Knockdown of MLO genes reduces susceptibility to powdery mildew in grapevine

Erysiphe necator is the causal agent of powdery mildew (PM), one of the most destructive diseases of grapevine. PM is controlled by sulfur-based and synthetic fungicides, which every year are dispersed into the environment. This is why PM-resistant varieties should become a priority for sustainable grapevine and wine production. PM resistance can be achieved in other crops by knocking out susceptibility S-genes, such as those residing at genetic loci known as MLO (Mildew Locus O). All MLO S-genes of dicots belong to the phylogenetic clade V, including grapevine genes VvMLO7, 11 and 13, which are upregulated during PM infection, and VvMLO6, which is not upregulated. Before adopting a gene-editing approach to knockout candidate S-genes, the evidence that loss of function of MLO genes can reduce PM susceptibility is necessary. This paper reports the knockdown through RNA interference of VvMLO6, 7, 11 and 13. The knockdown of VvMLO6, 11 and 13 did not decrease PM severity, whereas the knockdown of VvMLO7 in combination with VvMLO6 and VvMLO11 reduced PM severity up to 77%. The knockdown of VvMLO7 and VvMLO6 seemed to be important for PM resistance, whereas a role for VvMLO11 does not seem likely. Cell wall appositions (papillae) were present in both resistant and susceptible lines in response to PM attack. Thirteen genes involved in defense were less upregulated in infected mlo plants, highlighting the early mlo-dependent disruption of PM invasion.


INTRODUCTION
Vineyards are treated with an impressive amount of chemical compounds, particularly fungicides, to prevent yield losses due to fungal pathogens. In France, Italy, Spain and Germany, between 1992 and 2003, 73% of the fungicides were used for grapevine protection, a crop that covers only 8% of the agricultural land in those countries. 1 Worldwide, grapevine powdery mildew (PM), caused by the fungus Erysiphe necator, is a destructive disease. 2 E. necator Schw. (syn. Uncinula necator (Schw.) Burr.) is an obligate biotroph infecting all green tissues of grapevine and results in significant losses in yield and berry quality. Symptoms are a white or gray powder covering both leaf surfaces, and, after infection, the fruits show shriveling or cracking. 3 The quality of the fruit is severely damaged, due to increased acidity and decreased anthocyanin and sugar content. 4 PM can be controlled by frequent applications of fungicides, particularly those based on sulfur. However, due to the ecological drawbacks of fungicides, 5 the relative high costs (up to 20% of total grapevine production expenses 6 ) and the rapid appearance of resistant strains of the pathogen 7-9 because of its adaptative gene copy-number variation, 10 new alternatives to chemical treatments should be adopted. Resistant varieties are one of the best options. The use of PM-resistant cultivars could reduce production costs in California by 720 $ ha − 1 , with a significant reduction of fungicide usage. 6 Vitis vinifera is susceptible to PM, 11 whereas North American Vitis species, due to their co-evolution with E. necator, have variable degrees of resistance to the pathogen. 12 Their resistances have been transferred to V. vinifera but the acceptance of resistant hybrids by producers and consumers has been very limited because of the attachment to traditions and lower quality of resulting wine, although resistant cultivar suitable for wine production are becoming available. 6 One strategy to create crops resistant to diseases is based on the exploitation of R-genes that encode proteins that recognize pathogen effectors and trigger defense response, 13 such as the Vitis REN and RUN genes. 14 Resistance manifests as localized hypersensitive responses at the sites of attempted infection. 15 However, R-genes are frequently overcome by mutations of the pathogen. 16 An alternative approach is based on susceptibility genes (S-genes), which loss of function results in recessively inherited resistance. 17 Knockout of S-genes may, however, induce pleiotropic phenotypes in the plant. 18,19 A typical class of S-genes is represented by the MLO (Mildew Locus O) genetic factors that, when inactivated, results in recessive mlo resistance, as discovered in barley. 20 MLO genes are largely conserved across the plant kingdom and their loss of function resulted in PM resistance in Arabidopsis thaliana, 21 pea, 18 tomato, 22 wheat 23 and pepper. 24 Of the seven phylogenetic clades in which the MLO family is divided, 25,26 only two include S-genes: clade IV with all monocot S-genes 27,28 and clade V with all dicot S-genes. 21,22,29,30 Not all members of clades IV and V are S-genes, but candidates can be identified during early stages of PM infection because of their increased expression, as documented in tomato, 22 barley, 31 pepper, 24 grapevine 29,30 and apple. 26 In grapevine, of four clade V MLO genes, three (VvMLO7, VvMLO11 and VvMLO13) are upregulated early after PM infection, whereas VvMLO6, the fourth, is not transcriptionally responsive to the pathogen. 29,30 MLOs are membrane proteins with seven transmembrane domains 32 involved also in a variety of physiological processes in different tissues, such as root morphogenesis 33 and pollen tube reception by embryo sac. 34 The proposed function of MLO S-genes is the negative regulation of vesicle-associated and actindependent defense pathways at the site of PM penetration. 27 The secretory vesicle traffic controls pathogen penetration, allowing the formation of cell wall appositions called papillae, 2,33 which are associated with mlo resistance. 21,36 The development of DNA-editing tools is rapidly changing plant genetics and biotechnology, due to the possibility of inducing mutations in specific genes. [37][38][39] Before adopting a gene-editing approach to knockout candidate S-genes, the evidence that loss of function of MLO genes can reduce PM susceptibility is necessary. This paper reports the knockdown through RNA interference of VvMLO6, 7, 11 and 13 and its effect on PM infection in grapevine.

MATERIALS AND METHODS Constructs
Three hundred-to 600-bp fragments of genes VvMLO6, VvMLO7, VvMLO11 and VvMLO13 were amplified (Supplementary Table S1) and cloned in pENTR/SD-TOPO (Invitrogen). After checking the sequences, the fragments were inserted in the RNAi Gateway vector pK7GWIWG2D(II) 40 (http://www. psb.ugent.be/), as in Urso et al. 41 After sequencing both the strands, the constructs were inserted in Agrobacterium tumefaciens strain GV3101, as in Zottini et al. 42 A. tumefaciens-transformed cells were tested by PCR for the presence of constructs, using primers annealing to the 35S promoter (5′-CGCACAATCCCACTATCCTT-3′) and the MLO fragment (Supplementary  Table S1).
Plant material, gene transfer and screening of regenerants Plant material of the grapevine cultivar long-cluster Brachetto was cultivated in vitro as described by Dalla Costa et al. 43 This cultivar was chosen because it is easy to transform, 44,45 has high somatic embryogenesis efficiency and good efficiency of regeneration from the callus. 46 Somatic embryos were used for gene transfer. Gene transfer, regeneration and selection of transgenic plants were performed as in Dalla Costa et al. 43 Five different gene transfers were carried out: four aimed to silence the four MLO target genes and the fifth was a control consisting of the empty vector (pK2WG7). DNA was extracted from in vitro leaf tissue (Phytopure kit, GE Healthcare, Little Chalfont, Buckinghamshire, UK). Integration was proven using the primers described above. Transformed in vitro grown lines were moved to a woody plant medium 47 in growth chamber at 20-24°C and transferred in fresh medium once a month.

Greenhouse acclimation
Plants were first acclimated to greenhouse conditions in a growth chamber at 25°C, 16-h day/8-h night, relative humidity (RH) 70 ± 5%. One-month-old plants with at least two main roots 3-cm long were transferred in a 250-ml plastic cup containing wet autoclaved turf (Tercomposti Spa, Brescia, Italy) and sealed with parafilm, to preserve humidity. Every 7 days, holes were made in the parafilm cover to progressively reduce air humidity and promote the formation of the foliar cuticle. After 3 weeks, parafilm was completely removed and, after 1 more week, the plants were transplanted in 1 l pots kept in the greenhouse at 25°C, 16-h day/8-h night, humidity 70 ± 5%. E. necator and P. viticola inoculation and disease severity assessment The E. necator and P. viticola inocula were obtained from infected leaves of an untreated vineyard in northern Italy (Trentino region). Subsequent reproduction of the inoculum was carried out infecting the V. vinifera cultivar Pinot Noir, under greenhouse conditions. For E. necator inoculation, plants were dry inoculated onto target leaves gently brushing adaxial leaf surfaces with infected young leaves carrying fresh PM sporulation. 48 Inoculated plants were incubated in a greenhouse at 25 ± 1°C with 100% RH for 6 h to promote fungal penetration, and then kept at 25 ± 1°C and 70 ± 10% RH until the end of symptom evaluation. Disease severity was assessed on all leaves at 14, 22 and 30 days post inoculation (d.p.i.).
For P. viticola inoculation, fresh sporangia were collected by washing the abaxial surfaces, carrying freshly downy mildew (DM) sporulating lesions, with distilled water at 4°C. Abaxial leaf surfaces were sprayed with the inoculum suspension of P. viticola (2 × 10 5 sporangia per ml). Inoculated plants were incubated overnight in the dark at 25°C with 99-100% RH, and then maintained under controlled greenhouse conditions at 25 ± 1°C and 70 ± 10% RH. Six days after inoculation, plants were incubated overnight in darkness at 25°C with 99-100% RH to allow DM sporulation and assess the disease severity.
Severity of PM and DM was assessed visually on all leaves of each plant, according to the standard guidelines of the European and Mediterranean Plant Protection Organization. 49,50 For each leaf, disease severity was expressed as the proportion (percentage of 0-100%, with intervals of 5%) of the leaf area covered by white sporulation of PM or DM in relation to the total leaf area, and a mean value was calculated for each plant. Disease reduction was calculated as (disease severity in control plants − disease severity in transformed plants)/(disease severity in control plants) and expressed as a percentage. For PM severity, the area under disease progress curve was considered as a quantitative summary of disease intensity over time to analyze all time points together 51,52 using the following formula: area under disease progress curve = ∑[( where X i corresponds to the disease severity (%) at assessment i, X i+1 corresponds to the disease severity (%) at subsequent assessment i+1 and (t i+1 − t i ) corresponds to the number of days between the two consecutive assessments. PM severity was also assessed as the number of E. necator conidia produced from infected leaves as in Angeli et al. 53 Three leaves were collected from each replicate at 30 d.p.i. and four disks of 0.8 cm diameter for each leaf were cut, for a total of 12 disks per replicate. Leaf disks were transferred to 50-ml tubes containing 5 ml distilled water with 0.01% Tween-80 (Sigma-Aldrich, St. Louis, MO, USA). Tubes were vortexed for 1 min and the concentration of conidia per ml was determined by a hemocytometer count. The values obtained were converted in conidia per cm 2 of grapevine leaf. Two inoculation experiments were carried out and in each experiment three to nine biological replicates (plants) per line were analyzed in a randomized complete block design.

Histological analysis
Two inoculated leaves were collected from three replicates of each transgenic and control line at 3, 10 and 21 d.p.i. for hyphae visualization and histological analyses. Leaves were treated as described by Vanacker et al. 54 with the following modifications: small pieces of leaf with the adaxial surface up were laid on filter paper moistened with ethanol:glacial acetic acid (3:1, v/v) until the chlorophyll was removed. Leaf segments were transferred to water soaked filter paper for 2 h, incubated in lactoglycerol [lactic acid:glycerol:water 1:1:1 (v/v/v)] for 12 h and stored in lactoglycerol at room temperature. For microscopic analysis, leaf segments were mounted on microscope slides and a drop of aniline blue (0.1% (w/v) in lactoglycerol) was pipetted on their surface. Aniline blue staining does not fluoresce when in lactoglycerol and E. necator structures (hyphae, conidia and appressoria) were visualized using the bright-field illumination of a Leica LMD7000 microscope (Leica Microsystems, Wetzlar, Germany). After spore localization, fluorescence was used with a LMD filter (BP filter 380-420-nm excitation, 415 dichroic mirror, and BP 445-485-nm emission) to visualize the bright blue-green auto-fluorescence associated with infected cells and papillae (autofluorogenic phenolic compounds) formation. 54

RNA extraction and gene expression analysis
The first gene expression analysis of transgenic plants was carried out on in vitro grown lines to identify genotypes with reduced expression of target genes. Three biological replicates were collected from each line. The second analysis was carried out on acclimated transgenic plants, with leaf samples collected before inoculation, 24 h and 10 days post PM inoculation, the time of the last two samplings corresponding to the upregulation of MLO genes after infection. 29,30 Five biological replicates Knockdown of MLO genes in grapevine S Pessina et al.
were collected from each line. For each line at each time point, the third and fifth half leaves from the top were collected, frozen in liquid nitrogen and stored at − 80°C.
Total RNA was extracted with the Spectrum Plant Total RNA kit (Sigma-Aldrich). Following a treatment with the DNAse I (Sigma-Aldrich), the RNA was reverse-transcribed using the SuperScript III reverse transcriptase (Invitrogen, Life Technologies, Waltham, USA).
Quantitative PCR amplification (SYBR Green Supermix, Bio-Rad, Hercules, CA, USA) was carried out in a 15-μl volume (primers in Supplementary  Table S2) and the results were recorded by a CFX96 Touch Real-Time PCR detection system (Bio-Rad), run by CFX Manager software (Bio-Rad Laboratories, Hercules, CA, USA). The software applies comparative quantification with an adaptive baseline. Each sample was run in two technical replicates with the following parameters: 95°C 3 min, 40 cycles of 95°C 10 s and 55°C 30 s, with a final step at 95°C 10 s. Primers for genes VvMLO6, VvMLO11 and VvMLO13 were taken from Winterhagen et al., 30 whereas for VvMLO7 they were specifically designed (Supplementary Table S2). Primers for VvWRKY19, VvWRKY27, VvWRKY48 and VvWRKY52 were taken from Guo et al., 55 for VvEDS1 from Gao et al. 56 and for VvPR1, VvPR6 and VvLOX9 from Dufour et al. 57 The new primer pairs were designed with the NCBI Primer Designing Tool (http://www.ncbi.nlm.nih.gov/tools/primer-blast/; Supplementary Table S2). Complementary DNA samples diluted 10, 100, 1000 and 10 000 times were used to test calculate the efficiency of the primers pairs and the size of the PCR products was confirmed by agarose gel electrophoresis. Presence of a specific final dissociation curve was determined after every quantitative PCR run with progressive increments of temperature from 65 to 95°C (0.5°C each step, 5 s).
Reference genes were, as reported for grapevine, 58 Elongation Factor 1α, GAPDH and Actin. Reference genes stability was assessed with GeNorm (medgen.ugent.be/~jvdesomp/genorm/): the three genes had M-values o0.5, well below the threshold of 1.5 considered sufficient for stability. [59][60][61] Threshold cycles (Ct) were converted to relative expression following Hellemans et al. 62 and based on the average Ct of two technical replicates. For MLO genes, the reference Ct was the average of all samples; for other genes, the control EVB (empty vector Brachetto) at T = 0 was adopted.

Statistical analyses
Disease severity. Data were analyzed using the Statistica 9 software (StatSoft, Tulsa, OK, USA) and the package SPSS (IBM, Armonk, NY, USA). The smallest statistical unit considered was a plant. Severity values of all leaves were averaged, resulting in the value considered in further analyses. Normal distributions (Kolmogorov-Smirnov and Shapiro-Wilk tests, P40.05) were validated for variances homogeneity (Levene's test, P40.05) and subsequently used for one-way analysis of variance (ANOVA) with Tukey's post hoc test (Po0.05) at each time point. Data were transformed in arcsin(x) to meet the pre-requisites of ANOVA. In case of non-homogeneous variances, the Games-Howell's post hoc test was used.
In some cases, data from two experiments were pooled and the ANOVA applied independently for each time point (14, 22 and 30 d.p.i.). Area under disease progress curve data were treated as for severity data. Conidia counts were analyzed with the Kruskall-Wallis test (Po0.05).
Quantitative PCR data analysis. Values of relative expression were expressed in logarithms 26 to obtain normal distributions and homogeneity of variances of the residues, as assessed with Shapiro-Wilk (P ≤ 0.05) and Levene (P ≤ 0.05). Homoscedastic data were analyzed with Tukey's test (Po 0.05) and non-homoscedastic with Games-Howell test (Po0.05) using the statistical package SPSS (IBM).
Expression data from two experiments were analyzed independently and pooled. Differences were revealed by one-way ANOVA with Tukey post hoc test (Po 0.05). In addition, a two-way ANOVA with Tukey post hoc test

RESULTS
Gene transfer, selection and acclimation of MLO transgenic lines A total of five gene transfers were carried out. Four aimed to knockdown (KD) specific MLO genes (i = KD-VvMLO6, ii = KD-VvMLO7, iii = KD-VvMLO11, iv = KD-VvMLO13), the fifth to insert an empty vector. Thirty-seven regenerated lines were obtained, with 29 of them confirmed to contain the insert (Supplementary Table S3). The result of the PCR analysis of six lines is shown in Supplementary Figure S1. Twenty-six transgenic lines were propagated in vitro and tested for the silencing of MLO genes with quantitative PCR. Due to in vitro contaminations, three lines were lost before it was possible to test the level of expression. Gene knockdown was evident for three lines out of eight from gene transfer (iii) (KD-VvMLO11) and three out of nine from gene transfer (iv) (KD-VvMLO13), whereas the lines regenerated from gene transfers (i) (KD-VvMLO6) and (ii) (KD-VvMLO7) did not show any reduction of expression (data not shown). Regenerated lines were also tested for off-target silencing, showing that the RNAi fragments targeted other clade V MLO genes (data not shown). Six lines with various combinations of silenced genes were selected and indicated with acronyms TLB1 (transgenic line of Brachetto) to TLB6 (Supplementary Table S3). Lines from TLB1 to 3 came from gene transfer (iii) (KD-VvMLO11), lines from TLB4 to TLB6 from gene transfer (iv) (KD-VvMLO13; Supplementary Table S3). The control was the EVB line. In addition, TLB7, a regenerated line with no reduction of expression, was also included. All lines, including the control, will be referred in the text as 'transgenic lines'. Lines from TLB1 to 7 are further indicated as 'RNAi lines' and from TLB1 to 6 'mlo lines'.
The survival rate of plants to the acclimation process was~85%. Under greenhouse conditions, the transgenic plants showed normal growth and no pleiotropic phenotypes.
PM and downy mildew resistance of transgenic lines Two independent experiments of PM inoculation were carried out on the RNAi lines TLB1 to 7, and the transgenic control EVB. Three mlo lines, TLB4, 5 and 6, showed a reduction of E. necator infection 460% at 30 d.p.i. (Figure 1; Table 1). The disease reduction of TLB6 decreased with the progression of the infection (Table 1). TLB1, 2, 3 and 7 had a level of susceptibility to PM comparable to EVB (Figure 1; Supplementary Figure S2). The leaf phenotypes in Figure 1 visualize the differences between the different lines. All the mlo lines showed fewer conidia at 30 d.p.i. compared with EVB and the decrease was statistically significant for TLB4, TLB5 and TLB6 (Supplementary Figure S3). Compared with EVB plants, TLB4, 5 and 6 had a reduction of 93%, 95% and 72%, respectively. Conidia counts and disease severity were, as expected, correlated (R = 0.58; P ⩽ 0.01). The reduction of conidia in TLB 4, 5 and 6 (93, 95 and 72%) was higher than the reduction of PM symptoms (68.4, 76.6 and 65.1%), indicating that the leaf diseased area had a lower concentration of conidia in TLB 4, 5 and 6 compared with EVB.
Line TLB4 was characterized by histological analysis, demonstrating a reduced progression of PM infection compared with EVB ( Figure 2). In EVB, conidiophores appeared at 10 and 21 d.p.i. and they were present all over the leaf surface (Figure 2a), whereas on TLB4 leaves they were visible in a limited number only at 21 d.p.i. (Figure 2b). Histological analysis revealed the accumulation of autofluorogenic phenolic compounds possibly associated to host cell response and papillae formation in both TLB4 and EVB plants at 3 d.p.i. (Figure 3).
An experiment was designed to test the cross-reaction of mlo lines to fungal pathogens different from PM. Three mlo lines (TLB1, 3 and 4) and the EVB control were inoculated with the downy mildew causal agent Plasmopara viticola. None of the plants were resistant and all plants showed statistically comparable levels of susceptibility to the pathogen (Supplementary Figure S4).

Expression of MLO genes in the mlo transgenic lines
The lines TLB1 to 6 and the EVB control were considered in a gene expression analysis. The results concerned four genes member of clade V and supported the off-target cross-silencing, as well as,  (Figure 4). Lines TLB1, 2 and 3, all resulting from transformation (iii) (KD-VvMLO11), as expected had the target gene VvMLO11 silenced. TLB1 showed also knockdown of VvMLO13 and TLB3 of VvMLO6 (Table 2). Lines TLB4, 5 and 6 derived from transformation (iv) (KD-VvMLO13) showed more off-target silencing: in TLB4 and 6, all four clade V MLO genes were, to some degree, significantly knocked down, whereas in TLB5 the expression of genes VvMLO6, 7 and 11 was reduced ( Table 2).  Table S4). The reason to choose TLB4 over the other resistant lines was that, in this line, all four MLO clade V genes were knocked down and the knockdown was more intense than TLB5 and 6. In EVB, seven of the genes tested were upregulated at 10 d.p.i. In TLB4, fewer genes were upregulated and the increase of expression was limited in terms of fold change ( Figure 5; Supplementary Table S4). Moreover, three genes were downregulated in TLB4 after inoculation, namely, VvPR6 (pathogenesis related) at 1 d.p.i. and VvNPF3.2 (nitrate transporter/peptide transporter family) and VvALS1 (acetolactate synthase) at 10 d.p.i. It is noteworthy that, before the inoculation, there were no differences in expression between TLB4 and the control EVB ( Figure 5; Supplementary Table S4).

DISCUSSION
Loss-of-function mutations of MLO genes reduce susceptibility to PM in barley, 63 A. thaliana, 21 pea, 18 tomato, 22 wheat 23 and pepper. 24 Because in dicots not all Clade V MLO S-genes are implicated in PM susceptibility, 21,22,29,30 the aim of this work was to identify which of the clade V MLO genes of grapevine have a role in PM susceptibility and can be inactivated to develop resistant genotypes. Out of 26 transgenic lines tested, six from gene transfers (iii) (KD-VvMLO11) and (iv) (KD-VvMLO13) supported significant gene knockdown. In the regenerated lines obtained from gene transfers (i) (KD-VvMLO6) and (ii) (KD-VvMLO7), reduction of expression was not evident. It cannot be excluded  that this was due to the short RNAi fragments present in the constructs. 64 The detection of off-target silencing in five of the six mentioned lines was expected, as clade V MLO genes have high levels of sequence identity (36-60%, 46% on average 29,30 ). To find a balance between specificity (short RNAi fragments) and effectiveness (long RNAi fragments) is particularly difficult in gene families with high sequence similarity. 65 As the aim was to study the effect of the knockdown of four MLO genes quite similar to each other, we opted for long RNAi fragments, so that off-target silencing was not only expected but also desired. Knockout and knockdown of MLO genes may induce pleiotropic phenotypes, such as necrotic spots on leaves and reduced grain yield in barley, 20 slow growth in A. thaliana 21 and reduced plant size in pepper. 24 In grapevine, no pleiotropic phenotypes were observed under the experimental conditions adopted.
Lines TLB4, 5 and 6, which showed clear resistance to PM, allowed the investigation of the link between resistance and the expression of specific MLO genes. VvMLO11 expression was significantly reduced in susceptible and resistant mlo lines: it is concluded that its knockdown was not directly linked to grapevine susceptibility to PM. VvMLO6 was significantly silenced in the resistant lines TLB4, 5 and 6, and in the susceptible line TLB3. As for VvMLO11, the knockdown of VvMLO6 in both susceptible and resistant lines indicates that this should not be an S-gene. Similarly to VvMLO6, VvMLO13 was knocked down in the resistant lines TLB4 and 6, but also in the susceptible line TLB1. VvMLO7 was knocked down only in the three resistant lines TLB4, 5 and 6, but it was always knocked down together with other two or three MLO genes, as there was no line showing the knockdown of VvMLO7 only. Therefore, the knockdown of multiple MLO genes provided resistance to PM. VvMLO6 was also knocked down in all the resistant lines with a reduction of expression of 58-65%, whereas the reduction of its expression was of only 29% in the susceptible line TLB3, indicating that it can contribute to PM resistance. There are no information available about how the reduction of the expression of an S-gene affects disease severity: it could be a linear relationships, meaning that the reduction of expression causes a proportional reduction of disease severity, or there could be a threshold above which the knockdown, even if statistically significant, does not cause any reduction of disease severity. Given the weak knockdown of VvMLO6 in TLB3 (29%), it is possible that this hypothetical threshold was not surpassed, therefore we cannot rule out the possibility of a role for VvMLO6. This would be particularly surprising, as there are no precedence of MLO genes acting as S-genes without being upregulated upon PM inoculation, such as VvMLO6. 29,30 In conclusion, VvMLO6 and VvMLO7 are the main candidates for causing PM susceptibility in V. vinifera, with a possible additive activity. A similar scenario was observed in A. thaliana, where the simultaneous knockout of three MLO genes is necessary to obtain complete PM resistance: knockout of AtMLO2 results in a moderate level of resistance, whereas the knockout of AtMLO6 and AtMLO12, alone or combined, does not decrease the intensity of the infection. 21 When AtMLO2 is knocked out together with AtMLO6 or AtMLO12, the level of resistance rises to become complete when the three genes are knocked out together. 21 In grapevine, VvMLO7 is the best candidates to act as A. thaliana AtMLO2, whereas VvMLO6 and VvMLO11 might have an additive synergistic role in PM susceptibility, as their expression was also significantly reduced in all three resistant lines. In A. thaliana, the knockout of three MLO genes induces complete resistance, 21 a situation not observed in grapevine, in agreement with the incomplete silencing of MLO genes obtained by the RNAi approach. A complementation test carried out in A. thaliana mlo triple mutant showed that VvMLO11 and VvMLO13 induce susceptibility to PM, whereas VvMLO7 has only a partial effect and VvMLO6 has no effect at all. 66 However, single and double VvMLO11 and VvMLO13 knockdown mutants of V. vinifera obtained by RNAi did not show significant reduction of PM penetration. 15 Accordingly, our data indicated only a putative additive effect provided by VvMLO11 and no role at all for VvMLO13. The differences observed with the results carried out by Feechan et al. 66 can be explained by the fact that they operated in a heterologous system (A. thaliana) not reproducing with fidelity the PM infection of grapevine plants.
The precise mechanism through which the reduction of MLO genes expression results in resistance to PM pathogens is not completely clear. Resistance seems linked to secretory vesicles traffic 2,35 and to the formation of cell wall appositions called papillae. 21 These structures consists of a callose matrix enriched in proteins and autofluorogenic phenolics compounds, 54 and their formation depends on endomembrane transport. 67 Our results showed the accumulation of autofluorogenic compounds at the infection sites of transgenic and control lines that reflects the host response and the papilla formation. This first observation suggested that papillae were accumulated in both TLB4 and EVB lines. A difference in dimension was noted between the two lines, but further investigations are required to clear whether this is a direct consequence of MLO knockdown or a random event. Particularly, the use of specific staining protocols 68 would allow to better characterize the composition, dimensions and timing of papilla accumulation in resistant and susceptible lines. It is known that the defense response based on papillae differs between resistant and susceptible genotypes in timing of formation, composition and size. [67][68][69] Particularly, Chowdhury et al. 68 showed that the difference between effective and non-effective papillae is due to the higher concentration of callose, cellulose and arabinoxylan of the effective ones. The MLO protein has been proposed to be a negative regulator of vesicle-associated and actin-dependent defense pathways at the site of attempted PM penetration. 27 Furthermore, Miklis et al. 35 proposed that, once MLO proteins are under the control of the fungus, actin filaments serve the purpose of supplying nutrients for the growing hyphae through vesicular transport. The pathogen may be able to control the transport of material to the cell wall, with the purpose of changing the composition of the papillae and turning them from effective to non-effective.
The formation of papillae is not the only process instigated by the activity of MLO genes. To understand the effect of MLO knockdown on other genes involved in plant-pathogen interaction, the expression of 13 genes known to be differentially expressed after PM inoculation [55][56][57]66,[70][71][72][73][74] was analyzed. The knockdown of MLO genes in the TLB4 line did not affect the expression levels of the 13 considered genes as compared with the EVB line in absence of PM infection. At 1 d.p.i, the response of the considered genes was very limited in both the lines tested, suggesting that the infection was not sufficient to trigger the response. Conversely, some differences between TLB4 and EVB lines were observed at 10 d.p.i. As a general trend, seven genes (VvEDS1, VvPAD4, VvPR1, VvPR6, VvWRKY19, VvWRKY48 and VvWRKY52) were upregulated in EVB, reflecting the reaction of grapevine to PM. Conversely, only four genes (VvLOX1, VvNPF3.2, VvWRKY19 and VvWRKY52) were activated in TLB4, suggesting that   57 whereas VvEDS1 is a grapevine defense gene involved in the SA pathway, 56 a plant growth regulator that activates pathogenesisrelated genes and induces disease resistance. 71 This may indicate that only a severe E. necator infection triggers the plant defense   depending on SA. However, the expression profiles observed in this two lines need to be confirmed by further gene expression studies that comprise more transgenic control and knockdown lines. The knockout of MLO genes increased susceptibility to other pathogens in barley 75,76 and A. thaliana. 21 Not surprisingly, the infection with P. viticola, an obligate biotroph fungus such as E. necator, revealed that the knockdown of MLO genes did not change the susceptibility of grapevine to downy mildew, supporting the conclusion that MLOs S-genes are specific for E. necator and are not involved in the plant interaction with P. viticola.
This work provides crucial information that can be used in breeding grapevine varieties resistant to E. necator. The tagging via genome editing of the MLO genes identified in this paper should result in knockout mutants highly resistant to PM. Alternatively, the search in V. vinifera and in wild species of nonfunctional MLO alleles should contribute to the creation of durable resistance.

CONFLICT OF INTEREST
The authors declare that on 08/07/2015 a patent protecting VvMLO genes has been submitted to the European Patent Office (EPO). The patent application number is PCT/EP2015/065624.