An MTP Inhibitor That Normalizes Atherogenic Lipoprotein Levels in WHHL Rabbits

See allHide authors and affiliations

Science  23 Oct 1998:
Vol. 282, Issue 5389, pp. 751-754
DOI: 10.1126/science.282.5389.751


Patients with abetalipoproteinemia, a disease caused by defects in the microsomal triglyceride transfer protein (MTP), do not produce apolipoprotein B–containing lipoproteins. It was hypothesized that small molecule inhibitors of MTP would prevent the assembly and secretion of these atherogenic lipoproteins. To test this hypothesis, two compounds identified in a high-throughput screen for MTP inhibitors were used to direct the synthesis of a highly potent MTP inhibitor. This molecule (compound 9) inhibited the production of lipoprotein particles in rodent models and normalized plasma lipoprotein levels in Watanabe-heritable hyperlipidemic (WHHL) rabbits, which are a model for human homozygous familial hypercholesterolemia. These results suggest that compound 9, or derivatives thereof, has potential applications for the therapeutic lowering of atherogenic lipoprotein levels in humans.

Apolipoprotein B (apoB)–containing lipoproteins [chylomicrons, very low density lipoproteins (VLDL) and their respective metabolic products, chylomicron remnants, and low density lipoproteins (LDL)] promote coronary artery atherosclerosis, which is a leading cause of death in industrialized nations. MTP is a heterodimeric lipid transfer protein consisting of protein disulfide isomerase and a unique 97-kD subunit that is localized in the endoplasmic reticulum of hepatocytes and enterocytes (1–3). Defects in MTP cause abetalipoproteinemia (3–5), a disorder in which the production of VLDL and chylomicrons is disrupted. Patients with abetalipoproteinemia have plasma cholesterol levels of ∼40 mg/dl and plasma triglyceride levels of <10 mg/dl (6), whereas normal adults have levels of 180 to 220 and 100 to 150 mg/dl, respectively. These findings suggest that inhibitors of MTP might be therapeutically useful for inhibiting the production of VLDL and chylomicrons, thereby reducing the levels of atherogenic lipoprotein particles.

Abetalipoproteinemia is an extreme example of MTP inhibition and would not be the intended clinical end point for a drug. A related genetic disease, hypobetalipoproteinemia, is caused by mutations in apoB (6, 7). Heterozygous people with this disease have half the normal levels of apoB-containing lipoproteins and lack the clinical signs and symptoms of patients with abetalipoproteinemia. Family studies have shown that individuals with hypobetalipoproteinemia have a prolonged life span (8).

We performed a high-throughput screen of a large chemical library to identify inhibitors of MTP-mediated triglyceride transfer. This resulted in the discovery of BMS-200150 (compound 1 in Fig. 1), which inhibits the MTP-mediated transport of triglycerides between membranes in vitro (9) and inhibits the secretion of apoB-containing lipoproteins from HepG2 cells, a human liver-derived cell line. However, the compound was not active in animal models. Compound 2, an analog of BMS-200150 with an extended alkyl linker, was of comparable potency in both the lipid transfer and HepG2 apoB secretion assays (Fig. 1 and Table 1). A fluorenyl amide (compound 3), which is a much less potent inhibitor, was also identified in the high-throughput screening. Subsequently, nitrogen substitution of 3 with a short-chain alkyl group (for example, 4) was found to improve the potency of this series, making it comparable with the series of1 and 2.

Figure 1

Structures of MTP inhibitors.

Table 1

Comparative activity of compounds as MTP inhibitors measured in vitro, in cell culture, and in vivo. The in vitro and cell culture assays were performed as described in (9). The fasted rat assay was performed as described in Fig. 2 and (10). For intravenous (iv) dosing, the compound was directly incorporated into Triton WR1339. Human MTP was produced in Sf9 insect cells with the baculovirus expression system (4). IC50, concentration that produces half-maximal inhibition; ED50, dose that produces a half-maximal effect; IA@2, IA@15, and IA@100, inactive at doses 2, 15, and 100 mg/kg, respectively; ND, not done.

View this table:

On the basis of the structural relationships between 1,2, and 4, it was proposed that 4 binds to MTP at a subdomain of the binding site for 1 and2. To test this proposal, we prepared “hybrid” analogs5 and 6 (Fig. 1). With 5, the overlap between the fluorenyl amide of 4 and the isoindolone of1 was explored, resulting in a significant loss in inhibitory potency (Table 1). With 6, the overlap between the fluorenyl amide of 4 and the diphenylmethyl substituent of 2 was explored, resulting in a dramatic improvement (>100-fold) in potency in both the lipid transfer and HepG2 apoB secretion assays.

Hybrid 6 was tested for its ability to inhibit lipoprotein secretion in fasted rats. This assay was based on the ability of intravenously injected Triton WR1339 to prevent the catabolism of triglyceride-rich lipoproteins (10). Thus, after the administration of Triton, there was a linear increase in plasma triglyceride levels, which reflects the triglyceride secretion rate from the liver and intestine. The inhibition of hepatic lipoprotein production can be measured with fasted animals in which the liver is the primary source of plasma triglycerides, whereas in fed rats, there is substantial lipoprotein production in the intestine. In contrast to the earlier analogs, hybrid 6 was a potent and dose-dependent inhibitor of triglyceride secretion in vivo. The corresponding N-CH2CF3 analog, 7, resulted in a dramatic improvement in oral potency.

A modification of the isoindolone portion of 7indicated that a benzamide (inhibitor 8) was a suitable replacement for the isoindolone ring. An extensive optimization of this substituent was performed by an automated organic synthesis (11), which resulted in the identification of the 4′-CF3-biphenyl carboxamide, 9. This MTP inhibitor possesses subnanomolar potency in both the lipid transfer and HepG2 apoB secretion assays and inhibits lipoprotein secretion in fasted rats (Table 1). The intravenous and oral half-maximal effective dose (ED50) values were similar, suggesting that the compound was well-absorbed and bioavailable to the liver. Compound 9 was equally effective in inhibiting acute lipoprotein secretion in fasted or fed rats (Fig. 2), indicating that it inhibits both hepatic and intestinal lipoprotein secretion.

Figure 2

Inhibition by compound9 of triglyceride secretion in fasted and fed rats. Sprague-Dawley rats (∼200 g each, four per treatment group) were adapted to a reversed diurnal light cycle for 2 weeks. Before the experiment, the rats either were fasted (solid bars) or had free access to food (hatched bars) for 18 hours. The animals were orally dosed with compound 9 in a vehicle (10% M-pyrol, 80% water, 5% cremophore, and 5% ethanol) 1 hour before receiving an intravenous injection of Triton WR1339 (250 mg/kg). The triglyceride secretion rate was determined by calculating the amount of triglyceride that was accumulated in plasma during the 2.5 hours after the Triton injection. The standard assay was linear for at least 5 hours after the Triton injection. Plasma triglyceride levels were determined with a Roche Cobas blood chemistry autoanalyzer (a negative percentage indicates a decrease). The animals were cared for in accordance with institutional guidelines. Error bars indicate SEM. **, significantly different from vehicle-treated animals at P < 0.005.

The effect of compound 9 on plasma lipid levels was tested in hamsters which, unlike other rodents, transport a substantial proportion of their cholesterol on LDL. Hamsters that were fed a standard, low-fat, high-carbohydrate diet were treated once daily with doses of compound 9 of 1, 3, or 6 mg per kilogram of body weight (mg/kg) for 7 days. After the final drug treatment, the animals were fasted for 18 hours, and their plasma lipid and lipoprotein levels were then evaluated (Fig. 3). A dose-dependent decrease in the total plasma cholesterol was observed, with an ED50 value for cholesterol lowering of 2.4 mg/kg. The triglyceride, VLDL and LDL cholesterol, and high density lipoprotein (HDL) cholesterol levels all decreased in parallel with the decrease in total plasma cholesterol. The HDL decrease was unlikely to be due to a direct effect on HDL production because, in HepG2 cells, the ED50 for the inhibition of secretion of apolipoprotein AI (apoAI; the major structural protein of HDL) by compound9 is over 8000-fold higher than that for the inhibition of secretion of apoB (Table 1). At the 6 mg/kg dose, the total plasma cholesterol and triglyceride levels were decreased by 90 and 49%, respectively, in comparison to the control animals. When animals were treated with compound 9 for up to 3 weeks, there were similar findings. There were no clinically relevant differences between vehicle- and drug-treated animals in body weight, plasma aspartate aminotransferase (AST), alanine aminotransferase (ALT), or any other plasma chemistry variables that were tested (12).

Figure 3

Effect of compound9 on (A) total cholesterol, (B) triglycerides, (C) VLDL and LDL cholesterol, and (D) HDL cholesterol levels in hamsters (a negative percentage indicates a decrease). Male Golden Syrian hamsters (∼140 g each, four per treatment group) were adapted to and maintained on a reverse diurnal light cycle. They were dosed orally once a day with compound 9 in a vehicle (10% M-pyrol, 80% water, 5% cremophore, and 5% ethanol) and were allowed free access to a standard hamster diet (Purina 5001, which contains 0.02% cholesterol and 4.5% triglyceride). After 7 days of drug treatment, hamsters were fasted for 18 hours, after which plasma lipid levels and chemistries were determined with a Roche Cobas blood chemistry autoanalyzer. Hamster lipoprotein fractions were quantitated after the precipitation of apoB-containing lipoproteins with phosphotungstate and magnesium chloride. Drug effects were calculated in relation to the vehicle control group. Error bars indicate SEM. *, P < 0.05; **, P < 0.005.

Hamsters that were treated with compound 9 did not have steatorrhea, as indicated by the absence of fat (<0.01 g per 24 hours) in their stools. After killing these hamsters, we noted visual and biochemical evidence of fat accumulation in enterocytes. The liver triglyceride content of animals that were treated with compound9 reached a plateau after 1 week of treatment, never exceeded 25 mg per gram of tissue (13), and returned to control levels 48 hours after the termination of a 3-week treatment regimen. In comparison, fatty livers in humans contain up to 35% triglyceride by weight (14). Liver weight in the hamsters (as percent body weight) showed minimal change with treatment (15). Plasma transaminase levels did not rise significantly over a 3-week treatment period, and triglyceride accumulation was reversible, which suggest that the modest increase in hepatic lipid had no adverse effects in the hamster model.

We also investigated the effect of a 2-week treatment with compound9 in homozygous Watanabe-heritable hyperlipidemic (WHHL) rabbits whose hepatic LDL receptor activity is <5% that of normal rabbits, resulting in dramatically elevated levels of apoB-containing lipoproteins. WHHL rabbits are a model for human homozygous familial hypercholesterolemia (FH) (16). The elevated plasma cholesterol levels in FH patients (600 to 1200 mg/dl) cannot be normalized with current therapies, which typically lower LDL cholesterol levels by a maximum of 30%. The ED50 value for cholesterol lowering in WHHL rabbits that were treated with compound9 was 1.9 mg/kg. Triglyceride levels were lowered in parallel. At a dose of 10 mg/kg of 9, the plasma levels of atherogenic, apoB-containing, lipoprotein particles were essentially normalized (Fig. 4) with no alteration in plasma AST or ALT (17).

Figure 4

Effect of compound 9 on plasma lipid levels in WHHL rabbits (a negative percentage indicates a decrease). Five rabbits were treated orally for 14 days with compound 9(10 mg/kg). Plasma (A) total cholesterol and (B) triglyceride levels were measured 18 hours after the last dose. Prebleed, solid bars; 14-day treatment, hatched bars.

In the search for an understanding of lipoprotein metabolism and a means to treat atherosclerosis, much effort has been focused on hyperlipidemic states. This focus has resulted in the identification of genetic defects that produce loss of functions that cause hyperlipidemia. Our results highlight the importance of understanding hypolipidemic states. Identifying a loss-of-function gene defect for a genetic disorder that is the physiological opposite of the disease that one is attempting to treat allows one to identify a target that is theoretically amenable to pharmacological intervention (that is, a normal protein for which an inhibitor mimicking the effect of the mutation will produce the desired therapeutic outcome). With current technologies, it is easier to identify a small molecule inhibitor of a normal activity than to augment a deficient activity. This strategy yielded compound 9, which has the potential to be an effective anti-atherogenic agent and is now in clinical trials. Similar drug discovery strategies may be applicable to other diseases.

  • * To whom correspondence should be addressed. E-mail: Wetterau_John_R{at} (J.R.W.) and biller{at} (S.A.B.)

  • Present address: GelTex Pharmacueticals, 9 Fourth Avenue, Waltham, MA 02154, USA.

  • Present address: 14 Royal College of Surgeons, Dublin, Ireland.

  • § Present address: Department of Chemistry, 516 Physical Sciences 1, University of California, Irvine, CA 92697, USA.


View Abstract

Navigate This Article