Phospholipase Dα6 and phosphatidic acid regulate gibberellin signaling in rice

Abstract

Phospholipase D (PLD) hydrolyzes membrane lipids to produce phosphatidic acid (PA), a lipid mediator involved in various cellular and physiological processes. Here, we show that PLDα6 and PA regulate the distribution of GIBBERELLIN (GA)‐INSENSITIVE DWARF1 (GID1), a soluble gibberellin receptor in rice. PLDα6‐knockout (KO) plants display less sensitivity to GA than WT, and PA restores the mutant to a normal GA response. PA binds to GID1, as documented by liposome binding, fat immunoblotting, and surface plasmon resonance. Arginines 79 and 82 of GID1 are two key amino acid residues required for PA binding and also for GID1’s nuclear localization. The loss of PLDα6 impedes GA‐induced nuclear localization of GID1. In addition, PLDα6‐KO plants attenuated GA‐induced degradation of the DELLA protein SLENDER RICE1 (SLR1). These data suggest that PLDα6 and PA positively mediate GA signaling in rice via PA binding to GID1 and promotion of its nuclear translocation.

Phospholipase Dα6 and the lipid phosphatidic acid (PA) mediate gibberellin (GA) signalling in rice. PLDα6 and PA promote the nuclear localization of the gibberellin receptor GID1 and the degradation of the DELLA protein SLENDER RICE1 (SLR1), enhancing GA signaling.

Introduction

Gibberellins (GAs) are important hormones affecting plant growth and development throughout the life cycle, ranging from seed germination, stem elongation, leaf expansion, and flowering to fruit development (Sun, 2010; Hu et al, 2018). Genetic studies of GA‐deficient and GA‐response mutants have led to identification of key components in GA action and signaling. In rice, the GA receptor GIBBERELLIN‐INSENSITIVE DWARF1 (GID1) interacts with the aspartate–glutamate–leucine–leucine–alanine motif‐containing DELLA protein SLENDER RICE1 (SLR1) to form a GA‐GID1‐DELLA complex in a GA‐dependent manner. The GA‐GID1 binding stimulates the interaction of DELLA with rice Skp1‐Cullin1‐F‐box (SCF) SCF^GID2 protein, leading to the DELLA degradation via a ubiquitin‐dependent pathway and the consequent activation of GA response (Ueguchi‐Tanaka et al, 2005, 2007; Hirano et al, 2010). DELLAs are transcriptional factors in nuclei, whereas GID1 contains no apparent nuclear localization sequence, but how GID1 is localized to nuclei remains elusive.

Increasing results indicate that membrane lipids are rich sources for signaling messengers in plant response to hormones and stress conditions (Wang, 2004; Testerink & Munnik, 2005; Yao & Xue, 2018). Phospholipase D (PLD) hydrolyzes membrane lipids to generate phosphatidic acid (PA) that acts as lipid messengers (Mishra et al, 2006; Zhang et al, 2009; Yu et al, 2010; Zhang et al, 2012). PA can bind to target proteins to regulate biological processes (Min et al, 2007; Guo et al, 2012a; Yao et al, 2013). The PA binding may enhance or inhibit the catalytic activity of target proteins (Zhang et al, 2004; Guo et al, 2012b; Anthony et al, 2014), tether protein to subcellular membranes (Gao et al, 2013; McLoughlin et al, 2013), and/or promote the formation and/or stability of protein complex (Huang et al, 2006; Li et al, 2012, 2015). The rice genome contains 17 PLDs that can be subdivided into 3 groups, including the calcium‐dependent phospholipid‐binding C2‐PLDs, the polyphosphoinositide‐interacting PX/PH‐PLDs, and a putative signal peptide‐containing SP‐PLD (Li et al, 2007). PLDs play important and diverse roles in rice, such as responses to salt, cold, drought, and disease. OsPLDα1 is involved in salt tolerance through mediating the H⁺‐ATPase activity and transcription (Shen et al, 2011). In addition, OsPLDα1 affects cold stress response through its product PA regulating the expression of OsDREB1 (Huo et al, 2016). Overexpression of PLDα1 in upland rice improved drought tolerance by maintaining the photosynthetic apparatus integrity (Abreu et al, 2018). Two chloroplast‐localized PLDs, OsPLDα4 and OsPLDα5, regulate herbivore‐induced direct and indirect defenses (Qi et al, 2011). OsPLDβ1 mediates disease response and stimulates abscisic acid (ABA) signaling by activating the protein kinase SAPK to repress GAMYB expression and inhibit seed germination (Li et al, 2007b; Yamaguchi et al, 2009). Here, we report that PLDα6 and PA mediate the subcellular localization of the GA receptor GID1 and longitudinal cell growth in rice.

Results

Knockout of PLDα6 decreases GA sensitivity in rice

The rice genome has 17 putative PLDs that are designated as PLDα (8), PLDβ(2), PLDδ(3), PLDζ(2), PLDκ(1), and PLDφ(1). PLDαs, PLDβs, PLDδs, and PLDκ contain the calcium/lipid‐binding C2 domain and PLDζs have the PX and PH domains, whereas PLDφ has a signal peptide at the N‐terminus. All C2‐PLDs and PX/PH‐PLDs contain two HKD (HxKxxxxD) catalytic motifs except that PLDα7 has a mutation (RxKxxxxD) in the second HKD motif (Fig EV1A). PLD‐C2 domain (calcium/lipid binding) that consists of eight strands was also found in PLDαs (Hong et al, 2016), and all the strands except the second share high homology between PLDα6 and other PLDαs (Appendix Fig S1). PLDγs were identified in Arabidopsis but not in rice, whereas PLDκ and PLDφ were in rice but not in Arabidopsis. Comparison of the amino acid sequences of OsPLDs with those of different plant species suggests that PLDγ, PLDκ, and PLDφ may be duplicated lately as they are not found in the moss Physcomitrella patens (es) or the lycophyte Selaginella moellendorffii, and they differ between the dicot Arabidopsis thaliana and monocot plants. By comparison, PLDαs and PLDδs are found in lower and higher plant species, suggesting that they are original PLDs and conserved among plant species (Fig EV2).

Figure EV1. Sequence alignment of PLDs and gene expression of PLDαs in rice.

1. Alignment of deduced amino acid sequences of rice PLDαs. The sequences were aligned using the website http://www.ebi.ac.uk/Tools/clustalw2/index.html, and the conserved domains were identified using the website http://www.ncbi.nlm.nih.gov/Structure/cdd.shtml. The gray highlighting indicates the region of two conserved HxKxxxxD motifs. PLDα6 was marked with red color. “*” indicates positions which have a single, fully conserved residue. “:” indicates residues with strong similarities of physicochemical properties, and “.” indicates residues with weak similarities of physicochemical properties.
2. Expression of rice PLDαs in various tissues and in response hormones based on data from CREP (http://crep.ncpgr.cn/crep‐cgi/home.pl). Values are means ± SD (n = 5 independent experiments).
3. Expression of PLDα3 and PLDα6 in various tissues and leaf response to hormones. The transcript level of rice PLDαs was quantified by real‐time PCR (qRT–PCR) normalized to GAPDH. Values are means ± SD (n = 3 technical replicates). IAA, KT, and NAA (10 μM each) were sprayed on leaves, and 2 hours later, samples were collected for RNA extraction.

Figure EV2. Phylogenetic analysis of PLDs.

Plant PLD protein sequences were aligned using ClustalW 2, and the result was analyzed using MEGA 5. At, Arabidopsis thaliana (green); Os, Oryza sativa (red); Pp, Physcomitrella patens (purple); Sm, Selaginella moellendorffii (olive); and Ps, Picea sitchensis (light blue). Rice PLDα6 (red five‐pointed star) homologs were identified in eudicot, monocot, lycophyte, and bryophyte. PLDαs are highlighted with yellow background. All the sequences are obtained from UniProt database (https://www.uniprot.org/), and accession numbers are shown in Appendix Table S3. AtPLDγ, PsPLDδ, OsPLDκ, and OsPLDφ were marked with green, light blue, purple, and black solid circle, respectively.

Based on the rice CREP chip database (http://crep.ncpgr.cn/crep‐cgi/home.pl), four PLDαs were highly differentially expressed in rice tissues (Fig EV1B). The transcript of PLDα6 and PLDα5 was detected primarily in stems and roots, respectively, whereas that of PLDα2 and PLDα8 was detected mainly in inflorescence. In comparison, the transcript of PLDα1 and PLDα3 was high at almost all tissues (Fig EV1B). To verify the results, we performed real‐time PCR (qRT–PCR) and found that PLDα6 transcript was highest in stems, while it was detectable in other seedling tissues, including leaves, leaf sheath, roots, and inflorescence. In addition, the transcript level of PLDα6 in leaves was increased in response to GA₃ and naphthylacetic acid (NAA), but not to kinetin (KT), whereas that of PLDα3 was increased in response to all the hormones tested (Fig EV1C).

To investigate the function of rice PLDs, T‐DNA‐insertional mutants for various PLDs, including pldα1, pldα3, pldα6, and pldδ2, were isolated and tested for sensitivity to ABA, indole‐3‐acetic acid (IAA), and GA₃. The pldα6 mutant was less sensitive to GA₃ than wild type (WT) (Fig EV3). To confirm the function of PLDα6, we genetically complemented the pldα6 mutant with the native PLDα6 (PLDα6‐COM) (Fig 1A). The expression level of PLDα6 in pldα6 and PLDα6‐COM was verified by quantitative real‐time PCR. The lack of PLDα6 transcript in the mutant indicates that pldα6 is a knockout (KO) mutant, whereas PLDα6‐COM restored PLDα6 expression in the mutant to that of WT plants (Fig 1B).

Figure EV3. Phenotype observation of different rice PLD mutants in GA response.

Three‐day‐old seedlings of PLD mutants, pldα1, pldα3, pldα6, and pldδ2 were transferred to nutrient solutions containing 0 and 1 μM GA₃. Photographs were taken from seedlings 4 days after growth in the presence or absence of GA₃. Scale bar = 4 cm. The vertical dotted red line separates WT and pld mutant plants.

Figure 1. Decreased GA response in PLDα6‐KO plants.

• A
  T‐DNA insertion site in the PLDα6 gene and the complementation construct introduced into the T‐DNA insertion mutant. Boxes denote exons and lines introns.
• B
  PLDα6 transcript in WT, pldα6, and complementation line (PLDα6‐COM). Leaf samples from 4‐leaf stage rice (Dongjin background) were collected, and the expression levels of PLDα6 were analyzed by normalizing to that of GAPDH. Values are means ± SD (n = 3 biological repeats).
• C
  Seedling phenotype under GA treatment. After germination, five‐day‐old seedlings with the same growth stage were transferred to 0.5 MS liquid media without or with different concentrations of GA₃. PA from soybean was added to the media at a final concentration of 20 μM. Pictures were taken 7 days after transfer. The horizontal red line separates different plants, and the vertical red scale bar represents 2 cm.
• D
  Seedling length of WT, pldα6, COM, and PA‐treated plants grown on 0, 0.1, 1, and 10 μM GA₃ for 7 days. Values are means ± SD (n = 15 plants) from one representative of three independent experiments.
• E
  Fresh weight of 10 seedlings of WT, pldα6, COM, and PA‐treated plants grown on 0, 0.1, 1, and 10 μM GA₃ for 7 days. Values are means ± SD (n = 15 plants) from one representative of three independent experiments.

WT, pldα6, and COM plants displayed no overt morphological alterations under normal growth conditions at the early stage (4‐leaf‐old) of rice. However, at the mature stage, the longitudinal growth including plant height, panicle length, and flag leaf length in the pldα6 mutant was significantly reduced as compared to WT and COM when plants were grown in the field (Appendix Fig S2). In addition, pldα6 was delayed in flowering time reduced in spikelet numbers compared to WT and COM plants, whereas tiller number and leaf width were similar among WT, pldα6, and COM plants (Appendix Fig S2).

To further characterize the GA response, five‐day‐old seedlings germinated under normal conditions were transferred to liquid media containing 0, 0.1, 1, and 10 μM GA₃. After one week, WT and pldα6 plants displayed significant differences in seedling length and fresh weight in GA₃‐containing media (Fig 1C). The seedlings of pldα6 were shorter and lighter than those of WT (Fig 1D and E). In the presence of 0.1 μM GA₃, the plant height of pldα6 seedlings was increased by 12%, whereas that was increased by 30% in WT and COM, compared to corresponding plants without GA treatments. With increasing GA concentrations, the differences between pldα6 and WT became greater in the GA’s growth‐promoting effect. These results indicate that the loss of PLDα6 decreases GA₃ sensitivity in rice.

OsPLDα6 hydrolyzes phospholipids and affects PA content and lipid composition

To test whether PLDα6 encodes a functional PLD, the PLDα6 cDNA was tagged at the C‐terminus with polyhistidine (6xHis) and expressed in E. coli (Fig 2A). Purified PLDα6 displayed Ca²⁺‐dependent hydrolysis of phospholipids with the highest activity at the mM levels of Ca²⁺ toward phosphatidylcholine (PC). PLDα6 also hydrolyzed phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and phosphatidylserine (PS), and the highest activity was detected with PE (Fig 2B, Appendix Fig S3). The apparent production of PA associated with the empty vector pET28‐His control could due to background contaminants in the sample purified from the E. coli harboring the empty vector. To correct the background PA, we determined PA produced by subtracting PA presented with empty vector from PA produced in the presence of PLDα6.

Figure 2. PLDα6 production and hydrolysis of phospholipids.

1. Immunoblotting of His‐tagged PLDα6 (arrow) expressed in E. coli as separated on an 10% SDS–PAGE and blotted to a membrane.
2. Lipid hydrolyzing activity assayed in the presence of different phospholipids using purified PLDα6 from E. coli. Solid bars are activities assayed using purified PLDα6, whereas open bars were empty vector control that used an equal volume of eluents from bacteria containing the empty vector that was identically processed as those expressing PLDα6. Values are means ± SD (n = 3 biological replicates).

To examine the effect of PLDα6 on lipid metabolism in rice plants, we analyzed lipids from 2‐week‐old leaves of WT, pldα6 and COM, using mass spectrometry. The level of monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG) was similar among pldα6, WT, and COM plants with and without the GA₃ treatment. Without GA, PA content in pldα6 plants was ∼70% of that in WT plants, whereas the level of PC (130% of WT) and PE (115% of WT) was significantly higher in pldα6 than that in WT and COM plants (Fig 3A). The decreased PA level in pldα6 was due to primarily the decrease in 34‐carbon and 36‐carbon PA species, whereas the increased PC and PE resulted from elevated 34‐ and 36‐carbon species (Fig 3B). The results indicate that PLDα6 contributes to the production of basal PA and that PLDα6 prefers PC and PE as the main substrates in vivo under normal condition. In the presence of GA, the content of PE (118% of WT) and PS (194% of WT) in pldα6 was significantly higher, whereas that of PA in the mutant was lower than that of WT (Fig 3A). The decreased PA in pldα6 was due primarily to reduced 34:2‐, 34:3‐, 36:4‐, and 36:5‐PA species, whereas increased PE in pldα6 resulted from elevating 34:2‐, 34:3‐, 36:4‐, and 36:5‐PE species (Fig 3B). These data indicate that PE and PS may be the main source of PA in response to the GA₃ treatment.

Figure 3. Effect of PLDα6‐KO on lipid changes in response to GA.

1. Total lipid levels in WT, pldα6, and COM without and with 10 μM GA₃. Leaf samples (15 seedling each) from 4‐leaf stage rice (Dongjin background) were collected, and lipids were extracted and profiled using ESI‐tandem mass spectrometry. Values are means ± SD (n = 3 biological replicates). MGDG, monogalactosyldiacylglycerol; DGDG, digalactosyldiacylglycerol; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PS, phosphatidylserine; and PA, phosphatidic acid.
2. Phospholipid species in WT, pldα6, and COM. Leaf samples (15 seedling each) from 4‐leaf stage rice (Dongjin background) were collected and, the different phospholipid species including carbon number and double bond number were determined. Values are means ± SD (n = 3 biological replicates). GA₃ was dissolved in ethanol, and seedlings treated with the same ethanol concentration were used as control.

Data information: In panels A and B, * denotes significant at P < 0.05 compared with WT under the same treatment based on Student’s t‐test.

PA promotes GA response and interacts with GA receptor GID1

To probe how PLDα6 affects GA response, we tested whether the PLD lipid product PA could restore pldα6’s GA₃ response to that of WT. In the presence of GA₃ and PA, the length and fresh weight of pldα6 seedlings were comparable to those of WT and COM (Fig 1C–E). The cell length in the second leaf sheath of pldα6 was 60% of that WT and COM in the presence of 10 μM GA₃ and was recovered to that of WT when PA was supplied to the growth media (Appendix Fig S4A and B). The PA restoration of pldα6 seedling growth to that of WT suggests that PLDα6‐produced PA is likely responsible for the effect of PLDα6 on GA‐promoted growth.

To explore how PA mediates rice response to GA, we examined potential interactions between PA and GID1 and SLR1, two major components of GA perception and signal transduction in rice. Using the same amount of OsGID1 and OsSLR1 proteins produced and purified from E. coli (Fig 4A), GID1 displayed a strong binding signal to PA, whereas SLR1 and control pET28 protein gave no signal on a lipid‐protein blotting assay. No signal was detected for binding between GID1/SLR1 and other phospholipids, such as PC, PE, PG, and PS (Fig 4B). In addition, liposome binding was performed to verify the interaction between PA and OsGID1. GID1 was co‐precipitated with liposomes consisting of PA: PC (1:3 molar ratio), and the amount of GID1 associated with liposomes increased with increasing amounts of PA in the liposomes. In contrast, liposomes with PC alone failed to bind to GID1. Similar to the lipid blotting, no PA binding with SLR1 was detected in the liposomal assay (Fig 4C).

Figure 4. PA‐GID1 interaction and amino acid residues involved in the binding.

1. Immunoblotting of His‐GID1 and His‐SLR1 expressed in E. coli.
2. Lipid–protein blotting assay of PA, PC, PE, PG, and PS with GID1 and SLR1. Lipids (0.5 μg) were spotted on nitrocellulose strips. PA, PC, PE, and PG were from egg yolk, and PS was from porcine brain. Purified proteins (GID1 and SLR1, 0.5 mg/ml) were used, followed by immunoblotting with anti‐His‐tag antibodies and color development.
3. Liposomes were made from di18:1‐PC only or di18:1‐PA/PC (PA:PC = 1:3 mole ratio). 1× and 10× refer to the concentration of PC or PA/PC liposomes used. NL, no liposome was added to the binding mixture.
4. Schematic diagram showing serial deletions of GID1. GID1 fragments were expressed in E. coli and used for defining the PA‐binding region. CD1/2 denote conserved domain HGG and GXSXG. Catalytic triad including three conserved amino acids of GID1, S, D, and H (vertical red bars).
5. Immunoblotting of His‐GID1 proteins using constructs shown in (D). Proteins were separated by SDS–PAGE, followed by immunoblotting with anti‐His‐tag antibodies. The PA‐binding activity of different truncation mutations was analyzed by fat immunoblotting. The red arrowheads indicate truncated proteins with different molecular weights.
6. Sequence alignment of the PA‐binding fragment of GID1 with that of the PA‐binding motifs in chicken Raf1, abscisic acid‐insensitive 1 (ABI1), constitutive triple response1 (CTR1), and werewolf (WER) from Arabidopsis. Residues in bold are basic, potentially involved in PA binding and were mutated to Ala in GID1.
7. Immunoblotting of His‐GID1 mutants and lipid immunoblotting of PA binding by GID1 mutant proteins on a filter.
8. Liposomal binding of GID1 proteins to PA. Liposomes were made from di18:1‐PA/PC (PA:PC = 1:3 mole ratio). Liposomal associated proteins were subjected to SDS–PAGE and immunoblotting using anti‐His antibodies. The band intensity was analyzed by ImageJ, and the intensity of input was set as 100%.

Furthermore, the PA‐GID1 binding was verified by surface plasmon resonance (SPR). In the representative sensorgram, a strong increase in response units (RUs) occurred when PA‐containing liposomes were infused to a GID1‐containing chip PA, whereas only a slight increase was detected when liposomes containing PC, PS, or PG‐only were injected (Appendix Fig S5). Association (k _a) and dissociation (k _d) constants for PA were 117 M⁻¹ s⁻¹ and 1.37 × 10⁻⁵ s⁻¹, respectively, with a binding affinity (K _D = k _d/k _a) at 1.2 × 10⁻⁷ M. By contrast, k _a and k _d for PC were 60.9 M⁻¹ s⁻¹ and 7.17 × 10⁻⁴ s⁻¹, respectively, resulting in a KD of 1.2 × 10⁻⁵ M. Thus, the binding affinity of GID1 to PA was ˜100‐fold greater than that of PC. The results suggest that GID1 binds to PA in vitro with a high affinity.

Arg79 and Arg82 of GID1 are required for PA binding

To identify the protein region involved in PA binding, several deletion mutants of GID1 were constructed and expressed in E. coli. Rice GID1 shares a high homology with hormone‐sensitive lipase (HSL) which contains three conserved domains, HGG motif, GXSXG motif, and a catalytic triad (Ueguchi‐Tanaka et al, 2005). The N‐terminal truncated mutants covering residues 1–119 (F1) and 1–138 (F2) displayed the strongest binding signal to PA. The C‐terminal truncated mutant containing 51 to 354 amino acid (aa) (F5) also showed PA binding, whereas no binding was found between PA and the shorter C‐terminal fragments covering 139–354 aa (F4) and 120–354 aa (F3) (Fig 4D and E). These results suggest that the PA‐binding site is located in the N‐terminal 51–119 residues of GID1.

Basic amino acid residues involved in binding to PA have been documented (Zhang et al, 2004; Testerink & Munnik, 2005; Awai et al, 2006; Wang et al, 2006). There are six Arg residues in the putative PA‐binding region covering from 51 to 119 aa (Fig 4F). Therefore, we generated eight mutants with the basic Arg residues substituted with Ala, including six single mutants GID1_R51A, GID1_R52A, GID1_R58A, GID1_R79A, GID1_R82A, and GID1_R98A and two double mutants GID1_R51AR52A and GID1_R79AR82A. All GID proteins with a single‐site mutation still exhibited binding to PA as did GID without mutation. However, the mutant GID1_R79AR82A, but not GID1_R51AR52A, lost PA binding (Fig 4G and H). The results suggest that Arg79 and Arg82 of GID1 are two key amino acid residues for PA binding.

PLDα6 and PA promote OsGID1’s nuclear localization

To determine how PLDα6 and PA modulate GID1 functions, we examined the effect of PLDα6 and PA on the nuclear localization of GID1. We first assessed the subcellular localization of PLDα6, using a PLDα6 fused with the green fluorescent protein (GFP) at the C‐terminus transiently expressed in rice protoplasts. PLDα6‐GFP was detected in both nucleus and cytosol, and its nuclear localization was enhanced by GA supplementation. In contrast, the IAA treatment did not change the intracellular distribution of PLDα6‐GFP (Fig 5A). Subcellular fractionation also showed that more PLDα6 was detected in the nuclear fraction when cells were treated with GA₃ (Figs 5B and EV4A and B).

Figure 5. Subcellular localization of PLDα6.

1. PLDα6‐GFP distribution in rice protoplasts with different concentrations of GA₃ or IAA. Protoplasts from 12‐day‐old rice (ZH11 background) leaf sheath tissue were collected. Rice protoplasts were transfected with pM999‐PLDα6 for 12 h. After that, GA₃ or IAA was added to protoplasts, and 2 h later, confocal images of protoplasts are shown. pM999‐GFP refers to the empty vector with GFP only that was transformed as control and Ghd7‐RFP was a nucleus marker. Scale bar = 10 μm.
2. Immunoblotting of PLDα6 in subcellular fractions. Total (T), soluble (S), and nuclear (N) proteins were isolated from PLDα6:GFP expressed in rice protoplasts treated with 10 μM GA₃ or IAA. Equal amount of each sample was loaded for SDS–PAGE and blotting.

Figure EV4. Subcellular location of PLDα6 in tobacco.

1. PLDα6‐GFP distribution in tobacco with or without GA₃ or IAA treatment. Tobacco leaves were infiltrated with agrobacteria harboring the PLDα6‐GFP construct. Two to three days after infiltration, 10 μM GA₃ or IAA was sprayed on tobacco leaves, and 2 h later, confocal images of the leaf samples were taken. Scale bar = 20 μm.
2. Immunoblotting analysis of the subcellular distribution of OsPLDα6 in tobacco leaves when treated with GA₃ or IAA. Total proteins (T) were extracted from infected leaves, and soluble (S) and nuclear (N) fractions were isolated. Equal amounts of proteins from each sample were loaded for SDS–PAGE and performed a Western blot.

To test whether PLDα6 and PA affect GID1’s subcellular localization, GID1 fused with GFP was expressed in the protoplasts isolated from seedlings of WT and pldα6. Rice Ghd7 (a PSEUDO‐RESPONSE REGULATOR 7‐like protein) was used as a nuclear localization marker (Xue et al, 2009). GID1 in WT protoplasts was localized in the nucleus, whereas it was localized in both nucleus and cytosol of pldα6 protoplasts. The presence of added GA₃ did not affect the GID1’s subcellular distribution in either WT or pldα6 cells (Fig 6B and D).

Figure 6. Effect of PLDα6‐KO on subcellular localization of GID1.

1. Protoplasts from 12‐day‐old rice (ZH11 background) leaf sheath tissue were collected, and protoplasts were transfected with pM999‐GID1 and mutation constructs and imaged 12 h after transfection. All confocal images were scanned using similar laser gain and offset settings. Bars = 10 μm.
2. Subcellular location of GID1 in WT and pldα6 protoplasts with or without GA₃ treatments. Cells for 12 h after transformation were treated with or without 10 μM GA₃ for 2 h. Bars = 10 μm.
3. Immunoblotting of subcellular fractions of GID1 and mutations expressed in rice protoplasts. Total (T), soluble (S), and nuclear (N) protein fractions were isolated from protoplasts for 12 h after transformation. Equal amounts of proteins from each sample were loaded.
4. Immunoblotting of subcellular fractions of GID1 expressed in WT and pldα6 protoplasts. After 2‐h treatment with 10 μM GA₃, total (T), soluble (S), and nuclear (N) protein fractions were isolated from 10 samples of protoplasts. Equal amount of each sample was loaded for SDS–PAGE and immunoblotting.

To further determine whether the PA‐GID1 interaction is required for GID1’s nuclear localization, we expressed the non‐PA‐binding GID1_R79AR82A mutant fused with GFP and GID1‐GFP, as well as six single GID mutants and GID1_R51AR52A fused with GFP in rice protoplasts. The GFP signal in the mutant GID1_R79AR82A‐GFP was detected in the cytosol but not in the nucleus (Fig 6A). In contrast, the GFP signal in cells expressing GID1‐GFP, six single mutant‐ and GID1_R51AR52A‐GFP fusions was co‐localized with the nuclear marker Ghd7. To verify the subcellular distribution, we performed subcellular fractionation of rice protoplasts, followed by immunoblotting, and verified the loss of nuclear localization of the GID1_R79AR82A mutant (Fig 6C). In addition, we transiently expressed GID1_R79AR82A‐GFP and GID1‐GFP in tobacco leaves and verified the loss of GID1_R79AR82A’s nuclear localization (Fig EV5A and B). These data suggest that PLDα6 and PA are important for the subcellular distribution of GID1 and that the PA‐GID1 binding is required for its nuclear localization.

Figure EV5. Subcellular localization of OsGID1 and its mutations in tobacco leaves.

1. Tobacco leaves were infiltrated with agrobacteria harboring plasmids with GID1 or its mutants. Two to three days after infiltration, leaf samples were observed. DAPI staining indicates nucleus. All confocal images were scanned using similar laser gain and offset settings. Bars = 20 μm.
2. Immunoblotting of subcellular fractions of OsGID1 and its site‐directed mutations in tobacco leaves. Total proteins (T) were extracted from the infected leaves, and soluble (S) and nuclear (N) fractions were isolated. Equal amounts of proteins from each sample were loaded for SDS–PAGE and immunoblotting.

PLDα6 promotes SLR1 degradation in response to GA

In the GA signal transducing process, the GA receptor GID1 interacts with DELLA proteins, such as SLR1, and GA promotes the degradation of SLR1, activating GA response (Ueguchi‐Tanaka et al, 2005, 2007; Hirano et al, 2010). To test the effect of PA‐GID1 interaction on the degradation of downstream target SLR1, we constructed the SLR1‐GFP in the plant transient expression vector pM999‐GFP and transformed the construct into rice protoplasts. SLR1‐GFP was localized to nuclei in WT and pldα6 cells with or without GA treatments. However, the addition of GA promoted SLR1 degradation, but GA‐promoted SLR1 degradation in pldα6 was less than that in WT (Fig 7A). The SLR1‐GFP signal in WT cells was decreased for 3 h after a GA₃ treatment, and no GFP signal was detected after 9 h of a GA treatment. In comparison, the GFP signal in pldα6 cells still remained high 9 h after the GA₃ treatment, being one‐third of non‐GA₃ treatment control (Fig 7A). In addition, we verified the effect of PLDα6 on decreased SLR1 degradation by immunoblotting because more SLR1portein was detected in pldα6 cells than WT cells after the GA treatment (Fig 7B). Those results from microscopic and immunoblotting observations both suggest that PLDα6 promotes SLR1 degradation in response to GA.

Figure 7. SLR protein stability in WT and pldα6 protoplasts.

1. Protoplasts of WT and pldα6 from 12‐day‐old stage leaf sheath were collected and transfected with a pM999‐SLR1 construct, incubated for 12 h, and then treated with 10 μM GA₃ for 0, 3, and 9 h. The first column, fluorescence from GFP; the second, red fluorescence from the nucleus marker Ghd7; the third, bright field; and the fourth, overlay of the three channels. All confocal images were scanned using similar laser gain and offset settings. Bars = 10 μm.
2. Immunoblotting of SLR1‐GFP proteins from protoplast cells. Equal amounts of proteins from each sample were used for SDS–PAGE, followed by immunoblotting with anti‐GFP antibodies.

In addition, we tested whether the knockout of PLDα6 affected the expression of genes involved in GA signaling and metabolism. Without added GA, the transcript level of GID1 and SLR1 was higher in WT than pldα6 (Appendix Fig S6). In response to a GA treatment, the transcript level of GID1 decreased, whereas that of SLR1 increased in WT and pldα6. Gibberellin 3β‐hydroxylase 1 (GA3OX1) and gibberellin 20‐oxidase 1 (GA20OX1) are involved in the production of bioactive Gas, whereas gibberellin 2‐oxidase 1 (GA2OX1) catalyzes the inactivation of GAs. There was no difference between WT and pldα6 in the transcript level of GA3OX1 or GA20OX1 with or without GA treatments. Without GA treatment, the transcript level of GA2OX1 was slightly lower in pldα6 than WT but with added GA, the level of GA2OX1 was similar in pldα6 and WT (Appendix Fig S6). With the limited number of genes tested, the results could mean that the knockout of PLDα6 did not alter significantly the expression of genes involved in GA metabolism.

Discussion

The PLD family has multiple members, and several PLDs are involved in plant response to hormones in Arabidopsis, such as PLDα1 and PLDδ in response to ABA (Sang et al, 2001; Zhang et al, 2004) and PLDζ2 in auxin (Li & Xue, 2007). The PLDs are more diverse in rice than Arabidopsis (Li et al, 2007), but their roles and molecular mechanisms of action remain largely unknown. Results of this study indicate that rice PLDα6 plays a positive role in longitudinal growth through its PA‐mediated GA response. PA binds to the GA receptor GID1 that is found in the nucleus (Ueguchi‐Tanaka et al, 2005). GID1 contains no apparent nuclear localization sequence (NLS), but how GID1 is localized to nuclei remained unknown. Our study showed that the loss of PLDα6 compromised the nuclear localization of GID1 and that the PA‐GID1 interaction is required for GID1’s nuclear localization in rice protoplasts. Previously, PA was found to facilitate the nuclear localization of a R2R3 MYB transcription factor involved in root hair formation in Arabidopsis, but enzymes responsible for that PA production remained unknown (Yao et al, 2013).

In this study, we found that PLDα6 was translocated from cytosol to nuclei in response to GA treatment. Nuclear membranes contain various phospholipids, such as PC, PE, and PS that are substrates for PLDα6 to produce PA. Thus, the activity of PLDα6 could potentially increase PA in the nuclear envelope in response to GA. However, direct measurement of GA‐induced PA increase in the nuclear envelope is technically challenging because fractionation would activate lipolytic enzymes, such as PLDs, thus altering lipid composition. The functional significance of PA in GA response is supported by the supplementation of PA that restored the normal GA response in of the PLDα6‐KO mutant. In addition, the requirement of PA‐GID binding in GID localization further supports the PA function. Those results indicate PLDα6 positively mediates GA via the lipid mediator PA. PA may tether GID1 to the nuclear membrane to facilitate GID1’s translocation from cytosol to the nucleus.

Comparative lipid analysis between WT and pldα6 indicates that PLDα6 is involved in the basal and GA‐induced PA production. However, it is possible that other enzymes are involved in PA production in response to GA. Specifically, the transcript level of PLDα3 is induced several‐fold by GA in leaves and the highest in stem relative to other tissues. In addition, WT and pldα6 plants displayed no overt growth difference at early seedling stages, suggesting also potential functional redundancy between PLDα6 and other PLDs. However, mature pldα6 plants are shorter than WT plants, which could mean that PLDα6 is a major mediator of the GA response and that the functional redundancy could occur more at early than mature growth stages. The transcript levels of genes involved in GA metabolism displayed a similar change in WT and pldα6 in response to GA₃. The result implicates that the decreased GA response in the pldα6 mutant did not directly result from altered GA metabolism in plants. On the other hand, with or without GA treatments, the transcript level of GID1 and SLR1 in pldα6 was both lower than that in WT. This decrease could mean that due to subdued GA signaling in the pldα6 mutant, less GID1 and SLR1 proteins are needed.

Based on the results, we propose a model of interaction between PLDα6/PA and GID1 in GA signaling (Fig 8). PLDα6 is translocated from cytosol to nuclei in response to GA, resulting in the production of more PA in nuclear envelope membranes. PA binds to GID1, resulting in the tethering of GID1 to nuclear membranes and facilitating the movement of GID1 from cytosol into the nucleus and interaction with the DELLA SLR1. The loss of PLDα6 attenuated the degradation of SLR1, which is consistent with the current model that the GA‐GID1‐SLR1 interaction is required for the degradation of the suppressor SLR1. Taken together, present results indicate that the lipid mediator PA and PLDα6 are new modulators in GA signaling, and they promote the nuclear localization of GID1 and enhance the GA suppressor DELLA degradation to enhance GA signaling and response. Further investigations are needed to establish the in vivo relevance of PA‐GID1 interactions and its role in mediating nuclear localization.

Figure 8. Model of PLDα6/PA and GID1 interaction and function in GA signaling.

GA promotes PLDα6 translocation from cytosol to the nucleus, resulting in an increase of PA in nuclear membranes. PA binds to GID1, tethers it to the membrane, and facilitates its nuclear translocation and the degradation of the suppressor SLR1, enhancing rice response to GA. PLs, phospholipids.

Materials and Methods

Knockout mutant isolation and genetic complementation

A T‐DNA insert mutant in PLDα6, designated as pldα6, was identified from the stock at Salk Institute Genomic Analysis Laboratory (http://signal.salk.edu/). A PLDα6 homozygous T‐DNA insert mutant was isolated by PCR using the primer PLDα6‐1F/1R and border primer PLDα6‐2 (Appendix Table S1). A pair of PLDα6‐specific primer PLDα6‐3F/3R was used in RT–PCR to confirm the PLDα6 null mutant. To complement pldα6, the native PLDα6 plus 2 kb upstream of the start codon of PLDα6 was amplified and cloned into the pU2301 vector (Zhou et al, 2016). The plasmid was transformed into pldα6 plants by agrobacterium‐mediated transformation (Deng et al, 2019). The transformants were selected by hygromycin resistance and confirmed by PCR using the primer PLDα6‐4F/4R. The primers used are listed in Appendix Table S1.

Plant growth and treatments

WT, pldα6, and COM (PLDα6 complementation) plants were grown in 1/2 MS liquid media in growth chambers under 12‐h light/12‐h dark photoperiods (100 μmol m^‐1 s^‐1) at 28/23°C and 50% humidity. To screen for altered GA response, 3‐day‐old seedlings of WT and several PLD mutants (Appendix Table 2) were transferred to 0.5 MS liquid media with or without 1 μM GA₃ and seedling growth was measured a week after the treatment. For further GA treatment experiments, one‐week‐old rice seedlings were treated with various concentrations (0, 0.1, 1, 10 μM) of GA₃ and growth phenotypes were measured at different time intervals. The control seedlings were sprayed with the same volume of solution without GA₃ and shown as a mock.

Subcellular localization of PLDα6 and GID1

Rice PLDα6 and GID1 cDNAs were cloned by PCR amplification from a rice leaf cDNA pool using the primers PLDα6‐5F/5R and GID1‐1F/1R (Appendix Table S1). The PCR products were cloned into the pM999 that contains the p35S promoter and eGFP fusion at the C‐terminus. The PCR product of GID1 was used as the template to generate a series of site‐directed mutants in GID1. PLDα6, GID1, and GID1 mutants in pM999 were transfected into rice protoplasts for protein expression and subcellular localization. In brief, rice protoplasts were isolated from 15‐day‐old seedling (ZH11 background), and the pM999‐PLDα6, pM999‐GID1, and pM999‐GID1 mutant constructs were transfected to protoplasts by a polyethylene glycol (PEG)‐mediated transformation. After 12‐ to 16‐h incubation, GFP fluorescence was observed with a Lecia TCS SP2 confocal microscope. pM999‐Ghd7 was transfected into protoplasts as a nuclear marker. GA₃ of various concentrations (0, 0.1, 1, 10 μM) or 10 μM IAA was added to protoplasts to test its effect on the subcellular distribution.

Transient expression in tobacco leaves was also used for examining the intracellular distribution of PLDα6 and GID1. PLDα6‐GFP, GID1‐GFP, and GID1 mutant‐GFP constructs were introduced into Agrobacterium tumefaciens strain GV1301 by electroporation, and transformants were selected on LB plates containing 50 mg/ml kanamycin. Transformants were grown overnight in 5 ml liquid LB media and then centrifuged at 4,000 rpm for 10 min. The pellets were resuspended with 10 mM MgCl₂ plus 10 μl 100 mM acetosyringone to OD600 = 1.0 and used for infiltrating leaves on 4‐week‐old Nicotiana benthamiana plants. To facilitate the production of recombinant proteins, agrobacteria expressing the viral p19 protein that inhibits post‐transcriptional gene silencing was co‐infiltrated. The production of PLDα6‐GFP and GID1‐GFP proteins was visualized 2–3 days after infiltration, and the fluorescence images were observed using a Lecia TCS SP2 confocal microscope. DAPI (4´,6‐diamidino‐2‐phenylindole) was used for nuclear staining. For hormone treatments, the infiltrated leaves were sprayed with GA₃ or IAA at indicated concentrations for subcellular localization.

The subcellular fractionation analysis was performed using rice protoplasts and tobacco leaves as described with some modifications (Shen et al, 2019); the above infiltrated tobacco leaves or protoplast infected were homogenized with a chilled buffer (25 mM Tris–HCl, pH 7.5, 100 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM EGTA, 1% NP40, and 0.5 mM PMSF) (Deng et al, 2019). The homogenate was filtrated through 300 mesh sieves, and the filtered product was centrifuged at 1,500 g for 5 min at 4°C to obtain the crude supernatant and nuclear (pellet) fractions. The resulting supernatants were centrifuged at 12,000 g for 5 min at 4°C to obtain soluble fraction. Equal amounts of proteins from different fractions including total, nuclear, and soluble fraction were separated by 8% (w/v) SDS–PAGE and then transferred onto a polyvinylidene difluoride (PVDF) membrane for immunoblotting. After that, the membrane was incubated with anti‐GFP antibody (Sangon Biotech D110008) and then a secondary antibody conjugated with alkaline phosphatase (Sangon Biotech D110072) for 2‐h incubation before color development using a chemiluminescence method.

Expression and purification of His‐tagged PLDα6, GID1, and SLR1 protein

The cDNAs of PLDα6, GID1, and SLR1 were amplified with primers PLDα6‐6F/6R, GID1‐2F/2R, and SLR1‐1F/1R and then inserted into the pET28 vector. The deletion fragments GID1_1–119 , GID1_1–138 , GID1_51–354 , GID1_120–354, and GID1_139–354 were amplified using the pairs of primers GID1‐2F/GID1‐119R, GID1‐2F/GID1‐138R, GID1‐511F/GID1‐2R, GID1‐120F/GID1‐2R, and GID1‐139F/GID1‐2R (Appendix Table S1). To generate the site‐specific mutation of GID1, thee full‐length GID1 cDNA was used as the template for PCR amplification with mutant primers GID1‐512F/R, GID1‐52F/R, GID1‐58F/R, GID1‐79F/R, GID1‐82F/R, Gid98‐F/R, GID1‐5152F/R, GID1‐7982F/R, and GID1‐2F/2R (Appendix Table S1) to generate GID1_R51A , GID1_R52A , GID1_R58A , GID1_R79A , GID1_R82A , GID1_R98A , GID1_R51AR52A, and GID1_R79AR82A . All these GID1 mutation products were inserted into the pET28 expression vector. The plasmids were transformed into the Escherichia coli strain BL21 (DE3) and cultured in LB media. Bacteria harboring the plasmids grown to OD₆₀₀ ≈ 0.4–0.6 were induced with 0.4 mM isopropyl 1‐thio‐b‐D‐galactopyranoside (IPTG) for 4 h at 28°C. The expressed proteins were purified with 6xHis agarose beads (Novagen) according to the manufacturer’s instruction, and the amount of protein was determined using the dye‐binding protein assay kit (Bio‐Rad).

PLDα6 activity assay

PLDα6 activity was assayed using the condition previously described (Hong et al, 2008). Briefly, the reaction contained a buffer (50 mM CaCl₂, 100 mM MES, pH 6, 0.5 mM SDS) and 0.4 mM of lipid substrates (PC, PE, PG, or PS) which were dried under a stream of nitrogen and suspended in H₂O by sonication. Different Ca²⁺ concentrations from 0, 50 nM, 50 µM, to 50 mM were tested with PC as substrate. After the addition of purified PLDα6, the reaction was incubated for 30 min at 30°C and stopped by adding 1 ml of chloroform: methanol (1:2, v/v) and 0.2 ml of 1 M NaCl. The organic phase was dried under a stream of nitrogen and dissolved in 20 µl of chloroform. The product was loaded onto silica gel plate (Merck, TLC silica gel 60) and separated by the developing solvent chloroform:ethanol:triethylamine:water (10:11.3:11.7: 2.7 v/v); TLC plate was exposed to iodine to visualize lipids. Lipid spots corresponding to that of the PA standard were scraped from the TLC plate. Five µL of 5.4 µM 17:0 TAG was added to the sample as an internal standard, and the mixture was transmethylated in methanol containing 1% H₂SO₄ and 0.05% butylated hydroxytoluene at 90°C for 1 h. One milliliter of hexane and 1 ml of water were added, and the upper phase was removed for GC analysis. The amount of PA was determined by comparing the amount of fatty acids in PA with that in the internal fatty acid standard. PLD activities toward PC, PE, PG, and PS were calculated based on the PA quantification from the TLC plate, using PLD produced PA = PA detected with PLDα6 – PA with empty vector in the presence of a specific phospholipid (Peters et al, 2010).

Lipid–protein blotting and liposomal binding

The binding between protein and lipids on filters was performed as described (Stevenson et al, 1998; Cao et al, 2016) with some modifications. Lipids (5 µg) including PA, PC, PE, PG, and PS were spotted on a nitrocellulose filter, followed by incubation with purified His‐tagged protein to the final concentration of 0.5 mg/ml in PBST (0.1% Tween 20) overnight at 4°C. The filter was then washed and incubated with anti‐His antibody conjugated with alkaline phosphatase (Sigma). GID1 and SLR1 proteins that bound to lipids on filters were visualized by staining alkaline phosphatase activity.

Liposomal binding was performed as previously described with some modifications (Cao et al, 2016). Dioleoyl PC alone or mixed with dioleoyl PA (molar ratio 3:1) was dissolved in chloroform and dried under a stream of nitrogen. Lipids were rehydrated in a buffer (250 mM raffinose, 25 mM HEPES, pH 7.5, and 1 mM DTT) for 1 h at 42°C. Liposomes were produced using a liposome extruder (Avanti) to produce small unilamellar liposomes. Liposomes were diluted to 3.2 mM. For each assay, 320 nmol and 32 nmol of liposomes were incubated with GID1 and SLR1 proteins for 45 min at room temperature. A negative control used the binding mixture but without liposome added. Liposomes were pelleted at 14,000 g for 30 min, washed twice with the binding buffer, and pelleted again. Both liposome‐bound proteins and proteins remaining in the supernatants were detected by immunoblotting with anti‐poly His antibodies conjugated with alkaline phosphatase (1:10,000).

Surface plasmon resonance analysis

Surface plasmon resonance analysis was performed using a Biacore 2000 system as described with some modifications (Guo et al, 2012b). Purified His‐tagged GID1 (2 µM) was immobilized on the Biacore Sensor Chip NTA via Ni²⁺‐NTA chelation. For all experiments, running buffer (0.01 M HEPES, 0.15 M NaCl, 50 µM EDTA, pH 7.4) containing 500 µM NiCl₂ was injected to saturate the NTA with nickel. Di18:1‐PA/di18:1‐PC liposomes (200 µM) were suspended in the running buffer and injected in sequence over the surface of the sensor chip. The liposomes containing dioleoyl PC, PG, or PS only were used for control. The sensorgrams of association and dissociation for each protein–liposome interaction were determined and plotted by SigmaPlot 10.0. Kinetic constants, including association rate constant (k _a), an intermediate dissociation rate constant (k _d), and the equilibrium binding affinity constant (K _D), were analyzed using BIA evaluation software.

qRT–PCR analysis of gene expression

To monitor the expression pattern of rice PLDαs, seedling, leaf, leaf sheath, and root samples were collected from the 4‐leaf stage rice (Dongjin). Stem and inflorescence samples were collected from the heading stage. For hormone treatments, 10 μM GA₃, KT, and NAA were sprayed on the leaves of 4‐leaf stage rice seedlings, and 2 h later, leaves were collected for RNA extraction. To test the expression pattern of genes involved in GA signaling and metabolism, leaf tissues from 4‐leaf stage WT and pldα6 were collected. Total RNA from rice tissues was extracted using a TransZol reagent according to the manufacturer's instruction (Transgen Biotech) and treated with DNaseI (Thermo). cDNA synthesis was performed with TIANscript RT Kit (Transgen Biotech) from 5 μg of DNA‐free RNA and diluted to a final volume of 200 μl. A total of 4 μl of diluted cDNA was used for each quantitative RT–PCR (qRT–PCR) reaction. qRT–PCRs were prepared using SYBR Green Master Mix on a MyiQ single‐color real‐time PCR detection system (Bio‐Rad). GAPDH was used as a housekeeping gene to normalize the expression, and a 2^{−Δ C T} method was used to calculate the transcript level of genes tested. The primer sequences used for qRT–PCR are listed in Appendix Table S1.

Lipid analysis

Lipid profiling was carried out using the method described previously (Cao et al, 2016). Briefly, leaves from rice seedlings (4‐leaf‐old) were detached and immediately immersed in 4 ml of 75°C isopropanol (preheated) with 0.01% butylated hydroxytoluene (BHT) for 15 min, followed by the addition of 1.5 ml of chloroform and 0.5 ml of water. After shaking for 1–2 h, the solvent was transferred to a new clean tube. The leaves were re‐extracted with 5 ml chloroform: methanol (2:1, v/v) six times with agitation for 45 min each, and the extracts were combined and then washed with 1 M KCl, followed by another wash with water. The solvent was evaporated by nitrogen, and the remaining tissue was oven‐dried at 100°C and weighed. For each genotype and treatment, three leaf samples were extracted and analyzed separately. Lipid samples were introduced by continuous infusion into the ESI source on a triple quadrupole MS. Phospholipids and galactolipids were quantified by comparison of the peak for each lipid species to internal standards of the same class as described previously (Welti et al, 2002).

Author contributions

HC cloned expressed proteins, performed binding assays, activity assay, subcellular location, and phenotype analysis, and wrote the manuscript. RG made the complementary construct and rice transformation and field trail investigation and participated in article preparation. SY and YS analyzed the lipid data in vivo. WL and YZ isolated PLD mutants. QZ, XD, and PT participated in subcellular localization. SL, YH, and XW directed the project and article preparation.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

EMBO reports (2021) 22: e51871.

Data availability

No data that require deposition in a public database.