Identification of DGAT2 Inhibitors Using Mass Spectrometry

Abstract

Mass spectrometry offers significant advantages over other detection technologies in the areas of hit finding, hit validation, and medicinal chemistry compound optimization. The foremost obvious advantage is the ability to directly measure enzymatic product formation. In addition, the inherent sensitivity of the liquid chromatography/mass spectrometry (LC/MS) approach allows the execution of enzymatic assays at substrate concentrations typically at or below substrate K_m. Another advantage of the LC/MS approach is the ability to assay impure enzyme systems that would otherwise be difficult to prosecute with traditional labeled methods. This approach was used to identify inhibitors of diacylglycerol O-acyltransferase-2 (DGAT2), a transmembrane enzyme involved in the triglyceride (TG) production pathway. The LC/MS approach was employed because of its increased assay window (compared with control membranes) of more than sevenfold compared with less than twofold with a conventional fluorogenic assay. The ability to generate thousands of dose-dependent IC₅₀ data was made possible by the use of a staggered parallel LC/MS system with fast elution gradients. From the hit-deconvolution efforts, several structural scaffold series were identified that inhibit DGAT2 activity. Additional profiling of one chemotype in particular identified two promising reversible and selective compounds (compound 15 and compound 16) that effectively inhibit TG production in mouse primary hepatocytes.

Keywords

LC/MS DGAT2 triglycerides

Introduction

Triglycerides (TGs) are important molecules for eukaryotic fuel storage. Altered TG metabolism is associated with a variety of human diseases. TG is synthesized by two major pathways: the Kennedy pathway (de novo synthesis), which is present in most tissues, and the monoacylglycerol pathway, which is predominant in the small intestine to absorb dietary fats and is believed to be the major pathway to pack dietary fat into chylomicron particles. At the final step of both pathways, diacylglycerol (DG) and a fatty acid molecule are joined together to form TG.^1–3 Two enzymes, diacylglycerol O-acyltransferase-1 (DGAT1) and DGAT2, have been identified to share the same substrate specificity and catalyze this reaction, yet these two enzymes are structurally unrelated and share no homologs.^4,5 Furthermore, the knockout phenotypes for DGAT1 and DGAT2 are quite different. DGAT1 knockout mice exhibit normal growth patterns when fed with a chow diet and are resistant to diet-induced obesity. DGAT1 knockout mice have also been reported to have increased insulin sensitivity and increased leptin sensitivity.^6,7,⁸ DGAT2 knockout mice, on the other hand, have dramatically reduced levels of TG and fatty acid, suggesting a critical role of DGAT2 in fundamental TG synthesis.⁷ DGAT2 knockout mice die a few hours after birth; however, several antisense oligonucleotides (ASO) studies using different animal models have demonstrated the potential benefits of DGAT2 inhibition.^9–11 ASO knockdown of DGAT2 for 6 weeks has yielded reduction of liver TG content and body weight in C57BL/6j and ob/ob models.^9,12 In a separate study, knocking down of DGAT2 in diet-induced rats yielded a similar phenotype as well as increased insulin sensitivity.¹⁰

These encouraging results from ASO studies in rodents and potential benefits of DGAT2 inhibition have sparked some interest in recapitulating some of these results using small-molecule inhibition. For such a lead identification process, which includes high-throughput screening (HTS), hit triage, and structure-activity relationship (SAR) study, a robust assay is needed.

A scintillation proximity assay has been reported for DGAT1 with acceptable screening parameters.¹³ However, the assay conditions reported produced a small window (~1.5-fold) for DGAT2 compared with mock membranes. Recently, new in vitro approaches have been developed for DGAT2 screening assays, including the use of basic FlashPlate, which is based on scintillation proximity assay,^14,15 and mass spectrometry (HTMS).¹⁶

Here we describe additional attempts to generate a robust DGAT2 assay amenable to small-molecule screening and detailed enzymatic characterization for development of highly potent DGAT2 inhibitors. Several methods were investigated to find the most suitable for maximizing the identification and validation of a variety of chemical classes in a robust manner, but ultimately the described liquid chromatography/mass spectrometry (LC/MS) method was implemented.

Materials and Methods

Materials

All general chemical reagents were obtained from Sigma-Aldrich (St. Louis, MO) and Fisher Scientific (Pittsburgh, PA). All cell culture reagents were obtained from Invitrogen Corporation (Carlsbad, CA). Diolein (1-2 dioleoyl-sn-glycerol) was purchased from Avanti Lipids (Birmingham, AL). CPM (7-diethylamino- 3-(4′-maleimidylphenyl)-4-methylcoumarin) was purchased from Molecular Probes (Eugene, OR). The internal standard glyceryl tri(oleate-1,2,3,7,8,9,10-¹³C₇) was custom synthesized by Sigma-Aldrich.

Insect Cell Expression and Membrane Preparation

Human DGAT2 was overexpressed in Sf9 insect cells using the Bac-to-Bac Baculovirus Expression System from Life Technologies (Grand Island, NY) according to the manufacturer’s instructions. Sf9 insect cells were maintained in Grace’s insect cell culture medium with 10% heat-inactivated fetal bovine serum, 1% Pluronic F-68, and 0.14 µg/mL Kanamycine sulfate in Erlenmeyer flasks at 28 °C in a shaker incubator. After infection with untagged hDGAT2 baculovirus at multiplicity of infection of 0.1 for 48 h, cells were harvested. Cell pellets were resuspended in buffer containing 10 mM Tris-HCl (pH 7.5), 1 mM EDTA (pH 8.0), 250 mM sucrose, and Complete Protease Inhibitor Cocktail (Roche Diagnostics Corp., Indianapolis, IN) and sonicated on ice. Membrane fractions were isolated by ultracentrifugation (100,000 × g for 20 min), resuspended in the same buffer, and frozen (–80 °C) for later use. The protein concentration was determined with the BCA Protein Assay kit (Pierce Biotechnology Inc., Rockford, IL). Expression of protein levels was analyzed by immunoblotting with goat polyclonal DGAT2 antibody C-15 and donkey anti-goat IgG HRP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) followed by detection with ECL reagent (GE Healthcare, Piscataway, NJ). Human DGAT1, 2-acylglycerol O-acyltransferase 2 (MGAT2), and glycerol-3-phosphate acyltransferase-1 (GPAT1) enzymes were overexpressed in Pichia Pastoris and prepared as microsomal fractions at Banyu Pharmaceuticals Co. Inc. (Japan).

Fluorogenic Acyltransferase Assays

The acyltransferase activities of DGAT1, DGAT2, MGAT2, and GPAT1 were measured in the CPM assay, in which the byproduct (free CoA) generated in these reactions reacts with nonfluorescent CPM to generate a fluorescent adduct.¹⁷ For the GPAT1 assay, compounds were mixed together with its substrates, glycerol-3-phosphate (G-3-P) and oleoyl-CoA, in reaction buffer (50 mM HEPES buffer, pH = 7.5, 45 mM MgCl₂). The reaction was initiated by the addition of enzyme and incubated at room temperature for 60 min. The final concentrations of G-3-P and oleoyl-CoA were 250 µM and 30 µM, respectively. After incubation at room temperature for 90 min, the reaction was then quenched with 90 µM CPM. The plates were read on an EnVision plate reader (Perkin Elmer, Waltham, MA) using top read mode with excitation at 405 nm and emission at 480 nm.

DGAT1, DGAT2, and MGAT2 fluorogenic assays are similar to the GPAT1 assay in terms of assay format and signal detection. Both DGAT1 and DGAT2 assays used diolein and oleoyl-CoA as substrates. Their concentrations were 150 µM/20 µM in the DGAT1 assay and 100 µM/20 µM in the DGAT2 assay, respectively. The DGAT1 enzyme was diluted in buffer containing 100 mM Tris-HCl (pH 7.0), 100 mM sucrose, 100 mM MgCl₂, and 0.01 mg/mL of N-ethylmaleimide–treated bovine serum albumin at a final concentration 0.75 µg/mL. DGAT2 enzyme was diluted in 100 mM Tris-HCl (pH 7.0) and 5 mM MgCl₂ at 4.5 µg/mL. The reaction time was 60 min for both assays. The reaction mixture of MGAT2 assay contained 100 mM Tris-HCl (pH 7.0), 5 mM MgCl₂, 100 µM 2-monoolein, 30 µM oleoyl-CoA, 500 µM 1,2-Diacyl-sn-glycero-3-phosphocholine, and 0.7 µg/mL of MGAT2 membranes and was incubated for 90 min at room temperature before quenching.

DGAT2 Enzymatic Activity Assay

DGAT2 activity was determined by measuring the amount of enzymatic product triolein (1, 2, 3-Tri-(cis-9-octadecenoyl)-glycerol) using the membrane preparation mentioned above. The assay was carried out in deep-well 384 plates in a final volume of 40 µL at room temperature. The assay mixture contained the following: assay buffer (100 mM Tris-HCl, pH 7.0, 20 mM MgCl₂, 5% ethanol), 25 µM of diolein, 10 µM of oleoyl-CoA, and 10 ng/µL of DGAT2 membranes. The reaction was incubated for 90 min at room temperature before being quenched by 10% formic acid in isopropanol. An additional 75 µL of pentanol was added to the quenched reaction to perform liquid-liquid extraction. The plates were then sealed, vortexed, and centrifuged. Extracted triolein from the top layer was then analyzed by LC/MS/MS. The K_m values for the substrate were determined using the Michaelis-Menten equation. A JANUS automated workstation (Perkin Elmer) was used to add compounds, substrates, and enzymes. The Matrix Wellmate (Thermo Scientific, ON, Canada) was used to add the quenching solution and pentanol.

LC/MS/MS Method for Determining DGAT2 Enzyme Activity

LC/MS/MS analyses were performed using Thermo Fisher’s LX4-TSQ Vantage system. This system consists of four Agilent binary high-performance liquid chromatography pumps and a TSQ Vantage triple quadrupole MS/MS instrument. For each sample, 2 µL from the top organic layer of in-plate liquid-liquid extraction was injected onto a Thermo Betabasic C4 column (2.1 mm × 20 mm, 5 µm particle size). The samples were then eluted using the following isocratic conditions: mobile phase: methanol/water with 0.1% ammonium formate = 92/8 (v/v), flow rate: 1 mL/min, temperature: 25 °C. Data were acquired in positive mode using a HESI interface. The operational parameters for the TSQ Vantage MS/MS instrument were a spray voltage of 3000 V, capillary temperature of 280 °C, vaporizer temperature 400 °C, shealth gas 60 arbitrary units, Aux gas 40 arbitrary units, S-lens 113, and collision gas 1.2 mTorr. Selected reaction monitoring (SRM) chromatograms of triolein (Q1: 902.9 > Q3:603.5) and internal standard (Q1: 923.9 > Q3:617.5) were collected for 40 s. The peak area was integrated by Xcalibur Quan software. The ratios between the ¹²C-triolein generated in the reaction and ¹³C₂₁ spiked internal standard were used to generate percentage inhibition and IC₅₀ values. Compound percentage inhibition was calculated by the following formula: Inhibition % = 1 – [(compound response – low control)/(high control – low control)] × 100%. Potent compounds were titrated, and IC₅₀’s were calculated by a four-parameter sigmoidal curve-fitting formula.

DGAT2 Cell-Based Assay in Mouse Primary Hepatocytes

C57/BL6 mouse liver was perfused with EGTA/collagenase solution followed by excision of the tissue and further cell digestion. The cells were spun and washed twice with the complete cell medium (DMEM [high glucose]; Life Technologies)/10% fetal bovine serum (HyClone)/1 mM sodium pyruvate/100 nM insulin/100 nM dexamethasone. The cells were seeded in the same medium at 30,000 cells/well into a 96-well rat collagen I–treated plates (BD BioCoat) for overnight recovery at 37 °C with 5% CO₂. The overnight medium was aspirated and replaced with 100 µL of complete growth medium containing 0.4 mM oleic acid plus compound followed by 24 h treatment at 37 °C with 5% CO₂. The cell medium was collected, and dislodged cells were removed by passage through a filter plate (Millipore). A total of 20 µL of medium was transferred to 384 shallow-well plates (Abgene), followed by addition of 5 µL of 1 µM glyceryl tri(oleate-1,2,3,7,8,9,10-13C7) internal standard in isopropanol and 35 µL of 1-pentanol. After removing the media, cells were lysed in 100 µL of isopropanol, and 5 µL of the cell lysate was transferred to 384 deep-well plates (Abgene), followed by addition of 5 µL of 2 µM glyceryl tri(oleate-1,2,3,7,8,9,10-¹³C₇) internal standard in isopropanol, 40 µL of phosphate-buffered saline, and 60 µL of 1-pentanol. Plates were heat sealed with pierceable aluminum foil, vigorously vortexed, and subsequently stored at −20 °C overnight for phase separation prior to LC/MS analysis as described below.

LC/MS/MS Method for Determining Cell-Based Activity

Samples of medium and cell lysate were analyzed on a Xevo TQ triple quadrupole mass spectrometer equipped with an Acquity UPLC binary pump, column heater, and a 2777 autosampler (Waters, Milford, MA). Chromatographic separation was carried out using a 2.1 mm × 50 mm, 1.9 µm Hypersil GOLD C18 column (Thermo Scientific) maintained at 60 °C. A binary solvent system composed of 10 mM ammonium formate in 40% H₂O:60% acetonitrile (v/v, eluent A) and 90% isopropyl alcohol:10% acetonitrile (v/v, eluent B) was used for the separations. The column was initially conditioned with 90% solvent A. Immediately following injection, the solvent composition was ramped to 95% solvent B over 2.5 min. At 2.6 min, the solvent composition was returned to 90% solvent A and held for 0.4 min before the next injection. The flow rate was held constant throughout the run at 400 µL/min. A total of 5 µL of the upper pentanol phase was directly injected from the plates by setting an appropriate depth for the autosampler syringe. SRM transitions for ¹²C triolein and the ¹³C₂₁ glyceryl trioleate internal standard were acquired as described above.

Results

Expression DGAT2 in Sf9 Insect Cell

We expressed DGAT2 in a baculovirus system to obtain sufficient quantities of enzyme and to avoid the endogenous mammalian DGAT2 activity. The expression of DGAT2 was visualized by Western blot analysis, as shown in Figure 1A (right panel) using anti-DGAT2 antibody C-15 from Santa Cruz Biotechnology. A corresponding band of ~40 kDa was also observed in the companion Ponceau staining picture ( Fig. 1A , left panel). Difficulty in enriching DGAT2 led to the use of an unpurified membrane preparation as our enzyme source.

Figure 1.

Expression of DGAT2 in Sf9 cells and DGAT2 CPM assay. (A) Cells were harvested 48 or 72 h postinfection, and membranes were generated and analyzed by gel electrophoresis and Western blot. Left panel shows a Ponceau staining picture of different DGAT2 membrane preparations. The right panel shows the Western blot analysis of DGAT2 membrane preparations after expression in Sf9 cells using anti-DGAT2 polyclonal antibody C-15 from Santa Cruz Biotechnology. Lane 1: membranes prepared after 48 h infection. Lane 2: membranes prepared after 72 h. Lane 3: mock Sf9 cells as negative control. Lanes 4–6: repeat of 1–3 on a different preparation. (B) Schematic representation of DGAT2 CPM assay. This assay detects CoA, a by-product generated by DGAT2 using oleoyl-CoA and diacylglycerol as substrates. The free thiol of CoA can react with 7-diethylamino-3-(4-maleimidylphenyl)-4-methylcoumarin (CPM), a profluorescent coumarin maleimide derivative that becomes fluorescent upon reaction with thiols. (C) Enzyme titration curve. The experiment was performed with 100 µM diolein and 20 µM oleoyl-CoA at room temperature for 60 min. After addition of 30 µM CPM and incubation for 30 min at room temperature, plates were read using an Envision (Perkin Elmer) plate reader (450 ex, 460 em). Each point on the plot represents mean ± SD for n = 3 replicates. The assay window is only about twofold, mostly because of the very high background generated by Sf9 mock membranes.

Fluorogenic DGAT2 Activity Assay

Initially, we attempted to develop a fluorogenic-based DGAT2 activity assay, a thiol quantification method, as established for other acyl transferases such as DGAT1.^18,19 In this approach, the reaction is monitored via the free thiol group from the enzymatic by-product of free CoA. The mechanism is illustrated in Figure 1B . Using this assay, we were able to detect DGAT2 activity in the DGAT2-expressing membranes. However, the window for this assay was very low, as shown in Figure 1C . The small observed window is partly due to the low content of active DGAT2 and to the high background of mock membranes. We believe the high background detected in mock membranes is due to nonspecific acyltransferases or acyl-CoA hydroylase present in the Sf9 membranes, which are also capable of producing CoA using fatty-acid CoA as substrate.²⁰

LC/MS-Based DGAT2 Activity Assay Development and Validation

We developed an LC/MS based assay to avoid the high background generated using crude membrane preparations and to detect the triolein product directly. The triple quadrupole mass spectrometer was used to discriminate triolein with contaminated mass species with same m/z ratio by performing a MS/MS step. Fragmentation upon collision of triolein with inert argon gas generated the product ion specific to triolein. The two levels of selection conferred greater selectivity and sensitivity for triolein signal detection, therefore reducing the interference from coeluting background and increasing the signal-to-noise ratio. A stable isotopically labeled triolein (glyceryl tri[oleate-1,2,3,7,8,9,10-¹³C₇]) was used to normalize the MS signal and minimize any experimental variation. ¹³C-labeled triolein and unlabeled ¹²C triolein have identical physical and chemical properties, have the same extraction efficiency and co-eluted, and are ionized similarly, thus reducing the data variation.

Enzyme kinetic determinations were performed by titrating the two substrates, diolein and oleyl-CoA. As shown in Figure 2A , enzymatic activity increases with increasing concentrations of substrates and reaches saturation at about 20 µM of oleoyl-CoA and 100 µM of diolein. The apparent K_m calculated values are 6 µM for oleoyl-CoA and 25 µM for diolein. In consequent experiments, substrates were kept close to the apparent K_m values to ensure that DGAT2 enzyme is assayed in the linear range of activity.

Figure 2.

Enzymatic characterization of DGAT2 membrane by liquid chromatography/mass spectrometry assay. (A) Determination of apparent K_m values for DGAT2 substrates. Each substrate concentration dependency was determined at saturation levels of the partner substrate. (B) Membrane titration. Aliquots of the indicated membranes were incubated with 25 µM diolein and 10 µM oleoyl-CoA for 60 min. (C) Inhibition of DGAT2 activity by hexadecyl-CoA. The graph represents a typical dose-response inhibition curve using hexadecyl-CoA, a palmitoyl-CoA nonhydrolyzable analog. Aliquots of DGAT2 membranes were incubated with 25 µM diolein and 10 µM oleoyl-CoA for 60 min in the presence of the indicated concentrations of hexadecyl-CoA. Each point was expressed as the percentage inhibition and is an average of triplicate determinations.

Membrane titration and time courses were also studied by incubating DGAT2 membranes at different concentrations with substrates and stopping the reactions at different times by the addition of isopropanol containing 10% formic acid. Quantitation of peak areas showed a linear conversion of substrate to product through a 90 min incubation (data not shown). In reactions carried out at various concentrations of enzyme for 60 min, the linearity of DGAT2 activity was observed up to approximately 20 ng/µL ( Fig. 2B ). Based on these results, screening assays were set at 10ng/μL DGAT2 membranes, and the reaction time was set for 1 h, both within the linear range of the response curve. The substrate concentrations were used at 25 µM for diolein and 10 µM for oleoyl-CoA, close to their apparent K_m values. Under these conditions, the LC/MS assay has an increased assay window (compared with control membranes) of greater than sevenfold. DMSO up to 5% and ethanol up to 10% had no significant effect on the activity (data not shown). Hexadecyl-CoA, a palmitoyl-CoA nonhydrolyzable analog, was used as a reference compound to test the effects of DGAT2 inhibitors using this novel assay. As shown in Figure 2C , hexadecyl-CoA inhibits DGAT2 activity with an IC₅₀ of approximately 2 µM.

Library Screening with LC/MS-Based Assay

To determine the screening parameters for this assay, the following controls were included for each plate: (1) DGAT2 membranes/DMSO as high control, (2) DMSO/mock membranes or hexadecyl-CoA at 100 µM as low control, and (3) hexadecyl-CoA at different concentrations for the inhibition standard curve. The signal/noise ratio (the fold difference between the low and high controls) was determined to be approximately eightfold with a percentage coefficient of variation of 10%. Figure 3A shows a scatterplot of a representative assay plate in which the high control for this plate was 2.5% DMSO and low control was hexadecyl-CoA at 100 µM. Dose responses of hexadecyl-CoA are also shown in the left part of the scatterplot. The Z′ for this representative plate is 0.75, demonstrating reasonable assay performance. Figure 3B shows the data correlation between two separate runs on the same set of compounds, having an R² of 0.8, indicating reproducibility of the assay performance. The optimized LC/MS-based assay was used to evaluate a focused library from a Merck compound collection (~19,000 compounds). Compounds were screened at 10 µM using 384-well plates. Figure 3C shows the average Z′ factor for all the plates that went into this assay. The Z′ varies between 0.5 and 0.9, with an average of 0.7, indicating a good-quality assay performance. Hits were identified from this effort as compounds that showed >70% inhibition at 10 µM. The hit rate is approximately 5%. To confirm the validity of the hits, all hit compounds were retested for inhibitory activity. The confirmation rate was about 80%, and the correlation between the primary screen and confirmation is shown in Figure 3D .

Figure 3.

Robustness and reproducibility of DGAT2 liquid chromatography/mass spectrometry assay and screening campaign. (A) A scatterplot of a representative assay plate. The high control for this plate was 2.5% DMSO and low control was hexadecyl-CoA at 100 µM. Dose responses of hexadecyl-CoA are also shown in the left part of the scattering plot. The Z′ factor for this particular plate is 0.75. (B) One random compound plate was tested twice to check reproducibility. The line represents the linear correlation between these two sets of data. The R² for this test was 0.78. (C) Representative Z′ factors for almost 150 plates. The high control of these plates was 2.5% DMSO, and the low control was hexadecyl-CoA at 100 µM. (D) Hits confirmation. All hit compounds were retested in different plates. The confirmation rate was approximately 80%.

Hit Characterization

Additional counter screening assays were used to serve as filters to identify nonspecific inhibitors. A secondary CPM assay using an acyltransferase unrelated to DGAT2 was implemented to rule out any DGAT2 nonspecific inhibition. This target, glycerol-3-phosphate acyltransferase (GPAT1), has no structural homology to DGAT2. The GPAT1 fluorogenic assay was established with good throughput and an assay window of fourfold over mock membranes (data not shown). Compounds that showed nonspecific inhibitory activity against GPAT1 were eliminated, comprising 34% of the hits. Compounds that were not eliminated in this step were retested using fresh made solutions from powders in dose-response curves, not only in the DGAT2 LC/MS assay for confirmation but also in the DGAT1 and MGAT2 CPM assays to study selectivity. In addition, an LC/MS assay to identify covalent binders was performed using myoglobin as a surrogate target, measuring any change in the protein molecular weight due to labeling. The majority of the hits (87%) did not show any covalent binding to myoglobin. A hit triage diagram is shown in Figure 4 . Through this practice, several classes of compounds with diverse structures were initially identified. In particular, one initial hit compound, compound 1, with an IC₅₀ of 0.54 µM inhibiting DGAT2 and no selectivity against MGAT2, was chosen for optimization and exploration of SAR in an attempt to design analogs that display enhanced potency and selectivity.

Figure 4.

Schematic diagram of the high-throughput screening work flow. The process of the high-throughput screening for identification of active DGAT2 inhibitors is shown. The liquid chromatography/mass spectrometry (LS/MS) DGAT2 assay was used to screen approximately 19,000 compounds. Active compounds from the primary screen (5% of the initial compounds) were cherry-picked from mother plates and GPAT1 inhibitory potency was tested in an eight concentration dose-response. In this step, 34% of the initial hits were eliminated. The remaining hits were then tested as fresh powders for activity in DGAT2 LC/S, and DGAT1 and MGAT2 CPM assays. After this step, several actives were selected and clustered into structural groups.

Lead Characterization

The hit-to-lead optimization strategy led to the discovery of compound 15, with an IC₅₀ of 3 nM, and compound 16, with an IC₅₀ of 2 nM ( Fig. 5A ). Moreover, both lead compounds had an excellent selectivity for DGAT2 as no inhibition of GPAT1, DGAT1, and MGAT2 enzymes was observed at concentrations as high as 10 µM (data not shown). Further characterization of these compounds included the development of a semiautomated LC/MS enzymatic reversibility assay. The reversibility of inhibition was determined by measuring the recovery of enzymatic activity after a rapid and large dilution of the enzyme-inhibitor complex. In this assay, compound 16 was incubated with the DGAT2 enzyme at its IC₈₀ or its IC₂₀ concentration before it was diluted to its corresponding IC₂₀ concentration into reaction buffer containing the two DGAT2 substrates to initiate the reaction and monitor the enzymatic activity ( Fig. 5B ). Compound 16 is rapidly reversible as the progress curve (IC₈₀:IC₂₀) is linear with a slope equal to the slope of the control sample (IC₂₀:IC₂₀). As expected, a reduced slope was observed when the concentration of the compound was held constant at its IC₈₀ through the dilution.

Figure 5.

Characterization of compound 16. (A) Dose-response curve of a potent inhibitor (compound 16) in the DGAT2 liquid chromatography/mass spectrometry assay. The IC₅₀ for this compound was approximately 2 nM. (B) Reversibility test of compound 16. Upon dilution, the time course for IC₈₀:IC₂₀ showed a similar profile as the IC₂₀:IC₂₀ curve. Therefore, compound 16 is a reversible inhibitor. (C) Mode-of-action study of compound 16. The IC₅₀ of compound 16 increased with higher concentration of the substrate diolein. Each point on the plot represents mean ± SD for n = 2 replicates. Experiments were repeated three times.

We used enzyme kinetics to investigate the mechanism of action of compound 16. For each concentration of the substrate diolein, the percentage of inhibition was plotted with respect to the tested inhibitor concentration, and inhibition curves were analyzed to generate the corresponding IC₅₀ values. As Figure 5C indicates, there is a right shift in the IC₅₀ values of compound 16 as the concentration of the substrate diolein increases. The results are indicative of competitive inhibition with respect to the substrate diolein. Compound 15 also displayed the kinetic signature of a competitive inhibitor (not shown).

Lastly, both lead compounds, compound 15 and compound 16, were tested in an in vitro cell-based assay, and both compounds showed cellular activity. When primary murine hepatocytes were incubated with different concentrations of compound 15, compound 16, and the CP-346086 inhibitor of the microsomal TG transfer protein (MTP) as reference control,²¹ the TG content in the culture media and in the cells was significantly reduced ( Fig. 6 ). Both DGAT2 lead inhibitors were able to reduce TG secretion in mouse primary hepatocytes. However, the maximal effect is equivalent to only half of the maximum effect observed with the MTP inhibitor. In contrast, the DGAT2-specific inhibitors also reduced intracellular TG content, whereas the treatment with MTP inhibitor resulted in no change. This observation indicates that DGAT2 inhibition in primary mouse hepatocytes affects TG pools bound for secretion or for storage.

Figure 6.

Liquid chromatography/mass spectrometry detection of triolein in mouse primary hepatocytes after compound treatment. The values represent n = 12, Mean/error ± SEM. (A) The microsomal triglyceride transfer protein (MTP) compound exhibits complete inhibition of triglyceride secretion into the extracellular medium. Treatment of cells with specific inhibitors of DGAT2 resulted in approximately 50% reduction in triglyceride secretion. (B) Treatment with the MTP inhibitor resulted in almost no change (a slight increase) in intracellular triglycerides (consistent with mechanism), whereas treatment with specific inhibitors of DGAT2 resulted in a 20% reduction.

Discussion

Acyltransferases are critical enzymes in the de novo lipid synthesis pathway. Inhibition of these enzymes has been described in the literature as having potentially desirable phenotypes when their activities are inhibited by genetic manipulation. One such enzyme, DGAT2, is involved in TG synthesis. However, highly potent and selective small-molecule inhibitors of this target suitable for subsequent in vivo studies have not been reported.

A challenge for this target is the ability to generate a form of the protein suitable for classic HTS formats. DGAT2 is an integral membrane protein, rendering it to be used as a crude membrane preparation in order to maintain enzymatic activity. This has the consequence of generating high assay backgrounds, especially in detection formats that cannot discriminate between targeted and nontargeted activity. Here we report on the development of a novel, high-throughput LC/MS assay and hit characterization approach that fully enabled the development of the most potent DGAT2 inhibitors reported to date. Other assay formats to identify DGAT2 inhibitors have been described in the literature with similar or better throughput, but in our hands, the LC/MS-based approach to screen for and characterize DGAT2 inhibitors was deemed the most suitable to implement based on assay performance. By leveraging the throughput of the staggered gradient quad-pump LC/MS approach, sufficient throughput was achieved to support the screening campaign of a 19,000 compound library. An average Z′ of 0.7 was maintained during the screening, and confirmation of hits from the primary campaign was impressive at >80%. A cutoff of 70% inhibition at 10 µM was used, resulting in a hit rate of 5%. To verify specificity and selectivity of enzymatic inhibition, we developed a data-driven hit-triage approach, putting confirmed hits through multiple orthogonal counterscreens using a fluorogenic assay. In addition, covalent modifiers were eliminated using an LC/MS assay with myoglobin as a surrogate. Finally, dose dependence with Hill slope inspection (<1.5) led to the selection of compounds for more in-depth enzymatic characterization by the novel high-throughput LC/MS assay. For example, we developed a novel semiautomated approach to conduct enzymatic reversibility studies by LC/MS. Hundreds of compounds across multiple chemical series were investigated using the jump-dilution method, in which legitimate series were validated as being fully reversible. Finally, detailed characterization of the compound mode of inhibition was carried out with the LC/MS approach, and lead compounds validated as described above were found to be competitive with the substrate diolein. The selectivity profile was excellent, being inactive against DGAT1, MGAT2, and GPAT1. The throughput of the assay was sufficient to support SAR by inventory and substructure searching of the larger compound collection, resulting in the acquisition of thousands of dose-response curves in the postscreen phase. The campaign was highly successful, enabled by the high fidelity of the primary LC/MS assay, yielding eight distinct, selective, and enzymatically reversible chemical series from a relatively small library of 19,000 compounds. Ultimately, extremely potent and selective inhibitors of DGAT2 were identified, which demonstrate the utility of the overall approach. Less robust assays would have limited our ability to generate these validated chemical series in an efficient and quantitative manner. Although the development and execution of the primary screen was judged to be successful because of the identification of validated leads, if desired, the ability to assay larger compound collections is possible. For instance, the use of self-deconvoluting mixtures could extend the reach of the LC/MS approach to sample collections greater than a million compounds, without increasing total compound load on the target beyond the 30 to 50 µM range.

In conclusion, the application of LC/MS to difficult targets is a viable option in the hunt for small-molecule pharmacological agents. Our lead optimization strategy successfully identified a specific series of compounds more potent than any other reported DGAT2 inhibitors. Based on the fact that our lead compounds show on-target activity in murine hepatocytes as well, in vivo efficacy studies were performed and will be the subject of a subsequent publication.

Footnotes

Acknowledgements

The authors thank Drs. Thomas P. Roddy, Mike Kavana, Yusheng Xiong, and Brian Hubbard for their critical scientific discussions and helpful advice during the course of these studies.

Supplementary material for this article is available on the Journal of Biomolecular Screening Web site at .

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Bell

R. M.

Coleman

R. A.

Enzymes of Glycerolipid Synthesis in Eukaryotes. Annu. Rev. Biochem. 1980, 49, 459–487.

Yen

C. L.

Stone

S. J.

Koliwad

. Thematic Review Series: Glycerolipids. DGAT Enzymes and Triacylglycerol Biosynthesis. J. Lipid Res. 2008, 49, 2283–2301.

Liu

Siloto

R. M.

Lehner

. Acyl-CoA:Diacylglycerol Acyltransferase: Molecular Biology, Biochemistry and Biotechnology. Prog. Lipid Res. 2012, 51, 350–377.

Cases

Stone

S. J.

Zhou

. Cloning of DGAT2, a Second Mammalian Diacylglycerol Acyltransferase, and Related Family Members. J. Biol. Chem. 2001, 276, 38870–38876.

Cases

Smith

S. J.

Zheng

Y. W.

. Identification of a Gene Encoding an Acyl CoA:Diacylglycerol Acyltransferase, a Key Enzyme in Triacylglycerol Synthesis. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 13018–13023.

Buhman

K. K.

Smith

S. J.

Stone

S. J.

. DGAT1 Is Not Essential for Intestinal Triacylglycerol Absorption or Chylomicron Synthesis. J. Biol. Chem. 2002, 277, 25474–25479.

Stone

S. J.

Myers

H. M.

Watkins

S. M.

. Lipopenia and Skin Barrier Abnormalities in DGAT2-Deficient Mice. J. Biol. Chem. 2004, 279, 11767–11776.

Chen

H. C.

Smith

S. J.

Ladha

. Increased Insulin and Leptin Sensitivity in Mice Lacking Acyl CoA:Diacylglycerol Acyltransferase 1. J. Clin. Invest. 2002, 109, 1049–1055.

Liu

Millar

J. S.

Cromley

D. A.

. Knockdown of Acyl-CoA:Diacylglycerol Acyltransferase 2 with Antisense Oligonucleotide Reduces VLDL TG and ApoB Secretion in Mice. Biochim. Biophys. Acta. 2008, 1781, 97–104.

10.

Choi

C. S.

Savage

D. B.

Kulkarni

. Suppression of Diacylglycerol Acyltransferase-2 (DGAT2), but Not DGAT1, with Antisense Oligonucleotides Reverses Diet-Induced Hepatic Steatosis and Insulin Resistance. J. Biol. Chem. 2007, 282, 22678–22688.

11.

X. X.

Murray

S. F.

Watts

. Reduction of JNK1 Expression with Antisense Oligonucleotide Improves Adiposity in Obese Mice. Am. J. Physiol. Endocrinol. Metab. 2008, 295, E436–E445.

12.

X. X.

Murray

S. F.

Pandey

S. K.

. Antisense Oligonucleotide Reduction of DGAT2 Expression Improves Hepatic Steatosis and Hyperlipidemia in Obese Mice. Hepatology 2005, 42, 362–371.

13.

Seethala

Peterson

Dong

. A Simple Homogeneous Scintillation Proximity Assay for Acyl-Coenzyme A:Diacylglycerol Acyltransferase. Anal. Biochem. 2008, 383, 144–150.

14.

Kim

M. O.

Lee

S. U.

Lee

H. J.

. Identification and Validation of a Selective Small Molecule Inhibitor Targeting the Diacylglycerol Acyltransferase 2 Activity. Biol. Pharm. Bull. 2013, 36, 1167–1173.

15.

Kim

M. O.

Lee

Choi

. Discovery of a Novel Class of Diacylglycerol Acyltransferase 2 Inhibitors with a 1H-pyrrolo[2,3-b]pyridine Core. Biol. Pharm. Bull. 2014, 37, 1655–1660.

16.

Lang

Geisler

J. G.

. The Use of Stable Isotope-Labeled Glycerol and Oleic Acid to Differentiate the Hepatic Functions of DGAT1 and -2. J. Lipid Res. 2012, 53, 1106–1116.

17.

Chung

C. C.

Ohwaki

Schneeweis

J. E.

. A Fluorescence-Based Thiol Quantification Assay for Ultra-High-Throughput Screening for Inhibitors of Coenzyme A Production. Assay Drug Dev. Technol. 2008, 6, 361–374.

18.

Cao

Zhou

Peng

. Targeting Acyl-CoA:Diacylglycerol Acyltransferase 1 (DGAT1) with Small Molecule Inhibitors for the Treatment of Metabolic Diseases. J. Biol. Chem. 2012, 286, 41838–41851.

19.

Liu

Gorski

J. N.

Gold

S. J.

. Pharmacological Inhibition of Diacylglycerol Acyltransferase 1 Reduces Body Weight and Modulates Gut Peptide Release—Potential Insight into Mechanism of Action. Obesity (Silver Spring). 2013, 21, 1406–1415.

20.

Heckman

C. A.

Friedman

S. J.

Skehan

P. J.

. Localization and Partial Characterization of Acyltransferases Present during Rapid Membrane Formation in the Drosophila melanogaster Embryo. Dev. Biol. 1977, 55, 9–32.

21.

Chandler

C. E.

Wilder

D. E.

Pettini

J. L.

. CP-346086: An MTP Inhibitor That Lowers Plasma Cholesterol and Triglycerides in Experimental Animals and in Humans. J. Lipid Res. 2003, 44, 1887–1901.