FATP1 Is an Insulin-Sensitive Fatty Acid Transporter Involved in Diet-Induced Obesity

Antibodies and reagents.

BODIPY (4,4-difluoro-3a,4a-diaza-s-indacene) fatty acid (C₁-BODIPY-C₁₂) was obtained from Molecular Probes, and [¹⁴C]oleic acid and 2-deoxy-d-[1-³H]glucose (2-DG) were purchased from American Radiolabeled Chemicals. NSC119889 and NSC154020 were from Calbiochem (San Diego, CA). Wortmannin, U0126, PP2, and Go 6850 were from Tocris Bioscience (Ellisville, MO). Polyclonal antisera against the C termini of FATP1 to -6 were raised as described previously (44), mouse monoclonal anti-Glut4 and anti-caveolin 3 were purchased from BD Pharmingen, and the mouse monoclonal anti-CD36/FAT antibody was a gracious gift from Maria Febbraio.

Skeletal muscle fractionation.

One gram of insulin-treated and untreated flash-frozen soleus tissue was thawed in 30 ml of ice-cold sucrose buffer (250 mM sucrose, 20 mM HEPES-Tris, 1 mM EDTA, 100 μM phenylmethylsulfonyl fluoride, pH 7.4), minced with scissors, and homogenized with a Polytron device (setting 3; 30 s). This homogenate was filtered through two layers of cheesecloth to remove residual connective tissue and centrifuged (3,000 × g for 10 min at 4°C). The pellet was saved for preparation of the plasma membrane-enriched fraction. The resulting pellet was washed twice, resuspended in 10 mM Tris-HCl (ratio of 30 ml to 1 g [wt/vol] of the original skeletal muscle), and homogenized using a Teflon pestle. The suspension was transferred to a glass beaker, a 9- by 25-mm-diameter Teflon-coated magnet was added to the beaker, and the beaker was covered with Parafilm and stored at 4°C for 16 h. Thereafter, 50 mM lithium bromide (LiBr; 200 μl/10 ml of the suspension) was added to the beaker and magnetically stirred (setting 3; 2.5 h) to extract all contractile elements. The LiBr-treated suspension was transferred to a centrifuge tube, diluted in 20 ml of 10 mM Tris-HCl, and centrifuged (10,000 × g for 10 min at 4°C). The pellet was resuspended in 10 ml of the 10 mM Tris-HCl and centrifuged (6,000 × g for 10 min at 4°C). The pellet obtained was resuspended in 25% potassium bromide (KBr; 15 ml per 1 g [wt/vol] of muscle), followed by centrifugation (10,000 × g for 30 min at 4°C). The KBr-treated pellet was washed once with 250 mM sucrose buffer and was recentrifuged (17,000 × g for 20 min at 4°C) to obtain the final plasma membrane-enriched pellet. This pellet was resuspended in 500 μl of sucrose buffer and fractionated by a continuous Percoll gradient procedure (7). In the resulting Percoll-free fractionated samples, the protein content was determined by the bicinchoninic acid method, and 10 μg of each fraction was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and identified by Western blotting. Western blot signals were quantified by densitometry.

Subcellular localization of FATP1 in skeletal muscle.

Soleus muscle from overnight fasted mice was dissected 30 min after intraperitoneal injection of saline or 0.75 U insulin per kg of body weight. Muscle strips were fixed in 2% paraformaldehyde, thick sectioned (15 μm to 40 μm), and stained with antibodies as previously described for FATP4 (45). Sections were examined with a Zeiss LSM10 confocal microscope. Three-dimensional reconstructions of confocal image stacks were performed using Volocity (Improvision, England).

Cell culture and [¹⁴C]oleate uptake assay.

3T3-L1 fibroblasts (ATCC) were grown in Dulbecco modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (DMEM-FBS). Differentiated cells were prepared as described previously. In short, 3T3-L1 fibroblasts were grown 2 days after confluence in DMEM-FBS and then for 2 days in DMEM-FBS supplemented with 0.83 μM insulin, 0.25 μM dexamethasone, and 0.25 mM isobutylmethylxanthine. The medium was then changed to DMEM-FBS supplemented with 0.83 μM insulin for 2 days only, and then cells were maintained in DMEM-FBS alone for a further 3 to 5 days. Differentiated cells (at least 90% of cells showed lipid droplets) were used on day 8 after initiation of differentiation.

3T3-L1 adipocytes were detached with trypsin-EDTA and seeded onto a 24-well plate at ∼10,000 cells/well in 500 μl of DMEM-FBS the day before assay. After 4 h of preincubation with deprivation of serum, cells were treated with inhibitors (2× 50% inhibitory concentration [IC₅₀]) and insulin (100 nM) (30 min each at 37°C). Then each well was replaced with 200 μl fatty acid uptake solution (50 μM [¹⁴C]oleate in Hanks’ balanced salt solution [HBS] with 0.1% fatty-acid-free bovine serum albumin [BSA]) for 5 min followed by washing with cold HBS. The cells were lysed in 250 μl radioimmunoprecipitation assay buffer and 75 μl supernatant of lysate used for scintillation count, and counts per minute were normalized by protein concentration of each lysate.

Preparation of primary cells.

Primary adipocytes were prepared from epididymal fat pads as previously described (44), and hepatocytes were isolated by a dual perfusion method (48). Mouse ventricular cardiomyocytes were prepared essentially as described previously (8). Mice were subjected to overnight fasting so as to lower systemic insulin levels. The isolated cells were allowed to recover for ∼2 h at room temperature in Krebs-Henseleit bicarbonate medium supplemented with 1.0 mmol/liter CaCl₂ and 2% (wt/vol) BSA and equilibrated with 95% O₂ and 5% CO₂ at 37°C. Only when >80% of the cells had a rod-shaped appearance and excluded trypan blue were they used for subsequent fatty acid uptake studies.

Fluorescent fatty acid and [¹⁴C]oleic acid uptake assay.

Fluorescence-activated cell sorter-based short-term (1-min) fatty acid uptake assays with primary adipocytes, hepatocytes, and cardiomyocytes were performed as previously described (23, 38, 41). Long-chain fatty acid uptake assays with soleus muscle were performed by incubating muscle strips in RPMI with 50 nM insulin for 30 min or without. Each muscle strip was then dipped into HBS containing 0.2% fatty-acid-free BSA, 50 μM [¹⁴C]oleic acid (BSA, 55 mCi/mmol) for 1 min. After a thorough washing with ice-cold 0.2% BSA solution, muscle strips were homogenized and radioactivity in supernatants was determined by β-scintillation counting and normalized to protein amount.

Deoxyglucose uptake.

A 2-DG uptake assay with soleus muscle was done simultaneously with fatty acid uptake by adding 2-deoxy-d-[1-³H]glucose to a final concentration of 0.1 mmol/liter to the uptake mixture.

Diet studies.

FATP1-null mice were generated as previously described (28). For diet studies, 8-week-old male mice housed separately were fed ad libitum a high-fat diet containing 60% fat (D12492; Research Diets, NJ) or a low-fat diet (D12450) containing 10% fat. Weight was measured weekly, and food intake was measured twice a week. Glucose concentrations were monitored with a glucometer (Bayer, NY) using whole blood collected from transversely sectioned tails. Standard glucose tolerance tests (1) were performed in which mice fasted for 10 h before receiving an intraperitoneal injection of 2 mg of glucose per kg of body weight in unanesthetized mice following a 12-hour fast. Circulating glucose levels were then measured at indicated time points. Insulin tolerance tests were performed similarly by injecting 0.75 U insulin per kg body weight.

Lipid gavage.

Mice were given an intragastric 200-μl olive oil bolus containing 2 μCi [¹⁴C]oleic acid. Before (t = 0) and at 30, 60, 120, and 240 min after administration, blood samples (75 μl) were taken by orbital eye bleeding and tissues were harvested after the last time point. ¹⁴C content was measured by scintillation counting and normalized to mg protein content for tissues or μl volume for serum.

Serum analysis.

Serum free fatty acids and triglyceride concentrations were measured with colorimetric assays (Wako Chemicals, VA). Levels of plasma glucose were determined enzymatically using a commercially available kit (Thermo Electron, Australia). Plasma insulin levels were determined by enzyme-linked immunoassay using a mouse insulin kit (Alpco Diagnostics, NH). Fasting plasma leptin, adiponectin, and resistin levels were measured by enzyme-linked immunosorbent assay (Linco Research, MO).

Lipoprotein separation by FPLC.

One hundred fifty microliters serum from each animal in the feeding study (five male animals per group) was pooled according to diet and genotype and centrifuged twice at full speed. Total cholesterol was determined as for serum samples. Two hundred microliters of each serum pool was then applied to a fast protein liquid chromatograph (FPLC) (Pharmacia, Sweden), separated on a Superose 6 column (HR 16), and eluted with (wt/vol) 0.9% NaCl, 0.01% Tris, 0.01% EDTA, 0.02% sodium azide, pH 7.6. Up to 40 2-ml fractions were collected for each sample and assayed for protein (A₂₈₀) and cholesterol. Cholesterol concentrations were plotted versus fraction number. Peaks were automatically identified, integrated, and expressed as percentage of total signal. Absolute cholesterol amounts for very low density lipoprotein (VLDL), low-density lipoprotein (LDL), and high-density lipoprotein were calculated by multiplying the total plasma cholesterol value by the volume percentage of these specific fractions.

Tissue triglyceride analysis.

Liver and skeletal muscle samples were powdered in liquid nitrogen, and total lipids were extracted by the Folch method (16). Total triglycerides were assayed using a colorimetric kit (Thermo Electron, Australia).

Fatty acid oxidation in liver slices.

Liver slices (0.5 mm thick) were prepared using a McIlwain tissue slicer and placed in 25-ml Erlenmeyer center-well flasks with 2 ml Krebs-Ringer phosphate buffer. [¹⁴C]oleic acid bound to fatty-acid-free BSA in Krebs-Ringer phosphate buffer was added to a final concentration of 0.5 mM and 0.15 mM, respectively, and equilibrated for 30 s with a humidified 95%/5% O₂/CO₂ gas mix. Flasks were then capped with a rubber stopper containing a centered well enclosing a loosely folded filter paper moistened with 0.2 ml of 1 N NaOH solution. After incubation for 4 h at 37°C, the reaction was stopped by injecting 0.2 ml of H₂SO₄ (1 M), and the radioactivity trapped in the filter paper was determined by scintillation counting.

Real-time PCR.

Primers and a TaqMan 6-carboxyfluorescein probe for selected genes were designed using Primer Express software (Applied Biosystems, CA). Pooled total liver RNA samples of the FATP1^−/− and littermate wild-type mice, either on a regular diet or on a high-fat diet, were prepared using an RNeasy Mini Kit (QIAGEN, CA). One microgram of total RNA was reverse transcribed in a total volume of 20 μl using the GeneAmp RNA PCR kit (Applied Biosystems) following the manufacturer's instructions. Quantitative real-time reverse transcription-PCRs were performed using the ABI Prism 7700 sequence detection system (Applied Biosystems) using the TaqMan Universal PCR Master Mix (Applied Biosystems) and the universal thermal cycling parameters (2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C). Arbitrary units of target mRNA were corrected by measuring the levels of glyceraldehyde-3-phosphate dehydrogenase mRNA in the paralleled PCR.

The nucleotide sequences of primers and probes are shown in Table 1.

Insulin induces LCFA uptake and FATP1 translocation.

To test whether insulin-triggered LCFA uptake by a variety of tissues coincides with the expression of any particular FATP, we compared FATP protein levels to insulin-induced fatty acid uptake rates (Table 2). Data from this experiment were consistent with the hypothesis that FATP1 mediates the insulin-inducible LCFA uptake (Table 2). FATP4 is also expressed in insulin-sensitive tissues (Table 2) but has been shown to be the primary fatty acid transporter in the small intestine (45), a tissue that does not respond to insulin stimulation with increased LCFA uptake (Table 2). In soleus muscle, insulin increased both glucose and oleic acid uptake to comparable extents (Fig. 1A). Interestingly, chronic high-fat feeding impaired the insulin response in both transport systems (Fig. 1A), highlighting overlaps in the signaling pathways for glucose and fatty acid transporter translocation. These similarities were also obvious when 3T3-L1 adipocytes were preincubated with inhibitors (at a twofold IC₅₀ concentration) of translation (NSC119889, 3 mM [34], phosphatidylinositol 3-kinase [wortmannin, 5 nM] [35], AKT [NSC154020, 20 mM] [57], mitogen-activated protein kinase [U0126, 0.1 mM] [14], Src [PP2, 10 nM] [22] [may also inhibit insulin receptor kinase and protein kinase C {PKC } with an IC₅₀ of 10 μM for both {29, 38}], or PKC [Go 6850, 0.5 mM] [26]) (Fig. 1B). Similar results were obtained using respective concentrations of inhibitors at 10-fold their IC₅₀s. Importantly, inhibition of translation did not significantly impact the insulin-induced increase in fatty acid uptake. We have previously shown that insulin causes FATP1 translocation to the plasma membrane and enhanced LCFA uptake in adipocytes (44). To demonstrate that insulin-induced FATP1 translocation is not unique to adipocytes, we used continuous gradient sedimentation of insulin-stimulated and unstimulated skeletal muscle to analyze changes in the subcellular localization of FATP1 and GLUT4. A 30-min insulin stimulus induced a shift of FATP1 to lighter fractions, similar to changes observed for GLUT4 by us (Fig. 1C) and others (7). To visualize FATP1 protein in muscle, we used immunofluorescence microscopy of isolated soleus muscles incubated with or without insulin for 30 min. The sarcolemma was highlighted using an anti-caveolin 3 antibody, and costainings were performed using antibodies against FATP1 (44) and CD36 (16). Three-dimensional reconstruction of surfaces showed a robust sarcolemmal staining and a pronounced intracellular FATP1 vesicle population under insulin-free conditions (Fig. 1D). In contrast, staining of intracellular compartments was greatly reduced in insulin-treated muscle fibers (Fig. 1D). In contrast to FATP1, CD36 was expressed only by a subset of muscle fibers, as had been previously reported (54), and was largely confined to the sarcolemma regardless of insulin treatment.

FATP1 is required for insulin-induced LCFA uptake by adipocytes and muscle.

We have previously reported the generation of FATP1-null animals and used hyperinsulinemic-euglycemic clamp studies to show that skeletal muscle of FATP1-knockout mice displays partial protection from lipid infusion-induced insulin resistance and intramuscular accumulation of fatty acyl-CoA (28). However, changes in LCFA uptake in these animals remained undetermined. If FATP1 translocation is indeed causal to the increased LCFA uptake in peripheral insulin-sensitive tissues, one would expect the insulin response in adipocytes and skeletal muscle to be blunted in FATP1-knockout animals. To test this hypothesis, we isolated primary adipocytes and soleus muscle strips from FATP1-null mice and wild-type littermates. Using fluorescent and radiochemical-based LCFA uptake assays, we found that fatty acid uptake in unstimulated adipocytes and soleus muscle was not significantly altered in the absence of FATP1 (Fig. 2A and B). Most importantly, while adipose and muscle tissues from wild-type mice more than doubled their LCFA uptake after a 30-min stimulation with 50 nM insulin, the insulin responsiveness of FATP1-null tissues was greatly blunted (Fig. 2A and B). Insulin-stimulated fatty acid uptake was completely abolished in FATP1-null primary adipocytes (Fig. 2A) and diminished by 80% in soleus muscle (Fig. 2B).

Diminished LCFA uptake by muscle and adipose tissues in vitro also translated into changes in vivo, as the normally rapid drop of serum-free fatty acids after an intraperitoneal insulin injection was significantly delayed, particularly at early time points, in FATP1-null animals (Fig. 2C). Further, oil gavages containing [¹⁴C]oleate tracer demonstrated that, while lipid absorption was normal, removal of postprandial lipids was impaired in FATP1-KO mice (Fig. 2D). The delayed removal of postprandial lipids could be indicative of changes in fatty acid biodistribution. In line with this hypothesis, postgavage levels of [¹⁴C]oleate tracer in FATP1-KO animals were increased in tissues expressing FATP5, such as liver (24), or FATP6, such as heart (18), but markedly reduced in tissues normally relying on FATP1—most notably white adipose tissue (Fig. 2E) (Table 2).

Adipocytes display significant LCFA fluxes both into and out of cells. Since it was unknown if FATP1 is involved only in the uptake or also in the efflux of fatty acids, we wanted to test whether fatty acid efflux is impaired in FATP1-null adipocytes. To this end, wild-type and FATP1-null primary adipocytes were prepared from epididymal fat pads and stimulated with isobutylmethylxanthine to induce hormone-sensitive lipase activity. Lipolysis of triglycerides produces glycerol besides fatty acids, and glycerol uptake and efflux should be independent of FATP1. Therefore, the ratio between extracellular fatty acids and glycerol should be reduced if fatty acid efflux is impaired but increased if reuptake of fatty acids is affected by loss of FATP1 function. Importantly, this assumption holds true independently of possible changes in the amount of intracellular triglyceride stores, which could have arisen from the formation of fat pads in the absence of FATP1. Figure 2F shows that extracellular LCFA/glycerol ratios are indeed increased in the medium of FATP1-null adipocytes after 1 h of incubation with isobutylmethylxanthine. This strongly argues that FATP1 is predominantly involved in the uptake, but not efflux, of fatty acids. One alternative explanation for the observed data would be that the rate of glycerol phosphorylation, contributing to the glycerol 3-phosphate pool used for reesterification of triglycerides, is increased in FATP1-null adipocytes. However, we have no indication that this is indeed the case. Further, given the strong sequence conservation among the FATP family members (43), it is likely that FATP2 to -6 are also primarily involved in LCFA uptake rather than efflux.

FATP1-null mice are protected from diet-induced obesity and metabolic syndrome.

The delayed insulin-induced clearance of serum fatty acids that we observed could indicate that LCFAs in FATP1-null mice are redirected away from adipose tissue and muscle, possibly towards the liver. This redistribution could potentially alter energy expenditure and susceptibility to diet-induced obesity and diabetes. To test the latter hypothesis, we fed FATP1-null mice and wild-type littermates either a low-fat (10% cal from fat) or a high-fat (60% cal from fat, predominantly lard) chow. We repeated studies twice using six male animals per study group. As expected, wild-type mice on the high-fat diet gained significantly more weight than the low-fat chow-fed animals (Fig. 3A). In contrast, FATP1 knockout animals were protected from high-fat diet-induced weight gain and showed no significant difference in weight from either wild-type or knockout animals fed a low-fat chow (Fig. 3A). This was not due to a difference in feeding behavior since caloric consumption in knockout and wild-type animals was identical (Fig. 3B). Glucose levels, measured by tail bleeds in ad libitum-fed animals, started to rise after 8 weeks in the high-fat diet/wild-type group, indicating a loss of glycemic control. In contrast, glucose levels in the FATP1-null high-fat group remained normal (Fig. 3C). In line with these observations, the glucose tolerance test (Fig. 3D) and the insulin tolerance test (Fig. 3E), performed at the end of the 12-week feeding study, showed impaired insulin sensing in high-fat-fed wild-type but not FATP1-knockout animals. At this time point, we also determined fasting and fed serum values for insulin, glucose, triglycerides, and free fatty acids (Table 3). While the high-fat-fed wild-type animals had developed a pattern consistent with the onset of metabolic syndrome, FATP1-null animals were completely protected from these alterations (Table 3). This was also evident from the analysis of lipoprotein profiles (Fig. 3F) that showed for wild-type, but not FATP1-null, animals a conversion to a high-LDL/proatherosclerotic profile following high-fat feeding (Fig. 3F). Additionally, profiling of serum adipokines showed significantly lower leptin and resistin levels in FATP1-null mice fed a high-fat diet, while adiponectin levels were higher than those in wild-type littermates (Table 4).

Triglyceride and morphometric analysis of organs at the end of the feeding studies further supported the notion that chronic loss of FATP1 function leads to a redistribution of lipids among organs. Liver weights and triglyceride contents of FATP1-knockout animals were significantly increased in both high- and low-fat-fed animals (Fig. 4A and B), and mice displayed pronounced lipid droplet accumulation on the high-fat diet (Fig. 4C). Interestingly, β-oxidation rates in liver slices from FATP1 animals were clearly elevated (Fig. 4D)—possibly reflecting a fatty acid-induced activation of peroxisome proliferator-activated receptor α (PPAR-α) (30). While real-time PCR analysis of mRNA levels revealed similar expression of PPAR-α in livers of knockout and wild-type animals fed a high-fat diet, several PPAR-α target genes such as FABP1, CPT1, and UCP2 were strongly up regulated in the livers of FATP1-null mice (Table 5). Interestingly, expression of liver fatty acid transporters FATP2, FATP5, and CD36 revealed a small but significant up regulation of these transporters in the FATP1-knockout mice (Table 5).

Significantly, epididymal fat pads from FATP1-null animals were considerably smaller not only in the overall lighter high-fat group but also in the normal chow cohort (Fig. 5C). Interestingly, fat pad differences were primarily due to a significant reduction in adipocyte size (Fig. 5A and B) rather than a reduced cell number. Further, muscle triglyceride content was also reduced in both the basal and chronic high-fat groups (Fig. 5D). However, differences in muscle TG content did not affect motor strength (data not shown).