Thiazolidinediones Upregulate Uptake and Oxidation in Adipose Tissue of Diabetic Patients Guenther Boden,1,2 Carol Homko,2 Maria Mozzoli,1,2 Louise C. Showe,3 Calen Nichols,3 and Peter Cheung1,2

Thiazolidinediones (TZDs) are a new class of insulin- TZDs produce insulin sensitization, which takes place sensitizing drugs. To explore how and in which tissues primarily in skeletal muscle (6). Lowering plasma free they improve insulin action, we obtained fat and muscle fatty acid (FFA) levels improves insulin sensitivity (7). The biopsies from eight patients with type 2 diabetes before well-established TZD lowering of plasma FFAs has there- -fore been widely considered to be a major factor respon (5 ؍ and 2 months after treatment with rosiglitazone (n ,TZD treatment was associated sible for their insulin sensitizing effect (8–12). It is .(3 ؍ or troglitazone (n with a coordinated upregulation in the expression of however, not clear how TZDs lower FFA levels. Since they and synthesis of proteins involved in fatty acid ␤ seem to have little or no effect on basal rates of lipolysis uptake, binding, -oxidation and electron transport, (11,13,14), the decrease in plasma FFA levels is probably and oxidative phosphorylation in subcutaneous fat but not in skeletal muscle. These changes were accompa- due to an increase in FFA clearance (oxidation and/or nied by a 13% increase in total body fat oxidation, a 20% esterification). Supporting this notion, we have recently decrease in plasma free fatty acid levels, and a 46% found that TZD-induced lowering of plasma FFAs was increase in insulin-stimulated glucose uptake. We con- associated with an increase in total body fatty acid oxida- clude that TZDs induced a coordinated stimulation of tion (FOX) (14). This study, however, provided no infor- fatty acid uptake, oxidation, and oxidative phosphory- mation as to the location (fat, muscle, or both) or the lation in fat of diabetic patients and thus may have mechanism by which TZDs stimulated FOX. To further corrected, at least partially, a recently recognized explore these issues, we have now investigated the effects defect in patients with type 2 diabetes consisting of of TZD treatment on the expression of genes related to reduced expression of genes related to oxidative fatty acid uptake and oxidation in subcutaneous fat and in and mitochondrial function. Diabetes 54: 880–885, 2005 skeletal muscle of patients with type 2 diabetes.

RESEARCH DESIGN AND METHODS We studied eight patients with type 2 diabetes (Table 1). Two were treated hiazolidinediones (TZDs), a new class of insulin- with sulfonylureas, two with sulfonylureas and metformin, two received sensitizing drugs, are strong peroxisome prolif- sulfonylureas, metformin, and NPH insulin (10–25 units at bedtime), and two erator–activated receptor ␥ (PPAR␥) agonists received insulin twice daily. These medications were withheld at least 72 h (1). As their binding affinity to PPAR␥ correlates before hospital admission but were otherwise continued throughout the T studies. Informed written consent was obtained from all subjects after well with their insulin sensitizing activity, it is generally explanation of the nature, purpose, and potential risks of the studies. The accepted that TZDs exert their insulin sensitizing action study protocol was approved by the institutional review board of Temple through PPAR␥ (2,3). However, exactly how and where University Hospital. TZDs increase insulin sensitivity remains incompletely All patients were admitted to the general clinical research center at Temple ϳ understood. PPAR␥ is expressed at highest concentrations University Hospital the night before the study. At 8:00 A.M. on the day of the study, a fat and muscle biopsy was obtained from the subcutaneous fat of the in adipose tissue and at much lower concentrations in upper thigh and the lateral vastus muscle, respectively, and rates of fat and muscle (4,5), suggesting that the primary action of oxidation (FOX) were determined by indirect calorimetry. After that, the TZDs is on adipose tissue. This raises the question of how patients were started on TZDs. After 2 months on TZD therapy, they were readmitted for another fat and muscle biopsy and FOX measurement. The initial three patients received troglitazone (600 mg/day) and the other five From the 1Division of Endocrinology, Diabetes, and Metabolism, Temple patients received rosiglitazone (8 mg/day). The change was necessary because University School of Medicine, Philadelphia, Pennsylvania; the 2General troglitazone was withdrawn from the U.S. market after the study had begun. Clinical Research Center, Temple University School of Medicine, Philadelphia, Fat and muscle biopsies. Under local anesthesia (1% lidocaine in a field Pennsylvania; and the 3Wistar Institute, Philadelphia, Pennsylvania. block pattern), an incision (ϳ1 inch) was made through the skin at the lateral Address correspondence and reprint requests to Guenther Boden, MD, aspect of the upper thigh (ϳ15 cm above the patella) and ϳ200 mg of Temple University Hospital, 3401 N. Broad St., Philadelphia, PA 19140. E-mail: subcutaneous adipose tissue was mobilized and excised. Through the same [email protected]. ϳ Received for publication 14 July 2004 and accepted in revised form incision, biopsies ( 100 mg) were obtained from the lateral aspect of the 9 December 2004. vastus lateralis muscle. The excised fat and muscle samples were dropped Additional information for this article can be found in an online appendix at immediately into isopentane, kept at its freezing point (Ϫ160°C) by liquid http://diabetes.diabetesjournals.org. nitrogen, and stored at Ϫ80°C until used for RNA extraction. ACADM, acyl-CoA dehydrogenase; FFA, free fatty acid; FOX, fatty acid Array hybridization. The cDNA filter arrays were prepared by the Wistar oxidation; HCS, cytochrome C; PPAR␥, peroxisome proliferator–activated Institute Genomics facility (available at www.wistar.upenn.edu/genomics and ␥ receptor ; TZD, thiazolidinedione. ref. 16). A 7.5 ϫ 11.5–cm nylon filter (HA-04) carrying a total of 9,600 probes © 2005 by the American Diabetes Association. was used. Sequence-verified clones were purchased from Research Genetics The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance (Huntsville, AL). Total RNA was isolated from the aqueous phase by isopro- with 18 U.S.C. Section 1734 solely to indicate this fact. panol precipitation after homogenization in phenol-guanidine thiocyanate

880 DIABETES, VOL. 54, MARCH 2005 G. BODEN AND ASSOCIATES

TABLE 1 Soluble proteins, 35 ␮g per lane, were resolved in 8–16% SDS-PAGE and Characteristics of study subjects transferred to nitrocellulose membranes. CD-36 immunoblots were probed for 1 h with a monoclonal anti-CD36 antibody at 1:1,000 (Santa Cruz, Santa Cruz, Pre-TZD Post-TZD CA). Cytochrome C was detected with a monoclonal mouse anti-pigeon cyto- chrome C antibody (diluted 1:1,000; Imgenex, San Diego, CA). ␤-Actin was de- Sex (M/F) 5/3 5/3 tected with a monoclonal mouse anti-human ␤-actin antibody (diluted 1:4,000; Age (years) 53 Ϯ 3—Santa Cruz). A horseradish peroxidase conjugate goat anti-mouse IgM anti- Weight (kg) 95.2 Ϯ 6.2 95.5 Ϯ 6.4 body (A-8786m; Sigma, St. Louis, MO) was applied for1hat1:5,000. Immu- Body fat (kg) 35.9 Ϯ 3.7 34.0 Ϯ 4.0 noreactivity was detected by chemiluminescence. Acyl-CoA dehydrogenase BMI (kg/m2) 32.7 Ϯ 1.7 33.0 Ϯ 1.8 (ACADM) immunoblots were probed with a rabbit polyclonal anti-ACADM Duration of diabetes (years) 8.4 Ϯ 2.1 — antibody at 1:1,000 (AG Scientific, San Diego, CA) at 4°C overnight. A horse- Ϯ Ϯ radish peroxidase conjugate goat anti-rabbit IgG antibody was applied for 1 h HbA1c (%) 8.8 0.5 8.7 0.9 at 1:200,000. Immunoreactivity was detected by Femto Maximum Sensitivity Substrate and Cl-Xposure film (Pierce). The signal was scanned by Hewlett- (Tri-Reagent Molecular Research Center, Cincinnati, OH). Total RNA samples Packard Scanjet and quantitated by Scion Image (Scion, Frederick, MD). were amplified (aRNA) using a modified T7 protocol for fat (available at Statistical analysis. All data are expressed as means Ϯ SE. Statistical http://cmgm.Stanford.edu/pbrown/protocols) and muscle (MessageAmp; Am- significance was assessed using repeated-measures ANOVA and paired stu- bion, Austin, TX). Each sample (total 32) was hybridized to a single array, as dent’s two-tailed t test when applicable. materials were limited. The average relative error between arrays was Ͻ9% when they were hybridized in triplicate under our standard conditions (data RESULTS not shown). All hybridization steps were carried out in a hybridization oven with constant rotation: prehybridization 1 was in 3 ml of MicroHyb (Research Adipose tissue Genetics) and 1 ␮g/ml of salmon sperm DNA for1hat42°C then removed. A Effect of TZDs on expression. TZD treatment of total of 2.5 ml of MicroHyb with 1 ␮g/ml of denatured human Cot-1 DNA and the eight patients with type 2 diabetes (Table 1) was as- 1 ␮g/ml of PolydA (Prehyb2) was added and filters incubated for2hat42°C. The aRNA target was labeled with 33P (30005-000 Ci/mmol; Amersham Phar- sociated with a statistically significant twofold or greater macia Biotech) using reverse transcriptase. The denatured 33P target (0.8 ␮g) increase in the expression (by microarray) of 107 gene was added to Prehyb 2, and filters were incubated at 42°C for 18 h. Filters sequences and a twofold or greater decrease in the expres- were washed two times in 2ϫ SSC/1% SDS, one time in 0.5ϫ SSC/1% SDS, and sion of six gene sequences (online appendix [available at one time in 0.1ϫ SSC/0.5% SDS each for 30Ј at 55°C. The fat filters received an additional wash of 0.1ϫ SSC/0.1% SDS at 60°C then all filters exposed to a http://diabetes.diabetesjournals.org]). Of those that had phophorimager screen for 6 and 14 days for fat and muscle, respectively. increased, 95 included genes related to carbohydrate or Screens were scanned at 50-␮ resolution in a Storm PhosphorImager and protein metabolism, signal transduction, cell growth and visualized using ImageQuant (Molecular Dynamics). development, transcription factors, nuclear proteins, and Array analysis. Arrays were analyzed with ImaGene 5.1 (Biodiscovery, El others (online appendix), and 12 gene sequences were Segundo, CA), and the median pixel value was calculated for each spot after subtraction of the local background. The calculated values were exported to functionally related to either fatty acid uptake and binding Microsoft Excel and the values for each spot normalized by dividing the signal (fatty acid translocase/CD36, fatty acid binding proteins 4 reference, which is the calculated median pixel value for all 9,600 background- and 6 and fatty acid CoA ligase) or ␤-oxidation (acyl-CoA subtracted spots on the array (normalized median density). The dynamic dehydrogenase and acetyl-CoA acetyltransferase) or mito- range of array signals was on average 7–10,000 (16). The detection limit for these conditions and arrays was calibrated by quantitative PCR. Standard curves were generated using known amounts of a purified gene-specific TABLE 2 template. PCR values from the RNA were mapped to that standard curve. The Primer sequences of genes related to fatty acid uptake, binding, lowest detectable signal that was considered reliable was a normalized and oxidation median density three times the global array background a value of 0.15. PCR data for samples with 0.15 signal level on the arrays was equivalent to 0.03 Name Gene ID Sequence molecules/cell (16). A TZD effect was only considered significant if there was CD36 N39161 a statistically significant change of twofold or greater in gene expression Ј Ј between pre- and posttreatment. Forward 5 -AGGTCAACCTATTGGTCAAGC-3 Real-time RT-PCR. One-step real-time RT-PCR was performed with random Reverse 5Ј-AGATCATTTCTATCAGGCCAAGGA-3Ј hexamers and amplified by one-step RT-PCR (Titanium kit; Becton Dickinson FABP4 N92901 Biosciences, Palo Alto, CA) and with a LightCycler (Roche, Indianapolis, IN) Forward 5Ј-TTGACGAAGTCACTGCAGATGACA-3Ј equipped with an emission filter (at 530 nm) for SYBR green I fluorescence Reverse 5Ј-TCTCTCATAAACTCTCGTGGAAGT-3Ј measurement and analyzed by the Relative Quantification Software provided FACL3 AA424965 by the manufacturer. Reverse transcription: 50°C for 30 min; denaturation: Forward 5Ј-GTGTGACAATGGGGTACT-3Ј 95°C for 0 min; amplification: 95°C for 0 min, 55–65°C for 15 s, and 72°C for Reverse 5Ј-CACAGCTCCTCCCAAG-3Ј 13 s for 40 cycles; melting curve: 95°C for 0 min, 65°C for 10 s, and 95°C for ACAT1 AI871165 0 min, cooling 40°C for 30 s. Primer sequences are listed in Table 2. Cycle of Ј Ј threshold (Ct), relative concentration, standard curve, coefficient of correla- Forward 5 -AGCGAAGAGGCTCAAT-3 tion, and efficiency of each amplification was calculated with the Relative Reverse 5Ј-GTCAAATGACCAACAATCCTG-3Ј Quantification Software (as provided by the manufacturer). A standard curve ACADM N70794 of Ct versus concentrations was obtained using several dilutions of control Forward 5Ј-ATTGCAAAGGCATTTGCTGGAGAT-3Ј RNA (10–150 ng). Control RNAs were human adult adipose tissue RNA and Reverse 5Ј-GTGTTCACGGGCTACAATAAGTCT-3Ј skeletal muscle RNA (purchased from BioChain, Haywood, CA). Real-time CYS R52654 RT-PCR of an internal standard was run separately using primers of 18s rRNA Forward 5Ј-GCCACACCGTTGAAAAG-3Ј at a ratio of 4:6 (competimer to primer) (QuantumRNA 18s internal standards; Reverse 5Ј-AGATAAGCTATTAAGTCTGCCC-3Ј Ambion, Austin, TX). Data were presented as a ratio of the relative concen- NDUFC2 AI934779 tration of the amplicon after drug and the relative concentration of the Ј Ј amplicon before drug (A-to-B ratio) for each patient. This ratio was normal- Forward 5 -AACCCAGAACCCTTACG-3 ized by the A-to-B ratio of the internal standard 18s rRNA. The average Reverse 5Ј-CTCACAGCATACAGGTAGTC-3Ј normalized A-to-B ratio for each amplicon is presented in Table 3. ATP5J AA504465 Western blot. Fat tissues were homogenized in T-Per tissue protein extrac- Forward 5Ј-TCTGTCATTCGGTCAGCC-3Ј tion reagent and the halt protease inhibitor cocktail for 60 s (Pierce, Rockford, Reverse 5Ј-GGGAAATGTATTCATGTCTG-3Ј ء IL). The soluble proteins were collected after centrifugation at 10,000g for 15 18s rRNA min at 4°C. Protein concentrations were determined with Coomassie plus protein assay reagent using BSA as standard. *Competitimer primer (cat. no. 5103G; Ambion, Austin, TX).

DIABETES, VOL. 54, MARCH 2005 881 TZDs UPREGULATE FFA OXIDATION IN FAT

TABLE 3 Effect of TZDs on microarray expression of genes sequences related to FFA uptake, oxidation, and oxidative phosphorylation in subcutaneous fat and skeletal muscle Fat Muscle Accession Fold Fold Name no. Symbol increase P increase P Cellular transport FAT/CD36 N39161 CD36 3.2 Ϯ 0.9 Ͻ0.05 0.73 Ϯ 0.14 Ͻ0.05 Fatty acid binding protein 6 Al311734 FABP6 5.3 Ϯ 2.3 Ͻ0.05 1.08 Ϯ 0.33 NS Fatty acid binding protein 4 N92901 FABP4 2.8 Ϯ 0.8 Ͻ0.05 1.46 Ϯ 0.46 NS Fatty acid CoA ligase (long-chain 3) AA424965 FACL3 2.0 Ϯ 0.2 Ͻ0.001 1.46 Ϯ 0.26 NS Carnitine palmitoyl transferase W85710 CPT 1B 2.2 Ϯ 0.9 NS 1.48 Ϯ 0.31 NS ␤-Oxidation Acyl-CoA dehydrogenase (C 4-12) N70794 ACADM 2.2 Ϯ 0.3 Ͻ0.001 1.10 Ϯ 0.32 NS Acetyl-CoA acetyltransferase (thiolase) A187665 ACAT-1 2.0 Ϯ 0.4 Ͻ0.05 1.17 Ϯ 0.25 NS Oxidative phosphorylation Cytochrome c R52654 HCS 3.1 Ϯ 0.9 Ͻ0.05 1.85 Ϯ 0.68 NS NADH dehydrogenase (ubiquinone) 1, ␤-subcomplex 5 N93053 NDUFB5 3.1 Ϯ 1.6 Ͻ0.05 1.30 Ϯ 0.29 NS NADH dehydrogenase (ubiquinone) 1, ␤-subcomplex unknown AI934779 NDUFC2 2.1 Ϯ 0.5 Ͻ0.05 1.09 Ϯ 0.09 NS ATP synthase, F1 complex, ␤-polypeptide AA708298 ATP5B 2.4 Ϯ 0.6 Ͻ0.05 1.40 Ϯ 0.31 NS ATP synthase, F0 complex, subunit F6 AA504465 ATP5J 2.2 Ϯ 0.5 Ͻ0.05 1.13 Ϯ 0.31 NS ATP synthase, F0 complex, subunit F6 AA453765 ATP5FI 2.2 Ϯ 0.6 Ͻ0.05 1.40 Ϯ 0.18 0.03 chondrial electron transport and oxidative phosphoryla- two treated with troglitazone and two with rosiglitazone) tion (cytochrome c, NADH dehydrogenase and ATPase) after TZD treatment (P Ͻ 0.03) (Fig. 2). ACADM mass (Table 3). Individual changes in the expression of eight increased 2.0-fold (P Ͻ 0.05, Fig. 2) in fat from three pa- of those gene sequences are shown in Fig. 1 (upper tients (1.7-, 1.4-, and 3.0-fold, respectively; one treated with panels). troglitazone and two with rosiglitazone). Cytochrome C Real time RT-PCR confirmed the differential induction (HCS) mass increased 5.7-fold in fat from two patients of these eight gene sequences (Fig. 1, lower panels). There (8.4- and 3.0-fold, respectively) treated with rosiglitazone. were no consistent differences between troglitazone- and Skeletal muscle rosiglitazone-induced changes (Fig. 1). The results of Effects of TZDs on gene expression. None of the gene both were therefore combined for statistical evaluation sequences whose expression was increased in fat was sig- (Table 3). nificantly induced in skeletal muscle. Expression of FAT/ Effect on FAT/CD36, ACADM, and cytochrome C pro- CD36, in fact, was significantly reduced (Table 3). tein. Protein mass was determined with immuoblots in fat Plasma FFA, FOX, and glucose uptake. As reported biopsies for three proteins for which antibodies were previously, TZD treatment in these patients resulted in a commercially available. FAT/CD36 mass increased 3.8-fold 20% lowering of basal plasma FFA levels (P Ͻ 0.02), a 13% in four patients (2.2-, 4.1-, 7.7-, and 1.2-fold, respectively; increase in basal total body FOX (P Ͻ 0.05), and a 46%

FIG. 1. Individual microarray (upper panels) and real-time RT-PCR (lower panels) values of the expression of genes or gene sequence coding for en- zymes related to fatty acid uptake, ␤-oxidation, electron transport, and ox- idative phosphorylation in eight pa- tients with type 2 diabetes before and 2 months after treatment with TZDs. E, ؍ patients treated with rosiglitazone (n 5); F, patients treated with troglitazone .(3 ؍ n)

882 DIABETES, VOL. 54, MARCH 2005 G. BODEN AND ASSOCIATES

TABLE 4 Effects of TZDs on basal plasma FFA levels, rates of FOX, and insulin-stimulated rates of glucose disappearance TZD Pre Post P Plasma FFA (␮mol/l) 721 Ϯ 51 574 Ϯ 57 Ͻ0.02 FOX (␮mol ⅐ kgϪ1 ⅐ minϪ1) 2.87 Ϯ 0.27 3.25 Ϯ 0.28 Ͻ0.05 ␮ ⅐ Ϫ1 ⅐ Ϫ1 Ϯ Ϯ Ͻ Rd ( mol kg min )* 17.1 1.9 26.4 5.0 0.01 Ϯ ϭ Data are means SE (n 8). *Rd (rate of glucose disappearance) at the end of a 4-h hyperinsulinemic-euglycemic clamp. FOX was determined, and hyperinsulinemic clamps were performed as described (14,15). From Boden et al. (14).

fatty acid binding proteins 4 and 6 (proteins involved in high-affinity, saturable, fatty acid transport and utilization [17]) of fatty acid CoA ligase (a protein that makes fatty acid transport unidirectional by converting fatty acids, which have crossed the cell membrane, into acyl-CoA derivatives that prevent their efflux out of cells) (18), ACADM and acetyl-CoA acetyltransferase (the first and the last of the 4 ␤-oxidation ), HCS, the ␤-subcomplex of NADH-dehydrogenase, and the F0 and F1 complexes of ATPase (sequences that are part of the electron trans- port chain and of oxidative phosphorylation). TZD in- duction of these gene sequences was confirmed by real- time RT-PCR (Fig. 1). These results confirm, in human fat, previous findings by others in white adipose tissue of rodents that have shown that TZDs induced expression of FAT/CD36, fatty acid binding protein, fatty acid CoA ligase, and ACADM (12,18,19). The current study is, as far as we know, the first to demonstrate that TZDs induced expression of gene sequences that are part of the electron transport chain and of oxidative phosphorylation and, more importantly, that they promote a coordinated up- regulation of all these functionally related genes in human adipose tissue. TZDs had no significant effects in skeletal muscle on expression of any of these gene sequences. This suggested ␥ FIG. 2. FAT/CD36 and acyl-CoA dehydrogenase (ACADM) proteins in that these PPAR agonists had little if any direct action on subcutaneous fat from patients with type 2 diabetes before and after 2 fatty acid uptake and oxidation in skeletal muscle and is in months of treatment with TZDs. A: Individual FAT/CD4 immunoblots of line with the observation that PPAR␥ is expressed very four patients. B: Densitometric ratios of FAT to CD36 protein normal- ized to arbitrary units by assigning the value 1 to the 36-protein ratio little in muscle (4,5). Nevertheless, direct effects of TZDs C: Individual on skeletal muscle have been reported (20) and cannot be .(4 ؍ before treatment. Shown are means ؎ SE (n ACADM immunoblots of three patients. D: Densitometric ratios of ACADM protein normalized to arbitrary units by assigning the value 1 completely ruled out with this study because it is possible to the ACADM ratio before treatment. E: Individual cytochrome C that even smaller changes than those we considered (HCS) immunoblots of two patients. F: densitometric ratios of HCS significant in the expression of several functionally related ؍ ؎ ␤ protein normalized with -actin. Shown are means SE (n 3). genes can have biological effects (21,22) and because of the relatively short period of treatment (2 months) and the increase in insulin-stimulated glucose uptake (P Ͻ 0.01) fact that most patients were on other antidiabetic medica- (Table 4) (14). tions. The TZD-mediated induction of expression of genes DISCUSSION related to FFA uptake, ␤-oxidation, and electron trans- To explore where and how TZDs improve insulin sensitiv- port (FAT/CD36, ACADM, and HCS) in fat was accom- ity, we examined gene expression in subcutaneous fat and panied by increases of comparable magnitude of the re- skeletal muscle biopsies of eight patients with type 2 spective proteins. diabetes. The key findings were that 2 months of treatment It is currently believed that in adipose tissue, PPAR␥ is with TZDs was associated with a coordinated twofold or critically important for adipocyte differentiation and fat greater increased expression in fat but not in skeletal storage, whereas in the liver, PPAR␣ plays an important muscle of a number of genes coding for proteins related to role in fatty acid oxidation (23). Our results suggest that uptake, binding, ␤-oxidation, electron transport, and oxi- this concept should be expanded to include TZD stimula- dative phosphorylation of fatty acids. Specifically, we tion of FOX in adipose tissue. How TZDs stimulated FOX, found significantly increased expression of FAT/CD36, however, is not clear. It is possible that TZDs induced

DIABETES, VOL. 54, MARCH 2005 883 TZDs UPREGULATE FFA OXIDATION IN FAT expression of all these FFA uptake and oxidation-related factors contributed to the TZD effect on insulin sensitivity. genes simultaneously. It appears more likely, however, Such factors may include the TZD-mediated increase in that the primary event was induction of the transport- the expression and protein concentration of adiponectin related genes, particularly FAT/CD36 (24,25), and that the that occurred in these patients (14). Adiponectin is a increased fatty acid flux then enhanced expression of the protein that is exclusively produced by adipocytes and oxidation-related genes. In support of this notion, Ibrahim that has been shown to increase insulin sensitivity (32–36). et al. (25) have shown that muscle targeted overexpres- TZDs have also been shown to increase the number of sion of FAT/CD36 produced a large (five- to sixfold) small adipocytes that are more insulin sensitive than large increase in FOX. This indicated not only a tight coupling adipocytes (14,37). Moreover, TZDs have been shown to between FFA uptake and oxidation but also that FFA shift the balance of adipose tissue away from visceral to uptake was rate limiting (25). subcutaneous fat (38,39), which may be important because The demonstrated TZD-induced induction of genes in- visceral fat seems to cause more insulin resistance than volved in fatty acid uptake and oxidation in adipose tissue subcutaneous fat. Lastly, TZDs activate AMP kinase and raises several questions. First, to what extent could it lower malonyl-CoA, which increases FOX (40). account for the observed increase in total body FOX? In summary, we have shown that 2 months of treatment Second, to what extent could it account for the decrease in with TZDs induced a coordinated upregulation of tran- plasma FFA? And third, to what extent could it account for scripts of proteins involved in fatty acid uptake, binding, ␤ the improvement in insulin sensitivity in these patients -oxidation, electron transport, and oxidative phosphory- (14) (Table 4)? With respect to FOX, TZDs have been lation in adipose tissue of patients with type 2 diabetes. shown to not affect FFA uptake and oxidation by the liver These changes were associated with an increase in total body FOX, a decrease in plasma FFA levels, and an in- (19). The current study showed that they also have little or crease in insulin-stimulated glucose uptake (insulin sensi- no effect on muscle. This leaves fat, where a 2- to 2.5-fold tivity). We conclude that the TZD-induced expression of increase in fat oxidation (i.e., an increase similar to the fat uptake and oxidation genes and of FOX in human observed change in gene expression) could have been adipose tissue contributed to the TZD-mediated decrease sufficient to explain the observed 13% increase in total in plasma FFA levels and the increased insulin sensitivity. body FOX. This is based on the following consideration. In In addition, microarray and RT-PCR data suggested that vitro O2 consumption by adipose tissue from obese adults TZDs improved expression of multiple genes in fat related ⅐ Ϫ1 ⅐ Ϫ1 has been determined to be 0.39 ml kg min (26). to oxidative metabolism and mitochondrial function, which Thus, O2 consumption by fat in our patients can be esti- has recently been reported to be defective in the skeletal ϳ ⅐ Ϫ1 ⅐ mated to have been 14 ml O2/min (0.39 ml O2 kg muscle of patients with type 2 diabetes (22,41–43). Ϫ1 ϫ ϳ min 35.87 kg) or 5% of their total O2 consumption, which was 273 ml/min. These in vitro results, however, ACKNOWLEDGMENTS might not accurately reflect O in vivo adipocyte O 2 2 This work was supported by National Institutes of Health consumption. Grants R01-AG15353, R01-DK58895, R01-HL0733267, and With respect to plasma FFA, our results suggested that R01-DK066003 and a mentor-based training grant from the it was the TZD-mediated increase in adipose tissue FOX American Diabetes Association (to G.B.). We thank the that was responsible for the observed 20% decrease in Molecular Core Facility of the Center for Substance Abuse basal plasma FFA levels (14). The reason for this is that (grant no. P30-DA13429) for letting us use their Roche there is usually a close and positive correlation between Light-Cycler for the real-time RT-PCR measurement. plasma FFA levels and FOX. Therefore, had the decrease We thank the nurses of the general clinical research in plasma FFA been the primary TZD effect, FOX should center for help with the studies and for excellent patient have decreased. In addition, TZDs could have lowered care, Karen Kresge for outstanding technical assistance, FFAs by increased FFA esterification via increased glyc- and Constance Harris Crews for typing the manuscript. eroneogenesis (by induction of PEPCK) (27). The results of the current and other studies suggest that REFERENCES TZDs improve insulin sensitivity indirectly and by several 1. Saltiel AR, Olefsky JM: Thiazolidinediones in the treatment of insulin mechanisms. First, there is strong evidence that TZDs in- resistance and type II diabetes. Diabetes 45:1661–1669, 1996 crease insulin sensitivity by lowering plasma FFAs (7,12). 2. Lehmann JM, Moore LB, Smith-Oliver TA, Wilkison WO, Willson TM, These FFA-induced changes in insulin sensitivity in mus- Kliewer SA: An antidiabetic thiazolidinedione is a high affinity ligand for cle are now recognized to be associated with changes in peroxisome proliferator-activated receptor ␥ (PPAR-␥). J Biol Chem 270: 12953–12956, 1995 intramyocellular diacylglycerol content and protein kinase 3. Willson TM, Cobb JE, Cowan DJ, Wiethe RW, Correa ID, Prakash SR, Beck C activity, which can cause insulin resistance by serine KD, Moore LB, Kliewer SA, Lehmann JM: The structure-activity relation- phosphorylation of the insulin receptor substrate-1, result- ship between peroxisome proliferator-activated receptor ␥ agonism and ing in interruption of insulin signaling (28). Further, by the antihyperglycemic activity of thiazolidinediones. J Med Chem 39:665– selectively stimulating FFA uptake and FOX in fat, TZDs 668, 1996 4. Tontonoz P, Hu E, Graves RA, Budavari AI, Spiegelman BM: mPPAR can decrease FFA uptake in muscle (29), which would gamma 2: tissue-specific regulator of an adipocyte enhancer. Genes Dev increase insulin sensitivity (30). 8:1224–1234, 1994 FFA-mediated changes in insulin sensitivity are dose 5. Kruszynska YT, Mukherjee R, Jow L, Dana S, Paterniti JR, Olefsky JM: dependent and proportional (31). It is therefore not likely Skeletal muscle peroxisome proliferator-activated receptor-gamma ex- ϳ pression in obesity and non-insulin dependent diabetes mellitus. J Clin that the observed 46% increase in insulin-stimulated Invest 101:543–548, 1998 glucose uptake (Table 4) can be fully explained by the 20% 6. DeFronzo RA, Jacot E, Jequier E, Maeder E, Wahren J, Felber JP: The decrease in plasma FFA levels. This suggested that other effect of insulin on the disposal of intravenous glucose: results from

884 DIABETES, VOL. 54, MARCH 2005 G. BODEN AND ASSOCIATES

indirect calorimetry and hepatic and femoral venous catheterization. 26. Hallgren P, Sjostrom L, Hedlund H, Lundell L, Olbe L: Influence of age, fat

Diabetes 30:1000–1007, 1981 cell weight, and obesity on O2 consumption of human adipose tissue. Am J 7. Santomauro ATMG, Boden G, Silva MER, Rocha DM, Santos RF, Ursich Physiol 256:E467–E474, 1989 MJM, Strassmann PG, Wajchenberg BL: Overnight lowering of free fatty 27. Tordjman J, Chauvet G, Quette J, Beale EG, Forest C, Antoine B: acids with acipimox improves insulin resistance and glucose tolerance in Thiazolidinediones block fatty acid release by inducing glyceroneogenesis obese diabetic and nondiabetic subjects. Diabetes 48:1836–1841, 1999 in fat cells. J Biol Chem 278:18785–18790, 2003 8. Ghazzi MN, Perez JE, Antonucci TK, Driscoll JH, Huang SM, Faja BW, 28. Itani SI, Ruderman NB, Schmieder F, Boden G: Lipid-induced insulin Whitcomb RW: Cardiac and glycemic benefits of troglitazone treatment in resistance in human muscle is associated with changes in diacylglycerol, NIDDM: the Troglitazone Study Group. Diabetes 46:433–439, 1997 protein kinase C, and I␬B-␣. Diabetes 51:2005–2010, 2002 9. Chaput E, Saladin R, Silvestre M, Edgar AD: Fenofibrate and rosiglitazone 29. Martin G, Schoonjans K, Staels B, Auwerz AJ: PPAR␥ activators improve lower serum with opposing effects on body weight. Biochem glucose homeostasis by stimulating fatty acid uptake in the adipocytes. Biophys Res Commun 271:445–450, 2000 Atherosclerosis 137 (Suppl.):S75–S80, 1998 10. Maggs DG, Buchanan TA, Burant CF, Cline G, Gumbiner B, Hseuh WA, 30. Boden G: A fundamental and clinical text. In Diabetes Mellitus. 3rd Ed. Inzucchi S, Kelley D, Nolan J, Olefsky JM, Polonsky KS, Silver D, Valiquett LeRoith DE, Taylor SI, Olefsky JM, Eds. Philadelphia, Lippincott, Williams, TR, Shulman GI: Metabolic effects of troglitazone monotherapy in type 2 and Wilkins, 2004, p. 979–986 diabetes mellitus. Ann Intern Med 128:176–185, 1998 31. Boden G: Free fatty acids (FFA) a link between obesity and insulin 11. Mayerson AB, Hundal RS, Dufour S: The effects of rosiglitazone on insulin sensitivity, lipolysis, and hepatic and skeletal muscle content resistance. Front Biosci 3:169–175, 1998 in patients with type 2 diabetes. Diabetes 51:797–802, 2002 32. Maeda N, Takahashi M, Funahashi T, Kihara S, Nishizawa H, Kishida K, 12. Way JM, Harrington WW, Brown KK, Gottschalk WK, Sundseth SS, Nagaretani H, Matsuda M, Komuro R, Ouchi N, Kuriyama H, Hotta K, ␥ Mansfield TA, Ramachandran RK, Willson TM, Kliewer SA: Comprehensive Nakamura T, Shimomura I, Matsuzawa Y: PPAR ligands increase expres- messenger ribonucleic acid profiling reveals that peroxisome proliferator- sion and plasma concentrations of adiponectin, an adipose-derived pro- activated receptor ␥ activation has coordinate effects on gene expression tein. Diabetes 50:2094–2099, 2001 in multiple insulin-sensitive tissues. Endocrinology 142:1269–1277, 2001 33. Hotta K, Funahashi T, Arita Y, Takahashi M, Matsuda M, Okamoto Y, 13. Racette SB, Davis AO, McGill JB, Klein S: Thiazolidinediones enhance Iwahashi H, Kuriyama H, Ouchi N, Maeda K, Nishida M, Kihara S, Sakai N, insulin-mediated suppression of fatty acid flux in type 2 diabetes mellitus. Nakajima T, Hasegawa K, Muraguchi M, Ohmoto Y, Nakamura T, Ya- Metabolism 51:169–174, 2002 mashita S, Hanafusa T, Matsuzawa Y: Plasma concentrations of a novel, 14. Boden G, Cheung P, Mozzoli M, Fried SK: Effect thiazolidinediones on adipose-specific protein, adiponectin, in type 2 diabetic patients. Arterio- glucose and in patients with type 2 diabetes. scler Thromb Vasc Biol 20:1595–1599, 2000 Metabolism 52:753–759, 2003 34. Krssak M, Petersen KF, Dresner A, DiPeitro L, Vogel SM, Rothman DL, 15. Tappy L, Owen OW, Boden G: Effect of hyperinsulinemia on urea pool size Shulman GI, Roden M: Intramyocellular lipid concentrations are correlated and substrate oxidation rates. Diabetes 37:1212–1216, 1988 with insulin sensitivity in humans: a 1H NMR spectroscopy study. Diabe- 16. Kari L, Loboda A, Nebozhyn M, Rook AH, Vonderheid EC, Nichols C, Virok tologia 42:13–116, 1999 D, Chang C, Horng WH, Johnson J, Wysocka M, Showe MK, Showe LC: 35. Combs TP, Berg AH, Obici S, Scherer PE, Rossetti L: Endogenous glucose Classification and prediction of survival in patients with the leukemic production is inhibited by the adipose-derived protein Acrp30. J Clin phase of cutaneous T-cell lymphoma. J Exp Med 197:1477–1488, 2003 Invest 108:1875–1881, 2001 17. Abumrad N, el-Maghrabi M, Amri E, Lopez E, Grimaldi P: Cloning of a rat 36. Yamauchi T, Kamon J, Minokoshi Y, Ito Y, Waki H, Uchida S, Yamashita S, adipocyte membrane protein implicated in binding or transport of long- Noda M, Kita S, Ueki K, Eto K, Akanuma Y, Froguel P, Foufelle F, Ferre P, chain fatty acids that is induced during preadipocyte differentiation: Carling D, Kimura S, Nagai R, Kahn BB, Kadowaki T: The fat-derived homology with human CD36. J Biol Chem 268:17665–17668, 1992 hormone adiponectin reverses insulin resistance associated with both 18. Martin G, Schoonjans K, Lefebvre AM, Staels B, Auwerx J: Coordinate lipoatrophy and obesity. Nat Med 7:941–946, 2001 regulation of the expression of the fatty acid transport protein and 37. Okuno A, Tamemotto H, Tobe K, Ueki K, Mori Y, Iwamoto K, Umesono K, ␣ ␥ acyl-CoA synthetase genes by PPAR and PPAR activators. J Biol Chem Akanuma Y, Fujiwara T, Horikoshi H, Yazaki Y, Kadowaki T: Troglitazone 272:28210–28217, 1997 increases the number of small adipocytes without the change of white 19. Motojima K, Passilly P, Peters JM, Gonzalez FJ, Latruffe N: Expression of adipose tissue mass in obese Zucker rats. J Clin Invest 101:1354–1361, putative fatty acid transporter genes are regulated by peroxisome prolif- 1998 ␣ ␥ erator-activated receptor and activators in a tissue- and inducer- 38. deSouza CJ, Eckhardt M, Gagen K, Dong M, Chen W, Laurent D, Burkey specific manner. J Biol Chem 273:16710–16714, 1998 BF: Effects of pioglitazone on adipose tissue remodeling within the setting 20. Wilmsen HM, Ciaraldi TP, Carter L, Reehman N, Mudaliar SR, Henry RR: of obesity and insulin resistance. Diabetes 50:1863–1871, 2001 Thiazolidinediones upregulate impaired fatty acid uptake in skeletal 39. Miyazaki Y, Mahankali A, Matsuda M, Mahankali S, Hardies J, Cusi K, muscle of type 2 diabetic subjects. Am J Physiol Endocrinol Metab 285: Mandarino LJ, DeFronzo RA: Effect of pioglitazone on abdominal fat E354–E362, 2003 distribution and insulin sensitivity in type 2 diabetic patients. J Clin 21. Brown GC: Control of respiration and ATP synthesis in mammalian Endocrinol Metab 87:784–2791, 2002 mitochondria and cells. Biochem J 284:1–13, 1992 22. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J, 40. Ruderman N, Parentki M: AMP kinase and malonyl-CoA: targets for Puigserver P, Carlsson E, Ridderstrale M, Laurilla E, Houstis N, Daly MJ, therapy of the metabolic syndrome. Nature 3:340–351, 2004 Patterson N, Mesirov JP, Golub TR, Tamayo P, Spiegelman B, Lander ES, 41. Yang X, Pratley RE, Tokraks S, Bogardus C, Permana PA: Microarray Hirschhorn JN, Altshuler D, Groop LC: PGC-1␣-responsive genes involved profiling of skeletal muscle tissues from equally obese, non-diabetic in oxidative phosphorylation are coordinately downregulated in human insulin-sensitive and insulin-resistant Pima Indians. Diabetologia 45:1584– diabetes. Nat Genet 34:267–273, 2003 1593, 2002 23. Rosen ED, Spiegelman BM: PPAR␥: a nuclear regulator of metabolism, 42. Sreekumar R, Halvatsiotis P, Schimke JC, Nair KS: Gene expression profile differentiation, and cell growth. J Biol Chem 276:37731–37734, 2001 in skeletal muscle of type 2 diabetes and the effect of insulin treatment. 24. Hajri T, Abumrad NA: Fatty acid transport across membranes: relevance to Diabetes 51:1913–1920, 2002 nutrition and metabolic pathology. Ann Rev Nutr 22:383–415, 2002 43. Patti ME, Butte AJ, Crunkhorn S, Cusi K, Berria R, Kashyap S, Miyazaki Y, 25. Ibrahimi A, Bonrn S, Blinn WD, Hajri T, Li X, Zhong K, Cameron R, Kohane I, Costello M, Saccone R, Landaker EJ, Goldfine AB, Mun E, Abumrad NA: Muscle-specific overexpression of FAT/CD36 enhances fatty DeFronzo R, Finlayson J, Kahn CR, Mandarino LJ: Coordinated reduction acid oxidation by contracting muscle, reduces plasma triglycerides and of genes of oxidative metabolism in humans with insulin resistance and fatty acids, and increases plasma glucose and insulin. J Biol Chem 274: diabetes: potential role of PGC1 and NRF1. Proc Natl Acad Sci U S A 26761–26766, 1999 100:8466–8471, 2003

DIABETES, VOL. 54, MARCH 2005 885