Overexpression of 1-Acyl-Glycerol-3-Phosphate Acyltransferase-αEnhances Lipid Storage in Cellular Models of Adipose Tissue and Skeletal Muscle

Cell culture. 3T3-L1 cells were purchased from the American Type Culture Collection (Rockville, MD), maintained as fibroblasts, and differentiated into adipocytes as previously described(33). C2C12 myoblasts,obtained from Robert J. Schwartz, PhD, were maintained in Dulbecco's minimum essential medium (DMEM) supplemented with 10% fetal calf serum. After confluence, the culture medium was changed to DMEM with 2% horse serum to initiate myogenic differentiation. Before each assay, both 3T3-L1 adipocytes and C2C12 myotubes were serum starved for 3 h in Krebs-Ringer bicarbonate buffer solution with 25 mmol/l Krebs-Ringer bicarbonate HEPES buffer (KRBH)(pH 7.4) containing 100 μmol/l bovine serum albumin (BSA) (Sigma, St. Louis, MO) essentially fatty acid free, and 5 mmol/l glucose.

Stable cell lines. A 2,083-bp cDNA clone encoding human AGAT was isolated from a human adipocyte cDNA library (Clontech, Palo Alto, CA) using a human infant brain expressed sequence tag similar to yeast AGAT (GenBank no. T77083; Genome Systems, St. Louis, MO) as a hybridization probe. The cDNA clone contains an 852-bp open reading frame encoding a protein of 283 amino acids with a calculated molecular mass of 31.7 kDa. Membrane preparations from AGAT-deficient Escherichia coli strain JC-201(34) transformed with full-length AGAT cDNA exhibited AGAT activity (H.R. and H.J.P., unpublished data). This AGAT was identical to the α isoform described by West et al.(32).

The polymerase chain reaction (PCR) method was used to delete the predicted NH₂-terminal 30-amino acid leader sequence of AGAT-α protein and to introduce an EcoRI site into the upstream sequence and a NotI site into the downstream sequence of AGAT-α. The resulting PCR fragment was subcloned into the pcDNA3.1/HisC vector (Invitrogen,Carlsbad, CA); its sequence was verified by DNA sequencing. The recombinant plasmid, pHis-AGAT, encodes 293 amino acids containing a 40-amino acid NH₂-terminal epitope tag (including six histidines) and has a calculated mass of 32.3 kDa. This expression construct was transfected into 3T3-L1 fibroblasts and C2C12 myoblasts using SuperFect reagent (Qiagen,Valencia, CA). After selection in G-418, pools of 40-60 surviving colonies were expanded for study. Control cell lines were transfected with pcDNA/HisC vector without the insert, followed by G-418 selection. Cell line expression of AGAT-α was confirmed by Western blot analysis and immunofluorescence microscopy using a monoclonal anti-His₆ antibody (Clontech). The transfection and selection were repeated, and the independent G-418-resistant clones were amplified and verified for AGAT expression by reverse-transcriptase-PCR and/or Western blot analysis. Phenotypes from the two selections were not distinguishable, and data are representative of the two selections.

Western blotting. Either 30 or 60 μg of total cellular protein was separated by 12% SDS-PAGE and electroblotted onto a nitrocellulose membrane (Amersham, Arlington Heights, IL). The filter was incubated for 1 h at room temperature with monoclonal anti-His₆ antibody (1:5,000;Clontech), washed, and incubated with horseradish peroxidase—conjugated secondary antibody. Bound antibodies were detected by using the enhanced chemiluminescence Western blotting analysis system (Amersham).

Immunofluorescence microscopy. Cells were grown on cover slips,fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, washed with phosphate-buffered saline (PBS), blocked with 1% normal goat serum in PBS, and incubated for 1 h at room temperature with monoclonal anti-His₆ antibodies (1:500; Clontech). Cells were washed with PBS and incubated with fluorescein isothiocyanate (FITC) or rhodamine-conjugated goat anti-mouse IgG (Pierce, Rockford, IL). Cells were treated with 4,6-diamidino-2-phenylindole as a counterstain, washed extensively with PBS,and examined using a Zeiss Axiophot fluorescence microscope.

[³H]oleic acid uptake and incorporation into cellular lipids. After serum deprivation, cells in six-well dishes were incubated for 1 h in KRBH: 100 μmol/l BSA:5 mmol/l glucose with or without 174 nmol/l insulin. The assay was initiated by addition of 100 μmol/l[³H]oleic acid (OA) (5 μCi per well). At various times(t), aliquots of medium were counted in duplicate, and the uptake was calculated as (dpm_t=0 -dpm_t)/dpm_t=0 × 100%, where dpm is the number of disintegrations per min per aliquot. After 90 min, uptake was terminated by exhaustive washing with ice-cold PBS. Cells were lysed, and an aliquot was used to determine cell-associated radioactivity by liquid scintillation counting. Total cellular lipids were extracted according to a modified Folch method (35),dried under a stream of nitrogen, dissolved in CHCl₃:MeOH (2:1),mixed with lipid standards for TG, diglyceride (DG), LPA, and PA, and loaded onto two thin-layer chromatography (TLC) plates. One plate was developed in CHCl₃:MeOH:H₂O (65:25:4), which separates polar lipids,and the other in hexane:diethyl ether:acetic acid (80:20:2), which separates neutral lipids. Individual lipid spots were visualized by exposing the dried TLC plates to iodine vapor and/or a phosphor screen (Bio-Rad, Hercules, CA),scraped from the plate, and dissolved in scintillation fluid, and radioactivity was determined by counting.

Lipid and glycogen synthesis. After serum starvation, cells in six-well dishes were incubated for 15 min in KRBH/BSA without glucose. Insulin was added for an additional 15 min. The assay was started by adding 5 mmol/l[¹⁴C]glucose (1 μCi per well) to the culture medium, followed by incubation at 37°C for 1 h. Uptake was terminated by washing the cells three times with ice-cold PBS containing 10 mmol/l glucose. Cells were lysed using 1 ml RIPA (150 mmol/l NaCl:1% NP-40:0.5% deoxycholate:0.l% SDS, 50 mmol/l Tris pH 8.0) per well. Cell-associated radioactivity in 50 μl lysate was determined by liquid scintillation counting. To determine lipid-associated radioactivity, lipids were extracted from 450 μl cell lysate(35) and counted. Another aliquot of cell lysate (450 μl) was used to measure[¹⁴C]glycogen formation(36).

Lipolysis assay. AGAT-or mock-transfected 3T3-L1 adipocytes grown in six-well dishes were rinsed with PBS, incubated for 3 h with 600 μmol/l OA and 300 μmol/l BSA, and washed with DMEM/300 μmol/l BSA three times. DMEM:600 μmol/l BSA:25 mmol/l HEPES (500 μl) was added to the cells,which were incubated for 30 min with or without 10 μmol/l isoproterenol. At the end of the incubation, the medium was collected, concentrated, and assayed with kits for glycerol (Sigma) and NEFA (Wako Chemicals, Richmond, VA).

AGAT activity in cell extracts. Cell extracts were prepared as previously described (37), and total protein was measured using a kit (Bio-Rad). An aliquot of cell extract(7-10 μg) was added to a reaction mixture containing 50 μmol/l LPA(oleoly-sn-glycero-3-phosphate) and 50 μmol/l[¹⁴C]oleoyl-CoA (Amersham), followed by incubation for 3-6 min at 30°C. The reaction was terminated by spotting an aliquot onto a silica gel 60 TLC plate, which was developed in chloroform:methanol:water (65:25:4). Production of radiolabeled lipids was quantified as described above.

Statistical methods. Comparisons were performed using a two-tailed Student's t test. P ≤ 0.05 was considered significant. Data are means ± SE.

Stable cell lines overexpressing human AGAT-α. Stable 3T3-L1 and C2C12 cell lines expressing His₆-tagged human AGAT-α(AGAT-) or containing vector pcDNA3.1/HisC alone (mock-) were generated. Cellular localization of the His₆-AGAT fusion protein, which has an apparent molecular mass of 34 kDa (Fig. 1), was examined in 3T3-L1 fibroblasts and adipocytes and in C2C12 myoblasts and myotubes incubated with anti-His₆ antibody. Immunofluorescence microscopy showed distinct cytoplasmic staining in all cell types (Fig. 2, left panels);staining did not occur in mock-transfected cells(Fig. 2, right panels). Overexpression of AGAT-α did not significantly alter cell morphology,cell proliferation rate, or differentiation of fibroblasts and myoblasts into adipocytes and myotubes.

Cell extracts of AGAT- and mock-3T3-L1 adipocytes or AGAT- and mock-C2C12 myotubes were assayed for the conversion of [¹⁴C]oleoyl-CoA to[¹⁴C]PA. The endogenous AGAT activity in 3T3-L1 adipocyte extracts was nearly an order of magnitude higher than that in C2C12 myotube extracts(Table 1). [¹⁴C]PA formation in AGAT-C2C12 myotubes was 150% of that of the mock-transfected myotubes (P < 0.02). Measures of [¹⁴C]PA and[¹⁴C]phosphatidylethanolamine production in AGAT-overexpressing adipocyte extracts were 123% (P < 0.04) and 131% (P <0.02) of those of mock-transfected adipocyte extracts.

Effects of AGAT-α overexpression on [³H]OA and[¹⁴C]glucose metabolism in 3T3-L1 adipocytes. [³H]OA or [¹⁴C]glucose uptake by 3T3-L1 adipocytes and C2C12 myotubes was measured as a function of insulin concentration (data not shown); the maximal[³H]OA uptake (>90%) by 3T3-L1 adipocytes occurred at 10 nmol/l of insulin, and was not significantly different between 10 nmol/l and 1μmol/l of insulin. In C2C12 myotubes, the maximal [¹⁴C]glucose uptake observed was ≥174 nmol/l. When called for, insulin was used at 174 nmol/l.

[³H]OA uptake in the absence and presence of insulin was measured in AGAT-L1 and mock-L1 adipocytes. At 90 min, insulin treatment enhanced [³H]OA uptake in AGAT-L1 cells (65%, P = 0.04)and mock-L1 cells (170%, P = 0.004), and under both conditions,[³H]OA uptake was increased by AGAT-α overexpression (by 27%with insulin treatment and by 130% without insulin treatment)(Fig. 3).

Total cellular lipids were extracted at the end of the 90-min incubation and separated by TLC (Fig. 4A). Radioactivity associated with LPA, PA, DG, and TG was quantified, and the incorporation of [³H]OA into glycerolipids was assessed. There were no significant differences in cellular[³H]LPA or [³H]DG content in AGAT-L1 and mock-L1 adipocytes in the presence or absence of insulin(Fig. 4B). Insulin increased [³H]PA in both AGAT-L1 and mock-L1 adipocytes (40 and 107%); AGAT-α overexpression increased [³H]PA under both basal and insulin-stimulated conditions (96 and 33%). Insulin increased[³H]TG content in mock-L1 adipocytes (181%), but not in AGAT-L1 adipocytes; AGAT-α overexpression increased [³H]TG, but only under basal conditions (196%). Under all conditions studied, most[³H]OA was converted to TG.

Insulin significantly increased total glucose uptake in both AGAT-L1 and mock-L1 adipocytes (514 and 330%), and AGAT-α overexpression did not affect basal glucose uptake but significantly increased insulin-stimulated glucose uptake (56%) (Fig. 5A). These findings accorded with measurements of the conversion of glucose to cellular lipids, which significantly increased with insulin exposure in AGAT-L1 and mock-L1 cells (920 and 350%) and with AGAT-α overexpression under the condition of insulin stimulation (85%)(Fig. 5B).

Lipolysis and re-esterification. In the lipolysis assay,isoproterenol exposure increased glycerol release in both AGAT-L1 and mock-L1 adipocytes (37 and 40%) (Fig. 6A) and OA release in mock-L1 cells (120%)(Fig. 6B). The increase in OA release with isoproterenol in AGAT-L1 cells was substantial(105%) but not significant (Fig. 6B). AGAT-α overexpression affected neither basal nor isoproterenol-mediated glycerol release(Fig. 6A), but suppressed OA release under both conditions (-44 and -45%)(Fig. 6B). In the absence of NEFA re-esterification, the concentration of NEFA in the medium would be expected to be three times that of glycerol. Calculation of the percentage of NEFA relative to glycerol in the medium yielded values indicating that most of the mobilized NEFAs were re-esterified, a result enhanced by AGAT-α over-expression(Table 2).

Effect of AGAT-α overexpression on [¹⁴C]glucose and[³H]OA utilization in C2C12 myotubes. To identify the role of AGAT-α in the utilization of fuel molecules by skeletal muscle, the effects of AGAT-α overexpression on [¹⁴C]glucose and[³H]OA uptake in C2C12 myotubes were assessed in the absence and presence of insulin. Treatment of fully differentiated muscle cells with 174 nmol/l insulin did not significantly affect [³H]OA uptake in either mock- or AGAT-C2C12 myotubes (Fig. 7). Similar results were observed between 1 and 500 nmol/l insulin(data not shown). AGAT overexpression increased [³H]OA uptake by 34% (P = 0.027) in the presence of 174 nmol/l insulin, but not in the absence of insulin (Fig. 7).

Insulin exposure increased [¹⁴C]glucose uptake in both AGAT-C2C12 and mock-C2C12 myotubes (29 and 23%), although AGAT-αoverexpression did not alter either basal or insulin-stimulated glucose uptake(Fig. 8A). In the presence of insulin, AGAT-α overexpression altered the metabolic fate of intracellular glucose; glycogen synthesis decreased (-30%)(Fig. 8B), and glucose conversion to lipids increased (33%) (Fig. 8C).

Insulin stimulates cellular glucose uptake at multiple steps, including transport across the plasma membrane(38) and intracellular metabolism(39,40). Although overexpression of GLUT1 in skeletal muscle of transgenic mice increases cellular glucose uptake and glycogen content, intracellular glucose also accumulates (22). Thus,under conditions of increased glucose transport, cellular glucose metabolism in skeletal muscle may be rate limiting with respect to wholebody glucose disposal. Similarly, cellular metabolism of NEFA to glycerolipids, especially TG, might be the primary determinant of the rate of NEFA uptake in adipose tissue.

Although the enzymes involved in glycerolipid synthesis have been identified, the role of each enzyme in cellular TG accumulation and energy uptake is unknown. We established stable 3T3-L1 and C2C12 cell lines overexpressing human AGAT-α. These cells were used to evaluate the effects of AGAT-α overexpression on cellular energy uptake and TG synthesis.

In both the presence and absence of insulin, overexpression of AGAT-αin 3T3-L1 fat cells increased OA uptake and its incorporation into TG (Figs. 3 and 4B). However, OA uptake was greater in the presence of insulin, suggesting that the coordinate regulation of other insulin-responsive genes can enhance the effects of AGAT-α overexpression. These genes might include those for adipocyte fatty acid binding protein(41), FACS(42), glycerol-3-phosphate acyltransferase (GPAT)(43,44),and hormone-sensitive lipase (HSL). On the other hand, in C2C12 myotubes insulin did not affect OA uptake, and AGAT-α overexpression produced only modest increases in OA uptake, which was observed only in the presence of insulin. Because AGAT activity is seven- to ninefold higher in 3T3-L1 adipocytes than in C2C12 myotubes (Table 1), it is possible that adipocytes have a higher expression of other adipogenic genes, such as FATP, FACS, GPAT, PA phosphohydrolase, and DG acyltransferase, which are required for the conversion of NEFA to TG. It is also possible that C2C12 cells lack the mechanism(s) for activating the adipogenic genes that would support increased TG synthesis and NEFA uptake.

In adipocytes, intracellular lipolysis is catalyzed by HSL, which liberates both glycerol and NEFA; this activity is stimulated by isoproterenol. The cellular release of glycerol, which cannot be directly esterified by adipocytes(45,46),provides an estimate of HSL activity. In contrast, the NEFA liberated by HSL can be either released into the medium or re-esterified, so that differences in the amount of NEFA released and the amount expected on the basis of glycerol release reflect the effects of NEFA re-esterification. Our data(Table 2 and Fig. 6) show that the amount of NEFA released into the medium was low relative to glycerol under all conditions studied, indicating that most NEFA liberated by HSL are re-esterified. Under both basal and isoproterenol-stimulated conditions,AGAT-α overexpression decreased NEFA release. Because AGAT overexpression increases cellular TG (Fig. 4B), we conclude that most of the mobilized NEFA return to the intracellular TG pool, and that AGAT-α overexpression increases esterification of NEFA derived from both intracellular lipolysis(Table 1 and Fig. 6) and the extracellular medium (Fig. 3).

Cellular glucose metabolism was also altered by AGAT-αoverexpression, and the alterations were different in 3T3-L1 and C2C12 cells. In 3T3-L1 fat cells, AGAT-α overexpression increased glucose uptake and its conversion to lipids, but only with insulin stimulation(Fig. 5). The absence of an effect without insulin is likely because of a strict requirement for insulin-dependent activities in the pathway that converts extracellular glucose to intracellular 1-acyl-glycerol-3-phosphate or fatty acyl CoA, the substrates for AGAT. The insulin-dependent activities may include GLUT4,hexokinase, and the enzymes involved in lipogenesis, such as acetyl-CoA carboxylase, pyruvate dehydrogenase, and fatty acid synthase(47). Similarly, in C2C12 cells, AGAT-α overexpression had no effect on basal glucose uptake. In the presence of insulin, however, AGAT-α overexpression diverted glucose from glycogen synthesis to lipid synthesis(Fig. 8B and C), presumably because of increased entry of glucose-6-phosphate into the glycolysis pathway generating acetyl-CoA and glycerol-3-phosphate.

Because AGAT catalyzes the conversion of LPA to PA(31), AGAT overexpression in 3T3-L1 adipocytes would be expected to increase cellular PA content. However,in our experiments, that difference was small; the greatest difference was in cellular TG content. In contrast to the intact cells, most of the product formed in cell extracts was PA (Table 1), and AGAT overexpression only modestly increased PA and did not increase TG formation. This finding is consistent with a model in which AGAT is proximal to the other enzymes of glycerolipid synthesis. Disruption of cells could separate these enzymes and uncouple their activities. Consequently, PA accumulates in the extracts, but not in intact AGAT-L1 adipocytes.

In 3T3-L1 adipocytes, AGAT-α overexpression increases NEFA uptake and decreases NEFA efflux. Increased entry of NEFA into the glycerolipid synthesis pathway would lower the cytoplasmic NEFA concentration and in turn create a NEFA concentration gradient across the plasma membrane. This gradient could drive NEFA influx, which may be diffusive(48,49)or involve NEFA transport proteins(30). In a physiological context, increased AGAT activity in adipose tissue might decrease plasma NEFA concentrations. Future studies of this pathway should include tests of AGAT overexpression in vivo to determine whether it is an attractive target for pharmacological management of plasma NEFA and lipoprotein concentrations.

This work was supported by grants from the National Institutes of Health(HL-30914 and HL-56865) and the Ethel L. Walker Memorial Research and Education Fund for Arteriosclerosis. H.R. was a student in The Michael E. DeBakey Graduate Program in Cardiovascular Sciences.