Mechanisms of hepatic very low density lipoprotein overproduction in insulin resistance. Evidence for enhanced lipoprotein assembly, reduced intracellular ApoB degradation, and increased microsomal triglyceride transfer protein in a fructose-fed hamster model.

A novel animal model of insulin resistance, the fructose-fed Syrian golden hamster, was employed to investigate the mechanisms mediating the overproduction of very low density lipoprotein (VLDL) in the insulin resistant state. Fructose feeding for a 2-week period induced significant hypertriglyceridemia and hyperinsulinemia, and the development of whole body insulin resistance was documented using the euglycemic-hyperinsulinemic clamp technique. In vivo Triton WR-1339 studies showed evidence of VLDL-apoB overproduction in the fructose-fed hamster. Fructose feeding induced a significant increase in cellular synthesis and secretion of total triglyceride (TG) as well as VLDL-TG by primary hamster hepatocytes. Increased TG secretion was accompanied by a 4.6-fold increase in VLDL-apoB secretion. Enhanced stability of nascent apoB in fructose-fed hepatocytes was evident in intact cells as well as in a permeabilized cell system. Analysis of newly formed lipoprotein particles in hepatic microsomes revealed significant differences in the pattern and density of lipoproteins, with hepatocytes derived from fructose-fed hamsters having higher levels of luminal lipoproteins at a density of VLDL versus controls. Immunoblot analysis of the intracellular mass of microsomal triglyceride transfer protein, a key enzyme involved in VLDL assembly, showed a striking 2.1-fold elevation in hepatocytes derived from fructose-fed versus control hamsters. Direct incubation of hamster hepatocytes with various concentrations of fructose failed to show any direct stimulation of its intracellular stability or extracellular secretion, further supporting the notion that the apoB overproduction in the fructose-fed hamster may be related to the fructose-induced insulin resistance in this animal model. In summary, hepatic VLDL-apoB overproduction in fructose-fed hamsters appears to result from increased intracellular stability of nascent apoB and an enhanced expression of MTP, which act to facilitate the assembly and secretion of apoB-containing lipoprotein particles.

Insulin resistance is an extremely common pathophysiological trait that is implicated in the development of a number of important human diseases including Type 2 diabetes, atherosclerosis, hypertension, and dyslipidemia (1,2). Many studies have suggested that insulin resistance may be a factor in causing dyslipidemia (3)(4)(5)(6)(7). The insulin resistant state is commonly associated with lipoprotein abnormalities that are risk factors for coronary heart disease, including hypertriglyceridemia, high levels of VLDL, 1 low levels of high density lipoprotein cholesterol (8), and small, dense LDL (9). These metabolic abnormalities together with hypertension and Type 2 diabetes may cluster in the same individual, constituting a syndrome referred to as the metabolic Syndrome X (2). It has been suggested that the most fundamental defect in these patients is resistance to insulin-stimulated glucose uptake, which leads to compensatory hyperinsulinemia, enhanced VLDL secretion by the liver, and hypertriglyceridemia (10). Hypertriglyceridemia is the most common lipid abnormality in subjects with insulin resistance. Early kinetic studies suggested that the hypertriglyceridemia associated with insulin resistance is due to an increase in VLDL-TG production (11)(12)(13)(14), but the cellular mechanisms of this process have not been clearly determined.
Insulin has been shown to acutely inhibit the hepatic production of VLDL-TG in both in vitro and in vivo studies (reviewed in Refs. 15 and 16). Short-term hyperinsulinemia also inhibits hepatic secretion of apolipoprotein B (apoB) by perfused rat liver (17), primary rat hepatocytes (18 -20), human hepatocytes (21), as well as in human subjects in vivo (22,23) (reviewed in Refs. 15 and 16). Interestingly however, obese, chronically hyperinsulinemic and insulin-resistant human subjects were resistant to the acute inhibitory effects of insulin on VLDL apoB (22). Furthermore, stable isotope studies in humans have shown that VLDL1 production is acutely inhibited by insulin in normal subjects but not in insulin-resistant patients with Type 2 diabetes (24 -26). Primary rat hepatocytes, incubated in vitro with high concentrations of insulin for 3 days, no longer respond to insulin suppression of VLDL apoB secretion, and secrete higher basal levels of VLDL-apoB (27). A similar resistance to the acute suppressive effects of insulin has been observed in HepG2 cells (28). Secretion of VLDL by hepatocytes from hypertriglyceridemic and hyperinsulinemic Zucker fatty rats (fa/fa), is also resistant to the inhibitory effect of insulin (28,29).
Insulin regulates hepatic synthesis and secretion of apoB, either directly or indirectly, by its effects on lipid availability (15). Acute insulin exposure reduces the synthesis of apoB in cultured hepatocytes (20) and increases the rate of apoB degradation (20). Studies in our laboratory, using cell-free translation systems, have shown that insulin attenuates the rate of apoB mRNA translation (30,31). It has also been suggested that apoB availability may become a limiting factor in VLDL assembly and secretion in insulin-treated hepatocytes (32). Recent studies by Sparks and co-workers (33,34) have suggested that insulin inhibits apoB secretion through activation of phosphoinositide 3-kinase. Phosphoinositide 3-kinase activity which phosphorylates phosphoinositol in the 3Ј-position of the inositol ring (35) appears to be necessary for insulin-dependent inhibition of apoB secretion by rat hepatocytes (33,34). Insulin acts by causing the activation and localization of phosphoinositide 3-kinase to the site of apoB synthesis (34).
In the present study, we have employed a new animal model of chronic hyperinsulinemia and insulin resistance, namely, the fructose-fed Syrian golden hamster. The Syrian golden hamster has been used with increasing frequency in recent years to study hepatic lipid metabolism (36 -38). Hamsters develop hypertriglyceridemia, hypercholesterolemia, and atherosclerosis in response to a modest increase in dietary cholesterol and saturated fat (39,40). The hamster has attracted increasing attention as a model for lipoprotein research since its lipoprotein metabolism appears to closely resemble that in humans. The main plasma cholesterol carrier in the hamster is LDL (40,41). Furthermore, hamster liver produces VLDL containing only apoB-100 with a density close to that of human VLDL (42,43), unlike the rat, which has been used extensively for studies of VLDL metabolism and the effects of insulin resistance, and whose liver secretes both apoB-48 and apoB-100. Carbohydrate induced insulin resistance in rodents has been previously well documented. Reaven and colleagues (44 -47) were among the first groups to use sucrose or fructose feeding to induce insulin resistance in rats. Hamsters can also be made obese, hyperinsulinemic, hypertriglyceridemic, and insulin-resistant by fructose feeding (48). Fructose feeding appears to interfere with glucose utilization in vivo, inducing an insulin resistant state (48). It thus appears feasible to induce insulin resistance in the hamster and use the insulin-resistant hamster model to study the mechanisms controlling hepatic VLDL-apoB secretion.

MATERIALS AND METHODS
Male Syrian golden hamsters (Mesocricetus auratus) were purchased from Charles River (Montreal, PQ). Fetal bovine serum (certified grade), liver perfusion medium, hepatocyte wash medium, liver digest medium, and hepatocyte attachment medium were obtained from Life Technologies (Grand Island, NY). Rabbit anti-hamster apoB antiserum was prepared commercially by Lampire Biological Laboratories (Pipersville, PA) using hamster LDL prepared in our laboratory. Specificity of this commercial preparation of anti-apoB polyclonal antibody and lack of any cross-reactivity to other hamster apolipoproteins (apoA-I or apoE) was confirmed by immunoblotting analysis of purified plasma lipoprotein fractions. Anti-bovine MTP antibody was generously provided by Dr. David Gordon (Bristol-Meyers Squibb). Anti-transferrin apoB antibody was obtained from Sigma. Anti-3-hydroxy-3-methylglutaryl-CoA reductase antibody (polyclonal anti-peptide antibody) was generously provided by Dr. S. P. Tam, Queen's University.
Animal Protocols-Male Syrian golden hamsters were obtained from Charles River Canada (Montreal, PQ). All animals were housed in pairs and were given free access to food and water. After blood collection, animals were placed on either the control diet (normal chow) or fructose-enriched diet (hamster diet with 60% fructose, pelleted, Dyets Inc., Bethlehem, PA). The diet was continued for 2 weeks and hamster weight was monitored every 2 days. Plasma glucose, TG, and cholesterol levels were determined on an automated clinical chemistry ana-lyzer (Hitachi 705). Plasma insulin levels were determined by radioimmunoassay using a rat insulin kit from Linco Research (St. Louis, MO). This assay has 100% cross-reactivity to hamster insulin and the intraand interassay coefficient of variation were 6.8 and 10.6%, respectively.
Euglycemic Hyperinsulinemic Clamp Studies-At the end of the 2-week feeding period anesthesia was induced using isoflurane (4% in 100% oxygen followed by 2% isoflurane with O 2 enriched by mask throughout the surgical procedure), and catheters (PE 10 tubing) were inserted into the femoral vein (for infusion) and into the femoral artery (for blood sampling). The animals were allowed to awaken from anesthesia and were unrestrained in their cage. They were fasted from 6:00 p.m. that evening. Catheters were kept patent overnight with 1% citrate solution. At 9:00 a.m. two baseline blood samples (0.25 ml) were drawn at 10-min intervals followed by a primed-constant intravenous infusion of human biosynthetic insulin (Humulin R, Eli Lilly, Toronto, ON, Canada) (180 milliunits/kg bolus followed by 18 milliunits⅐kg Ϫ1 min Ϫ1 in 0.9% NaCl and 0.1% bovine serum albumin solution) for 2 h. The blood glucose level was maintained at the baseline value throughout the study by adjusting a 10% dextrose infusion according to frequent plasma glucose monitoring (approximately 0.1 ml every 10 min). Blood samples (0.25 ml) were drawn at 90, 100, 110, and 120 min to assess the steady state glucose and insulin levels. There was no significant decline in hematocrit throughout the study.
In Vivo VLDL Secretion Studies-In order to determine whether 2-week fructose feeding is associated with an in vivo increase in VLDL-apoB secretion, catheters were inserted in the femoral vein and artery of fructose (n ϭ 6) and control fed animals (n ϭ 7) of the same weight (119 Ϯ 5 versus 121 Ϯ 6 g, respectively, p ϭ 0.79) as described above and VLDL-apoB and VLDL-TG levels were measured in the fasting state (12 h) at 1, 30, 60, and 90 min after an intravenous bolus (600 mg/kg) of a 20% (w/v) solution of Triton WR-1339 (Sigma) in normal saline (NaCl 0.9%). Because Triton WR-1339 effectively blocks the activity of lipoprotein lipase in vivo and therefore blocks the VLDL particle clearance, the secretion rate of VLDL-apoB and VLDL-TG is proportional to the rate of increase in VLDL-apoB and VLDL-TG concentration over time (49 -53). This method has been previously used in the hamster by others (42,54,55). The total blood volume of the samples drawn was less than 1.5 ml/animal during the experiment.
Calculation of the in vivo VLDL-apoB and VLDL-TG secretion rates was performed by multiplying the slope of the concentration increase of VLDL-apoB (in g/ml/min) and VLDL-TG (in mol/ml/min), respectively, over time by the VLDL distribution volume estimated as 3.8 ml/100 g body weight (56). Linearity of the increase in VLDL-apoB and VLDL-TG concentration was assessed by the linear regression R squared value. A two-tailed unpaired t test was used to compare the slope of the VLDL-apoB and VLDL-TG concentration increase and the VLDL-apoB and VLDL-TG secretion rates between fructose-fed and control-fed hamsters.
VLDL (d Ͻ 1.006 g/ml) in the in vivo studies was isolated by ultracentrifugation of plasma samples at 110,000 rpm for 3 h at 16°C in a TI 110 rotor with a Beckman Optima TLX ultracentrifuge. VLDL-TGs were measured using a Roche Molecular Biochemicals colorimetric kit. VLDL-apoB was quantified by electroimmunoassay as described (57), with an anti-hamster apoB rabbit polyclonal antibody and a standard curve performed on each plate. ApoB standards were prepared by isolation of hamster LDL and protein quantification using Lowry's method (58). Triton WR-1339 added in vitro at concentration in the range of those used in vivo (ϳ15 mg/ml) to hamster samples did not interfere with this assay (data not shown). The CV of this assay is 10%.
Liver Perfusion and Isolation of Primary Hamster Hepatocytes-At the end of the 2-week feeding period, hamsters were fasted overnight and blood samples were collected for measurement of a number of analytes in plasma. Hamsters were then fed for another day and were sedated and anesthetized by intramuscular injection of acepromazine (1 mg/kg) and intrapretoneal injection of a mixture of ketamine (200 mg/kg) and xylazine (10 mg/kg). After achieving complete general anesthesia, lidocaine (10 units in 4 divided doses) was injected subcutaneously in the mid-abdominal line of the animal before surgical incision. The liver was perfused as described (59) with small modifications. Released hepatocytes from digested liver tissue were washed three times in hepatocyte wash medium and eventually transferred into culture medium (hepatocyte attachment medium containing 5% fetal bovine serum, 1.0 g/ml insulin, 1ϫ penicillin-streptomycin) and seeded in collagen-coated plates (1.5 ϫ 10 6 cells/35-mm plate). After 4 h or overnight incubation at 37°C, 5% CO 2 , attached cells were used to carry out the experiments.
Determination of the Synthesis and Secretion of Cellular and Secreted Lipids-Primary hepatocytes were pulsed for 3 or 18 h with 5 Ci/ml [ 3 H]acetate to assess the rates of synthesis and secretion of cholesterol, cholesteryl ester, and phospholipids. TG synthesis and secretion were monitored by labeling cells for 3-5 h with 5 Ci/ml [ 3 H]oleate bound to bovine serum albumin. Following labeling, cells were extracted with hexane/isopropyl alcohol (3:2) and the total lipid extract was dried, suspended in hexane, and applied to a thin layer chromatogram. The TLC plates were developed using a two-solvent system to separate polar lipids with chloroform/methanol/acetic acid/formic acid/H 2 O (70:30:12: 4:2) and neutral lipids with petroleum ether/ethyl ether/acetic acid (90:10:1). The lipids were stained with iodine vapor and identified based on the use of a set of known lipid standards (Sigma). The spots identified on the TLC plates were cut and counted using a scintillation counter.
Metabolic Labeling of Intact Primary Hamster Hepatocytes-Primary hamster hepatocytes were preincubated in methionine-free minimal essential medium at 37°C for 1 h and pulsed with 75-100 Ci/ml [ 35 S]methionine for 45-60 min. Following the pulse, the cells were washed twice and chased in hepatocyte attachment medium supplemented with 10 mM methionine. At various chase times duplicate or triplicate dishes were harvested, and cells were lysed in solubilization buffer (phosphate-buffered saline containing 1% Nonidet P-40, 1% deoxycholate, 5 mM EDTA, 1 mM EGTA, 2 mM phenylmethylsulfonyl fluoride, 0.1 mM leupeptin, 2 g/ml N-acetyl-leucyl-leucyl-norleucinal). The lysates were centrifuged for 10 min at 4°C in a microcentrifuge, and supernatants were collected for immunoprecipitation.
Permeabilization of Primary Hamster Hepatocytes-Primary hamster hepatocytes cultured in 35-mm dishes were depleted of methionine by incubation in methionine-free minimal essential medium for 1 h at 37°C under 5% CO 2 . Cells were pulsed with 100 Ci/ml [ 35 S]methionine for 45-60 min and then permeabilized as described (60,61). At various intervals, duplicate or triplicate dishes were washed, solubilized, and subjected to immunoprecipitation.
Analysis of Luminal and Membrane-associated ApoB Pools-Isolation of the microsomal fraction and the separation of the luminal and membrane components by carbonate extraction and ultracentrifugation was performed as described (62,63). Membrane and luminal fractions were then diluted with 800 l of a solubilization buffer containing 360 l of 5ϫC buffer (250 mM Tris-HCl, pH 7.4, 750 mM NaCl, 25 mM EDTA, 5 mM phenylmethylsulfonyl fluoride, 5% Triton X-100) and 410 l of phosphate-buffered saline supplemented with 450 KIU/ml Trasylol, 5 mM phenylmethylsulfonyl fluoride, and subjected to immunoprecipitation, SDS-PAGE, and fluorography.
Chemiluminescent Immunoblotting of MTP 97-kDa Subunit-Cell samples were subjected to chemiluminescent immunoblotting for the MTP 97-kDa subunit. Samples were analyzed by SDS-PAGE using a 10% polyacrylamide mini-gel (8 ϫ 5 cm). Following SDS-PAGE the proteins were transferred electrophoretically overnight at 4°C onto nitrocellulose membranes using a Bio-Rad Wet Transfer System. The membranes were blocked with 5% solution of fat-free dry milk powder, incubated with a rabbit anti-hamster MTP antiserum, washed, and then incubated with a secondary antibody conjugated to peroxidase. Membranes were then incubated in an ECL detection reagent for 60 s and exposed to Hyperfilm. Films were then developed and quantitative analysis was performed using an Imaging Densitometer.
Immunoprecipitation, SDS-PAGE, and Fluorography-Immunoprecipitation was performed as described previously (60). Immunoprecipitates were washed with wash buffer (10 mM Tris-HCl, pH 7.4, 2 mM EDTA, 0.1% SDS, 1% Triton X-100) and were prepared for SDS-PAGE by suspending and boiling in 100 l of electrophoresis sample buffer. SDS-PAGE was performed essentially as described (64). The gels were fixed, stained and soaked in Amplify (Amersham Pharmacia Biotech), before being dried, and exposed to Dupont autoradiographic film at Ϫ80°C for 1-4 days. ApoB bands were excised from the gel, digested in hydrogen peroxide/perchloric acid, and the radioactivity was determined by scintillation counting.
Calculations and Statistical Analysis for Glucose Clamp Studies-All the values are reported as mean Ϯ S.E. The baseline and clamp periods were the mean of times Ϫ10 and 0 min and the mean of times 90, 100, 110, and 120 min, respectively. The insulin sensitivity index (S I ) during the euglycemic hyperinsulinemic clamps was calculated using the formula (65): S I ‫؍‬ Cl glu /INS clamp , where Cl glu is the glucose clearance defined as the mean glucose infusion rate (Ginf)/mean plasma glucose during the last 30 min of the clamp, and where INS clamp is the mean plasma insulin level during the last 30 min of the clamp. S I is expressed in arbitrary units (liter 2 ⅐kg Ϫ1 min Ϫ1 ). Two-way ANOVA was used to compare the glucose, insulin, and Ginf curves of the fructose-fed and control groups during the last 30 min of the clamp. A two-tailed paired t test was used to compare the mean baseline versus mean clamp S I values. Fig. 1 shows the physiological changes observed in control and fructose-fed hamsters after a 2-week feeding period. Fructose-fed hamsters gained body weight at approximately the same rate as that for control hamsters over the 2-week feeding period (data not shown). Fructose-fed hamsters showed a significant elevation of plasma TG (p ϭ 0.0309) and an elevation of plasma cholesterol level that approached statistical significance (p ϭ 0.0550), following a 2-week period on a fructose-rich diet (Fig. 1, A and B). There was also a significant elevation (p ϭ 0.0110) of plasma insulin level (Fig. 1C). In addition, fructose feeding caused a significant elevation of plasma free fatty acids (p ϭ 0.0045) as shown in Fig. 1D. However, plasma glucose levels did not differ significantly (p ϭ 0.9452) between control and fructose-fed hamsters (Fig. 1E). Overall, fructose feeding induced significant elevation in plasma levels of TG, insulin, and free fatty acids. Fig. 2 shows the results of the euglycemic hyperinsulinemic clamp studies, which were performed in 9 fructose-fed hamsters and 10 control hamsters. The plasma glucose levels ( Fig.  2A) did not change from baseline and were significantly higher in the fructose-fed versus control animals during the last 30 min of the clamp (5.0 Ϯ 0.4 versus 3.9 Ϯ 0.3 mmol/liter, p Ͻ 0.01). Although the insulin levels ( Fig. 2B) tended to be higher in the fructose-fed versus control group during the last 30 min of the clamp (2394 Ϯ 441 versus 2002 Ϯ 272 pmol/liter), this difference was not significant. However, the Ginf (Fig. 2C) during the last 30 min of the clamp was significantly lower in fructose-fed versus control animals (26.0 Ϯ 6.5 mol kg Ϫ1 min Ϫ1 versus 39.5 Ϯ 9.4 mol kg Ϫ1 min Ϫ1 , p Ͻ 0.01). This difference in Ginf, especially in the face of higher levels of both glucose and insulin during the clamp in fructose-fed versus control hamsters, confirms that the former are more insulin resistant than the latter. This is shown by the calculation of S I (Fig. 2D) which was significantly lower in fructose-fed versus control hamsters (2.7 Ϯ 0.8 ϫ 10 6 versus 4.8 Ϯ 0.4 ϫ 10 6 liter 2 kg Ϫ1 min Ϫ1, p ϭ 0.03).

Evidence for Development of Insulin Resistance in Fructosefed Hamsters: Euglycemic Hyperinsulinemic Clamp Studies-
In Vivo Evidence of VLDL-ApoB Overproduction in Fructosefed Hamsters- Fig. 3A shows the increase of VLDL-TG over 90 min following the intravenous administration of Triton WR-1339. The increase in both fructose-fed and control hamsters was linear (mean R squared 0.98 Ϯ 0.01 and 0.91 Ϯ 0.05 for the fructose-fed and control group, respectively). VLDL-TG increase over time tended to be higher in fructose-fed hamsters but this difference was not statistically significant (0.051 Ϯ 0.010 versus 0.034 Ϯ 0.006 mol/ml/min in the fructose fed versus control group, respectively, p ϭ 0.13). Similarly, the VLDL-TG secretion rate was 30% higher in the fructose-fed than in the control group (0.23 Ϯ 0.03 versus 0.16 Ϯ 0.03 mol/min, respectively) although this difference did not reach statistical significance (p ϭ 0.14) (inset of Fig. 3A).
Evidence that Direct Incubation with Fructose Does Not Directly Affect Hepatic ApoB Secretion by Primary Hamster Hepatocytes-It was important to determine if fructose can directly induce the hepatic synthesis and secretion of apoB-100 in hamster hepatocytes since such a direct effect would complicate the interpretation of our data relating apoB overproduction to the development of fructose-induced insulin resistance. Freshly isolated hamster hepatocytes from control, chow-fed hamsters were incubated with different concentrations of fructose for a 24-h period and synthesis and secretion of apoB were monitored by pulse labeling with [ 35 S]methionine. Fig. 4A shows a dose-response study of the effect of fructose on hepatic apoB. Cellular accumulation and extracellular secretion of apoB were unaffected in the presence of increasing concentrations of fructose in the culture media. Even at the highest concentration of 3 mM, there was no significant effect on the synthesis or secretion of apoB in primary hamster hepatocytes. To further confirm a lack of direct effect of fructose on hamster apoB biogenesis, we incubated cultured hepatocytes for a period of up to 3 days with exogenous fructose at the highest concentration (3 mM). Cells incubated for 1, 2, or 3 days were then subjected to pulse-chase labeling (45 min pulse, 1-2 h chase) to determine the extent of hamster apoB secretion and its intracellular stability in hamster hepatocytes. Fig. 4, B-G, show the effects of fructose incubation for periods of 1-3 days. Panels B, D, and F

FIG. 2. Euglycemic hyperinsulinemic clamp studies in fructose-fed hamsters (white bars) versus normal (or control) chow-fed animals (black bars).
Mean glucose levels (A) were slightly but significantly higher in fructose-fed versus control animals during the last 30 min of the clamp period (p Ͻ 0.01). Mean insulin levels (B) were slightly but not significantly higher in the fructose-fed versus control hamsters during the clamp period. The glucose infusion rate (Ginf) (C) during the clamp period was significantly lower in fructose-fed versus control animals (p Ͻ 0.01). The calculated insulin sensitivity index (S I , see "Materials and Methods") (D) was also significantly lower in the fructose-fed versus control hamsters (p ϭ 0.03). Fructose-fed (n ϭ 9), control hamsters (n ϭ 10).
show hamster apoB secretion, while panels C, E, and G show the stability of apoB as assessed by the total apoB remaining in cells and media over a 2-h chase. There was no detectable stimulation of apoB secretion or stability with fructose treatment for up to 3 days. There was actually some inhibition of apoB secretion observed at day 2, but overall the entire experiment revealed no specific effect.
Effect of Fructose Feeding on Hepatic Synthesis and Secretion of Lipids-Primary hamster hepatocytes isolated from normal chow-fed and fructose-fed hamsters were used to determine the synthesis and secretion of cholesterol, cholesteryl ester, and TG. Fig. 5 shows the effect of fructose feeding on the hepatic synthesis and secretion of total lipids. There was a small decrease in cellular levels of cholesteryl ester, although this change was not statistically significant (Fig. 5A). However, the intracellular levels of TG and free cholesterol were both significantly increased in hepatocytes from fructose-fed hamsters (Fig. 5A). Analysis of radiolabeled lipids in culture media of primary hamster hepatocytes also revealed no significant change in cholesteryl ester secretion (Fig. 5B). Interestingly, however, the secretion of TG was significantly elevated in fructose-fed hamsters (Fig. 5B). Conversely, hepatocytes from fructose-fed hamsters secreted significantly lower levels of free cholesterol (Fig. 5B). The decline in free cholesterol secretion was accompanied by an increase in its intracellular levels, suggesting that fructose feeding of hamsters has an inhibitory effect on the release of free cholesterol from hepatocytes. In the case of TG, both the cellular and secreted levels were elevated, suggesting that fructose feeding enhanced the synthesis of TG and its secretion from the cell.
We also analyzed the secreted levels of core lipids associated with VLDL particles secreted by primary hepatocytes. Following radiolabeling of hamster hepatocytes, the cultured media was subjected to ultracentrifugation to isolate the VLDL fraction. The radiolabeled lipids associated with media VLDL were then analyzed by thin layer chromatography. Secretion of VLDL-TG was also significantly induced in fructose-fed hamsters whereas VLDL-cholesteryl ester secretion was unaffected by fructose feeding (data not shown). The observed increase in VLDL-TG secretion compared well with the increase in the intracellular and secreted levels of total TG reported in Fig. 5, A and B.
Overproduction of VLDL-ApoB in Hepatocytes from Fructosefed Hamsters-Primary hepatocytes isolated from hamster liver secrete apoB at a density of VLDL ( Fig. 6 and Ref. 42). To determine the effect of fructose feeding on VLDL-apoB secretion, we performed in vitro steady state labeling experiments in which hepatocytes from control and fructose-fed hamsters were radiolabeled for a 2-h period. Culture media containing secreted lipoprotein particles was then collected and subjected to ultracentrifugation to isolate VLDL. Radiolabeled apoB associated with VLDL particles was immunoprecipated and analyzed by SDS-PAGE and fluorography. Fig. 6 shows the immunoprecipitable VLDL-apoB secreted by control and fructose-fed hepatocytes. There was a highly significant (4.6-fold) elevation in the amount of VLDL-apoB secreted into the media in fructosefed hepatocytes. Increased VLDL-apoB levels suggest the secretion of a considerably higher number of VLDL particles by fructose-fed hepatocytes compared with control hepatocytes.
Turnover Rate of ApoB in Control and Fructose-fed Hepatocytes-We employed pulse-chase labeling experiments to assess the stability and secretion of apoB in hepatocytes isolated from control and fructose-fed hamsters. Isolated hepatocytes were pulsed for 45 min and then chased for 1 and 2 h. Cellular and media apoB was immunoprecipitated and analyzed by SDS-PAGE and fluorography. Fig. 7 shows the intracellular turnover and extracellular secretion of apoB in control and fructose-fed hepatocytes. A large percentage of newly synthesized, radiolabeled apoB disappeared from control cells over the 2-h chase with a small percentage appearing in the media (Fig.  7, A and B). The disappearance rate of apoB in fructose-fed hepatocytes was considerably slower, with only about 25% of apoB having been lost during the 2-h chase (Fig. 7A). The increased stability of apoB in fructose-fed hepatocytes was accompanied by a dramatic increase in the secreted level of newly-synthesized apoB. As shown in Fig. 7B, fructose-fed hepatocytes secreted about 20% of newly synthesized apoB compared with only about 5% in control cells.
Stability of ApoB in Permeabilized Primary Hamster Hepatocytes-Permeabilized cells have been used previously to investigate post-translational degradation of apoB, allowing for detection of specific degradation intermediates, including a 70-kDa fragment (61). We have recently applied the permeabilization protocol to primary hamster hepatocytes and have investigated hamster apoB stability and turnover in this cell model system. 2 Utilizing the permeabilized cell model system, we attempted to determine the effect of fructose feeding on the turnover of apoB. Control and fructose-fed hepatocytes were pulse-labeled, permeabilized, and then chased for a 2-3-h period. Fig. 8 shows the turnover of full-length hamster apoB-100 in permeabilized control and fructose-fed hepatocytes. Hamster apoB-100 was significantly more stable in fructose-fed hepatocytes as judged from the considerably higher intracellular level of apoB remaining in permeabilized cells after a 3-h chase. There was approximately a 2-fold higher level of apoB-100 remaining in fructose-fed hepatocytes following completion of the chase period (Fig. 8).
Effect of Fructose Feeding on Intracellular Assembly of ApoBcontaining Lipoproteins-To directly investigate the formation of apoB-containing lipoprotein particles in hamster hepatocytes, cells were pulse-labeled, chased for 0 and 1 h, and then subjected to subcellular fractionation. Nascent lipoproteins accumulated in the microsomal lumen were fractionated by sucrose gradient centrifugation and immunoprecipitated with anti-hamster apoB antibody. Fig. 9 illustrates the pattern of nascent lipoproteins accumulated in the lumen of control hepatocytes compared with that of lipoproteins detected in fructosefed hepatocytes. Luminal apoB-containing lipoproteins in both control and fructose-fed hepatocytes were predominantly recov- ered from the top of the gradient (fraction 12) and fractions 6 -8 which corresponded to densities of VLDL and LDL, respectively, as previously documented (62,63,66). There was, however, a considerable discrepancy as to the ratio of VLDL to LDL-like lipoproteins in control versus fructose-fed hepatocytes. Control cells had a significantly higher level of LDL-like lipoproteins with only a small pool fractionating with a density of VLDL (Fig. 9A). In contrast, most of the apoB-containing lipoproteins formed in the lumen of microsomes from fructosefed hepatocytes at 1-h chase had a VLDL-like density with only a minor fraction of the total pool of nascent lipoproteins exhibiting a density typical of LDL (Fig. 9B). Also intriguing was the detection of high density lipoprotein-size lipoproteins in the lumen of control hepatocytes but not that of fructose-fed hepatocytes. This observation suggests that a small pool of nascent hamster lipoproteins may form a dense, secretion-incompetent pool in normal hamster hepatocytes as previously reported in HepG2 cells (62,66). The absence of high density lipoproteinlike apoB-containing lipoproteins in microsomes of fructose-fed hepatocytes may in turn suggest a higher efficiency of lipoprotein assembly under this metabolic condition. Finally, when the radiolabeled apoB in all fractions of the gradient were combined, fructose-fed hepatocytes showed approximately a 2-fold higher level of total lumenal apoB after a 1-h chase. This clearly suggested an increased availability of labeled apoB in the microsomal lumen for assembly into VLDL particles in fructose-fed hepatocytes.
Evidence for Enhanced Expression of MTP in Fructose-fed Hepatocytes-Facilitated assembly of apoB-containing lipopro-tein particles in fructose-fed hepatocytes could be related to an increased mass and/or activity of MTP, the key factor involved in the lipoprotein assembly process. To test this hypothesis, a specific anti-hamster MTP antibody was used to estimate the protein mass of MTP in control and fructose-fed hepatocytes. Equal quantities of total cell lysate (1 g of cell protein) were analyzed by SDS-PAGE and then subjected to immunoblotting with the anti-hamster MTP antibody. Fig. 10 shows the immunoblotting analysis of lysates from control and fructose-fed hepatocytes. There was approximately 2-fold higher cellular protein mass of MTP in fructose-fed hepatocytes compared with control hepatocytes after correction for total protein concentration of the cell lysates analyzed. Fig. 10 illustrates the analysis of duplicate aliquots of hepatocyte cell lysates from two different control hamsters and two fructose-fed hamsters. This representative experiment was repeated once with similar results.

DISCUSSION
Although overproduction of VLDL-TG and VLDL-apoB has been well demonstrated in the insulin-resistant state in both humans and animal models, few data are available on the underlying cellular mechanisms involved, particularly those directly affecting the apoB protein itself. The majority of studies have focused on the acute effects of insulin, while the role of chronic hyperinsulinemia and insulin resistance in VLDL overproduction have been understudied. In the present study, we have simultaneously examined the specific impact of inducing an insulin-resistant condition on the rate of apoB expression at the potential regulatory steps of synthesis, intracellular degradation, and lipoprotein assembly. We employed a fructose-fed hamster model to investigate the above mechanisms in the state of insulin resistance. This model offers advantages over the more commonly used fructose-fed rat model, in that the metabolism of its apoB-containing lipoproteins is more similar to that of humans. It has been well documented that fructose feeding in rodents including hamsters (48), results in chronic hyperinsulinemia, an insulin resistance state, and hyperlipidemia. A previous study (48) clearly demonstrated the feasibility of inducing insulin resistance and chronic hyperinsulinemia in fructose-fed hamsters. The fructose protocol employed in the current study was also similar to those previously shown to induce insulin resistance in the rat (67). The data from in vivo hyperinsulinemic-euglycemic clamp studies presented in this article also support the induction of an insulin-resistant condition in the fructose-fed hamster model. The in vivo clamp study suggests whole body resistance to insulin action and reduced rate of in vivo glucose uptake.
Our in vivo Triton 1339 studies suggested the hepatic overproduction of VLDL-apoB in the fructose-fed hamster model. We also observed an increase in VLDL-TG production rate with fructose feeding although this was not statistically significant.
In vitro experiments with primary hamster hepatocytes further confirmed both VLDL-TG and VLDL-apoB overproduction in hepatocytes from fructose-fed hamsters. These abnormalities closely resemble those seen in insulin resistance and Type 2 diabetes in humans, in which increased VLDL production is the main abnormality of lipoprotein metabolism (23,25,68,69). This observation is important because some animal models of insulin resistance and hyperlipidemia, such as the ob and db mouse have been shown not to overproduce VLDL in vivo, an observation which limits the usefulness of these animals as a model of human pathophysiology (70). Although fructose feeding has been shown to induce an increase in VLDL-TG production in vivo in rats (50,51) these previous studies did not investigate the VLDL-apoB production rate.
The insulin-resistant, fructose-fed hamster model thus provided an excellent system to investigate the intracellular mechanisms that may mediate the considerable VLDL-apoB overproduction observed. A number of important observations were made which appear to explain the VLDL-apoB overproduction in this model. First, there was a significant enhancement of intracellular stability of newly synthesized apoB with only a minor fraction being sorted to intracellular degradation. The increased intracellular stability of apoB in fructose-fed hepatocytes was evident both in intact cells as well as in permeabilized cells. Turnover of nascent apoB was slowed in intact fructose-fed hepatocytes compared with control cells. This observation may or may not be related to an enhanced rate of apoB translocation across the endoplasmic reticulum membrane. Whether stimulated translocation of apoB across the endoplasmic reticulum membrane is responsible for the enhanced intracellular stability is currently unknown and awaits further analysis of apoB translocational status in normal and fructose-fed hepatocytes. We are currently investigating this question by analyzing translocational status of apoB in both isolated microsomes as well as permeabilized cells.
Further analysis of lipoprotein formation in hepatocytes derived from fructose-fed animals revealed a considerable stimulation of VLDL assembly under this metabolic condition. This was evident from reduced formation of LDL-like apoB-containing lipoproteins and increased accumulation of VLDL particles in fructose-fed hepatocytes. These observations argue for enhanced efficiency of VLDL assembly in the microsomal lumen of fructose-fed hepatocytes. Facilitated assembly of hamster VLDL may be related to an increased availability of core lipids, an increased availability of freshly translated apoB, and/or increased activity of MTP. Analysis of intracellular lipid biosynthesis revealed a significant increase in intracellular TG levels, which may in turn contribute to increased assembly of VLDL. In addition, intracellular stability of nascent apoB was also increased, making a higher pool of nascent apoB molecules available for VLDL assembly. Most interesting, however, was an increased mass of MTP detected in fructose-fed hepatocytes. MTP catalyzes the transfer of lipids to the apoB molecule and is an important factor involved in the assembly of apoB-containing lipoproteins (71,72). Inhibition of the activity of MTP blocks the assembly and secretion of apoB-containing lipoprotein particles (73). Thus it is reasonable to conclude that an increased intracellular mass of MTP can enhance the VLDL assembly process, leading to formation and secretion of an increased number of mature particles. Furthermore, the combination of an increased abundance of MTP, in the presence of both higher availability of TG as well as apoB, strongly favors the formation of VLDL particles and their secretion from the cell. Insulin is known to acutely diminish both the MTP mRNA level as well as the mass of MTP protein (74). The insulin effect was shown to be dose-and time-dependent and mediated through the insulin receptor (75). Despite acute inhibition of MTP mRNA levels, short-term insulin treatment (24 h) did not change MTP activity levels due to the slow turnover rate of MTP, t1 ⁄2 ϭ 4.4 days. These observations suggested that sustained changes in MTP mRNA levels would be required to affect MTP protein levels (75). The 5Ј ends of both human and hamster MTP genes contain a negative insulin response element whose activity is negatively regulated by insulin (76). Very recent studies in Otsuka Long-Evans Tokushima Fatty rat, an animal model of Type 2 diabetes, characterized by visceral obesity and hyperlipidemia, has shown enhanced expression of acyl-coenzyme A synthetase, and MTP mRNA in the absence of insulin resistance (77). These investigators suggested that the enhanced expression of both acyl-coenzyme A synthetase and MTP genes associated with visceral fat accumulation, prior to the development of insulin resistance, may be involved in the pathogenesis of hyperlipidemia in obese animal models with Type 2 diabetes (77). In contrast to these findings, MTP protein levels were found to be unaltered in the streptozotocin diabetic rat and 10-day-old suckling rats, animal models in which VLDL-TG secretion is markedly reduced (78). Thus, whether increased MTP causes the increased stability and assembly of VLDL in insulin resistance or is merely secondary to the increase in intracellular lipid synthesis is currently unknown.
Hepatic overproduction of VLDL in the state of insulin resistance may result from direct hepatic effects of insulin as well as indirect metabolic effects, such as increased availability of free fatty acids (FFA) for TG secretion (23). In the present study, we found significantly elevated plasma levels of free fatty acids in fructose-fed hamsters, suggesting that increased flux of FFA into the liver may contribute to VLDL overproduction. However, we did not measure in vivo FFA flux in fructosefed hamsters and cannot confirm the impact of plasma FFA elevation on in vivo VLDL production rates. It is also important to note that the rate of VLDL-apoB secretion from primary hamster hepatocytes was measured under identical concentrations of free fatty acids in the culture media, for both control and fructose-fed hepatocytes. Elevated FFAs in the presence of hyperinsulinemia may have induced hepatic enzymes responsible for channeling FFAs into secretory rather than oxidative pathways, which could have had lasting effects in the cultured hepatocytes.
In conclusion, the fructose-fed hamster model has allowed us to address a number of important questions regarding the intracellular mechanisms that modulate hepatic VLDL assembly and secretion. The evidence obtained in this model suggest that the hepatic overproduction of apoB observed in insulin FIG. 10. Immunoblotting analysis of intracellular mass of MTP in isolated hepatocytes. Control (C) and fructose-fed (F) hepatocytes were solubilized, and equal amounts of cell protein (1 g) were subjected to SDS-PAGE (10% (v/v) acrylamide resolving gel) and proteins were then transferred onto nitrocellulose membranes. Immunoblotting was performed to detect the 97-kDa MTP subunit with a rabbit antihamster MTP antiserum. A, the autoradiograph of the MTP immunoblot. In B, the MTP bands were quantitated by densitometric scanning and the mass of the 97-kDa MTP subunit detected were expressed as a percentage of the MTP mass detected in control cells. resistance may be caused by the combined effect of an increased expression of MTP, increased hepatocyte neutral lipid availability, and reduced degradation of apoB, which can in turn facilitate the assembly and secretion of apoB-containing lipoprotein particles. Precisely which of these factors occurs directly as a result of hepatic hyperinsulinemia or insulin resistance and which are secondary to the extrahepatic effects of insulin is not currently known. The hepatic effects may be due to a direct action of insulin or may be secondary to an increased lipid availability. The mechanisms for enhanced intracellular stability of apoB and increased extracellular secretion are currently unknown and will require further investigation particularly on whether fructose feeding affects apoB translocation. Enhanced activity of MTP may contribute to intracellular stability of apoB, but whether it is sufficient by itself to explain the VLDL overproduction is unknown. Further studies are required to fully investigate the mechanisms by which insulin resistance can influence either the expression or intracellular stability of MTP and thus exert a stimulatory effect on VLDL assembly and secretion. Of particular interest is the interaction of MTP abundance/activity, intracellular apoB stability, and core lipid availability in determining the efficiency of the VLDL assembly process.