WY14,643, a Peroxisome Proliferator-activated Receptor (cid:1) (PPAR (cid:1) ) Agonist, Improves Hepatic and Muscle Steatosis and Reverses Insulin Resistance in Lipoatrophic A-ZIP/F-1 Mice*

WY14,643 is a specific peroxisome proliferator-acti-vated receptor (cid:1) (PPAR (cid:1) ) agonist with strong hypolipi-demic effects. Here we have examined the effect of WY14,643 in the A-ZIP/F-1 mouse, a model of severe lipoatrophic diabetes. With 1 week of treatment, all doses of WY14,643 that were tested normalized serum triglyceride and fatty acid levels. Glucose and insulin levels also improved but only with high doses and longer treatment duration. WY14,643 reduced liver and muscle triglyceride content and increased levels of mRNA encoding fatty acid oxidation enzymes. In liver, the elevated lipogenic mRNA profile (including PPAR (cid:2) ) in A-ZIP/F-1 mice remained unchanged. These results suggest that WY14,643 acts by increasing (cid:3) -oxidation rather by than decreasing lipogenesis or lipid uptake. Hyperinsulinemic euglycemic clamp studies indicated that WY14,643 treatment improved liver more than muscle insulin sensitivity and that hepatic mRNA levels of gluconeogenic enzymes were reduced. Combination treatment with both WY14,643 and

Severe adipose tissue deficiency or lipoatrophy causes a metabolic syndrome known as lipoatrophic diabetes with insulin resistance, hypertriglyceridemia, and hepatic steatosis (1,2). Lipoatrophic diabetes is intriguing because it is typically obesity, an excess of triglyceride in adipocytes, and not a deficiency of adipose tissue that causes insulin resistance.
Here we have studied the A-ZIP/F-1 mouse, which closely mimics the severe human lipoatrophic phenotype, with a near complete lack of fat, insulin resistance, diabetes, hypoleptinemia, increased appetite, hypertriglyceridemia, and hepatic steatosis (3). The A-ZIP/F-1 mouse was produced by the expression of a dominant negative protein (A-ZIP/F) selectively in adipose tissue. The A-ZIP/F molecule heterodimerizes with and inactivates certain bZIP transcription factors, including C/EBP family members. The lack of adipose tissue causes the metabolic phenotype, because fat ablation by other methods gives a similar phenotype (4 -6) and because adipose tissue transplantation reverses it (7,8).
Treatment of lipoatrophy, whether mouse or human, has been difficult. Insulin and insulin secretagogues and dietary restriction (9) are only partially effective. Modest success has been achieved with insulin-sensitizing agents, the thiazolidinediones (5,10,11) and metformin (12). Recent experiments replacing the adipose hormone leptin show great promise (13,14).
Attempts at treatment of lipoatrophic diabetes can also yield insights into the causes and mechanisms underlying the metabolic complications. A number of lines of evidence point to elevated tissue triglyceride content in non-adipose tissue as a correlate, and possibly a cause, of insulin resistance (15). WY14,643 is a PPAR␣ 1 (peroxisome proliferator-activated receptor ␣) activator that increases fatty acid oxidation by increasing transcription of genes encoding peroxisomal and mitochondrial fatty acid ␤-oxidation enzymes (16). Here we have tested the hypothesis that WY14,643 will decrease insulin resistance in A-ZIP/F-1 mice.

EXPERIMENTAL PROCEDURES
Animals-Animal handling followed National Institutes of Health guidelines, and experimental procedures were approved by the NIDDK animal care and use committee. All A-ZIP/F-1 mice were 8 -12-week-old hemizygous males on the FVB/N background. Wild type controls were matched for age and sex. Mice were typically housed 2-4 per cage, kept on a 12-hour light/dark cycle (0600 -1800), and fed NIH-07 rodent chow (12.9 kcal % fat, Zeigler Brothers Inc., Gardners, PA) and water ad libitum. The treatment diet was powdered AIN-93G (Dyets, Bethlehem, PA) (17) with or without WY14,643 (ChemSyn Laboratories, Lenexa, KS) prepared daily using a coffee grinder. Mice were sacrificed in the non-fasted state between 0900 and 1200. Tissue was fixed in neutralized 10% formalin and processed by American Histolabs (Gaithersburg, MD).
Serum samples were prepared from tail (weeks 0 and 1, awake) or retro-orbital (week 2, anesthetized) blood in non-fasting mice between 0900 and 1200. Glucose levels were measured using a Glucometer Elite (Bayer Corp., Elkhart, IN). Fatty acid (no. 1383175, Roche Diagnostics), triglyceride and ␤-hydroxybutyrate (no. 339-11 and no. 310-A, respectively, Sigma), and insulin (no. SRI-13K, Linco Research Inc., St. * 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 with 18  Charles, MO) were assayed according to the manufacturers' instructions.
Tissue Triglyceride Assay-Extraction of tissue triglycerides with chloroform/methanol was modified from Burant et al. (5). After hydrolysis with base, triglycerides were measured radiometrically using a glycerol kinase assay (18).
Euglycemic Hyperinsulinemic Clamp-The clamp protocols are based on those of Jason Kim and Gerald Shulman 2 and were performed as described (20).
Statistical Analysis-Data are expressed as means Ϯ S.E. Statistical significance between the groups was determined with SigmaStat (SPSS, Inc., Chicago, IL) using Student's t test or two-way ANOVA.

WY14,643 Lowers Serum Triglyceride, Fatty Acid, and Glucose Levels in A-ZIP/F-1 Mice in a Dose-and Time-dependent
Manner-A-ZIP/F-1 mice were treated for 2 weeks with the PPAR␣ agonist WY14,643 ( Fig. 1). WY14,643 reduced the glucose levels slightly after 1 week and to nearly wild type levels after 2 weeks. The greatly elevated insulin levels remained unchanged. In contrast to the slow and partial glucose reduction, 1 week of WY14,643 treatment normalized the triglyceride and fatty acid levels in the A-ZIP/F-1 mice. In wild type FVB/N mice, WY14,643 affected only the triglyceride levels, which were reduced by 70% at both 1 and 2 weeks of treatment.
In an independent experiment, A-ZIP/F-1 mice were treated for 2 weeks with different doses (control, 0.01%, 0.03%, and 0.1% of diet) of WY14,643. The highest dose gave results nearly identical to those in Fig. 1, except that the glucose levels were actually normal by 2 weeks of treatment (data not shown). Insulin levels did not change. Only when treatment was extended to 4 weeks did insulin levels fall (see below and Fig. 6). In marked contrast to the glucose and insulin levels, 1 week of treatment, even at the lowest dose, completely normalized the triglyceride and fatty acid levels. Thus, PPAR␣ agonist treatment rapidly improves the lipid levels of A-ZIP/F-1 mice, but the reduction in glucose and insulin levels requires a higher dose and occurs with a slower time course.
WY14,643 Decreases Liver and Muscle Triglyceride Content-WY14,643 treatment changed body and liver weights (Fig. 2). In wild type mice, WY14,643 reduced body weight, probably because of reduced adipose tissue weight. The control A-ZIP/F-1 group lost weight, possibly because of worsened diabetes due to the diabetogenic nature of the powdered AIN-93G control diet. 2 In wild type mice and A-ZIP/F-1 mice, WY14,643 treatment increased liver weight, a known effect of the hepatocyte hypertrophy and hyperplasia caused by PPAR␣ agonists in mice (21). The increase in A-ZIP/F-1 body weight was quantitatively accounted for by the increase in liver weight.
Histologically, hepatocyte hypertrophy was caused by WY14,643 in both A-ZIP/F-1 and wild type mice (Fig. 3A). The WY14,643-treated A-ZIP/F-1 livers also showed decreased vacuolization, suggestive of reduced hepatic triglyceride levels. Indeed, liver triglyceride levels were reduced by 40% in the A-ZIP/F-1 mice and 41% in the wild type controls (Fig. 3B).
Muscle triglyceride levels were measured in quadriceps mus-cle. The levels in wild type muscle are problematic due to interspersed adipose tissue (5), 3 but the lack of WAT in the A-ZIP/F-1 mice allows accurate measurement of intramyocellular triglyceride. WY14,643 treatment of the A-ZIP/F-1 mice reduced the muscle triglyceride levels by 44% (Fig. 3C).
Hyperinsulinemic Euglycemic Clamp Analysis of WY14,643 Treatment-Liver and muscle both contribute to the insulin resistance of the A-ZIP/F-1 mouse (8). We used the hyperinsulinemic euglycemic clamp to investigate the improvement caused by WY14,643 treatment. The basal pre-clamp glucose values (after a ϳ13-hour fast) were elevated in the control A-ZIP/F-1 mice but were reduced to normal in the WY14,643treated A-ZIP/F-1 mice (Table I) were similar between wild type and A-ZIP/F-1 mice despite their high glucose levels.
Base-line EGP was slightly increased in the WY14,643treated wild type mice (Fig. 4A), possibly because of the increased liver size (Fig. 2). In untreated A-ZIP/F-1 mice, basal EGP was high, as expected for their diabetes. Base-line EGP in the WY14,643-treated A-ZIP/F-1 mice was reduced. WY14,643 treatment caused increased insulin suppression of EGP in both wild type and A-ZIP/F-1 mice during the clamp (Fig. 4B). Clamp EGP in the WY14,643-treated A-ZIP/F-1 mice was similar to that of the untreated wild type mice, reflecting the improvement in insulin sensitivity. WY14,643 treatment tended to increase whole body glucose uptake, a measure of muscle insulin sensitivity (Fig. 4C). However, this was quantitatively small, suggesting that the A-ZIP/F-1 muscle remained quite insulin-resistant. Taken together, these data show that WY14,643 treatment reduces liver insulin resistance more than muscle insulin resistance.
Liver Gene Expression Changes Caused by WY14,643 Treatment-A possible explanation for the reduced tissue triglyceride levels is increased fatty acid oxidation, a known effect of PPAR␣ agonists achieved via increased transcription of peroxisomal and mitochondrial ␤-oxidation genes (22). WY14,643 treatment increased liver CPT-1 and AOX mRNA levels about 1.5-and 7-fold, respectively, in both wild type and A-ZIP/F-1 mice (Fig. 5). Small but significant increases in AOX mRNA levels were observed in muscle (Fig. 5). The data suggest that increased liver, and possibly muscle, ␤-oxidation contributes to the lowered tissue triglyceride levels. In A-ZIP/F-1 mice, WY14,643 also reduced the liver mRNA levels of the gluconeogenic enzymes, phosphoenolpyruvate carboxykinase, and glucose-6-phosphatase (Fig. 5). A-ZIP/F-1 mice have an increased lipogenic mRNA profile (Fig. 5, PPAR␥, ACC, FAS) (11), and this was not changed by WY14,643 treatment. WY14,643 increased the lipogenic mRNA in the wild type mice. WY14,643 did not affect the liver mRNA levels of two genes important for insulin signaling, IRS-2 and SREBP-1, or of glycogen synthase or of muscle GLUT4 (Fig. 5).
Combined Rosiglitazone and WY14,643 Treatment of A-ZIP/ F-1 Mice-Rosiglitazone treatment of FVB/N A-ZIP/F-1 mice has little effect on serum glucose and insulin levels (11), the net result of decreased liver and improved muscle insulin sensitivity. 4 We reasoned that adding a PPAR␣ agonist would treat the liver insulin resistance and might prove synergistic to the PPAR␥ agonist. However, treatment with both rosiglitazone and WY14,643 did not lower glucose, insulin, triglyceride, or fatty acid levels more than did treatment with WY14,643 alone, even when extended to a period of 4 weeks (Fig. 6). WY14,643 treatment decreased liver triglyceride content, whereas rosigli-tazone treatment increased it. Treatment with both rosiglitazone and WY14,643 caused a 2.1-fold increase (p ϭ ns) in liver triglyceride as compared with WY14,643 alone and a 43% decrease (p ϭ ns) in muscle triglyceride levels (Fig. 7). DISCUSSION We have shown that the PPAR␣ agonist WY14,643 greatly improves the metabolic phenotype of lipoatrophic A-ZIP/F-1 mice. The circulating triglyceride levels are normalized by a low dose and by short treatment durations. In contrast, the improvement in blood glucose levels required a higher dose and prolonged treatment. Clamp studies indicate that the improvement in the diabetes is mostly due to improved insulin sensitivity in liver, whereas muscle remained relatively resistant.
The mechanisms by which the lack of adipose tissue causes insulin resistance and hyperlipidemia are partially understood. Leptin deficiency is a major component, causing insulin resistance and increased food intake (13,14,23,24). 5 Deficiency of other adipose hormones such as adiponectin/Acrp30 may also contribute (25). It is not known how the WAT-deficient status is communicated to the liver, but the result is abnormal metabolic regulation. This includes resistance to insulin-mediated suppression of glucose production but preservation of the insulinmediated stimulation of lipogenesis and suppression of ketogenesis. It has been proposed that reduced hepatic IRS-2 levels account for the inappropriate glucose production, whereas maintained hepatic SREBP-1c explains the ongoing lipogenesis (26). Livers of lipoatrophic mice also have elevated levels of PPAR␥ (11), which contributes to the hepatic triglyceride accumulation. 3 There is evidence that excess accumulation of triglyceride in non-adipose tissue causes the tissue to be insulin-resistant. Muscle overexpression of lipoprotein lipase increases muscle triglyceride content and decreases muscle insulin sensitivity (27,28). Similarly, liver overexpression of lipoprotein lipase increases liver triglyceride content and decreases liver insulin sensitivity (27). The correlation between muscle triglyceride levels and insulin resistance has also been documented in lipoatrophic humans (29) and exists for the lipoatrophic mice. Several manipulations of the A-ZIP/F-1 mouse increase hepatic and reduce circulating/muscle triglyceride levels while decreasing liver and increasing muscle insulin sensitivity. The manipulations are rosiglitazone treatment, 4 switching the genetic  background from FVB/N to C57BL/6J, 6 and liver-specific ablation of PPAR␥. 3 PPAR␣ agonists such as WY14,643 increase fatty acid oxidation, particularly in the liver but also to a lesser degree in other tissues (30). Mice lacking PPAR␣ develop hepatic steatosis upon fasting, confirming that PPAR␣ is important for ␤-oxidation (31,32). Although better studied for their lipid-lowering effects, PPAR␣ ligands also lower glucose and insulin levels (33). Our results extend this observation to lipoatrophic diabetes, confirming evidence from one patient (34).
The phenotype of the WY14,643-treated mice provides some clues to the mechanisms underlying the physiology of lipoatrophic diabetes. The slow kinetics of the improvement in insulin sensitivity compared with the more rapid hypolipidemic effect is consistent with the hypothesis that insulin sensitization is due to a reduction in tissue triglyceride lev-els, a slow step that may take weeks to completely reach a new steady state. The mechanism by which adipose deficiency is signaled to the liver to cause increased lipogenesis is unknown (but is reversed by adipose transplantation or leptin infusion). Although WY14,643 treatment reduces liver triglyceride content and insulin resistance, it does not reduce hepatic PPAR␥, ACC, or FAS mRNA levels, suggesting that WY14,643 does not interfere with the sensing of the low WAT signal by the liver.
Addition of rosiglitazone treatment did not significantly improve the insulin sensitivity of WY14,643-treated mice. As expected, rosiglitazone did lower muscle and increase liver triglyceride levels compared with mice treated with WY14,643 alone. The lack of measured improvement in insulin sensitivity may be due to the relatively crude measurement used (glucose and insulin levels, as opposed to clamps) or the high degree of improvement already seen with this dose of WY14,643 by itself.
In conclusion, PPAR␣ agonist treatment normalized the dyslipidemia and improved the insulin resistance of A-ZIP/F-1 mice. It is likely that the improvement was caused by increased