Loss of Function in Mlo Orthologs Reduces Susceptibility of Pepper and Tomato to Powdery Mildew Disease Caused by Leveillula taurica

Powdery mildew disease caused by Leveillula taurica is a serious fungal threat to greenhouse tomato and pepper production. In contrast to most powdery mildew species which are epiphytic, L. taurica is an endophytic fungus colonizing the mesophyll tissues of the leaf. In barley, Arabidopsis, tomato and pea, the correct functioning of specific homologues of the plant Mlo gene family has been found to be required for pathogenesis of epiphytic powdery mildew fungi. The aim of this study was to investigate the involvement of the Mlo genes in susceptibility to the endophytic fungus L. taurica. In tomato (Solanum lycopersicum), a loss-of-function mutation in the SlMlo1 gene results in resistance to powdery mildew disease caused by Oidium neolycopersici. When the tomato Slmlo1 mutant was inoculated with L. taurica in this study, it proved to be less susceptible compared to the control, S. lycopersicum cv. Moneymaker. Further, overexpression of SlMlo1 in the tomato Slmlo1 mutant enhanced susceptibility to L. taurica. In pepper, the CaMlo2 gene was isolated by applying a homology-based cloning approach. Compared to the previously identified CaMlo1 gene, the CaMlo2 gene is more similar to SlMlo1 as shown by phylogenetic analysis, and the expression of CaMlo2 is up-regulated at an earlier time point upon L. taurica infection. However, results of virus-induced gene silencing suggest that both CaMlo1 and CaMlo2 may be involved in the susceptibility of pepper to L. taurica. The fact that overexpression of CaMlo2 restored the susceptibility of the tomato Slmlo1 mutant to O. neolycopersici and increased its susceptibility to L. taurica confirmed the role of CaMlo2 acting as a susceptibility factor to different powdery mildews, though the role of CaMlo1 as a co-factor for susceptibility cannot be excluded.


Introduction
Powdery mildews are conspicuous plant pathogens that comprise approximately 500 species and infect more than 1500 plant genera [1]. In Europe, the largest application of fungicides is for controlling powdery mildew diseases in agricultural and horticultural production [2]. The powdery mildew pathogen Leveillula taurica (Lév.) G. Arnaud is a serious fungal threat to pepper as well as tomato production. Heavy epidemics of powdery mildew disease could cause a significant yield loss up to 2 to 4 kg/ m 2 in greenhouse pepper production [3].
Besides repeated application of fungicides, powdery mildew diseases could be controlled by using resistant cultivars. In practice, breeding for resistance is mainly done by introgressing the resistance trait from wild species into the cultivated crop. In tomato, the Lv gene is the only resistance (R) gene identifed so far conferring resistance to L. taurica, which was found in a wild tomato accession of Solanum chilense and mapped on chromosome 12 [4]. In the Capsicum genus, several studies have been carried out to search for resistance to L. taurica [5,6,7,8]. Five quantitative trait loci (QTLs) for resistance have been identified, with one of the QTLs, namely Lt-9.1, co-linearizing with the tomato Lv locus [9].
In principle, all resistance resources discovered from the wild accessions could be promising materials for isolation of potential R-genes and be used in resistance breeding. However, there are several weaknesses of using R-genes. First of all, the interspecific crossability barrier could restrict introgression of an R-gene from the resistant donor into cultivated species [10]. Even if the resistant donor can be easily crossed with the cultivated species, extensive backcrossing is required to remove undesirable traits. Secondly, transferring R-genes from one species into another does not guarantee that resistance conferred by the R-genes is retained in the receptor species in all cases. The function of an R-gene sometimes requires additional gene(s) in signaling pathways and/ or metabolites [11,12]. Thirdly, R-genes confer race-specific resistance, which could easily be overcome by new races of the pathogen in a short period.
Complementary to introgression of R-genes, a novel breeding strategy has been proposed, which is to disable plant susceptibility genes (S-genes) [13,14]. In order to infect a plant species a pathogen should be able to suppress the plant's innate immunity by exploiting effector molecules to establish effector-triggered susceptibility. Plant genes, which are required for triggering susceptibility to pathogens and play a negative role in defense responses, are referred to as S-genes [13]. Impairment of the function of plant S-genes results in recessive resistance. One representive example is the Mildew resistance Locus O (Mlo) gene identified in barley. Loss-of-function mutants of the barley Mlo gene give resistance to powdery mildew (Blumeria graminis f. sp. hordei) and have been used in European barley cultivation for more than 30 years [15]. In addition to barley, mutations in Mlo orthologues result in recessively inherited powdery mildew resistance in Arabidopsis [16], tomato [17] and pea [18,19]. In tomato, we have demonstrated that the recessive ol-2 gene conferring resistance to the powdery mildew pathogen Oidium neolycopersici contains a 19-bp deletion in the coding region of the tomato Mlo ortholog, SlMlo1 [17,20]. So far, no natural powdery mildew isolates could break down the mlo-based resistance, which thus represents a broad-spectrum resistance and has become a successful example of using S-genes in crop protection for durable resistance.
Mlo susceptibility genes are part of a large family (the Mlo gene family) encoding a class of plant-specific proteins anchored in the plasma membrane by seven transmembrane domains [21,22]. When functional MLO susceptibility proteins are lacking, powdery mildew fungi fail to enter their host. So far, mlo-based resistancerepresents a well-studied pre-penetration resistance conferred by cell wall appositions [16,23,24,25]. To date, 15, 17, 7, 9 and 12 Mlo paralogs have been reported in the genomes of Arabidopsis, grape, wheat, maize and rice, respectively [16,26,27,28,29]. Of these paralogs, members of two phylogenetic clades, one specific for monocots and the other for dicots, play a role in susceptibility to powdery mildews. These members are characterized by the presence of a tetra-peptide (D/E-F-S/T-F) motif in the cytoplasmic region at the C-terminus of the protein [17,26,30,31] and their response to powdery mildew infection at the early time points [17,32,33].
Besides O. neolycopersici, another powdery mildew fungus, L. taurica, can also infect tomato. O. neolycopersici develops its mycelium on the leaf surface and is distinguishable by the appearance of characteristic powder-like colonies on the adaxial side of the leaves. In contrast, L. taurica grows intercellularly inside the leaf. After a latency period of 3-4 weeks disease symptoms appear as chlorotic spots on the adaxial side of the leaves (see http://cals. arizona.edu/plp/plpext/diseases/vegetables/tomato/pm.htm) and fungal conidiophores appear as white powdery masses on the abaxial side of the leaf [34]. Based on the different colonization habit, O. neolycopersici is referred to as an epiphytic powdery mildew fungus, whereas L. taurica is referred to as endophytic. The aim of this study was to test, in pepper and tomato, whether Mlo homologs could have a role in the interaction with the endophytic powdery mildew species L. taurica.
When we started our work only one Mlo gene had been identified in pepper [30]. This CaMlo1 gene was considered to be a mildew-effective Mlo ortholog of SlMlo1 [18,30]. We isolated a second Mlo gene in pepper, CaMlo2. During our study, the isolation and characterization of another allele of the same CaMlo2 gene has been described by Kim and Hwang [35], accession number JN896629. Their results demonstrate that the CaMllo2 gene is involved in cell death response as well as formation of reactive oxygen species (ROS). In addition, they showed that the CaMlo2 gene was induced by the hemibiotrophic bacterial pathogen Xanthomonas campestris pv. vesicatoria (Xcv), the oomycete pathogen Phytophthora capsici, exogenous salicylic acid (SA), methyl viologen (MV), NaCl and drought stress treatment. Silencing of CaMlo2 could induce resistance against Xcv in pepper, while overexpression of CaMlo2 in Arabidopsis resulted in enhanced susceptibility to Pseudomonas syringae pv. tomato and Hyaloperonospera arabidopsidis.
In this study, we show that the tomato Slmlo1 mutant which is fully resistant to O. neolycopersici, is also partially resistant to L. taurica. In pepper, the newly isolated CaMlo2 gene was transcriptionally responsive to the penetration of L. taurica and silencing of either CaMlo1 or CaMlo2 reduced the susceptibility of pepper to L. taurica. Overexpression of CaMlo2 could restore the susceptibility of the tomato Slmlo mutant ol-2 to O. neolycopersici, and increase its susceptibility to L. taurica. We provide evidence indicating that at least one pepper Mlo homologue is involved in susceptibility to L. taurica.

L. taurica is sensitive to mlo-based resistance in tomato
Previously, we reported that loss-of-function in the SlMlo1 gene in tomato causes resistance against the powdery mildew pathogen O. neolycopersici [17]. In order to investigate whether L. taurica was principally amenable to mlo-based resistance, tomato lines carrying different alleles of the SlMlo1 gene were challenged with L. taurica. Plants of the breeding line ol-2 carrying the mutant allele of SlMlo1 showed less symptoms and a significant decrease in fungal DNA amount compared to plants of the susceptible control Moneymaker (MM) (Figure 1). In contrast, plants of the SlMlo1 overexpression line ol2::SlMlo1 were more susceptible and showed a marked increase in fungal DNA amount compared to MM plants. It could be inferred that SlMlo1 over-expression enhances the susceptibility to L. taurica in tomato and that loss-of-function in SlMlo1 results in partial resistance against L. taurica. The resistant control cv. Laurica, harbouring the Lv gene, showed no fungal sporulation and a clear HR (hypersensitive response) phenotype upon L. taurica infection (Figure 1). No HR could be seen macroscopically in plants of the breeding line ol-2, indicating that the resistance in these plants is likely not conferred by the Lv gene ( Figure 1). These data collectively indicate that in tomato L. taurica is sensitive to mlo-based resistance.

Isolation of the CaMlo2 gene in pepper
The sensitivity of L. taurica to mlo-based resistance in tomato promotes the possibility to use loss-of-function mutations in pepper Mlo ortholog(s) for resistance to L. taurica. When we started our work, CaMlo1 was the only identified Mlo homologue in pepper and considered as the ortholog of SlMlo1 [18,30]. By BLAST analysis, we inspected the publicly available pepper expressed sequence tag (EST) collections from the SOL Genomics Network (SGN) and identified one EST sequence (SGN-U202700) that is distinct from CaMlo1. The full-length cDNA sequence of this EST was obtained from cultivar Maor by Rapid Amplification of cDNA Ends (RACE), which was designated as CaMlo2. In the process of submitting this paper another report was published [35] in which the identification of another CaMlo2 allele was described (GenBank accession number JN896629). Furthermore, in the newly released pepper transcriptome database [36] a CaMlo2 sequence is present (accession number JW054099). The sequences of these three alleles were aligned ( Figure S1A). The cDNA sequence of the Maor CaMlo2 allele proved to be identical to sequence JW054099, while three SNPs were present compared with the JN896629 sequence. Protein sequences of all three CaMlo2 alleles were identical. The CaMlo2 cDNA sequences are clearly distinguishable from the CaMlo1 cDNA sequences (Figure S1B). Phylogenetic analysis comprising all Arabidopsis as well as four tomato and the two pepper MLO protein sequences, and MLO protein sequences from additional plant species (Figure 2), revealed that the newly isolated CaMLO2 is closer to SlMLO1 than CaMLO1. Figure 3 shows the alignment of MLO proteins from Arabidopsis, tomato, pepper and barley. The C-terminal D/E-F-S/T-F tetra-peptide sequence, one of several motifs characteristic of barley Mlo orthologs [30], is outlined in grey dotted lines. For this tetra-peptide sequence CaMLO2 contains D-F-T-F, which is identical to SlMLO1, while CaMLO1 contains a different sequence (E-F-S-F). This supports the assumption that CaMlo2 represents the pepper ortholog of SlMlo1.

Accumulation of CaMlo transcripts upon L. taurica infection
Previously, we showed that the expression of SlMlo1 is induced in tomato at early time points upon challenge with O. neolycopersici [17]. Data obtained by Piffanelli  In order to correlate the expression with the infection process, we consecutively tracked the infection progress of L. taurica on the pepper leaves with a digital microscope (Video S1) and performed qRT-PCR with more detailed time points ( Figure 5A and Figure S2). Based on the infection process of L. taurica, we could cluster time points into four stages. The first stage (B1) corresponds to the period from 0 to 3 hpi. In this stage fungal spores landed on the leaf surface and started the germination, defined as fungal germination stage ( Figure 5B-1). During this stage both CaMlo genes did not significantly change their expression level. The second stage (B2) includes four time points from 5 to 25-hpi. During this period, most spores formed a primary adhesion body (around 5 hpi), primary hyphae and secondary adhesion bodies (after 7 hpi). The adhesion bodies helped the fungus fix itself on the leaf surface and primary and secondary (branched) hyphae grew into the stomata ( Figure 5B-2). This stage is defined as stomata penetration stage in this study. During this stage, the expression of CaMlo2 was obviously induced at 5 hpi (the highest), 7 hpi and 25 hpi; in contrast, the expression of CaMlo1 did not significantly vary. The third stage (B3), defined as fungal growing stage, includes four time points from 30 to 96 hpi. During this , homozygous for the wild-type SlMlo1 allele; breeding line ol-2, homozygous for a Slmlo1 loss-of-function allele; cultivar Laurica, carrying the Lv gene for hypersensitive responsebased resistance. Yellow patches on the adaxial surface of ol-2::SlMlo1, MM and ol-2 leaves are due to L. taurica colonization and correspond to abaxial fungal sporulation. Necrotic lesions on the Laurica genotype are the consequence of Lv-gene-mediated hypersensitive response and do not correspond to fungal growth on the abaxial side of the leaves. Pictures were taken four weeks after fungal inoculation. The experiment was carried out twice yielding similar results. B. Fungal DNA quantification by real-time PCR on the same genotypes described above. Amount of L. taurica DNA was normalized by the plant reference gene elongation factor (SlEf) with the DDCt method. The relative pathogen biomass in MM plants is set as 1. In total, three leaves (4 th , 5 th and 6 th leaf) per plant were pooled together, and three plants per genotype were assayed. Bars indicate the standard error of the mean for three biological replicates. Asterisks indicate significant difference with the control cultivar MM, performed by Student's t-test. doi:10.1371/journal.pone.0070723.g001 stage, intercellular growth (including haustorium formation in mesophyll cells) of L. taurica took place ( Figure 5B-3) and the expression of CaMlo genes was quantified to be similar to the expression before fungal inoculation (with the exception of CaMlo2 30 hpi) ( Figure 5A). The last time point is 21 dpi (stage B4), when L. taurica finished the whole life cycle and released the next generation spores ( Figure 5B-4). Both CaMlo1 and CaMlo2 showed the highest expression at this time point. Taken together, these data show that the newly isolated CaMlo2 gene is responsive to L. taurica at the stomata penetration stage, whereas CaMlo1 expression is largely independent from the infection process, at least till four days after fungal inoculation. The gene expression levels were measured with two different primer pairs located in different regions of each gene and similar results were obtained.

Silencing of CaMlo genes in pepper by VIGS
In order to evaluate the potential role of the CaMlo genes in conferring powdery mildew susceptibility in pepper, we performed functional analysis via VIGS. In total, four constructs were generated for CaMlo1 and CaMlo2 silencing (two constructs per gene), namely VIGS:CaMlo1-a, VIGS:CaMlo1-b, VIGS:CaMlo2a and VIGS:CaMlo2-b. The sequences used for specific silencing of CaMlo1 or CaMlo2 are highlighted in Figure S1B. Plants infiltrated with the TRV2 empty vector (EV) were used as control. All EV-infiltrated plants showed symptoms when challenged with L. taurica, indicating that the TRV-based VIGS system does not influence the susceptibility of pepper to this fungus. Before fungal inoculation, CaMlo gene expression levels in plants infiltrated with the silencing constructs were assayed via qRT-PCR. For each construct except VIGS:CaMlo1-a, significant silencing of the target gene was observed ( Figure 6A). However, silencing of CaMlo1 in VIGS:CaMlo1-b plants was accompanied by a lower expression of CaMlo2, although this reduction was not significant. Similarly, a non-significant reduction of CaMlo1 expression was observed in VIGS:CaMlo2 plants ( Figure 6A). Compared with the EV control, plants silenced with each of the constructs except VIGS:CaMlo1-a showed a significant reduction of fungal colonies (visually measured) as well as fungal biomass (quantified by realtime PCR) ( Figures 6B and C). In general, a stronger reduction was observed in VIGS:CaMlo2 plants compared to VIG-S:CaMlo1 plants, suggesting that the newly isolated CaMlo2 gene may play a major role in susceptibility to L. taurica. Two independent VIGS assays were performed and similar data were obtained. It is worth to note that, in the VIGS experiments, eight out of 10 VIGS:CaMlo2-a plants showed a remarkable decrease in size when compared with plants infiltrated with other VIGS constructs ( Figure S3).

Complementation of SlMlo1 loss-of-function mutant by CaMlo2
To further confirm the role of CaMlo2 in powdery mildew pathogenesis, CaMlo2-overexpressing transgenic plants (35S::CaMlo2) were obtained by transforming the tomato Slmlo1 mutant (the ol-2 mutant). First, we tested cuttings of the T1 transformants with O. neolycopersici. Results showed that overexpression of CaMlo2 could restore susceptibility of Slmlo1 mutant to O. neolycopersici (Figure 7), providing further evidence for the role of CaMlo2 as a powdery mildew susceptibility factor. However, a fully restored susceptibility was not observed since the T1 transformants showed a significantly lower level of fungal biomass compared to cultivar Moneymaker (Figure 7), in addition to a 5-days delay of showing visible symptoms.
T2 seeds were obtained from transformants 1 and 6, showing the highest CaMlo2 expression level ( Figure 7A). T2 plants, together with the ol-2 mutant, were assayed for L. taurica susceptibility ( Figure 8A). As the T2 families segregate for presence of the 35S::CaMlo2 gene insertion(s), the CaMlo2 expression level of each T2 plant was measured by qRT-PCR. Figures 8B and 8C show that for both T2 families 1 and 6 the T2 plants showing expression of CaMlo2 have a significantly higher L. taurica biomass than the ol-2 plants. Therefore, we can conclude that CaMlo2 plays a role in susceptibility to both the epiphytic powdery mildew O. neolycopersici and the endophytic powdery mildew L. taurica.

Discussion
Previous studies documented that Mlo homologs are conserved across monocot and dicot plant species, and that mutations in certain Mlo homologs (e.g. barley Mlo, Arabidopsis AtMlo2, pea PsMlo1 and tomato SlMlo1) result in resistance to epiphytic powdery mildew species [16,26,27,28,29]. However, this is the first study that shows that Mlo homologs can be also involved in susceptibility to an endophytic powdery mildew fungus.
In tomato, we found that SlMlo1 overexpression increased susceptibility to L. taurica and in agreement with this the Slmlo1 mutant was resistant to this fungus. However, compared with the full resistance to O. neolycopersici, the Slmlo1 mutant was only partially resistant to L. taurica, In barley and pea, the mutation of a single Mlo gene suffices to confer full resistance against the adapted powdery mildews B. graminis f. sp. hordei and E. pisi, respectively. In contrast, in Arabidopsis, full penetration resistance to the adapted powdery mildew species G. cichoracearum and G. orontii is observed upon simultaneous loss-of-function of three Mlo homologues (AtMlo2, AtMlo6 and AtMlo12). Here our data indicate that, for the plant species tomato, full resistance to L. taurica may require mutations in other Mlo genes in addition to SlMlo1. Recently, the tomato genome sequence has become available from The Tomato Genome Consortium (http://solgenomics.net/organism/ Sola-num_lycopersicum/genome) [38], which provides us with the chance to perform a functional analysis of mildew-effective SlMlo paralogs to discover other tomato Mlo homologues which are potentially involved in the interaction between tomato and L. taurica.
In pepper, we isolated the CaMlo2 gene and demonstrated that its expression was induced at the fungal penetration stage of L. taurica. Studies in tomato, barley and Arabidopsis have documented that mildew-effective Mlo gene(s) respond to fungal penetration at very early time points [16,17,18,21,32,37]. In barley and tomato Mlo transcripts significantly increase 6 hours post inoculation (hpi) [17,32]. Though both CaMlo1 and CaMlo2 transcripts accumulated during L. taurica infection, only CaMlo2 transcripts showed a significant up-regulation at early time points (5 and 7-hpi), corresponding to the stage of first adhesion body formation. In contrast, CaMlo1 was found to be significantly upregulated only at 21 dpi (Figures 4 and 5, and Video S1). Although we did not analyze CaMlo1 and CaMlo2 expression levels between 4 dpi and 21 dpi we do not expect major changes during this interval, based on our study of the infection process of L. taurica on pepper [40]. We observed that L. taurica hyphae penetrated pepper leaf tissue starting from 6 hours post inoculation (hpi). Primary hyphae penetrated leaf tissue within 24-hpi, whereas tips of secondary branched hyphae penetrated the leaf 30-48 hpi. Conidiophores emerged from stomata 14-16 dpi. Subsequently, mature conidia became the source of secondary infection around 21 dpi. In tomato inoculated with O. neolycopersici SlMlo1 is induced only at early time points (6-36 hpi) [17]. Thus, it is expected that the Mlo Figure 3. Multiple amino acid sequence alignment of pepper CaMLO1 and CaMLO2, barley HvMLO, tomato SlMLO1, as well as Arabidopsis AtMLO2, AtMLO6, and AtMLO12. The alignment was generated by CLUSTALW using default parameters. Grey boxes indicate similar amino acids and black boxes indicate identical amino acids. The C-terminal D/E-F-S/T-F tetra-peptide sequence, one of several motifs characteristic of barley Mlo orthologs [30] is outlined in grey. For this tetra-peptide sequence, CaMLO2 contains D-F-T-F, which is identical to SlMLO1. doi:10.1371/journal.pone.0070723.g003   susceptibility gene should be expressed at the time points of fungal penetration stage. Therefore, in pepper inoculated with L. taurica we measured expression at time points between 0 hpi and 4 dpi to monitor primary infection, and in addition at 21 dpi to monitor secondary infection.
Together with the evidence obtained in phylogenetic analysis, the transcriptional response to fungal infection suggests that CaMlo2 is likely the ortholog of SlMlo1. CaMlo1 shows highest homology to a different SlMlo gene that is not involved in tomato susceptibility to O. neolycopersici (unpublished results). If CaMlo1 also responds at transcriptional level to L. taurica, it is likely to occur at later infection points during the intracellular fungal growing stage.
Surprisingly, knocking-down the expression of either CaMlo1 or CaMlo2 resulted in reduced susceptibility of pepper to L. taurica, suggesting that both homologues are involved in the susceptibility of pepper to this fungus. The alignment of CaMlo1 and CaMlo2 coding sequences ( Figure S1B) shows there is no continuous stretch of 21 identical nucleotides present, which is the minimal requirement for efficient silencing. Still, we cannot completely rule-out co-silencing effects ( Figure 6). Overexpression of CaMlo2 restored the susceptibility of the tomato Slmlo1 mutant to O. neolycopersici, and increased the susceptibility to L. taurica, thereby further indicating that CaMlo2 is a pepper susceptibility gene towards different powdery mildew species (Figures 7 and 8). Also, it supports the assumption that CaMlo2 is very likely the functional ortholog of SlMlo1. However, a more precise assessment of the biological role of CaMlo1 and CaMlo2 in the interaction with L. taurica needs future functional analyses. Currently, we are performing transformations to overexpress CaMlo1 in the Slmlo1 mutant and we are planning to retransform CaMlo2-expressing T2 plants with a CaMlo1 overexpression construct. In case overexpression of CaMlo1 increases susceptibility of the Slmlo1 mutant to powdery mildew we plan to monitor the expression of CaMlo1 during the intercellular fungal growing stage of L. taurica, i.e. between 4 and 21 dpi. With these materials and functional analysis, we hope to unravel the involvement of both CaMlo1 and CaMlo2 in the interaction between tomato or pepper and L. taurica. Meanwhile, we are carrying out an allele mining approach to search for natural mutations in CaMlo genes using publicly available germplasm.
O. neolycopersici is epiphytic and develops all structures except haustoria on the host surfaces. In contrast, L. taurica is endophytic because fungal hyphae grow intercellularly in the mesophyll. For complete resistance against O. neolycopersici, mutation of one Mlo homologue is sufficient [39]. In contrast, for complete resistance against L. taurica, our data suggest that in both tomato and pepper mutations in more than one Mlo gene may be required, as it is reported for Arabidopsis. A recent study showed that silencing of the CaMlo2 gene could induce resistance against Xanthomonas campestris pv. vesicatoria in pepper [35]. Thus, the specificity or redundancy of plant MLO functions in susceptibility to pathogens in different pathosystems requires careful attention.
In general, resistances against plant diseases are achieved by introduction of R-genes into susceptible plant genotypes. With the recently proposed new S-genes breeding strategy, resistance could be achieved by impairment of susceptibility factors that evolved in the disease development [13,14]. With increasing interest in the research topic on suppression of plant immunity, a considerable number of potential S-genes have been identified in Arabidopsis. However, it is largely unknown whether orthologs of Arabidopsis Sgenes in crop species exist and are functional to corresponding crop pathogens. Taking Mlo as the target gene, this study shows that there are functional orthologs present in tomato and pepper, demonstrating the potential use of S-genes identified in Arabidopsis in breeding crops with durable resistance.
However, pleiotropic effects of the achieved resistance by disabling S-genes, as observed when silencing CaMlo2, should also be considered for practical applications. By comparing features of Mlo-like genes of several monocot species, the dicot Arabidopsis and the moss Ceratodon purpureus, it is assumed that the origin of the Mlo gene family can be tracked back at least to the early evolutionary stages of land plant development [28]. The ancient presence of plant Mlo gene families implies their vital function for plant development. In addition to being a negative regulator of plant defence, functions have been discovered for other members of the plant Mlo gene family. Barley and Arabidopsis mlo mutants showed some pleiotropic effect in older plants and under certain conditions, suggesting that mildew-effective MLOs may be involved in other biological functions as well [13]. However, the tomato Slmlo1 mutant did not show an abnormal plant phenotype [17]. Thus, the functions of members of the Mlo gene family need to be further characterized.

Plant materials
Four tomato genotypes were used in this study: a breeding line ol-2, a homozygous T3 ol-2 line (35S::SlMlo1), Solanum lycopersicum cv. Moneymaker (MM) and cv. Laurica. The ol-2 line carries the mutated allele (a 19-bp deletion in the coding sequence) of SlMlo1 and is resistant to O. neolycopersici [17]. Homozygous T3 ol-2 line is the selfed progeny of a transgenic T2 plant over-expressing the SlMlo1 gene under the 35S promoter (35S::SlMlo1) in the ol-2 line and fully susceptible to O. neolycopersici [17]. MM carrying the natural SlMlo1 allele was used as the susceptible control. Cultivar Laurica was used as the resistant control which carries the wildtype SlMlo1 allele and the Lv gene derived from S. chilense LA1969.
Three pepper genotypes, cv. Maor and two anonymous cultivars A and B, were used which are all susceptible to L. taurica. The doubled-haploid line HV-12 (kindly provided by Dr.Alain Palloix, INRA, France) was used as the resistant control.
Leveillula taurica disease assay L. taurica was obtained in the greenhouse of a seeds company from The Netherlands and maintained on susceptible pepper plants in a climate chamber at Wageningen University, The Netherlands. The inoculation was performed on six weeks-old plants by a spray method. Conidiospores of L. taurica were washed off from heavily infected pepper leaves with water. Inoculum was adjusted to a final concentration of 2.5610 4 conidia/ml. The abaxial surface of plant leaves was spray-inoculated. After the inoculation, pepper plants were kept for 24 h in plastic cages with 100% relative humidity at 21uC and subsequently with a temperature of 21uC (day) / 19uC (night) and relative humidity of 65%. Inoculated tomato plants were kept in a greenhouse compartment without plastic cages, and thus without increasing the relative humidity to 100%.
For the analysis of CaMlo1 and CaMlo2 gene expression in pepper during infection with L. taurica two independent inoculations were performed. In the first one four genotypes (HV-12, Maor, cultivar A and cultivar B) were assayed, with two plants per genotype at five time points. In the second experiment three genotypes (Maor, cultivar A and cultivar B) were assayed, with three plants per genotype at 12 time points (including the five time points of the first experiment). For the disease assay of tomato transformants expressing CaMlo2 in a Slmlo1 mutant background (ol-2) T2 plants obtained from the self-pollination of T1 transformed plants 1 and 6 ( Figure 7A) were used. For each of the two families15 plants were inoculated, together with 10 ol-2 plants.
For scoring L. taurica symptoms, numbers of fungal colonies on the 4 th , 5 th and 6 th true leaves were counted. Furthermore, Disease Index (DI) was used to score the sporulation of L. taurica on each infected leaf as described in [40]. This DI system was based on a scale of 0 to 5; 0 = no visible sporulation, 1 = restricted chlorotic spots on the adaxial leaf surface with weak or no sporulation on the corresponding abaxial leaf areas, 2 = several isolated sporulation spots on the abaxial leaf area, 3 = numerous sporulation spots covering up to 40-50% of the abaxial leaf area, 4 = numerous coalescent sporulation spots covering up to 75% of the abaxial leaf area, and 5 = the whole abaxial surface of the leaf, and also parts of the adaxial leaf surface covered with dense sporulation. The third, fourth and fifth true leaves were scored for each plant, for both tomato and pepper.
In addition, a real-time PCR method was used to quantify the fungal biomass. Internal Transcribed Spacer (ITS) sequences of L. taurica infecting pepper were retrieved from GenBank and specific primers LV-F (59AGCCGACTAGGCTTGGTCTT39) and LV-R (59 GCGGGTATCCCTACCTGATT39) were designed by using the online software: Primer3 plus (http://www. bioinformatics.nl/cgi-bin/primer3plus/primer3plus. cgi). The DNA was isolated from infected leaves. For tomato, Ef primers SlEF-F (59GGAACTTGAGAAGGAGCCTAAG39) and SlEF-R (59CAACACCAACAGCAACAGTCT39) [41] were used as reference to normalize the plant DNA proportion by DDCt methods [42]. For pepper, CaActin primers CaActin-F (59ATCCCTCCACCTCTTCACTCTC39) and CaActin-R (59GCCTTAACCATTCCTGTTCCATTATC39) [43] were used as a reference.

RNA/DNA isolation, RACE and sequence analysis
RNA and DNA were isolated by using the RNeasy Kit and DNeasy Kit (Qiagen, Germany) respectively, according to the manufacturer's recommendations. The Rapid Amplification of cDNA Ends (RACE) reaction was conducted using the GeneRacer TM kit (Invitrogen). For obtaining the 59 Untranslated Region (UTR) of the CaMlo2 cDNA primer 59R (59TCTCTACAG-TAATGAACTCGAGACAAA39) was used together with the GeneRacer TM 59primer (59CGACTGGAGCACGAGGA-CACTGA39). For obtaining the 39UTR primer 39F (59CGTGGGAATAAGTCCAGCAT39) was used together with the GeneRacer TM 39primer (59GCTGTCAACGATACGC-TACGTAACG39). Amplicons for sequencing were generated from 50 ng genomic DNA template. PCR amplifications were performed in 20 ml reactions using 1 u of Taq Polymerase, 1x reaction buffer, 200 nM dNTP and 250 nM of each primer. Standard cycling conditions were: 4 min initial denaturation at 94uC, followed by 35 cycles of 30 sec denaturation at 94uC, 30 sec at the appropriate annealing temperature, and 30-60 sec extension at 72uC. Reactions were finished by 7 min incubation at 72uC. PCR products were examined for quality on ethidium bromide or GelRed TM -stained agarose gels. PCR products were directly sequenced on ABI377 or ABI3700 sequencers (Greenomics, Wageningen UR) using the dideoxy chain-termination method and ABI PRISM Reaction Kit. One of the amplification primers was used as sequencing primer. Sequences were aligned using CLUSTALW. Visualization of the alignment was performed using BOXSHADE 3.21 (http://www.ch.embnet.org/software/ BOX_form.html). A phylogenetic tree was constructed using the ''One Click'' method of Phylogeny.fr [44] on http://www. phylogeny.fr/version2_cgi/index.cgi. The radial view was obtained by TreeDyn.

Gene expression analysis
For gene expression analysis, the first experiment was conducted using a semiquantitative RT-PCR evaluating the expression of CaMlo1 and CaMlo2 at five time points after L. taurica inoculation: 0 hpi (hours post inoculation), 5 hpi, 25 hpi, 72 hpi and 21 dpi (days post inoculation). The 3 rd , 4 th and 5 th leaves were collected from two individual plants of four different genotypes (HV12, Maor, cultivar A and cultivar B), and pooled per plant. cDNA was generated from 1 mg of total RNA using a iScript cDNA synthesis kit from Bio-Rad.
In a second independent inoculation experiment the 3 rd , 4 th and 5 th leaves were collected from three individual plants of three different cultivars (cultivar A, cultivar B and Maor) at twelve time points: 0 hpi (hours post inoculation), 1 hpi, 3 hpi, 5 hpi, 7 hpi, 21 hpi, 25 hpi, 30 hpi, 47 hpi, 72 hpi, 96 hpi and 21 dpi (days post inoculation) after L. taurica inoculation. cDNAs was synthesized as described above. Quantitative real-time PCR was performed by using the SYBR H Green dye on Bio-Rad iCycler iQ H machine (Bio-Rad).
Primer sequences of CaMlo1 and CaMlo2 and the reference genes are shown in Table S1. Five pepper housekeeping genes were tested for expression stability in order to determine which could be used as a reference gene in qRT-PCR. These include CaActin [43], Ubiquitin-conjugating protein (UBI-3) [45], Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) [45], Elongation factor (Ef1a) [46], and Ubiquitin Extension Protein (UEP) [46]. Gene expression stability was essayed with the BestKeeper program [47]. CaActin, CaEF1a and CaUEP proved to be most stable. These three reference genes were used in qRT-PCR analyses of two pepper cultivars A and B at twelve different time points after inoculation. Gene expression levels at different time points were normalized by DDCt methods. All three reference genes yielded similar results for the two tested cultivars ( Figure S2). Therefore, each of them could be used as a suitable reference gene in subsequent experiments.

Virus Induced Gene silencing (VIGS) in pepper
The VIGS experiments were performed as described by Liu et al. [48]. Primer pairs (Table S2) were designed to amplify fragments suitable for specific silencing of CaMlo1 or CaMlo2 by using the online software Primer3 plus. Target fragments of CaMlo1 were amplified and cloned into the pGEMH-T Easy vector (Promega). Positive plasmids were digested with EcoRI and target fragments were excised from an agarose gel. The pTRV2(pYL156) vector [37] was digested with EcoRI and dephosphorylation was performed by using Thermosensitive Alkaline Phosphatase (TSAP) (Promega). Target fragments of CaMlo1 were ligated by T4 ligase into the linearized pTRV2 vector. Positive clones were confirmed by sequencing. Target fragments of CaMlo2 were cloned into the Gateway-compatible vector pENTR D-TOPO (Invitrogen) and subsequently recombined into pTRV2-Gateway VIGS vector. The cloned sequences of CaMlo1 and CaMlo2 are highlighted in Figure S1B.
The pTRV2 vectors carrying the target gene fragment were transformed into Agrobacterium strain GV3101 by electroporation. A 100-mL culture of Agrobacterium containing the target vectors was grown overnight at 28uC in YEP (yeast extract/bactopeptone) medium with antibiotics (50 mg/mL kanamycin and 50 mg/mL rifampicin). The cells were resuspended into infiltration medium MMA (150 mM acetosyringone, 10 mM MgCl and 10 mM MES, pH 5.7) with OD600 = 2. Cultures were kept at room temperature for 1 to 6 hours before agroinfiltration. Agrobacterium strains containing the pTRV1 vector and pTRV2 were mixed at a 1:1 ratio and co-infiltrated into two weeks-old cotyledons of pepper. For each pTRV2 construct 10 pepper plants of cv. A were infiltrated. Infiltrated plants were grown at 22uC with a 16 h-light/ 8 h-dark photoperiod cycle. Three weeks after agroinfiltration, three of the ten plants were used for assessing the silencing effect by real-time PCR with the primers CaMlo1-ge-F and CaMlo1-ge-R for CaMlo1 and primers CaMlo2-ge-F and CaMlo2-ge-R for CaMlo2 (see above). The remaining seven plants were inoculated with L. taurica. Powdery mildew symptoms were observed three weeks after inoculation. Tested plants were scored for DI, and numbers of powdery mildew colonies were counted on the 4 th , 5 th and 6 th leaves. After colony counting, these three leaves per plant were collected and pooled for DNA isolation and fungal quantification with qRT-PCR using primers LV-F and LV-R (see above).

Consecutive digital micrographs by a dissecting microscope
Conidia were collected from conidiophores using a probe of an electrostatic spore collector and were transferred to particular sites of test leaves [49]. Conidial growth on host epidermal cells were photographed at 0.5-1 hour intervals after inoculation using a CCD-camera of a high-fidelity digital microscope KH-2700 (Hirox, Tokyo, Japan). Photographs were treated using an image processing software (Adobe Photoshop ver.5) (Adobe Systems, CA, USA), and 110 image-analyzed photographic data were input to a Windows live movie maker software (Microsoft, WA, USA) to present the animated data of the conidial development.

Generation of transgenic ol-2 tomato plants expressing the CaMlo2 gene
In order to amplify and clone the full length coding sequence of the pepper CaMlo2 gene four different primer pairs were designed. An amplicon of the expected size was only obtained for CaMlo2 with primers CaMlo2_4fw (59caccATGGAGGCAACCCC-TACGTGG39) and CaMlo2_4rev (59 CTATTTGTTTC-CAAAAGTAAAATCTGACATT39). This amplicon was cloned into Gateway-compatible vector pENTR D-TOPO (Invitrogen). Cloning reactions were performed in E. coli strain One ShotH TOP10 according to the manufacturer's instructions. Presence of the right fragment was assessed by colony PCR, restriction enzyme digestion and sequencing. Then, the amplicon was transferred by LR recombination to the binary vector pK7WG2 [50], which harbours a 35S Cauliflower Mosaic Virus (CaMV) promoter and the marker gene for kanamycin resistance nptII, following the manufacturer's instructions (Invitrogen). Recombinant plasmids were cloned into E. coli One ShotH TOP10. Bacterial colonies were screened for positive recombinant plasmids by colony PCR, restriction enzyme digestion and sequencing. Binary vectors containing the expected insert were subsequently transferred into the AGL1-virG strain of Agrobacterium tumefaciens [51] by electroporation. Transformation of the tomato ol-2 mutant was performed according to [52]. The obtained T1 transformants and their corresponding T2 families were assessed for the expression of the target gene and the presence of the nptII marker gene by real-time qPCR, using the following primer pairs: CaMlo2_qPCR.2_fw (59  TCACCTTGGAGAGTGGTTGTTG39) with CaM-lo2_qPCR.2_rev (59GCGCAATTGC CAACACAAAG39) and nptII_fw (59ACTGGGCACAACAGACAATC39) with nptII_rev (59 TCGTCCTGCAGTTCATTCA G 39). As housekeeping gene the elongation factor a was used (primers SlEF-F and SlEF-R).

Disease assay with O. neolycopersici
Cuttings originating from 26 T1 transgenic plants (3 cuttings per T1 plant) were inoculated with the Wageningen isolate of O. neolycopersici, which has been maintained on cv. Moneymaker (MM) as described in [39]. Pathogen inoculation was carried out by using a inoculum suspension of 2.5610 4 conidia/ml. Three weeks after inoculation the disease severity was measured by realtime qPCR quantification of O.neolycopesici biomass (On). The infected leaves (the 2 nd or 3 rd leaf) sampled for each cutting. Plant and fungal DNAs were extracted by using the DNeasy DNA extraction kit (Qiagen, Germany). In total, 20 ng of DNA was used as template for amplification with the primer pair ON-F (59-CGCCAAAGACCTAACCAAAA-39) and ON-R (59-AGCCAA-GAGATCCGTTGTTG-39), designed on On-specific ITS sequences (GenBank accession number EU047564). The SlEF-F + SlEF-R primer pair was used as reference to normalize the plant DNA proportion by the DDCt method.
Two T1 transformants (plants 1 and 6) showing the highest CaMlo2 expression level and the most susceptible phenotype to O. neolycopersici (Figure 7), were selected to be self-pollinated. The resulting T2 families were used in a L. taurica assay as described in a previous paragraph. Figure S1 A. Sequence alignment of cDNAs of different alleles of the CaMlo2 gene. The alignment was generated by CLUSTAL 2.1 using default parameters. CaMlo2_cDNA_Maor indicates the cDNA sequence isolated from this study, JW054099 is a transcript from the pepper transcriptome database [36], and JN896629 indicates the cDNA identified in the study of Kim and Hwang [35]. Start and stop codons are indicated in green and red, respectively. Nucleotide differences are indicated in blue. B. Sequence alignment of coding sequences of CaMlo1 cDNAs AY934528 and JW061356 and CaMlo2 cDNAs JN896629 and JW054099. Identical nucleotides are boxed in black. Sequences used for VIGS experiments are highlighted in green and blue for CaMlo1 and yellow for CaMlo2. Primers used for qRT-PCR are indicated with blue arrows for CaMlo1 and yellow arrows for CaMlo2. (DOCX) Figure S2 Expression profile of CaMlo genes measured by real time qRT-PCR in pepper leaves upon Leveillula taurica infection using three different reference genes. A, cultivar A. B, cultivar B. Columns indicate transcript fold changes with respect to noninoculated plants (0 hours after inoculation (hpi)). Relative quantification was performed by using the DDCt method and the reference genes CaActin, CaEF1a and CaUEP. Samples were taken from three whole pepper leaves per plant (the 3 rd , 4 th and 5 th leaf) upon L. taurica infection at 0 hpi, 1 hpi, 3 hpi, 5 hpi, 7 hpi, 21 hpi, 25 hpi, 30 hpi, 47 hpi, 72 hpi, 96 hpi and 21 days post inoculation (dpi). Results are based on three individual pepper plants per time point. Bars refer to standard errors of the biological replicates and asterisks refer to significant differences with respect to non-inoculated plants, inferred by mean comparisons by Student's t-test. (TIF) Figure S3 Eight out of ten plants in which the CaMlo2 gene is silenced with the VIGS construct CaMlo2-a (left) show an obvious decrease in size compared with plants in which CaMlo1 is silenced (right).

(TIF)
Video S1 Infection progress of Leveillula taurica on pepper leaves. Germination of a primary conidium of L. taurica on the abaxial surface of a young pepper leaf (0 to 3 hours post inoculation, hpi), production of primary adhesion body followed by growing of primary hypha until the penetration of a stoma (3 to 24 hpi), and production of secondary adhesion bodies followed by emergence and growing of branched hyphae until penetration of stomata (24 to 48 hpi). (WMV)