. (2003)
Fungi‐like organism
	Phytophthora infestans	G‐protein β‐subunit encoding gene (Pigpb1)	Pigpb1 native promoter	Pigpb1 sense strand	Silenced transformants failed to sporulate	Latijnhouwers and Govers (2003)
		Cdc 14 coding gene (PiCdc14)	HAM34	Sense orientation	Transformants showed reduced sporulation	Ah Fong and Judelson (2003)
		G‐protein α‐subunit gene (Pigpa1)	Pigpa1 native promoter	Pigpa1 sense orientation	Transformants exhibited reduced zoospore production and infection on potato leaves	Latijnhouwers et al. (2004)
		GFP inf1 and cdc14	n/a	In vitro synthesized dsRNA	GFP, inf1 and cdc14 expression levels were reduced after exposure to dsRNA	Whisson et al. (2005)
		bZIP transcription factor (Pibzp1)	HAM34	Sense and antisense orientations	Silenced transformants showed abnormal zoospore movement, failed to develop appressoria and were incapable of infecting tomato leaflets	Blanco and Judelson (2005)
		Nuclear LIM interactor‐interacting factors (NIFC1 and NIFC2)	HAM34	Hairpin	Zoospore cyst germination was impaired by 60% in silenced NIFC transformants	Judelson and Tani (2007)
					Silencing occurred at the transcription level
		Inf1		Sense, antisense and hairpin RNA constructs	Hairpin most efficient method of silencing	Ah‐Fong et al. (2008)
		Putative glycosylated protein (Pihmp1)	n/a	In vitro synthesized dsRNA	Silenced lines showed loss of pathogenicity	Avrova et al. (2008)
		Putative ATP‐dependent DEAD‐box RNA‐helicase gene (Pi‐RNH1)	n/a	In vitro synthesized dsRNA	Pi‐RNH1‐silenced lines formed large aberrant zoospores that had multiple flagella and underwent partial cleavage	Walker et al. (2008)
		Four members of the CesA encoding for cellulose synthase genes	n/a	In vitro synthesized CesA1‐, CesA2‐, CesA3‐ and CesA4‐dsRNA	Silenced strains contained disrupted cell wall surrounding appressoria	Grenville‐Briggs et al. (2008)
					Cellulose content of the silenced strains was >50% lower than that of nonsilenced strains
		Effector protein (PiAVR3a)	HAM34	Hairpin	A PiAVR3a‐silenced line (CS12) was significantly reduced in pathogenicity on Solanum tuberosum cv Bintje (susceptible) and on Nicotiana benthamiana	Bos et al. (2010)
		Dicer‐like (Pidcl1), Argonaute (Piago1/2) and Histone deacetylase (Pihda1)	n/a	In vitro synthesized dsRNA	Stable Ns (nonsporulating, Picdc14‐silenced) transformant protoplasts were treated with dsRNA homologous to Pidcl1, Piago1/2 and Pihda1	Vetukuri et al. (2011)
					Regenerated lines of Pidcl1, Piago1/2 and Pihda1 showed an obvious sporulation
	P. parasitica var. nicotianae	A coding gene considered to be involved in cellulose‐binding (CB), elicitor (E) of defence in plants and lectin (L)‐like activities (CBEL)	Bremica lactucae HSP70	Antisense or sense constructs	Silenced transformants showed reduced attachment to cellulosic surfaces and cell wall thickening	Gaulin et al. (2002)
	P. sojae	Heterotrimeric G‐protein α subunit (PsGPA1)	HAM34	Antisense construct	Silenced PsGPA1 lines had abnormal zoospore chemotaxis, encystment and germination	Hua et al. (2008)
					Silenced transformants were unable to infect soybean
		C2H2 zinc finger transcription factor (PsCZF1)	HAM34 and native PsCZF1	Antisense construct	Hyphal growth rate of silenced transformants was reduced about 50% and oospore production, zoospore and cyst germination were impaired	Wang et al. (2009)
					Silenced strains lost virulence in soybean
		MAP kinase encoding gene (PsSAK1)	HAM34	Antisense construct	Silenced transformants showed faster encystment, reduced germination rate and longer germ tubes compared with wild‐type	Li et al. (2010a)
					Transformants were unable to colonize wounded and unwounded soybean leaves
		Putative seven‐transmembrane G‐protein‐coupled receptor (GPR11)	HAM34	Antisense construct	Zoospore release from sporangia was drastically impaired as well as zoospore encystment and germination	Wang et al. (2010)
					Silenced transformants lost pathogenicity to soybean
		PsYKT6, a conserved member gene of the soluble N‐ethylmaleimide‐sensitive factor attachment protein receptors (SNAREs)	HAM34	Antisense construct	Silencing of PsYKT6 revealed involvement of this gene in asexual development, sexual reproduction and pathogenesis on soybean cultivars	Zhao et al. (2011)
					Antisense constructs were stable in all three transformants
		Crinkling‐ and necrosis‐inducing proteins (CRN) (PsCRN63 and PsCRN115)	HAM34	Sense construct	Silenced transformants were unable to suppress host cell death	Liu et al. (2011)

^* Aspergillus nidulans promoters.

^† Hairpin = inverted repeats.

^‡ Delivery of synthetic siRNAs into Aspergillus flavus protoplasts resulted in significant reduction in aflD transcripts (Abdel‐Hadi et al., 2011).

^§ Glomerella cingulata.

^¶ Nonpathogenic fungus, but it is a relevant organism within Zygomycota.

n/a, not applicable.

As shown in Table 1, a variety of target genes have been used to test whether silencing is functional in any given PPF. Transgenes, such as β‐glucuronidase (GUS) and green fluorescent protein (GFP) (reporter genes), or endogenous fungal genes with an easily screened phenotype have proven to be useful as markers for PTGS (Caribé dos Santos et al., 2009; Fitzgerald et al., 2004; Kadotani et al., 2003; Laurie et al., 2008; Miyamoto et al., 2008; Nguyen et al., 2008; Whisson et al., 2005). Further, reporters, such as a GFP, can be fused to a target gene as an endogenous positive control to provide visible evidence that silencing is functional in a particular GFP‐expressing fungal transformant. This approach is particularly useful for high‐throughput screens. Exploring this strategy, Nguyen et al. (2008) developed a pSilent‐Dual1 (pSD1) vector to study 37 genes involved in calcium signalling in M. oryzae. Although this is a powerful approach, it requires the ability to create a GFP‐tagged strain. Alternatively, endogenous genes leading to easily scorable traits, such as melanin biosynthesis genes, are ideal and have been used in Bipolaris oryzae, Cochliobolus sativus, M. oryzae and Venturia inaequalis (Fitzgerald et al., 2004; Leng et al., 2011; Moriwaki et al., 2007; Nakayashiki et al., 2005). Other suitable target genes include those coding for hydrophobins as a result of the easily wettable phenotype of silenced transformants (Nakayashiki et al., 2005). Screens for silenced hydrophobin transcripts have been used in Cladosporium fulvum, M. oryzae and Moniliophthora perniciosa (Caribé dos Santos et al., 2009; Hamada and Spanu, 1998; Lacroix and Spanu, 2009; Nakayashiki et al., 2005). Other endogenous genes for screens include enzymes whose activities can be measured directly, such as polygalacturonase in Ophiostoma novo‐ulmi (Carneiro et al., 2010).

RNA silencing has been explored in a number of PPF and Phytophthora spp. (Table 1) amenable to genetic transformation. In general, constitutive promoters are effective for the expression of silencing constructs. The Aspergillus nidulans trpC and gpd are the promoters of choice in PPF within the Ascomycota. In addition, the oliC (A. nidulans) and gpdA (Glomerella cingulata) promoters have been used effectively in Botrytis cinerea and V. inaequalis, respectively (Fitzgerald et al., 2004; Patel et al., 2008). For PPF within the Basidiomycota, the Ustilago maydis HSP70 promoter has been used successfully in Moniliophthora perniciosa (Caribé dos Santos et al., 2009). Native and species‐specific promoters have also been employed, such as AvrL567 in Melampsora lini and an Ustilago‐specific HSP70 in Ustilago hordei (Laurie et al., 2008; Lawrence et al., 2010). Alternatively, in vitro‐synthesized double‐stranded RNA (dsRNA) molecules have been used successfully for RNA silencing. In this approach, sense and antisense strands of the target gene are synthesized by in vitro transcription and then mixed to form dsRNA molecules. This eliminates many cloning steps to construct the silencing vector prior to fungal transformation (Grenville‐Briggs et al., 2008; Whisson et al., 2005). In the Basidiomycota, Moniliophthora perniciosa, hydrophobin and peroxiredoxin transcripts were successfully silenced using in vitro‐synthesized dsRNA (Caribé dos Santos et al., 2009). In Phytophthora infestans, synthesized dsRNA was used to silence exogenous and endogenous genes, such as GFP, inf1 and cdc14 (Whisson et al., 2005). As Phytophthora spp. are diploid and homologous recombination has not been demonstrated conclusively, RNA silencing is currently the only option for functional gene analysis in Oomycetes. As listed in Table 1, the HAM34 promoter has been widely used in Phytophthora spp., as well as native promoters, such as Pigpa1 and Pigpb1. Although silencing mediated by synthesized dsRNA is typically transient, this does not appear to be the case in P. infestans, where silencing may persist throughout the pathogen's entire life cycle. The use of synthesized dsRNA is likely to be valuable for high‐throughput functional characterization of genes and/or silencing of gene families, such as those for the cellulose synthase gene family (CesA) in P. infestans (Grenville‐Briggs et al., 2008). Nevertheless, the main drawback of the use of synthesized dsRNA is the constant need to generate new silenced lines for long‐term studies.

As listed in Table 1, the source of the dsRNA precursor molecule is also critical. This precursor molecule can be transcribed in trans independently in sense or antisense orientations (Kadotani et al., 2003; Patel et al., 2008). Alternatively, a short spacer may be used between cis‐located sense and antisense sequences corresponding to the target to allow the self‐formation of a hairpin RNA molecule (Kadotani et al., 2003; Miyamoto et al., 2008; Nakayashiki et al., 2005). The hairpin structure has been used extensively with PPF (Table 1). Kadotani et al. (2003) showed that a hairpin‐generating construct more effectively silenced GFP in M. oryzae than did sense and antisense constructs. The principle constraint to the use of a hairpin‐like structure is that it requires the cloning of inverted fragments, which can be time consuming. To simplify cloning, Nguyen et al. (2008) developed the pSD1 vector, which contains opposing A. nidulans trpC and gpd promoters separated by a multiple cloning site. pSD1 offers the great advantage of a single‐step, nonoriented cloning procedure.

Although gene silencing has a number of advantages compared with gene knockout, studies to date have produced variable results in terms of efficiency. For example, in Moniliophthora perniciosa, the silencing efficiency of hydrophobin and peroxiredoxin transcripts varied from 18% to 97% and 23% to 87%, respectively (Caribé dos Santos et al., 2009). Results using RNA hairpins are also highly variable. For example, a RNA hairpin precursor used to transform the Ascomycota Ophiostoma novo‐ulmi resulted in three knockdown transformants having 6%, 22% and 31% expression relative to the wild‐type (Carneiro et al., 2010). A number of factors may explain the inconsistent efficiency of silencing. For instance, the genomic context of the inserted RNA silencing vector and/or number of integration events may influence the transcriptional activity of the precursor RNA molecule. An important drawback regarding the application of gene silencing to simultaneously silence multiple genes is that it may lead to the silencing of nontarget transcripts. As an illustration, the analysis of the expression levels of C. fulvum transformants carrying various hydrophobin silencing constructs resulted in a reduction in the expression levels of several off‐target transcripts (Lacroix and Spanu, 2009).

RNA silencing has been applied successfully to discover and characterize a number of fungal pathogenicity genes. In Alternaria alternata, the silencing of two host‐selective ACT‐toxins resulted in the complete loss of pathogenicity (Ajiro et al., 2010; Miyamoto et al., 2008). In M. oryzae, high‐throughput silencing of 37 genes involved in calcium signalling revealed that many genes affected hyphal growth, sporulation and pathogenicity (Nguyen et al., 2008). The silencing of mycotoxin‐specific genes in Fusarium graminearum resulted in reduced virulence on wheat (McDonald et al., 2005). In Melampsora lini, transformants silenced for the AvrL567 avirulence effector were able to cause disease on flax plants carrying the corresponding L6 immune receptor (Lawrence et al., 2010).

APPLICATIONS IN PDR USING HIGS

Plants naturally utilize the RNA silencing machinery to defend against invading viruses (Csorba et al., 2009; Harvey et al., 2011; Hu et al., 2011). HIGS has also been shown to be effective against other plant pathogens (Huang et al., 2006). In contrast with plant viruses, most of the nematode body resides outside of the plant cell; nevertheless, RNA silencing has been shown to target nematode parasitism genes (Urwin et al., 2002). Arabidopsis thaliana expressing a dsRNA targeting the root‐knot nematode parasitism gene 16D10 led to effective disease resistance against four major Meloidogyne spp. (Huang et al., 2006). The expression of MjTis11 (Meloidogyne javanica putative transcription factor) in tobacco resulted in the silencing of the target gene in feeding root‐knot nematodes (Fairbairn et al., 2007). Alternatively, dsRNA delivered by the viral infection of host plants has the potential to induce gene silencing in nematodes (Dubreuil et al., 2009).

RNA silencing has been implicated in resistance to bacterial pathogens, including crown gall disease caused by Agrobacterium tumefaciens (Escobar and Dandekar, 2003; Escobar et al., 2001, 2002; Viss et al., 2003). To induce gall formation, iaaM (tryptophan monooxygenase) and ipt (AMP isopentenyl transferase) genes are horizontally transferred from the bacterium to the host genome. Escobar et al. (2001) showed that the silencing of iaaM and ipt oncogenes in Arabidopsis thaliana and Lycopersicon esculentum resulted in high levels of resistance to crown gall. The same approach was also effective in producing resistance to crown gall in walnut (Julglans regia L.) and apple (Escobar et al., 2002; Viss et al., 2003). Other work has suggested that a number of small RNAs participate in the bacterium–plant interaction. miRNAs (miR393, miR160, miR167 and mR825) were upregulated in plants after challenge with bacterial pathogens (Katiyar‐Agarwal and Jin, 2010). In insects, a reduction in damage caused by the coleopteran western corn rootworm was obtained by expression of dsRNA V‐ATPase A in transgenic corn plants (Baum et al., 2007). For other examples, see Mao et al., 2007; Turner et al., 2006.

Examples of the application of RNA silencing in fungus–plant interactions are now beginning to emerge. Recently, Tinoco et al. (2010) showed that tobacco transformed to express a GUS hairpin‐structured dsRNA specifically silenced GUS transcripts in a GUS‐expressing strain of Fusarium verticillioides during infection. Other recent publications have shown that the in planta expression of fungal dsRNA results in HIGS in the biotrophic pathogens Blumeria graminis and Puccinia striiformis f.sp. tritici. In the barley–B. graminis pathosystem, 76 fungal genes found to be expressed during the interaction were subjected to RNA silencing by HIGS; 21% (16 of 76) of plants transiently transformed with each of the fungal RNA silencing constructs exhibited a reduction in the number of B. graminis spores able to develop haustoria (Nowara et al., 2010). Two genes coding for 1,3‐β‐glucanosyl‐transferase (GTF1 and GTF2) were further investigated. Virus‐induced gene silencing (VIGS), using Barley stripe mosaic virus (BSMV) to produce antisense GTF1 and GTF2 transcripts, led to a significant reduction in haustorium formation and the rate of secondary hyphae elongation, respectively. The analysis of independent T1 transgenic plants carrying inverted repeats (hairpin) targeting GTF1 showed that the level of dsRNA‐derived hairpin transcript decreased rapidly in noninoculated plants. However, some transgenic plants inoculated with B. graminis f.sp. hordei exhibited reduced symptoms compared with control plants. The same study also used HIGS to silence Avra10 transcripts produced by B. graminis f.sp. hordei. In susceptible barley hosts, Avra10 acts as a virulence effector, whereas, in plants containing the R gene Mla10, it is additionally recognized as an avirulence factor. As predicted, the expression of the silencing construct targeting the Avra10 transcript in susceptible plants resulted in a reduction in pathogen development (haustorium formation) (Nowara et al., 2010). Furthermore, the abundance of the Avra10 transcript was reduced in plants co‐bombarded with the cognate resistance gene Mla10 and the Avra10 silencing construct following fungal infection. The authors concluded that these data are consistent with gene silencing occurring within young haustoria. Also using VIGS mediated by BSMV, Yin et al. (2010) demonstrated HIGS in wheat targeting Puccinia striiformis f.sp. tritici (PST) genes. These authors initially showed that a highly abundant haustorial PST transcript (PSTha12J12) was silenced by the expression of dsRNA in VIGS‐mediated BSMV–wheat transformants. Further examination of other PST genes revealed that only genes highly expressed in haustoria were effectively silenced. Three predicted genes coding for secreted proteins and a fourth gene predicted to code for a chitinase with expression patterns similar to PSTha12J12 were silenced to a high level. Similarly two genes, one with unknown function and a Uromyces fabae hexose transporter homologue having high expression levels in purified haustoria cells and relatively high levels in urediniospores, were silenced. Other genes that were expressed constitutively in a variety of fungal cell types, such as an elongation factor‐1 homologue, predicted β‐tubulin, glyceraldehyde 3‐phosphate dehydrogenase and actin coding genes, were not silenced (Yin et al., 2010). Although VIGS mediated by BSMV in wheat appears to be restricted to the silencing of only pathogen genes highly expressed in haustoria, these recent reports using different pathosystems strongly suggest that RNA molecules are able to move from plants into fungal cells and effectively silence their target genes. We expect more examples to be forthcoming in the near future.

FUTURE PERSPECTIVES OF HIGS IN FUNGUS–PLANT INTERACTIONS

The application of PDR to engineer transgenic plants has led to valuable outcomes. The development of transgenic virus‐resistant plants represents the foremost example, and the deployment of such plants has been recognized as saving an entire commercial agricultural industry (Gonsalves, 1998). More recently, the coupling of PDR to HIGS has emerged as a feasible strategy to develop plants resistant to other important pathogens, including nematodes and fungi.

Although PDR was discovered more than 25 years ago, it is only now that promising applications of fungus‐derived resistance mediated by HIGS are ready for take off. In this concept, the source of resistance is derived from the parasite's own genome. The public availability of more than 100 fungal genome sequences and many plant genome sequences, combined with transcriptome data, now offers the scientific community an arsenal of candidate genes for testing and for the design of silencing constructs to avoid off‐target transcripts. Possible targets may include virulence determinants, genes essential for fungal infection development and known fungicide targets. The availability of host genome sequence data helps to ensure that HIGS constructs do not target and negatively affect the host.

The ability to identify fungal targets, as well as the means to develop stable and environmentally safe transgenic crop plants, holds great promise for the fulfilment of future food needs. The concept and use of transgenic pathogen‐resistant crops are well established (Collinge et al., 2010), and the recent growing acceptance of genetically modified plants around the world, including the European Union and China, is likely to attract the attention of government (funds) and researchers (scientific knowledge) to further develop PDR. Although the application of the concept of fungus‐derived resistance mediated by HIGS appears to be very encouraging, a series of challenges needs to be carefully addressed before application of this strategy can be fully implemented. We conclude by highlighting a few fundamental questions.

How do RNA molecules traffic from the plant into fungal cells?

A knowledge of the mechanisms involved in the movement of protein effector molecules from pathogens into host cells is now beginning to emerge (Dean, 2011; Rooney et al., 2005; Schornack et al., 2009; Tyler, 2009; White et al., 2009). In Arabidopsis, the intercellular and systemic movement of small RNAs is mediated through the plasmodesmata and phloem, respectively (Melnyk et al., 2011). Although virus particles are also capable of moving through plasmodesmata and in the phloem, the mechanism by which naked RNA moves between two organisms is unknown. It remains to be determined how RNA is delivered from the plant nucleus, across the plasma membrane and plant cell wall into fungal cells. Is it a passive or active process? If active, what host and pathogen genes are involved and how is the process regulated?

Can HIGS be effective for other types of fungal pathogen?

HIGS has been shown to be effective in silencing the transcripts of obligate biotrophic invading fungi, at least for B. graminis and Puccinia striiformis f.sp. tritici. Perhaps the intimate interaction of this group of pathogens allows sufficient opportunity for the trafficking of RNA molecules. Considering hemibiotrophic and necrotrophic fungi, is there adequate time for RNA movement from host into fungal cells before host cells die?

What are the most promising fungal targets?

In the gene‐for‐gene concept, pathogen avirulence (Avr) genes are constantly evolving to escape recognition by the corresponding host resistance (R) gene product (Stukenbrock and McDonald, 2009). To date, the number of successful HIGS candidates leading to reduced fungal growth development remains limited and includes effectors, a cell wall elongation gene, a chitinase, a hexose transporter and a gene with unknown function. Much work remains to be done to identify suitable fungal candidate genes. Fortunately, as we have described, opportunities exist to establish high‐throughput screening pipelines to identify strong candidates.

What are the best strategies for applying HIGS to control fungal disease?

A number of possible strategies are feasible to exploit HIGS for durable resistance. The use of (i) a silencing construct that targets entire gene families, (ii) multiple lines deployed in rotation to minimize the selection pressure on pathogens, (iii) a ‘polygenic’ HIGS line conferring resistance to single or multiple fungal pathogens or (iv) a combination of both classical R genes and HIGS cassettes in the same host, which may synergistically boost resistance, are all strategies worthy of investigation.