Two Abscisic Acid-Responsive Plastid Lipase Genes Involved in Jasmonic Acid Biosynthesis in Arabidopsis thaliana

Abscisic acid-inducible genes encode plastid lipases involved in releasing polyunsaturated fatty acids from chloroplast lipids as precursors for oxylipin production in Arabidopsis thaliana.

Abstract

Chloroplast membranes with their unique lipid composition are crucial for photosynthesis. Maintenance of the chloroplast membranes requires finely tuned lipid anabolic and catabolic reactions. Despite the presence of a large number of predicted lipid-degrading enzymes in the chloroplasts, their biological functions remain largely unknown. Recently, we described PLASTID LIPASE1 (PLIP1), a plastid phospholipase A₁ that contributes to seed oil biosynthesis. The Arabidopsis thaliana genome encodes two putative PLIP1 paralogs, which we designated PLIP2 and PLIP3. PLIP2 and PLIP3 are also present in the chloroplasts, but likely with different subplastid locations. In vitro analysis indicated that both are glycerolipid A₁ lipases. In vivo, PLIP2 prefers monogalactosyldiacylglycerol as substrate and PLIP3 phosphatidylglycerol. Overexpression of PLIP2 or PLIP3 severely reduced plant growth and led to accumulation of the bioactive form of jasmonate and related oxylipins. Genetically blocking jasmonate perception restored the growth of the PLIP2/3-overexpressing plants. The expression of PLIP2 and PLIP3, but not PLIP1, was induced by abscisic acid (ABA), and plip1 plip2 plip3 triple mutants exhibited compromised oxylipin biosynthesis in response to ABA. The plip triple mutants also showed hypersensitivity to ABA. We propose that PLIP2 and PLIP3 provide a mechanistic link between ABA-mediated abiotic stress responses and oxylipin signaling.

INTRODUCTION

Chloroplasts in plants and algae are organelles that carry out a number of crucial biological functions including photosynthesis. They also participate in plant immune responses through the production of defense signaling compounds such as the oxylipin jasmonic acid (JA). Membranes with a unique lipid composition are required for the proper functioning of the chloroplasts. Chloroplast membranes contain predominately galactolipids, monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG), as well as two anionic lipids, phosphatidylglycerol (PG) and sulfoquinovosyldiacylglycerol (Andersson and Dörmann, 2009; Benning, 2010; Boudière et al., 2014). In seed plants, lipid assembly begins in the plastids where fatty acid (FA) synthesis occurs. In Arabidopsis thaliana, two pathways are responsible for plant lipid assembly (Benning, 2009; Hurlock et al., 2014). De novo-synthesized FAs can either directly enter the plastid (prokaryotic) pathway or are exported to the endoplasmic reticulum (ER) where they are assembled into polar lipids (eukaryotic pathway). Lipid precursors synthesized in the ER can also return to the plastid and contribute to chloroplast lipid biosynthesis (Wang and Benning, 2012; Hurlock et al., 2014). Lipids synthesized by the two pathways are distinguished by their glyceryl sn-2-bound acyl groups due to the specificity of the respective acyltransferases (Frentzen et al., 1983; Frentzen, 1986). Plastid pathway-derived galactolipids contain 16-carbon acyl groups at the sn-2 glyceryl position, while ER-derived lipids contain 18-carbon acyl groups in this position.

To date, many of the genes and enzymes responsible for chloroplast lipid biosynthesis have been identified, yet still little is known about the role of enzymes involved in chloroplast lipid degradation. Turnover of glycerolipids requires lipases to release the acyl or the head groups from the three different positions of the glyceryl backbone (Troncoso-Ponce et al., 2013; Kelly and Feussner, 2016). Based on their preferred hydrolysis position, phospholipases are classified into three groups (Wang et al., 2012): Phospholipase C and phospholipase D cleave the phosphodiester bonds adjacent and distal to the sn-3 glyceryl position, respectively. Phospholipase A (PLA) releases FAs from glyceryl sn-1 or sn-2 positions; PLA₁ hydrolyzes at the sn-1 position, while PLA₂ hydrolyzes at sn-2. Free FAs released by PLAs from the membrane lipids can be subsequently oxygenated by either enzymatic reactions or chemical oxidation (Blée, 1998; Mueller, 2004; Mosblech et al., 2009). In plants, enzyme-mediated FA oxidation predominantly works on polyunsaturated FAs, e.g., linolenic acids (18:3; number of carbons:number of double bonds in the acyl chain) through the lipoxygenase pathway (Porta and Rocha-Sosa, 2002). Lipoxygenases catalyze the synthesis of FA hydroperoxides. These highly reactive oxylipins are subsequently converted through an array of pathways to structurally and functionally diverse compounds in response to various developmental and environmental cues (Göbel et al., 2002; Porta and Rocha-Sosa, 2002; Andersson et al., 2006; Mosblech et al., 2009). The best characterized lipid oxidation pathway is the allene oxide synthase pathway, which leads to the synthesis of 12-oxo phytodienoic acid (OPDA), JA, and various amino acid conjugates of JA, including the bioactive JA-isoleucine (JA-Ile) (Turner et al., 2002; Wasternack and Hause, 2013). JA production and signaling are induced by mechanical wounding and other forms of tissue damage caused by herbivorous insects and necrotrophic pathogens (Koo and Howe, 2009; Wasternack and Hause, 2013). JA serves several important developmental roles, including stamen development, growth inhibition (Feys et al., 1994; McConn and Browse, 1996; Major et al., 2017), and lignin biosynthesis (Ellis et al., 2002; Denness et al., 2011). DEFECTIVE IN ANTHER DEHISCENCE1 (DAD1) is a PLA₁ that is primarily present in the reproductive tissues, where it initiates JA biosynthesis critical for pollen maturation, anther dehiscence, and flower opening in Arabidopsis (Ishiguro et al., 2001). In leaf tissues, DONGLE encodes a DAD1-homologous lipase, which among others has been proposed to contribute to wound-induced JA production (Hyun et al., 2008; Ellinger et al., 2010). In addition, synergistic or antagonistic interactions between JA and other phytohormones including salicylic acid, ethylene, and abscisic acid (ABA) have been observed (Thomma et al., 1998; Song et al., 2014; Suzuki, 2016), suggesting that the interaction between different hormones is an important strategy for plants to fine-tune their responses to environmental cues.

It is known that JA biosynthesis is stimulated by applied ABA and during ABA-inducing abiotic stress conditions and that JA biosynthesis is compromised in ABA-deficient mutants (Creelman and Mullet, 1995; Adie et al., 2007; Fan et al., 2009; Avramova, 2017). Nevertheless, the underlying mechanisms by which JA synthesis is controlled by ABA are not well understood. During abiotic stress, rising ABA levels induce JA synthesis, which synergistically optimizes plant abiotic defenses by altering the plant transcriptome, possibly facilitating stomata closure, inhibiting cell division, and epigenetically modifying defense genes during dehydration responses (Murata et al., 2015; Riemann et al., 2015; Liu et al., 2016; Valenzuela et al., 2016). A basic-helix-loop-helix transcription factor MYC2 plays roles in both JA and ABA signaling and has been proposed as one of the factors linking ABA and JA signaling (Kazan and Manners, 2013; Aleman et al., 2016).

In Arabidopsis, over 50 lipases with unverified enzymatic activities and unknown biological functions are predicted to be located in chloroplasts (Ajjawi et al., 2010). Recently, we discovered a thylakoid associated lipase named PLIP1, which is a PLA₁ that releases polyunsaturated FAs from chloroplast PG, leading to the export of the FAs to the ER for seed oil biosynthesis (Wang et al., 2017). Sequence comparison and phylogenetic analyses identified two putative PLIP1 paralogs in the Arabidopsis genome, which we designated PLIP2 and PLIP3. Both PLIP2 and PLIP3 are located in the chloroplasts. PLIP2 and PLIP3 are not involved in seed oil biosynthesis, and expression of the respective genes was induced by ABA. Overexpression of PLIP2 and 3 stunted plant growth and led to oxylipin accumulation. The growth inhibition phenotype of the overexpression plants was fully restored by blocking JA signaling. The plip triple mutant also exhibited compromised oxylipin production in response to ABA. A role for PLIP2 and 3 in linking ABA and JA pathways is proposed.

RESULTS

PLIP2 and PLIP3 Are Putative Paralogs of PLIP1 in Arabidopsis

PLIP1 was included in the Chloroplast 2010 project studying chloroplast targeted proteins (Ajjawi et al., 2010) and later characterized as a phospholipase A₁ enzyme involved in acyl export from the chloroplasts and contributing to seed oil biosynthesis (Wang et al., 2017). Comparing the PLIP1 protein sequence against the Arabidopsis proteome using BLASTp with default settings in TAIR (http://www.arabidopsis.org) identified two putative paralogs of PLIP1 in Arabidopsis encoded by AT1G02660 and AT3G62590, which were designated PLIP2 and PLIP3, respectively. Phylogenetic analysis of the top 17 Arabidopsis proteins similar to PLIP1 together with other well-characterized plant lipases (Figure 1; sequence alignment in Supplemental File 1) showed that PLIP2 and 3 clustered with PLIP1 but were distinct from other predicted α/β hydrolases. However, PLIP2 and 3 peptide sequences only share 41% and 37% identity with PLIP1, respectively (57% and 51% similarity, respectively). PLIP2 and 3 are more closely related to each other, sharing 51% identity and 65% similarity.

Figure 1.

Phylogenetic Analysis of PLIP1-Similar Protein Sequences and Other Characterized Lipase Sequences in Arabidopsis.

The phylogenetic tree was built using the maximum likelihood method with PLIP1 and the top 17 Arabidopsis similar protein sequences identified from the BLASTp search, as well as five other known Arabidopsis lipases. Previously studied lipases are presented with their gene names, and others with their gene accession numbers. Sequences alignments are shown in Supplemental File 1. Bootstrap values (based on 1000 repetitions) are indicated at the tree nodes. The scale measures evolutionary distances in substitutions per amino acid.

PLIP2 and 3 Are Located in the Chloroplast

PLIP1 is imported into the chloroplast with an N-terminal transit peptide (Wang et al., 2017). PLIP2 and 3 were also predicted to be located in the chloroplast using different subcellular location prediction programs, e.g., TargetP (Emanuelsson et al., 2000) or PredSL (Petsalaki et al., 2006). To experimentally verify these predictions, the 3′-ends of the coding sequences for PLIP2 or 3 derived from an Arabidopsis wild-type (Col-0) cDNA were spliced to the 5′-end of the open reading frame of YFP to create a C-terminal fusion of YFP to the PLIPs. When the PLIP-YFP constructs were stably expressed in the wild type under the control of the CaMV 35S promoter, the YFP signals were observed specifically overlapping with the chlorophyll signals using confocal microscopy (Figure 2A). Interestingly, the chloroplast sizes were smaller in plants overexpressing either PLIP2 or 3 than those expressing the empty vector (EV) control. This phenotype likely results from the reduced size of PLIP-overexpressing plants, as described below.

Figure 2.

Subcellular Localization of PLIP2 and PLIP3 in Arabidopsis.

(A) Subcellular localization of YFP tagged PLIP2 and PLIP3 in leaf mesophyll cells of 3-week-old Arabidopsis Col-0 transformed with PLIP2-YFP or PLIP3-YFP driven by the 35S promoter or EV control using confocal laser scanning microscopy. Chlorophyll autofluorescence is shown in red, and YFP fluorescence is shown in yellow. Overlay of chlorophyll and YFP are shown as “Merge.” Representative images from one experiment are presented. Bars = 5 µm.

(B) Chloroplast import experiments with radiolabeled PLIP2 and PLIP3. Chloroplasts were treated with (+) or without (−) trypsin. Total chloroplast membranes (P) or soluble (S) fractions were analyzed by SDS-PAGE followed by fluorography. TP, translation products; p, precursor; m, mature form; MW, molecular weight markers.

The PLIP2 and 3 genes are predicted to encode 78,346 D and 73,044 D proteins, respectively. Based on the ARAMEMNON database (Schwacke et al., 2003), PLIP2 and 3 have similar predicted topologies consisting of four transmembrane domains with a very short transit peptide at the N terminus. A chloroplast import assay was performed to confirm the plastid location of PLIP2 and 3. PLIP2 and 3 cDNAs were translated in vitro in the presence of labeled methionine. The translation products were then tested for import into isolated pea (Pisum sativum) chloroplasts. As shown in Figure 2B, both PLIP2 and 3 were imported into chloroplasts. During the import, the precursor protein of PLIP2 was processed to a smaller protein, indicative of transit peptide cleavage. PLIP3 did not show detectable processing, possibly due to the small size of the transit peptide. The PLIP2 signal was present in both the pellet and supernatant fractions of the import reaction. Trypsin is a protease capable of penetrating the chloroplast outer but not the inner envelope membrane. Upon treatment of import reactions with trypsin, only the mature PLIP2 protein was retained in the pellet fraction, while the soluble fraction was not affected (Figure 2B). This suggests that imported PLIP2 is probably ubiquitously present in the chloroplast envelope membranes, stroma, and thylakoids. For PLIP3, the signal was specifically associated with the membrane fractions in the pellet and the signal intensity partially faded with trypsin treatment, indicating that PLIP3 is a membrane protein embedded in the chloroplast envelope membranes and thylakoids. Taken together, PLIP2 and 3 are chloroplast proteins, but likely have distinct subplastid locations.

PLIP2 and 3 Are Glycerolipid A₁ Lipases with Serine in Their Catalytic Sites

Both PLIP2 and 3 are annotated as triacylglycerol lipases in TAIR and contain conserved Lipase 3 domains similar to PLIP1. Lipases are characterized by a signature catalytic triad comprised of Ser-Asp-His and reduced occasionally to a Ser-Asp dyad. Sequence alignment of PLIP2 and 3 with classic lipases using NCBI’s conserved domain database (Marchler-Bauer et al., 2015) identified a conserved triad in PLIP2, while only a dyad was identified in PLIP3 (Supplemental Figure 1). To verify whether PLIP2 and 3 possess lipase activity, recombinant PLIP2 or 3 constructs were introduced into Escherichia coli or Saccharomyces cerevisiae, and protein production was investigated by immunoblotting against the 6× His tag included in the N terminus. Production of the recombinant PLIP2 protein and a mutant protein in which the putative catalytic residue Ser-428 was changed to Ala (PLIP2^S428A) was successful in E. coli using a protocol with a brief induction period and growth under low temperature (Figure 3A). Phosphatidylcholine (PC) was then provided to the microsome fractions isolated from E. coli cultures expressing PLIP2, PLIP2^S428A, or the EV as control. At the end of the reaction time, lipids were extracted and analyzed on a thin-layer chromatography (TLC) plate (Figure 3B). Degradation of PC and production of lyso-PC were observed with PLIP2 but not with PLIP2^S428A, suggesting that PLIP2 is a lipase with a conserved Ser-428 as its catalytic residue. For PLIP3, despite using different variations of the protocol, we were unable to detect production of the recombinant PLIP3 protein nor that of the point mutant PLIP3^S385A protein in either E. coli or in yeast by immunoblotting. However, when PC was provided to the cell culture lysates expressing PLIP3, PLIP3^S385A, or the EV, stronger lipase activity was observed for extracts from the culture expressing PLIP3 (Supplemental Figure 3A).

Figure 3.

Production of the Recombinant PLIP2 Proteins and Its Lipase Activity in Vitro.

(A) Detection of PLIP2 and PLIP2^S428A protein by immunoblotting against the N terminus 6× His tag (top panel) and Coomassie Brilliant Blue (CBB) staining (bottom panel) in the lysed E. coli cultures with or without isopropyl β-d-1-thiogalactopyranoside induction. The E. coli strain carrying the EV was included as a negative control.

(B) Thin-layer chromatogram of products of a representative in vitro lipase reaction using PC with E. coli microsomes carrying the wild type (PLIP2 + PC) or the mutant enzyme (PLIP2^S428A + PC). Substrate without enzyme (PC) or with E. coli microsome carrying the EV control (EV + PC) were included as controls. O, origin of sample loading. Lipids were visualized by iodine vapor staining.

Detection of lyso-lipids in the lipase assay implied that PLIP2 could only hydrolyze one of the two acyl-glyceryl ester bonds in PC. To determine the positional specificity of PLIP2, two defined molecular species of PC molecules with inverted acyl compositions were offered to PLIP2. At the end of the reaction time, lyso-PC was isolated by TLC and FA methyl esters derived from the lyso-lipid were analyzed by gas chromatography (Figure 4A). For PC with acyl groups composed of 18:1^∆9/16:0 (sn-1/sn-2) as substrates, 18:1^∆9 was selectively released while 16:0 was retained. The opposite pattern was observed for lyso-PC produced from 16:0/18:1^∆9-PC. These data suggested that PLIP2 preferentially hydrolyzes the sn-1 position and therefore is a glycerolipid lipase A₁. Using a similar approach, PLIP3 was also shown to be a glycerolipid lipase A₁ (Supplemental Figure 3B). To further investigate a possible acyl group preference of PLIP2 at the sn-1 glyceryl position, PLIP2 was offered different combinations of defined PC molecules carrying the same acyl groups at the sn-2 but acyl groups with different degrees of saturation at the sn-1 position (Figure 4B). Comparing 18:0/18:1 with 18:1/18:1 PC, PLIP2 showed slightly higher activity with 18:1/18:1 PC during the linear phase of the reaction (Supplemental Figures 2A and 2B), while no statistically significant difference was detected between 18:0/18:2 and 18:2/18:2. This result implied that, at least in vitro, PLIP2 does not have a strong preference for the degree of saturation of acyl groups and is capable of releasing both saturated and unsaturated acyl groups from the sn-1 position.

Figure 4.

In Vitro Lipase Activity of PLIP2.

(A) Gas-liquid chromatograms of fatty acid methyl esters derived from defined PC substrates or lyso-PC fractions from PLIP2 lipase reactions with different PC substrates. 15:0 was used as an internal standard.

(B) PLIP2 lipase activity on defined PC substrates (carbon number:double bond number; sn-1/sn-2) with different degree of saturation of the sn-1 acyl groups. PC containing 18:0/18:1 and 18:1/18:1 and PC containing 18:0/18:2 and 18:2/18:2 were compared, respectively. n = 4, ±sd. Student’s t test was applied (**P < 0.01; n.s., not significant).

(C) PLIP2 lipase activity toward commercially available galactolipids and phospholipids. Reactions were stopped after 40 min based on the linear activity range for a time course observed under identical conditions shown in Supplemental Figure 2. PLIP2^S428A was included as a negative control. PC and PG contained two oleic acids (18:1) each, while MGDG and DGDG were isolated from plants with a mixed fatty acid composition. n = 4, ±sd.

(D) PLIP2 lipase activity on lipids extracted from intact spinach chloroplasts. Lipid degradation at each time point is presented relative to the 0 time point recorded prior to the addition of protein. n = 3, ±sd. SQDG, sulfoquinovosyldiacylglycerol.

To examine the lipid head group preference of PLIP2, commercial galactolipids and phospholipids were provided to microsomal PLIP2 or PLIP2^S428A (Figure 4C). Compared with PLIP2^S428A, PLIP2 showed activity on all substrates tested, however, with slightly higher activity toward MGDG (Figure 4C). The commercially available substrates used did not reflect the in vivo acyl compositions and potentially competing lipid substrates present in the chloroplasts were not considered. Therefore, chloroplasts were isolated from spinach (Spinacia oleracea), which is an 18:3 plant with a similar lipid composition to that of Arabidopsis (Heinz and Roughan, 1983), and bulk membrane lipids were extracted and provided to PLIP2 or PLIP2^S428A (Figure 4D). During a time course, the galactolipids MGDG and DGDG showed faster degradation compared with that of PG and sulfoquinovosyldiacylglycerol, implying a preference of PLIP2 for galactolipids. For PLIP3, similar promiscuousness was observed for individual lipids (Supplemental Figure 3C). However, when PLIP3 was provided with the spinach chloroplast lipid mixture, PG was degraded slightly but significantly faster than the other three lipids (Supplemental Figure 3C). It should be noted that DGDG appears to have a tendency to autohydrolyze given the high level of apparent hydrolysis in both reactions with the respective inactivated point mutant proteins (Figure 4C; Supplemental Figure 3C).

PLIP2 Uses MGDG as Primary Substrate, While PLIP3 Uses PG

Plant glycerolipid lipases such as PLIP1 tend to be promiscuous with regard to the glycerolipid head groups in vitro and the same was observed here for PLIP2 and 3. A facile approach to investigate lipase substrate specificity and to identify their native substrates is to overproduce them in the plant, thereby providing the enzyme with a native lipid environment (Wang et al., 2017). Accordingly, we took advantage of the PLIP2- and 3-overexpressing plants that were used for the localization study (Figure 2). We generated over 20 independent overexpression lines for each of PLIP2 and PLIP3 constructs. Three lines for each construct were chosen for further studies based on the presence of the recombinant protein as detected by confocal microscopy. It should be noted that despite the use of the strong constitutive promoter, location of PLIP2-YFP and PLIP3-YFP was strictly limited to the chloroplasts in the overexpression lines. We performed lipidomic analysis to determine the relative abundance of polar lipids and the acyl composition of individual polar lipids in one EV control line and two PLIP2 overexpression (PLIP2-OX) lines (Figure 5A; Supplemental Figure 4), as well as two PLIP3-OX lines (Supplemental Figure 5). The lipid phenotype of PLIP2-OX and PLIP3-OX plants was very similar to that of the previously reported PLIP1-OX plants (Wang et al., 2017). Specifically, two chloroplast lipids (MGDG and PG) and ER lipids (phosphatidylinositol, phosphatidylethanolamine, and PC) showed statistically significant changes in their acyl compositions. MGDG exhibited primarily decreased 16:3-to-18:3 ratios, and PG showed decreased 16:1-to-16:0 ratios in all PLIP2-OX and PLIP3-OX plants. The 16-carbon acyl groups are predominantly present at the sn-2 glyceryl position of plastid pathway-derived lipids. These changes in acyl ratios suggest that PLIP2 and 3 might have a substrate preference for plastid pathway-derived MGDG and PG.

Figure 5.

The Steady State and Dynamic Lipid Phenotypes of the PLIP2-OX Plants.

(A) Relative acyl composition of MGDG in PLIP2-OX and EV control lines. Leaf samples harvested from one plant were pooled as one biological repeat; n = 4, ±sd. Student’s t test was applied to compare the empty vector (EV) control plants with each of the two PLIP2-OX plants (**P < 0.01).

(B) and (C) In vivo pulse-chase acetate labeling of lipids of leaves of the EV control and the PLIP2-OX1 plants grown for 4 weeks in soil. The [¹⁴C]-acetate labeling pulse lasted for 60 min. The fractions of label in all polar lipids were calculated as percentages of total incorporation of label in polar lipids (B). After the pulse, [¹⁴C]-acetate medium was replaced with nonlabeled acetate medium to initiate the chase with a duration of 3 d (C). The fractions of label in all polar lipids are given during a time course. n = 3, ±sd. PE, phosphatidylethanolamine.

The steady state compositional changes observed for the chloroplast lipids MGDG and PG led us to further investigate their synthesis and turnover employing in vivo with pulse-chase lipid labeling analysis in both the PLIP2-OX and PLIP3-OX plants, to identify their native substrates in vivo. For this purpose, [¹⁴C]-acetate was used to label the de novo synthesized FAs in detached leaves. In the PLIP2-OX leaves, MGDG was labeled ∼20% higher during the pulse phase (Figure 5B), suggesting that FA incorporation into MGDG is accelerated. During the chase phase (Figure 5C), the label in MGDG was lost within the first 10 h, and labeled MGDG eventually reached a level comparable to that in the EV control plants. The accelerated turnover of the MGDG pool in the PLIP2-OX leaves suggested that MGDG is likely the favored native substrate of PLIP2 in vivo. In contrast, for the PLIP3-OX plants (Supplemental Figure 6), pulse-chase labeling data showed that the labeling of the MGDG pool was not altered. Instead, PG was labeled almost twice as high as in the PLIP3-OX plants during the pulse phase, and the label in PG was rapidly lost during the chase phase, a pattern that resembled that previously observed for the PLIP1-OX plants (Wang et al., 2017). These data suggested that PLIP3 likely uses PG as its preferred in vivo substrate as was also concluded for PLIP1 (Wang et al., 2017).

The Growth and Lipid Phenotypes Are Not Correlated in PLIP-OX Plants

One of the striking phenotypes of the PLIP2 and 3 overexpression plants was that their vegetative growth was severely reduced under normal conditions (Figure 6A). This growth phenotype was similar to, but more severe than, that observed for the PLIP1-OX plants (Figure 6A). Despite the fact that overexpression of PLIPs significantly altered plant lipid composition, it seemed unlikely that these changes were the underlying cause for the growth phenotype because the lipid phenotypes were not correlated with the growth phenotype. To illustrate this fact, acyl group ratios of the presumed PLIP-preferred substrates MGDG (18:3/16:3) and PG (16:0/16:1) were chosen as representative indicators (Figure 6A, ratios for PG and MGDG at the bottom of the panel). Based on these indicators, PLIP1-OX plants exhibited the most severe lipid changes, but showed the mildest growth reduction among all the PLIP-OX plants. On the contrary, the PLIP2-OX plants exhibited severely reduced vegetative growth, but only showed mild lipid changes for both MGDG and PG. This inconsistency between the lipid and growth phenotypes was also visible in the pulse-chase labeling data (Figure 5; Supplemental Figure 6).

Figure 6.

Additional Phenotypes of the PLIP-OX Plants.

(A) Growth of 4-week-old soil-grown plants. One EV control line and two lines of PLIP1-OX, PLIP2-OX, and PLIP3-OX plants are shown. The relative mol% ratios of acyl groups in PG (16:0/16:1) and MGDG (18:3/16:3) in each line are indicated. Leaves harvested from one plant were pooled as one biological repeat. n = 4, ±sd.

(B) Morphology of the 24-d-old EV and PLIP2-OX1 plants grown on MS agar medium. Inserts show representative enlarged individual plants for details.

(C) Phase separation of the lipid extracts from the EV and PLIP2-OX plants in (B). Three independent extracts are shown for each genotype. The upper aqueous phases accumulating anthocyanin are indicated by an arrow. The image was taken with the plastic reaction tubes inserted into transparent glass tubes on a rack.

(D) Expression of JA responsive genes in the EV and the PLIP-OX plants in (B). Expression levels of indicated genes were determined by quantitative RT-PCR and normalized to the levels in the EV plants as fold changes. Analysis of JAZ10 includes JAZ10.1, 10.2, and 10.3, but not 10.4 splicing forms; two to three seedlings were harvested and pooled as one biological repeat. n = 3, ±sd. Student’s t test was applied to compare the level in the EV plants and every other genotype (**P < 0.01).

The reduced vegetative growth of PLIP2-OX and PLIP3-OX plants was evident even when they grew on nutrient rich agar plates; the PLIP2-OX plants are shown in Figure 6B as an example. Beside their reduced rosette sizes, these plants had shorter leaf petioles than the wild-type control plants. In addition, a red pigment, likely anthocyanin, accumulated at the center of the plants under normal growth conditions (Figure 6B). The elevated level of the presumed anthocyanin was particularly apparent during lipid extraction and phase partitioning as the red pigment stained the upper aqueous phase (Figure 6C). Shortened petioles and accumulation of anthocyanin are typical phenotypes of plants with activated JA signaling or plants that overproduce JA, e.g., the jazQ (Campos et al., 2016) and the dgl-D mutants (Hyun et al., 2008), respectively. JA is an 18:3-FA derived phytohormone, which upon perception by the COI1-JAZ receptor system reprograms the transcriptome to redirect metabolism from growth to defense, which is associated with reduced vegetative growth (Acosta and Farmer, 2010; Huot et al., 2014; Guo et al., 2018). In vitro characterization of PLIP2 and PLIP3 determined their glycerolipid lipase A₁ activity and capability of releasing polyunsaturated FAs (Figure 4). In addition, the presumed native substrates, MGDG for PLIP2 and PG for PLIP3, contain predominately 18:3 at the sn-1 position. Therefore, we hypothesized that the stunted vegetative growth of the PLIP-OX lines might be caused by the overproduction of JA and activation of the JA signaling pathway.

To test whether JA signaling is relevant to the observed growth inhibition, the expression of JA-responsive marker genes was examined by quantitative PCR (Figure 6D). All the analyzed JA marker genes were constitutively induced in the PLIP-OX plants, including genes involved in JA biosynthesis (LOX2), JA signal transduction (MYC2), downstream responses (VSP1), negative regulators of JA signaling (JAZ1 and JAZ10.1-3), as well as JA catabolism (CYP94C1). In addition, the relative gene expression levels appeared correlated with the extent of growth reduction. For example, the smallest PLIP2-OX plants showed the highest induction of the JA responsive genes, while PLIP1-OX plants only exhibited a mild induction consistent with its mild growth reduction (Figure 6D). These data provided evidence that JA-mediated defense responses are activated in the PLIP-OX plants.

Oxylipin Production and Signaling in PLIP2- and PLIP3-OX Plants Is Constitutive

We hypothesized that the constitutively activated JA responses in the PLIP2-OX and PLIP3-OX plants might be a consequence of increased levels of JAs, which could be produced from the FAs released by PLIP2 and 3. To investigate whether JA levels were altered in the PLIP2-OX and PLIP3-OX plants, a targeted metabolite analysis was conducted in leaf tissues (Figure 7A). Most of the oxylipins were maintained at very low basal levels in the wild-type plants under optimal growth condition that typically do not cause oxylipin accumulation (Figure 7A). However, the PLIP2-OX plants showed increased levels of JA, OPDA, methyl-JA (MeJA), and JA-Ile. Two JA catabolites, 12OH-JA and 12OH-JA-Ile, also accumulated to levels that were even higher than their non-hydroxylated forms, JA and JA-Ile, respectively (Figure 7A). This was consistent with the constitutive induction of JA-Ile catabolism genes in the PLIP2-OX plants (Figure 6D), e.g., CYP94C1, which encodes a hydroxylase involved in JA-Ile turnover (Heitz et al., 2012). Similar results were also observed for the PLIP3-OX plants (Supplemental Figure 7A).

Figure 7.

Phytohormone Quantification and Comparative Oxylipin Profiling in the Wild-Type and the PLIP2-OX Plants.

(A) Plants were grown on soil for 3 weeks before the leaves were harvested for hormone extraction and quantification by LC-MS/MS. The two y axes apply to compounds with different order of magnitudes in quantity on the left and right panels, respectively. Leaves harvested from one plant were pooled as one biological repeat. n = 4, ±sd. Student’s t test was applied to compare the level in the EV plants to each other PLIP2-OX line (**P < 0.01).

(B) Partial metabolite profiling chromatographs from 3-week-old wild-type and PLIP2-OX1 plant leaves. Numbers indicate the two metabolite peaks that are elevated in the PLIP2-OX1 plants compared with the wild type. The “1” represents the combined peak of Arabidopside A and D, and “2” represents Arabidopside B.

(C) Relative amount of the three major oxylipins accumulating in the PLIP2-OX1 plant identified by comparative metabolite profiling in (B). The indicated compound levels were normalized to the level of native 18:3/18:3 digalactosyldiacylglycerol, which remains unchanged in wild-type or the PLIP2-OX plants. n = 4, ±sd. Student’s t test was applied (**P < 0.01).

JA biosynthesis is only one of several oxidation fates for polyunsaturated FAs, e.g., 18:3 FAs. Oxylipins other than JA are also recognized as signaling molecules that contribute to various aspects of plant physiology (Porta and Rocha-Sosa, 2002). Therefore, we comprehensively examined whether additional oxylipins accumulate in the overexpression lines. Nontargeted metabolite profiling was preformed focusing on extracted oxidized complex lipids using liquid chromatography-QTOF-MS. Through comparative metabolic profiling, a few metabolites were found to highly accumulate in the PLIP2-OX plants but not in the wild type (Figure 7B). By determining their accurate mass and their molecular fragmentation patterns, these metabolites were tentatively identified as Arabidopsides (Figure 7C). Arabidopsides are structurally very similar to galactolipids except that their typical acyl groups are replaced with OPDA or dinor-OPDA at the sn-1 and sn-2 glyceryl positions (Göbel and Feussner, 2009; Mosblech et al., 2009) (see Supplemental Figure 7B for structures). Arabidopsides do not accumulate in plants grown under normal (i.e., nonstressing) conditions unless plants are wounded. However, in the PLIP2-OX plants, the levels of Arabidopsides A, B, and D were highly elevated. Arabidopsides have been proposed to give rise to OPDAs due to lipid-linked oxidation or they may serve as storage pools for OPDA (Kourtchenko et al., 2007; Nilsson et al., 2012). The latter is possibly a compensatory mechanism for fine-tuning free OPDA levels in the cells (Feussner et al., 2001; Stelmach et al., 2001). Similarly, accumulation of these compounds was also detected in the PLIP3-OX plants (Supplemental Figure 7B).

The targeted phytohormone measurement and the nontargeted metabolite profiling strongly indicated that JA metabolism and likely signaling were affected in the PLIP2-OX and PLIP3-OX plants. To directly test whether activated JA signaling specifically caused the reduced vegetative growth, one of each of the PLIP2-OX and PLIP3-OX lines was crossed with the coi1 mutant defective in CORONATINE INSENSITIVE1, which encodes a component of the JA-Ile coreceptor complex (Thines et al., 2007; Katsir et al., 2008). Indeed, the growth inhibition of the PLIP-OX plants was completely reversed in PLIP-OX; coi1 homozygous offspring (Figure 8A; Supplemental Figure 8A), while the lipid phenotypes remained the same as in the PLIP-OX plants (Figure 8B; Supplemental Figure 8B). Further pulse-chase labeling of the PLIP-OX coi1 homozygous 3-week-old seedlings showed similar lipid labeling patterns compared with those observed for leaves of 4-week-old, soil-grown, stunted PLIP-OX plants (Figure 8C; Supplemental Figure 8C). The PLIP2-OX2; coi1 plants exhibited moderately higher label incorporation into MGDG during pulse and early chase phases, which was more rapidly lost during the later chase phase (Figure 8C). In the PLIP3-OX2; coi1 plants, PG was labeled higher during the pulse phase and then rapidly lost its label during the first 10 h of the chase phase (Supplemental Figure 8C). These labeling data further corroborated that PLIP2 and PLIP3 favor different lipid substrates. Together, by applying a genetic approach, we were able to separate the primary lipid phenotype from the growth phenotype, confirming that the stunted growth phenotypes were caused by constitutive JA signaling due to oxylipin accumulation in the PLIP2-OX and PLIP3-OX plants and not primarily to a change in lipid composition or turnover.

Figure 8.

Rescue of Growth but Not Lipid Phenotypes in PLIP2-OX coi1 Homozygous Plants.

(A) Image of 4-week-old plants grown on soil and the rosette diameter of the indicated plants below the corresponding images. n = 6, ±sd. One-way ANOVA with post-hoc Tukey’s HSD test was applied. Rosette diameters indicated by different letters are significantly different (P < 0.01).

(B) Relative acyl group compositions of leaf monogalactosyldiacylglycerol in the indicated plants in (A). Leaves harvested from one plant were pooled as one biological repeat. n = 3, ±sd. One-way ANOVA with post-hoc Tukey’s HSD test was applied. Acyl groups in four genotypes with different letters indicate acyl levels are significantly different (P < 0.01). ANOVA statistical results are in Supplemental File 2.

(C) In vivo pulse-chase acetate labeling of lipids in the wild type and the PLIP2-OX2; coi1 homozygous rosettes of 3-week-old plants grown on MS agar-solidified medium. The [¹⁴C]-acetate-labeling pulse lasted for 60 min, then [¹⁴C]-acetate medium was replaced with nonlabeled acetate to initiate the chase with a duration of 24 h. The fractions of label in all polar lipids are given during the time course. The time point P before 0 indicates the 30-min pulse labeling phase. n = 3, ±sd. PE, phosphatidylethanolamine; SQDG, sulfoquinovosyldiacylglycerol.

To further test whether induction of JA signaling independent of changes in lipid composition causes stunted growth, lipidomic analysis was conducted on leaves of the jazQ mutant, which exhibits constitutive JA signaling and stunted vegetative growth as a consequence of mutations in multiple JAZ genes (Campos et al., 2016). That no significant differences in the lipid composition of jazQ and wild-type rosettes were detected (Supplemental Figure 9) indicates that growth restriction of this mutant is irrespective of the lipidome. This finding is consistent with the view that JA-induced growth stunting often results from regulatory interactions between growth and defense signaling (Guo et al., 2018).

PLIP2 and PLIP3 Contribute to Oxylipin Production in Response to ABA

The data presented thus far have only indirectly provided clues toward the biological functions of PLIP2 and 3. Taking into consideration the severely reduced vegetative growth of the PLIP2-OX and PLIP3-OX plants, it seemed plausible that PLIP2 and 3 are possibly only expressed under certain conditions or in specific tissues requiring increases in oxylipin levels. Based on data available at the Arabidopsis e-FP browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi; Winter et al., 2007), an online Arabidopsis transcriptomic database, PLIP2 has relatively high expression in pollen, while PLIP3 is primarily expressed in senescent leaf tissues, pollen, and embryos under normal growth conditions. Even though PLIP2 and 3 are not predominantly expressed in leaf tissues, their expressions in vegetative tissues can be specifically induced by ABA treatment as well as under ABA-mediated abiotic stress conditions based on the transcriptome database. To directly verify ABA induction of PLIP2 and 3 expression, 2-week-old wild-type seedlings were transferred to Murashige and Skoog (MS) agar plates containing 7 µM ABA or the equivalent amount of ethanol (Figure 9A). The expression of PLIP2 and 3, but not of PLIP1, was elevated within the first 2 h of exposure to ABA, and the response gradually attenuated during the following 48 h, confirming that PLIP2 and 3 are transcriptionally induced by ABA supplementation. Interestingly, PLIP2 and 3 are likely responsive to different abiotic stressors. Based on the transcriptome database, expression of PLIP2 seems to be specifically induced by cold, while PLIP3 is induced by osmotic stress (Figures 9B and 9C).

Figure 9.

Induction of PLIP Transcription by ABA.

(A) Expression of PLIP genes in response to 7 µM ABA or equivalent ethanol solution (Mock) treatment in 2-week-old seedlings determined by quantitative RT-PCR during a time course as indicated. Three to four seedlings were harvested and pooled as one biological repeat. n = 3, ±sd.

(B) and (C) Absolute expression values of PLIP2 (B) and PLIP3 (C) in response to abiotic stresses. Data were extracted from the Arabidopsis e-FB browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi).

(D) and (E) Jasmonate contents in the wild type, two plip-tri lines, and one plip-CRP line before and after 24-h 7 µM ABA addition. Three to four 2-week-old seedlings were pooled as one biological repeat. n = 3 to 4, ±sd. Two-way ANOVA with post-hoc Tukey’s HSD test was applied. Different letters indicate significant hormonal level difference in different genotypes before or after ABA treatment (P < 0.01). ANOVA statistical results are in Supplemental File 2.

Characterization of the PLIP2-OX and PLIP3-OX plants suggested that the activity of the two proteins promotes JA-Ile synthesis and signaling. Furthermore, the evidence provided above shows that the expression of these genes is induced by ABA, either in response to abiotic stress or directly when applied to the plants. A large body of evidence shows that JA-responsive genes are also induced by ABA or conditions that lead to increased ABA levels, although mechanistic details of this relationship remain underexplored (Adie et al., 2007; Valenzuela et al., 2016). To test the hypothesis that ABA induction of JA responsive genes is generally mediated by PLIP gene activation, we constructed plip triple mutant lines that are deficient in all three PLIP lipases. The triple mutant allowed us to address the potential functional redundancy of the three PLIPs. Two independent T-DNA insertion lines were obtained for PLIP2 and 3 each (Alonso et al., 2003). These T-DNA insertion lines were physiologically indistinguishable from the wild type under normal growth conditions (Supplemental Figure 10A), and their leaf lipid profile was indistinguishable from the wild type as well. We further constructed the double mutant between plip2-2 and plip3-1, which had no physiological phenotype under normal growth conditions. Since the PLIP3 and PLIP1 genes are closely linked on the same chromosome, we obtained the PLIP triple mutant using two different approaches, either by artificial microRNA to reduce PLIP1 expression or by CRISPR-Cas9 technology to disrupt PLIP1 in the plip2-2/3-1 double mutant background. We obtained 14 independent microRNA lines in total with reduced PLIP1 expression, and the final two representative lines, plip-tri1 and plip-tri2 had no expression of PLIP2 and 3, but ∼60% reduced expression of PLIP1 (Supplemental Figure 10B). These plants were also morphologically nearly identical to the wild-type plants under normal growth conditions and did not exhibit any lipid phenotype (Supplemental Figures 10C to 10I). For the CRISPR-Cas9 approach, we obtained only one homozygous line (plip-CRP). Sequencing results showed that the plip-CRP plants have a single nucleotide insertion after the 223rd nucleotide in the PLIP1 coding sequence, causing a frameshift (Supplemental Figure 11). To test whether PLIP2 and 3 share redundant roles with PLIP1 in seed oil metabolism, seed oil content of all the plip mutants, including the single, double, and triple plip mutants, was determined (Supplemental Figure 12). Decreased oil content is genetically exclusively linked to the inactivation of the PLIP1 locus, suggesting that the primary biological functions of PLIP2 and 3 are not in seed oil biosynthesis.

To test whether PLIPs contribute to oxylipin production in response to ABA, plants were first grown on filter papers placed on top of normal, agar-solidified MS medium for 2 weeks. Following transfer to MS agar-solidified medium containing 10 µM ABA, quantification by LC-MS showed that both JA and JA-Ile were significantly elevated after 24-h treatment (Figures 9D and 9E) in the wild type. However, their accumulation was abolished in the two plip-tri lines and one plip-CRP line (Figures 9D and 9E), confirming that PLIPs are mediating ABA-induced JA production.

PLIP1,2,3 Triple Mutants Are Hypersensitive to ABA

Aside from their inability to produce JA in response to ABA application, plip-tri mutants were hypersensitive to ABA. With ABA in the medium, the germination of the plip single mutants was delayed, and more severely so for the plip-tri seeds, although the seeds were able to germinate within 2 d without application of ABA (Figure 10A; Supplemental Figure 13). ABA hypersensitivity was also observed during seedling establishment for the plip-tri plants as they showed reduced growth and paler rosettes compared with the wild-type seedlings when grown on ABA-containing MS medium for 2 weeks (Figure 10B). As established above, PLIP2 and 3 hydrolyze chloroplast membrane lipids providing 18:3 FA precursors for JA production when induced by ABA. Furthermore, decreased endogenous ABA levels were observed in the PLIP2-OX and PLIP3-OX plants, which have elevated levels of oxylipins (Figure 7A). To explain these observations, we hypothesize that JA antagonizes ABA anabolism or signaling, providing a feedback regulatory loop for the cooperative action of ABA and JA under certain conditions.

Figure 10.

The plip-tri Plants Are Hypersensitive to ABA.

(A) Germination rates of the wild-type and plip-tri seeds with or without 2 µM ABA addition on agar-solidified medium. Observation of 100 seeds in one plate was treated as one biological repeat. n = 4, ±sd. Student’s t test was applied to compare the wild type with each of the remaining genotypes (**P < 0.01).

(B) Morphology of the 2-week-old wild-type and plip-tri seedlings grown on MS agar-solidified medium with 0.5 µM ABA or the equivalent ethanol solvent (Mock). Bars = 0.5 cm.

(C) ABA content in the 2-week-old wild-type, plip-tri, and plip-CRP seedlings during a 48-h time course following the addition of 10 µM MeJA to the MS agar-solidified medium. Three to five seedlings were pooled as one biological repeat. n = 3, ±sd. Student’s t test was applied to compare the wild type with each of the remaining genotypes (**P < 0.01).

To investigate how JA might affect ABA metabolism in wild-type and PLIP-compromised plants, we grew the plants on regular MS agar-solidified medium for 2 weeks and subsequently transferred the plants to medium containing 10 µM MeJA. During a time course from 1 to 48 h after transfer, endogenous ABA content in different genotypes was quantified (Figure 10C). ABA levels generally decreased with time in all the genotypes, suggesting JA might repress ABA accumulation. However, endogenous ABA levels were also found to be slightly lower in the PLIP-compromised plants compared with the wild-type plants following addition of 10 µM MeJA. This was not expected as ABA levels were also decreased in the PLIP-OX lines, which accumulate JA (Figure 7A).

DISCUSSION

PLIP2 and PLIP3 Share Biochemical Properties with PLIP1

PLIP2 and PLIP3 were discovered based on their sequence similarity to PLIP1. They are more closely related to PLIP1 compared with other α/β hydrolases in the Arabidopsis genome (Figure 1). All three contain a conserved Lipase 3 domain. Biochemical characterization of PLIP2 and 3 showed that they are also glycerolipid A₁ lipases (Figure 4; Supplemental Figure 3). Lipases tend to be promiscuous with regard to offered lipid substrates in vitro, but an in vitro competition assay on spinach chloroplast lipids indicated that PLIP2 and 3 favor different lipids, PLIP2 galactolipids, and PLIP3 PG (Figure 4; Supplemental Figure 3). By ectopically overexpressing the respective lipase cDNAs in the plant, the recombinant lipases under investigation gain access to their native lipid substrates in a quasi-native environment causing changes in the lipid composition indicative of their respective substrate preference in vivo. Lipidomic analysis of the PLIP2-OX and PLIP3-OX plants showed very similar steady state lipid changes (Figure 5A; Supplemental Figures 4 and 5) to that of the previously reported PLIP1-OX plants (Wang et al., 2017), corroborating their similarity in their enzymatic activities. Especially, the correlative changes for chloroplast lipids (MGDG and PG) with ER lipids (phosphatidylinositol, phosphatidylethanolamine, and PC) suggest that all three PLIPs can efficiently release polyunsaturated FAs that are subsequently exported from chloroplasts to the ER. However, it seems unlikely that PLIP2 and 3 work redundantly with PLIP1 in seed oil biosynthesis because the plip2 and 3 seeds did not show a change in seed oil content (Supplemental Figure 12).

Further pulse-chase labeling analysis suggested that different PLIP-OX plants are affected in different lipid pools in vivo (Figure 5; Supplemental Figure 6). This conclusion was further confirmed in the PLIP-OX; coi1 plants, which we analyzed to preclude the possibility that JA signaling might have an effect on lipid metabolism that could explain the stunted growth of the PLIP-OX lines (Figure 8; Supplemental Figure 8). In the PLIP2-OX plants, labeling and turnover of the MGDG pool was accelerated. On the other hand, the PG pool was labeled and turned over at an accelerated rate in the PLIP3-OX plants. These data imply that PLIP2 and PLIP3 preferentially access different substrates in vivo, PLIP2 the galactolipid MGDG, and PLIP3 the phospholipid PG. Acyl compositional analysis of MGDG showed reduced 16:3, a signature FA of the plastid-assembled MGDG in the PLIP2-OX plants, suggesting that prokaryotic MGDG is likely the native substrate of PLIP2. Although PLIP2 and PLIP3 can act on MGDG and PG in vitro, their distinct subchloroplastid location might give them preferential access to one or the other lipid explaining the observed apparent lipid preferences in vivo.

PLIP2 and PLIP3 Are Involved in JA Biosynthesis

Overexpressing any one of the three PLIP cDNAs causes a growth defect (Figure 6). However, this growth phenotype does not correlate with the changes in lipid composition and lipid metabolism. Hence, simple degradation of the membrane lipids due to increased PLIP lipase activity cannot explain this growth phenotype. Morphologically, the PLIP2-OX and PLIP3-OX plants resemble plants with activated JA responses (Figure 6). Indeed, JA-responsive marker genes were constitutively induced in the PLIP-OX plants, and their induction levels matched the severity of the growth reduction (Figure 6). Oxylipins including JA, its derivatives, and Arabidopsides accumulated in the PLIP2-OX and PLIP3-OX plants (Figure 7; Supplemental Figure 7). These results indicate that release of free FAs from the chloroplast membranes is likely the rate-limiting step for JA biosynthesis. Arabidopsides were previously observed at very high levels in vegetative tissues after wounding or during the pathogen-induced hypersensitive response (Andersson et al., 2006; Buseman et al., 2006). Even though some of the Arabidopside species have been shown to directly participate in plant defense (Kourtchenko et al., 2007), they are generally thought to provide a temporary, inert storage pool for OPDA (Koo et al., 2009; Mosblech et al., 2009). The biosynthesis of Arabidopsides is still a matter of debate, but in the PLIP2-OX and PLIP3-OX plants, formation of Arabidopsides possibly serves to adjust the levels of free OPDA as a result of increased 18:3 release by the PLIPs, implying that lipoxygenases would have to act on free FAs. After PLIP2 or PLIP3 release the polyunsaturated FAs, free OPDA readily accumulates in the plastids following oxidation of free 18:3 FA (Figure 11). Free OPDA is converted to biologically active oxylipins requiring strict control of its levels, but may also be detrimental to the cell in high enough quantities simply due its detergent properties. Presumably, a fraction of the free OPDA is converted to OPDA-acyl carrier protein, which serves as substrate for the plastid acyl transferases that either assemble membrane lipids de novo or catalyze acyl exchange on existing membrane lipids giving rise to the observed Arabidopsides. This mechanism likely provides a buffer for the fine-tuning of JA/JA-Ile levels along with catabolism of the hormone (Figures 6 and 7).

Figure 11.

Proposed Function of PLIP2 and 3 in ABA-Induced JA Biosynthesis.

ABA that is applied or synthesized upon sensing abiotic stressors induces the expression of PLIP2 and 3. Mature PLIP2 and 3 proteins are located in the chloroplasts and release polyunsaturated fatty acids (PUFAs) from the chloroplast membranes. PUFAs are metabolized to OPDA, which is subsequently exported from the chloroplast and eventually converted to JA and its bioactive derivative JA-Ile. JA-Ile triggers the degradation of JAZ proteins together with COI1 in the nucleus, derepressing transcription factors, e.g., MYC2, which initiates gene expression and turns on the JA responses, such as repressing the vegetative growth and turning on biotic defenses. JAZ, jasmonate ZIM domain (JAZ) transcriptional repressors; SCF^COI1, SKP1-CUL1-F-box (SCF) CORONATINE INSENSITIVE1 (COI1).

The growth repression phenotype of the overexpression lines was specifically caused by turning on JA biosynthesis leading to constitutive JA signaling, as introduction of the coi1 mutation into PLIP2-OX and PLIP3-OX plants, which abolishes JA perception in plants, restored their growth irrespective from direct effects on membrane lipids (Figure 8). Taken together, we hypothesize that PLIP2 and PLIP3 might be involved in initiating JA biosynthesis under conditions where these enzymes are overproduced, for example, following activation of their respective genes in response to environmental cues (Figure 11).

PLIP2 and PLIP3 Link ABA Signaling with JA Biosynthesis

As the three PLIPs possess similar biochemically characteristics, the question arises how they exert their distinct physiological roles. One possible explanation is their different subplastid locations. PLIP1 is specifically associated with the thylakoid membranes (Wang et al., 2017). Based on our current analysis, PLIP3 is a membrane protein associated with both the envelope and thylakoid membranes, while PLIP2 is found in the soluble and membrane fractions and seems to be ubiquitously present in envelope membranes, stroma and thylakoids (Figure 2). Possibly, PLIP2 and PLIP3 are enriched in microenvironments that are in close proximity to other enzymes required for oxylipin synthesis, like lipoxygenase, allene oxide synthase, and allene oxide cyclase to readily convert free FAs to OPDA and related oxylipins.

Another factor that diversifies their physiological functions might be the distinct transcriptional regulation of their respective genes. PLIP1 is specifically expressed during seed developmental stages and has much lower basal expression levels in the leaf tissues compared with PLIP2 and PLIP3. In fact, deficiency of PLIP2 and PLIP3 did not affect seed oil biosynthesis (Supplemental Figure 12). In addition, the expression of PLIP2 and PLIP3 in leaf tissues is ABA responsive, but that of PLIP1 is not (Figure 9). Moreover, despite the fact that both PLIP2 and 3 can be induced by supplied ABA, public transcriptomic data show that they are induced by different abiotic stresses in vivo (Figure 9), implying that each of the two genes responds to factors sensing different stressors.

Based on the ABA responsiveness of their genes and their potential in initiating JA biosynthesis when overproduced, we propose that PLIP2 and PLIP3 can mediate ABA induction of JA biosynthesis in response to certain environmental cues that activate ABA production (Figure 11). In fact, the plip triple mutant lines exhibited compromised oxylipin biosynthesis in response to supplied ABA (Figure 9), indicative of their potential in mediating ABA responsiveness of JA biosynthesis, when endogenous ABA levels rise naturally. JA production is most commonly induced by tissue damage caused by insect herbivory or infection by necrotrophic pathogens (Creelman et al., 1992; Trusov et al., 2006). A growing body of evidence suggests that JA biosynthesis is also stimulated by various abiotic stress conditions, including high salinity and conditions leading to ABA accumulation (Creelman and Mullet, 1995; Adie et al., 2007; Avramova, 2017). It is therefore conceivable that distinct lipases mediate the release of FA precursors of JA biosynthesis in response to different stresses and/or in specific cell types or tissues. Responsiveness of PLIP2 and 3 expression to ABA and their ability to initiate JA biosynthesis provide a mechanism linking plant abiotic and biotic stress responses. The transcription factor MYC2 is a potential component in both ABA- and JA-mediated physiological responses given its involvement in both JA perception and ABA signaling (Kazan and Manners, 2013). Thus far, it is still unknown how or even whether MYC2 contributes to ABA-induced JA production, although a recent study reports that MYC2 interacts directly with an ABA receptor (PYL6) to modulate ABA and JA responses (Aleman et al., 2016). Even though ABA transcriptionally induces MYC2, MYC2 activity is theoretically repressed by JAZ proteins until the biosynthesis of JA commences. Because the expression of PLIP2 and 3 is not induced by MeJA based on published transcriptomic data, MYC2 is likely not directly involved in the transcriptional induction of the two genes, PLIP2 and PLIP3. It is important to explore further the unique timing and range of responsiveness to different environmental cues for each PLIP gene. In nature, plants continually encounter different adverse environmental conditions. Possibly, the PLIP genes evolved allowing plants to cope with different abiotic/biotic stresses and integrate their protective responses for example during coinciding drought conditions and insect outbreaks (Suzuki, 2016). At this time, we did not observe compromised Botrytis resistance in the plip-tri mutant lines grown under normal conditions, but to implicate PLIPs in linking biotic and abiotic stresses, it would be important to test plip-tri mutant defense in response to a combination of abiotic and biotic stresses.

Another possible physiological function of PLIP2 and 3 may be their role in fine-tuning ABA-related developmental processes. Aside from their inability to induce JA biosynthesis in response to ABA application, the plip single and triple mutants were also hypersensitive to ABA during germination and early seedling development (Figure 10; Supplemental Figure 13). Application of JA to the wild-type and the plip-tri mutant seedlings led to a trend of gradually decreasing ABA levels in all the genotypes over time (Figure 10). Decreased ABA content was also observed in the PLIP2-OX plants (Figure 7). Therefore, one possible explanation for the ABA hypersensitivity phenotype might be that JA negatively regulates ABA biosynthesis. However, given that ABA levels were lower compared with the wild type for both, the PLIP-OX plants treated with MeJA and the plip-tri mutant, other possibilities cannot be precluded at this time. For example, besides ABA metabolism, it is important to investigate whether PLIP2 and PLIP3 are directly or indirectly involved in ABA signaling. It is also possible that other oxylipin products as a result of PLIP2 and PLIP3 activation serve as negative regulators of ABA signaling, a possibility that still needs to be further explored.

METHODS

Plant Material and Growth Conditions

All experiments were performed in the Arabidopsis thaliana Col-0 ecotype. The T-DNA insertion lines SALK_1234548 (plip2-1), SALK_134525 (plip2-2), SALK_006633 (plip3-1), and SALK_115539 (plip3-2) were obtained from the Arabidopsis Biological Resource Center, Ohio State University. For the overexpression lines, the coding sequences of PLIP2 or PLIP3 were inserted into the pENTR/TEV/D-TOPO vector (Life Sciences; catalog no. K240020) before recombining them into the pEarleyGate 101 vector (YFP at the C terminus) (Earley et al., 2006) with the Gateway LR Clonase Kit (Thermo Fisher Scientific; catalog no. 11791-020). The final constructs were introduced into Col-0 wild-type plants by Agrobacterium tumefaciens-mediated floral dip (Clough and Bent, 1998). Transformed seeds were initially screened for resistance to Basta, followed by confirmation by RT-PCR. Primers used for genotyping of T-DNA insertion lines or for RT-PCR analysis of overexpression lines are given in Supplemental Table 1. The MS medium-grown and the soil-grown Arabidopsis plants were incubated under 100 µmol m⁻² s⁻¹ (Sylvania FO/741/ECO) in a 16-h-light/8-h-dark cycle. All other growth conditions were the same as previously reported (Wang et al., 2017).

Phylogenetic Analysis

The top 17 PLIP1 putative paralog amino acid sequences were obtained by comparing the Arabidopsis PLIP1 protein sequence against the Arabidopsis proteome using the BLASTp program in TAIR BLAST 2.2.8 (http://www.arabidopsis.org/Blast/index.jsp) with the default parameters. The amino acid alignment was created using the MUSCLE program with default settings employing MEGA (version 7.0.21) software. The maximum likelihood phylogenetic trees were built using MEGA (version 7.0.21) and the bootstrap percentages were based on 1000 replicates.

Confocal Laser Scanning Microscopy

The confocal images were taken with a Nikon A1Rsi confocal microscope on leaves of 3-week-old Arabidopsis grown on MS agar plates. YFP fusion proteins were detected with the configuration A1CYFR using a 514-nm laser for excitation and a 530- to 600-nm band-pass filter for fluorescence emission. For the chlorophyll autofluorescence, a 647-nm laser was used for excitation and a 660-nm long-pass filter for emission detection. Images were merged and pseudocolored using NIS-Elements AR software (version 4.30.01 64 bit).

Chloroplast Import Assay

Isolation of pea (Pisum sativum) chloroplasts, import assays, and postimport trypsin treatment were done essentially as previously described (Xu et al., 2005). The constructs used for production of labeled proteins were pET41a-PLIP2 and pET41a-PLIP3. N-terminal 6× His tag and TEV cleavage sites were removed using a Q5 site-directed mutagenesis kit (New England Biolabs).

Recombinant Protein Production

The PLIP2 (or PLIP3) coding sequence was amplified from the Arabidopsis wild-type cDNA and inserted into the pGEM-T-Easy plasmid (Promega). It was then inserted into the pET41a plasmid at the BamHI and XhoI restriction sites (BamHI and SalI for PLIP3). The PLIP2^S428A and PLIP3^S4385A point mutation constructs were generated with a Q5 site-directed mutagenesis kit (New England Biolabs). Final pET41a-PLIP2, pET41a-PLIP2^S428A, pET41a-PLIP3, and pET41a-PLIP3^S385A constructs were introduced into the BL21 (DE3) Escherichia coli strain for protein production. Cultures grown in LB medium (containing 0.1% glucose) were inoculated with fresh E. coli colonies and grown to log phase (OD₆₀₀ 0.8) at 37°C. Protein production was then induced by adding isopropyl-β-d-thiogalactopyranoside to the final concentration of 0.2 mM, and the culture was transferred to 14°C. Cells were harvested from 1 liter culture after 1.5 h of induction by centrifugation at 5000 × g for 10 min and resuspended in 10 mL of resuspension buffer, followed by sonication to lyse cells (Wang et al., 2017). For cultures expressing PLIP3 and PLIP3^S385A, these lysates were used directly for assays. For PLIP2 and PLIP2^S428A, the lysates were centrifuged at 10,000g for 30 min, resuspended in 1 mL of 1× PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄, pH 7.4), and subjected to another 1-h centrifugation at 100,000g followed by removal of the supernatant. The pellet was then resuspended with 1 mL PBS buffer to give rise to microsomes.

PLIP Lipase Assays

The lipase assay was performed essentially according to Wang et al. (2017). For each lipase assay, 100 µg dried lipids were first resuspended in 300 μL reaction buffer (0.1 M PBS, pH 7.4, 4.2 mM Anzergent 3-12 [Anatrace]) and dispersed by sonication. Then, 50 μL PLIP2 microsomes (equivalent to ∼70 µg total proteins) or PLIP3 cell lysates (equivalent to ∼600 µg total proteins) were added to each reaction followed by another 10-s sonication to mix. The reaction was incubated at ambient temperature (∼22°C) for 1 h or as indicated for time courses (Supplemental Figure 2). The reaction was stopped by lipid extraction, followed by lipid separation by TLC and quantification by gas chromatography as described below.

Lipid Analysis

Spinach chloroplast isolation was done as previously reported (Roston et al., 2011). Lipid extraction, TLC of polar and neutral lipids, transesterification, and gas chromatography were done as described (Wang and Benning, 2011). For polar lipids, activated ammonium sulfate-impregnated silica gel TLC plates (TLC Silica gel 60; EMD Chemical) were used for lipid separation with a solvent consisting of acetone, toluene, and water (91:30:7.5 by volume). Lipids were visualized by brief exposure to iodine vapor on TLC plates. Acyl groups of the isolated lipids were then converted to methyl esters, which were subsequently quantified by a gas chromatography.

Pulse-Chase Labeling

The pulse-chase labeling experiments on 4-week-old soil-grown plant leaves were performed essentially according to Wang et al. (2017).

Hormone Measurements and Oxylipin Profiling

Hormone extraction and measurements were done based on Wang et al. (2017). 12OH-JA and 12OH-JA-Ile were added as standards. Their transitions from deprotonated molecules to characteristic product ions were monitored (12OH-JA, m/z 225.09 > 58.90; 12OH-JA-Ile, m/z 338.23 > 130.09), and the collision energies and source cone potentials were optimized for each compound with the QuanOptimize software before integration into the LC-MS/MS method.

For oxylipin profiling, oxidized lipids were extracted and analyzed using a Waters Xevo G2-XS Q-TOF mass spectrometer according to Wang et al. (2017). The metabolite chromatograms from the wild-type and the PLIP-OX leaf samples were normalized and aligned using the Progenesis QI software (version 2.2) to identify peaks accumulated in the PLIP-OX, but not the wild-type samples. Identification of the Arabidopside metabolites was based on their accurate masses and the signature OPDA or dn-OPDA fragment ions (m/z values are 291.19 and 263.16, respectively). The peak areas of the Arabidopsides were normalized to that of the 18:3/18:3-DGDG ([M-H]-, m/z 981.61).

Accession Numbers

Sequence data from this article can be found in the Arabidopsis TAIR database (https://www.arabidopsis.org/) under the following accession numbers: At3g61680 for PLIP1, At1g02660 for PLIP2, At3g62590 for PLIP3, At2g44810 for DAD1, At1g51440 for DALL2 (DAD1-LIKE LIPASE 2), At2g30550 for DALL3, At1g06800 for DALL4, At1g05800 for DONGLE, At5g04040 for SDP1 (SUGAR-DEPENDENT1), At1g45201 for ATTLL1 (ARABIDOPSIS THALIANA TRIACYLGLYCEROL LIPASE-LIKE1), At2g26560 for pPLA-IIα, At2g39220 for pPLA-IIIα, At5g24780 for VSP1, At3g45140 for LOX2, At1g32640 for MYC2, At1g19180 for JAZ1, At5g13220 for JAZ10, and AT2G27690 for CYP94C1.

Supplemental Data

• Supplemental Figure 1. Alignment of PLIP2 and PLIP3 with class 3 lipase protein sequences in the NCBI conserved domain database.

• Supplemental Figure 2. Time course of PLIP2 and 3 lipase activities with different lipid substrates.

• Supplemental Figure 3. In vitro lipase activity of PLIP3.

• Supplemental Figure 4. Analysis of polar lipids from the 4-week-old soil-grown PLIP2-OX plant leaves.

• Supplemental Figure 5. Analysis of polar lipids from the 4-week-old soil-grown PLIP3-OX plant leaves.

• Supplemental Figure 6. In vivo pulse-chase acetate labeling of polar lipids in the empty vector control and the PLIP3-OX plants.

• Supplemental Figure 7. Phytohormone quantification and oxylipin profiling in the wild-type and PLIP3-OX plants.

• Supplemental Figure 8. Phenotypes of PLIP3-OX; coi1 plants.

• Supplemental Figure 9. Analysis of polar lipids from the jazQ plants.

• Supplemental Figure 10. Growth and lipid phenotype of the plip single and triple mutants.

• Supplemental Figure 11. Genotyping of the plip-CRISPR line.

• Supplemental Figure 12. Seed oil content of the plip mutants.

• Supplemental Figure 13. Germination of the plip single mutants in the presence of ABA.

• Supplemental Table 1. Primers used in this study.

• Supplemental File 1. Sequence alignment of the PLIP1 homologous sequences in Arabidopsis.

• Supplemental File 2. ANOVA tables.