Hindbrain octadecaneuropeptide gliotransmission as a therapeutic target for energy balance control without nausea or emesis

Abstract

Glia play a dynamic role in central nutrient sensing and appetite regulation yet represent underexplored targets in treating dysregulated energy balance. Glia within the dorsal vagal complex of the hindbrain synthesize the anorexigenic peptide octadecaneuropeptide (ODN), the influence and therapeutic potential of which remain to be explored. We demonstrate that hindbrain-targeted ODN induced weight loss, counteracted glucoprivation, and improved glucose clearance in rats. Furthermore, blocking central ODN signaling attenuated the anorectic response to GLP-1R agonists in rats. Peripheral administration of an ODN derivative, TDN, improved insulin sensitivity assessed by hyperinsulinemic-euglycemic clamp in obese mice and induced weight loss without pica behavior, a proxy for nausea in rats, or emesis in the musk shrew, a vomiting mammalian model. Central ODN and TDN treatment in rats was not accompanied by changes in core body temperature, physical activity, or heart rate. This work highlights hindbrain ODN signaling as an important modulator of energy balance and demonstrates the potential for targeting this gliopeptide system to treat dysregulated feeding and metabolic activity without side effects.

Editor’s summary

Octadecaneuropeptide (ODN) is an anorexigenic peptide synthesized by glia in the hindbrain. Geisler et al. show that, in preclinical in vivo models, ODN improves glucose homeostasis and dose dependently reduces food intake without inducing nausea or vomiting. Similar beneficial results were observed through peripheral administration of a truncated version of ODN called TDN. This study indicates that ODN derivatives may be valuable for targeting weight and glycemic control without eliciting the common treatment side effects seen with GLP-1 receptor agonists. —Catherine Charneski

INTRODUCTION

Obesity and type 2 diabetes mellitus (T2D) are two of the most common and costly disorders worldwide today (1-3). Both disorders frequently coexist and are characterized by an upward resetting and homeostatic defense of the disturbed physiological variable that most defines the respective disease state: fat mass in obesity and blood glucose concentrations in T2D (4, 5). Central sensing of a surplus or deficit of energy elicits compensatory responses that promote homeostasis of both circulating and stored fuel, but this flexibility is perturbed in obesity and T2D (6-8). Two key central energy balance regulating centers, the dorsal vagal complex (DVC) of the hindbrain and the hypothalamus, are sensitive to gut-derived and circulating nutrient signals that engage integrated neurocircuitry governing energy and glucose homeostasis (9-12). Still, the mechanisms underlying the dysregulation of central sensing and response to changes in energy metabolism in obesity are incompletely understood.

Glial populations, notably astrocytes and tanycytes, which interface at the blood-brain and brain–cerebral spinal fluid barriers to monitor systemic energy status, have become increasingly recognized for their critical roles influencing appetite, body weight, nutrient sensing, and metabolic coordination (13, 14). Astrocytes are sensitive to glucose and mediate the counterregulatory response to glucose deprivation (15-17). Furthermore, depolarization of hindbrain astrocytes potently suppresses food intake, improves glucoregulatory responses, and is critical in the energy adaptation responses to high-fat diet (HFD) feeding (18, 19). Similarly, tanycytes actively sense glucose as well as lipids and amino acids and participate in appetite, body weight, and glycemic regulation (20-25). Recently, tanycytes have been shown to play a role in the central transport of insulin, leptin, and the glucagon-like peptide-1 receptor agonist (GLP-1RA) liraglutide (26-28), highlighting their importance in controlling access of key energy regulating hormones to their central sites of action.

Understanding how glia communicate nutrient-state–dependent information to broader neural networks will be vital in developing therapies that improve energy balance homeostasis in obesity. The endozepine octadecaneuropeptide (ODN) is an anorexigenic gliopeptide expressed densely by astrocyte and tanycyte populations in the hypothalamus and DVC around the ventricles and in circumventricular organs, giving it a key location to respond to changes in energy status (29-31). Expression of ODN’s endogenous precursor protein diazepam binding inhibitor (DBI) is nutrient state dependent, showing down-regulation during fasting and up-regulation upon refeeding (30, 32). Additionally, third ventricle tanycytic DBI mRNA expression and ODN release are directly stimulated by glucose, play a role in hypothalamic glucose sensing, and improve glycemic responses (31). ODN signaling in the hypothalamus has been more thoroughly explored (30-40), whereas hindbrain ODN action has received far less attention (29, 30).

Given the essential role that the DVC plays in metabolic and energy balance regulation (10, 12, 41-47), we set out to characterize the appetite, glucose sensing, and glycemic regulatory responses to hindbrain-targeted ODN. We further investigated whether DBI expression is regulated by GLP-1RAs and whether the anorectic effects of GLP-1RAs are partially mediated by downstream recruitment of ODN signaling. Additionally, we evaluated the therapeutic viability of targeting ODN signaling by assessing the impact of central and peripheral treatment with a modified ODN derivative, tridecaneuropeptide (TDN), on body weight, glycemic control, and the side effect profile as a potential compound for further development for the treatment for obesity and T2D. This work highlights that hindbrain endozepine signaling is downstream of GLP-1RAs and regulates systemic energy balance and that the ODN system is an attractive therapeutic target for weight and glycemic control without nausea or emesis.

RESULTS

Hindbrain ODN differentially regulates food intake and meal patterns in lean and obese rats

We assessed food intake and meal pattern responses to acute hindbrain-targeted ODN treatment on different diet conditions to evaluate which components of feeding microstructure were most responsible for ODN-induced anorexia. To compare the anorectic effect of hindbrain ODN signaling in the lean and obese states, we tested several doses of fourth intracerebroventricular (ICV)–injected ODN (0.2, 2, and 20 μg) in chow- and HFD-maintained rats. ODN acutely suppressed food intake only at 1 hour without affecting 24-hour body weight in lean rats (Fig. 1, A and B). Given the early but transient suppression of food intake, we further analyzed the impact of ODN on the first eating events of the dark cycle. ODN did not significantly delay consumption of the first meal, but the size of the first meal was reduced by 20 μg of ODN (Fig. 1, C and D). The time to consume the first bout of food was delayed after 2 μg of ODN, whereas the first bout size was unaffected by treatment (Fig. 1, E and F). Notably, in HFD-maintained rats, the first ODN dose dependently suppressed food intake 12 hours after injections, with 20 μg of ODN also decreasing 24-hour food intake and body weight (Fig. 1, G and H). Given the delayed onset of ODN in obese rats, we further analyzed the average meal duration and size in the latter half of the dark cycle and the light cycle. The highest dose of ODN decreased meal duration, but a decrease in meal size was not significant (P = 0.06; Fig. 1, I and J). Meal duration and size during the light cycle were not altered by ODN treatment (Fig. 1, K and L). The variability of responses within the treatment group was quite high, and some measures, such as the meal size during the light cycle (Fig. 1L), appeared to display a bimodal distribution. Full meal pattern data including gram intake, meal number, time in meals, meal duration, and meal size for intervals 1, 6, 12, 6 to 12, and 12 to 24 hours after injection are described for lean and obese rats (tables S1 and S2). Considering the diet-specific responses to ODN, we wondered how chow and HFD intake would be affected by exogenous delivery of the parent peptide DBI. Fourth ICV administration of recombinant DBI protein suppressed chow intake at 6 and 12 hours postinjection without affecting HFD intake or 24-hour body weight in either diet group (Fig. 1, M and N). These data support that the anorectic response to hindbrain-targeted ODN and DBI treatment differs in lean chow-fed and HFD-induced obese rats.

Fig. 1. Hindbrain ODN differentially regulates food intake and meal patterns in lean and obese rats.

Dose response of chow- and HFD-maintained rats injected at the fourth ventricle (4V injected) with ODN (0.2, 2, and 20 μg/2 μl) or vehicle. Cumulative chow intake (A), 24-hour body weight change (B), latency (C) and size (D) of the first meal, and latency (E) and size (F) of the first bout in chow-fed rats. Cumulative HFD intake (G), 24-hour body weight change (H), meal duration (I), and size (J) 6 to 12 hours postinjection, as well as meal duration (K) and size (L) 12 to 24 hours postinjection in HFD-fed rats. Cumulative intake (M) and 24-hour body weight change (N) in chow- and HFD-maintained rats 4V injected with recombinant DBI protein (RecDBI; 2 μg/2 μl) or vehicle. Data are means ± SEM. Each experiment was within subjects; [(A) to (F)] n = 9 per treatment, [(G) to (L)] n = 9 to 11 per treatment, and [(M) and (N)] n = 9 to 11 per treatment. Data were analyzed by one-way [(B) to (F), (H) to (L), and (N)] or two-way [(A) and (G)] ANOVA with Dunnett’s post hoc or by three-way ANOVA (M) with Bonferroni post hoc. NS, not significant; Veh, vehicle; h, hour.

Diverse hindbrain glial populations express DBI transcripts

To understand the role of endozepine signaling in energy balance control under physiologic and pathologic states, a comprehensive picture of the cell populations that express DBI is critical. Centrally, DBI and ODN-like immunoreactivity (IL) and DBI transcripts are most often visualized exclusively in glia (29, 30, 48-51), although, in some instances, their expression has been ascribed to neurons (52-54). In the hindbrain, ODN-IL has been found to colocalize with ependymal tanycytes and astrocytes (29, 30). We similarly visualized DBI protein immunoreactivity in vimentin-positive tanycytes and glial fibrillary acidic protein (GFAP)–positive astrocytes of the DVC (fig. S1A) and further profiled DVC DBI expression by DVC subnuclei. Most DVC DBI expression was present in the area postrema (AP; 56.4 ± 7.4%), with 36.5 ± 7.6% found in the nucleus tractus solitarius (NTS) and 7.1 ± 0.7% in the subpostrema border between the AP and NTS (fig. S1B). Within each subnucleus, DBI immunofluorescence labeled around two-thirds of both the AP (66.7 ± 6.4%) and subpostrema border (63.7 ± 7.5%), whereas only 15.9 ± 5.3% of the NTS showed DBI staining (fig. S1C). Thus, more than half of the AP by area expressed DBI, which accounted for the largest portion of total hindbrain DBI.

To assess whether the glial cell populations expressing DBI differed by DVC subnuclei, we analyzed the degree of DBI staining colocalized with vimentin, GFAP, and endothelial cell marker RECA in the AP, NTS, and subpostrema border (fig. S1, D to F). Notably, 73.7 ± 2.1% of DBI expression in the AP did not colocalize with any of these markers, indicating that, in the DVC subnuclei with most DBI expression, most DBI mRNA may not be found in astrocytes or tanycytes (fig. S1D). In the subpostrema border, which regulates the transport of molecules from the AP to the NTS, most (94.6 ± 8.2%) DBI was expressed in astrocytes and tanycytes, which serve the critical role of gating entry across the blood brain barrier (fig. S1F) (55, 56). These data support that the distribution of DBI expression and, thereby, functioning is specialized to the role of each subnuclei. GFAP is classically used as an astrocyte marker but can be present in tanycytes and radial glia, whereas vimentin can be found not only in tanycytes but also in astrocytes, radial glia, fibroblasts, and endothelial cells (57). Accordingly, to achieve a more precise characterization of DVC DBI expression patterns, we searched for DBI transcripts by cell cluster in our previously described single-nucleus transcriptomic dataset of combined AP/NTS tissues (58). DBI transcripts were most abundant within vimentin-positive tanycytes and Pdgfd-positive radial-glial–like cell clusters (fig. S1, G to I). Additional glial populations, including oligodendrocytes, oligodendrocyte precursor cells, vascular and leptomeningeal cells, and microglia, all expressed notable DBI transcripts and may account for the majority of AP DBI-expressing cells (fig. S1, G to I). These data support that DVC DBI gene expression is found in all glial populations examined with regional differences in the relative proportion of cell types expressing DBI, suggestive of subnuclei-specific functions of DBI signaling.

Hindbrain ODN suppresses the hyperphagic and hyperglycemic response to glucoprivation

Healthy systemic glucoregulation depends on the brain’s ability to accurately detect changes in circulating glucose concentration and coordinate peripheral metabolic activity to maintain euglycemia. Glucoprivic agents like 5-thio-d-glucose (5-TG) inhibit glycolysis and simulate hypoglycemia, acutely stimulating feeding and the release of counterregulatory hormones including corticosterone and glucagon (59, 60). Lateral ventricle–administered ODN has been shown to block the hyperphagic response to central glucoprivation, with the hypothalamus as the suggested site of action (31). However, the hindbrain is also sensitive to glucoprivation and is required to mount a full counterregulatory response (61, 62). We set out to determine whether ODN signaling in the hindbrain could suppress counterregulation induced by glucoprivation in the DVCs of lean rats. Fourth ICV injection of 5-TG (210 μg) at the onset of the dark cycle increased food intake at 1 and 24 hours after injection compared with vehicle, and ODN pretreatment suppressed 5-TG–mediated hyperphagia at the 4-hour time point (Fig. 2A). ODN pretreatment also blunted 5-TG–induced hyperglycemia 1 hour after injection, whereas body weight was not affected 24 hours after any treatment (Fig. 2, B and C).

Fig. 2. Hindbrain ODN suppresses the hyperphagic and hyperglycemic response to glucoprivation in rats.

Cumulative chow intake (A), 1-hour blood glucose (B), 24-hour body weight change (C), and 30-min plasma corticosterone (D), glucagon (E), and free fatty acids (F) in chow-fed rats 4V injected with ODN (20 μg/2 μl) or vehicle before 5-TG (210 μg/2 μl) or vehicle. Plasma glucagon in chow-fed rats 4V injected with ODN (20 μg/2 μl) or vehicle 15 min before intraperitoneal insulin [0.75 U/kg; (G)]. One-hour blood glucose in chow-fed rats after 4V injection of ODN (200 μg/2 μl), AntOP (100 μg/2 μl), or vehicle (H). Data are means ± SEM. Bars that do not share a common letter (a, b, and c) within a time point differ significantly (P ≤ 0.05). Each experiment was within subjects; [(A) to (C)] n = 11 or 12 per treatment, [(D) to (F)] n = 15 per treatment, (G) n = 15 per treatment, and (H) n = 14 per treatment. Data were analyzed by two-way ANOVA with Tukey’s post hoc [(A), (B), and (H)] or Šidák post hoc [(C) to (G)].

The ability of hyperglycemia to induce satiation has been inconsistently reported (63-65). Still, previous results report that pharmacological antagonism of the ODN G protein–coupled receptor (GPCR) blocks the hypophagic response to lateral ventricle–injected glucose, and we attempted to replicate these findings at the level of the hindbrain (31). We did not observe any changes in food intake after fourth ICV injection of glucose or the ODN GPCR antagonist (AntOP). Although blood glucose concentrations were lower 30 min after fourth ICV glucose injection relative to controls, this was not so in rats that received AntOP pretreatment before glucose (fig. S2, A to C).

To investigate the mechanism by which ODN interrupted the hyperglycemic response to hindbrain glucoprivation, we once more injected ODN before 5-TG at dark cycle onset, again observing an attenuation of 5-TG–induced hyperglycemia (fig. S2D), but this time, we collected plasma 30 min after drug treatments to assay circulating counterregulatory hormone concentrations. ODN pretreatment did not lessen the 5-TG–induced rise in corticosterone but did attenuate 5-TG–stimulated plasma glucagon (Fig. 2, D and E). Glucoprivic agents stimulate circulating free fatty acids (66), which were also suppressed by hindbrain ODN pretreatment (Fig. 2F). Furthermore, fourth ICV-injected ODN lowered plasma glucagon concentrations 30 min after intraperitoneal insulin (Fig. 2G), supporting that hindbrain ODN signaling modulates peripheral glucagon release in response to local glucoprivation induced by glucose analogs and insulin-induced systemic hypoglycemia. Given that hindbrain glucose sensing regulates blood glucose concentrations (12, 67), we next tested the possible regulation of basal glycemia by hindbrain ODN signaling. Under baseline-fed conditions, fourth ICV–injected ODN lowered blood glucose concentrations over the 1-hour testing period compared with vehicle (Fig. 2H). Conversely, fourth ICV administration of an antagonist to the ODN GPCR (AntOP) increased blood glucose over this period, significantly above vehicle at 10 min after injection (Fig. 2H). These findings suggest that ODN disrupts the monitoring of hypoglycemia in the hindbrain and, furthermore, that basal hindbrain endozepine tone serves to lower glycemia.

Hindbrain ODN signaling improves systemic glucose clearance without the incretin effect

Given that ODN signaling reduced blood glucose concentrations under glucoprivation and at baseline, we set out to investigate whether hindbrain ODN similarly could lower glycemia during a glucose challenge. In lean rats, we administered two doses of fourth ICV–injected ODN (20 or 200 μg) before an oral glucose tolerance test (OGTT) and found that the higher dose blunted the spike in blood glucose 15 and 30 min after glucose gavage and reduced the overall area under the curve (AUC) without enhancing glucose-stimulated plasma insulin (Fig. 3, A to C). Glucose tolerance was also markedly improved in diet-induced obese (DIO)–maintained rats by 200 μg of hindbrain-targeted ODN without augmenting glucose-stimulated insulin values (Fig. 3, D to F). Fourth ICV–injected ODN (200 μg) also increased glucose clearance during an insulin tolerance test in lean and DIO rats (Fig. 3, G to J). To test whether blocking endogenous hindbrain endozepine signaling would worsen insulin sensitivity, we pretreated with the ODN GPCR antagonist AntOP via fourth ICV injection and observed higher blood glucose concentrations 30 min after intraperitoneal insulin in DIO rats as well as an overall treatment effect of AntOP to elevate blood glucose during the insulin tolerance tests in lean and DIO rats (P = 0.014; Fig. 3, K to M). Thus, hindbrain ODN signaling bidirectionally regulates systemic glucose homeostasis with agonism improving and ODN-GPCR antagonism worsening glucose disposal.

Fig. 3. Hindbrain ODN signaling improves systemic glucose clearance in rats without the incretin effect.

OGTT (A), OGTT AUC (B), and glucose-stimulated insulin (C) in chow-fed rats 4V injected with ODN (20 or 200 μg/2 μl) or vehicle. OGTT (D), OGTT AUC (E), and glucose-stimulated insulin (F) in HFD-fed rats 4V injected with ODN (200 μg/2 μl) or vehicle. Insulin tolerance test (ITT) (G) and ITT AUC (H) in chow-fed rats and ITT (I) and ITT AUC (J) in HFD-fed rats 4V injected with ODN (20 or 200 μg/2 μl) or vehicle. ITT in chow-fed (K) and HFD-fed (L) rats and ITT AUC (M) after 4V injection with AntOP (200 μg/2 μl) or vehicle. Data are means ± SEM. Each experiment was within subjects; [(A) to (C)] n = 12 or 13 per treatment, [(D) to (F)] n = 11 per treatment, [(G) and (H)] n = 9 to 11 per treatment, [(I) and (J)] n = 11 to 20 per treatment, and [(K) and (M)] n = 8 to 10 per treatment. Data were analyzed by one-way ANOVA with Dunnett’s post hoc [(B), (C), (H), and (J)], by two-way ANOVA with Dunnett’s post hoc [(A), (G), and (I)] or Šidák post hoc [(D) and (K) to (M)], or by paired two-tailed t test [(E) and (F)].

Hindbrain DBI mRNA and protein expression are differentially regulated by postprandial signals

The above findings that ODN attenuates induction of the counter-regulatory response triggered by hindbrain 5-TG suggests that ODN may serve as a signal reflecting eu- or hyperglycemic status. DBI mRNA expression is strongly regulated by nutritional status, and fasting down-regulates, whereas refeeding up-regulates, DBI transcripts in the hypothalamus and DVC (30-32). Centrally administered glucose is also sufficient to restore hypothalamic fasting DBI expression to fed-state values, with the largest effect observed immediately around the third ventricle (31). Additionally, intraperitoneal insulin, but not leptin, in the fasted state has been shown to up-regulate DBI expression around the lateral ventricle (32). Yet, the influence of other postprandial hormones on DBI expression has not been explored. To examine whether hindbrain DBI expression is regulated directly by glycemia and investigate potential regulation by GLP-1RA, we analyzed the amount of DBI transcript expression in rats that were ad libitum chow-fed, fasted, or fasted plus received a fourth ICV injection of the GLP-1RA exendin-4 (Ex-4; 0.3 μg) or glucose (5.5 M). We analyzed the percent fluorescent area of DVC subnuclei expressing DBI transcripts with mild, moderate, or strong intensity at the obex, AP, and fourth ventricle coronal levels (representative annotations depicted in fig. S3A). Starting at the level of the obex caudal to the AP, DBI mRNA expression in the NTS was unaffected by treatment (Fig. 4A). Meanwhile, in the NTS at the AP and fourth ventricle levels, Ex-4 treatment elevated mild DBI expression above fasted control and fed rats (Fig. 4, B and C). DBI mRNA expression in the AP was not affected by treatment, but glucose increased the percentage of moderate DBI expression in the subpostrema border above fasted control and fed rats (Fig. 4, D and E). Along the fourth ventricle border, fasting down-regulated, whereas Ex-4 and glucose up-regulated, mild-intensity DBI expression compared with fasted controls (Fig. 4F). Thus, the regulation of DBI displayed regionally specific differences with fasted DBI expression being most strongly up-regulated by Ex-4 in the NTS and by glucose in the subpostrema border, whereas DBI expression in the fourth ventricle border was sensitive to both postprandial factors. Secondary analysis of the number of DBI transcripts per cell revealed a similar but slightly different pattern of expression (fig. S3, B to G; representative image analysis in Fig. 4G). DBI transcript copies per cell were increased by glucose over fasted controls in the NTS at the AP level and by glucose over fasted control and fed rats in the subpostrema border (fig. S3, C and F).

Fig. 4. Rat hindbrain DBI mRNA and protein expression are differentially regulated by postprandial signals.

Rats were ad libitum fed and 4V vehicle-injected (Ad Lib Veh) or 24 hours fasted and 4V injected with vehicle (Fast Veh), Ex-4 (Fast Ex-4; 0.3 μg/2 μl), or glucose (Fast Glucose; 2 μl of 5.5 M) 90 min before sacrifice. Mild, moderate, and strong fluorescent signals for DBI mRNA in the NTS at the obex (A), AP (B), and 4V (C) level and in the subpostrema border (D), AP (E), and 4V border (F). Mild, moderate, and strong fluorescent signals for DBI protein in the NTS at the obex (H), AP (I), and 4V (J) level and in the subpostrema border (K), AP (L), and 4V border (M). Representative DBI mRNA (G) and protein (N) analysis images. Data are means ± SEM. Each experiment was between subjects; [(A) to (G)] n = 5 to 9 per treatment and [(H) to (M)] n = 3 or 4 per treatment. Symbols * or # denotes significance between treatment groups at mild or moderate fluorescent intensity, respectively. Fluorescent signal was stratified by mild, moderate, or strong intensity in HALO analysis software and universally applied to all images. Data were analyzed by two-way ANOVA with Tukey’s post hoc [(A) to (M)].

We next examined the impact of fasting and Ex-4 treatment in the fasted state on DBI protein expression in the same DVC subnuclei, again analyzing by the percentage of area labeled with mild, moderate, or strong fluorescence (representative image analysis, Fig. 4N). This time, we found the opposite result, with fasting generally increasing and Ex-4 generally decreasing DBI protein back to fed values. Whereas the NTS at the obex was again unaffected by treatment, Ex-4 lowered mild DBI expression relative to fasted control in the NTS at the AP and fourth ventricle level (Fig. 4, H to J). The percentage of AP area showing modest DBI expression was increased by fasting and diminished by Ex-4 treatment in the fasted state (Fig. 4K).

ODN signaling contributes to the anorectic effect of GLP-1RAs

Having observed that, in select DVC subnuclei including the NTS and fourth ventricle border, Ex-4 treatment up-regulated DBI mRNA expression, we hypothesized that the anorectic effect of ODN may contribute to the food intake suppression and body weight loss induced by GLP-1RAs. We applied two strategies to block the effects of ODN, namely, antibody (Ab)–mediated neutralization of DBI and AntOP-mediated antagonism of the ODN GPCR, and determined the effect of GLP-1RAs in the absence of ODN signaling in chow and DIO rats. As expected, fourth ICV Ex-4 (0.3 μg) injection robustly suppressed food intake in chow rats at 6 and 24 hours postinjection, whereas fourth ICV pretreatment with the DBI Ab (3 μg) attenuated Ex-4–induced anorexia at 24 hours (Fig. 5A). A major side effect that limits the efficacy of direct GLP-1R targeting therapeutics is gastrointestinal distress including nausea and vomiting (68). To see how blocking ODN signaling may affect GLP-1RA–induced malaise, we simultaneously assessed pica behavior as kaolin intake, a well validated proxy for nausea in rodents (69). Ex-4 increased 1-, 3-, and 24-hour kaolin intake, which was not affected by Ab pretreatment (Fig. 5B). Ex-4–induced 24-hour body weight loss was 40% less with Ab pretreatment, although not significantly different (P ≥ 0.05) from Ex-4 alone (Fig. 5C). In DIO rats, Ex-4–mediated HFD intake suppression, kaolin intake, and body weight loss were not altered by Ab pretreatment (fig. S4, A to C). We then repeated the study using hindbrain-targeted AntOP (200 μg) pretreatment before fourth ICV Ex-4 and found that, in chow rats, Ex-4 treatment alone suppressed food intake at 3, 6, and 24 hours, whereas the intake of rats pretreated with AntOP before Ex-4 was never significantly different from that of vehicle controls and showed reduced Ex-4–induced intake suppression (by 22%) at 24 hours (Fig. 5D). AntOP pretreatment did not affect Ex-4–induced kaolin intake or body weight loss, although AntOP treatment before Ex-4 reduced Ex-4–mediated weight loss by 64%, which was not significantly different from that of vehicle controls (Fig. 5, E and F). In DIO rats, AntOP pretreatment attenuated Ex-4–induced anorexia at 24 hours without blunting the induction of kaolin intake, but the reduction in Ex-4–mediated weight loss by 32% was nonsignificant (fig. S4, D to F). Notably, AntOP treatment alone reduced 24-hour HFD intake and body weight loss compared with vehicle controls, despite lessening the anorectic effect of Ex-4 (fig. S4, D and F).

Fig. 5. ODN signaling contributes to the anorectic effect of GLP-1RAs in rats.

Cumulative chow intake (A), kaolin intake (B), and 24-hour body weight (BW) change (C) in rats 4V injected with vehicle or a DBI neutralizing Ab (3 μg/3 μl) before 4V treatment with vehicle or Ex-4 (0.3 μg/2 μl). Cumulative chow intake (D), kaolin intake (E), and 24-hour BW change (F) in rats 4V injected with vehicle or AntOP (200 μg/2 μl) before 4V vehicle or Ex-4 (0.3 μg/2 μl). Cumulative HFD intake (G), kaolin intake (H), and 24-hour BW change (I) in rats lateral ventricle injected with vehicle or AntOP (100 μg/2 μl) before intraperitoneal vehicle or liraglutide (50 μg/kg). Twenty-four-hour HFD intake (J), kaolin intake (K), and BW change (L) in rats 4V injected with vehicle or ODN (100 μg/2 μl) before intraperitoneal vehicle or liraglutide (25 μg/kg). Data are means ± SEM. Bars that do not share a common letter (a, b, and c) within a time point differ significantly (P ≤ 0.05). Data in (A) to (I) are within subjects, whereas data in (J) to (L) are between subjects. [(A) to (C)] n = 9 to 11 per treatment, [(D) to (F)], n = 9 to 11 per treatment, [(G) to (I)] n = 12 per treatment, and [(J) to (L)] n = 6 or 7 per treatment. Data were analyzed by two-way ANOVA with Tukey’s post hoc [(A) to (I)] or Šidák post hoc [(J) to (L)]. Lirag, liraglutide.

To next determine how blocking central ODN-GPCR signaling would affect peripheral delivery of a GLP-1RA, we tested pretreatment of AntOP (100 μg), this time administered at the lateral ventricle to bathe more of the brain, against intraperitoneal liraglutide. In chow rats, there was no effect of AntOP pretreatment on the anorectic or kaolin response to liraglutide (fig. S4, G to I). In DIO rats, liraglutide-induced suppression of HFD intake was blunted by AntOP pretreatment compared with vehicle pretreatment at 1 and 3 hours, whereas kaolin intake and body weight loss were unaffected (Fig. 5, G to I). To see whether ODN signaling could augment the anorectic response to GLP-1RA, we injected ODN (100 μg) fourth ICV before intraperitoneal liraglutide (25 μg/kg). ODN pretreatment enhanced liraglutide-induced 24-hour food intake suppression without altering kaolin intake or weight loss (Fig. 5, J to L). Together, these findings support that ODN may be downstream of GLP-1R agonism because blocking ODN signaling partially attenuated the anorectic effect of centrally and peripherally delivered GLP-1RA. Yet, ODN can still further suppress food intake by GLP-1RA, suggesting that GLP-1RAs may not maximally engage ODN signaling.

ODN and a TDN derivative do not induce nausea and emesis or affect heart rate, body temperature, and locomotion

The above findings show that blocking ODN signaling was unable to mitigate the nausea induced by GLP-1RAs, suggesting that if ODN is downstream of the GLP-1R, then it may be engaged through a distinct mechanism that does not overlap with the nausea-causing pathway. To better understand whether ODN signaling itself triggers gastrointestinal malaise, we turned to the musk shrew, a vomiting mammal that shows emetic sensitivity to GLP-1RA (70). Intraperitoneal ODN (5 mg/kg) suppressed food intake at 6 and 24 hours after injection without affecting body weight at 24 hours and caused no emetic episodes (Fig. 6, A to C). Given that ODN showed great therapeutic potential for improving glycemic and weight control in obesity and type 2 diabetes without emesis, we pursued creating an analog of ODN to assess its impact on energy balance control and for confirmation that targeting endozepine signaling is a worthwhile strategy for further compound development. We synthesized an ODN derivative with five amino acids cleaved off the N terminus appropriately named TDN (Fig. 6D), which has not been reported to be detected in vivo. To first determine that TDN did not cause emesis, we tested a peripheral injection of TDN (5 mg/kg) in shrews, which suppressed 24-hour food intake without altering 24-hour weight change, and again did not show any induction of emesis (Fig. 6, E to G). Fourth ICV injection of ODN or TDN (20 μg) also did not cause kaolin intake in rats (fig. S5A), supporting that ODN and TDN do not cause nausea and emesis as seen with GLP-1RAs. To assess other common off-target concerns with preclinical pharmacotherapies, we implanted telemetry devices to assess core body temperature, physical activity, and heart rate. Lateral ventricle ODN and TDN (20 and 100 μg) injection did not alter any of these physiological parameters averaged across the dark cycle, light cycle, and 24-hour day (Fig. 6, H to J). Data by hour during the 24 hours after treatment are shown in fig. S5 (C to H).

Fig. 6. ODN and TDN do not induce emesis in shrews or affect heart rate, body temperature, and locomotion in rats.

Cumulative food intake (A), 24-hour body weight change (B), and emetic events (C) in musk shrews intraperitoneally injected with vehicle or ODN (5 mg/kg). Peptide sequences of ODN, AntOP, and TDN (D). Cumulative food intake (E), 24-hour body weight change (F), and emetic events (G) in musk shrews intraperitoneally injected with vehicle or TDN (5 mg/kg). Average body temperature (H), locomotion (I), and heart rate (J) during the dark cycle (DC), light cycle (LC), or over 24 hours in chow-fed rats injected at the lateral ventricle with vehicle, ODN (20 or 100 μg/2 μl), or TDN (20 or 100 μg/2 μl). Data are means ± SEM. Each experiment was within subjects; [(A) to (C)] n = 9 per treatment, [(D) to (F)] n = 9 per treatment, and [(G) to (I)] n = 10 to 14 per treatment. Data were analyzed by two-way ANOVA with Šidák post hoc [(A) and (E)] or Dunnett’s post hoc [(H) to (J)] or by paired two-tailed t test [(B), (C), (F), and (G)].

Peripheral TDN improves glycemia and weight loss, mimicking central action

We next assessed the effect of central TDN on appetite and glycemic regulation. Lateral ventricle injection of TDN (200 μg) suppressed 24-hour food intake without an impact on body weight (Fig. 7, A and B). Fourth ICV injection of TDN had an overall treatment effect (P = 0.0064) to improve glucose clearance during an OGTT and lowered the OGTT AUC without heightening glucose-stimulated plasma insulin concentrations (Fig. 7, C to E). Insulin sensitivity during an insulin tolerance test was also improved by fourth ICV TDN (Fig. 7, F and G). Thus, centrally administered TDN, like ODN, suppresses food intake and improves glucose homeostasis. However, peripherally administered ODN (5 mg/kg) did not alter 24-hour food intake or body weight in lean or DIO mice, whereas TDN (5 mg/kg) induced hypophagia and weight loss at 24 hours in DIO mice with no effect in lean mice (fig. S6, A to H). These data might suggest that TDN has better brain penetrance than ODN or a different pharmacokinetic profile. In rat serum, ODN had an average half-life of 12.4 min compared with TDN with an average half-life of 21.2 min (fig. S6, I and J). To investigate the longitudinal impact of repeated TDN delivery and ensure that peripheral administration retains its satiation effects, we measured daily food intake and body weight in DIO mice for 9 days of daily intraperitoneal TDN injections (5 mg/kg). TDN suppressed HFD intake immediately on day 1, and this effect did not wane over the 9 days, causing cumulative intake to be reduced on days 3 to 9 (Fig. 7H and fig. S7A). Similarly, daily body weight loss persisted through the 9 days, reducing cumulative weight change on days 4 to 9 and resulting in a 4.7% body weight loss over 9 days (Fig. 7I and fig. S7B).

Fig. 7. Central TDN in rats and peripheral TDN in mice improves weight and glycemic control.

Cumulative HFD intake (A) and 24-hour body weight change (B) in rats injected at the lateral ventricle with TDN (200 μg/2 μl) or vehicle. OGTT (C), OGTT AUC (D), glucose-stimulated insulin (E), ITT (F), and ITT AUC (G) in HFD-fed rats 4V injected with TND (200 μg/2 μl) or vehicle. Cumulative HFD intake (H) and body weight change (I) in mice intraperitoneally injected daily with TDN (5 mg/kg) or vehicle. Blood glucose concentration (J), glucose infusion rate (K), endogenous glucose appearance [Ra; (L)], and % suppression Ra (M) in HFD-fed mice injected intraperitoneally with TDN (5 mg/kg) or vehicle before insulin infusion during a hyperinsulinemic-euglycemic clamp. Data are means ± SEM. Data in (B) to (H) are within subjects, and data in (I) to (N) are between subjects. [(B) and (C)] n = 6 per treatment, [(D) to (F)] n = 11 per treatment, [(G) and (H)] n = 9 per treatment, [(I) and (J)] n = 6 per treatment, and [(K) to (N)] n = 5 per treatment. Data were analyzed by two-way ANOVA with Šidák post hoc [(B), (D), (G), and (I) to (M)] or by paired [(C), (E), (F), and (H)] or unpaired (N) two-tailed t test.

To determine how peripheral TDN affects glucose homeostasis, we performed hyperinsulinemic-euglycemic clamps in DIO mice that received a single intraperitoneal injection of TDN (5 mg/kg) or vehicle before insulin infusion during the clamp. Starting body weight and blood glucose concentrations during the clamp were not different by treatment group (Fig. 7J and fig. S7C). The mice were highly insulin resistant, and plasma insulin concentrations during the clamp period only tended to be elevated compared with baseline (P = 0.067; fig. S7D). Still, the rate of glucose disappearance (Rd) was significantly elevated during the clamp period overall (P = 0.018; fig. S7E). Despite a relatively mild hyperinsulinemic clamp state, TDN treatment increased the glucose infusion rate during the clamp (Fig. 7K). Only TDN treatment lowered the rate of endogenous glucose appearance (Ra), suppressing Ra relative to baseline by 57% compared with 18% in vehicle-treated mice (Fig. 7, L and M). These data support that systemic administration of TDN favorably modulates glucose homeostasis by enhancing insulin-stimulated glucose disposal and inhibiting hepatic glucose production. Future studies will validate whether systemically administered ODN analogs replicate the effects of endogenous ODN signaling, for which this manuscript supports a key role of DVC-produced ODN in mediating appetite and metabolic responses to sensing of postprandial nutrient signals (fig. S8).

DISCUSSION

Hindbrain glia have been widely implicated in glucose and glucoprivation detection, counterregulatory hormone responsiveness, and appetite suppression (15-19, 71-73). Here, we demonstrate that hindbrain DBI transcription is up-regulated by glucose and Ex-4 in the fasted state, whereas hindbrain ODN treatment blunts the response to glucoprivation, improves glucose clearance, and reduces food intake. These collective findings support hindbrain ODN as a glucosensitive signal that communicates postprandial increases in fuel and coordinates the transition from an energy-depleted state to an energysurplus state by facilitating suppression of glucagon release and hepatic glucose production while promoting glucose disposal. Additionally, we report that ODN is a likely downstream mediator of GLP-1R activation and GLP-1RA anorectic effect. Taking therapeutic advantage of these actions, we determined that our ODN analog TDN effectively promoted acute and chronic weight loss and improved glucose tolerance after systemic administration without producing nausea/emesis or heart rate disturbances. These preclinical findings support further investigation of ODN signaling as a viable pharmacologic target for weight and glycemic control in obesity and T2D while addressing important gastrointestinal side effects that often exist alongside therapeutic efficacy.

The magnitude and time course of food intake suppression with hindbrain-targeted ODN treatment differed in chow-fed lean and HFD-fed obese rats, with DIO rats ultimately showing a stronger anorectic response and body weight loss at 24 hours. In chow-fed rats, ODN transiently suppressed food intake at 1 hour through reduction in meal size. Similarly, reductions in meal size and meal number have been reported after lateral ventricle ODN injection in chow-fed mice (30). Conversely, in DIO rats, we found that ODN had a delayed but more robust anorectic effect primarily driven by reduced meal size and meal duration in the latter half of the dark cycle. Likewise, peripherally administered TDN induced hypophagia only in DIO but not lean mice at 24 hours. These findings are consistent with the literature that ODN derivatives are more anorectic in obese models. Acute and chronic lateral ventricle administration of the ODN C-terminal octapeptide, OP, also caused more pronounced food intake suppression and weight loss in DIO compared with lean mice (30). Similarly, repeated daily lateral ventricle injections of a cyclized OP robustly reduced food intake and body weight in ob/ob mice (35). Although these studies support a persistent anorectic effect of ODN in chronically maintained obese models, they do not separate the effects on obesity versus diet. The influence of ODN on intake of palatable diets or specific macronutrients in the absence of obesity, or more broadly on reward aspects of ingestive behavior, has yet to be assessed.

In contrast to ODN, we found that, whereas injection of recombinant DBI protein (recDBI) suppressed 6-hour chow intake in lean rats, it had no effect in DIO rats. A potential interpretation of this finding could be that proteolytic cleavage of DBI to ODN may be impaired in obesity. Additionally, DVC and hypothalamic DBI expression is down-regulated in obesity (30), and, although most certainly from peripheral sources, plasma DBI is markedly lower in patients with obesity (74). Thus, persistent energy surplus may chronically lower DBI expression and impair posttranslational processing. Under these conditions, sensitivity for exogenous ODN signaling at its unidentified metabotropic receptor may be heightened, making its use as an anti-obesity treatment even more attractive, although how the expression profile of the ODN-GPCR is affected by obesity and how this could delay the onset of anorectic response are unclear.

Beyond appetite suppression, the collective findings here support that hindbrain ODN signaling regulates peripheral metabolic functions through suppression of pancreatic glucagon release and hepatic glucose production. We found that hindbrain ODN treatment selectively attenuated 5-TG and insulin induced plasma glucagon concentrations but did not affect corticosterone concentrations after glucoprivation. Previously, lateral ventricle OP injection in male rats was shown to prevent insulin-stimulated plasma glucagon and corticosterone, whereas, conversely, injection of an ODN-GPCR antagonist elevated baseline plasma glucagon and corticosterone, supporting a bidirectional role of ODN signaling on counterregulatory hormone concentrations (38). Lateral ventricle administration may have affected both corticosterone and glucagon by either targeting both the hypothalamus and hindbrain or engaging hypothalamus-to-hindbrain circuity (75). Whereas the ventromedial nucleus of the hypothalamus (VMN) particularly regulates glucagon and cortisol release (4, 76), DBI knockdown in the VMN exacerbated insulin-stimulated corticosterone but not glucagon concentrations (37). Of note, one of the first functions of DBI discovered was that pancreatic-derived DBI and ODN signaling through paracrine mechanisms inhibited glucagon secretion as well as glucose-stimulated insulin secretion (77, 78). In addition to the glucagon-blunting effects, we observed a trend of hindbrain ODN treatment to attenuate insulin concentrations after oral glucose gavage, and, conversely, astrocyte-specific DBI knockdown elevates glucose-stimulated insulin concentrations (35). Therefore, it is possible that central ODN signaling serves a complementary function to local pancreatic endozepine signaling, acting in parallel through peripheral and central mechanisms to inhibit the endocrine pancreas. We hypothesize that ODN may engage the well-established local NTS-to-DMX (dorsal motor nucleus of the vagus) circuit to modulate parasympathetic efferent nerve activity to the pancreas and liver (67, 75, 79).

TDN’s inhibition of hepatic glucose production, whether through direct hindbrain-to-liver circuits or secondary to the reduction in glucagon, and improvement in glucose tolerance support that hindbrain ODN signals an elevation in systemic glycemia and that action through its GPCR helps dispose of the postprandial rise in glucose. Consistent with the findings that lateral ventricle–administered ODN improves glucose tolerance (30, 31), we determined that hindbrain ODN treatment was sufficient to improve systemic glucose tolerance and insulin sensitivity, whereas antagonizing ODN-GPCR signaling could induce insulin resistance. Furthermore, hindbrain ODN-GPCR antagonism acutely elevated baseline glucose concentrations at dark cycle onset. Thus, endogenous ODN tone appears to be involved in maintaining euglycemia under basal conditions and promoting glucose disposal under postprandial conditions. Further studies will elucidate whether ODN-mediated improvements in glucose homeostasis are due to its regulation of pancreatic hormones or also involve insulin-independent mechanisms. We quantified plasma glucagon, corticosterone, and insulin concentrations at a single time point to capture the peak of release (61, 80); however, this does limit interpretation of dynamic hormonal responses that would provide more insight into the mechanisms of ODN’s regulation of glycemia.

The neuronal populations engaged by DVC ODN signaling are relatively unknown. Fourth and lateral ventricle OP injection induces cFos in the NTS, with one estimate being that, of NTS neurons labeled with cFos after OP treatment, ~5% coexpress with proopiomelanocortin (POMC), ~10% with tyrosine hydroxylase, ~20% with the GLP-1 preproglucagon neurons, and ~ 40% with nesfatin-1 (29, 30). Thus, activation of neurocircuits involving the anorexigenic neuropeptides POMC, GLP-1, nesfatin-1, and catecholaminergic NTS neurons may mediate the effects of hindbrain ODN. NTS catecholaminergic neurons, especially C2 populations, mediate the hyperphagia response to glucoprivation and may be sensitive to ODN modulation of this response (62, 81). In the hypothalamus, ODN-dependent activation of POMC neurons and α-MSH (α–melanocyte-stimulating hormone) signaling at MC4 receptors has been identified as an important downstream mechanism (31, 35, 82). Future necessary work will aim to elucidate DVC circuits activated by ODN and explore potential cross-talk between hypothalamic and hindbrain pathways engaged bidirectionally by ODN. Tracing experiments followed by projectionspecific inhibition during ODN treatment in the hindbrain or hypothalamus will help determine whether the feeding and glycemic responses initiated by ODN at one site require activation of the other energy balance center.

To understand ODN’s responsiveness to metabolic stimuli, we characterized the regulation of DVC DBI expression to fasting, glucose, and the GLP-1RA Ex-4. Glucose and Ex-4 up-regulated fasted DBI transcript expression with region-specific differences. Ex-4 up-regulated DBI expression in the NTS, whereas glucose up-regulated subpostrema border DBI expression. The strongest degree of nutritional regulation was observed in the tanycyte-rich fourth ventricle border, where DBI expression was down-regulated by fasting and up-regulated by both Ex-4 and glucose. This is consistent with findings in the hypothalamus that fasting decreased and glucose increased DBI mRNA expression most strongly around the border of the third ventricle (31). This regulation in ventricle borders suggests a specialized function of DBI in these tanycyte populations and potentially serves as an important source of DBI as a signal of changes in energy status. Notably, DBI expression in the AP, which represents the largest pool of DVC DBI, was not strongly regulated by nutritional status. Because almost three-quarters of AP-expressed DBI did not colocalize with GFAP or vimentin, this suggests that the cell populations expressing DBI in the AP may be less nutrient sensitive. Several identified response elements, including peroxisome proliferator response element and sterol response element (SRE), regulate DBI transcription in a tissue-specific manner (83-85). Insulin up-regulates DBI through SRE binding protein in hepatocytes and glia and likely contributes to DBI up-regulation upon feeding (30, 32, 86); however, how glucose and Ex-4 regulate DBI expression is unknown, and its transcriptional regulation in glia has not been well characterized.

We observed that DBI protein expression in the DVC was generally up-regulated by fasting and down-regulated by Ex-4, opposite that of DBI mRNA. We did not distinguish whether DBI protein staining was cytosolic or extracellular, representing stored or released protein. The half-life of DBI in the brain is not definitively known, although it is likely on the order of minutes. We would hypothesize that we are largely not capturing released DBI protein in the interstitial space with IHC analysis and that the DBI-positive area represents intracellular protein. DBI release is thought to be mediated by an unconventional pathway that does not involve classic endoplasmic reticulum–Golgi packaging and may involve transport by autophagy-associated vesicles (87-89). Membrane depolarization as well as adenosine 3′,5′-monophosphate (cAMP)–dependent protein kinase and protein kinase C signaling downstream of various neuropeptides can induce astrocyte DBI release (88, 90). We hypothesize that, in the fasted state, glial DBI may build up intracellularly and serve as a “readily releasable pool” that releases upon nutrient intake and cellular stimulation by depolarization or increased metabolism and adenosine 5′-triphosphate (ATP) production. In another model organism, Dictyostelium discoideum, DBI has been observed to accumulate in vesicle-like structures that are released in a burst upon stimulation (87). Glucose-dependent metabolism to ATP and subsequent depolarization is critical for tanycyte glucose-sensing mechanisms (91, 92), and GLP-1RAs stimulate cAMP and calcium signaling in NTS astrocytes (93). These mechanisms may be involved in stimulating DBI release in the postprandial state and mechanistically link ODN signaling with established players in the glial glucose-sensing mechanism.

We synthesized a series of truncated ODN analogs, one of which, designated TDN, was selected (vide infra) to specifically evaluate the preclinical therapeutic viability of targeting this peptide system through comparing the effects of central and peripheral drug administration on appetite and glycemic regulation and to identify potential off-target effects. We found that lateral ventricle and intraperitoneally injected TDN suppressed food intake and improved glucose disposal in DIO rodents. Most critically, 9 days of consecutive intraperitoneal TDN treatment suppressed food intake and induced substantial weight loss. Previously, daily lateral ventricle ICV injection of OP at 2 μg/day for 13 days was shown to induce consistent weight loss through the study duration (30). Another 15-day study using osmotic minipumps to continuously infuse lateral ventricle ICV ODN (10 ng/hour) reduced food intake through day 4, and, although a significantly lower body weight was maintained through the remainder of the study, the daily anorectic effect of ODN ceased, potentially because of the comparatively lower dose of 0.24 μg/day (36). We conclusively demonstrated that an ODN analog can reduce food intake and body weight when administered peripherally, and, although we cannot confirm a central site of action at this time, we hypothesize that intraperitoneal TDN is gaining access to the brain to exert its anorexic effects. We additionally report that the half-life of TDN in rat serum, although approximately twice that of ODN, is estimated to be 21.2 min. Evidence of brain penetrance and further exploration of the downstream mediators will be critical to evaluate the signaling profile of TDN. The doses of ODN and TDN given here are often supraphysiological concentrations, and, thus, we cannot rule out that the effects reported may represent pharmacological rather than physiological responses.

Consistent with previous reports that OP did not induce kaolin intake in mice (30), we observed that central ODN and TDN did not cause kaolin intake in rats, suggesting that these peptides do not induce nausea. We went further to examine drug action in a vomiting, mammalian model, the musk shrew, and detected no incidence of emesis with peripheral ODN or TDN, all while treatment still reduced food intake. We also observed no effect of central ODN or TDN on core body temperature, locomotion, or heart rate by light cycle assessed by telemetry probes. Looking at the individual hours, the high dose of ODN decreased locomotor activity at hour 8. Previous findings reported that ICV ODN signaling either increased or had no effect on dark cycle locomotor activity (30, 35), revealing no consistent finding of central ODN signaling on locomotion. The lower dose of ODN increased heart rate at two individual time points during the dark cycle, whereas the higher dose of ODN decreased heart rate at one time point in the light cycle; however, this seems likely because of individual variabilities given that no consistent or sustained effect was evident. More stringent evaluation of cardiac effects after ODN treatment will be needed. TDN did not alter these physiological variables at any time point.

Preclinical assessment of a drug compound’s effect on mood is also of great importance when evaluating the potential use of pharmacotherapies. Central treatment of ODN and DBI is known to induce anxiety-like behaviors in rodents and goldfish, but not pigs (94-97). However, global or astrocyte-specific DBI knockout did not increase anxiety-related behaviors but rather impaired the anxiolytic response to diazepam (98). Accordingly, the anxiogenic and other social effects of ODN/DBI are conclusively mediated through its action as an allosteric modulator of the γ-aminobutyric acid type A (GABA_A) receptor benzodiazepine binding (GAB_AAR-BZ) site (95, 96, 99, 100). Of note, peripheral DBI action through the GABA_AR-BZ site has been proposed to stimulate appetite and worsen cardiometabolic health (101, 102), an effect that theoretically is avoided by TDN from the existing data. In contrast, central ODN’s anorexigenic and glucose-sensing effects are maintained in the presence of GABA_AR-BZ antagonists but abolished by ODN-GPCR antagonists (31, 33, 35, 36), supporting that the ODN action at its GPCR and not GABA_AR-BZ site mediates the improvements in weight and glycemic control desired in a treatment for obesity and T2D. Thus, development of an ODN analog that is inactive at the GABA_AR-BZ site may be key in avoiding potential off-target anxiogenic and orexigenic effects.

Expanding the anti-obesity and anti-diabetes pharmacotherapeutic options with improved drug tolerability is necessary to improve weight and glycemic management. If ODN signaling is engaged downstream of GLP-1RAs, as suggested herein, then further understanding of this mechanism of action provides a route to create drug treatments that avoid the negative gastrointestinal adverse events of direct GLP-1R–targeting therapeutics. These data provide compelling evidence that hindbrain ODN signaling is physiologically relevant for energy balance control and demonstrate proof of concept that systemically administered ODN analogs suppress appetite, induce weight loss, and improve glucose homeostasis and thus may be of interest for further compound optimization for therapeutic application.

MATERIALS AND METHODS

Study design

This study was designed to explore the role of DVC ODN signaling in central glucose-sensing mechanisms and its regulation of systemic energy balance, as well as the validity of targeting ODN signaling as a therapeutic treatment for obesity/T2D. Pharmacological central (fourth and lateral ICV) and peripheral manipulations paired with behavioral, molecular, and imaging data were used to understand the regulation of endogenous DBI expression by postprandial factors; the impact of exogenous ODN signaling on appetite, body weight, and glucose homeostasis; and ODN’s contribution to GLP-1RA–induced hypophagia. We evaluated emesis in a vomiting musk shrew model; pica behavior modeling nausea in rats; and telemetry-monitored heart rate, body temperature, and locomotion to determine any side effects of treatment with ODN or its derivative TDN. Last, we performed hyperinsulinemic-euglycemic clamps and assessed the efficacy of repeated treatments with peripherally injected TDN to demonstrate systemic delivery as a viable route of administration. Data were excluded from animals with improper cannula placement as assessed via postmortem histological verification. Additionally, any mice that did not respond to the hyperinsulinemic-euglycemic clamp with an elevation in glucose disposal rate (Gd) were excluded from analyses. Samples for immunohistochemical and fluorescent in situ hybridization imaging were analyzed in a blinded fashion. Group sizes for all experiments were determined to minimize the numbers of animals required to perform an experiment with appropriate statistical power to detect statistically significant differences; power of 0.8 and an alpha of 0.05 were based on the following: (i) the minimal desired detectable differences in means of ~20%, (ii) expected ±5% SD of the dataset, and (iii) appropriate statistical test for each experiment. Additional materials and methods are available in the Supplementary Materials.

Animals

Three preclinical species—male Sprague-Dawley rats, male C57Bl/J6 or /6NTac mice, and male musk shrews (Suncus murinus)—were used for all studies. Animals were individually housed during all experiments and with 12-hour light:12-hour dark cycles in a temperature-controlled environment and food and water ad libitum unless otherwise specified. Rats were tested in the lean or DIO state and maintained on either chow [5001, LabDiet; calories provided by protein (28.9%), fat (13.6%), and carbohydrate (57.5%); 3.35 kcal/g] or a 60% HFD [D12492, Research Diets; calories provided by protein (26.2%), fat (34.9%), and carbohydrate (26.3%); 5.24 kcal/g]. Behavioral studies generally included 9 to 11 rats per group, 9 shrews per group, and 5 or 6 mice per group, with exact numbers for each experiment specified in the figure legends. All procedures were performed in accordance with National Institutes of Health’s Guide for the Care and Use of Laboratory Animals and approved by the University of Pennsylvania Institutional Animal Care and Use Committee.

Statistical analysis

Several studies including ODN plus liraglutide injections in rats, chronic intraperitoneal TDN injections in mice, and hyperinsulinemic-euglycemic clamps in mice were performed in a between-subjects manner with treatment groups stratified by body weight. All other experiments were performed in a within-subjects manner with the initial treatment order randomly selected and subsequent treatments randomized to avoid any effect of treatment order. All tissue analysis was performed in a blind fashion. All biobehavioral data were analyzed using ordinary or repeated-measures one-way or two-way analysis of variance (ANOVA) followed by Tukey’s post hoc tests or paired two-tailed Student’s t tests of normally distributed samples. Figure legends note statistical parameters for each experiment. All data are biological replicates and are expressed as means ± SEM with individual data points shown. For all statistical tests, P < 0.05 was considered significant. Data were graphed and analyzed using GraphPad Prism 9 software (GraphPad software).

Supplementary Material

This PDF file includes:

Methods

Fig. S1 to S11

Tables S1 and S2

References (103–107)

Other Supplementary Material for this manuscript includes the following:

Data file S1

MDAR Reproducibility Checklist