Preferential Channeling of Energy Fuels Toward Fat Rather Than Muscle During High Free Fatty Acid Availability in Rats

Adult male lean (Fa/?) Zucker rats (Charles River, Lecco, Italy) of 200 g body weight were housed at 24°C with a 0700–1900 light cycle and had free access to water and food pellets. All of the animals were studied in the morning after 6 h of fasting. The study protocol was approved by the institutional review board of the University of Padova.

Surgical procedure was as previously described (17). At the end of anesthesia, the animals were placed in individual cages with a single slot for the catheters, which were kept under continuous tension. The patency of the arterial catheter was maintained by means of a slow infusion of saline solution.

In a first series of experiments, four groups of six animals each were studied. Twelve hours after surgery, an infusion of 20% triglyceride emulsion (Intralipid; Pharmacia and Upjohn, Milan, Italy) plus heparin (20 U/ml) was started and continued at a rate of 0.7 ml/h for either 3 or 24 h in the two experimental groups, whereas the two control groups received saline solution plus heparin at the same infusion rate. Blood samples were collected before and after 3 and 24 h of infusion to determine blood glucose, plasma insulin, NEFA, and leptin concentrations. The rats were then submitted to a euglycemic-hyperinsulinemic clamp to evaluate overall and tissue glucose utilization (see below).

In a second series of experiments, four other groups of six animals each were infused following the same protocol as that described above. After a 3- or 24-h infusion period, the rats were killed, and tissues were removed for quantification of gene expression. FAT/CD36, GLUT4, TNF-α, PPAR-γ, leptin, UCP2, and UCP3 mRNA levels were assessed in various muscles and visceral fat by means of Northern blotting or reverse transcriptase–polymerase chain reaction (RT-PCR) analysis (see below).

The glucose utilization index was assessed at the end of a euglycemic-hyperinsulinemic clamp. A primed continuous infusion of human insulin (Actrapid HM; Novo Nordisk, Copenhagen) (dissolved in 0.9% saline solution) was administered at a rate of 3 mU/min for 120 min. Arterial blood was sampled at 5-min intervals throughout the clamp to determine plasma glucose concentrations. Starting 1 min after the beginning of the insulin infusion, glucose (20% wt/vol solution) was infused at a rate that was adjusted to maintain plasma glucose at preinfusion levels, as previously described (18).

Glucose utilization in vivo within the individual tissues was studied according to a previously described method (19,20). Briefly, a flash injection of 30 μCi of the nonmetabolizable glucose analog 2-deoxy-[1-³H]-d-glucose ([³H]-2DG) (Amersham Pharmacia, Arlington Heights, IL) was administered in 30 μl 0.9% NaCl solution through the femoral vein, and arterial blood samples were obtained at different times after the bolus administration. Upon completion of blood sampling, the rats were killed, and their adipose tissue and skeletal muscles were quickly removed, collected in liquid nitrogen, and kept frozen at −80°C for subsequent analysis. The glucose utilization index was derived from the amount of 2 deoxy-[1-H]-glucose-6-phosphate ([³H]-DGP) measured in the various tissues as previously described (19), thus using the accumulation of [³H]-DGP as an index of the glucose metabolic rate.

Total mRNA was isolated from adipose and muscle tissues using the RNAzol method (TM Cinna Scientific, Friendwood, TX), and Northern blots and hybridization were performed as previously described (21) using the cDNA probe of the rat GLUT4, UCP2, UCP3, and leptin genes. Quantitation was performed by scanning densitometry by Image Master VDS (Pharmacia-Biotechnology).

RT-PCR analyses were performed as described (16). Taq DNA polymerase (Promega) in 25 μl standard buffer (10 mmol/l Tris-HCl, pH 9, 50 mmol/l KCl, 0.1% Triton X-100, 2.5 mmol/l MgCl₂, and 200 μmol/l dNTPs) and 40 pmol of each specific sense and antisense oligonucleotide primer were used. The primer sequences chosen using the Gene-Works program (IntelliGenetics) were as follows: for FAT/CD36: 5′-AAG AGA GAT GAG GAA CCA GAG C-3′, 5′-AGT GAA GGT TCG AAG ATG GC-3′; for TNF-α: 5′-ATG AGG ACT GAA AGC ATG ATC CGG GAC GTG G-3′ and 5′-CAA TGA TCC CAA AGT AGA CCT GCC CAG ACT C-3′; for PPAR-γ: 5′-AAC TGC GGG GAA ACT TGG GAG ATT CTC C-3′ and 5′-AAT AAT AAG GTG GAG ATG CGA GCT CC-3′; and for leptin: 5′-CAC CAA AAC CCT CAT CAA GCA-3′ and 5′-AGC CTG CTC AGG GCC ACC ACC-3′. The primers for β-actin were added at the tenth cycle of each PCR amplification to avoid a plateau situation. All of the genes were amplified using 30 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s, followed by a 5-min final extension at 72°C. After amplification, 10 μl of the reaction mixture were separated by electrophoresis (1.2% agarose gel in Tris-acetate-EDTA buffer), visualized using ethidium bromide staining and a QuickImage-D system (Camberra Packard, Milan, Italy), and densitometrically analyzed using Phoretix 1D version 3.0. The number of cycles for the semiquantitative RT-PCR assay and the conditions of the reaction temperature were estimated to be optimal for a linear relationship between the amount of input template and the amounts of PCR product generated over a significant concentration range: 20–100 ng from total RNA. In particular, the linearity of the RT-PCR amplifications for all of the tested genes was measured at 15, 30, and 40 cycles (data not shown). The RT-PCR analyses were performed three times on the same sample from each of the six animals. The intra-assay coefficient of variation was <5%.

Plasma glucose was measured by means of the glucose oxidase method (Glucose Analyzer; Beckman Instruments) and insulin by means of radioimmunoassay using rat standards (Techno Genetics, Milan, Italy; Linco Research, St. Charles, MO). NEFAs and leptin were determined using commercial kits (Boehringer Mannheim, KK, Tokyo) (recombinant rat leptin; Linco Research).

The results are presented as mean values ± SE. Statistical significance was assessed by means of analysis of variance. All of the analyses were made using the Statview statistical package. The level of significance was set at P < 0.05.

Intralipid plus heparin infusion significantly increased plasma NEFA and insulin levels in comparison with the saline plus heparin–treated rats, but no significant differences in blood glucose were observed (Table 1). Intralipid treatment was also followed by a significant increase in serum leptin levels, which was evident after only 3 h of infusion (Table 1). No significant differences in food intake and body weight were noted between the two groups of animals (data not shown).

The glucose infusion rate was significantly lower in the Intralipid-infused rats than in the saline-treated rats after both 3 h of treatment (38.14 ± 1.74 vs. 18.38 ± 1.26 mg · kg⁻¹ · min⁻¹; P < 0.0002) and 24 h of treatment (38.14 ± 1.74 vs. 17.89 ± 1.44 mg · kg⁻¹ · min⁻¹; P < 0.0002). Figure 1 shows that the glucose utilization index was significantly reduced in the same animals in all of the oxidative muscles considered (the red portion of the quadriceps and gastrocnemius, and the soleus) after both 3 (Table 2) and 24 h (Fig. 1) of treatment. Increased NEFA availability was followed by an enhanced glucose uptake in visceral adipose tissue after 3 (Table 2) and 24 h (Fig. 1), but no difference was observed in white fiber–type muscles (white portion of quadriceps and gastrocnemius) or in the subcutaneous fat depot at either time point (Table 2 and Fig. 1). On the basis of these results, the following experiments were performed only in visceral fat and red fiber–type muscle (soleus).

Postinfusion RT-PCR analysis using specific primers of the RNA isolated from visceral fat, and red fiber–type muscle showed a marked increase in FAT/CD36 mRNA in comparison with the levels in the saline-treated animals after both 3 and 24 h in the soleus but only after 24 h in visceral fat (Fig. 2).

Figure 3A shows that the enhancement of glucose uptake observed after Intralipid plus heparin infusion was paralleled by increased GLUT4 expression in visceral fat after 24 h. GLUT4 mRNA levels were significantly decreased in the soleus muscle after both 3 and 24 h of Intralipid infusion (Fig. 4A).

Because it has been widely reported that TNF-α can affect adipose tissue and muscle metabolism at various levels and it has been demonstrated that TNF-α modulates GLUT4 expression (22), we measured TNF-α’s mRNA levels by RT-PCR after Intralipid plus heparin infusion. Figures 3B and 4B show that the levels of TNF-α mRNA were markedly higher in fat and soleus after both 3 and 24 h in comparison with those measured in the saline plus heparin–treated animals.

In addition, because the expression of PPAR-γ protein appears to be modulated by calorie intake of fat (23), we investigated whether increased levels of plasma fatty acids could affect PPAR-γ mRNA levels and found that these were high after both 3 and 24 h in red fiber–type muscle (Fig. 4C), but only after 3 h in visceral fat (Fig. 3C).

Moreover, the Intralipid infusion markedly increased the levels of leptin mRNA in fat after both 3 (not shown) and 24 h (Fig. 5A). Leptin gene expression in muscle was detected by RT-PCR only after 24 h of Intralipid infusion but was undetectable at baseline and after 3 h (Fig. 5B). Interestingly, mean plasma leptin levels were increased by the first 3-h Intralipid plus heparin in comparison with baseline (Table 1). These results are in line with the reported observation that high-fat feeding leads to increased leptin gene expression in rat adipose tissue and muscle (23).

Because adipose UCP2 has been found to be upregulated by a high-fat diet (24), we investigated whether increased plasma fatty acid levels affect the expression of UCP2 and UCP3 in the visceral adipose tissue and skeletal muscle of normal-weight rats. Figure 5C shows that the mRNA levels of UCP2 were increased 24 h after Intralipid plus heparin infusion in both fat and muscle with respect to those of saline-treated rats. Figure 5D shows that the mRNA levels of UCP3 were increased 24 h after Intralipid plus heparin infusion in muscle, but not in visceral fat, with respect to those of saline-treated rats. Both UCP2 and UCP3 mRNA levels were unchanged after 3-h Intralipid plus heparin infusion (data not shown).

The experimental approach used in this study mimics in a short time the increased circulating NEFA levels observed with a high-fat diet. As shown by the reduced glucose infusion rate during a euglycemic-hyperinsulinemic clamp (which was probably due to increased lipid use by red fiber–type muscles), insulin resistance develops as a result of the greater NEFA availability driven by Intralipid plus heparin infusion. Early enhancement of the gene expression of FAT/CD36 (which is suggested to be involved in NEFA uptake [25]) was evident in rat muscle after 3 h of infusion, and red fiber–type muscle insulin-dependent glucose uptake measured by [³H]-2DG technique decreased in accordance with the competitive action of NEFA on glucose metabolism (4–7).

A number of molecular mechanisms may be involved. First, GLUT4 mRNA levels were reduced in the red fiber–type muscles of the rats infused with Intralipid plus heparin after both 3 and 24 h, and TNF-α gene expression (which is extremely low even if measured by RT-PCR under baseline conditions) was markedly enhanced. This finding is relevant because, although high TNF-α expression has been demonstrated in the adipose tissue of insulin-resistant obese rodents (26) and humans (27), the fact that circulating TNF-α levels in insulin-resistant obese subjects were not detected in all studies (28) raised doubts as to whether adipose-released TNF-α really affects muscle metabolism. Our finding that high levels of circulating NEFAs can directly increase TNF-α mRNA levels in red fiber–type muscle suggests that the putatively increased TNF-α protein may act in a paracrine way in muscle tissues. Because an inverse linear relationship between the maximum glucose disposal rate and muscle TNF-α has been reported (29), this may constitute a critical point in the development of insulin resistance. It has been shown that TNF-α impairs insulin receptor signaling (30,31) and that TNF-α knockout mice have more GLUT4 protein expressed in muscle tissue (32). Our findings suggest that greater NEFA availability may shift red fiber–type muscle metabolism to preferential use of lipids rather than carbohydrates as fuel substrates. This may be at least partially due to the induction of TNF-α and the inhibition of GLUT4 gene expression.

There is a simultaneous increase in the insulin sensitivity of visceral but not subcutaneous fat. The Intralipid infusion significantly increased GLUT4 gene expression in visceral adipose tissue, thus suggesting that energy fuels may be preferentially partitioned to fat rather than muscle. It is interesting to note that both GLUT4 and TNF-α increased in visceral adipose tissue but not in red fiber–type muscles. GLUT4 favors glucose utilization, whereas TNF-α counteracts excessive fat storage by means of various mechanisms (e.g., by stimulating lipolysis [33] and decreasing the activity of adipose tissue lipoprotein lipase [34]). A similar pattern of events has been observed in the dynamic phase of several animal models of obesity in which fat accumulation is accompanied by increased glucose uptake and GLUT4 gene expression in adipose tissue, and insulin resistance develops at the muscle level (17).

The increased expression of PPAR-γ and leptin in both muscle and adipose tissue after Intralipid infusion may be a means of counterbalancing the development of insulin resistance. After Intralipid infusion, all of the studied tissues showed a higher expression of the PPAR-γ gene, a major adipogenic transcription factor that is poorly expressed in preadipocytes but turned on during adipogenesis (35) and that regulates the expression of most adipocyte genes (35). Three different groups have recently demonstrated that PPAR-γ is required for adipogenesis in vivo (36–38), and Barroso et al. (39) have recently reported that two loss-of-function PPAR-γ mutations are associated with severe insulin resistance and type 2 diabetes in humans. Although such mutations are rare, the implication that PPAR-γ is required for normal insulin sensitivity at least in some individuals is an important advance.

It can also be speculated that the high level of PPAR-γ expression observed by us may confer some protection against the development of the skeletal muscle insulin resistance associated with increased circulating NEFA levels. This is consistent with the fact that thiazolidinediones (TZDs) may improve insulin action in skeletal muscle by upregulating the expression of PPAR-γ (40). Furthermore, Kubota et al. (37) demonstrated that, although apparently normal on a regular diet, mice that are heterozygous for PPAR-γ deficiency are partially resistant to high-fat diet–induced obesity and markedly resistant to the insulin resistance that normally accompanies such obesity. Both PPAR-γ overactivity due to TZD stimulation and PPAR-γ underactivity due to haploinsufficiency may protect against NEFA-induced insulin resistance. Lowell (41) suggested that PPAR-γ can be seen as a “thrifty gene” that promotes fat storage to survive starvation when food is scarce and produces excessive fat storage leading to disease when food is plentiful. The mechanism by means of which the TZD-mediated stimulation of PPAR-γ improves insulin resistance has not been definitively established (42), but it is probably related to the transcription factor’s ability to increase the number of small adipocytes by stimulating adipogenesis and to decrease the number of large adipocytes, which are known to produce excess amounts of TNF-α and free fatty acids. Although not affecting adipogenesis, PPAR-γ haploinsufficiency limits further adipocyte hypertrophy, possibly by increasing leptin expression. Kubota et al. (37) have reported that leptin mRNA and circulating protein levels are elevated in high fat–fed PPAR-γ^+/− mice, possibly because of a decrease in the PPAR-γ–mediated inhibition of leptin gene expression.

Under our experimental conditions, high levels of circulating NEFAs induced leptin expression in both fat and muscle and increased plasma leptin levels. We did not measure body composition (in particular body fat content), but it seems unlikely that the rapid induction of leptin gene expression observed after Intralipid infusion was a consequence of an increase in fat mass. Although the effects of leptin on satiety are mainly due to its activation of hypothalamic receptors (43), there is also substantial evidence that leptin has important peripheral effects (44,45), including the stimulation of lipolysis with lipid oxidation in white fat pads ex vivo (46) and decreasing in vivo tissue triglyceride stores (47,48). It can therefore be hypothesized that, under our experimental conditions, leptin has an autocrine/paracrine function, i.e., it limits excess fat storage in adipose tissue. Furthermore, leptin mRNA, which is generally undetectable in the skeletal muscle of nonobese rodents (5,49), increased 24 h after an increase in plasma NEFA levels. In skeletal muscle, leptin may be important in the regulation of intermediate metabolism (particularly lipid oxidation and storage) (1,45,48), energy expenditure (50), and insulin signaling (51,52). Interestingly, it has been reported that, regardless of its effect on food intake, leptin leads to a marked improvement in insulin sensitivity and glucose disposal in a mouse model of lipodistrophy (53). Leptin expression in the fat and muscle of subjects with high circulating NEFA levels could therefore be involved in counterbalancing excessive fat accumulation and the development of insulin resistance.

Finally, our data demonstrate that Intralipid infusion increased UCP2 and UCP3 mRNA in both adipose tissue and muscle. These findings are in line with the enhanced UCP2 gene expression in both tissues observed in obese subjects (54,55), whose circulating NEFA levels are commonly high. It has been reported that PPAR-γ agonists induce UCP2 gene expression in adipocytes (56), and TNF-α administration increases UCP2 and UCP3 mRNA levels in rat skeletal muscle (57). This molecular event may be ultimately aimed at dissipating energy excess, but UCP2 and UCP3 may also be involved in fatty acid metabolism in the case of a large supply of nutrients (55,58).

In conclusion, our findings suggest that NEFAs are involved in the partitioning of calories to fat by modulating the expression of various genes: FAT/CD36, which enhances fatty acid flux into adipocytes and muscle cells; GLUT4, which is differently affected in skeletal muscle and adipose tissue, thus favoring the preferential use of glucose for triglyceride synthesis; TNF-α, which probably plays a crucial role in the development of insulin resistance in skeletal muscle; PPAR-γ, which promotes cell differentiation and fat storage and also modulates peripheral insulin sensitivity; and leptin and UCPs, which control energy intake and dissipation. Although it is difficult with our experimental approach to attribute the effects seen on the mRNA levels only to elevated plasma NEFAs because also insulin, leptin, and triglyceride levels are increased, nevertheless, it seems possible to speculate that the above mechanisms during a high-fat diet may unbalance energy homeostasis and channel the fuels toward adipose tissue, thus leading to obesity.

This work was supported by an educational grant from Knoll Farmaceutici (Italy) to the Center for Study and Research on Obesity, by Legge 46 (grant number 199402–1396/516) and the Ministero dell’ Università e della Ricerca Scientifica e Tecnologica ([MURST] grant number 9706240333) (to M.O.C.), by the Centro Nazionale delle Ricerche (grant number 93.04399.CT04) and MURST (to G.F.), and by the Swiss National Science Foundation (Bern) (grant number 31–53719.9 to F.R.-J).

We kindly thank Marilena Tormene and Sonia Leandri for expert technical assistance and Leopoldo Carpanese for animal care.