Polyunsaturated Fatty Acids and Acetoacetate Downregulate the Expression of the ATP-Binding Cassette Transporter A1
- Yoshinari Uehara123,
- Thomas Engel2,
- Zhengchen Li2,
- Christian Goepfert1,
- Stephan Rust2,
- Xiaoqin Zhou2,
- Claus Langer2,
- Christian Schachtrup4,
- Johannes Wiekowski2,
- Stefan Lorkowski2,
- Gerd Assmann12 and
- Arnold von Eckardstein123
- 1Institute of Clinical Chemistry and Laboratory Medicine, Central Laboratory, Westphalian Wilhelms-University, Münster, Germany
- 2Institute of Arteriosclerosis Research, Münster, Germany
- 3University Hospital Zürich, Institute of Clinical Chemistry, Zürich, Switzerland
- 4Department of Biochemistry, University of Münster, Münster, Germany
Low HDL cholesterol is a frequent cardiovascular risk factor in diabetes. Because of its pivotal role for the regulation of HDL plasma levels, we investigated in vivo and in vitro regulation of the ATP-binding cassette transporter A1 (ABCA1) by insulin and metabolites accumulating in diabetes. Compared with euglycemic control mice, ABCA1 gene expression was severely decreased in the liver and peritoneal macrophages of diabetic mice. Treatment with insulin restored this deficit. Incubation of cultivated HepG2 hepatocytes and RAW264.7 macrophages with unsaturated fatty acids or acetoacetate, but not with insulin, glucose, saturated fatty acids, or hydroxybutyrate, downregulated ABCA1 mRNA and protein. The suppressive effect of unsaturated fatty acids and acetoacetate became most obvious in cells stimulated with oxysterols or retinoic acid but was independent of the expression of the thereby regulated transcription factors liver-X-receptor α (LXRα) and retinoid-X-receptor α (RXRα), respectively. Unsaturated fatty acids and acetoacetate also reduced ABCA1 promotor activity in RAW264.7 macrophages that were transfected with a 968-bp ABCA1 promotor/luciferase gene construct. As the functional consequence, unsaturated fatty acids and acetoacetate inhibited cholesterol efflux from macrophages. Downregulation of ABCA1 by unsaturated fatty acids and acetoacetate may contribute to low HDL cholesterol and increased cardiovascular risk of diabetic patients.
Numerous clinical and epidemiological studies have demonstrated the inverse association between HDL cholesterol and the risk of coronary heart disease events (1). It is controversially discussed whether this relationship is causal or only an epiphenomenon of a more general atherogenic disorder. On the one hand, HDL exerts various potentially antiatherogenic properties. For example, HDL particles transport cholesterol from cells of the arterial wall to the liver and to steroidogenic organs, in which cholesterol is used for the synthesis of bile acids, lipoproteins, vitamin D, and steroid hormones (1). On the other hand, low HDL cholesterol is frequently found as a component of the metabolic syndrome in many populations, i.e., together with overweight or obesity, glucose intolerance or overt diabetes, hypertriglyceridemia, and hypertension, which by themselves contribute to the pathogenesis of atherosclerosis (2). Moreover, many individuals with low HDL cholesterol have elevated fasting plasma levels of insulin, are resistant to exogenous insulin in euglycemic clamp studies, and bear an increased risk for future type 2 diabetes (3–5). The pathogenesis of low HDL cholesterol in insulin resistance is not well understood but may provide an important key to answer the question of whether low HDL cholesterol is a causal factor in the pathogenesis of atherosclerosis in patients with insulin resistance or diabetes.
ATP-binding cassette transporter A1 (ABCA1) has previously been identified as a pivotal gene in the regulation of both HDL cholesterol plasma levels and the cellular cholesterol homeostasis, which is defective in patients with Tangier disease (1,6–9). In these patients and their heterozygous relatives, mutations in the ABCA1 gene cause gene dosage-dependent decreases in plasma levels of HDL cholesterol and in the capacity of skin fibroblasts and monocyte-derived macrophages to release cholesterol in the extracellular presence of apolipoproteins (1,6–9). As the clinical result, homozygous patients with Tangier disease accumulate macrophage-derived foam cells in various tissues and develop tonsil anomalies, hepatosplenomegaly, premature atherosclerosis, and peripheral neuropathy (10).
Cyclic AMP and ligands of the nuclear transcription factors liver-X-receptor α (LXRα) and retinoid-X-receptor α (RXRα), i.e., oxysterols and retinoids, respectively, have been identified as enhancers of ABCA1 gene expression (11–14). By contrast, interferon-γ (IFNγ) downregulates ABCA1 gene expression (15). Here we investigated the regulation of ABCA1 in a mouse model of diabetes. Since ABCA1 gene expression was substantially decreased in the liver and peritoneal macrophages of diabetic mice, we next investigated the effects of insulin as well as metabolites accumulating in diabetes on ABCA1 gene expression. Thereby, we identified free polyunsaturated fatty acids (PUFAs) and acetoacetate as strong suppressors of ABCA1 gene expression.
RESEARCH DESIGN AND METHODS
Palmitic acid, oleic acid, eicosapentaenoic acid (EPA), arachidonic acid, stearic acid, linoleic acid, 22(R)-hydroxycholesterol (22R-HC), 9-cis-retinoic acid (9-cis-RA), acetoacetate, and β-hydroxybutyrate were purchased from Sigma. Insulin was from Hoechst Marion Roussel. ApoA-I was isolated from delipidated HDL by fast-performance liquid chromatography as previously described (16)
The human hepatocyte cell line HepG2 and the murine macrophage cell line RAW264.7 (DSMZ; Braunschweig) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FCS (GIBCO-BRL). For all experiments, cells were maintained in serum-free DMEM containing 10 g/l BSA and was supplemented with additives (22R-HC, 9-cis-RA, insulin, glucose, fatty acids, or ketone bodies) at concentrations and time periods described in the results section and in the figure legends.
All animal experiments were approved by the local Animal Care and Usage Committee and the District Governor of Münster. Mice were caged in groups of three to four animals in a 12-h dark rhythm. They were fed a regular chow diet (Furina Altromin 1034; Lage, Germany). At the age of 10 weeks, 10 mice were made diabetic by intraperitoneal injection of 200 mg streptozotocin dissolved in citric buffer (pH 4.0) (17). The manifestation of diabetes was indicated by blood glucose levels >22 mmol/l (>400 mg/dl). After 2 weeks of being diabetic, the animals were killed by inhalation of CO2. Blood was obtained immediately thereafter by intracardial or intra-aortal puncture. Peritoneal macrophages and livers were obtained by lavage and careful preparation, respectively, and stored in liquid nitrogen. In some experiments the diabetic mice received intraperitoneal injections of PBS or 14 IU insulin per 100 g/body wt as well as subcutaneous injections of PBS or 14 IU insulin per 100 g/body wt every 6 h before being killed.
RNA isolation and RT-PCR.
Total RNA was extracted from cells of 9.6-cm2 flasks using the RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol and treated with DNase to remove residual contaminations with DNA. The RNA was reverse transcribed and the cDNA amplified by PCR using the Superscript II reverse transcriptase, Oligo-dT primers (Invitrogen) and Hotstar Taq-DNA polymerase (Qiagen). Sequences of the primers are listed in Table 1. After amplification (activation at 95°C for 15 min, denaturing at 94°C for 0.5 min, primer annealing at 60°C for 0.75 min, and chain elongation at 72°C for 0.75 min) through 16–35 cycles, the whole volume of each PCR product was separated by electrophoresis in 3% agarose gel containing 0.5 μg/ml ethidium bromide. Bands were visualized and photographed under ultraviolet light. Alternatively, ABCA1 gene expression was quantified by real-time PCR using the Applied Biosystems prism model 7700 sequence detection instrument (Weiterstadt, Germany). The amplification was detected using SYBRGreen as the fluorogenic probe specific for double-stranded DNA. Threshold cycle, Ct, which correlates inversely with the target mRNA levels, was measured as the cycle number at which the reporter fluorescent emission increases above threshold level using the manufacturer’s settings. The mRNA levels for ABCA1 were corrected for GAPDH.
A firefly luciferase reporter construct was generated by amplifying a 968-bp fragment of human ABCA1 promoter (–803 to +165 bp) (18) with KpnI and XhoI restriction sites out of genomic DNA. PCR product was digested with KpnI and XhoI, gel purified, and ligated to KpnI-XhoI-cut pGL3 Basic vector (Promega, Madison, WI). For transfection, 50–80% confluent RAW264.7 cells were grown in DMEM supplemented with 10% FCS, 100 units/ml penicillin G, and 100 mg/ml streptomycin. Cells were transiently transfected with 10 μg of the firefly luciferase reporter plasmids and cotransfected with 1 μg pRL-SV40 vector (Promega) for the control Renilla luciferase by electroporation at 950 μF and 220 V using 0.4 ml of 2 × 107 cells/ml per transfection. The cells from one electroporation were seeded into eight wells of a 24-well dish. After transfection, the cells were incubated with 22(R)-HC, 9-cis-RA, fatty acids, and/or acetoacetate for 18 h as indicated. The luciferase activities were measured using the dual luciferase assay system (Promega). The firefly luciferase activity was divided by the renilla activity to obtain a normalized value, the relative luciferase activity.
Immunoprecipitation and Western blotting of ABCA1.
Cells were washed in PBS and lysed in lysis buffer containing 1% (wt/vol) Triton-X 100 (Sigma), 10 mmol/l Tris (pH 7.4), 150 mmol/l sodium chloride, 0.5% (wt/vol) sodium deoxycholic acid, 50 mmol/l 2-mercaptoethanol, 10 μg/ml pepstatin A (Sigma), 10 mmol/l PMSF, and protease inhibitor cocktail (Roche, Mannheim, Germany) for 1 h on a rotating platform at 4°C. Then, 40 μg protein G agarose beads (Pierce, Rockford, IL) were added to 12 ml of the cell lysate. After incubation for 2 h on a rotating platform, protein G beads were separated by centrifugation at 13,000 rpm for 20 s. The cell lysate was then incubated with rabbit anti-ABCA1 antibody for 2 h at 4°C. After addition of 10 μg protein G agarose beads the cell lysates were incubated for additional 2 h. Immune complexes were precipitated by centrifugation at 13,000 rpm for 20 s and washed three times with Triton-Tris-buffered saline (50 mmol/l Tris/HCl pH 7.4, 150 mmol/l NaCl, and 1% Triton-X 100) and PBS. Immune precipitates were denatured by boiling for 5 min in a buffer with 20 g/l SDS and 100 mmol/l mercaptoethanol. Proteins were separated by SDS-PAGE and transferred onto a nitrocellulose membrane by electroblotting. Membranes were blocked by incubation for 1 h in a buffer with 50 mmol/l Tris, 150 mmol/l NaCl, 0.05% Tween 20, and 5% milk. ABCA1 was probed with an antibody that was raised in rabbits by immunization with a recombinant ABCA1 fragment encompassing amino acid residues 1,371 through 1,657 (T.E., unpublished observations). The immunoreaction was visualized after incubation of the nitrocellulose sheets with secondary horseradish peroxidase-conjugated anti-rabbit IgG antibodies (Dako) using a chemiluminescent substrate (Lumi-light; Roche Molecular Biochemicals) and a photo imaging system (Raytest, Straubenhardt, Germany).
At the state of near confluence, RAW264.7 cells were washed with PBS and incubated for 20 h at 37°C with 2,5 μCi [1,2-3H]-cholesterol (51.7 μCi/mmol, New England Nuclear), which were complexed with BSA. After radiolabeling, cholesterol pools were equilibrated for 15 h by incubation with serum-free DMEM/2 g/l BSA in the presence or absence of additives (oxysterol, fatty acids, or ketone bodies) as described in the results section. For cholesterol efflux, cells were washed with ice-cold PBS and incubated for 5 h with 300 μl DMEM containing 2 g/l BSA alone or 2 g/l BSA plus 50 μg apoA-I as well as the additives indicated in the results section. After incubation for 5 h, cells and media were harvested separately. Radioactivity was measured by scintillation spectrometry of the supernatant and of the cell monolayers after lysis with 500 μl 0.15 mol/l NaCl with 0.2 mol/l NaOH. Fractional cholesterol efflux was calculated as cpmmedium/(cpmmedium + cpmcells) × 100%.
Plasma concentrations of glucose, cholesterol, triglycerides, and HDL cholesterol were measured by the use of a Hitachi 917 autoanalyzer (Roche Diagnostics). Plasma levels of free fatty acids were determined by the use of reagents from Roche Diagnostics according to the manufacturer’s recommendations.
To test the possible contribution of dysregulated ABCA1 gene expression to low HDL cholesterol, we initially investigated the regulation of ABCA1 in mice that were made diabetic with streptozotocin (Table 2, Fig. 1). Compared with untreated mice, the liver (Fig. 1A) and peritoneal macrophages (Fig. 1B) of diabetic mice showed a marked decrease in the expression of ABCA1 mRNA. Intraperitoneal injection of insulin abolished the diabetic phenotype (Table 2) and restored the expression of ABCA1 in either liver (Fig. 1C) or peritoneal macrophages (Fig. 1D).
Our next experiments were aimed at the identification of the metabolic basis of the decreased ABCA1 gene expression in diabetes. We incubated HepG2 hepatocytes (Fig. 2A) or RAW264.7 macrophages (Fig. 2B) for 18 h with increasing dosages of insulin, glucose, fatty acids, or ketone bodies. Addition of insulin or glucose into the cell culture medium had no consistent effects on the expression of ABCA1 in either hepatocytes or macrophages. By contrast, the polyunsaturated EPA as well as the ketone body acetoacetate dose-dependently decreased the expression of ABCA1 in both cell lines (Figs. 2A and B). The suppressive effects of these fatty acids became most evident in those cells where ABCA1 expression was upregulated with an oxysterol, i.e., an activator of the nuclear transcription factor LXRα, which was previously shown to induce the ABCA1 gene (12,13). In HepG2 cells, the stimulatory effect of 22R-HC (12,13) on ABCA1 was inhibited by ∼80% (P < 0.01), 60% (P < 0.01), and 40% (P < 0.05) in the additional presence of physiological dosages of EPA (100 μmol/l), oleic acid (500 μmol/l), or acetotacetate (10 mmol/l), respectively (Fig. 3A). In RAW264.7 cells, the respective numbers were 95, 80, and 60% (all P < 0.01; Fig. 3B). Also, in the presence of 9-cis-RA, i.e., an activator of the nuclear transcription factor RXRα that forms heterodimers with LXRα and thereby also induces the ABCA1 gene (14), EPA, oleic acid, and acetoacetate suppressed ABCA1 gene expression in HepG2 and RAW by 50–60% (Figs. 3C and D, all significant at a level of P < 0.05 or P < 0.01, except acetoacetate in HepG2 cells). In none of these conditions did palmitic acid or hydroxybutyrate modulate ABCA1 gene expression (Fig. 3A–D). Likewise, and in agreement with data from Wang and Oram (19), we did not see any suppressive effect of fatty acids or acetoacetate on ABCA1 gene expression in macrophages that were stimulated with Br-cAMP (data not shown).
We verified the stimulatory effects of 22(R)-HC and 9-cis-RA as well as the inhibitory effects of EPA and acetoacetate in a luciferase reporter gene assay, which was performed on a construct that contained 0.968 kb of the human ABCA1 promotor (Fig. 4). As expected from previous reports (12–14), 22R-HC and 9-cis-RA increased the ABCA1 promotor activity by a factor up to 30 (Figs. 4A and B). EPA and acetoacetate suppressed these stimulatory effects dose dependently by up to 80% (EPA, Figs. 4C and D) and 50% (acetoacetate, Figs. 4E and F), respectively. Using this reporter gene assay, we compared the suppressive effects of various fatty acid species at a concentration of 100 μmol/l. Linoleic acid was most effective (−80%), followed by arachidonic acid (−75%), EPA (−70%), and oleic acid (−40%). Palmitic acid had a modest but significant suppressive effect (−10%), whereas stearic acid exerted a moderate but also significant stimulatory effect (+10%, Fig. 5).
Our observations raised the question as to whether unsaturated fatty acids and acetoacetate regulate the expression of ABCA1 indirectly via regulation of LXRα or RXRα gene expression. Such a mechanism was previously demonstrated to be responsible for the upregulation of ABCA1 by agonists of peroxisome proliferator-activated receptor (PPAR)-α (i.e., fibrates) and PPARγ (i.e., glitazones) (20). Neither oleic acid nor EPA nor acetoacetate regulated the expression of LXRα or RXRα (data not shown).
Immunoprecipitation and subsequent Western blot analysis clearly detected ABCA1 in RAW264.7 cells (Fig. 6A), which were cultivated in the presence of an oxysterol. However, the cellular ABCA1 protein concentration was severely decreased when these cells were incubated in the additional presence of free fatty acids or acetoacetate (Fig. 6A). The functional relevance of ABCA1 downregulation by free fatty acids and ketone bodies was investigated in cholesterol efflux experiments (Fig. 6B). Lipid-free apoA-I induced significant cholesterol efflux from RAW264.7 cells. As reported previously, cholesterol efflux to apoA-I was stimulated by oxysterols (11–13). The stimulatory effect of oxysterols was completely inhibited by 500 μmol/l oleic acid, 100 μmol/l EPA, and 10 mmol/l acetoacetate (Fig. 6B).
Interestingly, the expression of another important gene involved in cholesterol efflux, scavenger receptor BI, was not affected by either fatty acids or acetoacetate (data not shown).
Here we have demonstrated that ABCA1 is tremendously downregulated in an animal model of diabetes (Fig. 1). As the most likely explanation, we found suppressive effects of PUFAs, oleic acid, and acetoacetate on ABCA1 gene expression in vitro (Figs. 2–5). Interestingly, the saturated palmitic and stearic acids as well as the ketone body 3-hydroxybutyrate did not alter ABCA1 gene expression so that the inhibitory effects appear to be specific for unsaturated fatty acids and acetoacetate.
Suppression of ABCA1 gene expression by unsaturated fatty acids became most obvious in the additional presence of an oxysterol (Fig. 3) but did not occur in the presence of Br-cAMP (data not shown, 19). This raised the possibility that unsaturated fatty acids downregulate ABCA1 via suppression or inhibition of the nuclear transcription factor LXRα, which is activated by oxysterols and which upregulates ABCA1 (12,13,21). Free fatty acids were previously shown to interact with LXRα on both the transcriptional and the posttranslational level (22,23). Free fatty acids induce the transcription of the LXRα gene via activation of PPARs (22,24). However, unsaturated fatty acids do not suppress ABCA1 via transcriptional regulation of LXRα, since in our experiments oleic acid and EPA did not modulate the expression of LXRα (data not shown) and since PPARα, PPARγ, and PPARδ rather upregulate ABCA1 via stimulation of LXRα transcription (14,25). On the posttranslational level, unsaturated fatty acids antagonize the activation of LXRα by oxysterols and thereby inhibit the transcription of the sterol regulatory element-binding protein-1c (23,26). This posttranslational inhibition of LXRα well explains the transcriptional suppression of ABCA1 by unsaturated fatty acids. Interestingly, PUFAs turned out as more suppressive than the monounsaturated oleic acid, whereas saturated fatty acids exerted only moderate or even no suppressive effects at all. This order is in agreement with the effects of dietary fatty acids on HDL cholesterol. Compared with habitual Western diets being rich in saturated fatty acids, diets rich in PUFAs substantially lower HDL cholesterol, whereas diets rich in oleic acid have only a little HDL cholesterol-decreasing effect (27).
The suppressive effect of acetoacetate on ABCA1 gene expression was more obvious in macrophages than in hepatocytes (Figs. 3–5). Like PUFAs and oleic acid, acetoacetate did not suppress the expression of LXRα or RXRα. Downregulation of ABCA1 by acetoacetate is, to the best of our knowledge, the first example of a gene regulatory effect exerted by this ketone body. Previous in vivo and in vitro studies demonstrated that ketoacidosis decreases hepatic apoA-I gene expression in rats (28). However, this effect was linked to low pH rather than to ketone bodies (28). In vitro we also observed that acidosis suppresses ABCA1 gene expression (data not shown); however, in our experiments the addition of acetoacetate did not acidify the cell culture medium. Moreover, in those experiments, suppression of apoA-I was accomplished by butyrate (28), which did not suppress ABCA1 in our experiments. Thus, the suppressive effect appears to be specific for acetoacetate. Our data do not allow any conclusion of whether acetoacetate exerts the inhibitory effect directly by itself or indirectly by a metabolite or a signal elicited thereof. It is unlikely that the regulatory effect of acetoacetate is exerted by a metabolite because ketolysis is a hallmark of heart and skeletal muscle as well as kidney and brain, which in contrast to liver and macrophages have a high activity of the rate-limiting enzyme in ketolysis, i.e., succinyl-CoA-oxoacid transferase (29,30). Because acetoacetate but not β-hydroxybutyrate downregulates the expression of ABCA1 and because the suppressive effect of acetoacetate was more obvious in macrophages than in hepatocytes, it is interesting to note that acetoacetate by contrast to β-hydroxybutyrate generates oxygen radicals in cells (31). Because oxygen radicals are known to regulate gene expression via various signaling pathways (32), it will be important to investigate the effects of oxygen radicals on the expression of ABCA1.
Our findings contribute to the understanding of the pathogenesis of the low HDL cholesterol syndrome in patients with latent or manifest type 2 diabetes. Insulin was previously shown to stimulate hepatic apoA-I gene expression (33,34), to inhibit hepatic VLDL production (35,36), and to stimulate the release of lipoprotein lipase from adipose tissue (37). Insulin resistance hence reduces the hepatic production of nascent HDL and induces hypertriglyceridemia, which secondarily causes low HDL cholesterol levels via disturbed release of surface remnants (1,2,38) and by enhancing the exchange of triglycerides from VLDL against cholesteryl esters from HDL (1,2,39). Free fatty acids were previously shown to downregulate the expression of apoA-I (40) and to upregulate the expression of cholesteryl ester transfer protein (41). In addition, we have demonstrated here that unsaturated fatty acids and acetoacetate suppress the expression of ABCA1 in liver and in macrophages. By the latter effect, hyperacylemia and ketosis in diabetes and insulin resistance will probably not only interfere with the formation of HDL but also facilitate lipid accumulation in vascular macrophages.
Patients with type 1 diabetes have increased cardiovascular risk but normal or even slightly elevated levels of HDL cholesterol (42). This is, at first sight, in partial contrast to our finding of suppressed ABCA1 expression in liver and macrophages of streptozotocin-induced diabetic mice (i.e., a model for type 1 diabetes) as well as in cultivated cells treated with unsaturated fatty acids and acetoacetate (which are also elevated in patients with type 1 diabetes). However, patients with type 1 diabetes have abnormal HDLs, which are enriched with triglycerides so that in vivo the HDL-decreasing effect of low ABCA1 expression may be counterbalanced by other deregulations of HDL metabolism (42). In addition, low HDL cholesterol is the most frequent dyslipidemia in patients with type 1 diabetes, which becomes manifest especially in patients with insufficient euglycemic control (43). Thus, in patients with type 1 diabetes, HDL cholesterol decreases with increasing blood levels of glycated hemoglobin and increasing albuminuria (44,45), which both are associated with acylemia and ketosis.
In conclusion, here we have provided strong evidence that ABCA1 is suppressed in diabetes by unsaturated long chain fatty acids and acetoacetate. This may be an important etiological factor for the increased cardiovascular risk of patients with insulin resistance or overt diabetes by lowering HDL cholesterol and by impairing cholesterol efflux from vascular cells.
This work is supported by an International HDL Research Award from Pfizer and by grants from the European Commission (BMH4-CT98-36 and QLG1-1999-01007).
We also acknowledge the excellent technical assistance of Isabel Schaukal and Kathrin Tkocz.
Address correspondence and reprint requests to Dr. Arnold von Eckardstein, University Hospital Zürich, Institute of Clinical Chemistry, Rämistrasse 100, CH 8091 Zürich, Switzerland. E-mail:.
Received for publication 30 April 2002 and accepted in revised form12 July 2002.
9-cis-RA, 9-cis-retinoic acid; 22R-HC, 22(R)-hydroxycholesterol; ABCA1, ATP-binding cassette transporter A1; DMEM, Dulbecco’s modified Eagle’s medium; EPA, eicosapentaenoic acid; LXRα, liver-X-receptor α; IFNγ, interferon-γ; PPAR, peroxisome proliferator-activated receptor; PUFA, polyunsaturated fatty acid; RXRα, retinoid-X-receptor α.