Genetic ablation or chemical inhibition of phosphatidylcholine transfer protein attenuates diet-induced hepatic glucose production


  • Potential conflict of interest: Nothing to report.

  • Supported by National Institutes of Health grants to D.E.C. (DK56626 and DK48873) and R.G.-J. (DK45024), by Partners Healthcare and by an unrestricted gift from Johnson and Johnson. This work was also conducted with support from the Harvard Catalyst, The Harvard Clinical and Translational Science Center (NIH Award #UL1 RR 025758 and financial contributions from Harvard University and its affiliated academic health care centers) and the Harvard Digestive Diseases Center (P30 DK34854). J.M.S. was supported by a National Institutes of Health T32 grant (DK07477). E.F.S. was the recipient of an American Liver Foundation Irwin M. Arias Postdoctoral Research Fellowship Award.


Phosphatidylcholine transfer protein (PC-TP, synonym StARD2) is a highly specific intracellular lipid binding protein that is enriched in liver. Coding region polymorphisms in both humans and mice appear to confer protection against measures of insulin resistance. The current study was designed to test the hypotheses that Pctp−/− mice are protected against diet-induced increases in hepatic glucose production and that small molecule inhibition of PC-TP recapitulates this phenotype. Pctp−/− and wildtype mice were subjected to high-fat feeding and rates of hepatic glucose production and glucose clearance were quantified by hyperinsulinemic euglycemic clamp studies and pyruvate tolerance tests. These studies revealed that high-fat diet-induced increases in hepatic glucose production were markedly attenuated in Pctp−/− mice. Small molecule inhibitors of PC-TP were synthesized and their potencies, as well as mechanism of inhibition, were characterized in vitro. An optimized inhibitor was administered to high-fat-fed mice and used to explore effects on insulin signaling in cell culture systems. Small molecule inhibitors bound PC-TP, displaced phosphatidylcholines from the lipid binding site, and increased the thermal stability of the protein. Administration of the optimized inhibitor to wildtype mice attenuated hepatic glucose production associated with high-fat feeding, but had no activity in Pctp−/− mice. Indicative of a mechanism for reducing glucose intolerance that is distinct from commonly utilized insulin-sensitizing agents, the inhibitor promoted insulin-independent phosphorylation of key insulin signaling molecules. Conclusion: These findings suggest PC-TP inhibition as a novel therapeutic strategy in the management of hepatic insulin resistance. (HEPATOLOGY 2011;)

Phosphatidylcholine transfer protein (PC-TP) is a soluble, highly specific lipid binding protein with accentuated expression in the liver.1 PC-TP was identified, purified, and named based on its capacity to catalyze the intermembrane exchange of phosphatidylcholines in vitro.2 According to its characteristic lipid binding pocket that accommodates a single phosphatidylcholine molecule,3 PC-TP has been designated StARD2 within the steroidogenic acute regulatory protein-related transfer (START) domain superfamily.3, 4 Specificity for binding phosphatidylcholines is conferred by a uniquely structured phosphorylcholine head group binding site within the lipid binding pocket of PC-TP.3 Interestingly, evidence for a biological role for PC-TP in lipid transport in vivo is generally lacking,5 and we have instead reported the unanticipated finding that PC-TP regulates glucose metabolism6: Livers of chow-fed Pctp−/− mice exhibit increased insulin sensitivity, as evidenced by profound decreases in hepatic glucose production under hyperinsulinemic euglycemic clamp conditions.6 Although the molecular mechanism is not fully understood, PC-TP may control hepatic glucose homeostasis in response to variations in the fatty acyl composition of membrane phospholipids.5

Type 2 diabetes is characterized by excess hepatic glucose production due to insulin resistance, commonly in the setting of obesity.7 As evidenced by coding region polymorphisms in both humans and mice, PC-TP may play a pathogenic role. A Glu10Ala substitution in human subjects in the Quebec Family Study was correlated with a 3-fold lower risk of the atherogenic small dense low-density lipoprotein (LDL) phenotype,8 which is commonly associated with insulin resistance.9 In New Zealand Obese (NZO) mice, an Arg120His substitution in PC-TP appeared to play a protective role against the development of obesity-associated type 2 diabetes.5, 10 The current study was designed to provide a direct test of whether Pctp−/− mice are resistant to high-fat diet-induced increases in hepatic glucose metabolism and whether small molecule inhibition of PC-TP would recapitulate this phenotype.


ALT, alanine aminotransferase; AMPK, AMP-activated protein kinase; AUC, area under the curve; Clint, intrinsic clearance; Cmax, maximum plasma concentration; GSK3β, glycogen synthase kinase 3β; HEK, human embryonic kidney; Krel, relative affinity; NEFA, nonesterified fatty acids; NZO, New Zealand Obese; PA, phosphatidic acid; TCNQ, 7,7,8,8-tetracyanoquinodimethane; PC-TP, phosphatidylcholine transfer protein; PPARγ, peroxisome proliferator-activated receptor γ; Pyr-PC, 1-hexadecanoyl-2-(1-pyrenedecanoyl)-sn-glycero-3-phosphocholine; S6K, p70 ribosomal S6 kinase; START, steroidogenic acute regulatory protein-related transfer; Tm, melting temperature; tmax, time to maximum concentration.

Materials and Methods

Mice, Diet, and Treatments.

Pctp−/− and wildtype control mice seven generations backcrossed into FVB/NJ genetic background11 were housed in a standard 12-hour alternate light/dark cycle facility and fed a standard rodent diet 5001 (LabDiets, St. Louis, MO) with free access to drinking water. Protocols for animal use and euthanasia were approved by the institutional committees of the Harvard Medical School and the Albert Einstein College of Medicine. Experiments were conducted using male mice. Starting at 4-5 weeks of age, mice were fed a high-fat diet (60% kcal; D12492; Research Diets, New Brunswick, NJ) for periods ranging from 8 to 18 weeks prior to commencing experiments. Mice were weighed weekly and rates of food consumption were calculated per mouse from the weight of food withdrawn by all of the mice in the cage each week. Prior to selected experiments, mice were anesthetized by intraperitoneal (i.p.) injection of ketamine (87 mg/kg body weight [b.w.]) plus xylazine (13 mg/kg b.w.) (Webster Veterinary, Sterling, MA). In experiments designed to test the influence of PC-TP inhibition on glucose homeostasis, administration of compound A1 (see Supporting Information: Materials and Methods, for synthesis and assays of microsomal stability and pharmacokinetics) or vehicle was initiated concurrently with the high-fat diet. Compound A1 was prepared to a final concentration of 0.6 mg/mL in 4% dimethyl sulfoxide (DMSO) and 96% of 6% hydroxypropyl-β-cyclodextrin (Sigma Aldrich, St. Louis, MO) solution in sterile water. Mice were injected i.p. 5 days per week with 3 mg/kg compound A1 or the equivalent volume of vehicle (5 μL/g). For all experiments, mice were sacrificed after an overnight fast.

Analytical Techniques.

Plasma nonesterified fatty acid (NEFA), triglyceride, cholesterol, and phospholipid concentrations were determined using reagent kits from Wako (Richmond, VA), Sigma Aldrich, and Roche Diagnostics (Indianapolis, IN), respectively. Blood glucose was determined using a OneTouch Ultra glucose monitor (LifeScan, Milpitas, CA). Hepatic concentrations of triglycerides and cholesterol were measured enzymatically following hepatic lipid extraction.12 Plasma insulin, leptin, and adiponectin were determined by enzyme-linked immunosorbent assay (ELISA) as a service of the Joslin Diabetes and Endocrinology Research Center Specialized Assay Core (NIH 5P30 DK36836, Joslin Diabetes Center, Boston, MA). Plasma activities of alanine aminotransferase (ALT) and concentrations of bilirubin were determined using standard assays by Charles River Research Animal Diagnostic Services (Wilmington, MA). To assess protein expression, liver tissue or cell lysates were homogenized in RIPA buffer supplemented with protease and phosphatase inhibitors (Roche Diagnostics). Lysates were rotated slowly at 4°C for 30 minutes and then centrifuged at 12,000g for 10 minutes to remove cellular debris. Proteins were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subjected to immunoblot analysis. An anti-PC-TP rabbit polyclonal antibody was as described.13 Antibodies against pAkt(S473)/total Akt, pS6K(T389)/total S6K, pGSK3β(S9), pAMPK(T172)/total AMPK were from Cell Signaling Technology (Beverly, MA). An antibody against total GSK3β was from BD Transduction Laboratories (Woburn, MA), and an antibody against β-actin was from Sigma Aldrich. Detection was by enhanced chemiluminescence using a Western Lightning chemiluminescence reagent (PerkinElmer, Waltham, MA). For histologic analysis, sections of liver were fixed in 10% formalin. Samples were processed, sectioned, and stained by hematoxylin and eosin as a service of the Dana-Farber/Harvard Cancer Center Rodent Histopathology Core (supported in part by an NCI Cancer Center Support Grant NIH 5P30 CA06516).

Magnetic Resonance Imaging of Body Fat.

Images of mice were acquired with a GE/Omega 9.4 T vertical wide-bore spectrometer operating at a 1H frequency of 400 MHz and equipped with 50-mm shielded gradients (General Electric, Fremont, CA) and a 40-mm 1H imaging coil (RF Sensors, New York, NY). The following parameters were used to obtain transverse and sagittal images: echo time, 30 msec; field of view, 51.2 mm; number of averages, 2; slice thickness, 3 mm; gap 0.5 mm, repetition time, 0.4 sec; matrix size, 128 × 256 (interpolated to 256 × 256). Water images were acquired with the water peak on resonance, fat images were acquired with the lipid CH2 peak on resonance.14 Imaging data were analyzed using MatLab-based software (MathWorks, Natick, MA). Spectroscopy studies were conducted prior to imaging using the same coil and a routine pulse-acquire sequence (four signal averages, 5-second delay between scans). The areas under the fat and water peaks were integrated using the spectrometer software and were used to calculate the %fat/water and %fat mass.

Pyruvate and Glucose Tolerance Tests.

Pyruvate and glucose tolerance tests were performed after an overnight fast.15-17 Plasma glucose was measured at baseline. This was followed by i.p. injection of 2 mg/g b.w. of sodium pyruvate (0.4 g/mL in sterile phosphate-buffered saline [PBS]) for pyruvate tolerance tests or 1 mg/g b.w. D-glucose (20% wt/vol) for glucose tolerance tests. Blood glucose concentrations were measured periodically for up to 180 minutes. For experiments in which both were performed, mice were allowed to recover for 1 week between pyruvate and glucose tolerance tests.

Hyperinsulinemic Euglycemic Clamp Studies.

Studies were performed in conscious, unrestrained mice fitted with intravenous catheters.18 Briefly, food was removed 5 hours prior to commencing the studies, and infusions lasted for 90 minutes. Mice received a constant infusion of high-performance liquid chromatography (HPLC)-purified insulin (3.6 mU/min/kg b.w.) and [3H-3]-glucose (0.1 μCi/min; PerkinElmer). A glucose solution (10% wt/vol) was infused at a variable rate so that plasma glucose concentration was constant at 8 mM throughout the experiment. Plasma glucose concentrations and specific activities were used to determine rates of glucose uptake and hepatic glucose production.6

Structure-Activity Relationships of Small Molecule Inhibitors and Mechanism of PC-TP Inhibition.

As described in the Supporting Information: Materials and Methods, recombinant PC-TP, StARD7, and StARD10 were purified, and small molecule inhibitors were synthesized and used in assays of phosphatidylcholine transfer activity, PC-TP-inhibitor binding by surface plasmon resonance, displacement of pyrene-labeled phosphatidylcholine (Pyr-PC), thermal stability using ThermoFluor and peroxisome proliferator-activated receptor γ (PPARγ) activation.

Phosphorylation of Signaling Proteins in Cultured Cells.

Frozen primary human hepatocytes (CellzDirect/Invitrogen, Carlsbad, CA) were thawed using cryopreserved hepatocyte recovery medium, then plated in serum containing plating media for 6 hours to allow cells to adhere. After overnight starvation in serum-free medium, compounds A1 or B1 dissolved in DMSO was added (0.1% final concentration of DMSO) for 60 minutes. Negative controls included no addition and 0.1% DMSO. The positive control included 0.1% DMSO plus insulin (50 nM) for 30 minutes. Compound A1 was also tested in human embryonic kidney (HEK) 293E cells.19 HEK 293E were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 1% penicillin-streptomycin and then starved overnight. Cells were then exposed to compound A1 for 60 minutes.


Data are reported as means ± standard error of the mean (SEM). Differences between groups were analyzed using a two-tailed unpaired Student's t test.


Attenuation of Diet-Induced Hepatic Glucose Production in High-Fat-Fed Pctp−/− Mice.

In mice fed a high-fat diet, the lack of PC-TP expression did not influence growth or food consumption (Supporting Fig. 1A) and obesity was evidenced by body weights that exceeded chow-fed counterparts6 by up to 40%. After 8 weeks, fasting plasma glucose concentrations (Fig. 1A) were similar to those previously reported for chow-fed mice6 and were similarly decreased in Pctp−/− mice. The high-fat diet for 12 weeks was sufficient in wildtype mice to elevate plasma glucose concentrations, which did not rise in Pctp−/− mice until 18 weeks. As a result, the absence of PC-TP expression was associated with 25%, 46%, and 17% reductions in fasting plasma glucose concentrations at 8, 12, and 18 weeks of high-fat feeding, respectively. Hyperinsulinemic euglycemic clamp studies performed after 18 weeks revealed a 3.6-fold increase in glucose infusion rate in Pctp−/− mice (Fig. 1B). Whereas the rate of glucose uptake was unchanged, hepatic glucose production was decreased by 46%. In order validate an alternative, more facile procedure to determine rates of hepatic glucose production, mice fed the high-fat diet for 12 weeks were subjected to pyruvate tolerance tests.16 Reduced plasma glucose concentrations were observed in Pctp−/− mice during the course of the pyruvate tolerance tests (Fig. 1C), such that the mean values for area under the curve (AUC) were reduced 32% compared with wildtype mice.

Figure 1.

Pctp−/− mice are resistant to diet-induced glucose intolerance. (A) Reduced fasting plasma glucose concentrations in mice lacking PC-TP. Wildtype (n = 6-10) and Pctp−/− (n = 8-10) mice were fed a high-fat (60% kcal) diet for up to 18 weeks. Values for chow-fed wildtype and Pctp−/− mice of similar age to mice fed high-fat for 8 weeks are replotted from the previous study.6 (B) Decreased hepatic glucose production in Pctp−/− mice. Values determined by hyperinsulinemic euglycemic clamp studies for wildtype mice (n = 6) and Pctp−/− mice (n = 8) fed the high-fat diet for 18 weeks. (C) Pyruvate tolerance tests reflect changes in hepatic glucose production. Wildtype (n = 4) and Pctp−/− (n = 5) mice fed high-fat diet for 12 weeks were subjected to pyruvate tolerance tests. The barplot gives mean AUC values. *P < 0.05 versus wildtype mice.

Percentages of body fat, as well as other metrics of body and plasma compositions, were measured after 12 weeks on the high-fat diet. In contrast to our previous findings,6 the percentages of body fat in high-fat-fed mice did not differ between the two genotypes (inset to Supporting Fig. 1A). Moreover, liver and epididymal fat pad weights were similar (Table 1) and both macro- and microvesicular hepatic steatosis were equally present (Supporting Fig. 2). High-fat-fed Pctp−/− mice did not exhibit changes in leptin or adiponectin concentrations or in plasma or hepatic concentrations of insulin, NEFA, triglycerides, cholesterol, and phospholipids (Table 1).

Table 1. Characteristics of Fasting Wildtype and Pctp−/− Mice Following 12 Weeks of High-Fat Feeding*
 (n = 7)(n = 10)(n = 5)(n = 8)(n = 3)(n = 4)
  • *

    Mice were fed a high fat (60% kcal) diet for 12 weeks prior to sacrifice under fasting conditions. Mice treated with compound A1 or vehicle were injected i.p. 5 days per week for 12 weeks, starting when the high fat diet was initiated. Data are presented as mean (SEM).

  • Values for wildtype and Pctp−/− mice were compared using two-tailed unpaired Student's t test.

  • Values for mice treated with compound A1 and vehicle were compared using two-tailed unpaired Student's t test.

  • §

    P < 0.05 versus wildtype mice.

  • ||

    P < 0.05 versus compound A1 treated mice.

  • For untreated mice: wildtype, n = 6; Pctp−/−, n = 8.

  • #

    The phospholipid assay utilized choline oxidase, which detects plasma phosphatidylcholines and sphingomyelins. In a prior study (Wu and Cohen, Am J Physiol 2005;289:G1067-1074), we demonstrated that lack of PC-TP expression in mice did not influence plasma phospholipid concentration or composition, which principally comprises phosphatidylcholines (∼60%) with lesser proportions of sphingomyelins (∼15%).

 Body (g)40.4 (1.3)40.0 (0.7)29.1 (0.4)27.5 (0.6)29.1 (1.3)26.2 (0.6)
 Liver (g)1.48 (0.07)1.65 (0.10)1.28 (0.06)1.25 (0.05)1.37 (0.03)1.13 (0.06)
 Epididymal fat pad (g)1.26 (0.09)0.95 (0.11)0.94 (0.07)1.02 (0.07)1.00 (0.1)0.85 (0.12)
Plasma concentration      
 Glucose (mg/dl)300.6 (26.5)162.5 (10.2)§133.4 (9.9)||92.7 (4.4)110.7 (5.7)97.0 (8.5)
 Insulin (ng/ml)0.23 (0.03)0.24 (0.02)0.15 (0.01)0.15 (0.01)0.14 (0.02)0.14 (0.01)
 Leptin (ng/ml)64.6 (8.7)59.5 (7.2)3.4 (0.8)4.1 (0.8)5.4 (1.9)3 (0.9)
 Adiponectin (μg/ml)14.6 (3.9)8.6 (1.3)8.9 (0.5)9.8 (0.6)8.5 (0.5)8.7 (0.6)
 NEFA (μM)310 (40)428 (61)511 (75)417 (42)358 (117)400 (49)
 Triglycerides (mg/dl)60.7 (4.4)70.5 (7.6)21.9 (6.1)30.6 (5.1)11.8 (2)||34.2 (5.2)
 Cholesterol (mg/dl)140 (10)168 (16)86 (12)110 (11)87 (13)117 (8)
 Phospholipids (mg/dl)#187.6 (12)191.8 (15)120 (27)117.5 (29)97 (44)118.7 (37)
Hepatic concentration      
 Triglycerides (mg/g)97.2 (8.4)92.8 (7.6)59.9 (13.4)||98.3 (6.9)50.2 (5.0)62.7 (12.6)
 Cholesterol (mg/g)2.2 (0.1)2.2 (0.1)2.5 (0.2)||3.0 (0.1)2.6 (0.3)2.8 (0.3)

Identification of an Optimized Small Molecule Inhibitor of PC-TP.

In a high-throughput screening of 114,752 compounds, we previously identified six distinct small molecule inhibitors of the phosphatidylcholine transfer activity of PC-TP.20 To select an optimized molecule for a therapeutic trial in mice, we synthesized structural analogs around the two most potent inhibitors identified in the screen, A1 and B1 (Fig. 2). Structure-activity analyses using a fluorescence quench assay (Supporting Fig. 3) revealed molecular features that influence the median inhibitory concentration (IC50) values (Fig. 2). For the A series, at least one halogen group on the terminal ring at R1 was essential for inhibition. The addition of a methyl substituent to the aryl amide at R3 reduced inhibition more than 30-fold. Finally, the two methyl substituents at R5 were essential for inhibitory activity. For the B series, essential features for inhibition included a sulfur atom at position X, a Ph on the α-carbon of the amide at R2, as well as the nature of substituents on the terminal ring, particularly 3,5-dichloro at R4. Additionally, the introduction of methyl on the amide nitrogen at R3 eliminated inhibition. StARD10 activity was inhibited by selected compounds, but less effectively (Supporting Fig. 3B), with the IC50 values (Fig. 2) ranging from 1.5 to 10-fold greater than for PC-TP. StARD7 was only modestly inhibited by compounds A1 and B1 (IC50 ≈70 μM) and more weakly inhibited by other compounds at higher IC50 values that could not be quantified under conditions of the assay.

Figure 2.

Structure-activity relationships of PC-TP inhibitors. Structures and inhibitory characteristics of small molecules derived from compounds A1 and B1. “NA” denotes that a compound exhibited no activity and “-” denotes that no measurement was performed. Syntheses of small molecular inhibitors are illustrated in Supporting Information.

Compound A1 Binds PC-TP, Displaces Phosphatidylcholine, and Increases Thermal Stability.

We used surface plasmon resonance to demonstrate binding of representative inhibitors directly to PC-TP with KD values in the micromolar range (Fig. 3A). Compound A10, which demonstrated no inhibitory activity (Fig. 2), did not bind PC-TP. Because the parent compounds from each series (i.e., A1 and B1) exhibited both the lowest IC50 values and greatest specificity for PC-TP, these were tested for in vitro microsomal stability in order to determine their potential utilities in vivo. This revealed a 6.5-fold greater metabolic stability of compound A1 (compound A1, half-life (t1/2) = 230 minutes and intrinsic clearance (Clint) = 6.0 μL/min/mg protein; compound B1, t½ = 35.6 minutes and Clint = 39 μL/min/mg protein). Based on this result, we selected compound A1 (LDN-193188) for additional characterization. In a fluorescence competition assay, compound A1 displaced a fluorescent phosphatidylcholine from the lipid binding pocket of PC-TP (Fig. 3B). The Krel value for compound A1 was 20% of the Krel for phosphatidylcholine (inset to Fig. 3B). Finally, we demonstrated that compound A1 increased the thermal stability of PC-TP (Fig. 3C): The Tm of PC-TP, which was markedly increased when PC-TP was bound by phosphatidylcholine, was further increased when bound by compound A1 (inset to Fig. 3C).

Figure 3.

Compound A1 binds PC-TP, displaces phosphatidylcholine, and increases thermal stability. (A) Binding of compound A1 to PC-TP as assessed by surface plasmon resonance. Upper panel: Compound A1 (50-100 μM) was injected for 150 seconds to allow binding and then dissociation from PC-TP bound to a CM5 chip. Representative sensograms are shown overlaid and zeroed on the y-axis to the average baseline before injection. The injection start time for each sample was set to zero on the x-axis. Lower panel: Selected inhibitors were analyzed by surface plasmon resonance. Rate constants of association (ka) and dissociation (kd) were determined by nonlinear least-squares analysis of the sensogram data (SigmaPlot, Systat Software, San Jose, CA) and used to calculate equilibrium dissociation constants, KD = kd/ka. (B) Displacement of phosphatidylcholine from PC-TP by compound A1. Upper panel: Emission scans of Pyr-PC in the absence of PC-TP or with PC-TP plus compound A1. Lower panel: Relative affinity constants, Krel, were determined according to the equation33 ΔF(392 nM) = −1/Krel·([X]/[Pyr-PC])·ΔF(392 nM) + ΔFmax for X = phosphatidylcholine or compound A1 by linear regression of ΔF plotted against the product of ΔF(392 nM) and the molar ratio [X]/[Pyr-PC]. The inset bar plot displays values of Krel (n = 3-4 determinations) for phosphatidylcholine and compound A1. *P = 0.0001. (C) Stabilization of PC-TP by compound A1 using a thermal shift (ThermoFluor) assay. Fluorescence intensity (upper panel) and first derivative, dF/dT of the melting curves (lower panel) for PC-TP, PC-TP plus phosphatidylcholine and PC-TP with compound A1. Values of Tm were determined using LightCycler Protein Melting software (Roche), which identified Tm as minimum values of −dF/dT when plotted as functions of temperature. The inset bar plot shows the comparison of the Tm values (n = 3 determinations) of PC-TP plus phosphatidylcholine, and PC-TP plus compound A1. *P = 0.02.

Compound A1 Attenuates Diet-Induced Hepatic Glucose Production in Wildtype Mice.

Based on a pharmacokinetic analysis, which revealed a maximum plasma concentration (Cmax) = 54 μM, time to maximum concentration (tmax) = 0.5 h and t1/2 = 19.4 hours following a single 3 mg/kg i.p. dose of compound A1, we designed a dosing schedule in which 3 mg/kg of inhibitor or the equivalent volume of vehicle was administered i.p. daily for 5 days per week for 12 weeks. Mice administered the inhibitor exhibited similar weight gain and food consumption as vehicle-treated mice (Supporting Fig. 1B,C), although neither group gained as much weight as high-fat-fed mice that were not subjected to i.p. injection (Supporting Fig. 1A).

Treatment with compound A1 led to a 31% reduction in fasting plasma glucose concentrations (mg/dL) in wildtype mice, but did not significantly affect Pctp−/− mice (Table 1). The inhibitor improved glucose tolerance tests for wildtype mice (Fig. 4A), as evidenced by separation of the response curves and a 20% reduction in AUC compared with vehicle-treated controls. Pctp−/− mice did not respond to inhibitor treatment (Fig. 4B). Consistent with decreased hepatic glucose production, compound A1 also improved the pyruvate tolerance tests of wildtype but not Pctp−/− mice (Fig. 4C,D), with separation of response curves and a 20% reduction in AUC for wildtype mice, which did not achieve statistical significance. For both glucose and pyruvate tolerance tests, the similar peak glucose concentrations and AUC values in wildtype mice treated with compound A1 and in Pctp−/− mice treated with either compound A1 or vehicle suggests that PC-TP was completely inhibited by compound A1.

Figure 4.

Compound A1 improves glucose homeostasis in high-fat-fed wildtype but not Pctp−/− mice. Improved glucose and pyruvate tolerance tests in wildtype mice treated with compound A1. High-fat-fed wildtype mice and Pctp−/− mice were injected i.p. 5 days per week with 3 mg/kg of compound A1 for 12 weeks. Mice were subjected to glucose tolerance tests (A: wildtype: vehicle n = 4, compound A1 n = 8; B: Pctp−/−: vehicle n = 3, compound A1 n = 4) and pyruvate tolerance tests (C: wildtype: vehicle n = 4, compound A1 n = 7; D: Pctp−/−: vehicle n = 3, compound A1 n = 3). The inset bar plots provide AUC values for wildtype mice and Pctp−/− mice. *P < 0.05 versus vehicle treatment.

Treatment of mice with compound A1 did not alter plasma concentrations of insulin or NEFA (Table 1). Although body composition was not measured directly in these mice, there were no changes in epididymal fat pad weights or plasma concentrations of leptin and adiponectin. In livers of wildtype mice treated with compound A1, we observed increases in triglyceride and cholesterol concentrations together with nonsignificant increases in plasma triglyceride and cholesterol concentrations. Livers of mice treated with either compound A1 or vehicle exhibited microvesicular steatosis, but did not show histologic evidence of toxicity or inflammation (Supporting Fig. 2). Treatment with compound A1 was not associated with ALT elevations (Supporting Fig. 4A), and plasma bilirubin concentrations tended to decline in both genotypes (Supporting Fig. 4B).

Compound A1 Activates Insulin Signaling.

Figure 5A shows insulin-independent increases in the phosphorylation of Akt, p70 ribosomal S6 kinase (S6K), and glycogen synthase kinase 3β (GSK3β) following exposure of cultured primary human hepatocytes or HEK 293E cells to concentrations of compound A1 achieved in vivo. This suggested that compound A1 augments the intrinsic cellular response to insulin signaling due to specific inhibition of PC-TP. Similar effects of compound B1 in primary human hepatocytes (Supporting Fig. 5A) support this assertion, as does increased basal phosphorylation of Akt and S6K in livers of fasted wildtype but not Pctp−/− mice treated with compound A1 (Fig. 5B; Supporting Fig. 5B). The absence of changes in PC-TP expression in cultured cells or in livers of inhibitor treated wildtype mice (Fig. 5A; Supporting Fig. 5) indicates that small molecule inhibition does not reduce PC-TP expression or accelerate its degradation. Finally, we tested compounds A1 and B1 for activation of PPARγ (Supporting Fig. 6), but neither compound exhibited this activity.

Figure 5.

Compound A1 enhances insulin signaling. (A) Compound A1 increases phosphorylation of selected mediators of insulin signaling in cultured cells. Human hepatocytes (left panel) and HEK 293E cells (right panel) were serum-starved overnight, then treated with compound A1 for 1 hour and harvested for immunoblot analyses. Results are representative of two experiments for human hepatocytes and three experiments for HEK 293E cells. (B) Compound A1 increases phosphorylation of Akt and S6K in livers of wildtype mice. Homogenates of liver samples from mice fasted overnight were subjected to immunoblotting and bands (Supporting Fig. 5B) were quantified by densitometry. *P < 0.05 versus vehicle treatment.


Our interest in PC-TP as a therapeutic target was motivated by the unexpected initial finding of increased hepatic insulin sensitivity in chow-fed Pctp−/− mice.6 These mice exhibited reduced fasting plasma glucose concentrations and profound decreases in hepatic glucose production under conditions of a hyperinsulinemic-euglycemic clamp. In addition to increased Akt phosphorylation in cultured primary hepatocytes that lacked PC-TP expression, an increased percentage of body fat in Pctp−/− mice was associated with elevated plasma concentrations of both leptin and adiponectin. These findings suggested two potential mechanisms for increased hepatic insulin sensitivity: intrinsic sensitization of hepatocytes to insulin and adipokine-mediated sensitization of the liver to insulin action. The current study confirms and extends our observations in chow-fed mice by demonstrating that Pctp−/− mice are resistant to diet-induced glucose intolerance, but not to obesity. The high-fat diet eliminated genotype-dependent differences in body composition and adipokines, as well as plasma and hepatic concentrations of NEFA, triglycerides, and cholesterol. Therefore, the persistent decrease in hepatic glucose production was most likely attributable to intrinsic sensitivity of the liver to insulin action in the absence of PC-TP expression. When taken together with genetic evidence that PC-TP polymorphisms are protective against insulin resistance in humans8 and mice,10 these findings prompted us to examine whether pharmacological inhibition of PC-TP would recapitulate the same effects and serve as proof-of-concept for a novel therapeutic modality.

In order to identify an optimized small molecule inhibitor for a therapeutic trial in mice, we subjected the two most potent compounds identified in a small molecule screen20 to systematic structure-function analyses. These identified molecular features required for inhibition, but did not ultimately generate compounds with in vitro potencies beyond those observed for the parent molecules (i.e., compounds A1 and B1). To further characterize the inhibitors, we tested their specificity against other mammalian proteins that exhibit phosphatidylcholine transfer activity in vitro.5 We selected two closely related START domain proteins.21 StARD10 is overexpressed in some primary human breast cancers,22 and like PC-TP, it binds and transfers phosphatidylcholines in vitro.23 StARD10 activity was inhibited by compound A1, but less effectively. StARD7 is a highly expressed protein in a choriocarcinoma cell line24 and exhibits phosphatidylcholine transfer activity.25 It was only modestly inhibited by compounds A1 and B1, although precise IC50 values for these weakly active compounds could not be quantified under conditions of the assay. These findings suggest that small molecule inhibition was at least relatively selective for the structural characteristics of PC-TP.

An important unresolved question from the initial high-throughput screen20 was the mechanism of inhibition. The in vitro activity of PC-TP in the fluorescence quench assay reflects a multistep process20: The protein must first associate with a donor small unilamellar vesicle, exchange a phosphatidylcholine molecule, dissociate from the vesicle, associate with an acceptor small unilamellar vesicle, again exchange a phosphatidylcholine, and then dissociate. Therefore, it was possible that the small molecule inhibitors might not bind directly to PC-TP, but may have instead inserted into the membrane bilayer and disrupted membrane association of the protein. Because such a mechanism would likely reduce the therapeutic potential of an inhibitor in vivo, we explored whether the compounds bound directly to PC-TP. Surface plasmon resonance, which is a sensitive biophysical technique for the measurement of ligand-protein interactions,26 revealed that inhibitors of PC-TP bound the protein with KD values that were in good agreement with respective IC50 values. In further support of an inhibitory mechanism that involved direct binding to the protein, compound A1 displaced a fluorescent phosphatidylcholine analog from the PC-TP lipid binding pocket, displaying similar relative affinity for PC-TP as natural phosphatidylcholines. Compound A1 also increased the thermal stability of PC-TP, providing additional independent evidence for inhibitor-protein binding.

Because of its favorable metabolic and pharmacokinetic characteristics, we selected compound A1 for in vivo testing. Consistent with a PC-TP-dependent mechanism of glucose regulation, the administration of this compound to high-fat-fed mice led to significant reductions in fasting plasma glucose concentrations and improved glucose tolerance tests for wildtype, but not Pctp−/− mice. Due to prohibitive logistical issues, we were unable to perform hyperinsulinemic euglycemic clamp studies to determine whether inhibitor treatment reduced hepatic glucose production. We therefore used pyruvate tolerance tests as a more facile surrogate measure of hepatic glucose production. In a pyruvate tolerance test, the gluconeogenic substrate pyruvate is converted to glucose by the liver and exported into plasma.16 Having validated this approach against hyperinsulinemic euglycemic clamp studies using untreated high-fat-fed mice, pyruvate tolerance tests were utilized to demonstrate a PC-TP-dependent reduction in hepatic glucose production by compound A1.

Inhibitor treatment of wildtype mice was not associated with differences in plasma concentrations of NEFA, leptin, and adiponectin, or in other lipid-related parameters that might influence insulin sensitivity.27 These findings suggested that inhibitor-mediated improvements in glucose homeostasis were attributable to the activity of compound A1 in the liver. This possibility was tested in primary human hepatocytes and HEK 293 cells, both of which exhibited robust expression of PC-TP. Exposure of cells to PC-TP inhibitors resulted in dose-dependent, but insulin-independent activation of Akt, S6K, and GSK3β, which are key effectors of insulin signaling.28 The relevance of these findings in vivo was evidenced by increased basal levels of pAkt and S6K in wildtype, but not Pctp−/− mice treated with compound A1. The possibility that compound A1 enhances insulin signaling is consistent with our recent observations in HEK 293 cells following siRNA-mediated knockdown of PC-TP (Ersoy et al., HEPATOLOGY 2010;52:591, abstract).

Metformin and thiazolidinediones are commonly utilized insulin-sensitizing agents to manage type 2 diabetes. The absence of consistent increases in phosphorylation of AMP-activated protein kinase (AMPK) suggests a distinct activity of compound A1 from metformin.29, 30 Compound A1 also failed to activate PPARγ, arguing against a mechanism in common with thiazolidinediones.7, 31 In addition, the absence of ALT or bilirubin elevations and lack of histologic evidence of liver damage or inflammation indicated that chronic treatment of mice with compound A1 was not associated with hepatotoxicity. The reductions in plasma bilirubin concentrations were observed in both wildtype and Pctp−/− mice, suggesting an additional off-target effect of compound A1. Possible explanations include the induction of bilirubin metabolism or biliary secretion. Because the objective in measuring plasma concentrations was to evaluate potential hepatoxicity of compound A1, we did not pursue a mechanistic explanation in the current study.

An important limitation of this study is that treatment with both compound A1 and vehicle reduced weight gain together with fasting plasma glucose concentrations and degrees of glucose intolerance in both genotypes when compared with untreated mice. Although likely attributable to the frequency of i.p. injections performed during the 12-week treatment period, it is possible that the vehicle, which contained 2-hydroxypropyl-β-cyclodextrin, did exert metabolic effects. This could explain the noticeable difference in distribution of hepatic lipids in treated mice, which exhibited microvesicular hepatic steatosis despite similar tissue lipid concentrations as untreated mice in which we observed both macro- and microvesicular distributions of hepatic fat. Appreciating that the administration of 2-hydroxypropyl-β-cyclodextrin to mice can increase intracellular cholesterol transport,32 further study will be required to ascertain the specific influence of the vehicle on hepatic lipid distribution. We also noted tendencies toward increased hepatic and plasma concentrations of triglycerides and cholesterol in both wildtype and Pctp−/− mice treated with compound A1. The occurrence of these changes independent of PC-TP expression is suggestive of an off-target effect of the small molecule, the mechanism for which is not yet understood.

In summary, this study has served as proof of principle that genetic or chemical targeting of PC-TP in a mouse model can attenuate diet-induced glucose intolerance by sensitizing the liver to insulin action and reducing hepatic glucose production. If small molecule inhibitors of PC-TP prove to be capable of treating established type 2 diabetes, they could represent a novel approach to the management of this common disorder. Moreover, these compounds should be of value in efforts to dissect the molecular mechanisms by which PC-TP regulates hepatic glucose metabolism.


We thank Dr. Ji-Feng Liu at Aberjona Laboratories (Beverly, MA) for synthesizing inhibitor analogs and to Dr. Xin Teng, Brigham and Women's Hospital, for preparing sufficient amounts of compound A1 (LDN-193188) for in vivo studies. The authors also thank Drs. Jorge Plutzky and Gabriela Orasanu for assistance with the experiment to test activation of PPARγ, Drs. Ross Stein, David Brooks, and David Silver for helpful discussions, and Mr. James Macdiarmid for editorial assistance with the article.