Increased beta-oxidation in muscle cells enhances insulin-stimulated glucose metabolism and protects against fatty acid-induced insulin resistance despite intramyocellular lipid accumulation.

Skeletal muscle insulin resistance may be aggravated by intramyocellular accumulation of fatty acid-derived metabolites that inhibit insulin signaling. We tested the hypothesis that enhanced fatty acid oxidation in myocytes should protect against fatty acid-induced insulin resistance by limiting lipid accumulation. L6 myotubes were transduced with adenoviruses encoding carnitine palmitoyltransferase I (CPT I) isoforms or beta-galactosidase (control). Two to 3-fold overexpression of L-CPT I, the endogenous isoform in L6 cells, proportionally increased oxidation of the long-chain fatty acids palmitate and oleate and increased insulin stimulation of [(14)C]glucose incorporation into glycogen by 60% while enhancing insulin-stimulated phosphorylation of p38MAPK. Incubation of control cells with 0.2 mm palmitate for 18 h caused accumulation of triacylglycerol, diacylglycerol, and ceramide (but not long-chain acyl-CoA) and decreased insulin-stimulated [(14)C]glucose incorporation into glycogen (60%), [(3)H]deoxyglucose uptake (60%), and protein kinase B phosphorylation (20%). In the context of L-CPT I overexpression, palmitate preincubation produced a relative decrease in insulin-stimulated incorporation of [(14)C]glucose into glycogen (60%) and [(3)H]deoxyglucose uptake (40%) but did not inhibit phosphorylation of protein kinase B. Due to the enhancement of insulin-stimulated glucose metabolism induced by L-CPT I overexpression itself, net insulin-stimulated incorporation of [(14)C]glucose into glycogen and [(3)H]deoxyglucose uptake in L-CPT I-transduced, palmitate-treated cells were significantly greater than in palmitate-treated control cells (71 and 75% greater, respectively). However, L-CPT I overexpression failed to decrease intracellular triacylglycerol, diacylglycerol, ceramide, or long-chain acyl-CoA. We propose that accelerated beta-oxidation in muscle cells exerts an insulin-sensitizing effect independently of changes in intracellular lipid content.

Insulin resistance in skeletal muscle is frequently present in obesity and is an early characteristic of the development of type 2 diabetes mellitus. Consistent with the association between insulin resistance and obesity, multiple lines of inquiry suggest a close relationship between the development of disease and disordered lipid metabolism. In particular, in both rodent (1,2) and human (3)(4)(5)(6) studies, a correlation has been observed between the degree of insulin resistance in vivo and the triacylglycerol content of muscle cells, the primary target for insulin-stimulated glucose disposal. A possible causal relationship has been proposed in which the high plasma fatty acid levels frequently observed in the obese/insulin-resistant state drive muscle lipid accumulation that in turn causes a predisposition toward decreased insulin sensitivity and worsening of the disease, the so-called "lipotoxic" model of skeletal muscle insulin resistance. Evidence for this hypothesis comes from a series of in vivo model systems in which artificial elevations in plasma fatty acid levels over several hours induced lipid build-up and insulin resistance in muscle tissue (7)(8)(9)(10). Parallel observations have been made in vitro in which incubation of model muscle cells in tissue culture with fatty acids has analogous consequences (11)(12)(13).
Although the association between increased intramyocellular lipid and the development of insulin resistance is compelling, the mechanism of the effect is far from clear. For example, triacylglycerol accumulation in muscle cells is not invariably associated with insulin resistance. Notably the muscle of trained endurance athletes has been shown to be highly insulin-sensitive despite the presence of high levels of intramyocellular triacylglycerol (14), and shorter term (4-week) exercise training in humans appears to improve muscle insulin sensitivity in the absence of measurable changes in muscle triacylglycerol content (15). To account for this apparent discrepancy, it has been proposed that triglyceride accumulation within insulin-resistant muscle is merely a marker for some other harmful fatty acid-derived metabolite(s). In particular, diacylglycerol (16), ceramide (17), and long-chain acyl-CoA (18) are also elevated in certain insulin-resistant muscle models. Each of these has been implicated in mediating the negative effects of lipids on insulin signaling via activation of isoforms of protein kinase C (16 -21) and inhibition of activation of protein kinase B (PKB 1 ; Refs. 11, 13, and 22).
In addition to lipid accumulation, both in vivo studies in humans (23,24) and ex vivo biochemical analysis of human muscle biopsies (25)(26)(27)(28) have indicated that the obese/insulin-resistant phenotype is associated with decreased mitochondrial fatty acid oxidative capacity in skeletal muscle characterized by decreased myocellular mitochondrial number, unusual mitochondrial morphology, and lower than normal levels of mitochondrial enzymes, including citrate synthase (25)(26)(27)(28), succinate dehydrogenase (14,29), and carnitine palmitoyltransferase (CPT; Refs. 26 and 27). It is likely, therefore, that a decreased ability to oxidize fatty acids in the muscle of these individuals may exacerbate lipid accumulation particularly in the context of elevated plasma free fatty acid levels.
If this mechanism is correct, we hypothesized that, conversely, an intervention that promoted ␤-oxidation in muscle cells should protect against fatty acid-induced insulin resistance by ameliorating intramyocellular lipid accumulation. We tested this hypothesis by developing a strategy to accelerate ␤-oxidation in rat L6 myotubes, a muscle-derived cell line that displays a robust response to insulin (30). Specifically we overexpressed the enzyme CPT I, the first step in the CPT pathway that permits the entry of long-chain acyl-CoA into the mitochondrial matrix for ␤-oxidation (31). Located on the outer mitochondrial membrane, CPT I acts on cytosolic long-chain acyl-CoA, transferring the acyl group from CoA to carnitine. The acyl-carnitine formed can enter the matrix (via a specific transporter) where acyl-CoA is regenerated through the action of a distinct gene product, CPT II, that reverses the CPT I reaction. CPT I is tightly regulated by virtue of its sensitivity to inhibition by malonyl-CoA and is believed to be a major control point for ␤-oxidation in all tissues where fatty acid oxidation occurs (31). CPT I is known to exist in at least two isoforms, denoted L-CPT I and M-CPT I, each found in several tissue and cell types (31,32). 2 We determined the effect of CPT I overexpression on ␤-oxidation, lipid accumulation, the metabolic response to insulin, and the ability of the hormone to stimulate certain signaling pathways involved in insulin regulation of muscle glucose metabolism and believed to be relevant to fatty acid-induced insulin resistance. These studies demonstrated that overexpression of CPT I increases the rate of fatty acid oxidation by muscle cells and significantly improves the insulin response, metabolically and at the level of the insulin signaling pathway, and furthermore is protective against fatty acid-induced insulin resistance. However, remarkably these improvements were not associated with decreased accumulation of intramyocellular lipids. The data provide compelling evidence for a beneficial effect of increased ␤-oxidation on insulin sensitivity by a mechanism independent of lipid accumulation.
Stock cultures of other myocyte cell lines were grown, reseeded for experimentation, and induced to differentiate as above with the following variations. For C2C12 cells, Growth Medium was Eagle's minimum essential medium, 10% FBS, penicillin, streptomycin with Differentiation Medium being the same but with 1% FBS (11). For MC13 cells Growth Medium was Dulbecco's modified Eagle's medium, 10% FBS, 10% horse serum, 0.5% (v/v) chicken embryo extract, penicillin, streptomycin with each of the sera reduced to 1% in the Differentiation Medium (34).
Primary cells were prepared hind limb muscle of neonatal rats enriched for myoblast content and grown on collagen-coated dishes by the method of Springer et al. (35). Growth Medium was Ham's F-10 nutrient mixture containing 20% (v/v) FBS, 2.5 ng/ml basic fibroblast growth factor, and antibiotics as above. Differentiation Medium was Dulbecco's modified Eagle's medium, 5% (v/v) horse serum plus antibiotics.
Adenovirus Preparation and Transduction of L6 and C2C12 Cells-Adenoviruses (a kind gift of the laboratory of J. Denis McGarry, University of Texas Southwestern, Dallas, TX) contained the cDNAs encoding Escherichia coli ␤-galactosidase or rat L-CPT I, M-CPT I, or CPT II under the control of the early cytomegalovirus promoter (36). Viral stocks/cell lysates were prepared with the human embryonal kidney cell line 293A as host (36) and used directly for cell transduction. For functional titration of AdV-L-CPT I and all subsequent experiments with that virus, a concentrated viral stock was prepared (36). For all experiments, an equivalent titer of AdV-␤-gal, prepared in an identical fashion, was used to transduce control cells (lysate or concentrate as appropriate). Adenoviral transduction of L6 cells was performed after 6 days of differentiation or 4 days for C2C12. Cells were incubated at 37°C for 1 h in 1 ml of serum-free medium to which was added 100 l of lysate or a volume of concentrated virus as indicated in the figure legends. The medium was then aspirated, the cells were washed with phosphate-buffered saline (PBS), and fresh Differentiation Medium was added. In all cases, culture was continued for a further 24 h to allow expression of the virally encoded proteins before commencing each of the procedures described below.
Northern Blot Analysis-Total RNA was extracted from fully differentiated myotubes (after 6 days of differentiation) or adult rat heart using the TRIzol method (Invitrogen). RNA species were separated by electrophoresis on 1% agarose gels, and messages for L-or M-CPT I were detected using 32 P-labeled single-stranded DNA probes that hybridize efficiently with both the rat and mouse messages (32).
CPT Assay-Activities of CPT I and II were determined by a radiochemical assay in the direction of acylcarnitine formation as described previously (37). Briefly cells were harvested in a buffer containing 150 mM KCl, 5 mM Tris-HCl, pH 7.2, and broken with a glass homogenizer. The cell homogenate, in which the mitochondria remain largely intact, was used directly for assay of CPT I. For assay of CPT II, a portion of the homogenate was adjusted to 1% (w/v) with regard to the detergent octylglucoside, which inactivates CPT I and releases CPT II from the mitochondrial matrix in active form (38).
Fatty Acid Oxidation-Fatty acids (palmitate, oleate, or octanoate) were preconjugated with essentially fatty acid-free bovine serum albumin (BSA) to generate a stock solution of 25% (w/v) BSA, 4 mM fatty acid in serum-free medium (39 Twenty-four hours following viral transduction, the culture medium was replaced with fresh Differentiation Medium containing labeled fatty acids (triplicate wells for each condition), and culture continued for a further 18 h. Identical incubations were conducted on parallel plates that contained no cells. The medium was then collected, and tritiated water was determined by the vapor-phase equilibration method of Hughes et al. (40). Cells were washed three times with 2 ml of ice-cold PBS and collected in 1 ml of 1 N NaOH for measurement of protein content by the bicinchoninic acid method (Pierce).

2-[ 3 H]Deoxyglucose
Uptake-Palmitate was preconjugated with BSA as described for fatty acid oxidation experiments (above) but without isotope. A parallel solution was prepared using BSA in the absence of palmitate for addition to control cells. Twenty-four hours after viral 2 Recently an mRNA encoding a putative third isoform of CPT I has been identified in rat brain and testis. However, the catalytic function of the encoded protein is unclear (33). transduction, culture medium was replaced with fresh Differentiation Medium containing 1.25% (w/v) BSA, 0.2 mM palmitate or BSA alone. After 14 h, the medium was changed to serum-free medium with the same additions, and culture continued for a further 4 h. Cells were then incubated for 20 min in fresh medium in the absence or presence of 100 nM insulin. 2-[ 3 H]Deoxyglucose ([ 3 H]DOG) uptake was determined in triplicate wells for each condition by the method of Tremblay and Marette (41) over 5 min in the presence of 10 M (0.5 Ci/ml) 2-deoxyglucose, and protein was measured as above.
[ 14 C]Glucose Incorporation into Glycogen-Virally transduced L6 or C2C12 cells were preincubated with BSA or BSA/palmitate in Differentiation Medium, as described above, for 16 h followed by 2 h in serum-free medium with the same additions and a further 60 min in fresh medium containing 2 Ci/ml D-[U-14 C]glucose in the presence or absence of 100 nM insulin. Triplicate wells were used for each condition. After incubation with the isotope, the medium was removed, and the monolayers were washed three times with 2 ml of ice-cold PBS. To solubilize the cells, 500 l/well 1 N NaOH was added to each well, and the plates were incubated for 1 h at 50°C. A 300-l portion was then collected (the remainder was used to measure protein content), and 10 l of 50 mg/ml unlabeled glycogen were added as carrier. Glycogen was precipitated by the addition of 750 l ethanol (prechilled to Ϫ20°C) and incubation at Ϫ20°C for 30 min. Samples were centrifuged at 18,000 ϫ g for 10 min at 4°C, the pellets were redissolved in 250 l of 1 N NaOH by heating at 60°C for 30 min, and 200 l were taken for liquid scintillation counting.
Immunoblot Analysis of Signaling Molecules-Twenty-four hours after transduction with adenoviruses, L6 cells were incubated with or without 0.2 mM palmitate for 14 h in Differentiation Medium followed by 4 h in serum-free medium with the same additions and 20 min in the absence or presence of 100 nM insulin. The medium was then removed, the cells were rapidly washed twice with ice-cold PBS, and 150 l of lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride) was added. After 10 min on ice, the lysis buffer was collected, and triplicate wells were pooled. Samples were sonicated (three brief pulses using a Virsonic 100 sonicator; power setting, 5) followed by centrifugation at 18,000 ϫ g for 10 min at 4°C. Pellets were discarded, and the solubilized proteins (ϳ15 g/sample) were resolved by 10% SDS-PAGE and electrotransferred onto polyvinylidene difluoride filters for immunoblotting by conventional means. After probing with phosphorylation-specific antibodies the membranes were stripped and reprobed with antibodies against total kinase proteins and actin. Signals were detected by chemiluminescence (ECL Plus detection system, Amersham Biosciences) and exposure to x-ray film to produce bands within the linear range. Quantification was accomplished with NIH Image software.
Triacylglycerol Assay-Triacylglycerol was extracted and measured by a variation of the method of Danno et al. (42). L6 cells transduced with AdV-␤-gal or AdV-L-CPT I were incubated for 18 h in Differentiation Medium in the absence or presence of 0.2 mM palmitate. The monolayers were harvested by scraping (six wells for each condition pooled in 1 ml of ice-cold PBS), centrifuged at 18,000 ϫ g for 2 min at 4°C, and resuspended in 200 l of 2 M NaCl, 2 mM EDTA, 50 mM sodium phosphate, pH 7.4. Samples were then sonicated (two pulses of 5 s each, settings as above), and 20 l were removed for protein assay. Tertiary butanol (200 l) was added to the remaining material followed by vortex mixing for 30 s. This was followed by the addition of 100 l of Triton X-114:MeOH (1:1) and vortex mixing as before. The phases were separated by centrifugation at 18,000 ϫ g at room temperature, and 250 l of the upper phase were collected. The glycerol component of triacylglycerol was determined using a kit (GPO Trinder triacylglycerol kit, Sigma).
Ceramide and Diacylglycerol Assay-Cells (preincubated plus or minus palmitate as above) were washed twice with ice-cold PBS, and lipids were extracted using a modification of the method of Bligh and Dyer (43). Briefly cells were scraped with 1.0 ml of methanol (three pooled wells) containing 0.2 mg/ml of ␤-hydroxytoluene as antioxidant. Chloroform was added (2.0 ml), and sample containers were filled with N 2 gas and placed at 4°C overnight. The samples were then centrifuged at 500 ϫ g for 5 min, and the organic phase was collected, washed with 600 l of 0.15 N NaCl, and dried in vacuo. The pellet was dissolved in 500 l of 1.0 N NaOH for protein determination.
Diacylglycerol and ceramide were phosphorylated using sn-1,2-diacylglycerol kinase in the presence of [ 32 P]ATP, and the labeled products were separated by thin layer chromatography according to the method of Preiss et al. (44,45) as modified by Martin and colleagues (46). Plates were exposed to x-ray film, areas of the plates corresponding to phos-phorylated diacylglycerol and ceramide were scraped into scintillation vials, and radioactive content was determined with 1,2-dioleoyl-snglycerol and N-palmitoyl-D-sphingosine as standards (62-2000 pmol).
Long-chain Acyl-CoA Assay-L6 Cells were grown, differentiated, transduced with adenoviruses, and preincubated plus or minus palmitate as described above but using 10-cm tissue culture dishes. The monolayers were washed twice with ice-cold PBS and incubated for 10 min on ice with 10 ml of 1% (w/v) trichloroacetic acid. Cells were harvested by scraping and centrifugation, and pellets were washed twice with ice-cold PBS before storage at Ϫ80°C. For assay, pellets were resuspended in 250 l of reconstitution solution (10 mM K 2 HPO 4 , 10 mM dithiothreitol, pH 12) and heated to 55°C to hydrolyze longchain acyl-CoA, releasing free CoASH. Samples were then neutralized (pH 6 -7) and centrifuged in an airfuge at 20 p.s.i. for 20 min. Supernatants were analyzed by an NAD-linked enzymatic assay that measures the free CoASH with the use of ␣-ketoglutarate dehydrogenase (47). In this system, NADH fluorescence is measured (excitation, 340 nm; emission, 460 nm), and the increase in fluorescence is proportional to the conversion of CoASH to succinyl-CoA. For standardization, the concentration of CoASH in a standard solution was determined by measuring A 340 of reaction mixture in the absence of enzyme and then taking readings at 1-min intervals after adding enzyme until the reaction was completed. Enzyme blanks were determined by adding the enzyme to cuvettes containing distilled water rather than CoASH in the reaction mixture. After correction for the enzyme blank, the concentration of standard CoASH was determined from the difference between the initial A 340 without enzyme and the final A 340 of the completed reaction. The percent recovery was calculated using a known standard of palmitoyl-CoA.
Statistical Analysis-Data are expressed as mean Ϯ S.E. For each data group, the presence of statistically significant differences were determined using non-parametric analysis of variance (Kruskal-Wallis procedure) followed by individual pairwise comparisons by the Student-Newman-Keuls method. Statistical significance was considered to be present at p Ͻ 0.05.

Effects of Overexpression of CPT I Isoforms and CPT II on Enzyme Activity and Fatty Acid Oxidation Rates-Previous
studies with mouse C2C12 muscle cells in our laboratory had indicated that L-CPT I was the primary CPT I isoform expressed in those cells even after in vitro differentiation. Hence, before performing the adenoviral overexpression studies presented below, we determined whether this was also the case in L6 cells. Northern blot analysis of expression of CPT I isoforms in four culture models of skeletal muscle cells is shown in Fig.  1. In the undifferentiated or fully in vitro differentiated states, L-CPT I mRNA was the only CPT I mRNA detected in L6, C2C12, or MC13 cells. A faint signal for M-CPT I was observed only in differentiated primary rat skeletal myoblasts. Fig. 2A shows the activities of CPT I and CPT II in fully differentiated L6 myotubes 24 h after transduction with adenoviruses encoding rat L-CPT I, M-CPT I, or CPT II. Control cells received the same dose (multiplicity of infection, ϳ20) of AdV-␤-gal. There were no differences in CPT I or CPT II activities between cells that received AdV-␤-gal and untransduced cells (not shown). Transduction with the L-and M-CPT I viruses resulted in significant increases of 2-and 5-fold in CPT I activity, respectively, relative to AdV-␤-gal-transduced cells. AdV-CPT II caused a 12-fold increase in CPT II activity without affecting the level of CPT I.
The effects on oxidation of exogenous fatty acids is shown in Fig. 2B. The 2-fold overexpression of L-CPT I resulted in significant and proportional (2-fold) increases in oxidation of the long-chain fatty acids palmitate and oleate. In contrast, M-CPT I overexpression had no effect on the ␤-oxidation rate. As predicted, an increased activity of CPT II also had no effect. Oxidation of the medium-chain fatty acid octanoate, which enters the mitochondrion independently of the CPT pathway, was unaffected by overexpression of any of the CPT enzymes.
Since AdV-L-CPT I resulted in a smaller increment in activity than either the M-CPT I or CPT II viruses, we performed a functional titration to determine whether higher levels of activity could be achieved by increasing the viral dose. Fig. 3 shows the relationship between L-CPT I activity and palmitate oxidation in cells transduced with AdV-L-CPT I at increasing multiplicity of infection. The almost linear relationship observed in the lower portion of the curve confirms that the activity of CPT I is a major determinant of the fatty acid oxidation rate in these cells. However, at higher doses, the increase in oxidation with viral dose diminishes, presumably reflecting saturation of distal elements of the pathway. Accordingly we chose to use the highest dose shown for further experiments (control cells always receiving the same dose of AdV-␤-gal).
Effects of Fatty Acids and CPT I Overexpression on the Metabolic Response to Insulin-As shown in Fig. 4A, basal (noninsulin-stimulated) incorporation of [ 14 C]glucose into glycogen was not significantly affected by preincubation with palmitate or overexpression of L-CPT I. However, the capacity of insulin to increase incorporation was greater in L-CPT I-overexpressing cells than in control (AdV-␤-gal-transduced) cells (83 and 62% increases, respectively). Hence, under insulin-stimulated conditions, both total (Fig. 4A) and net insulin-stimulated glucose incorporation (Fig. 4B)  decreases in both total incorporation in the presence of insulin (ϳ25-40% decreases (Fig. 4A) and the net insulin-stimulated increment (ϳ60% decreases, Fig. 4B) indicating fatty acid-induced insulin resistance. However, when both interventions were present, i.e. palmitate preincubation and L-CPT I overexpression, the values for both total incorporation and the insulinstimulated component were greater than in palmitate-treated control cells. Similar experiments were also performed in differentiated mouse-derived C2C12 skeletal muscle cells (Fig. 5). Comparison of Figs. 4 and 5 reveals that an almost identical result was obtained. The smaller increment in the insulin-stimulated component of incorporation caused by L-CPT I overexpression in the absence of palmitate in C2C12 cells when compared with L6 cells may reflect the 1.7-fold greater endogenous activity of the enzyme in C2C12 cells (not shown).
As further confirmation of fatty acid-induced insulin resistance, uptake of [ 3 H]DOG was measured in L6 cells under the same culture conditions as [ 14 C]glucose incorporation into glycogen. As with glycogen synthesis, neither L-CPT I overexpression nor palmitate preincubation had significant effects on basal [ 3 H]DOG accumulation (Fig. 6A). Insulin increased uptake by 48% in control cells and 46% in L-CPT I-overexpressing cells. After palmitate preincubation, insulin increased uptake by only 25 and 39%, respectively. Consequently the hormonestimulated component (Fig. 6B) was significantly decreased only in control cells since the smaller effect of palmitate in L-CPT I-overexpressing cells did not achieve statistical significance. As with [ 14 C]glucose incorporation into glycogen above, the net insulin-stimulated component of [ 3 H]DOG uptake in L-CPT I-overexpressing cells that had been pretreated with palmitate was significantly greater than that in control palmitate-treated cells.
To confirm that the effects on the metabolic response to insulin were specifically the consequence of increased ␤-oxidation and did not result from nonspecific effects of overexpression of mitochondrial protein, the effect of insulin on incorporation of [ 14 C]glucose into glycogen was also measured in L6 cells that had been transduced with AdV-M-CPT I, a virus that resulted in substantial expression of enzyme activity but had no effect on ␤-oxidation (Fig. 2). In contrast to the effect of L-CPT I (Fig. 4), M-CPT I overexpression alone had no effect on insulin-stimulated incorporation and was not protective against palmitate-induced loss of the insulin response (Fig. 7).

Effects of Fatty Acids and CPT I Overexpression on
Signaling Pathways-We determined the effect of L-CPT I overexpression on the phosphorylation state of PKB (an important mediator of the effects of insulin on muscle glucose metabolism, Refs. 48 and 49) at two sites critical for activation, serine 473 and threonine 308. As above, experiments were performed with or without 0.2 mM palmitate preincubation for 18 h.
The data for phosphorylation of serine residue 473 are summarized in Fig. 8A. In AdV-␤-gal-transduced cells, palmitate preincubation caused a 20% decrease in the insulin-stimulated level of PKB phosphorylation. In L-CPT I-overexpressing cells without palmitate preincubation, insulin-stimulated PKB phosphorylation was similar to control cells, although basal phosphorylation displayed a tendency toward an increased level that did not achieve statistical significance. However, in L-CPT I-overexpressing cells, insulin-stimulated PKB phosphorylation was unchanged by palmitate preincubation and was significantly greater than in fatty acid-treated control cells. In contrast to the effects observed for serine 473, phosphorylation of threonine 308 showed no significant change with L-CPT I overexpression or palmitate preincubation (not shown). Neither base-line nor insulin-stimulated levels of phosphorylated glycogen synthase kinase 3 were affected by palmitate pretreatment or L-CPT I overexpression (Fig. 8B).
In the absence of palmitate preincubation, insulin significantly increased the level of phosphorylation of p38MAPK (Fig. 8C), and insulin stimulation was significantly greater in the presence of L-CPT I overexpression. Base-line p38MAPK phosphorylation was increased 5-fold by incubation with palmitate, and insulin had no further effect (Fig. 8C). After palmitate treatment, there was a tendency toward a decreased signal in L-CPT I-overexpressing cells that did not reach statistical significance.
Since the proinflammatory IKK/IB/NFB pathway has also been implicated in fatty acid-induced insulin resistance (50), we determined whether CPT I overexpression influenced the activity of this pathway. Specifically we measured the level of the NFB inhibitor IB, which becomes degraded upon activation of the pathway. As predicted, palmitate induced degradation of IB, but CPT I overexpression had no effect on the level of that protein in the absence or presence of fatty acid preincubation (Fig. 8D).
Effects of Fatty Acids and CPT I Overexpression on Intracellular Lipids-Incubation with 0.2 mM palmitate for 18 h resulted in ϳ3-fold increases in cellular triacylglycerol in AdV-␤-galor AdV-L-CPT I-transduced L6 cells (Fig. 9A). Palmitate preincubation also resulted in 5.8-and 3.7-fold increases in diacylglycerol, respectively, and 3-fold increases in ceramide content in both cases (Fig. 9, B and C). In the absence or presence of palmitate preincubation, the levels of each of these lipids were not significantly different between AdV-␤-gal-and AdV-L-CPT I-transduced cells. In contrast, palmitate preincubation had no effect on the cellular content of long-chain acyl-CoA (Fig. 9D). However, surprisingly, L-CPT I overexpression increased the content of this metabolite by 80 and 112% in the absence or presence of palmitate preincubation, respectively (although this achieved statistical significance only in the former case).

DISCUSSION
Characterization of the Model-To determine the effects of increased ␤-oxidation capacity on fatty acid-induced insulin resistance, we designed an approach (overexpression of CPT I) that would specifically promote flux through the mitochondrial fatty acid oxidation pathway in the absence of other metabolic perturbations. L6 cells were chosen as a model system since they display a robust response to insulin in terms of glycogen synthesis and glucose uptake (Figs. 4, 6, and 7; Ref. 30), are easily transduced with the adenovirus, and displayed a significant and reproducible diminution of the insulin response after preincubation with moderate (0.2 mM) concentrations of palmitate (Figs. 4, 6, and 7).
The Northern blot analysis presented in Fig. 1 revealed that the endogenous CPT I isoform expressed in several tissue culture models of muscle cells was L-CPT I. This observation is similar to what we have seen previously with in vitro differentiated rat preadipocytes, which also express M-CPT I only as a minor isoform, although M-CPT I is the dominant isoenzyme in mature rat adipocytes in vivo (32). That none of these tissue culture systems possesses significant levels of M-CPT I suggests that expression of that gene depends on conditions absent in the culture environment. A natural corollary of this finding is that the prevailing concentration of malonyl-CoA in L6 cells is presumably appropriate for regulation of L-CPT I (IC 50 ϳ 3-10 M) and too high to permit significant flux through M-CPT I (IC 50 ϳ 0.03 M) as supported by our finding that despite substantial overexpression of M-CPT I there was no effect on the oxidation of long-chain fatty acids. Also as predicted, a 12-fold overexpression of CPT II had no effect on oxidation. In contrast to both these observations, overexpression of L-CPT I resulted in increases in oxidation of palmitate and oleate (but not octanoate) that were approximately proportional to enzyme activity up to at least a 2-fold increase, demonstrating a high degree of control over the entry of long-chain fatty acids into the ␤-oxidation pathway (Figs. 2 and 3). Therefore, L-CPT I overexpression achieved a specific increase in the capacity for mitochondrial oxidation of long-chain fatty acids, and the effects of L-CPT I overexpression on several markers of the insulin sensitivity of the cells (below) can reliably be attributed to the increase in ␤-oxidation flux and were not a result of nonspecific alteration in mitochondrial function. That oxidation of octanoate was not affected also supports the notion that the total energy expenditure by the cells was not increased by L-CPT I overexpression.
Effects of L-CPT I Overexpression on Insulin Action in the Absence of Palmitate Preincubation-Overexpression of L-CPT I was alone sufficient to increase [ 14 C]glucose incorporation into glycogen in response to insulin (although not in its absence), although the concurrent increase in insulin-stimulated [ 3 H]DOG uptake did not achieve statistical significance (Figs. 4  and 6). Notably the enhancement of insulin-stimulated glycogen synthesis was not attributable to a change in the phosphorylation state of PKB or decreases in the levels of triacylglycerol, diacylglycerol, ceramide, or acyl-CoA (Figs. 8 and 9) and was, therefore, by a mechanism independent of these parameters frequently considered to be important in fatty acid-induced insulin resistance. That L-CPT I overexpression did not alter the lipid content of the cells is consistent with the data of Rubi et al. (51) who reported that overexpression of L-CPT I in INS1E cells (a pancreatic ␤-cell line) increased the ␤-oxidation rate without changing the cellular triacylglycerol content. In contrast, L-CPT I overexpression was alone sufficient to significantly enhance the ability of insulin to cause phosphorylation of p38MAPK, a signaling molecule believed to be involved in insulin stimulation of glucose uptake in muscle cells probably through increasing the intrinsic activity of plasma membrane Glut4 glucose transporters (30). Interestingly the ability of insulin to stimulate phosphorylation of p38MAPK is reported to be lost in the skeletal muscle of type 2 diabetic humans (53). In contrast, the IKK/IB/NFB pathway, known to be activated by palmitate, was unaffected by CPT I overexpression, precluding blockage of this pathway as an explanation for the beneficial effect of an increased capacity for ␤-oxidation.
Although the mechanism is uncertain, several other factors, such as adenosine (54,55), cellular redox potential (56), and cellular ATP levels (57), have been proposed to have effects on insulin-stimulated glucose usage in muscle cells. Also overexpression of uncoupling protein 3 in muscle cells has been shown to preferentially increase fatty acid oxidation (relative to glucose oxidation) in association with a decreased mitochondrial membrane potential, an increased ATP/ADP ratio, and enhanced glucose uptake (58) probably mediated by a pathway involving phosphoinositide 3-kinase but independent of PKB (59). Since accelerated ␤-oxidation may be expected to have a direct influence on the energy state of the cell, it is possible that changes in ATP, ADP, or reducing equivalents may be involved in linking the ␤-oxidation rate with the insulin response.
Increased ␤-oxidation may also be predicted to have mass action effects on the fate of cellular glucose. Thus, according to the Randle hypothesis (60), an increase in ␤-oxidation in muscle will decrease glucose oxidation with concurrent accumulation of glucose 6-phosphate, inhibition of hexokinase, and decreased glucose uptake. Non-oxidative metabolism (storage) of glucose is also, therefore, predicted to be decreased. Indeed a variety of model systems have largely confirmed the validity of the Randle mechanism at least in terms of the acute effects of fatty acids on muscle glucose metabolism and storage (61,62). It is likely that, in our model, an increased fatty acid oxidation rate may have inhibited glucose oxidation as predicted by the Randle cycle. However, the further prediction of the Randle cycle that glucose uptake and storage would also be inhibited is clearly inconsistent with our data since, in our studies, uptake and storage were in every case either unchanged or increased by L-CPT I overexpression. Additionally a mass action effect directing glucose 6-phosphate into glycogen synthesis rather than glycolysis would be expected to be evident in the absence of insulin as well as in its presence, and this was not observed. However, in previous studies, the ␤-oxidation rate was increased by increasing the availability of fatty acids, a situation different from the experiments presented here in which we measured the effects of greater ␤-oxidation rates without increasing the fatty acid supply. Considerable further experimentation will be required to determine whether the effect of increased ␤-oxidation on glycogen synthesis is mediated directly by alterations in metabolic fluxes or by unidentified signals that link the rate of fatty acid oxidation to the insulin signaling pathways. Whichever is the case, our data point to a distinct influence of the ␤-oxidation rate on insulin sensitivity in muscle, which is not dependent on fluctuations in intracellular lipids.

Effects of L-CPT I Overexpression on Markers of the Insulin Response in the Context of Fatty Acid-induced Insulin
Resistance-Our key finding of a distinct beneficial effect of enhanced ␤-oxidation on insulin sensitivity was further supported by analysis of the effect of L-CPT I overexpression in parallel with fatty acid-induced, lipid accumulation-associated insulin resistance.
The latter was induced by incubation with palmitate, which has been shown previously to cause insulin resistance at the level of glycogen synthesis in C2C12 cells (11), and with greater potency than unsaturated fatty acids. Preincubation with 0.2 mM palmitate for 18 h significantly decreased insulin stimulation of [ 14 C]glucose incorporation into glycogen and [ 3 H]DOG uptake (by ϳ50%; Figs. 4, 5, and 6), decreased insulin-stimulated serine phosphorylation of PKB, and induced degradation of IB (Fig. 8). No significant effect was observed on threonine phosphorylation of PKB, suggesting that the lipotoxic effect of fatty acids on PKB-mediated insulin signaling, in L6 cells at least, may be the result of a specific decrease in serine phosphorylation. However, phosphorylation at both sites is required for PKB activation (63). This is the first demonstration that palmitate can induce a par-tial insulin-resistant state in L6 cells, providing further evidence for the generality of the fatty acid effect seen in other models. In addition to its effects on PKB, palmitate markedly increased phosphorylation of p38MAPK and at the same time abolished any effect of insulin on that parameter (Fig. 8C). This is consistent with the observations in C2C12 cells where a severalfold increase in p38MAPK phosphorylation is reported in response to palmitate preincubation (11). Although at first paradoxical (that phospho-p38 should be increased by insulin in the one case and also increased in palmitate-induced insulin resistance in the other), our data are consistent with the recent study of de Alvaro et al. (64) that has shown that activation of different isoforms of p38MAPK are associated with insulin stimulation of glucose transport (p38MAPK␣) and tumor necrosis factor ␣-induced insulin resistance (p38MAPK␤) in differentiated primary cultures of neonatal rat skeletal myocytes. Thus, a possible explanation is that the beneficial effect of L-CPT I overexpression may be mediated by facilitating activation of the ␣-isoform and inhibiting activation of p38MAPK␤. In contrast, CPT I overexpression did not appear to act by inhibition of the IKK/IB/NFB pathway. Schmitz-Peiffer and coworkers (11) have reported inhibition of insulin-stimulated glycogen synthase kinase 3 phosphorylation after preincubation of C2C12 cells with 0.75 mM palmitate for 18 h. That palmitate preincubation did not influence that parameter in our system may be explained by the moderate concentration used in this study (0.2 mM).
In the context of L-CPT I overexpression, palmitate preincubation reduced insulin-stimulated [ 14 C]glucose incorporation into glycogen and [ 3 H]DOG uptake (Figs. 4, 5, and 6). However, under these conditions (L-CPT I overexpression combined with palmitate preincubation), the insulin-stimulated components of both metabolic parameters were significantly higher than in control cells preincubated with palmitate. Hence L-CPT I-mediated increases in fatty acid oxidation capacity partially compensated for the insulin resistance effects of fatty acids, a finding consistent with our original hypothesis. However, the data do not support the concept that this effect is due to prevention of palmitateinduced accumulation of intracellular lipids.
First, direct measurement of the cellular content of triacylg- lycerol, diacylglycerol, ceramide, and long-chain acyl-CoA demonstrated that the increased flux through ␤-oxidation was not sufficient to lower intracellular accumulation of these lipids (Fig.  9). Second, in the absence or presence of L-CPT I overexpression, palmitate caused similar proportional decreases in insulin-stimulated glucose incorporation into glycogen (Figs. 4B and 5B). Therefore, in both control and L-CPT I-overexpressing cells, palmitate decreased insulin sensitivity, and the beneficial effect of increased ␤-oxidation cannot be explained by a reduction in stored lipids. Possibilities of distinct fatty acid/acyl-CoA pools preferential for CPT I/␤-oxidation or lipid synthesis could be considered such that increased ␤-oxidation diminishes the oxidation pool and putative adverse effects on insulin action without affecting lipid biosynthetic pathways. However, in our system, palmitate exerted a similar proportional effect on insulin action with or without L-CPT I overexpression.
Further evidence for a mechanistic separation between the effects of fat oxidation and lipid accumulation can be gleaned from other systems. For example, in humans, exercise increases ␤-oxidation and insulin sensitivity in the absence of changes in cellular lipid content (14,15). In contrast, in animal models, leptin (65)(66)(67)(68), thiazolidinediones (69), and adiponectin (52,70,71) have been shown to increase ␤-oxidation and insulin sensitivity in skeletal muscle in conjunction with a decrease in fatty acid esterification and lipid content. These latter data have been interpreted in the context of the lipotoxic paradigm consistent with a model in which increased ␤-oxidation lowers intramyocellular lipids and hence protects against the harmful effects of fatty acids on insulin signaling. Our data do not contradict this interpretation but rather suggest that, independently of impact upon lipid accumulation, increased ␤-oxidation may have a role in increasing insulin sensitivity. An advantage of our model is that it permitted specific alteration of ␤-oxidation in the absence of confounding variables such as marked increases in fatty acid concentrations.
In summary, overexpression of CPT I increases the rate of ␤-oxidation in L6 muscle cells and enhances insulin stimulation of glucose disposal and phosphorylation of p38MAPK although without apparent effect on insulin stimulation of PKB phosphorylation. In the context of palmitate preincubation, L-CPT I overexpression mitigates the negative effects of the fatty acid on insulin-stimulated glycogen synthesis, glucose uptake, and phosphorylation of PKB. Intriguingly these effects of L-CPT I overexpression are independent of changes in intracellular triacylglycerol, diacylglycerol, ceramide, or long-chain acyl-CoA. Hence we propose that an increase in the rate of ␤-oxidation in muscle cells has a beneficial effect on insulinstimulated glucose disposal that operates independently of the lipotoxic model. Activation of flux through CPT I in muscle (perhaps by means of a malonyl-CoA antagonist) may be a useful therapeutic target in the treatment of insulin resistance.