Isolation of RNA and protein.

Total RNA was isolated using RNA-Bee (Tel-Test). Protein extracts were prepared in a buffer composed of Tris-Cl (20 mM, pH 7.5), EDTA (5 mM), EGTA (2 mM), NP-40 (1%), NaF (100 mM), NaVO₃ (2 mM), NaP₂O₇ (10 mM), and proteinase and phosphatase inhibitor cocktails. Western blot analyses were essentially performed as described previously (36, 53). Antibodies for detection of CRBP-III have been described elsewhere (36). Antibodies to detect AMP-activated protein kinase and acetyl-CoA carboxylase (ACC) and their phosphorylation status were obtained from Cell Signaling (Danvers, MA). Separation of brown adipose tissue (BAT) into stromal vascular fraction and adipocytes was accomplished by collagenase digestion before Western blot analysis (53).

Quantitative real-time PCR.

cDNA was synthesized from total RNA (3 μg) using SuperScript III reverse transcriptase (Invitrogen). The final reaction for the quantitative real-time PCR consisted of 0.5 μM gene-specific primer, 1× SYBR Green PCR Master Mix (Applied Biosystems), and 10 μl of diluted cDNA. The reaction was carried out on a Stratagene cycler. (See supplemental Table S1 in the online version of this article for primer sequences.)

Reporter constructs for CRBP-III promoter and luciferase assays.

A 3.1-kb fragment upstream of the site of transcription was amplified from mouse genomic DNA (C57/Bl6) using a high-fidelity PCR system (Expand, Roche Diagnostics) and cloned into the Zero Blunt TOPO PCR vector (Invitrogen). The fragment was completely sequenced, and the results were compared with the published sequence. The 3.1-kb promoter fragment was cloned into the pGL3 basic vector (Promega) and designated pGL3k, which was used as a template for generation of two shorter promoter constructs, 2.5 kb (pGL2.5k) and 2 kb (pGL2k). A PCR approach was used to introduce mutations of the putative PPARγ response elements (PPREs). Putative PPRE1 (mut1), PPRE3 (mut3), or PPRE1 and PPRE3 (mut1 and mut3) were mutated. The primers used to introduce the mutations were 5′-GATCCATAACTGTGTACACAGCCGAAATTTCTGGAG-3 for PPRE1mut and 5′- CTGACTGACTTACACAGCCACAGTTTCCCTCTTACGGG-3′ for PPRE3mut. All constructs were sequenced to ensure the correct mutations. 3T3-L1 cells were differentiated in six-well plates and used at day 7 of differentiation. Cells were transfected with 0.2 μg of firefly luciferase DNA constructs (pGL3k, pGL2.5k, or pGL2k). COS-7 cells were plated for 24 h in 12-well plates in complete medium and transfected with firefly luciferase DNA constructs and with pcDNA3.1 alone or equivalent concentrations of pcDNA3.1-PPARγ₂ and pcDNA3.1-RXRα, which overexpress PPARγ₂ and RXRα, respectively. The Renilla luciferase vector (0.05 μg) was included in all transfection procedures as a control. Transfections were carried out overnight; on the next morning, the medium was replaced with complete medium containing 1 μM rosiglitazone or vehicle (ethanol). After 24 h, the dual luciferase reporter assay (Promega) was performed using a luminometer (TD-20/20, Turner Designs). Results were expressed as relative luciferase activity, with normalization of firefly luciferase activity to Renilla luciferase activity.

EMSA.

Nuclear proteins were isolated from COS-7 cells transiently expressing PPARγ₂ and RXRα using a nuclear protein extraction kit (Pierce). The probes were prepared by annealing oligonucleotide and end-labeling the double-stranded DNA with [γ-³²P]ATP using T4 kinase (Promega). The PPRE of the adipocyte-binding protein aP2 promoter was used as a positive control (50). Oligonucleotides were as follows: aP2-PPRE (5′-GATCTGTGACCTTTGACCTAGTAAG-3′, 5′-CTTACTAGGTCAAAGGTCACAGATC-3′), CRBP-III PPRE1 (5′-GAAATTTGGGCAGAGTTCACAGTTA-3′, 5′-TAACTGTGAACTCTGCCCAAATTTC-3′), CRBP-III PPRE2 (5′-ATGGATCCCTCAAAGGTCAAGGAAT-3′, 5′-ATTCCTTGACCTTTGAGGGATCCAT-3′), and CRBP-III PPRE3 (5′-TCTCCCTTGGAAAAGGTCACATTCA-3′, 5′-TGAATGTGACCTTTTCCAAGGGAGA-3′). The DNA-protein binding assays were performed as recommended by the manufacturer (Promega). A supershift assay was performed using PPARγ monoclonal (E8, Santa Cruz Biotechnology) or RXRα polyclonal (D-20, Santa Cruz Biotechnology) antibody. The binding reaction was incubated with the indicated antibody for 10 min at room temperature before addition of the labeled nucleotides. Electrophoresis of the DNA-protein complexes was performed on 6% DNA retardation gels (Invitrogen), which were placed on Whatman paper and dried using a vacuum gel dryer. The signals were detected using a Storm PhosphorImager (Amersham Biosciences).

Chromatin immunoprecipitation assay.

Differentiated 3T3-L1 adipocytes were used for the chromatin immunoprecipitation assays. After cross-linking of DNA and proteins in situ, the soluble chromatin was prepared according to the manufacturer's instructions (Upstate), and the DNA was sheared. The protein-chromatin complex was immunoprecipitated with PPARγ antibodies, and PCRs that included regions of PPRE1 and PPRE3 were performed.

Animal studies.

Three-month-old male wild-type (C57BL/6) mice (Jackson Laboratory, Bar Harbor, ME) were fed a chow diet or a diet supplemented with rosiglitazone (10 mg/kg body wt) for 21 days. Generation of mice lacking CRBP-III (C-III-KO) has previously been reported (36). The mice were backcrossed to the C57BL/6 genetic background for five generations before the experiments. C-III-KO and CRBP-III wild-type mice were fed an HFD providing 60% of calories from fat (Research Diets). The mice were fed the HFD for 20 wk and then killed, the liver was perfused with PBS, and tissues were excised, flash frozen in liquid nitrogen, and stored at −70°C until analysis. Experiments involving mice were performed with the approval of the Institutional Animal Care and Use Committee at Columbia University.

Histology.

Liver, white adipose tissue, and BAT were placed in 10% buffered formalin overnight and then embedded in paraffin. Tissue sections (5 μm) were stained with hematoxylin-eosin. Liver sections were also fresh frozen in OCT compound, sectioned (7 μm), and stained with oil red O for detection of neutral lipids.

Retinoid measures.

Retinol and retinyl ester levels in liver and adipose tissue were determined as previously described (36).

Lipolysis assay of isolated white adipose tissue.

Lipolysis assays were performed essentially as previously described (35, 42). White adipose tissue from HFD-fed wild-type and C-III-KO mice was excised, washed in PBS, and cut into small (~50 mg) pieces. The explants were incubated in Krebs-Ringer buffer for 3 h at 37°C in the absence or presence of isoproterenol (10 μM). At the end of the incubation, the plates were placed on ice, and the medium was removed for glycerol and free fatty acid analyses using commercial kits (Sigma Aldrich and Wako). Tissue pieces were washed and saved for protein analysis. For each animal, the assay was performed in quadruplicate.

Hepatic triglyceride levels.

Triglycerides were extracted from liver using a Folch extraction. Briefly, livers were homogenized in PBS and incubated in chloroform-methanol (2:1) in a rotator at room temperature for 1 h, and the chloroform phase was dried under a mild stream of nitrogen. The triglycerides were reconstituted in cold isopropanol, and levels were determined using a commercial kit (Thermo Scientific).

Serum parameters.

Blood for determination of lipid parameters, glucose, insulin, and adipokine levels was obtained from mice fasted during the day for 6 h (9 AM–3 PM) after 18 wk on the HFD. Serum was obtained after the blood samples had been allowed to clot. Commercially available kits were used to measure serum levels of nonessential fatty acids (Wako), free glycerol (Sigma Aldrich), β-hydroxybutyrate (Stanbio), total insulin (Mercodia), glucose (Invitrogen), and total cholesterol and triglyceride (Thermo Scientific) levels. Serum leptin and resistin levels were determined using a Multiplex Assay (Millipore).

Cold tolerance testing.

Cold tolerance was assessed in wild-type and C-III-KO mice fed the HFD for 20 wk by exposure to 4°C for 2 h. Rectal temperature was measured before the cold exposure and then at 1-h intervals during the cold challenge using a thermocouple probe (model IT-23) attached to a thermocouple amplifier (Physitemp BAT-10 Type I, Pharma).

Body composition analysis.

Body composition was measured by MRI using an EchoMRI (Echo Medical Systems).

Metabolic studies.

Animals were individually housed in metabolic chambers maintained at 20–22°C on a 12:12-h light-dark cycle, with lights on at 7 AM. Metabolic measurements (oxygen consumption, food intake, and locomotor activity) were obtained continuously using a Comprehensive Lab Animal Monitoring System (Columbus Instruments) open-circuit indirect calorimetry system. Metabolic data were collected over 5 consecutive days following 3 days of adaptation to the metabolic cages.

CRBP-III promoter contains two functional PPREs.

PPARγ elicits its transcriptional effect through binding as a heterodimer with RXRα to specific PPREs in the noncoding sequences of genes. We cloned different-sized promoter fragments upstream of the transcriptional start site to understand whether a PPRE is present in the CRBP-III promoter. All promoter constructs (3, 2.5, and 2 kb) showed basal activity in fully differentiated 3T3-L1 adipocytes (Fig. 1D). However, treatment with rosiglitazone increased relative luciferase activity up to threefold over basal activity for a promoter fragment including the region 2.5 and 3 kb upstream of the transcriptional start site (Fig. 1D). In contrast, for the shorter (2-kb) promoter construct, no change relative to basal activity was observed after rosiglitazone treatment. Similar data were obtained when the promoter constructs were cotransfected with PPARγ and RXRα into COS-7 cells, where treatment with rosiglitazone only increased luciferase activity for the 3- and 2.5-kb constructs (see supplemental Fig. S1A). Taken together, these data indicate that a putative PPRE is present in a 500-bp region between 2.0 and 2.5 kb upstream of the transcriptional start site. Analyzing the DNA sequence of this 500-bp region, we identified three putative PPRE sites that contained the PPRE consensus sequence (22).

On the basis of EMSAs, PPARγ and RXRα bind to two PPRE sites (PPRE1 and PPRE3), forming a single DNA-protein complex (Fig. 1E). In contrast, there was no specific band for PPRE2. Binding of the labeled PPRE to the nuclear protein appears to be specific, inasmuch as it can be competed off with excess of specific competitor, but not with nonspecific DNA (Fig. 1E). When antibodies against PPARγ or RXRα were used, the DNA-protein complex supershifted to a higher molecular weight, further indicating that PPARγ and RXRα are binding to the specific DNA region in PPRE1 and PPRE3 (see supplemental Fig. S1B). Furthermore, on the basis of chromatin immunoprecipitation assay, PPARγ binds to the PPRE region including PPRE1 and PPRE3 in vivo (Fig. 1F). Because PPRE1 and PPRE3 are within close proximity (132 bp apart), we were unable to discern whether PPARγ binds to PPRE1 or PPRE3 or both. Therefore, we generated promoter constructs that contained mutations in one PPRE alone or in both. Mutations in either PPRE alone lead to a decreased response to rosiglitazone treatment. Importantly, the response to rosiglitazone was abolished only when both PPREs were mutated (Fig. 1G). Taken together, these data demonstrate that PPARγ agonists regulate CRBP-III expression in vivo and in vitro through two PPREs present in the promoter of CRBP-III.

Hepatic steatosis is reduced in HFD-fed C-III-KO mice.

PPARγ is essential for normal adipogenesis and critically required for maintaining normal glucose and lipid homeostasis (11, 20, 50). Since CRBP-III is regulated by PPARγ in adipose tissue, we sought to understand its role in regulating lipid homeostasis by studying mice with a gene-targeted deletion of CRBP-III (C-III-KO) (36).

Since excess dietary fat leads to an expansion of adipose tissue and, often, to perturbations in energy metabolism, we examined the effects of diet-induced obesity in C-III-KO compared with wild-type mice. Weight gain in HFD-fed animals was not different between the two groups (Fig. 2A). In addition, serum glucose and insulin levels did not differ between HFD-fed wild-type and C-III-KO mice (Table 1). Next, we examined liver, white adipose tissue, and BAT for differences between wild-type and C-III-KO mice. As anticipated, livers from HFD-fed wild-type mice contained abundant, large lipid droplets (Fig. 2B). In contrast, livers from C-III-KO mice contained significantly fewer lipid droplets (Fig. 2B). In agreement with these observations, hepatic triglyceride concentrations were significantly lower in C-III-KO than wild-type mice (Fig. 2C). Hepatic retinyl ester levels were not different between the two groups (Fig. 2D).

Open in a separate window

Fig. 2.

A: effect of high-fat diet (HFD) on body weight in WT and CRBP-III-deficient (C-III-KO) mice. B: lipid accumulation in livers from C-III-KO and WT mice, as judged by hematoxylin-eosin (top) and oil red O (bottom) staining. Images are representative of observations of 7 mice per group. C: triglyceride (TAG) content in livers is significantly lower in C-III-KO than WT mice fed HFD for 20 wk. Values are means ± SE (n = 9 for both genotypes). *P < 0.05 vs. WT. D: retinol and retinyl ester levels in the liver were not significantly different between WT and C-III-KO mice.

Since CRBP-III is not expressed in the liver but is expressed at high levels in adipose tissue (8, 36, 53), decreased hepatic triglyceride may be secondary to alterations in adipose tissue of C-III-KO mice. One possibility is that altered adipokine levels in C-III-KO mice may be responsible for the decreased fat accumulation in the liver. However, we detected no differences in levels for serum leptin and resistin (Table 1) and expression of adiponectin (see Table 3) between C-III-KO and wild-type mice. Similarly, genes involved in hepatic glucose and fatty acid metabolism were not altered in the liver of C-III-KO mice (Table 2). We detected no differences in the liver for expression of genes for glucose metabolism (GLUT2, phosphoenolpyruvate carboxykinase, and glucose-6-phosphatase), fatty acid synthesis (ACC, fatty acid synthase, stearoyl CoA-desaturase, and sterol regulatory element-binding protein-1c), or fatty acid oxidation (PPARα and acyl-coenzyme A dehydrogenase, long-chain). Similarly, in the liver, phosphorylation of AMP-activated kinase and its downstream target ACC were not different between HFD-fed C-III-KO and wild-type mice (Table 2). These data imply that there was no apparent change in fatty acid synthesis or oxidation in the liver of HFD-fed C-III-KO mice that would account for the decreased hepatic lipid accumulation.

Decreased serum free fatty acid levels and in vitro adipose lipolysis in HFD-fed C-III-KO mice.

In diet-induced obesity, the majority of free fatty acids used for hepatic triglycerol synthesis originate from adipose tissue (6). Lipolysis of adipose triglyceride leads to release of nonesterified free fatty acids and glycerol into the circulation, a process that is further increased in states of insulin resistance (24, 38). Therefore, we examined serum levels of free fatty acids and glycerol (Fig. 3, A and B). Serum free fatty acid and free glycerol levels were significantly lower in C-III-KO than wild-type mice. This is indicative of decreased free fatty acid efflux from the adipose tissue of C-III-KO mice. To further understand these changes, we performed in vitro lipolysis assays. We isolated white adipose tissue from HFD-fed wild-type and C-III-KO mice and measured glycerol and free fatty acid release under basal and isoproterenol-stimulated conditions in vitro. Under basal conditions, glycerol and free fatty acid release were significantly lower from adipose tissue of C-III-KO than wild-type mice (Fig. 3C). This finding is in agreement with the decreased levels observed in vivo. On stimulation of lipolysis by isoproterenol, significantly less glycerol was released from adipose tissue of C-III-KO than wild-type mice, whereas the decrease in free fatty acid release did not reach statistical significance (P = 0.08; Fig. 3C). Gene expression analysis of white adipose tissue from wild-type and C-III-KO mice revealed no differences in expression of the lipolytic enzymes adipose triglyceride lipase, hormone-sensitive lipase, or monoglyceride lipase (Table 3). In addition, expression of genes involved in fatty acid oxidation in white adipose tissue is not different between C-III-KO and wild-type mice (Table 3). In the liver, free fatty acids are reesterified to triglyceride but are also oxidized to ketone bodies (13, 15). Ketone body formation is another surrogate marker for increased fatty acid release from adipose tissue and for uptake and oxidation by the liver in states of insulin resistance, such as diet-induced obesity (15). The levels of the ketone body β-hydroxybutyrate were significantly lower in serum of C-III-KO mice (Fig. 3D), consistent with the idea that less free fatty acid is reaching the liver. Taken together, these data indicate that a decreased flux of free fatty acids from the adipose tissue to the liver may contribute to the decreased hepatic triglyceride content in C-III-KO mice.

Open in a separate window

Fig. 3.

Serum levels of nonessential free fatty acids (A) and free glycerol (B) are significantly lower in C-III-KO than WT mice. Values are means ± SE (n = 10 for each genotype). *P < 0.05 vs. WT. C: in vitro lipolysis assay of adipose tissue isolated from HFD-fed WT and C-III-KO mice. Glycerol and free fatty acid (FFA) release was significantly lower in the basal state in C-III-KO than WT mice. In the stimulated state, free glycerol release is significantly lower in C-III-KO mice, and the difference in free fatty acid release between WT and C-III-KO mice did not reach significance (P = 0.08). Values are means ± SE (n = 5 mice for each genotype performed in quadruplicate). *P < 0.05. D: serum levels of β-hydroxybutyrate are significantly lower in C-III-KO than WT mice. Values are means ± SE (n = 10 for each genotype). *P < 0.05 vs. WT.

White adipose tissue morphology of HFD-fed C-III-KO and wild-type mice showed the prevalence of large unilocular droplets, with no difference in adipocyte size at the end of the HFD (data not shown). Similar to the liver, no differences between wild-type and C-III-KO mice were detected for retinol or retinyl ester in the white adipose tissue (Table 4). In addition, expression in adipose tissue of genes related to retinoid metabolism, such as alcohol dehydrogenase type 1, retinal dehydrogenase type 1, CRBP-I, and retinol-binding protein, was not altered in C-III-KO compared with wild-type mice (Table 3). Although we did not directly measure RA levels, target genes activated by RA (short-chain dehydrogenase/reductase and tissue transglutaminase type 2) were not different between wild-type and C-III-KO mice in white adipose tissue (Table 3).

Body composition, respiratory exchange ratio (RER), and food intake are altered in C-III-KO mice. To gain further mechanistic insight into the role of CRBP-III deficiency in lipid metabolism, we assessed body composition, RER, and food intake in wild-type and C-III-KO mice. After 20 wk of HFD feeding, body weights were not different between wild-type and C-III-KO mice (Figs. 2A and ​and4A).4A). However, MRI analysis of body composition revealed a small, but significant, reduction in fat mass accompanied by an increase in lean mass in C-III-KO compared with wild-type mice (Fig. 4A). To determine the basis for the decreased adiposity in C-III-KO mice, we assessed food intake during the HFD feeding. As shown in Fig. 4B, food intake was lower in C-III-KO than wild-type mice. Thus food intake may contribute to the reduction in adiposity. Because CRBP-III is not expressed in the brain (see supplemental Fig. S2), the reduced food intake in C-III-KO mice is likely due to central nervous system processing of the peripheral metabolic consequences of CRBP-III deficiency. Furthermore, there was a nonsignificant trend toward reduced oxygen consumption (Fig. 4C) and energy expenditure: 8.1 ± 0.1 and 7.6 ± 0.2 (daytime), 8.9 ± 0.1 and 8.5 ± 0.2 (nighttime), and 8.5 ± 0.1 and 8.1 ± 0.2 kcal·kg⁻¹·h⁻¹ (total) for wild-type and C-III-KO mice, respectively.

Open in a separate window

Fig. 4.

Metabolic analysis of WT and C-III-KO mice fed HFD for 20 wk. A: analysis of body composition of WT and C-III-KO mice. Total body weight was not different between WT and C-III-KO mice, but fat mass was decreased and lean mass was increased in C-III-KO compared with WT mice. Values are means ± SE (n = 6 for each genotype). B: C-III-KO mice consume less food than WT mice. Values are means ± SE (n = 6 for each genotype). C: oxygen consumption (o₂) of HFD-fed WT and C-III-KO mice. Values are means ± SE (n = 4 for each genotype). D: respiratory exchange ratio (RER) was significantly higher for C-III-KO than WT mice during light and dark phases. Values are means ± SE (n = 4 for each genotype). E: activity levels were not different between WT and C-III-KO mice. Ambulatory and rearing activity were measured and combined to express total activity. Values are means ± SE (n = 4 for each genotype). *P < 0.05 vs. WT.

Inasmuch as CRBP-III is expressed in the muscle (53), we also performed gene expression studies in muscle. Soleus muscle expression of genes involved in fatty acid oxidation did not differ between wild-type and C-III-KO mice (see supplemental Table S2). This finding is further substantiated by results from indirect calorimetry. RER was slightly, but significantly, increased in C-III-KO compared with wild-type mice, indicating an increase in glucose, rather than fatty acid, oxidation (Fig. 4D). In addition, there was no difference in locomotor activity between wild-type and C-III-KO mice (Fig. 4E).

CRBP-III and whole body energy metabolism.

In the present study, the lack of CRBP-III appears to affect whole body energy balance. Specifically, lack of CRBP-III resulted in a reduction in food intake and a nonsignificant trend toward reduced energy expenditure and oxygen consumption. Body weight was similar between C-III-KO and wild-type mice fed the HFD, suggesting that decreased food intake in C-III-KO mice may be balanced by decreased energy expenditure. However, fat mass was significantly reduced in C-III-KO compared with wild-type mice. This decrease in fat mass was offset by an increase in lean body mass, resulting in no net change in total mass relative to that of wild-type mice. The increased lean body mass would be predicted to increase the metabolic load in C-III-KO mice, which would, in turn, mitigate against a more significant reduction in oxygen consumption and energy expenditure.

The small, but significant, increase in RER in C-III-KO mice indicates increased carbohydrate, rather than fat, oxidation. This interpretation is consistent with our observation that expression of genes important in fatty acid oxidation was unchanged in skeletal muscle of C-III-KO compared with wild-type mice. Overall, the lack of CRBP-III induces a distinctive profile of behavioral and metabolic consequences, consisting of a change in nutrient partitioning, reduced feeding, decreased fat mass, and increased lean mass, resulting in an overall maintenance of body weight.

CRBP-III and BAT.

BAT is characterized by multilocular lipid droplets and a high density of mitochondria in mice fed a chow diet. The central function of BAT is oxidation of fatty acids and dissipation of energy in the form of heat as part of adaptive nonshivering thermogenesis to maintain body temperature in small mammals (7). On the basis of our data, C-III-KO mice likely maintained a better oxidative capacity, as reflected in higher expression of key genes involved in fatty acid oxidation and oxidative phosphorylation in BAT from C-III-KO than wild-type mice fed the HFD. In addition, C-III-KO mice were able to maintain body temperature during a cold challenge, indicating preserved thermogenesis in BAT of C-III-KO compared with wild-type mice. On the basis of this observation, lower body weights of C-III-KO than wild-type mice fed the HFD may be have been expected. In addition to the protection against cold, BAT has been suggested to play an important role in energy metabolism, but a direct role in body weight regulation remains controversial (26). Ablation of BAT or inactivation of all three β-adrenergic receptors rendered mice thermogenically inactive and susceptible to diet-induced obesity (3, 18, 27). In contrast, other mouse models, including inactivation of UCP-1 or failure to induce UCP-1, render chow- or HFD-fed mice cold sensitive but not obese (12, 46–48). Several studies indicate that interscapular BAT may not play a direct role in regulating body weight but, rather, that diet-induced thermogenesis may be mediated by brown adipocytes or increased expression of BAT-specific genes in tissues such as skeletal muscle or white adipose tissue (1, 9, 26, 48). In the present mouse model, we may have a more preserved BAT, but it is not contributing to the phenotype, and/or its contribution may be blunted.

CRBP-III and lipolysis.

Significantly lower serum levels of free fatty acids and glycerol in C-III-KO than wild-type mice may have contributed to lower hepatic triglyceride accumulation in C-III-KO than wild-type mice. In addition to fatty acid reesterification to triglyceride in the liver, free fatty acids are used for ketogenesis, leading to increased ketone body levels in the circulation, which are significantly increased in states of insulin resistance and diabetes (15). Serum levels of ketone bodies were significantly lower in C-III-KO than wild-type mice, consistent with the notion that fatty acid flux to the liver is reduced.

Hepatic steatosis during obesity and insulin resistance are consequences of increased rates of lipogenesis in the liver and increased uptake of nonesterified fatty acids released by adipose tissue (6, 17, 24, 44). Excess triglyceride stored as cytosolic lipid droplets in hepatocytes contributes to the development of fatty liver and insulin resistance (6, 40, 41, 45). Since we found no evidence that changes in hepatic lipid synthesis or oxidation contributed to differences in hepatic steatosis, significantly lower serum free fatty acid levels in C-III-KO mice may offer an explanation for the lower hepatic triglyceride accumulation. Serum free fatty acid levels, especially under postabsorptive conditions, are mostly a function of fatty acid release from adipose tissue (13). During insulin resistance, white adipose tissue lipolysis is increased, leading to a rise in the release of free fatty acids and glycerol into the circulation. Consistent with decreased serum fatty acids in C-III-KO mice, in vitro lipolysis assays of adipose tissue from C-III-KO mice indicated diminished efflux of glycerol and free fatty acids from adipose tissue compared with wild-type mice in the basal state.