A novel sterol-binding protein reveals heterogeneous cholesterol distribution in neurite outgrowth and in late endosomes/lysosomes

Abstract

We identified a mushroom-derived protein, maistero-2 that specifically binds 3-hydroxy sterol including cholesterol (Chol). Maistero-2 bound lipid mixture in Chol-dependent manner with a binding threshold of around 30%. Changing lipid composition did not significantly affect the threshold concentration. EGFP-maistero-2 labeled cell surface and intracellular organelle Chol with higher sensitivity than that of well-established Chol probe, D4 fragment of perfringolysin O. EGFP-maistero-2 revealed increase of cell surface Chol during neurite outgrowth and heterogeneous Chol distribution between CD63-positive and LAMP1-positive late endosomes/lysosomes. The absence of strictly conserved Thr-Leu pair present in Chol-dependent cytolysins suggests a distinct Chol-binding mechanism for maistero-2.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-022-04339-6.

Introduction

Cholesterol (Chol) is the most abundant component of the mammalian plasma membrane. Small alteration of Chol level and/or distribution in the plasma membrane modulates the local biophysical properties of lipid membranes, leading to signal transduction, endocytosis, and adhesion. Mammalian cells acquire Chol by de novo biosynthesis or as extracellular nutrient by endocytosis of low density lipoprotein (LDL). In the LDL pathway, LDL-derived Chol-ester is transported to the late endosome/lysosome compartment where Chol-ester is hydrolyzed to free Chol. Free Chol in turn is shuttled to other organelles assisted by Niemann–Pick disease type C (NPC) 1 and NPC2 proteins [1]. Although the Chol de novo biosynthesis is primarily localized at the endoplasmic reticulum (ER), the Chol content of the ER membrane is rather low. The importance of Chol is illustrated by the stringent cellular control of its content and distribution. However, the detailed subcellular distribution of Chol remains poorly understood [2], in part due to a lack of appropriate visualization methods.

The fluorescent polyene antibiotic, filipin, has long been utilized to localize non-esterified sterols by fluorescent and electron microscopy [3–5]. While filipin remains highly useful tool to visualize Chol, it has been reported to also exhibit considerable affinity towards the ganglioside GM1 (Galβ1,3GalNAcβ1,4(NeuAcα2,3)Galβ1,4GlcCer) [6]. Filipin can bind to membranes containing > 15% Chol [3, 7], but rapid photobleaching of filipin renders fluorescence-based quantification difficult. Recently developed protein based Chol probes reveal a heterogeneous distribution and functional diversity of specific Chol pools [8, 9]. These probes are derivatives of non-toxic Chol-binding domain of bacterial Chol-dependent cytolysins (CDCs) [10, 11] and lipid-binding components of two-component pore forming toxins, aegerolysins, from mushrooms [12–14]. The non-toxic lipid-binding component of aegerolysins, such as pleurotolysin A2 [12] and ostreolysin [13] have initially been reported to bind to equimolar sphingomyelin (SM)/Chol complexes. Recently, an even higher affinity to equimolar ceramide phosphoethanolamine/Chol complexes has been demonstrated [15, 16]. Among CDC-derived probes, the Chol-binding domain 4 (D4) of perfringolysin O (PFO) and anthrolysin O (ALO) have been extensively characterized [17–23]. Membrane attachment of CDC-derived probes requires very high Chol concentration [7, 24, 25]. The high Chol content threshold and narrow dynamic range of these probes are, however, useful to detect small changes in membrane Chol concentration [26–29]. In addition, recently developed low Chol threshold mutants of D4 [25, 30, 31] are useful tools in cell biology. However, non-specific binding of a mutant has also been reported [32] as well as the influence of lipid and ionic environment on PFO and ALO binding [33, 34]. Thus careful control experiments are necessary to evaluate the obtained results [35].

Development of additional Chol-binding proteins with different structure remains an important avenue to further probe the distribution and function of different endogenous Chol pools. Previously, we reported a novel SM/Chol-binding protein, termed nakanori, isolated from the extracts of edible mushroom Grifola frondosa [36]. In the present study, we employed the same strategy, but shifted the focus towards small molecular weight proteins. The isolated protein specifically binds Chol with lower threshold compared to D4-derived probes. With the help of a fluorescently labeled version of this protein, we visualized an increase of cell surface Chol in neurite outgrowth and the heterogeneous distribution of Chol in late endosomes/lysosomes.

Materials and methods

Lipids

The following lipids were purchased from Avanti Polar Lipids (Alabaster, AL, USA): SM (porcine brain); l-α-phosphatidylcholine (egg; PC); 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC); 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC); l-α-phosphatidylethanolamine (egg; PE); 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE); l-α-phosphatidylinositol (bovine liver; PI); l-α-phosphatidylserine (porcine brain; PS); 1′3′-bis(1,2-dioleoyl-sn-glycero-3-phospho)-glycerol (oleoyl CL); l-α-phosphatidic acid (egg; PA); 5α-cholest-8(14)-en-3β-ol-15-one (15-ketocholestene); 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-DOPE). 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-DPPE) was purchased from Invitrogen (Rockford, IL, USA). Cholest-4-en-3β-ol (allocholesterol) and 5-cholesten-3α-ol (epicholesterol) were from Steraloids Inc. (Newport, RI, USA). Campesterol and stigmasterol were from Dr. Hubert Schaller. Other lipids were from Sigma-Aldrich (St. Louis, MO, USA).

Other materials

Methyl-β-cyclodextrin (MβCD) was from Cyclolab (Budapest, Hungary). Filipin was from Polyscience (Warrington, PA, USA). U18666A was from Enzo Life Sciences (Farmingdale, NY, USA). Anti-human LAMP1 mouse monoclonal antibody (mAb, H4A3) was from Developmental Studies Hybridoma Bank at the University of Iowa (Iowa, IA, USA). Anti-CD63 mAb was from Cymbus Biotechnology (Hampshire, UK). Anti-rat GM130 mAb, anti-human EEA1 mAb and anti-human Rab11 mAb were from BD Biosciences (San Jose, CA, USA). Anti-LC3 mAb was from Nano Tools Antikoerper Technik (Teningen, German). Alexa 488-anti mouse IgG and Alexa 546-anti mouse IgG were from Thermo Fisher Scientific (Waltham, MA, USA). Anti-penta His mAb was from Qiagen (Hilden, Germany). Horse radish peroxidase-conjugated anti-mouse IgG was from Cytiva (Marlborough, MA, USA). MβCD-cholesterol (MβCD/Chol complex) and Mowiol were from Sigma-Aldrich . Silver staining kit was from Cosmo Bio Co. (Tokyo, Japan). Other chemicals were from Nacalai tesque (Kyoto, Japan).

Mushroom Grifola frondosa and preparation of G. frondosa extract

Basidiomycete G. frondosa and G. frondosa extract were prepared as described [36].

Determination of partial amino acid sequence of maistero-1

Amino acid sequence of maistero-1 was determined as described [12]. Multilamellar vesicles (MLVs) composed of equimolar SM and Chol were prepared by hydrating a lipid film with phosphate buffered saline (PBS), vortex mixing and bath sonication. MLVs and G. frondosa extract were incubated in Eppendorf tubes at 37 ℃ for 30 min and the mixture was centrifuged at 12,000 rpm for 10 min at 22 ℃. The pellets were washed with PBS twice and subjected to SDS-PAGE followed by Coomassie brilliant blue staining. The protein bands around 10 kDa were excised and treated with 0.2 μg of Achromobacter protease I (a gift from Dr. Masaki, Ibaraki University) [37] at 37 ℃ for 12 h in 0.1 M Tris–HCl (pH 9.0) containing 0.1% SDS. For sequencing, peptides generated were extracted from the gel and separated on columns of DEAE-5PW (1 × 20 mm; Tosoh, Tokyo, Japan) and Inertsil ODS-3 (1 × 100 mm; GL Sciences Inc., Tokyo, Japan) connected in series with a model 1100 series liquid chromatography system (Agilent Technologies, Santa Clara, CA, USA). Peptides were eluted at a flow rate of 20 μL/min using a linear gradient of 0–60% solvent B in solvent A, where solvents A and B were 0.09% (v/v) aqueous trifluoroacetic acid and 0.075% (v/v) trifluoroacetic acid in 80% (v/v) acetonitrile, respectively. N-terminal amino acid sequence (KAaNSsYiDDafyirnqxnqx (peptide sequence written in uppercase letters were proven by Edman degradation, and those in lowercase letters indicate tentative identification)) and 3 internal sequences (VGFYR, AANSSYIDDA, YSGGDDSxFRL) were obtained by Edman degradation using a Procise cLC protein sequencing system (Applied Biosystems, Thermo Fisher Scientific) and by matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) on an Ultraflex MALD-TOF (Bruker Corporation, Billerica, MA, USA) in a reflector mode using α-cyano-4-hydroxycinnamic acid as a matrix.

Molecular cloning of maistero-1–3

Maistero-1–3 were identified based on in-house G. frondosa genome database published by Sato et al. [38]. Full-length cDNAs of maistero-1, maistero-2 and maistero-3 were obtained by 5’- and 3’-RACE using the SMARTer RACE cDNA Amplification Kit (Clontech, Mountain View, CA, USA) according to the manufacture’s protocol. Total RNA for RACE was isolated from G. frondosa strain M51 (Gf-N2) as described [39]. The RACE PCR was run under the following conditions; initial denaturation at 94 ℃ for 1 min, then 30 cycles of 94 ℃ for 30 s, 68 ℃ for 30 s and 72 ℃ for 3 min. Sequences for cDNA isolation are 5′-TCAACACCATGAGCGAGTCCTTC-3′ and 5′-AGAAGCTACTGAAACACCCAGG-3′ (maistero-1), 5′-ACCAAGGGCTGAAATGCTTTAC-3′ and 5′-GTTGCTACGTGAAACTGCGTGG-3′ (maistero-2), and 5′-GTAGAACTTACAATGGCATCC-3′ and 5′-CGATTGGTCTTAGGAAGTTTGG-3′ (maistero-3). Resultant PCR fragments were purified using Gel Extraction Kit (Qiagen) and cloned into the pMD-20 T vector (TaKaRa, Shiga, Japan), and then sequenced. Obtained cDNA of maistero-1, maistero-2 and maistero-3 were 510 bp (106 AA) (Genbank code LC579959), 335 bp (93 AA) (Genbank code LC579961) and 327 bp (101 AA) (Genbank code LC579960) length, respectively.

Construction, expression and purification of recombinant proteins

The cDNA fragments of maistero-1, maistero-1t (amino acid 13 to 106), maistero-2 and maistero-3 were cloned into the expression vector pCold I (TaKaRa) to obtain His-tagged recombinant proteins fused at N-terminus. For EGFP-fused proteins, the cDNA fragment of EGFP was amplified by PCR from pEGFP-C1 (Clontech) and subcloned into pCold I-maisteros at the site between His-tag and maisteros. Plasmids for substitution mutants of maistero-2 were generated by KOD-Plus-mutagenesis kit (TOYOBO, Osaka, Japan) according to the manufacturer’s instructions, using plasmids encoding pCold I-maistero-2 and pCold I-EGFP-maistero-2. The resulting plasmids were transformed into Escherichia coli strain BL-21(DE3) (BioDynamics Laboratory, Tokyo, Japan). Recombinant protein expression was induced by culturing cells at 15 ℃ overnight in the presence of 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG). After harvesting cells by centrifugation, the pellets were resuspended in buffer A (20 mM Hepes, 500 mM NaCl, 20 mM imidazole, pH 7.5) and lysed by sonication on ice, followed by centrifugation at 20,000g for 10 min at 4 ℃ to collect soluble lysates. His-tagged recombinant proteins were purified by HisTALON Superflow cartridge (1 mL, Clontech) using ÄKTA purifier system (Cytiva) according to the manufacturer’s instructions. Elution was performed using buffer A and buffer B (20 mM Hepes, 500 mM NaCl, 500 mM imidazole, pH 7.5) with a gradient 10%/min at flow rate 1 mL/min by monitoring at 280 nm. Eluted fractions were concentrated and  the buffer was exchanged into PBS using Amicon Ultra centrifugal filters (Millipore, Burlington, MA). The protein concentrations was determined by BCA protein assay. Recombinant mCherry-D4 protein was prepared as previously described [29].

Enzyme-linked immunosorbent assay (ELISA)

The binding of recombinant His-maistero-1–3, His-EGFP-maistero-1–3, and mutant proteins to various lipids (0.5 nmol/well) was measured by ELISA as described [40] with a slight modification. Anti-penta His antibody and horse radish peroxidase-conjugated anti-mouse IgG were used to detect the recombinant proteins bound to lipid.

Cell culture

All cells were obtained from American Type Culture Collection (ATCC) and maintained at 37 ℃ in a 5% CO₂, 95% air incubator. Chinese hamster ovary (CHO)-K1 cells were maintained in Ham’s F-12 medium supplemented with 10% fetal calf serum (FCS), 100 units/mL penicillin and 100 μg/mL streptomycin. HeLa S3 and Neuro 2a cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS, 100 units/mL penicillin and 100 μg/mL streptomycin. To induce the differentiation of Neuro 2a cells, cells were cultured in DMEM without FCS for 2 days. To introduce Chol, cells were cultured in the presence of 40 μM MβCD/Chol complex for 20 h.

Cell surface labeling

Cells were grown on glass coverslips as described above and fixed with 4% paraformaldehyde (PFA) in PBS for 20 min, quenched with 50 mM ammonium chloride for 10 min and blocked with 0.2% gelatin in PBS for 30 min. The cells were then treated with or without 20 mM MβCD for 30 min, followed by incubation with EGFP-maistero (20 μg/mL) and mCherry-D4 (8 μg/mL) for 1 h and additional incubation with 50 μg/mL filipin for 30 min. After washing with PBS, the specimens were mounted with Mowiol and observed under a confocal microscope. For His-maistero staining, cells were incubated with His-maistero-1 or -2 (20 μg/mL) after blocking, followed by incubation with anti-penta His antibody and Alexa 488-anti mouse IgG. All manipulations were done at room temperature.

Intracellular labeling

All manipulations were done at room temperature. HeLa S3 cells were grown on glass coverslips as described above and fixed with 4% PFA for 20 min and then quenched with 50 mM ammonium chloride for 10 min. After washing with PBS, cells were permeabilized by dipping the coverslips into liquid nitrogen for 2 s (freeze and thaw) [41] and then blocked with 0.2% gelatin for 30 min followed by incubation with antibodies of organelle markers for 1 h. After washing with PBS, cells were additionally incubated with EGFP-maistero-2 and Alexa 546 anti-mouse IgG for 1 h and the specimens were mounted with Mowiol and observed under a confocal microscope. To observe U18666A treated cells, cells on coverslips were cultured in the presence of 5 μM U18666A for 24 h and then fixed, quenched, permeabilized and blocked as described above. Cells were incubated with EGFP-maistero-2, mCherry-D4 and antibodies for 1 h followed by additional incubation with fluorescent second antibody and 50 μg/mL filipin for 1 h and then mounted and observed. For colocalization analysis, linescans were drawn using Image J software (plot profile function). For staining of tryptophan mutants of maistero-2, cells on coverslips were fixed, quenched, permeabilized and blocked as described above. Cells were incubated with EGFP-maistero-2 tryptophan mutants for 1 h and then mounted and observed.

Living cell labeling

Cells grown on glass-bottomed dishes were washed with FCS-free DMEM/Ham’s F-12 (1: 1) three times and incubated with EGFP-maistero-2 (20 μg/mL) in FCS-free DMEM/Ham’s F-12 medium at 37℃, 15℃ or on ice for 30 min. After washing with FCS-free DMEM/Ham’s F-12 medium, cells were fixed with 4% PFA for 20 min and observed using a confocal microscope. In Fig. S3, living cells were observed by confocal microscopy. In some experiments, cells were incubated with EGFP-maistero-2 at 37 ℃ for 30 min, washed, and additionally cultured in FCS-free medium at 37 ℃ for 30 min followed by fixation and observation.

Confocal microscopy

The stained cells were imaged using an LSM510 confocal microscope (Carl Zeiss, Oberkochen, Germany) equipped with a C-Apochromat 63XW Korr (1.2 NA) objective or TCS SP8 confocal microscope (Leica, Wetzlar, Germany) equipped with a 63 × HC PL APO CS2 oil immersion objective.

Quantitation of lipids

Cells on culture dishes were washed with cold PBS and harvested by scraping. Total lipids were extracted according to the method of Bligh and Dyer [42]. The amount of total phospholipids was determined by phosphorus assay [43]. For estimation of cholesterol content, total lipids were separated by thin layer chromatography (n-hexan:diethylether:acetic acid = 80:20:2). Sterols were stained by spraying ferric chloride solution and heating at 120℃ for 3 min [44]. The band intensity at Chol position was quantified using Image J software.

Estimation of EGFP-maistero bound to cells

CHO-K1 cells were seeded on 96-well plates and cultured for 1 day. After washing with PBS, cells were fixed, quenched, blocked, and incubated with EGFP-maistero-1 or -2 as described above. Bound EGFP-tagged proteins were detected by the same method as ELISA described above.

Liposome-binding assay

For liposome-binding assay, 0.5 mol % of NBD-DOPE was added to the lipid mixture as vesicle marker prior to lipid film preparation . MLVs in PBS were prepared as described above. The MLVs were subjected to probe sonication to prepare small unilamellar vesicles (SUVs). Vesicles containing total 250 nmol lipid, recombinant His-tagged proteins (0.2–0.3 nmol) and 1% (w/v) bovine serum albumin (BSA) were incubated in 0.5 mL PBS at 37 ℃ for 30 min. The suspension was mixed with 1 mL of 2.1 M sucrose in PBS, loaded at the bottom of an ultracentrifuge tube and overlaid with 1.5 mL of 1.2 M sucrose and 2 mL of 0.8 M sucrose. The gradient was centrifuged at 35,000 rpm at 4 ℃ for 1 h using a Beckman Coulter Optima ultracentrifuge equipped with MLS-50 rotor. Top liposome fraction (100–200 μL) was collected and normalized by measurement of fluorescence intensity of NBD (485/538 nm), and then subjected to SDS-PAGE and silver staining. The band intensity of His-tagged proteins was quantified using Image J software, and the relative binding was normalized against the value of His-maistero-2 bound to POPC/Chol (1:1) SUV.

Quenching of NBD-DPPE in the most outer leaflet of MLVs and SUVs

MLVs and SUVs composed of equimolar POPC and Chol in the presence of 0.5% NBD-DPPE were prepared as described above. Fluorescence intensity at 536 nm of vesicles containing a total of 0.1 mM lipids in PBS was measured at 20 °C every 1 s on a FluoroMax Plus spectrofluorometer (Horiba, Kyoto, Japan) at the excitation wavelength of 465 nm. Sodium dithionite (final 30 mM) was added where indicated.

Quartz crystal microbalance with dissipation monitoring measurement

Quartz crystal microbalance with dissipation monitoring (QCM-D) measurements were performed according to a published method [15].

Viability of cells exposed to His-maistero-2

To evaluate the cytotoxicity of maistero-2, CHO-K1 cells on 96-well plates were cultured in the presence or absence of His-maistero-2 for 26 h. 10 μL/well of cell count reagent SF (Nacalai tesque) was added to each well during the last 4 h of incubation. The absorbance at 490 nm with reference at 630 nm was measured using a microplate reader model 680 (Bio Rad, Hercules, CA). To observe the morphology of maistero-2 treated cells, HeLa S3 cells were grown on glass-bottomed dishes. After washing with FCS-free DMEM/Ham’s F-12 medium three times, cells were cultured in the presence or absence of 2 mg/mL His-maistero-2 in FCS-free DMEM/Ham’s F-12 medium for 1 or 24 h. Cells were then washed with PBS and fixed with 4% PFA for 20 min followed by observation under microscope.

Circular dichroism (CD)

CD spectra were recorded on J-1500 spectropolarimeter (Jasco, Tokyo, Japan) at 25℃ using a quartz demountable cell with a path length of 0.1 mm (Jasco, Tokyo, Japan). His-tagged recombinant proteins were dissolved to 0.1 mg/mL in 20 mM Hepes (pH 7.5) buffer containing 120 mM NaF. 20 scans were accumulated and averaged in the wavelength range from 188 to 245 nm with a step size of 0.05 nm, a bandwidth of 1 nm, a scan speed of 100 nm/min. A control spectrum of buffer was subtracted from each protein spectrum, and the mean residue ellipticity was calculated. Percentages of secondary structures were estimated using the Jasco protein secondary structure estimation program [45].

Results and discussion

As a first step, extracts from the mushroom G. frondosa were incubated with MLVs composed of equimolar SM and Chol. Following sedimentation by centrifugation, SDS-PAGE revealed the co-sedimentation of two major protein bands (Fig. 1A). The upper band (indicated by *) was confirmed to be identical with the previously reported protein, nakanori [36]. The lower band at around 12 kDa (indicated by an arrow) has not been previously identified. This lighter band was extracted and subjected to amino acid sequencing as described in “Materials and methods”. Comparison with the in-house G. frondosa transcriptome database identified a novel 106 amino acid polypeptide. We named this protein maistero-1 (maitake (Japanese name of G. frondosa) mushroom-derived sterol-binding protein-1). A BLAST search (data not shown) revealed no apparent sequence homology with already characterized proteins. However, two homologous sequences, maistero-2 and maistero-3, were identified in G. frondosa (Fig. 1B). Maistero-2 represents a pseudogene due to its terminal codon after the 7th amino acid, while maistero-3 is a 101 amino acid long polypeptide. In case of maistero-1, the N-terminal amino acid sequencing suggests truncation of the first 12 amino acids of the endogenous mushroom protein, herein referred to as maistero-1t.

Fig. 1.

Identification of Chol-binding proteins maistero-1, -2 and -3 from mushroom G. frondosa. A Screening of the SM/Chol-binding protein from G. frondosa. Lane 1: molecular weight marker; lane 2: mushroom extract; lane 3: supernatant fraction; lane 4: liposome pellet fraction. Arrow indicates the position of maistero-1. Asterisk indicates the position of nakanori. B Alignment of the amino acid sequence of maistero-1, -2 and -3. Identical and similar amino acids are shaded with black and gray, respectively. Predicted CRAC motifs in maistero-2 are marked with blue lines. Predicted CRAC motif only in maistero-3 is marked with a light blue line. C Binding of His-tagged maistero proteins to Chol measured by ELISA. Blue triangle: His-maistero-1; white triangle: His-maistero-1t; red circle: His-maistero-2; black square: His-maistero-3. Data are the means ± SD of three experiments

Next, His-tagged recombinant maistero-1, maistero-1t, maistero-2 and maistero-3 were prepared and their binding to pure Chol was measured by ELISA (Fig. 1C). Unlike previously reported nakanori, which binds only to SM and Chol mixtures [36], all maistero proteins bound to Chol in the absence of other lipids.

To assess the general suitability of maistero proteins for cell labeling, EGFP constructs of maistero-1, -2 and -3 were prepared. PFA-fixed CHO-K1 cells were co-labelled with EGFP-maistero and filipin (Fig. 2). EGFP-maistero-2 showed cell labeling, while maistero-1 and -1t (data not shown) failed to label the cells (Figs. 2A, S1). To confirm the Chol dependence of maistero labeling, CHO-K1 cells were treated with MβCD to reduce cellular Chol. Cells were then subjected to the same co-labeling conditions. 65% of cellular Chol was removed by MβCD (Fig. S2). As expected, filipin labeling was dramatically reduced in all cases, confirming decrease of cellular Chol (Fig. 2B). The small dot-like labeling by EGFP-maistero-3 was not affected by MβCD treatment. In contrast, similar to filipin labeling, EGFP-maistero-2 labeling was reduced in MβCD-treated CHO-K1 cells, indicating that EGFP-maistero-2 binds cellular Chol. Squared areas in Fig. 2A were enlarged in Fig. S3A. Filipin is membrane permeable and labels intracellular organelles. Filipin labeled pericentriolar compartment (arrowhead), which was not labeled with EGFP-maistero-2. Previously we showed that the filipin-labeled pericentriolar compartment was colocalized with a recycling endosome marker rab11 and endocytosed transferrin in CHO-K1 cells, indicating the compartment is a recycling endosome [46]. These results suggest that EGFP-maistero-2 selectively labels the plasma membrane under the experimental conditions. Figure S3A shows that maistero-2 labeled heterogeneous dot-like structures. Maistero-2 was non-toxic and endocytosed very slowly at 37 °C (see below). Irregular dot-like structures were also observed when living CHO-K1 cells were labeled with EGFP-maistero-2 at 15 °C (Fig. S3B), a condition endocytosis is decreased. He et al. [47] also showed heterogeneous dot-like structures when living CHO-K1 cells were labeled with ALO D4 and observed by nanoscale secondary ion mass spectrometry (NanoSIMS). The authors identified these structures as microvilli. These results suggest that, similar to D4, maistero-2 preferentially labels microvilli on the plasma membrane of CHO-K1 cells.

Fig. 2.

Binding of EGFP-maistero-1, -2 and -3 to CHO-K1 cells. Fixed CHO-K1 cells were treated with (B) or without (A) 20 mM MβCD for 30 min. Subsequently, the cells were doubly labeled with 20 μg/mL of the indicated EGFP-protein and 50 μg/mL filipin as described in “Materials and methods”. Representative confocal microscope images of at least three experiments are shown. Higher magnified images of the indicated regions (white square) are shown in Fig. S3. Filipin was shown by pseudocolor to make comparison easier

Next, the binding specificity of His-maistero-1 and His-maistero-2 was examined in more detail against defined lipid mixtures by ELISA (Figs. 3, 4). His-maistero-2 bound to POPC/Chol mixtures containing at least 30% Chol (Fig. 3A, B). Binding of His-maistero-1 was inefficient at 30% Chol (Fig. 3C, D). Similar results were obtained using EGFP-maistero-1 and -2 (Fig. S4), suggesting that the protein tags do not significantly affect binding of maistero protein to Chol-containing membranes. As observed with EGFP-maistero proteins, His-maistero-2 but not His-maistero-1 bound CHO-K1 cells (Fig. S5), confirming the independence of maistero binding from its respective protein tags. Binding of EGFP-D4 to egg PC/Chol liposomes has previously been reported to require at least 40% Chol content [7]. Maistero-2 has a lower Chol threshold for binding compared to EGFP-D4. The Chol threshold for PFO binding to POPC/Chol has also been reported to be around 40% [48], while in DOPC/Chol liposomes, it was decreased to 25% [33] and in POPC/POPE/SM (1:1:1)/Chol liposomes to 36.5% [25]. His-maistero-2 retained a Chol threshold of 30% in DOPC/Chol (Fig. 3E, F) and POPC/POPE/SM (1:1:1)/Chol (Fig. 3G, H) mixtures. In the plasma membrane, PC and SM are located in the outer leaflet whereas PE is in the inner leaflet [49–51]. We also examined POPC/SM (2:1)/Chol mixture. This mixture also gave similar binding threshold (Fig. 3K, L). In contrast, the binding threshold of His-maistero-1 to POPC/POPE/SM (1:1:1)/Chol and POPC/SM (2:1)/Chol was increased to 50% (Fig. 3I, J, M, N). Chol concentration in the plasma membrane of CHO-K1 cells has been reported to be around 30–35% [52, 53]. The high Chol threshold for maistero-1 binding in the presence of SM may explain its failure to label CHO-K1 cells. The different effect of the lipid composition on maistero and PFO binding hints at a dissimilar binding mechanism.

Fig. 3.

Binding of His-maistero-1 and -2 to Chol measured by ELISA. A, B Binding of His-maistero-2 to indicated POPC/Chol mixtures. C, D Binding of His-maistero-1 to indicated POPC/Chol mixtures. E, F Binding of His-maistero-2 to indicated DOPC/Chol mixtures. G, H Binding of His-maistero-2 to indicated POPC/POPE/SM (1:1:1)/Chol mixtures. I, J Binding of His-maistero-1 to indicated POPC/POPE/SM (1:1:1)/Chol mixtures. K, L Binding of His-maistero-2 to indicated POPC/SM (2:1)/Chol mixtures. M, N Binding of His-maistero-1 to indicated POPC/SM (2:1)/Chol mixtures. B, D, F, H, J, L, N The individual data point corresponding to 1.25 μg/mL His-maistero proteins was redrawn from left graph. Data are the means ± SD of three experiments

Fig. 4.

Binding specificity of His-maistero-2 by ELISA. A Binding of His-maistero-2 (1.25 μg/mL) to indicated lipids. SM, porcine brain sphingomyelin; PC, egg phosphatidylcholine; PS, porcine brain phosphatidylserine; PI, bovine liver phosphatidylinositol; PE, egg phosphatidylethanolamine; PA, egg phosphatidic acid; PG, 1,2-dioleoyl-sn-glycero-3-phsopho-rac-(1-glycerol); CL, oleoyl cardiolipin. B Binding of His-maistero-2 (1.25 μg/mL) to neutral lipids. Chol-Ole, cholesteryl oleate; TG, 1,2,3-trioctadecanoylglycerol; DG, 1-oleoyl-2-acetyl-sn-glycerol. C Binding of His-maistero-2 (1.25 μg/mL) to indicated sterols. A–C Relative binding was calculated with the value of Chol as 1. Data are the means ± SD of three experiments. D Structures of the sterols used in (C)

In the absence of Chol, maistero-2 did not bind to common phospholipids (Fig. 4A) and neutral lipids (Fig. 4B). The failure of maistero-2 to recognize cholesteryl ester prompted us to further explore the structural sterol features required for maistero-2 recognition (Fig. 4C, D). Loss of H-bond donating ability, such as in cholesteryl acetate, 5α-cholestan-3-one, coprostane and 5α-cholestane, was associated with a loss of maistero-2 binding. Increased B or D ring size of the sterol backbone due to oxidation, such as in 5α-cholest-8(14)-en-3β-ol-15-one and 6-ketocholestanol, abolished maistero-2 association, while the conjugated double bond system in the B ring, such as in ergosterol and 7-dehydrocholesterol, was associated with reduced maistero-2 interaction. Out of plane orientation of the 3-hydroxyl function, as in epicholesterol, was detrimental to maistero-2 interaction. Similarly, distortion of the A-ring strongly affected maistero-2 recognition. Dihydrocholesterol featuring a planar sterol backbone similar to Chol and allocholesterol with the flattened A-ring were very well recognized. In contrast, out of plane conformation of the A-ring compared to the B,C,D-ring skeleton of coprostanol and epicoprostanol was associated with a loss of maistero-2 binding (Fig. S6). This is in strong contrast to the reported binding of PFO [33] and streptolysin O [54] to coprostanol. Alterations of the sterol side chain due to alkylation or saturation, such as in sitosterol, campesterol, stigmasterol or desmosterol, exhibited partial to no effect on maistero-2 recognition. Taken together, maistero-2 sterol interaction exhibited a higher sensitivity to increased roughness of the usually smooth α-face, while it was more indifferent to side chain modifications.

The above outlined Chol dependence of maistero-2 binding observed by ELISA was further corroborated using liposomes. SUVs featuring different POPC/Chol ratios were subjected to the liposome-binding assay in the presence of His-maistero-2 (Fig. 5A). Consistent with ELISA results, His-maistero-2 bound liposomes containing 30% Chol or higher. In the absence of Chol, no His-maistero-2 binding to POPC SUVs was observed (Fig. 5B). Switching from POPC/Chol (1:1) SUVs to POPC/Chol (1:1) MLVs decreased His-maistero-2 binding to 53% (0.59–0.12 (relative binding to MLVs) vs 1–0.12 (relative binding to SUVs)). In Fig. 5C, D, NBD fluorescence of NBD-DPPE doped POPC/Chol (1:1) SUVs and MLVs was quenched by sodium dithionite. Sodium dithionite irreversibly quenches NBD fluorescence [55]. Since sodium dithionite penetrates membranes very slowly, NBD-DPPE in the most outer layer of the liposomes is quenched under the experimental condition [55–57]. 50% of NBD fluorescence was quenched in SUVs (Fig. 5C), consistent with SUVs composed of a single bilayer. In contrast, only 25% of fluorescence was quenched in case of MLVs (Fig. 5D). This reduction in sodium dithionite access to NBD-DPPE is consistent with the above reported reduced His-maistero-2 binding to MLVs compared to SUVs. These results indicate that His-maistero-2 binds Chol-containing liposomes in curvature-independent manner and His-maistero-2 does not penetrate membranes. The dissociation constant (Kd) of EGFP-maistero-2 to equimolar DOPC/Chol membranes was determined to be 60 nM by QCM-D (Fig. S7). This dissociation constant is similar to the previously reported Kd of 52 nM of D4 [58].

Fig. 5.

Binding of His-maistero-2 to liposomes. A, B His-maistero-2 associated to liposome fractions analyzed by SDS-PAGE. Relative binding was calculated with the value of POPC/Chol (1:1) SUV as 1. Representative gel images are shown from at least three experiments. A Binding to SUVs composed of indicated POPC/Chol ratios. B Binding to MLVs composed of POPC/Chol (1:1) or to SUVs composed of POPC/Chol (1:1) or pure POPC. Quenching of NBD-DPPE in the most outer leaflet of SUVs (C) and MLVs (D) composed of POPC/Chol (1:1) by sodium dithionite. Sodium dithionite was added to liposome solutions (red) or PBS (blue) at the timepoint marked with an arrow. Representative spectra from three experiments are shown

CHO-K1 cell viability was not affected by incubation with His-maistero-2 for 26 h (Fig. S8A). Similarly, HeLa S3 cells did not show altered morphology based on DIC imaging after 24 h incubation in the presence of 2 mg/mL His-maistero-2 (Fig. S8B). Together, this suggests that even high doses of maistero-2 do not exhibit significant toxic effects on mammalian cells. Efficient CHO-K1 cell surface labeling was observed within 30 min upon incubation with EGFP-maistero-2 at low temperature (on ice) prior to cell fixation (Fig. S8C). In contrast, HeLa S3 cells were not labeled by EGFP-maistero-2 on ice (Fig. S8D, left panels). Incubation of HeLa S3 cells with EGFP-maistero-2 at 37 °C resulted in weakly labeling of the cell surface and microvilli (Fig. S8D, middle panels). When HeLa S3 cells were subsequently chased for 30 min at 37 °C, a small amount of EGFP fluorescence was observed inside the cells (Fig. S8D, right panels), suggesting slow endocytosis of EGFP-maistero-2 by HeLa S3 cells.

Double labeling of fixed CHO-K1 cells with EGFP-maistero-2 and mCherry-D4 showed cell surface staining of both probes with partial overlap (Fig. 6A). Next, cell surface Chol was triple labeled in fixed neuroblastoma Neuro 2a cells with EGFP-maistero-2, mCherry-D4 and filipin. Undifferentiated Neuro 2a cells were weakly labeled with filipin, but not with mCherry-D4 (Fig. 6B). The Chol labeling threshold of filipin in PC/Chol membranes has been reported as 15% [3, 7], while the threshold of EGFP-D4 is 40% [7]. EGFP-maistero-2 with a threshold of 30% partially labeled the plasma membranes of a limited number of undifferentiated Neuro 2a cells (Fig. 6B). After differentiation in serum-free medium, cellular Chol was increased (Fig. 6D). Under this condition, all cells were labeled with EGFP-maistero-2 (Fig. 6C). An increase of cell surface Chol is also supported by a brighter filipin labeling. On the other hand, mCherry-D4 failed to label differentiated Neuro 2a cells, suggesting that the concentration of cell surface Chol remained lower than the labeling threshold of mCherry-D4. These results indicate that Neuro 2a differentiation is associated with an increase in cell surface Chol. This observation is consistent with recent reports that raft (sphingolipid and Chol enriched microdomains) associated Fyn stimulates neurite outgrowth in Neuro 2a cells [59]. Further increase of Chol by the addition of MβCD/Chol complex resulted in a strong labeling by mCherry-D4 (Fig. 6E). This indicates that the lack of Neuro 2a cell labeling with mCherry-D4 is indeed due to low Chol concentration in the plasma membrane. Taken together, this raises the possibility that maistero-2 could act as a bioprobe to visualize neurite outgrowth in Neuro 2a cells during differentiation.

Fig. 6.

Cell surface labeling by EGFP-maistero-2 and mCherry-D4. A Fixed CHO-K1 cells were doubly labeled with EGFP-maistero-2 and mCherry-D4 as described in “Materials and methods”. Fixed Neuro 2a cells, undifferentiated (B) and differentiated following withdrawal of FCS (C), were triply labeled with EGFP-maistero-2, mCherry-D4 and filipin as described in “Materials and methods”. D Relative amount of Chol normalized against phospholipids content was measured in undifferentiated (solid bar) and differentiated (open bar) Neuro 2a cells. Data are the means ± SD of three experiments. E Neuro 2a cells were doubly labeled with mCherry-D4 and filipin before (upper panels) or after (lower panels) introducing Chol by addition of MβCD /Chol complex. A–C, E Representative images of at least three experiments are shown

As reported above, cell surface labeling of HeLa S3 cells by EGFP-maistero-2 in living (Fig. S8D) and fixed (data not shown) cells was not efficient. However, EGFP-maistero-2 exhibited clear dot-like distribution in fixed and permeabilized HeLa S3 cells (Fig. 7). EGFP-maistero-2 colocalized with the late endosome/lysosome marker, LAMP1 (Fig. 7A), but not with another late endosome/lysosome marker, CD63 (Fig. 7B). CD63 is enriched in the internal membranes of late endosomes/multivesicular bodies [60–62]. In contrast, LAMP1 is enriched in the limiting membrane of late endosomes [62, 63]. LAMP1 is proposed to be enriched in protease active late endosomes/lysosomes [64]. No colocalization was observed of EGFP-maistero-2 with other organelle markers, such as GM 130 (cis Golgi), EEA1 (early endosomes), Rab11 (recycling endosomes) and LC3 (autophagosomes) (Fig. S9). Chol egress from the late endosome/lysosome is facilitated by NPC1 in concert with NPC2 [65–67]. NPC1 has been reported to be colocalized with CD63 [68], while only partial colocalization has been reported with LAMP1 [69]. Consequently, we hypothesize that maistero-2 labels the Chol rich late endosome/lysosome population prior to NPC1 mediated Chol egress. Indeed, blocking Chol release from the late endosome/lysosome by U18666A treatment [65] increased maistero-2 colocalization with CD63 (Fig. 7C). U18666A drastically reduced background fluorescence of maistero-2. This is consistent with the previous observation that U18666A dramatically decreased plasma membrane D4 labeling [7].

Fig. 7.

Colocalization of EGFP-maistero-2 with late endosome/lysosome markers. Fixed and permeabilized HeLa cells were doubly labeled with EGFP-maistero-2 and anti-LAMP 1 antibody (A) or anti-CD63 antibody (B). C Fixed and permeabilzed HeLa cells following 24 h treatment with 5 μM U18666A as described in “Materials and methods” were double labeled with EGFP-maistero-2 and anti-CD63 antibody. Each subpanel on the far right is a higher magnification representation of the area outlined in the corresponding white box. Each graph shows the fluorescence intensities of the green and red channel along the white line of the corresponding subpanel. Representative images of at least three experiments are shown

Double labeling of fixed and permeabilized HeLa S3 cells with EGFP-maistero-2 and mCherry-D4 showed only weak mCherry-D4 labeling, which predominantly colocalized with EGFP-maistero-2 (Fig. 8A, C). After U18666A treatment, a drastic increase in mCherry-D4 labeling was observed while maintaining colocalization with EGFP-maistero-2 and filipin-labelled organelles (Fig. 8B, D). Together, these results further demonstrate the applicability of EGFP-maistero-2 and mCherry-D4 colabeling to characterize local Chol levels.

Fig. 8.

Colocalization of EGFP-maistero-2 and mCherry-D4 in fixed and permeabilized HeLa cells. A Cells were doubly labeled with EGFP-maistero-2 and mCherry-D4 as described in “Materials and methods”. B Cells following 24 h treatment with 5 μM U18666A as described in “Materials and methods” were triple labeled with EGFP-maistero-2, mCherry-D4 and filipin. C (left) Higher magnification representation of the area outlined in the white box in (A). (right) Line scan analysis of the fluorescence intensities from the green and red channel along the yellow line indicated in the left panel. D (left) Higher magnification representation of the area outlined in the white box in (B). (Right) Line scan analysis of the fluorescence intensities from the green, red and blue channel along the yellow line indicated in the left panel. Representative images of at least three experiments are shown in (A and B)

While the exact binding mechanism of CDCs, including PFO and ALO, is not well understood, the importance of a strictly conserved Thr-Leu pair located in loop 1 for Chol binding is well-established [70]. No Thr-Leu pair could be identified in all three isolated maistero proteins. On the other hand, all three maistero proteins feature a conserved CRAC (Chol recognition/interaction amino acid consensus) motif -L/V-X_1-5-Y-X_1-5-R/K- [71, 72] close to the C-terminus (V66-K73) (Fig. 1B, shown in blue). In addition, maistero-2 features a second distinct CRAC motif (V23-R29) located at the N-terminus and maistero-3 also has a second CRAC motif, VFYIR (Fig. 1B, shown in light blue). To evaluate the potential contribution of the predicted CRAC motif of maistero-2 to Chol binding, V23, Y26, R29, V66, Y69 and K73 were point mutated. While the V23A, Y26F, R29A, R29L, V66A and K73A mutants retained binding to Chol in ELISA, both Y69F and Y69W mutants lost binding (Fig. 9A). Figure S10 shows that EGFP-Y69F and -Y69W did not label CHO-K1 cells, confirming ELISA results. This highlights the importance of Y69 to facilitate Chol-dependent membrane association.

Fig. 9.

Binding of maistero-2 point mutants to Chol. Binding of the indicated His-maistero-2 mutant (1.25 μg/mL) to Chol by ELISA, disrupting the putative CRAC motif (A) and probing tryptophan residue importance for membrane association (B). WT means His-maistero-2 (wild type). C Liposome-binding assay of indicated His-maistero-2 tryptophan mutants to SUVs composed of POPC/Chol (1:1) in the same way as in Fig. 5A, B. D Binding of EGFP-maistero-2 and its tryptophan mutants to fixed and permeabilized HeLa cells. EGFP-maistero-2 labeling to MβCD-treated HeLa cells is also shown. E Binding of His-maistero-2-W52A mutant to indicated POPC/Chol mixtures by ELISA. F Data point corresponding to 1.25 μg/mL protein in (E) was redrawn. Data are the means ± SD of three experiments in (A, B, E, F), and representative images of at least three experiments are shown in (C and D)

Trp residues are known to play an important role in protein–lipid interaction [12, 36, 73]. Maistero-2 features three Trp residues, namely, W34, W46 and W52 (Fig. 1B). Point mutants W34F and W34A failed to be expressed in our E. coli expression system, suggesting the importance of W34 for folding and protein stability. Maistero-2 point mutants W46F, W46A, W52F and W52A were successfully expressed and subjected to our ELISA Chol-binding assay (Fig. 9B). Mutant W46A exhibiting the most pronounced reduction of Chol binding during ELISA assay, while W46A and W52A mutants failed to bind to POPC/Chol (1:1) liposomes (Fig. 9C). EGFP-maistero-2 labeling of late endosomes/lysosomes in fixed and permeabilized HeLa S3 cells was abolished by MβCD pretreatment, suggesting Chol-dependent labeling (Fig. 9D). W46A and W52A mutants failed to label EGFP-maistero-2 positive organelles. In contrast, W46F and W52F not only bound to POPC/Chol (1:1) liposomes, but also labelled HeLa S3 cells in a similar pattern compared to EGFP-maistero-2, albeit at reduced intensity. To gain a better understanding of the partial binding of W52A in the ELISA assay, the required Chol threshold for binding was examined (Fig. 9E, F). Indeed, the Chol threshold for W52A binding was shifted to higher Chol content of 50%. Taken together, this suggests that W34 plays a structural role, while W46 and W52 are important for Chol-dependent binding of maistero-2. Interestingly, W52A exhibited a reduced binding to 100% Chol compared to membranes containing 70–90% Chol in ELISA (Fig. 9E, F). The mushroom-derived protein nakanori has been reported to require both SM and Chol for binding [36]. Similarly, aegerolysins require both Chol and SM [12, 13] or ceramide phosphoethanolamine [15, 74, 75] for membrane binding. Considering that a single amino acid substitution altered the lipid specificity of aegerolysin [14], Fig. 9E, F indicates that His-maistero-2-W52A preferentially binds to PC and Chol containing membranes.

Sequence alignment of maistero-2 with other Chol-binding proteins, such as D4 [58], nakanori (binds SM/Chol) [36] and the aegerolysin PlyA2 (binds SM/Chol and ceramide phosphoethanolamine/Chol) [12, 15], did not reveal any significant homology or conserved folds (Fig. S11). CD spectra of His-maistero-1 and -2 indicated that the proteins feature ~ 50% β-structure (β-sheet and β-turn), minor presence of α-helical structures and the remaining secondary structure seems aperiodic (Fig. S12). This denotes a quite distinct secondary structure of maistero proteins compared to primarily (~ 85%) β-structure featuring CDC-derived probes, such as D4 and ALO [76, 77]. In contrast, aegerolysins, that bind SM and Chol [12, 13], have been reported to exhibit ~ 66% β-structure and up-to 10% α-helical structure [78], suggesting potential similarity with maistero proteins. W46F (bound to Chol) and W46A (not bound to Chol) exhibited similar CD profiles compared to maistero-2. This suggests that the observed difference in binding activity of W46F and W46A is not caused by gross structural changes. Low Chol-binding mutants W52A, Y69F and Y69W exhibited a strong increase in β-sheet structure but overall β-structure showed only a small increase compared to maistero-2. Taken together, this suggests a structural role of these amino acids.

In summary, maistero-2 specifically binds to lipid membranes containing 3-hydroxy sterols. Maistero-2 binding is particularly sensitive to the size and conformation of the A, B and D ring of sterols, but not very sensitive to modifications of the isooctyl side chain commonly found in phytosterols. No characterized protein with apparent sequence homology could be identified. The absence of the strictly conserved Thr-Leu pair present in other Chol-binding proteins, such as D4 and ALO, indicates a potentially different mode of sterol recognition. The lower threshold for Chol detection of maistero-2 compared to D4 provides an interesting avenue to assess the Chol content of subcellular compartments. In addition, EGFP-maistero-2 has been demonstrated to be a suitable tool to visualize the increase of cell surface Chol during the differentiation of Neuro 2a cells.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ALO

Anthrolysin O

BSA

Bovine serum albumin

Chol

Cholesterol

CD

Circular dichroism

CDC

Cholesterol-dependent cytolysin

CHO

Chinese hamster ovary

CRAC

Cholesterol recognition/interaction amino acid consensus

DIC

Differential interference contrast

DMEM

Dulbecco’s modified Eagle’s medium

DOPC

1,2-Dioleoyl-sn-glycero-3-phosphocholine

EGFP

Enhanced green fluorescent protein

ELISA

Enzyme-linked immunosorbent assay

ER

Endoplasmic reticulum

FCS

Fetal calf serum

LDL

Low density lipoprotein

mAb

Mouse monoclonal antibody

MβCD

Methyl-beta-cyclodextrin

MLV

Multilamellar vesicle

NBD-DOPE

1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2–1,3-benzoxadiazol-4-yl)

NBD-DPPE

1,2-Dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2–1,3-benzoxadiazol-4-yl)

NPC

Niemann–Pick disease type C

PBS

Phosphate buffered saline

PC

Phosphatidylcholine

PFA

Paraformaldehyde

PFO

Perfringolysin O

POPC

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

POPE

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine

QCM

Quartz crystal microbalance

SM

Sphingomyelin

SUV

Small unilamellar vesicle

Author contributions

Conception and design of the research: AY-H, TK. Investigation: AY-H, MM, TI, ND, MS, FF, PG. Methodology: TI, ND, FF. Supervision: TK. Writing-original draft: AY-H, PG, TK. Writing-review and editing: AY-H, MM, TI, ND, MS, FF, YS, PG, TK.

Funding

Agence Nationale pour la Recherche (ANR-19-CE16-0012–02 to T.K.). Agence Nationale de Recherche sur le Sida et les Hépatites virale (18365 to T.K.). Ligue Contre le Cancer (to T.K.). Vaincre les Maladies Lysosomales (19/LBPH/S44 to T.K.). Seed Money, Assemblée du groupement européen de coopération territoriale (GECT) Eucor (to T.K.). Japan Society for the Promotion of Science (JSPS) (18K06648 to A. Y-H). RIKEN Integrated Lipidology Program (to A. Y-H, P.G. and T.K.). RIKEN Glycolipidologue Initiative Program (to T.K. Y.S. and P.G.). Institut national de la santé et de la recherche médicale (Inserm) (to T.K.). Centre national de la recherche scientifique (CNRS) (to T.K.). Université de Strasbourg (to T.K.)

Data and materials availability

The data sets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

Authors declare that they have no competing interests.

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors agreed on the final version of the manuscript.