DIACYLGLYCEROL ACYLTRANSFERASE and DIACYLGLYCEROL KINASE Modulate Triacylglycerol and Phosphatidic Acid Production in the Plant Response to Freezing Stress

DGAT1 and DGK2/3/5 modulate the conversion of DAG to TAG and PA, respectively, and determine cold stress tolerance in Arabidopsis.

Abstract

Plants accumulate the lipids phosphatidic acid (PA), diacylglycerol (DAG), and triacylglycerol (TAG) during cold stress, but how plants balance the levels of these lipids to mediate cold responses remains unknown. The enzymes ACYL-COENZYME A:DIACYLGLYCEROL ACYLTRANSFERASE (DGAT) and DIACYLGLYCEROL KINASE (DGK) catalyze the conversion of DAG to TAG and PA, respectively. Here, we show that DGAT1, DGK2, DGK3, and DGK5 contribute to the response to cold in Arabidopsis (Arabidopsis thaliana). With or without cold acclimation, the dgat1 mutants exhibited higher sensitivity upon freezing exposure compared with the wild type. Under cold conditions, the dgat1 mutants showed reduced expression of C-REPEAT/DRE BINDING FACTOR2 and its regulons, which are essential for the acquisition of cold tolerance. Lipid profiling revealed that freezing significantly increased the levels of PA and DAG while decreasing TAG in the rosettes of dgat1 mutant plants. During freezing stress, the accumulation of PA in dgat1 plants stimulated NADPH oxidase activity and enhanced RbohD-dependent hydrogen peroxide production compared with the wild type. Moreover, the cold-inducible transcripts of DGK2, DGK3, and DGK5 were significantly more up-regulated in the dgat1 mutants than in the wild type during cold stress. Consistent with this observation, dgk2, dgk3, and dgk5 knockout mutants showed improved tolerance and attenuated PA production in response to freezing temperatures. Our findings demonstrate that the conversion of DAG to TAG by DGAT1 is critical for plant freezing tolerance, acting by balancing TAG and PA production in Arabidopsis.

Cold stresses, including chilling and freezing, are major environmental factors that severely restrict plant growth, development, productivity, and ecological distribution. Most temperate plants have evolved complex strategies, such as transcriptome reprogramming and biochemical membrane remodeling, to survive low temperatures (Thomashow, 1999). The INDUCER OF CBF EXPRESSION (ICE)-C-REPEAT BINDING FACTOR/DRE BINDING FACTOR1 (CBF/DREB1) transcriptional cascade is a well-established cold signaling pathway in Arabidopsis (Arabidopsis thaliana; Thomashow, 1999; Chinnusamy et al., 2007). The basic helix-loop-helix transcription factor ICE1 directly targets cis-elements in the CBF3/DREB1a and CBF2 promoters (Chinnusamy et al., 2007). In contrast, ICE2 (the paralog of ICE1) is associated with the transcriptional regulation of CBF1/DREB1b (Fursova et al., 2009).

Besides the core ICE-CBF/DREB1 module, plant cells undergo significant alterations of both plastidic and extraplastidic membrane lipids in response to cold temperatures, thus preventing membrane injury and cellular dehydration (Uemura et al., 1995; Welti et al., 2002; Moellering et al., 2010). Under chilling and freezing conditions, unsaturated fatty acid and phospholipid (PL) levels increase rapidly to maintain the fluidity and integrity of membrane structure, thereby enhancing plant tolerance to such stresses (Kuiper, 1970; Welti et al., 2002). In Arabidopsis, two phospholipase Ds (PLDs), PLDα1 and PLDδ, are involved in the cold-induced hydrolysis of membrane PLs to produce phosphatidic acid (PA; Welti et al., 2002; Li et al., 2004; Rajashekar et al., 2006). The PLDα1 knockout mutants have reduced cold-responsive accumulation of PA and enhanced freezing tolerance (Welti et al., 2002; Rajashekar et al., 2006), but the PLDδ knockout mutants have decreased freezing tolerance (Li et al., 2004). The opposite freezing sensitivities observed between the PLDα1 and PLDδ knockout mutants are likely caused by PLDα1-triggered reactive oxygen species (ROS) production and PLDδ-mediated ROS mitigation (Zhang et al., 2003, 2009b). Interestingly, both PLDα1- and PLDδ-mediated plant cold sensitivities are independent of the CBF/DREB pathway (Welti et al., 2002; Li et al., 2004; Rajashekar et al., 2006), which indicates an alternative mechanism in regulating cold tolerance in plants. Previous findings further reveal that a deficiency of SENSITIVE TO FREEZING2 (SFR2), a gene encoding a galactolipid-remodeling enzyme located in the chloroplast outer envelope membrane, confers an extremely freezing-sensitive phenotype (Thorlby et al., 2004; Fourrier et al., 2008; Moellering et al., 2010). Under freezing conditions, SFR2 transfers the galactosyl residues from monogalactosyldiacylglycerol (MGDG) to different galactolipid acceptors to produce oligogalactolipids and diacylglycerol (DAG).

Increasing evidence suggests that the polar lipid PA, which lacks a head group, can form a destabilized hexagonal II (H_II)-type phase with MGDG or DAG (Steponkus, 1984; Uemura et al., 1995; Thomashow, 1999; Welti et al., 2002; Moellering et al., 2010; Moellering and Benning, 2011), damaging the cell membrane during freezing-induced dehydration. Conversely, the removal of these lipids may prevent H_II membrane formation and improve recovery from cold stress.

DAG is likely converted to triacylglycerol (TAG) by the enzyme ACYL-COENZYME A:DIACYLGLYCEROL ACYLTRANSFERASE (DGAT), which may play an important role in the removal of excess DAG after freezing (Moellering et al., 2010; Moellering and Benning, 2011). Moreover, recent studies suggest that, in response to cold stresses, DAG can be phosphorylated by DIACYLGLYCEROL KINASE (DGK) to produce PA (Gómez-Merino et al., 2004; Arisz et al., 2013). Our recent work indicates that the disruption of three genes encoding lipase-like proteins, SENESCENCE-ASSOCIATED GENE101 (SAG101), ENHANCED DISEASE SUSCEPTIBILITY1 (EDS1), and PHYTOALEXIN DEFICIENT4 (PAD4), confers salicylic acid (SA)-mediated freezing tolerance (Chen et al., 2015b). The mutants showed significantly lower levels of DAG and PA and higher levels of TAG than the wild type, suggesting that SA is likely involved in the regulation of cold-responsive lipid remodeling in plants. Consistent with this, the application of SA suppresses the cold-induced expression of DGATs and DGKs in wild-type leaves (Chen et al., 2015b). It is conceivable, therefore, that DAG may act as a common substrate for PA and TAG production upon cold exposure, the homeostasis of which may determine plant tolerance to such stresses. However, it is still unknown how the enzymes downstream of DAG contribute to the balance of TAG and PA production under freezing conditions.

The Arabidopsis genome has three DGAT genes (DGAT1–DGAT3) and seven DGK genes (DGK1–DGK7). DGAT1 is involved in TAG biosynthesis and lipid deposition in Arabidopsis seeds (Routaboul et al., 1999; Saha et al., 2006; Shockey et al., 2006; Hernández et al., 2012). However, the precise cellular functions of Arabidopsis DGKs remain elusive. In this study, we show that the dgat1 null mutants (dgat1-1 and dgat1-2) were more sensitive to chilling and freezing stresses compared with the wild type. Upon freezing exposure, lipid profile analysis revealed that TAG decreased significantly and DAG and PA accumulated more in the dgat1 mutants in comparison with the wild type. In addition, we observed that the knockout mutants of DGK2, DGK3, and DGK5 conferred enhanced tolerance to freezing stress. These results, together with the finding that freezing-induced PA accumulation declines significantly in the dgk mutants, suggest that the modulation of DAG conversion to TAG and PA by DGATs and DGKs, respectively, plays an important role in plant responses to cold stress.

RESULTS

DGAT1 Expression Is Cold Inducible

To evaluate the potential involvement of DGAT1 in the response to cold, 4-week-old wild-type (ecotype Columbia-0) plants were subjected to chilling (4°C) conditions, and the temporal pattern of DGAT1 expression was determined using reverse transcription quantitative PCR (RT-qPCR). The transcript of PHOSPHOLIPID:DIACYLGLYCEROL ACYLTRANSFERASE1 (PDAT1), another DAG acyltransferase in Arabidopsis, also was tested as a control. The results showed that DGAT1 expression was up-regulated from 3 to 96 h upon cold exposure, with levels increasing 197- and 43-fold in the shoots and the roots, respectively, at 96 h (Supplemental Fig. S1A). However, in the shoot tissues, cold stress (4°C) had little effect on the level of PDAT1 transcript, which was only elevated slightly at 1, 12, 24, and 48 h following cold treatment (Supplemental Fig. S1A). Consistent with our findings, data from a publicly available microarray database (http://www.bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi; Kilian et al., 2007) showed that the expression of DGAT1, but not PDAT1, was induced in response to cold stress (Supplemental Fig. S1B). Therefore, it appears that DGAT1 may play a role in the response to cold in Arabidopsis.

The dgat1 Mutants Exhibit Reduced Tolerance to Chilling and Freezing Stresses

To investigate the function of DGAT1 in plant responses to cold stress, two previously characterized DGAT1 mutants, dgat1-1 and dgat1-2 (Routaboul et al., 1999; Xu et al., 2012), were obtained and used for phenotypic analysis (Supplemental Fig. S2). Under normal growth conditions, the 3-week-old dgat1-1 and dgat1-2 mutants showed no visible growth defects in comparison with the wild type (Supplemental Fig. S3A, top). However, when 3-week-old plants were exposed to 4°C for another 3 weeks, the rosettes of both dgat1-1 and dgat1-2 mutants became browner and yellower compared with the wild-type rosettes under the same conditions (Supplemental Fig. S3A, bottom). Trypan Blue and 3,3′-diaminobenzidine (DAB) staining showed that more significant cell death and hydrogen peroxide (H₂O₂) accumulation occurred in the cold-treated dgat1-1 and dgat1-2 rosettes in comparison with the wild type (Supplemental Fig. S3, B and C).

The sensitivities of dgat1-1 and dgat1-2 mutant plants to freezing temperatures were tested further and compared with those of the wild-type plants. Four-week-old wild-type, dgat1-1, and dgat1-2 plants that were nonacclimated (NA) or cold acclimated (CA; 4°C for 3 d) were further exposed to various freezing temperatures. When the NA plants were treated at −6°C and −8°C for 40 min followed by a 5-d recovery period in normal conditions, the dgat1-1 and dgat1-2 mutants were more sensitive to the freezing stress than wild-type plants (Fig. 1A). Statistical analysis of the survival rates confirmed that, after the 5-d recovery period, most of the dgat1 mutant plants died, with survival rates of ∼45% and ∼20% at −6°C and −8°C, respectively (Fig. 1B). However, most of the wild-type plants (75%) survived at temperatures as low as −6°C, and 45% survived at −8°C (Fig. 1B). After acclimation at 4°C for 3 d, the freezing tolerance of wild-type plants increased (Fig. 1E), as indicated by the increase in survival rate from 45% in NA to 65% in CA at −8°C (Fig. 1F). Survival also increased in the dgat1 mutants at −8°C, from 20% in the NA treatment to ∼30% in the CA treatment, but it was only ∼10% in the CA treatment at −10°C (Fig. 1, E and F). The reduced freezing tolerance of the dgat1-1 and dgat1-2 plants compared with the wild type was further confirmed by measuring their dry weight upon exposure to NA or CA freezing temperatures (Fig. 1, C and G).

Figure 1.

dgat1 mutants display decreased freezing tolerance. A, Images of NA wild-type (WT), dgat1-1, and dgat1-2 plants before (CK) and after 40 min at −6°C and −8°C freezing temperatures followed by a 5-d recovery period at normal growth conditions. B and C, Survival rate (B) and dry weight (C) of NA wild-type, dgat1-1, and dgat1-2 plants after freezing treatment (−6°C and −8°C) followed by a 5-d recovery. D, Electrolyte leakage of NA wild-type, dgat1-1, and dgat1-2 plants upon freezing exposure (−3°C, −4°C, −5°C, −6°C, −7°C, and −8°C). E, Images of CA wild-type, dgat1-1, and dgat1-2 plants before (CK) and after 40-min treatments at −8°C and −10°C freezing temperatures followed by a 5-d recovery under normal growth conditions. F and G, Survival rate (F) and dry weight (G) of CA wild-type, dgat1-1, and dgat1-2 plants after the freezing treatment (−8°C and −10°C) followed by a 5-d recovery period. H, Electrolyte leakage of CA wild-type, dgat1-1, and dgat1-2 plants after freezing treatments (−5°C, −6°C, −7°C, −8°C, −9°C, and −10°C). The experiments were repeated three times independently. Data are means ± sd (n = 3 biological replicates). For each experiment, five independent technical replicates (grouped by three different plants) were analyzed for each genotype. Asterisks indicate significant differences from the wild type (**, P < 0.01, Student’s t test).

To verify the damage to membrane integrity after freezing treatment, we analyzed electrolyte leakage in rosettes collected from the NA and CA freezing-treated wild-type, dgat1-1, and dgat1-2 plants. Under NA freezing conditions, the dgat1-1 and dgat1-2 mutants exhibited significantly higher ionic leakage than wild-type plants at −6°C, −7°C, and −8°C (Fig. 1D). The ionic leakage rates also were elevated significantly in the CA freezing-treated dgat1-1 and dgat1-2 rosettes at −8°C, −9°C, and −10°C compared with wild-type plants (Fig. 1H). These findings suggest that mutation of DGAT1 attenuates the plant tolerance to chilling and freezing stresses.

H₂O₂ and SA Accumulate in the dgat1 Mutants upon Exposure to Freezing Temperatures

Freezing stress triggers the accumulation of ROS and cell death in plant rosette leaves (Iba, 2002; Chen et al., 2015b). To further investigate the increased sensitivity of dgat1 mutants to freezing stresses involving ROS and cell death, we employed DAB and Trypan Blue staining assays to examine the levels of H₂O₂ and cell death in wild-type, dgat1-1, and dgat1-2 mutant plants during CA, freezing, and postfreezing recovery conditions. Under 22°C or CA conditions, the wild-type and mutant leaves showed few significant signals in both staining assays (Fig. 2). In contrast, after exposure to freezing temperatures (−8°C for NA and −10°C for CA plants), elevated H₂O₂ accumulation and more severe cell death were detected in the dgat1-1 and dgat1-2 mutants in comparison with the wild-type leaves (Fig. 2, A and B).

Figure 2.

Cell death, H₂O₂ levels, and SA levels in the rosettes of wild-type (WT) and dgat1 plants upon freezing treatment. A and B, Trypan Blue (A) and DAB (B) staining showing cell death and ROS accumulation in the rosettes of wild-type, dgat1-1, and dgat1-2 plants before (22°C) and after treatment. Bars = 1 mm. C, H₂O₂ accumulation in the rosettes of wild-type, dgat1-1, and dgat1-2 plants. FW, Fresh weight. D, SA levels in the leaves of wild-type, dgat1-1, and dgat1-2 plants. Rosettes of 4-week-old wild-type, dgat1-1, and dgat1-2 plants, NA or CA for 3 d, were transferred to −8°C (for NA plants) or −10°C (for CA plants) and subsequently recovered for 12 h at 4°C. Rosettes were collected for Trypan Blue staining, DAB staining, and H₂O₂ and SA content measurements. E, Phenotypes of 11-d-old wild-type, dgat1-1, and dgat1-2 seedlings before (CK) and after freezing treatment (NA −8°C) and treatment with 500 μm GSH, following a 5-d recovery at normal growth conditions. F and G, Survival rate (F) and relative chlorophyll content (G) of NA wild-type, dgat1-1, and dgat1-2 seedlings after medium freezing treatment (−8°C) followed by a 5-d recovery at normal growth conditions. The relative chlorophyll content under NA −8°C treatment was expressed as a percentage of the value for the same genotype grown under normal growth conditions (CK, set to 100%). nd, No significant difference. The experiments were repeated three times with more than 15 plants used for each genotype. Data are means ± sd (n = 3 biological replicates). Asterisks indicate significant differences from the wild type (**, P < 0.01, Student’s t test).

The accumulation of ROS in the dgat1 mutants in response to freezing exposure was further confirmed using an Amplex Red-coupled fluorescence quantitative assay. Consistent with the DAB staining, the H₂O₂ levels showed no significant difference between the wild type and dgat1 mutants at 22°C and CA conditions (Fig. 2C). Under NA −8°C, CA −10°C, and postfreezing recovery conditions, H₂O₂ accumulated to a greater degree in the dgat1-1 and dgat1-2 mutants in comparison with the wild-type plants (Fig. 2C).

Since freezing stress induces the accumulation of SA, which contributes to freezing-induced ROS production and plant freezing sensitivity (Miura and Ohta, 2010; Kim et al., 2013; Chen et al., 2015b), we next measured the endogenous SA levels in wild-type and dgat1 mutant rosettes under various temperature conditions. Consistent with previous findings, the freezing treatment induced SA levels in wild-type rosettes compared with the 22°C or CA conditions (Fig. 2D). In contrast, in response to both the NA −8°C and CA −10°C treatments, SA levels were much higher in dgat1-1 and dgat1-2 mutants in comparison with the wild type. After the 12-h recovery period following freezing treatment, SA levels increased further and remained significantly higher in the dgat1 mutants (P < 0.01; Fig. 2D).

In plant cells, the oxidative stress imposed by ROS accumulation activates the production of antioxidants, including glutathione (GSH), to maintain cellular redox homeostasis (Foyer and Noctor, 2011). Given that elevated levels of H₂O₂ were detected in dgat1 mutants under freezing stress, we further investigated the effects of GSH application on the sensitivity of freeze-treated wild-type and dgat1 mutant seedlings. Under normal growth conditions, the dgat1-1 and dgat1-2 seedlings exhibited little morphological difference from wild-type seedlings with 500 μm GSH application for 11 d (Fig. 2E). As expected, the increased sensitivities of dgat1 mutants to the NA freezing treatment were remedied significantly by supplying GSH in Murashige and Skoog (MS) growth medium (Fig. 2, E, and F). The recovery of freezing-sensitive phenotypes of dgat1 mutants by the addition of GSH was further confirmed by measuring the relative chlorophyll contents in the wild type and dgat1 mutants after freezing treatment. The dgat1 mutants accumulated significantly less chlorophyll than the wild type, but this difference was not observed when GSH was supplied (Fig. 2G).

The dgat1 Mutants Show Decreased Expression of CBF2 and Its Target Genes

To further elucidate the association of freezing-sensitive phenotypes of dgat1 mutants with the CBF-dependent signaling pathway, we analyzed the expression levels of CBFs (CBF1, CBF2, and CBF3) and their target genes, COR47, RD29A, and KIN1, by RT-qPCR. The cold-inducible expression of CBF2 at 3, 6, 12, and 24 h, and its targets COR47, RD29A, and KIN1 at 6, 12, and 24 h after cold treatment, were significantly lower (P < 0.01) in the dgat1 mutants compared with wild-type plants (Supplemental Fig. S4). In contrast, CBF1 and CBF3 expression was only down-regulated in the dgat1 mutants 3 h after cold treatment and showed similar levels to wild-type plants afterward (Supplemental Fig. S4). Therefore, the lower expression of CBF2, COR47, RD29A, and KIN1 may contribute to the increased sensitivity to chilling and freezing stresses in the dgat1 mutants.

Changes in Glycerolipid Levels in the Wild Type and dgat1 Mutants upon Freezing Exposure

DGAT1 is a DAG acyltransferase that catalyzes the conversion of DAG to TAG in plants (Katavic et al., 1995; Routaboul et al., 1999; Zou et al., 1999; Zhang et al., 2009a). Thus, we next asked whether the disruption of DGAT1 alters the glycerolipid profiles of Arabidopsis rosettes under freezing conditions. To this end, we used electrospray ionization-tandem mass spectrometry to analyze the profiles of galactolipids and PLs in wild-type, dgat1-1, and dgat1-2 rosettes under normal growth conditions (22°C) or after exposure to freezing temperatures (NA −8°C or CA −10°C). Under normal conditions (22°C), few differences were detected between the wild-type and dgat1 plants (Table I; Fig. 3). After NA −8°C or CA −10°C treatment, the total contents of PA were significantly higher in the dgat1 mutants compared with the wild type (Table I). When specific lipid species were analyzed, the levels of PA species, including 34:6-PA, 34:5-PA, 34:4-PA, 34:3-PA, 34:2-PA, 34:1-PA, 36:6-PA, 36:5-PA, and 36:4-PA under both NA −8°C and CA −10°C treatment, were significantly higher in dgat1 than in wild-type rosettes (P < 0.01; Fig. 3). In addition, total PE levels were elevated slightly in the dgat1 mutants under freezing treatment compared with the wild type (Table I). Specifically, the levels of 36:5-PE under NA −8°C and 34:3-PE, 34:2-PE, and 36:5-PE under CA −10°C were higher in both dgat1-1 and dgat1-2 mutants (Fig. 3). In addition, 36:5-PS under NA −8°C and 36:5-PS and 36:2-PS under CA −10°C increased in the dgat1 mutants, although total PS contents showed no significant differences from the wild type in all conditions (Table I; Fig. 3).

Table I. Total lipid content in each head group class in leaves of wild-type, dgat1-1, and dgat1-2 plants upon NA or CA treatment followed by freezing.

The lipid contents in rosettes of plants grown at 22°C were measured as controls. Values are means ± sd (nmol mg⁻¹ dry weight; n = 4). Significant differences between dgat1-1 and dgat1-2 and the wild type in each group are highlighted in boldface (*, P < 0.05 and **, P < 0.01, Student’s t test).

Lipid Class	22°C			NA −8°C			CA −10°C
	Wild Type	dgat1-1	dgat1-2	Wild Type	dgat1-1	dgat1-2	Wild Type	dgat1-1	dgat1-2
MGDG	204.9 ± 6.7	207.5 ± 8.5	198.7 ± 9.6	128.7 ± 7.8	134.1 ± 9.2	131.6 ± 8.8	132.8 ± 8.8	133.8 ± 10.9	128.9 ± 11.3
DGDG	39.1 ± 5.7	37.5 ± 4.6	38.3 ± 4.2	53.2 ± 5.7	50.8 ± 4.4	51.3 ± 4.9	45.4 ± 5.1	52.5 ± 3.7	50.1 ± 4.2
PG	15.5 ± 2.1	15.6 ± 1.7	15.1 ± 1.4	13.5 ± 2.8	12.7 ± 2.6	14.2 ± 3.2	14.7 ± 2.1	15.7 ± 1.8	15.9 ± 2.2
PC	15.0 ± 2.3	13.8 ± 2.7	14.4 ± 1.8	9.3 ± 1.2	8.7 ± 1.1	8.9 ± 1.1	8.7 ± 1.1	8.3 ± 0.9	8.6 ± 1.0
PE	2.5 ± 0.4	2.3 ± 0.2	2.3 ± 0.3	3.9 ± 0.2	4.4 ± 0.2*	4.1 ± 0.3	3.5 ± 0.3	4.0 ± 0.2*	4.1 ± 0.2*
PI	5.8 ± 0.8	5.4 ± 0.4	4.9 ± 0.5	5.5 ± 0.7	5.9 ± 0.4	6.1 ± 0.7	5.8 ± 0.7	6.1 ± 0.6	5.7 ± 0.7
PS	0.27 ± 0.05	0.30 ± 0.06	0.32 ± 0.04	0.33 ± 0.04	0.36 ± 0.06	0.33 ± 0.03	0.31 ± 0.05	0.28 ± 0.04	0.29 ± 0.06
PA	0.07 ± 0.04	0.09 ± 0.03	0.08 ± 0.04	9.74 ± 1.1	18.03 ± 2.7**	17.85 ± 2.6**	11.65 ± 0.9	23.50 ± 1.1**	23.56 ± 1.9**
LysoPG	0.18 ± 0.03	0.26 ± 0.05	0.19 ± 0.03	0.26 ± 0.04	0.28 ± 0.02	0.26 ± 0.03	0.21 ± 0.04	0.27 ± 0.03	0.22 ± 0.02
LysoPC	0.09 ± 0.03	0.07 ± 0.02	0.06 ± 0.03	0.57 ± 0.07	0.55 ± 0.05	0.61 ± 0.05	0.61 ± 0.09	0.57 ± 0.07	0.66 ± 0.09
LysoPE	0.07 ± 0.04	0.05 ± 0.02	0.06 ± 0.01	0.52 ± 0.05	0.54 ± 0.03	0.56 ± 0.03	0.44 ± 0.06	0.48 ± 0.07	0.51 ± 0.07

Figure 3.

Lipid profiles in the rosettes of 4-week-old wild-type (WT), dgat1-1, and dgat1-2 plants before (22°C) and after NA or CA treatment followed by freezing exposure (NA −8°C or CA −10°C). Alterations are shown in the compositions of galactolipids (MGDG and digalactosyldiacylglycerol [DGDG]) and PLs (Phosphatidylglycerol [PG], phosphatidylcholine [PC], phosphatidylethanolamine [PE], phosphatidylinositol [PI], phosphatidylserine [PS], and PA) of wild-type, dgat1-1, and dgat1-2 rosettes before (22°C) and after NA or CA treatment followed by freezing stress (−8°C or −10°C). Values represent means ± sd (n = 4) of four independent samples, with each sample collected from the rosettes of three plants. Asterisks indicate significant differences from the wild type (*, P < 0.05 and **, P < 0.01, Student’s t test).

The DGAT1-Mediated Freezing Response Requires RbohD

Previous findings suggest that PA can bind directly to the NADPH oxidase RbohD, leading to abscisic acid (ABA)-mediated ROS production and stomatal closure (Zhang et al., 2009b). Because of the elevation of H₂O₂ and PA contents in the dgat1 mutants under freezing treatment, we proposed that higher H₂O₂ production in the dgat1 mutants may result from the stimulation of RbohD activity by the accumulated PA. To test this possibility, we generated a dgat1 rbohD double mutant by crossing dgat1-1 to the rbohD knockout line (Chen et al., 2015a). Phenotypic analyses showed that, upon freezing exposure, the rbohD single mutant exhibited improved freezing tolerance compared with the wild type (Fig. 4, A and B). In contrast to the freezing-hypersensitive phenotype of the dgat1-1 mutant, the dgat1 rbohD double mutant showed similar freezing tolerance to the wild-type plants (Fig. 4, A and B), suggesting that the freezing response of dgat1 mutants requires functional RbohD.

Figure 4.

The increased freezing sensitivity of dgat1 mutants requires RbohD. A, Four-week-old wild-type (WT), dgat1-1, dgat1 rbohD, and rbohD plants before (CK) and after freezing treatment (NA −8°C and CA −10°C) followed by a 5-d recovery period at normal growth conditions. B, Survival rate of wild-type, dgat1-1, dgat1 rbohD, and rbohD plants after freezing treatment (NA −8°C and CA −10°C) followed by a 5-d recovery. C and D, NADPH activity (C) and H₂O₂ levels (D) in the rosettes of wild-type, dgat1-1, dgat1 rbohD, and rbohD plants before and after freezing treatment (NA −8°C and CA −10°C) followed by a 5-d recovery. NADPH oxidase activity is presented as ΔA₄₇₀ per milligram of protein per minute in C. FW, Fresh weight. Letters a and b indicate lower and higher survival rate, NADPH activity, and H₂O₂ level in the freeze-treated mutants, respectively, compared with wild-type plants. The experiments were repeated three times with more than 15 plants used for each genotype. Data are means ± sd (n = 3 biological replicates). Asterisks indicate significant differences from the wild type (**, P < 0.01, Student’s t test).

To determine whether the freezing-induced accumulation of PA in the dgat1 mutants affects ROS production, we measured NADPH oxidase activity in wild-type, dgat1-1, dgat1 rbohD, and rbohD rosettes under normal conditions (22°C) or after NA −8°C and CA −10°C treatments. In wild-type plants, both NA and CA freezing treatments stimulated NADPH oxidase activity compared with normal conditions (Fig. 4C). Under freezing treatments, dgat1-1 and rbohD mutants displayed significantly higher and lower NADPH oxidase activities, respectively, than wild-type plants (P < 0.01; Fig. 4C). Particularly, NADPH oxidase activity in the dgat1-1 mutant was reduced to wild-type levels by introducing the rbohD mutation (Fig. 4C). Moreover, the rbohD mutation also attenuated the accumulation of H₂O₂ under freezing treatment in the dgat1 rbohD double mutant relative to dgat1-1 (Fig. 4D).

To verify whether the PA-triggered accumulation of ROS and cell death require a functional RbohD, both wild-type and rbohD rosettes were treated with PA or PC for 24 h, and the levels of H₂O₂ and cell death were detected by DAB and Trypan Blue staining, respectively. Application of PA, rather than PC, significantly induced H₂O₂ production and cell death in the wild-type leaves (Supplemental Fig. S5). By contrast, H₂O₂ production and cell death were abolished in the rbohD rosettes, suggesting that PA-promoted cell death requires a functional RbohD. Together, these results suggest that PA-RbohD-triggered H₂O₂ production contributes to the decreased freezing tolerance of dgat1.

Disruption of DGAT1 Suppresses the Freezing-Induced Conversion of DAG to TAG

Under chilling or freezing temperatures, PA is produced through either the hydrolysis of PLs by PLDs or the phosphorylation of DAG by DGKs (Welti et al., 2002; Arisz et al., 2013). The increased accumulation of PA in dgat1 mutants did not accompany a decrease of other PLs, such as PC, PE, PI, or PS. Therefore, we hypothesized that the loss of function of DGAT1 might attenuate the freezing-triggered conversion of DAG to TAG, which results in PA accumulation through the DGK pathway. To test this hypothesis, we measured the levels of DAG and TAG in wild-type and dgat1 rosettes grown under normal conditions (22°C) and after NA −8°C or CA −10°C treatment. At 22°C, the total contents of DAG and TAG showed little difference between the wild type and dgat1 mutants (Fig. 5A). After NA −8°C or CA −10°C treatment, DAG and TAG contents were elevated substantially in the wild type compared with the untreated control (Fig. 5A). In contrast, after freezing treatments, the dgat1 mutants showed a higher level of DAG and a lower level of TAG than the wild type (Fig. 5A). As a result, a higher DAG-TAG ratio was observed in the freeze-treated dgat1 mutants in comparison with the wild type (Fig. 5B). Specifically, the levels of DAG with acyl chains of 18:3-16:3, 18:3-16:1, 18:3-16:0, 18:3-18:3, 18:3-18:2, 18:2-16:2, 18:2-16:1, 18:2-16:0, 18:2-18:2, and 18:1-16:3 were drastically higher in the dgat1 mutants compared with the wild type after both NA −8°C and CA −10°C treatments (Fig. 5C). In contrast, the levels of freezing-induced TAG with acyl chains containing 16:0, 18:1, 18:2, and 18:3 were significantly lower in the dgat1 mutants in comparison with the wild-type plants (P < 0.01; Fig. 5D). The higher DAG and lower TAG contents in the dgat1 mutants under freezing stresses imply that the decreased freezing tolerance from DGAT1 loss of function is due primarily to the attenuated freezing-induced DAG-to-TAG conversion and enhanced PA production through the DGK pathway.

Figure 5.

Profiles of DAG and TAG in rosettes of 4-week-old wild-type (WT), dgat1-1, and dgat1-2 plants before (22°C) and after NA or CA treatment followed by freezing exposure (−8°C for NA and −10°C for CA). A and B, Relative levels (signal mg⁻¹ dry weight) of total DAG and TAG in wild-type, dgat1-1, and dgat1-2 rosettes before (22°C) and after NA or CA followed by freezing treatment. The DAG-TAG ratio of each treatment is presented in B. C and D, Molecular species of DAG (C) and TAG (D) in wild-type, dgat1-1, and dgat1-2 rosettes before (22°C) and after freezing treatment (NA −8°C and CA −10°C). Data are means ± sd (n = 4) of four independent samples, and each sample was collected from the rosettes of three plants. Asterisks indicate significant differences from the wild type (**, P < 0.01, Student’s t test).

Previous findings suggest that, under freezing temperatures, the increased DAG and TAG in Arabidopsis rosettes is likely due to their conversion from MGDG by SFR2 (Thorlby et al., 2004; Fourrier et al., 2008; Moellering et al., 2010). To verify the source of freezing-induced DAG and TAG, we further analyzed the levels of these two lipids in two SFR2 knockout mutants, sfr2-3 and sfr2-4 (Moellering et al., 2010), under freezing conditions. We observed that, in response to freezing treatment, the total levels of DAG were significantly lower in the sfr2 mutants compared with wild-type rosettes (Supplemental Fig. S6A). More specifically, DAG species with 16:3 and 18:3 acyl chains, which are the characteristics of MGDG acyl composition, increased significantly in the wild type but remained unchanged in the sfr2 mutants after freezing treatment (Supplemental Fig. S6C). Consistent with the alterations of DAG, total TAG levels also showed significantly lower levels in the sfr2 mutants compared with wild-type rosettes upon freezing exposure (Supplemental Fig. S6A). Moreover, TAG species with 18:3 acyl chain were almost absent in sfr2 mutants after freezing treatment (Supplemental Fig. S6D). We conclude that the accumulated DAG levels under freezing temperatures are largely converted from MGDG by SFR2.

DGK2, DGK3, and DGK5 Deletion Confers Enhanced Freezing Tolerance and Decreased PA Production

To investigate the potential role of DGKs in DGAT1-mediated freezing tolerance, we first analyzed the expression levels of DGKs and PLDs in wild-type and dgat1 plants in response to cold treatments. The transcripts of all DGK genes were up-regulated by brief exposure to chilling temperatures (Fig. 6A), which confirmed previous findings (Arisz et al., 2013; Chen et al., 2015b). We also observed that the expression levels of DGK1, DGK2, DGK3, and DGK5 were further elevated by freezing treatment and postfreezing recovery (Fig. 6A). Moreover, the expression levels of DGK2, DGK3, and DGK5 were significantly higher in both dgat1-1 and dgat1-2 mutants at 12, 24, and 72 h after 4°C exposure, after freezing treatments, and following the postfreezing recovery period compared with the wild-type plants (P < 0.01; Fig. 6A). In contrast, transcript levels of DGK1, DGK4, and DGK6 were up-regulated in the wild type and dgat1 mutants under freezing and postfreezing conditions, but DGK1 expression at 12 and 24 h and DGK4 and DGK6 expression at 12 h were higher in the mutants than in the wild type (Fig. 6A). In comparison, the cold-induced expression levels of PLDα1 and PLDδ, used as controls, showed no significant difference between the wild-type and dgat1 plants in response to cold treatments (Fig. 6A). These results suggest that DGK2, DGK3, and DGK5 may share a predominant role in the DGAT1-mediated freezing response.

Figure 6.

Knockout of DGK2, DGK3, and DGK5 confers enhanced freezing tolerance. A, Expression profiles of cold-responsive DGKs and PLDs in wild-type (WT), dgat1-1, and dgat1-2 plants under cold (4°C) or freezing (−10°C) temperatures followed by recovery. Rosettes of 4-week-old wild-type, dgat1-1, and dgat1-2 plants treated at 4°C for 0, 6, 12, 24, and 72 h, freezing (−10°C) for 40 min (CA −10), and following recovery at 4°C for 6 h (CA-R) were collected for total RNA extraction. Hierarchical cluster analyses were used in the transcript levels of six DGKs (DGK1, DGK2, DGK3, DGK4, DGK5, and DGK6), PLDα1, and PLDδ determined by RT-qPCR. Three biological replicates were conducted with similar results, and representative data from one experiment are shown. Data are means ± sd (n = 3) of three technical replicates. The relative gene expression values were plotted with the heatmap 2.0 package in R, with red and blue colors representing up- and down-regulation, respectively. Asterisks indicate significant differences in dgat1-1 and dgat1-2 plants compared with the wild type (**, P < 0.01, Student’s t test). B, NA and CA wild-type, dgk2-1, dgk3-1, and dgk5-1 seedlings before (CK) and after freezing treatment (NA −8°C or CA −10°C) followed by a 5-d recovery period at normal growth conditions. C, Survival rate and dry weight of NA and CA wild-type, dgk2-1, dgk3-1, and dgk5-1 plants after freezing treatment (NA −8°C or CA −10°C) followed by a 5-d recovery period.

Under freezing temperatures, MGDG is converted to oligogalactolipids and DAG; the latter is likely released to the cytosol during freezing-induced membrane shrinkage (Moellering et al., 2010; Moellering and Benning, 2011). To determine this possibility, we further examined the subcellular localization of DGK2, DGK3, and DGK5 proteins. By expression in Arabidopsis protoplasts, we observed that all of the DGK2-GFP, DGK3-GFP, and DGK5-GFP fusions were localized in the cytosol (Supplemental Fig. S7), suggesting that they are available for the cytosolic conversion of MGDG-derived DAG to PA.

The functions of DGKs in freezing tolerance were validated by analyses of the DGK2, DGK3, and DGK5 knockout mutants. To this end, lines with T-DNA insertions in DGK2, DGK3, and DGK5 were identified. The transcripts of DGK2, DGK3, and DGK5 were undetectable in the respective dgk2-1, dgk3-1, and dgk5-1 lines (Supplemental Fig. S8), suggesting that all lines contain knockout alleles. Phenotypic analyses showed that, under normal growth conditions, the 4-week-old dgk2-1, dgk3-1, and dgk5-1 mutants exhibited little morphological difference from wild-type plants (Fig. 6B). However, when the NA and CA plants were exposed to freezing temperatures (−8°C or −10°C), all the dgk mutants showed enhanced tolerance to freezing stress (Fig. 6B). The freezing-tolerant phenotypes of the dgk mutants were further confirmed by higher survival rates and dry weights than wild-type plants during NA −8°C or CA −10°C followed by a 5-d recovery (Fig. 6C).

To evaluate the function of DGKs in the phosphorylation of DAG for PA synthesis under freezing conditions, we next analyzed the lipid profiles in wild-type, dgk2-1, dgk3-1, and dgk5-1 rosettes under normal growth conditions (22°C) or after exposure to freezing temperatures (NA −8°C or CA −10°C). Under normal growth conditions (22°C), the rosettes of wild-type and dgat1 plants showed few significant differences in either galactolipids or PLs (Fig. 7A; Supplemental Fig. S9). After NA −8°C or CA −10°C treatment, the total contents of PC, PE, PI, and PS declined in both the wild type and dgk mutants (Fig. 7A; Supplemental Fig. S9). As expected, the contents of PA were significantly lower in the dgk mutants compared with the wild type (Fig. 7A). Particularly, the levels of PA species, including 34:6-PA, 34:3-PA, 34:2-PA, 34:1-PA, 36:6-PA, 36:5-PA, 36:4-PA, and 36:3-PA, were significantly lower in the dgk mutants compared with wild-type rosettes under NA −8°C and/or CA −10°C treatments (Fig. 7B). As a result of the attenuated freezing-inducible PA accumulation, the dgk mutants showed inhibited NADPH oxidase activity and lower H₂O₂ production in comparison with wild-type plants (Fig. 7, C and D). Compared with wild-type plants, all the single dgk2-1, dgk3-1, and dgk5-1 mutants showed enhanced freezing-tolerant phenotypes (Fig. 6B), decreased PA level (Fig. 7, A and B), and reduced H₂O₂ production (Fig. 7D) in response to freezing stress. Therefore, DGK2, DGK3, and DGK5 proteins are predominantly functional in converting DAG to PA upon freezing exposure.

Figure 7.

Total lipids and PA species of 4-week-old wild-type (WT), dgk2-1, dgk3-1, and dgk5-1 plants before (22°C) and after NA or CA treatment following freezing exposure (NA −8°C or CA −10°C). A, Alteration in the compositions of PLs (PC, PE, PI, PS, and PA) of wild-type, dgk2-1, dgk3-1, and dgk5-1 rosettes before (22°C) and after NA or CA following freezing stress (−8°C or −10°C). B, Concentrations of PA species of wild-type, dgk2-1, dgk3-1, and dgk5-1 rosettes before (22°C) and after NA or CA following freezing stress (−8°C or −10°C). C and D, NADPH activity (C) and H₂O₂ levels (D) of wild-type, dgk2-1, dgk3-1, and dgk5-1 plants before (CK) and after freezing treatment (NA −8°C or CA −10°C). NADPH oxidase activity is presented as ΔA₄₇₀ per milligram of protein per minute. FW, Fresh weight. E, A working model showing the role of DGAT1 in the plant response to freezing by modulating DAG homeostasis and ROS production.

DISCUSSION

Arabidopsis DGAT1 acts as an acyl-CoA:DAG acyltransferase in converting DAG to TAG during seed oil deposition (Katavic et al., 1995; Routaboul et al., 1999; Zou et al., 1999; Zhang et al., 2009a). Previous findings suggest that the freezing-induced accumulation of the PL PA forms a destabilized H_II-type phase with MGDG or DAG, which damages cell membrane integrity under freezing temperatures, thus attenuating the plant tolerance to such stresses (Steponkus, 1984; Uemura et al., 1995; Thomashow, 1999; Welti et al., 2002; Moellering et al., 2010; Moellering and Benning, 2011). Recent studies reveal that, upon freezing exposure, the galactolipid MGDG is converted to oligogalactolipids and DAG, the latter of which is likely further catalyzed to TAG for membrane stabilization (Moellering et al., 2010; Moellering and Benning, 2011). In response to chilling or freezing, DAG also can be phosphorylated by DGKs to produce PA (Gómez-Merino et al., 2004; Arisz et al., 2013; Chen et al., 2015b). These findings indicate that DGATs and DGKs potentially function in regulating lipid remodeling in the plant response to low temperatures. However, how plant cells integrate the dynamics of these lipids to mediate sensitivity to cold stresses remains unknown.

Here, we presented several lines of evidence to support the idea that Arabidopsis DGAT1 and DGK2/3/5 contribute to the freezing response by modulating the homeostasis of DAG, TAG, and PA. First, the transcript of DGAT1 was induced by chilling treatment, and the DGAT1 null mutants, dgat1-1 and dgat1-2, were more sensitive to chilling and freezing stresses (Fig. 1; Supplemental Figs. S1 and S3). Second, the dgat1 mutants showed lower levels of cold-inducible expression of CBF2 and its regulons (Supplemental Fig. S4). Third, the decreased freezing tolerance of dgat1 mutants was associated with the enhanced accumulation of ROS under freezing treatment, which required a functional RbohD (Figs. 2 and 4). Fourth, lipid profiling revealed that, in response to freezing temperatures, the dgat1 mutants showed significantly higher levels of PA and DAG but significantly lower levels of TAG compared with wild-type plants (Figs. 3 and 5). Finally, knocking out DGK2, DGK3, and DGK5 reduced DAG-derived PA production and improved plant tolerance to freezing stress (Figs. 6 and 7). Taken together, our findings reveal that DAG is an essential substrate for the DGAT1 and DGK2/3/5 enzymes in the remodeling of cold-responsive lipids in Arabidopsis.

When temperatures decline as low as 0°C, ice crystals form in the extracellular space, leading to dehydration and plant cell damage (Thomashow, 1999; Yamaguchi-Shinozaki and Shinozaki, 2006). Under chilling and freezing conditions, plants have developed several physiological and biochemical strategies, including remodeling of the plastidic and extraplastidic membrane lipid compositions, to prevent cold-induced cell damage (Uemura et al., 1995; Welti et al., 2002; Li et al., 2004, 2008; Moellering et al., 2010; Chen and Thelen, 2013). For instance, upon cold exposure, the amounts of unsaturated fatty acids or PLs increase drastically, thus enhancing membrane fluidity and integrity. In contrast, the freezing-inducible accumulation of lipids with small or lacking head groups, such as MGDG, DAG, and PA, tends to form a nonlamellar H_II-type phase, which results in the shrinkage of membrane structure and membrane ionic leakage (Kuiper, 1970; Verkleij et al., 1982; Welti et al., 2002; Moellering et al., 2010).

Previous studies reveal that Arabidopsis SFR2, a galactolipid:galactolipid galactosyltransferase, catalyzes the freezing-inducible conversion of MGDG to DGDG and other oligogalactolipids in the chloroplast outer envelope. As a result, it reduces inverted H_II-type structure formation and improves plant freezing tolerance (Thorlby et al., 2004; Fourrier et al., 2008; Moellering et al., 2010). Furthermore, the freezing-hypersensitive mutant sfr2 displays significantly lower TAG levels compared with wild-type plants (Supplemental Fig. S6; Moellering et al., 2010). Consistent with this, our recent work shows that mutants of three lipase-like regulators in the SA signaling pathway, sag101, eds1, and pad4, exhibit enhanced freezing tolerance, along with lower DAG and higher TAG levels under freezing treatments (Chen et al., 2015b). These findings suggest that, by modulating cell membrane stability and fluidity, the ratio of DAG to TAG lipids represents an important mechanism in the regulation of plant freezing response. The dgat1 mutants were hypersensitive to chilling and freezing (Fig. 1; Supplemental Fig. S3) and showed elevated levels of freezing-inducible DAG but lower TAG levels (Fig. 5). This result provides direct genetic evidence to show the predominant functions of DGAT1 in catalyzing the conversion of DAG to TAG in response to freezing. It is conceivable that, under freezing exposure, the DGAT1-mediated conversion of DAG to the more neutral lipid TAG acts as a protective strategy for enhancing plant tolerance to freezing stresses. Interestingly, PDAT1, another DAG acyltransferase that catalyzes the DAG-to-TAG conversion in Arabidopsis, is involved in TAG production when plants suffer heat stress (Mueller et al., 2017), suggesting that DGAT1 and PDAT1 play primary roles in the plant responses to these different abiotic stresses.

In addition to MGDG and DAG, the lipid molecule PA can form a nonlamellar lipid structure under freezing conditions, which confers increased sensitivity to cold stresses in Arabidopsis (Wang, 2004; Browse, 2010). Freezing-induced PA production can be derived from either the PLD pathway by hydrolysis of PLs or the DGK pathway by phosphorylation of DAG (Arisz et al., 2013). In Arabidopsis, two PLDs, PLDα1 and PLDδ, play distinct roles in the regulation of plant tolerance to freezing (Welti et al., 2002; Li et al., 2004; Rajashekar et al., 2006). In particular, knockout of PLDα1 reduces the hydrolysis of PLs to PA and leads to enhanced tolerance to freezing (Welti et al., 2002; Rajashekar et al., 2006). However, the PLDδ null mutants show increased sensitivity to freezing, and PLDδ overexpressors show the opposite phenotype (Li et al., 2004). A possible explanation is that PLDα1-mediated PA accumulation can activate RbohD and facilitate H₂O₂ production, whereas PLDδ is likely involved in ROS clearance (Zhang et al., 2003, 2009b). In this study, we found that, in the freeze-treated dgat1 mutants, the transcripts of DGK2, DGK3, and DGK5 were elevated compared with wild-type plants (Fig. 6A). Moreover, disruption of DGK2, DGK3, and DGK5 conferred enhanced freezing tolerance (Fig. 6B). By lipid profiling analyses, we further revealed that these three DGKs contributed redundantly to the freezing-inducible accumulation of PA, as indicated by the reductions in PA in the dgk mutants upon freezing exposure (Fig. 7, A and B). Therefore, we presented strong evidence to link DGK-catalyzed PA production to the DGAT1-mediated DAG-to-TAG conversion pathway in plant responses to freezing stress (Fig. 7E). Under freezing temperatures, DAG likely functions as a common substrate in balancing the downstream actions of DGAT1 and DGKs to produce the neutral TAG and polar PA, two lipids that play opposite roles in maintaining cell membrane integrity under freezing treatment. Thus, the DGK pathway functions either as a parallel mechanism or as a complementary mechanism to that of PLDs to induce PA production in response to cold stresses.

ROS are important second messengers for the regulation of plant adaption to various environmental cues, including cold temperatures (Apel and Hirt, 2004; Laloi et al., 2004). Particularly, the activities of several key enzymes in the MAPK signaling cascade, such as MEKK1-MKK1/2, are ROS responsive. In the Arabidopsis response to ABA, MKK1 directly regulates the expression of the reduction enzyme CAT1 and determines the cellular redox homeostasis (Xing et al., 2008). In contrast, MKK2 mediates plant tolerance to cold and salt stresses (Teige et al., 2004; Xing et al., 2008; Jammes et al., 2009). More recent studies reveal that two downstream ROS-responsive MPK kinases, MPK3 and MPK6, are involved in cold responses by modulating the protein stability of ICE1 (Li et al., 2017; Zhao et al., 2017), suggesting that the direct regulation of key components in cold signaling involves ROS. The increased ROS production in the dgat1 mutants (Fig. 2, B and C) is possibly associated with the enhancement of MPK3 and MPK6 activity, leading to the degradation of ICE1. CBF2 is one of the downstream targets of ICE1 (Chinnusamy et al., 2003; Ding et al., 2015); this may explain why the transcript levels of CBF2 and its regulons, COR47, RD29A, and KIN1, were significantly lower in dgat1 mutants than in the wild type under chilling conditions (Supplemental Fig. S4). Furthermore, increasing evidence indicates that a plant’s tolerance to freezing stresses is tightly associated with its intracellular ROS level (Iba, 2002), as the accumulation of ROS can trigger irreversible oxidative damage and cause cell death (Apel and Hirt, 2004). In particular, two mutants of LESION SIMULATING DISEASE RESISTANCE1 (LSD1) display chilling-sensitive phenotypes and elevated H₂O₂ levels, which can be fully suppressed by mutation of EDS1 or PAD4 in the lsd1 background, indicating that LSD1 may function in plant cold responses by modulating ROS levels (Huang et al., 2010). In Arabidopsis, ROS are primarily generated by the action of the NADPH oxidases RbohD and RbohF (Kaur et al., 2014). Previous studies suggest that PA can interact with RbohD and stimulate its activity, which facilitates ABA-responsive stomatal closure, indicating that PA is a lipid activator for RbohD-mediated ROS production (Zhang et al., 2009b). We further observed that the freezing-induced accumulation of PA in dgat1 was associated with elevated H₂O₂ contents (Figs. 2 and 3), suggesting that PA likely contributes to ROS production by activating RbohD activity in response to freezing. Consistent with this, we found that knocking out RbohD in the dgat1-1 background attenuated the increased sensitivity and enhanced H₂O₂ production upon freezing exposure (Fig. 4).

CONCLUSION

Our results demonstrate that freezing can rapidly stimulate the accumulation of DAG in Arabidopsis (Fig. 7E), possibly through the conversion of MGDG by the galactolipid-remodeling enzyme SFR2 or the hydrolysis of polyphosphoinositides by PLCs (Moellering et al., 2010; Arisz et al., 2013). In response to freezing, DGAT1 predominantly functions in catalyzing the subsequent conversion of DAG to TAG, preventing the formation of the destabilized H_II-type phase, as a protective mechanism to improve plant tolerance to such stresses (Fig. 7E). Freezing also induces the hydrolysis of PLs to produce PA (Welti et al., 2002; Li et al., 2004; Rajashekar et al., 2006). The elevation of PA may either disrupt membrane permeability or stimulate RbohD activity for ROS production and subsequently increase plant sensitivity to freezing (Fig. 7E). In parallel with the PLD pathway, cold-inducible DGK2/3/5 also contribute to PA generation by phosphorylating DAG (Fig. 7E). Therefore, it is conceivable that DAG may function as a common substrate necessary for both DGAT1- and DGK2/3/5-mediated TAG and PA production, respectively, the balance of which determines plant survival during freezing stress. The dynamic remodeling of DAG, TAG, and PA lipids is likely an important protective strategy enhancing plant survival of freezing temperatures.

MATERIALS AND METHODS

Plant Materials, Growth Conditions, and Treatments

The Arabidopsis (Arabidopsis thaliana) ethyl methanesulfonate-generated mutant dgat1-1 (CS3861) and T-DNA insertional mutant dgat1-2 (SALK_039456) were obtained from the Arabidopsis Biological Resource Center (http://www.arabidopsis.org). Characterization of the dgat1-1 homozygous mutant was carried out following Zou et al. (1999). For identification of the T-DNA insertion site in the dgat1-2 mutant, the DGAT1-specific primers XS1844/XS1845 and the T-DNA left border primer LBa1 (Supplemental Table S1) were used for PCR amplification. The dgk2-1 (SAIL 718_G03), dgk3-1 (SALK_082600), and dgk5-1 (SAIL 1212_E10) mutants were identified by PCR using primers (Pa/Pb and Pb/LB1 for dgk2-1, Pc/Pd and Pd/LBa1 for dgk3-1, and Pe/Pf and Pf/LB1 for dgk5-1).

Seeds of Arabidopsis wild type (ecotype Columbia-0) and mutants were surface sterilized with 20% bleach (v/v) containing 0.1% Tween 20 (v/v) for 20 min and then washed three times with sterilized water followed by sowing on MS medium supplemented with 1% Suc. Seeds were dark treated at 4°C for 3 d and subsequently transferred to a plant growth room under a 16-h-light (23°C)/8-h-dark (21°C) cycle. After germination for 10 d, the seedlings were transplanted to soil in the plant growth room until treatment.

Chilling and freezing treatments were carried out as described previously (Chen et al., 2008) with minor modifications. For cold acclimation, 4-week-old plants were transferred from the growth room to a cold chamber (4°C) for 3 d of acclimation under a normal light/dark cycle. For the freezing treatment, all 4-week-old NA or CA soil-grown plants or 11-d-old seedlings grown on MS medium plates were transferred to a growth chamber (Blue Pard LRH-250CA) with temperatures reduced steadily from 4°C to −2°C (2°C h⁻¹). Ice crystals were then placed on the soil or plates to avoid supercooling when the temperature reached −2°C. The temperature remained at −2°C for 2 h and continued to lower until reaching the final temperatures. After staying at the final temperatures (−6°C and −8°C for NA plants, −8°C and −10°C for CA plants) for 40 min, plants or seedlings were thawed overnight at 4°C. Plants were photographed after a 5-d recovery under normal growth conditions. For GSH application experiments, seeds were sown on MS medium plates containing 500 μm GSH or MS medium plates as controls for 11 d and then subjected to freezing treatment.

Electrolyte Leakage Measurement

Measurements of ionic leakage were conducted as described by Chen et al. (2008). Rosettes of plants were collected after NA or CA freezing treatments at the indicated times and thawed at 4°C overnight. Ionic leakage was measured using a conductivity meter (Mettler Toledo S220-USP/EP). The leaked ionic strength and total ionic strength were recorded for calculation of relative ionic leakage rates.

Total RNA Extraction and RT-qPCR

TRIzol (Invitrogen) was used for total RNA extraction according to the manufacturer’s recommendations. Reverse transcription was performed using the PrimeScript RT reagent kit with gDNA Eraser (Takara) to synthesize first-strand cDNA from 2 μg of total RNA. qPCR was performed by the SYBR Green method using the StepOnePlus Real-Time PCR system (Applied Biosystems) in 96-well blocks. The cycle threshold (Ct) value of each sample was calculated using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). The RT-qPCR primers used are listed in Supplemental Table S1. All RT-qPCRs were performed using total RNA samples isolated from three independent replicate samples with similar results.

Trypan Blue and DAB Staining

Trypan Blue and DAB staining were carried out as described previously (Xiao and Chye, 2011). Briefly, rosettes from NA and CA freezing plants were collected and boiled in Trypan Blue staining buffer for 1 min followed by incubating at room temperature for 10 min. Then, the leaves were transferred to tubes containing 70% chloral hydrate (w/v) for destaining. For DAB staining, the collected leaves were immersed in 1 mg mL⁻¹ DAB staining buffer (pH 3.8) for 5 h in the dark. Ethanol (95%) was boiled and used for chlorophyll clearing afterward.

Measurement of H₂O₂ and Chlorophyll Contents

The measurement of H₂O₂ contents was carried out as described by Yuan et al. (2017) using the Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Molecular Probes). Briefly, rosettes from NA and CA freezing plants were ground in liquid nitrogen, and 200 mL of reaction buffer was used for H₂O₂ extraction. A microplate reader (Tecan) was employed to detect the absorbance of H₂O₂ at 560 nm. The H₂O₂ concentration was calculated using a standard curve. Chlorophyll was extracted from Arabidopsis rosettes using 1 mL of N,N-dimethylformamide (Sigma; D4551) and kept at 4°C in the dark for 2 d. Absorbance at 664 and 647 nm was recorded. Total chlorophyll was calculated as described by Thompson et al. (2005). The chlorophyll contents of wild-type samples from normal growth conditions were set to 100%, and the relative chlorophyll contents in other samples were calculated accordingly.

Lipid Profiling

Total lipid extraction was performed as described previously (Welti et al., 2002). The profiles of membrane lipids were determined by electrospray ionization-tandem mass spectrometry as described previously (Xiao et al., 2010). Data shown in this study are means with sd from four independent samples, and each sample was pooled from the rosettes of five plants. DAG and TAG levels were separated and quantified following Peters et al. (2014).

Measurement of NADPH Oxidase Activity

The plasma membrane was isolated according to Qiu et al. (2002). NADPH oxidase activity measurements were conducted following Zhang et al. (2009b). Briefly, membrane vesicles were resuspended in 50 mm Tris-HCl buffer (pH 7.5). Then, the solution was mixed with the reaction buffer containing 50 mm Tris-HCl (pH 7.5) and 0.5 mm reduction of the tetrazolium dye, XTT. Fifty micromolar NADPH was added to initiate the reaction. After incubation at room temperature for 10 min, XTT formazan production at A₄₇₀ was detected using a microplate reader (Tecan). NADPH activity was represented as ΔA₄₇₀ per milligram of protein per minute.

Plasmid Construction, Transient Expression, and Microscopy Analyses

To construct the DGK2-GFP, DGK3-GFP, and DGK5-GFP vectors, full-length coding sequences of DGK2, DGK3, and DGK5 were inserted into BamHI-digested pUC119 and fused with the N terminus of eGFP. Transient expression assays in Arabidopsis protoplasts were carried out according to Qi et al. (2017). Briefly, protoplasts from 4-week-old wild-type rosettes were isolated and transfected with DGK2-GFP, DGK3-GFP, and DGK5-GFP vectors. Protoplasts were incubated for 12 h, and GFP signal was detected by confocal microscopy.

Accession Numbers

The sequence data discussed in this article can be found in the Arabidopsis Genome Initiative database under the following accession numbers: DGAT1 (AT2G19450), PDAT1 (AT5G13640), CBF1 (AT4G25490), CBF2 (AT4G25470), CBF3 (AT4G25480), COR47 (AT1G20440), RD29A (AT5G52310), KIN1 (AT5G15960), DGK1 (AT5G07920), DGK2 (AT5G63770), DGK3 (AT2G18730), DGK4 (AT5G57690), DGK5 (AT2G20900), DGK6 (AT4G28130), PLDα1 (AT3G15730), PLDδ (AT4G35790), SFR2 (AT3G06510), and RbohD (AT5G47910).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Expression patterns of DGAT1 and PDAT1 under cold stress (4°C).

• Supplemental Figure S2. Identification of dgat1 mutants.

• Supplemental Figure S3. DGAT1 knockout mutants show increased sensitivity to chilling stress.

• Supplemental Figure S4. Cold-inducible expression of CBFs and their regulons in the wild type and dgat1 mutants.

• Supplemental Figure S5. PA-induced H₂O₂ production and cell death require RbohD.

• Supplemental Figure S6. Profiles of DAG and TAG in rosettes of 4-week-old wild type, sfr2-3, and sfr2-4 before (22°C) and after NA or CA treatment followed by freezing exposure (−8°C for NA and −10°C for CA).

• Supplemental Figure S7. DGK2-GFP, DGK3-GFP, and DGK5-GFP fusions are localized in the cytosol.

• Supplemental Figure S8. Identification of dgk2-1, dgk3-1, and dgk5-1 mutants.

• Supplemental Figure S9. Lipid profiles in the rosettes of 4-week-old wild type, dgk2-1, dgk3-1, and dgk5-1 before (22°C) and after NA or CA treatment following freezing exposure (NA −8°C or CA −10°C).

• Supplemental Table S1. Primer sequences used in this study.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• CAT1 Gramene: AT1G20630

• CAT1 Araport: AT1G20630

• ICE1 Gramene: AT3G26744

• ICE1 Araport: AT3G26744

• MPK3 Gramene: At3g45640

• MPK3 Araport: At3g45640

• MPK6 Gramene: At2g43790

• MPK6 Araport: At2g43790

• NADPH CHEBI: CHEBI:16474

• PHYTOALEXIN CHEBI: CHEBI:26115

• RBOHD Gramene: AT5G47910

• RBOHD Araport: AT5G47910

• application CHEBI: CHEBI:33232

• chlorophyll CHEBI: CHEBI:28966

• EDS1 Gramene: AT3G48090

• EDS1 Araport: AT3G48090

• ethanol CHEBI: CHEBI:16236

• lipid CHEBI: CHEBI:18059

• pad4 Gramene: AT3G52430

• pad4 Araport: AT3G52430

• sag101 Gramene: AT5G14930

• sag101 Araport: AT5G14930

• sfr2 Gramene: AT3G06510

• sfr2 Araport: AT3G06510

• trypan blue CHEBI: CHEBI:78897