SGT1 regulates wounding- and herbivory-induced jasmonic acid accumulation and Nicotiana attenuata’s resistance to the specialist lepidopteran herbivore Manduca sexta

Authors

Errata

This article is corrected by:

  1. Errata: Erratum Volume 190, Issue 3, 809, Article first published online: 17 February 2011

Author for correspondence:
Jianqiang Wu
Tel: +49 3641 571122
Email: jwu@ice.mpg.de

Summary

  • SGT1 (suppressor of G-two allele of SKP1) is a conserved protein in all eukaryotes and is crucial for resisting pathogens in humans and plants. We studied whether SGT1 is involved in the induced defense response of a native tobacco (Nicotiana attenuata) to its natural herbivore, Manduca sexta.
  • We diminished NaSGT1 transcription in N. attenuata using virus-induced gene silencing (VIGS) and analysed the induced defense responses after wounding and M. sexta elicitation.
  • Silencing NaSGT1 highly attenuates wounding- and herbivory-induced amounts of jasmonic acid (JA) and JA-isoleucine but elevates the concentration of salicylic acid. Chemical profiling reveals that NaSGT1-silenced plants are also compromised in their ability to accumulate JA precursors produced in chloroplasts. We show that the reduced JA accumulation in NaSGT1-silenced plants is independent of the elevated salicylic acid levels. NaSGT1-silenced plants have decreased contents of defensive metabolites and have compromised resistance to M. sexta larvae. Transcript analyses after methyl jasmonate (MeJA) treatment revealed that NaSGT1 is important for the normal regulation of MeJA-induced transcriptional responses.
  • This work demonstrates the importance of SGT1 in the regulatory network that deploys defense responses against herbivores, and highlights the significance of SGT1 in plants’ responses to JA.

Introduction

Plants have evolved sophisticated strategies to adapt their physiology to changing environmental conditions. The interplay among complex signaling networks, including various pathways regulated by phytohormones such as salicylic acid (SA), jasmonic acid (JA), ethylene and ABA, dramatically influences plants’ stress responses (Chow & McCourt, 2006; Pieterse et al., 2009; Santner & Estelle, 2009). By tightly controlling the accumulation and perception of phytohormones, plants counteract environmental stresses. In recent years, new genetic and biochemical tools have greatly advanced our understanding of the biosynthesis, signal transduction, and physiological functions of phytohormones in various plant species. Among these, JA plays a crucial role in plant development and adaptation to environmental stresses.

Jasmonic acid is important for plants’ defense against insects and microbial pathogens, resistance to drought, ultraviolet radiation, ozone and other abiotic stresses (Conconi et al., 1996; Glazebrook, 2005; Howe & Jander, 2008; Browse, 2009). It is also involved in plant senescence and reproductive development (Feys et al., 1994; Li et al., 2004; Buchanan-Wollaston et al., 2005). JA biosynthesis and signaling are embedded in a complex network that includes crosstalk with other phytohormones such as SA and auxin (Harms et al., 1998; Spoel et al., 2003; Diezel et al., 2009; Grunewald et al., 2009), light signaling (Moreno et al., 2009) and mitogen-activated protein kinase (MAPK) pathways (Brodersen et al., 2006, Wu et al., 2007). Recently, significant progress has been made in understanding the molecular basis of JA signaling. The F-box protein COI1 (CORONATINE-INSENSITIVE 1) is part of a Skp/Cullin/F-box complex [SCF(COI1)] that acts as an ubiquitin ligase and is required for physiological responses mediated by JA (Xu et al., 2002; Chini et al., 2007; Paschold et al., 2007; Thines et al., 2007; Katsir et al., 2008; Yan et al., 2009). JARs (JASMONATE RESISTANT) conjugate JA with isoleucine to form JA-Ile (Staswick & Tiryaki, 2004). The binding of JA-Ile to SCF(COI1) facilitates the ubiquitination of JAZs (JASMONATE ZIM DOMAIN proteins), which are negative regulators of JA-induced transcriptional changes, and thereafter the degradation of JAZs through a 26S proteasome-mediated proteolytic pathway, and thus finally activates JA-induced responses (Chini et al., 2007; Thines et al., 2007; Katsir et al., 2008).

Both JA/JA-Ile and COI1 are required for herbivory-induced defense reactions in plants. In some plant species, induced resistance to insect herbivores is activated when tissues are wounded and certain molecules in herbivore oral secretions (OS) are recognized (reviewed in Howe & Jander, 2008; Wu & Baldwin, 2009). Recognition of herbivory triggers a set of diverse physiological responses that enhance plants’ defense levels against the attacking herbivores (Howe & Jander, 2008; Wu & Baldwin, 2009). The interaction between Nicotiana attenuata, a wild tobacco plant that grows in western North America, and its lepidopteran herbivore Manduca sexta has been intensively studied. After M. sexta attack, fatty acid-amino acid conjugates (FACs) present in the OS of M. sexta are rapidly recognized by N. attenuata; FACs amplify and modify wound-induced responses in N. attenuata, including the activation of MAPKs and the initiation of the biosynthesis of JA and JA-Ile (Wu et al., 2007; Kallenbach et al., 2010). In N. attenuata, JA/JA-Ile and COI1 control the accumulation of various anti-herbivore secondary metabolites (Halitschke & Baldwin, 2003; Paschold et al., 2007; Wang et al., 2007), including trypsin proteinase inhibitors (TPIs), which inhibit the digestion of proteins in insect midguts, phenolic compounds, and diterpeneglycosides (DTGs) (Zavala et al., 2004; Paschold et al., 2007; Wang et al., 2007; Jassbi et al., 2008; Kaur et al., 2010; Heiling et al., 2010). Clearly, the accumulation and signaling of JA/JA-Ile play a pivotal role in plant resistance to herbivores.

SGT1 (SUPPRESSOR OF G-TWO ALLELE OF SKP1) is conserved in all eukaryotes (Shirasu, 2009). It is required for immune responses to pathogens in humans and plants (Muskett & Parker, 2003; Mayor et al., 2007). It is thought to confer at least partial resistance to pathogens through its interaction with HSP90 (reviewed in Shirasu, 2009). In plants, SGT1 and HSP90 mediate the stability of NB-LRR type R proteins (Lu et al., 2003; Boter et al., 2007). RAR1 physically binds to SGT1 and HSP90 and thereby stabilizes the interaction between them (Boter et al., 2007). RAR1 is required for resistance conferred by some, but not all R genes, whose functions are dependent on SGT1and HSP90 (Liu et al., 2004; Boter et al., 2007). Furthermore, SGT1 also interacts with components of SCF–ubiquitin ligase complexes in yeast and plants, suggesting that it functions in the proteasome-mediated protein degradation pathway (Kitagawa et al., 1999; Azevedo et al., 2002; Liu et al., 2002). Auxin is perceived by its receptor TIR1, an F-box protein; the binding of auxin to SCF(TIR1) triggers degradation of Aux/IAA proteins which are negative regulators of auxin-induced transcriptional responses (Tan et al., 2007). In Arabidopsis, it was found that SGT1 is required for SCF(TIR1)-mediated auxin responses; moreover, sgt1b mutants are slightly insensitive to jasmonate-induced inhibition on root growth; given that COI1 is also a part of SCF–ubiquitin ligase complex, a role for SGT1 in SCF(COI1)-mediated JA responses has been suggested (Gray et al., 2003; Lorenzo & Solano, 2005).

The roles of SGT1 in plant resistance to pathogens have been intensively studied. However, whether SGT1 is also involved in induced resistance to leaf-chewing herbivores has not yet been examined. In this work, we investigated whether SGT1 is important for herbivory-induced defenses in N. attenuata. We used virus-induced gene silencing (VIGS) to knock down the transcript levels of NaSGT1 and analysed the role of NaSGT1 in modulating herbivory-induced responses in N. attenuata. Silencing NaSGT1 diminishes wounding- and herbivory-induced accumulation of JA and JA-Ile. The reduced herbivory-elicited JA concentrations in NaSGT1-silenced plants do not result from the antagonistic effect of SA, and these plants have normal levels of MAPK activity. Accordingly, NaSGT1-silenced plants have decreased amounts of defensive metabolites after herbivory, and M. sexta larvae gain more weight on these plants than on empty vector plants. Furthermore, NaSGT1 is required for the normal regulation of methyl jasmonate (MeJA)-induced transcriptional responses in N. attenuata. This study highlights the important role of SGT1 in plant resistance to leaf-chewing herbivores.

Materials and Methods

Plant growth, virus-induced gene silencing, and sample treatments

Nicotiana attenuata seeds were from a line maintained in our laboratory that was originally collected in Utah, USA, and inbred for 30 generations in the glasshouse. Seed germination and plant rearing are described in Krügel et al. (2002). Plants were grown at 20–22°C under 16 h of light. To analyse secondary metabolites and trypsin protease inhibitor activity, plants were transferred to a glasshouse set at 26°C during the time-course of the experiment. A VIGS system was used to silence the accumulation of NaSGT1 transcripts following a VIGS procedure optimized for N. attenuata (Ratcliff et al., 2001; Saedler & Baldwin, 2004).

Leaves were wounded with a pattern wheel; immediately thereafter either 10 μl of water or 10 μl of M. sexta oral secretions (OS, 1 : 5 diluted) was applied to the wounds (W + W and W + OS treatment). For MeJA treatments, MeJA was dissolved in heat-liquefied lanolin at specified concentrations; 10 μl of the solid paste was applied to leaves, and pure lanolin was applied as controls. Samples were harvested at indicated times, frozen in liquid nitrogen and stored at −80°C until use.

Cloning of NaSGT1 cDNA sequence and sequence alignments

The open reading frame of NaSGT1 was amplified from total cDNA prepared from 1 h W + OS-treated N. attenuata leaf tissue by PCR, using Phusion High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland) and primers SGT1-1 (5′-ATG GCG TCC GAT CTG GAG ATT A-3′) and SGT1-2 (5′-GAT TTC CCA TTT CTT CAG CTC-3′). The PCR product was cloned into a pJET1.2/blunt Cloning Vector (Fermentas, Vilnius, Lithuania) and sequenced. The NaSGT1 sequence was further confirmed by blasting a N. attenuata transcriptome database obtained by 454 sequencing.

The SGT1 sequences in N. benthamiana, tomato (Solanum lycopersicum) and Arabidopsis were retrieved from GenBank. Alignments of nucleotide and amino acid sequences were done using the Clustal W algorithm (DNASTAR, Lasergene 8, Madison, WI, USA).

Analysis of JA, JA-Ile and SA

One milliliter of ethyl acetate spiked with 200 ng of D2-JA, 40 ng of D4-SA and 40 ng of 13C6-JA-Ile (the internal standards for JA, SA and JA-Ile/JA-Leu, respectively) was added to each sample (c. 150 mg). Samples were homogenized on a FastPrep homogenizer (Thermo Electron, Waltham, MA, USA). After being centrifuged at maximum speed for 10 min at 4°C, supernatants were transferred to 2-ml Eppendorf tubes and evaporated to dryness on a vacuum concentrator (Eppendorf, Hamburg, Germany). The residue was resuspended in 0.5 ml of 70% methanol (v : v) and centrifuged to clarify phases. The supernatants were analysed on an high-pressure liquid chromatography (HPLC)-mass spectrometry (MS)/MS (1200L LC-MS system; Varian, Foster City, CA, USA).

Secondary metabolite analysis

Extraction and analysis of secondary metabolites was modified from Keinanen et al. (2001). A 200 mg aliquot of tissue in FastPrep tubes containing 0.9 g of FastPrep matrix (Sili GmbH, Warmensteinach, Germany) was homogenized in 1 ml of 40% methanol containing 0.5% acetic acid on a FastPrep homogenizer (Thermo Electron) for 45 s and then samples were centrifuged at 4°C for 12 min at maximum speed. The supernatants were transferred into 1.5-ml Eppendorf tubes, centrifuged and finally transferred to glass vials. Analysis was done on an HPLC (HPLC 1100 series; Agilent, Foster City, CA, USA), installed with an ODS Inertsil C-18 column (3 μm, 150 × 4.6 mm i.d.; GL Sciences, Tokyo, Japan). 0.25% H3PO4 in water and acetonitrile were used as the mobile phase. Solutions of nicotine, caffeoylputrescine, chlorogenic acid were used as external standards for quantification.

Trypsin proteinase inhibitor activity analysis

The TPI activity was quantified using a radial diffusion assay protocol described by Jongsma et al. (1994).

Protein extraction and in-gel kinase activity assay

Leaf tissue pooled from four replicate leaves was ground in liquid nitrogen and 250 μl of extraction buffer (100 mM Hepes pH 7.5, 5 mM EDTA, 5 mM EGTA, 10 mM Na3VO4, 10 mM NaF, 50 mM β-glycerolphosphate, 1 mM phenylmethylsulfonyl fluoride, 10% glycerol, complete proteinase inhibitor cocktail tablets (Roche)) was added to 100 mg tissue. Leaf tissue was then completely suspended by vortexing. After being centrifuged at 4°C for 20 min, supernatants of samples were transferred to fresh tubes. Protein concentrations were measured using the Bio-Rad Protein Assay Dye Reagent with BSA (Sigma) as a standard. Kinase activity assay was performed as described in Zhang & Klessig (1997).

Manduca sexta bioassay

Freshly hatched M. sexta larvae obtained from in-house colonies were placed on rosette leaves of plants (1 larva/plant, 37 replicated plants) 3 wk after agroinfiltration for VIGS. Mass gain was measured after the number of days indicated.

RNA extraction and quantitative real-time PCR

Total RNA was extracted using the TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. Half a microgram of total RNA was reverse-transcribed using oligo(dT)18 and SuperScript reverse transcriptase II (Invitrogen). Quantitative real-time PCR (qPCR) was performed on an Mx3005P Multiplex qPCR system (Stratagene, Santa Clara, CA, USA) using the qPCR Core kit for SYBR Green I (Eurogentec, Seraing, Belgium). An NaActin gene was used an internal standard to correct the variation of cDNA concentrations. All qPCR primers used are listed in the Supporting Information, Table S2.

Southern blotting analysis

Both DNA extraction and Southern blotting were done following the procedure described in Wu et al. (2006).

Two-dimensional gel proteomics, mass spectrometry and peptide analysis

Protein extraction, purification and two-dimensional gel electrophoresis were done according to Giri et al. (2006). Each protein sample was extracted from pooled leaves of six replicated plants. Three replicated gels loaded with different protein samples were run and evaluated for each plant type (VIGS-EV and VIGS-NaSGT1) and treatment. Gel image scanning and measurement of intensity values of protein spots, peptide analysis and database search were done as described in Giri et al. (2006).

Analysis of free linolenic acid, 13-hydroperoxy-linolenic acid and OPDA

Extraction and analysis was done according to Kallenbach et al. (2010).

Statistical analysis

Statistical analyses were done using statview (SAS Institute Inc., Cary, NC, USA) and R (http://www.r-project.org).

Sequence data from this article can be found in the GenBank/EMBL database under accession number GU265726 (NaSGT1).

Results

Silencing NaSGT1 in N. attenuata

Two copies of SGT1, AtSGT1a and AtSGT1b, exist in the Arabidopsis genome. Mutations in AtSGT1b but not in AtSGT1a decrease plant resistance to pathogens, and sgt1a sgt1b double mutants are lethal (Austin et al., 2002; Azevedo et al., 2006). Two isoforms of SGT1 were also found in tomato, and silencing these SGT1 homologues with VIGS leads to different degrees of plant growth defects and reduces Mi-1-mediated resistance to nematodes and aphids (Bhattarai et al., 2007). Similarly, two SGT1 homologues exist in N. benthamiana; silencing NbSGT1-1 results in reduced growth and compromised pathogen defense mediated by various R genes (Peart et al., 2002).

An N. attenuata NaSGT1 cDNA sequence was obtained by cloning, whose coding sequence showed 99%, 97% and c. 70% similarity to SGT1 in N. benthamiana, tomato and Arabidopsis (70% to AtSGT1a, 69% to AtSGT1b), respectively (Fig. S1a). Similar levels of homology were seen for their deduced amino acid sequences (Fig. S1b). Although cloning and scrutinizing a N. attenuata transcriptome database prepared by 454 sequencing of transcripts from all plant parts and developmental stages did not reveal any additional sequences of SGT1 (data not shown), Southern blotting analysis indicated that there is likely another SGT1 isoform in the N. attenuata genome (Fig. S2). To investigate the function of NaSGT1 in N. attenuata’s defense responses to its specialist herbivore M. sexta, we used a VIGS system optimized for N. attenuata to knock down the transcript levels of NaSGT1 (Ratcliff et al., 2001; Saedler & Baldwin, 2004). N. attenuata inoculated with Agrobacterium tumefaciens transformed with pTV00 empty vector and pTV-NaSGT1 containing a 316 bp fragment of NaSGT1 (Fig. S3) formed VIGS-EV and VIGS-NaSGT1 plants, respectively (Fig. 1a). To determine whether the pTV-NaSGT1 construct effectively silenced NaSGT1 transcript levels, VIGS-EV and VIGS-NaSGT1 plants were wounded with a pattern wheel and 20 μl of M. sexta OS was applied immediately to the wounds (W + OS); this treatment effectively elicits herbivory-induced responses in N. attenuata (Halitschke et al., 2001). For comparison, 20 μl of water was applied to wounds (W + W). In VIGS-EV plants, 3 h after W + W and W + OS treatment, NaSGT1 transcript levels elevated more than twofold, and W + W tended to induce slightly higher levels of NaSGT1 than did W + OS treatment (t-test; P = 0.07). Compared with VIGS-EV plants, NaSGT1 transcript levels reduced by 61% to 73% in VIGS-NaSGT1 plants (Fig. 1b), confirming the effectiveness of the VIGS. Furthermore, given the high similarity between the two N. benthamiana SGT1 genes, we assume that our silencing approach knocked down the expression of both genes in N. attenuata, if in fact the SGT1 homologue exists and is expressed. Silencing NaSGT1 altered plant growth, leading to decreased rosette diameter (Fig. 1c). In later stages, NaSGT1-silenced plants also lost apical dominance and had stunted growth (data not shown), suggesting that the role NaSGT1 plays in plant development in N. attenuata is similar to its role in other solanaceous plants. We performed all analyses of NaSGT1-dependent defense responses on the leaves of rosette-stage plants (Fig. 1a).

Figure 1.

 Silencing NaSGT1 with virus-induced gene silencing (VIGS). (a) Plants at rosette stage. Nicotiana attenuata plants were inoculated with Agrobacterium carrying VIGS construct, pTV00, pTV-NaSGT1, or pTV-NaPDS to form VIGS-EV, VIGS-NaSGT1 and VIGS-NaPDS (phytoene desaturase) plants, respectively. VIGS-NaPDS plants were used to visually determine the degree of gene silencing, because they have a photo-bleaching phenotype. (b) Mean (± SE) NaSGT1 transcript levels in VIGS-EV and VIGS-NaSGT1 plants. Leaves from five replicated plants at the rosette stage were wounded with a pattern wheel, treated with 10 μl of water (W + W) or with 10 μl of Manduca sexta oral secretions (W + OS), and harvested at the times indicated. Transcript levels of NaSGT1 in these plants were analysed with qPCR. (c) Mean (± SE) rosette radiuses were measured 20 d after plants were inoculated with Agrobacterium-carrying VIGS constructs. Asterisks represent significant differences between VIGS-EV and VIGS-NaSGT1 plants (Student’s t-test; *, < 0.05; **, < 0.01; ***, < 0.001; N = 5).

Silencing NaSGT1 reduces the accumulation of wounding- and herbivory-induced JA and JA-Ile

The role of SGT1 in plant defense against pathogens has been intensively studied (Muskett & Parker, 2003). However, whether SGT1 is important for plants’ induced resistance to phytophagous insects has not been explored. Mechanical wounding and herbivory quickly activate JA and JA-Ile biosynthesis in N. attenuata and JA/JA-Ile accumulation is essential for inducing most anti-herbivory defense compounds (Halitschke & Baldwin, 2003; Wang et al., 2007; Heiling et al., 2010). To determine if NaSGT1 is important for the elicitation of JA and JA-Ile, we analysed their concentrations in VIGS-EV and VIGS-NaSGT1 plants after W + W and W + OS treatment. One hour after W + W treatment, JA and JA-Ile concentrations in NaSGT1-silenced plants were reduced by 65% and 66%, respectively, compared with those in VIGS-EV plants (Fig. 2a). Similarly, 1 h after W + OS treatment, JA and JA-Ile concentrations in VIGS-NaSGT1 plants were only 32% and 36% as high as their concentrations in VIGS-EV plants (Fig. 2a). These data indicate that N. attenuata’s highly elevated concentrations of JA/JA-Ile in response to wounding and herbivory are NaSGT1-dependent.

Figure 2.

 Silencing NaSGT1 diminishes wounding- and herbivory-induced jasmonic acid (JA) accumulation. (a) Mean (± SE) levels of JA and JA-Ile. Leaves were wounded with a pattern wheel and treated with 10 μl of water (W + W) or 10 μl of Manduca sexta oral secretions (W + OS). Samples were harvested at indicated times. JA and JA-Ile content were analysed on an high-pressure liquid chromatography (HPLC)-mass spectrometry (MS)/MS (n = 5). (b) Concentrations (± SE) of several precursors of JA. Leaves were treated with W + OS and samples were collected at times indicated (n = 6 to 10). Asterisks represent significant differences between VIGS-EV and VIGS-NaSGT1 plants (Student’s t-test; *, < 0.05; **, < 0.01; ***, < 0.001).

As the physical interaction of SGT1 with HSP90 and RAR1 is required for plant resistance to pathogens, we also analysed herbivory-induced concentrations of JA and JA-Ile in NaHSP90-1- and NaRAR1-silenced plants (Fig. S4). The concentrations of W + OS-induced JA and JA-Ile were reduced only in VIGS-NaHSP90-1 plants. Thus, herbivory-induced JA accumulation in N. attenuata requires NaSGT1 and NaHSP90-1 but not NaRAR1, although it is still not known whether NaSGT1 binds to NaHSP90-1 and NaRAR1 in N. attenuata and whether these physical interactions are required for the normal regulation of herbivory-induced accumulation of JA and JA-Ile. As NaHSP90-1-silenced plants displayed strong developmental defects such as dwarfism and necrosis of apical meristems after they reached the early elongation stage, their defense responses against insects were not further investigated.

In order to understand the mechanism underlying the impaired herbivory-induced accumulation of JA in VIGS-NaSGT1 plants, we analysed the concentrations of the JA precursors, free linolenic acid (LA), 13-hydroperoxy-linolenic acid (18-3 OOH) and 12-oxo-phytodienoic acid (OPDA) within 30 min after W + OS treatment (Fig. 2b). The LA concentrations were reduced by 40% before and after W + OS elicitation in VIGS-NaSGT1 plants compared with those in VIGS-EV plants. The 18-3 OOH concentrations were also reduced, although to a lesser extent than those of LA (Fig. 2b). In VIGS-EV plants, 15 min after W + OS treatment, OPDA concentrations increased 10-fold; remarkably, in VIGS-NaSGT1 plants, OPDA concentrations remained the same after W + OS treatment. Thus, NaSGT1 is clearly required for the normal accumulation of herbivory-induced OPDA. The same extraction method also confirmed the attenuated amounts of JA 30 min after W + OS treatment in VIGS-NaSGT1 plants (Fig. 2b). Therefore, NaSGT1 is important for the normal accumulation of various herbivory-induced JA precursors and JA.

Impaired JA accumulation in VIGS-NaSGT1 plants is independent of MAPK activity and high SA levels

Two MAPKs, NaSIPK and NaWIPK, are rapidly activated by M. sexta herbivory, and silencing NaSIPK and NaWIPK compromises wounding- and herbivory-induced JA/JA-Ile accumulation in N. attenuata (Wu et al., 2007; Meldau et al., 2009). To test if silencing NaSGT1 impairs MAPK activity, leading to reduced JA and JA-Ile accumulation, we treated VIGS-EV and VIGS-NaSGT1 plants with W + OS and examined the levels of NaSIPK and NaWIPK activity with an in-gel kinase assay. Levels of NaSIPK and NaWIPK activity were not altered in NaSGT1-silenced plants compared with those in VIGS-EV plants (Fig. 3a). NaSIPK- and NaWIPK-silenced plants also show highly diminished transcript levels of genes involved in JA and JA-Ile biosynthesis (Wu et al., 2007). However, in VIGS-NaSGT1 plants, the transcript levels of genes involved in JA and JA-Ile biosynthesis did not differ markedly from the transcript levels of the same genes in VIGS-EV plants (Fig. S5). These data suggest that NaSGT1 mediates the accumulation of herbivory-induced JA and JA-Ile in a pathway that is either downstream or parallel to the MAPK cascade; moreover, the reduced amounts of JA and JA-Ile do not result from a mis-regulation of the transcript levels of their biosynthetic genes in NaSGT1-silenced plants.

Figure 3.

 Mitogen-activated protein kinase (MAPK) activity and elevated salicylic acid (SA) levels in VIGS-NaSGT1 plants do not account for plants’ attenuated jasmonic acid (JA) levels. (a) Silencing NaSGT1 does not compromise MAPK activation. Plants were wounded with a pattern wheel and 10 μl of Manduca sexta oral secretions (OS) was applied to the wounds (W + OS) and leaves of five replicate plants were harvested at indicated times. Kinase activity was measured with an in-gel activity assay using myelin basic protein as the substrate. (b) Mean (± SE) SA levels in VIGS-EV and VIGS-NaSGT1 plants. Leaves were wounded and treated with 10 μl of water (W + W) or treated with W + OS, and were harvested at indicated times. Asterisks represent significant differences between VIGS-EV and VIGS-NaSGT1 plants (Student’s t-test; *, < 0.05; **, < 0.01; ***, < 0.001; N = 5). (c) and (d) Mean concentrations (± SE) of SA and JA. Wild-type (WT) plants and plants expressing 35S:NahG were used to generate VIGS-EV and VIGS-NaSGT1 plants. One hour after W + OS treatment, samples were collected and their SA and JA content were analysed; untreated samples served as controls (C). Different small letters represent statistically significant differences based on the minimum adequate model (Fig. 3c: ANOVA, F3,26 = 167.84, < 0.0001; Fig. 3d: ANOVA, F3,25 = 147.91, < 0.0001).

The suppression of JA biosynthesis by high SA levels has been reported in different plant species (Spoel et al., 2003; Mur et al., 2006; Diezel et al., 2009). We analysed basal, W + W- and W + OS-induced SA concentrations in VIGS-EV and VIGS-NaSGT1 plants. Basal SA concentrations in VIGS-NaSGT1 plants were twice as high as those in VIGS-EV plants (Fig. 3b). One hour after W + W and W + OS treatment, concentrations of SA were four times higher in VIGS-NaSGT1 plants than in VIGS-EV plants, suggesting that NaSGT1 is important for SA homeostasis in N. attenuata (Fig. 3b). To test if the high concentrations of SA account for the suppressed JA accumulation after herbivory, we silenced NaSGT1 in plants that were transformed with bacterial salicylate hydroxylase under the 35S promoter (35S:NahG). Analysis of SA concentrations indicated that NahG effectively abolished the accumulation of SA in all plants (Fig. 3c). Reducing SA concentrations in VIGS-EV plants resulted in a c. 40% increase in JA accumulation 1 h after W + OS treatment (Fig. 3d). However, reducing SA levels by > 90% in VIGS-NaSGT1 plants did not significantly increase JA levels 1 h after W + OS treatment, demonstrating that the high SA levels do not account for the drastically diminished herbivory-elicited JA accumulation in NaSGT1-silenced plants.

NaSGT1-silenced plants have decreased amounts of herbivory-induced defensive secondary metabolites and have reduced resistance to the specialist herbivore, M. sexta

Transgenic N. attenuata plants with reduced JA or JA-Ile concentrations have compromised resistance to herbivores because of their impaired ability to induce anti-herbivore defensive compounds (Halitschke & Baldwin, 2003; Wang et al., 2007). As silencing NaSGT1 impaired the accumulation of wounding- and herbivory-induced JA and JA-Ile, we examined whether this was correlated with altered concentrations of defensive metabolites. Diterpeneglycosides are highly abundant defensive metabolites in N. attenuata whose concentrations increase after M. sexta herbivory in a jasmonate-dependent manner (Jassbi et al., 2008; Heiling et al., 2010). Caffeoylputrescine (CP) and TPIs are both anti-herbivory compounds that are induced by JA and JA-Ile and M. sexta attack (Zavala et al., 2004; Kaur et al., 2010). We analysed the concentrations of CP and DTGs and the activity of TPIs in NaSGT1-silenced plants 3 d after W + W and W + OS treatment. Basal levels of CP were slightly elevated in NaSGT1-silenced plants (Fig. 4a, Students t-test, P = 0.08) but CP levels were not further induced by W + W and W + OS treatment; by contrast, VIGS-EV plants showed increased levels of CP after these treatments. The DTG concentrations were not altered by W + W treatment and VIGS-NaSGT1 and VIGS-EV plants did not differ in DTG concentrations in control and W + W-treated leaves, whereas the W + OS treatment increased DTG levels in VIGS-EV plants but not in VIGS-NaSGT1 plants: DTG levels were 50% of those in VIGS-EV plants (Fig. 4b). Similarly, basal TPI activity was higher in NaSGT1-silenced plants than in VIGS-EV plants, but no difference was found after W + W treatment (Fig. 4c). After W + OS treatment, TPI activity in VIGS-NaSGT1 plants was only 50% of that in VIGS-EV plants (Fig. 4c). Thus, in line with the low JA levels in W + W- and W + OS-treated VIGS-NaSGT1 plants, these treatments did not induce accumulation of defensive secondary metabolites, while CP, DTG and TPI concentrations were increased after these treatments in VIGS-EV plants.

Figure 4.

 Silencing NaSGT1 compromises resistance of Nicotiana attenuata to Manduca sexta. (a–c) Levels of defensive metabolite caffeoylputrescine, diterpeneglycoside (DTG) and trypsin proteinase inhibitors (TPI) activity in VIGS-EV and VIGS-NaSGT1 plants. Plants were wounded with a pattern wheel and treated with 10 μl of water (W + W) or 10 μl of M. sexta oral secretions (W + OS). Three days after treatment, samples were harvested and analyzed; non-treated samples served as controls (C) (N = 5). (d) M. sexta larval performance on VIGS-EV and VIGS-NaSGT1 plants. Manduca sexta neonates were placed on rosette leaves 3 wk after Agro-inoculation of VIGS constructs, and their mass was recorded at the times indicated. Asterisks represent significant differences between VIGS-EV and VIGS-NaSGT1 plants (Student’s t-test; *, < 0.05; **, < 0.01; ***, < 0.001; N = 37).

We next examined if silencing NaSGT1 compromises the resistance of N. attenuata to its specialist herbivore M. sexta. M. sexta neonates were placed on rosette-staged VIGS-EV and VIGS-NaSGT1 plants and larval mass gain was measured. After 9 d, the mass of M. sexta larvae was c. 1.5 times greater in those that fed on NaSGT1-silenced plants than in those that fed on VIGS-EV plants (Fig. 4d). On day 11, mass was 40% higher in larvae that fed on VIGS-NaSGT1 plants than on VIGS-EV plants. We conclude that silencing NaSGT1 reduces N. attenuata’s resistance to its specialist herbivore, M. sexta; this is most likely because of diminished herbivory-induced JA and JA-Ile concentrations, which result in insufficiency of induced anti-herbivore defensive metabolites.

Silencing NaSGT1 alters N. attenuata’s transcriptional response to methyl jasmonate

We next examined the possibility that silencing NaSGT1 compromises the activity/stability of SCFCOI1 complex and thus decreases JA-induced responses in N. attenuata. As our attempts to create transformed plants that were stably silenced in NaSGT1 were unsuccessful, assays of MeJA-induced root growth inhibition for N. attenuata were not carried out. Apart from inhibition of plant root growth, jasmonate application elicits various transcriptional changes. We therefore analysed changes in gene transcript accumulation after MeJA treatment in VIGS-EV and VIGS-NaSGT1 plants. We treated the leaves of VIGS-EV and VIGS-NaSGT1 plants with lanolin pastes (10 μl) alone or lanolin pastes containing 5, 50 and 150 μg of MeJA; leaves were harvested at different time-points after treatment, and transcript accumulations of several genes that are known to be highly induced by MeJA in N. attenuata were analysed using quantitative real time-PCR. As most of the physiological changes induced by JA require JA perception through COI1, we also created VIGS-NaSGT1 and VIGS-EV plants in COI1-silenced N. attenuata background (irNaCOI1 plants) (Paschold et al., 2007) and induced these plants with 150 μg MeJA in 10 μl lanolin, so that we could examine whether NaSGT1 mediates the transcript accumulation of JA-inducible genes in a COI1-independent pathway. Compared with no treatment, lanolin treatment alone did not alter transcript levels of the genes examined (data not shown). The induction of all genes that we measured required JA perception through COI1 (Fig. 5). Compared with those in VIGS-EV plants, silencing NaSGT1 reduced the transcript levels of several genes that are upregulated by MeJA treatment, including threonine deaminase (NaTD) involved in converting Thr to Ile for JA-Ile biosynthesis and NaLOX2, which is important for fatty acid peroxidation (Fig. 5). Conversely, the relative transcript levels of NaTPI, which is frequently used as a JA-inducible marker in solanaceous plants, were higher after MeJA treatment in VIGS-NaSGT1 plants than in VIGS-EV plants. Elevated transcript levels were seen for an α-dioxygenase (Na α-DOX) gene in VIGS-NaSGT1 plants. The transcript levels of NaHPL (hydroperoxide lyase), which is involved in C6 metabolism, did not differ in VIGS-NaSGT1 and VIG-EV plants. These data demonstrate that NaSGT1 is required for COI1-dependent transcriptional regulation of MeJA-induced genes in N. attenuata.

Figure 5.

 Silencing NaSGT1 alters Nicotiana attenuata plants’ transcriptional responses to methyl jasmonate (MeJA). Relative transcript levels of jasmonic acid (JA)-inducible genes in VIGS-EV and VIGS-NaSGT1 plants after the application of MeJA. Leaves of VIGS-EV and VIGS-NaSGT1 plants were treated with 10 μl of lanolin paste containing different amount of MeJA. Transcript levels of NaTPI, Na α-DOX, NaHPL, NaTD and NaLOX2 were examined by quantitative real-time PCR using cDNA prepared from RNA isolated from five biological replicated plants.

Silencing NaSGT1 leads to altered levels of pathogenesis-related proteins and chaperone proteins

Given the remarkably altered defense against herbivores in NaSGT1-silenced plants, NaSGT1 might modulate the abundance of many proteins related to defenses. We employed a quantitative proteomic approach to examine the difference of protein abundance in VIGS-EV and VIGS-NaSGT1 plants. VIGS-EV and VIGS-NaSGT1 plants were treated with W + OS three times in 2-h intervals, leaf samples were harvested 6 h after the initial treatment and leaves from non-treated plants were harvested as controls. We extracted total proteins from these samples and separated them using two-dimensional electrophoresis (2-DE). Protein spots that differentially accumulated in all three biological replicates were sequenced and blasted in protein databases. In untreated samples, among the roughly 1200 protein spots detected on these gels, only a few differences were found between gels running samples from VIGS-EV and from VIGS-NaSGT1 plants. Very similar results were obtained from W + OS-treated samples (data not shown). Four proteins accumulated in higher amounts in VIGS-NaSGT1 plants than in VIGS-EV plants (Tables 1, S1; Fig. S6), all of which belonged to the pathogenesis-related (PR) protein family. Three of these proteins, namely α-galactosidase, osmotin and a hevein-like protein, were detected in VIGS-NaSGT1 plants but not in VIGS-EV plants (Tables 1, S1; Fig. S6). A protein spot identified as an endochitinase was found to have > 10-fold higher intensity in gels loaded with samples from VIGS-NaSGT1 plants than in gels loaded with samples from VIGS-EV plants (Table 1; Fig. S6). To further understand the molecular basis for the accumulation of these stress-related proteins, their transcript levels were also analysed. Relative transcript accumulation of NaEndochitinase showed several hundred-fold increase and NaOsmotin and NaHevein were c. 50 and 40 times increased in VIGS-NaSGT1 plants compared with VIGS-EV plants; however, the transcript levels of the gene encoding α-galactosidase showed no differences between untreated control VIGS-EV and VIGS-NaSGT1plants; higher levels of transcripts were seen in W + OS-induced plants (Fig. S7). Given that similarly higher amounts of α-galactosidase were found VIGS-NaSGT1 plants regardless of treatments, possibly the abundance of α-galactosidase was regulated on a posttranscriptional level.

Table 1.   Differentially accumulated proteins in VIGS-EV and VIGS-NaSGT1 plants
Protein name, accession number, and speciesPeptide(s)Mean (± SE) of normalized spot intensities
VIGS-EVVIGS-NaSGT1P
  1. *n.d. = not detected.

Endochitinase
P08252
Nicotiana tabacum
(P)SCHDVLLG(R)
(Y)APGFGTSGDTTA(R)
(Y)SNNGTGVSPGDNLDCGNQ(R)
(T)STYNLLGVSPGDNLDTDNE(R)
(G)PLQLSHNYGGYGPANS(T)
594 ± 1047663 ± 10580.001
Hevein-like protein
AAO63574
Hevea brasiliensis
(V)TNTGTGAKVSD(R)n.d.*250 ± 68.60.01
Osmotin
Q01591
Solanum lycopersicum
(Q)GTWVLNAP(R)
(N)NCPYTVWAASTLPGN(R)
n.d.*1476 ± 5710.03
α-Galactosidase
AAP04002
Carica papaya
(A)PLLLGCDL(R)n.d.*131.33 ± 5.36≤ 0.001
Trigger factor-like chloroplast chaperone
D1HZU8
Vitis vinifera
(N)VPEDLLVGYVG(K)
(L)SEQQLASLSSP(R)
(A)AVESLLK(R)
354 ± 39177 ± 7.50.01
NaCPN (herbivory-induced chaperone)
A1BQV8
Nicotiana attenuata
(E)VELEDPVENLGASLV(R)
(M)VAEYENC(K)
(Y)GYNAATG(K)
(V)VAAGANPLQKN(K)
(G)YLSPYFVTDNE(K)
678 ± 38320 ± 380.001

Only two proteins had consistently reduced levels in all replicated samples fromVIGS-NaSGT1 plants (Tables 1, S1; Fig. S6). Blasting peptide sequences of these protein spots revealed their homology to chaperone proteins (Tables 1, S1). One of them showed similarity to a chloroplast trigger factor-like chaperone which was identified in a proteomic analysis of Arabidopsis chloroplasts (Peltier et al., 2006); this type of chaperone is highly conserved and facilitates the folding of nascent proteins derived from ribosomes in prokaryotes and eukaryotes (Hartl & Hayer-Hartl, 2002). The other chaperone, NaCPN, was identified recently in a proteomic study of N. attenuata as a protein that is induced by M. sexta attack (Giri et al., 2006). Silencing NaCPN in N. attenuata using VIGS decreases the accumulation of defensive metabolites after herbivory and results in increased M. sexta larval mass gain (Mitra et al., 2008). In untreated plants, no difference of relative transcript levels of both chaperones were found between VIGS-EV and VIGS-NaSGT1 plants, although slight decreases were detected after W + OS treatment (Fig. S7). Therefore, the reduced protein levels of these two chaperones in VIGS-NaSGT1 plants were possibly modulated on a posttranscriptional level.

These data suggest that NaSGT1 is involved in modulating the abundance of certain defense-related proteins and the accumulation of these proteins may be regulated at transcriptional and posttranscriptional levels. Alternatively, higher SA concentrations in NaSGT1-silenced plants could cause the accumulation of these defense-related proteins.

Discussion

SGT1 is a highly conserved protein among all eukaryotes. In plants and humans, SGT1 is important for immune responses triggered by pathogen elicitors (Mayor et al., 2007; Shirasu, 2009). Here we investigated the role of SGT1 in herbivory-induced defense reactions in N. attenuata using a reverse genetic approach. In N. attenuata, NaSGT1 is important for the wounding- and herbivory-induced accumulation of JA and JA-Ile, the major hormonal regulators of plant defense against phytophagous insects. Accordingly, silencing NaSGT1 results in decreased levels of defensive metabolites and compromises N. attenuata’s resistance to M. sexta larvae. Moreover, NaSGT1 is also involved in mediating transcriptional responses to MeJA, indicating the importance of SGT1 in JA signaling.

Treating Arabidopsis with cell wall protein (CWP) fraction purified from non-pathogenic biocontrol agent Pythium oligandrum upregulates several JA-inducible genes and increases plant resistance to pathogen Pseudomonas syringae pv. tomato DC3000; these CWP-induced responses in Arabidopsis are SGT1-dependent (Kawamura et al., 2009). Wang et al. (2010) showed that SGT1 is important for the process of cell death during both compatible and incompatible plant-pathogen interactions in N. benthamiana. El Oirdi & Bouarab (2007) demonstrated that silencing SGT1 in N. benthamiana compromises the hypersensitive response induced by Botrytis cinera, a necrotrophic pathogen. The authors hypothesize that B. cinera promotes NbSGT1 expression to exploit the antagonistic effects between SA and JA (El Oirdi & Bouarab, 2007). Our phytohormone analysis indicated that SGT1 is an important regulator of basal and herbivory-induced SA and JA concentrations in plants. Whether SGT1 also mediates the homeostasis of pathogen-elicited SA and JA levels − thus knocking out (or down) SGT1 alters plant resistance to pathogens − deserves further study. At least three scenarios may account for the increased basal levels of SA in VIGS-NaSGT1 plants. Silencing NaSGT1 alters SA biosynthesis in the chloroplasts in a manner that is independent of JA-mediated suppression. Higher basal concentrations of SA were also found in NaCOI1-silenced N. attenuata, suggesting that impaired JA signaling increases SA concentrations (Kallenbach et al., 2010); thus higher SA contents in NaSGT1-silenced plants may result from reduced COI1 activity in the pathway of JA–SA interactions. Agrobacterium and tobacco rattle virus used in VIGS may also contribute to increased SA concentrations in VIGS-NaSGT1 plants.

Chemical analyses of precursors of JA revealed that NaSGT1 is associated with the concentrations of herbivory-induced free LA. More importantly, herbivory-induced OPDA accumulation is greatly compromised when NaSGT1 is silenced, although the concentrations of 18-3 OOH, the precursor of OPDA, were only slightly reduced. Therefore, it is likely that the reduced activity of enzymes that convert 18-3 OOH to OPDA accounts for most of the decreases in OPDA accumulation. Transcript levels of NaAOS (allene oxide synthase) and NaAOC (allene oxide cyclase), the genes involved in these conversion steps, do not markedly differ in VIGS-NaSGT1 and VIGS-EV plants (Fig. S5). This suggests that NaSGT1 regulates the herbivory-elicited accumulation of OPDA in a post-transcriptional manner.

Mitogen-activated protein kinases, namely SIPK and WIPK, and SA are known to regulate JA accumulation in plants (Harms et al., 1998; Spoel et al., 2003; Schweighofer et al., 2007; Seo et al., 2007; Wu et al., 2007; Diezel et al., 2009). However, activity levels of SIPK and WIPK are not altered in VIGS-NaSGT1 plants. Therefore, SGT1 is not required for the herbivory-induced activation of SIPK and WIPK and likely mediates the accumulation of JA in a pathway independent of MAPK signaling. The possibility that high SA concentrations resulted from silencing NaSGT1 accounted for the decreased JA levels in these plants can be excluded, as silencing NaSGT1 in plants expressing 35S:NahG, which had highly decreased concentrations of SA, still resulted in greatly diminished levels of herbivory-induced JA. The enzymes that convert LA to OPDA are located in chloroplasts. How NaSGT1 is involved in the regulation of JA biosynthesis in plastids requires further investigation.

In all eukaryotes studied to date, SGT1 interacts with SCF-ubiquitin ligase complexes (Kitagawa et al., 1999; Azevedo et al., 2002; Liu et al., 2002). In plants, auxin and JA-Ile are recognized by SCFTIR1 and SCFCOI1 ubiquitin ligase, respectively, resulting in proteasomal degradation of negative transcriptional regulators (Chini et al., 2007; Tan et al., 2007; Thines et al., 2007; Yan et al., 2009). Mutation in Arabidopsis AtSGT1b partially reduces plants’ sensitivity to auxin and MeJA (Gray et al., 2003; Lorenzo & Solano, 2005) and decreases the transcript levels of JA-inducible genes after being treated with cell wall protein fraction purified from P. oligandrum (Kawamura et al., 2009). Uppalapati et al. (2010) demonstrated that SGT1 is required for the induction of chlorosis and cell death elicited by the JA-Ile analogue, coronatine (produced by P. syringae pv. tomato DC3000), and for coronatine-induced Arabidopsis root growth inhibition and the expression of AtLOX2 and AtCORI1, two coronatine-inducible genes. These data suggest that SGT1 may be important for the function of multiple SCF-regulated pathways, including SCFCOI1. However, our transcriptional analyses demonstrated that silencing NaSGT1 does not generally decrease plant transcriptional responsiveness to MeJA in N. attenuata; instead, NaSGT1 appears to be involved in fine-tuning JA-induced transcript accumulation, although the mechanism is unclear. Silencing NaCOI1 in both VIGS-EV and VIGS-NaSGT1 plants greatly decreases transcript levels of all the JA-inducible genes that we examined after MeJA treatment. This implies that SCFCOI1 retains its activity in the absence of NaSGT1, and probably NaSGT1 is involved in modulating the stability of transcription factors that target JA-inducible genes. Furthermore, in N. attenuata, abolishing JA signaling by silencing COI1 leads to decreased levels of W + OS-induced JA (Paschold et al., 2008). The likely unaltered COI1 function in NaSGT1-silenced plants also suggests that the decreased W + W- and W + OS-induced JA concentrations do not result from feedback mechanisms within JA signaling circuits. Elegant studies have revealed that SCFCOI1 directly binds to JAZ protein and facilitates its degradation by ubiquitination (Chini et al., 2007; Thines et al., 2007). Directly examining the activity of SCFCOI1 by determining the stability of JAZ–GUS (glucuronidase) or JAZ–GFP (green fluorescent protein) fusion proteins in WT (or empty vector) and SGT1-deficient plants will shed light on this question. Whether SGT1 is generally positively associated with other JA-induced responses (i.e. transcriptional changes) in Arabidopsis and other plant species, how NaSGT1 is involved in modifying MeJA-induced transcriptional changes in N. attenuata and whether NaSGT1 is also required for auxin-induced responses merit future study.

In addition to SCF ubiquitin ligases, SGT1 physically interacts with chaperone proteins (e.g. HSP90 and HSP70) (Kitagawa et al., 1999; Azevedo et al., 2002; Shirasu, 2009). All these proteins are involved in regulating the activity of their targets by influencing target proteins’ stability and proper folding. Using a proteomic approach, we detected surprisingly few proteins whose abundance is altered in NaSGT1-silenced plants. Therefore, we speculate that SGT1 is mainly involved in regulating the abundance and activity of regulatory proteins, as these have relatively low concentrations. The proteomic analysis revealed that VIGS-NaSGT1 plants accumulate large amounts of PR proteins whose transcript levels are also greatly elevated. It is possible that the over-accumulated SA in VIGS-NaSGT1 plants accounts for the high abundance of these PR proteins. Interestingly, the flowers of Arabidopsis sgt1b mutants have enhanced resistance to the flower pathogen Fusarium culmorum (Cuzick et al., 2008). As many PR proteins, such as endochitinases, possess antifungal properties, it is tempting to speculate that the increased resistance might result from elevated PR protein content in flowers of Arabidopsis sgt1b mutants. Furthermore, our transcriptional analyses revealed that the abundance of the differentially accumulated abundant proteins between VIGS-EV and VIGS-NaSGT1 plants are regulated on transcriptional and post-transcriptional levels, demonstrating the complex mode of regulation of protein concentrations by NaSGT1.

In summary, we show that SGT1 is a key element in the regulatory network of plants in response to herbivory. SGT1 is required for herbivory-induced SA and JA homeostasis and normal MeJA-induced transcriptional responses, and is needed for the biosynthesis of secondary metabolites that defend plants against phytophagous insects (Fig. 6). Given the critical role of SGT1 in plant and human resistance to pathogens, and its well-conserved distribution in all eukaryotes, we propose that SGT1 plays a fundamental role in eukaryotes’ resistance to biotic stresses. In plants, whether SGT1 is also implicated in abiotic stress resistance and how SGT1 is involved in plants’ defense against herbivores are interesting questions to explore.

Figure 6.

 A model summarizing the function NaSGT1 in Nicotiana attenuata’s resistance to Manduca sexta. Herbivory by M. sexta elicits biosynthesis of jasmonic acid (JA), which is further converted to JA-isoleucine conjugate (JA-Ile). Binding of JA-Ile to the COI1 receptor further induces herbivore defense responses. NaSGT1 is required for herbivory-induced JA accumulation and modulates some but not all of jasmonate-induced COI1-dependent defense responses (indicated by the dashed arrow). Furthermore, NaSGT1 negatively regulates the concentration of salicylic acid (SA) and pathogenesis-related (PR) proteins.

Acknowledgements

We thank M. Coon, D. Sonntag, Dr. M. Schöttner, D. Yang and H. Wünsche for technical assistance, M. Kallenbach and Dr G. Bonaventure for help with free fatty acid and peroxide analysis, E. Wheeler for editorial assistance, and A. van Doorn, Dr E. Gaquerel and Dr I. Galis for valuable comments. We thank Dr A. Muck for excellent assistance on proteomic analyses and the Max Planck Society for funding.

Ancillary