Transcriptome Analysis of H2O2-Treated Wheat Seedlings Reveals a H2O2-Responsive Fatty Acid Desaturase Gene Participating in Powdery Mildew Resistance

Hydrogen peroxide (H2O2) plays important roles in plant biotic and abiotic stress responses. However, the effect of H2O2 stress on the bread wheat transcriptome is still lacking. To investigate the cellular and metabolic responses triggered by H2O2, we performed an mRNA tag analysis of wheat seedlings under 10 mM H2O2 treatment for 6 hour in one powdery mildew (PM) resistant (PmA) and two susceptible (Cha and Han) lines. In total, 6,156, 6,875 and 3,276 transcripts were found to be differentially expressed in PmA, Han and Cha respectively. Among them, 260 genes exhibited consistent expression patterns in all three wheat lines and may represent a subset of basal H2O2 responsive genes that were associated with cell defense, signal transduction, photosynthesis, carbohydrate metabolism, lipid metabolism, redox homeostasis, and transport. Among genes specific to PmA, ‘transport’ activity was significantly enriched in Gene Ontology analysis. MapMan classification showed that, while both up- and down- regulations were observed for auxin, abscisic acid, and brassinolides signaling genes, the jasmonic acid and ethylene signaling pathway genes were all up-regulated, suggesting H2O2-enhanced JA/Et functions in PmA. To further study whether any of these genes were involved in wheat PM response, 19 H2O2-responsive putative defense related genes were assayed in wheat seedlings infected with Blumeria graminis f. sp. tritici (Bgt). Eight of these genes were found to be co-regulated by H2O2 and Bgt, among which a fatty acid desaturase gene TaFAD was then confirmed by virus induced gene silencing (VIGS) to be required for the PM resistance. Together, our data presents the first global picture of the wheat transcriptome under H2O2 stress and uncovers potential links between H2O2 and Bgt responses, hence providing important candidate genes for the PM resistance in wheat.


Introduction
Hydrogen peroxide (H 2 O 2 ) is a well known toxic molecule and is also a specific component of several biotic and abiotic signaling pathways [1]. In plants, H 2 O 2 production causes oxidative stress during external stimuli such as chilling [2], drought [3], salinity [4], UV irradiation [5], ozone exposure [6], heavy metal [7], and wounding [8]. H 2 O 2 is also produced upon phytohormone treatments such as abscisic acid (ABA) [9] and jasmonic acid (JA) [8], as well as during elicitor and pathogen challenges [10]- [11]. In almost all cases, H 2 O 2 seems to be positively employed by plants to activate certain stress-responsive genes that help them cope with adverse environmental changes.
Global gene expression profiling experiments on H 2 O 2 -treated plants have revealed a large number of genes in Arabidopsis and tobacco that are mostly involved in response to oxidative stress [12]- [14]. Such treatments, however, not only affect genes involved in reactive oxygen species (ROS) detoxification, but also drive the expression of genes involved in signal transduction, transcriptional regulation and protein, carbohydrate or lipid metabolism, illustrating the complexity of the transcriptional responses to H 2 O 2 . Recently, a proteomics investigation of proteins that are differentially accumulated responding to exogenous H 2 O 2 was performed in rice [15]. The study identified proteins that are involved in various cellular responses and metabolic processes, redox homeostasis, signal transduction, protein synthesis and degradation, photosynthesis and photorespiration, and carbohydrate/energy metabolism.
Production of ROS has been observed during pathogen infection and is also a hallmark of successful recognition of infection and activation of plant defense systems. Oxidative burst is one of the fastest defense responses activated in plants to resist pathogen and parasite attacks. It consists of the production of ROS, mainly H 2 O 2 , at the first site of pathogen invasion: the plant cell wall, and incurs a number of events before the transcription-dependent defense mechanisms are activated [16]. Subsequently, H 2 O 2 works as a selective signal for the induction of a subset of defense genes including phosphorylation cascades [17]- [18], cyclic oxylipins of the jasmonate type [19], phytoalexins and secondary metabolites [20], as well as genes associated with programmed cell death [21] and plant hormone signaling [22]. Phyto-oxylipins are metabolites produced in plants by the oxidative transformation of unsaturated fatty acids via a series of diverse metabolic pathways and are believed to play a pivotal role in plant defenses as signal molecules and/or protective compounds [23]. Along with salicylic acid (SA), JA, and ethylene (Et) are hormones usually associated with the induction of defenses where they antagonistically interact with SA [24]. ROS are proposed to be the central component of a self-amplifying loop that regulates the interaction between SA, JA and Et to mediate the response to ozone and possibly some defense and cell death processes [25]- [26].
Wheat (Triticum aestivum L.) is one of the most important crops to feed the world's population. Each year, wheat powdery mildew (PM) causes serious yield loss worldwide. H 2 O 2 has been shown to accumulate in the mesophyll cells during the early stage of the wheat-PM incompatible interaction [11], [27]- [28], suggesting that H 2 O 2 signaling plays a role in PM defense. In Arabidopsis, the co-regulation of H 2 O 2 signaling and defense responses was reported. For example, H 2 O 2 regulates the coordinated action of the Arabidopsis TGA1 (TGACG motifbinding factor 1) and NPR1 (non expressor of pathogenesis-related genes 1) genes that are required for defense gene expression and systemic acquired disease resistance [18], [29]. Similarly, a serine/ threonine kinase gene, Stpk-V introgressed from Haynaldia villosa, a wild relative of wheat, is recently reported to be induced by both Blumeria graminis s.sp. tritici (Bgt) and exogenous H 2 O 2 , and confers the resistance to the powdery mildew (PM) in wheat [30]. These studies demonstrate the coordination of the H 2 O 2induced signaling pathways and that of pathogens. The recent advance in next-generation sequencing technology provides a powerful tool for sequence-based transcriptome analysis in species with large genomes such as wheat. However, a global analysis of H 2 O 2 -responsive genes in wheat is still lacking, especially H 2 O 2 -triggered defense related genes and molecular pathways.
In the present work, we performed a transcriptome analysis of H 2 O 2 -treated wheat seedlings in the PM resistant (PmA) and susceptible (Cha and Han) lines. Using the mRNA tag sequencing approach, we identified differentially expressed genes with consistent expression patterns in all three wheat lines as well as those with specific expression patterns in PmA. We found that the basal H 2 O 2 -responsive genes in wheat were involved in various cellular responses and metabolic processes, with functions toward cell defense, signal transduction, photosynthesis, carbohydrate metabolism, lipid metabolism, redox homeostasis, and transport. In PmA, Gene Ontology analysis revealed the enrichment of 'transport' activities, while MapMan classification unraveled activated genes in PmA for JA and Et signaling pathway. Further, eight genes were found to be coregulated by H 2 O 2 treatment and Bgt inoculation, among which a fatty acid desaturase gene (TaFAD) was shown to be involved in the PM resistance. Our work is the first transcriptome-wide analysis of wheat genes responding to H 2 O 2 treatment and provides candidate genes that may deserve further investigation for PM resistance in wheat.

Results
The morphological and physiological changes of wheat seedlings under H 2 O 2 treatment H 2 O 2 can act as signaling molecules at low concentrations by diffusing into cells and be rapidly and specifically perceived by a series of target proteins before being scavenged by antioxidative defense mechanisms [15]. These H 2 O 2 signals are transmitted to downstream signaling molecules and together modulate various metabolic and defense pathways in plants [9], [13], [31]- [32]. To study the effects of H 2 O 2 on wheat seedlings and discover genes that are potentially involved in biotic responses, we set out to investigate the H 2 O 2 -triggered transcriptome profile changes in one PMresistant (PmA) and two PM-susceptible lines (Cha and Han; see Materials and Methods; Figure 1A). As shown in Figure 1B, the growth of 9-day-old wheat seedlings in 10 mM H 2 O 2 for 6 hours (h) did not cause evident changes in morphology, which was in contrast to the severely curled leaves of rice seedlings under similar conditions [15]. To confirm that the application of exogenous H 2 O 2 indeed elevated the cytosolic H 2 O 2 level, we determined the endogenous H 2 O 2 concentration in the treated leaves. The results showed that the internal H 2 O 2 level indeed increased by more than 200% over the control in all of the three wheat lines ( Figure 1C). In contrast, the net photosynthetic rate (Pn) in all of the three wheat lines were decreased ( Figure 1D). These measurements demonstrate that the current treatment condition was sufficient to cause physiological changes in the wheat cells. We then carried out an mRNA tag profiling analysis using RNA samples extracted from the seedling leaves. The two libraries derived from the 0 h and 6 h time points were named as follows: for PmA, PK and P6, respectively; for Han, HK and H6, respectively; and for Cha, CK and C6, respectively.
The mRNA tag raw data processing, gene association, and differential expression analysis An average total of 11,841,476 high quality tags (clean tags) were obtained from each of the six libraries. After removing unmappable reads, 9,201,359 tags were shown to have at least one match (with less than 1 bp mismatch considering the existence of homoeo-alleles) in the reference tag database generated from 274,754 PlantGDB sequences (Release 163b). The unmatched tags may arise from the absence of ESTs in the current datasbase or from sequencing errors, and were not pursued further. More than 30,000 genes that were tagged by 291,026 unambiguous tags for each library were further analyzed for differential expression analysis (see Materials and methods for the data processing details, Table S1), with an average number of tags per gene as 137.9. A total of 6,157, 6,876, and 3,268 transcripts were found to be differentially expressed between 0 h and 6 h H 2 O 2 treatment in PmA, Han, and Cha respectively (p,0.001 and FDR,0.001). In PmA, the number of up-regulated transcripts (4,008) was nearly 2-fold that of the down-regulated transcripts (2,148), whereas the numbers of up-and down-regulated transcripts in Han (3,780 vs 3,095) and Cha (1,780 vs 1,487) were similar (Table S2). A BlastN search between tagged ESTs and sequences presented on Affymetrix wheat GeneChips showed that only a third of these differentially expressed transcripts were represented by the latter methodology, demonstrating that the high-throughput sequencing technology-based gene expression methodology provides more comprehensive information than the traditional array-based approach.
A correlation analysis showed that the correlation coefficients (R 2 ) between libraries of the same wheat line were higher (.0.95) than those between different wheat lines, regardless whether the plants were treated with H 2 O 2 or not. This result indicates significant variations in basal gene expression levels between the wheat lines of different genetic backgrounds (Table S3). This observation also cautions us to take all three of the lines into consideration when identifying genes that specifically respond to H 2 O 2 . A total of 28 differentially expressed genes between PK and P6, as measured in tag numbers, were then verified using qRT-PCR. The result showed .86% (24 out of 28 genes) consistency between the two methods (Table S4).

Functional classification of differential expression genes with consistent expression patterns in all three wheat lines
To reduce the potential variation derived from different genetic backgrounds, we firstly looked at the genes with consistent differential expression patterns in all three of the lines. A total of 260 genes fell in this category, with 135 genes up-regulated and 125 down-regulated ( Figure 2A, Table S5). We then manually classified these genes using functional categories as reported [15]. The numbers of up-and down-regulated genes were significantly different (Fisher's exact test P,0.05) in the functional categories related to cell rescue/defense, photosynthesis, and carbohydrate metabolism ( Figure 2B, 2C). For example, nearly a third (30 ESTs, or 29%) of the up-regulated genes were annotated as to be involved in cell rescue/defense, significantly more than the ones down regulated (9 ESTs,or 11% (Table 1). In contrast, the expressions of genes for other two categories, i.e. photosynthesis and carbohydrate metabolism, were all repressed. Twelve genes (14%) were associated with photosynthesis including those for the light harvest center such as chlorophyll A-B binding proteins (Ta-0112922, 036375, 053588, 129259, 146211, 19895, 2098165442, 3264165442, 98245) and photosynthesis I and II reaction center proteins (Ta-2040165448, 132758). The consequent repression of carbohydrate metabolism genes included key sucrose synthesis genes such as beta-amylase (Ta-155349), trehalose-6-phosphate synthase (Ta-2355165445), and ribulose bisphosphate carboxylase small chain precursor (Ta-3117165446; Table S5).
In addition, four categories contain similar numbers of induced and repressed genes. These categories were signal transduction, lipid metabolism, redox homeostasis, and transport. Genes for signal transduction, for example, were comprised of 27 (26%) induced genes and 16 (19%) repressed ones (Table S5)  genases (Ta-01383, 1776165447), fatty acid hyroxylase (Ta-45483), and acyl carrier protein (Ta-15083). There were nine genes (4 up-and 5 down-regulated) that were involved in redox homeostasis. The genes in this category encode dehydrogenases (Ta-21469, 31030), tropinone reductase 2 proteins (Ta-36863, Ta-060010), glutathione S-transferase (GST, Ta-66576), glutathione peroxidase (Ta-19785), peroxisomal membrane protein (Ta-121223), peroxidase precursor (Ta-010988), and thioredoxin (Ta-2210165449). There were seven genes that were involved in transport activity (2 up-and 5 down-regulated). The genes in this  Ta-01816), peptide transporter (Ta-149275), and trafficking particle complex subunit (Ta-95913). The genes in other category were mainly were involved in protein biosynthesis and degradation (Table S5). In addition, a group of 38 genes (15 up-and 23 downregulated) were annotated to encode putative, hypothetical, or expressed proteins of unknown functions. Finally, we compared these H 2 O 2 -regulated genes with those differential genes from several other wheat microarray works. As shown in Table 2, a number of H 2 O 2 regulated genes also participated in other biotic and abiotic stresses, such as heat, drought, and pathogen infections (leaf rust, yellow rust, and Fusarium pseudograminearum), confirming the diverse functions of H 2 O 2 in wheat stress responses. There seem to be more common genes differentially expressed under our condition and under drought and heat conditions [33]. This may be attributed to the similar time points for sample collection in the two experiments.

Gene Ontology analysis suggests H 2 O 2 -enhanced transport activities in PmA
In barley and wheat, H 2 O 2 can accumulate in the mesophyll cells during the early stage of the host-PM incompatible interaction, where it acts as an important signaling molecule to activate defense systems [11], [27]- [28]. Since PmA is resistant to Bgt isolate E09 while Han and Cha are susceptible, we studied in PmA H 2 O 2 responding genes that were ''subtracted'' with their expression patterns in Han and Cha. We defined three classes of PmA-specific (PmA-H 2 O 2 ) genes: class I comprises genes whose expression was induced in PmA (U) but suppressed (D) or remained unchanged (N) in both Han and Cha (UDD/UNN); class II includes genes that were suppressed in PmA (D) but induced or unchanged in Han and Cha (DUU/DNN); class III contains genes whose expression was not affected in PmA but changed in Han and Cha (NUU/NDD) (Table S6, S7). A total of 2,982 genes fell in these categories. Among them, nearly 60% of the genes (1,763) belong to class I genes (fold change $2, p,0.001, and FDR,0.001), while classes II and III genes comprise 28% (846) and 12% (373) respectively.
Gene Ontology (GO) analysis showed that class I genes were mostly enriched in functions associated with the biological process (BP) terms ''localization'', ''response to stimulus'', and ''metabolic process'', the molecular function (MF) terms ''transporter activity'', ''catalytic activity'' and ''binding'', and the cellular component (CC) terms ''plasma membrane'', ''vacuole'' and ''membrane-bounded organelle'' (Table 3). A total of 112 genes were associated with ''localization'' including 43 transporters, 18 protein targeting proteins, and 11 cell vesicle transporters (data not shown), among which a total of 21 genes were annotated as related to the vesicle-mediated transport (Table S8). Genes with ''transporter activity'' were 17.5 fold more over-represented than in the Arabidopsis ATH1 Genome Array ATH1 ( Table 3) that contains more than 22,500 probe sets representing approximately 24,000 genes. Transporter genes are known to be associated with defense responses by participating in the formation of multivesicular bodies and cell wall-associated paramural bodies that have been shown to be involved in secretion of building blocks for cell wall appositions [41]. These vesicle bodies not only arrest fungal penetration but also can cause hypersensitive cell death through blocking plasmodesmata. However, whether H 2 O 2 enhanced membrane transport, as observed here in wheat, plays a role in pathogen defense or not may need further experimental investigation.
GO analysis showed that class II genes were also enriched with ''localization'' and ''metabolic process'' (BP), ''catalytic activity'' and ''binding'' (MF), and ''plasma membrane'' (CC). The difference between the class II and the class I genes was the enrichment of the GO terms ''chloroplast stroma'' (CC, 10 fold enrichment) and ''thylakoid membrane'' (CC, 8.3 fold enrichment) in class II genes, suggesting significant effects on chloroplast functions by H 2 O 2 . The composition of the class III genes (up-or down-regulated only in both Han and Cha) was enriched with the BP term ''cellular component biogenesis'' and the CC term ''chloroplast stroma'' (10 fold enrichment). In contrast, the BP ''localization'' was not enriched among the genes of this category, demonstrating a difference in membrane transport activities between the PM susceptible and resistant lines. Therefore, these data point to possible roles of H 2 O 2 in pathogen defense, with enhanced vesicle transportation as one of the possible approaches.

MapMan analysis showed H 2 O 2 -enhanced JA/Et signaling pathway in PmA
To study the possible roles of H 2 O 2 responding genes in biotic response, we overlaid classes I, II, and III genes to the Arabidopsis GeneChip template in the MapMan program [42] and analyzed them in the biotic stress overview ( Figure S1). Figure 3A summarizes 10 major categories of genes that were associated with biotic response according to the MapMan classification. Among these, the most abundant genes (102 ESTs) were involved in protein degradation of various functions such as subtilases, cysteine protease, aspartate protease, serine protease, metalloprotease, AAA type, ubiquitin E1, ubiquitin E2 and ubiquitin E3 (Table S9, S10). Ubiquitination-associated proteins have been shown to play important roles in plant-microbe interactions [43]. The second group of genes (75 ESTs) was associated with signaling including more than 30 receptor kinases, calcium/calmodulin dependent kinases, and ras-related proteins ( Figure 3A; Table  S10). Protein kinases are well known to play a central role in the pathogen recognition and subsequent activation of plant defense [44]. The biotic stress overview also displays genes involved in redox homeostasis (31 ESTs), secondary metabolism (31 ESTs), cell wall (23 ESTs), heat shock proteins (13 ESTs), pathogenesisrelated proteins (PRs; 4 ESTs), and 31 ESTs that may be overlapping with additional abiotic responses. A total of 30 genes were involved in hormone signaling according to their annotation. As shown in Figure 3B, except for salicylic acid (SA), putative signaling genes for auxin, abscisic acid (ABA), JA, Ethylene (Et), and brassinolides (BR) were all affected by H 2 O 2 among PmA-H 2 O 2 genes. For ABA, Auxin, and BR, some genes were up regulated whereas others were down regulated. In contrast, all JA and Et signaling genes were up  (Table S11). Besides, two additional JA biogenesis-related genes Ta-80972 and Ta-04807 were also confirmed by qRT-PCR to be up regulated by H 2 O 2 (Table S11). Ta-80972 encodes a cytochrome P450 that is highly similar to a rice allene oxide synthase (AOS), a key enzyme in the oxylipin pathway leading to AOS-derived jasmonates, while Ta-048072 encodes the enzyme 12-oxophytodienoate (OPDA) reductase that is involved in JA biosynthesis by catalyzing the reduction of 10, 11-double bonds of OPDA to yield 3-oxo-2-(29pentenyl)-cyclopentane-1-octanoic acid (OPC-8:0). These results suggest that it is not only the signaling, but also the biogenesis of JA that may have been enhanced in the H 2 O 2 -treated wheat.

Co-regulation of H 2 O 2 responsive genes by Bgt inoculation in PmA
Several studies demonstrate that H 2 O 2 -induced genes can also be regulated by pathogen infections, including powdery mildew [18], [29]- [30]. To identify H 2 O 2 regulated genes that may also respond to wheat Bgt infection, 19 PmA-H 2 O 2 ESTs that were annotated as related to defense response were selected for expression study using qRT-PCR assay between PmA and its susceptible isogenic line Beijing837 (Bj; Table 4). Eight of these genes were found to have similar expression patterns between H 2 O 2 treatment and Bgt inoculation. Among them, five were class I genes including Ta-048072, 147978, 019566, 92123, and 02061 that encode 12-oxophytodienoate reductase (JA pathway), ethylene-responsive transcription factor (Et pathway), calcium/calmodulin dependent kinases (signaling), heavy metal-associated domain containing protein (cell rescue/defense), and stripe rust resistance Yr10-associated (cell rescue/defense) respectively. In the mean time, two class II genes Ta-138137 and Ta-2126165445 that respectively encode a helix-loop-helix DNA-binding domain containing protein (signaling) and an ABC transporter family protein (transport) were co-suppressed by both treatments. These H 2 O 2 and Bgt co-regulated genes may be involved in the PM responses. The last EST Ta-0109540 represents a gene encoding a fatty acid desaturase (named TaFAD) and is a class III gene that was significantly repressed by H 2 O 2 in the two PM susceptible lines Han and Cha (Table 4). In PmA, the fold change of the TaFAD expression under 0 and 6 h H 2 O 2 treatments was calculated as close to 0.5, but was not significant under our statistic threshold (p,0.001;). qRT-PCR assay showed that under Bgt inoculation, TaFAD was significantly (p,0.05) repressed by Bgt infection in the PM susceptible isogenic line Bj (6 h:0 h ratio 0.2560.08) when the change in PmA was marginal (6 h:0 h ratio 0.5360.09). A more detailed analysis over a 48 h Bgt inoculation course showed that TaFAD was indeed significantly repressed by Bgt infection in Bj (p,0.05), whereas in PmA the modest down regulation was statistically insignificant (Figure 4). These data demonstrate that the capability to maintain a constant expression level of TaFAD might be essential for PM resistant wheat lines.
In Arabidopsis and rice, FADs have been shown to be involved in disease resistance by modulating JA and SA signaling [48]- [49]. To study whether TaFAD may play a role in wheat PM resistance, we silenced TaFAD in PmA using virus-induced gene silencing  [51]. The results showed that down-regulation of TaFAD caused the compatibility between Bgt and PmA ( Figure 5). The reduction of TaFAD expression resulted in the growth of Bgt spores on the leaves of the BSMV:TaFAD plants and enhanced PM penetration efficiency (PE, 28% vs. 0%; Table S12). To ensure that no additional genes were affected by ''off target'' silencing, we also tested the expression level of TaFAD6, the next closest gene of TaFAD in wheat, and indeed no co-suppression was observed ( Figure 5A). These data thus suggest that the constant expression of TaFAD is indeed required for the PM resistance in PmA. Since FADs have been shown to be involved in disease resistance by modulating JA and SA signaling, we then investigated wheat genes homologous to functionally characterized rice marker genes in JA and SA signaling pathways in these VIGS plants. The SAsignaling gene used was the acidic pathogenesis-related protein gene PR1a (Table S13), whereas the JA-signaling genes encode a lipoxygenase (LOX, Ta-2498165446), an allene oxide synthase 2 (AOS2, Ta-134687), and a pathogenesis-related protein PR1b which acts in both the JA-and SA-dependent pathways [52]- [53]. As shown in Table S13, the JA pathway related genes, Ta-134687 (AOS2), Ta249865446 (LOX), and Ta-141027 (PR1b) were significantly down-regulated in BSMV:TaFAD plants, whereas the expression level of PR1a remained unchanged. Therefore, we deduce that the loss of PM resistance in BSMV:TaFAD plants was likely caused by the failure to maintain proper JA signaling. BSMV:HSP90 was used as a positive control for successful silencing of the disease resistance system because HSP90 is essential for the plant hypersensitive reaction (HR) to diverse pathogens as shown by previous studies [50]- [51], [54]- [56]. In BSMV:TaHSP90 VIGS plants, AOS2 and LOX were significantly up-regulated, whereas PR1a were down-regulated (Table S13), suggesting different regulatory modes of TaFAD and TaHSP90. Overall, these observations suggest the potential modulation of TaFAD on JA signaling during wheat powdery mildew infection, which deserves further investigation.

Discussion
In plant cells, H 2 O 2 plays a dual role as a toxic by-product of normal cell metabolism and as a regulatory molecule in stress perception and signal transduction [15]. A complex interplay between H 2 O 2 and other signaling molecules is known to exist in plant cells, which explains the versatility of H 2 O 2 functions in different scenarios. However, the knowledge about the transcriptome changes by exogenous H 2 O 2 treatment in wheat is limited, especially the knowledge about the H 2 O 2 -trggered defense related genes and signaling pathways. In this work, we performed a mRNA tag analysis of the wheat seedling transcriptome after 6 h 10 mM H 2 O 2 treatment using the next-generation sequencing technology. We found that H 2 O 2 caused differential expression of genes associated with important biological processes such as cell rescue/defense responses, photosynthesis, and carbohydrate metabolism. Further analysis in the PM resistant wheat line provided additional clues about the H 2 O 2 -triggered molecular pathways and the possible links of H 2 O 2 signaling and Bgt defense.

The effects of exogenous H 2 O 2 on gene expression profiles in wheat
In rice, the exposure of seedlings to exogenous H 2 O 2 caused leaves to curl severely [15]. By contrast, the wheat seedlings did not exhibit evident morphological changes after growth in H 2 O 2 solution for 6 hours. Consistent with this, we found only three percent (9 out of 260) differentially expressed genes related to redox detoxification in the transcripts with identical expression patterns in all three wheat lines, suggesting that dramatic oxidative stress had not taken place under the current experimental condition. Despite this, exposure of wheat seedling roots to H 2 O 2 stress did result in increased endogenous H 2 O 2 accumulation with impaired photosynthesis in wheat leaves. A total of 12 genes were associated with the photosynthetic process and were all down-regulated, together with 6 carbohydrate metabolism genes, indicating that H 2 O 2 can effectively repress photosynthesis and related carbohydrate metabolism in wheat.
Several studies show that exogenous H 2 O 2 can initiate signal transduction processes in treated plants that render them to acquire tolerance to various abiotic and biotic stresses. In Arabidopsis, for example, genes responding to exogenous H 2 O 2 are also involved in wilting, UV irradiation, and elicitor challenge response, indicating that H 2 O 2 can mediate cross-tolerance toward other stresses [12]. Extra H 2 O 2 caused by high light in a catalase-deficient Arabidopsis mutant can regulate the transcription of two clusters of genes that encode heat shock proteins (HSPs) known to be involved in stress responses [14]. Similarly, treatment of winter wheat with low concentrations of H 2 O 2 and catalase inhibitor may enhance its tolerance to low temperature [57]. We found significant activation of putative cell rescue/defense genes than repression in the same functional category, indicating that, as in a number of other studies, H 2 O 2 can stimulate defense responsive genes in wheat. Further comparative analysis showed that H 2 O 2 -responsive genes also participated in the response to other biotic and abiotic stresses, such as heat, drought and pathogen infections, indicating that at least some of these genes were indeed involved in defense responses.

The characteristics of H 2 O 2 responses in the powdery mildew resistant wheat
The accumulation of H 2 O 2 in the mesophyll cells has been observed in barley and wheat during the early stage of the host-PM incompatible interaction, indicating that H 2 O 2 works as an important signaling molecule in defense system activation [11], [27]- [28]. Two events were observed during this process: the accumulation of vesicles and vesicle-like materials at the cells in contact with microorganisms [41], [58] and the subsequent hypersensitive response (HR) in the epidermis cells directly nearby the penetration sites [11], [27]. We studied the differential gene responses in the PM resistant line PmA by GO enrichment analysis and found the ''localization'' was the most enriched BP term, together with the most enrichment GO MF term ''transporter activity'' and GO CC term ''Plasma membrane'' and ''vacuole''. The enrichment of these GO terms suggest that H 2 O 2 enhanced the membrane trafficking, supporting the hypothesis that trafficking of membrane-bound solutes, such as H 2 O 2, is essential not only for signaling but also for accommodating the cellular volume changes associated defense response [59].
It is nowadays impossible to discuss ROS and redox signaling in plants without considering plant hormones and related signal molecules because these compounds act together with redoxmodulated signaling pathways to process and transmit environmental cues into appropriate responses. Compounds strongly interacting with redox processes include classical hormones such as auxin, ethylene, and ABA, as well as defense-related signals such as SA and JA [60]- [65]. JA is synthesized from a-linolenic acid in chloroplast membranes. The three chloroplast-located enzymes 13-lipoxygenase (13-LOX), 13-allene oxide synthase (13-AOS), and the AOC catalyze the first half of JA biosynthesis up to the intermediate product cis-(+)-12-oxophytodienoic acid (OPDA). The Arabidopsis AOS promoter is activated by a variety of signals including jasmonic acid, wounding, OPDA, and SA, indicating that regulation of the AOS gene might exert a major control on JA signaling [66]. As for OPDA, it alone is sufficient to induce defense responses. The methylation of jasmonic acid to MeJA is catalyzed by an S-adenosyl-L-methionine:jasmonic acid carboxyl methyltransferase (JMT) in Arabidopsis. The expression of JMT itself is sufficient to induce some JA-dependent responses. In our study, we simultaneously detected the up-regulation of JA signaling and biogenesis genes including Ta-80972 (AOS), Ta-048072 (OPDA), Ta-07168, 15852, 2498165446 (LOX), and Ta-89552 (JMT). We also detected the activation of an ethylene receptor (ETR, Ta-2002165441) and an ethylene-responsive transcription factor (ERF, Ta-147978) by H 2 O 2 treatment in PmA. It has been shown that JA, alone or often in combination with Et, mainly works in defense to insect wounding and necrotrophic pathogen attack [45]- [47]. Our results suggest that H 2 O 2 may play a role in biotic stress by enhancing JA and/or Et signaling pathways. Whether these responsive genes also play any function in response to the infection of the biotrophic pathogens such as wheat PM is worth further investigation.

The potential links between H 2 O 2 signaling and Bgt defense
The recent identification of a serine/threonine kinase gene, Stpk-V as a PM resistant gene in wheat, provides further evidence that H 2 O 2 and Bgt co-regulated genes can be involved in disease resistance [30]. This gene however was not detected in our work, probably because it was an introgressed gene from Haynaldia villosa and did not exist in the wheat line used here. In our study, eight genes exhibited similar expression patterns under H 2 O 2 treatment and Bgt inoculation. These representative genes belong to various functional categories such as cell rescue/defense, signaling, JA/Et signaling pathways, transport, and lipid metabolism, and may participate in the PM defense. We show that the H 2 O 2 responsive fatty acid desaturase gene TaFAD is indeed involved in the PM resistance, indicating that, as in other plants, fatty acids play important roles in wheat pathogen defense. In Arabidopsis, oleic acid (18:1) has been implicated to participate in SA and JAmediated defense pathways [48], [67]- [70]. It has been shown that suppressing the gene for stearoyl acyl carrier protein fatty acid desaturase (SACPD) enhances the resistance of Arabidopsis (SSI2), soybean, and rice to multiple pathogens [48]- [49], [70]. In addition, another fatty acid desaturase gene FAD7 is required for the accumulation of the systemic acquired resistance-inducing activity [71], suggesting that different FAD family members may play distinct roles in plant defense. Therefore, the role of TaFAD in wheat PM resistance is not accidental since this gene is significantly repressed by both H 2 O 2 and Bgt in the PM susceptible lines Han, Cha and Bj while maintained a relative stable expression level in the PM resistant line PmA. The observation that silencing of TaFAD causes the loss of PM resistance in PmA and the significant down-regulation of the JA signaling pathway supports the idea that the maintenance of TaFAD function is crucial for the defense response, probably due to its potential regulatory role in modulating JA signaling. Whether or how H 2 O 2 regulates the TaFAD expression in PmA during Bgt infection needs further investigation.

An overview of the transcriptome changes caused by exogenous H 2 O 2 in wheat
Several issues that are intrinsic to the hexaploid nature of the bread wheat need to be taken into consideration. For example, the mRNA tag numbers should represent the collective levels of the three homoeologous alleles if they were all expressed. Although consensus primers can be applied during qRT-PCR when sequences for all three alleles are available, allele-specific studies will result in more precise conclusions. Despite this, our analysis for the first time provides a global insight about the transcriptomic response to exogenous H 2 O 2 treatment in wheat seedlings. Combining our observations and the previous studies, we present in Figure 6 several major cellular and metabolic processes in response to H 2 O 2 treatment in wheat. First, the application of exogenous H 2 O 2 may increase the cytosolic H 2 O 2 level causing the disruption of the redox homeostasis that can be relayed by MAPKs and type-2C protein phosphatase (PP2C) or perceived by the chloroplasts where both photosynthesis and carbohydrate metabolism are consequently suppressed. Second, the MAPK cascade and/or the retrograde signals from the chloroplast are then transmitted into the nucleus where the activation of transcription factors may initiate defense related genes. Third, the enhanced JA and Et signaling, accompanied with largely up regulated lipid metabolism (which may provide needed JA precursors), may positively regulate genes for defense response where H 2 O 2enhanced trafficking of membrane-bound solutes may also play an important role. Finally, we propose that possible links of H 2 O 2 signaling and Bgt defense may exist in the processes of signaling, transport, JA/Et pathway, and lipid metabolism (indicated by stars). Genes involved in these aspects of H 2 O 2 response may deserve further study for their roles in the PM resistance in wheat.

Plant Growth and Treatments
Am6 is a synthetic amphiploid derived from a cross between Triticum durum (AABB) accession DR147 and Aegilops tauschii (DD) accession Ae39 [72]. The line used for this study (PmA) is a BC5F3 progeny between Am6 and the cultivar Beijing837 with a novel powdery mildew resistant gene. PmA is resistant to the popular Blumeria graminis f. sp. tritici (Bgt) race No. 15 in the Beijing area, whose virulence type is E09, while Beijing837 is susceptible. Hanxuan10 is a drought tolerant cultivar and Chadianhong is known to be salt tolerant, both of which are susceptible to the Bgt race No. 15. Wheat seedlings were grown in a chamber at 22uC and a photoperiod of 16 h (60 mmol m 22 s 21 photon flux density). For H 2 O 2 treatment, 9-day-old seedlings were transferred to a Hoagland's solution containing 10 mM H 2 O 2 . For PM inoculation, the Bgt isolate E09 conidia on heavily diseased leaves were shaken off over a settling tower onto the wheat primary leaves, harvested at 0 and 6 h after treatment, and stored at 280uC until use. Inoculated plants were checked later to ensure proper development of powdery mildew on the leaves of the susceptible control.

Measurements of Photosynthesis Efficiency and H 2 O 2 Levels
Net photosynthetic rates of the first and second leaves of wheat seedlings were measured using a LI-COR 6400 portable gas analysis system with a light-emitting diode light source (LI-COR Inc., Lincoln, NE), with seven duplicates. H 2 O 2 accumulation in wheat leaves was measured according to a previously described method [73]. Briefly, leaves were grounded to a fine powder and extracted with 5 mM titanium sulfate. The oxidation of titanium sulfate was recorded by reading the absorbance at 410 nm. The readings were converted to corresponding concentrations using a standard calibration plot.

Tag Library Construction and the Initial Bioinformatic Processing
Wheat seedling leaves at 0 h and 6 h after H 2 O 2 treatment were used for mRNA tag library construction, as described by [74]. Sequencing was performed by Beijing Genomics Institute, Shenz-hen, China (http://sequencing.genomics.com.cn/). For the tag number counting, a reference tag database was generated using 274,754 PlantGDB sequences (Release 163b). For convenience, the prefix ''PUT-163b-Triticum_aestivum'' was replaced by a shorter one, ''Ta''. A total of 217,691 sequences were found to have GATC sites, which generated a total of 425,312 reference tags, with340,097 unambiguous tags (79.96%). The tags derived from wheat mRNA libraries were counted for redundancy and, therefore, copy numbers, using in-house Perl scripts. Numbers of all the tags on one EST or assembly were used to represent its expression levels.

mRNA Tag Data Analysis
Tag numbers for each gene were normalized with the total number of tags in the corresponding library. The statistical difference between corresponding genes in differently libraries was measured using a previously described method [75]. Expression level changes were calculated using the log2 ratios of transcripts per million mapped reads (TPM) between conditions (P6/PK, H6/HK, C6/CK). The sets of genes were selected for further analysis after the following filters: (1) TPM log2 ratios were either $1 for up regulation or #1 for down regulation; (2) the copy numbers for each condition should be $12; (3) the FDR for differential expression was set to be ,0.001. For convenience, transcripts with zero detected tags in one condition were arbitrarily designated as 0.01 TPM and marked, whereas others were normalized using the total tag numbers in one library (minimum 12 copies for tags to be included for analysis). For MapMan analysis, wheat EST sequences were compared using Blastx against Arabidopsis sequences represented on Affymetrix GeneChips. Fold change numbers were then transferred to the best Arabidopsis matches which were mapped to the Arabidopsis template in the MapMan program for display [42]. The manual functional classification of genes followed Wan and Liu [15]. Gene Ontology enrichment analysis was performed using Arabidopsis proteins as templates at AGRIGO website (http://bioinfo.cau. edu.cn/agriGO/).

Virus Induced Gene Silencing Assays
The plasmids utilized in these experiments were based on the constructs described by Holzberg et al. [76]. The virus-induced gene silencing of TaFAD (PlantGDB EST Ta-0109540) and TaHSP90 (TaHSP90.3-D1, GQ240789.1; TaHSP90.3-A1, GQ240787.1) [51] was performed using a d RNA vector, pSS031-1. TaHSP90 was used as a positive control for powdery mildew resistance. A 231-bp fragment of TaFAD was amplified from the plasmid pTaFAD with the forward primer, CTAGCTAGCGGGGTCTTCTGGTA-CAGC, and the reverse primer, CTAGCTAGCGACACGC-TACTCTTTCCTTT. A 355-bp fragment of TaHSP90 was amplified from the plasmid pTaHSP90 with the forward primer, CTAGCTAGCGAGACCTTCGCCTTCCAG, and the reverse primer, CTAGCTAGCCACCGAACTGCCCAATCA. The underlined bases are the NheI restriction sites. Plants were infected with BSMV using a modified protocol [50], [76]. Briefly, capped transcripts were prepared from three linearized plasmids that contain the tripartite barley stripe mosaic virus (BSMV) genome using the mMessage mMachine T7 in vitro transcription kit (Ambion, Austin, TX), following the manufacturer's protocol. These in vitro transcription reactions typically resulted in a final concentration of 1 to 1.5 mg/ml of RNA.

Bgt-wheat Interaction Assays
The method to estimate Bgt infection efficiency is largely in accord to Li et al. [11]. Wheat leaves of three centimeters were aligned on the surface of 0.5% agarose with 50 mg/L 2-[(4chlorophenenyl)methyl]-1H-benzimidazol and sprayed with Bgt spores (isolate E09) using an air compressor and nozzle. After 5 days, the leaves were bleached with trichloroacetic acid (1.5 g L 21 ) in ethanol:chloroform (4:1 v/v), stained with aniline blue (1 g L 21 ), and observed under a light microscope for the formation of elongated secondary hyphae. A segment of the same leaf was kept at 280uC for mRNA extraction and subsequent quantitative real time PCR (qRT-PCR). For VIGS plants, newly emerging leaves of 14 d after the viral inoculation (usually the fourth leaves) were used. At least 8 VIGS plants were tested for each VIGS vector with three independent biological replicates.

Measurements of Gene Expression Levels by qRT-PCR
For mRNA tag data and gene silencing confirmation, RNA was extracted using Trizol reagent (TIANGEN, China) and qRT-PCR experiments were performed on an ABI PrismH 7300 (Applied Biosystems, USA). The number of transcripts was normalized with the constitutively expressed glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA [77]- [78], which was tested as the most stable reference gene for the wheat seedling powdery mildew infection assay (data not shown).. The qRT-PCR assays were repeated three times, each with three biological replicates. To test the silencing efficiency, qRT-PCR primers were designed as: TaFAD forward, TACGTCGAGCCCGAGGACCG; TaFAD reverse, TGCCCCAAAATGCCCTCTTGCT; TaFAD6 (Ta-1294165443) forward, GGGAGAAGTCACCACCAA; and Ta-FAD6 reverse, GACCGAAAGCATACGAAG. RNA from the BSMV:GFP-treated plants was used as controls.