The Antipsychotic Dopamine 2 Receptor Antagonist Diphenylbutylpiperidines Improve Glycemia in Experimental Obesity by Inhibiting Succinyl-CoA:3-Ketoacid CoA Transferase

It is imperative that new treatments continue to be developed for type 2 diabetes (T2D), as the majority of individuals with T2D will eventually require multiple therapies to control their glycemia (1,2). Recent observations have demonstrated that high-fat diet (HFD)–induced obesity in mice increases the expression/activity of succinyl CoA:3-ketoacid CoA transferase (SCOT) in skeletal muscle, the rate-limiting enzyme of ketone oxidation (3). It was determined that a muscle-specific knockout (KO) of SCOT (SCOT^MuscleKO) in mice improves glycemia in response to experimental obesity, implying that this metabolic adaptation may contribute to the pathology of obesity-induced T2D. Intriguingly, it was demonstrated with in silico modeling that pimozide, an antipsychotic drug from the diphenylbutylpiperidine (DPBP) family, was a SCOT inhibitor. In vivo studies confirmed that pimozide inhibited SCOT, while recapitulating the glucose-lowering effect observed in obese SCOT^MuscleKO mice. Moreover, pimozide failed to improve glycemia in obese SCOT^MuscleKO mice, confirming that its ability to improve glycemia depended on SCOT inhibition and not its canonical antipsychotic actions where it primarily inhibits the dopamine 2 (D₂) receptor (4).

Pimozide and other members of the DPBP family, were synthesized as D₂ receptor antagonists in the late 1960s to treat schizophrenia and other mental disorders (5–7). Pimozide is currently used to manage tics in Tourette syndrome, as typical antipsychotic agents have been predominantly supplanted by atypical antipsychotics (e.g., lurasidone) (8). In relation to previous findings demonstrating that pimozide improves glycemia in obesity through SCOT inhibition, our aim was to determine whether the ability of pimozide to inhibit SCOT is preserved among the DPBPs, which may suggest new therapeutic utility in repurposing this drug class for the potential treatment of T2D.

An expanded version of our research design and methods can be found in the Supplementary Material.

All animal procedures were approved by the institution’s health sciences animal welfare committee, with animals receiving care according to guidelines from the Canadian Council on Animal Care. To induce experimental obesity, 8-week-old male C57BL/6J mice were fed either a low-fat diet (LFD) (D12450J; Research Diets) or an HFD (D12451; Research Diets) for 12 weeks. Male mice were randomly assigned to receive either vehicle control (corn oil) or pimozide, fluspirilene, or penfluridol (10 mg/kg once every 48 h) by oral gavage for 4 weeks. All animals were subjected to intraperitoneal (IP) glucose tolerance tests (IPGTTs) and IP insulin tolerance tests (IPITTs) at 2 and 3 weeks posttreatment, respectively. At study completion, animals were killed with euthanyl (12 mg), following which their tissues were extracted and snap frozen in liquid nitrogen for storage at −80°C.

The DPBP structures were prepared in the framework of the AMBER99SB force field using the University of California, San Francisco, Chimera Dock Prep tool (9). The docking was performed using AutoDock Vina (10), followed by two sets of molecular dynamics (MD) simulations, one for the apo-SCOT (Protein Data Bank ID: 3DLX) and the other for the SCOT-drug complex using GROMACS 2021 (11). The system was neutralized to the physiological salt concentration and the energy minimized using the AMBER99SB-ILDN force field (12). The system was equilibrated to 1 bar and heated to 300 K before performing the 40-ns production run for apo-SCOT and SCOT-drug complex using the periodic boundary. Root mean square deviation (RMSD), ligand positional RMSD, root mean square fluctuation (RMSF), and hydrogen bonding analyses were calculated using GROMACS. All figures were generated using ChimeraX, and Pymol and Drug Discovery Studio Visualizer were used for protein-drug interaction analysis (13).

After an overnight or 6-h fast, IPGTTs and IPITTs were performed following administration of glucose (2 g/kg) or insulin (0.5 units/kg), respectively, with the Contour Next blood glucose monitoring system (Bayer) used to assess blood glucose levels sampled from tail vein whole blood.

Frozen tissues were homogenized in PBS (pH 7.2) with protease inhibitors (Halt Protease Inhibitor Cocktail; Thermo Fisher Scientific). The resulting lysate was centrifuged at 20,000g for 20 min at 4°C, and the supernatant was collected for measuring the rate of acetoacetyl CoA formation as previously described (14).

All values are presented as mean ± SEM. Multiple groups were compared using a one- or two-way ANOVA, followed by Bonferroni post hoc analysis. Differences were considered significant when P < 0.05. GraphPad Prism 9 software was used for all data analyses.

All data associated with this study are available from the corresponding author upon reasonable request.

The binding modes of penfluridol and fluspirilene were compared with that of pimozide with MD, followed by an MD simulation to determine the stability and dynamics of the DPBP-enzyme complex. Analysis of the binding modes revealed that all DPBPs, analogous to pimozide, promptly aligned and established a hydrogen bond with two critical binding pocket residues, namely Lys368 and Gly322 (Fig. 1A–D).

RMSD analysis revealed that these compounds did not impose any significant distortion on the backbone of the SCOT (Fig. 1E), and the three DPBPs were sufficiently stable inside the oxyanion pocket of SCOT during MD simulation (Fig. 1F). The RMSF of the side chains analysis of the drug-enzyme complex and apo-SCOT structure confirmed the initial docking poses, indicating that the DPBPs under investigation linger in contact with the residues mentioned above throughout the MD simulation (Fig. 1G). Hydrogen bond interactions were also computed throughout the MD simulation, highlighting their importance in DPBP-SCOT interactions (Fig. 1H).

We next used a cell-free assay system to quantify acetoacetyl CoA formation by recombinant SCOT in the presence of its endogenous substrates (Fig. 1I). Consistent with our in silico model, all three DPBPs bind to SCOT and reduce SCOT’s ability to synthesize acetoacetyl CoA (Fig. 1J).

Male mice were subjected to experimental obesity through HFD supplementation (obese mice), whereas their lean counterparts were provided an LFD (lean mice). Mice were treated with either vehicle control, pimozide, penfluridol, or fluspirilene (each at 10 mg/kg every 48 h through oral gavage) during the final 4 weeks (Fig. 2A). All three DPBPs elevated circulating β-hydroxybutyrate (β-OHB) levels in response to an overnight fast, though there was negligible impact on ad libitum/random-fed β-OHB levels (Fig. 2B). The improvement in glycemia following an IPGTT previously observed with pimozide (3) was mirrored by penfluridol and fluspirilene (Fig. 2C) while associated with lower circulating insulin levels (Fig. 2D). Conversely, treatment with all three DPBPs did not improve glycemia following an IPITT (Fig. 2E and F). Treatment with the DPBPs does not induce hypoglycemia, as glucose profiles were unaffected following an IPGTT in lean mice (Supplementary Fig. 1A and B). DPBP administration in obese mice demonstrated no impact on fat mass, lean mass, or body weight (Fig. 2G–I), which was recapitulated in lean mice (Supplementary Fig. 1C–E). Furthermore, all three DPBPs inhibited SCOT activity in skeletal muscle and brain lysates collected at study completion (Fig. 2J–L).

Consistent with pimozide failing to improve glycemia in obese SCOT^MuscleKO mice (3) (Supplementary Fig. 2A–C), all three DPBPs failed to improve glucose tolerance in obese SCOT^MuscleKO mice, whereas they alleviated glucose intolerance in obese human α-skeletal actin-Cre (HSA^Cre) mice (Fig. 3A–C). This improvement in glycemia for all three DPBPs in obese HSA^Cre mice was associated with lower circulating insulin levels (Fig. 3D), while all three DPBPs also increased circulating β-OHB levels following an overnight fast (Fig. 3E).

Ketones are an essential fuel source for the brain during prolonged periods of starvation (15), and brain energy metabolism can regulate whole-body glucose homeostasis (16,17). Thus, we generated brain-specific SCOT KO (SCOT^BrainKO) mice that lack SCOT activity within the brain (Supplementary Fig. 2D–F). Treatment of obese nestin^Cre littermates with all three DPBPs once again improved glucose tolerance, whereas treatment with all three DPBPs failed to improve glucose tolerance in obese SCOT^BrainKO mice (Fig. 3F–H). Although treatment with all three DPBPs was still associated with lower circulating insulin levels in the obese nestin^Cre mice, DPBP-treated obese SCOT^BrainKO mice consistently had higher circulating insulin levels after glucose administration than their nestin^Cre mice counterparts (Fig. 3I). In addition, circulating β-OHB levels following an overnight fast were elevated in the obese SCOT^BrainKO mice or in response to treatment with the DPBPs in the nestin^Cre littermates (Fig. 3J).

To assess the potential contribution of the canonical actions of DPBPs as D₂ receptor antagonists toward our observations, we next treated obese mice with a single IP injection of a structurally unrelated D₂ receptor antagonist, lurasidone (10 mg/kg) (18,19) (Fig. 4A). When compared to treatment with pimozide, we observed that lurasidone had no impact on circulating β-OHB levels and was unable to improve glycemia during an IPGTT (Fig. 4B and C). In addition, lurasidone was devoid of actions to diminish SCOT activity (Fig. 4D and E). Similarly, treatment of obese mice with lurasidone (10 mg/kg) for 4 weeks yielded identical findings to acute treatment whereby no change in glucose tolerance, circulating β-OHB levels, or SCOT activity was observed (Fig. 4F–I).

In this study, we demonstrated that the D₂ receptor antagonist DPBPs, originally developed to treat schizophrenia, also inhibit SCOT activity and induce glucose lowering in overnight-fasted obese mice. While this may not be representative of obese individuals with T2D where improvements in glycemia need to ideally be present throughout, we have observed that pimozide also induces glucose lowering with short-term fasting and in the ad libitum state (3). With the use of state-of-the-art in silico modeling techniques, we show that the DPBPs tightly bind to the catalytic pocket of SCOT to inhibit its enzymatic activity.

Our previous studies were limited because they did not rule out a possible role for the canonical antipsychotic actions of pimozide as a D₂ receptor antagonist in mediating glucose lowering. We posited that the glucose-lowering actions of pimozide and other DPBPs would be independent of these actions, since glycemia was improved in numerous studies following D₂ receptor activation (20–22). Furthermore, the partial D₂ receptor agonist bromocriptine exhibits antihyperglycemic effects in people with T2D (23). Nonetheless, to rule out these actions, we treated mice with a structurally unrelated antipsychotic agent that inhibits D₂ receptors, lurasidone. We specifically chose lurasidone as its distinct chemical structure should dictate that it does not inhibit SCOT activity (24). The inability of lurasidone to improve glucose tolerance is consistent with our data illustrating that the DPBP-mediated alleviation of obesity-induced hyperglycemia depends entirely on the ability of the DPBPs to inhibit SCOT.

We observed some discrepancies compared with our previous findings with pimozide (3), in particular, treatment with the DPBPs did not improve insulin tolerance in obese mice. This may be explained by a reduced fasting period (6 h) versus the overnight fast implemented in our previous study (3) and is consistent with our data following a 5-h fast, demonstrating that pimozide treatment does not improve skeletal muscle insulin sensitivity in obese mice during a hyperinsulinemic-euglycemic clamp. We posit that noninsulin-mediated glucose disposal accounts for DPBP-mediated glucose lowering. This may be attributed to increased glucose oxidation in muscle secondary to stimulation of pyruvate dehydrogenase (PDH) activity, the rate-limiting enzyme of glucose oxidation. Indeed, it has been suggested that increased mitochondrial oxidation of glucose is sufficient to improve glycemia independent of insulin signaling (25), and we did observe that DPBP treatment of obese mice decreased PDH phosphorylation (indicative of increased PDH activity) in gastrocnemius muscle but not in the brain (Supplementary Fig. 3A–C).

One of our surprising observations is that reductions in brain SCOT activity may contribute to the glucose-lowering actions of the DPBPs. Although the glucose clearance profiles of obese SCOT^BrainKO mice were comparable to that of obese nestin^Cre mice, the DPBPs failed to improve glycemia in SCOT^BrainKO mice. As our previous findings revealed that pimozide inhibits SCOT in a noncompetitive manner (3), one potential explanation for the lack of effect in SCOT^BrainKO mice may be due to the higher circulating ketones in these mice. Subsequent elevations in muscle acetoacetate would overcome the DPBP-mediated inhibition of SCOT, supporting the DPBPs’ inability to promote glucose lowering. An alternative explanation could relate to the inhibition of brain SCOT activity playing a regulatory role in whole-body glucose homeostasis. This regulatory role appears to require simultaneous inhibition of brain and muscle SCOT activity/ketone oxidation, since the DPBPs remain capable of inhibiting brain SCOT activity in SCOT^MuscleKO mice or inhibiting muscle SCOT activity in SCOT^BrainKO mice. While a potential brain-muscle ketone metabolism axis regulating glycemia appears intriguing, there are several concerns regarding nestin^Cre mice and perturbed metabolic phenotypes that make such an interpretation ambiguous. Nestin^Cre mice fed an HFD demonstrate improved insulin sensitivity with no improvement in glucose levels (26). In support of this, our comparisons of nestin^Cre mice, floxed SCOT mice, and their Cre/floxed SCOT–negative wild-type littermates fed an HFD indicate that nestin^Cre mice do exhibit improved insulin sensitivity with no change in fasting glucose levels, as well as a generalized growth defect (Supplementary Fig. 4A–F). As the DPBPs were still capable of improving glucose tolerance in nestin^Cre mice but not in SCOT^BrainKO mice, this suggests that brain SCOT activity and ketone oxidation regulate glycemia even if nestin^Cre mice harbor these metabolic phenotypes. However, Cre expression driven by the nestin promoter is also leaky, with reported expression in muscle tissues (26). Although we observed normal SCOT expression in gastrocnemius muscles from SCOT^BrainKO mice, it is possible that SCOT expression may have been decreased in other muscles, which could also account for the failure of the DPBPs to induce glucose lowering in these mice. As such, further interrogation of SCOT^BrainKO mice is still necessary to determine whether brain ketone oxidation plays a critical role in the regulation of glycemia in obesity.

In summary, our study has demonstrated that not only pimozide but the DPBP drug class in general are also SCOT inhibitors. Through these actions and not their ability to inhibit D₂ receptors, they induce potent glucose lowering in obesity. While increases in circulating ketone levels secondary to SCOT inhibition may lead to ketoacidosis as a potential adverse effect, the extent of this increase appears comparable to that observed in individuals consuming ketogenic diets and is more reflective of a nutritional ketosis (15). As the DPBPs appear to be relatively safe in humans, where they have been previously approved for the treatment of schizophrenia/psychosis, this drug class may have utility in being repurposed for the treatment of T2D, though careful monitoring of circulating ketones will be necessary.

Acknowledgments. The in silico modeling studies required access to the Compute Canada and SHARCNET online servers.

Funding. This work was supported by a Canadian Institutes of Health Research project grant (to J.R.U.). J.R.U. is a Tier 2 Canada Research Chair (Pharmacotherapy of Energy Metabolism in Obesity).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.A.T.D., A.A.G., K.Y., R.A.F., C.T.S., K.G., J.S.F.C., J.J.C., F.E., C.L., and R.A. performed the experiments. S.A.T.D., C.A.V.-M., P.A.C., J.N.M.G., R.A., and J.R.U. reviewed and edited the manuscript. S.A.T.D., R.A., and J.R.U. designed the experiments. S.A.T.D. and J.R.U. analyzed the data and wrote the manuscript. All authors read and approved the manuscript for submission. J.R.U. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 82nd Scientific Sessions of the American Diabetes Association, New Orleans, LA, 3–7 June 2022.