Genetic Deletion and Pharmacological Inhibition of Phosphodiesterase 10A Protects Mice from Diet- Induced Obesity and Insulin Resistance

300 Diabetes Volume 63, January 2014

Andrea R. Nawrocki,1 Carlos G. Rodriguez,1 Dawn M. Toolan,2 Olga Price,1 Melanie Henry,1 Gail Forrest,1 Daphne Szeto,1 Carol Ann Keohane,1 Yie Pan,1 Karen M. Smith,3 Izzat T. Raheem,4 Christopher D. Cox,4 Joyce Hwa,1 John J. Renger,2 and Sean M. Smith2

Genetic Deletion and Pharmacological Inhibition of Phosphodiesterase 10A Protects Mice From Diet- Induced Obesity and Insulin Resistance

Phosphodiesterase 10A (PDE10A) is a novel reversed hyperinsulinemia. These data demonstrate therapeutic target for the treatment of that PDE10A inhibition represents a novel schizophrenia. Here we report a novel role of antipsychotic target that may have additional PDE10A in the regulation of caloric intake and metabolic benefits over current medications for energy homeostasis. PDE10A-deficient mice are schizophrenia by suppressing food intake, resistant to diet-induced obesity (DIO) and alleviating weight gain, and reducing the risk for the associated metabolic disturbances. Inhibition of development of diabetes. weight gain is due to hypophagia after mice are fed Diabetes 2014;63:300–311 | DOI: 10.2337/db13-0247 a highly palatable diet rich in fats and sugar but not PHARMACOLOGY AND THERAPEUTICS a standard diet. PDE10A deficiency produces a decrease in caloric intake without affecting meal Atypical antipsychotic medications constitute the front- frequency, daytime versus nighttime feeding line treatment for schizophrenia; however, they have behavior, or locomotor activity. We tested THPP-6, a high rate of discontinuation resulting from dissatis- a small molecule PDE10A inhibitor, in DIO mice. faction with efficacy and a general lack of tolerability (1). THPP-6 treatment resulted in decreased food intake, Several atypical antipsychotics induce weight gain and body weight loss, and reduced adiposity at doses produce adverse metabolic effects that contribute to that produced antipsychotic efficacy in behavioral morbidity and drive patient noncompliance with pre- models. We show that PDE10A inhibition increased scribed medications (2). Increases in weight gain vary and whole-body energy expenditure in DIO mice fed range from modest changes with aripiprazole to signifi- a Western-style diet, achieving weight loss and cant increases with olanzapine (3,4). Metabolic risks as- reducing adiposity beyond the extent seen with food sociated with these medications include increased insulin restriction alone. Therefore, chronic THPP-6 resistance, hyperglycemia, dyslipidemia, and type 2 di- treatment conferred improved insulin sensitivity and abetes (3–5). Therefore, there is a pressing need to

1In Vivo Pharmacology, Merck Research Laboratories, Rahway, NJ Corresponding author: Sean M. Smith, [email protected]. 2 Neuroscience, Merck Research Laboratories, West Point, PA Received 12 February 2013 and accepted 29 September 2013. 3In Vivo Pharmacology, Merck Research Laboratories, West Point, PA This article contains Supplementary Data online at http://diabetes 4Medicinal Chemistry, Merck Research Laboratories, West Point, PA .diabetesjournals.org/lookup/suppl/doi:10.2337/db13-0247/-/DC1. © 2014 by the American Diabetes Association. See http://creativecommons .org/licenses/by-nc-nd/3.0/ for details. diabetes.diabetesjournals.org Nawrocki and Associates 301 identify novel therapeutics to treat schizophrenia that do the Neo cassette between exons 16 and 17 of the not induce weight gain or have an increased risk for PDE10A gene. After backcrossing into C57Bl/6Ntac, producing metabolic adverse effects in patients. 99.6% of the C57Bl/6 genome was confirmed. Animals Cyclic nucleotide phosphodiesterases (PDEs) are were individually housed under a 12-h light/dark cycle a family of enzymes that selectively degrade cyclic AMP and fed ad libitum a standard diet (Teklad 7012; Harlan (cAMP) and cyclic GMP (cGMP). There are 11 different Teklad, Indianapolis, IN), a high-fat diet (HFD; D12492: families of PDEs that vary in their substrate specificity, 60% kcal from fat; Research Diets, New Brunswick, NJ), kinetic properties, modes of regulation, intracellular lo- or a Western-style diet containing 32% kcal from milk fat calization, and tissue expression patterns (6–8). PDE 10A and corn-oil (D12266B) with ad libitum water. All pro- (PDE10A) is a dual-substrate PDE that degrades both tocols for the use of these animals were approved by the cAMP and cGMP. PDE10A is encoded by a single gene Merck Research Laboratories Institutional Animal Care that is highly expressed in the brain and has limited and Use Committee. expression in the peripheral tissues (6,9). In the brain, PDE10A is highly concentrated in the striatum and is Metabolic Phenotyping 2 2 expressed at lower levels in other brain regions (10–12). At 8 weeks of age, male PDE10A / and wild-type (WT) In peripheral tissues, measurable levels of PDE10A ex- mice (n =10–13 per group) were switched on HFD ad pression are limited to the testis and pancreatic islet cells libitum for 10 weeks. Whole-body composition was (9,10,13,14). quantified by magnetic resonance analysis (Echo Medical Substantial evidence has been generated supporting Systems, Houston, TX). Leptin and adiponectin (Meso the hypothesis that PDE10A may be a novel therapeutic Scale Discovery, Gaithersburg, MD), insulin (PerkinElmer, target for the treatment of schizophrenia. PDE10A- Waltham, MA), and glucose (OneTouch Ultra, LifeScan, deficient mice show reduced exploratory behavior when Milpitas, CA), were measured using commercially available placed in a novel environment and have a blunted re- assays. Triglycerides, cholesterol, aspartate aminotrans- sponse to the psychomotor activating effects of ferase (AST), and alanine aminotransferase (ALT) levels N-methyl-D-aspartate receptor (NMDA) receptor antag- were measured using an automated analyzer (Roche, onists phencyclidine and MK-801 (15,16). Selective Indianapolis, IN). PDE10A inhibitors decrease psychomotor activity, re- THPP-6 [2-(6-chloropyridin-3-yl)-4-[(2S)-1- verse deficits in prepulse inhibition, and inhibit condi- methoxypropan-2-yl]oxy-N-(6-methylpyridin-3-yl)-7,8- tioned avoidance responding in rodents, models that dihydropyrido[4,3-d]pyrimidine-6(5H)-carboxamide] are predictive of antipsychotic activity in the clinic was synthesized in-house (18) and was dosed by oral (17–21). PDE10A inhibitors are also efficacious in pre- gavage in 5% Tween80/0.25% methylcellulose. AM251 clinical assays that test cognitive domains impaired in [N-(Piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4- schizophrenia (17,20,22,23) and models of negative dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide] symptoms (17). was purchased from (Tocris, Ellisville, MO). For glucose Here we describe a novel role for PDE10A in the tolerance tests (GTTs), 6-h fasted mice were given regulation of caloric intake and energy homeostasis. Us- 1.5 g/kg glucose intraperitoneally (i.p.), and glucose was 2 2 ing PDE10A knockout (PDE10A / ) mice and a selective measured by glucometer (OneTouch Ultra) at 0, 20, 40, small molecule PDE10A inhibitor, we investigated the 60 and 120 min. Blood (2.5 mL) was collected into 13 role of PDE10A in regulating body weight, feeding be- PBS/EDTA (12.5 mL) for insulin determination. havior, locomotor activity, and energy expenditure in Data were analyzed using repeated-measures ANOVA mice fed a standard diet or a highly palatable diet rich in with Bonferroni post hoc analysis or unpaired Student fats and carbohydrates. We show that pharmacological t tests. Homeostasis model assessment of insulin re- inhibition of PDE10A activity leads to a dose-dependent sistance (HOMA-IR) was calculated as [fasting insulin suppression of food intake and increased energy expen- (mU/L) 3 fasting glucose (mmol/L)]/22.5. diture in obese mice with subsequent reduction of adiposity and improved insulin sensitivity. Thus, we Indirect Calorimetry 2 2 describe a novel role for PDE10A in the homeostatic Male PDE10A / mice and their littermates were ana- regulation of energy metabolism, possibly by integrating lyzed at 10 weeks of age (n = 12 per group). They were peripheral satiety and adiposity signals with central acclimated to individual boxes within the OxyMax sys- regulation of food intake. tem (Columbus Instruments, Columbus, OH). Total and ambulatory locomotor activity, food intake, and Vo and RESEARCH DESIGN AND METHODS 2 VCO2 were measured in 30-min intervals, and the re- Animals spiratory exchange ratio (RER) was calculated as VCO2/ Experiments were performed in lean and diet-induced VO2. Heat (energy expenditure) was calculated as CV 3 0.75 obese (DIO) C57Bl/6NTac mice (Taconic, Germantown, VO2 subject 3 body weight , with CV = 3.815 + 1.232 2 2 NY). PDE10A / mice were generated by replacing 3 RER (24). Total locomotor activity was expressed as a fragment of 2,756 nucleotides (bp 1,947 to 2,000) with one count per two consecutive x-axis infrared beam 302 PDE10A Inhibition Decreases Weight Gain Diabetes Volume 63, January 2014 breaks, and fine movement was defined a single-beam factor), and group differences were further examined break. Food intake and locomotor activity data were ex- using post hoc comparisons (Fisher least significant cluded 30 min before and after handling of the mice. difference). For pair-feeding studies, C57Bl/6 mice were fed an Statistical Analysis HFD (D12492) for 16 weeks and then switched to Data are expressed as means 6 SEM. Statistical analysis D12266B for 4 weeks before study. Mice were acclimated was conducted using ANOVA or unpaired Student to pair-feeding for 1 week. The average feeding activity t test, as indicated. Statistical significance was defined pattern of the group treated with 10 mg/kg THPP-6 was as P , 0.05. imposed on the pair-fed group in 10-min intervals, and pair-feeding was achieved by automatically controlling RESULTS access to the food hopper. PDE10A2/2 Mice Are Protected From HFD-Induced Pharmacokinetic Studies Weight Gain and Associated Metabolic Disturbances Plasma concentration of THPP-6 was sampled from 2 to To explore a functional role of PDE10A in energy me- 2 2 24 h after oral administration and quantified using tabolism and feeding behavior, we challenged PDE10A / a SCIEX API5000 Q-Trap mass spectrometer (Applied mice and WT controls with an HFD, which was high in Biosystems, Concord, ON, Canada). Plasma samples were saturated fats (60% kcal from lard and soybean oil, 20% separated on an Acquity UPLC HSS T3 C18 column kcal carbohydrates). Diet-induced body weight gain was F P , (50 mm 3 2.1 mm 3 1.8 mm), with a mobile phase inhibited [ANOVA, (1,21) = 41.26, 0.01] in 2 2 consisting of solvent A (water with 0.1% formic acid) and PDE10A / mice and associated with a reduction in ca- solvent B (acetonitrile with 0.1% formic acid). THPP-6 loric intake (Fig. 1A, Table 1). Cumulative food intake 2 2 and the internal standard were monitored in the positive was significantly reduced in PDE10A / compared with F P , B ion mode at the transition from m/z 469.10 to 134.90 WT mice [ANOVA, (1,21) = 28.30, 0.01; Fig. 1 ]. DIO and m/z 309.2 to 205.1, respectively. The concentration resistance correlated with reduced fat (t test, P , 0.01) of THPP-6 in the samples was determined using Multi- and lean mass (t test, P , 0.01; Fig. 1C) and reduced 2/2 Quant 2.1 based on standard curve and quality control leptin levels in PDE10A mice (16.8 6 3.4 ng/mL) samples. versus WT controls (65.04 6 4.1 ng/mL; Fig. 1D). Feeding of an HFD leads to modest diet-induced in- In Vitro PDE Activity Assays sulin resistance and dyslipidemia in C57Bl/6 mice. After PDE activity was determined in duplicate at room tem- 10 weeks on the HFD, WT control mice developed hy- perature using an IMAP FP kit (Molecular Devices, Sun- perglycemia (286 6 13 mg/dL) and mild hyper- nyvale, CA) (20). Human PDE10A2 and rhesus monkey insulinemia (6.5 6 1.7 ng/mL; Table 1). Deletion of PDE2A3 enzymes were prepared from cytosolic fractions PDE10A was associated with a normalization of the of transiently transfected AD293 cells, as described (20). metabolic profile evidenced by improved fed plasma All other PDEs (PDE1A, PDE3A, PDE4A1A, PDE5A1, glucose (232 6 7 mg/dL; t test vs. WT, P , 0.01) and PDE6C, PDE7A, PDE8A1, PDE9A2, and PDE11A4) were insulin levels (1.1 6 0.3 ng/mL; t test vs. WT, P , 0.05; glutathione S-transferase–tagged human enzymes Table 1). Improved insulin sensitivity was indicated by PDE10A2/2 expressed in insect cells and were obtained from BPS a reduction in HbA1c (Table 1). mice showed Bioscience (San Diego, CA). The apparent inhibition lower total cholesterol levels in plasma and a trend to- K constant ( i) for THPP-1 against the 11 PDEs was de- ward reduced triglyceride levels in plasma compared with K termined using the apparent M values for each enzyme, controls, indicative of improved lipid metabolism (Table as described by Smith et al. (20). 1). A significant reduction of AST levels (t test, P , 0.01) PDE10A2/2 Measurement of MK-801–Induced Psychomotor suggested overall improved liver function in Activity mice. Male C57Bl/6 mice were used to examine the influence of Diet-Dependent Suppression of Food Intake in 2/2 THPP-6 on MK-801–induced locomotor activity. After PDE10A Mice habituation, animals were given vehicle (10% Tween To investigate the involvement of PDE10A in the regu- 80/90% water) or THPP-6 (1, 3, 10, or 30 mg/kg orally), lation of food intake and whole-body energy metabolism, 2 2 followed 120 min later by vehicle (saline) or MK-801 we placed PDE10A / mice in metabolic rate chambers (0.25 mg/kg i.p.). Animals were placed in locomotor ac- and determined feeding behavior, locomotor activity, tivity monitors (21 3 42 cm; Kinder Scientific, Julian, energy expenditure, and RER in mice when fed a regular CA) 90 min after injection of THPP-6 and left in the chow and after switching them to the HFD. No genotypic activity monitors 90 min after being given MK-801. Total differences were found in cumulative food intake or daily distance traveled was used to assess the influence of food intake when mice were fed a regular chow diet THPP-6 administration on activity during habituation (Fig. 2A–C). As observed previously, introduction of and after MK-801 treatment. Locomotor activity was a highly palatable HFD suppressed food intake in 2/2 analyzed with one-way ANOVA (group as a between-subjects PDE10A mice [ANOVA, F(1,19) = 4.44; P , 0.05], diabetes.diabetesjournals.org Nawrocki and Associates 303

Figure 1—PDE10A2/2 mice are resistant to DIO. A: Changes in body weight of male PDE10A2/2 (n = 10) and their littermate controls (WT, n = 13) after being fed an HFD (60% kcal from fat) for 10 weeks. Initial body weights were 24.2 6 0.5 g for WT and 20.4 6 0.7 g for PDE10A2/2 mice. B: Weekly HFD consumption expressed as cumulative food intake. C: Body composition at the end of the study was assessed as the absolute amount of fat and lean mass as determined by quantitative nuclear magnetic resonance (qNMR) analysis. D: Plasma leptin levels after 10 weeks of HFD feeding. *P < 0.05 vs. WT control.

2 2 resulting in resistance to diet-induced body weight gain in PDE10A / mice within 24 h after switching to the F P , A B F P , fi [ANOVA, (1,22) = 58.77; 0.001; Fig. 2 and ]. The HFD [ANOVA, (1,22) = 17.79, 0.001]. These ndings suppression of food intake was not associated with suggest that a shift in nutrient partitioning toward in- a change in circadian behavior or meal patterns; however, creased fatty acid oxidation relative to glucose oxidation 30-min food intake was significantly reduced [ANOVA, likely contributed to obesity resistance (Supplementary D F(1,22) = 20.11; P , 0.01; Fig. 2D]. Total distance traveled Fig. 1 ). fi and ne locomotor activity was unaffected by genotype Inhibition of PDE10A Causes Weight Loss and or diet (Fig. 2E and F). Leptin levels (16.3 6 1.74 vs. Metabolic Improvements in Obese Mice 6 t P , 4.3 0.66 ng/mL, respectively; test, 0.001) and A small molecule PDE10A inhibitor, THPP-6, was used to 6 6 insulin (6.5 1.7 vs. 1.1 0.3 ng/mL, respectively; further characterize the effects of PDE10A inhibition on t P , PDE10A2/2 test, 0.05) were also reduced in mice body weight, glucose, and lipid metabolism in DIO mice. compared with WT mice. THPP-6 (Fig. 3A) is a potent inhibitor of PDE10A (Ki = Using indirect calorimetry, we investigated the im- 0.92 6 0.08 nmol/L) as determined by a fluorescence fi plication of a PDE10A de ciency on whole-body energy polarization assay measuring the ability of test com- expenditure and nutrient utilization. Net metabolic rate pounds to inhibit hydrolysis of cAMP (25). THPP-6 is was unchanged on the chow diet or after switching greater than 300-fold selective over the other 10 PDE 2/2 PDE10A mice to an HFD (Supplementary Fig. 1A families (Table 2) and did not have any significant ac- and B). We hypothesize that potential effects on energy tivity when tested against a panel of more than 150 expenditure may have been masked by compensatory receptors, enzymes, and ion channels (Panlabs Screen, decreases in energy expenditure due to reduced caloric MDS Pharma). The antipsychotic potential of THPP-6 intake. The RER ( CO2-to-O2 ratio) was similar mice fed was evaluated in the psychostimulant-induced locomotor the chow diet (Supplementary Fig. 1C) but was reduced activity assay, and THPP-6 produced a full attenuation of 304 PDE10A Inhibition Decreases Weight Gain Diabetes Volume 63, January 2014

Table 1—PDE10A2/2 mice are protected from HFD-induced intake and body weight loss independently of the source hyperglycemia and hyperinsulinemia or the amount of the fat in the diet, that is, when fed 10 weeks’ HFD lard- (60% kcal from fat) or milk fat–based (32% kcal PDE10A2/2 from fat) diets (Figs. 4 and 6). DIO mice (initial body WT 6 (n =9) (n = 12) P weight, 43.9 0.6 g) were treated with vehicle, 10 mg/kg DBody weight (g) 22.1 6 0.9 10.8 6 1.1 ,0.01** THPP-6, or were vehicle pair-fed to the average food intake of the THPP-6–treated group for 13 days (n = 7 per Cumulative food intake (g) 187.1 6 3.8 170.9 6 5.2 ,0.01** group). Caloric intake was indistinguishable between THPP-6–treated and pair-fed groups (average 212% Ambient glucose (mg/dL) 286 6 13 232 6 7 ,0.01** cumulative reductions compared with vehicle group; Fig. 6A). Despite equal food intake, body weight loss was Insulin (ng/mL) 6.457 6 1.7 1.120 6 0.31 ,0.05* significantly greater in the THPP-6–treated group, in- Total cholesterol (mg/dL) 226.8 6 8.7 148.9 6 7.3 ,0.01** dicating a feeding-independent mechanism (Student t test; Fig. 6B). Notably, the first dose of THPP-6 caused Triglyceride (mg/dL) 77.2 6 4.4 71.1 6 7.2 a marked suppression of food intake (Supplementary Fig. 2) that was sustained for ;8 h postdose, followed by Adiponectin (mg/mL) 28.01 6 0.97 27.02 6 1.23 a compensatory increase on days 2 and 3. Food intake then stabilized for the remainder of the study (85% of HbA (%) 4.98 6 0.03 4.87 6 0.03 ,0.05* 1c the control groups). Lean mass was not significantly ALT (units/L) 80 6 9.3 25 6 1.7 ,0.01** different between the THPP-6–treated mice and their 6 6 AST (units/L) 80 5.4 69 8.6 pair-fed group (Fig. 6C); however, loss in adipose tissue was greater in the THPP-6 group than in the pair-fed *P , 0.05 vs. age-matched WT control. **P , 0.01. group, suggesting that PDE10A inhibition promotes utilization fat independent of food intake. Pair-feeding allowed us to uncover a PDE10A-related increase in energy expenditure (Fig. 6D). Energy expen- MK-801 (NMDA receptor antagonist) activity in mice at diture was decreased in the pair-fed group as result of a dose (10 mg/kg, orally) that yielded a total plasma reduced food intake but was reversed by THPP-6 treat- concentration of 3.46 6 0.91 mmol/L (Fig. 3B). DIO ment to normal levels. The differential drug effect was mice were dosed with THPP-6 daily before the feeding highly reproducible after 5 days of dosing and in- period began. Steady-state plasma concentrations of dependent of food intake and body weight. THPP-6 THPP-6 were 0.28 6 0.03, 1.98 6 0.18, and 13.12 6 treatment did not independently affect RER and tracked 1.51 mmol/L after oral administration of 3, 10, and closely with food intake (Fig. 6E and Supplementary 30 mg/kg THPP-6, respectively. Fig. 2). Ambulatory activity was comparable to vehicle THPP-6 treatment caused significant body weight (Fig. 6F), with the exception of a transient reduction F loss within 3 days at all doses [ANOVA, (4,35) = 60.92, after the first dose that subsided quickly at steady-state P , 0.01;Fig.4A]comparabletoefficacy with a CB1 concentrations of the drug. Our results indicate that neutral antagonist, AM251, at full target engagement PDE10A inhibition affects body weight and adiposity due (Fig. 4A). THPP-6 treatment also decreased food intake to a combination of hypophagia and increased resting F P , B [ANOVA, (4,29) = 16.04, 0.01; Fig. 4 ]andreduced and active metabolic rate. In addition, the PDE10A in- fat mass (Fig. 4C and D). Reduced leptin [ANOVA, hibitor THPP-6 causes hypophagia and increases energy F P , (4,35) = 30.02, 0.01] and an increase in total adi- expenditure unrelated to the source of lipids and/or ponectin (Fig. 4E and F)wasobservedwithTHPP-6but composition of the diets (Figs. 4 and 6). notAM251treatment[ANOVA,F(4,35) = 12.59, P , 0.001]. Body weight loss was associated reduced fasting insulin DISCUSSION F P , A [ANOVA, (4,35) =8.57, 0.001; Fig. 5 ] and HOMA-IR Schizophrenia is a debilitating disorder estimated to ef- [ANOVA, F(4,35) =6.5,P , 0.001; Fig. 5B]. GTT results were fect up to 1% of the general population (26). In addition not affected by THPP-6 in this study (Fig. 5C). However, to the well-defined psychiatric symptoms, schizophrenic glucose-stimulated insulin release was reduced during the patients have an increased prevalence of obesity (27,28). F P , D GTT [ANOVA, (4,35) = 9.48, 0.001; Fig. 5 ]. In A number of atypical antipsychotic medications have also summary, PDE10A inhibition improved insulin sensitiv- been reported to induce significant weight gain and in- ity by mechanisms that are likely secondary to body crease the likelihood of developing type 2 diabetes and weight loss. cardiovascular disease (3,4,29). Therefore, developing In a separate experiment, we investigated the effect of novel antipsychotic medications that do not induce feeding of a Western-style diet on PDE10A-mediated weight gain or that have the potential to reverse weight food intake in obese mice. We confirmed that equivalent gain and metabolic effects produced by marketed anti- doses of THPP-6 caused comparable suppression of food psychotic drugs is of utmost importance. diabetes.diabetesjournals.org Nawrocki and Associates 305

Figure 2—Deletion of PDE10A suppresses food intake when mice are fed a highly palatable HFD but not a standard low-fat diet (chow). Daily measures of body weight (A) and food intake (B)inPDE10A2/2 and littermate controls (WT) are shown. Food intake on chow (C)or HFD (D) was measured in 30-min intervals after 2 days of acclimation to the newly introduced diet. Locomotor activity was comparable in PDE10A2/2 and WT mice fed chow (E) or the HFD (F). The shaded areas indicate active period (lights off). *P < 0.05 vs. WT control.

In this study, we describe a novel role for PDE10A in changes correlated with improved HbA1c levels and the regulation of caloric intake and energy expenditure. demonstrate a protective effect against the development We found that deletion of PDE10A in mice leads to a pro- of insulin resistance associated with obesity. found suppression of food intake after exposure to highly Using indirect calorimetry, we show that PDE10A palatable Western-style diets containing an excess of fats regulates feeding behavior and also influences overall and sugars. This significant reduction of food consumption metabolic rate. Under conditions of HFD feeding, 2 2 conferred resistance to diet-induced weight gain in PDE10A / mice exhibited a small but significant shift in 2/2 PDE10A mice and protection against the development RER toward increased lipid oxidation, whereas total heat of comorbidities typically associated with obesity such as was unchanged. The increase in lipid oxidation at the 2 2 insulin resistance and dyslipidemia. PDE10A / mice relative cost of glucose utilization is reflective of a state had normalized lipid profiles and a significant reduction of reduced caloric intake when increased mobilization of in fasting insulin and glucose-stimulated insulin secre- lipids is necessary to maintain energy balance (24). Such tion compared with obese DIO controls (Table 1). These an increase in lipid oxidation may have contributed to 306 PDE10A Inhibition Decreases Weight Gain Diabetes Volume 63, January 2014

Figure 3—A: Chemical structure of the PDE10A inhibitor THPP-6. Papp, apparent permeability; Pgp, P-glycoprotein substrate; PPB, plasma protein binding. B: THPP-6 dose-dependently attenuated psychomotor activity induced by the NMDA receptor antagonist MK- 801. Mean plasma concentrations of THPP-6 for each dosing group are listed above the dataset on the graph. **P < 0.01 vs. vehicle (V) + MK-801. a preservation of lean mass and a reduction of fat mass increased overall energy utilization. The mechanism for 2 2 in PDE10A / mice. From these findings we speculate this increase in metabolic rate is yet to be dissected on 2 2 that effects on total energy utilization in PDE10A / a molecular level, and greater understanding of the link mice were masked by a compensatory decrease in energy between PDE10A inhibition and peripheral oxidative ca- expenditure associated with reduced caloric intake. The pacity is needed to fully interpret the utility of this net balance therefore contributing to the obesity re- mechanism. In any case, enhanced nutrient combustion in 2 2 sistance in PDE10A / mice is likely a combination of the absence of cardiovascular effects (data not shown) and fi reduced intake and increased metabolism. in combination with signi cant caloric restriction may be Findings based on genetically deficient PDE10A mice of great therapeutic interest. were mirrored pharmacologically with a PDE10A in- Pharmacological inhibition of PDE10A was also asso- fi hibitor, THPP-6. Chronic THPP-6 treatment of DIO mice ciated with an improved metabolic pro le in insulin- PDE10A2/2 caused significant body weight loss and suppression of resistant DIO mice. Similar to mice, fasting food intake at similar plasma exposures that produce insulin levels were reduced, and there was a trend toward antipsychotic-like activity in animal models. Body com- improved glucose disposal with reduced glucose-stimulated position analysis indicated that body weight loss was insulin release during a GTT. It is noteworthy that fasting driven by a reduction in body fat while lean mass was maintained. Inhibition of PDE10A activity produced body weight loss significantly greater than data from the pair-fed group, revealing a previously not appreciated role of PDE10A in the regulation of energy expenditure. Table 2—Selectivity of THPP-6 for PDE10A compared with Importantly, these effects were sustained after multiple other human PDE families doses and chronic THPP-6 treatment caused metabolic PDE family Substrate Selectivity vs. PDE10A improvements in the absence of effects on locomotive PDE1A cGMP .1,0003 behavior. PDE2A cAMP .1,0003 Our data also indicate that the type of substrates used PDE3A cAMP .1,0003 by the body is not affected by PDE10A inhibition. Unlike PDE4A cAMP .1,0003 PDE10A2/2 the results from the mice, RER was over- PDE5A cGMP .3003 lapping with the pair-fed group, suggesting that the com- PDE6A cGMP .3003 pound did not affect substrate preference. However, the PDE7A cAMP .1,0003 lack of an effect may be attributable to food restriction and PDE8A cAMP .1,0003 cannot be dissected in this particular paradigm. Our pair- PDE9A cGMP .1,0003 feeding experiments unequivocally demonstrate a food- . 3 independent component of PDE10A inhibition leading to PDE11A cAMP 300 diabetes.diabetesjournals.org Nawrocki and Associates 307

Figure 4—Chronic dosing of a PDE10A inhibitor THPP-6 inhibited food intake in obese mice, leading to body weight loss and reduced adiposity. A: Daily change in body weight relative to baseline (average, 44 g). B: Cumulative food consumption over 13 days. Fat mass (C) and lean mass (D) were determined by quantitative magnetic resonance analysis on day 13. Plasma leptin (E) and total adiponectin (F) were measured after DIO mice received 13 days of THPP-6 treatment (n = 8 per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs. vehicle.

insulin was reduced to a greater extent upon inhibition of peripheral tissue, in our hands, non–brain-penetrant PDE10A than with similar body weight loss associated antagonists do not have insulin-sensitizing properties in- with AM251 treatment at full-target engagement, sug- dependently of body weight loss (data not shown). To that gesting superior insulin-sensitizing efﬁcacy. The CB1 re- effect, reductions in fasting insulin correlated with an in- ceptor antagonist AM251 is thought to drive metabolic crease in circulating total adiponectin levels after THPP-6 improvements predominantly through central regulation treatment that was absent in mice treated with our an- of appetite. Although CB1 receptors are expressed in orexic control compound AM251. Adiponectin levels 308 PDE10A Inhibition Decreases Weight Gain Diabetes Volume 63, January 2014

Figure 5—Pharmacological inhibition of PDE10A inhibition improved glucose metabolism by reducing fasting and postprandial insulin. Fasting insulin (A) and glucose (B) after 2 weeks of treatment with THPP-6 are shown. HOMA-IR was calculated for each treatment group. C: GTT after 13 days of treatment with THPP-6 in obese mice. Glucose (1.5 g/kg) was injected intraperitoneally after a 6-h fast, and blood was collected from the tail vein for glucose measurements. The inset shows area under the curve (AUC) values for 0–120 min of glucose challenge. D: Reduced glucose-dependent insulin response after chronic THPP-6 treatment measured 20 min postglucose challenge (n = 8 per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs. vehicle control.

inversely correlate with adipose tissue inflammation, he- THPP-6–mediated effects on glucose metabolism and patic glucose output and peripheral glucose uptake, and insulin levels are conferred predominantly by body adiposity (30). Taken together, our data indicate a poten- weight loss and not through direct regulation of insulin tial for additional body weight–independent benefits re- secretion. lated to PDE10A inhibition on handling of glucose that Preclinical data pointing to nutrient-dependent effects should be further investigated. However, whether PDE10A on feeding behavior suggest that PDE10A plays a role in activity directly impacts metabolic parameters independently central orexigenic pathways and/or with a reward of CNS-mediated effects remains to be examined. mechanism related to the palatability of various types of PDE10A mRNA is expressed in pancreatic islet cells, nutrient and caloric content of food. This is supported by and researchers have reported that small-molecule the abundant expression of PDE10A in the brain and PDE10A inhibitors function as insulin secretagogues in limited distribution in peripheral tissue. Specifically, vitro (31,32). Furthermore, Cantin et al. (31) demon- PDE10A is expressed at high levels in the mesolimbic and strated that PDE10A inhibitors improved glucose toler- mesocortical dopamine systems, including the caudate, ance and reduced insulin secretion in Wistar rats after nucleus accumbens, hippocampus, and prefrontal cortex a glucose challenge, suggesting that the effects of THPP-6 (10,11). Preclinical studies have linked cues from highly in DIO and HFD studies may be partly related to altered palatable foods to multiple aspects of dopaminergic sig- insulin secretion. In this study, we examined the effects naling. Mesolimbic dopaminergic neurons that innervate of THPP-6 on glucose-stimulated insulin release and the nucleus accumbens have been associated with moti- showed that THPP-6 did not significantly affect glucose- vation for highly palatable food and related reward stimulated insulin release in isolated mouse islets (data behaviors (33–36). Furthermore, mesocortical dopami- not shown). These data support the hypothesis that nergic neurons innervate brain regions that regulate the diabetes.diabetesjournals.org Nawrocki and Associates 309

Figure 6—THPP-6 increases energy expenditure and causes body weight loss in mice fed a Western-style diet independent of food intake. Effect of vehicle, 10 mg/kg THPP-6, or pair-feeding to THPP-6 on daily food intake (A) and body weight (B)isshown.Lines indicate P < 0.05 vs. vehicle, *P < 0.05, **P < 0.01, ***P < 0.001 THPP-6 vs. pair-fed group. C: Changes in body composition measured by quantitative nuclear magnetic resonance NMR at baseline and after 13 days of THPP-6 treatment. *P < 0.05, **P < 0.01, ***P < 0.001 vs. vehicle. Energy expenditure (D), respiratory quotient (E), and X-ambulatory activity (F)areshownin1-hbinsfor baseline, day 1, and day 2 of dosing and for the last 4 days of steady-state dosing. Baseline and THPP-6 treatment periods are indicatedbyshadedareas.

emotional response to eating, and dopaminergic neurons that PDE10A is positioned to play an important role in that innervate the caudate influence sensory-motor the central regulation of feeding behavior. aspects of feeding (34,36,37). In addition to the direct In summary we have described a novel role for effects of palatable foods on central dopaminergic sys- PDE10A in the regulation of feeding behavior and the tems, a number of peripheral metabolic signals, including regulation of metabolic rate. We have demonstrated that leptin, insulin, and ghrelin, influence dopaminergic sys- genetic deletion and pharmacological inhibition of tems (34,38,39). Taken together, these studies suggest PDE10A both have significant effects on body weight and 310 PDE10A Inhibition Decreases Weight Gain Diabetes Volume 63, January 2014 food intake in rodents and on improved insulin sensi- 14. Soderling SH, Bayuga SJ, Beavo JA. Isolation and characterization of tivity. Furthermore, PDE10A inhibition increased energy a dual-substrate phosphodiesterase gene family: PDE10A. Proc Natl Acad expenditure, leading to reduced adiposity and greater Sci U S A 1999;96:7071–7076 weight loss than seen under conditions of equal caloric 15. Siuciak JA, Chapin DS, Harms JF, et al. Inhibition of the striatum-enriched intake. These findings suggest that PDE10A inhibitors phosphodiesterase PDE10A: a novel approach to the treatment of psy- may be novel therapeutics for the treatment of schizo- chosis. Neuropharmacology 2006;51:386–396 phrenia with the potential to address weight gain and 16. Siuciak JA, McCarthy SA, Chapin DS, et al. Genetic deletion of the metabolic issues associated with this disorder. striatum-enriched phosphodiesterase PDE10A: evidence for altered striatal function. Neuropharmacology 2006;51:374–385 17. Grauer SM, Pulito VL, Navarra RL, et al. Phosphodiesterase 10A inhibitor Funding. All work in the manuscript was funded by Merck Research activity in preclinical models of the positive, cognitive, and negative Laboratories. symptoms of schizophrenia. J Pharmacol Exp Ther 2009;331:574–590 Duality of Interest. All authors are employees of Merck Research Laboratories. No other potential conflicts of interest relevant to this article were 18. Raheem IT, Breslin MJ, Fandozzi C, et al. Discovery of tetrahydropyrido- reported. pyrimidine phosphodiesterase 10A inhibitors for the treatment of schizo- – Author Contributions. A.R.N. and S.M.S. wrote and edited the phrenia. Bioorg Med Chem Lett 2012;22:5903 5908 manuscript. C.G.R., D.M.T., O.P., M.H., G.F., D.S., C.A.K., Y.P., K.M.S., I.T.R., 19. Schmidt CJ, Chapin DS, Cianfrogna J, et al. Preclinical characterization and C.D.C. generated data for the manuscript and contributed to discussion. of selective phosphodiesterase 10A inhibitors: a new therapeutic ap- J.H. and J.J.R. contributed to discussion and edited the manuscript. S.M.S. is proach to the treatment of schizophrenia. J Pharmacol Exp Ther 2008; the guarantor of this work and, as such, had full access to all the data in the 325:681–690 study and takes responsibility for the integrity of the data and the accuracy of 20. Smith SM, Uslaner JM, Cox CD, et al. The novel phosphodiesterase 10A the data analysis. inhibitor THPP-1 has antipsychotic-like effects in rat and improves cog- References nition in rat and rhesus monkey. Neuropharmacology 2013;64:215–223 1. Tsutsumi C, Uchida H, Suzuki T, et al. The evolution of antipsychotic switch 21. Wadenberg ML, Soliman A, VanderSpek SC, Kapur S. Dopamine D(2) receptor and polypharmacy in natural practice—a longitudinal perspective. Schiz- occupancy is a common mechanism underlying animal models of antipsy- ophr Res 2011;130:40–46 chotics and their clinical effects. Neuropsychopharmacology 2001;25:633–641 2. Weiden PJ, Mackell JA, McDonnell DD. Obesity as a risk factor for anti- 22. Rodefer JS, Murphy ER, Baxter MG. PDE10A inhibition reverses subchronic psychotic noncompliance. Schizophr Res 2004;66:51–57 PCP-induced deficits in attentional set-shifting in rats. Eur J Neurosci 3. Newcomer JW. Metabolic risk during antipsychotic treatment. Clin Ther 2005;21:1070–1076 2004;26:1936–1946 23. Rodefer JS, Saland SK, Eckrich SJ. Selective phosphodiesterase inhibitors 4. Newcomer JW. Antipsychotic medications: metabolic and cardiovascular improve performance on the ED/ID cognitive task in rats. Neuropharma- risk. J Clin Psychiatry 2007;68(Suppl. 4):8–13 cology 2012;62:1182–1190 5. Casey DE, Haupt DW, Newcomer JW, et al. Antipsychotic-induced weight gain 24. Tschöp MH, Speakman JR, Arch JR, et al. A guide to analysis of mouse and metabolic abnormalities: implications for increased mortality in patients energy metabolism. Nat Methods 2012;9:57–63 – – with schizophrenia. J Clin Psychiatry 2004;65(Suppl. 7):4 18; quiz 19 20 25. Huang W, Zhang Y, Sportsman JR. A fluorescence polarization assay for 6. Bender AT, Beavo JA. Cyclic nucleotide phosphodiesterases: molecular cyclic nucleotide phosphodiesterases. J Biomol Screen 2002;7:215–222 – regulation to clinical use. Pharmacol Rev 2006;58:488 520 26. van Os J, Kapur S. Schizophrenia. Lancet 2009;374:635–645 7. Conti M, Beavo J. Biochemistry and physiology of cyclic nucleotide 27. Fontaine KR, Heo M, Harrigan EP, et al. Estimating the consequences of phosphodiesterases: essential components in cyclic nucleotide signaling. anti-psychotic induced weight gain on health and mortality rate. Psychiatry Annu Rev Biochem 2007;76:481–511 Res 2001;101:277–288 8. Reneerkens OA, Rutten K, Steinbusch HW, Blokland A, Prickaerts J. Se- 28. Wirshing DA. Schizophrenia and obesity: impact of antipsychotic medica- lective phosphodiesterase inhibitors: a promising target for cognition en- tions. J Clin Psychiatry 2004;65(Suppl. 18):13–26 hancement. Psychopharmacology (Berl) 2009;202:419–443 29. Parsons B, Allison DB, Loebel A, et al. Weight effects associated with 9. Fujishige K, Kotera J, Michibata H, et al. Cloning and characterization of antipsychotics: a comprehensive database analysis. Schizophr Res 2009; a novel human phosphodiesterase that hydrolyzes both cAMP and cGMP 110:103–110 (PDE10A). J Biol Chem 1999;274:18438–18445 30. Scherer PE. Adipose tissue: from lipid storage compartment to endocrine 10. Coskran TM, Morton D, Menniti FS, et al. Immunohistochemical localization – of phosphodiesterase 10A in multiple mammalian species. J Histochem organ. Diabetes 2006;55:1537 1545 Cytochem 2006;54:1205–1213 31. Cantin LD, Magnuson S, Gunn D, et al. PDE-10A inhibitors as insulin – 11. Seeger TF, Bartlett B, Coskran TM, et al. Immunohistochemical localization secretagogues. Bioorg Med Chem Lett 2007;17:2869 2873 of PDE10A in the rat brain. Brain Res 2003;985:113–126 32. Heimann E, Jones HA, Resjö S, Manganiello VC, Stenson L, Degerman E. 12. Xie Z, Adamowicz WO, Eldred WD, et al. Cellular and subcellular locali- Expression and regulation of cyclic nucleotide phosphodiesterases in hu- zation of PDE10A, a striatum-enriched phosphodiesterase. Neuroscience man and rat pancreatic islets. PLoS ONE 2010;5:e14191 2006;139:597–607 33. Hoebel BG, Hernandez L, Schwartz DH, Mark GP, Hunter GA. Microdialysis 13. Loughney K, Snyder PB, Uher L, Rosman GJ, Ferguson K, Florio VA. Iso- studies of brain norepinephrine, serotonin, and dopamine release during lation and characterization of PDE10A, a novel human 39,59-cyclic nu- ingestive behavior. Theoretical and clinical implications. Ann N Y Acad Sci cleotide phosphodiesterase. Gene 1999;234:109–117 1989;575:171–191; discussion 192–193 diabetes.diabetesjournals.org Nawrocki and Associates 311

34. Palmiter RD. Is dopamine a physiologically relevant mediator of feeding 37. Palmiter RD. Dopamine signaling in the dorsal striatum is essential for behavior? Trends Neurosci 2007;30:375–381 motivated behaviors: lessons from dopamine-deﬁcient mice. Ann N Y Acad – 35. Salamone JD, Correa M. Motivational views of reinforcement: implications Sci 2008;1129:35 46 for understanding the behavioral functions of nucleus accumbens dopa- 38. Figlewicz DP, Benoit SC. Insulin, leptin, and food reward: update 2008. Am mine. Behav Brain Res 2002;137:3–25 J Physiol Regul Integr Comp Physiol 2009;296:R9–R19 36. Smith GP, Schneider LH. Relationships between mesolimbic 39. Lenard NR, Berthoud HR. Central and peripheral regulation of food intake dopamine function and eating behavior. Ann N Y Acad Sci 1988;537: and physical activity: pathways and genes. Obesity (Silver Spring) 2008;16 254–261 (Suppl. 3):S11–S22