(A–C) Data from male WT and Abcg1^–/– mice, age 3 months. (D–F) Data from female WT and Abcg1^–/– mice, age 3 months. (A, D) Mice were fasted overnight, and blood glucose levels were measured (n = 8 for males, 12 for females). (B, E) Fasted mice were injected with glucose i.p., and blood glucose levels were monitored over 90 minutes (n = 8 for males, 6 for females). (C, F) Plasma insulin levels were measured at fasting, and 15 minutes after i.p. glucose injection (n = 12–14 for males, 5–13 for females). Data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

(A and B) Pancreata from WT and Abcg1^–/– mice were fixed, sectioned, and immunostained for insulin, and the total β cell area as determined by insulin-positive staining was quantified (n = 8). Scale bars: 500 μm. (C) Islets were isolated from WT and Abcg1^–/– mice, and glucose- and potassium-induced insulin secretions were measured in vitro (n = 9–22, groups of 50 islets). (D and E) Isolated islets were loaded with the calcium dye fura-2, and [Ca²⁺]_i changes in response to glucose perifusion were measured over time. (D) Representative curves of [Ca²⁺]_i oscillations from WT and Abcg1^–/– islets. (E) Average net increases in [Ca²⁺]_i from basal to glucose-stimulated (n = 33–43 islets). (F) Insulin content from isolated islets (n = 10). (G) mRNA was harvested from isolated islets, and qRT-PCR was performed for transcript levels of the unprocessed insulin pre-mRNA, the mature processed insulin mRNA (transcripts for both insulin genes I and II were captured), and the regulatory enzyme glucokinase (n = 5–10). (H and I) First- and second-phase insulin secretion from WT and Abcg1^–/– islets was measured by successive static incubations (first phase, 5 minutes in 30 mM KCl; second phase, 10 minutes in 28 mM glucose) (n = 10–11, groups of 50 islets). Data are presented as mean ± SEM. *P < 0.05; **P < 0.01.

(A) MIN6 cells were labeled with [³H]cholesterol overnight, and [³H]cholesterol efflux to extracellular acceptors over 4 hours was measured (n = 6). (B) Western blotting showing siRNA-mediated knockdown of ABCG1 in MIN6 cells (approximately 90% by densitometry). (C, D) MIN6 cells were treated with siRNA as in B and a time-course (C) and dose-response curve (D) of HDL-mediated [³H]cholesterol efflux was measured (n = 4; representative of 3 independent experiments). (E) ApoAI-mediated cholesterol efflux was measured in MIN6 cells treated with ABCG1 siRNA. (F) Islets were isolated from WT and Abcg1^–/– mice and total cholesterol (TC), free-cholesterol (FC), and cholesterol-ester (CE) content were measured by gas chromatography (150 islets per sample, n = 14 mice per group). (G) qRT-PCR was performed on isolated islets for cholesterol- (SREBP-2, LDLR, HMG-CoA R, HMG-CoA S) and oxysterol-regulated (ABCA1, SREBP-1c) mRNA (n = 5–10 mice per genotype). (H) Triglyceride content was measured in isolated islets from WT and Abcg1^–/– mice (200 islets per sample, n = 8 mice per genotype). LDLR, LDL receptor; HMG-CoA R, HMG-CoA reductase; HMG-CoA S, HMG-CoA synthase. Data are presented as mean ± SEM. ***P < 0.001.

(A) MIN6 cells were homogenized, organelles were separated by differential centrifugation, and fractions were blotted for ABCG1, the granule marker secretogranin (SG), and the plasma membrane (PM) marker Na/K-ATPase. Mito, mitochondria. (B, C) The granule fraction from A was subfractionated on 2 different continuous density gradients, iodixanol (B), and sucrose (C), and blotted for ABCG1; SG and insulin (granules); syntaxin 6 and VAMP4 (immature granules); SCAMP3 (TGN and endosomes); and LAMP1 (late endosome/lysosome). (D) Approximately 1500 mouse islets were homogenized and fractionated, with subfractionation of the granules on a discontinuous sucrose gradient, collected in 2 fractions: that which floated above 1.5 M sucrose (<1.5 M) and that which pelleted through 1.5 M sucrose (>1.5 M). PNS, post-nuclear supernatant; PMS, post-mitochondrial supernatant. (E) MIN6 cells were surface labeled with biotin and the relative amount of ABCG1 on the cell surface was quantified by Western blotting. The plasma membrane marker Na/K-ATPase and the cytoplasmic protein γ-adaptin are shown as positive and negative controls. Numbers over bars represent mean ± SEM (n = 6, 3 independent experiments). (F–K) MIN6 cells were costained for insulin (red) and ABCG1 (green) and imaged by confocal microscopy. Scale bars: 10 μm; 2 μm (insets). Panels F–H and I–K are from 2 different fields. ***P < 0.001.

(A–D) WT and Abcg1^–/– islets were isolated, and transmission electron microscopy was performed to analyze granule morphology. Scale bars: 1 μm. Panels C and D are enlarged from the highlighted areas in A and B. (E and F) Quantification of membrane-bound granule areas. (E) Mean granule area ± 95% CI. (F) Relative frequency distribution of granule areas (n granules: WT = 1945, –/– = 1654; n images: WT = 7, –/– = 6). (G) MIN6 cells were treated with control or ABCG1 siRNA. 60 hours following knockdown, cells were homogenized, and the postnuclear supernatant was subjected to velocity gradient centrifugation. (H and I) Measurements of granule cholesterol. MIN6 cells were treated with control or ABCG1 siRNA, then incubated with [³H]cholesterol overnight to label the granule population. Following incubation, granule fractions were isolated and subfractionated on continuous density gradients. Subfractions were assayed for [³H]cholesterol and phospholipid. (H) Total cholesterol content of the granule fraction; 4 independent experiments. (I) Western blotting for the immature granule marker VAMP4 and insulin in the subfractions. The plot below represents the [³H]cholesterol relative to phospholipid across the gradients; 2 independent experiments. (J) Assessment of relative plasma membrane cholesterol. MIN6 cells were treated with control or ABCG1 siRNA and labeled overnight with [³H]cholesterol. Cells were then cooled to 10°C to prevent membrane recycling, and surface accessible cholesterol was extracted with methyl-β-cyclodextrin. Extraction rate was monitored over 15 minutes and normalized to total cellular [³H]cholesterol (n = 8; 2 independent experiments). Data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

(A and B) MIN6 cells were treated with control or ABCG1 siRNA and 36 hours after treatment were incubated overnight with 20 μg/ml soluble cholesterol (+Chol-CD). (A) [³H]Cholesterol was included in the media and chol-CD, and granule [³H]cholesterol content relative to phospholipid was assayed as in Figure I. (B) Velocity gradient centrifugation of MIN6 cells with knockdown of ABCG1, treated overnight with chol-CD. (C–F) WT (C) and Abcg1^–/– islets (D) were treated overnight with chol-CD, and then transmission electron microscopy was performed to assess granule morphology. Scale bars: 1 μm. (E) Frequency distribution of granule areas from chol-CD–treated islets shown in C and D (n granules: WT = 2345, –/– = 2126; n images: WT = 6, –/– = 5). (F) Mean granule areas (± 95% CI) from vehicle (as shown in Figure ) and chol-CD–treated WT and Abcg1^–/– islets. (G) Glucose-stimulated insulin secretion in WT and Abcg1^–/– islets treated with vehicle or chol-CD (n = 11–14, groups of 50 islets; 3 independent experiments). All fractionation studies are representative of 2–3 independent experiments. Data are presented as mean ± SEM. **P < 0.01; ***P < 0.001.

(A–C) Islets were isolated from hyperglycemic db/db, normoglycemic db/db, and db/+ control littermates, and mRNA (A, C) and protein (B) were harvested (n = 10 [db/+]; 2 [db/db normoglycemic]; 5 [db/db hyperglycemic]). (A) qRT-PCR for ABCG1 and LXRβ, the main transcriptional regulator of ABCG1 in islets. (B) Western blot for ABCG1. db/db norm, normoglycemic; db/db hyper, hyperglycemic. (C) Linear regression analysis of the correlations between islet mRNA levels of ABCG1 and LXRβ, and random blood glucose. (D and E) Isolated mouse islets were cultured for 1, 3, or 5 days in either 25 mM D-glucose or the nonmetabolizable L-glucose control (5.5 mM D-glucose + 19.5 mM L-glucose), mRNA was isolated, and qRT-PCR was performed (n = 3, representative of 3 independent experiments). (D) Relative expression of LXRβ and ABCG1 after 5 days in culture. (E) Time course of ABCG1 expression in response to glucose in culture. (F) MIN6 cells were treated for 24 hours with 10 μM pioglitazone, and ABCG1 mRNA levels were measured (n = 6, from 2 independent experiments). (G) Isolated human islets from a cadaveric nondiabetic donor were treated for 24 hours with 5 μM pioglitazone, and ABCG1 mRNA levels were measured (n = 4). (H) Db/db mice were treated with either vehicle or pioglitazone (Pio-db/db, 20 mg/kg body weight) for 6 weeks, islets were isolated, and ABCG1 mRNA was measured. BLKS is the control background strain (n = 5, BLKS control; 3, db/db; 7, Pio-db/db). Data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

ABCG1 localizes to the regulated secretory pathway and acts to maintain secretory granule cholesterol, perhaps by redistributing granule membrane cholesterol to the inner leaflet, thereby regulating membrane curvature and restricting cholesterol exit via carrier-mediated diffusion to other cellular sites. This activity is analogous to that of other intracellular ABC transporters, which move their respective substrates into the lumen of the organelles in which they reside. In ABCG1 deficiency, cholesterol redistributes out of the granules; the granules are subsequently enlarged, and their release is impaired. SG, secretory granule; ISG, immature secretory granule, TGN, trans-Golgi network; CMD, carrier-mediated diffusion.