Optimized virus-induced gene silencing in Solanum nigrum reveals the defensive function of leucine aminopeptidase against herbivores and the shortcomings of empty vector controls


Author for correspondence:
Ian T. Baldwin
Tel: +49 3641 571101
Fax: +49 3641 571102
Email: Baldwin@ice.mpg.de


  • • Virus-induced gene silencing (VIGS) enables high-throughput analysis of gene function in plants but is not universally applicable and requires optimization for each species. Here a VIGS system is described for Solanum nigrum, a wild relative of tomato and potato and a valuable model species for ecogenomics.
  • • The efficiency of the two most widely used Tobacco rattle virus (TRV) vectors to silence phytoene desaturase (PDS) in S. nigrum was tested. Additionally, the infiltration method and growth temperatures for gene silencing were optimized and the suitability of different control vectors evaluated. Using leucine aminopeptidase (LAP), a herbivore-induced protein, silencing efficiency and the applicability of silenced plants for herbivore feeding assays were assessed.
  • • Vacuum infiltration of seedlings with Agrobacterium carrying the vector, pYL156, proved the most efficient means of silencing genes. Empty-vector controls decreased plant growth but control vectors carrying a piece of noncoding sequence did not. Silencing LAP significantly increased the larval mass of Manduca sexta that fed on silenced plants.
  • • This VIGS protocol proved highly successful for S. nigrum, which should include control vectors carrying noncoding sequence as control treatments. Silencing LAP provided the first experimental evidence that LAP has a defensive function against herbivores.


Recently virus-induced gene silencing (VIGS) has been established as an important tool in plant sciences with which to investigate gene function. Plants are infected with genetically modified viruses that carry a short sequence of a target gene in their genome. When the virus replicates, dsRNA is formed that triggers posttranscriptional gene silencing in the plant (for reviews describing the mechanism see Lu et al., 2003; Burch-Smith et al., 2004). Consequently, the plant defends itself against the virus by degrading the accumulating viral genome and in the process, silences its own gene. Inserting plant gene fragments into the viral vectors follows standard protocols and infecting a large number of plants for experiments requires much less time than any other functional genetic approach. In addition to its speed, VIGS has distinct advantages with plant species that are difficult to transform or genes that are essential for plant survival. Helped by the availability of steadily growing plant sequence databases, VIGS is emerging as the method of choice for rapid functional genomic screens, especially in ‘nonmodel organisms’ where complete genome sequence information and mutant collections are not available. These advantages drove the development of many viral vectors and VIGS-protocols for several plant species (for reviews see Burch-Smith et al., 2004; Robertson, 2004), including various Nicotiana and Solanum species, Arabidopsis thaliana, Aquilegia vulgaris, Papaver somniferum, Hordeum vulgare, among others (Holzberg et al., 2002; Liu et al., 2002a; Brigneti et al., 2004; Hileman et al., 2005; Burch-Smith et al., 2006; Gould & Kramer, 2007). Since the ground-breaking work of David C. Baulcombe's group (Ratcliff et al., 2001) and additional refinements by Liu et al. (2002b), the Tobacco rattle virus (TRV)-based vectors have emerged as the most effective systems for VIGS in solanaceous species.

Our group uses Solanum nigrum (black nightshade, Solanaceae) as an ecological model species (Schmidt et al., 2004). Its phylogenetic proximity to tomato and potato allows us to use genetic tools and databases that have been established for these crops. Moreover, S. nigrum is a wild species that has not been under artificial selection for particular traits (e.g. that enhance yield). This makes the plant particularly interesting as an ecogenetic model system because its traits can be studied in the context of its natural habitat and the selection pressures that have shaped the plant's genome (Kant & Baldwin, 2007). A report by Brigneti et al. (2004) demonstrated that TRV-based VIGS was in general applicable to the genus Solanum, albeit with different efficiencies depending on the species, conditions, and vector used. However, upon closer examination, the species used in this study was likely not S. nigrum s.str. but rather another diploid and outcrossing Solanum species (Solanum nigrescens) and in preliminary studies, the procedure was found not to work in our genotype, S. nigrum ssp. nigrum Sn30. The section Solanum, centering on S. nigrum L., represents a species complex that is taxonomically challenging. Its complexity has various causes such as high phenotypic plasticity, high intraspecific genetic variation, polyploidy and natural hybridization (Edmonds & Chweya, 1997). Thus it was necessary to reassess the taxonomic classification of the genotype used by Brigneti et al. (2004) and to find and optimize a procedure for VIGS in the S. nigrum genotype Sn30 (in the following referred to as S. nigrum).

In this first VIGS protocol for S. nigrum, we evaluate different inoculation procedures and the influence of temperature on silencing efficiency; compare the two available TRV-vectors from Ratcliff et al. (2001) and Liu et al. (2002b) and develop an appropriate control vector; and assess silencing efficiency and persistence of silencing for a highly expressed gene on transcript and protein levels.

We used phytoene-desaturase (PDS) for all optimization experiments, and selected leucine aminopeptidase (LAP) as target gene for a final proof of principle. Leucine aminopeptidase-A is a well-described enzyme that accumulates in tomato leaves after wounding, herbivore challenge or various elicitor treatments (Matsui et al., 2006). Moreover, an LAP clone also ranks among the highest upregulated genes in S. nigrum after attack by larvae of Manduca sexta or treatment with methyl jasmonate, a potent elicitor of herbivore-induced responses (Schmidt et al., 2005). Although plant LAPs have been intensively studied and characterized especially in tomato (Solanum lycopersicum), their function and role in defense are still not completely resolved (Walling, 2006). Walling (2006) postulated a regulatory role for LAP in the wound signaling cascade; others have hypothesized that LAP has a direct antinutritive effect in insect guts (Chen et al., 2005; Matsui et al., 2006). Its abundance after herbivory and the availability of an activity assay made LAP an ideal target for demonstrating how effectively the VIGS system can be used to investigate the function of defense-related genes in S. nigrum.

To assess the possible defensive function of a gene, assays with suitable target organisms are required. The tobacco hornworm M. sexta (Sphingidae) is a solanaceous specialist and is commonly used in herbivore assays with tomato and tobacco as well as with S. nigrum (Schmidt & Baldwin, 2006). We used M. sexta larvae in a cut-leaf assay and present data which clearly demonstrate that LAP influences herbivore performance in S. nigrum.

Materials and Methods

Plant material and growth

All experiments were done with the S. nigrum L. inbred line Sn30 (Schmidt et al., 2004). Additionally we grew the Solanum sp. (genotype CGN21367, Center for Genetic Resources, Wageningen, The Netherlands) that was used in the Brigneti et al. (2004) study for a taxonomic comparison.

Seeds were germinated on Gamborg B5 plant agar in Petri dishes, as described by Schmidt et al. (2004), using a 1 m KNO3 solution for overnight incubation instead of 3.5 mm Ca(NO3)2·4H2O. The seedlings were used for vacuum infiltration 7–10 d after sowing. Vacuum-infiltrated seedlings and seedlings that were later used for sap inoculation or syringe infiltration were planted individually in 9 × 9 × 9.5 cm pots filled with a peat-based substrate (Tonsubstrat; Klasmann, Geeste-Groß Hesepe, Germany) and covered with transparent plastic hoods. The plants were grown in a climate chamber (16 h light : 8 h dark, 20°C light : 17°C dark; c. 85% humidity) and kept for 2 d under indirect light; subsequently, the direct lights were switched on and the plastic hoods removed. To compare VIGS performance at different temperatures, the plants were transferred after infiltration to climate chambers maintained at 17, 21, and 26°C, with the same light and humidity conditions as already described.

Isolating SnPDS, SnPDSi and SnLAP-N and silencing construct generation

Primers were designed from tomato sequences (see the Supplementary Material, Table S1) to obtain a LAP-N fragment (SnLAP-N), an exonic and an intronic PDS fragment (SnPDS and SnPDSi) by standard polymerase chain reaction (PCR) on cDNA and gDNA from S. nigrum (NCBI Genbank accessions: SnLAP-N EU262259; SnPDS EU434622; SnPDSi EU434623). Based on these sequences, antisense VIGS-constructs of approx. 300 bp were created using a forward primer containing an XhoI restriction site and a reverse primer with an EcoRI restriction site (Table S1). The XhoI–EcoRI fragments obtained were ligated into the vector pYL156 (described in Liu et al., 2002b), resulting in derivatives called pTRV-SnLAP-N, pTRV-SnPDS and pTRV-SnCV (containing SnPDSi), respectively. The SnPDS-fragment was also ligated into the binary vector pTV00 (Ratcliff et al., 2001) to form pTV-SnPDS. Ultimately, the vectors were cloned into Agrobacterium tumefaciens strain GV3101.

Agrobacterium cultivation and infiltration methods

Five milliliters of LB (Luria–Bertani)–MES (2-(N-morpholino) ethanesulfonic acid) medium (20 g LB broth l−1, 1.95 g MES l−1 (Carl-Roth GmbH, Karlsruhe, Germany); 20 µm acetosyringone (Sigma-Aldrich, Taufkirchen, Germany) complemented with 50 mg l−1 kanamycin were inoculated with the respective Agrobacterium strain from glycerol stock and incubated overnight at 28°C and 200 rpm. The next evening, 200 ml LB–MES medium was inoculated with 300 µl from the preculture and grown overnight under the same conditions. After reaching an optical density at 600 nm (OD600) of c. 0.6, cultures were harvested in a 50 ml tube by centrifugation at 2000 g and 4°C for 10 min. The pellet was resuspended in 20 ml infiltration medium (10 mm MgCl2, 10 mm MES, 200 µm acetosyringone) and kept in the dark for 3 h at room temperature. Each suspension containing a construct of interest (derivatives of pYL156 or pTV00) was mixed with the same volume of a culture containing TRV RNA1 (pYL196 or pBINTRA6).

For vacuum infiltration 7- to 10-d-old seedlings were transferred from the Petri dishes to the bacteria solution (up to 30 seedlings in a total volume of 40 ml). Vacuum was applied in a desiccator until it reached a pressure of 60–70 mbar and then slowly released. Unlike in species such as A. thaliana and tomato (Ekengren et al., 2003; Wang et al., 2006), in S. nigrum it is important to release the vacuum slowly, as initial experiments led to high levels of necrosis and mortality when seedlings were repressurized too rapidly (data not shown). Sap inoculation was performed as described by Brigneti et al. (2004), using 14-d-old seedlings for the inoculation. Syringe infiltration was done by directly injecting the Agrobacterium solution with a needle-less syringe into the first three leaves (c. 21-d-old seedlings).

Monitoring bleaching and plant heights

The PDS silencing efficiency of the vectors pTRV-SnPDS and pTV-SnPDS was assessed by determining the ratio of plants showing signs of bleaching at 10, 15, 20 and 25 d post inoculation (dpi) and documented by photographing all plants. Because of its higher silencing efficiency, the pTRV-vector was used for all following experiments.

The height of plants treated with pTRV-SnLAP-N, pTRV-SnPDS, pTRV-SnCV or pYL156 (empty vector) was measured as the distance from the cotyledonary node to the apex 15, 20, and 25 dpi.

Plant treatments

At 21 dpi 75 µg methyl jasmonate (MeJA) in 10 µl lanolin was applied to the stem between the first and second node above the cotyledons. Three days later the fully expanded leaf at node three was harvested and flash-frozen in liquid nitrogen to determine silencing efficiency and LAP activity.

To determine the abundance of viral RNA, whole shoots were harvested and flash-frozen in liquid nitrogen, and the RNA amount of TRV RNA1 and RNA2 quantified by real-time PCR (QRT-PCR).

Quantitative real-time PCR

To analyse transcript abundance after silencing and to gauge the virus load, QRT-PCR was used with S. nigrum elongation-factor 1α (SnEF1α; NCBI Genbank accession number: AY574951) as internal control and the comparative 2−ΔΔCt method for relative quantification (Livak & Schmittgen, 2001). Total RNA was extracted from flash-frozen and ground plant material using a modified TRI reagent protocol (The Institute for Genomic Research, Online Protocol, 2003; URL: http://www.tigr.org/tdb/potato/images/SGED_SOP_3.1.1.pdf). Reverse transcription of 500 ng of total RNA was performed using SuperScript II RNaseH-Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) either with a poly-T primer to obtain cDNA from plant mRNA or with random hexamer primers to additionally transcribe viral RNA. Twenty nanograms of cDNA were applied to 20 µl SYBR green reactions (qPCR Core Kit for SYBR Green I; Eurogentec, Saraing, Belgium), which were run on an ABI PRISM 7700 sequence detection system (Applied Biosystems, Foster City, CA, USA; cycler conditions: 10 min at 95°C, 40 cycles of 30 s at 95°C and 30 s at 60°C). Primers were designed using primer 3 v.0.4.0 (Rozen & Skaletsky, 2000), checked for specificity using a melting curve analysis (ABI PRISM 7700 Dissociation Curve Software) and are summarized in the Supplementary Material, Table S1.

Native polyacrylamide gel electrophoresis (PAGE) and in-gel LAP activity assay

Protein extraction and in-gel LAP activity assay were performed according to Gu et al. (1996) with minor changes. A total of 600 µl of glycerol-free extraction buffer was added to 200 mg of flash-frozen and ground leaf material, vortexed for 5 min and centrifuged for 20 min at 12 000 g. Protein concentration of the supernatant was determined by Bradford reaction and 160 µg of soluble protein was loaded to each lane of a 7.5% native polyacrylamide gel. The gel was run for 4 h at 18°C and 30 mA and then incubated in 100 ml of staining solution containing 50 mm sodium phosphate (pH 6.0), 30 mg of l-Leu-β-naphtylamide (Sigma, St Louis, MO, USA; dissolved in 1 ml DMSO), 100 mg FastBlue salt (Fluka) and 0.5 mm MnCl2. Leucine aminopeptidase from porcine kidney (Sigma) served as positive control.

Caterpillar performance assay

Two days after treatment with MeJA, the leaf at node three above the cotyledons was excised and transferred to a water-filled tray in the glasshouse (16 h light at 26–28°C, 8 h dark at 22–24°C; supplemental light from Master Sun-T PAI Agro 400 or Master Sun-T PIA Plus 600W Na-lamps (Philips, Turnhout, Belgium), 45–55% humidity). The following day, freshly hatched M. sexta neonates were placed on the leaves; these leaves were replaced every other day with the next, similarly pretreated, leaf from the same plant. At days 4, 7 and 11, larval mass was recorded. The experiment was started with 10 larvae for each of the vectors B-EV and B-CV and 20 larvae for B-LAP, one larva per leaf. After day 7 the larvae's rapid consumption of leaf material required us to halve the number of caterpillars by removing both the lightest and the heaviest individuals in each group. To investigate the stability of silencing under glasshouse conditions, the residual leaves were flash-frozen when leaves were exchanged on day 4 and checked for LAP transcripts as described earlier.

Statistical analysis

The software package SPSS 15.0 for Windows (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses. Data in Fig. 2 were analysed by univariate anova followed by a Scheffépost-hoc test. The caterpillar mass data in Fig. 4a were not normally distributed and thus analysed with a Kruskal–Wallis test for each time point, followed by three Bonferroni-corrected Mann–Whitney U tests to determine which groups differed significantly.

Figure 2.

Solanum nigrum plant height and abundance of viral RNA after treatment with different silencing vectors. (a) Mean (± SE) heights of LAP-silenced plants (B-LAP) and of plants treated with two different control vectors: empty vector B2 (B-EV), and vector B2 with an insert of intronic PDS sequence (B-CV). Untreated wild-type (WT) and wild-type plants infiltrated with untransformed Agrobacterium tumefaciens (Agro) served as additional controls. Height was measured as the distance from the cotyledonary node to the apical meristem 20 d post inoculation (dpi). Different letters indicate a significant difference (univariate anova, F5, n = 11.048, P < 0.001, followed by a Scheffépost-hoc test, P < 0.05; B-EV, B-CV, B-LAP: n = 30; Agro, WT: n = 5). (b) Mean (± SE) relative abundance of TRV RNA1 and RNA2 in shoots of plants treated with B-EV (closed bars), B-CV (tinted bars), or B-LAP (open bars) 24 dpi. Different letters indicate a significant difference (univariate anova; RNA1: F2,11 = 1.908, P = 0.165; RNA2: F2,11 = 14.999, P < 0.001, followed by a Scheffépost-hoc test, P < 0.05). The experiments were conducted twice with similar results.

Figure 4.

Manduca sexta performance on LAP-silenced plants and LAP-transcript abundance in detached leaves after herbivory. (a) Mean (± SE) mass of individual M. sexta larvae reared on LAP-silenced (B-LAP open bars) and control plants (B-EV, closed bars; B-CV, tinted bars), measured 4, 7 and 11 d after hatching. Different letters indicate a significant difference of P < 0.01 (computed for each day individually: Kruskal–Wallis, P < 0.01, followed by Bonferroni-corrected Mann–Whitney U, P < 0.01, for n see the Materials and Methods section). (b) Mean (± SE) relative LAP mRNA levels in methyl jasmonate-treated and detached leaves of LAP-silenced (B-LAP; n = 20) and control plants (B-EV and B-CV; n = 10) after 2 d of caterpillar feeding.


Identification of the Solanum sp. genotype CGN21367

Genotypes Sn30 and CGN21367 – the latter was used by Brigneti et al. (2004) – differed strikingly in their growth habits. Additionally, flow cytometry indicated different ploidy levels of the two genotypes: Sn30 is hexaploid, as Schmidt et al. (2004) also note, but CGN21367 is diploid (unpublished). Finally, CGN21367 originates from Costa Rica, where S. nigrum L. only rarely occurs, and a taxonomic examination revealed that this genotype represents S. nigrescens (S. Knapp, pers. comm.) we propose that this genotype represents another Solanum species, which could also explain its different response to the described VIGS method.

Optimization of plant stage, inoculation method and temperature for VIGS

In initial experiments, the influence of plant stage, inoculation technique and temperature on silencing efficiency were assessed. We used the TRV-vector developed by Liu et al. (2002b), harboring a fragment of PDS isolated from S. nigrum, and observed the amount of photobleaching after PDS silencing. The common practice of infiltrating three to four leaf-stage seedlings with a syringe proved inefficient. Plants showed no or very late silencing (see the Supplementary Material, Fig. S1a). Vacuum infiltration of 1-wk-old seedlings (Hileman et al., 2005) and sap inoculation of 2-wk-old seedlings (Brigneti et al., 2004) both greatly increased silencing efficiency (Fig. S1b,c). Ultimately we experimented with vacuum infiltration; because it allows genes to be silenced earlier in their development, vacuum infiltration seems to silence slightly more efficiently and it does not require the additional Nicotiana benthamiana plants needed for virus culture, as in the case of sap inoculation. Vacuum-infiltrated seedlings showed bleaching within 10 dpi and a survival rate of c. 97%, despite the early inoculation.

Temperature is an important variable that influences several aspects in VIGS, such as Agrobacterium performance, viral spread, accumulation of short-interfering RNAs and eventually plant growth (Burch-Smith et al., 2004; Robertson, 2004; Wang et al., 2006). The PDS-silenced plants grown at 17, 21 and 26°C were compared and similar growth and silencing were found at the lower temperatures, while plants at 26°C displayed faster growth but heterogeneous photobleaching (see the Supplementary Material, Fig. S2). All subsequent experiments were carried out at 20°C, which is within the natural temperature range of S. nigrum growth.

Comparing two TRV vectors by silencing PDS

Brigneti et al. (2004) have shown that the TRV vectors developed by Ratcliff et al. (2001) and Liu et al. (2002b) have different silencing efficiencies, depending on plant species. We compared the performance of the two groups of vectors in Solanum nigrum in a similar manner as Brigneti et al. (2004) by silencing PDS. For reasons of simplicity we will refer to the TRV vectors of Ratcliff et al. (2001) as A and to the vectors of Liu et al. (2002b) as B. The TRV genome consists of two parts, RNA1 and RNA2, and therefore we refer to the two vectors developed by each group of authors as ‘A1’ and ‘A2’, ‘B1’ and ‘B2’, respectively. When Brigneti et al. (2004) compared the efficiency of all four possible vector combinations, the mixtures A1B2 and A2B1 never exceeded the performance of A1A2 or B1B2. As a consequence, we compared only A1A2 to B1B2 in our experiments. Vector B-PDS leads to an earlier silencing with a higher frequency (100%) when compared with A-PDS (c. 70%; Fig. 1a). Moreover, silencing spreads more uniformly and to a higher extent with vector B-PDS (Fig. 1b). Interestingly, Brigneti et al. (2004) did not observe such a dissimilar performance of the two vectors in Solanum sp. CGN21367. To ensure that this difference was not caused by changing the infiltration method, we repeated the experiment using sap infiltration, with a very similar result (Fig. S1d). Thus, vector B seems to be more suited for VIGS in Solanum nigrum and was used in all subsequent experiments.

Figure 1.

PDS silencing efficiency of two different vectors. (a) Percentage of plants showing photobleaching after treatment with either the vector A-PDS (open bars; Ratcliff et al., 2001) or B-PDS (tinted bars; Liu et al., 2002b) 10, 15, 20 and 25 d post inoculation (dpi). Ten to 11 plants per vector were used and the experiment was conducted twice with similar results. (b) Representative examples of Solanum nigrum treated with vectors A-PDS, B-PDS, or the control vector B-CV, 15, 25 and 55 dpi. Bars: 1 cm, 15 and 25 dpi; 5 cm, 55 dpi.

The appropriate control: empty vectors vs control vectors

In TRV-VIGS experiments, the control usually consists of plants that are treated with a combination of RNA1 and RNA2, with the latter containing no insert at the multiple cloning sites where usually a gene fragment is placed. Surprisingly, we observed in our experiments that such empty-vector (EV) plants were slightly smaller than plants treated with vectors harboring gene fragments for silencing. To our knowledge, such an effect has not been described before. To control for differences that could result from differences in vector size, we cloned a unique intronic sequence of PDS (SnPDSi) into the empty vector B2, resulting in a new control vector (B-CV). As expected, the intronic insert did not silence PDS (data not shown). We compared plants treated with B-EV or B-CV with plants silenced in the expression of leucine aminopeptidase (B-LAP) because in a study with stably silenced tomato lines, LAP was shown not to affect plant growth (Pautot et al., 2001). At 20 dpi and before any additional treatment, B-EV plants were significantly smaller than plants treated with the vectors B-CV or B-LAP (Fig. 2a). Plants that were infiltrated with untransformed A. tumefaciens (Agro) resembled noninfiltrated wild type plants: both grew taller than virus-infected plants. Silencing target genes other than LAP in additional experiments confirmed these results (data not shown).

The observed growth difference between the empty-vector B-EV and the control-vector B-CV could be a result of unequal viral loads, which in turn might result from differences in viral replication efficiency caused by the insertion. We tested this hypothesis by measuring the abundance of viral RNA1 and RNA2 in the shoot. The RNA2 levels were significantly higher in B-EV plants than in B-CV and B-LAP plants (Fig. 2b). No significant difference was apparent for RNA1.

Silencing LAP-N: transcript levels, enzyme activity and herbivore performance

Plants were pretreated with methyl jasmonate (MeJA) to strongly induce the expression of LAPs. After 3 d leaf material was harvested and the LAP transcript abundance and enzyme activity were determined. In LAP-silenced plants LAP mRNA levels were reduced to c. 5% of the empty vector level (Fig. 3a). In plants treated with B-CV, levels were slightly higher than in B-EV but not to a statistically significant degree.

Figure 3.

Silencing leucine aminopeptidase (LAP): transcript abundance and enzyme activity. (a) Mean (± SE) relative LAP mRNA levels in leaves of LAP-silenced (B-LAP) and control plants (B-EV and B-CV) 24 d post inoculation (dpi) and 3 d after methyl jasmonate treatment (B-EV, B-CV: n = 5; B-LAP: n = 10). (b) In-gel LAP activity after native polyacrylamide gel electrophoresis. Crude extracts from three biological replicates of the same samples as in (a) were loaded; LAP from porcine kidney served as a positive control. Activity was visualized as described in the Materials and Methods section.

It is known that LAP occurs in several isoforms in tomato (Gu et al., 1996; Tu et al., 2003). To gain a better understanding of the number of active and silenced LAPs, we compared protein extracts with an in-gel activity assay after native PAGE (Fig. 3b). The expression of the most active LAP was silenced. The two faint bands might indicate other low-level-expressed LAPs.

In a feeding assay with M. sexta larvae, we investigated the impact of LAP on a herbivore. As M. sexta develops very slowly at the optimal temperature for VIGS (20°C), we performed the assay with excised leaves supplied with water at c. 26°C in the glasshouse. The plants were pretreated with MeJA 3 d in advance to induce high levels of LAP. After 11 d of feeding, the caterpillars that fed on LAP-silenced plants were nearly three times heavier than those that fed on controls (Fig. 4a). Gene silencing was stable in leaves that had been excised, transferred to higher temperatures and maintained in water and challenged by caterpillars for 2 d (Fig. 4b).


With a growing number of plant species under investigation, tools for rapid reverse genetic approaches are indispensable. Solanum nigrum has proven to be a valuable model system in molecular ecology and large-scale transcriptional profiling has identified numerous genes involved in plant–plant and plant–insect interactions (Schmidt et al., 2005). However, gene function analysis was hampered because until now it has required the time-consuming process of stable plant transformation. Because VIGS allows for rapid functional screens, a targeted selection of candidate genes for stable transformation is possible, as is the opportunity to transiently manipulate genes that are essential for plant survival. Here we describe the development of a VIGS method for S. nigrum in the context of what is known from other systems. Finally, we validate the use of VIGS by silencing LAP and demonstrate that this gene plays a central role in the antiherbivore defense.

Optimizing VIGS for S. nigrum

Several studies (Brigneti et al., 2004; Hileman et al., 2005; Gould & Kramer, 2007; Senthil-Kumar et al., 2007) have shown that TRV–VIGS is applicable to various plant species beyond the Solanaceae. Nevertheless, because the silencing efficiency greatly varies among even closely related species, additional fine-tuning of the method may be required. This fact is reinforced by our initial experiments. When we tried to adopt a VIGS method published for a species of the S. nigrum complex, our S. nigrum accession resisted all silencing attempts. Our study demonstrates how the TRV–VIGS method can be adapted for S. nigrum and highlights the important parameters that most likely have to be considered when applying the method to other plant species.

The mechanisms underlying VIGS depend on the interaction of TRV and, in most cases, Agrobacterium with the host on multiple levels. Plant–virus and plant–microbe interactions are highly dynamic and thus differences in the degree of susceptibility between closely related plant species may evolve rapidly. Hence, a VIGS system may have difficulty efficiently infecting and propagating the TRV. Wang et al. (2006) defined the optimal conditions for Agrobacterium culture when VIGS is to be used with A. thaliana (e.g. bacterial density, acetosyringone concentration, preincubation time, etc.) and these are likely generally applicable to other plant systems and have been integrated into this protocol. In addition, most authors emphasize the choice of vector, the inoculation method, plant age, and plant growth temperature as the key parameters (Brigneti et al., 2004; Saedler & Baldwin, 2004; Burch-Smith et al., 2006; Wang et al., 2006).

Vector choice is crucial for successful silencing. Our experiments indicate that vector B silences plants more quickly and efficiently than does vector A in S. nigrum. It seems as if the modifications in vector B, a double CaMV 35S promoter and a self-cleaving ribozyme at the 3′-end, lead to improved virus propagation and silencing, without increasing the severity of the disease symptoms. However, whether these changes are advantageous for silencing or not might depend on the host species: in Solanum tuberosum they seem to have a negative effect (Brigneti et al., 2004) and in Nicotiana attenuata VIGS works highly efficiently when vector A is used (Saedler & Baldwin, 2004; Wu et al., 2007).

Different inoculation techniques vary greatly in their efficiency. In S. nigrum, photobleaching could not be observed when vector A-PDS was syringe infiltrated. Only when vacuum infiltration and sap inoculation were used did silencing symptoms become clearly visible. This might be the result of higher transformation or transfection rates of these methods, but their main advantage is probably that they allow much younger plants to be treated. It appears that VIGS works best on young plants (compare also Burch-Smith et al., 2006) and vacuum infiltration provides the means to inoculate young seedlings. Sap inoculation requires larger leaves, which delays the silencing, and its advantage of bypassing the Agrobacterium-mediated transformation step seems to be negligible if we compare the growth of Agrobacterium-inoculated plants and untreated wild-type plants. Vacuum infiltration allows a large number of seedlings to be treated with several constructs at one time for large-scale screening, thus obviating the need to also grow N. benthamiana, as sap inoculation requires. The infiltration of seedlings grown on Petri dishes or filter paper also facilitates the procedure when compared with an upside-down infiltration of soil-grown plants, as is commonly done with tomato (Ekengren et al., 2003; van Schie et al., 2007). The only other method that allows young seedlings to be treated is the so called ‘Agrodrench’ procedure (Ryu et al., 2004). However, we found this much less effective than vacuum infiltration (unpublished) and it required larger volumes of bacterial suspension. In conclusion, vacuum infiltration of seedlings is our method of choice for plants that require very early inoculation or for studying gene function early in plant development.

In order to obtain and maintain a high degree of silencing, temperature seems to be an important but species-specific factor. Several Solanum spp. favor temperatures between 16 and 20°C (Liu et al., 2002a; Brigneti et al., 2004), whereas A. thaliana and N. benthamiana show better results at higher temperatures (23°C and 25°C, respectively). Thus appropriate tests are advisable when optimizing VIGS for other plants species.

A new standard for control vectors

Empty vectors (EVs) are commonly used for control treatments in VIGS experiments. However, our results indicate that inoculating plants with EV constructs compared with vectors that carry a gene fragment negatively affects plant growth. This effect became especially apparent when seedlings rather than older plants were used for infiltration. The increase in viral genome size effected by inserting a gene fragment may reduce the replication efficiency of the virus and thus reduce the stress on the plant, as has been described for the Sendai virus in mammals (Sakai et al., 1999). Consistent with this hypothesis, we found RNA2, the TRV genome that carries the insertion, to be more abundant in EV plants, whereas RNA1 levels were only slightly increased. Thus a short insert of approx. 300 bp is sufficient to reduce virus replication efficiency. In addition, the insertion could negatively affect the performance of the virus in other ways (e.g. by hampering encapsidation), and more research is necessary to explain the effect in detail. However, if the difference in virus genome size affects plant growth, we hypothesize that many other processes may be affected and conclusions drawn from such experiments may be confounded. As a consequence, we suggest the use of a CV instead of an EV in VIGS experiments, especially when vacuum infiltration is used with seedlings.

A defensive role for LAP in response to herbivores

The LAPs belong to the metallopeptidases and hydrolyse residues at the N-terminus of peptides or proteins (for a recent review see Matsui et al., 2006). They constitute a diverse group of proteins occurring in bacteria, plants, and mammals. In plants, two classes of LAPs are recognized based on their isoelectric point (pI): LAP-N (neutral) and LAP-A (acidic). The former seems to be ubiquitous for angiosperms, being constitutively expressed in most tissues, whereas LAP-A is probably restricted to a group within the Solanaceae to which S. nigrum also belongs (Chao et al., 2000) and accumulates in response to wounding, herbivory, pathogen attack, salt stress and water deficiency (Matsui et al., 2006). The cDNA clone SnLAP-N that we isolated from S. nigrum shows the highest similarity to LapN from tomato. However, when a fragment of SnLAP-N was used for VIGS, MeJA-induced LAP activity was silenced, which in tomato is usually caused by LAP-A. Two additional faint bands on the activity gel could indicate activity of constitutive LAPs, which were obviously not affected by VIGS. Since LapA and LapN have a high sequence similarity in tomato, it is possible that transcripts of both constitutive and induced LAPs were silenced and measured with the construct and primers based on SnLAP-N. Although further work is necessary to understand if the expression patterns of LAP-A and LAP-N found in tomato are different from those in S. nigrum, LAP exemplified the potential of VIGS. As we demonstrated on the mRNA and the protein levels, the efficiency of the silencing method used on a highly expressed gene such as LAP was similar to the efficiency of stable silencing methods using antisense or inverted-repeat constructs.

The function of plant LAPs is still under discussion and it is likely that the different classes play various roles. Although LAPs were initially classified according to their proteolytic properties, they may act as transcriptional regulators, as has been suggested for microbial LAPs (Matsui et al., 2006). Given that LAP activity increases after herbivory (Gu et al., 1996; Chao et al., 1999) and that it is highly abundant and stable in the insect midgut (Chen et al., 2005), two explanations for the defensive function of LAPs are possible. (1) Leucine aminopeptidase serves as a direct defense in the insect midgut either by damaging it, as shown for a cysteine protease from maize (Pechan et al., 2002), or by reducing the availability of free amino acids through liberation of arginine from peptides and its subsequent degradation by threonine deaminase (Chen et al., 2005). (2) Leucine aminopeptidase acts as a transcriptional or post translational regulator of the jasmonic-acid-mediated signaling cascade and thus influences the overall defense response of the plant, as suggested by Walling (2006) and Matsui et al. (2006). In the same review Walling (2006) announced that evidence supporting both hypotheses will soon be published in addition to data on improved herbivore performance on LAP-A-silenced tomato plants. Interestingly in a former study on LAP-A-silenced plants, such a phenotype could not be found (Pautot et al., 2001). Silencing LAP in S. nigrum may lead generally to a different response than in tomato. However, we propose that the different result in our experiment is rather caused by the pretreatment with MeJA that strongly upregulated LAP activity before the larvae fed on the plant material. This could have amplified the response of the caterpillars and revealed the defensive function of LAP. The appearance of a strong phenotype after treatment with MeJA suggests that a LAP acts downstream of JA in the signaling cascade, although the possibility of another or the same LAP-isoform also acting upstream cannot be excluded. Publicly available microarray data of LAP-A-silenced tomato plants (Walling, TIGR Solanaceae Genomics Resource: http://www.tigr.org/tdb/sol/sol_expression.shtml) point to a regulatory role of LAP late in the JA signaling cascade. By contrast, the relatively slow but steady increase in LAP activity after wounding and herbivore attack and the findings of Chen et al. (2005) concerning stability and abundance in the insect gut make a direct defensive function plausible. To conclude, plant LAPs seem able to contribute to plant defense on multiple levels and are likely not restricted to a single function.


We have presented the first protocol for VIGS in S. nigrum that results in efficient silencing. The methodological modifications address common problems of VIGS in nonmodel species and thus we hope that this work will facilitate the future adoption of the procedure for other species. Moreover, we have demonstrated that EVs can have unintended effects; replacing them with CVs should be considered. Rapid screens for gene function using VIGS can be useful to reduce the number of candidate genes for more detailed studies. We demonstrate the value of the procedure by silencing a highly expressed gene (LAP) and demonstrate its function in an herbivore performance assay. Although the ultimate functions of LAPs in plants remain to be determined, here we demonstrate for the first time the considerable impact of LAPs on herbivores.


We thank D. C. Baulcombe and S. P. Dinesh-Kumar for providing the TRV VIGS vectors, CGN Wageningen for providing seeds of S. nigrescens. CGN21367, Chris van Schie and Jan W. Kellmann for helpful discussions on VIGS methodology and plant viruses, Sandra Knapp for her invaluable efforts of bringing light into Solanum taxonomy and identifying the S. nigrescens genotype, Tamara Krügel for carefully culturing and hand-pollinating S. nigrescens and for making herbarium specimens, Antje Wissgott and Klaus Gase for design and generation of the VIGS constructs, Emily Wheeler for editorial assistance, and the Max Planck Society for financial support.