Quantifying anomalous intestinal sterol uptake, lymphatic transport, and biliary secretion in Abcg8−/− mice


  • Helen H. Wang,

    1. Department of Medicine, Liver Center and Gastroenterology Division, Beth Israel Deaconess Medical Center, Harvard Medical School and Harvard Digestive Diseases Center, Boston, MA
    Search for more papers by this author
  • Shailendra B. Patel,

    1. Division of Endocrinology, Metabolism, and Clinical Nutrition, Medical College of Wisconsin, Milwaukee, WI
    2. Clement J. Zablocki Veterans Medical Center, Milwaukee, WI
    Search for more papers by this author
  • Martin C. Carey,

    1. Division of Gastroenterology, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School and Harvard Digestive Diseases Center, Boston, MA
    Search for more papers by this author
  • David Q.-H. Wang

    Corresponding author
    1. Department of Medicine, Liver Center and Gastroenterology Division, Beth Israel Deaconess Medical Center, Harvard Medical School and Harvard Digestive Diseases Center, Boston, MA
    • Gastroenterology Division, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, DA 601, Boston, MA 02215
    Search for more papers by this author
    • fax: 617-975-5071

  • Presented in part at the Annual Meeting of the American Gastroenterological Association, Chicago, Illinois, in 2005 and published as an abstract in Gastroenterology 2005;122:A403.

  • Potential conflict of interest: Dr. Patel is on the speakers' bureau of Merck-Schering Plough.


Sitosterolemia is caused by mutations in either ABCG5 or ABCG8, but simultaneous mutations of these genes have never been observed. To explore whether ABCG8, the sterol efflux (hemi-)transporter, plays a major role in determining intestinal absorption efficiency and hepatic secretion rates of cholesterol and sitostanol, we performed direct measurements of the absorption and lymphatic transport of these sterols in mice with chronic biliary and lymphatic fistulae, as well as the transport rates of radiolabeled cholesterol and sitostanol from plasma high-density lipoprotein (HDL) into bile in male Abcg8−/− and wild-type mice. We observed that the absorption and lymphatic transport rates of radiolabeled cholesterol and sitostanol were increased by ≈40% and ≈500%, respectively, in Abcg8−/− mice in the setting of constant intraduodenal infusion of micellar taurocholate and lecithin. Both strains displayed identical intestinal Npc1l1 expression levels and small intestinal transit rates. After 45 minutes of intraduodenal infusion, acute intestinal uptake rates of trace [14C]cholesterol and [3H]sitostanol were essentially similar in both groups of mice with intact biliary secretion. Furthermore, in wild-type mice, mass transport rate of [3H]sitostanol from plasma HDL into bile was significantly faster than that of [14C]cholesterol; however, no [3H]sitostanol and only traces of [14C]cholesterol were detected in bile of Abcg8−/− mice. Conclusion: Deletion of the Abcg8 gene alone significantly increases the mass of intestinal cholesterol and sitostanol absorption and reduces but does not eliminate hepatic secretion of cholesterol. Moreover, the mutation has no influence on acute uptake of cholesterol and sitostanol by the enterocyte nor small intestinal transit time. (HEPATOLOGY 2007;45:998–1006.)

Sitosterolemia is a rare autosomal, recessively inherited disorder that is characterized mainly by elevated plasma levels of plant sterols (phytosterols) and with normal or only moderately increased cholesterol levels.1 Sitosterolemic patients display hyperabsorption of cholesterol and phytosterols and reduced secretion of these sterols into bile.2, 3 Patel et al.4 were the first to map the sitosterolemia locus, STSL, to human chromosome 2p21, between D2S2294 and D2S2298. Using a positional cloning approach as well as microarray analysis of murine complementary DNAs from intestines and livers of mice treated with a liver X receptor (LXR) agonist, 2 groups5, 6 independently identified the adjacent genes, ABCG5 and ABCG8, encoding adenosine triphosphate–binding cassette (ABC) hemi-transporters that are mutated in sitosterolemia. Unlike other ABC transporters that encode proteins with 12-transmembrane domains, these 2 proteins function as a unique heterodimer forming a 12-transmembrane protein complex essential for sterol transport activity.7 Heterodimerization of ABCG5 and ABCG8 is also necessary for their translocation from endoplasmic reticulum to the plasma membrane.8 Based on these observations, as well as the similarity of the clinical and biochemical phenotypes, it has been proposed that both ABCG5 and ABCG8 function as obligate heterodimers in the transmembrane “flipping” of sterols.9

A mouse model with disruption of both Abcg5 and Abcg8 results in a phenotype similar to that of human disease.10 However, human sitosterolemia is caused by a disruptive mutation in only 1 of the 2 ABC genes, but not in both simultaneously. Mice with disruption of Abcg8 or Abcg5 alone have been described.11, 12 In the absence of a functional ABCG8, biliary sterol secretion is severely impaired, but sterol absorption has not been characterized. Moreover, in mice deficient in ABCG5 alone, biliary sterol secretion under baseline conditions seemed to be impaired, but upon stimulation with an LXR agonist, secretion was claimed to be restored to normal. Additionally, a real distinction between cholesterol and phytosterols in terms of either their intestinal absorption or biliary sterol secretion has not been elucidated. In a study where cholesterol absorption was measured by a fecal dual-isotope ratio method in Abcg5−/−/Abcg8−/− mice, sitostanol was employed injudiciously as the “nonabsorbable” standard for correction10 because it is absorbed well in humans with sitosterolemia.1–3 In addition, the difficult and laborious biliary sterol secretion studies in humans pose a dilemma, because patients with mutations in either ABCG5 or ABCG8 are rare and geographically scattered. Despite the aforementioned studies and issues, little information is available to date on whether ABCG5 alone, ABCG8 alone, or both ABCG5 and ABCG8 are responsible for selectively pumping cholesterol and sitostanol out of enterocytes and hepatocytes and whether the 2 half-transporters function as homodimers or heterodimers.

We hypothesized, therefore, that disruption of just one of the half-transporters would result in a significant effect on intestinal absorption and biliary secretion of sterols such as occurs in humans with sitosterolemia.1–3 Our results show that deletion of the Abcg8 gene alone significantly increases the mass of cholesterol and sitostanol absorbed from the small intestine. Also, loss of ABCG8 function impairs all biliary secretion of sitostanol and reduces, but does not eliminate, hepatic secretion of biliary cholesterol. Our study indicates that this protein (together with its heterodimeric partner, ABCG5) is key to preferential secretion of phytosterols into bile and is also responsible for hepatic secretion of the majority of cholesterol into bile. Furthermore, deletion of this hemi-transporter neither influences small intestinal motility nor interferes with acute uptake rates of cholesterol and sitostanol by the small intestine via the canonical NPC1L1 transporter.


ABC, adenosine triphosphate–binding cassette (transporter); HDL, high-density lipoprotein; LXR, liver X receptor.

Materials and Methods


Medium-chain triglyceride was purchased from Mead Johnson (Evansville, IN) and its fatty acid composition includes 67% octanoic acid (8:0) and 23% decanoic acid (10:0), followed by <6% shorter than C8 fatty acids and <4% longer than C10 fatty acids. The synthetic LXR agonist T0901317 {N-(2,2,2-trifluoro-ethyl)-N-[4-(2,2,2-trifluoro-1-hydroxy-1-trifluoromethyl-ethyl)phenyl]benzenesulfonamide} was purchased from Cayman Chemical (Ann Arbor, MI). Radioisotopes [4-14C]cholesterol and [51Cr]sodium (as Na2[51Cr]O4) were purchased from PerkinElmer Life and Analytical Sciences (Boston, MA), and [5,6-3H]sitostanol was obtained from American Radiolabeled Chemicals (St. Louis, MO).

Animals and Diets.

Abcg8−/− mice in a C57BL/6J background were generated by targeting disruption of the Abcg8 gene.11 The wild-type mice on the same C57BL/6J background were purchased from the Jackson Laboratory (Bar Harbor, ME). Male mice at 8-10 weeks of age and with a mean (±SD) body weight of 32.2 ± 4.5 g were fed normal rodent chow (Harlan Teklad Laboratory Animal Diets, Madison, WI) containing trace (<0.02%) amounts of cholesterol. All procedures were performed in accordance with current NIH guidelines and were approved by the Institutional Animal Care and Use Committee of Harvard University.

Measurement of Absorption and Lymphatic Transport of Cholesterol and Sitostanol.

To exclude the effects of endogenous bile on intestinal absorption of the sterols, we established a mouse model (n = 5 per group) with a chronic biliary fistula and employed intraduodenal infusion of model “bile.”13, 14 After the mesenteric lymphatic duct was cannulated with a polyethylene catheter (PE-10), its end was externalized through the abdominal wall and connected to a heparinized tube. Exactly 2.5 μCi of [14C]cholesterol and 5.0 μCi of [3H]sitostanol dissolved in 100 μl of medium-chain triglyceride dispersed with 1.6 μmol taurocholate and 0.4 μmol lecithin, which constitutes a model “micellar” bile, were infused into the small intestine through a duodenal catheter. Because the constituent fatty acids of medium-chain triglyceride oil have essentially a neutral effect on cholesterol metabolism,15 this oil was selected as the vesicle for oral dosing procedures and intraduodenal infusion in cholesterol absorption experiments.13, 14, 16 Measurement of absorption and lymphatic transport of cholesterol and sitostanol was performed for a total of 12 hours. The two radioactive isotopes were extracted from collected lymph and counted. To maintain lymph flow, a continuous intraduodenal infusion of model micellar “bile” containing 16 mM taurocholate and 4 mM lecithin with medium-chain triglyceride was performed at a rate of 300 μl/hour for 12 hours. In our in situ experimental system, mice were anesthetized during the experiments, which allowed hourly collection of fresh lymph samples without the effects of stress on the animals, such as may be induced by restraining cages. Continuous anesthesia was maintained with an intraperitoneal injection of pentobarbital at a dose of 25 mg/kg every 2 hours. Moreover, it is easier to maintain anesthetized mice in a physiological state of hydration with an intravenous infusion of 0.9% NaCl at 100 μl/hour throughout the experimental period.

Cholesterol Balance Studies.

To further determine whether deletion of the Abcg8 gene alone increases cholesterol absorption, we measured percent cholesterol absorption and total mass of cholesterol absorbed from the small intestine by the cholesterol balance method.14, 17 Because biliary lipid secretion in mice is continuous and constant during the 24-hour period, biliary cholesterol outputs during the first hour of interruption of the enterohepatic circulation were determined. For calculation of daily biliary cholesterol secretion, this first hour of biliary cholesterol output was normalized to 24 hours. To investigate whether LXRα plays a regulatory role in cholesterol absorption via ABCG8, we studied mice (n = 5 per group) to which the LXR agonist was administered intragastrically via gavage at 10 mg/kg/day, or a control vehicle in a formulation of propylene glycol/Tween 80 (4:1, vol/vol) for 2 weeks.

Hepatic Secretion of Biliary Sterols.

Plasma high-density lipoprotein (HDL) rather than very low-density lipoprotein and low-density lipoprotein preferentially provides cholesterol and sitostanol to the liver for secretion as such into bile.18, 19 Exactly 20 ml of plasma from wild-type mice was incubated with a Whatman filter paper impregnated with [3H]sitostanol and [14C]cholesterol at 4°C for 12 hours. The HDL (1.050 ≤ d ≤ 1.210 g/ml) fraction was isolated in a Beckman Model L8-80M preparative ultracentrifuge using a type SW41 Ti rotor at 4°C and 285,000g for 24 hours. The radiolabeled HDL was dialyzed overnight against 0.9% NaCl and 0.01% EDTA (pH 7.4), and then passed through a 0.22 μm Millipore filter to remove particulate cholesterol and proteins. To determine the transport rate of radiolabeled cholesterol and sitostanol from plasma HDL into bile, the common bile duct was cannulated following cholecystectomy.20 Mice (n = 5 per group) were injected intravenously with a bolus of mouse HDL labeled with identical quantities (5.0 μCi) of [3H]sitostanol and [14C]cholesterol. Immediately after injection, samples of hepatic bile were collected by gravity at 30-minute intervals for 6 hours and total radioactivity in the samples was counted.

Quantitative Real-Time PCR Assay.

The duodenal, jejunal, and ileal tissues were harvested and the tissues from 4 mice were pooled per group. Primers and probes for Abcg5, Abcg8, Lxrα, Lxrβ, Npc1l1, Sr-b1, Abca1, and Fatp4 were described elsewhere.17, 21 To obtain normalized values, the target amount was divided by the endogenous reference Gapdh as the invariant rodent control. Real-time PCR assays in triplicate were performed according to published methods.17, 21

Measurement of Intestinal Sterol Uptake.

After the duodenum was cannulated with a PE-10 catheter, the catheter was externalized through the incision and implanted subcutaneously.17 After 24-hour recovery from surgery and another 12-hour fasting period (with water ad libitum), exactly 2 μCi [14C]cholesterol and 2 μCi [3H]sitostanol dissolved in 100 μl of medium-chain triglyceride were injected into the nonanesthetized mice (n = 5 per group) via the duodenal catheter. After injection, mice were allowed to move freely in the cage. Exactly 45 minutes after instillation, the entire small intestine was removed and flushed with 0.5% taurocholate buffer solution. After wet-weighing, the small intestine was cut into 3 segments with duodenum/jejunum/ileum length ratios of 1:3:2. The radiolabeled sterols were extracted and counted. The radioactivity was used to calculate intestinal sterol uptake in vivo, which is expressed as disintegrations per minute (dpm)/g tissue/45 minutes.17

Measurement of Small Intestinal Transit Time.

Because sluggish small intestinal motility increases the efficiency of cholesterol absorption,22, 23 we explored whether deletion of Abcg8 alone influences small intestinal transit rate according published methods.13 Because loss of ABCG8 function could result in a significant increase in intestinal absorption of radiolabeled sitostanol, we used [51Cr]sodium (i.e., Na2[51Cr]O4) as a nonabsorbable reference marker in this study. Small intestinal transit time was calculated using the following equation:

geometric center = Σ (fraction of [51Cr] per segment × segment number).

Statistical Methods.

All data are expressed as the mean ± SD. Statistically significant differences between groups of mice were assessed via Student t test or Mann-Whitney U test. If an F value was significant, comparisons between groups of mice were further analyzed using a multiple comparison test. Analyses were performed with SuperANOVA software (Abacus Concepts, Berkeley, CA). Statistical significance was defined as a 2-tailed P value of <0.05.


Effects of Abcg8 on Absorption and Lymphatic Transport of Cholesterol and Sitostanol.

Because differences in biliary lipid outputs can have marked effects on intestinal cholesterol absorption,14 we studied mice with chronic biliary fistulae but in the setting of intraduodenal infusion of model bile including trace amounts of radiolabeled cholesterol and sitostanol. This scheme prevents variations in bile flow and quantity of natural bile secretion from confounding the determination of intestinal absorption of the dietary sterols. Intraduodenal infusion of the micellar lipid solution produced a steady, continuous lymph flow that was constant (250-290 μl/h) and similar in Abcg8−/− and wild-type mice over the 12-hour experimental period (Fig. 1A). However, at 12 hours after instillation of [14C]cholesterol, cumulative radioactivities in lymph were 56 ± 9% in Abcg8−/− mice (Fig. 1B), significantly (P < 0.01) higher than those in wild-type mice (40 ± 4%). Cumulative (12-hour) radioactivities of sitostanol in the lymph of Abcg8−/− mice (25 ± 3%) were significantly (P < 0.00001) enhanced by greater than 6-fold compared with the values in wild-type mice (4 ± 0.4%).

Figure 1.

Comparison of the lymphatic transport of cholesterol and sitostanol between Abcg8−/− and wild-type mice with chronic biliary fistulae and in the setting of constant intraduodenal infusion of 16 mM taurocholate, 4 mM lecithin, and trace amounts of radiolabeled cholesterol and sitostanol (model “bile”) mixed with medium-chain triglyceride. (A) Lymph flow rates and (B) cumulative radioactives in lymph.

Table 1 shows that there were no differences in food or calorie consumption between the two groups. On the chow diet, biliary cholesterol outputs per diem in Abcg8−/− mice were significantly reduced compared to wild-type mice. Furthermore, daily fecal total neutral steroid excretion was significantly lower in Abcg8−/− mice than in wild-type mice. An input–output analysis indicates that the daily cholesterol mass absorbed by Abcg8−/− mice was significantly higher than in wild-type mice. The calculated percent cholesterol absorption was significantly higher in Abcg8−/− mice than in wild-type mice, consistent with the results measured by the lymph fistula method (Fig. 1B).

Table 1. Cholesterol Mass Balance Data in Chow-Fed Abcg8−/− and Wild-Type Mice During Metabolic Steady State Conditions
MouseCholesterol Intake (mg/day)Biliary Cholesterol (mg/day)Steroid Excretion (mg/day)Absorbed Cholesterol (mg/day)Cholesterol Absorption (%)
  • *

    P < 0.001,

  • P < 0.0001,

  • P < 0.05, and

  • §

    P < 0.01 compared with wild-type mice.

Mean ± SD0.81 ± 0.032.14 ± 0.102.58 ± 0.100.37 ± 0.0545 ± 5
Mean ± SD0.82 ± 0.031.30 ± 0.08*1.60 ± 0.080.52 ± 0.0963 ± 10§

Regulation of Intestinal Sterol Efflux Transporters by LXRα.

Figure 2 shows that feeding T0901317 to wild-type mice increases expression levels of Lxrα (Fig. 2A), Abcg5 (Fig. 2B), and Abcg8 (Fig. 2C) significantly in the jejunum and ileum and to a lesser extent in the duodenum. These data confirm previous observations.17 In contrast, the relative messenger RNA levels for Lxrβ (data not shown) were very low in the mouse small intestine, thus functionality was not explored further. Additionally, the sterol balance studies reveal that fecal total neutral steroid excretion (Fig. 2D) was significantly increased—and percent cholesterol absorption (Fig. 2E) was significantly reduced—by the synthetic LXR agonist. In contrast, in Abcg8−/− mice, LXR agonist treatment up-regulated expression levels of intestinal Lxrα (Fig. 2A), but did not influence the relative messenger RNA levels for intestinal Abcg5 and Abcg8, fecal total neutral steroid excretion, or percent cholesterol absorption (Fig. 2B-E). In line with our results, it has been shown that increasing LXR agonist doses promotes fecal neutral steroid excretion markedly in wild-type mice but not so in Abcg5−/− and Abcg5−/−/Abcg8−/− mice.12, 24 These observations indicate that the intestinal sterol hemi-transporter ABCG8 (with its partner ABCG5) constitute a critical efflux pump for cholesterol from the small intestinal enterocyte into the lumen for fecal elimination.

Figure 2.

Experiments establishing the role of the “LXRα-ABCG8 control loop” in regulation of intestinal cholesterol absorption. Expression levels of (A) Lxrα, (B) Abcg5, and (C) Abcg8, as well as (D) fecal neutral steroid excretion and (E) percent cholesterol absorption.

Sterol Uptake Pattern by the Small Intestine.

At 45 minutes after dosing, uptake of radiolabeled cholesterol (Fig. 3A) and sitostanol (Fig. 3B) by the enterocytes was comparable in both wild-type and knockout mice in the setting of intact biliary lipid secretion. Furthermore, the small intestine absorbed significantly (P < 0.001) more [14C]cholesterol than [3H]sitostanol, regardless of whether the Abcg8 gene was deleted or not. These findings support the concept that the topologic uptake mechanism is also selective for entry of sterols into enterocytes. The present results parallel data observed in other inbred strains of mice with normal expression of the intestinal Abcg5 and Abcg8 genes.17

Figure 3.

Comparison of acute uptake rates of radiolabeled (A) cholesterol and (B) sitostanol by the small intestine, assayed 45 minutes after dosing.

Expression of Intestinal Sterol Transporters.

Figure 4A shows that the expression levels of the intestinal sterol influx transporter Npc1l1 is essentially similar between Abcg8−/− and wild-type mice, consistent with identical acute uptake rates of cholesterol and sitostanol by the enterocyte in these mice (Fig. 3). In addition, expression levels of Sr-b1 (Fig. 4B) and Abca1 (Fig. 4C) in the small intestine were comparable in the 2 groups of mice. Moreover, deletion of the Abcg8 gene did not influence expression levels of the gene that encodes the fatty acid transport protein-4 (FATP4) in the intestine, which quantitively controls dietary triglyceride assimilation, a parameter that is basically similar in both groups of mice (Fig. 4D).

Figure 4.

Relative expression levels of (A) Npc1l1, (B) Sr-b1, (C) Abca1, and (D) Fatp4 in the small intestine.

Effects of Abcg8 on Small Intestinal Motility.

Small intestinal transit time can be a powerful factor in determining absorption of intestinal sterols.23 Figure 5 shows the identical luminal distributions of [51Cr]sodium along the small intestine of Abcg8−/− and wild-type mice. Mean values for the geometric centers of the distribution profiles of radioactivity were 11.2 ± 1.7 and 12.7 ± 2.2 for wild-type and Abcg8−/− mice, respectively. These results indicate that loss of ABCG8 function does not influence small intestinal transit times.

Figure 5.

Lack of effect on small intestinal motility by deletion of the Abcg8 gene. Arrows indicate the geometric centers in (A) wild-type mice (11.2 ± 1.7) and (B) Abcg8−/− mice (12.7 ± 2.2), suggesting that small intestinal transit times are similar in both groups of mice.

Effects of Abcg8 on Biliary Secretion of Both Sterols.

In wild-type mice (Fig. 6A), the HDL-derived radioactivity of sitostanol appeared rapidly in bile and reached its highest values, with the cumulative percentage of the injected dose being 5.2 ± 0.7 at 1 hour after injection. In contrast, radioactivity of biliary cholesterol was significantly (P < 0.05) less compared with that of sitostanol. Subsequently, secretion rates of both [3H]sitostanol and [14C]cholesterol decreased gradually over the final 4 hours of bile collection. Furthermore, disruption of Abcg8 induced a significant decrease in the HDL-derived radioactivity of cholesterol in bile with peak values appearing 2 hours after injection. Thereafter, secretion rates of [14C]cholesterol in Abcg8−/− mice diminished over the final 4 hours. No sitostanol radioactivity was recovered in bile of Abcg8−/− mice over the 6-hour period of biliary secretion experiments, confirming the vital role of ABCG5/ABCG8 in phytosterol secretion by the liver. In wild-type mice, the total sitostanol radioactivity (21.7 ± 1.9%) recovered in bile at 6 hours is significantly (P < 0.001) higher than that of [14C]cholesterol (14.9 ± 1.5%) (Fig. 6B). Deletion of the Abcg8 gene alone significantly (P < 0.0001) reduced hepatic secretion of [14C]cholesterol (5.6 ± 0.7%) and completely abolished hepatic secretion of [3H]sitostanol. At the end of the study, we measured the distributions of [14C]cholesterol and [3H]sitostanol in plasma, liver, bile, and carcass, and found that radioactivities of sitostanol in plasma and liver and radioactivities of cholesterol in liver were significantly higher in Abcg8−/− mice compared with wild-type mice (Fig. 7). At 6 hours after intravenous injection, 14.1 ± 1.4 nmol of [14C]cholesterol was transferred into bile from radiolabeled plasma HDL (a 94.3-nmol dose) in wild-type mice. This is significantly (P < 0.001) higher than the 5.3 ± 0.6 nmol obtained in Abcg8−/− mice. Moreover, in wild-type mice, 108.5 ± 9.5 pmol of [3H]sitostanol was transferred from radiolabeled plasma HDL into bile from a 500-pmol loading dose. Under similar experimental conditions, no radiolabeled sitostanol was detected in bile of Abcg8−/− mice.

Figure 6.

Transport rates of (A) radiolabeled cholesterol and sitostanol from plasma HDL into bile and (B) cumulative radioactivity recovered in bile. Of special note is that no radiolabeled sitostanol is detected in bile of Abcg8−/− mice.

Figure 7.

Percent distributions of (A) [14C]cholesterol and (B) [3H]sitostanol in plasma, liver, collected bile, and carcass at the end of a 6-hour biliary secretion study. Abbreviation: N.D., not detected.


In this study, to exclude the effect of biliary lipids on intestinal sterol absorption, we studied intestinal absorption of cholesterol and sitostanol by directly measuring their lymphatic transport in Abcg8−/− and wild-type mice with chronic biliary fistulae but in the setting of a constant intraduodenal infusion of model bile. We observed that cumulative radioactivities of sitostanol and cholesterol at 12 hours after duodenal instillation are significantly increased in lymph of Abcg8−/− mice compared with wild-type mice. Additionally, the significant, steady state increase in intestinal cholesterol absorption in Abcg8−/− mice compared with wild-type mice was confirmed by cholesterol mass balance analyses—an independent, validated, and accurate benchmark method.13, 14 Furthermore, expression levels of the intestinal sterol influx transporter Npc1l1 and acute uptake rates of cholesterol and sitostanol by the small intestine are basically similar in the 2 groups of mice, suggesting that these pathways are not affected by loss of ABCG8 function. These results show that ABCG8 promotes efflux of both cholesterol and sitosterol from the enterocyte into the intestinal lumen for excretion in feces. Because ABCG8 functions as a heterodimer, we would predict that Abcg5−/− mice would show a phenotype very similar to that described here. Furthermore, we found that there is an LXRα–ABCG8 positive feedback loop for the regulation of cholesterol absorption, because in response to a synthetic LXR agonist, deletion of the Abcg8 gene results in loss of its regulatory function on intestinal cholesterol absorption.

It is appropriate to emphasize that compared with wild-type mice, the net absorption efficiency of cholesterol in Abcg8−/− mice is increased by ≈40% (from 40 ± 4% to 56 ± 9%); however, the intestinal absorption efficiency of sitosterol is inordinately increased by ≈500% (from 4 ± 0.4% to 25 ± 3%). These results are consistent with the observations that plasma levels of sitosterol (18.5 ± 5.0 mg/dl) and campesterol (8.8 ± 2.1 mg/dl) in Abcg8−/− mice are significantly higher than the values (sitosterol = 0.5 ± 0.5 mg/dl and campesterol = 1.1 ± 0.7 mg/dl) in wild-type mice.11 In contrast, the plasma cholesterol level (35.0 ± 6.0 mg/dl) in Abcg8−/− mice is significantly lower than the value (73.0 ± 14.0 mg/dl) in wild-type mice.11 Furthermore, our studies indicate that because both sterols use the NPC1L1 influx transporter, there is a selective regulatory mechanism distinguishing between cholesterol and sitostanol at the enterocyte level under normal physiological conditions. This mechanism displays exquisite specificity, although the structure of sitostanol differs from cholesterol solely in the ethylation of its side chain. Our results suggest that intestinal ABCG8 together with its partner ABCG5 plays a more important regulatory role in sitostanol absorption than in cholesterol absorption. In a survey of multiple inbred strains of mice, a negative correlation was found between cholesterol absorption efficiency and expression levels of Abcg5 and Abcg8 in jejunum and ileum but not in duodenum.17 This attests to the concept that variability in intestinal Abcg5 and Abcg8 expression and activity is an important determinant of overall differences in cholesterol absorption efficiency among murine strains. Furthermore, the essential importance of small intestinal transit rates as a determining factor of cholesterol absorption often ignored in prior investigations, has been validated in previous reports of human studies22 and in mouse studies by us.23 We have now used a highly accurate methodology to verify that loss of ABCG8 function does not influence small intestinal transit rates, indicating that this variable is not responsible for differences in intestinal absorption efficiency of sterols in Abcg8−/− mice. Overall, our broad experimental approaches indicate that ABCG8 with its partner ABCG5 provides a barrier to cholesterol and sitostanol accumulation in the body by promoting partial efflux of cholesterol and nearly complete efflux of sitostanol from the enterocyte into the intestinal lumen for fecal elimination.

Also, we observed that the transport efficacy of cholesterol from plasma HDL into bile is partially reduced, whereas that of sitostanol is blocked totally in Abcg8−/− mice. Similar observations have been reported in patients with sitosterolemia.2, 3 Furthermore, as noted before in human studies,2, 3 sitostanol transport from plasma into bile is significantly faster than that of cholesterol, just as we found in wild-type mice. Moreover, the total amount of hepatic sitostanol secreted in healthy mice over the 6-hour period of biliary experimentation is significantly higher than that of cholesterol. These observations indicate that ABCG8 with its partner ABCG5 plays a critical role in hepatic secretion of biliary sitostanol by facilitating the rate-limiting step in its transport from plasma to bile. As a result, most sitostanol absorbed from the intestine can be rapidly cleared from the circulation, avoiding any appreciable accumulation in the body. Conclusive evidence supporting the liver's key role in maintaining low body pools of phytosterols is the successful elimination of these sterols to near-normal levels in a sitosterolemic patient who received a healthy liver transplant.25 However, we found that deletion of Abcg8 alone significantly reduces but does not eliminate hepatic secretion of biliary cholesterol. More recently, we observed that gallbladder biles of Abcg8−/− and Abcg5−/−/Abcg8−/− mice can become lithogenic in response to a lithogenic diet, inducing cholesterol-supersaturated bile as well as precipitating cholesterol monohydrate crystals and forming gallstones.26 These findings strongly support the notion that there is an ABCG5/ABCG8-independent pathway for hepatobiliary secretion of cholesterol, at least in mice. A similar suggestion has also been reported for mice deficient in Abcg5 alone.27

Intestinal absorption of cholesterol in patients with sitosterolemia is increased by ≈30% (from ≈46% to ≈60%) and intestinal absorption of sitosterol is increased by ≈800% (from <5% to ≈45%) as measured by plasma dual-isotope labeling methods.2, 3, 28, 29 These increments in humans are consistent with the present studies in Abcg8−/− mice as determined by direct measurements of the absorption and lymphatic transport of these sterols. Additionally, the 60% reduction in biliary cholesterol output in the present study is in agreement with the ≈65% decrease in Abcg8−/− mice as measured using a biliary washout technique11 and by an ≈55% decrease in mice with deletion of Abcg5 alone.27 Although reduced cholesterol concentrations and absence of detectable phytosterols in bile were noted in sitosterolemic patients, biliary sterol secretion studies were not performed in these patients.2, 3 Our quantitative analysis of murine biliary sterol secretion indicates that an ABCG5/ABCG8-dependent pathway has a major effect (55%-65%) on the regulation of hepatic cholesterol secretion and that at least the remaining (35%-45%) biliary cholesterol secretion by the murine liver is through an ABCG5/ABCG8-independent transporter. Taken together, our findings strongly support the notion that ABCG5 and ABCG8 may each have symmetrical and equal roles in controlling biliary secretion and intestinal absorption of cholesterol and phytosterols in mice. These and other studies in the literature should prove important experimental models for exploring the assimilation and disposition of body pools of cholesterol and noncholesterol sterols in humans as well as revealing why mutations in either of ABCG5 and ABCG8 induce the same syndrome in patients with sitosterolemia.