Tryptophan-derived metabolites are required for antifungal defense in the Arabidopsis mlo2 mutant.

Arabidopsis (Arabidopsis thaliana) genes MILDEW RESISTANCE LOCUS O2 (MLO2), MLO6, and MLO12 exhibit unequal genetic redundancy with respect to the modulation of defense responses against powdery mildew fungi and the control of developmental phenotypes such as premature leaf decay. We show that early chlorosis and necrosis of rosette leaves in mlo2 mlo6 mlo12 mutants reflects an authentic but untimely leaf senescence program. Comparative transcriptional profiling revealed that transcripts of several genes encoding tryptophan biosynthetic and metabolic enzymes hyperaccumulate during vegetative development in the mlo2 mlo6 mlo12 mutant. Elevated expression levels of these genes correlate with altered steady-state levels of several indolic metabolites, including the phytoalexin camalexin and indolic glucosinolates, during development in the mlo2 single mutant and the mlo2 mlo6 mlo12 triple mutant. Results of genetic epistasis analysis suggest a decisive role for indolic metabolites in mlo2-conditioned antifungal defense against both biotrophic powdery mildews and a camalexin-sensitive strain of the necrotrophic fungus Botrytis cinerea. The wound- and pathogen-responsive callose synthase POWDERY MILDEW RESISTANCE4/GLUCAN SYNTHASE-LIKE5 was found to be responsible for the spontaneous callose deposits in mlo2 mutant plants but dispensable for mlo2-conditioned penetration resistance. Our data strengthen the notion that powdery mildew resistance of mlo2 genotypes is based on the same defense execution machinery as innate antifungal immune responses that restrict the invasion of nonadapted fungal pathogens.

Powdery mildews represent a group of widespread ascomycete phytopathogens that can colonize a broad range of angiosperm plant species. Attempted fungal penetration into epidermal host cells triggers multifaceted plant defense-related responses, such as the transcriptional activation of PATHOGENESIS-RELATED (PR) genes, de novo cell wall biosynthesis underneath fungal contact sites (papillae), as well the biosynthesis and local extrusion of antimicrobial molecules (for review, see Eichmann and Hü ckelhoven, 2008). The toxic principles that ultimately cause abor-tion of fungal pathogenesis, however, remain largely enigmatic.
Pathogen-induced plant secondary metabolites with antimicrobial activity, also known as phytoalexins, are low M r compounds that are structurally diverse and often restricted in their occurrence to a limited number of plant species (Glawischnig, 2007). At present, the only phytoalexin known in Arabidopsis (Arabidopsis thaliana) is the indole derivative 3-thiazol-2#-yl-indole, also known as camalexin (Tsuji et al., 1992). Camalexin is derived from indole-3acetaldoxime (Glawischnig et al., 2004), which in turn is synthesized from Trp by the functionally redundant CYTOCHROME P450 monooxygenases CYP79B2 and CYP79B3 (Hull et al., 2000;Mikkelsen et al., 2000). The reaction catalyzed by CYP79B2 and CYP79B3 is the sole entry point for a range of biosynthetic pathways leading to diverse indolic metabolites, including camalexin and indole glucosinolates (1-thiob-D-glucosides; Supplemental Fig. S1). Consistently, cyp79B2 cyp79B3 double mutant plants are unable to accumulate these metabolites (Zhao et al., 2002;Glawischnig et al., 2004).
In a genetic screen for camalexin-deficient mutants, phytoalexin deficient3 (pad3) was isolated (Glazebrook and Ausubel, 1994) and subsequently shown to be defective in another cytochrome P450 enzyme, CYP71B15, which catalyzes the final step of camalexin biosynthesis (Zhou et al., 1999;Schuhegger et al., 2006;Supplemental Fig. S1). pad3 mutants have been extensively used to study the role of camalexin in plantpathogen interactions, leading to the proposition that camalexin contributes to resistance against necrotrophic pathogens but not against biotrophs (for review, see Glazebrook, 2005). For instance, attack by the powdery mildew fungus Golovinomyces orontii, one of several powdery mildew species that are virulent on Arabidopsis (Micali et al., 2008), does not trigger camalexin accumulation, and pad3 mutants do not exhibit enhanced susceptibility at the macroscopic level (Reuber et al., 1998).
In barley (Hordeum vulgare) and Arabidopsis, penetration attempts of powdery mildew fungi typically induce the localized formation of callose-containing cell wall appositions (papillae; Aist, 1976;Zeyen et al., 2002). Callose is a polymeric (1/3)-b-D-glucan that is synthesized by plasma membrane-resident GLUCAN SYNTHASE-LIKE (GSL) proteins. Callose deposition in papillae has been implicated in the highly efficient powdery mildew resistance of barley mildew resistance locus o (mlo) mutants, supposedly by contributing as a physical barrier to impede fungal entry into host cells (Skou, 1982(Skou, , 1985Bayles et al., 1990). However, recent genetic studies in Arabidopsis challenged a decisive role for papillary callose in antifungal defense. The POWDERY MILDEW RESISTANT4 (PMR4) gene, originally identified in a genetic screen for Arabidopsis mutants that are resistant to adapted powdery mildews, was found to encode GSL5. In rosette leaves, GSL5-generated callose is primarily deposited at wound sites and in pathogen-triggered papillae, suggesting that a lack of callose deposition in papillae does not compromise antifungal defense (Jacobs et al., 2003;Nishimura et al., 2003).
In the genetic screen leading to the identification of pmr4, further pmr mutants were isolated (Vogel and Somerville, 2000). One of these is defective in PMR2, which is allelic to MLO2, encoding a member of the seven-transmembrane-domain MLO protein family (Devoto et al., 2003). Mutant versions of the founder of this protein family, barley Mlo, are known to confer broad-spectrum resistance to powdery mildew fungi (Jørgensen, 1992;Bü schges et al., 1997). Arabidopsis MLO2 and its closest homologs, MLO6 and MLO12, were found to be required for full susceptibility against powdery mildews in Arabidopsis . While mutations in MLO2 alone confer partial resistance against G. orontii and Golovinomyces cichoracearum, additional mutations in MLO6 and MLO12 resulted in full immunity, which is characterized by early termination of fungal pathogenesis before successful penetration of the host cell wall . This infection phenotype is reminiscent of fully resistant barley mlo single mutants. Collectively, these findings suggest that distantly related powdery mildew species rely on functionally conserved host proteins in dicot and monocot plants for successful pathogenesis (Panstruga, 2005). Partial mlo2 resistance in Arabidopsis depends on three PENETRATION (PEN) genes that were originally discovered based on their requirement for effective extracellular defense responses against the nonadapted powdery mildews Blumeria graminis f. sp. hordei and Erysiphe pisi (Collins et al., 2003;Lipka et al., 2005;Stein et al., 2006). PEN1 encodes a plasma membraneresident syntaxin (t-SNARE) involved in exocytosis (Collins et al., 2003;Kwon et al., 2008). PEN2 codes for an atypical myrosinase (Bednarek et al., 2009) and cofunctions with the plasma membrane-resident PEN3 ATP-binding cassette multidrug transporter in a parallel extracellular defense pathway, presumably by targeted delivery of indole glucosinolate-derived antimicrobial metabolites into the apoplastic space (Lipka et al., 2005;Stein et al., 2006;Bednarek et al., 2009). Besides their supposed antimicrobial capacity, indolic glucosinolates may have an additional role as signaling molecules during innate immune responses (Clay et al., 2009).
Mutations in MLO genes result not only in resistance against powdery mildew fungi but also in additional, developmentally controlled pleiotropic phenotypes. Spontaneous accumulation of callose in leaf mesophyll cells and early leaf chlorosis/necrosis that is reminiscent of senescence was observed both in barley and Arabidopsis mlo mutants (Wolter et al., 1993;Piffanelli et al., 2002;Consonni et al., 2006). In Arabidopsis, this phenotype is fully dependent on salicylic acid (SA) accumulation but independent of jasmonic acid (JA) and ethylene (ET) biosynthesis and signaling , demonstrating separate requirements for the desired disease resistance trait and undesired leaf chlorosis/necrosis in mlo mutant plants.
Here, we employed comparative global gene expression analysis and performed targeted metabolite profiling to obtain deeper insights into the molecular basis of the pleiotropic phenotypes in the Arabidopsis mlo2 single mutant and mlo2 mlo6 mlo12 triple mutant. We found aberrant accumulation patterns of indolic secondary metabolites in the single and triple mutants during the appearance of the mlo2-conditioned early leaf senescence phenotype. Genetic analysis revealed a critical contribution of indolics, including camalexin and glucosinolates, in mlo2-mediated resistance. days (16 h of light) start to show leaf chlorosis and necrosis at around 6 weeks after sowing, this appearance arises considerably later (from 9 weeks onward) in plants grown in short-day conditions (10 h of light; data not shown).
To find out whether this phenotype is an authentic senescence process, we measured plant photosynthetic performance (photochemical efficiency of PSII [F v /F m ]) to assess functional leaf longevity (Maxwell and Johnson, 2000;Oh et al., 2003;Kusaba et al., 2007). The F v /F m ratio reflects the quantity of light energy absorbed by PSII that is used for photosynthesis (photochemical efficiency). The optimal value for this parameter is expected to be around 0.83 for leaves of young and healthy plants and decreases with plant age, owing to senescence (Woo et al., 2001;Kim et al., 2006). Photosynthetic performance was measured on defined rosette leaves (leaf 7) of intact plants every 2 to 6 d starting from 24 d after sowing until day 46, corresponding to a rosette of 50% final size (growth stage 3.50) to midflowering (growth stage 6.50; Boyes et al., 2001), respectively. Excellent photosynthetic performance (F v /F m . 0.8) was observed for both wild-type and mlo2 mlo6 mlo12 plants at the beginning of the time course (24-38 d; Fig. 1A; Supplemental Fig.  S2A), indicating that photochemical efficiency is not constitutively impaired in the mlo triple mutant. From day 40 onward, the F v /F m ratio declined in both genotypes, although somewhat faster in the mlo2 mlo6 mlo12 triple mutant ( Fig. 1A; Supplemental Fig.  S2A). However, this difference was not statistically significant.
We determined the chlorophyll content in ecotype Columbia (Col-0) wild type and mlo2 mlo6 mlo12 mutant plants throughout development by processing the same leaves that were used for measuring photosynthetic performance. Chlorophyll levels continuously decreased from day 28 onward in both wild-type and mutant plants. However, chlorophyll decay occurred faster in mlo2 mlo6 mlo12 compared with the wild type, signifying accelerated chlorophyll catabolism in the mlo triple mutant ( Fig. 1B; Supplemental Fig. S2B). Taken together, these results reveal that typical senescence-associated physiological markers such as photosynthetic performance and chlorophyll content exhibit a trend toward a more rapid decline in the mlo2 mlo6 mlo12 triple mutant, suggesting that mlo-conditioned leaf chlorosis and necrosis in Arabidopsis might be mechanistically related or identical to leaf senescence.
A developmental program that, despite some differences, largely phenocopies the authentic senescence processes of intact plants in time lapse is induced upon placing detached leaves in the dark (so-called darkinduced senescence; Weaver and Amasino, 2001). In comparison with natural senescence, this procedure has the advantage that the initiation of senescence is synchronized among individuals and genotypes. Thus, to further test whether the early leaf chlorosis and necrosis of mlo mutants resembles genuine senes-cence, we comparatively analyzed natural whole plant development and artificial dark-induced senescence of detached leaves in Col-0 wild-type, mlo2 single mutant, and mlo2 mlo6 mlo12 triple mutant plants. At the whole plant level, 7-week-old mlo2 mutants showed the previously reported early leaf chlorosis and necrosis that is exacerbated in mlo2 mlo6 mlo12 plants Fig. 1C, top row). To trigger dark-induced senescence, the fifth leaf of 4-week-old plants grown in long-day conditions was detached and kept in the dark at 22°C. After 4 d of incubation, leaves of Col-0 plants were still largely green, while leaves of the mlo2 single mutant and mlo2 mlo6 mlo12 triple mutant plants exhibited mild and variable (mlo2) or pronounced and consistent (mlo2 mlo6 mlo12) chlorosis (Fig. 1C, bottom row; see also Figs. 4B and 6C below). This finding further strengthens the notion that the mlo2-associated phenotype represents authentic but untimely leaf senescence and in addition provides further support for accelerated progression of the senescence process in the mlo2 and mlo2 mlo6 mlo12 mutants.

Transcript Accumulation of MLO2 Peaks around the Onset of Leaf Senescence
The expression pattern of Arabidopsis MLO genes, including MLO2, was recently studied by compiling data from transgenic promoter::GUS reporter lines, reverse transcription (RT)-PCR analysis, and publicly accessible microarray experiments (Chen et al., 2006). To expand this work, we used a transgenic line harboring an MLO2 promoter::GUS reporter gene construct to temporally and spatially resolve MLO2 promoter-driven gene expression during vegetative development of Arabidopsis plants grown in shortday (10 h of light) conditions. At 3 weeks after sowing, we observed GUS activity predominantly in the cotyledons and the margins of rosette leaves ( Fig. 2A). Intensity and coverage of GUS staining in rosette leaves increased with leaf and plant age, peaking in 6-week-old plants, approximately around the onset of spontaneous callose deposition in the mlo2 mutant . Thereafter, we noted an overall slight decrease in GUS staining intensity. Consistent with publicly accessible microarray data (Schmid et al., 2005;Winter et al., 2007), younger rosette leaves consistently showed less GUS staining throughout vegetative development than older leaves (Fig. 2). In sum, these findings are reminiscent of the observed kinetics of Mlo transcript accumulation in barley (Piffanelli et al., 2002) and corroborate a potential functional role for these monocot and dicot Mlo orthologs in senescence-associated physiology.

Transcriptomic Approaches to Unravel the Molecular Basis of mlo-Associated Pleiotropic Phenotypes
To gain further knowledge on the molecular mechanisms that underlie the untimely leaf senescence phenotype of the mlo2 mlo6 mlo12 mutant, we performed global gene expression profiling of leaf material collected at different developmental stages using Affymetrix ATH1 GeneChips. Spontaneous callose deposition is a convenient marker for the onset of mlo-conditioned leaf senescence and was observed from 6 weeks onward in healthy plants grown in short-day conditions (10 h of light; Consonni et al., 2006). Therefore, we collected mature rosette leaves from unchallenged wild-type and triple mutant plants at 5, 6, and 7 weeks after sowing, signifying time points before (5 weeks), at (6 weeks), and after (7 weeks) the onset of spontaneous callose deposition in the mutant. During this time period, no other signatures of leaf senescence, such as chlorosis and necrosis, were visible in the wild type or the mlo triple mutant. Total RNAs extracted from these leaves were labeled with fluorescent dyes and employed in a single hybridization experiment using Affymetrix ATH1 oligonucleotide arrays (for details, see "Materials and Methods").
Consistent with the occurrence of callose deposits at 6 and 7 weeks, unsupervised clustering of the six data sets showed that transcript accumulation in leaves of wild-type and mlo2 mlo6 mlo12 mutant plants was most similar in 5-week-old plants and exhibited increasing differences in 6-and 7-week-old plants (Supplemental Fig. S3). We analyzed the data to identify candidate genes that exhibit differential transcript levels between wild-type and mutant plants during the developmental period considered. Based on two distinct computational approaches, we generated two result lists, one sorting all 22,810 genes (list A) and the other representing a set of 98 selected genes (list B; Supplemental Table S1). List A is focused on genes more highly expressed in the mutant compared with the wild-type plants when comparing weeks 6 and 7 with week 5. For list B, only those genes were selected with elevated transcript levels specific to the mutant plants at week 7 compared with week 5, while making sure fold changes and absolute intensities are high (for details about the selection criteria and computation procedure, see "Materials and Methods").
In list B as well as in the top approximately 120 genes of list A, we noted a high prevalence of genes whose products are related to the biosynthesis of Trp and Trp-derived (indolic) metabolites. Consistently, Figure 1. Developmentally controlled mlo2-associated leaf chlorosis and necrosis resembles an authentic leaf senescence program. A, Timecourse analysis of photosynthetic performance (F v /F m ) of Col-0 wildtype (black circles) and mlo2 mlo6 mlo12 (white circles) mutant plants. Data represent means 6 SD of four independent rosette leaves (leaf 7) measured at the time points indicated (days after sowing). Plants were grown in long-day conditions. The experiment was repeated twice with similar results (Supplemental Fig. S2; data not shown). B, Time-course analysis of chlorophyll content in leaves of Col-0 wild-type (black circles) and mlo2 mlo6 mlo12 (white circles) mutant plants. Data represent means 6 SD of two independent rosette leaves (leaf 7) collected at the time points indicated (days after sowing) and measured with three technical replicates. Plants were grown in long-day conditions. Note that the very same leaves were taken for measuring chlorophyll content and for determining photosynthetic performance. The experiment was repeated once with similar results (Supplemental Fig. S2). FW, Fresh weight. Asterisks denote statistically significant differences (P , 0.05; Student's t test) from the Col-0 wild type. C, Habitus of representative unchallenged (pathogen-free) plants at 7 weeks after sowing (top row) and macroscopic phenotypes of detached leaves (leaf 5) from 4-week-old plants dark treated for 4 d (bottom row). Plants were grown in long-day conditions. The experiment was repeated four times with similar results. we found a significant enrichment of genes with the Gene Ontology (GO) term "indole derivative biosynthetic process" in list A (adjusted P = 0.009) and in list B (P = 0.03; for details about the calculation procedure, see "Materials and Methods"). With respect to Trp biosynthesis, these comprised the genes encoding the anthranilate synthase ASA1, phosphoribosylanthranilate transferase (PAT1), the Trp synthase a-subunit TSA1, and indole-3-glycerol phosphate synthase (IGPS). With regard to indole/camalexin/glucosinolate biosynthesis, the genes encoding the cytochrome P450 monooxygenases CYP79B2, CYP71B15/PAD3, CYP71A13, CYP81F2, and CYP83B1 as well as the sulfotransferase SOT16 and the UDP-glucosyltransferase UGT74B1 were included. Finally, the transcription factor MYB51, a key regulator of the genes encoding indole glucosinolate biosynthesis enzymes (Gigolashvili et al., 2007), was found in the top of list A (Table I;  Supplemental Table S1). Together, these findings indicate a potential role for indolic secondary metabolites in the mlo2-associated deregulated leaf senescence phenotype. Among the identified genes in list B, two are known to be specifically related to early leaf senescence (WRKY53 and SAG13; Lohman et al., 1994;Hinderhofer and Zentgraf, 2001;Miao et al., 2004), suggesting that a senescence-related process might have become initiated in the mlo2 mlo6 mlo12 mutant but not yet in wild-type plants (Table I; Supplemental Table S1).

Indolic Metabolites in
To validate the altered transcript levels of this set of genes in the mlo2 mlo6 mlo12 triple mutant, we exemplarily compared expression levels of a subset of these by real-time RT-PCR in mature rosette leaves of mutant and wild-type plants. Besides genes that were found at both the top of list A and in list B (CYP79B2, PAD3, and WRKY53), we included SAG13, a characteristic marker of early leaf senescence (Lohman et al., 1994), which occurred in list B but was ranked at a lower position in list A (Table I). Consistent with the microarray data, at the late time point (7 weeks) all tested genes showed considerably higher transcript accumulation in the mlo triple mutant compared with the wild type ( Fig. 3A; Supplemental Fig. S4). In sum, the results of the quantitative RT-PCR analysis corroborate the microarray data, indicating elevated transcript levels of genes of the Trp biosynthesis and metabolism pathways as well as initiation of a senescence-related developmental program in mlo2 mlo6 mlo12 mutant plants.
We extended the quantitative RT-PCR analysis to investigate whether a similar transcript pattern for the selected genes also occurred in plants grown in long-day conditions. In this experiment, mature rosette leaves were collected at early (4 weeks) and late (5 weeks) time points from plants grown in a pathogenfree environment with a 16-h-light/8-h-dark cycle. The expression patterns of the genes considered were similar to those observed in short-day-grown plants, indicating that the elevated transcript accumulation of these genes in the mlo2 mlo6 mlo12 triple mutant is independent of short-or long-day growth conditions but dependent on the developmental stage and the genetic background (Fig. 3A).

Comparative Proteomic and Metabolomic Analysis of Wild-Type and mlo2 mlo6 mlo12 Plants
To study whether the observed changes in the transcriptome would concur with detectable alterations in Figure 2. Transcript accumulation of MLO2 peaks around the onset of leaf senescence. A, Time-course analysis of MLO2 promoter-driven GUS expression in a transgenic Col-0 wild-type plant during vegetative development. Plants were grown in short-day conditions (10 h of light), and entire plants of the indicated ages were stained for GUS activity. The photographs depict exemplary plants from one experiment; similar results were obtained in two independent replicate experiments. B, Schematic representation of AtMLO2 expression data on publicly accessible microarray databases. The cartoon represents leaves at various developmental stages, color coded with the respective AtMLO2 expression level (according to the reference color bar shown at the bottom). Microarray source data are from the AtGenExpress project (Schmid et al., 2005). The pictograph is a screenshot from the Arabidopsis eFP browser (http://bbc.botany.utoronto.ca/efp/cgi-bin/ efpWeb.cgi; Winter et al., 2007). c, Cauline leaf; s, senescent leaf.
the leaf proteome, we performed comparative twodimensional gel electrophoresis of total soluble protein extracts obtained from rosette leaves of 7-week-old plants grown under short-day conditions. This analysis revealed no obvious differences between the mlo2 mlo6 mlo12 triple mutant and the Col-0 wild type, demonstrating that at this level of resolution deregulated leaf senescence does not correlate with altered protein steady-state levels (Supplemental Fig. S5).
Finally, we conducted a comparative metabolite profile of the mlo2 mlo6 mlo12 mutant and wild-type plants. Guided by the results of the microarray analysis, we focused this approach on indolic metabolites. Leaf material (leaf 7) was collected from plants grown in short-day conditions at 5, 6, 7, and 8 weeks after sowing. Crude leaf extracts were analyzed by HPLC coupled to a diode array detector or a fluorescence detector. Consistent with the elevated PAD3 transcript levels (see above), we found a trend toward higher camalexin concentrations in the triple mutant than in the wild type from 6 weeks after sowing onward (Fig.  3B). Similarly, an unknown derivative of camalexin was found at higher levels in the triple mutant at all time points tested (statistically significant at 8 weeks; Supplemental Fig. S6A). A differential pattern similar to the one observed for camalexin and its derivative was also detected for another Trp-derived compound, indol-3-yl-methylamine (I3A; statistically significant at 7 and 8 weeks), which was recently found to be an indole glucosinolate metabolism product that is indirectly linked to plant defense responses (Bednarek et al., 2009). Unlike I3A, two indole glucosinolates, indol-3-yl-methyl glucosinolate (I3G) and its 4-methoxylated derivative 4-methoxy-indol-3-yl-methyl glucosinolate (4MI3G), had lower levels in the triple mutant compared with the wild type (statistically significant at 5, 6, and 8 weeks [I3G] and 6 and 8 weeks [4MI3G]). Taken together, these results indicate that a globally altered profile of Trp-derived metabolites precedes deregulated leaf senescence in the mlo triple mutant, raising the possibility that perturbed levels of indolics could be responsible for the latter phenotype.
To find out to what extent the developmentally controlled hyperaccumulation of the Trp-derived metabolites camalexin and I3A is different in the mlo2 mlo6 mlo12 triple mutant and the mlo2 single mutant, we examined leaves of 7-week-old plants of these mutant genotypes and of Col-0 wild type. This analysis revealed similar trends but higher absolute levels in the triple mutant than in the mlo2 single mutant for both metabolites (Fig. 4A). This finding is reminiscent of the quantitative differences in powdery mildew resistance and premature leaf senescence in these two genotypes Fig. 1C) and opens up the possibility to dissect the genetic requirements for the altered accumulation of indolic compounds in the more genetically tractable mlo2 single mutant. To examine the contribution of known biosynthetic enzymes to the altered levels of indolics in the mlo2 mutant, we generated the mlo2 pad3 double mutant and the mlo2 cyp79B2 cyp79B3 triple mutant. The pad3 single mutant and mlo2 pad3 double mutant revealed quantities of camalexin and its derivative that were close to the detection limit, whereas I3A, I3G, and 4MI3G levels were variable but similar overall to Col-0 wild-type plants and the mlo2 single mutant ( Fig. 4A; Table I. Selected genes identified by microarray analysis Genes related to Trp biosynthesis/metabolism and early leaf senescence were selected by manual inspection of the top of list A (top 150 genes) and list B (Supplemental Table S1).  S6B). These results demonstrate that the Trp-derived compounds hyperaccumulating in the mlo2 mutant originate from their conventional biosynthetic route and that the respective double and triple mutants can be used to assess the role of indolic compounds in mlo2-associated premature leaf senescence. It was recently shown that PEN2 acts as a myrosinase that hydrolyzes both nonsubstituted and 4-methoxylated indole glucosinolates and that is essential for the pathogen-inducible accumulation of the glucosinolate hydrolysis product, I3A (Bednarek et al., 2009). To test whether the mlo2-associated increase in levels of this compound is also dependent on PEN2 activity, the mlo2 pen2 double mutant was analyzed. This double mutant indeed exhibited levels of I3A comparable to the pen2 single mutant (Fig. 4A), demonstrating the requirement of PEN2 activity for the mlo2-mediated I3A hyperaccumulation.
We also examined whether a mutation in the wellcharacterized senescence-related gene WRKY53, which was found to have elevated transcript levels in the mlo2 mlo6 mlo12 triple mutant ( Fig. 2A), was sufficient to restore wild-type levels of Trp derivatives in a mlo2 genetic background. Therefore, we generated a mlo2 wrky53 double mutant and performed metabolic profiling. This revealed mlo2-like metabolite levels, suggesting that the perturbation of indolic compounds in the mlo2 mutant is unlikely to be a direct or an indirect consequence of altered WRKY53 transcript accumulation (Fig. 4A). However, resolution of this analysis is limited, since the differences in indolic metabolite levels between the wild type and the mlo2 mutant were not statistically significant.

mlo2-Associated Early Leaf Senescence Can Be Uncoupled from the Accumulation of Indolic Metabolites
To determine the role of Trp-derived compounds in mlo2-associated early leaf senescence, both whole plant and dark-induced senescence phenotypes were analyzed for the set of mutants described above. No obvious differences were observed between mlo2 single and mlo2 pad3 double mutants, whereas a slight enhancement of leaf senescence symptoms was found in the mlo2 cyp79B2 cyp79B3 triple mutant (Fig. 4B). Notably, a mutation in the gene encoding the senescence regulator WRKY53 did not suppress mlo2-associated leaf chlorosis and necrosis (Fig. 4B), indicating that  Fig. S3). B, Secondary metabolite profile of leaves of the mlo2 mlo6 mlo12 triple mutant compared with the Col-0 wild type. Rosette leaf 7 of independent plants grown in short-day conditions was collected at 5, 6, 7, and 8 weeks after sowing and used for metabolite analysis. The graphs show the average 6 SD of a representative experiment comprising three leaf samples. The experiment was repeated once with similar results. The left y axis refers to camalexin and I3A, whereas the right y axis applies to I3G and 4MI3G. A statistically significant difference from Col-0 is indicated either by an asterisk (P , 0.01; Student's t test) or the number sign (P , 0.05; Student's t test). FW, Fresh weight.
the pathway leading to mlo2-conditioned premature senescence is either not under the control of WRKY53 or that redundancy in gene functions masks the con-tribution of this transcriptional regulator in the context of mlo2. Early leaf senescence of mlo2 mutants thus differs at least in this respect from conventional leaf are not the primary cause of the early leaf senescence phenotype. A, Leaf 7 from nine independent plants grown in short-day conditions was collected at 7 weeks after sowing, and HPLC analysis was performed on metabolite extracts. The graphs show averages 6 SD of a representative experiment comprising three technical replicates. One typical experiment out of three is shown. Asterisks denote statistically significant differences (P , 0.05; Student's t test) from the Col-0 wild type, and number signs denote statistically significant differences (P , 0.05; Student's t test) from the mlo2 mutant. FW, Fresh weight. B, Habitus of representative unchallenged (pathogen-free) plants at 7 weeks after sowing (top row) and macroscopic phenotypes of detached leaf 5 from 4-week-old plants dark treated for 4 d (bottom row). Plants were grown in long-day conditions. The experiment was repeated once (except for Col-0, mlo2, and mlo2 mlo6 mlo12, which had five replicates) with similar results.
senescence (Hinderhofer and Zentgraf, 2001). Taken together, these data suggest that altered levels of Trp-derived metabolites and/or aberrant WRKY53 activity are not the major cause of mlo2-associated early senescence.

mlo2-Mediated Pathogen Resistance Depends on the Biosynthesis of Indolic Secondary Metabolites
Given the key role of indolic metabolites in defense against nonadapted powdery mildew fungi (Bednarek et al., 2009), we next investigated the functional contribution of Trp-derived compounds in mlo2-mediated powdery mildew resistance. We challenged the mlo2 pad3 double mutant and the mlo2 cyp79B2 cyp79B3 triple mutant with the adapted pathogen G. orontii and quantified infection success by microscopic analysis of fungal host cell entry and conidiophore production . The mlo2 pad3 double mutant displayed partially restored host cell entry and conidiation compared with the mlo2 single mutant, suggesting a requirement for camalexin for full mlo2 resistance. Remarkably, the triple mutant mlo2 cyp79B2 cyp79B3 exhibited an infection phenotype that was similar to the Col-0 wild type, indicating that both camalexin and other indolic metabolites are major determinants of mlo2 resistance against the adapted powdery mildew pathogen (Fig. 5, A and B). With respect to host cell entry, similar results were also obtained with the nonadapted pea (Pisum sativum) powdery mildew pathogen Erysiphe pisi (Fig. 5C). However, this plant-pathogen constellation in addition reveals the effect of the cyp79B2 cyp79B3 double mutant in the context of MLO2, which owing to saturating fungal entry levels is masked upon challenge with the adapted powdery mildew pathogen G. orontii (Fig. 5, compare B and C). In sum, these data indicate that Trp-derived compounds, including camalexin, are crucial for preinvasion and postinvasion resistance against both adapted and nonadapted powdery mildew fungi in wild-type plants as well as in mlo2 mutant plants.
The phytoalexin camalexin is known to contribute to defense against necrotrophic fungi (Glazebrook, 2005). It was recently shown that sensitivity of Botrytis cinerea to camalexin is isolate specific and that this necrotrophic fungus also exhibits variable sensitivity to indolic glucosinolates (Kliebenstein et al., 2005). We inoculated 4-week-old wild-type and mutant plants with a camalexin-sensitive isolate of B. cinerea. Compared with the susceptible Col-0 wild type, the mlo2 single mutant and mlo2 mlo6 mlo12 triple mutant were highly resistant, as no spreading lesions were detected upon pathogen treatment (Fig. 6). To test whether this phenotype was also dependent on indolic metabolites, the double mutant mlo2 pad3 and the triple mutant mlo2 cyp79B2 cyp79B3 were inoculated with the same B. cinerea isolate. Consistent with the effect of these mutations in a wild-type background on camalexin-sensitive B. cinerea strains (Kliebenstein et al., 2005), mutations in these genes fully reverted mlo2-conditioned resistance (Fig. 6), suggesting that Trp-derived compounds play a major role in mlo2-mediated resistance against both biotrophic and necrotrophic fungi.
mlo2 Resistance Is Independent of PMR4/GSL5-Mediated Callose Deposition mlo2 mutants spontaneously accumulate callose depositions during later developmental stages, before the onset of macroscopically visible signs of early leaf senescence . We reasoned that the PMR4/GSL5 callose synthase, which mediates wound and papillary callose deposition in Arabidopsis rosette leaves (Jacobs et al., 2003;Nishimura et al., 2003), could be responsible for this phenotype. To test this hypothesis, the mlo2 pmr4 double mutant was generated and spontaneous callose deposition was analyzed in a time-course experiment using plants that were not challenged by any pathogen. Lack of the mlo2-characteristic callose deposits in rosette leaves of 6-week-old mlo2 pmr4 double mutant plants demonstrates that the PMR4 glucan synthase-like polypeptide is accountable for the developmentally controlled biosynthesis of callose in mlo2 (Fig. 7A).
pmr4 plants are macroscopically resistant against powdery mildew attack (Jacobs et al., 2003;Nishimura et al., 2003). In these studies, however, it was not quantitatively resolved at which stage of fungal pathogenesis that resistance in the pmr4 mutant becomes effective. We challenged control and pmr4 mutant plants with G. orontii and microscopically scored host cell entry and conidiophore formation. This revealed a fungal host cell entry rate in the pmr4 mutant that is comparable to wild-type plants. In contrast, further development of fungal colonies was found to be severely impaired, resulting in a significant reduction of conidiophore formation (Fig. 7B). A defense response that becomes effective at the postpenetration stage is thus the primary reason for the macroscopic resistance phenotype of the pmr4 mutant. To investigate whether local pathogen-triggered PMR4-dependent callose deposition in cell wall appositions is required for the mlo2-conditioned powdery mildew resistance, the mlo2 pmr4 double mutant was assessed upon G. orontii challenge. Intriguingly, no differences between mlo2 single and mlo2 pmr4 double mutant plants were observed with respect to both penetration rate and the level of conidiophore formation (Fig. 7B). This finding indicates that mlo2 resistance is fully independent of PMR4-mediated callose deposition.
pmr4 mutants were described to constitutively hyperaccumulate SA, and mutations in components of the SA pathway, such as npr1 and pad4, as well as expression of the bacterial salicylate hydroxylase gene, NahG, were reported to suppress resistance of pmr4 mutants against G. cichoracearum (Nishimura et al., 2003). To investigate whether powdery mildew resis-tance retained in the mlo2 pmr4 double mutant was the result of pmr4-dependent SA hyperaccumulation, mlo2 pmr4 plants were crossed with the sid2 mutant, which is defective in defense-associated SA biosynthesis (Wildermuth et al., 2001), to generate a mlo2 pmr4 sid2 triple mutant. Both macroscopic and quantitative microscopic analyses of mlo2 pmr4 sid2 plants challenged with the adapted G. orontii revealed a mlo2-like phenotype, demonstrating that the resistance retained in the mlo2 pmr4 double mutant is independent of SA accumulation (Fig. 7B).
Since callose deposition strictly precedes macroscopically visible signs of early leaf senescence in the mlo2 mutant Supplemental Fig. S7), we reasoned that a causal relationship may dictate the sequential order of the two events. If true, then a genetic block in callose deposition should alleviate the early leaf senescence phenotype of mlo2 plants. We first assessed the pmr4 mutant with respect to natural and dark-induced leaf senescence and observed early leaf chlorosis and necrosis that is reminiscent of the mlo2 mutant phenotype (Fig. 7C). This phenomenon was exaggerated in the mlo2 pmr4 double mutant, suggesting that the phenotypes of the mlo2 and pmr4 mutants are additive and excluding a direct causal link between callose deposition and mlo2-conditioned early leaf senescence. To test whether the exaggerated phenotype of the mlo2 pmr4 double mutant is an effect of the combined SA hyperaccumulation that is known to take place in mlo2 as well as pmr4 mutant plants (Nishimura et al., 2003;Consonni et al., 2006), we analyzed whole rosette and dark-induced senescence of the mlo2 pmr4 sid2 triple mutant. Indeed, lack of SID2 function partially suppressed both types of senescence in the mlo2 pmr4 double mutant, suggesting that SA (hyper)accumulation is in part the cause for this phenotype (Fig. 7C).

mlo2-Mediated Powdery Mildew Resistance Requires Trp-Derived Secondary Metabolites
In this study, we found that mlo2 single and mlo2 mlo6 mlo12 triple mutants accumulate aberrant levels of a subset of Trp-derived compounds during vegetative development (Figs. 3B and 4A). The consistently lower glucosinolate (I3G) but higher glucosinolate breakdown product (I3A) levels in the mlo2 mlo6 mlo12 mutant compared with wild-type plants point to enhanced glucosinolate turnover being one of the defects in the mlo triple mutant (Figs. 3B and 4A). A perturbed profile of indolic metabolites was also recently detected in systemic leaves of the aux1 mutant, Figure 5. Indolic secondary metabolites are essential for mlo2mediated powdery mildew resistance. A, Representative photographs depicting macroscopic infection phenotypes of wild-type and mutant plants upon challenge with G. orontii. Images were taken at 10 d postinoculation. B, Quantitative analysis of G. orontii host cell entry (determined at 48 h postinoculation; gray bars) and conidiophore formation (determined at 7 d postinoculation; black bars). Results represent means 6 SD of four to eight independent experiments per genotype. Statistically significant differences from the Col-0 wild type are indicated by asterisks (*** P , 0.01, * P , 0.05; Student's t test), and statistically significant differences from mlo2 are indicated by number signs ( ### P , 0.01, # P , 0.05; Student's t test). C, Quantitative analysis of E. pisi entry into Arabidopsis epidermal cells determined at 7 d postinoculation. Results represent means 6 SD of three to six samples per genotype derived from at least three independent experiments. Statistically significant differences from the Col-0 wild type are indicated by asterisks (*** P , 0.01, * P , 0.05; Student's t test), and statistically significant differences from mlo2 are indicated by number signs ( ### P , 0.01, # P , 0.05; Student's t test). Among plant secondary metabolites, few indole derivatives have been functionally associated with biotic stress responses. These include the indolic glucosinolates, a class of metabolites restricted to the Brassicales that has been implicated in combating insects (for review, see Halkier and Gershenzon, 2006). Additionally, in Arabidopsis, the phytoalexin camalexin was shown to contribute to resistance against necrotrophic fungi, such as Alternaria brassicicola and B. cinerea (Thomma et al., 1999b;Ferrari et al., 2003;for review, see Glazebrook, 2005). In contrast, the biotrophic fungus G. orontii was reported to be insensitive to camalexin (Reuber et al., 1998). Recently, however, glucosinolate breakdown products with presumptive antimicrobial activity were found to be required for penetration resistance against nonadapted powdery mildew fungi in Arabidopsis, suggesting that at least some indolic compounds contribute to defense against a subset of biotrophic pathogens (Bednarek et al., 2009). In addition to their presumptive direct antimicrobial activity, these metabolites may also have a signaling role in plant innate immunity (Clay et al., 2009). To investigate the contribution of Trp-derived molecules in mlo2-associated phenotypes in detail, we generated and analyzed an informative set of Trp metabolism mutants in combination with the mlo2 single mutant.

Indolic Metabolites in
We reasoned that altered accumulation patterns of indolic compounds that become apparent during vegetative development (Figs. 3 and 4) might also contribute to the powdery mildew resistance of the mlo2 mutant. Experimental support for this hypothesis comes from the partially and fully restored susceptibility (host cell entry and conidiation) to G. orontii in the mlo2 pad3 and mlo2 cyp79B2 cyp79B3 mutants, respectively (Fig. 5, A and B). Furthermore, mutations in PAD3 or CYP79B2/CYP79B3 resulted in moderately (pad3) or substantially (cyp79B2 cyp79B3) increased penetration rates of the nonadapted powdery mildew pathogen E. pisi, both in the presence and absence of MLO2 (Fig. 5C). These findings demonstrate that indolic metabolites, including camalexin, contribute to both preinvasive and postinvasive defense against adapted as well as nonadapted powdery mildew fungi in Arabidopsis. Notably, genetic analyses in the context of systemic acquired resistance indicated that CYP79B2/CYP79B3 are also required for the establishment of this type of plant immunity, suggesting that indolic metabolites (and possibly the indole-derived phytohormone auxin) contribute either to long-distance signaling or the execution of systemic immune responses (Truman et al., 2010).
This broad indole-mediated antifungal capacity is seemingly masked by the high pathogen success rate in compatible powdery mildew interactions, possibly as a consequence of the ability of adapted pathogens to detoxify these compounds or to suppress pathways that lead to their accumulation. This would explain why the cyp79B2 cyp79B3 mutant exhibits only a Figure 6. Camalexin is essential for mlo2-mediated resistance against B. cinerea. A, Representative photographs depicting macroscopic infection phenotypes on leaves of wild-type and mutant plants upon challenge with a camalexin-sensitive strain of B. cinerea. Detached leaves of 4-week-old plants grown in 16-h-light/8-h-dark cycles (low light intensity and humidity) were inoculated with two 5-mL droplets of a suspension containing 5 3 10 5 spores mL 21 . Images were taken at 2 d postinoculation. The experiment was repeated twice with similar results. B, Quantitative assessment of lesion diameter determined at 2 d postinoculation with B. cinerea. Results originate from one experiment and represent means 6 SD of at least nine lesions per genotype. Statistically significant differences from the Col-0 wild type are indicated by asterisks (*** P , 0.01, * P , 0.05; Student's t test), and statistically significant differences from mlo2 are indicated by number signs (P , 0.01; Student's t test). The experiment was repeated twice with similar results. subtle, if any, effect on colonization by the adapted powdery mildew pathogen G. orontii (Fig. 5, A and B), while it causes a dramatic increase in plant cell entry by the nonadapted pea powdery mildew pathogen (Fig. 5C). The indolic toxic compounds that, besides camalexin, contribute to defense against powdery mildews currently remain elusive. Metabolomic analyses in the context of powdery mildew-nonhost interactions suggest that a conversion product of 4-methoxylated indole glucosinolate could be the major toxic principle (Bednarek et al., 2009).
It is remarkable that resistance in the mlo2 mutant is already effective in young plants (5 weeks or younger), before considerable differences in the accumulation of indolic metabolites and the abundance of transcripts that encode for the respective biosynthetic enzymes become evident (Supplemental Fig. S7). One possibility is that substantial cell type-specific differences that may exist between young wild-type and mlo2 seedlings are diluted in whole leaf-based transcriptomic and metabolomic analyses. Alternatively, despite similar steady-state levels in unchallenged plants, metabolic pathways leading to indolic compounds might be more rapidly activated in mlo2 mutant plants upon challenge with powdery mildew fungi.
Camalexin Is Required for mlo2-Mediated Resistance to B. cinerea Among plant pathogens with a necrotrophic lifestyle, B. cinerea is one of the best studied. In Arabidopsis, the well-characterized JA-and ET-dependent Figure 7. The GSL5/PMR4 callose synthase is required for spontaneous callose deposition but dispensable for penetration resistance of the mlo2 mutant. A, Representative micrographs of spontaneous callose deposition in unchallenged (pathogen-free) plants grown in long-day conditions. Leaves were collected at the time points indicated. The experiment was repeated once and additionally performed once with plants grown in short-day conditions, yielding similar results. Bar = 50 mm. B, Quantitative analysis of G. orontii host cell entry (determined at 48 h postinoculation; gray bars) and conidiophore formation (determined at 7 d postinoculation; black bars). Results represent means 6 SD of three to eight independent experiments per genotype. Statistically significant differences from the Col-0 wild type are indicated by asterisks (*** P , 0.01, * P , 0.05; Student's t test), and statistically significant differences from mlo2 are indicated by number signs ( ### P , 0.01, # P , 0.05; Student's t test). C, Habitus of representative unchallenged (pathogen-free) plants at 7 weeks after sowing (top row) and macroscopic phenotypes of detached leaf 5 from 4-week-old plants darktreated for 4 d (bottom row). In both experiments, plants were grown in long-day conditions. The experiment was repeated twice with similar results.
Indolic Metabolites in Atmlo2-Mediated Antifungal Defense Plant Physiol. Vol. 152, 2010 defense signaling pathways have been shown to contribute to basal resistance against B. cinerea, since mutants in these pathways showed increased susceptibility to this pathogen (Thomma et al., 1998(Thomma et al., , 1999aFerrari et al., 2003). Additionally, the Arabidopsisspecific phytoalexin camalexin contributes to resistance against B. cinerea, as indicated by the increased susceptibility of camalexin-deficient mutants (Ferrari et al., 2003;Kliebenstein et al., 2005;van Baarlen et al., 2007). However, not all isolates of B. cinerea are camalexin sensitive (Thomma et al., 1999b;Ferrari et al., 2003); different strains vary in their camalexin tolerance, and this difference determines the ability of the fungal pathogen to proliferate and produce lesions on camalexin-deficient mutants (Kliebenstein et al., 2005). Inoculation of Arabidopsis mlo2 and mlo2 mlo6 mlo12 mutants with the necrotrophic fungus B. cinerea resulted in severely reduced disease symptoms compared with Col-0 wild-type plants (Fig. 6). This finding is of note since it is, to the best of our knowledge, the first report of enhanced resistance of a mlo mutant against a pathogen other than powdery mildews (Jørgensen, 1977;Consonni et al., 2006). Rather conversely, it was previously observed that barley and Arabidopsis mlo mutants tend to be more susceptible and/or to show more disease symptoms upon challenge with hemibiotrophic and necrotrophic pathogens, respectively (Jarosch et al., 1999;Kumar et al., 2001;Consonni et al., 2006). The Botrytis strain used in our work is camalexin sensitive (Ferrari et al., 2003), suggesting that the increased resistance observed in mlo plants might be caused by altered constitutive or pathogen-triggered accumulation patterns of the phytoalexin in these mutants. This hypothesis was supported by the reinstated susceptibility of mlo2 pad3 and mlo2 cyp79B2 cyp79B3 double and triple mutants (Fig.  6). Given this finding, it would be interesting to study the infection phenotypes of the Arabidopsis mlo mutants upon challenge with a broader range of phytopathogens, preferentially those that are known to be sensitive to camalexin.

mlo2-Associated Leaf Chlorosis/Necrosis Corresponds to a Premature Senescence Program
It has been previously speculated that the developmentally controlled leaf chlorosis and necrosis of barley and Arabidopsis mlo mutants corresponds to an authentic senescence process. A comparative timecourse analysis performed in the context of a distinct study revealed a premature decay of leaf pigments concomitant with a decline of glyceraldehyde-3-phosphate dehydrogenase and an increase in ubiquitin transcript accumulation in barley mlo mutant plants (Piffanelli et al., 2002). These results suggest that the early leaf cell-death phenotype of barley mlo mutants reflects genuine senescence. We expanded this type of analysis to the Arabidopsis mlo2 mlo6 mlo12 triple mutant and found, in addition to conservation of accelerated chlorophyll catabolism across plant clades, a trend toward an earlier decrease in photochemical efficiency in the mutant plants compared with wild-type plants (Fig. 1,  A and B). Moreover, we observed a genotype-specific resemblance of whole plant leaf chlorosis phenotypes upon induction of synchronized artificial (dark-induced) senescence in detached leaves (Fig. 1C). Taken together, these findings strongly suggest that the early leaf chlorosis in barley and Arabidopsis mlo mutant plants reflects the premature initiation and possibly accelerated progression of an authentic leaf senescence program. This senescence program might be based on perturbed autophagic processes, since autophagydeficient mutants atg2 and atg5 resemble the SAdependent early leaf chlorosis/necrosis phenotype of mlo mutants (Yoshimoto et al., 2009).
We further investigated the effect of premature leaf senescence in the mlo2 mlo6 mlo12 mutant at both the transcriptional and proteomic levels. Among the differentially expressed genes are WRKY53 and SAG13, two marker genes of early leaf senescence. WRKY53 is a transcriptional regulator with a well-established role in the control of leaf senescence; mutants in WRKY53 show delayed onset of senescence, while WRKY53 overexpression results in early senescence (Hinderhofer and Zentgraf, 2001;Miao et al., 2004). Genetic epistasis analysis suggests that hyperaccumulation of WRKY53 transcripts in the mlo2 mutant background is not responsible for mlo2-associated leaf senescence (Fig. 3B). Transcription of WRKY53 is activated by SA (Dong et al., 2003), and the function of WRKY53 in regulating leaf senescence is dependent on the JA/SA equilibrium (Miao and Zentgraf, 2007). Since 7-week-old mlo2 mlo6 mlo12 mutants were previously shown to possess high levels of SA , elevated transcript accumulation of WRKY53 in the mlo2 mlo6 mlo12 mu- Based on the results of metabolite profiling, we speculated that the developmentally controlled hyperaccumulation of indolic metabolites might drive mlo mutants into early leaf senescence. However, even a complete block in the biosynthesis of indolic metabolites (such as in the mlo2 cyp79B2 cyp79B3 triple mutant) did not reverse the early senescence phenotype of mlo2 mutants, suggesting that deregulated accumulation of these compounds is not the primary cause for the early senescence phenotype of mlo2 mutants. It remains a future challenge to further disentangle the molecular events that lead to this mlo-associated phenotype.
Callose Biosynthesis Is Not Required for mlo2 Resistance Cell wall appositions have long been thought to locally reinforce the cell wall during pathogen attack and were assumed to represent an essential first physical barrier to prevent fungal ingress (Schmelzer, 2002). In this context, callose was for a long time believed to provide a structural reinforcement of papillae (Smart et al., 1986). Nevertheless, recent experiments have questioned the role of callose in plant defense against diverse pathogen classes (Jacobs et al., 2003;Nishimura et al., 2003;Galletti et al., 2008).
Analysis of the mlo2 pmr4 double mutant revealed that PMR4 activity is required for spontaneous callose deposition in the mlo2 mutant (Fig. 7A). However, mlo2-mediated resistance against G. orontii was not affected in this double mutant, suggesting that local callose deposition at attempted fungal entry sites is not obligatory for mlo2-mediated penetration resistance against powdery mildews in Arabidopsis (Fig. 7B). The powdery mildew resistance retained in the mlo2 pmr4 double mutant is not because of a pmr4-conditioned hyperactivation of SA-dependent defense reactions, since an additional mutation in the SA biosynthesis gene, SID2, likewise resulted in unaffected mlo2-like resistance in the mlo2 pmr4 sid2 triple mutant (Fig. 7B). In sum, these data suggest that, contrary to common belief, callose deposition in papillae is not causative for antifungal penetration resistance. CONCLUSION We previously hypothesized that resistance conferred by innate immune pathways and mlo-mediated powdery mildew resistance share common defense execution machinery . The results obtained in this study substantiate this notion and further highlight the crucial contribution of indolic metabolites in both preinvasive and postinvasive antifungal defense in Arabidopsis. The fact that mutations in PEN2 restore powdery mildew susceptibility in the mlo2 mutant to a similar extent as a complete block of biosynthesis of indole-type secondary metab-olites (cyp79B2 cyp79B3 mutant) suggests that PEN2 catalyzes a rate-limiting step in this defense pathway that leads to the major toxic principle(s). The finding that transcripts of the very same genes that are crucial for antifungal defense accumulate during the establishment of early leaf senescence in the mlo2 mutant provides additional support for the molecular link and potential cross talk between these two processes (Schenk et al., 2005). We propose a model in which MLO proteins negatively regulate both pathways (Fig.  8). Further work is required to unravel how this hypothesized negative regulation is exerted and how the various processes modulated by MLO proteins are interconnected.

Phytopathogens
Powdery mildews used in this study were the isolates of G. orontii and E. pisi kept at the Max-Planck Institute for Plant Breeding Research (Lipka et al., 2005). The camalexin-sensitive B. cinerea strain was originally isolated from Brassica oleracea (Ferrari et al., 2003;J. Plotnikova, unpublished results). Inoculation of Arabidopsis with B. cinerea was performed as described previously (Ferrari et al., 2003).

Microarray Experiment and Data Analysis
Mature rosette leaves (six per genotype and time point) were harvested at 5, 6, and 7 weeks after sowing. Total RNA was extracted using the Trizol reagent and purified with the RNeasy Plant RNA Purification kit (Qiagen). Copy RNA (cRNA) was prepared following the manufacturer's instructions (www.affymetrix.com/support/technical/manual/expression_manual.affx). Labeled cRNA transcripts were purified using the sample cleanup module (Affymetrix). Fragmentation of cRNA transcripts, hybridization, and scanning of the high-density oligonucleotide microarrays (Arabidopsis ATH1 genome array; Affymetrix) were performed according to the manufacturer's GeneChip Expression Analysis Technical Manual. A single experiment with one microarray per time point and genotype was performed. Quality of the data was evaluated at probe level by examining the arrays for spatial effects, distribution of absent and present calls, and the intensity of spike-in controls. We used the robust multiarray average procedure (Irizarry et al., 2003) to correct for background effects and chip effects and to summarize the probe values into probe set values, resulting in 22,810 normalized expression values per array. The raw and normalized data have been deposited in the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) and are accessible through Gene Expression Omnibus record GSE17875. R/Bioconductor (Gentleman et al., 2004) was used to preprocess the raw microarray data.
To identify candidate genes with potentially altered transcript accumulation in the mlo2 mlo6 mlo12 triple mutant during development, we performed two different computations. In one procedure, we first calculated the ratio of the mean expression values at 6 and 7 weeks to the expression values at week 5 for both the wild type and the mutant. We then calculated the ratio of the respective values for the mutant and the wild type and generated a list ranked by descending ratios based on this result. The produced list (list A) contains all 22,810 genes and provides a full overview of the genes for which the average transcript levels in weeks 6 and 7 increased disproportionately in the triple mutant in comparison with week 5, irrespective of the actual absolute transcript levels. The reported ratios and expression values in list A are on an arithmetic scale, while computations were performed on a logarithmic scale. In the second procedure, we generated a hit list (list B) that represents a fraction of the original gene number (22,810 genes) by subsequently applying the following criteria. First, we selected genes that at 7 weeks showed an at least 2-fold higher transcript accumulation in the mutant than in the wild type (M7 . 23 C7; 326 genes in total). Within these 326 genes, we then selected those that exhibited at least 2-fold microarray values at 7 weeks compared with 5 weeks in the triple mutant (M7 . 23 M5; 181 genes). Next, we selected those for which the increase in transcript levels in the mutant from 5 to 7 weeks was at least 2-fold higher compared with the respective increase in the wild type [(M7/M5) . 23 (C7/C5); 135 genes]. Finally, we eliminated those genes for which all six microarray values were below 40, resulting in a final list with a total of 98 genes. This list (list B) represents genes with elevated transcript levels specific to the mutant at week 7 in comparison with week 5 while ensuring that absolute fold changes and transcript levels are high.
To find differentially regulated pathways, we first obtained predefined sets of genes with the same GO term (http://www.geneontolgy.org) as defined in the ath1121501 annotation library available from R/Bioconductor. We then used a Wilcoxon rank test, as implemented in the function "geneSetTest" of the Limma package (Smyth, 2005), to test whether genes with the same GO term are more highly ranked in list A compared with randomly selected genes. A Fisher's exact test, as implemented in GOstat (Beißbarth and Speed, 2004), was used to test for overrepresented GO terms in the 98 genes of list B. Correction for multiple testing was done in both cases by controlling the false discovery rate (Benjamini and Hochberg, 1995).

RNA Isolation, cDNA Synthesis, and Quantitative RT-PCR
Total RNA was isolated using the Trizol reagent according to the manufacturer's instructions (Invitrogen). The isolated RNA was further purified using RNeasy mini columns (Qiagen) according to the manufacturer's instructions. cDNA was synthesized using SuperScript RTII reverse transcriptase according to the manufacturer's instructions (Invitrogen).

Protein Extraction and Two-Dimensional PAGE
Total protein extracts from rosette leaves of 7-week-old plants grown in short-day conditions were prepared as described (Noir et al., 2009). Briefly, plant material was ground in liquid nitrogen, and 1 mL of extraction buffer (50 mM Tris, pH 8.0, 10 mM EDTA, 10 mM dithiothreitol, 0.5% CHAPS, and one protease inhibitor cocktail tablet [Roche]) was added. Proteins were extracted by repeated cycles of vortexing and chilling. Cell debris was removed by centrifugation, and the supernatant was subjected to TCA/acetone precipitation. Two-dimensional PAGE and protein staining were performed as described (Noir et al., 2009) except that 100 mg of protein was loaded per gel. SeeBlue Plus2 prestained standard was used for estimating M r values. Falsecolored protein pattern overlays were generated using Proteomeweaver twodimensional analysis software version 4.0 (Bio-Rad).

Extraction of Secondary Metabolites and HPLC Analysis
Rosette leaves (leaf 7) of Col-0 and mlo2 mlo6 mlo12 triple mutant plants were collected and frozen in liquid nitrogen. After addition of dimethyl sulfoxide (50 mL 20 mg 21 fresh weight), the tissue was homogenized using zirconia beads (1 mm; Roth) in a Mini-Beadbeater-8 (Biospec Products) and centrifuged for 15 min at 20,000g. The supernatants were collected and subjected to HPLC on an Agilent 1100 HPLC system equipped with diode array detector and fluorescence detector. Samples were analyzed on an Atlantis T3 C18 column (150 mm 3 2.1 mm, 3 mm; Waters) using 0.1% trifluoroacetic acid as solvent A and 98% acetonitrile/0.1% trifluoroacetic acid as solvent B at a flow rate of 0.25 mL min 21 at 22°C (gradient of solvent A: 100% at 0 min, 100% at 2 min, 90% at 9 min, 72% at 30 min, 50% at 33 min, 20% at 40 min, and 100% at 41 min). Camalexin was analyzed on a Zorbax Extend-C18 column (100 3 2.1 mm, 3.5 mm; Agilent) using water as solvent A and 98% acetonitrile as solvent B at a flow rate of 0.3 mL min 21 at 22°C (gradient of solvent A: 96% at 0 min, 96% at 3 min, 70% at 20 min, 20% at 33 min, and 0% at 34 min). The concentrations of the metabolites of interest were quantified based on the comparison of their peak areas with those obtained during HPLC analyses of known amounts of the respective compounds purified from plant tissue (I3G) or synthetic (I3A, camalexin) standards.

Analysis of Chlorophyll Content
Approximately 20 wild-type and mutant plants were grown on soil in long-day conditions. Three independent samples of leaf 7 were taken for each genotype and time point, starting from 28 d after sowing, with intervals ranging from 2 to 6 d. Chlorophyll was isolated from leaf tissue by homogenization in liquid nitrogen and subsequent extraction into 80% (v/v) acetone saturated with KOH. After centrifugation (10 min, 13,000g), chlorophyll concentrations were determined spectrophotometrically at 664, 647, and 750 nm (Strain et al., 1971;Pruzinská et al., 2005).

F v /F m Measurements
F v /F m was measured in vivo using a pulse amplitude modulation 101/103 fluorometer (Walz), starting from 28 d after sowing, at intervals of 2 to 6 d. After 30 min of incubation in the dark, saturating pulses of white light (0.8 s, 6,000 mmol photons m 22 s 21 ) were applied and F v /F m was calculated (Varotto et al., 2000;Pesaresi et al., 2002;Oh et al., 2003;Ihnatowicz et al., 2004). Leaf 7 of four independent plants grown in long-day conditions was used per genotype and time point.

Dark-Induced Senescence
Leaf 5 from 4-week-old Arabidopsis plants grown in long-day conditions were excised and placed on moisturized filter paper in petri dishes with the adaxial side facing up. The plates were kept in darkness at 22°C and controlled conditions for 4 d (Pruzinská et al., 2005;Guo and Gan, 2006).

Callose Deposition
For visualization of callose, samples were stained with aniline blue as described (Adam and Somerville, 1996). Data were obtained from two independent experiments with five plants each per genotype and experiment.

Quantification of Fungal Infection Success
For visualization of epiphytic fungal structures, specimens were briefly stained in 0.6% Coomassie Brilliant Blue (in ethanol) and then rinsed with water. Due to the small size of respective haustoria, differentiation of elongating secondary hyphae served as an approximation of penetration success in the powdery mildew species G. orontii and E. pisi. Data were obtained from a minimum of three independent experiments with five plants each per genotype and experiment.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Scheme of the Trp and indole biosynthetic pathway.
Supplemental Figure S2. Independent biological replicate of the analysis of photosynthetic performance and chlorophyll content shown in Figure 1A.
Supplemental Figure S3. Unsupervised hierarchical cluster of the microarray data.
Supplemental Figure S4. Independent biological replicate of the RT-PCR analysis shown in Figure 3A.
Supplemental Figure S5. Two-dimensional gel electrophoresis reveals no differences between the proteomes of wild-type and mlo2 mlo6 mlo12 mutant plants.
Supplemental Figure S6. Accumulation of an unknown camalexin derivative in wild-type and mutant plants.
Supplemental Figure S7. Timing of events and phenotypes in the mlo2 or mlo2 mlo6 mlo12 mutant.
Supplemental Table S1. Genes identified as differentially regulated in the mlo2 mlo6 mlo12 mutant throughout development.