Small‐Molecule Neuromedin U Receptor 2 Agonists Suppress Food Intake and Decrease Visceral Fat in Animal Models

Abstract Obesity is a growing public health concern, with 37.5% of the adult population in need of therapeutics that are more efficacious with a better side effect profile. An innovative target in this regard is neuromedin U, a neuropeptide shown to suppress food intake and attenuate weight gain in animal models. These effects of neuromedin U on feeding behavior are thought to be related to agonism at the centrally expressed neuromedin U receptor 2 (NMUR2). As peptides present unique challenges that limit their therapeutic potential, the discovery of small‐molecule NMUR2 agonists is needed to validate the targets therapeutic value, but to date, none have been evaluated in any animal model of disease. We therefore assessed two small‐molecule NMUR2 agonists for their in vitro signaling and their in vivo efficacy. The NMUR2 agonists were synthesized and both NMUR2 agonists, NY0116 and NY0128, decreased cAMP while stimulating calcium signaling in stably expressing NMUR2 HEK293 cells. When small‐molecule NMUR2 agonists were tested in vivo, acute administration significantly decreased high‐fat diet consumption. Repeated administration of the compounds decreased body weight and more specifically, decreased the percentage of visceral adipose tissue (VAT) in obese mice. These results have confirmed small‐molecule NMUR2 agonists are efficacious in animal models to decrease fat content, food intake, and body weight, suggesting NMUR2 is a promising therapeutic target for metabolic disorders.


| INTRODUCTION
Obesity remains a troubling health problem that demands a more mechanistic appreciation of the molecular and neural basis of food consumption. 1 The complex nature of obesity, a disorder partially characterized by overconsumption of energy-dense foods, creates challenges for developing therapeutics. 2 Although 80 million people in the United States are obese, 3 the available medications and surgical interventions suffer from modest efficacy and/or numerous side effects. 4 Our objective is to identify and validate new druggable targets via pharmacological approaches that exploit an enhanced understanding of mechanisms underlying feeding behavior. Neuromedin U (NMU) is a neuropeptide, synthesized in the lateral hypothalamus and expressed in all mammals. 5 NMU-deficient mice are hyperphagic, obese, and have decreased metabolic activity. 6 Conversely, systemic administration of NMU peptide suppresses food intake 7 and causes weight loss. 8 NMU gene variants have also been associated with obesity, 9 whereas studies have also shown NMU to be upregulated in calorie restricted animals. 10 Together, these data suggest that NMU plays an important role in regulation of feeding and body weight.
To leverage the desirable antiobesity effects of NMU, it is important to understand the primary sites of action for NMU. There are two high-affinity receptors for NMU, neuromedin U receptor 1 (NMUR1) and NMUR2. These receptors also bind the endogenous neuropeptide neuromedin S (NMS). NMUR1 is enriched in the periphery, and may be involved in regulation of the immune system and responses to pain. 11,12 NMUR2, however, is enriched in the brain, specifically the hypothalamus, confirmed by qRT-PCR and mRNA analyses. [13][14][15] NMUR2 is highly expressed in regions associated with food reward 16 and mediates the central effects of NMU (and NMS) on food intake and body weight. 8,17 NMUR2 may represent a druggable target and a potential access point for the control of important neural pathways underlying food intake and ultimately obesity. NMUR2 is a G-proteincoupled receptor, suggested to couple to both G i/o and G q and was de-orphaned over 10 years ago. 7,18 Following the discovery that NMUR2 (aka FM-4, TGR-1) is stimulated by NMU, NMUR2 was shown to play an important role in regulating food intake and body weight. Knockout of NMUR2 in mice causes hyperphagia, weight gain, and decreased metabolic activity. 17 Although not all studies observed the same effect on food intake after NMUR2 knockout. 8 Furthermore, our laboratory discovered that NMUR2 regulates specific types of feeding behavior in animal models. That is, knockdown of NMUR2 in the paraventricular nucleus of the hypothalamus potentiates binge-type eating, increases consumption of a highfat diet, and stimulates weight gain. 19 Recent studies have also shown that NMU administered to the paraventricular nucleus of the hypothalamus also decreases intake 20 and motivation for highfat diet. 5 Taken together, NMUR2 activation shows therapeutic promise for treating obesity.
Several small-molecule agonists for NMUR2 were discovered by Meng et al. 21 via a high-throughput screen. A series of the original compounds were investigated in live-cell assays, allowing a preliminary structure-activity relationship. In particular, two structurally similar compounds, NY0116 and NY0128 (chemical structures shown in Figure 1A) containing a lipophilic trityl motif and a hydrophilic guanidine scaffold were demonstrated as agonists against human NMUR1 (hNMUR1) and human NMUR2 (hNMUR2) via intracellular calcium mobilization assays. 21 In this previous work, NY0116 had EC 50 values of 27.76 μmol/L for hNMUR1 and 13.61 μmol/L for hNMUR2, whereas NY0128 had EC 50 values of 29.99 μmol/L for hNMUR1 and 10.30 μmol/L for hNMUR2. 21 However, the potential of these smallmolecule agonists to suppress food intake or alter body composition in animals has not been evaluated, creating uncertainty about the potential of NMUR2 as a target for obesity. Here we have filled a major gap in our understanding of NMUR2 as a therapeutic target by performing the first test of small-molecule NMUR2 agonists in any animal model of disease. These results support NMUR2 as a druggable target that can suppress food intake and improve body composition.

| Ethics statement
All experiments were carried out in accordance with all national and local guidelines and regulations, with the Guide for the Care and Use of Laboratory Animals 27, and with the approval of the Institutional Animal Care and Use Committee at The University of Texas Medical Branch. All efforts were made to minimize animal suffering and to reduce the number of animals used.

| Compounds
NY0116 and NY0128 were synthesized in-house following a reported synthetic route 21 with optimized procedures. The chemical structures were fully characterized using H1 and C13 NMR and mass spectrometry techniques, and compound purity was verified using HPLC analysis.

| Tango beta-arrestin activity assay
The tango beta-arrestin assay was performed according to previously

| Micro-CT scan
Mouse whole-body scans were performed using a Siemens Inveon

| Serum collection
Post micro-CT scan, mice were exposed to CO 2 and decapitated.
Following decapitation, 500 μL of trunk blood was collected and allowed to incubate at room temp for >15 minutes. Following incubation, the blood samples were then centrifuged at 1500 g for 15 minutes. Following centrifugation, plasma was removed from the sample and the serum was stored at −80°C until further analysis.
Serum analysis was conducted by the Texas A&M Veterinary Medicine Diagnostic Laboratory. Cholesterol, creatinine, blood urea nitrogen, aspartate aminotransferase, and alanine aminotransferase and glucose levels were analyzed in the serum. Due to hemolysis of samples, some samples were unable to be accurately analyzed and were removed from further analysis.

| Data analysis and statistics
All data were analyzed using GraphPad Prism 7.0 software (La Jolla, CA

| NMUR2 agonists NY0116 and NY0128 potently inhibit cAMP
The pharmacological activity of NMUR2 agonists NY0116 and NY0128 ( Figure 1A) were evaluated in a G i mode cAMP assay using HEK293 cells stably expressing hNMUR2. Cells were stimulated with 1 nmol/L isoproterenol to elevate cAMP levels, which is required to see an inhibition of cAMP via G i -coupled GPCRs. 24 Figure 1C) and 0.71 ± 1.36 nmol/L for NY0128 ( Figure 1D) (Means ± SEM, n = 15-16 independent experiments) ( Table 1). To validate the inhibition of cAMP by these NMUR2 agonists was G imediated, the stable NMUR2 expressing cells were pretreated with 150 ng/mL pertussis toxin for 18 hours to uncouple G i proteins. Pertussis toxin entirely reversed the observed cAMP inhibition, indicating the potent agonism of these ligands requires NMUR2-G i coupling ( Figure 1E). To further insure the inhibition was due to activation of NMUR2, cells not expressing NMUR2 were also evaluated (Figure 1F). There was no effect on cAMP at concentrations of compounds that decreased cAMP in the NMUR2 stably expressing cells.
T A B L E 1 It shows the EC50 and Emax (relative to NmU-8) values of NmU-8, NY0116, and NY0128 when evaluated in Gi mode.

| NMUR2 agonists NY0116 and NY0128 mobilize calcium
To further assess the pharmacology of NY0116 and NY0128, full dose responses (1 nmol/L to 100 μmol/L) of ligands were tested in calcium mobilization assays by assessing calcium dye fluorescence in living cells (Figure 2A). Robust responses to both ligands were observed indicating G q -mediated NMUR2 signaling. EC 50 values estimated from regression curves were 32.7 ± 1.1 μmol/L for NY0116 and 16.9 ± 0.9 μmol/L for NY0128 (Means ± SEM, n = 4). Although the observed potencies for both ligands were low, they are similar to the micromolar potencies previously reported for NMUR2 calcium signaling. 21 These calcium mobilization effects were not due to off-target compound activity as no significant increase in calcium mobilization was observed in wild-type HEK293 cells lacking the NMUR2 (Figure 2B). In addition, vehicle treatment, with up to 0.5% DMSO (DMSO concentration at 100 μmol/L drug point), showed no effect on calcium mobilization in either NMUR2 stable or wild-type HEK293 cells.
As an additional functional assessment of NMUR2 signaling, the agonists were tested in a beta-arrestin recruitment and activity assay. Contrary to receptor-mediated G i /cAMP or G q /calcium signaling, NY0116 and NY0128 did not recruit beta-arrestin, even when the ligands were tested up to 10 μmol/L ( Figure 2C). NY0128 also significantly decreased basal levels of beta-arrestin recruitment and activity, whereas NY0116 did not ( Figure 2D) ( Table 2).

| Pharmacokinetics study
Plasma concentrations for NY0116 peak at 10 hours (n = 3 for all time points, except at 10 hours when n = 2) ( Figure 3A). Plasma concentrations for NY0128 peak at 4 hours postinjection (n = 3 for all time points, except at 45 min when n = 2) ( Figure 3B). According to Lipinski's "Rule of Five" for drug-like properties, 25 Figure 3C). Pharmacokinetic analysis of NY0116 ( Figure 3D) and NY0128 ( Figure 3E) concentrations in rats' wholebrain homogenates and the brain-to-plasma ratio of NY0116 ( Figure 3F) and NY0128 ( Figure 3G) shows that the compounds accumulate in the brain over time.

| Acute NY0116 and NY0128 decrease food intake
We evaluated the effects of NY0116 and NY0128 on intake of a standard diet and a high-fat diet. NY0116 had no effect on consumption of a standard diet (main effect of drug treatment F  Figure 4C). NY0128 also suppressed intake of a high-fat diet in a T A B L E 2 It shows the EC50 and Emax (relative to NMS) values of NMS, NY0116, and NY0128 when evaluated in Gq mode. The EC50 and Emax values for NMS in a beta-arrestin recruitment assay are also provided   Figure 4D).

| Repeated NY0116 and NY0128 decrease bodyweight, VAT, and cholesterol on a high-fat diet
Compounds were administered over 14 days at doses of 3, 10, and 30 mg/kg. Both NY0116 and NY0128 treatments significantly decreased bodyweight (F(5,42) = 10.69, P < .001) compared to vehicle, whereas mice were maintained on a high-fat diet ( Figure 5A-B).
To compare the amounts of VAT between the groups, 23 we examined the cross section at the L5 vertebrae, using Inveon Research Workplace for quantification. Representative images of the cross sections quantified are shown in Figure 5C. There was a significant main effect (F(5,44) = 26.86, P < .001) and the high dose of NY0128 significantly decreased VAT percentage compared to vehicle ( Figure 5D). Food intake was measured daily with no significant effect seen during the study ( Figure 5E). Serum chemistry analysis showed a significant decrease (F(7,45) = 32.48, P < .001) in cholesterol ( Figure S1) at the highest dose of NY0128, with no change in blood urea nitrogen (BUN) ( Figure S1). There were also no observed drug-induced changes in creatinine and aspartate aminotransferase (AST) ( Figure S1).

| DISCUSSION
In this study, we evaluated the potency and efficacy of two small molecule NMUR2 agonists in vitro and in vivo, previously only evaluated in an in vitro model measuring calcium accumulation. NMUR2 has been shown to signal through both the G i and G q G-protein    In order to understand the pharmacological properties of the compounds for future studies, initial pharmacokinetic studies were conducted, with rats receiving NY0116 or NY0128 (30 mg/kg).
Both compounds had appreciable plasma concentration values, with NY0116 reaching the highest concentration in both the plasma and the brain. NY0116 also reached maximum plasma concentration twice as fast as NY0128, while also having a half-life twice as long. Although plasma levels of NY0128 stayed lower than NY0116, NY0128 had an increased brain-to-plasma ratio, with compound accumulating in the brain up to 10 hours post administration. The brain exposure is critical to the work conducted in this paper, because we are targeting a centrally expressed receptor, 13,29 therefore it is crucial that our compounds cross the blood brain barrier. The presence of the compounds in the brain suggests the compounds have access to the intended target, NMUR2. The concentrations in the brain homogenate are comparable to the EC50s from the cell assays. Furthermore, the timing-course of brain access and inhibition of food intake occur at the same time, thus providing evidence that the compounds are acting on NMUR2 in the brain. This is particularly important when using these tool compounds as scaffolds to design other small molecule NMUR2 agonists. The ability to penetrate the blood brain barrier, the potency, and the efficacy of these compounds provides us with an advantageous starting point in developing a pharmacotherapy targeting NMUR2.
Previous studies have shown that NMUR2 plays an important role in feeding, with one showing central administration of NMU significantly decreased food intake 20 leading us to investigate the effect of the compounds on food intake. A single dose of compound significantly decreased consumption of high-fat diet, 24 hours after treatment. NY0128 also showed a decrease at the 8 hour time point of high-fat diet consumption, as well as at the 8 and 24 hour time points of standard diet consumption. These data suggest that both compounds affect food intake similar to NMU.
Lastly, the compounds showed a decrease in adiposity after chronic subcutaneous treatment. Compound was administered once per day for 14 days. Excitingly, the high dose of NY0128 not only prevented weight gain but also resulted in weight loss while also reducing cholesterol levels compared to vehicle. This is complemented by the NY0128 30 mg/kg reduction in VAT and together suggests that part of the observed weight reduction with NY0128 is due to decreased VAT. The highly efficacious results seen with NY0128 in vivo further compliments our in vitro data, where NY0128 is the more potent and efficacious compound. The concentrations found effective in vitro are similar to the brain exposure of NY0128. This further suggests that NY0128 is acting on the centrally expressed NMUR2 to promote fat loss and decrease food intake. Taken together, these data suggest that NY0116 and NY0128 are promising tool compounds in targeting NMUR2 as a pharmacotherapy for obesity.

| CONCLUSION
We provide strong evidence that two small-molecule agonists for NMUR2 activate both G i -mediated cAMP inhibition and G q -mediated calcium signaling, achieve appreciable brain levels after a single subcutaneous dose, and decrease high-fat diet intake in rats. Although NY0116 and NY0128 are not selective for NMUR2 (also act on NMUR1), our laboratory and others have demonstrated that NMUR2 in particular is a key regulator of feeding behavior, 7,19,30 and therefore the likely target for these agonists. Furthermore, despite being fed a high-fat diet, repeated administration of NY0128 reduces body weight, VAT, and cholesterol in obese mice. Although the NMUR2 agonists presented here require further optimization, these small molecules demonstrate promising effects as lead drug candidates on feeding, fat content and body weight composition. Taken together, this work indicates NMUR2 agonism has pharmacotherapeutic potential for addressing metabolic disorders, such as obesity.

ACKNOWLEDG EMENTS
We thank Dr. David Konkel for critically editing the manuscript and Caitlin R. Benzon