NPC1L1-dependent intestinal cholesterol absorption requires ganglioside GM3 in membrane microdomains

Intestinal cholesterol absorption is a key regulator of systemic cholesterol homeostasis. Excessive dietary cholesterol and its intestinal uptake lead to hypercholesterolemia, a major risk factor for cardiovascular disease. Intestinal cholesterol uptake is mediated by Niemann-Pick C1-like 1 (NPC1L1), a transmembrane protein localized in membrane microdomains (lipid rafts) enriched in gangliosides and cholesterol. The roles of gangliosides, such as monosialodihexosylganglioside (GM3) and its synthesizing enzyme GM3 synthase (GM3S), in NPC1L1-dependent cholesterol uptake have not been examined previously. Here, we examined NPC1L1-dependent cholesterol uptake in a cell model as well as in wild-type and apoE-deficient mice fed normal or high-cholesterol diets. We showed that NPC1L1-dependent cholesterol uptake was impaired in GM3S-deficient cells and that GM3S deficiency promoted resistance to hypercholesterolemia in both wild-type and apoE-deficient mice fed the high-cholesterol but not the normal diet. Our findings suggest that GM3 and related gangliosides are essential for NPC1L1-mediated intestinal cholesterol absorption and are potential targets for hypercholesterolemia therapy.

hypercholesterolemia, and the intestinal cholesterol absorption rate are reduced in GM3S-deficient mice. Our findings suggest that gangliosides, particularly GM3, are potential targets for hypercholesterolemia therapy.

Animals
C57BL/6 mice and apoE-deficient mice (B6.KOR/stm Slc-Apoe shl ) were from Japan SLC, Inc. (Hamamatsu, Japan). GM3S (St3gal5)deficient mice were generated in our lab as described previously (19). To generate ApoE shl /GM3S +/ mice, ApoE shl mice were crossed with GM3S / mice. ApoE shl /GM3S / mice and littermate controls were generated by heterozygous mating. Mice were analyzed for the GM3S genotype by PCR and for apoE protein expression by immunoblotting as described previously (20,21). Mice were fed a regular chow diet (CE-2; CLEA Japan, Tokyo, Japan) or high-cholesterol diet (Research Diets; New Brunswick, NJ) ad libitum. All animal experiments were approved by appropriate institutional review board committees at Tohoku Medical and Pharmaceutical University.

Generation of GM3S-deficient HEK293T cells
A single exon of the human ST3GAL5 (GM3S) gene containing the coding sequence for sialyl motif L was selected for the design of targeting guide RNAs using an online CRISPR design tool (23). Guide oligos (5′-CACCGCAAGACCTGTCGGCGCTGTG-3′ and 5′-AAACCACAGCGCCGACAGGTCTTGC-3′) were annealed and inserted into plasmid pSpCas9(BB) (Addgene, Cambridge, MA). The plasmid was introduced into HEK293T cells using Lipofectamine 2000 (Thermo Fisher) per the manufacturer's instructions, clonal cell lines were obtained by single-cell cloning, and expression levels of gangliosides were evaluated by TLC.

Measurement of cellular cholesterol
Cells were seeded in DMEM in 60-mm dishes at a density of 8 × 10 5 cells/dish and transfected at 24 h with the indicated plasmid using Lipofectamine LTX (Thermo Fisher) per the manufacturer's instructions. The medium was replaced by cholesterol-depleting medium at 48 h, cells were cultured overnight, and cholesterol-MCD complex was added to the medium the next day. For ezetimibe treatment, cells were preincubated with 30 µM ezetimibe for 30 min, incubated with cholesterol-MCD complex for 60 min, and washed with PBS, and total cellular lipids were extracted as described by Bligh and Dyer (24). Lipid extracts were dried by an N 2 stream and resuspended in 1 ml 1% Triton X-100 in chloroform, chloroform was evaporated by an N 2 stream, and detergentsolubilized lipids were resuspended in 1 ml distilled water. Total cholesterol and phospholipid concentrations were determined using LabAssay TM cholesterol and phospholipid test kits (Wako Pure Chemical Industries, Osaka, Japan).

Visualization of living cells and fluorescence quantification
For time-lapse fluorescence imaging, cells were seeded onto 35-mm glass-bottom dishes (Greiner Bio-One, Frickenhausen, Germany) coated with poly-l-lysine (Sigma-Aldrich) and transfected at 24 h. The medium was replaced by cholesterol-depleting medium at 48 h, and cells were cultured overnight. Cholesterol-MCD complex was added to the medium the next day with or without 30 µM ezetimibe. Living cells were visualized by confocal laser scanning microscopy (FluoView FV1000; Olympus, Tokyo, Japan) for 60 min. The relative intensity of plasma membranelocalized NPC1L1-GFP turbo was quantified as described previously (6). The intensity at various time points was normalized relative to the intensity at time zero (defined as 100%). Fluorescence intensity was calculated using the FluoView software program (Olympus).

Blood cholesterol and lipoprotein analyses
Plasma total cholesterol was analyzed using the Cholesterol E-Test kit (Wako Pure Chemical Industries) per the manufacturer's instructions. Serum lipoproteins were analyzed by an HPLC system at Skylight Biotech (Akita, Japan) according to the procedure described by Usui et al. (25).

Lipid and LC/MS/MS analyses
Lipids were analyzed as described previously (26). Intestinal samples were obtained by perfusing mice with saline and scraping off small-intestinal mucosa with a plastic spatula. LC/MS/MS analysis was performed as described previously (27).

Intestinal cholesterol absorption rate
The fecal dual-isotope ratio method (28)

Statistical analysis
Data were expressed as mean ± SD, and means were compared by Student's t-test or ANOVA followed by Tukey's post hoc test.

RESULTS
To evaluate the possible involvement of gangliosides (GM3) in NPC1L1-mediated cholesterol absorption, we first examined cholesterol content in control HEK293T and GM3S-deficient (GM3S KO) cells. The two cell lines were transfected with NPC1L1-GFP turbo , and cellular cholesterol content was measured. GM3S deficiency had no effect on endogenous cholesterol levels (Fig. 1). In NPC1L1expressing control cells, the cholesterol level was increased by cholesterol supplementation, and the increase was blocked by pretreatment with ezetimibe. In NPC1L1-expressing Fig. 1. GM3S deficiency inhibits cholesterol uptake via an NPC1L1-dependent pathway. NPC1L1-expressing control HEK293T cells and GM3S KO cells were incubated in cholesterol-depleting medium to reduce cellular cholesterol. For cholesterol replenishment, cholesterol-MCD complex (30 µg/ml) was added directly to the medium, and cells were cultured for 60 min with or without ezetimibe pretreatment. Total lipid extraction was performed, and cholesterol and phospholipid contents were measured. Cholesterol content was normalized relative to phospholipid content. **P < 0.01.

Fig. 2.
Cholesterol-dependent internalization of NPC1L1-GFP turbo is ameliorated by GM3S depletion. A: Control HEK293T and GM3S KO cells were seeded in 0.001% poly-l-lysine-coated 35-mm glass-bottom dishes on day 0 and transfected with NPC1L1-GFP turbo on day 1. The medium was replaced with medium containing 5% lipoprotein-deficient serum and 2 µM compactin on day 2 to deplete cellular cholesterol, and cells were supplemented with 60 µg/ml cholesterol with or without ezetimibe treatment on day 3. Time-lapse images were taken by confocal microscopy. Representative images are shown. B: Quantification of plasma membrane-localized NPC1L1-GFP turbo in the cells shown in panel A. Intensity at time zero was defined as 100%. *P < 0.05 for comparison of control versus GM3S KO.
GM3S KO cells, cholesterol uptake was significantly lower than in control cells, and ezetimibe pretreatment had no notable effect (Fig. 1). Previous studies indicate that the dynamic translocation of NPC1L1 between the cell surface and intracellular region is essential for NPC1L1-mediated cholesterol uptake and that cholesterol is required for the active endocytosis of NPC1L1 (6, 10).
Next, experiments were performed to evaluate the effect of GM3S deficiency on NPC1L1 translocation. In control cells, cholesterol supplementation following cholesterol depletion resulted in the translocation of NPC1L1 from the plasma membrane to the intracellular region. In contrast, NPC1L1 translocation was much lower in GM3S KO cells and at a level similar to that of ezetimibe-treated cells (Fig. 2). These findings indicate the involvement of GM3 in NPC1L1-dependent cholesterol absorption.
We accordingly hypothesized that experimentally induced hypercholesterolemia in GM3S KO (GM3S / ) mice can be ameliorated by inhibiting NPC1L1-mediated intestinal cholesterol uptake. To test this hypothesis, we crossed apoE-deficient, spontaneously hyperlipidemic mice (ApoE shl ) with GM3S / mice and examined plasma cholesterol levels. Plasma cholesterol was not significantly reduced in GM3S / mice, whereas levels in ApoE shl / GM3S / mice were strikingly lower than the high levels in ApoE shl mice (Fig. 3A). Next, we examined the possible resistance of GM3S / mice to diet-induced hypercholesterolemia. Plasma cholesterol levels were increased by a high-cholesterol diet in WT mice but not in GM3S / mice (Fig. 3B). These findings indicate that GM3S / mice were resistant to hypercholesterolemia induced by either apoE deficiency or a high-cholesterol diet.

/GM3S
/ serum. A 5 l serum sample was injected onto two tandem gel permeation columns and eluted with TSK eluent LP-1 at a flow rate of 0.7 ml/min. Pink lines represent cholesterol, and blue lines represent triglyceride. Serum total cholesterol and total triglyceride levels are 587 ± 65 and 59 ± 33 mg/dl (A) and 259 ± 75 and 35 ± 11 mg/dl (B), respectively. Lipoprotein subclasses determined from observed elution times are presented. C-F: Chylomicron, VLDL, LDL, and HDL, respectively (n = 3 per group). *P < 0.05 and **P < 0.01. Plasma lipoprotein profiles were obtained for ApoE shl and ApoE shl /GM3S / mice. In ApoE shl /GM3S / mice, cholesterol content was significantly reduced in chylomicron, VLDL, and LDL fractions but not in the HDL fraction (Fig. 4). The reduction of cholesterol content was most striking for the chylomicron fraction (Fig. 4C), indicating defective intestinal cholesterol absorption in these mice.
We next examined the GSL composition of intestinal mucosa, where NPC1L1-mediated cholesterol absorption occurs. It has been reported that intestinal GM3S expression level is high in neonatal mice and declines during the course of development (29). We detected GM3 expression in WT and ApoE shl mice by TLC and LC/MS/MS analyses. Trace amounts of GM3 molecular species were also detected in GM3S KO mice (Figs. 5A, 6), likely as the result of a newly identified transcriptional variant in GM3S / mice generated by targeting exon 3 of the GM3S gene (30). Levels of neutral GSLs in intestinal mucosa were not notably altered in GM3S / mice (Fig. 5B).
To test the hypothesis that resistance to hypercholesterolemia in GM3S / mice is due to impaired NPC1L1 function, we compared the intestinal cholesterol absorption rates of ApoE shl versus ApoE shl /GM3S / mice based on the oral administration of radiolabeled cholesterol. Uptake of cholesterol from the intestine was significantly lower in ApoE shl /GM3S / than in ApoE shl mice (Fig. 7). It has been demonstrated that the oral administration of cholesterol in mice induces the translocation of NPC1L1   from the intestinal epithelial surface to the intracellular region (31,32). We examined the possibility that GM3S deficiency impairs cholesterol-dependent NPC1L1 translocation in vivo by immunostaining of intestinal NPC1L1. In both ApoE shl and ApoE shl /GM3S / mice, in the absence of cholesterol feeding, NPC1L1 localized mainly at the apical side of enterocytes (Fig. 8). Cholesterol feeding induced NPC1L1 internalization in ApoE shl but not in ApoE shl / GM3S / mice. Taken together, these findings clearly indicate that GM3 and/or related gangliosides are essential for NPC1L1-dependent intestinal cholesterol absorption.

DISCUSSION
The protein NPC1L1 is known to be localized in detergent-resistant, ganglioside-enriched microdomains (11,12,32). The role of gangliosides in NPC1L1-dependent cholesterol absorption is unknown. Our previous studies have shown that GM3 plays key roles in certain metabolic disorders and that the inhibition of GM3 biosynthesis may help ameliorate metabolic imbalance (27,33,34). Results from the present study indicate that GM3S deficiency promotes resistance to hypercholesterolemia by inhibiting NPC1L1mediated cholesterol uptake. NPC1L1-expressing GM3S KO cells displayed cholesterol uptake significantly lower than that of control cells (Fig. 1) and impairment of the cholesterol-dependent translocation of NPC1L1 from the plasma membrane to the intracellular region (Fig. 2). Consistent with these findings, GM3S / mice showed reductions of intestinal cholesterol uptake and cholesteroldependent translocation of NPC1L1 (Figs. 7, 8). Plasma cholesterol levels in WT, GM3S / , ApoE shl , and ApoE shl / GM3 / mice are summarized in Fig. 3A. GM3S-deficient mice were resistant to hypercholesterolemia induced by the high-cholesterol diet, and the hypercholesterolemia characteristic of ApoE shl mice was significantly ameliorated in ApoE shl /GM3S / mice. On the other hand, plasma cholesterol levels were similar for WT and GM3S / mice fed a normal diet. Taken together, these observations suggest functional involvement of GM3S in the exogenous pathways of cholesterol metabolism, including intestinal NPC1L1 activity.
Developmental changes in intestinal GSL composition have been found to be synchronized with expression levels of intestinal nutrient transporters (29). The knockdown of intestinal glucosylceramide synthase resulted in retarded growth and early death in mice because of defects in intestinal intracellular vesicular transport (35). These studies suggest that GSLs are physiologically important for intestinal nutrient absorption, but they did not address the role of GSLs in the NPC1L1 pathway. It has been shown that NPC1L1 requires a cholesterolenriched membrane microdomain to function as a cholesterol transporter (11,12,32).
In the present study, cholesterol uptake by NPC1L1 was reduced in GM3S-deficient cells and mice. We therefore conclude that NPC1L1 requires not only cholesterol but also GM3 (or related gangliosides) to form functional membrane microdomains for cholesterol transport. Two possibilities can be considered: i) gangliosides interact directly with NPC1L1 via electrostatic interactions with multiple oligosaccharide chains to facilitate conformational change leading to translocation from lipid rafts to clathrincoated pits, and ii) gangliosides are required for the association of NPC1L1 with proteins such as flotillins. The present findings provide novel insights into the mechanism of NPC1L1-mediated cholesterol absorption, which can be regulated by membrane lipid composition as well as by protein-protein interactions. The detailed mechanisms whereby GM3 and related gangliosides function in NPC1L1-mediated cholesterol absorption remain to be elucidated.
Gangliosides, particularly GM3, and its synthesizing enzyme GM3S appear to be potential targets for hypercholesterolemia therapy. Genetic variation in NPC1L1 is closely associated with interindividual variation in response to ezetimibe (36). Moreover, the loss of ezetimibe-binding mutations in the extracellular loop of NPC1L1 has been reported (9). Taken together, one can speculate that mutations in the region lead to unresponsiveness to the ezetimibe treatment. Membrane lipid modification such as GM3S inhibition can be used regardless of the binding affinity of compounds to the NPC1L1 and provide an alternative therapy for nonresponsive individuals.
The authors are grateful to all members of the Inokuchi and Fukase laboratories for valuable scientific discussions and technical support, to the Tohoku Medical and Pharmaceutical University Center for Laboratory Animal Science for their services, and to S. Anderson for English editing of the manuscript. / mice with or without oral cholesterol administration. Frozen sections were stained with anti-NPC1L1 antibodies. Arrows: plasma membrane-localized NPC1L1. Arrowheads: internalized NPC1L1.