Plant materials and growth conditions<\/p>\n
The hot1 mutant was isolated from an ethyl methanesulfonate-mutagenized M2 population of the indica cultivar Shuihui527 (R527). Approximately 3,000 M1 plants were grown in the paddy field, and one M2 family exhibiting an obviously reduced seed setting rate was identified during the 2017 summer season in Wenjiang, Chengdu, when field temperatures at the heading stage frequently exceeded 35\u2009\u00b0C. Heat tolerance assay at the seedling stage (14-day-old plants, 45\u2009\u00b0C for 60\u2009h followed by at least 3 days of recovery) confirmed the heat-sensitive phenotype, and the mutant was designated hot1. For subsequent gene mapping, an F2 population was generated by crossing hot1 with R527.<\/p>\n
For functional analyses of OsALA5, several transgenic lines were generated in different genetic backgrounds. A complementation line (OsALA5-C) was produced by introducing the OsALA5 genomic sequence driven by its native 2-kb promoter region into the hot1 mutant background. In the O. sativa cv. Kasalath (Kas) genetic background, two independent knockout lines (OsALA5-KO1 and OsALA5-KO2) were generated using a CRISPR\u2013Cas9 system, and the overexpression lines (OsALA5-OE) were obtained by expressing OsALA5 driven by the CaMV 35S promoter. Additionally, a complementation line (CL) carrying a native promoter-driven OsALA5\u2013GFP fusion construct was created in the OsALA5-KO1 background for subcellular localization and a series of subsequent physiological analyses. All transformation procedures and molecular verifications are described in \u2018Generation of transgenic plants\u2019. Unless otherwise stated, rice plants were cultivated in the experimental field of Sichuan Agricultural University (Chengdu, China).<\/p>\n
To assess evolutionary conservation of OsALA5, we examined the corresponding Arabidopsis orthologues using both T-DNA and CRISPR\u2013Cas9 mutants. A. thaliana Col-0 served as the wild type. T-DNA insertion mutants37<\/a> of ALA9 (ala9, SALK_073953, At1g68710), ALA10 (ala10-2, SALK_024877, At3g25610), ALA11 (ala11-2, SALK_107029, At1g13210) and ALA12 (ala12, SALK_111498, At1g26130) were obtained from the Nottingham Arabidopsis Stock Centre (NASC), and the homozygous lines were verified by PCR-based genotyping following the guidelines provided by the Arabidopsis Information Resource (TAIR; https:\/\/www.arabidopsis.org<\/a>). Double mutants (ala9\/12 and ala10-2\/11-2) were generated by crossing and verified by PCR. Additional CRISPR\u2013Cas9 knockout lines alleles (ala10-1, ala11-1 and ala10-1\/11-1) were produced and confirmed by sequencing. Arabidopsis seeds were surface-sterilized and germinated on 0.5\u00d7 Murashige and Skoog agar plates at pH 5.8 as described previously51<\/a>. Ten-day-old seedlings were transferred to soil and grown at 22\u2009\u00b0C under long-day conditions (16\u2009h light:8\u2009h dark, 150\u2009\u00b5mol\u2009m\u22122\u2009s\u22121) until harvest. Primers used for vector construction, genotyping and mutation confirmation are listed in Supplementary Table 1<\/a>.<\/p>\n Heat tolerance assay at the seedling stage<\/p>\n For heat tolerance analysis at the rice seedling stage, healthy seeds were surface-sterilized with 3% sodium hypochlorite for 30\u2009min, rinsed thoroughly with sterile distilled water, and soaked at 37\u2009\u00b0C for 3 days to promote uniform germination. Germinated seeds were then placed into bottom-cut 96-well PCR plates and hydroponically cultured in Hoagland\u2019s modified nutrient solution (Coolaber) in a controlled growth chamber (WIPGC-B2P84, BPC600H, Fujian Jiupo Biotechnology) under normal condition (28\u2009\u00b0C, 14\u2009h light:10\u2009h dark photoperiod, 150\u2009\u00b5mol\u2009m\u22122\u2009s\u22121 light intensity, and 65% relative humidity). Fourteen-day-old seedlings were transferred to 45\u2009\u00b0C for the indicated durations under the same light and humidity conditions and then returned to normal condition for at least three days of recovery.<\/p>\n Seedling survival was evaluated following established criteria52<\/a>: seedlings that resumed growth and produced new green leaves were scored as alive, whereas those that remained fully bleached and failed to regrow were scored as dead. Survival rates were calculated as the percentage of living seedlings among the total number tested. For the PC spraying assays under heat stress, 14-day-old Kas and OsALA5-KO1 seedlings were prepared as described above. Seedlings were evenly sprayed with 3\u2009ml of buffer control (2% Tween-20 and 2% glycerol), 10\u2009\u03bcM PC (18:0\/18:0) or 10\u2009\u03bcM PC (18:1\/18:1). After spraying, seedlings were held under normal growth condition for ~10\u2009min to permit absorption and then subjected to heat treatment at 45\u2009\u00b0C for the indicated durations. Survival rates were determined using the same criteria described above. For Arabidopsis heat tolerance assays, 10-day-old seedlings grown on 0.5\u00d7 Murashige and Skoog agar medium were prepared as described in \u2018Plant materials and growth conditions\u2019. Plates were transferred to a 45\u2009\u00b0C growth chamber for the indicated durations and then returned to normal condition (22\u2009\u00b0C, 16\u2009h light:8\u2009h dark) for at least 3 days of recovery. Survival rates were determined using the same criteria as for rice. For all seedling stage heat tolerance assays, approximately 30 seedlings were grown per pot, and three independent pots were used per genotype for each biological replicate. Each experiment was independently repeated at least three times with consistent results. Data were analysed using GraphPad Prism (v8.0.2). The exact heat treatment durations and conditions for all genotypes and assays are summarized in Supplementary Table 2<\/a>.<\/p>\n Gene identification<\/p>\n MutMap analysis was performed for gene mapping as previously described16<\/a>. In brief, an F2 population of 300 individuals was generated by crossing the hot1 mutant with its wild-type parent R527. Genomic DNA from 30 F2 individuals displaying the extreme heat-sensitive phenotype was extracted and pooled in equal amounts for whole-genome resequencing (average depth ~25\u00d7 per sample). Genomic DNA from R527 was resequenced and used as the wild-type reference. The SNP index was calculated for each site to identify genomic regions associated with the mutant phenotype. Candidate SNPs showing complete linkage with the heat-sensitive trait were validated by Sanger sequencing of PCR-amplified genomic fragments. Primers used for SNP validation are listed in Supplementary Table 1<\/a>.<\/p>\n OsALA5 immunoblotting and immunoprecipitation ATPase assay<\/p>\n For OsALA5 immunoblotting and immunoprecipitation analyses, membrane proteins were extracted from 14-day-old hydroponically grown seedlings of wild-type R527 and the hot1 mutant using the procedure described in \u2018Co-immunoprecipitation assay\u2019. Immunoblotting was performed as described previously53<\/a> using anti-OsALA5 monoclonal antibody (generated as described in \u2018Monoclonal antibody generation against OsALA5\u2019) and anti-actin (Sangon Biotech, D191048) as a loading control; both antibodies were diluted 1:3,000. Signals were developed using ECL substrate and imaged using an e-Blot Touch Imager Pro. For ATPase measurements, OsALA5 complexes were immunoprecipitated as described in \u2018Co-immunoprecipitation assay\u2019. The resulting immunoprecipitates were then subjected to ATPase activity measurement following the procedure described in \u2018ATPase activity assay\u2019. ATPase activity measurements were obtained from four individual repeats with consistent results. Data were analysed using GraphPad Prism (v8.0.2).<\/p>\n For heat-induced protein accumulation analysis of OsALA5, 14-day-old Nipponbare (Nip) seedlings were exposed to 45\u2009\u00b0C and sampled at 0, 5\u2009min, 10\u2009min, 30\u2009min, 1\u2009h and 3\u2009h after treatment. Membrane proteins were extracted using the procedure described above, and OsALA5 immunoblotting was performed under the same conditions.<\/p>\n Generation of transgenic plants<\/p>\n For complementation of the hot1 mutant, the full-length genomic sequence of OsALA5 driven by its native 2-kb promoter amplified from R527 was cloned into the binary vector pCAMBIA1300. The construct was introduced into hot1 to generate the complementation line (OsALA5-C). More than 20 independent transgenic lines were obtained, and 1 T2 line plant with one copy insertion was used for subsequent analyses. For gene knockout in the Kas background, two independent OsALA5-knockout lines (OsALA5-KO1 and OsALA5-KO2) were generated using the CRISPR\u2013Cas9 system. The backbone vectors were gifts from Y. Liu, and the vector construction followed the approaches described by Ma et al. (2015)54<\/a>. The guide RNA site targeting the first exon of OsALA5 was designed using CRISPR-P 2.0 (http:\/\/crispr.hzau.edu.cn\/CRISPR2\/<\/a>) and assembled into the pYLCRISPR-Cas9P35S-H binary vector. More than 30 independent transgenic lines were obtained, and two homozygous T3 lines with stable phenotypes were used for subsequent analyses. For overexpression, the OsALA5 CDS was cloned downstream of the CaMV 35S promoter in the pCAMBIA2300 vector and transformed into Kas. More than 15 independent transgenic lines were obtained, and three homozygous T3 lines with higher expression levels were used for subsequent analyses.<\/p>\n To determine the subcellular localization of OsALA5 in vivo, the OsALA5 CDS was fused in-frame with GFP under the native promoter and cloned into pCAMBIA2300. The construct was introduced into the OsALA5-KO1 background, generating the complementation line (CL). More than 15 independent transgenic lines were obtained, and a single-copy T2 line was used for subsequent analyses. For all rice transformations, recombinant plasmids were introduced into Agrobacterium tumefaciens strain EHA105 and subsequently transformed into rice calli derived from mature embryos. Transformed calli were selected on Murashige and Skoog medium containing the corresponding selection antibiotic, regenerated, and transferred to soil. Genotyping of knockout lines was performed by Sanger sequencing. Complementation lines were verified by PCR for promoter-CDS insertion, and overexpression lines were validated by qPCR.<\/p>\n For the generation of Arabidopsis knockout lines, CRISPR\u2013Cas9 constructs targeting ALA10 and ALA11 were generated following the approach described by Ma et al. (2015)54<\/a>. Single guide RNAs (sgRNAs) were designed using the CRISPR-P 2.0 and assembled into the pYLCRISPR-Cas9P35S-H binary vector using a dicot-compatible assembly strategy. Constructs were delivered into A. thaliana Col-0 plants via the floral-dip method55<\/a> using A. tumefaciens GV3101. T0 seedlings were selected on 0.5\u00d7 Murashige and Skoog agar medium supplemented with 50\u2009mg\u2009l\u22121 hygromycin B (sigma), transferred to soil, and grown under long-day conditions (22\u2009\u00b0C, 16\u2009h light:8\u2009h dark). Target-site mutations were confirmed by PCR and Sanger sequencing using primers spanning the sgRNA recognition regions. More than 15 independent transgenic lines were obtained, and two homozygous T3 lines with stable phenotypes were used for subsequent analyses. All primers used for vector construction, genotyping and mutation confirmation are listed in Supplementary Table 1<\/a>.<\/p>\n Yeast strains and heat stress treatments<\/p>\n The S. cerevisiae mutant strain ZHY709 (MATa his3 leu2 ura3 met15 dnf1\u0394 dnf2\u0394 drs2::LEU2)56<\/a>, which lacks endogenous plasma membrane P4-ATPase activity, was used for co-immunoprecipitation, lipid uptake and FLIM\u2013FRET assays involving OsALA5 and OsALIS2. BY4741 (MATa his3 leu2 ura3 met15; EUROSCARF) served as the wild-type control. Plasmid construction for heterologous expression in yeast followed previously described procedures25<\/a>. In brief, full-length OsALA5 and OsALIS2 cDNAs were cloned into the yeast expression vectors pMP4062 and pMP3864, respectively. These vectors, designed for GAL-promoter-driven expression of plant P4-ATPases in yeast were gifts from R. L. L\u00f3pez-Marqu\u00e9s. Single knockout strains dnf1\u2206 (YDR093W, 6162) and dnf2\u2206 (YDR093W, 4028) were obtained from the Yeast Knock-Out Collection39<\/a>, and the dnf1\u2206 dnf2\u2206 double mutant was generated by homologous recombination using a PCR-based deletion strategy39<\/a>,57<\/a>. Yeast transformations were performed using the standard lithium acetate method58<\/a>, and transformants were selected on synthetic defined medium lacking the appropriate auxotrophic markers.<\/p>\n For heat tolerance assay, yeast cells of BY4741, dnf1\u0394, dnf2\u0394 and the dnf1\u0394 dnf2\u0394 double mutant were cultured overnight in YPD medium at 30\u2009\u00b0C with shaking (200\u2009rpm) until OD600\u2009\u2248\u20090.3. Cells were then diluted to OD600\u2009=\u20090.3 in fresh YPD and exposed to 55\u2009\u00b0C for 1\u2009h to induce heat stress, while control cultures were maintained at 28\u2009\u00b0C. After treatment, tenfold serial dilutions (100 to 10\u22123) were spotted onto YPD agar plates and incubated at 30\u2009\u00b0C for 2 to 3 days to assess growth recovery. For flow cytometric analysis of membrane integrity, control or heat-treated cells were resuspended in PBS (pH 7.4) containing 5\u2009\u00b5g\u2009ml\u22121 propidium iodide (IP50308, Solarbio) and incubated for 10\u2009min in the dark. Propidium iodide fluorescence was measured in the FL2-A channel using a BD Accuri C6 Plus flow cytometer controlled by BD Accuri C6 Plus software (BD Biosciences). No additional post-acquisition gating was applied, and all acquired yeast events were analysed under identical acquisition settings. Representative raw FSC-A\/SSC-A and propidium iodide area (PI-A) plots are provided in the Supplementary Fig. 2<\/a>. Membrane integrity loss was quantified as the mean propidium iodide fluorescence intensity (FL2-A) of the acquired yeast events. Each assay included four independent biological replicates with consistent results. To validate the flow cytometric results, aliquots of heat-treated cells (55\u2009\u00b0C, 30\u2009min) were stained with propidium iodide and imaged using a Zeiss LSM 800 confocal laser-scanning microscope controlled by ZEN software (Zeiss, v2.3) under identical acquisition settings. The percentage of propidium iodide-positive cells was calculated from at least three independent biological replicates. All data were analysed with GraphPad Prism (v8.0.2). Primers used for yeast strain construction, verification, and plasmid assembly are listed in Supplementary Table 1<\/a>. The exact heat treatment durations and conditions are summarized in Supplementary Table 2<\/a>.<\/p>\n RNA extraction and qPCR<\/p>\n Fourteen-day-old seedlings grown hydroponically in Hoagland\u2019s nutrient solution were used for all expression analyses. For expression analysis of OsALA5-OE lines, total RNA was extracted using Plant Total RNA Isolation Kit (FOREGENE). Approximately 500\u2009ng total RNA was reverse-transcribed using the RT Easy II Kit (with gDNase; FOREGENE) to synthesize first-strand cDNA. qPCR was performed on a Bio-Rad CFX96 system using AceQ qPCR SYBR Green Master Mix (Vazyme) and OsALA5-specific primers. UBQ5 (AK061988<\/a>) was used as the internal reference for normalization. For heat-induced expression analysis of OsALA5, 14-day-old Nip seedlings were exposed to 45\u2009\u00b0C and sampled at 0, 5\u2009min, 10\u2009min, 30\u2009min, 1\u2009h and 3\u2009h after treatment. Total RNA extraction, first-strand cDNA synthesis, and qPCR were performed as described above. Relative expression levels for the OsALA5-OE lines and heat-induced expression analyses were calculated using the 2\u2212\u0394\u0394Ct method based on Ct values obtained from the Bio-Rad CFX96 system. In each experiment, one control sample was used as the calibrator (\u0394\u0394Ct\u2009=\u20090), and its relative expression value was set to 1.0.<\/p>\n For expression analyses of haplotype accessions and NIL of OsALA5, RNA extraction and cDNA synthesis followed the same procedures as above. qPCR was conducted using a QX400 Real-Time PCR System (Sichuan JLM Technology). Relative expression values for these analyses were obtained directly from the instrument software after normalization to the internal reference gene. The accessions used for haplotype expression analysis are listed in Supplementary Table 3<\/a>, and all primers used for qPCR are listed in Supplementary Table 1<\/a>. Relative expression levels were calculated using the 2\u2212\u0394\u0394Ct method, and data were analysed with GraphPad Prism (v8.0.2).<\/p>\n Phylogenetic tree analysis<\/p>\n To identify the potential ALIS proteins in rice, full-length protein sequences of Arabidopsis ALIS1\u2013ALIS5 were retrieved from TAIR (https:\/\/www.arabidopsis.org\/<\/a>) and used as queries for BLASTP searches against the O. sativa genome in JGI Phytozome (https:\/\/phytozome-next.jgi.doe.gov\/<\/a>). Six homologous sequences were identified in rice but could not be assigned in a one-to-one manner to the Arabidopsis members; these were therefore designated OsALIS1\u2013OsALIS6 according to chromosomal position: LOC_Os02g07750 (OsALIS1), LOC_Os03g02830 (OsALIS2), LOC_Os03g57170 (OsALIS3), LOC_Os05g45370 (OsALIS4), LOC_Os06g45430 (OsALIS5) and LOC_Os09g38768 (OsALIS6). To retrieve OsALIS2 orthologues in other representative species, the full-length OsALIS2 protein sequence was subsequently used as a query for BLASTP searches against the corresponding genomes in JGI Phytozome. For phylogenetic analyses, full-length protein sequences were aligned using ClustalW version 2.1 via the GenomeNet web server, and a phylogenetic tree was constructed in MEGA 5.1 with the Jones-Taylor-Thornton (JTT) substitution model and 1,000 bootstrap replicates. Representative species included O. sativa (Os), Triticum aestivum (Ta), Zea mays (Zm), Sorghum bicolor (Sb), Hordeum vulgare (Hv), Setaria italica (Si), A. thaliana (At), Glycine max (Gm), Arachis hypogaea (Ah), Chenopodium quinoa (Cq), Helianthus annuus (Ha), Saccharum officinarum (So), Cucumis sativus (Cs), Solanum lycopersicum (Sl), Solanum tuberosum (St), Malus domestica (Md), Nymphaea colorata (Nc), Physcomitrium patens (Pp), Marchantia polymorpha (Mp), Selaginella moellendorffii (Sm) and Chlamydomonas reinhardtii (Cr). To examine the evolutionary conservation of OsALA5 across plant species, putative OsALA5 orthologues were retrieved from the representative species above using BLASTP searches in JGI Phytozome and analysed using the parameters described above. Species names and corresponding gene identifiers are listed in Supplementary Table 4<\/a>.<\/p>\n Bimolecular fluorescence complementation assay<\/p>\n To identify the potential \u03b2-subunit(s) interacting with OsALA5, BiFC assay was performed as described previously59<\/a>. In brief, full-length coding sequences of OsALA5 and the six OsALIS homologues were amplified and subcloned into the pCAMBIA-35S-YN and pCAMBIA-35S-YC vectors. All constructs were verified by Sanger sequencing and introduced into A. tumefaciens strain GV3101. Agrobacterium cultures carrying the corresponding YN and YC constructs were mixed in equal volumes (OD600\u2009=\u20090.5 in infiltration buffer: 10\u2009mM MES, 10\u2009mM MgCl2, 150\u2009\u00b5M acetosyringone, pH 5.6) and co-infiltrated into N. benthamiana leaves. After incubation for 48\u2009h under standard growth conditions (25\u2009\u00b0C, 16\u2009h light:8\u2009h dark), YFP fluorescence in leaf epidermal cells was imaged using a Zeiss LSM 800 confocal laser-scanning microscopy (excitation 514\u2009nm\/emission 525\u2013560\u2009nm). Each BiFC assay was independently repeated at least three times with consistent results. Primers used for BiFC vector construction are listed in Supplementary Table 1<\/a>.<\/p>\n Monoclonal antibody generation against OsALA5<\/p>\n A mouse monoclonal antibody against OsALA5 was custom-generated by Absea Biotechnology. The antigen corresponded to amino acids 116\u2013308 of OsALA5 (cytoplasmic loop region). The antigenic fragment was expressed in E. coli as a His-tagged recombinant protein, purified by Ni-NTA affinity chromatography, and used to immunize BALB\/c mice following the manufacturer\u2019s standard protocol. Hybridoma clones producing OsALA5-specific antibodies were identified by ELISA and immunoblotting, subcloned by limiting dilution, isotyped, and expanded for Protein G-based purification. Antibody specificity was validated by immunoblotting as previously described53<\/a>. In brief, total proteins extracted from 14-day-old seedlings of wild-type and OsALA5-KO1 plants were resolved by SDS\u2013PAGE, transferred to PVDF membranes, and incubated with the anti-OsALA5 antibody (1:3,000 dilution) followed by a horseradish peroxidase (HRP)-conjugated secondary antibody (Proteintech, SA00001-1). Actin (Sangon Biotech, D191048) served as a loading control. Signals were visualized using an enhanced chemiluminescence (ECL) substrate (Bio-Rad, 1705061) and imaged on an e-Blot Touch Imager Pro. The validated monoclonal antibody was subsequently used for plasma membrane fraction analysis and immunolocalization assays.<\/p>\n Co-immunoprecipitation assay<\/p>\n To examine the physical interaction between OsALA5 and OsALIS2 in yeast, full-length coding sequences of both genes were cloned into the pMP4062 (HIS3) and pMP3864 (URA3-Flag) vectors, respectively. The recombinant plasmids were co-transformed into S. cerevisiae strain ZHY709, which lacks endogenous plasma membrane P4-ATPase activity, using the lithium acetate method. Transformants expressing only OsALA5 or OsALIS2 alone served as negative controls. Positive transformants were selected on synthetic defined (Solarbio) dropout medium and cultured in 30\u2009ml YPG medium (Coolaber) at 28\u2009\u00b0C with shaking (150\u2009rpm) for 36\u2009h.<\/p>\n Cells were harvested by centrifugation (3,000g, 5\u2009min, 4\u2009\u00b0C), and total membrane proteins were isolated using the Minute Plasma Membrane Protein Isolation Kit (Invent Biotechnologies). Membrane proteins were solubilized in Minute Non-Denatured Protein Solubilization Reagent (Invent Biotechnologies). Approximately 2\u2009mg of total membrane protein was incubated overnight at 4\u2009\u00b0C with the monoclonal anti-OsALA5 antibody in TBS-T (Tris-buffered saline, 0.05% Tween-20, pH 7.5). Subsequently, 20\u2009\u03bcl of Protein A\/G agarose beads (Thermo Scientific), prewashed three times in TBS-T, were added and incubated for 2\u2009h at 4\u2009\u00b0C with gentle rotation. Beads were collected using a DynaMag-2 magnetic rack (Invitrogen), washed five times with TBS-T, and bound proteins were eluted with 0.1\u2009M glycine (pH 2.8) for 30\u2009min at 4\u2009\u00b0C. Eluates were resolved by SDS\u2013PAGE and analysed by immunoblotting53<\/a> using anti-Flag (Proteintech, 66008-4-Ig) and anti-OsALA5, both diluted 1:3,000. Primers used for vector construction are listed in Supplementary Table 1<\/a>.<\/p>\n Two-phase partitioning<\/p>\n To determine the subcellular distribution of OsALA5, plasma membrane and other subcellular fractions were isolated using a PEG\/dextran 2PP21<\/a> procedure with minor modifications. Fourteen-day-old rice seedlings (~200 plants, fresh weight ~40\u2009g fresh weight) were rinsed 3 times with ice-cold deionized water, blotted dry, and homogenized on ice in pre-chilled homogenization buffer (1:5, w:v) containing 50\u2009mM Tris-HCl (pH 7.5), 250\u2009mM sucrose, 2\u2009mM EDTA, 150\u2009mM NaCl, 0.5% BSA, 1.5\u2009g\u2009l\u22121 PVP 360, and 1\u00d7 protease inhibitor cocktail. Homogenates were lysed on ice for 15\u2009min, filtered twice through double-layer gauze, and centrifuged at 700g for 10\u2009min at 4\u2009\u00b0C to obtain the nuclear pellet. The supernatant was centrifuged at 85,000g for 1\u2009h at 4\u2009\u00b0C in a Beckman Optima XE-90 ultracentrifuge, yielding a microsomal pellet and a cytosolic supernatant. The microsomal pellets were gently resuspended in 3\u2009ml resuspension buffer (250\u2009mM sucrose, 5\u2009mM phosphate buffer pH 7.8, 1\u00d7 protease inhibitor cocktail) and subjected to 2PP separation.<\/p>\n For phase partitioning, the resuspended microsomes were layered onto a pre-equilibrated, ice-cold two-phase system (final preparation per batch: 20\u2009g of 20% Dextran T-500, 10.3\u2009g of PEG-3350, 3.3\u2009ml of Na3PO4 solution, and 17.9\u2009ml of ddH2O). After gentle mixing (~30 inversions), samples were centrifuged at 2,000g for 10\u2009min at 4\u2009\u00b0C to resolve upper and lower phases. Approximately 90% of the upper phase was transferred to fresh lower phase, mixed, and re-partitioned; the original tube was replenished with fresh upper phase. This partitioning step was repeated for three times. Pooled upper phases were diluted 2\u20133\u00d7 in resuspension buffer and centrifuged at 85,000g for 1\u2009h at 4\u2009\u00b0C to yield the plasma membrane pellets. Pooled lower phases were processed in parallel to obtain intracellular organelle (endomembrane) fractions. All membrane pellets (plasma membrane and organellar) were solubilized in membrane protein solubilization buffer (25\u2009mM Tris-HCl pH 7.5, 150\u2009mM NaCl, 1.5% DDM) on ice prior to analysis. Finally, the subcellular fractions (nuclear, cytosolic, organellar, and plasma membrane) were analysed by immunoblotting as described53<\/a>, using 2\u2009\u00b5g total protein per lane. OsALA5 distribution was detected with an anti-OsALA5 monoclonal antibody (1:3,000). Fractional purity was assessed with specific marker proteins: H+-ATPase (plasma membrane marker; Agrisera, AS07 260; 1:5,000), BiP (endoplasmic reticulum marker; Agrisera, AS09 481; 1:5,000), UGPase (cytosolic marker; Agrisera, AS14 2813; 1:5,000), and Histone H3 (nuclear marker; Proteintech, 17168-1-AP; 1:5,000). HRP-conjugated Goat Anti-Mouse IgG(H\u2009+\u2009L) (Proteintech, SA00001-1; 1:3,000) and HRP-conjugated Goat Anti-Rabbit IgG(H\u2009+\u2009L) (Proteintech, SA00001-2; 1:3,000) were used as secondary antibodies.<\/p>\n Free-flow electrophoresis<\/p>\n Plasma membrane pellets obtained from 2PP were gently resuspended in the resuspension buffer prior to FFE purification. The resuspended plasma membrane vesicles (\u22650.5\u2009mg\u2009ml\u22121 total protein; \u2265100\u2009ml per batch) were further purified by continuous zone FFE (ZE-FFE)21<\/a>,60<\/a>,61<\/a> using an FFE-32 instrument (Biochine Biotechnology). The electrophoresis medium (chamber buffer) consisted of 10\u2009mM triethanolamine, 10\u2009mM acetic acid, 5\u2009mM glucose, 250\u2009mM sucrose, and 0.5\u2009mM MgCl2 (pH 6.5, adjusted with 1\u2009N NaOH), with an osmolarity of ~270 mosm and conductivity of ~5.9\u2009\u00d7\u2009102\u2009\u00b5mho. The electrode buffer contained 100\u2009mM triethanolamine and 100\u2009mM acetic acid (pH 6.5). All buffers were freshly prepared and degassed prior to use. Electrophoresis was conducted at a field strength of ~130\u2009V\u2009cm\u22121 (\u2248500\u2009V total) with a current limit of 100\u2009mA, while maintaining the chamber at 6\u201310\u2009\u00b0C. Chamber buffer flowed at 2.9\u2009ml per fraction per hour, and the plasma membrane suspensions were injected through channel 8 at 404\u2009\u03bcl\u2009min\u22121. Approximately 12\u2009ml plasma membrane sample was processed per run. Individual fractions were collected and monitored by UV absorbance at 280\u2009nm. Fractions corresponding to the major UV peak (fractions 8\u201319) were analysed to identify plasma membrane-enriched material. Equal amounts of protein (2\u2009\u00b5g per fraction) were analysed by immunoblotting using antibodies against specific organelles: BiP (endoplasmic reticulum marker; Agrisera, AS09 481; 1:5,000), COXII (mitochondrial inner membrane marker; Agrisera, AS04 053\u2009A; 1:5,000), IDH (mitochondrial matrix marker; Agrisera, AS06 203\u2009A; 1:5,000), V-ATPase (vacuole marker; Agrisera, AS07 213; 1:5,000), Tic40 (chloroplast marker; Agrisera, AS10 709; 1:5,000), Sec21p (Golgi apparatus marker; Agrisera, AS08 327; 1:5,000), H+-ATPase (plasma membrane marker; Agrisera, AS07 260; 1:5,000), and the anti-OsALA5 monoclonal antibody (1:3,000). Microsomal fractions obtained before and after 2PP served as controls to assess enrichment and depletion across subcellular compartments.<\/p>\n Subcellular localization<\/p>\n To determine the subcellular localization of OsALA5 in vivo, a complementation line expressing OsALA5 fused to GFP under the control of its native promoter (proOsALA5::OsALA5\u2013GFP; the complementation line) was generated as described above. Transgenic seedlings exhibiting restored growth phenotype (Extended Data Fig. 3d,e<\/a>) and confirmed expression of the OsALA5\u2013GFP fusion protein (Extended Data Fig. 3c<\/a>) were used for microscopy. Roots of 6-day-old seedlings were incubated in 5\u2009\u03bcM FM4-64 (US Everbright) for 5\u2009min (ref. 62<\/a>), briefly rinsed, and imaged on a Zeiss LSM 800 confocal microscope (GFP: excitation 488\u2009nm\/emission 500\u2013540\u2009nm; FM4-64: excitation 488\u2009nm\/emission 640\u2013750\u2009nm). For plasmolysis assays, roots were treated with 10% (w\/v) mannitol for 1\u2009h and imaged in the same solution following standard procedures63<\/a>.<\/p>\n To characterize the subcellular localization of OsALIS2, the \u03b2-subunit that interacts with OsALA5, transient expression assays were performed in rice protoplasts and N. benthamiana epidermal cells as described previously64<\/a>,65<\/a>. In brief, the full-length coding sequence of OsALIS2 was fused to GFP under the CaMV 35S promoter (35S::OsALIS2\u2013GFP). In rice protoplasts, OsALIS2\u2013GFP was co-expressed with HDEL\u2013mCherry (endoplasmic reticulum marker)66<\/a> or stained with FM4-64 prior to imaging. For localization analysis in N. benthamiana, A. tumefaciens GV3101 suspensions (OD600\u2009=\u20090.5 in 10\u2009mM MgCl2, 10\u2009mM MES, 150\u2009\u03bcM acetosyringone, pH 5.6) carrying 35S::OsALIS2\u2013GFP were infiltrated into 4-week-old leaves; co-localization assays were performed by co-infiltrating HDEL\u2013mCherry or PIP2A\u2013mCherry (plasma membrane marker)66<\/a> at equal optical densities. Fluorescence was examined 48\u2009h after infiltration using a Zeiss LSM 800 with sequential acquisition of GFP (excitation 488\u2009nm\/emission 500\u2013540\u2009nm), mCherry (excitation 561\u2009nm\/emission 580\u2013650\u2009nm), and FM4-64 (excitation 488\u2009nm\/emission 640\u2013750\u2009nm) channels under identical settings.<\/p>\n To examine the spatial association between OsALA5 and its \u03b2-subunit OsALIS2, co-localization assays were conducted in both rice protoplasts and N. benthamiana epidermal cells. Full-length OsALA5 and OsALIS2 coding sequences were fused to GFP and mCherry (35S::OsALA5\u2013GFP and 35S::OsALIS2\u2013mCherry), and introduced using the same transformation methods as above. Fluorescence was acquired sequentially on a Zeiss LSM 800 using identical laser power and detector settings across samples (GFP: excitation 488\u2009nm\/emission 500\u2013540\u2009nm; mCherry: excitation 561\u2009nm\/emission 580\u2013650\u2009nm). All plasmids were confirmed by Sanger sequencing, and primers used for vector construction are listed in Supplementary Table 1<\/a>.<\/p>\n Protein expression and purification<\/p>\n Full-length coding sequences of OsALA5 and OsALIS2 were cloned into the mammalian expression vector pCAG with C-terminal Flag (DYKDDDDK) and His8 tags for heterologous expression in HEK293F cells. The OsALA5(D433N) mutant was generated by in-fusion cloning using primers introducing a G-to-A substitution at nucleotide position 1297. All constructs were confirmed by Sanger sequencing, and primers are listed in Supplementary Table 1<\/a>. Protein expression and purification were performed following established procedures for membrane transporters67<\/a> with minor modifications. HEK293F cells were purchased from Sino Biological and cultured in SMM 293T-II medium (M293TII, Sino Biological) at 37\u2009\u00b0C with 5% CO2 and 130\u2009rpm shaking. HEK293F cells were not further authenticated or tested for mycoplasma contamination after purchase. Plasmids encoding OsALA5 alone, OsALA5\u2009+\u2009OsALIS2 or OsALA5(D433N)\u2009+\u2009OsALIS2 were transiently transfected using polyethylenimine (PEI), followed by sodium butyrate induction. Cells were collected by centrifugation and lysed in buffer containing 25\u2009mM Tris-HCl (pH 7.4), 150\u2009mM NaCl, and protease inhibitors (Cocktail, Roche). Membrane proteins were solubilized in 1.5% (w\/v) n-dodecyl-\u03b2-d-maltoside (BLUEPUS) with 0.3% (w\/v) cholesteryl hemisuccinate for 2\u2009h at 4\u2009\u00b0C, and insoluble material was removed by ultracentrifugation (12,000g, 60\u2009min, 4\u2009\u00b0C). Flag\u2013His8-tagged proteins were purified using anti-Flag M2 affinity resin (Sigma-Aldrich), washed in buffer containing 25\u2009mM Tris-HCl (pH 7.4), 150\u2009mM NaCl, and 0.05% DDM, and eluted with Flag peptide. For downstream biochemical assays\u2014including PLO assays, ATPase activity measurements and proteoliposome reconstitution\u2014the eluates were passed through desalting columns to remove Flag peptide and equilibrate the samples into assay buffer (25\u2009mM Tris-HCl, pH 7.4; 150\u2009mM NaCl; 0.05% DDM). Protein concentration was quantified by BCA assay, and purified proteins were aliquoted, flash-frozen in liquid nitrogen, and stored at \u221280\u2009\u00b0C until use.<\/p>\n Protein lipid overlay assay<\/p>\n PLO assays were conducted as described previously22<\/a> with minor modifications. To assess the binding of OsALA5 to major membrane phospholipid classes, six plant phospholipids\u2014phosphatidic acid (PA 16:0\/16:0), phosphatidylserine (PS 16:0\/16:0), phosphatidylethanolamine (PE 16:0\/16:0), PC (16:0\/16:0), phosphatidylglycerol (PG 16:0\/16:0), and phosphatidylinositol (16:0\/16:0) (Echelon Biosciences)\u2014were dissolved in chloroform at 1\u2009mg\u2009ml\u22121 and spotted (1\u2009\u00b5l) onto PVDF membranes (n\u2009=\u20093 per lipid). Spotted membranes were air-dried for 1\u2009h at room temperature and blocked in TBS-T (20\u2009mM Tris-HCl, pH 7.5; 150\u2009mM NaCl; 0.05% Tween-20) containing 3% (w\/v) BSA for 1\u2009h. Purified OsALA5\u2013OsALIS2 complex (30\u2009\u00b5g in blocking buffer) was applied to the membranes and incubated for 5\u2009min at either 28\u2009\u00b0C or 45\u2009\u00b0C. Membranes were washed three times (5\u2009min each) in TBS-T, incubated overnight at 4\u2009\u00b0C with anti-OsALA5 monoclonal antibody (1:3,000 in TBS-T\u2009+\u20093% BSA), washed, and then probed with HRP-conjugated anti-mouse IgG (Proteintech; 1:3,000) for 1\u2009h at room temperature. Signals were developed using ECL substrate and imaged using an e-Blot Touch Imager Pro. To evaluate binding specificity across PC molecular species, PC (16:0\/16:0), PC (16:0\/18:1), PC (16:0\/18:2), PC (18:0\/18:0), PC (18:0\/18:1), and PC (18:1\/18:1) (Echelon Biosciences) were spotted onto PVDF membranes following the same spotting and blocking procedures. Incubation, washing, antibody probing, and signal detection steps were identical to those described above.<\/p>\n NBD-PC uptake assay in yeast cells<\/p>\n NBD-PC uptake assays were performed as described previously28<\/a> with minor modifications. NBD-labelled PCs (NBD-PC 16:0\/12:0 and NBD-PC 18:1\/12:0; Avanti Polar Lipids) were dissolved in DMSO to 4\u2009mM and stored at \u221220\u2009\u00b0C. Yeast strains BY4741, ZHY709 and ZHY709 expressing OsALA5\u2013OsALIS2 were grown overnight in selective SG medium (Solarbio) at 28\u2009\u00b0C (200\u2009rpm) to mid log phase (OD600\u2009\u2248\u20090.5), washed once with assay buffer (10\u2009mM Tris-HCl, pH 7.4; 150\u2009mM NaCl), and resuspended to OD600\u2009=\u20090.5. To ensure equal cell input across samples, an aliquot of each culture was stained with FM4-64 (US Everbright) and fluorescence was recorded using a FlexStation 3 microplate reader controlled by SoftMax Pro software (Molecular Devices, v7.1) at excitation 510\u2009nm\/emission 750\u2009nm under identical settings. For NBD-PC uptake measurements, 200\u2009\u00b5l of cell suspension was transferred into a temperature-controlled black 96-well plate in FlexStation 3, and baseline fluorescence was recorded for 30\u2009s (excitation 460\u2009nm\/emission 535\u2009nm). NBD-PC (16:0\/12:0 or 18:1\/12:0) was then added to a final concentration of 4\u2009\u00b5M, rapidly mixed, and fluorescence was monitored continuously for 300\u2009s to measure incorporation into the plasma membrane. To quench NBD-PC in the exoplasmic plasma membrane leaflet, freshly prepared sodium dithionite (100\u2009mM in 1\u2009M Tris-HCl, pH 10.0) was added to a final concentration of 10\u2009mM, and fluorescence was recorded for another 300\u2009s. Finally, 2\u2009\u00b5l 30% Triton X-100 was added to permeabilize cells and allow dithionite access to all remaining NBD-labelled lipids. All uptake assays were performed at either 28\u2009\u00b0C or 45\u2009\u00b0C, and fluorescence was acquired every 10\u2009s under identical detector settings. Relative NBD fluorescence was calculated as: FI(t)\/FI(0)\u2009=\u2009(Ft\u2009\u2212\u2009FTriton)\/(F0\u2009\u2212\u2009FTriton). F0 is the initial fluorescence before dithionite addition, Ft is fluorescence at time t, and FTriton is the residual signal after Triton X-100 treatment. Curve fitting and plotting were performed using GraphPad Prism (v8.0.2). All experiments were repeated three times with consistent results.<\/p>\n Artificial liposome-based NBD-PC transport assay<\/p>\n Artificial liposome-based NBD-PC transport assays were performed as described previously29<\/a> with minor modifications. Two types of NBD-labelled liposomes were generated from POPC:POPG:NBD-PC at a 25:25:1 molar ratio, in which the fluorescent lipid component was either NBD-PC (16:0\/12:0) or NBD-PC (18:1\/12:0) (Avanti Polar Lipids). Lipid mixtures were dissolved in chloroform, dried under nitrogen, desiccated for 1\u2009h, hydrated in transport buffer (25\u2009mM Tris-HCl, pH 7.4; 150\u2009mM NaCl; 1\u2009mM ATP) to 5\u2009mg\u2009ml\u22121, subjected to \u226510 freeze\u2013thaw cycles, and extruded through 200\u2009nm polycarbonate membranes (Avanti) for 20 times to obtain unilamellar liposomes. For protein incorporation, liposomes were incubated with 0.5% (w\/v) DDM for 20\u2009min at 4\u2009\u00b0C, followed by addition of purified OsALA5\u2013OsALIS2 or OsALA5(D433N)\u2013OsALIS2 at a lipid:protein mass ratio of 100:1. The mixture was rotated overnight at 4\u2009\u00b0C, and detergent was removed using Bio-Beads SM-2 (Bio-Rad). Proteoliposomes were collected by ultracentrifugation (100,000g, 1\u2009h, 4\u2009\u00b0C), and resuspended in transport buffer. Liposome size distribution was verified by dynamic light scattering, confirming uniform vesicle populations, and successful protein reconstitution was validated by immunoblotting using Flag and OsALA5 antibodies.<\/p>\n For transport measurements, 200\u2009\u00b5l of proteoliposomes (or mock liposomes) were equilibrated in a black 96-well plate of FlexStation 3 (Molecular Devices) at either 28\u2009\u00b0C or 45\u2009\u00b0C. ATP (1\u2009mM final concentration; Sigma) was added immediately before fluorescence acquisition. Baseline fluorescence was recorded for 30\u2009s (excitation 460\u2009nm\/emission 535\u2009nm), and fluorescence was monitored continuously for 300\u2009s. Sodium dithionite (100\u2009mM in 1\u2009M Tris-HCl, pH 10.0) was subsequently added to 10\u2009mM to selectively quench outer leaflet NBD-PC, and fluorescence was recorded for another 300\u2009s. Finally, 2\u2009\u00b5l 30% Triton X-100 was added to permeabilize liposomes and enable dithionite quenching of all remaining NBD-labelled lipids. Relative NBD fluorescence was calculated as: FI(t)\/FI(0)\u2009=\u2009(Ft\u2009\u2212\u2009FTriton)\/(F0\u2009\u2212\u2009FTriton). F0 is fluorescence before dithionite addition, Ft is fluorescence at time t, and FTriton is the residual fluorescence after Triton X-100 treatment. Curves were generated using GraphPad Prism (v8.0.2). Experiments were independently repeated three times with consistent results.<\/p>\n ATPase activity assay<\/p>\n ATPase activity was measured using the ATPase Colorimetric Assay Kit (Innova Biosciences) as described previously67<\/a> with minor modifications. To evaluate PC-dependent ATP hydrolysis under normal (28\u2009\u00b0C) and heat stress (45\u2009\u00b0C) conditions, purified OsALA5\u2013OsALIS2 or OsALA5(D433N)\u2013OsALIS2 complexes (5\u2009\u00b5g) were incubated with 2\u2009\u00b5g of the indicated PC species in a 50\u2009\u00b5l reaction containing 25\u2009mM HEPES-KOH (pH 7.4), 150\u2009mM KCl, 10\u2009mM MgCl2, 0.01% (w\/v) DDM, and 1\u2009mM ATP. Reactions were carried out for 5\u2009min at either 28\u2009\u00b0C or 45\u2009\u00b0C. Following incubation, released inorganic phosphate was quantified using the ATPase Colorimetric Assay Kit according to the manufacturer\u2019s instructions, and absorbance was measured at 650\u2009nm using a Biomate 3S spectrophotometer (Thermo Scientific). Each reaction was measured in technical triplicate, and all experiments were independently repeated three times with consistent results.<\/p>\n Characterization and orientation analysis of plasma membrane vesicles<\/p>\n Plasma membrane vesicles were isolated from the FFE fractions enriched for the H+-ATPase and OsALA5 (fractions 10\u201313; Extended Data Fig. 3b<\/a>). Fractions from Kas, OsALA5-KO1, and the complementation line under control (28\u2009\u00b0C) and heat stress (45\u2009\u00b0C for 1\u2009h) conditions were pooled and centrifuged at 100,000g for 1\u2009h at 4\u2009\u00b0C. Pellets were gently resuspended in the 2PP resuspension buffer (250\u2009mM sucrose, 5\u2009mM phosphate buffer pH 7.8, 1\u00d7 protease inhibitor cocktail) and kept on ice for further analysis. Plasma membrane vesicle morphology and purity were examined by transmission electron microscopy (TEM). Vesicle suspensions were adsorbed onto glow-discharged carbon-coated grids, negatively stained with 2% (w\/v) uranyl acetate, air-dried, and imaged using a Talos L120C TEM (Thermo Fisher) at 120\u2009kV. TEM analyses were independently repeated three times with consistent results.<\/p>\n As leaflet-resolved lipidomic analysis requires right-side-out plasma membrane vesicles, we next examined vesicle orientation using two established leaflet-specific probes. Annexin-CF594 (excitation 593\u2009nm\/emission 614\u2009nm; Biotium), which binds phosphatidylserine on the cytoplasmic plasma membrane leaflet68<\/a>, was used to identify inside-out vesicles. Concanavalin A-FITC (ConA-FITC, excitation 488\u2009nm\/emission 500\u2013540\u2009nm; Alpha Diagnostic International), which binds glycoproteins exposed on the exoplasmic plasma membrane leaflet, was used to identify right-side-out vesicles61<\/a>. Microsomes isolated before 2PP and after 2PP served as reference controls. Microsomes and FFE-purified plasma membrane vesicles (fractions 10\u201313) were incubated simultaneously with Annexin-CF594 and ConA-FITC in binding buffer (10\u2009mM HEPES, 150\u2009mM NaCl, 2\u2009mM CaCl2, pH 7.4) for 20\u2009min in the dark, washed twice with the same buffer, and imaged using a Zeiss LSM 800 confocal microscope under identical acquisition settings.<\/p>\n Lipidomics analysis<\/p>\n Plasma membrane vesicles were isolated from the FFE fractions 10\u201313, corresponding to H+-ATPase- and OsALA5-enriched plasma membrane fractions as verified by immunoblotting and vesicle orientation analysis. Fractions from Kas, OsALA5-KO1 and the complementation line (CL) under control (28\u2009\u00b0C) or heat stress (45\u2009\u00b0C, 1\u2009h) conditions were pooled, pelleted by ultracentrifugation (100,000g, 1\u2009h, 4\u2009\u00b0C), and gently resuspended in the 2PP resuspension buffer. Vesicles were kept on ice for subsequent analyses.<\/p>\n Because leaflet-resolved lipidomics requires complete removal of exoplasmic phospholipids while preserving the cytoplasmic leaflet, PLA2 digestion conditions were empirically optimized. Plasma membrane vesicles (60\u2009\u00b5g total protein; Kas, fractions 10\u201313) were incubated at 37\u2009\u00b0C in a digestion buffer (10\u2009mM Tris-HCl, pH 7.4; 10\u2009mM CaCl2; 5% (w\/v) BSA) containing PLA2 (0.5 U mg\u22121 membrane protein, corresponding to 0.03 U PLA2 per reaction; P6534, Sigma) for 0, 5, 10, 20, 30, 40, 50, 60 and 120\u2009min. Reactions were terminated by immediate addition of aristolochic acid (1.5\u2009\u00b5M; MCE) to inactivate PLA2. Digestion kinetics were monitored by quantifying surface-exposed phosphatidylserine, total phosphatidylserine, and lysophosphatidylcholine (lysoPC). Surface phosphatidylserine was measured directly from intact vesicles, whereas total phosphatidylserine was measured after dissolving vesicles in methanol:chloroform (3:1, v\/v). Phosphatidylserine was quantified using a phosphatidylserine ELISA kit (MBS731866; MyBioSource), and lysoPC using a lysoPC ELISA kit (MBS2031889; MyBioSource). All measurements were repeated at least twice independently, with consistent results. Based on digestion curves (Extended Data Fig. 6d\u2013f<\/a>), a 10-min PLA2 digestion was selected for all subsequent leaflet-resolved lipidomic analyses. For lipid extraction, PLA2-treated and untreated plasma membrane vesicles (60\u2009\u00b5g total protein per sample) were inactivated with hot isopropanol according to a modified protocol69<\/a>. Extraction solvent (chloroform:methanol: 300\u2009mM ammonium acetate = 30:41.5:3.5, v\/v\/v) was added, and samples were incubated at 4 \u00b0C for 30\u2009min at 1500\u2009rpm. After centrifugation, the clear supernatant was transferred to fresh tubes. The inactivation and extraction steps were repeated once, and pooled lipid extracts were dried in a SpeedVac (Genevac). Dried lipids were stored at \u221280 \u00b0C until liquid chromatography\u2013mass spectrometry analyses.<\/p>\n Lipidomic analyses were performed at LipidALL Technologies using a Shimadzu Nexera 20AD-HPLC coupled with SCIEX QTRAP 6500 PLUS as described previously70<\/a>. For normal phase analysis of polar lipids, individual species were separated using a TUP-HB silica column (internal diameter 150\u2009\u00d7\u20092.1\u2009mm, 3\u2009\u00b5m) under the following conditions: mobile phase A (chloroform:methanol:ammonium hydroxide, 89.5:10:0.5) and mobile phase B (chloroform:methanol:ammonium hydroxide:water, 55:39:0.5:5.5), the gradient started with 2% B that was maintained for 2\u2009min, then increased to 20% B over 1\u2009min, which was increased to 45% B over 3\u2009min and then increased to 75% B over 1.5\u2009min, before increasing again to 100% B over 1\u2009min. The gradient was maintained at 100% B for 4.5\u2009min, before returning to 2% B over 0.5\u2009min and equilibrated for another 1.5\u2009min prior to the next injection. For reverse phase LC\/MS, lipids were analysed using a modified version of reverse phase (RP)-HPLC\/ESI\/MS\/MS as reported previously70<\/a>. In brief, separation of the aforementioned lipids was carried out on a Phenomenex Kinetex 2.6 \u00b5m-C18 column (internal diameter 4.6\u2009\u00d7\u2009100\u2009mm) using an isocratic mobile phase chloroform:methanol:0.1\u2009M ammonium acetate (100:100:4) at a flow rate of 300\u2009\u00b5l\/min for 10\u2009min. Quantification of individual lipid species were carried out by referencing to spiked internal standards; namely d9-PC32:0 (16:0\/16:0), d7-PE33:1 (15:0\/18:1), d31-PS (d31-16:0\/18:1), d7-PA33:1 (15:0\/18:1), d7-PG33:1 (15:0\/18:1), d5-CL72:8 (18:2)4, d7-LPC18:1, d7-LPE18:1, DMPS, DMPA, DMPG, MGDG 34:0, DGDG 36:0, d5-DAG17:0\/17:0, and d5-DAG18:1\/18:1 obtained from Avanti Polar Lipids and LIPID MAPS. Dioctanoyl phosphatidylinositol (PI) (16:0-PI) was purchased from Echelon Biosciences and used together with d7-PI33:1(15:0\/18:1) (Avanti Polar Lipids) for phosphatidylinositol quantification. TAGs were quantified using TAG (14:0)3-d5, TAG (15:0)3-d29, and TAG (18:0)3-d5 obtained from CDN isotopes obtained from CDN isotopes. Free fatty acids were quantified using d31-16:0 (Sigma-Aldrich). Six biological replicates were analysed for each genotype and treatment. Data were processed using GraphPad Prism (v8.0.2).<\/p>\n Membrane fluidity analysis<\/p>\n Freshly isolated rice protoplasts were used to assess plasma membrane fluidity. Protoplasts were prepared from 14-day-old hydroponically grown seedlings of Kas, OsALA5-KO1, and the complementation line as described above. Cell density was quantified using a haemocytometer (Solarbio), and suspensions were adjusted to 1\u2009\u00d7\u2009106 cells per ml in W5 solution. For each measurement, 100\u2009\u00b5l of protoplast suspension (1\u2009\u00d7\u2009105 cells) was dispensed into wells of a black 96-well plate. Membrane fluidity was monitored using two established fluorescent probes, TMA-DPH and ANS, following previously described procedures33<\/a>,71<\/a>. TMA-DPH (Sigma) was prepared as a 2\u2009\u00d7\u200910\u22123\u2009M stock in tetrahydrofuran and diluted in 1\u00d7 PBS to 2\u2009\u00d7\u200910\u22126\u2009M as the working solution. Protoplasts were incubated with 2\u2009\u03bcM TMA-DPH for 45\u2009min at 30\u2009\u00b0C in the dark to allow probe incorporation. ANS (Sigma) was prepared as a 4\u2009\u00d7\u200910\u22123\u2009M stock in ethanol and added to protoplasts to a final concentration of 5\u2009\u00d7\u200910\u22125\u2009M; fluorescence acquisition was initiated 1\u2009min after dye addition.<\/p>\n Temperature treatments were applied during fluorescence acquisition. The microplate reader FlexStation 3 (Molecular Devices) was pre-equilibrated to either 28\u2009\u00b0C or 45\u2009\u00b0C, and protoplast suspensions were incubated for 5\u2009min in the chamber to achieve thermal equilibration before recording. Fluorescence was detected at excitation 360\u2009nm\/emission 430\u2009nm for TMA-DPH and excitation 400\u2009nm\/emission 510\u2009nm for ANS. Membrane fluidity was expressed as fluorescence anisotropy (r) for both probes. For ANS, fluorescence intensity was recorded directly and used as the fluorescence anisotropy readout. For TMA-DPH, fluorescence anisotropy was calculated as: r\u2009=\u2009(Ivv\u2009\u2212\u2009G\u2009\u00d7\u2009Ivh)\/(Ivv\u2009+\u2009G\u2009\u00d7\u2009Ivh). Ivv and Ivh are vertically and horizontally polarized emissions under vertical excitation, and G is the grating correction factor. Higher anisotropy reflects lower membrane fluidity. All experiments were independently repeated three times with consistent results, and data were analysed using GraphPad Prism (v8.0.2).<\/p>\n Differential scanning calorimetry assay<\/p>\n DSC was performed to determine the gel-to-fluid phase transition of reconstituted liposomes using a Nano DSC system (DSCRun, TA Instruments, v4.7.1)34<\/a>,48<\/a>. Liposomes were prepared following the same procedure described above for liposome construction and protein reconstitution, except that a 1:1 molar mixture of PC (18:0\/18:0) and PC (18:1\/18:1) was used, and 1\u2009mM ATP was included in the hydration buffer to allow ATP encapsulation during vesicle formation. In brief, 50\u2009mg phospholipids were dissolved in chloroform, dried under nitrogen, and vacuum-desiccated for 1\u2009h. The dried lipid film was hydrated in ATP-containing reconstitution buffer, subjected to \u226510 freeze\u2013thaw cycles, and extruded through 200\u2009nm polycarbonate filters to generate unilamellar vesicles. Purified OsALA5\u2013OsALIS2 complexes or the OsALA5(D433N)\u2013OsALIS2 complex were incorporated into liposomes using the same proteoliposome reconstitution strategy described above. Empty liposomes processed in parallel served as controls. To remove unencapsulated ATP, proteoliposomes were washed twice with ATP-free transport buffer (25\u2009mM Tris-HCl, pH 7.4; 150\u2009mM NaCl) by ultracentrifugation (100,000g, 1\u2009h, 4\u2009\u00b0C). DSC scans were conducted from 20 to 60\u2009\u00b0C at a rate of 1\u2009\u00b0C min\u22121. Each sample was analysed in three independent biological repeats, yielding reproducible thermograms. Data were processed using NanoAnalyze (TA Instruments, v3.11.0), and phase-transition curves were plotted in GraphPad Prism (v8.0.2). For clarity, only the main phase-transition region (35\u201355\u2009\u00b0C) is shown in the final presentation.<\/p>\n Electrolyte leakage analysis<\/p>\n Electrolyte leakage was quantified as described previously12<\/a>,72<\/a>. In brief, 14-day-old hydroponically grown seedlings of Kas, OsALA5-KO1 and the complementation line were exposed to either 28\u2009\u00b0C or 45\u2009\u00b0C for 6\u2009h. After treatment, 0.5\u2009g of leaf tissue was excised, briefly rinsed with deionized water, and transferred into 50\u2009ml tubes containing 20\u2009ml deionized water. Samples were incubated at 28\u2009\u00b0C for 2\u2009h, and the initial conductivity (R1) of the bathing solution was measured using a DDSJ-308F conductivity meter (LeiCi). Tubes were then heated in a boiling water bath for 15\u2009min to release total electrolytes, cooled to room temperature, and the final conductivity (R2) was recorded. Electrolyte leakage (EL) was calculated as: EL (%)\u2009=\u2009(R1\/R2)\u2009\u00d7\u2009100%. Values represent mean\u2009\u00b1\u2009s.d. from at least three independent biological replicates. Data were analysed using GraphPad Prism (v8.0.2).<\/p>\n Trypan blue staining<\/p>\n Trypan blue staining was performed as described previously36<\/a>. In brief, 14-day-old hydroponically grown seedlings of Kas, OsALA5-KO1 and the complementation line were exposed to either 28\u2009\u00b0C or 45\u2009\u00b0C for 6\u2009h. After treatment, leaves were immersed in 0.04% Trypan Blue solution (Solarbio; diluted in PBS, pH 7.4), heated in a boiling water bath for 10\u2009min, and then incubated at room temperature for 12\u2009h. Stained tissues were transferred to 25\u2009mg\u2009ml\u22121 chloral hydrate (Sigma) solution and destained for 24\u2009h, with several solution exchanges to reduce background staining. Samples were subsequently stored in 50% (v\/v) glycerol and imaged using a Leica S6D stereomicroscope.<\/p>\n Haplotype analysis<\/p>\n Haplotype identification of OsALA5 was performed using sequence polymorphisms across the 1.2-kb promoter region and the full-length genomic sequence retrieved from 2,236 Asian cultivated rice accessions in the RiceVarMap v2.0 database (http:\/\/ricevarmap.ncpgr.cn\/<\/a>)41<\/a>,73<\/a>. Accessions lacking complete sequence coverage in either region were excluded, and only haplotypes represented by at least two accessions were retained for subsequent analysis. Transcript abundance of representative haplotypes was measured in selected accessions by qPCR. RNA extraction, cDNA synthesis, and qPCR followed the procedures described in the corresponding Methods subsection. Expression values were normalized to internal reference genes, and relative expression levels were compared across haplotypes to identify alleles with distinct transcriptional activities. Promoter activity analysis of representative haplotypes was conducted in rice protoplasts as described previously52<\/a>. Full-length promoter sequences of Hap1, Hap2, and Hap7 were amplified from corresponding genomic DNA and cloned in the pGreenII 0800-LUC vector containing both firefly luciferase (LUC) and Renilla luciferase (REN). Constructs and corresponding empty vectors were independently transformed into rice protoplasts. After 16\u201318\u2009h incubation at 28\u2009\u00b0C in the dark, promoter activity was quantified as the LUC\/REN ratio using the Dual Luciferase Reporter Assay Kit (Beyotime). All assays were performed with at least three independent biological replicates. Data were analysed using GraphPad Prism (v8.0.2). Primers are listed in Supplementary Table 1<\/a>.<\/p>\n Heat tolerance assessment of the Hap7 allele<\/p>\n A NIL carrying the OsALA5Hap7 allele (NIL-OsALA5Hap7) was generated from a cross between Nip (recurrent parent, Hap2) and Kas (donor parent, Hap7). Introgression of the Kas-derived chromosomal segment encompassing OsALA5 was confirmed by targeted sequencing. Fourteen-day-old hydroponically grown seedlings of Nip and NIL-OsALA5Hap7 were used for transcript analyses, and OsALA5 mRNA transcript abundance was quantified by qPCR following the procedures described in \u2018RNA extraction and qPCR\u2019.<\/p>\n Seedling stage heat tolerance was performed using the standard heat stress protocol described in \u2018Heat tolerance assay at the seedling stage\u2019 subsection. Heading-stage heat tolerance was assessed as previously described protocol with minor modifications52<\/a>. Field-grown plants were transferred to pots approximately one week before heading and allowed to stabilize. Panicles predicted to flower synchronously were tagged, and plants were moved into controlled-environment chambers (KOOLAND Technology) one day before flowering. Chambers were programmed with diurnal temperature regimes of 22\u201332\u2009\u00b0C (control) or 32\u201340\u2009\u00b0C (heat stress), with relative humidity maintained at 65\u201380%, following the profiles in Extended Data Fig. 10e<\/a>. Plants were exposed to the respective regimes for 2 days during flowering and returned to the field until harvest. Seed setting rate and grain yield per plant were quantified from tagged panicles. Whole-life-cycle heat tolerance was assessed by transferring 14-day-old seedlings into chambers set to 22\u201328\u2009\u00b0C (control) or 30\u201336\u2009\u00b0C (heat stress), following the profiles shown in Extended Data Fig. 2a<\/a>, with relative humidity maintained at 65\u201380%. To assess Hap7 performance across natural thermal environments, multi-location field trials were conducted in Chengdu, Sichuan (30.67\u00b0 N, 104.06\u00b0 E), Yongchuan, Chongqing (29.36\u00b0 N, 105.93\u00b0 E), and Changsha, Hunan (28.23\u00b0 N, 112.94\u00b0 E). These sites represent distinct agroclimatic regions, with Yongchuan and Changsha experiencing frequent summer heat episodes typical of humid subtropical climates, thereby providing contrasting thermal environments for evaluating heat responses under field conditions. Air temperature was recorded by on-site automated weather stations installed at each field site. At maturity, agronomic traits\u2014including plant height, tiller number, panicle traits, 1,000-grain weight, seed setting rate, and grain yield per plant\u2014were assessed for both chamber-based whole-life-cycle and field trials. All data were analysed with GraphPad Prism (v8.0.2). Primers are listed in Supplementary Table 1<\/a>. The exact heat treatment durations and conditions for all genotypes and assays are summarized in Supplementary Table 2<\/a>.<\/p>\n FLIM\u2013FRET assay<\/p>\n To examine heat-induced conformational change in OsALA5, FLIM\u2013FRET assays were performed in yeast cells co-expressing OsALA5 and OsALIS2 using an established method32<\/a>,74<\/a>. Using the structurally characterized yeast P4-ATPase Drs2p as a reference, OsALA5 was aligned and visualized with the ESPript v3.2 web server (https:\/\/espript.ibcp.fr\/ESPript\/cgi-bin\/ESPript.cgi<\/a>) to map the TM1\u2013TM4 region surrounding the predicted substrate-binding pocket. Guided by this alignment, a recombinant mRuby2\u2013OsALA5\u2013Clover construct was generated by inserting mRuby2 (amino acids 1\u2013237) at amino acid 105 of OsALA5 and Clover (amino acids 1-239) at amino acid 351 of OsALA5, corresponding to the extracellular loops between the M1\u2013M2 and M3\u2013M4 transmembrane regions. The fusion sequence was cloned into the yeast expression vector pMP4062 and transformed into the S. cerevisiae ZHY709 strain expressing OsALIS2 using the lithium acetate method. A donor-only construct (OsALA5\u2013Clover) generated in the same vector and transformed into the same background served as the control for FRET. Yeast strains carrying the respective constructs were grown in YPG medium to an optical density of OD600\u2009\u2248\u20090.5 and used for FLIM\u2013FRET measurements.<\/p>\n FLIM measurements were conducted using the STELLARIS 8 FALCON system (Leica). Samples were excited using a 500\u2009nm white-light laser (66% intensity), and fluorescence was collected using an HC PL APO CS2 63\u00d7\/1.40 oil objective. Emission from Clover (donor) was detected using HyD S detectors with a 516\u2013563\u2009nm detection window. Images were acquired with a 7.7\u2009\u00b5s pixel dwell time, 0.1\u2009\u00b5m pixel resolution (zoom 3.77, 512\u2009\u00d7\u2009512), and 8\u00d7 line accumulation to obtain sufficient photon counts for lifetime fitting. Measurements were performed under control (28\u2009\u00b0C) and heat stress (45\u2009\u00b0C, 5\u2009min) conditions. FRET efficiency was calculated based on the reduction in donor lifetime, and donor-acceptor distances were estimated according to F\u00f6rster theory (R0\u2009=\u20096.3\u2009nm) for the Clover\u2013mRuby2 pair. At least 30 cells were analysed for each condition. Data were processed using GraphPad Prism (v8.0.2), and primers used for construct generation are listed in Supplementary Table 1<\/a>.<\/p>\n Statistics and reproducibility<\/p>\n Statistical analyses were performed using GraphPad Prism. Data are presented as mean \u00b1 s.d. unless otherwise stated. The indicated n values denote biological replicate numbers or numbers of independent samples, as specified in the figure legends. Statistical tests used for each panel are stated in the corresponding figure legends. For comparisons between two groups, two-sided Student\u2019s t-tests were used. For multiple comparisons, one-way or two-way ANOVA followed by Tukey\u2019s, Dunnett\u2019s or \u0160\u00edd\u00e1k\u2019s multiple comparisons tests was used, as indicated in the figure legends. Exact P values are shown in the figure panels where applicable. For panels with extensive multiple comparisons, letter-based significance groups are used, and the corresponding significance thresholds are defined in the legends. Representative results were obtained from at least three independent experiments with similar results unless otherwise stated.<\/p>\n Software and data processing<\/p>\n Fluorescence data from the FlexStation 3 microplate reader were collected using SoftMax Pro software (Molecular Devices, v7.1). Confocal fluorescence images were acquired using ZEN software (Zeiss, v2.3). Flow cytometry data were collected using BD Accuri C6 Plus software (BD Biosciences). FLIM\u2013FRET data were acquired using LAS X software (Leica Microsystems, v4.7.0) on a STELLARIS 8 FALCON system. Phylogenetic analyses were performed using ClustalW version 2.1 via the GenomeNet web server for sequence alignment and MEGA (v5.1) for tree construction. DSC data were collected using DSCRun software (TA Instruments, v4.7.1) and analysed using NanoAnalyze software (TA Instruments, v3.11.0). Statistical analyses and graph generation were performed using GraphPad Prism (v8.0.2). Microsoft Excel 2019 was used for data processing and table preparation. Microsoft PowerPoint 2019 and Adobe Illustrator 2023 were used for figure preparation.<\/p>\n Reporting summary<\/p>\n