Applications and advantages of virus-induced gene silencing for gene function studies in plants

Authors

  • Tessa M. Burch-Smith,

    1. Department of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, CT 06520-8104, USA,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Jeffrey C. Anderson,

    1. Department of Molecular Biology and Genetics, Cornell University, Ithaca, NY 14853-2703, USA,
    2. Boyce Thompson Institute for Plant Research, Tower Rd., Ithaca, NY 14853-1801, USA, and
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Gregory B. Martin,

    1. Boyce Thompson Institute for Plant Research, Tower Rd., Ithaca, NY 14853-1801, USA, and
    2. Department of Plant Pathology, Cornell University, Ithaca, NY 14853-4203, USA
    Search for more papers by this author
  • S. P. Dinesh-Kumar

    Corresponding author
    1. Department of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, CT 06520-8104, USA,
    Search for more papers by this author

*(Fax: +1 203 432 6161; e-mail savithramma.dinesh-kumar@yale.edu).

Summary

Virus-induced gene silencing (VIGS) is a recently developed gene transcript suppression technique for characterizing the function of plant genes. The approach involves cloning a short sequence of a targeted plant gene into a viral delivery vector. The vector is used to infect a young plant, and in a few weeks natural defense mechanisms of the plant directed at suppressing virus replication also result in specific degradation of mRNAs from the endogenous plant gene that is targeted for silencing. VIGS is rapid (3–4 weeks from infection to silencing), does not require development of stable transformants, allows characterization of phenotypes that might be lethal in stable lines, and offers the potential to silence either individual or multiple members of a gene family. Here we briefly review the discoveries that led to the development of VIGS and what is known about the experimental requirements for effective silencing. We describe the methodology of VIGS and how it can be optimized and used for both forward and reverse genetics studies. Advantages and disadvantages of VIGS compared with other loss-of-function approaches available for plants are discussed, along with how the limitations of VIGS might be overcome. Examples are reviewed where VIGS has been used to provide important new insights into the roles of specific genes in plant development and plant defense responses. Finally, we examine the future prospects for VIGS as a powerful tool for assessing and characterizing the function of plant genes.

Overview of developments leading to discovery of VIGS

Viruses are among the most damaging of plant pathogens and elicit a variety of responses from their hosts. Plant biologists have studied these responses for many years in attempts to engineer crops that are resistant to viruses and thereby decrease yield losses caused by these pathogens. One early observation was that virus-infected plants were subsequently resistant to infection by the same or a closely related strain of the same virus (McKinney, 1929). This phenomenon was termed ‘cross-protection’. Early attempts at engineering resistance relied on the introduction of viral genetic material into the plant genome, a technique described as pathogen-derived resistance (PDR) (for review see Beachy, 1997; Wilson, 1993). Some plants engineered for PDR exhibited a recovery phenotype; they initially developed disease symptoms upon virus infection but these symptoms were absent in newly emerging tissue (Lindbo et al., 1993). The underlying molecular basis of cross-protection, PDR and recovery was later found to be post-transcriptional gene silencing (PTGS) (Ratcliff et al., 1997).

PTGS is an epigenetic phenomenon that results in the sequence-specific degradation of endogenous mRNAs. PTGS was first described in plants and referred to as co-suppression (van der Krol et al., 1990; Napoli et al., 1990). Since then, it has also been identified in several other organisms and is termed quelling in fungi and RNAi in animals (Cogoni et al., 1996; Fire et al., 1998; Romano and Macino, 1992; Vaucheret et al., 2001). PTGS is characterized by the absence of accumulated transcript from either an endogenous gene or a stably integrated transgene when a sequence bearing homology to that gene or transgene is introduced.

Considerable effort has gone into elucidating the mechanism of PTGS and there are many excellent reviews on the topic (McManus and Sharp, 2002; Tijsterman et al., 2002; Vaucheret et al., 2001) so we shall only briefly describe it here as far as is relevant to the current discussion. The triggering molecule of PTGS is double-stranded RNA (dsRNA) (Fire et al., 1998; Klahre et al., 2002; Metzlaff et al., 1997; Waterhouse et al., 1998). This dsRNA is cleaved to produce small guide molecules called short interfering RNA (siRNA) (Hamilton and Baulcombe, 1999; Martinez et al., 2002). The antisense strand of the siRNA associates with the RNAi silencing complex (RISC) to target homologous RNA for degradation (reviewed in Bartel, 2004). Also associated with this process is the methylation of transgenes undergoing silencing by PTGS (English et al., 1996; van Houdt et al., 1997; Ingelbrecht et al., 1994; Jones et al., 1999; Sijen et al., 1996). The significance of this RNA-mediated epigenetic change is still not clear although it is thought to help maintain silencing (Aufsatz et al., 2002). There has been much debate about the origin of the initiating dsRNA. In the case of viral infection in plants, the dsRNA is believed to be associated with the replicative phase of the life cycle of RNA viruses (Dalmay et al., 2000; Waterhouse et al., 1998). For transgenes, the dsRNA may be generated by the action of a host RNA-dependent RNA polymerase (RdRp) (Dalmay et al., 2000). Transgenes that insert as or rearrange to form inverted tandem repeats may also produce dsRNA after transcription (Waterhouse et al., 1998). Transgenes that are designed to produce dsRNA are also very potent PTGS inducers (Chuang and Meyerowitz, 2000; Smith et al., 2000).

There are several lines of evidence pointing to the evolution of PTGS as an antiviral mechanism in plants (Vaucheret et al., 2001). One compelling observation is that viruses by themselves can be triggers for PTGS in certain plant species (Al-Kaff et al., 1998; Marathe et al., 2000; Ratcliff et al., 1997, 1999). In addition, a number of Arabidopsis mutants that are compromised in PTGS demonstrate increased susceptibility to virus infection (Dalmay et al., 2001; Mourrain et al., 2000). Interestingly, viruses are not passive in the face of this plant defense and they have evolved proteins that can act as suppressors of PTGS (Anadalakshmi et al., 1998; Beclin et al., 1998; Brigneti et al., 1998; Kasschau and Carrington, 1998; Voinnet et al., 1999). This is perhaps the most convincing argument that PTGS acts as an antiviral mechanism in plants.

Development of VIGS technology

The term ‘virus-induced gene silencing’ (VIGS) was first used by A. van Kammen to describe the phenomenon of recovery from virus infection (van Kammen, 1997). However, the term has since been applied almost exclusively to the technique involving recombinant viruses to knock down expression of endogenous genes (Baulcombe, 1999; Ruiz et al., 1998). The discovery of PTGS of endogenous genes by recombinant viruses carrying a near-identical sequence was made in 1995 (Kumagai et al., 1995). Because it allows the targeted downregulation of a particular gene through the degradation of its transcripts, the potential of VIGS as a tool for the analysis of gene function was quickly recognized (Baulcombe, 1999).

A DNA fragment with a minimum of 23 nucleotides bearing 100% identity to a targeted transgene appears to be required in order for silencing to occur (Thomas et al., 2001). However, a 23-nucleotide long sequence is often not sufficient to initiate silencing and longer identical sequences must sometimes be used (Ekengren et al., 2003; Thomas et al., 2001). It is possible that other factors including the nucleotide composition of the targeting sequence (Thomas et al., 2001) and the thermodynamic properties of the siRNA and target sequence pair (Khvorova et al., 2003; Schwarz et al., 2003) are also important for determining the efficacy of silencing by a given targeting sequence.

Many vectors for use in VIGS have been developed over the past 8 years (Table 1). The earliest of these vectors was based on the model RNA virus, tobacco mosaic virus (TMV) (Kumagai et al., 1995). Transcripts of recombinant virus carrying a sequence from the phytoene desaturase (PDS) gene were produced in vitro and inoculated onto Nicotiana benthamiana plants to successfully silence PDS. In early studies, VIGS was usually performed in the wild tobacco species N. benthamiana that is highly susceptible to virus infection and thus gives ready silencing because of good infection. A more recent VIGS vector is based on potato virus X (PVX) (Ruiz et al., 1998). Although this vector is more stable than the TMV-based vector, PVX has a more limited host range than TMV, with only three plant families having members that are susceptible to PVX infection compared with nine families for TMV (Brunt et al., 1996). Furthermore, both TMV and PVX-based vectors cause disease symptoms on inoculated plants, thus making interpretation of some subtle PTGS phenotypes difficult (Ratcliff et al., 2001). In addition, these viruses are excluded from the growing points or meristems of their hosts, which precludes effective silencing of genes in those tissues (Hull, 2002.; Ratcliff et al., 2001). The significance of this exclusion is still unclear as a VIGS vector based on the tomato golden mosaic DNA virus (TGMV) was used to successfully silence a meristem gene, proliferating cell nuclear antigen (PCNA) in N. benthamiana, although this virus is excluded from the meristem (Peele et al., 2001). This TGMV-based vector had been previously used to silence a non-meristematic gene as well as a foreign transgene (Kjemtrup et al., 1998).

Table 1.  Summary of features of VIGS vectors
Vector SpeciesaViral symptoms EfficiencybAdopted for large-scale studiesReference
  1. TMV, tobacco mosaic virus; PVX, potato virus X; TGMV, tomato golden mosaic virus; TRV, tobacco rattle virus; SVISS, satellite virus-induced silencing system; BSMV, barley stripe mosaic virus; CbLCV, cabbage leaf curl virus.

  2. aPlant species in which silencing has been shown to be effective.

  3. bEfficiency is the percentage of leaf area that shows evidence of silencing and the degree of silencing that occurs.

TMVN. benthamianaSevereFair; better with antisense
Improved with inclusion of direct inverted repeats
YesKumagai et al. (1995), Lacomme et al. (2003), Fitzmaurice et al. (2002)
PVXN. benthamianaModerateFair; not persistentYesRuiz et al. (1998), Lu et al. (2003)
TGMVN. benthamianaVariableFair; varies with gene silenced, size of insertNoKjemtrup et al. (1998), Peele et al. (2001)
TRVN. benthamianaMildGood; persistent and consistent between experimentsYes, available for GATEWAY cloningRatcliff et al. (2001), Liu et al. (2002b)
TomatoMildVariable; depends on inoculation techniqueYes, available for GATEWAY cloningLiu et al. (2002a)
SVISSTobaccoMildGood; persistent and consistent between experimentsNoGossele et al. (2002)
BSMVBarleyModerateFair; depends on target sequence; not persistent
Improved with use of direct inverted repeats of targeting sequence
YesHolzberg et al. (2002), Lacomme et al. (2003), Fitzmaurice et al. (2002)
CbLCVArabidopsisModerate in older plantsFair; effects variable with gene silenced and vector designNoTurnage et al. (2002)

The limitations of host range and meristem exclusion were overcome with the development of VIGS vectors based on tobacco rattle virus (TRV) (Liu et al., 2002b; Ratcliff et al., 2001). TRV is able to spread more vigorously throughout the entire plant, including meristem tissue, yet the overall symptoms of infection are mild compared with other viruses. The improved TRV VIGS vectors, pYL156 and pYL279, result in more efficient silencing of endogenous genes (Liu et al., 2002a,b). These vectors differ from the earlier TRV vector by having a duplicated 35S promoter and a ribozyme at the C-terminus for more efficient production of viral RNA, as well as a number of amino acid changes in the viral sequence itself (Liu et al., 2002b). Unlike the other VIGS vectors described so far, pYL156 and pYL279 are not limited to efficacy in N. benthamiana and they are being successfully used for silencing in tomato and other species (Ekengren et al., 2003; Liu et al., 2002a) (see below).

The number of plant species amenable to VIGS experiments is increasing with the development of new virus vectors. A two-component system (satellite virus-induced silencing system, or SVISS) has been described that allows VIGS in tobacco, a species that is widely used for plant biology but which has proven recalcitrant to the use of other VIGS vectors (Gossele et al., 2002). This system consists of a satellite TMV (STMV) and a helper virus, TMV strain U2. A VIGS vector has also been developed for silencing in barley, a monocotyledonous plant (Holzberg et al., 2002). This barley stripe mosaic virus (BSMV)-derived vector has been used to silence PDS and to date is the only example of VIGS in a monocot (see below). Indeed, the development of VIGS for functional genomics in monocots is significant because of the difficulty in applying other loss-of-function approaches requiring transformation to these species. A modified BSMV VIGS vector has been developed that has a greater efficiency of silencing as demonstrated by a more severe silencing phenotype. This new vector contains 40–60 base pair direct inverted repeats that generate dsRNA upon transcription of the VIGS vector (Lacomme et al., 2003). The same researchers have made similar improvements to the TMV VIGS vector. A vector based on a DNA virus has been developed and used to silence an endogenous gene in Arabidopsis (Turnage et al., 2002). This vector, derived from cabbage leaf curl virus (CbLCV), is to date the only vector reported for use directly for transient VIGS in Arabidopsis.

Methodology of VIGS

The most powerful aspect of VIGS as a tool for gene function studies and high-throughput functional genomics is the minimal amount of time and effort required to identify a loss-of-function phenotype for a gene of interest (Figure 1). Initially, single gene sequences were subcloned individually into viral genomes and plants were inoculated by rubbing leaves with viral RNA produced by in vitro transcription reactions (Kumagai et al., 1995). Although this approach is well-suited for studies looking at a limited number of genes, inoculating the plants in this manner is time-consuming and can yield variable results. Recent efforts to streamline the cloning process and subsequent inoculation of the virus have made it possible to go from gene sequence to phenotype in planta within 1 month, allowing a single lab to screen thousands of individual genes for a phenotype of interest in a high-throughput manner (Lu et al., 2003).

Figure 1.

Method for high-throughput virus-induced gene silencing (VIGS). VIGS is performed by cloning a short stretch of sequence (usually 100–500 base pairs) from a candidate gene or random cDNAs into a virus genome under the control of the CaMV 35S promoter within a binary vector. The virus construct is transformed into Agrobacterium and inoculated into seedlings by a toothpick, vacuum infiltration, or syringe infiltration. The virus spreads from the site of Agrobacterium transformation into the upper vasculature and leaves of the growing plant, triggering a plant defense mechanism that suppresses both viral replication and expression of the endogenous host gene that was targeted. After 2–3 weeks, the plants are scored for loss-of-function phenotypes associated with suppression of the target gene.

Constructing the virus vector

To simplify VIGS protocols, several different approaches to subcloning and virus inoculation have been developed. For larger library screens, virus vectors containing Gateway® (Invitrogen Corp., Carlsbad, CA, USA) cloning sites are now available and allow for efficient subcloning of a large number of cDNAs from EST libraries (Liu et al., 2002a). The upper limit for the size of insert sequence may depend on size constraints of the virus to move from cell to cell and may vary between viruses and plant species. In our experience, the practical limit is around 1.5 kB of gene sequence that can be inserted into PVX and TRV for silencing in N. benthamiana. Above this limit, the virus may not spread or may lose the insert at a higher rate. While the lower limit on insert size has been experimentally determined to be 23 nucleotides of exact identity for transgene silencing (Thomas et al., 2001), we generally use fragments in the range of 300–500 nucleotides with multiple stretches of more than 23 nucleotide identity to ensure efficient silencing of endogenous genes.

However, as VIGS theoretically requires only a minimal amount of target sequence, one plausible approach that could be taken would be to simply synthesize oligonucleotides corresponding to the gene of interest for subcloning into a VIGS vector. These oligonucleotides would contain a minimum of 23 base pairs of homology but longer sequences should yield more efficient silencing.

The region of a gene to target for silencing must be carefully considered for every VIGS experiment. Because PTGS will potentially silence any transcript containing at least 23 nucleotides identity to the targeted sequence, it is important to determine at the outset, through BLAST searches or DNA gel blot analyses, whether or not the gene to be suppressed is member of a gene family. For single copy genes, presumably any region of the open reading frame (ORF) can be used for silencing. To suppress a single gene within a family, one must carefully select regions that do not contain stretches of 23 nucleotide of exact identity. One possible approach to be considered is the use of UTR sequence to discriminate between genes with highly conserved ORFs. Conversely, it may be practical to choose regions that are conserved between genes to co-suppress several members of a gene family and overcome possible functional redundancy (Lu et al., 2003). The flexibility of VIGS to target either single or several genes will likely be an extremely valuable tool for dissecting overlapping functions within gene families.

Recent work has demonstrated that the simultaneous silencing of several distinct genes is possible by including multiple gene sequences in the virus. Peele and coworkers used a TGMV VIGS vector to co-silence a magnesium chelatase subunit (su) and proliferating cell nuclear antigen (PCNA) (Peele et al., 2001), whereas a CbLCV VIGS vector was used to silence Chlorata 42 and PDS in Arabidopsis (Turnage et al., 2002). As the sophistication of VIGS experiments continues to increase, it may become possible to suppress the expression of multiple, distinct genes at the same time to elucidate epistatic relationships as has been carried out for lignin biosynthesis using an antisense approach (Abbott et al., 2002). Furthermore, ligating combinations of random distinct gene sequences in the same vector could be one approach to limiting the number of plants required for a large-scale screen.

Inoculation techniques

While several different approaches to subcloning and virus inoculation exist, the basic principles of VIGS remain the same (Dinesh-Kumar et al., 2003; Ekengren et al., 2003; Liu et al., 2002a,b). Essentially, an infectious viral RNA or DNA containing sequence from a gene of interest is introduced into young plants (Figure 1). As the plant grows, the virus spreads from the site of inoculation into the developing regions of the plant and triggers PTGS. With TRV on N. benthamiana and tomato, mild viral symptoms typically develop within 7–10 days after inoculation and by 2–3 weeks the virus has spread to fully expanded upper leaves. Silencing is coincident with viral spread and is usually greatest 2–3 weeks after inoculation.

To facilitate faster virus inoculation, RNA virus genomes have been placed under the control of the CaMV 35S promoter into binary vectors for Agrobacterium-mediated expression in plant cells (Figure 1, Table 1). By delivering the virus into the plant cell in this manner, one avoids the laborious process of producing viral transcripts in vitro. Different approaches to inoculating plants with Agrobacterium have been developed depending on the plant species and type of experiment. With PVX on N. benthamiana, sufficient infection can occur by picking the Agrobacterium colony with a toothpick and stabbing the toothpick into the seedling leaf (Lu et al., 2003). This is an efficient way to inoculate a large number of plants and is suitable for high-throughput screens. However, in our experience, this approach does not work well with TRV on tomato, and we find it necessary to either syringe-infiltrate, vacuum-infiltrate, or spray seedlings with Agrobacterium in order to achieve good silencing results (Ekengren et al., 2003; Liu et al., 2002a). In the case of bipartite viruses like TRV, co-expression of both viral RNAs is achieved by simply mixing two Agrobacterium strains, each containing one of the RNAs, prior to inoculation. Conditions such as choice of Agrobacterium strain and inoculation technique will have to be optimized as VIGS is developed for other plant species.

Environmental effects on VIGS

Successful gene silencing depends upon a dynamic interplay between virus spread and plant growth, both of which can be influenced by environmental conditions. It is important to note that a substantial amount of effort must be invested initially to determine optimal inoculation levels and plant growth conditions. However, once these factors are optimized for a particular greenhouse or growth chamber it is fairly simple to obtain reproducible levels of silencing between plants and experiments. Temperature is one of the most important factors for good viral spread and effective silencing. In our experience the development of good silencing phenotypes with TRV on tomato occurs at greenhouse temperatures at or below 21°C. With TRV in N. bethamiana, temperatures around 25°C are desirable. In this regard, growth chambers are particularly well-suited for maintaining reproducibility of silencing between VIGS experiments.

Experimental controls for VIGS experiments and assessing efficacy of silencing

It is critical to include appropriate controls in every VIGS experiment to monitor the effectiveness of silencing. A common positive control is silencing of the PDS gene, which results in photobleaching of the silenced regions and is a readily visible phenotype (Figure 2) (Kumagai et al., 1995). Several studies have also used the magnesium chelatase gene as a positive control (Kjemtrup et al., 1998; Peele et al., 2001). To account for phenotypes potentially caused by the virus itself, it is also essential to include plants inoculated with an empty virus as a negative control. Even so, one should consider that the virus could contribute to an observed phenotype. Another desirable internal positive control for VIGS that awaits development would be a sequence from an endogenous gene that is included in every vector to produce a benign phenotype. This would allow the researcher to quickly identify regions of the plant where silencing occurs.

Figure 2.

Some phenotypes identified by VIGS experiments.
(a) Tomato leaf infected with TRV virus only.
(b) PDS gene silenced in tomato using TRV construct. Photobleached tissue indicates regions of silencing.
(c) TRV::Prf silenced tomato leaf treated with an avirulent Pseudomonas syringae strain. Visible disease symptoms are the result of loss of resistance because of suppression of the Prf gene (Ekengren et al., 2003).
(d) Nicotiana benthamiana plant infected with PVX virus only.
(e) A high-throughput VIGS screen of a random cDNA library.
(f) PDS gene silenced in N. benthamiana using a TRV construct.
(g) Silencing of the CTR1 gene in tomato results in a dwarf phenotype with severe epinasty.
(h) Silencing using a tomato PP2Ac gene in N. benthamiana results in the formation of localized regions of cell death. Photograph shows lesions on leaf revealed by trypan blue staining (from He et al., 2004).
(i–l) Abnormal developmental phenotypes identified in large-scale VIGS screen of random cDNAs in N. benthamiana. (i) Silencing of an unknown gene results in elongated internodes and an abnormal leaf shape. (j) Suppression of a gene encoding a putative nuclear shuttle protein produces a plant with shorter stature, multiple branches, and small, ruffled leaves. (k) Silencing of an unknown gene results in a dwarf phenotype. (l) Suppression of a gene encoding a putative clathrin-associated protein results in chlorosis.

The degree and specificity of VIGS is typically determined by reverse transcription followed by polymerase chain reaction (RT-PCR) because this technique requires a minimal amount of tissue and is more sensitive than RNA gel blot analysis for detecting low-abundance transcripts. RT-PCR products amplified from gene-silenced and virus only infected tissue are compared to determine the relative reduction in transcript level (Ekengren et al., 2003; Liu et al., 2004b). The abundance of a RT-PCR product is normalized to the level of a constitutively expressed gene such as actin. To avoid amplifying the gene sequence in the virus, primers for RT-PCR are designed to anneal to a region of the targeted transcript outside of the sequence cloned in the VIGS vector. Subsequent cloning and sequencing of PCR products ensures that the PCR reaction is specifically amplifying the targeted transcripts(s) and not those from closely related genes.

‘Fast-forward’ genetics using VIGS

Improvements to vectors and inoculation techniques have opened the door to large-scale VIGS experiments. With VIGS, the closest approximation to a traditional forward genetics screen is subcloning random cDNA clones into a virus vector and screening for interesting phenotypes. The significant advantage to this approach, termed ‘fast-forward genetics’, is that gene sequences that produce interesting phenotypes can be quickly isolated and identified (Baulcombe, 1999). Large-scale functional genomics screens have been undertaken with TMV- and BSMV-based VIGS vectors in N. benthamiana and barley, respectively (Fitzmaurice et al., 2002). A recent report describes the silencing of nearly 5000 cDNAs in N. benthamiana to assess their functional significance in defense signaling and identified a role for HSP90 genes in disease resistance (Lu et al., 2003). Our laboratories have taken similar approaches and we have identified several candidate genes in the Pto and N disease resistance pathways, including a chloroplast carbonic anhydrase gene that, when silenced, suppresses a Pto-mediated hypersensitive response (Slaymaker et al., 2002; G.B.M and S.P.D.-K., unpublished data). For this forward genetics approach to be successful, it is important to normalize the cDNA library to eliminate high abundance transcripts that would otherwise dilute unique sequences (Lu et al., 2003). Moreover, careful consideration of the source and treatment of tissue used to construct the library is essential in order to enrich for relevant genes.

Advantages of VIGS

The easiest and most effective way to determine the function of a gene or protein is to attenuate the expression of the gene or to generate a mutant that does not encode a functional protein. With the availability of whole genome sequences, there is increasing need for large-scale analysis to determine the functions of thousands of genes. Several approaches have traditionally been used for loss-of-function analysis (Table 2). These techniques have proven very useful but each has inherent disadvantages. VIGS is one example of a new technology that avoids many of the limitations of traditional approaches to functional analysis. As described above, VIGS is most commonly used as an Agrobacterium- or in vitro transcription-based transient assay to rapidly generate phenotypes (Dinesh-Kumar et al., 2003; Ekengren et al., 2003; Liu et al., 2002a,b; Ratcliff et al., 2001). Here we will compare VIGS with other commonly used loss-of-function methods.

Table 2.  Comparison of VIGS with other functional genomics approaches
MethodDescriptionTransformation requiredSpace requirementsCause of loss of functionUsed to study entire familiesCost
  1. VIGS, virus-induced gene silencing; TILLING, targeting induced local lesions in genomes; T-DNA, transfer- DNA; EMS, ethylmethane sulfonate; SNP, single-nucleotide polymorphism; PTGS, post-transcriptional gene silencing; UTR, untranslated region.

VIGSPlants infected with virus carrying fragment of endogenous geneNoSmall for single gene studies
Large for large- scale screen
PTGS of gene homologous to targeting sequenceYesLow; limited sequencing to ascertain specificity;
PCR to confirm silencing
Chemical/ physical mutagenesisSeeds or plants treated with chemical mutagen (e.g. EMS) or radiation (e.g. fast-neutron)NoLargePoint mutations (EMS)
Deletions (fast-neutron)
NoHigh; mapping and sequencing
TILLINGChemical mutagenesis with mutations identified by SNP analysisNoLargePoint mutations (EMS)NoHigh; extensive PCR and sequencing
T-DNA insertionPlants transformed with Agrobacterium have T-DNA inserted into their genomesYesLargeEctopic activation of neighboring genes or disruption of coding sequence or UTRNoModerate; sequencing and PCR
Transposon activationPlants transformed with transposon that is mobilized to produce insertions or excision footprintsYesLargeDisruption of coding sequenceNoModerate; sequencing and PCR

Other functional genomics approaches

The most established technologies used for loss-of-function studies in plants are chemical mutagenesis and the use of transposons or Agrobacterium T-DNA insertions to create disruptions in coding sequences. These technologies have been widely and successfully used and are still the method of choice for the model plant Arabidopsis. There are, however, a few important points to consider that limit the adoption of these approaches in other plant species. First, these techniques require the generation of large populations to screen for mutations in a gene of interest (Bouche and Bouchez, 2001; Parinov and Sundaresan, 2000). Even with new technologies such as Targeting Induced Local Lesions In Genomes (TILLING) (reviewed in Henikoff and Comai, 2003) and the availability of large T-DNA collections, identifying a knockout mutation in a specific gene is not trivial. Second, the creation of T-DNA- or heterologous transposon-mutagenized populations requires the large-scale generation of transgenic plants, a time-consuming process that is limited to a handful of plant species. Third, the presence of large gene families and gene duplications in plant genomes means that many point mutations and insertions do not generate obvious phenotypes (Bouche and Bouchez, 2001). As a result, many mutations are missed and the function of the gene under study remains unknown. Finally, when a mutation is identified in T-DNA or transposon insertion lines, they may not occur in isolation but in the presence of second, or sometimes multiple other insertions. In most instances, this is undesirable as it may complicate the interpretation of the knockout phenotype (Henikoff and Comai, 2003). To isolate the desired mutation away from other unlinked mutations it is often necessary to perform several backcrosses, a process that can be laborious and time-consuming. VIGS, which avoids most of these difficulties, can therefore be seen as a complementary approach to traditional approaches as discussed in the following sections.

VIGS compared with other functional genomic approaches

VIGS is rapid.  A significant advantage of VIGS as a functional genomics tool is that it can identify a loss-of-function phenotype for a specific gene within a single generation. Because the gene of interest is directly targeted in VIGS there is no need for screening large populations to identify a mutation in a specific gene. In a forward genetics VIGS screen, only a single plant is needed to identify a phenotype and an initial result can be rapidly scaled up and repeated. Also, the gene responsible for an interesting phenotype can be quickly sequenced from the VIGS vector and identified.

VIGS avoids plant transformation.  VIGS is a transient method that does not rely on the generation of transgenic plants, a procedure that is difficult in many plant species. As VIGS is carried out in a young plant or mature seedling, loss-of-function phenotypes that would otherwise result in death at early stages of development are often, but not always, avoided. This potentially means that more genes with roles in development can be studied by VIGS.

VIGS overcomes functional redundancy.  The problem of redundancy because of the presence of gene families can also be addressed through the use of VIGS. By using a targeting sequence derived from the most highly conserved region of a gene family, it is possible to silence all or most members of a given family. Conversely, specific members of a gene family can be targeted by selecting unique sequence stretches in family members.

VIGS allows rapid comparisons of gene function between species.  Another advantage of VIGS is that it readily allows the comparison of gene function between different species. The TRV-VIGS vector designed by Liu et al. can be used in both N. benthamiana and tomato to determine gene function (Ekengren et al., 2003; Liu et al., 2002a,b). This vector was recently used to show that certain MAPKs and COI1 function in disease resistance to a bacterial pathogen in tomato as well as to a virus in N. benthamiana (Ekengren et al., 2003; Liu et al., 2004b).

VIGS works in different genetic backgrounds.  VIGS can also be used to rapidly test the function of a gene in multiple genetic backgrounds of a single species. In contrast, most mutant populations have been generated in a single genotype (e.g. Columbia-0 for Arabidopsis) and assessment of a mutant allele in different ecotypes requires multiple backcrosses. This is an important issue because some mutations can produce different phenotypes in the context of different genotypes. The versatility of VIGS in this regard means that a more thorough analysis of gene function can be performed and the potential problems of genotype-specific effects can be avoided.

Other gene silencing methods

Apart from VIGS, other technologies based on PTGS have been employed to generate knock down expression of a gene of interest. The earliest of these methods is the use of antisense RNA (Schuch, 1991). However, this approach requires the generation of stable transgenic lines and is often unreliable. Although the generation of stable transgenic RNAi lines also requires laborious transformation procedures, this approach has marked advantages over antisense technology (Waterhouse et al., 2001). Stable RNAi lines are produced by transforming plants with constructs that generate a hairpin RNA (hpRNA) (Smith et al., 2000). These constructs generate efficient, reliable and reproducible results (Smith et al., 2000; Wesley et al., 2001). In addition, they can be used universally in all plant species as long as they are set in the appropriate transcriptional context. A number of vectors have been designed that allow the high throughput cloning of gene sequences into hpRNA constructs (Helliwell and Waterhouse, 2003; Wesley et al., 2001). Recently, a system for placing PTGS inducing fragments under an inducible promoter has been described (Guo et al., 2003). However, the bottleneck of transformation still remains and hinders this approach from being useful for large-scale screens.

Limitations of VIGS and how they can be addressed

Despite its advantages, certain limitations are inherent in VIGS as a technique for loss-of-function studies. Importantly, VIGS seldom results in the complete suppression of expression of a target gene. Therefore, because it is possible that a decreased transcript level will still be sufficient to produce enough functional protein, a phenotype might not be observed in the silenced plant. As a result, VIGS cannot rule out the involvement of a gene in a particular functional context if a phenotype is not apparent. VIGS can also miss phenotypes that are masked by functional redundancy between gene family members, especially in random screens where virus inserts are not ‘pre-designed’ to target conserved gene sequences. However, this is unlikely to occur in a reverse genetics approach where the availability of genome sequences and large EST collections can aid in the identification of gene families prior to the VIGS experiment.

Another limitation of VIGS is that it often does not result in uniform silencing of the gene throughout an infected plant, and the levels of silencing can vary between plants and experiments. This can complicate the interpretation of results, especially if the silencing does not produce a readily visible phenotype. One solution to this problem could be the development of internal positive controls for VIGS vectors that will ‘mark’ silenced regions with a visible phenotype (see Methodology of VIGS).

The dependence of VIGS upon a pathogen–host interaction also presents several disadvantages. Inoculation of a plant with the virus alone can alter plant development, especially overall height and leaf morphology. This is particularly true for VIGS using PVX and TMV. As a result, it is possible that subtle phenotypes as a result of suppression of a gene could be masked by virus symptoms. PVX and TMV are also excluded from meristematic tissue and therefore may not be effective in assessing the function of genes involved in shoot, leaf, flower, and fruit development. However, the use of TRV has largely overcome these limitations.

Finally, a frequently stated concern is that VIGS (and PTGS approaches in general) might inadvertently result in the suppression of non-target genes, a possibility that is difficult to rule out when working with plant species that lack sequenced genomes. While this criticism might have some merit, it should not overshadow the enormous potential of VIGS for providing rapid insights into gene function. As outlined below, impressive advances have been made in several fields that have utilized VIGS for functional studies. The likelihood of unintentional co-silencing will decrease as genome sequencing data and large EST collections become available for more species. Furthermore, it is important to note that VIGS relies strictly upon regions of exact nucleotide identity and will not target genes simply based on family relationships. Several studies have successfully targeted individual members of gene family for silencing without affecting the transcript levels of the most closely related sequences and therefore have demonstrated that a high degree of specificity can be achieved with VIGS (Ekengren et al., 2003; He et al., 2004; Liu et al., 2004b).

New insights into gene function using VIGS technology

Since its advent 5 years ago, VIGS has become an important tool for the analysis of gene function. Because of its broad applicability, it has been used to silence a wide variety of genes encoding proteins that function in numerous cellular and developmental processes in plants.

Use of VIGS for understanding plant defense against pathogens

Perhaps the most common application of VIGS is in the study of plant defense, where it has been used to study the function of genes already known to be involved in defense and also to identify new genes. For example, a role for Rar1 in disease resistance was first identified in barley through the isolation of mutants (Shirasu et al., 1999). Later, VIGS was used to show that Rar1 plays a similar role in tobacco (Liu et al., 2002b). Similarly, VIGS has been used in N. benthamiana to establish the roles of SGT1 in R gene-mediated resistance to bacteria and viruses (Liu et al., 2002c; Peart et al., 2002b), and in non-host resistance to fungi and bacteria (Bouarab et al., 2002; Peart et al., 2002b). In addition, silencing EDS1 (Liu et al., 2002b; Peart et al., 2002a) and NPR1/NIM1 (Liu et al., 2002b) compromised N-mediated resistance to TMV. VIGS has also been used to study the roles played by several kinases in defense against TMV mediated by the N gene in N. benthamiana. These kinases include NPK1 (Jin et al., 2002), WIPK, SIPK, NtMEK1 and NtMEK2 (Jin et al., 2003; Liu et al., 2003, 2004b; Sharma et al., 2003).

Recently, VIGS was employed to characterize the role of protein phosphatase type 2A (PP2A) catalytic subunits in plant defense (He et al., 2004). A specific tomato PP2A gene, PP2Ac1, was identified by expression profiling to be upregulated in response to pathogen attack. Silencing in N. benthamiana using a tomato PP2Ac sequence resulted in upregulation of defense-related gene expression, the spontaneous formation of localized cell death lesions in leaves and stems, increased pathogen resistance, and accelerated R gene-dependent signaling. Furthermore, only the targeted subset of transcripts from the highly conserved PP2Ac gene family were decreased in these plants.

VIGS has also been used to identify genes in tomato that are required for R gene-mediated resistance to the bacterial pathogen Pseudomonas syringae (Ekengren et al., 2003). This work identified a role in Pto-mediated resistance for the MAPKKs MEK1 and MEK2, the MAPKs NTF6 and WIPK, the defense-related NPR1, and the transcription factors TGA1a and TGA2.2. Some of these genes were previously known to function in defense pathways in other species.

In addition to the candidate gene approach taken by these studies, VIGS has also been used to establish a role for genes previously unknown to function in disease resistance. Plant P58IPK was shown to be required for viral pathogenesis because virus titer is reduced in P58IPK-silenced N. benthamiana plants (Bilgin et al., 2003). Interestingly, silencing of P58IPK leads to host cell death upon infection with viruses. Using a similar strategy, HSP90 was shown to be required for N-mediated resistance to TMV (Liu et al., 2004c; Lu et al., 2003) as well as for the induction of a hypersensitive response (HR) triggered by a non-host pathogen and by the fungal elicitor INF1 (Kanzaki et al., 2003). Other genes with functions in defense that have been studied by VIGS include the COP9 signalosome in N mediated resistance to TMV (Liu et al., 2002c), NbrbohA and B in non-host resistance to Phytophthora infestans and INF1 (Yoshioka et al., 2003), and NtCDPK2 in Cf4 and Cf9-mediated fungal resistance (Romeis et al., 2001).

Use of VIGS for understanding metabolic pathways or plant development

Other areas of plant biology have also been studied using VIGS. Several metabolic pathways have been analyzed by VIGS including plant sterol synthesis (Darnet and Rahier, 2004), the jasmonate-induced production of inhibitors because of insect attack (Saedler and Baldwin, 2004) and the photosystem II in photosynthesis (Saitoh and Terauchi, 2002). Burton et al. (2000) used a PVX-based VIGS vector to silence a cellulose synthase gene (CesA) in N. benthamiana. The Arabidopsis genome sequence has revealed the existence of a large number of genes that bear homology to known CesA genes. VIGS was used instead of traditional knockout approaches to determine the function of CesA-1 homologs in N. benthamiana. They were able to successfully observe a phenotype specifically caused by silencing of CesA-1 although it has 80% nucleotide homology to CesA-2.

VIGS also shows potential as a tool for the study of genes involved in development. The difficulty with obtaining knockouts of genes with developmental roles is obvious in that many mutations result in severe deformities or lethal phenotypes. For instance, meristem genes like NFL (Ratcliff et al., 2001) and floral organ identity genes like DEFICIENS (Liu et al., 2004a), the N. benthamiana orthologs of Arabidopsis LEAFY and AP3, respectively, have been silenced using VIGS. This holds much promise for studying the regulation of plant growth and differentiation that originates in the meristem and floral organs while avoiding the problems of sterility that often accompany mutations in these genes.

Future prospects

VIGS shows much promise as a tool for gene function studies and for high-throughput functional genomics in plants, and it has already begun to fulfill some of this promise. However, there is still a great potential for this approach that remains to be tapped. At present, the most reliable and effective VIGS vectors have a limited host range. They are for the most part functional only in N. benthamiana and other members of the Solanaceae family like tomato (Table 2). The development of vectors with wider host ranges, and in particular those that can facilitate efficient VIGS in Arabidopsis and rice will be very useful to the plant research community. The utility of the existing vectors, especially TRV-based ones, would be greatly improved with the increased availability of gene sequences for plants in which these vectors are functional.

Using VIGS to target a specific gene requires sequence information. This is a limitation when working with N. benthamiana as there is only a relatively small EST database available for this species (http://www.tigr.org). At present, this problem is addressed by using databases of ESTs from closely related species like tomato (Ekengren et al., 2003; Liu et al., 2002a,b, 2004b,c; Lu et al., 2003) and potato (Yoshioka et al., 2003). Excitingly, there has been progress towards the sequencing of the tomato gene-space by an international collaboration (http://www.sgn.cornell.edu/solanaceae-project/). Such a development will address some of the concerns about the specificity of silencing in solanaceous species and make it much easier to identify targets and design primers to carry out VIGS.

VIGS should also be more widely used for large-scale screens to identify interesting phenotypes. In addition, the development of sequenced and arrayed TRV::cDNA libraries would allow a gene-of-interest to be identified in an on-line database and quickly silenced. VIGS could also be used to study the function of all members of a gene class, for example, protein kinases or phosphatases. Thus as VIGS becomes adapted to more plant species and researchers become more familiar with the technique, we expect to see it become a common and widespread tool for gene function studies and functional genomics in plant biology.

Acknowledgements

We thank Dr Peter Moffett, Dr Kerry Pedley, Dr Sophia Ekengren, Robert Abramovitch, Jennifer Miller, Jeffrey Caplan and an anonymous reviewer for helpful comments on the manuscript; Dr Joyce Van Eck, Sanika Kulkarni, Michael Schiff, and Yi Yu Dong for photographs of VIGS phenotypes. This work was supported, in part, by grants from the NSF Plant Genome Program (DBI-0116076, GBM; and DBI-0211872; SPD-K).

Ancillary