Plant parasitic nematodes cause significant damage to crops on a worldwide scale. These nematodes are often soil dwelling but rely on plants for food and to sustain them during reproduction. Complex interactions occur between plants and nematodes during the nematode life cycle with plant roots developing specialized feeding structures through which nematodes withdraw nutrients. Here we describe a novel method for delivering macromolecules to feeding nematodes using a virus-based vector [tobacco rattle virus (TRV)]. We show that the parasitic nematode Heterodera schachtii will ingest fluorescent proteins transiently expressed in plant roots infected with a TRV construct carrying the appropriate protein sequence. A prerequisite for this delivery is the presence of replicating virus in root tips prior to the formation of nematode-induced syncytia. We show also that TRV vectors expressing nematode gene sequences can be used to induce RNAi in the feeding nematodes.
Plant parasitic nematodes cause significant damage to crops grown throughout the world. The economic impact of the damage caused by nematodes, and the costs of their control, has been estimated at between $70 and 100 billion each year (Sasser and Freckman, 1987; Koenning et al., 1999). The majority of plant parasitic nematodes are soil organisms that feed on plant roots (Davis et al., 2004).
The most economically important plant parasitic nematodes are the sedentary endoparasites, in particular the root-knot (Meloidogyne spp.) and the cyst-forming nematodes (including Heterodera and Globodera spp.) (Koenning et al., 1999). These two groups of nematodes have similar life cycles (Williamson and Gleason, 2003; Davis et al., 2004). The eggs of cyst nematodes lie dormant in the soil within cysts formed from the body of the adult female. The second stage juveniles enclosed in the eggs hatch in response to host cues (Jones et al., 1998) and locate and invade roots of host plants. Root knot nematodes tend to have much broader host ranges and therefore hatch spontaneously. Cyst nematodes enter and move through the root tissue by piercing cells using their sharp mouthpiece (stylet) until they reach the vascular system. Root knot nematodes move intracellularly until they reach the cortex, move down the root and enter the vascular cylinder at the root apex. Both groups of nematodes secrete a cocktail of plant cell wall-degrading enzymes to assist this migration process (Smant et al., 1998). The nematodes then induce the formation of specialized feeding structures in the host tissue by altering the development of the root cells surrounding them. Cyst nematodes induce the formation of a syncytium, a large multinucleate cell formed by breakdown of plant cell walls and fusion of adjacent protoplasts. Root knot nematodes induce giant cells, formed through repeated rounds of nuclear division and cell growth in the absence of cytokinesis (reviewed in Gheysen and Fenoll, 2002). Despite their different ontogeny, many of the functions and cellular features of the two types of feeding cell are similar. Both have multiple enlarged nuclei, small vacuoles and show proliferation of smooth endoplasmic reticulum, ribosomes, mitochondria, and plastids. In both cases formation of the feeding site involves large changes in gene expression within the root tissue (Gheysen and Fenoll, 2002). The nematodes remain at these feeding sites and develop through a further three moults into adult males or females. Female cyst nematodes develop into cysts that contain several hundred eggs, whereas female root-knot nematodes produce egg masses containing several hundred eggs directly into the rhizosphere.
Current methods of nematode control include the use of organophosphates that target invasive stage juveniles or methyl bromide applied as a soil fumigant (Hutchinson et al., 1999; Chitwood, 2002). Methyl bromide is a major ozone-depleting agent and its use is being phased out under the Montreal protocol, while use of organophosphates is becoming increasingly restricted due to both legislative and consumer pressure. Alternative methods for controlling plant parasitic nematodes are therefore being sought, and much effort has been put into developing genetically modified plants resistant to plant parasitic nematodes (Atkinson et al., 2003).
Many different strategies that could provide resistance to plant parasitic nematodes have been suggested. These include expression in plants of protease inhibitors or Bt endotoxins that target the nematode, or of factors that target the developing feeding site (Atkinson et al., 2003). In addition, many thousands of novel peptides are available that may be of use for nematode control (Pomilio et al., 2006). Generating such peptides and screening them for activity against free-living nematodes is a relatively simple process and one that is amenable to high throughput analysis. However, endoparasitic nematodes do not feed until the feeding site is established. Screening for peptides against such pathogens therefore requires production of transgenic plants that express candidate proteins (Liu et al., 2005). This process is expensive, time-consuming, and represents a bottleneck in the pipeline from peptide discovery to functional verification.
Viral vectors have been developed as a powerful means of transiently expressing proteins in planta (Scholthof et al., 1996). Vectors have been developed from both DNA and RNA (Gleba et al., 2004) viruses, including potato virus X, tobacco mosaic virus (TMV), wheat streak mosaic virus and tobacco rattle virus (TRV). More recently, TRV has been shown to be particularly useful for expressing proteins in root tissues (MacFarlane and Popovich, 2000), with the presence of the 2b gene of this virus increasing viral root tropism (Valentine et al., 2002, 2004).
Viral vectors have also been used to induce post-transcriptional gene silencing in plants (Angell and Baulcombe, 1997; Ruiz et al., 1998), including plant roots (Faivre-Rampant et al., 2004; Ryu et al., 2004; Saedler and Baldwin, 2004; Valentine et al., 2004). The process involves the sequence-specific degradation of RNA (Agrawal et al., 2003), and is similar to RNAi in animals and quelling in fungi. During silencing, sequence-specific 22- to 24-nt-long double-stranded RNA is produced. RNAi in nematodes has been mainly studied using the free-living Caenorhabditis elegans, where RNAi can be induced by soaking nematodes in dsRNA. dsRNA produced during silencing in plants using a GFP transgene has also been used to induce silencing of a GFP transgene in C. elegans, by injecting RNA extracted from plants undergoing silencing into the nematode's gut (Boutla et al., 2002). Although soaking does not readily induce RNAi in parasitic nematodes, H. schachtii and other cyst nematodes can be stimulated to take up dsRNA from solution, and this has been shown to induce silencing (Urwin et al., 2002). These techniques have subsequently been used to investigate gene function in plant parasitic nematodes (e.g. Chen et al., 2005). This area has been reviewed in Bakhetia et al. (2005) and, more recently, plants expressing dsRNA have been shown to provide resistance against plant parasitic nematodes (Huang et al., 2006).
Here we describe a system that uses TRV-based vectors for delivering peptides to cyst nematodes via the syncytium, and demonstrate that proteins delivered in this way can be taken up by feeding nematodes. We show also that double-stranded RNA can be delivered to feeding nematodes by the same route, and that the effects of targeting an endogenous nematode RNA can be detected. The use of viral vectors has the potential to enable the screening of peptides, or other macromolecules, for nematicidal activity in vivo prior to the construction of transgenic plant material.
Results and discussion
Uptake of virally expressed proteins by feeding nematodes
A pathosystem consisting of TRV, Arabidopsis and the beet cyst nematode H. schachtii was used to demonstrate the feasibility of delivering proteins to nematodes via the feeding site. Arabidopsis was chosen as it is a host for H. schachtii in an in vitro culture system (Wyss and Grundler, 1992), and has thin roots that allow detailed observations of the nematode feeding sites to be made at all life-cycle stages. TRV is an RNA virus with a genome that consists of two RNAs. The viral vectors were modified to express one of two fluorescent proteins; dsRED (Gross et al., 2000) or the smaller monomeric red fluorescent protein [mRFP; (Campbell et al., 2002) (Figure 1a)]. Although the length of the genes encoding dsRED and mRFP is approximately 700 bp for both proteins, mature dsRED tends to form tetramers and therefore has a larger Stokes radius than mRFP (Campbell et al., 2002). This is important as cyst nematodes produce a feeding tube that excludes molecules above 28 kDa (Urwin et al., 1997). The vectors also included the 2b gene, which improves the movement of the vector into the roots of Arabidopsis (Valentine et al., 2004).
Initial experiments examined whether H. schachtii would develop on TRV-infected plants and take up proteins expressed from the viral genome. Red fluorescence, indicating presence of protein expressed by TRV, could be seen in the leaves of Arabidopsis plants inoculated with the TRV constructs 2–4 days after inoculation, and could be seen in the roots of plants from approximately 5 days after inoculation onwards (data not shown). Nematodes were able to locate and invade these plants and went on to develop to the adult stage. The time frame for this development was comparable to that on A. thaliana plants not infected with TRV (Figure 1b). In addition, we found no significant difference in the number of nematodes attaching and developing on virus infected plants (29 ± SE 1.9 attached nematodes per plant) compared with uninfected plants (30.0 ± SE 3.5 attached nematodes per plant). Plants of different ages were inoculated with virus and the degree of virus movement subsequently tracked. Virus movement was more efficient in younger plants (Figure 1c).
To visualize virus replication in roots, we used Arabidopsis plants expressing GFP under control of the companion cell-specific promoter SUC2 (SUC2:GFP). These plants were used to aid location of nematode and syncytia as during early syncytial development de novo production of phloem results in GFP accumulation around the syncytium (Hoth et al., 2005). In virus-infected plants, nematodes established feeding sites in roots containing replicating virus. As the nematode-feeding site developed, expression of virally encoded mRFP or dsRed protein was clearly visible in syncytia that developed close to GFP-labelled phloem poles. Figure 2 shows the distribution of dsRED expression in a syncytium due to TRV.2b.dsRED infection. The proportions of red fluorescent syncytia per virally infected root varied from 5 to 18% and depended on the proportion of plant roots that contained virus particles at the time that the feeding site was established (Figure 1d). Plants infected with TRV.2b.dsRED exhibited high levels of fluorescence in the syncytia and very low levels of fluorescence (detectable only with confocal microscopy) in some of the nematodes (data not shown). However, when plants were infected with the TRV.2b.mRFP construct, mRFP could be seen very clearly in the digestive system of many feeding nematodes using either standard fluorescence or confocal microscopy (Figure 3). No GFP uptake was visible, this is consistent with the result of Urwin et al. (1997). The difference in the levels of uptake is likely to be linked to the differences in relative size of the fluorophores (Urwin et al., 1997; Campbell et al., 2002). At 11 dpi, approximately 5% of the total nematodes exhibited red fluorescence in the digestive system indicating uptake of the mRFP protein. This represented uptake of protein in 25% of those nematodes attached to roots in which replicating virus was present, as assessed by scoring mRFP-labelled roots. Our observations suggested that nematodes did not take up mRFP continuously. This discontinuous uptake could be due to the fact that plant parasitic nematodes are known to feed in cycles (Wyss, 2002). Ingestion of cytosol from the syncytium is interrupted by phases in which secreted proteins are introduced into the syncytium by the nematode. It is therefore likely that mRFP would be visible in the digestive system for only a short time after each ingestion phase, and that this fluorescence would disappear as the proteins in the digestive system were broken down, with further subsequent fluorescence observed after the next ingestion phase. Thus, the figure of 25% above is probably an underestimate of the proportion of nematodes that were feeding at any one time. mRFP was observed in the digestive systems of nematodes at all developmental stages in roots from sedentary J2 through to J4, and adult females (Figure 3). The fluorescence was visible using a stereo microscope (Figure 3a–h) or using confocal microscopy (Figure 3i–l). Fluorescence was visible throughout the nematode gut.
RNAi in feeding nematodes
TRV vectors were constructed that included either a single length of H. schachtii glyceraldehyde-3-phoshate dehydrogenase (GAPDH) sequence, replacing the dsRED sequence, or a hairpin consisting of two 70 bp fragments of GAPDH sequence cloned in opposite convergent orientation creating a hairpin either before or after the dsRED sequence (Figure 4a). The development of nematodes on Arabidopsis infected with these viral clones was followed, and the size of the nematodes at 14 dpi was recorded. Nematodes exposed to the constructs containing GAPDH sequence showed a significant 10–15% decrease in size across the entire female population (P = 0.016) (Figure 4b). We next tested whether any decrease in the H. schachtii GAPDH mRNA was detectable in nematodes exposed to the constructs. Semiquantitative RT-PCR, in which expression levels of the GAPDH mRNA were compared with an endogenous control (ATPase), showed a significant reduction in the level of GAPDH mRNA in those nematodes exposed to the constructs containing GAPDH sequence compared to the endogenous control gene (Figure 4c,d). The data were repeated for each of the sequences for at least three different plants and shown to be statistically significant (P < 0.05) (Figure 4d).
Unpublished experiments on the effectiveness of hairpin vs. sense or antisense fragments to generate silencing have concluded that TRV carrying a sequence fragment larger than 150 bp in length in the sense or antisense orientation can silence with a comparable efficiency, compared with short hairpins (60 bp inverted repeats). In this respect, the situation with TRV is comparable to BSMV as a comparable silencing response was observed with larger sense or antisense fragments and shorter hairpins using BSMV vectors (Lacomme et al., 2003), as opposed to TMV where short hairpins trigger a strong silencing response (in comparison to larger antisense fragments, Lacomme et al., 2003). It is likely that not all viruses will tolerate the presence of small inverted repeats (because it may impede viral replication, increase recombination events leading to a faster elimination of the inverted-repeat sequence). Another explanation maybe differences in the ratio of positive strand vs. negative strand replicative intermediates between different viruses. In the case of TMV, the positive strand is in large excess to the negative strand (100–1), therefore dsRNA generated by subgenomic RNA expression of hairpin is far more abundant than any other forms of dsRNA (C. Lacomme, pers. comm.).
These experiments suggest that dsRNA sequence, derived from nematode DNA sequence and expressed using viral constructs, can induce a reduction in gene expression, presumably due to RNAi, in the feeding nematodes. The actual level of RNA reduction observed in these samples is influenced both by the level of RNAi induced in an individual nematode and the proportion of nematodes exposed to the virus-encoded dsRNA. Experiments using virally encoded dsRed or mRFP showed that between 5 and 30% of the nematode-feeding sites induced in plants infected with these viral constructs contained replicating virus (see above). Assuming that a similar proportion of feeding sites were infected with the gene silencing constructs, it is likely that only a proportion of the nematodes included in the RNA extractions from plants infected with these constructs were exposed to the virus and/or silencing signal, and thus exposed to the optimal conditions required for RNAi induction. The mechanism and efficiency of transfer of silencing trigger from the feeding site to the nematode are also unknown. These factors may explain why the RNAi effect was not more marked in the feeding nematode population.
Here, we have demonstrated the utility of TRV as a vector for delivering proteins to feeding cyst nematodes, and have shown also that virally encoded factors can be used to induce a reduction in expression of target genes in feeding nematodes. These results provide a basis for testing novel peptides and dsRNAs for their efficacy in nematode control without the need for generation of transgenic plants, and may present a new way in which to study parasitic nematode gene function using RNAi. Further analysis will be needed to assess the range of parasitic nematode tissues that can be targeted by this method of RNAi induction.
Viral construct synthesis
All RNA2 constructs were based on a TRV clone containing the 2b gene as described in Valentine et al. (2004). TRV.2b.mRFP was constructed by direct replacement of the dsRED using NcoI and EagI restriction sites. H. schachtii genes were obtained from mRNA extracted from J2s of H. schachtii using a Micro-FastTrack™ 2.0 Kit (Invitrogen) following the manufacturer's instructions. Individual H. schachtii genes were then amplified by RT-PCR using primers designed to the consensus sequences within the Globodera rostochiensis GAPDH sequence (EMBL AF004522). 5′ primers contained either an NcoI or an EagI restriction site and 3′ primers included an XhoI or an EagI restriction site. Viral clones were produced that included either approximately 500 bp of sequence encoding a H. schachtii gene or a hairpin sequence of approximately 80 bp encoding a dsRNA targeting a H. schachtii gene. In both cases, these sequences were inserted into either the NcoI or the EagI restriction site, leaving the dsRED sequence intact. Hairpins were produced as direct inverted repeats as described in Lacomme et al. (2003).
Plant growth and inoculation
Nicotiana benthamiana were grown under glasshouse conditions at 22 °C with a 16-h day length. Infectious sap was produced by inoculating N. benthamiana with transcripts of RNA2 constructs, produced in vitro using the Ambion Message Machine TM kit mixed with RNA1 obtained as previously described (Vassilakos et al., 2001). The two RNAs were resuspended in 0.1% bentonite solution in phosphate buffer and rubbed on to N. benthamiana using carborundum. Leaves showing signs of systemic virus infections were ground in sterile distilled water (SDW) at a ratio of 1 g leaf tissue to 1 mL SDW to produce infectious sap. Sap was stored frozen until use.
Seeds of A. thaliana (C24) were surface sterilized by rinsing them in 70% ethanol followed by 95% ethanol prior to sowing single seeds in 12-well multiwell sterile plates on 0.2 Knop medium (Sijmons et al., 1991). Plates were sealed with micropore tape and initially incubated at 4 °C for 2–5 days to break seed dormancy before being transferred to an incubator and kept at 22 °C with a 12 h day length.
Arabidopsis plants between 10 and 14 days old were inoculated with TRV either using the method described by Valentine et al. (2004) or by mixing transcribed RNA2 with infectious sap prepared as described above.
Maintenance and inoculation of plants with H. schachtii
H. schachtii was maintained on Sinapsis alba grown on 0.2 Knop medium (Sijmons et al., 1991). Cysts were transferred to 3 mm ZnCl2 to induce hatching of J2 juveniles that were collected 5–8 days later and sterilized by rinsing in 0.5% HgCl2 followed by several washes with SDW. Fifty J2s were inoculated on to each Arabidopsis plant previously infected with TRV constructs. Where the TRV construct carried a visible marker such as dsRED, only plants expressing the marker in systemic leaves or roots were inoculated. Unless otherwise stated, plants were inoculated with nematodes 7–8 days after virus inoculation.
Movement of viruses carrying fluorescent markers and uptake of proteins by nematodes was monitored in situ using a Coolview digital camera (Photonic Science) attached to a Leica fluorescence stereomicroscope (MZFLIII) or a Bio-Rad MRC 1000 confocal laser scanning microscope (CLSM; Bio-Rad, Hemel Hempstead, UK). Growth of nematodes was monitored using a Coolview digital camera on a Leica microscope. Image pro plus software (Media Cybernetics, Silver Spring, MD, USA) was used to gather quantitative data on nematode development.
mRNA transcript levels
mRNA levels in developing females were measured by competitive RT-PCR. Primers for the RT-PCR reactions were designed against areas of the coding sequence not contained in any of the viruses used for the RNAi induction. Whole roots with all associated nematodes were frozen in liquid nitrogen 15 days after inoculation of nematodes onto plants. Plant and nematode RNA was extracted simultaneously using a Micro-FastTrack 2.0 Kit (Invitrogen) as described above. Equal quantities of the reverse primers for ATPase and GAPDH were added to each RNA extraction to prime the reverse transcription reaction using Superscript II (Invitrogen). PCR was performed using Expand DNA Polymerase and reactions included sets of primers for both ATPase and GAPDH. Ten microlitres of each reaction was removed at three cycle intervals from 21 cycles to 42 cycles, and the quantities of PCR product were assessed after agarose gel electrophoresis and ethidium bromide staining. Relative levels of GAPDH and ATPase PCR products were calculated for each PCR using the gel analysis plug-in for image j image analysis software (Rasband and Image, 1997–2006). GAPDH levels were calculated relative to the internal ATPase levels before expressing as a percentage of the dsRED viral control GAPDH level.
This work was supported by the Large Scale Biology Corporation, the Scottish Executive Environment and Rural Affairs Department (SEERAD) and Scottish Enterprise via their Proof of Concept programme. We would also like to thanks Mylnefield Research Services Ltd for their assistance and Dr Christophe Lacomme for helpful discussions.