In vivo trans-specific gene silencing in fungal cells by in planta expression of a double-stranded RNA

Background Self-complementary RNA transcripts form a double-stranded RNA (dsRNA) that triggers a sequence-specific mRNA degradation, in a process known as RNA interference (RNAi), leading to gene silencing. In vascular plants, RNAi molecules trafficking occur between cells and systemically throughout the plant. RNAi signals can spread systemically throughout a plant, even across graft junctions from transgenic to non-transgenic stocks. There is also a great interest in applying RNAi to pathogenic fungi. Specific inhibition of gene expression by RNAi has been shown to be suitable for a multitude of phytopathogenic filamentous fungi. However, double-stranded (ds)RNA/small interfering (si)RNA silencing effect has not been observed in vivo. Results This study demonstrates for the first time the in vivo interference phenomenon in the pathogenic fungus Fusarium verticillioides, in which expression of an individual fungal transgene was specifically abolished by inoculating mycelial cells in transgenic tobacco plants engineered to express siRNAs from a dsRNA corresponding to the particular transgene. Conclusion The results provide a powerful tool for further studies on molecular plant-microbe and symbiotic interactions. From a biotechnological perspective, silencing of fungal genes by generating siRNAs in the host provides a novel strategy for the development of broad fungi-resistance strategies in plants and other organisms.


Background
The genetic interference phenomenon [RNA interference (RNAi)] was described in Caenorhabditis elegans in which double-stranded RNA (dsRNA) induces individual sequence-specific posttranscriptional gene silencing. The spreading silencing effect has been demonstrated, in which the interference is observed in a broad region of the animal after the injection of dsRNA into the extracellular body cavity. Fire et al. [1] demonstrated that RNAi abolished expression of targeted genes in C. elegans by injecting dsRNA. Subsequently, others have shown that the effect also occurs when C. elegans is fed the bacterium Escherichia coli, which transcribes the recombinant dsRNA [2]. It also results from simply soaking the animals in dsRNA preparations [3]. Data obtained from in vivo and in vitro studies are now being used to engineer resistance against parasitic nematodes in transgenic plants [4][5][6].
Cellular boundaries play a pivotal role in this integration by maintaining a level of cell autonomy while enabling communication between cells for coordinated gene expression and metabolism. In vascular plants, trafficking of RNAi molecules occurs between cells and systemically throughout the plant [7,8]. Such RNA trafficking breaks the boundaries of our traditional thinking of RNAs as functioning solely within the cells in which they are produced, and ushers in a new frontier of plant biology [9]. Intercellular and systemic movement occurs via plasmodesmata, which provide the continuity of cytoplasm and endoplasmic reticulum between adjacent cells and the phloem. Analyses of vascular exudates from oilseed rape (Brassica napus) showed that phloem sap contained a large number of small (sm)RNAs, predominantly of 21 and 24 nucleotides in length [10]. In addition, RNAi signals can spread systemically throughout a plant, even across graft junctions from transgenic stocks to non-transgenic scions [11,12]. Moreover, recent experiments described by Tomilov et al. [13] demonstrated that the movement of RNAi molecules between a parasite and its host plants.
The gus silencing signal generated by lettuce roots was functional in its parasite Triphysaria versicolor, translocating across the haustorium interface. In nematodes gene silencing may also be triggered by a diet composed of transgene-encoded RNAi plants [4,14]. The same phenomenon has been observed in herbivorous insects fed on a plant engineered to express dsRNAs targeting vital insect genes [15,16].
Plants have been genetically manipulated by introducing constructs encoding self-complementary hairpin RNA (hpRNA) to efficiently silence genes [17][18][19]. The expressed transcripts form a dsRNA that triggers a sequence-specific messenger (m)RNA degradation (RNAi). Briefly, the dsRNA is recognized by DICER-like enzymes, which cleaves the molecule into a series of 21-23 bp duplexes [called small interfering (si)RNAs], that complexes with a ribonucleoprotein complex (RISC). The duplex is unwound to give single stranded siRNA leading to activation of the RISC, which searches for homologous mRNA transcripts by a base-pairing mechanism, leading to mRNA degradation. In addition, siRNA can affect the chromatin structure of targeted genes, resulting in transcriptional inhibition [20].
Here we show that the gus gene expression can be specifically silenced in Fusarium verticillioides (= F. moniliforme) interacting with a transgenic plant engineered with a gus gene-inferring cassette [hairpin (hp)GUS].

Results and Discussion
In this study, we hypothesized that a plant expressing dsRNA could result in in vivo trans-specific suppression of the corresponding target gene in fungal cells attached to plant tissues. Tobacco plants expressing the gus gene were re-transformed with an interfering intron-hairpin construct ( Figure 1A). None of the re-transformed lines (GUS-RNAi lines) presented an observed gus expression in the leaves ( Figure 1B) and no significant phenotypical differences were observed in the transgenic lines when compared with wild-type tobacco. Polymerase chain reaction (PCR) analysis confirmed the presence of both transgenes in the GUS-RNAi lines ( Figure 1C). Northern analysis was carried out to detect the siRNA in the transgenic tobacco lines showed siRNA bands of expected size range only in the GUS-RNAi lines ( Figure  1D). No signal was observed in either gus-expressing or non-transgenic plants. The constitutive expression of the gus dsRNA in these transgenic tobacco lines provides gus siRNA molecules for absorption by F. verticillioides feeding cells and subsequent RNAi of the constitutive gus transgene in fungal mycelium and conidial cells. In addition, uptake of unprocessed dsRNA and others RNAi molecules could be also occurs.
F. verticillioides fungal strain S68 (a niamutant) was transformed to express the GUS and nitrate reductase coding genes. Only colonies harbouring the nitrate reductase gene (nia + ) were able to grow on MM plates and all co-transformants showed mitotic stability in respect of the GUS + phenotype. One line over-expressing the gus gene was selected to inoculate both GUS-RNAi and non-transformed tobacco lines ( Figure 2A). Leaves detached from in vitro plants (n = 12 per line; repeated twice) were inoculated with F. verticillioides culture, applied to the adaxial surface. After 11 days the agar plug was removed and leaves containing penetrative food-absorbing structures that remained attached to leaf surface were analysed for gus gene expression. When the F. verticillioides expressing the gus transgene was inoculated on leaves of GUS-RNAi lines, expressing dsRNA corresponding to the gus gene, a small number of fungal structures exhibiting the gus gene expression (7 ± 3 blue spots/cm 2 ) were observed by GUS protein assay ( Figure 2B). Non-transgenic tobacco leaves were unable to cause an evident gus gene silencing in transformed fungus structures and a large number of fungal structures exhibiting GUS expression was observed (221 ± 25 blue spots/cm 2 ) ( Figure 2C). After GUS assay, leaf surfaces were observed under scanning electron microscope. The analysis revealed that only germinating spores remained on the leaf surfaces ( Figures 2D-E). Almost all spores observed were penetrating into leaf stomata. These phenotypes convincingly indicate that the silencing signal translocated across the germinated spores from tobacco into fungal cells. Recently, the translocation of RNAi molecules across the haustorium interface was demonstrated between a parasite and its host plants [13]. As a consequence, transgenic lettuce plants expressing a gus dsRNA induced specific gene silencing in their parasitic plant Triphysaria versicolor.
Infection by F. verticillioides can result in highly variable disease symptoms ranging from asymptomatic plants to severe rotting and wilting. In the more virulent pathogen F. oxysporum the whole process proceeds much faster and the transition from the initial symptomless phase to the necrotrophic phase occurs within a few days [39]. In contrast, the less aggressive pathogen F. verticillioides develops more slowly, leaving the plant more time to respond and restrict fungal growth [40]. In this study, the infection started from fungal conidia that initiated penetration to palisade and spongy mesophyll through stomata~11 days after inoculation. We speculate that this fungus characteristic could be an important condition for plant-pathogenic fungi interaction-mediated gene silencing. These interaction types establish a long-term feeding relationship with a living host plant cell, allowing the nutrient-absorbing cells to uptake dsRNA molecules from their host. Indeed, this mode of plant colonization was confirmed when early stages of the F. verticillioides-maize interaction were characterized by using green fluorescent protein-expressing transgenic fungus isolates. Conidia were found only inside a cell in which an infection was established, while the surrounding cells appeared to be normal [40]. The differences in susceptibility could reflect the resistance of some cells or stages to the consequences of absorbed dsRNA.
In this work, transgene silencing in F. verticillioides was achieved by the interaction of fungal cells with a transgenic tobacco engineered for both transcribing and silencing the gus transgene expression, allowing the phenomenon of transitive silencing cycles of 'degradative-PCR' to take place in the plant cells. The 'degradative PCR' model proposes that siRNAs act as primers to transform the target mRNA into dsRNA. The nascent dsRNA is degraded by Dicer generating new siRNAs in a circle of dsRNA synthesis and degradation [41,42]. Further investigation should be carried out in order to determine whether this is important. It is known that gene silencing is dose-dependent and that more intact long hpRNAs could cause more pronounced silencing [43].
Ten fungal colonies were re-isolated from both GUS-RNAi and non-transformed tobacco lines and established in vitro. The gus gene expression was quantified in all re-isolated colonies and results revealed that two colonies isolated from GUS-RNAi plant lines presented a reduction of approximately 62% (RT1) and 96% (RT2) in the gus gene expression, compared with colonies isolated from non-transformed leaves ( Figure 3A). All colonies isolated from non-transformed plants revealed normal GUS expression (activity raging from 1312 to 1378 nmol 4-MU.min -1 .mg.protein -1 ). We further investigated the stability of gene silencing in mycelia cells over several passages in the absence of the RNAi trigger molecules from GUS-RNAi plants. The colonies showing a reduction of GUS expression were cultured in order to reach 70%-80% of confluence for eight passages ( Figure 3B). GUS fluorimetric analyses revealed that the colony that presented 62% GUS expression reduction (RT1) remained expressing the gus gene in a similar pattern. However, the colony that presented 96% GUS expression (RT2) resumed normal GUS activity after the seventh passage. Steady-state levels of GUS mRNA in mycelia from the seventh-passage were estimated by semi-quantitative reverse transcription (RT)-PCR. gus-specific primers were used to distinguish cDNA from genomic GUS sequences. Results have shown a reduction in the steady-state level of gus transcript in colony mycelia of RT1 and RT2, which exhibit partial degrees of transgene silencing. In addition, after GUS-silencing phenotype reversion, the colony RT2 exhibits a gus transcript level similar to the control mycelia (which were not attached to hpGUStobacco leaves; Figure 4). Moreover, northern analyses showed siRNA bands of expected size range in both RT1 and RT2 silenced colonies and no signal was observed for the fungus colony isolated from non-transformed leaves (N) and RT2 colony that resumed normal GUS activity (R; Figure 4D).
It has been shown that silencing phenotypes induced by RNAi can persist for generations. Despite numerous reports on RNA silencing in a variety of organisms, it is still not understood whether systemic smRNA signalling occurs in filamentous fungi. Nevertheless, silencing was shown to be a reversible dominant trait, operative in heterokaryotic strains containing a mixture of transgenic and non-transgenic nuclei, which suggests that a diffusable, trans-acting mobile RNA signal is involved [38,44]. In the filamentous fungus M. perniciosa, the gfp-silenced phenotype persisted for a period of 120 days after the transfection with dsRNAs [38]. In contrast, P. infestans treated with gfpdsRNA exhibited gene expression partially recovered after 4 days [45]. In Caenorhabditis elegans a single episode of RNAi in the nematode induced transgene silencing effects that were inherited over 80 generations in the absence of the original trigger [46]. In mice, microinjection of microRNAs into fertilized eggs also induced a heritable silencing phenotype, associated with the zygotic transfer of RNA molecules [47]. In vascular plants, recent studies suggest that smRNAs can move between cells (through plasmodesmata and endoplasmic reticulum) and can spread systemically via the long-distance transport systems (phloem) [48][49][50][51][52]. It was demonstrated that plant siR-NAs do not only act at the site of synthesis, but are additionally mobile between cells [53]. There are indications that siRNAs can move 10-15 cells without amplification -probably as molecules of the 21-nt class -whereas movement over greater distances requires an amplification of the original signal [53]. In addition, certain cell types, such as neurons and sperm, are known to be more resistant to RNAi. Athough, so far, no cell type n fungi has been reported to be resistant to RNA silencing, this may possibly be the case with certain fungal cell types. Moreover, we can not exclude the possibility of mutation or an off-target effect on a gene related to silencing mechanism, or paramutation phenomenon associated with the reversion of gus gene normal expression.
In fungi, transgene gene silencing or instability of the silencing effect has been observed to associated with modifications in the organization of the transgene loci [31,44,[54][55][56]. In order to investigate whether the gus gene silencing in fungal cells would be associated to changes in the status of the transgene in the genome, Southern analyses were carried out with fungal colonies that were re-isolated from both GUS-RNAi and nontransformed tobacco lines. Results have shown that all isolates presented a simple and unaffected integration pattern ( Figure 5), demonstrating that the transgene locus was stably maintained in the fungal colonies reisolated from GUS-RNAi plants and after several cultivation passages.

Conclusions
This work provides an example of RNA-mediated transfer of information between organisms and between plant and pathogen. Further studies should be carried out in order to access whether such RNA-mediated interference-transfer mechanisms participate in natural interactions. The use of siRNAs (RNAi) has become a powerful tool for specifically down-regulating gene expression and it has been demonstrated to be of great importance in basic and applied molecular plant-pathogen, mixotrophic and symbiotic interaction studies. From a biotechnological perspective, our results of in planta RNAi silencing of a gene expressed in a parasite fungus open new perspectives for the development of broad fungi-resistance strategies in plants and other organisms.

Plasmid construct and transgenic plants
Tobacco (cv Xanthi) was transformed as previously described by Horch et al. [57] with the vector pBI121 (Clontech) to express the gus gene. Transgenic lines presenting high level gus gene expression were re-transformed with the vector pC1302GUSi ( Figure 1A). pC1302GUSi is an intron-spliced hpRNA vector constructed to promote gus gene silencing in re-transformed lines. In order to construct the pC1302GUSi, intron 3 from the malate synthase gene from Arabidopsis thaliana (ms-i3; GenBank accession number AB005235) was PCR-amplified from genomic DNA by using Platinum Taq DNA Polymerase High Fidelity (Invitrogen, CA, USA) and specific primers (5'-CTCTAGAGGCGCGCCGG-TACCCCGGATCCATCAGCCTGTTCAAA-3' and 5'-TGAGCTCTCGCGATGGGGCCCACTAGTTTTATG-GTCCATTTTC-3'), generating sites for XbaI, AscI, KpnI, BamHI and SpeI, ApaI, NruI, SacI, respectively. The 438pb fragment was cloned into pCR2.1-TOPO (Invitrogen) and sequenced. The ms-i3 fragment was then removed with SacI and XbaI and inserted into the vector pUC19-35SAMVNOS [58] in order to generate the vector pSIU. A 627-bp fragment from gus gene coding sequence was removed with SspI and EcoRV from the vector pCam-bia3301 (Cambia, Brisbane, Australia) and cloned into the SmaI site from the pBlueScript KS+ (Fermentas, Ontario, Canada). The gus fragment was finally directionally cloned into the sites NruI and KpnI/XbaI from pSIU in order to generate sense and antisense arms, flanking the malate synthase gene intron 3 from Arabidopsis thaliana (Gen-Bank accession No. AB005235). The gus inferring cassette (hpGUS) was removed with PvuII and cloned into SmaI from pCambia1302 (Cambia) generating the vector pC1302GUSi.

Transgenic Fusarium
F. verticillioides strain S68 (a niamutant) was previously co-transformed with the vectors pNOM102 containing the gus gene under control of the gpd promoter from A. nidulans, and pNH24 carrying the F. oxysporum nitrate reductase gene (nia). The line TG8 exhibiting the nia + phenotype conferred by pNH24 was selected on MM with nitrate as the sole nitrogen source and purified by isolation of uninucleated conidia according to standard techniques. Only colonies harbouring the pNH24 vector (nia + ) are able to grow on MM plates. The line TG8 was previously characterized and presented a single copy of the gus gene integrated into the genome and mitotic stability in respect of the GUS + phenotype [59].

Inoculation of transgenic plants with gus + Fusarium
One transgenic line of F. verticillioides over-expressing the gus gene was selected to inoculate both re-transformed and non-transformed tobacco lines. The fungus was grown on MM medium [59] at 26°C. Inoculation was carried out according to Dias et al. [58]. A mycelial agar plug 3 mm in diameter was cut from the growing margins of a 6-day-old F. verticillioides culture and applied to the adaxial surface of a leaf detached from 6week-old in vitro plants (n = 12 per line; repeated twice). After 11 days at 26°C and under 90%-100% relative humidity, the agar plug was removed and leaves containing penetrative food-absorbing structures that remained attached to leaf surface were analysed for gus gene expression.
Fungal colonies were re-isolated from both GUS-RNAi and non-transformed tobacco lines. Small portions from mycelium were removed under stereomicroscope and cultured on MM medium (as described above) in order to reach 70%-80% of confluence for eight passages. GUS histochemical and fluorimetric assays were carried out in each passage. The qualitative β-glucuronidase activity assay was carried out according to Couteaudier et al.

GUS assays and scanning electron microscopy
The GUS histochemical assay was performed as described by Jefferson [61]. Stained materials were examined using a Zeiss Axiophot stereomicroscope and photographed using a Zeiss AxioCam ICC3 high-resolution digital camera system. Explants were also examined using a scanning microscope. For scanning microscopy, leaves were fixed with 3% glutaldeihyde in 0.1 M sodium cacodylate buffer at pH 7.2 for 15 h at 4°C and postfixed in 1% osmium tetroxide for 1 h. After washing in the same buffer, the specimens were dehydrated in a graded acetone series, critical-point dried with CO2, sputtercoated with a thin layer of gold and observed under a Zeiss DSM 962 SEM operating at 15 kV.

RT-PCR expression analysis
Mycelium cells were used for total RNA extraction as described, with Trizol (Invitrogen) as recommended by the manufacturer. Total RNA was used to produce cDNA using the reverse transcriptase Superscript III (Invitrogen), according to the protocol suggested by the manufacturer. PCRs were carried out as described [62], except that 20 ng of cDNA was used as a template, in reactions with 32 cycles of amplification. The number of amplification cycles was previously optimized in order to stop the reaction at the exponential stage, ensuring that amplification was semi-quantitative. Primers 5'-ATCACGCAGTTCAACGCTGAC-3' (GUS671) and 5'-TTGGGCAGGCCAGCGTATCGT-3' (GUS251) were used to amplify a 420 bp fragment from the gus gene. As an internal control, primers 5'-GGAAGTAAAAGTCG-TAACAAGG-3' (ITS5) and 5'-TCCTCCGCTTATTGA-TATGC-3' (ITS4) were utilized to amplify a fragment of the 5.8S rRNA housekeeping gene. PCRs with total RNA presented no amplified fragments. Experiments were repeated three times.

RNAi analysis
The mycelia of the fungi were cultivated on static liquid MM medium for 14 days at 26°C. Total RNA was isolated by extraction with Micro-to-Midi Total RNA Purification System (Invitrogen) as recommended by the manufacturer. Total RNA from plant leaves was isolated according to Bonfim et al. [19]. siRNA analysis was carried out according to Bonfim et al. [19]. RNAs were hybridized with a DNA probe corresponding to the 627 pb-fragment from gus gene, excised with EcoRV and SspI from pCambia2301. Probes were labelled with α 32 P dCTP using a random primer DNA labelling kit (Amersham Pharmacia Biotech, NJ, USA) according to the manufacturer's instructions. Three oligomers (18, 24 and 38 nucleotides) were used as molecular size markers. A GUS-sequence-related 21-nucleotide oligomer was used as a positive control. The bands were visualized with a fluorescent image analyser (FLA-3000) (FUJIFILM, Tokyo, Japan).

Southern blot analysis
Genomic DNA was isolated as described by Chow et al. [63]. Southern blotting was carried out as described [64]. Genomic DNA (20 μg) was digested with SphI, separated on a 1% agarose gel, and transferred to a nylon membrane (Hybond-N+; Amersham Pharmacia Biotech). Hybridizations were carried out using the PCR amplified 420-bp fragment from the gus gene, labelled with α 32 P deoxycytidine triphosphate (3000 Ci mol -1 ) using a random primer DNA-labelling kit (Amersham Pharmacia Biotech) according to the manufacturer's instructions. The bands were visualized with a fluorescent image analyser (FLA-3000; FUJIFILM).