In vitro delivery method

In C. elegans dsRNA can be delivered directly into the worm through microinjection, by feeding with Escherichia coli expressing the target gene dsRNA or by soaking worms in a solution of dsRNA (Kamath and Ahringer, Reference Kamath and Ahringer2003). Microinjection is technically difficult and has not been routinely achieved for plant parasitic nematodes. This, coupled with their obligatory parasitic nature, limits the options for in vitro delivery. They do not normally feed outside of the host plant and for sedentary plant parasitic nematodes, J2s and eggs are the only stages of the life-cycle that can survive outside the host root. The majority of in vitro studies have therefore been undertaken with these stages, using the soaking method of delivery. Soaking could potentially lead to uptake of dsRNA via a number of routes, including transcuticular, ingestion through the secretory/excretory pore or amphids that are exposed to the external environment. For J2s, oral ingestion via the stylet is considered the most likely route. Although J2 cyst and root-knot nematodes are non-feeding, the addition of a pharyngeal neurostimulant, such as octopamine, resorcinol, serotonin or carbamoylcholine chloride to the soaking solution has been used to induce dsRNA ingestion by J2s in a number of species including Globodera pallida (Urwin et al. Reference Urwin, Lilley and Atkinson2002), G. rostochiensis (Chen et al. Reference Chen, Rehman, Smant and Jones2005), Heterodera schachtii (Vanholme et al. Reference Vanholme, Van Thuyne, Vanhouteghem, De Meutter, Cannoot and Gheysen2007), H. glycines (Urwin et al. Reference Urwin, Lilley and Atkinson2002; Lilley et al. Reference Lilley, Goodchild, Atkinson and Urwin2005b; Alkharouf et al. Reference Alkharouf, Klink and Matthews2007; Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2007, Reference Bakhetia, Urwin and Atkinson2008), Meloidogyne incognita (Bakhetia et al. Reference Bakhetia, Charlton, Atkinson and McPherson2005a; Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005; Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006; Dubreuil et al. Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007; Shingles et al. Reference Shingles, Lilley, Atkinson and Urwin2007) and M. javanica (Adam et al. Reference Adam, Phillips, Jones and Blok2008; Gleason et al. Reference Gleason, Liu and Williamson2008). In some studies, uptake of the dsRNA along the pharynx and intestinal tract has been monitored through either the co-ingestion of the fluorescent dye fluorescein isothiocyanate (FITC) (e.g. Urwin et al. Reference Urwin, Lilley and Atkinson2002; Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006; Sukno et al. Reference Sukno, McCuiston, Wong, Wang, Thon, Hussey, Baum and Davis2007) or by the use of fluorescently-labelled dsRNAs (e.g. Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005; Adam et al. Reference Adam, Phillips, Jones and Blok2008). Although stimulants are frequently included to induce ingestion, adverse effects have occasionally been reported from their use. Adam et al. (Reference Adam, Phillips, Jones and Blok2008) found that overnight incubation of M. javanica J2s in either 0·5% or 1% resorcinol was lethal, whereas 0·5 mg/ml carbamoylcholine chloride stimulated oral uptake over the same time period with no detrimental effects.

Addition of pharyngeal stimulants is not always necessary to achieve an RNAi effect in cyst and root-knot nematode J2s (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007; Dalzell et al. Reference Dalzell, McMaster, Johnstone, Kerr, Fleming and Maule2009, Reference Dalzell, McMaster, Fleming and Maule2010a). FMRFamide-like peptides (FLPs) are a family of neuropeptides that modulate sensory and motor function in C. elegans and plant parasitic nematodes (Kimber and Fleming, Reference Kimber and Fleming2005). C. elegans flp genes are refractory to RNAi, however silencing of five flp genes in G. pallida caused severe effects on nematode migration (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). An effect on motility was observed in worms soaked for 24 h in 0·1 mg/ml dsRNA diluted in DEPC-treated water only. Knockdown of the target flp genes was confirmed by RT-PCR. The effect is not limited to dsRNA or to Globodera spp. as soaking in siRNAs targeting either G. pallida flp-12 or M. incognita flp-18 also resulted in RNAi-mediated silencing (Dalzell et al. Reference Dalzell, McMaster, Fleming and Maule2010a). It was proposed that dsRNA/siRNA was gaining entry through the amphids, secretory/excretory pore or the cuticle. Urwin et al. (Reference Urwin, Lilley and Atkinson2002) reported the occasional presence of the dye FITC in the amphids of cyst nematode J2s following soaking in the absence of octopamine. FITC has also been observed in the amphids of M. incognita incubated in collagenase or M9 buffer (Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005). The phenomenon of dye uptake by amphidial and phasmid neurons is well documented in C. elegans (Hedgecock et al. Reference Hedgecock, Culotti, Thomson and Perkins1985) and other free-living nematodes (Srinivasan et al. Reference Srinivasan, Durak and Sternberg2008). A recent study has shown the potential for larger molecules to be taken up from the amphidial pouches of cyst nematodes and transported along chemosensory neurons. A fluorescently labelled peptide that binds nicotinic acetylcholine receptors was taken up from an aqueous environment by the primary cilia of chemoreceptive sensilla of G. pallida before undergoing retrograde transport to their neuronal cell bodies and a limited number of connecting neurons (Wang et al. Reference Wang, Jones, Urwin and Atkinson2011). dsRNA/siRNA may potentially be transported in a similar manner. The success of RNAi targeted to flp genes in the absence of pharyngeal stimulants in G. pallida was suggested to reflect an increased sensitivity of neuronal tissue to RNAi in plant parasitic nematodes. A G. rostochiensis gene encoding a secreted amphid protein of unknown function (gr-ams-1) was targeted by dsRNA (Chen et al. Reference Chen, Rehman, Smant and Jones2005). Although octopamine was included in the soaking solution, gr-ams-1 was more susceptible to RNAi than a gland cell expressed endoglucanase. Possibly the two genes were being targeted via different routes, with more efficient dsRNA uptake from the amphids. Previous attempts to target other tissues of J2s with dsRNA in the absence of a pharyngeal stimulant failed to cause an RNAi effect. Soaking H. glycines and G. pallida in just dsRNA targeted to an intestinal cysteine proteinase did not reduce transcript abundance as determined by virtual Northern analysis (Urwin et al. Reference Urwin, Lilley and Atkinson2002). Similarly, resorcinol was required for suppression of 16D10 transcript in the subventral pharyngeal glands of M. incognita (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006).

RNAi effects have been achieved by soaking eggs of Meloidogyne spp. in either dsRNA or siRNA (Fanelli et al. Reference Fanelli, Di Vito, Jones and De Giorgi2005; Dalzell et al. Reference Dalzell, Warnock, Stevenson, Mousley, Fleming and Maule2010b). Eggs of M. artiellia contained within the gelatinous matrix were incubated in dsRNA targeted against a chitin synthase gene involved in the synthesis of the chitin layer in eggshells (Fanelli et al. Reference Fanelli, Di Vito, Jones and De Giorgi2005). The treatment resulted in a reduction in chitin synthase transcripts and a delay in the hatching of juveniles, along with reduced staining of chitin in the eggshells. This suggested that both the gelatinous matrix and the egg shells with their chitin layer are permeable to dsRNA. Two key factors in the processing of micro (mi)RNAs, the RNase III enzyme drosha and its cofactor pasha, were silenced in embryos of M. incognita by soaking undifferentiated eggs in siRNA solutions (Dalzell et al. Reference Dalzell, Warnock, Stevenson, Mousley, Fleming and Maule2010b) again showing that RNA can cross the egg shell. It remains to be seen if eggs of other species are similarly susceptible or if those of cyst nematodes would require pre-stimulation, as the permeability of the egg shell is known to change in preparation for hatching (Jones et al. Reference Jones, Tylka, Perry, Perry and Wright1998).

Migratory parasites that do not have an obligate sedentary phase can be targeted for RNAi at later stages of the life-cycle. Mixed stages of R. similis were incubated with dsRNA targeting a xylanase gene expressed only in adults. Octopamine was used to stimulate ingestion and reduction in transcript was correlated with subsequent reduced infection rates (Haegeman et al. Reference Haegeman, Vanholme and Gheysen2009). A 24 h incubation was necessary for ingestion of FITC by almost 100% of soaked B. xylophilus, but octopamine was not required to achieve this (Cheng et al. Reference Cheng, Dai, Xiao and Xie2010). This is the only plant parasitic nematode species for which non-soaking methods of dsRNA delivery have been evaluated in vitro. Soaking and electroporation of L2/L3 larvae resulted in similar RNAi effects for four essential genes but intestinal microinjection of adult females was found to be the most efficient delivery route, based on a reduced hatching rate in the F1 generation (Park et al. Reference Park, Lee, Lee, Oh, Jeong, Woo, Kim, Paik and Koo2008).

A number of general factors are commonly recognised as improving the efficiency of RNAi by soaking. Many studies include 3 mM spermidine and 0·05% gelatin in the soaking buffer, as recommended for work with C. elegans (Maeda et al. Reference Maeda, Kohara, Yamamoto and Sugimoto2001). Improved efficacy arising from the addition of spermidine has been confirmed for the cyst nematodes H. glycines (Sukno et al. Reference Sukno, McCuiston, Wong, Wang, Thon, Hussey, Baum and Davis2007) and G. rostochiensis (Chen et al. Reference Chen, Rehman, Smant and Jones2005). Nematode samples are routinely agitated during soaking with some studies specifically demonstrating that this is essential for efficient uptake and viability of the nematodes (Adam et al. Reference Adam, Phillips, Jones and Blok2008; Cheng et al. Reference Cheng, Dai, Xiao and Xie2010).

Characteristics of the inducing molecule

The majority of in vitro RNAi experiments carried out with plant parasitic nematodes have used dsRNA as the inducing molecule with sizes ranging from 42 bp (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006) to >1000 bp (Urwin et al. Reference Urwin, Lilley and Atkinson2002; Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2007). The discrepancies between studies in terms of the type of gene targeted, incubation time, dsRNA concentration and method for assessment of silencing make it difficult to draw meaningful conclusions as to either the optimum size of dsRNA or its preferred sequence characteristics. The few studies that compare different dsRNAs targeting the same gene may shed some light on the matter. Gp-flp-6 of G. pallida was silenced by both a 316 bp dsRNA and its 227 bp 3′ region, with the shorter molecule consistently proving more potent. The 88 bp 5′ region was insufficient to induce any silencing effect (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). Possibly the position within the targeted gene was more important than the length of the dsRNA as both 271 bp and 42 bp dsRNAs covering the full-length transcript or the coding region of the oesophageal gland peptide 16D10, led to >90% transcript reduction in M. incognita J2 (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006). A 267 bp dsRNA targeting the 3′ region of the H. glycines pel-1 pectate lyase transcript induced more potent silencing than a 285 bp dsRNA targeted to the 5′ region of the same gene (Sukno et al. Reference Sukno, McCuiston, Wong, Wang, Thon, Hussey, Baum and Davis2007) and a similar effect was reported for two dsRNAs homologous to 5′ and 3′ regions of a xylanase gene from R. similis (Haegeman et al. Reference Haegeman, Vanholme and Gheysen2009). In this case, no silencing or significant phenotypic effect was observed with the dsRNA targeting the 5′ catalytic domain.

More recently, synthetic 21 bp siRNAs were reported to induce efficient silencing of G. pallida and M. incognita flp genes in J2 worms (Dalzell et al. Reference Dalzell, McMaster, Fleming and Maule2010a). As with dsRNAs, efficacy of individual siRNAs targeting different regions of the same transcript was highly variable. This was the first example of siRNAs being used as silencing triggers in plant parasitic nematodes and was followed by further evidence of their efficacy when drosha and pasha transcripts were subjected to successful knockdown in eggs of M. incognita (Dalzell et al. Reference Dalzell, Warnock, Stevenson, Mousley, Fleming and Maule2010b). The use of siRNAs for functional analysis has the potential advantage of increasing target specificity and thus minimising off-target effects or allowing selective silencing of gene-family members. However, there is currently insufficient knowledge of the RNAi mechanisms of plant parasitic nematodes to determine if the production of secondary siRNAs by RdRps might compromise this aspect.

Off-target and non-specific effects

Use of siRNAs as RNAi triggers was explored partly as a way of circumventing problems of non-specific phenotypic effects observed with high soaking concentrations of long dsRNAs (Dalzell et al. Reference Dalzell, McMaster, Johnstone, Kerr, Fleming and Maule2009). After the initial demonstrations of RNAi in plant parasitic nematodes, most studies have included a control incubation with non-nematode dsRNA to assess any effects that may be unrelated to specific gene silencing. A green fluorescent protein (gfp) sequence is commonly used, but other controls have been soybean Rubisco subunit (Alkharouf et al. Reference Alkharouf, Klink and Matthews2007), plasmid vector sequence (Kang et al. Reference Kang, Lee, Koh and Lee2011) and tomato chloroplast ribosomal protein (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). Despite the use of non-target dsRNA concentrations of up to 4 mg/ml (e.g. Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005) there were, until recently, no reports of any adverse phenotypic or gene-silencing effects related to these non-nematode dsRNAs. Concentrations of control dsRNA >2 mg/ml were found to be detrimental to B. xylophilus after a 24 hr incubation (Cheng et al. Reference Cheng, Dai, Xiao and Xie2010). Dalzell et al. (Reference Dalzell, McMaster, Johnstone, Kerr, Fleming and Maule2009) reported an unexpected aberrant phenotype and reduced motility for M. incognita J2s treated with nine different non-nematode derived dsRNAs (including gfp) at a concentration of 0·1 mg/ml for 24 hrs. An increased concentration of 1 mg/ml elicited a similar response after 4 hrs. The same phenotypic effect was observed in G. pallida, although the concentration of non-nematode dsRNA required to induce a phenotype was 10-fold higher (1 mg/ml dsRNA for 24 h). No pharyngeal stimulants were used in the experiments. Despite the phenotypic effects, there was no general impact on transcript abundance in either nematode, in agreement with previous studies. There was a significant recovery of normal phenotype 24 or 48 hrs after removal from the dsRNA solution. This may partly account for the lack of reported effects in earlier studies, where RNAi phenotypes were frequently assessed only after parasitism of plants, at least 10 days after treatment (e.g. Bakhetia et al. Reference Bakhetia, Charlton, Atkinson and McPherson2005a; Lilley et al. Reference Lilley, Goodchild, Atkinson and Urwin2005b; Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2007; Dubreuil et al. Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007). The study of Dalzell et al. (Reference Dalzell, McMaster, Johnstone, Kerr, Fleming and Maule2009) underlined the limited extent of our knowledge regarding the mechanisms of RNAi in plant parasitic nematodes. One hypothesis for the observed effect is that rate-limiting components of the RNAi pathway become saturated, thus inhibiting processing of endogenous small RNAs and affecting normal gene regulation.

Unexpected, but potentially related effects were reported when H. glycines was exposed simultaneously to two distinct dsRNAs aimed at silencing two different genes expressed in the dorsal pharyngeal gland cell (Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2008). Contrary to expectation, no additive or synergistic effects were observed and phenotype penetrance was actually reduced for four genes when targeted in combination rather than individually. This result was consistent with reduced transcript suppression following the combinatorial RNAi treatment and was suggested to reflect competition between siRNAs for RISC binding. An unexpected increase in transcript abundance for the dorsal gland gene dg13 when targeted in combination with a second gene, dg14, was also observed (Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2008).

Effect of incubation time

The duration of soaking in dsRNA/siRNA can have a considerable effect on the efficacy of RNAi. RNAi effects have been observed following exposure of preparasitic J2 nematodes to dsRNA for time periods ranging from 1 hr to 7 days with 16–24 hrs becoming the most common soaking period used. In general, increasing the incubation time leads to greater ingestion/uptake and subsequent transcript reduction and enhanced phenotypic effects, particularly for cyst nematodes (Chen et al. Reference Chen, Rehman, Smant and Jones2005; Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007; Sukno et al. Reference Sukno, McCuiston, Wong, Wang, Thon, Hussey, Baum and Davis2007). A 4 hr incubation is sufficient to induce RNAi for some cyst nematode genes (Urwin et al. Reference Urwin, Lilley and Atkinson2002; Lilley et al. Reference Lilley, Goodchild, Atkinson and Urwin2005b) but soaking for 24 hr was necessary to induce silencing in G. rostochiensis (Chen et al. Reference Chen, Rehman, Smant and Jones2005). For RNAi of the flp-12 gene of G. pallida, incubation in dsRNA for 12 hr or more was required to elicit an aberrant phenotype. A dramatic increase in the number of worms displaying an RNAi effect occurred between 18 and 24 hr when almost complete inhibition of migration was observed. For other flp genes, prolonged incubations of 2 or 7 days were required to achieve this maximum effect (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). It is likely that the required incubation times are related to the speed and efficiency of uptake in different nematodes and different experimental systems. For Meloidogyne spp. effective silencing has been consistently observed following a 4 hr incubation of J2 in dsRNA (Bakhetia et al. Reference Bakhetia, Charlton, Atkinson and McPherson2005a; Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005; Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006; Shingles et al. Reference Shingles, Lilley, Atkinson and Urwin2007). In one study with M. incognita, just 1 hr incubation was required to cause 90% transcript reduction of a glutathione-S-transferase gene, with a subsequent reduction in the number of egg masses produced by treated nematodes (Dubreuil et al. Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007). Bakhetia et al. (Reference Bakhetia, Charlton, Atkinson and McPherson2005a) noted that ingestion of FITC by J2 M. incognita after 4 hrs was considerably more extensive than that observed for cyst nematodes over the same time period, with fluorescence frequently visible along the entire intestinal lumen. Possibly the combination of two neurostimulants (resorcinol and serotonin) used by Dubreuil et al. (Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007) further enhanced the rate of ingestion.

Effect of dsRNA/siRNA concentration

RNAi soaking experiments with plant parasitic nematodes typically use dsRNA at concentrations from 1–5 mg/ml. Only a few studies have investigated concentration-dependent effects in an attempt to optimise in vitro RNAi techniques. G. pallida J2 were incubated for 24 hr in a 10-fold serial dilution of dsRNA targeting the flp-12 gene, with concentrations from 0·1 mg/ml to 1 pg/ml. (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). The migratory ability of nematodes, assessed as an indication of RNAi efficacy, was profoundly impaired at 0·1 mg/ml, with a significant effect still evident at 0·1 μg/ml. Nematodes incubated in ⩽10 ng/ml dsRNA were not significantly affected. Later experiments targeting the same gene found the transcript suppression induced by 21 bp siRNAs to be even more potent with motility still significantly inhibited at 10 ng/ml siRNA. A similar potency was observed when the M. incognita flp-18 gene was silenced by siRNA (Dalzell et al. Reference Dalzell, McMaster, Fleming and Maule2010a). In all cases there appears to be a threshold concentration below which a rapid decrease in efficacy is observed. Conflicting data arose from experiments to silence two pharyngeal gland genes of H. glycines. Whilst 5 mg/ml of the most potent dsRNA caused greater silencing of Hg-pel-1 than 2·5 mg/ml, the effect was reversed using a less potent dsRNA targeted at the same gene. Transcript suppression of Hg-4E02 was also greater with 2·5 mg/ml dsRNA than with 5 mg/ml (Sukno et al. Reference Sukno, McCuiston, Wong, Wang, Thon, Hussey, Baum and Davis2007).

In the light of these studies, and considering the non-specific phenotypes reported when using higher concentrations of non-nematode dsRNAs, it would seem prudent for further investigations to be carried out for a range of nematodes, target genes and tissue types. It is possible that lower concentrations of RNAi inducers could be routinely used in functional studies, providing reassurance that non-specific effects were minimised.

Persistence of the RNAi effect

Following removal of the RNAi trigger, the timing and duration of transcript suppression seem highly variable, possibly relating to target gene expression pattern, and initial level and turnover rate of the endogenous transcript. In most studies, transcript depletion tends to be assessed immediately after the soaking period, however this may not be when maximum silencing occurs. A pharyngeal gland-expressed calreticulin (mi-crt) gene of M. incognita displayed maximum transcript repression 20 hr after removal from a 4 hr exposure to dsRNA whereas transcript of polygalacturonase (mi-pg-1) was unaffected at the same time point and displayed optimal silencing after 44 hr. The transcript level of both genes had returned to normal after a 68 hr recovery period (Rosso et al. Reference Rosso, Dubrana, Cimbolini, Jaubert and Abad2005). For a third gene (Mi-gsts-1) expressed in the same cells, silencing was detectable immediately after the 1 hr dsRNA incubation and remained stable for 28 hrs, with transcript returning to normal levels after 48 hrs (Dubreuil et al. Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007). Similar results have been observed for cyst nematodes. Transcript repression of a β-1,4-endoglucanase was observed immediately following a 16 hr dsRNA treatment of H. glycines J2 and after a 5 day recovery period. Transcripts then increased unexpectedly at 10 days before returning to pre-treatment levels by 15 days after dsRNA exposure (Bakhetia et al. Reference Bakhetia, Urwin and Atkinson2007). Reduced motility of treated J2s was used to monitor persistence of Gp-flp-12 gene silencing. In this case, there was a small but insignificant recovery of phenotype after 24 hrs and a significant but not complete recovery after 6 days (Kimber et al. Reference Kimber, McKinney, McMaster, Day, Fleming and Maule2007). A delayed phenotypic effect may arise if the protein target is particularly stable. For instance, whilst immediate suppression of the Mi-gsts-1 transcript was apparent, GST enzyme activity was still present after 16 hrs and was not abolished until 24 hr post-treatment (Dubreuil et al. Reference Dubreuil, Magliano, Deleury, Abad and Rosso2007).

In C. elegans, an RNAi phenotype can be inherited; however, only when germline genes are targeted does the RNAi effect extend beyond the F1 generation (Grishok et al. Reference Grishok, Tabara and Mello2000). RNAi-induced silencing of a germline expressed green fluorescent protein (gfp) transgene was maintained over 80 generations (Vastenhouw et al. Reference Vastenhouw, Brunschwig, Okihara, Müller, Tijsterman and Plasterk2006). It appears that RNAi effects can also be inherited in plant parasitic nematodes without the constraint of germline expression. The M. javanica gene Cg-1 encodes a putative avirulence factor, with nematodes expressing this gene unable to develop on tomato plants carrying the Mi-1 resistance gene. Following treatment of M. javanica J2s with dsRNA, some individuals were able to multiply on resistant plants. Their progeny displayed increasingly successful reproduction over at least five generations whilst selection was maintained on resistant plants. The penetrance of the phenotype was reduced following growth on susceptible plants suggesting that the RNAi effect was lost in some progeny but persisted in the population due to the strong selection pressure (Gleason et al. Reference Gleason, Liu and Williamson2008).

Efficacy of host plant-delivered RNAi

To date, the approach has shown potential against both cyst and root-knot nematodes, despite early concerns that the feeding tubes of sedentary endoparasitic species would prevent the ingestion of dsRNA and possibly siRNAs (Bakhetia et al. Reference Bakhetia, Charlton, Urwin, McPherson and Atkinson2005b).

Some proof-of-concept studies have shown extremely promising results; a high level of resistance to root-knot nematode was achieved by targeting a parasitism gene expressed in the subventral gland cells of M. incognita (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006). dsRNA complementary to the 16D10 gene was expressed in transgenic A. thaliana resulting in a significant reduction [63–90%] in the number of galls and a corresponding reduction in total egg production. The high level of homology between the 16D10 sequences of different Meloidogyne species led to broad-range resistance against M. incognita, M. javanica, M. arenaria and M. hapla. Almost complete resistance to Meloidogyne infection was reported in tobacco plants expressing dsRNA corresponding to a nematode-specific collagen gene (patent application by Michaeli et al. Reference Michaeli, Kenigsbuch, Livneh, Levy and Khayat2005), splicing factor or integrase (Yadav et al. Reference Yadav, Veluthambi and Subramaniam2006) and of four M. incognita genes targeted from transgenic soybean roots, two reduced gall number by >90% (Ibrahim et al. Reference Ibrahim, Alkharouf, Meyer, Aly, Gamal El-Din, Hussein and Matthews2011). Plants that expressed dsRNA targeting a M. incognita tyrosine phosphatase gene supported 92% fewer galls than control plants with a similar reduction reported when a mitochondrial stress-70 protein precursor was targeted. Of the four genes examined, the lowest reduction in gall number was 57% when L-lactate dehydrogenase was targeted. In all cases, female nematodes that did develop within the galls were smaller than the controls (Ibrahim et al. Reference Ibrahim, Alkharouf, Meyer, Aly, Gamal El-Din, Hussein and Matthews2011). These latter studies all targeted genes presumed to be essential for nematode development based on the lethal RNAi phenotypes of their C. elegans orthologues. This may explain the potent phenotypic effects observed.

Partial resistance to Meloidogyne, based on a reduction in established nematodes, resulted from targeting either a dual oxidase gene (Miduox) with a probable role in cuticle formation or a subunit of signal peptidase (Mispc3), a protein complex required for the processing of secreted proteins. Crossing transgenic Arabidopsis lines expressing the two separate dsRNAs provided higher levels of resistance to M. incognita than either parent plant. The number of nematodes able to establish on the F2 plants was not different from the parental lines, however the proportion of nematodes reaching the adult female stage was lower when both dsRNAs were expressed together indicating further impediment of nematode development (Charlton et al. Reference Charlton, Harel, Bakhetia, Hibbard, Atkinson and McPherson2010). Not all host-delivered RNAi results in a resistance phenotype of Meloidogyne. Partial silencing of a putative transcription factor of M. javanica (MjTis11) did not significantly affect either nematode development or fecundity, perhaps because the degree of silencing was insufficient or the gene product does not have an essential, non-redundant role (Fairbairn et al. Reference Fairbairn, Cavallaro, Bernard, Mahalinga-Iyer, Graham and Botella2007).

Cyst nematodes can also be targeted by expression of dsRNAs in plant roots. Nematodes infecting transgenic soybean plants expressing a hairpin construct targeting the major sperm protein (MSP) gene of the soybean cyst nematode H. glycines displayed an overall 68% reduction in egg production (Steeves et al. Reference Steeves, Todd, Essig and Trick2006). However, the experimental set-up precluded clear interpretation of these results (Gheysen and Vanholme, Reference Gheysen and Vanholme2007). The in planta delivery of dsRNA from Arabidopsis to feeding H. schachtii reduced transcript abundance of a range of targeted parasitism genes in the pharyngeal gland cells although in some cases this reduction was only small and either non-significant or non-specific (Patel et al. Reference Patel, Hamamouch, Li, Hussey, Mitchum, Baum, Wang and Davis2008, Reference Patel, Hamamouch, Li, Hewezi, Hussey, Baum, Mitchum and Davis2010; Sindhu et al. Reference Sindhu, Maier, Mitchum, Hussey, Davis and Baum2009). For six of eight genes tested there was a significant reduction in female numbers of between 23–64% but with considerable variation between lines for some constructs (Sindhu et al. Reference Sindhu, Maier, Mitchum, Hussey, Davis and Baum2009). Variable, non-significant effects were also observed when a fibrilin gene of H. glycines was targeted from chimeric soybean plants whilst in the same study, RNAi of a coatomer subunit of this nematode resulted in a significant reduction in egg production (Li et al. 2010a). Soybean composite plants engineered to silence either of two ribosomal proteins, a spliceosomal protein or synaptobrevin, of H. glycines by RNAi resulted in 81–93% fewer females developing on the transgenic roots (Klink et al. Reference Klink, Kim, Martins, MacDonald, Beard, Alkharouf, Lee, Park and Matthews2009). A similarly high reduction in egg production was achieved by targeting H. glycines mRNA for splicing factor prp-17 or an uncharacterized gene cpn-1 from soybean roots (Li et al. Reference Li, Todd, Oakley, Lee and Trick2010b).

The success of RNAi in planta can depend on a number of factors, such as the use of a promoter able to drive a sufficiently high level of dsRNA expression and the pattern of expression of the target gene. There is evidence to suggest that a high level of hairpin RNA expression in feeding cells is required to produce gene silencing in feeding nematodes (Fairbairn et al. Reference Fairbairn, Cavallaro, Bernard, Mahalinga-Iyer, Graham and Botella2007). Although based on the analysis of a relatively small number of studies, it appears that in planta RNAi elicits more potent phenotypes when it targets nematode genes involved in essential cellular processes. The only parasitism gene target to confer a similarly high level of control is the 16D10 peptide of M. incognita. Possibly the majority of parasitism genes have combinatorial roles in the parasitic interaction or have redundant functions as members of gene families. Whilst off-target silencing in the nematode may not be problematic for RNAi-based resistance strategies, a lack of silencing specificity may influence results when RNAi is used as a tool for functional analysis. In contrast to in vitro studies, few of the in planta experiments (Sindhu et al. Reference Sindhu, Maier, Mitchum, Hussey, Davis and Baum2009; Charlton et al. Reference Charlton, Harel, Bakhetia, Hibbard, Atkinson and McPherson2010; Patel et al. Reference Patel, Hamamouch, Li, Hewezi, Hussey, Baum, Mitchum and Davis2010) have included a non-nematode dsRNA control as standard. There is also the additional constraint that expressed dsRNAs should have no homology with sequences in the host plant.

Confirmation of an in planta RNAi effect

Although a number of studies have reported a phenotype for nematodes parasitizing plants expressing dsRNAs, it has proved difficult to clearly demonstrate that the phenotype is due to an RNAi effect. Most commonly, qRT-PCR has been used to demonstrate a decrease in transcript level of the target gene, although the desired down-regulation has not always been detected (e.g. Patel et al. Reference Patel, Hamamouch, Li, Hussey, Mitchum, Baum, Wang and Davis2008). Ibrahim et al. (Reference Ibrahim, Alkharouf, Meyer, Aly, Gamal El-Din, Hussein and Matthews2011) reported that control gene transcripts were also reduced in nematodes parasitising dsRNA-expressing plants, this may reflect the fact that the few nematodes that had succeeded in developing to the point of assay were starting to lose viability. Given that only five studies have shown the accumulation of target gene siRNAs in the plant roots (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006; Steeves et al. Reference Steeves, Todd, Essig and Trick2006; Fairbairn et al. Reference Fairbairn, Cavallaro, Bernard, Mahalinga-Iyer, Graham and Botella2007; Sindhu et al. Reference Sindhu, Maier, Mitchum, Hussey, Davis and Baum2009; Li et al. Reference Li, Todd, Oakley, Lee and Trick2010b), the route by which gene silencing occurred cannot be determined. Other studies have used RT-PCR of the intron or spacer region of the hairpin construct to demonstrate the presence of unprocessed transcript (e.g. Patel et al. Reference Patel, Hamamouch, Li, Hussey, Mitchum, Baum, Wang and Davis2008, Reference Patel, Hamamouch, Li, Hewezi, Hussey, Baum, Mitchum and Davis2010). Charlton et al. (Reference Charlton, Harel, Bakhetia, Hibbard, Atkinson and McPherson2010) detected low levels of dsRNA in transgenic lines by northern blot analysis but reported a failure to detect corresponding siRNAs, despite observing an RNAi phenotype in the nematodes.

In C. elegans and plants, primary siRNAs derived from cleavage of exogenous dsRNA initiate the de novo synthesis of secondary siRNAs from the target transcript. These molecules are more abundant than the primary siRNAs and highly effective in gene silencing. With in planta RNAi, the target transcript is not present in the plant cell and this amplification step cannot occur. Simultaneous expression of the nematode gene and its corresponding hairpin construct in the same transgenic plant may lead to a higher level of nematode resistance, especially if siRNAs are the main form of RNA ingested by the feeding nematodes. When an Arabidopsis line expressing the Meloidogyne 16D10 transgene was crossed with a line expressing the 16D10 dsRNA, the F1 plants contained higher levels of 16D10 siRNAs (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006).