Expression of a Grapevine NAC Transcription Factor Gene Is Induced in Response to Powdery Mildew Colonization in Salicylic Acid-Independent Manner

Tissue colonization by grape powdery mildew (PM) pathogen Erysiphe necator (Schw.) Burr triggers a major remodeling of the transcriptome in the susceptible grapevine Vitis vinifera L. While changes in the expression of many genes bear the signature of salicylic acid (SA) mediated regulation, the breadth of PM-induced changes suggests the involvement of additional regulatory networks. To explore PM-associated gene regulation mediated by other SA-independent systems, we designed a microarray experiment to distinguish between transcriptome changes induced by E. necator colonization and those triggered by elevated SA levels. We found that the majority of genes responded to both SA and PM, but certain genes were responsive to PM infection alone. Among them, we identified genes of stilbene synthases, PR-10 proteins, and several transcription factors. The microarray results demonstrated that the regulation of these genes is either independent of SA, or dependent, but SA alone is insufficient to bring about their regulation. We inserted the promoter-reporter fusion of a PM-responsive transcription factor gene into a wild-type and two SA-signaling deficient Arabidopsis lines and challenged the resulting transgenic plants with an Arabidopsis-adapted PM pathogen. Our results provide experimental evidence that this grape gene promoter is activated by the pathogen in a SA-independent manner.


Results and Discussion
Numerous studies have demonstrated that transcriptome remodeling induced by obligate plant pathogens is mediated to a great extent by SA signaling 27 . PM pathogens have been shown, however, to induce changes in the transcriptome well beyond SA-induced gene expression 13 . To distinguish host transcriptome changes triggered exclusively by SA from those triggered more broadly by E. necator colonization, we conducted two separate global leaf transcriptome analyses using the Vitis Affymetrix GeneChip platform. In the first experiment, we compared the leaves with fully established PM colonies to healthy reference leaves, and found that transcript abundance was at least 1.5-fold higher or lower for 373 genes in PM-infected leaves relative to healthy reference leaves (Supplementary Table S1). Whereas the SA was below the threshold of detection in control leaves, SA accumulated in subsamples of PM-infected leaves to 0.92 ± 0.68 μ g/g fresh weight. These SA levels were similar to those measured in PM-infected grapevine leaves at 2 days post-inoculation (dpi) 10 , indicating that SA levels remained high even when PM colonies became well established on grapevine leaves. This suggests that defense signaling was active in leaves supporting mature, well established PM colonies.
In the second experiment, we assayed MeSA-treated grapevine leaves in comparison with control leaves. The total SA concentration was significantly higher in the MeSA-treated plants (26.33 ± 12.48 μ g/g fresh weight) than in control leaves where SA was undetectable. We found that 481 genes responded to the MeSA treatment with at least 1.5 fold-change in expression, and 179 of them were a subset of the PM-regulated gene list. This suggests that a subset of PM-responsive genes may be regulated via SA signaling.
The Vitis Affymetrix GeneChip included nine probe sets of fungal origin with a nearest homology to genes of ascomycetous fungi (Supplementary Table S1). All of these nine genes were identified in our microarray results as exclusively PM-dependent and were among the genes with highest expression rates (8-to 284-fold). The hybridization of these probe sets by transcripts in exclusively PM-treated samples confirmed that E. necator inoculum was absent in MeSA-treated and control samples proving that experimental treatments were carried out appropriately (Fig. 1).
The relative transcriptional change of those genes that were found to be modulated by PM infection only, MeSA treatment only, or both PM infection and MeSA treatment are displayed in Fig. 2. The microarray probe sets and grapevine transcripts in these categories as well as their nearest Arabidopsis homologues are listed in Supplementary Table S1. [VIT_12s0028g00860] were significantly up-regulated by PM, whereas a gene encoding a lipid transfer protein [VIT_04s0008g05640] was suppressed. Concerning gene regulation influenced by SA, the expression of PRP1 increased, whereas expression of the ADAGIO PROTEIN 1 [VIT_01s0011g05810] was suppressed. However, not all expression changes that were significant in the microarray data could be confirmed as significant changes in the qPCR analysis. For example, although the FAH1 [VIT_07s0031g01380] was registered as up-regulated in both microarray and qPCR experiments, the change was significant (p = 0.0005) only in the microarray data.
Genes induced by both MeSA treatment and PM colonization. Among the 179 transcripts that responded in a similar way to PM and to MeSA, we found genes that function in biotic stress signaling as well as in primary and secondary metabolism (Fig. 1). We refer to these genes as the PM-and SA-regulated gene set.
The key signaling molecule for systemic acquired resistance is MeSA, a mobile form of SA 7 . Gene SAMTBSCMT [VIT_04s0023g02240] was found to be up-regulated by PM as well as by MeSA treatment. SAMTBSCMT encodes a salicylate O-methyltransferase which catalyzes the formation of MeSA from SA and regulates MeSA formation at the site of infection; MeSA is then delivered to the systemic uninfected region of the plant where it can be converted back to SA by SABP2 (SA binding protein 2) to fulfill its function 28 . We found that most of the typical defense-associated genes responded to MeSA treatment. During pathogen attack, the receptor-like protein kinases (RLKs) are the first key regulator proteins of pathogen-associated molecular patterns-triggered immunity (PTI). Among the identified kinases, many of them belong to leucine-rich repeat domain-containing RLKs, which regulate a wide variety of defense responses 29 . From the identified 25 PM-responsive RLKs, 15 were stimulated by MeSA. Three of these were homologous to the Avr9/Cf-9 Rapidly Elicited 256 gene of tobacco, which is one of the key regulators of the HR during biotic stress 30 . Another key defense signaling gene that was found both MeSAand PM-inducible is Enhanced Disease Susceptibility1 (EDS1 [VIT_17s0000g07420]). Albeit, EDS1 is an upstream regulator of SA, previous studies demonstrated that abundant SA may feedback-regulate the EDS1/PAD4 complex in Arabidopsis 28 . It has recently been shown that V. vinifera EDS1 is induced in response to SA and that its orthologue from a PM-resistant V. aestivalis grape variety has a distinct expression pattern 31 .
Although NIMIN-1 acts as a negative regulator of SA/NPR1 signaling 33 , we found a gene, probably encoding the grape orthologue of NIMIN-1 [VIT_07s0005g02070], which was up-regulated in response to both treatments. NPR1, TGA2, 3, 5, and/or 6 control WRKY transcription factor genes, which may positively or negatively regulate the defense response 34 . The grape orthologue of WRKY18_2 [VIT_04s0008g05760], an Arabidopsis gene known to positively and negatively regulate SA/EDS1-mediated resistance against Pseudomonas syringae and G. orontii, respectively 35 , was stimulated by both treatments. In addition, we also found two Myb-type transcription factor (TF) genes, namely, MYB108 [VIT_05s0077g00500] and MYB14_3 [VIT_05s0049g01020], to be MeSA-inducible. The MYB108 TF belongs to the R2-R3-type MYB family, members of which are known to be involved in the . Bivariate plot of expression fold-change for genes that respond to PM colonization and/or MeSA treatment in grapevine leaves. Fold-change of expression induced by PM (vertical axis) plotted against foldchange of expression induced by MeSA (horizontal axis) for genes that respond at the 99% significance level with at least 1.5-fold up-or down-regulation allowing for 5% false discovery rate. Graph does not include data point for the fungal gene derived probe set (1615715_at). Green: PM-responsive only; blue: MeSA-responsive only; red: both treatment-responsive.
Scientific RepoRts | 6:30825 | DOI: 10.1038/srep30825 SA-signaling pathway 36 . MYB108 is closely related to the ABA-dependent BOTRYTIS SUSCEPTIBLE1 gene, which is a negative regulator of cell death triggered by wounding or pathogen attack 37 .
The following pathogenesis-related (PR) genes were regulated via both SA, . PR genes were expressed over the course of the infection process with a steady increase starting at early infection stages 10 . Due to their expression pattern, they all were allocated to the same cluster 10 . It is likely that the regulation of PR genes during PM infection is indicative of the coordination of the defense response via SA signaling, as it was found for other plant-pathogen interactions 26 . Fungal infection-triggered PR protein secretion may be assisted by chaperone proteins 38 . The expression of chaperone genes calnexin 1 (CNX1 [VIT_00s0283g00030]) and endoplasmin (SHD [VIT_18s0001g14500]) was found to be up-regulated in PM-and SA-dependent manner, as it was earlier shown for their orthologues in Arabidopsis 13 .
PM infection along with SA signaling may also induce cross-linking of molecules in the plant cell wall and/or deposition of lignin as part of PTI 39 , which is indicated by the enhanced expression of OMT1 (caffeic acid O-methyltransferase [VIT_16s0098g00850]), a gene known to be involved in lignin synthesis 40 41 all responded to both PM and SA. HSPs are involved in abiotic stress signaling and their role in plant responses to pathogen attack has yet to be fully understood. However, these proteins were found to be also active under oxidative stress, as reactive oxygen species (ROS) and photorespiratory H 2 O 2 induces their expression 25,42 . The earliest response after PM infection in V. vinifera is an oxidative burst, and rapid up-regulation of genes involved in protection from ROS 10 . The pathogen-triggered ROS could explain that these heat shock protein encoding genes are up-regulated by both treatments. Furthermore, low levels of H 2 O 2 act as a signal for defense gene expression 43 , which is supported by the PM and MeSA-dependent up-regulation of a reticuline oxidase precursor transcript. Reticuline oxidase (BBE) catalyzes H 2 O 2 production by using hexose sugars and it mediates basal resistance against pathogens 44 . However, the MSS1 (sugar transport protein 13 [VIT_11s0016g03400]) was also up-regulated by both treatments.
Defense responses along the SA-mediated pathway include redox signaling which is based on the glutathione (GSH) and disulphite (GSSG) ratio. Glutathione S-transferases (GSTs) have both conjugase and peroxidase activity, therefore, GSTs use GSH and reduce H 2 O 2 amount, thereby increasing GSSG levels 45 . Indeed, glutaredoxins (GRXs), namely, GRX480 [VIT_10s0003g00390] and a cytosol localized GSTU8 [VIT_08s0007g01400], which lower H 2 O 2 and elevate GSSG levels were found to be up-regulated by both PM colonization and MeSA treatment. It has been demonstrated that AtGRX480 mediates redox regulation by TGA factors during stress, and linked to SA-dependent pathway 46 . We found in grapevine, however, that another glutaredoxin [VIT_07s0104g01390] was markedly repressed by both treatments.
Among the PM-responsive ATP binding cassette (ABC) transporters, we identified three genes, which were up-regulated by MeSA. One identified transporter probably belongs to the C, the two others to the G family (ABCG7 [VIT_00s0625g00020/VIT_03s0017g01280]). Notably, the expression of the G family members were induced to very high levels by MeSA (6-and 25-fold). Members of the G family are known to mediate the export of cuticular lipids, with PEN3 being a key player in the defense response in Arabidopsis 47,48 .
We found several PM-stimulated secondary metabolism-related genes which play a role in the biosynthesis of antimicrobial compounds. The genes encoding HMG-CoA-synthase (MVA1 [VIT_02s0025g04580]) and HMG-CoA -reductase (HMGR1 [VIT_03s0038g04100]) were activated by both PM and MeSA treatments. The MVA1 and HMGR1 proteins are components of the isoprenoid biosynthesis pathway and involved in the synthesis of mevalonate 49 . Mevalonate is the precursor of phytosterols which play a key role in innate immunity and restrict the nutrient efflux into the apoplastic space where nutrients may be taken up by the pathogen 50 . Moreover, it has been demonstrated previously that the over-expression of Brassica juncea HMG-CoA-Synthase1 in Arabidopsis resulted in the constitutive expression of PRP1, PR2 and PR5 along with suppression of H 2 O 2 -induced cell death 51 , which is in agreement with our findings in grapevine presented here.
Genes involved in aromatic amino acid and phenylpropanoid biosynthetic pathways, such as prephenate and DMR6 genes [VIT_16s0098g00860/VIT_13s0047g00210] were found to be inducible by both MeSA and PM. This is concordant with the notion that flavonoids and their anthocyanin derivatives, have anti-fungal activity in grape varieties 52 . However, the Arabidopsis AtDMR6 gene was found to provide susceptibility to downy mildew 53 . Transcription of the flavonoid biosynthetic gene, CYP706A4 (encoding flavonoid 3′-hydroxylase [VIT_00s1682g00020]) as well as the cytokinin glucosyltransferase gene, UGT85A2 [VIT_00s0324g00070] were down-regulated by both treatments.
SA antagonizes JA signaling in various biotic stresses, and it was found that increased SA levels along with repression of JA-signaling resulted in resistance against biotrophic pathogens, but provided susceptibility to necrotrophs 54 . This cross-talk may be partially dependent on the cellular redox status, while overexpression of GRX480 induced PR-1, but repressed PDF1.2 55 . Confirming this relationship, MeSA as well as PM, induced the expression of JAZ1_2 [VIT_09s0002g00890] in our study. JAZ proteins were shown to repress transcription of JA-responsive genes 56 . However, synergism was also observed between these two signaling pathways as SA signaling does not always repress JA biosynthesis 57 . We found three genes, LOX2 (lipoxygenase [VIT_06s0004g01510]), CYP74A (allene oxide synthase [VIT_18s0001g11630]), and OPR2 (12-oxophytodienoate reductase 2 [VIT_18s0041g02020]) participating in JA synthesis, which were up-regulated in response to both treatments. This is consistent with a recent study which demonstrated that LOX expression in cucumber was stimulated not only by PM and SA, but also by JA and ABA 58 .
The basal defense of susceptible plants also implicates processes that lead to cell wall fortification in response to pathogen attack. We identified two cell wall-related genes, EXPA8 [VIT_13s0067g02930] and a pectate lyase [VIT_17s0000g09810], which were down-regulated by both treatments. Expansins unlock the network of wall polysaccharides and pectate lyases degrade the pectin component of cell wall 59,60 , therefore, their repression maintains cell wall integrity. PM-induced repression of these grapevine genes via SA-signaling suggests a regulation by the plant to boost structural resistance against the invading pathogen. In Arabidopsis, down-regulation of the pectate lyase-like gene PMR6 was shown to enhance resistance to PM 61 . Thus, PM-induced repression of these grapevine genes suggests that their down-regulation may also contribute to enhanced resistance.
Overall, expression of most genes modulated by both MeSA and PM were part of the SA-mediated defense response The majority of the MeSA-and PM-responsive transcripts are downstream of SA in the signaling cascade (as the NIMIN1-1, WRKY or PR proteins), but some upstream regulators (EDS1) are also known to participate in a feedback-regulatory loop with SA.
Genes induced by PM colonization but not by SA treatment. Among the PM-regulated genes in grapevine, 185 candidates were identified which were not triggered solely by MeSA, indicating that elevated SA levels alone cannot substitute for regulation by PM. These 185 genes are referred to as the "PM-dependent" gene set. These include numerous genes that are involved in primary metabolism, including the pathways of carbohydrate, protein, and fatty acid metabolism (Fig. 1, Supplementary Table S1). Since PMs are obligate biotrophic pathogens, they must rely on their host as carbon and nitrogen source and, therefore, modulate plant metabolic processes to fulfill their needs. However, previous results demonstrated that carbohydrates also may have signaling function in defense responses as the increased content of soluble sugar induced the expression of PR genes in Arabidopsis 62 . Beside the activation of defense-genes, sugar accumulation is also expected to decrease photosynthesis 63 65 . Interestingly, RNAi silenced tip1-1 Arabidopsis plants revealed an increased apoplastic carbohydrate content 66 , suggesting that AQUAPORIN TIP1_3 suppression in infected grapevine may support the sugar availability for the pathogen. In addition, the transcription of a germin-like protein-encoding gene [VIT_17s0000g05360] was also induced by PM. Such proteins were found to catalyze H 2 O 2 production 67 .
Among the PM-dependent gene set, several transcription factors were identified, among them a NAC-type transcription factor (NAC042_5 [VIT_12s0028g00860]). Based on the expression pattern reported earlier 10 , this gene belongs to the same cluster as genes for pinoresinol forming dirigent protein (DIRPR [VIT_02s0025g00750]), dicyanin blue copper protein (BCB [VIT_09s0002g06890]) and isoflavone methyltransferase [VIT_12s0028g01940] 10 . These latter genes were also PM-dependent, albeit their cluster also contains PR genes 10 which were PM and MeSA-inducible in our current dataset. Two other transcription factors that belong to the WRKY family (WRKY71_2 [VIT_12s0028g00270] and WRKY21_2 [VIT_00s2547g00010]) were in the PM-dependent gene set. Previous studies demonstrated that WRKY71 is involved in the defense response and that it is an upstream regulator of NPR1 in rice 68 . The WRKY IId subfamily members, including WRKY21, were found to interact with Ca 2+ /calmodulin binding transcription factors 69 and mediate the defense response. However, the transcription of a calmodulin-binding protein [VIT_01s0026g01790] was found to be up-regulated by both MeSA and PM.
Two typical defense associated genes, namely PR10 [VIT_05s0077g01530] and Bet v I allergen [VIT_05s0077g01540], were strongly expressed (8-and 13-fold up-regulation) only in response to PM. Although most PR transcripts were found to be MeSA-inducible, these genes responded only to PM. They were grouped in a cluster along with genes encoding stilbene synthases and the cytochrome P450 84A1 (FAH1) 10 . Several studies proved that PR-10 proteins, which have RNase, DNase, and anti-fungal activity, play a role in defense responses and cell death and that they are regulated by WRKY TFs 70,71 . It was shown that the expression of the V. vinifera PR10.1 was transcriptionally regulated by the WRKY33 TF due to Plasmopara viticola infection 72 . Further, it was demonstrated that the Asparagus PR10 was responding to pathogen infection and H 2 O 2 independently from SA 73 , which suggests that these proteins mediate defense responses upstream or independent of SA signaling in grapevine also. Some genes involved in the biosynthesis of stilbenoids, flavonoids and phenylpropanoids were found to be regulated in a PM-dependent manner. The genes encoding DAHP-and EPSP-synthases [VIT_00s0391g00070 and VIT_15s0048g00350] were found to be up-regulated. The corresponding proteins catalyze the synthesis of aromatic amino acids, which are precursors of flavonoids and stilbenoids in the shikimate pathway 74  which likely regulates stilbene biosynthesis, was 3-to 7-fold up-regulated by PM in grapevine. Previously, a MYB14 was found to be co-expressed with STSs and to specifically interact with the promoters of STS41 and STS29 in grapevine 75 . Stilbenes in Vitis species were proposed to be part of the plant arsenal against E. necator 76 . The gene, encoding chalcone-flavonone isomerase (TT5 [VIT_13s0067g03820]) involved in flavonoid biosynthesis was also regulated in a PM-dependent manner. A similar response was found for a putative DFRA [VIT_08s0040g00440] and a UGT75C1 [VIT_05s0062g00740] gene, which are involved in secondary metabolism. UGTs along with cytochrome P450 monooxygenases play a key role in creating the structural diversity of triterpenoid saponins 77 , which are antifungal compounds 78  ) in response to the presence of pathogen. Corroborating our findings, the Arabidopsis FAH1 was also found to be up-regulated by PM independently of SA signaling 13 . The cytochrome P450 gene CYP716A1, protein product of which is involved in antimicrobial saponin biosynthesis, was up-regulated 9-fold in response to PM, supporting the premise that it mediates plant defense 79 . Geraniol 10-hydroxylase [VIT_15s0048g01490] is involved in terpenoid indole alkaloid biosynthesis 80 , and its gene is homologous to AtCYP76C1, which was also down-regulated independently from SA in response to PM in Arabidopsis 13 . Since these genes were not inducible by SA, the inducer is likely to be another signal. CYP87A3 (61% identity to CYP87A2) was previously reported to be responsive to auxin 81 which may also act as a defense signal during pathogen attack and it may have an antagonistic regulatory role to SA 82 .
Four of the six PM-responsive dirigent-like protein genes [VIT_06s0004g01020/VIT_02s0025g00750/ VIT_06s0004g01010/VIT_06s0004g00990], which play a role in lignin synthesis 83 were strongly up-regulated (3-to 13-fold) in a PM-dependent manner, which is in agreement with previous reports 84 . Another lignin biosynthetic gene encoding a cinnamoyl-CoA reductase [VIT_02s0012g01570], was found to be regulated by PM only. The protein product of this gene promotes the H-, S-, and G-lignin formation in the monolignol pathway.
The acyl-CoA-binding domain 3 proteins (ACBP3) are proposed to be involved in lipid metabolism. However, the Arabidopsis ACBP3 also regulates the NPR1-dependent defense in response to the biotrophic bacterium P. syringae, and overexpression of ACBP3 resulted in enhanced PR expression, cell death and H 2 O 2 production 85 . We found that the ACBP3 [VIT_07s0129g00430] grapevine gene is not MeSA-inducible, but it is triggered by the pathogen. In contrast, other genes encoding enzymes involved in the lipid metabolism (3-oxoacyl-[ACP] reductase [VIT_01s0010g02670], a probable sulfotransferase [not registered in Ensembl]) and in lipid transfer/binding [VIT_04s0008g05640/VIT_11s0016g05840/VIT_04s0008g05640] were at least 6-fold down-regulated by PM, in accordance with previously reported results 10 . Since lipids have a signaling function during pathogen attack, the PM-mediated down-regulation of the expression of such genes may halt activation of defense responses.
Observation of the NAC042_5 promoter regulation. Among the most dramatically regulated PM-dependent genes there was NAC042_5 which codes for NAC-like transcription factor 42. The expression of NAC042_5 was induced 7-fold in response to the PM fungus, but was unchanged in response to MeSA treatment (Supplementary Table S1), suggesting PM-specific and SA-independent regulation. To confirm that the transcription of NAC042_5 was indeed SA-independent, a pNAC042_5::GUS translational fusion reporter was constructed and inserted in the genome of wild-type A. thaliana Wassilewskija (WS-0), SA-signaling mutant WS-nim1-1, and SA-deficient transgenic WS-nahG plants. In the designation of this construct, pNAC042_5 denotes a 3896 bp-long stretch sequence of the pNAC042_5 promoter (National Center for Biotechnology Information GenBank accession number: KU297673) and the first 8 amino acids of the NAC042_5 polypeptide.
GUS staining of non-inoculated homozygous transgenic plants demonstrated that all three types of transgenic Arabidopsis (WS-0, WS-nim1-1, and WS-nahG) showed a similar basal GUS expression independent of PM challenge. Earlier studies demonstrated that members of the Vitis NAC gene family regulate organ development in grapevine species, and that their expression differed in various developmental stages 86,87 . The expression of NAC042_5 did not demonstrate strict tissue-specificity, as its promoter was active in the shoot apical meristem, young developing shoots and leaves, siliques, trichomes, vascular tissues, and lateral shoot buds (Fig. 3). Promoter activity in this broad variety of organs could be explained by a more general transcriptional regulator function.

NAC042_5 promoter activity in response to PM infection. To quantify PM-induced transcriptional
activity directed by the NAC042_5 promoter, we inoculated and mock-inoculated the pNAC042_5::GUS transgenic Arabidopsis lines with Oidium neolycopersici following the method described by Huibers and co-workers 88 . By 14 dpi, the inoculation led to fully developed conidium-producing PM colonies in all lines, and all mock-treated plants remained PM-free. PM infection advanced faster and produced more extensive colonies in plants of the nim1-1 and the nahG genetic background than in wild-type plants, which is likely due to the higher disease-susceptibility of nim1-1 and nahG plants. Leaf tissues with 14 day-old PM colonies and mock-inoculated control leaves were used for a pNPG spectrophotometric assay to quantify GUS activity. Statistical analysis of the GUS assay data revealed that the interaction effect of treatment and assay time is significant (p < 0.0001), and this significance of the PM-infection occurred at 0 and 30 min. At subsequent time points during the spectrophotometric assay, the variability of absorbance values increased with time and the absorbance values were also correlated in the PM-infected samples. After adjusting for the dependence and the varying variability, the estimated rate of change is 1.34 times of the median, which is significant. The confidence limit for the rate is 1.112 and 1.568 times of the median. We also detected a marginally significant (p = 0.0485) effect of the interaction between the treatment and genetic background which probably reflects the more intense growth of the PM pathogen in the highly susceptible nim1-1 and nahG lines than in the wild-type 89 . Values from plants of independent lines for each type of transgenic plant with a similar basal expression are displayed (Fig. 4). As SA signaling is abrogated in nim1-1 and nahG plants, these results provide evidence that the NAC042_5 promoter is responsive to PM infection in an SA-independent manner. Two recent studies presented that another Vitis NAC transcription Scientific RepoRts | 6:30825 | DOI: 10.1038/srep30825 factor gene, NAC1, was activated by E. necator along with increased expression of defense-associated genes, such as PDF1.2, VSP1, PR1, PR2 PR4 and PR5 86,90 . However, in contrast to the NAC042_5 used for our investigations, the NAC1 gene was found to be SA-inducible, indicating that the expression of the various grapevine NAC genes is regulated by different signaling pathways.
The pNAC042_5::GUS reporter lines were also investigated by histochemical staining in response to PM infection and the staining of these leaves revealed a marked increase in GUS activity at the sites where PM colonies developed (Fig. 5a). In mock-inoculated control leaves, GUS-staining was mostly limited to trichomes (Fig. 5b). To confirm that GUS-staining was indeed caused by the growth of O. neolycopersici colonies, we also stained the fungus with the dye cotton blue. Robust GUS-staining was always associated with the presence fungal structures (Fig. 5c-e) and never occurred in their absence. On mock-inoculated leaves, only few confined GUS spots were visible, but this was clearly distinguishable from the robust GUS-staining detected at fungal infection sites (Fig. 5a,c,d). This indicates that the reporter gene was strongly expressed in only those areas of the leaf where the pathogen had direct contact with the plant tissue (Fig. 5c-e). Higher magnification revealed that GUS expression severely increased mostly in those cells, in which the fungus developed haustoria (Fig. 5d,e). This PM-dependent increase in GUS activity was found in all three types of transgenic plants (with nim1-1, nahG, and WT background), which provides further evidence that NAC042_5 expression does not require SA signaling.
The Arabidopsis gene ANAC042/JUB1, the ortholog of the grape NAC042_5, was also shown to be up-regulated exclusively in the immediate vicinity of cells invaded by haustoria in the G. orontii-Arabidopsis interaction 91 . The same study found that PR-1 expression was 137-fold up-regulated in cells closely associated with  the haustorium 91 . Furthermore, the ABC transporter gene PEN3, which mediates penetration resistance, also showed infection site-specific transcription 47 . Similarly, PUX2 and DMR6, genes which support mildew development in Arabidopsis, were also up-regulated at the site of infection 53, 91 . Additional examples of SA-independent PM-responsive genes are PMR5 and PMR6, which are required for the accommodation of the fungal haustorium at later stages colonization 12,61 .

Conclusions
In grapevine, PM colonization triggers changes in expression of a broad range of genes, many of which were not responsive to an increase in SA levels alone. This suggests that PM colonization activates regulatory networks that are more extensive than the SA-mediated defense system. Furthermore, genes with no known defense-related function have also been observed to change in expression. One of these is an NAC-type transcription factor gene (NAC042_5). The cloned promoter of this gene was activated in tissues colonized by the PM fungus O. neolycopersici in nim1-1-mutant and nahG-transgenic A. thaliana lines which are SA signaling-impaired and SA-deficient, respectively. These results provide experimental evidence that PM colonization activates regulatory mechanisms that are independent of SA-mediated regulation.

Methods
Grapevine plant material, growth conditions, and PM-/MeSA-treatments. The experiments were performed with one-year-old greenhouse-cultivated potted V. vinifera L. cv. 'Cabernet Sauvignon' grapevines with a single actively growing herbaceous shoot on each vine. To prepare PM-colonized tissues, two unfolded, but still expanding leaves were mock-inoculated or inoculated with E. necator conidia under greenhouse conditions. Inoculation was done by touching the upper surface of the leaf with a detached grapevine leaf covered with E. necator colonies actively producing conidia. To prepare leaf tissues for SA-induction, leaves at the same developmental stages were mock-inoculated by touching the leaves with detached PM-free healthy grapevine leaves. To prepare healthy reference leaf tissues, plants were treated in the same manner, including mock-inoculation. Three dpi, all grapevines were transferred to a PGR15 plant growth chamber (Conviron) with conditions of 85% RH, 14/10 h diurnal cycle, and 26 °C temperature. PM-inoculated plants were cultivated in the growth chamber for eight additional days until 11 dpi, at which time the PM-colonized leaves were harvested for RNA extraction. Plants prepared for SA-induction were cultivated in the growth chamber for seven days, at which time they were treated with 15 μ M of methyl salicylate (MeSA, SA analogue), evaporated in the atmosphere of the growth chamber under airflow generated by a computer fan for 24 hours. SA-induced mock-inoculated leaves were harvested at the completion of this 24-hour treatment (11 dpi). Reference plants were cultivated under identical growth chamber conditions with their mock-inoculated leaves harvested at 11 dpi. Thus, the PM-colonized and reference samples differed only in the presence/absence of PM treatment, whereas the SA-induced and reference samples differed only in the presence/absence of MeSA treatment. Leaves from all treatments were harvested at 11 dpi and immediately flash-frozen in liquid nitrogen. Each treatment was done in three biological repeats, that is, each experiment was repeated three times in 14-day intervals with dedicated biological material. Each repeat consisted of ten potted vines. For RNA extraction, two young leaves were harvested from each vine of the ten-vine repeat and pooled into a single sample.

Measurement of SA concentration.
For SA concentration measurements, the method described by Fung and co-workers 10 was applied. Leaf subsamples were vacuum-dried and 0.5 g of the sample was extracted and suspended in 300 μ l of 20% methanol. Five microliters of the sample were used for analysis with the Agilent HPLC 1100 Series instrument with diode array detector (4.6 × 75 mm Zorbax SB-C18 3.5 μ m and Zorbax High Pressure Reliance Cartridge Guard Columns, Agilent). The flow rate was 1.2 ml/min and three technical replicates were analyzed for each sample. The reference curve consisted of a dilution series of sodium salicylate in a concentration range between 1 to 100 ng/μ l.
Total RNA extraction from grapevine leaves. Leaf tissues were ground in liquid nitrogen and homogenized in extraction buffer (2% Hexadecyltrimethyl Ammonium Bromide/CTAB, 1% SDS, 2.5 M NaCl, 0.5 M Tris, 50 mM EDTA, 5% beta-mercaptoethanol, and 3% polyvinyl poly-pyrrolidone). The samples were stored at -80 °C until processing. For RNA isolation the frozen samples were thawed at 45 °C and centrifuged (13,000 rpm, 20 min, 4 °C). The supernatant was replenished with 1/2 volume of chloroform, vortexed, and centrifuged (13,000 rpm, 15 min, 4 °C). The supernatant was supplemented with 1/5 volume of 12 M LiCl and incubated for 2 hours at 4 °C. After centrifugation (13,000 rpm, 30 min, 4 °C) the supernatant was discarded and the pellet was washed twice with 80% ethanol and dissolved in RNase-free water. The samples were treated with 1 μ l Turbo DNase I (Ambion) in 40 μ l reactions, and RNA was purified using an RNeasy MiniElute Cleanup column (Qiagen) following the manufacturers' guidelines.

Microarray experiment.
To analyze gene expression changes in response to PM colonization and SA, we employed the Affymetrix GeneChip V. vinifera (Grape) Genome Array following the manufacturer's guidelines. Briefly, 4 μ g total RNA was used to synthesize double stranded cDNA using the One Cycle cDNA Synthesis kit, then this cDNA was used to produce biotin-labeled cRNA through an in vitro transcription (IVT) reaction. The labeled cRNA was fragmented (heated at 94 °C for 35 min to break RNA molecules to 35-to 200-nucleotide fragments) before hybridization to Genechip probes. Hybridization was performed at 45 °C for 16 hours, followed by a washing and staining process of the array, which was performed on an Affymterix Fluidic Station 450. Fluorescence was amplified with streptavidin-phycoerythrin staining, followed by the addition of a biotinylated antibody (anti-streptavidin) solution, and by a final streptavidin-phycoerythrin staining. The prepared chip was then scanned by a GSC3000 laser scanner and the intensity values were processed using the GeneChip Operating Software version 1.2 of Affymetrix. Following Affymetrix guidelines, we performed background corrections and calculated expression values. Normalization was performed using the robust multiarray averaging method. Normalized intensity values, as well as raw GeneChip images have been deposited in the Gene Expression Omnibus database in GenBank (accession number: GSE53824).
The Affymetrix GeneChip contained nine homologous probe sets derived from ascomycetous fungi with close DNA sequence homology to Blumeria, Marssonina, and Ajellomyces species which served as internal controls in our inoculation experiments.

Statistical analysis of microarray results.
Intensity values of the microarray experiment were log2-transformed and submitted to exploratory analysis. An ANOVA model with balanced single factor was applied for evaluating data using the statistical package S-plus. The error term is assumed to be normally distributed with mean zero and constant variance. The genes with at least 1.5-fold change compared to the control (p-value < 0.01 and False Discovery Rate 5%) were selected for further analysis.
Annotation of Affymetrix probe sets. The probes which showed at least 1.5 fold-change compared to the control were annotated. The annotation of probes was performed by blasting the Affymetrix GeneIDs (downloaded from http://www.affymetrix.com/estore/) to the EST (Expressed Sequence Tags) database of NCBI GeneBank (GeneBank and GeneIndex IDs), and the most analogous ESTs were searched for homologs among five species (V. vinifera, A. thaliana, A. lyrata, S. tuberosum, S. lycopersicum) using the BLASTx algorithm (Supplementary Table S1). The homology between query and database sequences was perceived to be informative only if the E-value was less than 1e −10 . The identified transcripts then were analyzed by MapMan, KEGG and Ensembl databases [92][93][94] to categorize the role of genes in metabolic pathways or other processes (Supplementary  Table S1).
Reverse transcription-quantitative real-time PCR analysis of selected genes. Selected genes were independently validated by quantitative real-time PCR (qPCR) to evaluate expression changes detected by the microarray experiment. After RNA isolation, cDNA was synthesized using the Taqman Reverse Transcription Reagent kit (Life Technologies) following the manufacturer recommendations. Based on the DFCI EST database and on the reference genome sequence provided by Genoscope (http://www.genoscope.cns.fr), gene-specific primers were designed for the following grapevine target genes (Genes and primers in Supplementary Data S2): NAC042_5; PR genes PRP1 and Bet v I allergen; FAH1; LTP (lipid transfer protein); EXPA1 (expansin A1 [VIT_14s0108g01020]); and ADAGIO PROTEIN 1 (FKF1); the ACTIN 1 served as a reference. For qPCR analysis, the SYBR Green Reagent kit (Life Technologies) and the real-time thermal cycler Mx3005P (Stratagene) were used. All samples were run in triplicates under identical reaction settings: the initial activation step of AmpliTaq Gold ® was 95 °C for 10 min and followed by 40  using the qPCR analysis software package MxPro-Mx3005P version 3.0 (Stratagene) and the DART-PCR version 1.0 software tool 95 as recommended. R 0 values of target genes were normalized to R 0 values of the reference gene. Statistical significance was determined by Student's t-test to compare the treatment-induced response to the control.

Construction of pNAC042_5::GUS transgenic Arabidopsis and inoculation with PM.
Based on the grape reference genome sequence, the NAC042_5 promoter region was isolated using the primers 5′ -CACC TCA ATC ACA CTC AAA AAC CA-3′ (Forward) and 5′ -AGT GCT AGT CTT CTC CAC CTC CAT-3′ (Reverse). The amplified DNA fragment (National Center for Biotechnology Information GenBank accession number: KU297673) was cloned into the pGWB633 binary vector 96 using the pENTR vector system and the Gateway ® Cloning technology following the guidelines of Invitrogen TM Life Technologies. In the pGWB633 binary construct, which was confirmed by sequencing, the NAC042_5 promoter controls the GUS reporter gene.
The T-DNA of the pGWB633 contains the bar gene for selection of positive transformants. The pGWB633 plasmid with the NAC042_5 promoter construct was transferred into Agrobacterium tumefaciens GV3101 (pMP90) strain and, subsequently, the bacteria were used for transformation of A. thaliana via the flower dip method 97 . Three Arabidopsis lines with Wassilewskija ecotype background were selected for transformation: wild type (WS-0), nim1-1 mutant (SA signaling deficiency, non-inducible immunity mutant), and nahG transgenic (lack of SA signal, it contains the salicylate hydroxylase gene from Pseudomonas putida) 89,98 . Following transformation, the seeds that developed from dipped flowers were harvested and sowed to grow the T1 generation under the following growth conditions: cool white light illumination, 16/8 h diurnal cycle at 24 °C degree. For selection of positive transgenic plants, we applied a solution of 60 mg/l glufosinate-ammonium containing herbicide and 0.01% Silvet L-77 on soil grown seedlings at ten to eleven days after germination 96 . The application was repeated three times until herbicide-sensitive and resistant plants could be unambiguously differentiated. The glufosinate-ammonium-resistant T1 plants were allowed to self-pollinate and produce a T2 generation. T2 seedlings were also treated with herbicide to look for the homozygous lines. The resulting plants were allowed to produce T3 progeny, which were used for inoculation experiments. Three week-old plants were mock-inoculated or inoculated with O. neolycopersici conidia under growth chamber conditions. Inoculation was done by touching the upper surface of the leaf with a detached tomato leaf covered with O. neolycopersici colonies actively producing conidia. This inoculation method was performed for histochemical assays.
For spectrophotometric measurements, four week-old plants were inoculated by spraying a conidial suspension 88 . Control treatment of plants was accomplished by a mock-inoculum spray using healthy tomato leaves. The mock-inoculated and inoculated plants were cultivated under the following growth conditions: cool white light illumination, 16/8 h diurnal cycle at 24 °C degree. pNPG measurements to quantify promoter activity in Arabidopsis. At 14 dpi, six individuals were collected from each line of the PM-inoculated and mock-treated plants. The infected leaf tissues were excised and were ground in extraction buffer (50 mM NaPO 4 pH 7.0, 10 mM β -mercaptoethanol, 0.1% Triton X-100). The extract was incubated after addition of 1 mM 4-nitrophenyl β -D-glucuronide (pNPG) at 37 °C for 2 h 99,100 and its conversion by β -glucuronidase was measured in a spectrophotometric assay at 405 nm at 30-min intervals (as repeated measurements) using a Nanodrop 1000 instrument 99 . To determine if GUS activity was different between the PM-inoculated and mock-treated tissues, the absorbance values were transformed to natural logarithm values (to obtain a reasonably normal distribution), and analyzed using a mixed linear model implemented by the software package SAS. The mixed linear model was as follows: Log (observation) = effect of gene + effect of treatment + effect of time + interaction effect of gene and treatment + interaction effect of gene and time + interaction effect of treatment and time + interaction effect of gene, treatment and time + error.
Histochemical GUS assay to localize GUS expression in Arabidopsis leaf tissue. At 11 dpi, the plants were investigated by histochemical GUS assay 101 . The leaves were incubated overnight at 37 °C in the assay solution (100 mM NaPO 4 buffer; pH 7.0, 10 mM EDTA, 1% Triton X-100, 0.3% H 2 O 2 , 0.5 mg/ml X-Gluc/5-bromo-4-chloro-3-indolyl-β -D-glucuronic acid cyclohexylammonium salt) and the chlorophyll from the samples was removed by repeatedly washing with 70% ethanol. To stain fungal tissue, leaves were dipped for 30 s into cotton blue solution (Thermo Scientific; 30X dilution in 70% ethanol), rinsed with distilled water, and subsequently investigated using a stereo-and light-microscope.