Genetic dissection of Verticillium wilt resistance mediated by tomato Ve1.

Vascular wilt diseases caused by soil-borne pathogens are among the most devastating plant diseases worldwide. The Verticillium genus includes vascular wilt pathogens with a wide host range. Although V. longisporum infects various hosts belonging to the Cruciferaceae, V. dahliae and V. albo-atrum cause vascular wilt diseases in over 200 dicotyledonous species, including economically important crops. A locus responsible for resistance against race 1 strains of V. dahliae and V. albo-atrum has been cloned from tomato (Solanum lycopersicum) only. This locus, known as Ve, comprises two closely linked inversely oriented genes, Ve1 and Ve2, that encode cell surface receptor proteins of the extracellular leucine-rich repeat receptor-like protein class of disease resistance proteins. Here, we show that Ve1, but not Ve2, provides resistance in tomato against race 1 strains of V. dahliae and V. albo-atrum and not against race 2 strains. Using virus-induced gene silencing in tomato, the signaling cascade downstream of Ve1 is shown to require both EDS1 and NDR1. In addition, NRC1, ACIF, MEK2, and SERK3/BAK1 also act as positive regulators of Ve1 in tomato. In conclusion, Ve1-mediated resistance signaling only partially overlaps with signaling mediated by Cf proteins, type members of the receptor-like protein class of resistance proteins.

Vascular wilt diseases caused by soil-borne pathogens are among the most devastating plant diseases worldwide (Tjamos and Beckman, 1989). Vascular wilts are particularly notorious since, in the vascular system of host plants, the pathogens cannot be reached by many fungicides and few fungicides exist to cure plants once they are infected. Because of extremely persistent resting structures, such as microsclerotia, vascular wilt fungi survive in soil for many years, and the only effective control measure, soil fumigation, is expensive and has harmful environmental effects (Rowe et al., 1987;Fradin and Thomma, 2006). Their high economic impact, combined with the absence of curative treatments, justifies increased attention for vascular wilt diseases. However, to design novel control strategies, understanding the biology of vascular pathogens is of fundamental importance.
Four fungal genera, Ceratocystis, Fusarium, Ophiostoma, and Verticillium, contain the main vascular wilt pathogens (Agrios, 2005). Most vascular pathogens are characterized by narrow host ranges; the exceptions are fungi of the genus Verticillium. While V. longisporum infects various hosts that belong to the Cruciferaceae, including cabbage (Brassica oleracea var capitata), cauliflower (Brassica oleracea), and rapeseed (Brassica napus), V. dahliae and V. albo-atrum are responsible for monocyclic vascular wilt diseases in over 200 dicotyledonous species, including economically important crops (Pegg and Brady, 2002;Fradin and Thomma, 2006). Triggered by root exudates, microsclerotia in the soil germinate and penetrate the roots through the root tip or via wounds and sites of lateral root formation. After crossing the root endodermis, the fungus enters the xylem and produces conidia that are transported by the water stream throughout the plant. Once senescing, tissues become colonized and microsclerotia are produced that are released in the soil during decomposition of plant materials. Little is known about the genetics and molecular biology of Verticillium-host interactions. Recently, transcriptome profiling has been undertaken to study compatible, incompatible, and tolerant interactions to identify genes that play a crucial role in host defense (Robb et al., 2007;. Intriguingly, it was recently demonstrated that posttranscriptional gene silencing governs basal defense against Verticillium in Arabidopsis (Arabidopsis thaliana; Ellendorff et al., 2009).
In several plant species, including alfalfa (Medicago sativa), cotton (Gossypium hirsutum), potato (Solanum tuberosum), strawberry (Fragaria vesca), sunflower (Helianthus annuus), and tomato (Solanum lycopersicum), sources of genetic resistance to Verticillium have been described (Schaible et al., 1951;Lynch et al., 1997, Bae et al., 2008. However, a locus responsible for resistance against Verticillium has been cloned only from tomato (Kawchuk et al., 2001). This Ve locus governs resistance against race 1 strains of V. dahliae and V. alboatrum, and strains that are not contained by this locus are assigned to race 2 (Schaible et al., 1951;Diwan et al., 1999). The Ve locus contains two closely linked inversely oriented genes, Ve1 and Ve2, that, upon independent heterologous expression in potato, were shown to provide resistance against a race 1 V. alboatrum strain (Kawchuk et al., 2001). Both Ve1 and Ve2 were found to encode cell surface receptor proteins that belong to the extracellular Leu-rich repeat (eLRR) receptor-like protein (RLP) class of disease resistance proteins (Kawchuk et al., 2001;Wang et al., 2008). The largest group of eLRR-containing cell surface receptors comprises the receptor-like kinases that contain an eLRR domain, a single-pass transmembrane domain, and a cytoplasmic kinase domain, with over 200 representatives in the Arabidopsis genome (Shiu and Bleecker, 2001). The second largest group of eLRRcontaining cell surface receptors, represented by 57 members in the Arabidopsis genome, is formed by the RLPs, which differ from receptor-like kinases in that they lack a cytoplasmic kinase domain and carry only a short cytoplasmic tail that lacks obvious signaling motifs other than the putative endocytosis motif found in some members Wang et al., 2008). This class of resistance protein was identified originally as Cf resistance proteins that provide resistance in tomato against the foliar leaf mold pathogen Cladosporium fulvum (Jones et al., 1994;Thomma et al., 2005) but also includes the apple (Malus domestica) HcrVf proteins that confer resistance to the scab fungus Venturia inaequalis (Vinatzer et al., 2001;Belfanti et al., 2004;Malnoy et al., 2008). In addition to racespecific resistance proteins, the RLP family harbors receptors that act in basal defense and nonhost resistance, including the tomato LeEIX genes that encode receptors for the ethylene-inducible xylanase produced by Trichoderma biocontrol fungi (Ron and Avni, 2004) and Arabidopsis AtRLP52 and AtRLP30 that provide resistance against the powdery mildew pathogen Erysiphe cichoracearum and nonhost resistance toward Pseudomonas syringae pv phaseolicola, respectively (Ramonell et al., 2005;Wang et al., 2008).
The interaction between tomato and C. fulvum has been the most extensively used model to study the molecular basis (and the evolution) of recognition specificity in RLP-type disease resistance proteins Thomas et al., 1997;Parniske and Jones, 1999;van der Hoorn et al., 2001avan der Hoorn et al., , 2001bvan der Hoorn et al., , 2005Wulff et al., 2001;Seear and Dixon, 2003;Kruijt et al., 2004;Thomma et al., 2005). Also, the genetics of RLP-mediated disease resistance signaling has been most extensively studied exploiting the tomato Cf genes. Using transcriptomics approaches based on AFLPs, the transcriptional response of tobacco (Nicotiana tabacum) suspension cells heterologously expressing the tomato resistance gene Cf-9 was monitored upon addition of the C. fulvum effector Avr9 (Durrant et al., 2000). Similarly, the transcriptome of tomato Cf-4 seedlings heterologously expressing C. fulvum Avr4 was monitored (Gabriëls et al., 2006). Subsequent analysis of candidate genes has revealed several components that are required for the Cf-mediated hypersensitive response or resistance against C. fulvum. These include the thioredoxin CITRX (Rivas et al., 2004), the protein kinase ACIK1 , the NB-LRR protein NRC1 (Gabriëls et al., 2006(Gabriëls et al., , 2007, the U-box protein CMPG1 (González-Lamothe et al., 2006), the mitogen-activated protein (MAP) kinases LeMPK1, LeMPK2, and LeMPK3 (Stulemeijer et al., 2007), and the F-box protein ACRE189/ACIF1 (van den Burg et al., 2008). Although the use of tomato has been successful so far, it may be anticipated that unraveling the genetics of RLP signaling would be facilitated by the use of the model plant Arabidopsis. However, despite significant efforts, so far no race-specific disease resistance proteins of the RLP class have been identified in Arabidopsis Wang et al., 2008).
Here, we describe the functional analysis of Ve1 and Ve2 in resistant and susceptible tomato plants. We show that Ve1, but not Ve2, provides resistance against race 1 strains of V. dahliae and V. albo-atrum and not against race 2 strains. Furthermore, the signaling cascade downstream of Ve1 in tomato is shown to overlap only partially with the Cf-mediated signaling cascade.

Sequence Analysis of the Ve Locus in Resistant and Susceptible Tomato Genotypes
Verticillium resistance in most tomato cultivars is based on the introduction of the dominant Ve locus that was identified in the tomato accession Peru Wild in the 1930s (Schaible et al., 1951). To study the composition of the Ve locus in resistant and susceptible tomato genotypes, the coding sequences (CDSs) of Ve1 and Ve2 homologs were amplified from genomic DNA of the tomato cultivars MoneyMaker (LA2706), which is susceptible to race 1 strains of Verticillium, and Motelle (LA2823) and VFN8 (LA1022), which are resistant to those strains. Furthermore, the homologs were also amplified from the isogenic lines Craigella GCR26 (LA3247) and Craigella GCR218 (LA3428), which are susceptible and resistant to race 1 Verticillium strains, respectively. The Ve1 and Ve2 CDSs, 3.1 and 3.4 kb, respectively, were amplified successfully from all genotypes, and the sequences were compared with the previously published Ve sequences (Kawchuk et al., 2001) for Ve1 genomic DNA (accession no. AF272367; VFN8), Ve1 cDNA (AF272366; Craigella), Ve2 genomic DNA (AF365929; VFN8), and Ve2 cDNA (AF365930; Craigella). Between the two published Ve1 sequences (AF272366 and AF272367), five single nucleotide polymorphisms (SNPs) were identified in the coding region, four resulting in a single amino acid change, while one mutation was silent (Table I, positions 246, 610, 706, 1,548 and 1,888). Interestingly, in the ve1 CDS amplified from the susceptible genotypes, these five SNPs also were found, suggesting that these SNPs are not causing the susceptibility of these genotypes. In addition, four SNPs were identified in the various Ve1 alleles that all resulted in amino acid substitutions (Table I). Remarkably, two of these SNPs (Table I, positions 29 and 35) were identified in all sequenced genotypes, while a third SNP was found in the Ve1 alleles from the resistant genotypes but was absent from the two published Ve1 sequences (Table I, position 380). As these mutations do not discriminate the resistant from the susceptible genotypes, they are unlikely to be the basis of susceptibility in Craigella CGR26 or MoneyMaker. We finally identified a single nucleotide deletion at position 1,220 resulting in a predicted premature stop codon. As a consequence of this deletion, a truncated Ve1 protein of 407 amino acids is predicted in the susceptible cultivars, whereas the protein in resistant cultivars is 1,053 amino acids. Intriguingly, this mutation was present in all susceptible cultivars but not in the resistant cultivars.
For Ve2, eight SNPs were identified, of which six led to predicted amino acid substitutions while two were silent (Table I). Remarkably, two of these SNPs (Table I, positions 3,380 and 3,383) leading to a predicted amino acid change from two Phes into two Sers were identified in all of the sequenced genotypes. In addition to these two SNPs, five SNPs were identified in the Ve2 alleles from the resistant genotypes that were absent from the two published Ve2 sequences, while one SNP was identified only in MoneyMaker. We were not able to identify a single mutation for Ve2 that discriminated between resistant and susceptible genotypes.
To further analyze the Ve locus, the intergenic region between Ve1 and Ve2 was amplified from the resistant tomato genotypes Motelle and Craigella GCR218 and the susceptible genotype Craigella GCR26. In addition to a number of SNPs, approximately in the middle of this intergenic region of 3.4 kb a significant deletion of 36 nucleotides was found in the susceptible Craigella genotype. Subsequently, the intergenic region of the three genotypes was analyzed using the PlantCARE software (http://bioinformatics.psb.ugent.be/webtools/ plantcare/html/; Lescot et al., 2002) to identify putative cis-acting regulatory elements (Supplemental Table S1; Supplemental Fig. S1). In addition to TATA boxes, putative regulatory elements were identified, such as a Box-W1 domain with a putative function in fungal elicitor responsiveness and several TC-rich repeats that are involved in defense and stress responses. Furthermore, a putative ethylene-responsive element was identified in the resistant Craigella genotype but not in the resistant Motelle or susceptible Craigella genotypes. Most importantly, however, no differences in regulatory elements were observed between the resistant and susceptible genotypes.

Ve Expression Analysis in Resistant and Susceptible Tomato Genotypes
The expression of the Ve genes in root, stem, and leaf tissues from susceptible and resistant tomato cultivars MoneyMaker and Motelle at 2 weeks after inoculation with a race 1 V. dahliae strain or mock inoculation was assessed with real-time PCR. Transcripts of the Ve genes were detected in all samples. In the compatible interaction on MoneyMaker plants, transcription of ve1 and ve2 was clearly increased by V. dahliae challenge. Also in the incompatible interaction, transcript accumulation of Ve1 and Ve2 was increased, albeit only moderately, which may reflect a rather localized response, because the fungus is halted at an early stage of the infection process ( Fig. 1). Both genes follow a similar transcription pattern, although the level of Ve2 expression is slightly lower than that of Ve1 (Supplemental Fig. S2). Subsequently, expression of the Ve genes was assessed in the stems of the resistant and susceptible Craigella isogenic lines in time-course experiments ( Fig. 1). This analysis demonstrated that the peak of induction for both genes occurred faster in the incompatible interaction than in the compatible interaction. Several studies show that Verticillium species enter the xylem vessels of the root and start sporulating after 2 to 5 d (Gold and Robb, 1995;Heinz et al., 1998;Chen et al., 2004). After 1 week, sporulation results in colonization of stem vessels coinciding with fungal elimination as a consequence of plant defense.
In compatible interactions, the pathogen is able to overcome this elimination (Gold and Robb, 1995;Heinz et al., 1998;Chen et al., 2004;. Also in the Craigella lines, both genes follow a similar expression pattern, with a slightly higher level of Ve1 transcription when compared with Ve2 ( Fig. 1). In any case, these results indicate that lack of Ve gene expression cannot explain Verticillium compatibility with susceptible tomato genotypes.

Silencing Reveals Differential Activity of Ve1 and Ve2
Based on the sequence analysis and the expression study, it can be concluded that Ve1 and Ve2 expression is induced in resistant as well as susceptible tomato genotypes and that no single mutation in the CDS of Ve2 discriminates resistant and susceptible tomato genotypes. However, a single point mutation in Ve1, resulting in a premature stop codon, was found in all susceptible genotypes and was absent in all resistant genotypes. This suggested that Ve1, but not Ve2, governs Verticillium resistance in tomato.
To investigate the role of Ve1 and Ve2 in Verticillium resistance, we used virus-induced gene silencing (VIGS). VIGS is a well-established method for gene functional analysis in interactions of plants with various foliar pathogens (Burch-Smith et al., 2004). How- ever, VIGS has not been used so far to study interactions with vascular pathogens. Three recombinant tobacco rattle virus (TRV) vectors (Liu et al., 2002a) were designed to target Ve gene expression. While TRV:Ve was designed to target expression of Ve1 and Ve2 simultaneously, TRV:Ve1 and TRV:Ve2 were designed to target expression of Ve1 and Ve2 individually, respectively (Supplemental Fig. S3). As a control, an empty TRV construct (TRV:00) was used. Target specificity of the different constructs was verified by assessment of Ve1 and Ve2 silencing in tomato (Supplemental Fig. S4). Subsequently, the recombinant TRV vectors were inoculated with Nicotiana benthamiana leaf sap containing the recombinant virus (Brigneti et al., 2004) using at least 10 plants per construct of the resistant cultivars VFN8 and Motelle. One week later, half of the plants were inoculated with a race 1 V. dahliae strain, while the other half were mock inoculated. Two weeks after inoculation, Verticillium resistance was assessed by comparing the degree of stunting (height of the plant, length of the leaves, diameter of the stems) that has occurred in host plants, an indicator of disease progression. Upon Verticillium inoculation of TRV:00-treated plants, little stunting was observed when compared with mock-inoculated plants ( Fig. 2A; Table II), indicating that TRV inoculation by itself does not compromise Ve-mediated Verticillium resistance in VFN8 or Motelle plants. In contrast, Verticillium inoculation of TRV:Ve-treated VFN8 and Motelle plants resulted in clear and consistent stunting that was not observed in mockinoculated TRV:Ve-treated plants ( Fig. 2A; Table II). This confirms that the Ve locus is responsible for Verticillium resistance and, importantly, that VIGS can be used as a tool to investigate gene function in resistance signaling against this vascular fungus. Selective targeting of only Ve2 by means of the TRV:Ve2 construct resulted in slight stunting after Verticillium inoculation, similar to that in Verticillium-inoculated TRV:00-treated plants ( Fig. 2A; Table II). Interestingly, clearly compromised Verticillium resistance was observed after selective targeting of Ve1 expression using TRV:Ve1 ( Fig. 2A; Table II). These findings were confirmed by fungal recovery from stem sections of the inoculated plants (Fig. 2B) and confirm the hypothesis that Ve1, but not Ve2, mediates Verticillium resistance in VFN8 and Motelle plants. To confirm our finding that Ve1, but not Ve2, mediates Verticillium resistance in tomato, stable overexpression lines were generated in the susceptible tomato cultivar MoneyMaker expressing either Ve1 or Ve2 driven by the cauliflower mosaic virus (CaMV) 35S promoter (Supplemental Figs. S5 and S6). For Ve1, the Motelle/VFN8 allele was used (P35S:Ve1; Supplemental Fig. S6) because it was shown to provide resistance in our VIGS analysis and this genotype was used previously to engineer Verticillium-resistant potato (Kawchuk et al., 2001). For Ve2, the Craigella GCR26 allele was used (P35S:Ve2; Supplemental Fig.  S3) that most closely matches the allele used to engineer Verticillium-resistant potato (Kawchuk et al., 2001). As shown in Table I, we have not been able to identify the exact same Ve2 allele used by Kawchuk et al. (2001). However, the only polymorphism that is present in the Craigella GCR26 allele is present in all Ve2 alleles analyzed. For each construct, at least 10 transgenic lines were obtained, of which, after determination of diploidy levels and the copy number of the transgene, lines with one-or two-copy inserts were chosen for further analysis. For each of the constructs, a minimum of five T2 plants of a minimum of two different lines were challenged with each of five different race 1 Verticillium isolates, three belonging to V. dahliae and two to V. albo-atrum (Table III). Intriguingly, while all plants carrying the P35S:Ve1 transgenes were found to exhibit robust Verticillium resistance, all plants carrying P35S:Ve2 transgenes were as susceptible as MoneyMaker plants toward these race 1 isolates, showing typical wilt symptoms that included stunting, chlorosis, wilting, and necrosis ( Fig. 3A; Table III). Furthermore, when challenged with race 2 isolates belonging to V. dahliae and V. albo-atrum, all transgenic plants showed typical symptoms of Verticillium disease (Table III). All findings were confirmed in subsequent analyses using the T3 generation of the transgenic lines. Moreover, the disease phenotypes were corroborated by assessing Verticillium colonization of the transgenic plants through measurement of fungal recovery from stem sections (Fig. 3B).
In addition to the lines with constitutive Ve expression, stable transgenic MoneyMaker lines were generated expressing either the same Ve1 or Ve2 CDS but driven by the endogenous promoter isolated from Motelle (PVe1:Ve1 and PVe2:Ve2, respectively; Supplemental Fig. S5). For PVe2:Ve2, the full intergenic region was used, while for Ve1, only half the intergenic region adjacent to the Ve1 CDS was used (Supplemental Fig.  S5). Subsequent Verticillium assays on transgenic lines in the T2 and T3 generations revealed that, when driven by the Motelle promoter, Ve1 but not Ve2 conferred resistance toward race 1 isolates of V. dahliae and V. albo-atrum but not toward race 2 isolates ( Fig.  3A; Table III). These disease phenotypes were corroborated by assessment of Verticillium colonization of the transgenic plants through measurement of fungal recovery from stem sections (Fig. 3B).

Characterization of Ve-Mediated Signaling by VIGS
So far, little is known about the genetic requirements for Ve signaling. The only gene implicated in downstream signaling is the tomato homolog of Arabidopsis Eds1 (for Enhanced Disease Susceptibility1; Aarts et al., 1998;Hu et al., 2005). To further identify genes required for Ve-mediated resistance, a set of candidate genes was selected, some of which have previously been implicated in RLP signaling mediated by the tomato Cf genes against C. fulvum. In addition to Eds1, this set included genes encoding the disease signaling components RAR1 (for Required for Mla12 Resis-tance1; Liu et al., 2002b), SGT1 (for Suppressor of the G2 allele of SKP1; Peart et al., 2002), NDR1 (for Nonrace-specific Disease Resistance; Ekengren et al., 2003), and NPR1 (for Nonexpressor of Pathogenesis-Related genes1; Liu et al., 2002b) but also the Ser/Thr protein kinase ACIK1 , the F-box protein ACIF1 (van den Burg et al., 2008), the U-box protein CMPG1 (González-Lamothe et al., 2006), the MAP kinase kinase MEK2 (Ekengren et al., 2003), the MAP kinases LeMPK1, LeMPK2, and LeMPK3 (Stratmann and Ryan, 1997;Stulemeijer et al., 2007),  and the NB-LRR protein required for hypersensitive response-associated cell death NRC1 (Gabriëls et al., 2007). All of these TRV constructs have been described and used previously for effective silencing in tomato (Peart et al., 2002;Ekengren et al., 2003;Rowland et al., 2005;Gabriëls et al., 2006;González-Lamothe et al., 2006;Stulemeijer et al., 2007;van den Burg et al., 2008). As expected, silencing of Eds1 resulted in a clear and consistent decrease of resistance in Motelle tomato plants ( Fig. 4A; Table II), indicated by stunting (reduced plant height, leaf length, and stem diameter), confirming the previously described involvement of Eds1 in Ve-mediated signaling (Hu et al., 2005). Clear and consistent loss of Verticillium resistance in Motelle plants also was observed upon treatments with recombinant viruses targeting Mek2, Nrc1, Acif1, and Ndr1 ( Fig. 4A; Table II), indicating their requirement for Ve-mediated disease resistance. In contrast, recombinant viruses targeting Cmpg1, Mpk1, Mpk2, and Rar1 did not compromise Verticillium resistance in Motelle plants ( Fig. 4B; Table II), while viruses targeting Mpk3 and Npr1 caused a slightly higher number of stunted plants when compared with the empty vector control, suggesting that these components could make a minor contribution to disease resistance (Table II).
It was shown recently that the Arabidopsis Somatic Embryogenesis Receptor Kinase3 (SERK3)/Brassinosteroid (BR)-Associated Kinase1 (BAK1) takes part in an elicitor-dependent complex with Flagellin Sensing2 (FLS2) to initiate a defense response upon elicitation with the bacterial pathogen-associated molecular pattern (PAMP) flagellin or its peptide derivative flg22 (Chinchilla et al., 2007;Heese et al., 2007). In N. benthamiana, Serk3/Bak1 is also required for flagellintriggered immunity (Heese et al., 2007). In addition, in Arabidopsis as well as N. benthamiana, Serk3/Bak1 is required for full responses to unrelated PAMPs, basal defense, and restriction of pathogen infection (Heese et al., 2007;Kemmerling et al., 2007). Therefore, we attempted to silence the tomato gene encoding SERK3/BAK1 using two different TRV constructs to target different regions of NbSerk3 (Heese et al., 2007). As a control, TRV constructs targeting NbFls2 and NbSerk2 (Colcombet et al., 2005;Heese et al., 2007) were included. Treatment of Motelle tomato plants with the two constructs targeting Serk3/Bak1 or the construct targeting Serk2 resulted in slight stunting and weakly distorted leaves. These results are consistent with the phenotype of N. benthamiana upon treatment with these constructs (Heese et al., 2007). Interestingly, treatment with the two different recombinant viruses targeting expression of Serk3/Bak1, but not with viruses targeting expression of Fls2 or Serk2, clearly compromised Verticillium resistance ( Fig.  4; Table II). This result suggests that, in addition to PAMP-triggered immunity, Serk3/Bak1 also functions in race-specific disease resistance. Many crop species contain genes for tolerance or partial resistance, but not complete resistance, to Verticillium species (Fradin and Thomma, 2006). Tomato is an exception in which resistance to race 1 Verticillium isolates is conferred by a single dominant locus that was introduced in cultivated varieties in the 1950s (Schaible et al., 1951;Diwan et al., 1999) and that is still carried by most commercial tomato varieties. This article describes the functional analysis of Ve1 and Ve2 in resistant and susceptible tomato plants. We were not able to identify a single mutation for Ve2 that discriminated between resistant and susceptible genotypes. However, sequence analysis revealed that Ve1 encodes a truncated protein in all susceptible genotypes that were analyzed. This suggested that solely Ve1 determines the resistance of tomato toward race 1 strains of Verticillium. This hypothesis was verified through two lines of evidence. First, VIGS of Ve1 but not of Ve2 compromised Verticillium resistance in Motelle and VFN8 plants that harbor the Ve locus. Second, trans-genic tomato plants expressing either Ve1 or Ve2 showed that Ve1 expression, and not Ve2 expression, resulted in resistance against race 1 strains of V. dahliae and V. albo-atrum, irrespective of whether expression was driven by the endogenous promoter or the constitutive CaMV 35S promoter.
Previously, the Ve locus was cloned from tomato and used for heterologous expression in susceptible potato (Kawchuk et al., 2001). Our study revealed a number of sequence differences for the Ve1 and Ve2 alleles that were sequenced by Kawchuk et al. (2001). Support for the veracity of the sequences from our study is provided by Acciarri et al. (2007), who similarly reported on the sequencing of Ve1 and Ve2 alleles from resistant and susceptible Italian tomato genotypes. That study confirms the polymorphisms found at positions 246, 380, 610, 706 and 1,220 in Ve1 and at positions 1,811, 2,771, 2,893, and 2,934 in Ve2 in our study. Remarkably, Kawchuk et al. (2001) reported that both Ve1 and Ve2 provided resistance against a race 1 strain of V. alboatrum, irrespective of whether expression was driven by the endogenous promoter or the constitutive CaMV 35S promoter. Possibly, the Ve2 protein is no longer active in tomato while it is still able to connect to a Figure 4. Characterization of Ve-mediated signaling in resistant tomato by VIGS. Motelle (Ve/Ve; resistant) plants were agroinfiltrated with recombinant TRV vector carrying a fragment of candidate disease signaling genes. Two weeks after agroinfiltration, the plants were mock inoculated (control) or inoculated with a race 1 strain of V. dahliae (JR2). Photographs were taken at 14 d after V. dahliae inoculation, and compromised resistance is evident from a stunted appearance of V. dahliae-inoculated plants when compared with control plants. A, VIGS of tomato genes required for Ve-mediated Verticillium resistance. B, VIGS of tomato genes that do not play a major role in Ve-mediated Verticillium resistance. disease signaling cascade in potato, for instance, through the presence of auxiliary components in potato that confer functionality. Also, in contrast to Ve1, Ve2 contains a PEST motif that is typically observed in many rapidly degraded proteins (Hershko and Ciechanover, 1998). Therefore, the protein stability of Ve2 may be significantly reduced in tomato when compared with Ve1. Alternatively, the single race 1 V. albo-atrum strain that was used on potato contains an elicitor that is not generally carried by most race 1 isolates. Loci with active (demonstrated resistance specificities) and nonactive (unknown functions) homologs of RLP-type resistance genes are found commonly, not only in tomato (Dixon et al., 1996;Parniske et al., 1997) but also in apple (Malnoy et al., 2008). It has been speculated that members with unknown functions are a source to generate new recognition (R gene) specificities , which may also be true for the Ve locus.

Genetic Analysis of Ve-Mediated Signaling in Tomato
Interestingly, VIGS using recombinant viruses that target Ve1 expression resulted in compromised V. dahliae resistance, demonstrating that this transient assay can be used to investigate defense against a vascular pathogen. Apart from Eds1 (Hu et al., 2005), little is known about the genetic requirements for Ve signaling. In Arabidopsis, a differential requirement for Eds1 and Ndr1 was shown, particularly for cytoplasmic disease resistance proteins of the NB-LRR class (Aarts et al., 1998). Although exceptions exist for this class of resistance proteins, Eds1 generally mediates signaling initiated by the TIR-NB-LRR subclass, whereas Ndr1 mediates signaling initiated by the CC-NB-LRR subclass (Century et al., 1995;Aarts et al., 1998). Previously, Eds1 but not Ndr1 was found to play a role in Cf4-mediated signaling (Gabriëls et al., 2007). Intriguingly, both Eds1 and Ndr1 are required for Ve1 resistance, which represents, to our knowledge, the first example of a membrane-anchored resistance protein with extracellular LRRs that requires both of these genes that are more commonly associated with NB-LRR resistance.
In addition to Eds1 and Ndr1, also the MAP kinase kinase gene Mek2, the NB-LRR protein encoding Nrc1, and the F-box protein encoding Acif1 are required for Ve signaling, which was confirmed by performing fungal recovery assays from stem sections of the inoculated plants showing enhanced Verticillium outgrowth. These components have been implicated in Cf-mediated signaling as well (Gabriëls et al., 2007;van den Burg et al., 2008;Fig. 5). Furthermore, tomato Mek2 has been implicated in tomato resistance against Pseudomonas syringae mediated by Pto, while the NB-LRR protein encoding Nrc1 is required for the hypersensitive response induced by diverse R proteins, including LeEix, Pto, Rx, and Mi (Ekengren et al., 2003;Gabriëls et al., 2007). Recombinant TRV targeting expression of the U-box protein CMPG1, the MAP kinases MPK1 to MPK3, and the disease signaling components RAR1 (Liu et al., 2002b) and NPR1 (Liu et al., 2002b;Ekengren et al., 2003) did not consistently compromise Verticillium resistance in Motelle plants. Although we have not verified the VIGS efficiency of these genes, our results indicate that they do not play a major role in the resistance response. All of these, except NPR1, have been found to play a role in Cf signaling (González-Lamothe et al., 2006;Gabriëls et al., 2007;Stulemeijer et al., 2007), suggesting that the Ve1 and Cf proteins differ significantly in their requirements for downstream signaling components (Fig. 5). This is further substantiated by the recent observation that the C. fulvum-induced transcriptional changes in tomato show little overlap with those induced by V. dahliae in compatible as well as incompatible interactions ).

SERK3/BAK1 May Form a Receptor Complex with Ve1 in Tomato
For VIGS of all genes tested in this study, silencing constructs were employed that have been published previously and have been shown to be effective in silencing the tomato genes that were targeted. The only exception was the construct used to target the expression of Serk3/Bak1 and the corresponding con- Figure 5. The Ve1 and Cf proteins differentially require downstream signaling components. Signaling components that are required for Ve1mediated Verticillium resistance (in the blue circle) and those required for Cf-mediated Cladosporium resistance (in the red circle) are shown. The components in the overlap of the two circles are required for both resistance responses. The asterisks indicate that Serk3/Bak1 has not been tested for requirement in Cf signaling, while Sgt1 has not been tested for requirement in Ve1 signaling. trol constructs to target the related SERK family member Serk2 and the expression of the receptor for the bacterial PAMP flagellin and its peptide derivative flg22 Fls2. In Arabidopsis, AtSERK3/BAK1 takes part in an elicitor-dependent complex with FLS2 (Chinchilla et al., 2007;Heese et al., 2007), and in N. benthamiana, Serk3/Bak1 is also required for flagellintriggered immunity (Heese et al., 2007). In addition, in Arabidopsis as well as in N. benthamiana, Serk3/Bak1 is required for full responses to unrelated PAMPs, basal defense, and restriction of pathogen infection (Heese et al., 2007;Kemmerling et al., 2007). In our study, we observed weakly distorted leaves in tomato plants silenced with either of the NbSerk3 constructs. These results are consistent with the leaf phenotypes upon silencing of NbSerk3 in N. benthamiana plants and upon SERK3/BAK1 knockout in Arabidopsis, which were attributed to defects in brassinosteroid perception (Heese et al., 2007). This suggests that, indeed, the true tomato Serk3/Bak1 homolog had been silenced. So far, tomato Serk3/Bak1 has not been identified, but several studies have exploited N. benthamiana sequences to successfully target genes in the close relative tomato (Gabriëls et al., 2006(Gabriëls et al., , 2007. Interestingly, in our study, both of the NbSerk3 constructs that target different regions of Serk3, but not the Serk2 or the Fls2 construct, clearly compromised Verticillium resistance. Since SERK3/BAK1 is a coreceptor that physically associates with BRI1 for BR-dependent signaling and with FLS2 for flagellin-induced immunity, this may indicate that tomato SERK3/BAK1 physically associates with the RLP Ve1 to initiate Verticillium immunity. Future experiments will be directed to investigate this possibility.

MATERIALS AND METHODS
All experiments have been performed a minimum of three times yielding similar results.

Plant Manipulations
Tomato (Solanum lycopersicum) was grown in soil in the greenhouse at 21°C/19°C during 16-h/8-h day/night periods, respectively, with 70% relative humidity and 100 W m 22 supplemental light when the intensity dropped below 150 W m 22 . For Verticillium inoculations, 10-d-old tomato plants were uprooted and the roots were rinsed in water. Subsequently, the roots were dipped for 3 min in a suspension of 10 6 conidia per mL of water and harvested from 1-to 2-week-old Verticillium cultures on potato dextrose agar (Oxoid). Control plants were treated similarly, but their roots were dipped in water without conidiospores. After replanting in fresh soil, disease development was monitored up to 28 d after inoculation. The following isolates were used: V. dahliae ST14.01, JR2, CBS381.66 (all race 1), CBS321.91, and M050414 (both race 2) and V. albo-atrum 5431, CBS385.91 (both race 1), CBS451.88, and VA1 (both race 2).

Cloning of Ve Sequences
To amplify the CDSs of Ve1 and Ve2, the primer pairs Ve1F-Ve1R and Ve2F-Ve2R, respectively (Supplemental Table S2), were used with Expand High-Fidelity PCR system enzyme mix (Roche). PCR products were sequenced in both directions (Supplemental Table S2) and with overlapping sequencing products directly or, alternatively, cloned into a pBluescript variant with BamHI and AscI restriction sites, after which multiple clones from independent PCRs were sequenced (BaseClear). For constitutive expression, the Ve1 and Ve2 CDSs were cloned into a binary vector pmog800 variant (Honée et al., 1998), resulting in P35S:Ve1 and P35S:Ve2 (Supplemental Fig. S4).
The region between the inversely oriented Ve CDSs was PCR amplified (Supplemental Table S2) and sequenced (BaseClear). Furthermore, two constructs were designed (Supplemental Fig. S4): one containing the Ve1 CDS and half the intergenic region (IR) adjacent to the Ve1 CDS (PVe1:Ve1), and one containing the complete IR fused to the Ve2 CDS (PVe2:Ve2). The primer pair Ve1ProRegF and VeProReg3R (Supplemental Table S2) was used with Expand High-Fidelity PCR system enzyme mix (Roche) to amplify half of the IR and part of Ve1, containing an endogenous PstI restriction site. The PCR fragment was cloned into pGEM-T Easy (Promega) and sequenced (Supplemental Table  S2). Subsequently, the IR fragment was excised using ApaI and PstI and directionally cloned into the binary vector pGREEN (Hellens et al., 2000). Next, a PstI-SmaI fragment of the P35S:Ve1 construct containing the Ve1 sequence and the terminator from the potato proteinase inhibitor II (PiII) gene was cloned into the vector, resulting in PVe1:Ve1. For Ve2, the complete IR was obtained using two PCR-amplified fragments. The first IR fragment was amplified with the primer combination VeProRegF and VeProReg3R (Supplemental Table S2) and partially overlapped with the second IR fragment that was amplified with the primer pair VeProReg3F and Ve2ProRegR (Supplemental Table S2), with an endogenous SpeI restriction site in the region of overlap. The second IR fragment also partially overlapped with part of the Ve2 CDS, with an endogenous PstI restriction site in the region of overlap. Both fragments were cloned into pGEM-T Easy, sequenced, and excised using ApaI, SpeI, and PstI. Both fragments were cloned into ApaI-and PstI-digested pGREEN. Subsequently, a PstI-SmaI fragment of the P35S:Ve2 construct containing the Ve2 sequence and the PiII terminator was cloned into the vector, resulting in PVe2:Ve2. All constructs were introduced into Agrobacterium tumefaciens strain LBA4404 by electroporation and used for tomato transformation.

Engineering of Transgenic Plants
Tomato transformation was performed as described previously (van Esse et al., 2008). The ploidy level of transgenic tomato plants was determined as described (Jacobs and Yoder, 1989). Subsequently, diploid plants were retained and the transgene copy number was determined by quantitative real-time PCR using the qPCR Core kit for SYBR Green I (Eurogentec) with genomic DNA (Supplemental Table S3; Ingham et al., 2001). The single-copy tomato gene encoding protein phosphatase 5 was used as a reference to determine the number of copies of the neomycin phosphotransferase II transgene selection marker (Supplemental Table S2). Real-time PCR conditions consisted of an initial denaturation step of 10 min at 95°C, followed by denaturation for 15 s at 95°C, annealing for 30 s at 60°C, and extension for 30 s at 72°C for 40 cycles. Only one-or two-copy transgenes were used in this study.

VIGS Experiments
For all VIGS experiments, the binary TRV constructs pTRV-RNA1 and pTRV-RNA2 were used (Liu et al., 2002a). The inserts to generate TRV:Ve1 and TRV:Ve were amplified from the P35S:Ve1 plasmid using the primer pairs Ve1F-Ve1VIGSspeR and VeVIGSF2-VeVIGSR1, respectively, while the insert for TRV:Ve2 was amplified from the P35S:Ve2 plasmid using the Ve2F-Ve2VIGSspeR primer pair (for primer sequences, see Supplemental Table  S2). PCR fragments were cloned into pTRV:RNA2 (pYL156) using BamHI and KpnI. The constructs were transformed to A. tumefaciens GV3101 by electroporation.
TRV vectors were agroinfiltrated as described (Liu et al., 2002a) into cotyledons of 9-d-old Motelle (Ve/Ve) or VFN8 (Ve/Ve) plants, and after 2 weeks the plants were inoculated with race 1 V. dahliae. Alternatively, TRV vectors were agroinfiltrated into a leaf of 3-to 4-week-old Nicotiana benthamiana plants, and 3 to 6 d after agroinfiltration leaf sap was collected by grinding the agroinfiltrated leaves in 50 mM phosphate buffer (pH 7.2). Subsequently, 9-d-old Motelle plants were virus inoculated by rubbing the cotyledons with 6 to 12 mL of the leaf sap and inoculated with a race 1 strain of V. dahliae 1 week after treatment.

Expression Analyses
Target specificity of the constructs TRV:Ve, TRV:Ve1, and TRV:Ve2 was determined in the MoneyMaker overexpression lines expressing either Ve1 or Ve2 driven by the CaMV 35S promoter. Two weeks after virus inoculation, RNA was isolated from whole plants using the RNeasy kit (Qiagen) and used for cDNA synthesis using an oligo(dT) primer (Supplemental Table S2) and the SuperScript III reverse transcriptase kit (Invitrogen), according to the manufacturers' instructions. To analyze expression of the Ve alleles, real-time PCR was conducted with Ve-specific primers (Ve1QPCRF2-Ve1QPCRR1 for Ve1 and Ve2SeqF7-Ve2R for Ve2) with tomato actin as an internal standard (Supplemental Table S2) using the qPCR Core kit for SYBR Green I (Eurogentec). Real-time PCR conditions consisted of an initial denaturation step of 10 min at 95°C, followed by denaturation for 15 s at 95°C, annealing for 30 s at 60°C, and extension for 30 s at 72°C for 30 cycles.
Ve expression analyses in resistant and susceptible tomato genotypes as well as in Ve transgenic tomato lines were performed similarly.

Fungal Recovery Assay
Two weeks after Verticillium inoculation, a stem section immediately above the cotyledons was taken from three inoculated plants, surface sterilized for 15 min in 70% ethanol, followed by 15 min in 10% hypochlorite, rinsed three times with sterile water, and sliced. In total, for each plant, 10 slices were transferred onto potato dextrose agar supplemented with chloramphenicol (34 mg L 21 ) and incubated at 22°C. The frequency of stem slices from which Verticillium grew out is a measure of the susceptibility of the plant.
Sequences described in this study have been deposited in GenBank with accession numbers FJ464553 to FJ464565.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Sequence alignment of the Ve locus intergenic region of a resistant and a susceptible isogenic tomato line.
Supplemental Figure S2. Real-time PCR of Ve1 and Ve2 expression in susceptible and resistant tomato cultivars.
Supplemental Figure S3. Alignment of fragments used to target the expression of Ve genes.
Supplemental Figure S4. Specificity of fragments used to target the expression of Ve genes.
Supplemental Figure S5. Constructs used for transgenic expression of Ve1 and Ve2.
Supplemental Figure S6. Reverse transcription-PCR of Ve1 and Ve2 expression in transgenic tomato lines.
Supplemental Table S1. Putative regulatory elements identified in the intergenic region of the Ve locus of susceptible and resistant tomato genotypes.
Supplemental Table S2. Primers used in this study.
Supplemental Table S3. Copy number determination in transgenic tomato.