Protein Profiles of Lipid Droplets during the Hypersensitive Defense Response of Arabidopsis against Pseudomonas Infection

Abstract

Lipid droplets (LDs) have classically been viewed as seed storage particles, yet they are now emerging as dynamic organelles associated with developmental and stress responses. Nevertheless, their involvement in plant immunity has still been little studied. Here, we found LD accumulation in Arabidopsis thaliana leaves that induced a hypersensitive response (HR) after Pseudomonas infection. We established a protocol to reproducibly isolate LDs and to analyze their protein content. The expression of GFP fusion proteins in Nicotiana benthamiana and in transgenic Arabidopsis lines validated the LD localization of glycerol-3-phosphate acyltransferase 4 (GPAT4) and 8 (GPAT8), required for cutin biosynthesis. Similarly, we showed LD localization of α-dioxygenase1 (α-DOX1) and caleosin3 (CLO3), involved in the synthesis of fatty acid derivatives, and that of phytoalexin-deficient 3 (PAD3), which is involved in camalexin synthesis. We found evidence suggesting the existence of different populations of LDs, with varying protein contents and distributions. GPAT4 and GPAT8 were associated with LDs inside stomata and surrounding cells of untreated leaves, yet they were mainly confined to LDs in guard cells after bacterial inoculation. By contrast, α-DOX1 and PAD3 were associated with LDs in the epidermal cells of HR-responding leaves, with PAD3 mostly restricted to cells near dead tissue, while CLO3 had a more ubiquitous distribution. As such, the nature of the proteins identified, together with the phenotypic examination of selected mutants, suggests that LDs participate in lipid changes and in the production and transport of defense components affecting the interaction of plants with invading pathogens.

Accession number: The mass spectrometry proteomics data have been deposited with the ProteomeXchange Consortium via the PRIDE (Vizcaíno et al. 2016) partner repository under the dataset identifier PXD011499 and 10.6019/PXD011499.

Introduction

Lipid bodies, also referred to as lipid droplets (LDs), are storage organelles that contain neutral lipids within a phospholipid monolayer (van der Schoot et al. 2011, Chapman et al. 2012). These particles are found ubiquitously in most plant cells, and they have primarily been considered to be involved in the storage of metabolic energy. However, more dynamic roles for LDs are now being recognized, including their participation in developmental and stress responses (Ischebeck 2016, McLachlan et al. 2016, Gao et al. 2017, Pyc et al. 2017a, Huang, 2018, Shimada et al. 2018, Shao et al. 2019, Kretzschmar et al. 2020). Also, proteins that influence the formation and size of LDs have been characterized, such as Arabidopsis CGI-58, LD-associated proteins (LDAPs), LDAP-interacting proteins (LDIPs), Seipins or PUX10 (James et al. 2010, Horn et al. 2013, Cai et al. 2015, Gidda et al. 2016, Pyc et al. 2017b, Deruyffelaere et al. 2018, Kretzschmar et al. 2018, Taurino et al. 2018). Moreover, even when proteomics analyses of LDs from senescent leaves, chloroplasts (plastoglobules), fruits, seeds, seedlings and algae have provided valuable information regarding the composition of these cellular particles (Vidi et al. 2006, Ytterberg et al. 2006, Lundquist et al. 2012, Horn et al. 2013, Davidi et al. 2015, Brocard et al. 2017, You et al. 2019, Kretzschmar et al. 2020), there are still little data about the composition and function of LDs in the response to stress.

Beyond their role as an energy source, vertebrate LDs have been found to participate in protein trafficking and to display antioxidant activity (Saka and Valdivia 2012, Bailey et al. 2015). Moreover, an emerging role in the coordination of immune responses has been attributed to LDs in mammals (Bozza et al. 2011), particularly as they contribute to the generation of eicosanoids (prostaglandins, leukotrienes and lipoxins), important lipid mediators of inflammation that are synthesized by fatty acid oxygenases [e.g. prostaglandin-endoperoxide synthases and lipoxygenases (LOXs)]. Mammalian prostaglandin-endoperoxide synthases are dual function enzymes that possess cyclooxygenase and peroxidase activities. They share sequence and functional homology with plant α-dioxygenases (α-DOXs), enzymes that catalyze the stereospecific dioxygenation of fatty acids (Sanz et al. 1998, Hamberg et al. 1999). Significantly, α-DOX activity increases when a hypersensitive response (HR) is activated in Pseudomonas-infected leaves (Ponce de León et al. 2002, Hamberg et al. 2003). Recombinant α-DOX proteins have been shown to catalyze the formation of unstable 2(R)-hydroperoxides, which in vitro can be spontaneously converted into a mixture of a one-carbon-shortened fatty aldehyde (83%), a 2-hydroxy acid (15%) and a one-carbon-shortened fatty acid (2%) (Hamberg et al. 1999). However, in plants infected by bacteria, the α-DOX1 generated fatty acid 2-hydroperoxylinolenic acid (2-HPOT) is mainly reduced by plant peroxygenases producing 2-hydroxylinolenic acid (2-HOT, 90–95%: Hamberg et al. 2003). Later, it was shown that the CLO3 peroxygenase catalyzes the reduction in 2-HPOT into 2-HOT in Arabidopsis (Shimada et al. 2014). Interestingly, this study also showed that α-DOX1 co-localizes with CLO3 in LDs of Colletotricum higginsianum-infected leaves. These characteristics are reminiscent of the LD activities reported in vertebrates, suggesting that plant LDs may also participate in immunity and influence plant–pathogen interactions.

Biochemical and functional analyses have shown the enzymatic interaction of α-DOX1 and CLO3 with LOXs during the biosynthesis of plant oxylipins (Hamberg et al. 2003, Vellosillo et al. 2007, Vicente et al. 2012, Blée et al. 2014). Arabidopsis LOX1 and LOX2 have 9-LOX and 13-LOX activities, respectively (Bannenberg et al. 2009a), and they are both present in a pool of proteins isolated by immunoprecipitation of CLO3:GFP in senescent transgenic Arabidopsis leaves (Shimada et al. 2014). Moreover, lipid bodies containing LOX have been detected in cucumber seed LDs (Hause et al. 2000). Accordingly, LOX proteins may also be components of LDs.

Here, we have optimized an LD extraction protocol to study the LDs in leaves undergoing a hypersensitive response (HR) after infection with the avirulent bacterium Pseudomonas syringae pv. tomato (Pst) DC3000 avrRpm1, a process during which α-DOX1 is strongly activated (Ponce de León et al. 2002, Hamberg et al. 2003). As a positive control, we examined the formation of LDs in senescent leaves, known to possess high levels of α-DOX1 and to produce LDs containing α-DOX (Obregón et al. 2001, Bannenberg et al. 2009b, Shimada et al. 2014, Brocard et al. 2017). We performed a proteomics analysis of isolated LDs and assessed the localization of selected proteins in bacterially infected leaves. These studies showed that LDs formed in different cell types may vary in their protein composition, allowing us to consider the role of LD proteins in plant defense and the functional significance of these organelles.

Results

Efficient isolation of Arabidopsis LDs from senescent and Pst DC3000 avrRpm1-infected leaves

Confocal microscopy of Arabidopsis leaves stained with BODIPY (a neutral lipid-specific fluorescent dye) identified LDs that formed during aging and after Pst DC3000 avrRpm1 infection. While LDs were occasionally seen in leaves of 4-week-old plants, the number of LDs increased in leaves of 5- and 6-week-old plants (∼2.5- and ∼1.8-fold, respectively, above values in leaves of 4-week-old plants) (Fig. 1A, C). Moreover, LDs rose in 4-week-old plants 24 h after Pst DC3000 avrRpm1 inoculation, increasing to higher levels at 48 and 72 h after bacterial infection (∼2-fold above values in noninfected leaves; Fig. 1B, D). We did not detect a significant variation in the diameter of LDs in the samples examined (Fig. 1E, F).

Fig. 1

Formation of LDs in wild-type Arabidopsis plant leaves. (A) Fluorescence visualization of LDs in BODIPY-stained leaves of plants grown for 4, 5 and 6 weeks; scale bar = 20 μm. (B) Fluorescence visualization of LDs in BODIPY-stained leaves of 4-week-old plants after Pst DC3000 avrRpm1 infiltration (10⁶ cfu/ml). The time after bacterial inoculation is indicated (hpi), and representative images from the results obtained in independent experiments are shown; scale bar = 20 μm. (C) Number of LDs in BODIPY-stained leaves of plants grown for 4, 5 and 6 weeks. (D) Number of LDs in BODIPY-stained leaves of 4-week-old plants at different times after Pst DC3000 avrRpm1 infiltration (10⁶ cfu/ml). Each bar represents the mean of LD values (± SEM) in five biological replicates estimated as described in the section Material and Methods. Asterisks indicate significant differences between LD values in untreated 4-week-old plants and the remaining samples examined: **P < 0.01; *P < 0.05, Student’s t-test. (E) Diameter of LDs in BODIPY-stained leaves of plants grown for 4, 5 and 6 weeks. (F) Diameter of LDs in BODIPY-stained leaves of 4-week-old plants at different times after Pst DC3000 avrRpm1 infiltration (10⁶ cfu/ml). No significant differences were found with control 4-week-old plants (Wilcoxon–Mann–Whitney nonparametric test P < 0.05).

We used senescent leaves of 6-week-old plants to establish a protocol allowing isolation of highly LD-enriched fractions from limited amounts of tissue. The yield and purity of LDs in these fractions were monitored by BODIPY staining and flow cytometry analysis, testing the use of different compounds for LD purification (Tzen et al. 1997, Lin et al. 2012). To our purposes, it was critical to use freshly collected tissue and to add Tween-20 to the extraction buffer (see section Materials and Methods for a detailed description). This optimized protocol was used to isolate LDs from senescent leaves (6-week-old plants) and from the leaves of young plants (4 weeks) in response to Pst DC3000 avrRpm1 infection, both 24 and 72 h postinoculation. LD fractions formed a milky compact layer containing 1–1.5-μm diameter round particles that stained strongly with BODIPY (Fig. 2A). Flow cytometry analyses of unstained and BODIPY-stained LD aliquots showed that the size and complexity of the particles in the samples examined were similar. Thus, all samples consisted of neutral lipid-containing organelles that were mostly free of other cellular structures or organelles (Fig. 2B, C). LDs were examined and quantified in three biologically independent replicates from infected and senescent leaves to estimate the recovery. Fewer LDs were purified from samples prepared 24 h after bacterial infection, whereas 2.7- and 3.8-fold more LDs were isolated from samples obtained 72 h postinoculation and from senescent leaves, respectively (Fig. 2D). The estimated LD recovery based on the volume and flow cytometry assay conditions was ∼450 and ∼1,200 LDs/mg tissue from Pseudomonas-infected leaves 24 and 72 h postinoculation, respectively, and ∼1,700 LDs/mg from senescent leaves (Fig. 2E).

Fig. 2

Analysis of purified lipid droplet fractions. (A) Fluorescence visualization of the LD fractions purified from senescent or Pst DC3000 avrRpm1-infected leaves: scale bar = 20 μm. Higher magnification of an LD from senescent leaves; scale bar = 2 μm. (B) Representation of the two-dimensional plots of LD samples showing the size of the events detected vs. the complexity of the samples. A representative experiment of three performed is shown. (C) Flow cytometry of unstained (grey) and BODIPY-stained (green) LD samples, a representative experiment of three performed is shown. (D) Representation of the relative LD numbers in senescent and Pst DC3000 avrRpm1-infected leaves. (E) Estimated number of LDs recovered from the tissues examined. (F) Comparison of LD proteomes defined in senescent and Pst DC3000 avrRpm1-infected leaves 24 and 72 h postinoculation (hpi). Shown are Venn diagrams of proteins in each LD proteome and percentages of common and different proteins with respect to total numbers of compared proteome.

Identification of LD proteins

We used shotgun proteomics to analyze the protein composition of LDs from bacteria-infected leaves. Equal amounts of proteins from LD fractions of 4-week-old plants infected for 24 and 72 h with Pst DC3000 avrRpm1 (three independent samples) were trypsin-digested and analyzed by non-biased mass spectrometry (MS) analyses. LD fractions from 6-week-old senescent Arabidopsis leaves were also examined and used as a control in these experiments. The MS and MS/MS data obtained were used to identify the proteins in the Arabidopsis thaliana section of UniProtKB, considering peptides with individual Mascot ion scores 30 and an average peptide count of at least two in the three replicates examined (Supplementary Table S1). Accordingly, 70 and 81 proteins were identified in LD fractions from 24 and 72 h Pseudomonas-infected leaves, respectively, whereas 72 proteins were found in LD samples from senescent leaves; the LD proteomes of infected leaves were highly coincident with the senescence proteome (49% at 24 h and 43% at 72 h; Fig. 2F). Gene Ontology analyses of the three proteomes identified showed enrichment in terms associated with transport (3- to 4-fold enrichment) and of those that respond to biotic stress and abiotic stress (2.6- to 2.8-fold enrichment), relative to the entire Arabidopsis proteome (Supplementary Table S2). By contrast, terms related to development and regulatory processes were underrepresented (e.g. DNA and RNA metabolism, signal transduction and transcription; Supplementary Table S2).

The proteomes defined contain most of the proteins previously demonstrated to be LD components in non-seed tissues of dicotyledonous plants such as CLO3, LDAP1, PUX10, α-DOX1, LDAP3 and ULP (unknown lipid droplet protein)/LDIP (Horn et al. 2013, Shimada et al. 2014, Gidda et al. 2016, Brocard et al. 2017, Pyc et al. 2017b, Kretzschmar et al. 2018). Likewise, we identified proteins that had not been previously characterized as plant LD proteins but that have been detected in LDs of other organisms. This is the case of GPAT4 (glycerol-3-phosphate acyltransferase 4), which is found in Drosophila LDs during oleate loading (Wilfling et al. 2013), FLOT1 (Flotilin1) that has been detected in the LDs of mammalian cells (Browman et al. 2007) and SYT122 (Syntaxin122), the equivalent of the Syntaxin-5 detected in animal cell LDs (Boström et al. 2007). In line with this, we found numerous RAB proteins (RAB1, RAB2C, RAB18, RABG3G and RAB21A) that could parallel the Rab proteins shown to be associated with mammalian LDs (Li and Yu, 2016). Furthermore, we found chloroplast proteins such as Rubisco activase, chlorophyll a/b-binding protein or LHCB proteins from the light-harvesting complex (Supplementary Table S1) that could be bona fide LD components as found in Parachlorella kessleri (You et al. 2019) or carryover contaminants.

In addition to these, the proteomes defined contain proteins that had not been associated with LD; their validation as LD components also requires further experimental evidence. Of interest to our studies are important components of plant defense responses, such as CYP71A12, CYP71A13 and phytoalexin-deficient 3 (PAD3), enzymes involved in the synthesis of the antimicrobial compound camalexin (Nafisi et al. 2007). Moreover, we found GPAT4 and GPTA8, enzymes required for the biosynthesis of cutin (Li et al. 2007), or membrane proteins like PEN3 (Penetration 3), HIR1 (Hypersensitive Induced Reaction 1), HIR2 or HIR4, which are recruited to sites of pathogen attack (Qi and Katagiri 2012, Underwood et al. 2017).

Analysis of protein localization in LDs of Nicotiana benthamiana leaves

Because formation of LDs in response to biotic stress has been little investigated (Coca and San Segundo 2010, Shimada et al. 2014), we generated GFP fusions of selected proteins (previously shown to participate in different defense-related responses) and evaluated their localization in control Nicotiana benthamiana leaves and in leaves undergoing an HR due to co-infiltration with the P. syringae pv. syringae B728a (Pss B728a) strain. The proteins GPAT4, GPAT8 and PAD3 associated with plant LDs in this study were examined. In these experiments, we also tested the LD localization of α-DOX1 and CLO3 previously shown to co-localize in LDs of C. higginsianum-infected Arabidopsis leaves (Shimada et al. 2014). Moreover, although it was not detected in our proteomes, we evaluated the localization of LOX1, which is known to enzymatically interact with α-DOX and CLO3 during the oxygenation of fatty acids (Hamberg et al. 2003, Blée et al. 2014) and to coprecipitate with CLO3 in senescent leaves (Shimada et al. 2014).

In N. benthamiana not inoculated with Pss B728a, GFP fusion proteins adopted a reticulated pattern of fluorescence consistent with an endoplasmic reticulum (ER) or cytoplasmic localization (Fig. 3A). Also, GFP-fluorescent particles were seen in leaves expressing α-DOX1:GFP or CLO3:GFP. Co-inoculation of the GFP constructs with low doses (10⁴ cfu/ml) of Pss B728a was generally associated with weaker GFP fluorescence and the visualization of GFP-tagged organelles consistent with LDs in leaves expressing α-DOX1:GFP, CLO3:GFP, GPAT4:GFP, GPAT8:GFP and PAD3:GFP (Fig. 3A). By contrast, such structures were not evident in LOX1:GFP expressing leaves after Pss B728a co-infiltration (Fig. 3A).

Fig. 3

Localization of GFP fusion protein in Nicotiana benthamiana leaves. (A) Images of CLO3:GFP, α-DOX1:GFP, GPAT4:GFP, GPAT8:GFP, PAD3:GFP and LOX1:GFP fusion proteins (green) localization in the leaves of N. benthamiana infiltrated with Agrobacterium containing constructs (left panels) and co-infiltrated with the HR-inducing bacteria Pss B728a (right panels); scale bar = 20 μm. Images of LDs at higher magnifications are shown; scale bar = 5 μm. (B) Images of CLO3:GFP, α-DOX1:GFP, GPAT4:GFP and GPAT8:GFP fusion proteins (green) in leaves of N. benthamiana co-infiltrated with Agrobacterium containing the specific constructs and the HR-inducing bacteria Pss B728a. LDs were visualized by Nile Red staining, and the GFP and Nile Red images were merged to show co-localization of the GFP fusion proteins and LDs. Shown are representative images of the samples examined; scale bar = 7.5 μm. Images of LDs at higher magnifications are shown; scale bar = 2 μm. (C) Images of PAD3:GFP fusion protein localization (green) and Nile Red-stained LDs (red) in leaves of N. benthamiana co-infiltrated with Agrobacterium containing the specific construct and the HR-inducing bacteria Pss B728a. Merged images show variable localization; week overlap (upper panels) or extensive co-localization (lower panels) near dead cells; scale bar = 10 μm.

As an evaluation of LD localization, we stained Pss B728a-infected N. benthamiana leaves with the lipid-specific fluorescent dye Nile Red and assessed the co-localization of green and red fluorescence. Confocal microscopy showed that CLO3:GFP, α-DOX1:GFP, GPAT4:GFP or GPAT8:GFP fusion proteins were localized in red-stained LDs (Fig. 3B). The distribution of PAD3:GFP did not fully overlap with LDs; PAD3:GFP was irregularly distributed along the plasma membrane (PM) and occasionally co-localized with red LDs within the inoculated areas (Fig. 3C).

CLO3 is a well-characterized LD marker in vegetative tissues (Shimada et al. 2014, Brocard et al. 2017) and the most abundant protein in our proteomes. Therefore, we generated a CLO3:RFP fusion protein to test its co-localization with the GFP fusion proteins in Pss B728a-infected leaves. The simultaneous expression of CLO3:RFP and CLO3:GFP fusion proteins in Pss B728a-infected N. benthamiana leaves yielded red fluorescence that co-localized with CLO3:GFP in structures consistent with LDs (Supplementary Fig. S1). Similarly, CLO3:RFP co-localized with α-DOX1:GFP, GPAT4:GFP or GPAT8:GFP fusion proteins when they were co-expressed in Pss B728a-infected leaves. PAD3:GFP and CLO3:RFP proteins did not fully overlap; however, both proteins co-localized in LDs formed in the vicinity of dead cells and vessels within the inoculated areas (Supplementary Fig. S1).

Differential distribution of CLO3, α-DOX1, GPAT4, GPAT8 and PAD3 in Arabidopsis

The expression of GFP fusion proteins in N. bethamiana supported the localization of CLO3, α-DOX1, GPAT4, GPAT8 and PAD3 in LDs of tissues undergoing an HR; however, the transient nature of this type of analysis limits the examination of LD dynamics. Therefore, we generated stable transgenic Arabidopsis lines overexpressing the abovementioned GFP fusion proteins and analyzed their localization in untreated and MgCl₂-inoculated leaves (used as an inoculation control), and in leaves responding to the HR-inducing strain Pst DC3000 avrRpm1. These analyses confirmed the presence of CLO3:GFP, α-DOX1:GFP, GPAT4:GFP, GPAT8:GFP and PAD3:GFP in GFP-tagged structures consistent with LDs. Moreover, inspection of transgenic lines provided additional data regarding LDs’ formation and protein localization (Fig. 4). In CLO3:GFP or α-DOX1:GFP overexpressing plants, we found green fluorescent particles in epidermal and stomata cells of untreated leaves, the appearance of which varied following Pst DC3000 avrRpm1 inoculation (Fig. 4A). In the case of α-DOX1:GFP, we found round particles that were preferentially associated with epidermal cells, whereas CLO3:GFP fluorescence organelles of irregular size were seen inside both stomata and epidermal cells; the heterogeneous appearance of the CLO3:GFP containing particles is likely a reflection of their aggregation. Examination of untreated leaves in GPAT4:GFP and GPAT8:GFP expressing plants showed GFP-tagged organelles within the guard cells and in cells near the stomata. In these plants, bacterial infection caused a visible reduction in the amount of GFP-fluorescent particles, which were preferentially confined to stomata cells (Fig. 4A). As found in transient expression analysis, PAD3:GFP fluorescence in transgenic Arabidopsis plants was distributed in patches along the PM and occasionally, as round-green particles in bacteria-infected leaves (Fig. 4A). Moreover, LOX1 did not appear to be associated with LDs in LOX1:GFP Arabidopsis transgenic lines (Fig. 4A). No variations in GFP fusion protein localization were observed in MgCl₂-inoculated leaves compared to untreated plants (Supplementary Fig. S2), thus indicating that the changes in the distribution of GFP fusion proteins described above were activated in response to bacterial infection.

Fig. 4

Localization of GFP fusion proteins in transgenic Arabidopsis lines. (A) Images of the CLO3:GFP, α-DOX1:GFP, LOX1:GFP, GPAT4:GFP, GPAT8:GFP and PAD3:GFP fusion proteins (green) in the transgenic Arabidopsis lines. Images from untreated leaves (control) or leaves 72 h after Pst DC3000 avrRpm1 inoculation are shown. The GFP images are representative examples of the samples examined; scale bar = 20 μm. (B) Representative images of CLO3:GFP, α-DOX1:GFP, LOX1:GFP, GPAT4:GFP, GPAT8:GFP and PAD3:GFP in leaves of transgenic Arabidopsis lines 3 d after Pst DC3000 avrRpm1 inoculation. LDs were visualized by Nile Red staining, and the GFP and Nile Red images were merged to show the co-localization; scale bar = 10 μm.

Co-localization of GFP fusion proteins and LDs in Arabidopsis plants

As a validation of LD localization, we stained transgenic Arabidopsis lines with the lipid-specific fluorescent dye Nile Red and assessed the co-localization of green and red fluorescence. In noninfected leaves, CLO3:GFP, α-DOX1:GFP, GPAT4:GFP and GPAT8:GFP fusion proteins formed particles that partially co-localized with red-stained LDs, whereas PAD3:GFP or LOX1:GFP did not co-localize with LDs in these conditions (Supplementary Fig. S3). However, the co-localization of CLO3:GFP, α-DOX1:GFP, GPAT4:GFP, GPAT8:GFP and PAD3:GFP with LDs increased after bacterial infection where a preferential LD localization of these GFP fusion proteins was apparent (Fig. 4B). Again, GPAT4:GFP and GPAT8:GFP were mainly found in stomata LDs, whereas α-DOX1:GFP and PAD3:GFP co-localized with LDs in epidermal cells and CLO3:GFP was detected in both stomata and epidermal LDs. As expected, LOX1:GFP fluorescence did not co-localize with red-stained LDs in infected plants either (Fig. 4B).

Nile Red–GFP co-localization was also tested in LDs after their isolation from wild-type and transgenic plants (Supplementary Fig. S4). Green fluorescence co-localized with red fluorescent LDs prepared from bacteria and MgCl₂-treated leaves of CLO3:GFP, α-DOX1:GFP, GPAT4:GFP and GPAT8:GFP plants, whereas no GFP fluorescent particles were found in LDs from wild-type or LOX1:GFP plants. In the case of PAD3:GFP, green fluorescence co-localized with red LDs when prepared from leaves infected by bacteria but not from control plants (Supplementary Fig. S4). We noted that there was a higher proportion of red- and GFP-fluorescence co-localization in LD samples from both CLO3:GFP and α-DOX1:GFP bacterial treated plants. Hence, these appear to be abundant LD proteins that are present in a large proportion of these organelles.

Analyses of plant susceptibility to pathogen infection

LD-protein association and the observed changes in protein accumulation during the activation of plant defense led us to examine the responses of gpat4/gpat8 (Li et al. 2007), dox1/dox2 (Bannenberg et al. 2009b) and pad3-1 (Nafisi et al. 2007) mutants to pathogen infection. Due to the unavailability of CLO3 loss-of-function mutants on the Col-0 background, the function of CLO3 was not directly investigated here. Previous studies with α-DOX1 indicated that it participates in the establishment of the HR after Pseudomonas infection (Ponce de León et al. 2002, Hamberg et al. 2003, Hanano et al. 2015, Hong et al. 2017) and in limiting the growth of C. higginsianum (Shimada et al. 2014). In addition, it was reported that the resistance of pad3 plants to Pseudomonas infection is not compromised, yet both pad3-1 and gpat4/gpat8 mutant plants are more susceptible to the fungus Alternaria brasiccicola (Li et al. 2007, Nafisi et al. 2007). Here, we extended these analyses by testing the response to necrotrophic fungi like Plectospharella cucumerina and Botrytis cinerea. In addition, we examined the apoplastic and stomatal defense against P. syringae by infiltration or surface inoculation of different bacterial strains.

We found that P. cucumerina lesions were more intense in gpat4/gpat8 mutants than in wild-type plants and that pad3-1 showed a similar trend (P = 0.066). By contrast, the lesions were significantly smaller in dox1/dox2 plants, reflecting their enhanced resistance to P. cucumerina (Fig. 5A;Supplementary Fig. S5). In response to B. cinerea infection, dox1/dox2 plants behaved like wild-type controls (Fig. 5B), whereas gpat4/gpat8 plants were highly resistant to this pathogen. In line with previous studies, pad3-1 mutants showed enhanced susceptibility to B. cinerea (Glazebrook 2005). After apoplastic or surface inoculation of Pst DC3000 (virulent) or Pst DC3000 AK87 (coronatine-deficient), no differences in bacterial growth were observed between dox1/dox2, pad3-1 and wild-type plants, whereas gpat4/gpat8 mutants were more susceptible to surface inoculation with these bacterial strains (Fig. 5D–F; Supplementary Fig. S5). By contrast, after infiltration, Pst DC3000 avrRpm1 (avirulent) growth was mildly, yet significantly enhanced in dox1/dox2 plants relative to the wild-type plants and a similar trend was evident in pad3-1 and gpat4/gpat8 mutants, although the differences were not significant (Fig. 5).

Fig. 5

Response of wild-type, dox1/dox2, pad3-1 and gpat4/gpat8 plants to pathogen infection. (A) Response of plants to Plectospharella cucumerina infection. Each bar represents the mean P. cucumerina lesion diameter in four leaves of 25 plants/genotype (± SEM). (B) Response of plants to Botrytis cinerea infection; each bar represents the mean of the B. cinerea lesion diameter in four leaves of 25 plants/genotype (± SEM). (C) Response of plants to syringe inoculation of the avirulent Pst DC3000 avrRpm1. (D) Response of plants to syringe inoculation of the virulent strain Pst DC3000. (E) Response of plants to spray inoculation of Pst DC3000 onto the leaf surface. (F) Response of plants to spray inoculation of Pst DC3000 AK87 onto the leaf surface. For bacterial infections (C–F), each bar represents the bacterial growth (± SEM) in plants 72 h after inoculation. In all cases, asterisks indicate significant differences between the wild type and mutants: ***P < 0.001; *P < 0.05, Student’s t-test.

Discussion

It has recently become clear that LDs are not restricted to plant seeds but rather that they also accumulate in vegetative tissues, as well as in response to biotic and abiotic stress (Coca and San Segundo 2010, Shimada et al. 2014, Gidda et al. 2016, Ischebeck 2016, McLachlan et al. 2016, Brocard et al. 2017, Gao et al. 2017). Nonetheless, the role of LDs beyond that of lipid storage organelles and their contribution to the plant’s responses to stress have been little studied.

We previously showed that the biosynthesis of oxylipins via the α-DOX pathway is induced when an HR is provoked by infection with Pst DC3000 avrRpm1 (Ponce de León et al. 2002), as well as in senescent leaves (Obregón et al. 2001, Bannenberg et al. 2009b). The association of the α-DOX enzyme with LDs (Shimada et al. 2014, Brocard et al. 2017) suggests that these lipid-containing organelles could be produced after Pst DC3000 avrRpm1 infection. Consistent with this, we showed the accumulation of LDs after Pst DC3000 avrRpm1 inoculation, reaching high levels 72 h later. In agreement with previous studies (Shimada et al. 2014, Brocard et al. 2017), we also found a developmental increase in LDs that can be associated to senescence (6-week-old plants). The reproducible formation of LDs, together with the optimized isolation method used here, allowed us to obtain highly enriched fractions and high yields of these organelles.

The accumulation of LDs in Pst DC3000 avrRpm1-infected and aging leaves supports a role beyond that of passive lipid storage particles (van der Schoot et al. 2011, Chapman et al. 2012, Brocard et al. 2017, Pyc et al. 2017a, Huang, 2018, Shimada et al. 2018, Kretzschmar et al. 2020). Accordingly, the protein content reported for seed LDs (Jolivet et al. 2004, Vermachova et al. 2011, Kretzschmar et al. 2020), structures that serve as an energy source for germination, is different from that found here for LDs from leaves undergoing HR or aging. In addition, we noted that oleosins playing a structural role in seed LDs were absent in the LD proteomes characterized here. The absence of oleosins in LDs of vegetative tissues was previously described in Arabidopsis, and it was suggested that, in these tissues, caleosins and LDAP contribute to maintaining LD structure (Horn et al. 2013, Gidda et al. 2016, Brocard et al. 2017, Shao et al. 2019). Conversely to seed LDs, the ∼45% agreement found here between proteins from LDs of HR and aging leaves is indicative of common activities in these LDs.

Our LD proteomes contain the majority of proteins previously proved to be in non-seed LDs from plants, such as CLO3, α-DOX1, LDAP, ULP/LDIP and PUX10. In addition, GPAT4, FLOT1 and SYT122 homologs were previously identified in LDs from mammals (Boström et al. 2007, Wilfling et al. 2013) highlighting the value of our data to identify plant LD proteins and suggesting that plant LDs share a functional characteristic with those of other organisms. Consistent with the role of mammal LDs in protein trafficking (Saka and Valdivia 2012), the presence in our LD proteomes of PM proteins, which are important components of the plant defense response (e.g. PEN3, SYT, HIR1, HIR2, HIR4), suggests that LDs are involved in trafficking to PM during plant immune responses. In addition, LDs could participate in the delivery of proteins involved in lipid exchange or modification. Human SYT2 acts in lipid transfer between the ER and the PM (Schauder et al. 2014), and Arabidopsis SYT1 aggregates at ER/PM contact sites after mechanical damage (Pérez-Sancho et al. 2015). This raises the exciting possibility that LD proteins may modify the protein and lipid contents of the PM to drive plant immunity, although further studies will be needed to evaluate this possibility.

The involvement of LDs in plant immunity was previously indicated by their formation in response to fungal elicitors or infection by C. higginsianum, with LD-associated proteins contributing to fungal resistance (Coca and San Segundo 2010, Shimada et al. 2014). The increase in LDs when tissues respond to Pst DC3000 avrRPM1 further suggests the participation of these organelles in plant defense. These LDs contain proteins like CYP71A12, CYP71A13 and PAD3, all of which are involved in the biosynthesis of the antimicrobial camalexin (Nafisi et al. 2007), a nonpolar compound that could accumulate or be transported in the lipid core of LDs. Further support to a role in plant defense was the finding of CLO3 and α-DOX1 that act sequentially in the production of oxylipins contributing to plant immunity (Hamberg et al. 2003, Shimada et al. 2014, Hanano et al. 2015). Even though it has been reported in mammals (Bozza et al. 2011), we did not detect LOX proteins in the LDs characterized, which could reflect a lesser contribution of plant LDs to oxylipin production compared with vertebrates.

In addition to well-characterized defense proteins, our proteomes contain GPAT4 and GPTA8, which are required for the biosynthesis of cutin (Li et al. 2007). GPAT-bearing LDs could thus contribute to the modification of this lipid polymer participating in the protection against pathogen attack. In line with this idea are the findings showing LD localization of OsGLIP1 and OsGLIP2 lipases that modulate lipid homeostasis and immunity in rice (Gao et al. 2017). Moreover, the alteration in wax components in CLO3-overexpressing Arabidopsis leaves (Hanano et al. 2015) and the reduced cuticle thickness in α-DOX silencing pepper (Hong et al. 2017) are consistent with the involvement of LDs in the modification of epidermal lipid polymers.

Presently, we cannot exclude that some of the proteins in our proteomes might not be bona fide LD components but contaminants co-purifying with LDs during isolation. This could be the case of chloroplast components such as Rubisco activase, chlorophyll a/b-binding protein or LHCB proteins from the light-harvesting complex (Supplementary Table S1). We note, however, that homolog proteins were found in LDs of salt-stressed P. kessleri cells, a response in which chloroplast degradation materials might be incorporated into LDs to help in membrane reconstruction (You et al. 2019). Whether the chloroplast proteins found in our proteomes are present in LDs of stressed vegetative plant tissues should be further investigated.

Nonetheless, we show that PAD3, CLO3, α-DOX1, GPAT4 or GPAT8 GFP fusions are present in LDs of N. benthamiana or Arabidopsis leaves that induced an HR after Pseudomonas infection, which validates their LD localization and supports the involvement of LDs in plant immunity. This is also sustained by the defense phenotypes of mutants in the LD proteins characterized, although their responses vary considerably depending on the pathogen. In agreement with previous results, the dox1/dox2 mutant is more susceptible to Pst DC3000 avrRpm1 but not to Pst DC3000, which is consistent with the role of α-DOX1 in protecting plant tissues from HR-related cell death (Ponce de León et al. 2002, Hamberg et al. 2003, Hanano et al. 2015, Hong et al. 2017). The same mutant is, however, more resistant to P. cucumerina and shows a wild-type response to B. cinerea, which argues against any dependence on α-DOX1 for the production of antifungal compounds (Shimada et al. 2014). Likewise, pad3-1 is highly susceptible to B. cinerea, which is consistent with the participation of PAD3 in the biosynthesis of the antifungal camalexin (Glazebrook 2005) but responds like wild-type plants to Pseudomonas infection. Finally, we found that gpat4/gpat8 mutants are highly susceptible to the necrotroph P. cucumerina, which penetrates the leaf surface during infection (Ramos et al. 2013), but resistant to B. cinerea, a common feature of other cutin-deficient mutants (Bessire et al. 2007, Tang et al. 2007, Serrano et al. 2014). Moreover, gpat4/gpat8 plants show enhanced susceptibility after surface inoculation of Pst DC3000 and coronatine-deficient Pst DC3000 AK87, but not when bacteria are inoculated into the apoplast, thus indicating a failure to close stomata. It should be noted that, although the dominant LD localization of these proteins during infection suggests that the defense phenotypes of corresponding mutants are due to LD functions, we cannot discard that they might also be caused by protein activities independent of LDs.

The different responses of mutants to pathogen infection imply a differential role of LD proteins in plant immunity, which could also be related to the existence of different LD populations. Our studies with GFP fusion proteins endorsed this idea and helped us to further understand the dynamics of these organelles. GPAT4, GPAT8 or PAD3 GFP fusion proteins accumulate in LDs of N. benthamiana leaves undergoing HR, whereas α-DOX1 and the LD marker CLO3 associate with LD-like particles in both basal and HR-responding conditions (Fig. 3). Moreover, Pseudomonas infection improved LD localization of these proteins in Arabidopsis but often in different cell types (Fig. 4). For instance, in untreated leaves, GPAT4 and GPAT8 are found in LDs within the stomata and in adjacent cells, yet they become restricted to stomata LDs after bacterial inoculation. This correlates well with the enhanced susceptibility of gpat4/gpat8 to surface inoculation of Pseudomonas and is in line with the presence of specific LD population in stomata (McLachlan et al. 2016). The restriction of GPAT4 and GPAT8 to stomata LDs is consistent with the reduction in GPAT4 and GPAT8 expression found during HR activation (Supplementary Fig. S6). These changes in GPTA4 and GPTA8 levels might cause specific cutin modifications affecting the response of plants to pathogen infection, a possibility that deserves further investigation. A different situation was found for α-DOX1 and PAD3, which are mainly associated with LDs in epidermal cells of HR-responding leaves (in the case of PAD3, this is especially evident in cells close to HR dead tissues) (Figs. 3, 4). This suggests that LD populations near infection zones participate in the control of pathogen invasion as previously proposed (Shimada et al. 2014). The enhanced visualization of α-DOX1- and PAD3-containing LDs in HR-responding leaves is consistent with the increased expression of α-DOX1 and PAD3 during Pst DC3000 avrRpm1 infection (Supplementary Fig. S6), thus supporting the involvement of these LDs in plant defense. The existence of common and population-specific LD proteins also agrees with the fact that most LDs purified from CLO3:GFP plants display GFP fluorescence, which is in accordance to its structural role (Lin et al. 2012), whereas, from GFP-tagged α-DOX1, GPAT4, GPAT8 or PAD3 plants, fluorescence was only visible in a variable proportion of the isolated LDs (Supplementary Fig. S4). In addition to its structural role, the induced expression of CLO3 after bacterial infection (Supplementary Fig. S6) suggests the defensive action of CLO3 during Pseudomonas infection.

Irrespective of the function fulfilled by their specific proteins, our studies support a role for LDs beyond that of passive lipid storage particles. We showed that LDs may vary in protein composition and cellular localization, which reveals dynamic organelles that could be involved in protein trafficking, lipid homeostasis and production of defense compounds, potentially playing different roles depending on the cells and tissues in which they form. These variable characteristics would influence the participation of LDs in immune processes, thereby helping to define the outcome of the plant–pathogen interactions.

Materials and Methods

Plants and growth conditions

Wild-type A. thaliana and mutants dox1/dox2, gpat4/gpat8 and pad3-1 were derived from the Arabidopsis Columbia (Col-0) ecotype. dox1/dox2 is a double knockout mutant derived from a cross of the T-DNA insertion lines SALK_005633 (for α-DOX1) and SALK_029547 (for α-DOX2) (Bannenberg et al. 2009b), gpat4/gpat8 is a double knockout mutant derived from a cross of the T-DNA insertion lines SALK_106893 (for GPAT4) and SALK_095122 (for GPAT8) (Li et al. 2007) and pad3-1 is an EMS mutant (Glazebrook and Ausubel, 1994) used as a reference plant to test camalexin function. For in planta analyses, seeds were sown on soil, stratified (3 d, 4°C) and grown in chambers under standard conditions: 22°C, 70% relative humidity, with a 14-h light/10-h dark photoperiod and 250 μE/m²/s fluorescent illumination.

LD isolation

LDs were isolated from the leaves of untreated 6-week-old plants and of 4-week-old plants 24 and 72 h after Pst DC3000 avrRpm1 (10⁶ cfu/ml) inoculation. Different compounds and isolation procedures for LD purification were tested (Tzen et al. 1997, Lin et al. 2012), and the experimental conditions improving the LD yield were selected. In brief, freshly collected leaves (2.5 g) were cut into small pieces and ground using a pestle and mortar containing sea sand (Merck; Merck & Co., Inc, Kenilworth, NJ, USA) as an abrasive agent, along with 5 ml of cold 0.6 M sucrose extraction buffer: 10 mM sodium phosphate (pH 7.5), 150 mM NaCl, 0.1% (v/v) Tween 20, 1 mM PMSF, proteinase inhibitor cocktail (Sigma, St Louis, MO, USA). The homogenate was filtered through three layers of Miracloth (Calbiochem, San Diego, CA, USA) and placed in a centrifuge tube, covered with an equal volume of 0.4 M sucrose extraction buffer and ultracentrifuged for 90 min at 100,000 × g in a swinging-bucket rotor (Beckman Coulter Optima L-100 XP Ultracentrifuge, SW55 Ti Rotor). The LDs in the top layer were collected and placed in a centrifuge tube, and a volume of extraction buffer containing 0.2 M sucrose was layered onto the sample. The tubes were again ultracentrifuged at 100,000 × g for 90 min, and the LDs in the top layer were collected and then concentrated by centrifugation for 60 min in Eppendorf tubes at 10,000 × g. The lower fraction without LDs was removed with a Pasteur pipette, obtaining the purified LDs in a 100-μl final volume. For the estimation of recovery and purity of extracted LDs, aliquots were stained with BODIPY 493/503 (0.5 μg/ml) for 30 min, washed with water for 5 min and used for microscopy or flow cytometry examination (as described below).

Flow cytometry analysis

For the flow cytometry analysis, purified LD fractions were diluted in 10 mM sodium phosphate (pH 7.5) and analyzed on a Gallios cytometer (Beckman Coulter, CA, USA) using Kaluza software (Beckman Coulter). Unstained and BODIPY-stained aliquots from three independent LD preparations were examined in each case. The samples were excited with a 488-nm laser, and the BODIPY fluorescence emission was captured with a 525/40-nm band-pass filter. Latex Beads Microspheres (0.5 μm; Beckman L500, ref. 6602788) and Flow-Check Pro Fluorospheres (3, 6 and 10 μm size; Beckman A63493) were used to examine the size of the cellular particles in the samples examined. Measurements were taken over 90 s to determine the number of lipid bodies in all samples.

Analysis of the LD protein composition

Samples were loaded onto a 12% SDS–PAGE gel, and trypsin digestion was performed using an automatic robot (Proteineer; Bruker, Bremen, Germany) as described in Shevchenko et al. (1996). In brief, bands were excised from the gel, cut into pieces, washed with 50 mM ammonium bicarbonate, reduced with 10 mM DTT and alkylated with 55 mM iodoacetamide. Proteolytic digestion was performed by incubating for 18 h at 37°C with 10 μg/ml of recombinant sequencing grade trypsin (Promega, Madison, WI, USA). The peptides extracted were dried by speed-vac centrifugation and stored at −20°C until examination. Peptides were identified by nano-LC–ESI-MS/MS analysis (Liquid hormatography Electrospray ionization Tandem Mass Spectrometric), using an Eksigent 1D-nano HPLC coupled to aTripleTOF 5600 Time of Flight mass spectrometer analyzer (SCIEX, Framingham, MA, USA). Trypsin-digested peptides were loaded on a silica-based reversed-phase column (Waters nanoACQUITY Ultra Perfomance Liquid Chromatography, UPLC 75 µm × 15 cm, 1.7 µm particle size) and switched online with a trap column (Acclaim PepMap 100, 5 µm particle diameter, 100 Å pore size). A 0.1% formic acid solution in 98% water/2% acetonitrile was delivered at 3 µl/min with a loading pump (Scharlab, Barcelona, Spain), using 0.1% formic acid (Fluka, Buchs, Switzerland) as mobile phase A and 0.1% formic acid in 100% acetonitrile as mobile phase B. The flow rate was 250 nl/min, operated under gradient elution conditions (96% A:4% B for 5 min, a linear increase to 30% B over 110 min, a linear increase to 40% B over 10 min, a linear increase to 95% B over 5 min, isocratic conditions of 95% B for 5 min and a return to the initial conditions over 10 min). Automatic data-dependent acquisition using dynamic exclusion allowed full scan (m/z 350–1,250) MS spectra to be obtained, followed by tandem MS CID (Collisium Induced Dissociation) spectra of the 25 most abundant ions. The MS and MS/MS data obtained were used to search an A. thaliana database containing 31,478 sequences (downloaded from UniProtKB, February 2016). Database searches were performed using the Mascot server v.2.6, and the search parameters were set as follows: carbamidomethyl cysteine as a fixed modification and acetyl (protein N-term) NQ deamidation; Glu to pyro-glutamic and oxidized methionine as variable modification. Peptide mass tolerance was set at 25 ppm and at 0.1 Da for MS and MS/MS spectra, respectively, and two missed cleavages were allowed. Peptides with individual Mascot ion scores indicating identity or extensive homology (P < 0.05) were used for protein identification, and only proteins with at least one unique peptide were considered. We used AgriGO v2.0 http://systemsbiology.cau.edu.cn/agriGOv2/ to obtain an overview of Gene Ontology functional term enrichment for the proteomes defined in this study.

Fusion protein construction, transient expression in Nicotiana benthamiana plants and generation of transgenic Arabidopsis plants

Total RNA from wild-type Arabidopsis plants 24 h after Pst DC3000 avrRpm1 (10⁶ cfu/ml) inoculation was isolated according to Logemann et al. (1987) and used for cDNA amplification. cDNA fragments for the preparation of chimeric constructs were amplified using the Expand High Fidelity polymerase (Roche, Mannheim, Germany); the forward and reverse primers used in each case are shown in Supplementary Table S3. The PCR fragments amplified were inserted into appropriate GATEWAY plasmids. The pGWB5 vector was used to prepare the α-DOX1:GFP, CLO3:GFP, LOX1:GFP, GPAT4:GFP, GPAT8:GFP or PAD3:GFP chimeric proteins (Nakagawa et al. 2007), and the pUBC-RFP-Dest vector was used to generate the CLO3:RFP construct (Grefen et al. 2010). The distribution of fusion proteins was examined by transient expression in N. benthamiana, performed as previously described (van Herpen et al. 2010). To activate an HR response, Agrobacterium was co-infiltrated with low doses of P. syringae pv. syringae B728a (10⁴ cfu/ml) in a co-infiltration mix. Tissues were examined 3 d after inoculation. Agrobacterium tumefaciens carrying the chimeric constructs prepared were used to transform wild-type Arabidopsis plants and independent transgenic lines expressing the α-DOX1:GFP, CLO3:GFP, LOX1:GFP, GPAT4:GFP, GPAT8:GFP or PAD3:GFP fusion proteins were selected. Homozygous lines expressing α-DOX1:GFP, CLO3:GFP, LOX1:GFP or PAD3:GFP were identified, although no homozygous lines expressing GPAT4:GFP and GPAT8:GFP could be obtained and, thus, plants from heterozygous lines were used in our studies.

Analysis of gene expression

Total RNA was prepared from untreated leaves of 4-week-old wild-type plants and from leaves collected 3 d after Pst DC3000 avrRpm1 (10⁶ cfu/ml) or 10 mM MgCl₂ inoculation. RNA extraction and quantification of gene expression were performed as described previously (Logemann et al.1987, Marcos et al. 2015). Accession numbers for the genes analyzed and primers used for these analyses are described in Supplementary Table S4.

Microscopic visualization and quantification of LDs

Fluorescent images were obtained on a Leica TCS-SP8 confocal system equipped with a white laser (WLL2) for 470–670 excitation and LAS X v. 2.0 software (Leica Microsystem, Wetzlar, Germany) using a 20×/0.75 NA and 63×/1.20 NA water immersion objectives. Images were saved as 1,024 × 1,024-pixel images. To visualize LDs in Arabidopsis wild-type plants, leaves were cut in small pieces, stained for 30 min with BODIPY 493/503 (2 μg/ml; Thermo Fisher, Waltham, MA, USA) and then washed in water for 5 min before imaging. Sequential scanning was performed with a 488-nm excitation laser line, and the emission signal was collected in a spectrum of 500–540 nm. LDs were quantified in BODIPY-stained leaves of Arabidopsis wild-type plants using the microscopy image analysis software Imaris (https://imaris.oxinst.com). An intensity threshold for LD selection was set manually by visual inspection of representative three-dimensional (3D) images that was then applied to all pictures examined. Round particles corresponding to LDs were thus identified and highlighted with yellow dots (see Supplementary Fig. S7). For each type of samples, 3D images from five or more leaves from independent experiments were examined and the amount of LDs per tissue volume was analyzed statistically by Student’s t-test. The size of LDs in BODIPY-stained leaves was estimated using Fiji (https://fiji.sc/). Briefly, a Gaussian Blur (σ = 1) was applied to stacked projections, then a maximum entropy intensity threshold was applied to identify high fluorescence areas and, of these, round particles in the range of 0.4–2 μm diameter were identified (software settings: area of 0.15–3.14 μm² and circularity between 0.5 and 1) (see Supplementary Fig. S8). The diameter of at least 300 LDs from three or more independent leaves was estimated in each condition. As LD sizes did not follow a normal distribution, data were analyzed statistically by the Wilcoxon–Mann–Whitney nonparametric test.

The distribution of GFP fusion proteins in N. benthamiana leaves was examined in leaves expressing α-DOX1:GFP, CLO3:GFP, LOX1:GFP, GPAT4:GFP, GPAT8:GFP or PAD3:GFP. Co-infiltration with low doses of P. syringae pv. syringae B728a (10⁴ cfu/ml) was used to activate an HR response. To assess the co-localization of GFP fusion proteins and LDs, leaves of Pseudomonas-infected N. benthamiana plants were cut in pieces, stained for 30 min with Nile Red (2 μg/ml; Life Technologies, Carlsbad, CA, USA) and then washed with water for 5 min before microscopic examination. Sequential scanning was performed with a 488-nm (for GFP) and 550-nm (for Nile Red) excitation laser line. Emission band-widths between 497 and 530 nm (using a gated-hybrid emission detector) and 562 and 621 nm (using a PMT detector) were used for GFP and Nile Red detection, respectively. Agrobacterium containing CLO3:RFP was added to the infiltration mix to examine the protein co-localization of CLO3:RFP and GFP fusions. Sequential scanning was performed with a 488-nm (for GFP) and DPSS 561-nm (for RFP) excitation laser line. The distribution of GFP fusion proteins in Arabidopsis plants (4 weeks) was examined in untreated transgenic lines and in plants 3 d after MgCl₂ or Pst DC3000 avrRpm1 (10⁶ cfu/ml) inoculation. The co-localization of GFP-fused proteins and LDs in untreated or infected Arabidopsis leaves inoculation was examined by staining with 2 μg/ml Nile Red, as described above. Purified LD fractions were prepared from wild-type and transgenic Arabidopsis lines 3 d after Pst DC3000 avrRpm1 (10⁶ cfu/ml) or 10 mM MgCl₂ inoculation to assess the GFP and LD co-localization. Samples were diluted for flow cytometry quantification, and aliquots with similar LD numbers were stained for 30 min with Nile Red (0.5 μg/ml; Life Technologies) and washed before microscopic visualization, as described above.

In vivo analyses of pathogen infection

Plectospharella cucumerina infections were performed and plant responses were examined as described in López Sánchez et al. (2016). Plants were kept at 100% humidity during fungal infection and the lesions diameter caused by the fungal infection was measured 9 d after inoculation. Botrytis cinerea was cultured for 3.5 weeks at 23°C in Petri dishes with PDA medium supplemented with Arabidopsis leaves and with 8 h light. Fungal spores were suspended in sterile water, filtered through 2 layers of miracloth, centrifuged at 4,000 × g for 10 min and washed with water. Spores were collected by centrifugation at 11,000 × g for 5 min, suspended in 0.5 PDB + 0.01 M sucrose, adjusted to 2.5 × 10⁴ spores/ml and incubated in the dark for 2.5 h to induce fungal germination. Subsequently, 6 µl of the suspension was used to drop inoculate four leaves from 25 plants/genotype. Plants were kept at 100% humidity during fungal infection, and the lesions diameter caused by the fungal infection was measured 3 d after inoculation. In both fungal infections, the data reported are the means and standard errors of the values obtained from 25 plants per genotype (n = 25). Similar results were obtained in two independent experiments. For bacterial assays, strains Pst DC3000 avrRpm1 (avirulent), Pst DC3000 (virulent) or Pst DC3000 AK87 (coronatine deficient) were grown overnight in Petri dishes with King’s B medium (28°C) and collected by washing with 10 mM MgCl₂ prior to inoculation. Bacterial inoculations were performed by syringe inoculation into the apoplast (10⁵ cfu/ml) or spraying onto the leaf surface (10⁸ cfu/ml), as described in Izquierdo et al. (2018). In each experiment, three leaves from at least six plants/genotype were inoculated and bacterial growth was evaluated up to 3 d later. Disks of 0.6 cm² were excised from each infected leaf using a core borer pooled in triplicates and homogenized in sterile water. Six replicates were used for each time interval examined. Bacterial populations were determined by plating appropriated dilutions from each sample in King’s medium. The results reported are the means and standard errors of six biological replicates from 3 independent experiments (n = 18).

Supplementary Data

Supplementary data are available at PCP online.

Acknowledgments

We are grateful to Mary Paz González for critical reading of the manuscript and stimulating discussions, J. Ohlrogge and C. Nawrath for generously providing the gpat4/gpat8 seeds, J. Glazebrook for the pad3-1 seeds and V. Pastor and V. Flors for the B. cinerea fungi used in our study. The authors thank Mª. del Carmen Moreno and Sara Escudero for the flow cytometry analyses, Sylvia Gutierrez and Ana Oña for the confocal microscopy studies, Alberto Paradela for the proteomics analyses, I. Poveda for photography and Mark Sefton for editorial assistance.

Funding

European Comission Marie Sklodowska-Curie Actions Individual Fellowships (MSCA-IF-2016-746136 to A.L.); Spanish Ministry of Economy and Competitiveness/Fondo Europeo de Desarrollo Regional (BIO2015-68130-R); and Ministry of Science, Innovation and Universities (RTI2018-097102-B-100 to C.C.).

Disclosure

The authors have no conflicts of interest to declare.

Author notes

3

Joint first author

4

The author responsible for the distribution of materials related to the findings of this study is Carmen Castresana