Non-host resistance of Arabidopsis thaliana against Phytophthora infestans, the causal agent of late blight disease of potato, depends on efficient extracellular pre- and post-invasive resistance responses. Pre-invasive resistance against P. infestans requires the myrosinase PEN2. To identify additional genes involved in non-host resistance to P. infestans, a genetic screen was performed by re-mutagenesis of pen2 plants. Fourteen independent mutants were isolated that displayed an enhanced response to Phytophthora (erp) phenotype. Upon inoculation with P. infestans, two mutants, pen2-1 erp1-3 and pen2-1 erp1-4, showed an enhanced rate of mesophyll cell death and produced excessive callose deposits in the mesophyll cell layer. ERP1 encodes a phospholipid:sterol acyltransferase (PSAT1) that catalyzes the formation of sterol esters. Consistent with this, the tested T-DNA insertion lines of PSAT1 are phenocopies of erp1 plants. Sterol ester levels are highly reduced in all erp1/psat1 mutants, whereas sterol glycoside levels are increased twofold. Excessive callose deposition occurred independently of PMR4/GSL5 activity, a known pathogen-inducible callose synthase. A similar formation of aberrant callose deposits was triggered by the inoculation of erp1psat1 plants with powdery mildew. These results suggest a role for sterol conjugates in cell non-autonomous defense responses against invasive filamentous pathogens.
The oomycete Phytophthora infestans, the causal agent of late blight disease of potato, fails to colonize Arabidopsis thaliana. Elucidating the basis of this non-host pathogen interaction is of interest for understanding mechanisms underlying resistance against P. infestans. Arabidopsis mounts efficient extracellular pre- and post-invasive resistance responses to prevent colonization by this oomycete (Lipka et al., 2005). Components of non-host resistance in Arabidopsis were identified by genetic approaches. The isolation of penetration (pen) mutants, which permit enhanced entry of non-adapted Hordeum vulgare (barley) and Pisum sativum (pea) powdery mildews [Blumeria graminis f. sp. hordei (Bgh) and Erysiphe pisi, respectively] revealed the existence of inducible extracellular defense responses acting early during fungal pathogenesis (Collins et al., 2003; Lipka et al., 2005; Stein et al., 2006). PEN2 is a peroxisome-associated myrosinase that acts in the same pathway as the plasma membrane-resident ATP-binding cassette (ABC) transporter PEN3 (Lipka et al., 2005; Stein et al., 2006). PEN2 metabolizes a particular indole glucosinolate, 4-methoxyindol-3-ylmethylglucosinolate, and releases tryptophan-derived antimicrobial metabolites that are assumed to be exported via PEN3 at pathogen contact sites (Lipka et al., 2008; Bednarek et al., 2009). In addition, PEN2-derived metabolites might also act as defense signaling molecules (Clay et al., 2009). Because of the multilayered nature of non-host resistance, the loss of extracellular pre-invasion resistance alone does not result in the unrestricted growth of tested non-adapted pathogens. EDS1, PAD4 and/or SAG101 were shown to be required for post-invasive defenses against Bgh and E. pisi in Arabidopsis, but not against P. infestans (Lipka et al., 2005; Stein et al., 2006; Westphal et al., 2008).
Induced defense responses of non-host resistance are believed to rely on the recognition of pathogen-associated molecular patterns (PAMPs) by membrane-resident pattern recognition receptors, which activate PAMP-triggered immunity (Schwessinger and Zipfel, 2008). The evolution of strain-specific pathogen effectors, of which at least some suppress PAMP-triggered immunity, is thought to be critical for effective pathogenesis on plants. Plants can recognize the presence or the activity of some of these effectors, encoded by avirulence (AVR) genes, by intracellular immune receptors of the NLR type, thereby triggering effector-triggered immunity (Schwessinger and Zipfel, 2008). Studies of non-host resistance in closely related plant species suggest that some AVR genes can function both in effector-triggered immunity and non-host resistance (Schulze-Lefert and Panstruga, 2011). Thus, non-host resistance appears to involve the combined action of pattern recognition receptors and NLR-type receptors.
Sterols have been implicated in pathogen defense; however, little is known about their actual role in disease resistance. Stigmasterol was identified as a susceptibility factor in Arabidopsis, as inhibition of its synthesis results in enhanced resistance to Pseudomonas syringae (Griebel and Zeier, 2010). Lateral membrane compartments, which are proposed to represent signaling platforms for a specific set of plasma membrane proteins (Morel et al., 2006; Lefebvre et al., 2007), are enriched in sterols and have been implicated in pathogen perception and signal transduction (Mongrand et al., 2010). Notably, plants are able to recognize fungal sterols that act as PAMPs by eliciting defense responses (Granado et al., 1995). Interestingly, Phytophthora spp. are unable to synthesize sterols and depend on the acquisition of these compounds from their hosts (Hendrix, 1970). Phytophthora secretes highly conserved 10 kDa proteins with homology to non-specific lipid transfer proteins, called elicitins. Upon binding of sterols to elicitins, these stimulate plant defense responses (Osman et al., 2001). Infestin 1 (INF1), the elicitin of P. infestans, interacts with a lectin-like receptor kinase protein of Nicotiana tabacum (tobacco), which is required for the elicitation of a host cell death-associated resistance response (Kanzaki et al., 2008).
We describe in this work a genetic screen for alterations in non-host responses of Arabidopsis to P. infestans. This resulted in the identification of the enhanced response to Phytophthora 1 (erp1) mutant that is defective in the gene encoding phospholipid:sterol acyltransferase 1 (PSAT1). We show that a lack of sterol ester formation correlates with the erp1 phenotype. This suggests a role for sterol conjugates in modulating the defense responses of Arabidopsis.
Isolation of erp mutants
To identify genes involved in non-host resistance against P. infestans, we re-mutagenized seeds of pen2 plants that are impaired in pre-invasion resistance to the oomycete pathogen (Lipka et al., 2005). Approximately 70 000 M2 plants were screened for altered infection phenotypes to P. infestans. Unlike glabrous1 (gl1) plants, pen2 plants mount a localized host cell death response to P. infestans (Westphal et al., 2008). Fourteen independent mutants with an enhanced response to challenge from P. infestans were identified, and named enhanced response to Phytophthora (erp). Allelism tests revealed that two of the mutants were affected in the same gene, and these mutant alleles were designated erp1-3 and erp1-4. Macroscopic and microscopic inspection of inoculated leaves (Figure 1a) revealed that pen2-1 erp1-3 and pen2-1 erp1-4 plants show a stronger cell death response than pen2-1 plants. In addition, pen2-1 erp1-3 and pen2-1 erp1-4 plants displayed trypan blue-stained globular structures in the mesophyll cell layer, which were not visible in pen2-1 leaves (Figure 1a). These structures, as well as the enhanced cell death response in mesophyll cells compared with gl1, were also observed upon inoculation of backcrossed homozygous erp1 single mutants (Figure 1a), demonstrating that these phenotypes do not depend on the absence of PEN2. The determination of pathogen biomass in the leaves of P. infestans-infected mutants revealed that there was no significant alteration in pathogen growth of the pen2-1 erp1-3 mutant compared with pen2-1 (Figure 1d).
A mapping population was generated by crossing pen2-1 erp1-3 with Landsberg erecta (Ler). ERP1 was mapped to the upper arm of chromosome 1. Whole-genome sequencing of the pen2-1 erp1-3 mutant resulted in the identification of two single nucleotide polymorphisms (SNPs) in the region of the erp1-3 mutation: one located in an intron of At1g03960 and the other causing a premature stop codon in At1g04010. The latter gene consists of 15 exons and encodes PSAT1, a protein of 633 amino acids with phospholipid:sterol acyltransferase (PSAT) activity (Banas et al., 2005). The mutation in erp1-3 changed a tryptophan to a stop codon, and sequencing this gene in erp1-4 revealed a similar mutation (Figure 1b). The mutations are predicted to truncate the protein after 590 and 156 amino acids, respectively. Two T-DNA insertion lines of At1g04010, SALK_037289 carrying the T-DNA insertion in the first intron (erp1-1) and SALK_117091 with an insertion in the sixth exon (erp1-2; Figure 1b) were analyzed for their response to P. infestans. In contrast to Col-0 plants, homozygous mutant segregants of the insertion lines displayed a host cell death phenotype upon inoculation with the oomycete (Figure 1c). Importantly, globular structures were observed in mesophyll cells of the T-DNA insertion mutants upon inoculation with P. infestans (Figure 1c). Thus, the disruption of At1g04010 caused the same enhanced response phenotype as observed in the erp1 single mutants. Moreover, there was no complementation of the erp1 infection phenotype in F1 plants of a cross between erp1-3 and the T-DNA insertion line erp1-2, suggesting that the same gene is impaired in both mutants (Figure S1).
Further evidence that the phenotype of erp1 is caused by the mutation in At1g04010 was obtained by complementation assays. The expression of an ERP1-GFP fusion protein in erp1-3 plants resulted in the restoration of the wild-type response (Figure 4a), suggesting that the loss of At1g04010 function is responsible for the erp1 phenotype.
Characterization of the erp1 phenotype
During the first weeks of plant development, erp1 mutants did not show obvious morphological alterations compared with wild-type plants. However, as reported before (Bouvier-Navé et al., 2010), leaf senescence occurred earlier in erp1 plants (Figure S2a), and an extensive spontaneous chlorosis and necrosis was observed after 8 weeks (Figure S2b).
In microscopic analyses, the globular structures were detected at the cell walls of mesophyll cells, both under bright field and after staining with toluidine blue (Figure 2a,b). Electron microscopy and immunogold labeling using a specific β-1,3-glucan antibody revealed that these structures are cell wall-associated, and thus represent extracellular deposits containing the β-1,3-glucan callose (Figures 2c, S3). Notably, no oomycete structures were observed in the vicinity of these callose-containing structures in the mesophyll cell layer.
To test whether the wound- and pathogen-induced glucan synthase-like protein GSL5 is responsible for the enhanced callose formation, erp1-3 was crossed with pmr4, a gsl5 mutant (Jacobs et al., 2003; Nishimura et al., 2003). To verify the loss of GSL5 activity, leaves were wounded by cutting across the leaf blade with a scalpel, and stained with the dye aniline blue that fluoresces in the presence of callose.
In contrast to erp1-3 and pen2-1 erp1-3 plants, pmr4-1, pmr4-1 erp1-3 and pmr4-1 pen2-1 erp1-3 plants were unable to form callose in response to wounding (Figure 2d). After inoculation with P. infestans, however, microscopic analyses revealed aniline blue fluorescence in mesophyll cells not only of erp1-3 and pen2-1 erp1-3, but also of pmr4-1 erp1-3 and pmr4-1 pen2-1 erp1-3 genotypes (Figure 2f,g). As the pmr4-1 erp1-3 and pmr4-1 pen2-1 erp1-3 mutants exhibited reduced growth (Figure 2h), which correlated with a reduced leaf cell size (Figure S4), callose depositions in pmr4-1 erp1-3 and pmr4-1 pen2-1 erp1-3 mutants appeared smaller in size. However, the formation of callose-containing papilla-like structures in pmr4-1 erp1-3 and pmr4-1 pen2-1 erp1-3 plants (Figure 2e,g) suggests that GSL5 is not essential for the synthesis of P. infestans-induced callose deposits in erp1 plants.
Decreased sterol ester levels in erp1 mutants
PSAT1 catalyzes the transfer of fatty acids from the preferred acyl donor phosphatidylethanolamine to sterols, leading to the formation of sterol esters (Banas et al., 2005). To determine whether the mutations in erp1 led to a loss of PSAT acitivity, the levels of free sterols, sterol glycosides as well as sterol esters and acylated sterol glycosides were determined by quadrupole time-of-flight mass spectrometry (Q-TOF-MS; Wewer et al., 2011). In contrast to gl1, Col-0 and pen2-1, all erp1 mutants had significantly reduced levels of sterol esters and acylated sterol glycosides (Figure 3a). In particular, sterol ester levels in erp1-3, pen2-1 erp1-3, pen2-1 erp1-4 and erp1-1 were at or below the detection limit. erp1-2, which contains a T-DNA insertion in the sixth exon, still contained detectable levels of sterol esters, which is in accordance with previously published data suggesting that this line does not represent a complete knock out (Bouvier-Navé et al., 2010). Whereas free sterol levels in the erp mutants were not consistently altered, sterol glycoside levels were twofold higher than in the wild type, based on increased levels of sitosterol glycoside (Figure 3a, S5). The ratio of stigmasterol to sitosterol did not differ significantly in untreated control and erp1 plants, and increased in all plant lines upon challenge with P. infestans (Figure S6). Moreover, sterol ester levels increased about twofold in P. infestans-inoculated wild-type and pen2 plants, which seems predominantly to result from an increased incorporation of 18:2 and 18:3 fatty acids (Figure S7). As this was not observed in erp1 plants, PSAT1 appears to be the main enzyme responsible for sterol ester formation in response to infection with P. infestans in Arabidopsis. The P. infestans-induced increase in unsaturated C18 fatty acids in sterol esters is in accordance with the reported acyl specificity of PSAT1 (Banas et al., 2005).
The loss of sterol esters in erp1 plants is in accordance with the reported PSAT activity of the protein encoded by At1g04010 (Banas et al., 2005; Bouvier-Navé et al., 2010). However, in contrast to the twofold reduction in sterol ester levels in leaves reported by Bouvier-Navé et al. (2010), sterol ester levels in three of the four erp1 mutants analyzed here were at the detection limit. Thus, under our growth conditions, PSAT1 provides crucial sterol ester synthesis activity in leaves.
Sterol esters are presumed to be stored in lipid droplets (Gondet et al., 1994), which can be stained by Nile Red, a dye suitable for staining neutral lipids. Lipid droplets were detectable in gl1 and pen2-1 leaves, but not in erp1-3 or pen2-1 erp1-3 genotypes (Figure 3b). Thus, the absence of Nile Red-stainable lipid droplets correlates with the highly reduced levels of sterol esters, and represents an additional phenotype of erp1 mutant plants.
Subcellular localization of the ERP1 protein
To determine the subcellular localization of ERP1, gl1 and erp1-3 plants were transformed with the coding region of ERP1 fused to the coding region of GFP under the control of the 35S promoter. Expression of the fusion protein was detected by an anti-GFP antibody and, in the erp1-3 background, resulted in full suppression of the erp1 phenotype, indicated by reduced cell death rates in independent complemented transgenic lines compared with erp1-3 or a transgenic line that showed no expression of the fusion protein (Figure 4a). Moreover, ERP1-GFP expressing lines showed Nile Red-stainable lipid droplets (Figure 4c). GFP fluorescence in erp1-3 plants transformed with the Pro35S:ERP1-GFP construct was localized to spherical structures, which were evenly distributed throughout leaf epidermal and mesophyll cells, and moved within the cells (Figure 4b). These structures did not represent lipid droplets, as no overlap of GFP fluorescence and Nile Red staining was observed (Figure 4c). A similar pattern of GFP fluorescence was observed in root cells and root hairs (Figure S8), and after transient expression of Pro35S:ERP1-GFP in Nicotiana benthamiana (Figure 4d). Co-expression of markers for spherical organelles such as peroxisomes (Px-RK-mCherry) and the Golgi apparatus (G-RK-mCherry) showed no overlap of GFP and mCherry fluorescence, indicating that ERP1 is not present in these compartments.
Response of erp1 to other pathogens
In response to inoculation with the non-adapted barley pathogen Bgh, fungal entry or cell death 48 h post inoculation (hpi) in epidermal cells of erp1 plants were indistinguishable from wild-type plants (Figure 5a). However, aberrant cell death of mesophyll cell clusters was observed in the erp1 mutants within 9 days of inoculation with Bgh, leading to a macroscopically detectable necrosis phenotype that was even more pronounced in the pen2-1 erp1-3 mutant (Figure 5b). Moreover, in Bgh-inoculated erp1 plants, trypan blue- and aniline blue-stainable structures in the mesophyll cell layer, similar to the aberrant callose depositions observed after P. infestans inoculation, were detected 6 days post inoculation (dpi; Figure 5c). Similar structures were observed in erp1-3 plants infected with the adapted pathogen Golovinomyces cichoracearum (Figure S9). These results suggest that the loss of sterol esters does not only affect responses to the non-adapted pathogens P. infestans and Bgh, but also affects responses to host-adapted pathogens such as G. cichoracearum.
In contrast, no consistent differences in pathogen growth were observed on erp1 mutant plants and control lines infected by virulent Pseudomonas syringae pv. tomato (Pst) DC3000, avirulent Pst DC3000 avrRPM1 or the non-adapted Pseudomonas syringae pv. phaseolicola (Pph; Figure 6a). To test if the effector-triggered cell death reaction induced by Pst DC3000 avrRPM1 is accompanied by a deregulated callose deposition in the mesophyll, infiltrated leaves were stained with aniline blue. Callose depositions were observed in the epidermis of Col-0, gl1, erp1-2 and erp1-3 plants, which were undetectable in pmr4-1 and pmr4-1 erp1-3 plants (Figure 6b,c). In the mesophyll cell layer, callose depositions around the mesophyll cells were dependent on GSL5 (Figure 6d). In addition, GSL5-independent, small callose-containing structures were detected in erp1-2 and erp1-3 plants, which, however, did not resemble the large aberrant callose depositions observed after fungal infection.
Alterations in sterol homeostasis in erp1 plants
The identification of PSAT1 as the gene affected in the erp1 mutant suggests a role of sterol esters for pathogen-inducible responses of Arabidopsis. The enzyme encoded by ERP1, PSAT1, catalyzes the formation of sterol esters by transferring acyl groups from the sn2 position of phosphatidylethanolamine to sterols, as determined with microsomal fractions from PSAT1-overexpressing Arabidopsis plants (Banas et al., 2005). Although a function of the second product of the PSAT reaction, lysophosphatidylethanolamine, cannot be ruled out, it is generally assumed that the loss of sterol ester formation is responsible for phenotypes in mutants with impaired PSAT activity (Banas et al., 2005; Bouvier-Navé et al., 2010). Postulated functions for sterol esters so far include their role as a transport or storage form of sterols (Dyas and Goad, 1993). As such, they might be involved in sterol sequestration if more sterols are present in the cell than required. This has been shown in a tobacco mutant with a 10-fold higher sterol content, where excess sterols accumulated as sterol esters and were compartmentalized in cytoplasmic lipid bodies (Maillot-Vernier et al., 1991; Gondet et al., 1994), as well as in tobacco plants overproducing sterols as a result of the expression of a cDNA encoding the sterol biosynthetic enzyme 3-hydroxy-3-methylglutaryl CoA reductase (Schaller et al., 1995). In accordance with these observations, the lack of sterol esters in erp1 mutant plants correlates with the absence of Nile Red-stainable lipid bodies (Figure 3b).
The determination of the levels of free sterols, sterol glycosides and acylated sterol glycosides in erp1 mutants revealed a role of PSAT1 in sterol homeostasis. The inability of erp1 plants to synthesize sterol esters correlates with reduced levels of acylated sterol glycosides and enhanced levels of sterol glycosides (Figure 3a). In Arabidopsis, sterol glycosides are synthesized by the UDP-glucose:sterol glucosyltransferases (SGTs) UGT80A2 and UGT80B1, and are degraded by sterol glycoside hydrolases (Grille et al., 2010). In a double knock-out mutant of the two SGT genes (UGT80A2 and B1) , highly reduced levels of sterol glycosides and acylated sterol glycosides correlate with increased free sterol and sterol ester levels in inflorescences and siliques, suggesting a regulatory relationship between the levels of free sterols and conjugated sterols (DeBolt et al., 2009). Genes encoding enzymes catalyzing the synthesis of acylated sterol glycosides, on the other hand, are not known (Grille et al., 2010). Both PSAT1 and the elusive sterol glycoside acyltransferase require phosphoglycerolipids as acyl donors (Banas et al., 2005; Grille et al., 2010). The concomitant decrease of sterol esters and acylated sterol glycosides might suggest a direct involvement of PSAT1 in acylated sterol glycoside synthesis; however, PSAT1 acylates the 3β-OH group of sterols, whereas acylated sterol glycosides are acylated at the C6 of glucose.
Several Arabidopsis sterol biosynthesis mutants exhibit strong defects in embryonic patterning or growth, which demonstrates the importance of a well balanced phytosterol content (reviewed in Schaller, 2003). Despite the strong reduction of sterol ester levels in erp1 mutant plants and concomitant alterations in the quantity of other sterol lipids, the molecular species composition of free sterols, sterol glycosides and acylated sterol glycosides is comparable with those determined in leaves of wild-type plants (Figures S5 and S7). Thus, a lack of PSAT1 activity does not affect sterol biosynthesis in general. In accordance with this, erp1 mutant plants show no developmental defects when grown under optimal conditions, except for the early leaf senescence. These findings, together with the observation of reduced growth in growth competition experiments of the are1 are2 yeast mutant (Zweytick et al., 2000), which is impaired in sterol ester synthesis, indicate the functional importance of sterol ester formation during developmental aging and stress adaptation.
Role of sterol esters in the pathogen response
Our genetic screen aimed at the identification of genes involved in non-host resistance to P. infestans. In the pathogen-challenged erp1 mutant, the lack of sterol ester accumulation correlates with an enhanced number of mesophyll cells undergoing cell death, and with the formation of massive callose depositions in the mesophyll cell layer. This aberrant response was observed after infection by the non-adapted pathogens P. infestans and Bgh, as well as by the host-adapted pathogen G. cichoracearum. The enhanced response does not correlate with obvious alterations in the growth of the pathogen. Thus, we did not observe significantly enhanced growth of P. infestans in the erp1 mutants compared with their respective wild types (Figure 1), nor did we detect enhanced oomycete colonization of the mesophyll cell layer. Similarly, there were no differences in cell death and the number of entry sites in response to Bgh inoculation (Figure 5). These observations show that the deregulated callose depositions are not a consequence of enhanced growth and entry attempts of pathogens in the mesophyll cell layer. It should rather be postulated that a diffusible signal induces the aberrant callose depositions in the mesophyll cell layer. This diffusible signal would originate from the primary infection attempts of the filamentous pathogens in the epidermal cell layer. The nature of this signal is not known. It might be speculated that reactive oxygen species generated at the infection sites contribute to this cell non-autonomous signal; however, these also accumulate in response to bacterial infections (Torres et al., 2006), and do not induce the formation of aberrant callose depositions (Figure 6). As the massive callose depositions in the mesophyll cell layer are formed in response to inoculation with invasive filamentous pathogens, but not after bacterial infection, it is possible that damage-induced molecular patterns, generated by the invading pathogen, might act as signaling molecules.
The transient silencing of a sterol biosynthesis gene in tobacco resulted in enhanced membrane permeability, as deduced from increased staining of nuclei with a membrane-impermeable dye (Wang et al., 2012). Assuming that the altered sterol homeostasis in erp1 plants also leads to alterations in the sterol content of plasma membranes, it is thus possible that erp1 epidermal cells release larger quantities of the proposed diffusible signal(s).
Alternatively, erp1 mesophyll cells might have an enhanced responsiveness to the putative diffusible signal(s). Phytosterols constitute structural components of plasma membranes, which modulate acyl chain ordering and water permeability (Schuler et al., 1991). In particular, membrane microdomains, which are defined as heterogenous and highly dynamic areas of the plasma membrane, are enriched in sterols and sphingolipids (Mongrand et al., 2004). Free sterols and sterol glycosides have been reported to occur in detergent-resistant membranes (DRMs) isolated from Arabidopsis seedlings (Laloi et al., 2007). Thus, lateral membrane compartments are supposed to provide platforms for signaling events (Mongrand et al., 2010). A role of sterols for plant pathogen responses might be concluded from the observation that Filipin staining of barley leaf epidermal cells infected with Bgh shows an enrichment of sterols at sites of pathogen attack (Bhat et al., 2005). In Arabidopis, enhanced levels of stigmasterol in pathogen-infected leaves (Griebel and Zeier, 2010; Wang et al., 2012; Figure S6) lead to alterations in membrane composition (Griebel and Zeier, 2010). As stigmasterol stimulates H+-ATPase activity in plasma membranes in vitro (Grandmougin-Ferjani et al., 1997), increased stigmasterol levels might have a direct effect on membrane function. We thus hypothesize that the altered sterol homeostasis in erp1 mutant plants might have an impact on cellular signaling events in response to pathogen infection.
The massive accumulation of callose in response to infection with P. infestans might be the result of de novo synthesis or the inhibition of callose degradation, a mechanism that has been suggested to regulate callose accumulation at plasmodesmata (Zavaliev et al., 2011). The callose synthase-like protein GSL5, responsible for wound- and pathogen-induced callose synthesis (Jacobs et al., 2003; Nishimura et al., 2003), is not required for erp1-associated callose deposition (Figure 2). Interestingly, the non-adapted barley pathogen Bgh can penetrate pmr4 mutants slightly better than the wild type, suggesting that GSL5 has a function in entry resistance in the non-host resistance of Arabidopsis against Bgh (Jacobs et al., 2003). GSL5 has also been shown to be required for callose deposition in response to inoculation with the non-adapted pathogen Colletotrichum truncatum (Shimada et al., 2006). A second callose synthase gene, GSL6, is activated in response to treatment with salicylic acid (SA) , as well as after infection with Hyaloperonospora parasitica in an NPR1-dependent manner (Dong et al., 2008). Whether GSL6 plays a role in callose synthesis in erp1 plants remains to be determined. In GSL6-RNAi plants, callose formation in response to pathogen infection is not impaired (Jacobs et al., 2003). Microarray studies do not show major changes in GSL5 or GSL6 gene expression in response to P. infestans (http://www.genevestigator.com). Non-host resistance against Bgh does not involve changes in GSL6, and only minor changes in GSL5 expression (http://www.genevestigator.com).
In conclusion, the identification of PSAT1 as a gene involved in the response of Arabidopsis to infection by the non-adapted pathogens P. infestans and Bgh, as well as the adapted powdery mildew pathogen, G. cichoracearum, suggests a previously unreported role of sterol conjugates for pathogen defense responses in Arabidopsis.
Plant lines and growth conditions
The Arabidopsis accessions and mutants used in this study are summarized in Table S1. Table S2 lists the primers and enzymes used for genotyping. Plants to be infected with P. infestans (Si-Ammour et al., 2003) were grown in a phytochamber at 22°C and 60% humidity under short-day conditions with 8 h of light (~160 μMol m−2 sec−1). For infection experiments with Bgh (isolate K1), plants were grown in growth chambers with 8 h of light (~150 μMol m−2 sec−1) at 22°C and 16 h of dark at 18°C and 65% humidity.
Arabidopsis mutant screen
pen2-1 seeds were treated with ethyl methanesulfonate (EMS; 0.3% v/v) and M2 seeds were harvested as pools. The localized cell death response phenotype of about 70 000 M2 plants was scored visually 3 days after inoculation with P. infestans (5 × 105 zoospores ml−1; 10-μl droplets). The phenotype of putative mutants was verified by testing the corresponding M3 lines.
Mapping and cloning of erp1
A mapping population was generated by crossing erp1 with Ler-0. Positional cloning was carried out according to the methods described in Lukowitz et al. (2000). After inoculation with P. infestans, leaves of plants showing a mutant phenotype were harvested for DNA preparation (Weigel and Glazebrook, 2009) and polymerase chain reaction (PCR)-based marker techniques. F2 lines were re-tested for their P. infestans response phenotype in the F3 generation. Markers within the mapping interval were generated using the multiple SNP query tool (MSQT on http://www.weigelworld.org; Warthmann et al., 2007) and the primer design software Primer3 (Rozen and Skaletsky, 2000; http://frodo.wi.mit.edu/primer3).
Whole genome sequencing of erp1 was performed on an Illumina Genome Analyzer 1G and yielded 63.5 million 43-bp, single-end, reads. Filtering and short read alignments against the reference sequence were performed with the short-read analysis pipeline SHORE (Ossowski et al., 2008), resulting in an average genome coverage of 14 × . Base calling was performed for all genomic positions with a minimum requirement of three reads. The SNP within the coding sequence of At1g04010 led to the loss of a BseNI restriction site and was used to design a CAPS marker (Table S2), and analysis of F2 plants confirmed that it was linked to the erp1 phenotype. The erp1 single mutants erp1-3 and erp1-4 were isolated from back-crosses with gl1. The F2 progeny was inoculated with P. infestans and plants showing a localized cell death response phenotype were genotyped for the pen2-1 and erp1-3 or erp1-4 mutation. F2 lines with PEN2erp1 genotype and a localized cell death-response phenotype were selected for further analysis. Sequencing of amplified cDNA confirmed the specific SNPs, leading to a G → A transition, which results in a stop codon, in both erp1-3 and erp1-4. Primers used for cDNA sequencing were 5′-TCATTCACTCTCTTGGTGCAA-3′ and 5′-GCAGATGTGTTCCTCAATCTACA-3′ for erp1-3, as well as 5′-TGTCCATACACTCCGTTGGA-3′ and 5′-GCTCCAACAGCGAAATAAGC-3′ for erp1-4. Two T-DNA insertion lines for At1G04010, SALK_037289 and SALK_117091 were obtained from the Nottingham Arabidopsis Stock Centre (NASC, http://arabidopsis.info). These lines, annotated as psat1-1 (SALK_037289) and psat1-2 (SALK_117091) by Bouvier-Navé et al. (2010) were renamed erp1-1 and erp1-2, respectively, according to the TAIR nomenclature guidelines (http://www.arabidopsis.org). Detection of the T-DNA insertion was performed by using the ERP1 gene-specific oligonucleotide sequences 5′-ATGGGAGCGAATTCGAAATCAGTAAC-3′ (erp1-1) and 5′-TTTGTTGGCAAGGCTTCAG-3′ (erp1-2) in combination with the T-DNA left border-specific primer Lba1 5′-TGGTTCACGTAGTGGGCCATCG-3′.
Five-week-old Arabidopsis plants were infected with the GFP-expressing P. infestans isolate CRA208m2 (Si-Ammour et al., 2003). P. infestans zoospores were obtained as described by Lipka et al. (2005). For the inoculation of Arabidopsis plants, 10-μl droplets of zoospores with concentrations of 5 × 104 up to 5 × 105 zoospores ml−1 were applied to rosette leaves. Inoculated plants were transferred to a phytochamber with 16 h of light (~140 μMol m−2 sec−1), 20°C and 100% humidity for 3 days. Genomic DNA of infected Arabidopsis leaves was isolated from a pool of 20 leaf disks of 3 mm in diameter, harvested from two plants per genotype. P. infestans biomass was determined by quantitative real-time PCR, as described by Eschen-Lippold et al. (2007). In each experiment, three biological replicates were sampled. A minimum of two technical replicates of the PCR assay was used for each sample.
Blumeria graminis f. sp. hordei (Bgh) was propagated on barley (cultivar Ingrid, line I-10). Arabidopsis plants were inoculated using a settling tower, as described by Zimmerli et al. (2004). To score pathogen invasion and cell death in the epidermis, three leaf samples of each genotype were taken at 2 dpi and analyzed microscopically under UV light (Lipka et al., 2005). One hundred interaction sites were scored per leaf. Microscopic analyses were carried out at 6 dpi, and macroscopic phenotypes were monitored at 9 dpi.
The growth of G. cichoracearum (strain UCSC1) and infection of A. thaliana was performed as described by Adam et al. (1999). Microscopic analyses were carried out at 7 dpi. A minimum of nine leaves of different plants from each genotype was monitored in each experiment.
Bacteria were infiltrated into Arabidopsis rosette leaves using a needleless syringe. For the determination of bacterial growth, P. syringae pv. tomato (Pto DC3000 and Pto DC3000 avrRPM1) was infiltrated at 1 × 105 cfu ml−1 and P. syringae pv. phaseolicola (Pph) was infiltrated at 1 × 108 cfu ml−1. Immediately after infiltration (0 dpi) and at 4 dpi, bacterial titers were determined. In a minimum of three experiments, three leaf disks of 9 mm in diameter of three different plants per genotype were sampled and ground in 0.2 ml of 10 mm MgCl2. Bacterial growth was determined by plating serial dilutions (three technical replicates) on nutrient agar plates and counting the cfu after incubation for 2 days at 28°C.
For callose assays, Pto DC3000 avrRPM1 was infiltrated at 5 × 105 cfu ml−1 and aniline blue staining was performed 2 dpi.
To monitor the cell death response, pathogen-infected leaves were stained with trypan blue (0.1% trypan blue in lactophenol/ethanol) and destained with chloral hydrate solution (2.5 mg ml−1) according to Koch and Slusarenko (1990). Diaminobenzidine (DAB)-staining of H2O2 was performed as described by Thordal-Christensen et al. (1997). Stained leaves were stored in 50% glycerol solution and examined by light microscopy using an Axioplan 2 microscope (Zeiss, http://corporate.zeiss.com) equipped with a digital camera (AxioCam MRc5).
Callose depositions were visualized using aniline blue (0.01% in 150 mm KH2PO4, pH 9.5) as described by Adam and Somerville (1996). Stained leaves were stored in 50% glycerol solution in the dark and subsequently examined for fluorescence using a Nikon AZ100 stereomicroscope with standard filter block for ultraviolet fluorescence UV-2A (excitation 330–380 nm, dichroic 400 nm, emission 420 nm). The general staining of cell structures was achieved using 0.1% toluidine blue and 2.5% NHCO3 in water at 80°C for 10 s. Electron microscopy and detection of β-1,3-glucans was performed as described by Eschen-Lippold et al. (2012). Leaf disks (7 mm in diameter) were stained with Nile Red (#72485; Sigma-Aldrich, http://www.sigmaaldrich.com) as described by De Domenico et al. (2007).
Molecular cloning and generation of transgenic lines
To create a C-terminal ERP1-GFP fusion, the full-length coding sequence was amplified by PCR using the forward primer 5′-ATGGGAGCGAATTCGAAATCAGTAAC-3′, as well as the reverse primer 5′-TATGTACTGGAGAAGCATATTAGCAGAT-3′ and cloned into pEarleyGate103 (Earley et al., 2006). After transfer of the constructs into Agrobacterium tumefaciens (GV3101), Arabidopsis plants (gl1 and erp1-3) were transformed by floral-dip transformation (Zhang et al., 2006). Transgenic plants were selected on kanamycin-containing medium or after spraying with BASTA®(glufosinat-ammonium; Bayer, http://www.bayer.com). BASTA-resistant plants were screened for fluorescence using an Axioplan 2 microscope equipped with an epifluorescence filter set. Protein gel blot analyses were performed (Ausubel et al., 2003) using a 1:2000 dilution of an anti-GFP antibody (A11122; Invitrogen, http://www.invitrogen.com) to determine the expression of the ERP1-GFP fusion protein in transgenic plants.
Detection of intracellular GFP, mCherry and Nile Red fluorescence was performed with a Zeiss LSM 710 inverted confocal microscope. The 488-nm Argon ion laser line was used to excite GFP and Nile Red, whereas the excitation wavelength for mCherry was 561 nm. Emmision of GFP, Nile Red and mCherry was collected at 490–530, 550–600 and 585–645 nm, respectively. Aniline blue was excited using the 405-nm laser diode and the fluorescence was captured at 470–520 nm. Image processing was performed using zen 2009 software supplied with the microscope.
The peroxisome and Golgi marker proteins used in this study are described in Nelson et al. (2007). Co-localization was analyzed in N. benthamiana plants that transiently expressed ERP1-GFP under the control of the constitutively active 35S promoter from the Cauliflower mosaic virus.
Plasmolysis and FM4-64 staining of roots of seedlings was performed with 0.5 M NaCl solution containing 4 μm FM4-64. Confocal microscopy of roots and image processing was performed using a Leica TCS SP5 and the las af software provided (Leica, http://www.leica-microsystems.com).
Quantification of sterol lipids
Leaves of 5-week-old Arabidopsis plants were covered with 10-μl droplets containing 5 × 105P. infestans zoospores ml−1 or sprayed with water. At least three leaves of five different plants each [100–200 mg fresh weight (FW)] were harvested 3 dpi and ground in liquid N2. The extraction of total lipids, solid phase extraction, derivatization of free sterols and quantification of sterol lipids using Q-TOF-MS/MS were performed as described by Wewer et al. (2011).
Sequence data from this article can be found in the Arabidopsis Genome Initiative or the EMBL/GenBank data libraries under the following accession numbers: PSAT1/ERP1 (At1g04010), PEN2 (At2g44490), PMR4 (At4g03550) and GL1 (At3g27920).
We thank Felix Mauch (University of Fribourg, Switzerland) for providing the P. infestans isolate CRA208m2. We thank Shauna Somerville (University of California, Berkeley, USA; pmr4-1) and the SALK-Institute (T-DNA Insertion lines) for providing Arabidopsis seeds and Marina Häußler for technical assistance. We thank Bettina Hause, Hagen Stellmach and Jens Müller (IPB Halle) for their help with confocal imaging. This work was supported by the Deutsche Forschungsgemeinschaft within the Priority Program 1212.