Hsp90 Gene Is Required for Mi-1-Mediated Resistance of Tomato to the Whitefly Bemisia tabaci

The Mi-1 gene of tomato (Solanum lycopersicum) confers resistance against some nematodes and insects, but the resistance mechanisms differ depending on the harmful organism, as a hypersensitive reaction (HR) occurs only in the case of nematodes. The gene Rme1 is required for Mi-1-mediated resistance to nematodes, aphids, and whiteflies, and several additional proteins also play a role in this resistance. Among them, the involvement of the chaperone HSP90 has been demonstrated in Mi-1-mediated resistance for aphids and nematodes, but not for whiteflies. In this work, we studied the implication of the Hsp90 gene in the Mi-1 resistance against the whitefly Bemisia tabaci by means of Tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS). The silencing of the Hsp90 gene in tomato Motelle plants carrying the Mi-1 gene resulted in a decrease in resistance to whiteflies, as oviposition values were significantly higher than those on non-silenced plants. This decrease in resistance was equivalent to that caused by the silencing of the Mi-1 gene itself. Infiltration with the control TRV vector did not alter Mi-1 mediated resistance to B. tabaci. Similar to the Mi-1 gene, silencing of Hsp90-1 occurs partially, as silenced plants showed a significant but not complete suppression of gene expression. Thus, our results demonstrate the requirement of Hsp90 in the Mi-1-mediated resistance to B. tabaci and reinforce the hypothesis of a common model for this resistance to nematodes and insects.


Introduction
Bemisia tabaci (Hemiptera: Aleyrodoidea) is one of the most damaging agricultural pest species worldwide, with a high polyphagia, virus transmission capacity, and insecticide resistance problems [1][2][3]. B. tabaci is a complex of cryptic species [4,5], with MED and MEAM being two of the most damaging ones, formerly known as biotypes B and Q, respectively [6]. Higher fitness of biotype Q in the context of intensive insecticide use has resulted in frequent competitive displacement of biotype B in many locations, including Spain [7][8][9]. Problems associated with traditional chemical pest control make plant resistance an essential tool for the integrated pest management of B. tabaci. Although sources of resistance can be found in the wild relatives of cultivated plants, so far, the only cloned resistance (R) gene against whiteflies is the tomato Mi-1 gene [10]. This gene also provides resistance to root-knot nematodes (Meloidogyne spp.) and other phloem feeding insects such as the potato aphid, Macrosiphum euphorbiae, and the tomato psyllid Bactericerca cockerelli [11][12][13][14]. Mi-1-mediated resistance works by following a gene-for-gene interaction, resulting in a hypersensitive reaction (HR) in the case of nematodes, but not in the case of insects [15]. Factors in the epidermis and/or mesophyll of Mi-1-resistant tomato inhibit the whiteflies involvement of Hsp90 in the Mi-1-mediated resistance to whitefly B. tabaci, which is the objective of this work.

Evaluation of the Effects of Agroinfiltration with the Vector TRV
Motelle and Moneymaker, the resistant and susceptible tomato cultivars, respectively, were agro-infiltrated with the TRV-VIGS vector and infested with B. tabaci in a non-choice assay to test the effect of TRV vector on the whitefly fecundity. The results of these tests are shown in Figure 1. Considering each tomato cultivar separately, the number of eggs on the plant leaflets was similar, regardless of infiltration or not with the "empty" pTRV1:pTRV2 vector (TRV). No statistically significant differences due to TRV infiltration were found for either Moneymaker (p > 0.05) or Motelle (p > 0.05) plants. On the contrary and as expected, significant differences were obtained in the number of eggs on the control (not infiltrated) plants of the cultivars Moneymaker and Motelle (p < 0.001). The difference between Moneymaker and Motelle was maintained after infiltration with the empty TRV vector (p < 0.05).
plants (Motelle cultivar) [28], its potential implication in such a resistance to whiteflies was unknown. The same methodology (TRV-based VIGS) was used now to evaluate the involvement of Hsp90 in the Mi-1-mediated resistance to whitefly B. tabaci, which is the objective of this work.

Evaluation of the Effects of Agroinfiltration with the Vector TRV
Motelle and Moneymaker, the resistant and susceptible tomato cultivars, respectively, were agro-infiltrated with the TRV-VIGS vector and infested with B. tabaci in a nonchoice assay to test the effect of TRV vector on the whitefly fecundity. The results of these tests are shown in Figure 1. Considering each tomato cultivar separately, the number of eggs on the plant leaflets was similar, regardless of infiltration or not with the "empty" pTRV1:pTRV2 vector (TRV). No statistically significant differences due to TRV infiltration were found for either Moneymaker (p > 0.05) or Motelle (p > 0.05) plants. On the contrary and as expected, significant differences were obtained in the number of eggs on the control (not infiltrated) plants of the cultivars Moneymaker and Motelle (p < 0.001). The difference between Moneymaker and Motelle was maintained after infiltration with the empty TRV vector (p < 0.05).   The PCR cycles (35,40,45,50) used are indicated.

Mi-1 Gene Silencing in Tomato Plants
Tomato cultivar Motelle plants were agroinfiltrated with TRV-Mi an B. tabaci in a non-choice assay and whitefly fecundity was evaluated. The was counted on the leaflets of Motelle plants infiltrated with TRV-Mi a compared to those on Motelle and Moneymaker plants infiltrated with th TRV. These results are shown in Figure 3. Significant differences (p < 0.05 between Motelle plants infiltrated with the empty vector TRV and Mot trated with the TRV-Mi vector. This demonstrates that silencing of the M resistance against whiteflies, until similar levels to that of the highly sus Moneymaker infiltrated with the empty vector.    The PCR cycles (35,40,45,50) used are indicated.  To confirm silencing of the Mi-1 gene, its expression was analyzed by semi-quantitative ( Figure 4) and real-time quantitative PCR ( Figure 5). Only the Motelle plants were To confirm silencing of the Mi-1 gene, its expression was analyzed by semi-quantitative ( Figure 4) and real-time quantitative PCR ( Figure 5). Only the Motelle plants were analyzed by semi-quantitative PCR as no expression of the gene was previously obtained in Moneymaker plants ( Figure 2). Figure 4 shows that, in the leaflets of Motelle TRV-Mi plants (which had a large number of eggs), the expression of the Mi-1 gene was not clearly detected until cycle 45, while in Motelle plants infiltrated with the empty vector (Motelle TRV), Mi-1 expression was detected from cycle 35. This indicates a partial silencing of Mi-1 expression; namely, the expression values of Mi-1 do not reach zero because its silencing occurs only in a subset of the cells of the analysed tissue, as it is observed in the mosaic pattern of photobleaching when silencing the phytoene desaturase (PDS) gene ( Figure 6).

Mi-1 Gene Silencing in Tomato Plants
12, x FOR PEER REVIEW analyzed by semi-quantitative PCR as no expression of the gene was previou in Moneymaker plants ( Figure 2). Figure 4 shows that, in the leaflets of Mo plants (which had a large number of eggs), the expression of the Mi-1 gene w detected until cycle 45, while in Motelle plants infiltrated with the empty ve TRV), Mi-1 expression was detected from cycle 35. This indicates a partial sil 1 expression; namely, the expression values of Mi-1 do not reach zero becaus occurs only in a subset of the cells of the analysed tissue, as it is observed pattern of photobleaching when silencing the phytoene desaturase (PDS) ge       The expression of the Mi-1 gene in leaflets of the above-me was also compared by RT-qPCR ( Figure 6). Moneymaker TRV pl control. Significantly (p = 0.001) lower levels of expression of Mi leaflets from Motelle TRV-Mi-1 agroinfiltrated plants than in th plants. Significant differences (p = 0.002) were also obtained in th Motelle TRV-Mi and Moneymaker TRV control plants. Although duced in Mi-1 silenced plants, these levels did not reach zero, a plants, indicating a partial silencing of the Mi-1 gene as demonstr tative PCR analysis.  To confirm the silencing of the Hsp90-1 gene in TRV-SlHsp90-1 agroinfiltrated plants, leaflets with a high number of eggs were used in RT semi-quantitative and RT qPCR to evaluate Hsp90-1 expression. Figure 8 shows the results of the semi-quantitative PCR.

Hsp90-1 Gene Silencing in Tomato Plants
In the control plants (Motelle TRV), the expression of Hsp90-1 is observed from cycle 21 of the PCR, while in the plants agroinfiltrated with TRV-SlHsp90-1, 27 cycles of PCR were needed to obtain a detectable band corresponding to the expression of Hsp90-1.
The quantitative analysis of Hsp90-1 expression was performed using RT qPCR in the same samples, as well as in Moneymaker TRV plants and Motelle plants agroinfiltrated with TRV-Mi ( Figure 9). To confirm the silencing of the Hsp90-1 gene in TRV-SlHsp90-1 agroinfiltrated p leaflets with a high number of eggs were used in RT semi-quantitative and RT qPC evaluate Hsp90-1 expression. Figure 8 shows the results of the semi-quantitative PC The quantitative analysis of Hsp90-1 expression was performed using RT qPCR  To confirm the silencing of the Hsp90-1 gene in TRV-SlHsp90-1 agro leaflets with a high number of eggs were used in RT semi-quantitative evaluate Hsp90-1 expression. Figure 8 shows the results of the semi-qua

Effects of Agroinfiltration with Empty TRV Vector and TRV-Mi
Our results show that agroinfiltration with the empty TRV v defence nor Mi-1-mediated resistance to B. tabaci, as oviposition v in Moneymaker (mi-1/-mi-1) or Motelle (Mi-1/Mi-1) plants treate thermore, the expected differences between the susceptible Mo Motelle plants were maintained. This supports the suitability of

Effects of Agroinfiltration with Empty TRV Vector and TRV-Mi
Our results show that agroinfiltration with the empty TRV vector did not alter basal defence nor Mi-1-mediated resistance to B. tabaci, as oviposition values were not changed in Moneymaker (mi-1/-mi-1) or Motelle (Mi-1/Mi-1) plants treated with this vector. Furthermore, the expected differences between the susceptible Moneymaker and resistant Motelle plants were maintained. This supports the suitability of the TRV vector for VIGS studies with whiteflies. These results agree with those obtained with aphids [26] and nematodes [28], confirming that the infiltration of empty TRV does not alter the resistance mediated by the Mi-1 gene to insects and nematodes. Similarly, TRV infiltration of other plant species did not affect resistance to pathogens mediated by other R genes, such as the N gene, conferring resistance to tobacco mosaic virus (TMV) [58], or the SacMi gene, involved in resistance against M. incognita in Solanum aculeatissimum [59].
Mi-1 expression analysis by semi-quantitative PCR of plants agroinfiltrated with the empty TRV vector showed the same pattern previously observed by Li et al. [26] in younger plants of the same cultivars: no Mi-1 expression was obtained in Moneymaker, while expression in Motelle was observed from cycle 35 of the PCR.
Silencing of the Mi-1 gene in Motelle plants caused a partial loss of resistance to B. tabaci, as has been previously observed for aphids and nematodes using the same TRV vector [26,28]. In Motelle with the Mi-1 gene silenced, the oviposition of B. tabaci females increased compared with non-silenced Motelle and was similar to that of susceptible Moneymaker. This loss of resistance is partial because the silencing by TRV in tomato is not uniform and is patchy throughout the infiltrated leaf, in contrast to the efficiency of VIGS in Nicotiana benthamiana, where the silencing is more uniform [41,58,60]. The silencing of the PDS gene that produces a visible photobleaching phenotype in the leaves [58] allowed to verify this patchy silencing in the foliar tissues. This patchiness necessitates a larger number of tomato plant replicates for the analysis of any gene whose silencing does not produce a visible phenotype in the plant.
Associated with the partial loss of resistance to B. tabaci, semi-quantitative PCR and qPCR analyses confirmed the partial silencing of Mi-1, manifested as a significant reduction in the gene expression levels in plants agroinfiltrated with the TRV-Mi-1 construct.

The Gene Hsp90-1 in the Mi-1-Mediated Resistance to B. tabaci
It was previously known that Mi-1-mediated resistance to aphids and nematodes in tomato requires the participation of Hsp90-1 [28]. Here, our results demonstrate for the first time that the chaperone Hsp90-1 is also involved in Mi-1-mediated resistance to whiteflies. By silencing the Hsp90-1 gene in the resistant tomato Motelle, a partial loss of resistance to B. tabaci was detected, as occurred with the silencing of the Mi-1 gene in the same tomato cultivar, with very similar oviposition values, significantly higher than those on control Motelle plants. Similarly, Hsp90 is also involved in the tomato resistance mediated by the I-2 gene to Fusarium oxysporum [46] and in other pathosystems, such as the resistance mediated by the Pto, Rx, N, and Tm-2 2 genes to Pseudomonas syringae pv. tomato (Pst), potato virus X (PVX), and TMV in Nicotiana benthamiana [41,61]. Other examples are the resistance mediated by RPS2 against Pst in Arabidopsis [42] or the PsoRPM2-mediated resistance to Meloidogyne incognita in tobacco [62]. Thus, the requirement of HSP90 seems common in many cases of resistances mediated by plant R genes. As HSP90 is a highly conserved protein in most living organisms, involved in many biological processes [31][32][33][34], it could also be relevant for the basal response of plants to damaging organisms. However, the study of the possible involvement of the Hsp90 gene in the basal response of susceptible tomatoes to B. tabaci is outside the scope of the present work.
As noted above for the plants with a silenced Mi-1 gene, the partial loss of resistance to B. tabaci observed in the Motelle plants with silenced Hsp90-1 was correlated with a significant decrease in the expression levels of this gene, verified by means of semiquantitative PCR and qPCR. This reduction in Hsp90-1 expression confirmed the results from a previous work in which the same gene was silenced [28]. Quantitative analysis was also used after VIGS to confirm Hsp90 silencing in wheat [52], to evaluate the function of SlSERK3A and SlSERK3B in bacterial and nematode innate immunity [63], or to study the roles of the HsfAs in Mi-1.2-mediated resistance [30].
Previous studies have determined that some chaperones are constitutively expressed, while others are induced in response to stress. AtHSP90.1 is induced in Arabidopsis by p. syringae (Pst DC3000) infection, while three other AtHSP90 isoforms are not [42]. However, during the interaction of tomato/B. tabaci, the expression of Hsp90-1 did not vary in Motelle or Moneymaker plants agroinfiltrated with the empty vector TRV or in plants silenced for Mi-1, suggesting constitutive levels for Hsp90-1 and a regulation independent of Mi-1. This invariance in the expression levels of Hsp90-1 had been observed previously in untreated Motelle plants when comparing non-infested and aphid-infested plants, suggesting that infestation also does not produce variation in the expression levels of Hsp90-1 [28]. Similarly, previous studies have shown that Mi-1 expression is also constitutive, with the same level of expression in different organs of the plant, in different developmental stages, and after the attack of nematodes and aphids [64,65]. Although Hsp90-1 expression is constant, it is known that HSP90 protein levels in tomato are reduced after whitefly infestation [66]. Using polyclonal antibodies, a decrease in HSP90 protein levels has been shown after infestation with both viruliferous (TYLCV-carrying) and non-viruliferous whiteflies, indicating that the stress suffered by the plant translates into a greater consumption of the HSP90 protein, with this reduction in protein levels being more pronounced in plants resistant to TYLCV [66]. Taken together, it could be suggested that the biotic stress caused by infestation with sucking insects in tomato plants does not alter the expression of the Hsp90 gene, but it does cause a response that reduces the accumulated levels of this protein.
We have observed in this work that Hsp90-1 silenced plants had a slightly smaller size than the rest of the agroinfiltrated plants. A similar phenomenon has been observed in the silencing of Hsp90 in N. benthamiana [41], suggesting that HSP90 is involved in signalling pathways that affect both plant growth and immunity processes. In wheat too, silencing of certain Hsp90 genes caused a more pronounced inhibition of wheat seedling growth and even plant death [52].
Several activation models have been described regarding the role of Hsp90 in R genemediated resistance [28,47,67]. In Mi-1-mediated resistance, the model based on the "gatekeeper hypothesis" [68] proposes an interaction between Mi-1, HSP90, and SGT1 proteins that "guard" the RME1 protein [28]. Previous works had already shown that Rme1 is required for Mi-1-mediated resistance to nematodes, aphids, and whiteflies [25,29], but not for I-2 gene-mediated resistance to F. oxysporum f.sp. lycopersici [29] or for Pto-mediated resistance to Pst [25]. In the model proposed by Bhattarai et al. (2007) [28] for Mi-1 genemediated resistance to aphids and nematodes, HSP90 interacts with SGT1 but not RAR1, and RME1 is the target protein for nematode and insect avirulence proteins, which would modify RME1. This modification would cause a conformational change in the Mi-1 protein that would allow the union of HSP90 and SGT1, to form a signalosome that would activate the signalling pathway and defence in the plant. Our previous and current results with B. tabaci reinforce the hypothesis of a common model for nematodes and insects in the interaction of Mi-1, RME1, and HSP90 proteins. From this point on, further evaluation of the expression of other marker genes downstream to Mi-1 will allow to deepen the knowledge on the entire pathway in the complex process of the Mi-1-mediated resistance of tomato against nematodes, aphids, and whiteflies.

Plant Material and Growth Conditions
Tomato (Solanum lycopersicum L.) near isogenic lines Motelle and Moneymaker were used in this study. Motelle is homozygous dominant for the Mi-1 R gene (Mi-1/Mi-1) and Moneymaker is homozygous recessive (mi-1/mi-1). These cultivars differ only in the presence of a 650 kb introgressed region from Solanum peruvianum containing the Mi-1 gene, in chromosome 6 of Motelle [69,70]. Seeds were germinated in sterile vermiculite (number 3, Projar, Spain) inside a growth chamber under controlled conditions: 24 • C and 20 • C (16 h day/8 h night) at 70% relative humidity. Plants were watered when needed and supplied every 15 days with a 20/20/20 nutritive complex (Nutrichem 60; Miller Chemical, Hanover, PA, USA) at 3 g·L −1 . Approximately 10 days after germination, when the first true leaves began to emerge, cultivation of the agro clones was started, as described below (see Section 4.3). Around 15 days after germination, the first two true leaves had generally already developed ( Figure 10) and agroinfiltration of these leaves and cotyledons was carried out. During and after infiltration, care was always taken to keep the plants in separate trays for each genotype and treatment, in order to avoid possible accidental transmission of the virus by contact or during irrigation. About 8 weeks after infiltration, bioassays were carried out by infesting the plants with whiteflies.

VIGS Constructs
The Tobacco Rattle Virus (TRV) vector was used for gene silencing. It is made up of pTRV1 (TRV1), which contains the replicative part of the virus, and pTRV2 (TRV2), in which the gene to be silenced is inserted. This vector has been extensively described [58].

VIGS Constructs
The Tobacco Rattle Virus (TRV) vector was used for gene silencing. It is made up of pTRV1 (TRV1), which contains the replicative part of the virus, and pTRV2 (TRV2), in which the gene to be silenced is inserted. This vector has been extensively described [58].
The pTRV2-SlHsp90-1, pTRV2-Mi-1, and pTRV2-PDS constructs were used to silence the Hsp90-1, Mi-1, and PDS genes, respectively. Moreover, an empty pTRV2, without introduced genetic material, was also used to check if the infiltration itself produces any alteration in the behaviour of B. tabaci. The silencing of Mi-1 has been carried out successfully before [26], but its effect on resistance to whiteflies has not been studied. The silencing of the PDS gene aimed to evaluate the efficacy of the silencing, as it produces a visible phenotype of photobleaching due to the suppression of the activity of phytoene desaturase, which participates in the biosynthesis of carotenoids [26,58].
The pTRV1:pTRV2-PDS (TRV-PDS) and pTRV1:pTRV2-SlHsp90-1 (TRV-SlHsp90-1) constructs were obtained from Dr. S. Dinesh Kumar [35,58]. The pTRV1:pTRV2-Mi-1 (TRV-Mi-1) construction was carried out by cloning a 300 bp fragment of the Mi-1 cDNA and using the primers C1/2Do and C2S4 for their amplification [66]. The amplified fragment was cloned into pGEM T-Easy vector and restricted using SphI and then treated with T4 DNA polymerase. The resulting fragment was digested with SacI and introduced into pTRV2. The plasmid formed was transformed into a culture of Agrobacterium tumefaciens GV3101. Each of the agroclones (TRV1 and TRV2) obtained came from an isolated colony and were preserved in glycerol at −80 • C.

Growth of Agroclones and Agroinfiltration
As mentioned above, the growth of agroclones started approximately 10 days after germination of tomato seeds. Firstly, agroclones were grown on Petri dishes with 1.5% agar in LB medium supplemented with 50 mg·L −1 Kanamycin and 25 mg·L −1 Rifampicin at 28 • C for 24-48 h. Two days before the agroinfiltration, an isolated colony of each agroclone was introduced into a test tube containing 2 mL of LB medium supplemented with 50 mg·L −1 Kanamycin and 25 mg·L −1 and incubated overnight at 28 • C with horizontal agitation (200-250 rpm). Then, 2 ml of each culture was added to a 125 mL Erlenmeyer flask containing 25 mL of LB medium supplemented with Kanamycin (50 mg·L −1 ), Rifampicin (25 mg·L −1 ), Acetosyringone (20 µM), and MES (10 mM). Erlenmeyer flasks were incubated overnight at 28 • C with horizontal agitation (200-250 rpm).
The day of agroinfiltration (approximately 15 days after germination, as mentioned at the beginning) depended on whether the plants had already developed their first two true leaves. This day, the cultures were centrifuged at 2800 rpm at room temperature for 10 min. The pellet was resuspended in a freshly prepared infiltration buffer (10 mM MgCl 2 , 10 mM MES, and 200 µM Acetosyringone). The concentration was measured in a spectrophotometer at OD 600nm and adjusted to an absorbance value of 2. Then, the agroclones were gently stirred for 3-5 h in the dark. After this incubation period, the leaves were infiltrated as follows.
Firstly, TRV1 was mixed with each TRV2 ('empty' or with the gene of interest to be silenced) in a 1:1 ratio to obtain a total volume of 8 mL, usually enough to infiltrate 10 plants. The mixture was aspirated with a needleless syringe and the infiltration was carried out by gently pressing with the syringe on the underside of the leaves and cotyledons so as not to cause any mechanical damage to the plants ( Figure 11).
Ten days after infiltration, plants were ready for transplanting into pots. The temperature at infiltration was reduced to 19 • C and maintained during the next 8 weeks to delay plant development, thus favouring the virus distribution. So, these 10-week-old plants reached a degree of development similar to that of 8-week-old plants that had grown at 24 • C and 20 • C (day/night), which is important as tomato resistance mediated by Mi-1 to B. tabaci is dependent on plant development [71].
Firstly, TRV1 was mixed with each TRV2 ('empty' or with the g silenced) in a 1:1 ratio to obtain a total volume of 8 mL, usually en plants. The mixture was aspirated with a needleless syringe and the ried out by gently pressing with the syringe on the underside of the le so as not to cause any mechanical damage to the plants ( Figure 11). Ten days after infiltration, plants were ready for transplanting in ature at infiltration was reduced to 19 °C and maintained during the n plant development, thus favouring the virus distribution. So, these reached a degree of development similar to that of 8-week-old plant 24 °C and 20 °C (day/night), which is important as tomato resistance B. tabaci is dependent on plant development [71].

Bioassays with Plants Infested with B. tabaci
Moneymaker and Motelle plants were used, both uninfiltrated pTRV1:pTRV2-'empty' (TRV). In addition, Motelle plants infiltrated Mi-1 (TRV-Mi-1) and with pTRV1:pTRV2-SlHsp90-1 (TRV-SlHsp90 These six types of plants were infested in non-choice bioassays with MED (Mediterranean species) B. tabaci. A population of these white several generations in our laboratory, free from any plant pathogen tomato cultivar Marmande. Infestation was performed inside a grow and 20 °C (16 h day/8 h night) at 70% relative humidity. Briefly, each oped leaflets from the upper part of the tomato plant was placed into a a 50 mL Falcon tube following the methodology described in Rodrigu female whiteflies were taken from the rearing colony by vacuum aspir into each tube containing a leaflet. Four replicated plants were used type and treatment and three independent assays were carried out. tubes and the whiteflies were removed and the number of eggs counted. The statistical analysis of the data was performed by the n kal-Wallis test and Dunn's post-test, as more than two groups were c Immediately after egg counting, infested leaflets were cut with individually in Eppendorf tubes containing glass beads, frozen in Figure 11. Agroinfiltration of tomato seedlings in the underside of leaves, using a needleless syringe for virus-induced gene silencing.

Bioassays with Plants Infested with B. tabaci
Moneymaker and Motelle plants were used, both uninfiltrated and infiltrated with pTRV1:pTRV2-'empty' (TRV). In addition, Motelle plants infiltrated with pTRV1:pTRV2-Mi-1 (TRV-Mi-1) and with pTRV1:pTRV2-SlHsp90-1 (TRV-SlHsp90-1) were also used. These six types of plants were infested in non-choice bioassays with adult females of the MED (Mediterranean species) B. tabaci. A population of these whiteflies was reared for several generations in our laboratory, free from any plant pathogen, on the susceptible tomato cultivar Marmande. Infestation was performed inside a growth chamber at 24 • C and 20 • C (16 h day/8 h night) at 70% relative humidity. Briefly, each of three well-developed leaflets from the upper part of the tomato plant was placed into a cage made up from a 50 mL Falcon tube following the methodology described in Rodriguez et al. [23]. Twenty female whiteflies were taken from the rearing colony by vacuum aspiration and deposited into each tube containing a leaflet. Four replicated plants were used for each plant genotype and treatment and three independent assays were carried out. After two days, the tubes and the whiteflies were removed and the number of eggs on each leaflet was counted. The statistical analysis of the data was performed by the non-parametric Kruskal-Wallis test and Dunn's post-test, as more than two groups were compared.
Immediately after egg counting, infested leaflets were cut with a scalpel and placed individually in Eppendorf tubes containing glass beads, frozen in liquid nitrogen, and stored at −80 • C until RNA extraction for gene expression analyses. In addition, other noninfested leaflets were collected from the same plants, individually placed in hermetically sealed plastic bags, and kept on ice until they were used to extract genomic DNA in order to confirm the plant genotype in terms of the presence/absence of the Mi-1 gene.

Genomic DNA (gDNA) Extraction and PCR Amplification
Genomic DNA was extracted from three young leaflets from each plant, following the protocol described by Peterson et al. [72] with slight modifications. The extracted genomic DNA was quantified prior to amplification. For PCR, 50-100 ng of DNA was used and the amplification was carried out in a programmable thermal cycler Thermal Blok II ® (Lab-Line Instruments ® , Dubuque, IA, USA). The reaction was as follows: 3 min at 94 • C, followed by 30 cycles of three steps each: 30 s at 94 • C, 30 s at 64 • C, and 30 s at 72 • C. Finally, a cycle of 5 min at 72 • C. The amplification of Mi-1 from gDNA was performed using the primers PMiF3 (5 -GGTATGAGCATGCTTAATCAGAGCTCTC-3 ) and PMiR3 (5 -CCTACAAGAAATTATTGTGCGTGTGAATG-3 ), designed by El Mehrach et al. [73].

RNA Extraction and cDNA Synthesis and Amplification
Total RNA was extracted from the leaflets collected in the B. tabaci infestation bioassays. Total RNA extraction was performed using Trizol (Invitrogen, Waltham, MA, USA), in a three-step process, including (1) homogenization, separation, precipitation, washing, and dissolution; (2) treatment with DNase RQI (Promega); and (3) re-extraction of the RNA using phenol/chloroform. cDNA synthesis was performed using 5 µg of the total RNA extracted, which was reverse-transcribed using Superscript II reverse transcriptase (Invitrogen). The synthesized cDNA was amplified by PCR for semi-quantitative analysis and those samples with greater interest were subsequently amplified by qRT-PCR (qPCR). The semi-quantitative PCR amplification consisted of 5 min at 95 • C for a variable number of three-step cycles: 45 s at 94 • C, 45 s at 60 • C, and 1 min at 72 • C. For the amplification of Mi-1, 35, 40, 45, and 50 cycles were performed [26], while for the amplification of Hsp90-1, 21, 24, 27, 30, and 35 cycles were used [28]. The expression of the endogenous Ubi3 gene, which has been used in previous VIGS works [26,28], was used as a control. The amplified products were analysed by electrophoresis on Ethidium bromide stained with 1.5% (w/v) agarose gels.
By this semi-quantitative PCR, the cDNA extracted from leaflets of Moneymaker and Motelle plants infiltrated with the empty vector (pTRV2-'empty') was amplified. cDNA was also amplified from Motelle plants silenced for the Mi-1 gene or for the Hsp90 gene, selecting those leaflets in which the number of B. tabaci eggs was higher than that observed in Motelle plants not silenced but infiltrated with the empty vector. The sequences of primers used are detailed in previous works [26,28].
The most interesting samples amplified by semi-quantitative PCR, which confirmed the silencing of the genes, were quantified using real-time quantitative PCR (qRT-PCR or qPCR). This procedure measures the concentration of cDNA in a sample, using a probe that emits fluorescence (SYBR Green ® , Alameda, CA, USA) when bound to nucleic acid. The quantification of the samples was carried out in the 7900HT Fast Real Time PCR system (Applied Biosystems, Waltham, MA, USA) of the Genomics Service of the "Albert Sols" Institute (http://www.iib.uam.es/portal/web/genomica), and Ubiquitin was used as an endogenous control gene for quantification. Two biological samples (leaflets) per plant type were used and each of the qPCR reactions was performed in triplicate for each biological sample. Leaflets of Mi-1 or Hsp90 silenced Motelle plants, in which the number of B. tabaci eggs was higher than that in Motelle leaflets of not silenced plants but infiltrated with the empty vector (pTRV2-'empty'), were analysed. Leaflets from Moneymaker plants were also tested as controls. The qPCR conditions were 10 min at 95 • C, followed by 40 cycles of 15 s at 95 • C and 1 min at 60 • C. Finally, the dissociation curve was carried out, for 15 s at 95 • C, followed by 15 s at 60 • C and finally 15 s at 95 • C. The primers used in this study were specifically designed using the Primer Express ® program (Applied Biosystems) and their specificity was checked against the SGN and NCBI databases. The primers are detailed in Table 1. The relative expression of each gene was calculated using the ∆∆CT method, comparing the data with the reference gene (Ubi3). Finally, the mean of the triplicates and the associated error were calculated. The data obtained were analyzed using the Tukey test to compare all of the samples with each other. Table 1. Primers designed for qRT-PCR of Mi-1, Hsp90-1, and Ubi3 genes.

Ubi3
Ubi3-RT- Funding: This research was funded by a Project (AGL2007-65854/AGR) from the Plan Nacional I+D+I, Spanish Ministry of Education and Science. Clara I. Rodríguez Álvarez was financially supported by a fellowship/contract (AP2006-1035) from the Spanish FPU Program.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.