Cholesterol sulfate as a negative regulator of cellular cholesterol homeostasis

Abstract

Cholesterol sulfate (CS), one of the most abundant cholesterol derivatives, recently emerged as a key regulatory molecule in several physiological processes. Here, we demonstrate multiple mechanisms by which CS reduces intracellular cholesterol levels. CS promotes the proteasomal degradation of 3-hydroxy-3-methylglutaryl-CoA reductase reductase by enhancing insulin-induced gene-mediated ubiquitination, thereby inhibiting cholesterol synthesis. In addition, CS blocks low-density lipoprotein receptor endocytosis, reducing low-density lipoprotein cholesterol uptake. CS further suppresses the proteolytic activation of sterol regulatory element-binding protein 2, a master transcription factor governing cholesterol synthesis and uptake. Using in vitro and in vivo models, we show that CS lowers cholesterol by targeting both the cholesterol synthesis and uptake pathways, while also modulating an important feedback loop via sterol regulatory element-binding protein 2. These findings highlight the potential of CS as a modulator of cholesterol metabolism, offering new therapeutic insights into cholesterol-related disorders.

INTRODUCTION

Although cholesterol sulfate (CS) is a lesser-known cholesterol derivative, it is the most abundant steroid sulfate in human plasma (Fig. 1A) (Strott and Higashi, 2003). Structurally, it arises through the addition of a sulfate group (−SO₄^2-) to the hydroxyl group of cholesterol's sterol ring, a reaction catalyzed by the enzyme cholesterol sulfotransferase SULT2B1b (Strott and Higashi, 2003). Conversely, CS can then revert to cholesterol via a hydrolysis reaction catalyzed by the enzyme steroid sulfatase (STS). SULT2B1b is predominantly expressed in the gastrointestinal tract and epidermis, whereas STS is broadly expressed, with particularly high levels in the liver (Fig. S1). Thus, despite its widespread distribution across various tissues, local CS concentrations can vary significantly depending on the levels and activity of both SULT2B1b and STS.

Fig. 1.

Cholesterol sulfate (CS) lowers intracellular cholesterol levels. (A) Chemical structure of CS. (B) Intracellular cholesterol levels in 3 different cell lines after various incubation times with CS. (C) Filipin staining to visualize intracellular cholesterol levels in Huh-7 cells after a 24-h incubation with 50 μM CS in full media or with 10 μM lovastatin in dulbecco’s modified eagle medium (DPEM) supplemented with 5% LDS and 50 μM mevalonate as the cholesterol depletion (Chol Dep) condition. Scale bars, 20 μM. (D) Quantification of the filipin staining signal. (E) Effect of CS on de novo cholesterol synthesis in the indicated cell types. (F) Effect of CS on cholesterol uptake from the medium in the indicated cell types. Asterisks indicate statistical significance: *P < .05, **P < .01, and ***P < .001.

Initially regarded as a metabolite and circulating reservoir for its unsulfated counterpart, CS itself was recently found to act as a critical regulatory molecule. The most dramatic role for CS was observed in the epidermis, where it regulates keratinocyte differentiation, barrier function, and desquamation (Elias et al., 2014). Notably, elevated levels of CS in the skin due to loss-of-function mutations in the STS gene are associated with X-linked ichthyosis (XLI, OMIM 308100) (Hernandez-Martin et al., 1999), while reduced levels of CS due to mutations in SULT2B1 are associated with autosomal-recessive congenital ichthyosis (OMIM 242300) (Heinz et al., 2017). These human genetic disorders are characterized by dry, scaly, and rough skin (Hernandez-Martin et al., 1999) and underscore the importance of CS in maintaining the skin’s barrier function. CS has also been implicated in other physiological processes, including sperm capacitation (Langlais et al., 1981), blood clotting (Blache et al., 1995), and immune responses, particularly as a regulator of T-cell receptor signaling (Wang et al., 2016) and Dedicator of cytokinesis protein 2 (Dock2) (Sakurai et al., 2018).

Cholesterol homeostasis is essential for proper cellular function, membrane integrity, and the modulation of various signaling pathways (Luo et al., 2020). It is tightly regulated by networks of feedback loops at the transcriptional, translational, and post-translational levels. Sterol regulatory element-binding protein 2 (SREBP-2) is a major player in cholesterol homeostasis. When cellular cholesterol levels are low, SREBP2 moves from the endoplasmic reticulum (ER) to the Golgi, where it undergoes proteolytic activation and functions as a transcription factor, orchestrating the expression of genes crucial for both cholesterol synthesis and uptake. 3-Hydroxy-3-methylglutaryl-CoA reductase (HMGCR), the rate-limiting enzyme for cholesterol synthesis, is tightly governed by transcriptional, translational, and post-translational feedback mechanisms, including accelerated ER-associated degradation (ERAD). Cellular uptake of cholesterol relies predominantly on receptor-mediated endocytosis pathways involving the low-density lipoprotein (LDL) receptor. LDL-cholesterol (LDL-C) uptake is also regulated by multiple mechanisms at the transcriptional and post-translational levels.

Many cholesterol derivatives naturally contribute to cholesterol homeostasis by influencing cholesterol synthesis, uptake, and storage (Mutemberezi et al., 2016, Radhakrishnan et al., 2007). For instance, 25-hydroxycholesterol (25-HC) is a critical feedback regulator in cholesterol homeostasis that inhibits cholesterol synthesis by promoting the degradation of HMGCR and suppressing SREBP2 processing. Similarly, CS regulates intracellular cholesterol homeostasis (Williams et al., 1985, Williams et al., 1987). Notably, CS inhibits HMGCR activity in fibroblasts, although the underlying mechanism remains elusive.

Here, we leverage in vitro and in vivo models to clarify the role of CS in regulating cholesterol homeostasis. We show that CS reduces intracellular cholesterol levels by promoting HMGCR degradation and reducing LDL-C uptake. CS also disrupts SREBP2 processing, limiting its ability to activate genes encoding cholesterol-related enzymes. Our findings reveal previously unknown mechanisms of CS-mediated cellular cholesterol regulation and suggest potential therapeutic approaches for cholesterol-related pathologies.

MATERIALS AND METHODS

Media, Chemicals, and Antibodies

Dulbecco’s modified eagle medium (DMEM) (#11965092), fetal bovine serum (FBS) (#26-140-079), penicillin streptomycin (#15140122), and dulbecco’s phosphate-buffered saline (DPBS) (#14190250) were purchased from Thermo Scientific. CS (#C9523), Cholesterol (#C8667), Methyl-beta-cyclodextrin (#C4555), Filipin III (#F4767), Lovastatin (#PHR1285), Pravastatin (#P4498), Atorvastatin (#PHR1422), MG132 (#M7449), Potassium bromide (#60089), Polybrene (#TR-1003), Puromycin (#P9620), dimethyl sulfoxide (DMSO) (#472301), methanol (#8222831000), and Mevalolactone (#M4667) were purchased from Sigma-Aldrich. 25-HC (#11097) was purchased from Cayman Chemical Company. Anti-T7 tag (#D9E1X) and anti-LAMP1 (#D2D11) were purchased from Cell Signaling Technology. Anti-α-tubulin (#12G10) was purchased from the Developmental Studies Hybridoma Bank. Anti-HA (#H9658), Anti-FLAG M2 horseradish peroxidase (#A8592), and Anti-FLAG M2 Magnetic Beads (#M8823) were purchased from Sigma-Aldrich. Anti-SREBP2 (#ab30682) and Anti-LDL receptor (LDLR) (#ab52818, MAB2148) were purchased from Abcam and R&D Systems. Anti-HMGCR A9 (#CRL-1811) was purchased from the American Type Culture Collection. Anti-ATP1A1 (#sc-28800) and Anti-Ubiquitin (#P4D1, sc-8017) were purchased from Santa Cruz. CS and 25-HC were dissolved in a 1:1 mixture of DMSO and methanol to prepare a 25 mM stock solution. This stock solution was then further diluted in cell culture media to obtain the desired working concentrations.

Cell Culture

Huh-7, HEK 293T, and MEF cells were cultured in a complete growth medium (DMEM, 10% FBS, and 1% penicillin streptomycin). When indicated, Huh-7 cells were incubated with DMEM supplemented with 5% FBS or 5% lipid-depleted serum (LDS).

Generation of Stable Huh-7 Cell Lines

pCMV-HMG-Red-T7 (MBA-84, American Type Culture Collection) was subcloned into EcoRI and BamHI sites in the pLVX-EF1α-IRES-Puro (EIP) lentiviral vector. K89R (AAG to CGT) and K248R (AAA to CGT) mutations were introduced to generate the mutant HMGCR vector. For lentivirus preparation, transfer plasmid (HMGCR-T7), psPAX2, and pMD2.G vectors were transfected at a 3:4.5:2.5 mass ratio into Lenti-X 293T cells using X-tremeGENE HP DNA Transfection Reagent (XTGHP-RO, Roche). Three days after transfection, the supernatants were collected, filtered, and concentrated using Amicon Ultra-15 Centrifugal Filters Ultracel-100K (UFC910024, Merck-Millipore). Then, Huh-7 cells were infected with the concentrated supernatant supplemented with 8 μg/ml polybrene in a complete growth medium. Three days after infection, puromycin selection (0.5-2 μg/ml) was started. Subsequent analyses using stable cell lines were performed after at least 2 weeks of puromycin selection.

LDS Preparation

LDS was generated as previously described (Luu et al., 2017). Briefly, heat-inactivated serum was adjusted to the desired density with potassium bromide and transferred to ultracentrifuge tubes (Beckman Coulter) using a 14 G needle. The tubes were capped, sealed, and subjected to ultracentrifugation (24 h, 250,000 × g, 10°C). Following centrifugation, the LDS fraction was collected from the bottom layer and dialyzed against multiple changes of 150 mM sodium chloride solution.

Cellular Cholesterol Quantification

Cells were seeded at 10⁶ cells/plate in 6-cm dishes and incubated for 24 h in a complete growth medium. The cells were then treated with either CS or vehicle control. Following treatment and harvesting in radioimmunoprecipitation assay buffer, the cell lysates were subjected to cholesterol quantification using the Amplex Red Cholesterol Assay Kit (Invitrogen) according to the manufacturer’s instructions. Briefly, 5 μl of diluted lysate (1:10 dilution) was added to each well of a 96-well plate. This was followed by the addition of the assay reagents. Fluorescence intensity, proportional to cellular cholesterol content, was measured using the Varioskan LUX Multimode Microplate Reader (Thermo Scientific). The protein concentration of each lysate was determined using the BCA Protein Assay Kit (#23227, Thermo Scientific) for use in normalization.

Filipin Staining and Quantification

Cells were seeded at 5 × 10⁴ cells/well on poly-D-lysine-coated cover glasses and incubated for 24 h in a complete growth medium. Treated groups were exposed to either CS or vehicle control for an additional 24 h. For cholesterol depletion, the cells were treated for 24 h with 5% LDS-supplemented DMEM containing 10 µM lovastatin and 50 µM mevalonolactone. All cell groups were then fixed with 4% paraformaldehyde for 30 min at room temperature. Subsequently, the cells were incubated with a 50 µg/ml Filipin solution in phosphate-buffered saline (PBS) for 1 h. Fluorescence intensity, reflecting intracellular cholesterol accumulation, was captured using a Zeiss LSM 700 (Carl Zeiss) confocal microscope and quantified using ImageJ software (National Institutes of Health [NIH]).

HMG-CoA Reductase Activity Assay

HMG-CoA Reductase activity was measured using the HMG-CoA Reductase Assay Kit (CS1090, Sigma-Aldrich) according to the manufacturer’s protocol. Briefly, 1× assay buffer, pravastatin (positive control) or CS, NADPH, HMG-CoA, and HMGCR were added sequentially up to a 200 μl reaction volume at 4℃. Then, absorbance at 340 nm was read using a VersaMax (Molecular Devices) at 37℃ for 20 min.

Western Blot Analysis

Following transfection or treatment, cells were washed 3 times with 1× PBS and harvested by centrifugation. The cells were suspended by pipetting in 500 μl NP-40 lysis buffer (50 mM Tris-Cl pH 8.0, 150 mM NaCl, 1% NP-40) and immediately centrifuged at 16,000 × g for 10 min to collect the pellet. Next, the pellet was resuspended in 500 μl NP-40 lysis buffer supplemented with a protease inhibitor cocktail (cOmplete, #1183617001, Roche) using an IKA-T10 homogenizer (IKA-Werke GmbH & Co), rotated for 1 h, and centrifuged at 16,000 × g for 30 min. Lysate protein concentrations were measured with the BCA assay kit. Equal amounts (30 μg) of lysate were separated via sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and transferred to PVDF membranes (Merck-Millipore). The membranes were blocked for an hour in 5% skim milk in 1× phosphate-buffered saline with tween 20 (PBST) and then incubated overnight at 4°C with primary antibodies diluted in 1× PBS. After 3 30-min washes with 1× PBST, the membranes were hybridized with a secondary antibody labeled with horseradish peroxidase. Finally, after 3 30-min washes with 1× PBST, the protein bands were visualized using an enhanced chemiluminescence (#34580, Thermo Scientific) detection system and CP-BU NEW x-ray film (AGFA Healthcare).

RNA Isolation and Real-Time PCR

Total RNA was extracted from cells using the TRIzol reagent (Invitrogen) according to the manufacturer's protocol. About 1 μg of purified RNA was then reverse transcribed into cDNA using the RevertAid First Strand cDNA Synthesis Kit (#EP0442, Thermo Scientific). The resulting cDNA was diluted 3-fold, and 1 μl of the resulting dilution was used for each real-time PCR reaction. Amplification of target genes was performed using specific primers (sequences are listed in Table S2) and a 2× SYBR green-based PCR master mix on a QStudio3 real-time PCR system (Thermo Scientific). The quantitative polymerase chain reaction results were normalized using the GAPDH house-keeping gene.

In Vitro Polyubiquitination Assay

Huh-7 cells stably expressing T7-tagged HMGCR were treated with vehicle or CS in the presence of 20 μM MG132 for 5 h. Then, the cells were lysed using radioimmunoprecipitation assay buffer supplemented with a protease inhibitor cocktail (cOmplete). Lysates were precleared with 40 μl of Pierce protein A/G agarose beads (#20421, Thermo Fisher Scientific) for 2 h at 4°C and then incubated with 0.18 μg of T7 antibody overnight at 4°C. The lysates were then immunoprecipitated with 50 μl of Pierce protein A/G agarose beads for 2 h at 4°C, washed 5 times with lysis buffer, and eluted with 2× Laemmli Sample Buffer (#1610747, Bio-Rad Laboratories) at 95°C for 5 min. The supernatants were immunoblotted with antibodies against Ubiquitin (1:100) or T7 (1:1,000).

Immunoprecipitation

HEK 293T cells were seeded at 2 × 10⁶ cells per well onto 6-cm plates. After 24 h, the cells were transfected with 0.1 μg of pcDNA3.1-FLAG-human INSIG1 (insulin-induced gene 1) and 1 μg of pcDNA3.1-HA-human HMGCR or pcDNA3.1-HA-human SCAP (SREBP cleavage-activating protein; wild-type, L343A, V355A, or I348F) using X-TREMEGENE HP DNA transfection reagent in complete DMEM supplemented with 10% FBS. Six hours post transfection, the cells were transferred to DMEM containing 5% LDS, 10 μM lovastatin, and 50 μM mevalonate. After 16 h, the cells were treated with DMEM (5% LDS), 10 μM lovastatin, and 50 μM mevalonate for an additional 8 h in the absence or presence of either 10 μM 25-HC or 25 μM CS. The cells were then washed with DPBS and harvested using a cell scraper. The cells were suspended by pipetting in 500 μl NP-40 lysis buffer and immediately centrifuged at 16,000 × g for 10 min to collect the pellet. Next, the pellet was resuspended in 500 μl NP-40 lysis buffer supplemented with a protease inhibitor cocktail (cOmplete) using an IKA-T10 homogenizer (Germany), rotated for 1 h, and centrifuged at 16,000 × g for 30 min. Then, 2% of the resulting supernatant was used as input for SDS-gel electrophoresis. The remaining supernatant was incubated overnight at 4°C with anti-FLAG magnetic beads. These beads were washed 3 times with lysis buffer and eluted with 1× Laemmli buffer.

Animal Care and Use

Eight-week-old C57BL/6 mice were administered daily oral doses of vehicle (N = 4), CS (50 mg/kg, N = 4), atorvastatin (20 mg/kg, N = 4), or CS + atorvastatin (CS, 50 mg/kg; atovastatin, 20 mg/kg; N = 4). These doses of CS or atorvastatin were dissolved in mixtures of PBS, DMSO, and Cremophor (4:4:2) and delivered at a volume of 200 μl per mouse. After 10 days of daily treatments, the mice were sacrificed and their livers and blood were collected. All mice were maintained under standard animal housing conditions with 12-h light-dark cycles and free access to food and water. All animal experiments were approved by the Animal Care Committee of Yonsei University College of Medicine (#2023-0127).

Mouse Liver Fractionation for HMGCR Western Blots

Buffer A1 was prepared by combining 10 mM Hepes-potassium hydroxide (pH 7.6), 1.5 mM MgCl₂, 10 mM KCl, 5 mM ethylenediaminetetraacetic acid, 5 mM ethylene glycol tetraacetic acid, and 250 mM sucrose, and then adjusting the pH to 7.6 with 10 N potassium hydroxide. HMGCR solubilization buffer was prepared by mixing 62.5 mM Tris-HCl (pH 6.8), 15% SDS, 8 M urea, 10% glycerol, and 100 mM DTT. SDS lysis buffer was prepared by combining 10 mM Tris-HCl (pH 6.8), 100 mM NaCl, 1% SDS, 1 mM ethylenediaminetetraacetic acid, and 1 mM ethylene glycol tetraacetic acid. Prior to liver collection, Buffer A1 with freshly added protease inhibitors was prechilled on ice. Liver samples (40-50 mg) were placed in tubes containing Buffer A1 and homogenized using a TissueLyser II (Qiagen). The samples were then centrifuged at 4,000 rpm for 5 min at 4°C, and the resulting postnuclear supernatant (PNS) was transferred to a prechilled ultracentrifuge tube. The PNS was centrifuged at 100,000 × g for 30 to 60 min at 4°C. The supernatant (cytosolic fraction) was aspirated, and the membrane pellet was resuspended in 100 to 150 μl of SDS lysis buffer. The mixture was shaken for 15 to 20 min or until fully dissolved. A 2 μl sample was taken for a BCA assay to measure protein concentration. An equal volume of HMGCR solubilization buffer and the appropriate amount of 4X sample buffer were added to the membrane samples. The samples were then incubated in a 37°C water bath for 20 to 30 min before being loaded onto an SDS-PAGE gel.

Serum Biochemistry

Mouse serum was prepared from blood by centrifugation at 2,000 rcf for 20 min and stored at −80℃ until analysis. Serum lipid profiles were analyzed with a Cobas c502 module (Roche Diagnostics).

DiI-LDL Uptake Assay

Cells were seeded at 5 × 10⁴ cells/well onto poly-D-lysine-coated cover glasses and incubated for 24 h. The treatment groups were then exposed to either vehicle or CS at the desired concentration for either 4 or 24 h. CS + Chol samples were incubated with CS for 22 h and then 5 µM Chol/methyl-beta-cyclodextrin (1:7) was added and incubated for an additional hour. After removal, 5% LDS was added. LDL conjugated with the fluorescent probe DiI (L3482, Invitrogen) was added to all groups to reach a final concentration of 25 µg/ml and then incubated for 2 h at 37°C. The cells were then washed 3 times in a buffer containing 50 mM Tris-HCl (pH 7.4), 0.15 M NaCl, and 2 mg/ml of BSA, before being fixed again with 4% paraformaldehyde. DiI fluorescence intensity, reflecting LDL uptake, was captured using a Zeiss LSM 700 confocal microscope (Carl Zeiss) or a Flow cytometer (BD FACS Aria Fusion, BD Biosciences) and quantified using ImageJ software (NIH) or Flowjo software (FlowJo LLC), respectively.

Immunofluorescence Staining

Cells were seeded at 5 × 10⁴ cells/well on poly-D-lysine-coated coverslips and incubated for 24 h. The treatment groups were then exposed to either vehicle or CS at the desired concentration for either 4 or 24 h. After washing with DPBS, the cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton-X, and incubated with primary antibodies against LDLR (1:100) and LAMP1 (1:200) at 4°C overnight. This was followed by another 1 h of incubation with fluorescence probe-conjugated second antibodies at room temperature. For measuring surface staining, the cells were incubated with primary antibodies after fixation without permeabilization. The slides were then washed with PBS and mounted using VECTASHIELD Antifade Mounting Medium (#H-1000-10, Vector Laboratories). Fluorescent images were acquired using a Zeiss LSM 700 confocal microscope (Carl Zeiss).

RNA-seq and Analysis

Raw sequence reads produced by the sequencer were cleaned using Trim Galore version 0.6.5. The cleaned reads were aligned to the mouse reference genome (GRCm39) using STAR aligner v2.7.11a (Dobin et al., 2013). Quantification of gene expression was performed using the featureCounts function of the Rsubread R package v2.18.0 (Liao et al., 2013). Differentially expressed genes were identified using the DESeq2 R package v1.44.0 according to their adjusted P-values (Benjamini-Hochberg procedure). Genes were considered significant at P < .05 and | fold change | > 2. Significant differentially expressed genes were analyzed for gene ontology (GO) enrichment with the clusterProfiler R package v4.12.0 (Wu et al., 2021). Bubble plots and enrichment plots were generated using the ggplot2 R package v3.5.1. The ComplexHeatmap R package v2.20.0 (Gu, 2022) was used to draw heatmaps for leading-edge genes. All statistical analyses and visualizations were performed under R v4.4.0 in the RStudio environment.

In Silico Docking

The cryo-EM structures of INSIG and SCAP were published on the Protein Data Bank RSCB under PDB IDs 6M49 and 7ETW. To fill the residues missing from SCAP, the 3D structure of the INSIG/SCAP complex was created with the SWISS-MODEL server and its templates. The co-ligand was removed and the 3D structure of the complex was exported into a dockable pdbqt for molecular docking using the Auto Dock Tool (version 1.5.6). The 3D structure of 25-HC was extracted from the 3D structure of the complex (PDB ID: 6M49), and the 3D structure of CS was downloaded from the Pubchem Database. Compound geometry was optimized using Avogadro 1.2.0. After that, the ligands were converted to the dockable pdbqt format using Open Babel 3.1.1. Molecular docking was carried out using AutoDock Vina (version 1.2). After setting the grid box, the sterol-binding site was set in the middle of INSIG and SCAP. The molecular interaction between protein complex and ligands was visualized using BIOVIA Discovery Studio Visualizer 2020.

Molecular Dynamic Simulation

The complex of 25-HC with INSIG/SCAP was selected and then the Cryo-EM structure was accessed from the RCSB (PDB ID: 6M49). The missing residues were filled using SWISS-MODEL. The conformation of CS with the strongest docking score was selected for molecular dynamic simulation. The molecular dynamic simulation protocol was prepared as previously described (Cheng et al., 2022). The system was set up using the CHARMM-GUI membrane bilayer builder server (Jo et al., 2008). The protein complex was placed in a bilayer of palmitoyloleoyol-phosphatidylcholine (POPC). The complex was neutralized by adding Cl^- ions. The CHARMM36m forcefield was used for all simulations and the TI3P model was used for water. The simulations were performed at 310 K and a pressure of 1 atm. Finally, a dynamic simulation of the complex was performed for 200 ns. All simulations were carried out using GROMACS 2020. GROMACS was also used for the analysis of the molecular dynamic simulation trajectories. The root mean square deviations for the atomic position of the selected ligands and for the proteins were calculated by fitting protein backbone atoms with the gmx_rms subprogram. The VMD molecular graphics 1.9.3 program was used for contact frequency analysis with a cutoff distance of 4 Å.

Statistics

All data are presented as means ± standard error of the mean. For comparing 2 groups of data, unpaired Student’s t-tests and Mann-Whitney U-tests were used. For multiple comparisons, 1-way analysis of variances with Tukey post-hoc tests were used. The details of these statistical analyses are shown in Table S3.

RESULTS

CS Reduces Cellular Cholesterol Levels

To assess the impact of CS on cellular cholesterol homeostasis, we investigated its effect on steady-state cholesterol levels across various cell types at different time points. We opted for a concentration of 25 µM CS because it showed no discernible toxicity in any of several tested cell lines (Fig. S2). Exposure to CS for approximately 24 to 48 h significantly reduced cellular cholesterol levels by 20% to 30% across all tested cell lines (Fig. 1B). Given this consistency, we selected Huh-7 human liver cells for further investigation. Next, we used the cholesterol indicator filipin to validate the effect of CS on Huh-7 cholesterol levels. Treatment of Huh-7 cells with CS for 24 h significantly reduced filipin staining to levels commensurate with their levels of cholesterol depletion (Fig. 1C and D).

Next, we sought to identify the aspects of cholesterol regulation influenced by CS. Initially, we assessed the effect of CS on cholesterol synthesis. First, we depleted cellular cholesterol in Huh-7 cells by incubating them in LDS to prevent cholesterol uptake and inhibiting their synthesis of cholesterol with statin treatment. Statins are pharmacological inhibitors of HMGCR. We then removed the statin to allow the cells to resume de novo cholesterol synthesis while still being maintained in LDS for an additional 12 h (Fig. 1E). Although the control cells showed the expected increase in cholesterol via de novo synthesis, those that received 12 h of CS treatment showed cholesterol levels similar to those receiving an additional 12 h of statin treatment (Fig. 1E). This suggests CS inhibits de novo cholesterol synthesis. Next, we examined the effect of CS on cholesterol uptake. After again depleting cellular cholesterol levels via incubation with LDS and a statin, we switched the culture media to one containing FBS and the statin. This allowed the cells to take up LDL-C from their extracellular environment (Fig. 1F). Interestingly, we found the addition of CS partially blocked the cholesterol uptake permitted by the switch to FBS-containing media (Fig. 1F). Thus, our results suggest CS inhibits both de novo cholesterol synthesis and cholesterol uptake.

To address the possibility that CS affects cellular cholesterol levels by also promoting cholesterol export, we evaluated the transcript levels of the major cholesterol export transporters ABCA1 and ABCG8 (Fitzgerald et al., 2010, Graf et al., 2003). Neither removing the statin to allow cells grown in LDS to begin de novo cholesterol synthesis nor switching cells from LDS to FBS to enable cholesterol uptake affected ABCA1 transcript levels, but both of these treatments increased ABCG8 transcript levels (Fig. S3). Next, we found that the addition of CS did not trigger any additional effect on ABCA1 transcript levels. Although CS did not affect ABCG8 transcript levels in cells after statin removal (cholesterol synthesis condition), it blocked the increase in ABCG8 transcript levels associated with the switch back to FBS (cholesterol uptake condition). These data make it doubtful that cholesterol export contributes to the lower cellular cholesterol levels observed following CS treatment (Fig. S3). Instead, our results support a role for CS in modulating intracellular cholesterol levels by inhibiting both de novo cholesterol synthesis and uptake.

CS Promotes Proteasomal Degradation of HMGCR

While CS inhibits HMGCR enzyme activity in fibroblast cell lines, its mechanism remains unclear (Williams et al., 1985). To determine whether CS directly inhibits HMGCR activity, we performed HMGCR enzyme assays using the purified catalytic domain of HMGCR. In contrast to pravastatin, which completely inhibits HMGCR activity, CS had no effect (Fig. 2A). We next investigated whether CS regulates HMGCR activity at the transcriptional or post-translational levels. When we treated Huh-7 cells with CS, we were unable to see any discernible change in HMGCR transcript levels (Fig. S4). CS treatment did, however, reduce endogenous HMGCR protein levels in a dose-dependent manner akin to 25-HC, albeit with diminished potency (Fig. 2B and C) (Elsabrouty et al., 2013, Jo et al., 2011). We confirmed that CS reduces HMGCR protein in Huh-7 cells stably expressing T7-tagged exogenous HMGCR with an estimated IC₅₀ of 5.6 µM (Fig. 2D and E). Thus, CS reduces HMGCR protein levels at the post-translational level.

Fig. 2.

Cholesterol sulfate (CS) reduces HMGCR via ubiquitin-dependent proteasomal degradation. (A) HMGCR activity assay using the purified HMGCR catalytic domain. (B) Western blot analysis of endogenous HMGCR in Huh-7 cells after a 16-h incubation with 25-HC or CS in a complete growth medium. (C) Quantification of HMGCR protein from the blot shown in (B). (D) Western blot analysis of HMGCR in Huh-7 cells stably expressing EF1α-driven T7-tagged HMGCR after a 5-h treatment with CS or 25-HC. (E) Dose-response curve showing HMGCR protein levels resulting from the various concentrations of CS shown in (D). (F) Western blot analysis of the ubiquitination of HMGCR in Huh-7 cells after a 5-h treatment with CS in the presence of 20 μM MG132. Lysates from Huh-7 cells stably expressing EF1a-driven T7-tagged HMGCR were immunoprecipitated with anti-T7 antibody and the pellets were probed with anti-ubiquitin and anti-T7 antibodies. (G) Western blot analysis of HMGCR in Huh-7 cells stably expressing EF1α-driven T7-tagged HMGCR after a 5-h treatment with CS or 25-HC in the presence or absence of 20 μM MG132. (H) Western blot analysis of HMGCR protein in Huh-7 cells stably expressing EF1a-driven T7-tagged wild-type and K89R/K248R mutant HMGCR after a 5-h CS treatment. (I) Quantification of HMGCR protein in the blot shown in (H). (J) A cartoon illustrating the interaction between INSIG1 and HMGCR. The figure shows how sterols enhance the interaction between INSIG1 and HMGCR. (K) Immunoprecipitation experiments in HEK 293T cells to confirm CS-promoted INSIG1-HMGCR interaction. (L) Quantification of the amount of HMGCR co-immunoprecipitated with INSIG1 and shown on the blot in K. Asterisks indicate statistical significance: *P < .05, **P < .01, and ***P < .001.

Considering the tight regulation of HMGCR by cellular sterols such as 25-HC via the INSIG-mediated ubiquitin-proteasome degradation pathway (Sever et al., 2003), we examined the effect of CS on HMGCR protein levels in the presence and absence of the proteasome inhibitor MG132. Remarkably, cotreatment with MG132 abolished CS-induced HMGCR reduction (Fig. 2F). Furthermore, CS also increased the polyubiquitination of HMGCR (Fig. 2G). INSIG-associated ubiquitin ligases mediate HMGCR ubiquitination on lysines 89 and 248, and mutation of these 2 residues (K89R and K248R) prevents sterol-induced ubiquitination and ERAD of HMGCR (Goldstein et al., 2006). We found that CS treatment reduced the protein levels of wild-type HMGCR but not K89R/K248R double mutant HMGCR (Fig. 2H and I).

The sterol-induced interaction between HMGCR and INSIGs culminates in the ubiquitination and subsequent proteasomal degradation of HMGCR (Sever et al., 2003). We next asked whether CS instigates the interaction between INSIG1 and HMGCR (Fig. 2J). We found that like 25-HC, CS induces the interaction between INSIG1 and HMGCR, but with less potency (Fig. 2K and L). Together, these results suggest CS suppresses HMGCR activity by accelerating INSIG-mediated polyubiquitination and, therefore, proteasomal degradation of HMGCR.

CS Prevents Statin-Induced Accumulation of HMGCR

Statins are among the most frequently prescribed drugs worldwide, but compensatory increases in HMGCR protein due to negative feedback limit their therapeutic efficacy (Pineda and Cubeddu, 2011, Preiss et al., 2011). To validate our findings regarding the promotion of HMGCR degradation, we further investigated whether CS can mitigate statin-induced HMGCR accumulation both in cultured cells and in vivo. In Huh-7 cells, statin treatment led to an almost 2-fold increase in HMGCR protein levels (Fig. 3A and B). Cotreatment with CS, however, effectively abolished the statin-induced accumulation of HMGCR protein, reducing it even beyond control levels in a dose-dependent manner. Next, we treated mice with vehicle, CS alone, atorvastatin alone, or a combination of atorvastatin and CS for 10 days, and then assessed hepatic HMGCR protein levels via immunoblotting. We found that HMGCR protein levels in the livers of mice treated with CS alone were not significantly different from vehicle controls. In contrast, while atorvastatin-treated mice showed 13-fold higher hepatic HMGCR protein levels compared to those of vehicle-treated mice, the addition of CS reduced this increase by 50% (7.5-fold increase compared to vehicle; Fig. 3C and D). All 4 groups of mice exhibited similar plasma cholesterol and triglyceride profiles (Fig. 3E).

Fig. 3.

Cholesterol sulfate (CS) inhibits statin-induced HMGCR accumulation. (A) Western blot analysis of HMGCR in Huh-7 cells stably expressing EF1α-driven T7-tagged HMGCR. Cells were pretreated for 16 h with dimethyl sulfoxide (DMSO) or 10 μM lovastatin in media supplemented with 5% delipidated serum and 50 μM mevalonate. Then, the cells were incubated for an additional 5 h in the presence or absence of CS. (B) Quantification of HMGCR protein from the blot shown in (A). (C) Western blot analysis of liver HMGCR after oral administration of the indicated drugs (CS, 50 mg/kg; atorvastatin, 20 mg/kg) daily for 10 days. (D) Quantification of HMGCR protein from the blot shown in (C). (E) Plasma lipid profiles of mice after oral administration of the indicated drugs (CS, 50 mg/kg; atorvastatin, 20 mg/kg) daily for 10 days. Total-C, total cholesterol; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol. Asterisks indicate statistical significance: *P < .05, **P < .01, and ***P < .001.

CS Prevents LDL-Cholesterol Uptake

Next, we investigated the effects of CS on cholesterol uptake in Huh-7 cells. In peripheral cells, cholesterol is primarily taken up via LDLR-mediated endocytosis of LDL-C (Luo et al., 2020). Pretreatment with CS for 24 h significantly reduced fluorescently conjugated LDL uptake, as analyzed by both fluorescent microscopy (Fig. 4A-C) and flow cytometry (Fig. 4D and E). Since cholesterol depletion impairs LDL uptake (Rodal et al., 1999, Subtil et al., 1999), we wondered whether the decrease in LDL uptake we observed was due to CS-induced cholesterol reduction (Figs. 1B and S5A). When we supplemented Huh-7 cells with exogenous cholesterol (Fig. 4A), we found that although cholesterol supplementation restored intracellular cholesterol levels (Fig. S5A), it permitted only a partial rescue of LDL uptake (Fig. 4D and E). This indicates that CS reduces LDL uptake both through cholesterol depletion and additional mechanisms.

Fig. 4.

Cholesterol sulfate (CS) prevents LDL cholesterol uptake. (A) Schematics showing the experimental protocols for examining the effects of a 24-h CS treatment on DiI-LDL uptake in Huh-7 cells. (B) Representative fluorescence images of DiI-LDL uptake in Huh-7 cells incubated with CS for 24 h. (C) Quantification of the fluorescence intensity of DiI-LDL as shown in (B). (D) Flow cytometry of Huh-7 cells treated with DiI-LDL according to the protocol shown in (A). (E) Quantitative analysis of DiI-LDL fluorescence intensity in Huh-7 cells as shown in (D). (F) Schematics showing the experimental protocols for examining the effects of a 4-h CS treatment on DiI-LDL uptake in Huh-7 cells. (G) Representative fluorescence images of DiI-LDL uptake in Huh-7 cells incubated with CS for 4 h. (H) Quantification of the fluorescence intensity of DiI-LDL as shown in (G). (I) Flow cytometry of Huh-7 cells treated with DiI-LDL according to the protocol shown in (F). (J) Quantitative analysis of DiI-LDL fluorescence intensity in Huh-7 cells as shown in (I). (K) Representative fluorescence images of Huh-7 cells incubated with CS for either 4 or 24 h and then probed for total LDLR and LAMP1. (L) Quantification of the LDLR fluorescent signal as shown in (K). (M) Representative fluorescence images of Huh-7 cells incubated with CS for either 4 or 24 h and then probed for cell surface LDLR. (N) Quantification of cell surface LDLR fluorescent signal as shown in (M). Scale bar, 20 µm. Asterisks indicate statistical significance: *P < .05, **P < .01, and ***P < .001.

Since the structure of sterols influences clathrin-mediated endocytosis (CME), the primary pathway for LDL uptake (Anderson et al., 2021), we hypothesized that CS may directly inhibit LDL uptake by affecting CME. We found that a short 4-h CS treatment that did not alter cellular cholesterol levels (Fig. S5B) still significantly reduced LDL uptake in Huh-7 cells (Fig. 4F-J). Cholesterol supplementation did not rescue this effect, indicating that, in addition to its secondary effect on CME via cholesterol depletion, CS directly interferes with CME-mediated LDL uptake.

Next, we investigated the effect of CS on CME by monitoring LDLR endocytosis. CS did not alter total LDLR expression, as confirmed by immunostaining of LDLR in permeabilized Huh-7 cells and by western blot analysis (Figs. 4K and L and S5C). Via immunostaining of nonpermeabilized Huh-7 cells, however, we found increased plasma membrane LDLR levels in CS-treated cells compared to controls (Fig. 4M and N). This result further supports the hypothesis that the reduced LDL uptake we observed following CS treatment is due to the inhibition of LDLR CME.

CS Inhibits Statin-Induced Expression of SREBP2 Target Genes in the Liver

To explore the effects of CS on cholesterol homeostasis at the transcriptional level, we conducted a gene set enrichment analysis. This analysis compared the liver transcriptomes of mice treated orally with vehicle, CS alone, atorvastatin alone, or a combination of atorvastatin and CS for 10 days (Table S1). We found that the livers of statin-treated mice showed significant enrichment in cholesterol and sterol biosynthesis genes, reflecting the well-established ability of statins to enhance the activity of SREBP2, a master transcription factor regulating cholesterol homeostasis (Fig. 5A-D). Although we did not see any significant changes between the vehicle and CS-treated groups, the combination of atorvastatin and CS significantly suppressed the statin-induced upregulation of cholesterol and sterol biosynthesis genes (Fig. 5A-D). This finding suggests CS counteracts the statin-induced upregulation of SREBP2 target genes (Fig. 5A-D).

Fig. 5.

Cholesterol sulfate (CS) inhibits SREBP2 processing. (A) Bubble plot depicting significantly altered gene ontology (GO) terms from a gene set enrichment analysis (GSEA) of differentially expressed genes in the liver in response to CS, atorvastatin, or the combination of CS and atorvastatin compared to vehicle control. Red and gray dots indicate GO terms significantly or nonsignificantly enriched, respectively. The size of the dot indicates the number of genes contributing to that term. The blue horizontal dashed line indicates the significance threshold for the enrichment results (adjusted P-value < .05). A maximum of 5 of the most significantly enriched terms are labeled for each treatment. (B and C) Enrichment plots for 2 GO terms: sterol biosynthetic process (GO:0016126, B) and cholesterol biosynthetic process (GO:0006695, C) are shown (green, red, and blue lines indicate CS, atorvastatin, or CS + atorvastatin, respectively). (D) Heatmap visualizing the expression pattern of leading-edge genes contributing to the enrichment of the sterol biosynthetic process and cholesterol biosynthetic process terms. (E) The relative messenger ribonucleic acid (mRNA) expression levels of SREBP2-regulated cholesterol biosynthesis genes. Above are the schematics showing the experimental procedures. The doses used were 10 μM for lovastatin, 10 μM for 25-HC, and 25 μM for CS. (F) SREBP2 proteolytic processing in Huh-7 cells. Huh-7 cells were treated as shown in E. Immunoblots were incubated with anti-SREBP2 antibodies. Below is a quantification of nuclear SREBP2 protein (active form) based on the above western blot results made using ImageJ. Asterisks indicate statistical significance: **P < .01 and ***P < .001.

CS Inhibits the Processing of SREBP2

Next, we wanted to clarify the mechanism by which CS suppresses the statin-induced upregulation of SREBP2 target genes in the liver. SREBP2 activity is intricately regulated by intracellular cholesterol levels, particularly within the ER membrane (Goldstein et al., 2006). Decreasing intracellular cholesterol levels trigger the translocation of SREPB2 precursor (fSREBP2) proteins from the ER to the Golgi apparatus, where they undergo proteolytic cleavage to generate processed SREBP2 (nSREBP2). This processed form is then translocated to the nucleus where it upregulates the transcription of its target genes. We decided to investigate the impact of CS on SREBP2 activity in Huh-7 cells. When we depleted cellular cholesterol with LDS and a statin, which block cholesterol uptake and synthesis, respectively, we observed an increase in SREBP2 target gene transcripts, such as SREBF2, HMGCS1, HMGCR, SQLE, and LDLR (Fig. 5E). Cotreatment with CS partially attenuated this increase in SREBP2 target gene expression in a manner reminiscent of our liver bulk RNA-seq data with less potency than 25-HC. Therefore, we next measured SREBP2 activity at the protein level. Although depletion of cellular cholesterol increased the abundance of nSREBP2, the addition of CS reduced nSREBP2 levels (Fig. 5F). This was like the effect we observed with 25-HC, which suggests CS inhibits SREBP2 cleavage.

CS Promotes Interaction Between INSIG1 and SCAP

SCAP facilitates the transport of SREBP2 from the ER to the Golgi apparatus in response to decreased intracellular cholesterol levels. In contrast, increased cholesterol levels inhibit the transport of SREBP2 as SCAP/SREBP2 complexes bind INSIG1 or INSIG2. 25-HC enhances the interaction between SCAP and INSIGs, thereby preventing the translocation of SREBP2 to the Golgi apparatus (Yan et al., 2021). A recent cryo-electron microscopy (cryo-EM) analysis of the complex between human SCAP and INSIG2 revealed that 25-HC is positioned within the luminal membrane leaflet between SCAP segments S4 to S6 and INSIG2 transmembrane helices 3 and 4 (Yan et al., 2021).

To explore the potential interactions of CS with the SCAP/INSIGs complex, we performed a molecular docking simulation using the cryo-EM structure of the SCAP/INSIG2 complex in the presence of 25-HC (PDB ID: 6M49). This simulation indicated that CS occupies the sterol-binding pocket at the SCAP/INSIG2 interface, adopting a similar orientation and interaction profile as 25-HC (Fig. 6A-C). More specifically, the hydrocarbon tail of CS extends toward the cytosolic side, while its 3β-sulfate group projects into the luminal side (Fig. 6C).

Fig. 6.

Cholesterol sulfate (CS) promotes interaction between INSIG1 and SCAP. (A) 3D structure of the INSIG2-SCAP complex with CS made via docking. (B and C) Magnified view of the structure of 25-HC (B) and CS (C) surrounded by 2 proteins, simultaneously illustrating the positions of 3 crucial amino acid residues on SCAP that influence the interactions of the INSIG2-SCAP complex with 25-HC and CS. (D) Root mean squared deviations (RMSDs) of the dynamics for CS’s binding to pocket sites created in the INSIG2-SCAP complex during a 200-ns simulation. (E) Contact frequency for 25-HC and CS in the binding pocket created in the INSIG-SCAP complex during a 200-ns simulation. (F) Immunoprecipitation verifying the key interacting residues on SCAP in the INSIG1-SCAP complex that promote binding of CS and 25-HC. (G) Quantitative analysis of SCAP protein co-immunoprecipitated with INSIG1 as shown in (F). (H) Immunoprecipitation verifying the key interacting residues on SCAP in the INSIG1-SCAP complex that promote the binding of CS and that are unimportant for the binding of 25-HC. (I) Quantitative analysis of SCAP protein co-immunoprecipitated with INSIG from the results shown in (H). There is no statistically significant difference between groups labeled with the same letter.

By performing further molecular dynamics simulations, we found a consistent root mean square deviation value throughout the simulation period, indicating that both 25-HC and CS interact stably with the SCAP/INSIG2 complex (Fig. 6D and S6). In a contact frequency analysis, we identified key SCAP residues frequently interacting with ligands. Apart from Ile³⁴⁸, most SCAP residues showed a similar contact frequency with both 25-HC and CS (Fig. 6E).

To assess the functional impact of CS binding on SCAP-INSIG interactions, we next conducted an immunoprecipitation assay with SCAP and INSIG1. CS treatment resulted in the co-immunoprecipitation of SCAP with INSIG1, albeit with reduced efficiency compared to 25-HC (Fig. 6F and G). Notably, mutations at residues L343A and V355A, both of which were previously implicated in 25-HC-mediated SCAP and INSIG2 interactions (Yan et al., 2021), significantly reduced SCAP retention in the presence of 25-HC and CS (Fig. 6F and G). In the molecular dynamic simulation, Ile³⁴⁸ exhibited frequent contact with 25-HC but not with CS (Fig. 6E). When we substituted Phe for Ile³⁴⁸ (I348F), we saw a significant reduction in 25-HC-induced, but not CS-induced, SCAP-INSIG1 binding (Fig. 6H and I). These results suggest that CS enhances SCAP-INSIG interactions in a manner similar to 25-HC, although it may also mediate slightly different interaction patterns.

DISCUSSION

We have demonstrated that CS can reduce cellular cholesterol levels via multiple mechanisms. CS inhibits cholesterol synthesis by promoting HMGCR degradation and decreases LDL-C uptake by blocking LDLR endocytosis. In addition, CS prevents the translocation of SREBP2 into the Golgi apparatus, thereby disrupting the negative feedback loop that normally functions to increase cholesterol uptake and synthesis.

Oddly enough, CS was reported to inhibit cholesterol synthesis at the level of HMGCR in intact skin fibroblasts, but not in fibroblast lysates (Williams et al., 1985). Our results provide an explanation for this discrepancy. Unlike the statins, which are well-known HMGCR inhibitors, CS does not directly affect HMGCR enzyme activity (Fig. 2A). Instead, it promotes interaction between HMGCR and INSIGs, leading to INSIG-associated ubiquitin ligase-mediated ubiquitination of HMGCR (Fig. 2). This ubiquitination marks HMGCR for ERAD, a process that occurs only in the context of intact cells.

Statins, which lower LDL-C and reduce the risk of atherosclerotic cardiovascular disease, are the most frequently prescribed drugs in the world (Stone et al., 2014). The efficacy of statins can be reduced, however, when the liver increases the production of HMGCR to compensate for the reduced production of sterol and nonsterol isoprenoids secondary to statin-induced HMGCR inhibition (Jiang et al., 2018). We found that CS reduces statin-induced HMGCR protein levels (Fig. 3), suggesting that increasing plasma CS alongside statin treatment may help mitigate the side effects of high-intensity statin therapy. It is possible, however, that CS-induced inhibition of the proteolytic activation of SREBP2 could hinder the cholesterol-lowering effect of statins, because LDLR upregulation via SREBP2 activation in the liver is essential for statin efficacy (Fig. 5). CS may also blunt statin efficacy in the liver by inhibiting LDL-C uptake (Fig. 4). Thus, further investigation of the interactions between CS and statins in vivo is warranted.

CS seems to inhibit LDL-C uptake via 2 mechanisms (Fig. 4). Cholesterol replenishment partially rescues the DiI-LDL uptake defect induced by long-term CS treatment, indicating that this phenotype is partly a secondary effect of CS’s cholesterol-lowering activity, along with its additional effects (Fig. 4A-E). This was further supported by experiments with short-term CS treatment, which failed to alter intracellular cholesterol levels despite reducing DiI-LDL uptake (Fig. 4F-J). Some have reported that the increased charge of CS compared with cholesterol impedes sterol exchange between the membrane leaflets and phase separation, leading to reduced CME (Anderson et al., 2021, Bacia et al., 2005). Thus, the biophysical properties of CS likely impede LDL-C uptake via inhibition of LDLR endocytosis.

Recently, CS was suggested to have therapeutic potential for the treatment of ulcerative colitis because of its supposed ability to promote cholesterol biosynthesis (Xu et al., 2022). While CS may exert differential effects depending on the specific cells or tissues involved, our findings, along with previous studies (Williams et al., 1985, Williams et al., 1987), strongly suggest CS inhibits cholesterol synthesis. We demonstrated that CS promotes the degradation of HMGCR, the rate-limiting enzyme in cholesterol synthesis, both in vitro and in vivo. Mechanistically, CS functions as a ligand for various sterol-binding proteins, including Niemann Pick C2 (Infante et al., 2008), HMGCR, INSIGs, and SCAP, thereby negatively regulating cholesterol synthesis and modulating feedback mechanisms in a manner similar to 25-HC.

25-HC has also drawn significant attention, not only for its role in cholesterol metabolism at the cellular level but also for its involvement in broader processes such as metabolism, viral and bacterial infection, and various neurodegenerative diseases (Cyster et al., 2014). As a result, 25-HC has been suggested as a potential target for clinical development as a natural metabolite. High concentrations of 25-HC, however, have several negative effects in various contexts (Schroepfer, 2000). In contrast, CS exerts similar effects, albeit with less potency than 25-HC in terms of intracellular cholesterol homeostasis. In addition, no significant adverse phenotypes have been reported with CS, even in XLI patients, whose plasma CS levels can reach 90 µM (roughly 30 times that of healthy individuals) (Sanchez-Guijo et al., 2015). Further investigations into the physiological roles and therapeutic potential of CS in vivo will be necessary to uncover the full scope of its effects.

AUTHOR CONTRIBUTIONS

L.B.N. and S.K. conducted most of the experiments. K.N.T. and J.L. did the in silico analysis. C.J. and J.L. performed the FACS analysis. J.T.S. and S.J.M. designed and supervised the project and wrote the paper.

CRediT Authorship Contribution Statement

Seok Jun Moon: Writing – review and editing, Writing – original draft, Visualization, Supervision, Funding acquisition, Conceptualization. Jeong Taeg Seo: Writing – original draft, Supervision, Conceptualization. Le Ba Nam: Visualization, Investigation, Formal analysis, Data curation. Tan Khanh Nguyen: Visualization, Methodology. Sung-Jin Kim: Visualization, Methodology, Investigation. June-Yong Lee: Methodology, Investigation. Chang-Yun Jeong: Methodology, Investigation. Jun-Seok Lee: Methodology, Formal analysis, Data curation.

DECLARATION OF COMPETING INTERESTS

Seok Jun Moon reports financial support was provided by National Research Foundation of Korea. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

ORCID

Le Ba Nam: 0000-0002-3784-8852.

Sung-Jin Kim: 0000-0003-4115-0403.

Tan Khanh Nguyen: 0000-0002-6558-273X.

Chang-Yun Jeong: 0009-0001-8868-8465.

June-Yong Lee: 0000-0002-4476-725X.

Jun-Seok Lee: 0000-0003-3641-1728.

Jeong Taeg Seo: 0000-0003-2697-0251.

Seok Jun Moon: 0000-0001-7282-2888.

Appendix A. Supplementary material

Supplementary material

.

Supplementary material

.