Hepatic Expression of Microsomal Triglyceride Transfer Protein and In Vivo Secretion of Triglyceride-Rich Lipoproteins Are Increased in Obese Diabetic Mice

Fifty-four male and 34 female C57BL/6 mice from 5 to 7 weeks of age were obtained from M&B (Ry, Denmark) and were fed a standard laboratory diet (Altromin no. 1314; Altromin, Rugarden, Denmark). The mice were housed at the Panum Institute, University of Copenhagen, in a temperature-controlled (21–23°C) facility with a 12-h light/dark cycle. Female leptin-deficient Lep^ob/Lep^ob (ob/ob) mice (n = 20) and heterozygote Lep^ob/+ (ob/+) controls (n = 20) of the Umea strain were obtained from M&B, maintained under similar conditions at the Panum Institute or at Novo Nordisk A/S (Bagsvaerd, Denmark), and killed at 10 weeks of age. The Danish Government body that supervises animal experiments (Dyreforsoegstilsynet) approved the study protocols.

To induce diabetes in C57BL/6 mice, we injected STZ (40 μg/g per injection; Sigma, Vallensbaek Strand, Denmark) intraperitoneally seven times (once a day for 4 or 5 consecutive days and once a day for an additional 2–3 consecutive days after 2–3 days of recovery). STZ was dissolved in citrate buffer (0.05 mol/l [pH 4.5], 4°C) to a final concentration of 4 μg/μl. The STZ solution was kept on ice and used within 20 min of preparation. Control mice received the citrate-buffer vehicle (200 μl/mouse per day). To maintain hyperglycemia in the 9-week STZ treatment group, STZ was injected for an additional 5 days (40 μg/g per injection) during week 7. Mice that were used to study hepatic gene expression and lipid content were killed by cervical dislocation after 3 (n = 20), 6 (n = 20), or 9 weeks (n = 28). Equal numbers of males and females and STZ-treated and vehicle-treated control mice were studied at each time point. Liver biopsies were placed in liquid nitrogen immediately upon removal and stored at −141°C. Because differences in gene expression, plasma, and tissue lipid levels between STZ-treated mice and controls were similar at 3, 6, and 9 weeks, data from the three time points were pooled and are presented together. Hepatic triglyceride transfer activity and in vivo triglyceride secretion were studied in male mice 3 weeks after STZ induction of diabetes.

All procedures were performed on ice. Microsomal proteins were prepared by homogenization of ∼100-mg liver samples in 1 ml buffer (50 mmol/l Tris-HCl, 50 mmol/l KCl, 5 mmol/l EDTA, and freshly added Protease Inhibitor [Roche A/S]) with a PT 1200 Polytron. The protein concentration in homogenates was determined with the BCA-protein assay (Pierce, Copenhagen, Denmark) using BSA as a standard. After the protein concentration was adjusted to 1.75 mg/ml, liver homogenates were subjected to ultracentrifugation in a Beckman Optima LE-80K ultracentrifuge with the 50.4 Ti rotor head at 100,000g for 60 min. The supernatant containing the microsomal fraction was added to 1/10 volume of 0.54% sodiumdeoxycholate (pH 7.5) and incubated on ice for 30 min followed by overnight dialysis at 4°C against buffer (15 mmol/l Tris [pH 7.4], 40 mmol/l NaCl, 10 mmol/l EDTA, and 0.02% NaN₃). Triglyceride transfer activity in the microsomal protein fraction was measured as transfer of ¹⁴C-triglycerides from labeled donor vesicles to acceptor vesicle that contained only unlabeled triglycerides, exactly as described by Wetterau et al. (28,29). Initial evaluations of the assay showed that triglyceride transfer activity in normal mouse liver microsomal extracts increased linearly with an increasing amount of protein up to ∼25 μg protein. Consequently, we chose to analyze triglyceride transfer activity in microsomal extracts corresponding to 10 μg total protein. In protein purification and triglyceride transfer procedures, samples from ob/ob and ob/+-control mice were alternated and analyzed together. Samples from STZ- and vehicle-treated mice were likewise analyzed together by alternating samples from the two groups of mice.

Ob/ob, ob/+, STZ-treated, and vehicle-treated mice (n = 10 in each group) were fasted from 10:00 p.m. The next day, the mice were anesthetized with hypnorm:diazepam:saline (1:1:2, 6 ml/kg s.c.), and Triton WR-1339 (500 mg/kg in 200 μl sterile 0.9% NaCl; Sigma) was injected trough a tail vein; intravenous injection was successful in nine mice from each group. Blood samples (∼75 μl) were taken 1, 30, 60, and 120 min after injection from the retro-orbital plexus. Triglyceride secretion rate (in μmol triglycerides per hour per gram body weight) was calculated as the increase in plasma triglyceride concentration per hour multiplied by the plasma volume (i.e., 3.5% of the body weight) and divided by the body weight. Because triglyceride secretion rates were similar in the ob/+ and vehicle-treated control mice (0.068 ± 0.005 vs. 0.065 ± 0.007 μmol · h⁻¹ · g⁻¹), data for control mice are presented together.

The glucose concentration in tail blood was determined with the HUMACHECK^plus glucose meter (HUMAN, Wiesbaden, Germany). Blood samples for plasma insulin and lipid analyses were drawn into tubes that contained Na₂EDTA. Plasma aliquots were stored at −80°C until analyses. Plasma insulin was determined with enzyme-linked immunosorbent assay by Dr. B. Rolin (Novo Nordisk A/S) (30). Enzymatic kits were used to determine plasma concentrations of free fatty acids (WAKO NEFA C kit; TriChem Aps, Frederikssund, Denmark), triglycerides (GPO-TRINDER; Sigma), total cholesterol (CHOD-PAP; Roche, Mannheim, Germany), and HDL cholesterol (Roche). HDL cholesterol was measured after precipitation of apoB-containing lipoprotein with phosphotungstic acid and MgCl₂.

Liver biopsies (∼150 mg) were minced with scissors in 1 ml methanol before adding 1.5 ml methanol and 5 ml chloroform. After thorough mixing, 2.5 ml methanol was added and the mixture was vortexed and centrifuged for 15 min at 4°C at 2,000g. The supernatant containing the lipid extract was transferred to a new vial. The remaining tissue extract was washed in 1 + 1 ml chloroform-methanol and centrifuged. The supernatant of the wash was transferred to the vial with the lipid extract, and the wash was repeated. After evaporation of solvents under N₂, the lipids were redissolved in toluol (1 μl/mg wet weight of tissue) and quantified by thin-layer chromatography (TLC) (31). Briefly, TLC plates (DC-Fertigplatten SIL G-25, Macherey-Nagel, 20 × 20 cm) were impregnated with Na₂EDTA (1 mmol/l, pH 5.5) by ascending development, dried, and washed in chloroform:methanol:water (60:40:10). After activation of the TLC plates at 110°C for 30 min, 1-μl tissue extracts and a dilution series of a standard were applied. The standard contained triglyceride, free cholesterol, cholesterol esters, sphingomyelin, phosphatidylcholine, phosphatidylserine, phosphatidylinositol, phosphatidylethanolamine, and cardiolipin (all from Sigma). The plates were developed in a six-step procedure using chloroform:methanol:water (60:40:10), chloroform:methanol:water (65:40:5), chloroform:methanol:ethylacetat:2-propanol:ethanol:0.25% KCl (22:15:35:5:20:9), ethanol:toluol:diethylether (3:60:40), diethylether:n-heptan (8:94), and pure n-heptan. After development, the plates were placed in 10% cupric sulfate (wt/vol) in 8% phosphoric acid (vol/vol) for 10 s, dried with a hair dryer, and baked for 2 min at 200°C. For quantification, the TLC plates were scanned at 300 × 300 dpi with an HP flatbed 4c/T scanner (Hewlett Packard, Copenhagen, Denmark). The digitalized images were analyzed with the MULTI-ANALYST software (Bio-Rad, Copenhagen, Denmark). All samples were analyzed at least twice on different TLC plates. The interassay coefficients of variation of lipid determination were 11% for triglyceride and ∼14% for the other lipid fractions (n = 205–228 duplicate analyses).

Frozen liver samples (20–100 mg) were homogenized with a Polytron PT1200CL (Buch & Holm, Herlev, Denmark) in TRIzol reagent (Life Technologies, Taastrup, Denmark). Total RNA was extracted according to the manufacturer’s manual and suspended in RNase-free H₂O. The RNA concentration was calculated from the absorbance at 260 nm (A₂₆₀). The integrity of the RNA was always ensured by electrophoresis on a 1% agarose gel.

First-strand cDNA was synthesized at 37°C for 60 min: 1 μg RNA, M-MULV reverse transcriptase (40 units; Roche A/S, Copenhagen, Denmark), and random hexamer primers (0.02 A₂₆₀) were mixed in 10-μl reactions. The primers used for PCR were as follows: apoB, 5′-ATGGGAAGAAACAGGCTTGA-3′ and 5′-TTCTGTCCCACGAATTGACA-3′; MTP, 5′-TGAGCGGCTATACAAGCTCAC-3′ and 5′-CTGGAAGATGCTCTTCTCGC-3′ (12); β-actin, 5′-ACTGGGACGACATGGAGAAG-3′ and 5′-GGGGTGTTGAAGGTCTCAAA-3′; SREBP-1a, 5′-GCCGGCGCCATGGACGAG-3′ and 5′-ATGTCGTTCAAAACCGCTGTGTC-3′; SREBP-1c, 5′-CGCGGAAGCTGTCGGGGTAG-3′ and 5′-AAATGTGCAATCCATGGCTCCGTGGTC-3′; and SREBP-2, 5′-CACAATATCATTGAAAAGCGCTACCGGTCC-3′ and 5′-TTTTTCTGATTGGCCAGCTTCAGCACCATG-3′ (32) (all from SIGMA-GENOSYS, Pampisford, U.K.).

The expected length of each PCR product was confirmed by electrophoresis on 2% agarose gels. Primer pairs for apoB, MTP, β-actin, glyceraldehyde-3-phosphate dehydrogenase, and SREBP-2 spanned introns. We did not observe amplification of genomic DNA in any of the cDNA preparations. The specificity of each RT-PCR reaction was also ensured by DNA sequencing the RT-PCR products in both directions with the ABI-Prism 310 sequencer and terminator ready reaction mix (PE Applied Biosystems, Naerum, Denmark). In some cases, the PCR products were cloned into a plasmid before sequencing; in other cases, PCR products were sequenced directly after purification on JET quick-spin columns (Genomed, Bad Oeynhausen, Germany).

We used the LightCycler instrument (software version 3.39) and DNAmaster SYBR GREEN kit (Roche A/S, Hvidovre, Denmark) to perform quantitative real-time PCR analysis of hepatic cDNA preparations. The PCR reactions (20 μl) contained 2 μl SYBR GREEN I mix, 2–3 mmol/l MgCl₂, 10 pmol of each primer, cDNA synthesized from 20 ng total RNA, and PCR-grade H₂O. The melting point of the PCR products, the optimal MgCl₂ concentration, and annealing temperature of each PCR were carefully established. The amount of a specific mRNA transcript in each liver sample was determined from the time point of the log-linear increase in amplified DNA during the PCR. The fit-point option of the LightCycler software was always used in the quantifications. The relation between the time point of the log-linear increase in amplified DNA and the relative concentration of an mRNA transcript were determined in each run by analyzing dilution series (liver cDNA synthesized from 100 ng, 20 ng, 2 ng, and 0.2 ng total RNA) of a liver cDNA pool. This reference cDNA pool was made by mixing liver cDNA from three normal C57BL/6 male mice. We were initially concerned that the effectiveness of the PCR could be lower in standards with the highest amount of cDNA. Such an effect could potentially introduce an error in the mRNA quantifications. Therefore, in control experiments, we generated dilution series of mouse liver cDNA in which yeast cDNA had been included to obtain the same amount of cDNA in each sample (i.e., cDNA synthesized from 200 ng total RNA). However, the addition of yeast cDNA did not change the effectiveness of three different PCRs (data not shown). Consequently, all standard curves were generated by diluting the cDNA pool with H₂O. The real-time PCR assays were sufficiently sensitive to quantify MTP, apoB, and β-actin mRNA in <20 pg total liver RNA and SREBP-1a, -1c, and -2 mRNA in ∼200 pg total liver RNA. To assess the imprecision of the mRNA quantifications, we quantified MTP, apoB, and β-actin mRNAs on five different occasions using the same liver cDNA preparation made from 20 ng total RNA. The interassay coefficients of variation (i.e., SDs divided by the mean) were 5.0%, 3.5%, and 4.8% for MTP, apoB, and β-actin mRNA, respectively. In each sample, mRNA levels were normalized by the β-actin mRNA content.

Differences between groups were analyzed with Student’s t test with correction for different variance whenever appropriate. The analyses were done with the GraphPad Prism software (San Diego, CA).

ob/ob mice were hyperglycemic and had markedly elevated plasma insulin levels compared with ob/+ control mice (Table 1). The total plasma cholesterol concentration was 2.3-fold higher in the ob/ob mice than in the ob/+ control mice. This difference almost exclusively reflected an increased concentration of VLDL+LDL cholesterol (Table 1). The hepatic triglyceride content was increased ∼13-fold in the ob/ob mice (Table 2). This triglyceride accumulation was accompanied by an ∼twofold increase in cholesterol ester and a relative reduction in other lipid fractions (Table 2).

The hepatic MTP mRNA level was increased by 45% in the ob/ob mice compared with the ob/+ control mice (P = 0.006), whereas hepatic apoB mRNA levels were similar in the two groups (Fig. 1A). Analysis of triglyceride transfer activity in microsomal liver proteins revealed a 54% (P < 0.0001) increased activity in ob/ob mice compared with ob/+ control mice (Fig. 2). Furthermore, the in vivo triglyceride secretion rate from the liver was increased 70% in ob/ob mice compared with controls (0.113 ± 0.008 vs. 0.066 ± 0.004 μmol/h per g body wt; P < 0.0001; Fig. 3). Because plasma triglyceride levels were not increased in ob/ob mice (Table 1 and Fig. 3), these data suggest that the clearance of triglycerides from the blood is increased in ob/ob mice.

To explore further the putative effects of diabetes on hepatic MTP expression, we induced diabetes in lean C57BL/6 mice by repeated injections of STZ. The plasma insulin level was lower in the STZ-treated mice than in the vehicle-treated control mice (Table 1). However, the diabetic response varied between STZ-treated mice. Therefore, the STZ-treated mice were arbitrarily divided at the median blood glucose concentration into mice with fasting blood glucose ≥13.1 mmol/l and mice with fasting blood glucose <13.1 mmol/l. STZ-treated mice with blood glucose ≥13.1 mmol/l had 75% higher plasma triglycerides (P < 0.0005) and 17% higher total plasma cholesterol (P < 0.01) levels compared with vehicle-treated control mice (Table 1). Plasma lipid levels were similar in STZ-treated mice with blood glucose <13.1 mmol/l and vehicle-treated control mice. Consistent with previous studies of STZ-treated C57BL/6 mice that were fed a standard diet (33), the concentration of hepatic triglycerides and cholesterol esters were lower in both groups of STZ-treated mice when compared with the vehicle-treated control mice (Table 2).

There was no difference in hepatic MTP and apoB mRNA levels between STZ-treated diabetic and vehicle-treated control mice, irrespective of the blood glucose concentration (Fig. 1B). To assess hepatic microsomal triglyceride transfer activity and in vivo triglyceride secretion rates in STZ-treated mice, we studied another group of STZ-treated or vehicle-treated male mice 3 weeks after induction of diabetes. The mean blood glucose concentration was 22.0 ± 1.0 mmol/l in STZ-treated mice and 8.0 ± 0.2 mmol/l in vehicle-treated control mice. Hepatic microsomal triglyceride transfer activity was 14% lower (P = 0.04) and in vivo triglyceride secretion rates were 42% lower (0.039 ± 0.003 vs. 0.066 ± 0.004 μmol · h⁻¹ · g⁻¹; P = 0.0002) in STZ-treated diabetic mice compared with control mice (Figs. 2 and 3). The small difference in MTP activity between STZ-treated and vehicle-treated control mice (P = 0.04) should be interpreted with caution because of the multiple two-group comparisons made in this study. Because STZ-treated mice had increased plasma triglycerides (Table 1 and Fig. 3), an unchanged (or perhaps slightly decreased) triglyceride secretion rate suggests that triglyceride clearance from the blood is attenuated in STZ-treated mice.

Reporter-gene studies suggest that MTP gene expression can be regulated by SREBP transcription factors (17). To assess whether the observed changes in MTP expression in ob/ob mice might be paralleled by changes in SREBP gene expression, we measured the hepatic mRNA levels of SREBP genes. In accordance with a previous report (20), mRNA levels of SREBP-1a and -1c were increased 1.5- and 2.0-fold, respectively, in the ob/ob mice (Fig. 4A; P = 0.03 for SREBP-1a and P = 0.002 for SREBP-1c). SREBP-2 mRNA levels were similar in ob/ob and ob/+ control mice (Fig. 4A). In STZ-treated mice with blood glucose ≥13.1 mmol/l, SREBP-1c mRNA was increased by 54% (P = 0.02), whereas the STZ mice with blood glucose <13.1 mmol/l had intermediate SREBP-1c mRNA levels (Fig. 4B). SREBP-1a and -2 mRNA levels were similar in STZ-treated and vehicle-treated control mice (Fig. 4B).

Cell culture studies suggest that insulin and glucose negatively regulate MTP gene expression in liver cells (15). These in vitro observations suggest that hepatic MTP gene expression might be decreased in individuals with hyperinsulinemic (type 2) diabetes. However, several metabolic studies have shown that individuals with type 2 diabetes display increased hepatic production of VLDL (21,24,34). The principal aim of the present study was to determine the effect of obesity-associated diabetes in mice on hepatic MTP gene expression, microsomal triglyceride transfer activity, and secretion of triglyceride-rich lipoproteins. Previous studies of mice with genetically altered lipoprotein secretion from the liver (i.e., apoB and MTP knockout mice [12,35,36]) have suggested that an ∼50% reduction in the expression levels of the apoB or MTP gene is accompanied by a significant lowering of the plasma concentration of apoB-containing lipoproteins. Therefore, secondary hyperlipidemia in diabetes could be caused by relatively small changes (e.g., 25–50%) in hepatic gene expression levels. Although such small changes in mRNA expression can be of definite physiological importance, they are difficult to detect with standard techniques. Northern blot and RNase protection assays are often only semiquantitative and cannot efficiently be applied to a large number of tissue samples. The present study demonstrates that the real-time PCR technology is a sensitive and reproducible technology for quantification of mRNA in multiple tissue samples. The studies revealed that hepatic MTP mRNA expression is increased by 45% in hyperinsulinemic diabetic ob/ob mouse and that this difference in gene expression is paralleled by increases in microsomal triglyceride transfer activity and secretion of triglycerides from the liver. These findings support previous observations in vitro and in vivo of tight correlations between MTP gene expression and MTP protein content in liver cells (12,14,25,26). Thus, the results suggest that increased MTP gene expression confers accelerated production of triglyceride-rich apoB-containing lipoproteins in the liver of obese diabetic mice.

It is unknown what causes increased MTP expression in the liver of obese diabetic mice. A lipid enrichment of the diet stimulates the hepatic expression of MTP mRNA in hamsters in a dose-dependent manner (37), and the genetically obese OLEFT rat, which has increased fatty acid levels in the portal vein, also displays increased hepatic MTP activity and lipoprotein secretion (38). This has led to the suggestion that prolonged increases in flux of fatty acid to the liver may stimulate MTP expression (38). The present data are in accordance with this idea because an elevation of MTP expression in ob/ob mice occurred in the setting of increased triglyceride clearance from the blood as well as increased triglyceride stores in the liver. However, in cultured liver cells, addition of oleic acid did not affect MTP expression (15). Thus, additional studies are needed to resolve how MTP gene expression might be regulated in the mouse liver in vivo.

We quantified hepatic SREBP mRNA levels because a recent study proposed that SREBPs might regulate MTP gene expression in liver cells (17). Because SREBP-1c gene expression was increased and SREBP-2 gene expression was unchanged in both obese and lean diabetic mice, the selective increase in MTP expression in obese diabetic mice seems to be unrelated, at least in a direct manner, to changes in SREBP-1c gene expression. Whereas an increase in SREBP-1 gene expression was previously reported in ob/ob mice (20), we were surprised to find that SREBP-1c mRNA expression was also increased in STZ-treated mice with blood glucose >13.1 mmol/l. In a study of STZ-treated diabetic rats, Shimomura et al. (18) found that hepatic triglyceride levels increased by 60% and SREBP-1c mRNA levels were reduced by 80% shortly (42 h) after a single high-dose STZ injection into fasted animals; these acute effects of STZ were completely preventable by insulin treatment. We suspect that differences in study designs might explain the divergent results. In the present study, diabetes was induced in mice by repeated low-dose injections of STZ. After 3–9 weeks, this treatment caused a mild decrease in plasma insulin levels, a reduction in liver triglyceride stores, and a 54% increase in SREBP-1c mRNA levels in mice with severe hyperglycemia, whereas mice with mild hyperglycemia had an intermediate phenotype. The results are compatible with the idea that prolonged exposure to a high blood glucose level stimulates SREBP-1c gene expression in the mouse liver in vivo, even in the presence of slightly decreased plasma insulin levels. In accordance with this idea, a recent study described a glucose-responsive element in the SREBP-1 gene promoter region that confers concentration-dependent increases in SREBP-1 gene expression after prolonged exposure of mouse liver cells to glucose (39). SREBP-1 encodes membrane-bound precursors that need to undergo proteolytic cleavage (40) as well as phosphorylation (41) to induce transcriptional activation of target genes. Thus, it remains to be tested how an increased SREBP-1c mRNA expression in STZ-treated diabetic mice might affect the level of the active NH₂ fragment of SREBP-1c.

Studies of rabbits and hamsters treated with MTP inhibitor drugs have shown that an ∼50% reduction in MTP activity effectively lowered plasma lipoprotein levels and had only minor (but definitive) effects on triglyceride deposition in liver cells (7,9). Even though simple mild hepatic steatosis often has a benign course in nondiabetic individuals, ∼20% of individuals with type 2 diabetes develop nonalcoholic steatohepatitis that ultimately results in cirrhosis in 20–30% of the cases (42). Recent observations propose a link between hepatic MTP gene expression levels and development of nonalcoholic steatohepatitis: indexes of nonalcoholic steatohepatitis in individuals with type 2 diabetes are more frequent in carriers of the G allele of the −493 G/T polymorphism in the MTP promoter than in carriers of the T allele (43). The G allele confers a 50% lower promoter activity compared with the T allele (44). Physiologically, the present finding of increased MTP gene transcription in the fatty livers of type 2 diabetic ob/ob mice may reflect an attempt of the liver to limit triglyceride accumulation. Thus, even though the present studies suggest that MTP inhibition may be a rational approach to pharmacological treatment of hyperlipidemia in individuals with type 2 diabetes, it is possible that reducing hepatic MTP activity might accelerate development of nonalcoholic steatohepatitis in this patient group.

The studies were supported by The Danish Heart Foundation (grant no. 99-2-2-16B-22734), The Danish Medical Research Council (grant no. 9903236), and Novo Nordisk A/S (a scholar stipend to M.L.).

We thank Dr. Bidda Rolin, Novo Nordisk A/S, for ob/ob mice and plasma insulin analysis; Dr. Martin Raabe, Bayer, for helpful hints on MTP activity assays; and Karen Rasmussen for skillful technical assistance.