Ezetimibe Promotes Brush Border Membrane-to-Lumen Cholesterol Efflux in the Small Intestine

Abstract

Ezetimibe inhibits Niemann-Pick C1-like 1 (NPC1L1), an apical membrane cholesterol transporter of enterocytes, thereby reduces intestinal cholesterol absorption. This treatment also increases extrahepatic reverse cholesterol transport via an undefined mechanism. To explore this, we employed a trans-intestinal cholesterol efflux (TICE) assay, which directly detects circulation-to-intestinal lumen 3H-cholesterol transit in a cannulated jejunal segment, and found an increase of TICE by 45%. To examine whether such increase in efflux occurs at the intestinal brush border membrane(BBM)-level, we performed luminal perfusion assays, similar to TICE but the jejunal wall was labelled with orally-given 3H-cholesterol, and determined elevated BBM-to-lumen cholesterol efflux by 3.5-fold with ezetimibe. Such increased efflux probably promotes circulation-to-lumen cholesterol transit eventually; thus increases TICE. Next, we wondered how inhibition of NPC1L1, an influx transporter, resulted in increased efflux. When we traced orally-given 3H-cholesterol in mice, we found that lumen-to-BBM 3H-cholesterol transit was rapid and less sensitive to ezetimibe treatment. Comparison of the efflux and fractional cholesterol absorption revealed an inverse correlation, indicating the efflux as an opposite-regulatory factor for cholesterol absorption efficiency and counteracting to the naturally-occurring rapid cholesterol influx to the BBM. These suggest that the ezetimibe-stimulated increased efflux is crucial in reducing cholesterol absorption. Ezetimibe-induced increase in cholesterol efflux was approximately 2.5-fold greater in mice having endogenous ATP-binding cassette G5/G8 heterodimer, the major sterol efflux transporter of enterocytes, than the knockout counterparts, suggesting that the heterodimer confers additional rapid BBM-to-lumen cholesterol efflux in response to NPC1L1 inhibition. The observed framework for intestinal cholesterol fluxes may provide ways to modulate the flux to dispose of endogenous cholesterol efficiently for therapeutic purposes.

Fig 1. Inhibition of NPC1L1 increases trans-intestinal cholesterol efflux (TICE).

A, An example of the TICE assay settings. B, An illustrated protocol for TICE assay. Assay time course is indicated from left to right. Arrow heads show the time points when reagents were given to mice. C, Decay per minute (DPM) counts in 5 μl blood obtained at the indicated time points after ³H-cholesterol intravenous infusion via the jugular vein. The DPM counts for individual mice were normalized against that at 60 min. Each symbol indicates an individual assay result. Infused tracer seemed to be equivalent in the circulation approximately 30 min after the infusion. Open symbols, vehicle controls; closed symbols, mice treated with ezetimibe. D, Ezetimibe treatment increased TICE. Open circles, vehicle; closed circles, 50 μg ezetimibe treated. Plots and error bars show mean and SEM (n = 8), respectively. E, Plots for ³H-DPM counts in sera (left, open circles) and those in total perfusates (right, gray circles). F, Comparison of the area under the concentration-time curve (AUC) of D. Each plot shows an individual assay result. P value was obtained using the Student’s t-test.

Fig 2. Ezetimibe increases brush border membrane-to-lumen cholesterol efflux.

A, An example of jejunal cannulation for the luminal perfusion assay (129^+Ter/SvJ mouse). B, An illustrated protocol for the luminal perfusion assay. Assay time course is indicated from left to right. Arrow heads show time points when reagents were given to mice. C, Upper; Plots for ³H-decay per minute (DPM) counts in the perfusate (right, gray circles) and the perfused intestinal segment (left, open circles). Lower; ³H-DPM in the perfusate was divided by that in the intestinal segment perfused to obtain a ratio (%), which was then converted to a common logarithm, and compared using Dunnett’s test. Circles show vehicle controls or ezetimibe treatment (1–50 μg). Triangles, methyl-β-cyclodextrin (MβCD), a cholesterol absorbent; 20 mg/ml in the perfusion buffer. Each plot shows an individual assay result. D, Intestinal ³H-DPM abundance did not change apparently during luminal perfusion assay. Intestinal segments were obtained 3 h after the labeling with ³H-cholesterol (open circles) or after luminal perfusion assay (gray circles) for DPM count (separate experiments from Fig 2C). Bars show median DPM of each group. Each plot shows an individual assay result. The data for the perfused columns were obtained from the data shown in C, which were divided with the respective intestinal lengths perfused (cm). E, Elution pattern of ³H-cholesterol from the intestinal lumen with 0–50 μg/mouse ezetimibe in luminal perfusion assays of C. The inset shows fractional perfusate/intestine ³H-DPM ratios (fractional efflux efficiency).

Fig 3. Intestinal cholesterol efflux is a biologically active phenomenon.

A, An illustrated protocol for the ex vivo cholesterol efflux assay and for the recovery of enterocytes. Assay time course is indicated from left to right. Open arrow heads show time points when reagents were given to mice. Closed arrow heads indicate events for processing of the small intestine. B, Comparison of luminal perfusion assay and ex vivo cholesterol efflux assay on the elution of cholesterol from intestinal segments. TC/PC indicates taurocholic acid and phosphatidylcholine. Bars indicate the mean percentages of ³H-decay per minute (DPM) counts of perfusate/intestinal segment in the luminal perfusion assay and ³H-DPM of elution/intestinal segment in ex vivo cholesterol efflux assay. Box lines, median (center), 25th percentile (lower), and 75th percentile (upper); error bars, 10th and 90th percentiles. Numbers in the parentheses show assay replicates. C, There was no significant difference in alkaline phosphatase (ALP) activity in the perfusates with or without ezetimibe treatment. Data were compared using the Mann—Whitney U-test. Each plot shows the result from an individual luminal perfusion assay. D, Left, green indicates ALP activity, an enterocyte-specific marker in the small intestine, using a fluorogenic ALP substrate, ELF-97. Right, Kinetics of cholesterol efflux in murine primary enterocytes. Enterocytes from mice treated with (closed circles) or without (open circles) ezetimibe (50 μmol/l) were incubated for up to 2 h. Data are shown as mean ± SEM of triplicate assays.

Fig 4. Effect of ezetimibe on cholesterol transit in mice.

A, An illustrated protocol for the ³H-cholesterol distribution assay. B, Upper; ³H-decay per minute (DPM) count distribution 3 h after ³H-cholesterol was given orally to C57BL/6J mice. Bars indicate mean and the standard deviation (n = 5 for vehicle and n = 4 for ezetimibe). Open bars, vehicle; gray bars, ezetimibe (50 μg). The ³H-DPM abundance in each portion was shown as % as given ³H-DPM as 100% (left, black bar). In the vehicle treatment, almost all the tracer infused was recovered; thus, we did not measure the tracer in the cecum and data of the cecum is absent for the vehicle. In the ezetimibe treatment, the given tracer count was not sufficiently recovered. We then measured the cecum and detected approximately 20% of the given tracer in them. B, Lower; the pie charts show summaries of ³H distribution in the small intestine (tissue), intestinal tract (lumen), and absorbed (the sum of the serum and the liver). C, Dose-dependent inhibitory effect of ezetimibe on fractional cholesterol transit into the serum and the liver. The reduction in tracer activity in enterocytes reached a plateau at 5 μg ezetimibe, whereas reductions in the liver and serum were greater than that in enterocytes with 5 μg and 50 μg ezetimibe. Changes in the distribution are shown with vehicle treatment as 100%. The significance of individual differences was evaluated by using Dunnett’s test. * p < 0.05; ** p < 0.01 vs. vehicle. Bars indicate mean ± SEM (n = 6).

Fig 5. Cholesterol enters HepG2 and differentiated Caco-2 cells by NPC1L1-independent manner mainly.

A, (a) Transient overexpression of NPC1L1 in the plasma membranes examined by Western blotting analysis. Upper, QuickBlue staining; lower, immunostaining for NPC1L1. (b) ABCG5 (upper) and ABCG8 (lower) protein expressions in HepG2 cells. Insets, these two images were adjusted to increase visibility of the bands. B, Medium-to-cell cholesterol transit in HepG2 cells. Transfection and ezetimibe treatment were performed as indicated as shown in the bottom. Medium-to-cell ³H-cholesterol transit efficiency (%) was estimated as described in the “Materials and Methods”. Each plot shows an individual assay result. Bars indicate means. Alphabetical differences among the groups (in parentheses) indicate significant difference between the groups (p < 0.05) using Tukey's Honestly Significant Difference test. C, Cholesterol (upper) and protein (lower) abundance in HepG2 cells. No significant difference was observed among the groups using Tukey's Honestly Significant Difference test. Bars show mean ± SEM (n = 4). D, (a) An illustration of Caco-2 cell culture system used in this study. Caco-2 cells were grown on filter membranes to allow the development to an enterocyte-like phenotype (see text for the detailed methods). The supernatants in culture inserts and wells were designated as apical and basolateral media, respectively. Lipid micelles were added to the apical medium. (b) Absorptive epithelial cell morphology of differentiated Caco-2 cell monolayers with a cylinder-like cell shape, the development of microvilli, and the distal localization of nuclei demonstrated by an electron microscopic analysis. (c) NPC1L1 was localized to the apical membrane in differentiated Caco-2 cells. Confocal microscopic analysis showed that NPC1L1 colored in green was localized to the plasma membrane (the upper image), especially in the brush border area (the lower image). 7-amino-actinomycin D was used for counterstaining of nuclei (red). E, Ezetimibe had little effect on medium-to-cell ³H-cholesterol transit in differentiated Caco-2 cells. Open circles, vehicle (1% ethanol); closed circles, 50 μmol/l ezetimibe. Data were shown as mean ± SEM of triplicate assays. F, In contrast to human duodenum (a) and murine jejunum (b), the gene expressions of ABCG5 and ABCG8 were absent in differentiated Caco-2 cells (c). Gene expression levels of five major membrane sterol transporters were analyzed by quantitative RT-PCR. An RNA sample of human duodenum was assayed in triplicate. Murine jejunal RNA samples were obtained from three C57BL/6J mice and assayed in duplicate. Three separate total RNA samples were obtained from independent wells of differentiated Caco-2 cells and assayed in duplicate. Data were shown as mean ± SEM of analytical triplicate (a) or biological triplicate (b, c) assays.

Fig 6. Cholesterol efflux in ABCG5 and ABCG8 double knockout (DKO) mice.

A, Quantitative RT-PCR showed that NPC1L1 gene expression level did not differ apparently with ABCG5/G8 deletions in mice. Gene expressions were normalized against 18s expression levels. Then the relative expression levels were compared between the two genotypes, WT (n 10) and ABCG5/G8 DKO (n 6) with the levels in the WT mice as the references. Data are shown as mean ± SEM. B, Ezetimibe (50 μg)-induced increase in cholesterol efflux was partially abolished in ABCG5/G8 DKO mice. Upper, Plots for ³H-decay per minute (DPM) counts in the perfusate (right, gray circles) and the perfused intestinal segment (left, open circles). Lower; ³H-DPM in the perfusate was divided by that in the intestinal segment perfused to obtain a ratio (%), which was then converted to a common logarithm. Bars indicate medians. The difference between wild-type (WT) and DKO mice was compared using the Student’s t-test. Each plot shows an individual assay result. Alphabetical differences (in parentheses) indicate significant difference between the groups (p < 0.05) using Tukey's Honestly Significant Difference test. C, ABCG5/G8 DKO partially abolished the inhibitory effect of ezetimibe on the intestinal cholesterol transit. Bars indicate medians. Each plot shows an individual assay result. The difference between the WT and the DKO mice was compared using the Mann—Whitney U-test. D, Comparison of mRNA abundance by quantitative RT-PCR between the jejunal samples of C57BL/6J and 129^+Ter/SvJ. Data are shown as mean ± SEM (n 5). E, Characteristic comparisons of C57BL/6J and 129^+Ter/SvJ mice in luminal perfusion assay. Upper, efflux efficiency; lower, intestinal ³H-DPM abundance. The difference between the two strains was compared using the Mann—Whitney U-test. Bars indicate medians. Each plot shows an individual assay result.

Fig 7. An inverse correlation between BBM-to-lumen cholesterol efflux and fractional lumen-to-circulation transit.

A, Comparison of fractional cholesterol transit (Fig 4C) and cholesterol efflux efficiency (Fig 2C) in ezetimibe-treated C57BL/6J mice. B, Comparison of fractional cholesterol transit and cholesterol efflux efficiency between WT and ABCG5/G8 DKO mice presented in Fig 5A, lower, and 5B. A and B, The mean relative efflux efficiency (%) and the median relative absorption (%) versus vehicle are plotted. Bars for the X-axis and Y-axis indicate the SEM and 50% coefficient interval, respectively.

Fig 8. Poorly absorbable sitosterol and sitostanol show greater efflux efficiencies than cholesterol.

A, Time-course uptake of ³H-cholesterol (circles), ³H-sitosterol (squares), and ³H-sitostanol (triangles) by differentiated Caco-2 cells. Plots are shown as mean of duplicate assays. B, An illustrated protocol for the luminal perfusion assay with sitosterol or sitostanol tracers. Assay time course is indicated from left to right. Open triangles show time points when reagents were given to mice. C, Upper; Plots for ³H-decay per minute (DPM) counts in the perfusate (right, gray circles) and perfused intestinal segment (left, open circles). Lower; Comparison of efflux efficiency of three tracers: cholesterol, sitosterol, and sitostanol. D, Change in ³H abundance per cm of intestinal segment during luminal perfusion assay with ³H-sitosterol or ³H-sitostanol. Bars indicate medians. Each plot shows an individual assay result. E, Comparison of sterol absorption efficiency (cholesterol absorption efficiency as 100%) with efflux efficiency for cholesterol, sitosterol, and sitostanol. Sterol absorption efficiencies were obtained from Igel et al. (Ref. 10), in which absorption efficiencies for sitosterol and sitostanol were estimated as 7% and 1%, respectively. Efflux efficiencies were obtained from C. F, Efflux efficiency of three sterol tracers in ABCG5/G8 double knockout (DKO) mice. Upper; Plots for ³H-DPM counts in the perfusate (right, gray circles) and perfused intestinal segment (left, open circles). Lower; Effect of ABCG5/G8 DKO on sterol efflux efficiency. P values were obtained using the Student’s t-test (sitostanol and cholesterol) or the Mann—Whitney U test (sitosterol) for comparison of efflux efficiency. Each plot shows an individual assay result.

Fig 9. Brush border membrane as a major platform for cholesterol bidirectional flux.

A, Balance out of cholesterol translocation in the brush border membrane (BBM). The BBM is focused (the circle in the upper panel); the gray square indicates lipid bilayer, or microvillus. T, total input cholesterol incorporated into the BBM from the apical side. Cholesterol (T) diffuses into the BBM via the pathway (a). x, the amount of cholesterol transferred to endosomes for further processing via the pathway (b) and considered to be transferred eventually to the circulation, which of the following processes were not identified in this presentation. T—x, estimated amount of cholesterol effluxed into the lumen via the pathway (c). Thus, ‘x’ and ‘T—x’ should be inversely correlated. B, An illustration of the hypothesized bidirectional intestinal cholesterol flux system. Cholesterol diffuses into the BBM (pathway d; Fig 6); NPC1L1 mediates cholesterol movement from the BBM to the endosomal processing [e; Chang T-Y and Chang C. (2008, Ref. 24)]; Cholesterol that diffuses into the BBM can be caught by ABCG5/G8 in cholesterol-rich microdomains (f). ABCG5/G8 promotes elimination of sterols from the BBM when required or the sterols become in excess (g; Fig 6). Collectively, the BBM provides a buffering space for the bidirectional flux of cholesterol. C, A proposed route for TICE. Endogenous cholesterol circulates into the BBM (pathway, h). Ezetimibe prevents internalization of the cholesterol from the BBM (e) (Ref.24). The cholesterol in the BBM diffusively exits to the lumen or is pumped out by ABCG5/G8 (d and g).