Plastid Phosphatidylglycerol Homeostasis Is Required for Plant Growth and Metabolism in Arabidopsis thaliana

A unique feature of plastid phosphatidylglycerol (PG) is a trans-double bond specifically at the sn-2 position of 16C fatty acid (16:1t- PG), which is catalyzed by FATTY ACID DESATURASE 4 (FAD4). To offer additional insights about the in vivo roles of FAD4 and its product 16:1t-PG, FAD4 overexpression lines (OX-FAD4s) were generated in Arabidopsis thaliana Columbia ecotype. When grown under continuous light condition, the fad4-2 and OX-FAD4s plants exhibited higher growth rates compared to WT control. Total lipids were isolated from Col, fad4-2, and OX-FAD4_2 plants, and polar lipids quantified by lipidomic profiling. We found that disrupting FAD4 expression altered prokaryotic and eukaryotic PG content and composition. Prokaryotic and eukaryotic monogalactosyl diacylglycerol (MGDG) was up-regulated in OX-FAD4 plants but not in fad4-2 mutant. We propose that 16:1t-PG homeostasis in plastid envelope membranes may coordinate plant growth and stress response by restricting photoassimilate export from the chloroplast.


Introduction
Plant fatty acids (FAs) are mainly synthesized de novo in the plastid stroma [1], then exported into the cytoplasm. According to acyl-ACP synthesis rates and specificity of thioesterases in Arabidopsis plastids, oleic acid (18:1) is the major FA exported from the plastid, followed by palmitic acid (16:0) and trace amounts of stearic acid (18:0) [2,3]. Plastid membrane lipids are synthesized by both the prokaryotic and eukaryotic pathways [4]. The prokaryotic pathway assembles the synthesized FAs de novo in the plastid; the eukaryotic pathway assembles the plastid-exported FAs into lipids within the endoplasmic reticulum (ER). Part of the ER-assembled lipids are trafficked back to the plastid, where they are converted to monogalactosyldiacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), and sulfoquinovosyldiacylglycerol (SQDG). Due to the difference in substrate specificities of acyltransferases at the ER and the plastid envelope [5], glycerolipids originating from the prokaryotic pathway have a 16-carbon acyl chain at the sn-2 position of the glycerol backbone, while glycerolipids assembled on the ER carry an 18-carbon acyl chain at the same position [6].
In plant cells, plastid phosphatidylglycerol (PG) is the major phospholipid of the plastid envelope and thylakoid membranes and is also a minor component in other cellular membranes [7]. The plastid, ER, and mitochondria organelles are each capable of synthesizing PG from phosphatidic acid (PA) [8,9]. The plastid pathway for PG synthesis is located in the inner envelope (IE) membrane [10] (Figure 1). CDP-DAG synthase transfers the cytidylyl group from CTP to PA to form CDP-DAG [11]. PGP synthase 1 (PGP1) converts CDP-DAG and glycerol-3-phosphate (G3P) into phosphatidylglycerol phosphate (PGP) [9], then plastid PGP phosphatase 1 (PGPP1) converts PGP into PG [8]. The plant ER also possesses all enzyme activities for de novo PG synthesis [12], and PGP2 could be responsible for PG synthesis in ER [13]. The PC synthesized in ER could be transported into plastid inner envelope membrane through TGD complex, then converted into PG. Mitochondrial PG synthesis is mediated by a PGP1 isoform that targets to the mitochondria [14]. In addition, ER-assembled PG could be imported into the mitochondria to support cardiolipin synthesis in the inner mitochondria membrane [13,14]. Thus, PGP2 shows functional redundancy with the mitochondria PGP1 isoform even though they locate to different subcellular compartments. A unique feature of plastid PG is a trans-double bond at the sn-2 position of the 16C fatty acid (16:1t-PG), introduced by FAD4 using PG (18:x/16:0) substrates but not PA or PGP as substrates [15,16]. FAD4 is a thylakoid-associated protein facing the stromal side [17] (Figure 1), and chloroplast PG is produced at the IE membrane [9,13,17]. Thus, plastid PG synthesis and desaturation are spatially separated. Plastid PG has been implicated in multiple functions, including chloroplast protein import [18], stabilization of PSI trimers [19], PSII core dimers [20], and light-harvest complex II. A leaky mutation for plastid PGP1 showed a reduction in total PG content of~30%, leading to a pale green leaf color and impaired photosynthetic light energy utilization [9]. Complete knockout of the PGP1 gene leads to the abolishment of chloroplasts and loss of photoautotrophy [14]. In the sqd2pgp1-1 double mutant, the fraction of total anionic lipids (PG + SQDG) was reduced by about one-third and accompanied by multiple defects, including a pale-yellow leaf color, compromised photoautotrophic growth, reduced numbers of mesophyll cells, and an altered chloroplast ultrastructure. Thus, the authors concluded that anionic lipids PG and SQDG play critical roles in the proper structure and function of photosynthetic membranes for seed plants [21]. Unlike PGP1 knockout plants [14], FAD4 abolishment showed no significant effect on the stability of the chlorophyll-protein complexes to temperature-induced dissociation [21]. Knockdown of a chloroplast PGPP1 reduced plastid PG (34:4) and the other 34:x molecular species of PG without affecting the molecular species of PG 32:0 and 32:1 [16]. This mutant shifted glycolipid synthesis towards chloroplast-derived species of DGDG 34:3 and reduced DGDG 36:6 which is derived from ER-assembled precursors, though levels of PC, PE, and PI were not affected [16]. The pgpp1-1 mutant plants showed reduced chlorophyll content, which did not affect quantum yield [16]. Interestingly, the thylakoid-associated redox protein PRXQ is required for FAD4 activity [17]. The prxq mutant exhibited a reduction in 16:1t-PG [22] and increased sensitivity to oxidants [23]. Thus, 16:1t-PG is proposed to be involved in sensing of the redox status of chloroplasts [23].
Since 16:1t-PG is found throughout the plant kingdom, synthesized in green tissue, and associated with photosynthetic machinery, it has been proposed that 16:1t-PG has a fundamental role in eukaryotic photosynthesis [24]. However, complete abolishment of 16:1t-PG does not affect photosynthetic antenna function under standard growth conditions [25]. Genetic dissection of the lipid bilayer composition provides essential in vivo evidence for the role of individual lipid species in membrane function [14]. Lipidomics are increasingly applied to study pathway perturbations in various settings that implicate dysregulation in lipid metabolism [26,27], leading to a novel understanding of the connections between lipids and phenotypes [26,[28][29][30]. To offer additional insights about the in vivo role of the FAD4 gene and 16:1t-PG, we generated FAD4-overexpressing plants (OX-FAD4s). We measured their growth rates, and quantitated polar lipids changes by lipidomic profiling. Our data suggest that FAD4 is involved in the synthesis of PG (36:7); in addition, 16:1t-PG homeostasis in plastid envelope membranes could regulate plant growth by effecting photoassimilate export from the plastid.

Plant Growth
There are three FAD4 knockout lines (fad4-1, fad4-2, and fad4-3) reported before [15]. FAD4 protein possesses three histidine motifs (170QGHH173, 229HAWAH233, and 258HAEHH262) which could be responsible for the activity of membrane-bound desaturases. Thus, for a complete knockout of FAD4 activity, it is ideal to delete all the three histidine motifs. The fad4-1 presumably produce a truncated protein of 177 amino acids which still retain the first histidine motif; the T-DNA in fad4-2 (SAIL_1250_C12) was inserted in amino acid residue 132 of the inferred amino acid sequence; while the T-DNA in fad4-3 mutant was inserted in amino acid residue 287 in which all three histidine motifs were retained. We reasoned that the fad4-2 mutant is the only line that all histidine motifs are removed; thus, it was selected in this study. Plants were grown in a composite of peat soil: vermiculite: perlite (8:1:1, v/v/v); temperature 22 • C; light intensity 160 µmol photon m −2 s −1 ; humidity 60%.

Overexpression of the FAD4 Gene in Arabidopsis
FAD4 CDS was amplified from Col total RNA by RT-PCR using a pair of gene-specific primers, with Bam HI and Xho I recognition sequences added into the forward and reverse primers, respectively. The primer sequences were: 5 -CGGGATCCATGGCTGTATCACTTCC-3 and 5 -CCGCTCGAGTTATGCTTGGTTGTTGG-3 . The amplified PCR fragment was digested by Bam HI and Xho I, then was cloned into the pMGmubi vector, in which the FAD4 gene was driven by a soybean ubiquitin promoter. The plasmid construct was introduced into Agrobacterium tumefaciens strain GV3101, then transformed into Col by the floral dip method. T 1 seeds were surface sterilized and germinated on 1x MS plates containing 50 µg mL −1 kanamycin. Kanamycin-resistant seedlings were transferred into soil and allowed to self-fertilize. T 2 seeds were harvested from individual plants, then germinated on 1x MS plates containing 50 µg mL −1 kanamycin, six resistant seedlings per line were transferred into soil and self-fertilized, T 3 seeds were harvested individually. To identify the homozygote transgenic lines, T 3 seeds were sterilized and germinated on 1x MS plates containing 50 µg mL −1 kanamycin, the lines that show uniform kanamycin resistance were judged as homozygote, then used for physiological and biochemical analysis.

Plant Growth Parameter Measurement
Seeds of Col, fad4-2, OX-FAD4_2 and OX-FAD4_5 were surface sterilized and sawed onto 1x MS plates, kept in 4 • C freezer for 2 days, then moved into continuous white light for 9 days. The seedlings were transplanted into soil and grown under continuous light for an additional 15 days before photography. The projected leaf area was calculated by using Image J software. The number of leaves were counted from the individual plant. The above-ground rosette leaves from individual plants were harvested and weighed by a balance (Jing-Tian, Shanghai, China) to obtain a fresh weight, then dried overnight in a 105 • C oven to obtain the dry plant weight.

Total Lipid Extraction
Col, fad4-2, and OX-FAD4_2 plants were grown in soil under continuous light for three weeks, above-ground tissues were harvested, total lipids were immediately extracted using the single step extraction method, as described previously [31], five biological replicates were prepared. This extraction method divides plant mass into lipid fraction and insoluble residue. The lipid fraction was completely dried under a gentle nitrogen stream to obtain dry lipid weights, then stored at −80 • C before lipidomic profiling; the insoluble residue was dried overnight in a 105°C oven to obtain dry residue weight. Total plant dry mass = dry lipid weight + dry residue weight.

Lipidomic Profiling
Phospholipid and galactolipid internal standards were used for individual lipid class identification and quantification [32]. At least two different lipid species per lipid class were used. Internal standards and their acquisition information are provided in Table S1. To prepare the samples for MS analysis, for each sample, the volume corresponding to 7.5 to 9.6 µg dried lipid was transferred to a vial containing the internal standards indicated in Table S1. Samples were brought to 1 mL by adding chloroform: methanol: 300 mM ammonium acetate in water (30/66.5/3.5, v/v/v) for mass spectrometric analysis.
The samples were analyzed on an electrospray ionization tandem mass spectrometer (ESI-MS/MS) (Waters Xevo TQS mass spectrometer; Waters Corporation, Milford, MA, USA) using sequential precursor and neutral loss scans (Table S2) and processed, as described previously [33]. The source temperature was 150 • C, desolvation temperature was 250 • C, cone gas flow was 150 L/h, collision gas (argon) flow was 0.14 mL/min, nebulizer gas was at 7 Bar, the LM 1 resolution was set at 3.2, and the HM resolution was set at 15.5. With the analytical platform used (Waters Xevo TQS), the intensity observed for galactolipid species varies in relation to the number of double bonds in the acyl chains. The data were corrected for the response variation using the response factors indicated in Table S3.

Statistical Analysis
Average and standard errors were calculated in Excel software (Microsoft, Seattle, WA, USA). One-way ANOVA was used to determine the significance based on Duncan's multiple range tests in SPSS software (V22; IBM, Armonk, NY, USA). Gao et al. (2009) demonstrated that the FAD4 enzyme introduces a ∆ 3-trans double bond to palmitic acid esterified at the sn-2 position of PG [15] and is the key enzyme for PG (34:4) (PG18:3/16:1t) synthesis through the prokaryotic pathway. To offer additional insights about the functions of FAD4 and its products, we overexpressed the FAD4 CDS sequence driven by a soybean ubiquitin promoter in the Col background. To confirm whether transgenic FAD4 was properly expressed, total RNA was isolated ( Figure 2A) and reverse transcribed, then RT-qPCR was conducted. The FAD4 expression levels in OX-FAD4 lines were 7-10 fold higher than that of the WT control ( Figure 2B). Surprisingly, the FAD4 level in fad4-2 knockout plants was similar to the WT control when F1/R1 primer sets were used for RT-qPCR ( Figure 2B,C). To test whether T-DNA insertion in fad4-2 affects its FAD4 mRNA processing, the primer set F2/R1, which flanks the T-DNA insertion site, was used for regular RT-PCR amplification ( Figure 2C). The F2/R1 primer set specifically amplified a fragment with an expected size from the WT control but failed from the fad4-2 sample ( Figure 2C). These data suggest that the T-DNA insertion in fad4-2 produces a long FAD4 chimera mRNA, which may translate into a defective FAD4 protein. These data are in accordance with Gao et al. (2009) who reported that fad4-2 showed similar defects on the PG profile as other fad4 mutant alleles [15].

Overexpression or Knockout of FAD4 Enhances Plant Growth
To characterize whether FAD4 expression levels affect plant growth, WT, fad4-2, and two OX-FAD4s (OX-FAD4_2 and OX-FAD4_5) were grown under continuous white light condition. OX-FAD4_2 and OX-FAD4_5 showed similar expression levels of FAD4 ( Figure 2B); thus, were chosen for phenotypic characterization in order to minimize the potential gene dosage effects on phenotypic variations. Col plants were visibly smaller compared to fad4-2 or OX-FAD4s ( Figure 3A). Physiological parameters, including leaf number per plant, projected leaf area, fresh plant weight, and dry plant weight, were measured. At same stage, fad4-2 and OX-FAD4s plants had more leaves ( Figure 3B), suggesting that the disruption of FAD4 expression enhanced plant vegetative growth rates. The projected leaf area, fresh weight, and dry weight of fad4-2 and OX-FAD4s also were higher than the WT control ( Figure 3C-E).

Disrupting FAD4 Homeostasis Reduced Total Lipid Content but Increased Polar Lipid Proportion
To dissect the functions of FAD4 for 16:1t-PG and other potential lipid synthesis, we decided to choose fad4-2 as one representative FAD4 knockout line, and OX-FAD4_2 as one representative FAD4 overexpression line. We reasoned that through a small scale multiple parallel comparison of lipid changes among WT, fad4-2 and OX-FAD4_2, we might be able to validate its established functions and discover novel roles. Thus, the total lipids were extracted from Col, fad4-2, and OX-FAD4_2 by the single step extraction method, as described previously [31]. The total lipid to dry plant mass ratios from fad4-2 and OX-FAD4_2 were reduced by~25% compared to the WT control ( Figure 4A). In WT, this ratio was~0.4, which is higher than expected for plant vegetative tissue. Considering that the solvent used for lipid extraction contains chloroform: isopropanol: methanol: water (22.5/25/31.1/2.6, v/v/v/v), we speculate that most polar metabolites could also be extracted into the so-called "total lipids". Under such a scenario, the residue left after lipid extraction mainly represents insoluble plant materials such as cell wall components (cellulose, semicellulose, and lignin, etc.). Multiple factors could be accountable for the reduced ratio in fad4-2 and OX-FAD4 plants, including reduced polar metabolite, elevated cellulose or lignin synthesis, reduced polar lipid or non-polar lipid content. Additionally, this could be due to developmental differences between mutant and wild type. Polar lipids from Col, fad4-2, and OX-FAD4_2 were analyzed by lipidomic profiling of 157 polar lipid species, the ratio of total polar lipid to the total lipid dry mass were calculated. We found that the polar lipid proportions in fad4-2 and OX-FAD4_2 were 50% and 73% higher than that of WT control ( Figure 4B), indicating that polar lipid synthesis in fad4-2 and OX-FAD4_2 were not decreased.

Lipidomic Profiling Confirmed FAD4 Enzyme Activities for Prokaryotic PG Synthesis
To compare polar lipid changes among WT, fad4-2, and OX-FAD4_2, each polar lipid species was normalized to plant dry weight and data expressed as nmol mg −1 plant dry weight (Table S4). In addition, polar lipid values were transformed into mol% by normalizing to the total polar lipid content of the sample, which represents the overall polar lipid composition of the plant cell (Table S5).

Disruption of FAD4 Expression Reduced LysoPG Contents
Even though individual molecular species of PG (32), PG (34) and PG (36) from fad4-2 and OX-FAD4_2 showed predictable changes compared to WT plants ( Figures 5 and 6), the total subpool sizes of PG (32), PG (34), PG (36) did not show any significant differences ( Figure 7A). The total PG pool sizes as well as Mol% distributions were also similar among WT, fad4-2 and OX-FAD4_2 plants ( Figure 7B, upper panel). However, lysoPG pool sizes in fad4-2 and OX-FAD4_2 were consistently lower than that of WT regardless of the normalization method ( Figure 7B, lower panel).

Disruption of FAD4 Expression Did Not Affect PC, PE, PI, PS and PA Contents or Compositions
To evaluate the potential roles of FAD4 for other polar lipid metabolism, the contents of PC, PE, PI, PS, and PA were compared among WT, fad42, and OXFAD4_2 plants. Our data demonstrated that these polar lipids were not significantly affected by knockout or overexpression of the FAD4 gene in Arabidopsis (Figure 9).
Our lipidomic profiling experiment demonstrated that disturbing FAD4 expression effectively altered PG content and composition (Table S4, Figures 5 and 6). Even though PG (34:4) synthesis was almost completely abolished in the fad4-2 mutant ( Figure 5B) and the PG (34:4) content increased by 70% in the OX-FAD4_2 plants ( Figure 5B), most other polar lipid classes exhibited similar trends between fad4-2 and OX-FAD4_2 (Figure 9), suggesting the high substrate specificity of FAD4 enzyme. One exception is prokaryotic and eukaryotic MGDG synthesis, which was differential between fad4-2 and OX-FAD4_2 lines ( Figure 8D). Considering that the trans-double bond containing PGs including PG (32:1), PG (34:4), and PG (36:7) were the major PG lipids up-regulated in OX-FAD4_2 compared to WT (Figures 5 and 6), we speculate that trans-double bond-containing PGs could influence MGD1 or MGD2 activities on the plastid envelope membrane.
Interestingly, we found that knockout or overexpression of FAD4 led to enhanced plant growth and accompanied with reduced total soluble metabolites in vivo ( Figure 4A). These data were in accordance with the previous observation that biomass is negatively correlated with the intermediates of central metabolic pathways [45]. Plastids are the only site for photosynthesis, which provides the sole carbon source to feed plant growth. The photosynthetic end product (3-PGA) exits the plastid stroma to the cytoplasm through TPT translocator, then converts into sucrose and transport into other sink organs. Meanwhile, plastids are also the site for several anabolic pathways, including fatty acids, transit starch, shikimate pathway, aromatic amino acids, and phytohormones (GA, ABA), etc. Since plants function as integrated systems, the distribution of metabolites between growth, and production of defense and storage compounds has to be tightly regulated [45]. It is speculated that photoassimilate export could be the first check point to balance plant growth and defense. Previous research demonstrated that plant growth is limited to a submaximum level to enable plants to cope with unfavorable conditions [46]. To support this, 16:1t-PG has been associated with stress response. For example, thylakoid membrane-associated PRXQ mutation led to~75% reduction of 16:1t-PG [17]. Accordingly, the Arabidopsisi prxq mutant plants showed increased sensitivity to oxidants [23]; in copea, a cold-sensitive plants, 16:1t-PG was negatively associated with the robustness of photosynthesis and contributed to chilling sensitivity [47]. These studies demonstrated that 16:1t-PG plays important roles in coordinating plant metabolism and stress responses. Our data suggest that 16:1t-PG on plastid membranes could regulate the export of photoassimilate or other metabolites from chloroplast that affect plant growth. Although the underlying mechanisms still await further characterization, here we propose two potential routs for 16:1t-PG to fulfill such role: 1) plastid 16:1t-PG homeostasis could be a critical factor to determine membrane permeability coefficients for metabolites; 2) plastid 16:1t-PG could interact with plastidic translocators or membrane associated proteins, including the recently discovered CTI family of envelope membrane proteins, which directly affect acetyl-CoA carboxylase activity [48]. Disruption of PG homeostasis on plastid membranes could lead to accelerated photoassimilate export; thus, altering sink/source ratios and enhancing plant growth [49]. In return, the enhanced plant growth draws upon photoassimilate leading to lower apparent soluble metabolite levels (Figures 3 and 4A).

Conclusions
In this study, we generated FAD4 overexpression plants (OX-FAD4s), then evaluated the growth performance among WT, fad4-2 knockout mutant, and OX-FAD4_2 plants. We showed that knockout or overexpression of FAD4 led to enhanced plant growth, the total extractable soluble metabolite contents were negatively correlated with their enhanced growth. Lipidomic profiling of polar lipids showed FAD4 is involved in PG 32:1, PG 34:4 and PG 36:7 synthesis; while prokaryotic and eukaryotic MGDG was up-regulated only from OX-FAD4 plants but not fad4-2 knockout plants. This study provides novel insights about the roles of FAD4 on plastid PG homeostasis, plant growth and metabolism.