Cell lines, plasmids, antibodies and reagents

HEK293T (American Type Culture Collection (ATCC)) and FreeStyle 293-F (Thermo Fisher Scientific) cells were grown in DMEM containing 10% fetal calf serum (FCS) and FreeStyle 293 Expression Medium (Thermo Fisher Scientific), respectively. Mouse interleukin 3 (IL-3)-dependent Ba/F3 cells (ATCC) and their transformants were cultured in RPMI-1640 medium supplemented with 10% FCS and 50 U per ml recombinant IL-3. The C127I cells expressing mouse IL-3 were described previously65. Soluble FasL was produced as described using COS-7 cells66. The recombinant protein of mCherry-D4, a fluorescence-conjugating cholesterol-binding domain of perfringolysin O, and NT-Lys-RFP, a truncated nontoxic form of SM-binding toxin lysenin fused to red fluorescent protein (RFP), were prepared as described39,53,67. The Tmem63b-deficient Ba/F3 lines (Tm63bnull) were established by a CRISPR–Cas9 system using the pX330 vector68 carrying the following complementary sequences: 5′-CACCGCGTAGTCCCAGGCCACCTTC-3′ and 5′-AAACGAAGGTGGCCTGGGACTACGC-3′ (target sequences are underlined). The targeted loci was amplified with primers (5′- GCTCGGGCTCTTTCTTACCT-3′ and 5′-GCTGCTTTCCAACTCCTGTC-3′) and gene editing was verified by Sanger sequencing (Supplementary Fig. 1a).

The pMXs-puro and pGag-pol-IRES-bsr plasmids were from T. Kitamura (Foundation for Biomedical Research and Innovation at Kobe)69. The pCMV-VSV-G, pCMV-VSV-G-RSV-rev and pCAG-HIVgp plasmids were from H. Miyoshi (Riken Bioresource Center). The pCold I-His-NT-Lys and pCold I-His-mCherry-D4 plasmids were described previously31,67. The pAdVAntage plasmid was purchased from Thermo Fisher Scientific. The LentiCas9-Blast vector32, mouse CRISPR knockout pooled Library (GeCKO version 2)70 and pX330 plasmid were obtained from Addgene. Human complementary DNA (cDNA) for CDC50A (NM_018247.3), ATP11A (NM_015205.2) and ATP11A carrying the Q84E substitution (c.250C>G) was previously described31. The mouse Tmem63b cDNA (NM_198167.3; carrying synonymous single-nucleotide polymorphisms rs13465680 and rs13465681) was prepared by reverse transcription–PCR with RNA from Ba/F3 cells using the following primers: 5′-GCCTTAATTAAGCCACCATGCTGCCGTTCTTGCTG-3′ and 5′-GGCGAATTCCTGGTGAATCTCATTCTCTATG-3′ (PacI and EcoRI sites are underlined), tagged with Flag, EGFP or mCherry at the C terminus, introduced into pMXs-puro between PacI and EcoRI sites and used to infect Ba/F3 cells. The mTMEM63B mutants were prepared by NEBuilder HiFi DNA Assembly (New England Biolabs) or recombinant PCR using mTmem63b cDNA with primers carrying mutated nucleotides71 and mutations were certified by Sanger sequencing. The ΔCys-all cDNA flanked by PacI and EcoRI sites was synthesized by FASMAC, digested with PacI and EcoRI and inserted into the pMXs-puro vector for mCherry-tagged ΔCys-all expression. The synthesized cDNA was also used as a PCR template for ΔCys-IL1, ΔCys-hook 1 and ΔCys-hook 2 fragments, which were assembled using a HiFi Assembly kit. The sequence was verified by Sanger sequencing.

A horseradish peroxidase-conjugated rabbit anti-GFP antibody, rabbit anti-RFP antibody and mouse anti-DDDDK (Flag) antibody were purchased from Medical and Biological Laboratories. Hoechst 33342 and GolgiSeeing were from Dojindo and Funakoshi, respectively. PlasMem Bright, LysoTracker green and ER-Tracker green were purchased from Thermo Fisher Scientific. 1-palmitoyl-2-oleoyl-sn-glycero-3-PtdCho and N-stearoyl-d-erythro-sphingosylphosphorylcholine were purchased from Avanti Polar Lipids. MαCD was from FUJIFILM Wako Chemicals. FuGENE6 was purchased from Promega. Propidium iodide and the lactate dehydrogenase (LDH) cytotoxicity assay kit were purchased from Nacalai Tesque. Cinnamycin was purchased from Cayman Chemical. PLC, PLD, SMase and fatty acid-free BSA were purchased from Merck.

Cell transformation

Ba/F3 cells were transformed as described previously15,31. In brief, the pMXs-puro or pMXs-neo vector carrying cDNA for human CDC50A, ATP11A or mouse TMEM63B was transfected into HEK293T cells by FuGENE 6 (Promega) together with pGag-pol-IRES-bsr, pCMV-VSV-G and pAdVAntage. The retroviruses were concentrated by centrifugation or the Retro-X concentrator (Takara) and used to transform Ba/F3 cells. The transformants were selected by culturing the cells with 1 µg ml−1 puromycin. If necessary, the puromycin-resistant cells were subjected to sorting by FACSAria III (BD Biosciences) or CytoFLEX SRT (Beckman Coulter) for the expression of EGFP or mCherry.

Genome-wide CRISPR screening

Genome-wide CRISPR–Cas9 screening with Ba/F3 cells was performed as described70,72. In brief, Ba/F3 cells were coinfected with retroviruses carrying cDNA for Flag-tagged human CDC50A and mCherry-tagged human ATP11A-Q84E and the transformants were selected in the presence of 1 mg ml−1 G418. A lentivirus carrying cDNA for Cas9-Flag was produced by transfecting HEK293T cells with lentiCas9-Blast, pCAG-HIVgp and pCMV-VSV-G-RSV-rev and was used to infect Ba/F3 cells expressing hCDC50A and hATP11A-Q84E. The transformants were selected in the presence of 15 µg ml−1 blasticidin S and subjected to a limiting dilution. The clone expressing hCDC50A-Flag, Cas9-Flag and mCherry-hATP11A-Q84E was identified by western blotting with anti-Flag (DDDDK) or anti-mCherry (RFP) antibodies.

The lentivirus for the sgRNA library was produced by transfecting 1 × 107 HEK293T cells using FuGENE 6 with 17 µg of GeCKO version 2 library (A + B) DNA, 8 µg of pCAG-HIVgp and 5 µg of pCMV-VSV-G-RSV-rev. The virus was concentrated by centrifugation at 6,000g at 4 °C for 16 h and used to infect Ba/F3 cells expressing hCDC50A–Flag, Cas9-Flag and mCherry–hATP11A-Q84E at a multiplicity of infection of 0.3. Approximately 3 × 107 cells were placed in 24-well culture plates at 2 × 105 cells per well (a total of 150 wells) and spin-infected at 700g at 30 °C for 1 h in the presence of 10 µg ml−1 polybrene. The infected cells were then cultured for 3 days in culture medium containing 1 µg ml−1 puromycin and 15 µg ml−1 blasticidin. After tenfold dilution, the cells were further cultured for 2–3 days in the presence of puromycin and blasticidin S, followed by 2 days of culture without the antibiotics.

The mutated cells were then subjected to the flippase assay. In brief, 1 × 107 cells were washed twice with Hanks’ balanced salt solution (HBSS), suspended in 4 ml of HBSS containing 2 mM CaCl2 and 1 mM MgCl2 and preincubated at 15 °C for 10 min. NBD-PC was added to the cells at a final concentration of 500 nM, incubated at 15 °C for 6 min and put on ice. Then, 4 ml of HBSS containing 5 mg ml−1 fatty acid-free BSA was added to the solution and the cells were collected by centrifugation at 500g at 4 °C for 2 min. The cells were resuspended in prechilled HBSS containing 5 mg ml−1 fatty acid-free BSA and subjected to cell sorting with a FACSAria III. The cell population with a high NBD-PC signal (about 0.1%) was collected and cultured for 3–5 days. This sorting procedure was repeated for the cells with high NBD-PC flippase activity.

Genomic DNA was purified from the sorted cells by QIAamp DNA mini kit (Qiagen). Using 12 µg of DNA as the template, the lentiviral sequences were amplified by PCR with PrimeSTAR Max DNA polymerase (Takara Bio) and primers (5′-AATGGACTATCATATGCTTACCGTAACTTGAAAGTATTTCG-3′ and 5′-CTTTAGTTTGTATGTCTGTTGCTATTATGTCTACTATTCTTTCC-3′). The amplicons were subjected to a second PCR using PrimeSTAR HS DNA polymerase (Takara Bio) to connect adaptor sequences for next-generation sequencing with a mixture of forward primers (NGS-Lib-Fwd-1–10)73 and a common reverse primer (5′-CAAGCAGAAGACGGCATACGAGATGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTTCTACTATTCTTTCCCCTGCACTGT-3′). The PCR product was separated by electrophoresis, excised and purified using the Wizard SV gel and PCR clean-up system (Promega). The PCR product was quantified with a Quant-iT PicoGreen double-stranded DNA assay kit (Thermo Fisher Scientific) and subjected to deep sequencing by MiSeq (Illumina) using the MiSeq reagent Kit v3 (Illumina). The read sequences were assigned to the sgRNA sequences targeted to the respective genes and the abundance of each unique sgRNA was calculated using software custom-designed at Amelieff.

Incorporation of NBD lipids and detection of phospholipids and cholesterol

The incorporation of NBD lipids was assayed as described previously67. In brief, 5 × 105 cells were incubated at 4 °C with 0.1–0.01 µM NBD-PC or 1–0.2 µM NBD-SM in 500 µl of HBSS containing 1 mM MgCl2 and 2 mM CaCl2 (HBSS++). An aliquot (100 µl) of cells was mixed with 150 µl of HBSS containing 50 mg ml−1 fatty acid-free BSA, incubated for 1–2 min on ice and analyzed using CytoFLEX (Beckman Coulter). To detect PtdSer, SM and cholesterol on the cell surface, cells were incubated at 4 °C for 30–50 min in annexin V buffer containing 500-fold diluted annexin V–Cy5 (BioVision) or in HBSS containing 10 µg ml−1 NT-Lys-RFP or mCherry-D4. After counterstaining with 10 µg ml−1 propidium iodide, cells were analyzed by CytoFLEX. The PtdEtn exposure was examined as described previously41. In brief, cells were incubated at 4 °C for 40 min with 0.5 µM cinnamycin in 500 µl of annexin V buffer containing 5 mg ml−1 BSA and 10 µg ml−1 propidium iodide and analyzed by flow cytometry for propidium iodide positivity using CytoFLEX. In some cases, LDH released to the medium was evaluated using the LDH cytotoxicity assay kit according to the manufacturer’s instructions.

To examine the effect of the lipid composition of the PM on the incorporation of NBD lipids and the distribution of phospholipids on the cell surface, cells were pretreated in HBSS++ with 10 mM MβCD at 15 °C for 15 min or in annexin V buffer (10 mM HEPES–KOH buffer (pH 7.4) containing 140 mM NaCl and 2.5 mM CaCl2) with 50 U per ml PLD at 37 °C for 30 min, 1.0 U per ml PLC at 37 °C for 30 min or 0.5 U per ml SMase at 15 °C for 30 min and used in the assay described above.

Metabolic labeling of cells with [14C]choline and lipid extraction

Labeling cells with [14C]choline and detection of [14C]PtdCho and [14C]SM were essentially performed as described previously31. In brief, cells were cultured for 2 days in RPMI-1640 containing 10% FCS and 1 μCi of choline chloride [methyl-14C] (55 mCi mmol−1; American Radiolabeled Chemicals). The MαCD-mediated lipid exchange to extract lipids from the PM outer leaflet was performed as described previously54. In brief, 5.7 mg of SM was suspended in 4.5 ml of RPMI-1640 medium prewarmed to 70 °C and incubated at 70 °C for 5 min. After adding 500 μl of 400 mM MαCD, the solution was incubated at 37 °C for 1 h and centrifuged; the supernatant was used as SM-loaded MαCD. Cells were suspended in SM-loaded MαCD solution and incubated at 20 °C for 1 h. After centrifugation, the supernatants were mixed with a hexane–isopropanol mixture (3:2, v/v) and the organic phase was collected. Lipids in the water phase were re-extracted, combined with the first extract, dried under a nitrogen stream and dissolved in 25 μl of a methanol–chloroform mixture (1:1, v/v). Extracted lipids (8,000 disintegrations per minute) were separated on a high-performance TLC silica gel 60 plate (Merck) with a mixture of chloroform, methanol and ammonium hydroxide (65:25:5, v/v/v) as a solvent and visualized with an imaging plate detector (FLA-7000, FUJIFILM). Total cellular lipids were extracted by mixing with a hexane–isopropanol mixture (3:2, v/v) and analyzed as above.

Confocal microscopy analysis

The PM was stained with PlasMem Bright green and red, diluted 100-fold and 200-fold, respectively. For intracellular organelle staining, cells were incubated with 50 nM LysoTracker green (Thermo Scientific), 1 µM ER-Tracker green (Thermo Scientific) or 0.4 µM GolgiSeeing (Funakoshi). To detect SM-rich domains and cholesterol on the cell surface, cells were stained with 10 µg ml−1 NT-Lys-RFP and mCherry-D4, respectively. The cells were then counterstained with 3.6–8.9 µM Hoechst 33342 and transferred to a glass-bottom dish (Matsunami Glass); images were acquired using an LSM 710 (Carl Zeiss) or FV3000 (EVIDENT) confocal laser-scanning microscope, followed by image processing with ImageJ.

SDS–PAGE and western blotting

Cells were washed with PBS and lysed in radioimmunoprecipitation assay buffer (50 mM HEPES–NaOH (pH 8.0), 150 mM NaCl, 1% Nonided P-40, 0.1% SDS and 0.5% sodium deoxycholate) containing proteinase inhibitor cocktail (Nacalai Tesque). After removing insolubilized aggregates, the lysates were incubated at room temperature for 20 min in SDS sample buffer (62.5 mM Tris-HCl buffer (pH 6.8), 2% SDS, 10% glycerol, 5% β-mercaptoethanol and 0.001% bromophenol blue), separated by electrophoresis on a 7.5% or 10% polyacrylamide gel (Nacalai Tesque) and transferred to PVDF membranes (Merck Millipore). The membrane was blocked with TBS containing 0.05% Tween-20 and 5% skim milk and probed with anti-GFP (1:3,000), anti-RFP (1:3,000) or anti-DDDDK (Flag) (1:1,000) antibody, followed by detection with ChemiDoc Touch MP (Bio-Rad) or LAS4000 (GE Healthcare) using Chemi-Lumi One Super (Nacalai Tesque). Proteins on PVDF membranes were stained with Ponceau S or Coomassie brilliant blue (CBB). Precision Plus protein standards (Bio-Rad) and YesBlot western marker I (SMOBIO) were used as molecular weight markers.

Protein expression and purification for cryo-EM analysis

Mouse Tmem63b (mTMEM63B; UniProt Q3TWI9) cDNA tagged with a tobacco etch virus (TEV) cleavage site, EGFP and a Twin-Strep-tag at the C terminus was cloned into a pEG BacMam vector74. Bacmids were generated in Escherichia coli DH10Bac and used to transfect Sf9 cells (Thermo Scientific) with the Bac-to-Bac system for baculovirus production. FreeStyle 293-F cells (Thermo Scientific) at 2 × 106 cells per m were infected by concentrated baculovirus to a volume ratio of 100:1 (equivalent to 1:10 in the original concentration) and cultured at 37 °C for 18 h in FreeStyle medium (Thermo Scientific), followed by incubation at 30 °C for 48 h in the presence of 5 mM sodium butylate. Cells were collected by centrifugation, resuspended in buffer containing 50 mM Tris-HCl buffer (pH 8.0), 150 mM NaCl and protease inhibitor cocktail and lysed by sonication (Tomy/UD-211; output: 5, duty: 50, time: 30 min). The lysate was centrifuged at 26,000 r.p.m. for 1 h using an ultracentrifuge (Beckman Coulter, Optima XE-90, fixed-angle rotor 45Ti) to collect the membrane fraction, which was homogenized and stored at −180 °C until use.

mTMEM63B was purified from the membrane fraction by solubilizing with 1% LMNG and 0.1% CHS or 1.5% DDM and 0.3% CHS. After removing the insoluble fraction by centrifugation at 40,000 r.p.m. for 30 min, the supernatant was incubated for 1 h with 1 ml of anti-GFP nanobody-coupled CNBr Sepharose. The resin was loaded onto a column, washed with ten column volumes of buffer containing 0.02% LMNG and 0.002% CHS or 0.03% DDM and 0.006% CHS and treated at 4 °C overnight with 150 µg of TEV protease. mTMEM63B in the flowthrough fractions was purified by a Superose 6 Increase 3.2/300 Column (Cytiva) using the AKTA Pure 25M system (Cytiva) with a buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl and 1.0 mM 2-mercaptoethanol, supplemented with 0.003% LMNG and 0.0003% CHS or 0.03% DDM and 0.003% CHS. For the mTMEM63B sample solubilized with DDM–CHS, the protein was incubated at 4 °C for 10 min with anti-mTMEM63B Fab at a molar ratio of 2:1 before the size-exclusion chromatography. The purified mTMEM63B protein was concentrated to 5 mg ml−1 (LNMG–CHS sample) or 18 mg ml−1 (DDM–CHS sample) for the cryo-EM analysis. For the Fab-bound samples, the purified TMEM63B sample was mixed with anti-mTMEM63B YN9303-24 Fab at twofold molar excess (DDM–CHS sample) or 1.2-fold molar excess (LMNG–CHS sample) and then incubated at 4 °C for 10 min before the size-exclusion chromatography. The purified samples were concentrated to 5 mg ml−1 (LNMG–CHS sample), 3.5 mg ml−1 (LMNG–CHS sample with Fab), 11 mg ml−1 (DDM–CHS sample) or 18 mg ml−1 (DDM–CHS sample with Fab) for the cryo-EM analysis.

Antibody generation

All the animal experiments conformed to the guidelines of the Guide for the Care and Use of Laboratory Animals of Japan and were approved by the Kyoto University Animal Experimentation Committee (no. Med Kyo 22055). Mouse monoclonal antibodies against mTMEM63B were raised as previously described75. In brief, a proteoliposome was prepared by reconstituting purified mTMEM63B into phospholipid vesicles consisting of a 10:1 mixture of chicken egg yolk PtdCho (Avanti Polar Lipids) and the adjuvant lipid A (Sigma-Aldrich) to facilitate an immune response. The MRL/lpr mice (Japan SLC) were injected three times with the proteoliposome at 2-week intervals. Hybridomas were generated with NS-1 myeloma (ATCC) as described previously76. Biotinylated proteoliposomes were prepared by reconstituting mTMEM63B with a mixture of egg yolk PtdCho and 1,2-dipalmitoyl-sn-glycero-3-PtdEtn-N-(cap biotinyl) (16:0; Avanti Polar Lipids) and immobilized onto streptavidin-coated microplates (Nunc). Hybridoma clones were screened by ELISA on immobilized biotinylated proteoliposomes, allowing for positive selection of antibodies that recognized the native conformation of mTMEM63B. Positive hybridomas were further screened with SDS-denatured mTMEM63B to exclude the antibodies against linear epitopes. The complex formation between mTMEM63B and monoclonal antibodies was examined using fluorescence-detection size-exclusion chromatography and flow cytometry; one monoclonal antibody (clone YN9303-24) showed specific binding to the intracellular regions of mTMEM63B and was used for the cryo-EM analysis (Supplementary Fig. 3). The sequence of the Fab was determined by 5′-RACE with total RNA isolated from the hybridoma as described77.

Cryo-EM sample preparation, data collection and processing

First, 3 µl of protein solution (2.5 mg ml−1 or 5 mg ml−1 for the LMNG sample; 3.5 mg ml−1 for the LMNG sample with Fab; 5.5 mg ml−1 or 11 mg ml−1 for the DDM sample; 9 mg ml−1 or 18 mg ml−1 for the DDM sample with Fab) was applied to a Quantifoil grid (Quantifoil Au 1.2/1.3, Cu 0.6/1.0 or Cu/Rh 1.2/1.3 holey carbon 300 mesh) that was glow-discharged in advance at 10 mA for 50 s. All grids were blotted for 3 s at 6 °C and 100% humidity with a blot force of 10 and plunge-frozen into liquid ethane using a Vitrobot (Mark IV, Thermo Fisher). To solve the preferred orientation bias problem, a final concentration of 0.1% CHAPS or 1.5 mM F-FC8 was added to the drop of LMNG sample with Fab before applied to the grids.

Videos were collected at 300-kV acceleration voltage with a Titan Krios G4 (Thermo Fisher Scientific) equipped with a K3 detector (Gatan) in counting mode with correlative double sampling using automated EPU software (Thermo Fisher Scientific) in RIKEN Yokohama. Images were acquired at ×105,000 magnification with a pixel size of 0.83 Å. Videos were collected with 48 frames using a defocus range of −1.6 µm to −0.8 µm and the total exposure dose was 50.9 e− per Å2.

All data were processed using CryoSPARC (version 4.2.1)78. Videos were motion-corrected and the contrast transfer function was estimated using a patch-base method. The data processing for the LMNG–CHS sample (4,878 videos), LMNG–CHS sample with Fab (18,370 videos: 8,191 videos from the grid without detergents and 7,175 and 3,004 videos from the grids supplemented with F-FC8 and CHAPS, respectively), DDM–CHS sample (4,068 videos) and DDM–CHS sample with Fab (10,051 videos) are summarized in Extended Data Figs. 2–5. In brief, particles were automatically picked and extracted with a 256-pixel or 300-pixel square box size, which was Fourier-cropped to a 64-pixel square. The two-dimensional (2D) classification with different settings (initial batch size, number of classes, etc.) was concurrently performed to select better-looking micelle-shaped particles. The selected particles were used to reconstruct initial three-dimensional models and the best-looking model was used as a reference for further refinement. After several rounds of heterogeneous refinement, selected particles were re-extracted with a 300-pixel square box size and Fourier-cropped to a 160-pixel square (LMNG–CHS and DDM–CHS with Fab) or with a 256-pixel square box size and Fourier-cropped to a 128-pixel square (LMNG–CHS with Fab and DDM–CHS). Additional rounds of heterogeneous refinement were performed several times and final particles were selected and subjected to nonuniform refinement, yielding 3.4-Å, 3.6-Å, 6.9-Å and 3.5-Å resolution maps for LMNG–CHS, LMNG–CHS with Fab, DDM–CHS and DDM–CHS with Fab, respectively, determined by the gold-standard Fourier shell correlation (FSC = 0.143).

Modeling

Mode building and refinement were performed for the structures in LMNG–CHS and in DDM–CHS with Fab. The AlphaFold2-predicted model (AlphaFoldDB Q3TWI9) was globally fitted to each map using ChimeraX 1.5 and used as the initial model79. The model was then manually rebuilt using Coot 0.9.9.8 (ref. 80) and adjusted using an ISOLDE plugin in ChimeraX81; refmac-based reciprocal space refinement was performed against the halfmaps by Servalcat82, adjusting a weight parameter (--weight_auto_scale) to prevent model clashing in the less-resolved region. The densities for ICD in the LMNG–CHS structure, the YN9303-24 Fab fragment in the DDM–CHS structure and IL2 in both structures were weak or not observed in the cryo-EM maps and, thus, omitted in the model. The final model was validated by MolProbity83, EMRinger84 and Model-Map Q score in ChimeraX85. The data collection, refinement parameters and model statistics are summarized in Table 1.

Lipidomic analysis

Lipids were extracted according to Bligh and Dyer’s method with some modifications86. In brief, cells (1 × 106) were washed with prechilled PBS and snap-frozen in liquid nitrogen. The cell pellet (Ba/F3, Tm63bnull or Tm63bnull mTMEM63B cells with five biological replicates from each cell) was mixed with 1 ml of ice-cold methanol, transferred into a 2-ml tube and mixed with 10 µl of the internal standard (IS) solution A (mouse SPLASH Lipidomix MS standard, Avanti Polar Lipids) and 10 µl of IS solution B (Avanti Polar Lipids). Specifically, 10 µl of IS solution A contained 1.0 nmol of PtdCho (15:0/[2H7]18:1), 0.070 nmol of PtdEtn (15:0/[2H7]18:1), 0.20 nmol of PtdSer (15:0/[2H7]18:1), 0.05 nmol of PtdGly (15:0/[2H7]18:1), 0.20 nmol of phosphatidylinositol (15:0/[2H7]18:1), 0.10 nmol of phosphatidic acid (15:0/[2H7]18:1), 2.5 nmol of cholesteryl ester ([2H7]18:1), 0.15 nmol of diacylglycerol (15:0/[2H7]18:1), 0.35 nmol of triacylglycerol (15:0/[2H7]18:1/15:0) and 0.20 nmol of SM (d18:1/[2H9]18:1). On the other hand, 10 µl of IS solution B contained 0.10 nmol of free fatty acids ([13C16]16:0) and 3.0 nmol of cholesterol ([2H7]cholesterol). The sample was vigorously mixed by vortexing for 1 min, sonicated for 5 min and incubated on ice for 5 min. The extracts were centrifuged at 16,000g for 5 min at 4 °C and the supernatant was transferred to clean tubes. The protein concentration in the pellet was determined using a Pierce BCA protein assay kit (Thermo Fisher Scientific). Aliquots (400 µL) of the supernatant were dried under a stream of nitrogen, dissolved in 100 µl of methanol and chloroform (1:1) and analyzed by supercritical fluid chromatography (SFC) with a diethylamine (DEA) column coupled with triple-quadrupole MS87,88.

The conditions for the SFC (Nexera UC system, Shimadzu) were as follows: column, ACQUITY UPC2 Torus DEA column (3.0-mm inner diameter, 100 mm, 1.7 µm particle size; Waters); injection volume, 2 µl; column temperature, 50 °C; mobile phase A, supercritical carbon dioxide; mobile phase B (modifier), make-up pump solvent consisting of methanol and water (95:5, v/v) with 0.1% (w/v) ammonium acetate; flow rate of mobile phase, 1 ml min−1; flow rate of make-up pump, 0.1 ml min−1; back-pressure regulator, 10 MPa. The gradient conditions were as follows: 1% B, 0–1 min; 1–75% B, 1–24 min; 75% B, 24–26 min; 1% B, 26–30 min. The triple-quadrupole MS (LCMS-8060, Shimadzu) analysis conditions were as follows: polarity, positive and negative ionization; electrospray voltage, 4 kV in positive ion mode and ‒3.5 kV in negative ion mode; nebulizer gas flow rate, 3.0 L min−1; drying gas flow rate, 10 L min−1; desolvation line temperature, 250 °C; heat block temperature, 400 °C; detector voltage, 2.16 kV. The multiple reaction monitoring (MRM) parameters per time period were as follows: limit on number of MRM transitions, 150; dwell time, 2 ms; pause time, 2 ms; polarity switching time, 5 ms. Optimized MRM parameters are shown in Supplementary Table 1. The analytical platform for lipidomic analysis was controlled using LabSolutions (version 5.99 SP2, Shimadzu), and data analysis was performed using Multi-ChromatoAnalysT (version 1.3.4.0, Beforce). Lipid identification was performed on the basis of retention time and specific MRM transitions indexed to IS for each lipid class. Quantitative levels of lipids were calculated using peak areas relative to an IS and corrected for the total protein amount of each sample (Supplementary Table 1).

Detection of palmitoylated peptides

Shotgun proteomic analysis was performed as described previously89, with modifications. The purified mTMEM63B (1.0 µg of protein) was incubated at 37 °C for 30 min with 40 mM TCEP in 3.0 µl of 6.7 mM Tris-HCl buffer (pH 8.0) containing 0.01% DDM, 0.001% CHS, 50 mM NaCl and 0.33 mM β-mercaptoethanol. IAA was added to each sample at a final concentration of 80 mM and incubated at room temperature for 30 min. Three volumes of prechilled (−30 °C) acetone was added to the sample and the sample was kept at −30 °C for 2 h. The proteins were collected by centrifugation at 7,000g for 5 min at 4 °C, washed twice with 90% prechilled acetone, air-dried and dissolved by sonication in 2 µl of a solution of trypsin and Lys-C (250 ng µl−1) in 50 mM ammonium bicarbonate. The mixture was incubated at 37 °C for 2 h while shaken with a Thermomixer comfort (Eppendorf) at a setting of 300 r.p.m., diluted with two volumes of 0.5% (v/v) TFA and subjected to nanoflow liquid chromatography (nano-LC) high-resolution tandem MS/MS analysis.

The nano-LC–MS/MS system was composed of a Dionex Ultimate 3000 nano-RSLC system and a Q-Exactive HF high-performance benchtop quadrupole Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with a Dream spray electrospray ionization source (AMR). An Acclaim PepMap C18 column (Thermo Fisher Scientific; 300-µm inner diameter, 5 mm) with a particle size of 5 µm was used as the precolumn for sample trapping. The loading pump was run at 1 µl min−1 with water, acetonitrile and TFA (98/2/0.1, v/v/v) and 0.5 µL was injected per sample. After loading, the sample was switched online to the packed nano-LC column of 2-µm-particle-size L-column2 ODS (CERI; 50-µm inner diameter, 200 mm). The nano-LC conditions were as follows: mobile phase, water and FA (100/0.1, v/v; solvent A) and water, acetonitrile and FA (80/20/0.1, v/v/v; solvent B); flow rate, 200 nl min−1; column temperature, 40 °C. The gradient conditions were as follows: 5%–40% B, 0–100 min; 40%–99% B, 100–120 min; 99% B, 120–130 min; 99%–5% B, 130–131 min; 5% B, 131–150 min.

The MS analysis conditions were as follows: polarity, positive ionization; spray voltage, 1.8 kV; capillary temperature, 275 °C; S-lens level, 50; probe heater temperature, 350 °C; mass resolution, 60,000; automatic gain control (AGC) target (the number of ions used to fill the C-trap), 3,000,000; maximum injection time (IT), 110 ms; MS scan range, 350–1,500 m/z. The data-dependent MS/MS (dd-MS/MS) spectra (top seven) were acquired with higher-energy collision dissociation. The dd-MS/MS conditions were as follows: mass resolution, 60,000; AGC target, 200,000; maximum IT, 138 ms; isolation window, 1.2 m/z; normalized collision energy, 28 eV; ion selection abundance threshold, 140,000; charge exclusion, 1 and more than 7; dynamic exclusion, 15 s.

Proteome Discover version 2.4 (Thermo Fisher Scientific) was used for data processing, including peak detection, peak area calculations and peak identification. The MS/MS spectra were used to search the SwissProt database with decoy sequences to identify proteins using the Sequest HT algorithm (Thermo Fisher Scientific). The search was conducted with the following parameter settings: enzyme used, trypsin; allowed number of missed cleavages, 3; dynamic modifications, carbamidomethylation of cysteine and palmitoylation of cysteine. Assigned MS/MS spectra were filtered with a peptide confidence value of ‘high’ to obtain a false discovery rate of less than 1% with the following parameters: minimum number of peptides required for protein identification, 1; minimum number of amino acids required for protein identification, 4.

Statistics and reproducibility

The cell-based and biochemical experiments were independently conducted at least two times, as described in the figure legends, and similar results were confirmed. Analyses with western blots (SDS–PAGE) and confocal fluorescence microscopy were performed at least twice, confirming the results, and representative images are shown. P values were calculated using a two-sided Student’s t-test. Error bars indicate the s.d.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.