The role of oxytocin in regulation of appetitive behavior, body weight and glucose homeostasis

Abstract

Obesity and its associated complications have reached epidemic proportions in the US and worldwide, highlighting the need for new and more effective treatments. While the neuropeptide oxytocin (OXT) is well recognized for its peripheral effects on reproductive behavior, release of OXT from somatodendrites and axonal terminals within the central nervous system (CNS) is also implicated in the control of control of energy balance. In this review we summarize historical data highlighting the effects of exogenous OXT as a short-term regulator of food intake in a context specific manner and the receptor populations that may mediate these effects. We also describe what is known about the physiological role of endogenous OXT in the control of energy balance and whether serum and brain levels of OXT relate to obesity on a consistent basis across animal models and humans with obesity. We describe recent data on the effectiveness of chronic CNS administration of OXT to decrease food intake and weight gain or to elicit weight loss in diet-induced obese (DIO) and genetically obese mice and rats. Of clinical importance is the finding that both chronic central and peripheral OXT treatment evoke weight loss in obese animal models with impaired leptin signaling at doses that are not associated with visceral illness, tachyphylaxis, or adverse cardiovascular effects. Moreover, these results have been largely recapitulated following chronic subcutaneous or intranasal treatment in DIO nonhuman primates (rhesus monkeys) and obese humans, respectively. We also identify plausible mechanisms that contribute to the effects of OXT on body weight and glucose homeostasis in rodents, nonhuman primates, and humans. We conclude by describing ongoing challenges that remain in order for OXT-based therapeutics to be used as a long-term strategy to treat obesity in humans.

1. INTRODUCTION

OXT is expressed in neurons located in both parvocellular and magnocellular divisions of the paraventricular nucleus (PVN) as well as in the magnocellular division of the supraoptic nucleus (SON) in rats (77, 151, 189, 196, 197, 230, 231), mole-rats (192), mice (28, 192, 220, 228), hamsters (208), nonhuman primates (3, 55, 86, 184, 218) and humans (35, 53, 97, 167, 227). OXT is also expressed to a lesser extent in magnocellular neurons in the preoptic area (mole-rat) (192), anterior hypothalamus (mole-rat) (192), anterior commissural nuclei (mouse) (28), accessory nuclei (mouse/rat) (28, 189), and periventricular nuclei (mouse) (28). In addition, it is found in neurons (undetermined if magnocellular or parvocellular) of other forebrain areas that include the dorsal hypothalamic area (hamsters) (208), bed nucleus of the stria terminalis (BNST; mole-rat/rat) (189, 192), mediobasal preoptic area (mouse) (28), medial amygdala (mole-rat/rat) (189, 192), septal region (rat) (189) and caudal subzona incerta (hamsters) (248).

Mature OXT and the carrier neurophysin are derived from cleavage and modification of the OXT/neurophysin 1 prepropeptide during axonal transport (24). Both are stored in the axon terminals prior to release (187). The predominant function of neurophysin appears to be to target, package and store OXT within secretory granules in preparation for release (for review see (54)). OXT is released distally through axon terminals within the arcuate nucleus (ARC) (123), ventral tegmental area (VTA) (207), parabrachial nucleus (PBN) (193), nucleus of the solitary tract (NTS) (189, 196), dorsal motor nucleus of the vagus (189, 196), and spinal cord (196), and it is also released locally from somatodendrites within the hypothalamus (SON and PVN). In addition to local release from the PVN and SON (180, 260, 275, 276) (for review see (117)), the extent to which it may be released in other areas with more limited expression is not clear.

OXTR receptors (OXTR) are widely distributed and largely overlap in areas linked to the control of energy balance in both mice and rats, namely the basal ganglia [e.g. nucleus accumbens (NAcc) and central amygdala (mouse (66, 193, 269)/rat (240, 243, 246), hypothalamus [ARC, MPA, and ventromedial hypothalamus (VMH)/mouse (40, 56, 66, 193, 269)/rat (95, 243, 246)], midbrain VTA [mouse (170)/rat (243)], hindbrain PBN [mouse (193)], AP [mice (56, 193, 269)], DMV [rat (240, 243, 251), NTS [mouse (56, 269)/rat (11, 161, 162, 251)] and spinal cord [mouse (257)/rat (186)]. OXTR are also found in the SON and PVN (270) of OXT neurons in the rat (45) but it is unclear if this is the case in the mouse. Unlike the rodent model, there appears to be a more limited distribution of OXTRs in CNS areas linked to the control of energy balance in nonhuman primates [NAcc, preoptic area, VMH, DMV, and spinal cord) (20, 42, 200)] and humans (central amygdala, anterior hypothalamus, MPA, PVN, VMH, AP, NTS and spinal cord) (21, 114, 115). The overlapping patterns of expression in the basal ganglia, hypothalamus, hindbrain and spinal cord suggest potentially important roles of nuclei within these areas in controlling body weight (BW) that are well conserved across species. Less is known about the source of OXT that reaches OXTR that are expressed in brain sites that have little to no OXT fiber innervation (VMH, amygdala), but OXT has been proposed to reach these receptors by diffusion [reviewed in (250, 251)] from magnocellular neurons in the SON (reviewed in (109, 195).

In this review, we focus on the role of OXT in the control of energy balance based on data from rodents, nonhuman primates and humans. We highlight its well-characterized effects in the short-term control of food intake and the receptor populations that may mediate these effects. We also describe what is known about the role of endogenous OXT signaling in the control of energy balance and whether circulating and central levels of OXT reflect the state of obesity in animal models and humans. In addition, we 1) summarize recent data on the effectiveness of chronic CNS administration of OXT to decrease food intake and weight gain or elicit weight loss in diet-induced obese (DIO) and genetically obese rodents, nonhuman primates and humans, 2) identify potential mechanisms that may mediate these effects, and 3) discuss ongoing challenges that remain in order for OXT-based therapeutics to be used as a long-term strategy to treat obesity in humans.

2. Oxytocin and short-term control of food intake: key findings from laboratory rodent studies

Extensive evidence derived from animal studies focusing on short-term regulation of meal size indicates that OXT terminates feeding by affecting various facets of physiological responses that collectively contribute to cessation of eating behavior. This role arises from the strategic localization of OXT neurons in neural and neuroendocrine pathways that mediate, among others, gut-brain interactions, hormonal changes, and reward processing.

2.1. Receptor populations thought to contribute to the effects of OXT on short-term food and fluid intake

Well before the first reports on acute anorexigenic properties of OXT were published, it had been established that lesions of the hypothalamic PVN where the OXT cells are located and the disruption of the hindbrain-PVN connectivity (along with severing OXT fibers between the PVN and caudal brain stem) (90), promote excessive feeding and BW gain in rats (107, 209, 211). Those studies served as the cornerstone of the initial experiments examining the effects of OXT injections on food consumption. In 1989 and 1990, Arletti and colleagues showed in a comprehensive manner that intracerebroventricular (ICV) administration of OXT in rats caused short-lived hypophagia reversible by an ICV OXTR antagonist (d(CH₂)⁵Tyr(Me)²-[Orn⁸]-vasotocin) (4, 5). The anorexigenic effects were also recapitulated after peripheral (intraperitoneal; IP) injections. OXT-treated freely-feeding animals reduced intake of standard laboratory chow by approximately 60% (IP OXT) and 40% (ICV OXT). OXT lowered chow intake also in schedule-fed rats given food for 3 hours per day. Latency to begin a meal was increased, whereas meal duration, decreased, by the treatment. These findings were corroborated by Olson who found that lateral ventricular OXT and OXTR agonist [e-L-beta-MePhe²]OXT reduced deprivation-induced chow consumption and time spent feeding in rats accustomed to receiving a 1-hour meal (152). Over the years, OXT has been found to be effective in generating early termination of food intake when administered subcutaneously (sc), intravenously (IV), intranasally (IN) and into the lateral, third (3V) and fourth ventricle (4V; as a strategy to more specifically target hindbrain OXTRs). Considering the debate over whether peripherally administered OXT crosses the blood-brain barrier and enters the brain in rodents (67, 76, 130, 144, 190, 214, 236), pigs (185), sheep (87), nonhuman primates (29, 33, 44, 106, 134) and humans (108, 223), both the central and peripheral pools of the OXT receptor have been implicated in mediating OXT-driven hypophagia (16, 34, 67, 75, 76, 93–95, 105, 121, 138, 147, 161, 162, 189, 191, 275–277). In addition, it has been suggested that peripheral OXT may also act, in part, via hindbrain OXTRs: 4V pretreatment with an OXTR antagonist attenuated the effects of IP OXT on short-term food intake in rats (67). Based on 1) the findings by Ho (67), Iwasaki (75, 76), Wu (258, 259), and Liu (112), and 2) reports indicating that OXTRs are expressed in the nodose ganglion (255) and GI tract (183), peripheral OXT might also promote hypophagia through these peripheral OXTR populations. Importantly, as shown in hunger discrimination studies utilizing operant methodology, while OXT is effective in producing early termination of a meal that has already begun, it does not reduce the feeling of hunger (61).

Some (4, 5) but not all (e.g. (93, 138, 152)) studies have shown a reduction in water intake in animals treated with OXT, but this peptide’s effect on food consumption is thought not to arise from changes in drinking behavior. Recent data suggest that only discrete subpopulations of the OXTR-expressing neurons, such as those located in the PBN, promote changes in water intake (193).

2.2. Use of OXTR antagonists as a tool to examine the physiological role of OXT in the control of short-term food intake

While the vast majority of studies have shown reversibility of OXT’s effects on feeding with OXTR antagonists, there is no consensus as to whether OXTR blockade by itself increases consumption. For example, one study found that lateral ventricular administration of the antagonist [(CH₂)₅¹, Phe(Me)², Thr⁴, Orn⁸] OXT did not potentiate deprivation-induced feeding of chow in rats (152), Olszewski reported no effect of an IP BBB-penetrant antagonist, L-368,899, on chow intake in mice (156), and Klockars reported no change in chow consumption after direct intraparenchymal injection of L-368,899 (94). On the other hand, it was reported that both IP and ICV administration of the OXTR antagonist, d(CH₂)⁵Tyr(Me)²-Orn⁸]-vasotocin, increased food intake and reduced latency to begin a meal in schedule-fed rats (4, 5). Others (11, 17) found that 3V administration of the OXTR antagonists, [d(CH₂)⁵,Tyr(Me)²,Orn⁸]-vasotocin and [d(CH₂)51,Tyr(Me)²,Orn⁸]-oxytocin, respectively, stimulated chow consumption in rats. An additional paper reported that 4V injection of the antagonist stimulated chow intake, and, in contrast to some of the previous studies following 3V administration, found that 3V administration failed to stimulate chow consumption in 6-h fasted rats (17), indicating that mixed results have been obtained following 3V administration of OXTR antagonists in rat models. Consistent with the effects following 4V administration in rats, Blouet and Schwartz extended these findings and found that 4V administration of the OXTR antagonist, [d(CH₂)⁵,Tyr(Me)²,Orn⁸]-vasotocin, stimulated chow consumption in mice through an increase in meal size and a decrease in meal latency (19). However, considering the two studies in which ICV administration of an OTR antagonist stimulated food intake in rats that were deprived for 21 h (4, 5), Baskin also found that 3V administration of the antagonist in 16-h fasted rats, produced hyperphagia (17). It is not clear at this point if differences in timing of administration may account for these differences, as a more potent effect on daytime chow intake was found when [d(CH₂)⁵,Tyr(Me)²,Orn⁸]-vasotocin was administered into the 3V at the start of the light cycle in mice (276). While the effects of 3V and 4V OXTR blockade on food intake are largely consistent between rats and mice, differences with respect to length of fast, strain, or OXTR expression may explain, in part, why the OXTR antagonist may work more optimally under some conditions (276) but not others.

In the early stages of research on OXT and feeding, concerns were raised that OXT-driven hypophagia might be only a pharmacological phenomenon secondary to select homeostatic functions of the hormone. Indeed, intravenous infusion of hypertonic saline is known to increase OXT release in rats and humans (41, 221) and, simultaneously, inhibit feeding in rats (41). Plasma toxicity that leads to termination of food intake and underlies development of conditioned taste aversion (CTA) increases the number of c-Fos (marker of neuronal activation) positive OXT neurons (68, 149, 154, 155, 238). OXTR blockade prevents CTA acquisition (159, 263, 264) and drugs that block OXT neuronal activation oftentimes alleviate aversions (158, 256), although central or peripheral administration of OXT at anorexigenic doses does not cause pica or CTA (76, 147, 275) (behavioral readouts of visceral illness in rodent models). Furthermore, OXT is rapidly released into the periphery in response to electrical stimulation of gastric afferents (242) or a large degree of stomach distension regardless of energy load (143, 188). Overall the data suggest that, while one can view OXT as being involved in protecting internal milieu in a manner completely independent from feeding, in reality it acts as an integral component of homeostatic mechanisms typically triggered by adverse effects stemming from ingestive behavior (excessive stomach distension, sodium imbalance, and toxicity).

2.3. What have we learned about the role of OXT neurons in the control of short-term food intake?

What instills confidence in defining OXT as an anorexigen is that the activity of the OXT system changes in a hunger-satiation continuum and that there is a functional link between OXT and other neural and endocrine feeding regulators. Plasma OXT levels are elevated at satiation-driven end of a meal in female (ovariectomized) and male rats (116, 252). An increase in the percentage of c-Fos immunoreactive PVN and SON OXT neurons has been detected in mice and rats that have eaten a satiating amount of a calorie-dense chow or calorie-dilute liquids compared to animals anticipating presentation of a meal (79, 156, 265). In contrast, recent studies in mice using optical fiber photometry indicate that chow diet failed to alter activity of PVN OXT neurons in mice (50) raising the possibility that macronutrient composition of the diet may be important in activation of PVN OXT neurons. Consistent with this, sated rats exhibit higher hypothalamic OXT gene expression than deprived counterparts (98, 169, 241), and OXTR mRNA levels are modified by energy deprivation as well as by macronutrient composition of food [e.g. (94, 95, 156)]. Intragastric administration of nutrients that promote satiety (e.g., select amino acids), also triggers OXT gene expression and OXT neuronal activation (51) in mice. Importantly, OXT neurons or terminals co-express other anorexigens, including corticotropin-releasing hormone (CRH) (9), nesfatin-1 (96, 124) and CCK (127), and OXT appears to contribute to and/or potentiate the hypophagic action of nesfatin-1, CCK, and leucine (15, 18, 19, 124, 153) in rats and mice. OXT neuronal activation is modified by feeding regulatory peptides either directly (via synapses formed with OXT cells) or indirectly (as an element of broader circuits). For example, terminals containing cocaine-amphetamine related transcript (CART) are found in close apposition to magnocellular and parvocellular PVN OXT neurons (253) and central administration of CART increases c-Fos in both magnocellular and parvocellular PVN OXT neurons (253). In addition, OXT cells express the receptor for leptin and administration of leptin at doses that support hypophagia activates OXT neurons (17, 59, 169, 249, 261). Anorexigenic glucagon-like peptide-1 and adrenomedullin increase OXT neuronal activity and/or promote neurohypophyseal release of OXT (60, 164, 237, 279). Another anorexigenic signal, alpha-melanocyte stimulating hormone (alpha-MSH), an endogenous melanocortin 3/4 receptor (MC3/4R) agonist, also stimulates the release of OXT from hypothalamic dendrites (194) and activates c-Fos in PVN OXT neurons (160). Similarly, the synthetic MC3/4R agonist, melanotan II (MTII), increases c-Fos in PVN OXT neurons (135, 168) where MC4Rs are co-localized with OXT neurons in rodents (52, 113) and humans (210). However, alpha-MSH also suppresses OXT release from the nerve terminals by inhibiting the electrical activity of OXT neurons (194), but the extent to which this contributes to OXT’s effects on energy balance remains to be determined. On the other hand, orexigenic peptides, including Agouti-related protein and opioids, tend to suppress activity of the OXT system, thereby likely delaying the onset of satiation (6, 157). Finally, chemo- and optogenetic manipulation of glutamate-releasing ARC neurons expressing the OXTR, promotes rapid satiety in mice (40).

2.4. Role of OXT in feeding reward

It is well accepted that OXT decreases feeding in the process of facilitating satiation and/or protecting homeostasis. This role is mediated mainly by the hypothalamic-brainstem circuit (15, 95, 147). Importantly, OXT regulates feeding also via mechanisms related to reward, consistent with the OXTR being expressed in the NAcc, amygdala, BNST and VTA (56, 78, 181, 217, 226).

Studies on eating for pleasure have revealed that centrally acting OXT reduces appetite for carbohydrates and for palatable non-carbohydrate sweeteners, such as saccharin. Peripherally administered BBB-penetrant OXTR blocker, L-368,899, elevates consumption of calorie-dilute and non-caloric solutions containing carbohydrates or saccharin, but it fails to modify intake of Intralipid in mice and rats (63, 156, 212). OXT injected directly in the VTA, NAcc core and basolateral amygdala (BLA) in rats decreases the intake of episodically presented palatable sweet solutions, and this effect is reversible by pretreatment with an OXTR antagonist, L-368,899 (94, 138). OXT knockout (KO) mice drink excessive amounts of sucrose- and saccharin-sweetened water (2, 139). In a two-bottle choice test in which water is a control tastant, KO mice prefer sucrose, Polycose and even a “bland” cornstarch emulsion, but their preference for Intralipid is unchanged (132, 205). Intake of a similar amount of calorie-matched sugar water vs. fat emulsion produces a higher number of c-Fos positive PVN OXT neurons in sucrose-fed mice, and the activation is higher at the end than at the start of consumption regardless of the macronutrient content (156). The aforementioned experiments utilizing fluid consumption strongly suggest that OXT is particularly effective in reducing intake of carbohydrates and sweet solutions. One should note, however, that data also indicate that the effect of OXT on feeding appears to be more dependent on energy density of food rather than macronutrient composition or flavor. In line with this notion, genetic ablation of OXT cells in the hypothalamus generates obesity in high-fat diet-fed animals (261). ICV treatment with an OXTR antagonist increases feeding in mice (275, 276) and rats (Blevins, unpublished) given a high-fat high-sugar diet. Finally, chronicity of palatable diet presentation may be a mitigating factor in meal-end responsiveness of OXT neurons to food load. Accordingly, occasional episodic intake of a high-sugar diet in rats engaged more PVN OXT neurons than low-sugar food in rats, however daily habitual intake of sugary food promoted a significant decline in the level of activation of OXT cells (133).

2.5. Use of chemogenetic and optogenetic approaches to understand the role of PVN OXT in the control of food intake

While the existing work highlights an important role for specific OXTR populations in the control of food intake in rats and mice, based on recent lentiviral, chemogenetic and optogenetic strategies in mice the picture is becoming less clear with respect to the role of PVN OXT neurons in the control of food intake. As mentioned earlier, Zhang and Cai found that lentiviral reduction in PVN OXT expression resulted in increased food intake in both chow-fed and high fat diet-fed mice (275). In contrast, Xi and Wu found that diphtheria-toxin elicited ablation of PVN OXT had no effect on food intake in chow-fed (262) or high fat diet-fed mice (261). In addition, chemogenetic activation of PVN OXT neurons failed to suppress baseline food intake (193) or food intake following stimulation of PVN OXT neurons at the start of the dark cycle (228). Both chemogenetic and photo-activation of PVN OXT also failed to suppress refeeding of low-fat diet or chow after a 24-h fast in mice (6, 193). On the other hand, chemogenetic or tetanus toxin-elicited silencing of PVN OXT neurons failed to stimulate food intake (chow) (50, 111). While it will be important to determine if chemogenetic or optogenetic manipulation of PVN OXT neurons is sufficient to 1) impact OXT release and/or elicit c-Fos from axonal terminals in the NTS (175, 176), these findings are at least consistent with 1) a minor role of PVN OXT neurons to suppress the refeeding response following extended fasts in rat models (212). They are also consistent with data from mice with global deficiencies in OXTR or OXT signaling that show no defects in weekly food intake (81) or daily chow intake (27, 235). Given that OXT’s effects can be optimized based on time of day (276) and composition of the diet (18, 34, 120, 121, 191, 275–277), future studies should take into consideration timing of the stimulus (early vs late light cycle), macronutrient content of diet, and whether more can be gained by manipulating either OXTR populations (193) or OXT fibers (193) that innervate specific OXTR populations before more definitive conclusions can be made with respect to the importance of endogenous OXT signaling in the mouse model.

In addition, given the heterogeneity with respect to PVN OXT populations and outgoing projections, targeting specific PVN OXT neuronal populations that project to areas with distinct roles such as the spinal cord (thermoregulation and energy expenditure) (228) or hindbrain (food intake, thermoregulation and energy expenditure) (162, 191) may contribute to different outcomes across studies. Current findings indicate there might be species differences with respect to the origin (rostral vs caudal) of the more predominant descending parvocellular PVN OXT neuronal projections to the hindbrain. In particular, parvocellular PVN OT neurons that project to the NTS are located predominantly in the caudal parvocellular PVN in rats (189). While it remains to be determined if this is the case in mice, there appear to be few OT projections from the rostral parvocellular PVN to the NTS in mice and those in the rostral PVN appear to project to spinal cord (228). These findings raise the possibility that parvocellular PVN OT neurons located more caudally project to subsets of NTS OXT receptor (OXTR)-expressing neurons in both mice and rats. It is also possible that, in the mouse model, parvocellular PVN OXT neurons do not innervate the hindbrain as densely compared to the rat model although existing studies provide indirect evidence in support of such a projection in mice (19, 128, 193, 260) and suggest that caudal hindbrain OXT receptors are relevant in the control of energy balance in both mice (19, 128) and rats (11, 15, 67, 161, 162) (covered in more detail below and in sections 2.1 and 3). Recently, projections from the 1) PVN and SON OXT neurons to the ARC (123), PVN OXT neurons to the VMH (141) and PBN (193) and ARC OXTR-expressing neurons to the PVN (40) have been described in the mouse model. While it appears that PVN OXT neurons do not project to the VMH in rats, further studies will need to determine in rats whether 1) OXT neurons project from the PVN and SON to the ARC and PBN and 2) ARC OXTR-expressing neurons project to the PVN.

3. Effects of OXT treatment on weight loss in obese animal models

Recent studies have shown that central, systemic, or IN administration of OXT decreases food intake, weight gain and/or BW in DIO rodents (34, 101, 120, 121, 137, 191, 206, 216, 275, 276), genetically obese rodents (1, 8, 76, 98, 124, 137, 177), ovariectomized rodents (74), dihydrotestosterone-induced rat model of polycystic ovary syndrome (PCOS) (73), as well as in DIO nonhuman primates (16). In sections 3.1–3.2, we will review the potential mechanisms that contribute to the effects of chronic OXT in rodent and nonhuman primate models of obesity.

3.1. Does chronic OXT treatment affect BW by increasing energy expenditure?

3.1.1. Effects of acute and chronic OXT treatment on energy expenditure

OT-elicited reductions in food intake do not appear to fully explain OT-elicited weight loss. Namely, the effectiveness of OXT to decrease energy intake in DIO or genetically obese rodents appears to wane over time despite an intact response of OT to continue to reduce weight gain or BW (18, 34, 120, 121, 191). Whether the waning of the feeding effect is associated with OXT-elicited reductions in receptor binding (43, 72, 171) following chronic administration (further discussed in 7.2) remains to be determined. This is evident when rodents receive 1× daily repeated peripheral injections (120, 121) as well as chronic peripheral (1, 18, 34, 120, 121) and central infusions (18, 34). In addition, indirect evidence from pair-feeding (food in control group is matched to that of OXT-treated animals) studies in mice and rats, where the amount of weight loss following OXT exceeds that of pair-fed control animals (1, 18, 34), suggests that other mechanisms such as increased EE may also contribute to OXT-elicited weight loss. Furthermore, recent studies have shown that single or repeated acute injections of OXT increase energy expenditure in both rodents (147, 275, 276) and DIO nonhuman primates (16). Chronic infusion of OXT may also be able to prevent the counter-regulatory mechanisms that promote weight regain in DIO rodents following sustained weight loss, in part, by maintaining the level of energy expenditure to that of vehicle-treated control rats (18) and mice (121). Thus, chronic OXT treatment (repeated injections or minipump infusions) may be able to produce prolonged weight loss, in part, by increasing energy expenditure and/or preventing the reduction in energy expenditure associated with prolonged weight loss.

3.1.3. Role of endogenous OXT on energy expenditure and BAT thermogenesis

PVN OXT neurons are anatomically positioned to control energy expenditure by outgoing sympathetic projections to interscapular BAT (IBAT) (151), stellate ganglia (77) [sympathetic ganglia known to innervate IBAT (151)], as well as white adipose tissue [WAT; epididymal (208, 220) and inguinal WAT (208)]. Cold exposure activates PVN OXT neurons (81) and rostral medullary raphe (RMR) neurons that express OXTRs (82). Furthermore, data from animals with impaired or defective OXT signaling (e.g. OXTR and OXT null mice, mice with viral or diphtheria toxin-elicited reductions in PVN OXT neurons, or mice treated with an OXTR antagonist) show reductions in energy expenditure (261, 275, 276) and/or defects in cold-induced thermogenesis (80–82, 235, 262), the latter of which have been shown to be rescued by OXT treatment (262). In addition, histological assessment revealed the presence of large lipid droplets in IBAT in OXTR null mice which suggests hypo-activity of IBAT tissue (235). Furthermore, chemogenetic activation of rostral PVN OXT neurons, some of which project to the thoracic spinal cord, induces c-Fos in thoracic spinal cord cholinergic neurons, increases energy expenditure and tends to increase IBAT temperature (228), although the extent to which these effects are attributed to endogenous OXT remains to be determined. In addition, viral expression of OXTRs into the RMR (82) and hypothalamus (80) in OXTR null mice restores deficits in cold-induced thermogenesis; expression of OXTRs in RMR also normalized lipid droplet size in IBAT tissue to that of control mice (82). Together, these data suggest that OXTRs in these areas may be critical in regulating BAT thermogenesis during cold exposure. Collectively, these findings support the hypothesis that endogenous OXT signaling can control BAT thermogenesis and energy expenditure, but the extent to which its effects on energy expenditure require BAT thermogenesis and/or browning of WAT and the OXTR populations that are required to mediate these effects remain to be determined.

3.1.4. Role of exogenous OXT on BAT thermogenesis and browning of WAT

We know from recent studies that 3V or 4V administration of OXT elevates IBAT temperature in rats and mice and that these effects extend to DIO rats (191). Furthermore, chronic 3V infusion of OXT elevates IBAT temperature at a time that coincides with the onset of OXT-elicited weight loss in DIO rats (191). Consistent with these findings, 4V OXT also elevated core temperature in rats (162). These findings are particularly relevant given that changes in IBAT temperature tend to precede and contribute to changes in core temperature in cases of fever and stress (84). The effects of OXT on IBAT temperature appear to be related to non-shivering BAT thermogenesis as chronic OXT administration is not associated with elevations in locomotor activity in rats (18, 34, 74) or mice (121). Chronic subcutaneous (sc) OXT treatment was also associated with increased uncoupling protein-1 (marker of brown adipocytes) immunostaining in sc WAT (177). It will be helpful if future studies could evaluate the impact of chronic OXT treatment on additional markers of browning in WAT as well as WAT temperature as a more functional readout of WAT thermogenesis. While these collective findings indicate that exogenous OXT increases BAT thermogenesis, the extent to which BAT thermogenesis and browning of WAT is required for chronic OXT to increase energy expenditure and evoke weight loss is an ongoing area of investigation.

3.2.2. In vitro and in vivo data

OXT increased glycerol in 3T3-L1 adipocytes (268), which are known to express OXTRs (198, 268), suggesting that OXT stimulates lipolysis, in part, through a direct effect. Similarly, OXT increased glycerol release from epididymal fat pads ex vivo (34). OXT also elevated serum glycerol and reduced serum triglycerides after a 14-day treatment period in rats (34). Blevins extended these findings in a translational nonhuman primate model where chronic 2× daily administration of OXT increased both serum free fatty acids and glycerol and reduced serum triglycerides following a 4-week treatment period in DIO rhesus monkeys (16). A further study showed that chronic OXT increased epididymal WAT expression of lipoprotein lipase (Lpl) and fatty acid transporter (Fat), both of which are linked to uptake of triglycerides and fatty acids, respectively. While chronic OXT treatment did not impact enzymes linked to lipogenesis and storage of triglycerides (acetyl-coenzyme A carboxylase alpha, fatty acid synthase, and diacylglycerol O-acyltransferase homolog 1), it did increase enzymes linked to lipolysis (patatin-like phospholipase domain containing 2 and hormone-sensitive lipase) (34). Chronic central or peripheral OXT treatment is also associated with reductions in respiratory quotient in DIO mice (121) and rats (18, 34) relative to vehicle treatment (18, 34, 121) or pair-fed control animals (34). Taken together, these findings suggest that increased lipolysis and lipid oxidation may contribute to OXT-elicited weight loss.

3.3. How does chronic OXT treatment impact body composition?

Chronic central (lateral ventricle, 3V) or peripheral OXT treatment elicited a relative reduction (post vs pre-intervention) in fat mass or decreased total fat mass (post-intervention) compared to vehicle in lean (34) and DIO rats (18, 34, 137, 191), DIO C57BL/6J (191) or C57BL/6 mice (216, 276), db/db mice (177), and ob/ob mice (1) without impacting lean mass. In regards to targeting specific fat depots, OXT treatment appears to reduce visceral (specific visceral fat depot not determined) (120), sc (120), mesenteric (121) and epididymal fat (1, 121) in DIO C57Bl/6J (121) and ob/ob mice (1), as well as in female ovariectomized Wistar rats (74). It has also been shown to reduce adipocyte area from mesenteric (121), epididymal (8, 38, 121), sc (74, 177), visceral (74), perirenal (177), and epicardial (177) depots in db/db mice (177), C57Bl/6J mice (121) obese Zucker rats (8) or female ovariectomized Wistar rats (74). Chronic sc OXT treatment also reduced liver weight and fat in hepatocytes in DIO C57BL/6J mice (121) but was found to have no effect on liver triglyceride content in ob/ob mice (1). In contrast, some studies have reported that chronic systemic or central (4V) OXT administration elicited a relative reduction in lean mass compared to vehicle in lean C57Bl/6J (1) and DIO C57BL/6 mice (216) as well as in DIO CD IGS rats (191) with or without any relative reductions in fat mass. However, chronic 4V OXT infusions produced a relative reduction in fat mass without impacting lean mass compared to 4V vehicle in DIO Long-Evans rats (191) raising the possibility that differential effects may be attributed, in part, to rodent strain, dosing and/or route of administration. While there are exceptions, overall OXT treatment appears to preferentially reduce fat mass while preserving lean mass across rodent models.

3.4. Does chronic OXT treatment produce sex-specific effects on energy balance in animal models?

While an earlier study reported no sex-specific effects of acute OXT administration (central or peripheral) on food intake in male and female rats (12), the majority of studies that have tested the effects of chronic OXT on energy balance have focused predominantly on male animal models. One previous study found that chronic OXT reduced both food intake and weight gain in female Sim1 haploinsufficient mice but these effects were not observed in female wild-type mice (98). Maejima recently compared the effects of chronic sc administration of OXT on weight loss in both male and female DIO C57Bl/6J mice (120) and found that OXT reduced BW and fat mass (sc and visceral) to a similar extent in both sexes. Seelke recently reported that female DIO prairie voles may be more sensitive than males to the effects of IN OXT on BW although the sample size in this study was small (206). In addition, chronic OXT treatment was found to reduce food intake and BW in female ovariectomized rats (74). Further studies will need to confirm these effects in other male and female models of obesity, including nonhuman primates and humans.

3.4.1. Are there sex differences with respect to the effects of endogenous OXT signaling on energy balance?

There have been somewhat inconsistent results from the few studies that have examined the effects of global loss in OXT or OXTR signaling on energy balance in male and female rodents. While some have found that both male (27, 81) and female OXT null mice (27) develop adult onset obesity (27, 81), only male OXTR deficient mice become obese (235). In addition, selective ablation of PVN OXT neurons in Oxytocin-IresCre:Rosa26^iDTR/+ mice predisposes male but not female mice to diet-induced obesity relative to control Rosa26^iDTR/+ mice (261). Further studies will be required in order to provide more clarity on whether there are sex differences in the role of endogenous OXT in regulating energy balance.

3.5. Does OXT treatment produce adverse side effects in animal models?

3.5.1. Visceral illness

Previous studies have indicated that chronic systemic administration of OXT, at doses that are effective in reducing weight gain or evoking weight loss, does not appear to be associated with nausea or food aversion in rodent models. In particular, chronic peripheral administration of OXT (50, 100, 200 nmol/day) failed to significantly alter 2-h saccharin preference ratios in fluid choice tests (saccharin intake/water intake; a paradigm in which a reduction in saccharin preference ratios is a readout of a CTA) and OXT (50 nmol/day) also failed to increase kaolin consumption (index of pica behavior) (18). Chronic central delivery of OXT (3V, 4V) did not impact kaolin consumption in DIO mice or rats (18, 191). These findings extend previous reports showing that acute bolus injections of OXT into the CNS [3V (4 μg), VMH (0.1, 1 nmol)] or periphery (200 μg/kg, IP), at doses that reduced food intake, failed to impact kaolin consumption (275) or elicit a CTA (76, 147, 275) in mice or rats. In addition, no evidence of diarrhea or vomiting was found following daily sc OXT treatment over a 4-week period in rhesus monkeys (16). Taken together, these findings suggest that peripheral or central delivery of OXT, at doses that evoke weight loss, are not associated with visceral illness.

3.6. Effects of OXT treatment on glucose homeostasis in animal models

While the majority of studies to date largely provide support for a therapeutic role of OXT to stimulate insulin secretion, insulin sensitivity or glucose tolerance, there are also cases of conflicting data that may be related to animal model, route of administration, or dosing. Chronic systemic infusions [sc; 1.6 μg/kg/day (~0.0563 nmol/day)] or repeated 1× daily systemic administration (IP; 1 mg/kg) over a 1 to 6 week period resulted in an improvement in glucose tolerance in high-fat diet (HFD)-fed mice (121, 276, 277). Similar results were obtained even in the absence of any weight loss when HFD-fed mice received 2× central injections of OXT [3V; 4 μg] during an overnight fast (277). Others have shown that chronic central (ICV; 1.6 nmol/day) or systemic (sc; 50 nmol/day) infusions at the end of the 14-day infusion period either improved glucose tolerance or increased insulin sensitivity in HFD-fed rats (34). Acute IN (0.1, 1 or 10 μg) or central (ICV; 0.4 and 4 μg) administration failed to impact glucose tolerance but acute IP administration of OXT (IP; 40 and 400 μg/kg) improved glucose tolerance in C57BL/6J mice (122). Similarly, acute systemic administration of OXT (IP; 2 mg/kg) improved glucose tolerance in both lean and obese C57BL/6 mice (216) and these effects appear to be mediated, in part, by increased insulin secretion (216). In addition, acute injections of IN OXT over a 7-day period were able to reduce fasting levels of insulin and also appeared to reduce fasting levels of glucose (not significant) in a DIO prairie vole model (206). Finally, 2× daily sc OXT administration over a 4-week period (0.2 mg/kg for initial 2 weeks; 0.4 mg/kg for final 2 weeks) decreased fasting blood glucose in a more translational DIO rhesus monkey model (16). However, chronic ICV (1.6 nmol/day) or 3V infusion (16 nmol/day) failed to exert a significant effect on 1) blood glucose and insulin in the fed state (34) as well as 2) fasting blood glucose (18, 191) and insulin (191) at the end of the 14- to 28-day infusion periods in DIO rats and mice. In contrast, chronic systemic infusions of OXT (sc; 50 nmol/day) over 14 days appeared to worsen fasting blood glucose and fasting insulin at the end of the treatment period in C57BL6/J mice (1). In addition, the same group reported that chronic systemic infusions of OXT (sc; 50 nmol/day) over 14 days appeared to worsen both fasting blood glucose at the end of treatment and daily basal blood glucose in the fed state during the infusion period in ob/ob mice (1). Similarly, chronic systemic infusions of OXT [sc; 3.6 μg/100 g or ~8.22 (lean) or 12.5 nmol/day (obese)] over 15 days failed to impact fasting blood glucose but worsened fasting insulin in both lean and obese Zucker rats (8). In addition, this study found that OXT infusions over 13 days impaired glucose tolerance in obese Zucker rats (8). To summarize, the majority of data from animal models appear to support a promising effect of exogenous OXT to improve glucose tolerance. However, additional studies that 1) include weight-restricted controls in both genetically obese and DIO rodents and DIO nonhuman primates, and 2) explore whether differences in route of administration, strain, and/or dosing will be helpful in clarifying the role of OXT signaling on glucose homeostasis. Given that OXTRs are present on α cells and β cells in pancreatic islets (229), OXT may act, in part, to stimulate insulin and glucagon release from mouse islets (49) and rat pancreas (36, 37) through a direct effect, but may also stimulate insulin secretion through an indirect effect by activating vagal cholinergic neurons that innervate β cells (14). There is also evidence to suggest that OXT may increase insulin sensitivity, in part, by increasing insulin-sensitive glucose transporter 4 mRNA in fat (38).

4. OXT levels in animal models of obesity

4.1. Serum measurements in animal models

It is important to acknowledge that measurements of endogenous levels of OXT in serum or plasma remain controversial and assay validation continues to be an ongoing need (110, 119, 129). Many laboratories have used radioimmunoassays (RIAs), enzyme-linked immunosorbent assays (ELISAs) or mass spectrometry to measure circulating levels of OXT. Many of the newer commercially available ELISA kits have generated higher levels of OXT which may be due, in part, to interference from other substances, including OXT fragments and/or OXT degradation products (233). The extent to which these may be biologically active (and potentially impact energy balance) remains to be determined. Recent work with commercially available ELISA kits has resulted in endogenous OXT levels being up to two orders of magnitude higher than RIAs in cases where extraction was not incorporated (129). Extraction before ELISAs helps to eliminate some of the interfering substances and results in yields that are similar to those obtained with the more traditional RIAs (233). However, the discarded substances may include OXT bound to proteins (22) as well as OXT fragments and/or OXT degradation products. While absolute OXT levels cannot be compared across studies due to assay differences, measurements can be useful for comparing relative levels of peripheral OXT between groups. Future studies should also consider the impact of circadian variations that occur with respect to circulating OXT (276) and how circulating OXT levels may be impacted by 1) factors that impact central release of OXT (34), 2) feeding (116, 252, 265) and 3) obesity in DIO (276) and genetically obese animal models (48, 177, 204, 273).

4.2. “Genetic” models

Endogenous OXT levels have been measured in different genetic rodent models of obesity with divergent results.

4.2.1. Otsuka Long-Evans Tokushima Fatty (OLETF) and Zucker rats

The OLETF rat has a null-mutation in a gene that expresses a receptor (cholecystokinin 1 receptor) for the satiety signal cholecystokinin (CCK) and it presents with overeating-induced obesity, hyperleptinemia and predisposition to type 2 diabetes (85). OLETF rats showed significantly higher plasma OXT levels compared to the lean, Long-Evans Tokushima Otsuka (LETO) control strain (204, 273). Pair-feeding (diet restriction) from weaning can dramatically normalize OLETF rats to lower mean BW and plasma OXT levels, similar to those of lean LETO controls (202, 203).

Recently, levels of OXT gene expression in the PVN were measured on postnatal day 23 (PND 23) and PND 90 by in-situ hybridization in OLETF and LETO rats (89). PVN OXT expression was lower in OLETF rats both at PND 23 and PND 90, compared to lean LETO controls. This pattern is the opposite of what was found for dorsomedial hypothalamus (DMH) neuropeptide Y (Npy) gene expression. This raised the possibility that DMH NPY down-regulates OXT expression in the PVN. As discussed in the current review, PVN OXT has been postulated to mediate the effects of leptin on meal size (17). Thus, the signaling pathway of DMH NPY regulation of PVN OXT in modulating food intake merits further investigation.

In Zucker fatty rats, obesity is a consequence of spontaneous mutation (fa) in the gene encoding the leptin receptor, resulting in hyperphagia (174) and characteristics of type 2 diabetes (83). Obese Zucker rats had 40% lower plasma OXT levels, assessed by enzyme immunoassay (EIA), compared to lean Zucker rats (48).

4.2.2. ob/ob and db/db mice

The leptin deficient ob/ob mouse model shows extreme obesity together with high basal glycemia and insulinemia, thus presenting a model of an advanced stage in the development of type 2 diabetes. However, in males, their plasma OXT levels were similar to those of lean controls (1).

The leptin receptor-deficient db/db mouse is characterized by an obese phenotype with hyperglycemia and subsequent hyperinsulinemia as a result of two mutant copies of the leptin receptor gene. Male obese and diabetic db/db mice had lower (46%) serum OXT levels compared with control db+/mice (177).

In summary, circulating levels of OXT in genetic models of obesity are either increased (OLETF), decreased (Zucker rats, db/db mice) or not different (ob/ob mice) from lean controls. Whether differences in leptin signaling, genetic background, and/or OXT-degrading enzymes (48) contribute to these divergent results is an area for future research. Central expression/levels have been studied less frequently; obese OLETF rats showed lower OXT gene expression in the PVN. There are not enough data available on central expression levels to reach conclusions at this point.

4.3. High-fat/cafeteria DIO rodent models

HFD

Rats maintained on a chow or “cafeteria” diet weighed 14% more than chow-fed controls and had significantly higher fasting levels of plasma OXT measured by RIA (148). In contrast, lower levels of plasma and serum OXT have been reported at study termination in male HFD-fed DIO mice relative to chow-fed mice (275, 276). However, Maejima (120) did not see a decrease of plasma OXT in HFD-fed DIO mice at study termination relative to chow-fed mice [note that this may have been due to length of time the mice were on HFD (8 weeks (males) or 12 weeks (females) prior to study termination and blood collection relative to the other study (6 months) (275)]. In a study from our (Blevins) lab (137), we also did not see a decrease in serum OXT in male HFD-fed DIO rats (or in DIO mice (Blevins, unpublished) relative to low fat/chow diet-fed controls (though there was a non-significant tendency in the direction of a decrease in the rats). Comparing our (Blevins) null results and previous reports (275, 276), our mice were on the HFD for about the same amount of time.

Regarding brain OXT levels, in a recent, yet unpublished study in our (Weller) lab, male Wistar rats received 60% HFD from weaning (PND 21) to adulthood (PND 120). Preliminary data show that compared to rats raised on standard chow, these rats showed 3-fold higher OXT mRNA expression (measured by real time qPCR) in the hypothalamic PVN (Cohen-Or, Gerberg, Weller & Meiri, unpublished). However, using a much shorter period of exposure to this 60% HFD, 10 days, from PND 20 to PND 29 in male Sprague Dawley rats, significantly lower OXT levels (measured by ELISA) were found in the infralimbic medial prefrontal cortex, compared to chow-fed controls (267). Note that there were no significant BW differences between the two groups in this study. The two latter studies differ in many aspects, including brain area assessed, method of measurement and duration of HFD exposure, so clearly further research is warranted. Furthermore, Zhang and Cai used an ex vivo OXT release assay to determine if HFD feeding could affect OXT release in hypothalamic PVN and found that KCl-induced depolarization elicited OXT release in PVN slices of chow-fed wild type mice, but this effect was blunted by HFD feeding (276).

Summarizing this section, the overall data seem to be mixed in the cafeteria and HFD-induced obese rodent models, but this may be, at least in some cases, due to length of time on the high fat diet, time of blood collection (276), and/or species/strain differences.

5. Endogenous oxytocin levels and relationship to markers of energy homeostasis in humans

As in animal models, there is controversy regarding measurement of peripheral OXT levels in humans (119); however these assessments can be useful in comparing relative between-group levels of circulating OXT as well as the relationship between OXT and related clinical endpoints, such as appetitive behavior and markers of energy homeostasis (102). Consistent with the role of OXT as a signal of energy availability, serum OXT levels are positively associated with body mass index, leptin levels, and body fat in non-diabetic females across the weight spectrum (104, 199). Circulating OXT levels are low in states of energy deficit, for example in low-weight women with anorexia nervosa (104, 136) or normal-weight females with exercise-induced hypothalamic amenorrhea and low percent body fat (103) compared to controls. There is also evidence for higher peripheral OXT levels in individuals with excess energy stores, including overweight/obese women and men (166, 199, 222, 234, 254). However, in a number of studies of obese adults with a high prevalence of type 2 diabetes or insulin resistance (46, 125, 182) and children with insulin resistance (13), OXT levels were low; this may be directly related to impaired glucose homeostasis as individuals with type 1 or type 2 diabetes have low OXT levels (99, 182). Some studies in individuals with type 2 diabetes or insulin resistance and overweight/obesity have found that lower OXT levels are associated with increased features of metabolic syndrome (13, 46, 125, 272). However, one study of older men (mean weight 80 kg) -- a minority of whom had type 2 diabetes -- found that higher OXT levels were associated with greater weight and metabolic syndrome (234), and another of men and women across a wide spectrum of weights with insulin resistance in those with obesity reported higher OXT levels were associated with greater body mass index and insulin resistance (Homeostatic Model Assessment of Insulin Resistance HOMA-IR)] (166). Consistent with these data, a recent study of 721 adults across the weight spectrum (excluding participants with diabetes) found a positive association between OXT levels and HOMA-IR, which remained significant after adjusting for sex, age and body mass index (BMI) (254). Further research will be important to explain these divergent results and how metabolic status and BW impact OXT secretion. Whether low OXT levels in some individuals with overweight or obesity, e.g., in the context of type 2 diabetes, contribute to weight gain is an important area for future investigation.

Studies examining endogenous OXT levels in response to caloric intake in humans also show discrepant results, with no change (131, 222), increased (150), or decreased (7) postprandial levels reported, possibly related to differences in meal size, macronutrient or fiber content, meal form (liquid vs. solid) or in some cases, small sample size. A recent investigation of 55 healthy women and girls fed a balanced mixed meal showed a 20% postprandial decrease in OXT levels (7). Further, a smaller percent decrease in postprandial serum OXT was associated with a greater sense of fullness and less hunger after the 400 kcal mixed meal, suggesting that endogenous OXT secretion may modulate subjective appetite in humans (7).

6. Effects of IN oxytocin administration on energy homeostasis and metabolism in humans

Recent translational studies have shown that intranasal (IN) administration of supraphysiologic OXT decreases food intake and/or BW in obese humans (69, 105, 239, 277). A single dose of 24 IU IN OXT in men reduces caloric consumption without impacting subjective appetite (25, 105, 219, 239). Importantly OXT reduces food intake in fasted (105, 239) and fed (25, 165, 239) states with preferential effects in reducing sweet (25, 165, 239), fatty (105) and salty (25) foods, suggesting effects on both homeostatic and hedonic eating. Mechanistic studies using functional magnetic resonance imaging (fMRI) have confirmed that OXT modulates homeostatic (e.g., hypothalamus) (178, 245) and hedonic (e.g., ventral tegmental area, orbitofrontal cortex, insula) (178, 219) neurocircuitry in response to food images. In addition, fMRI studies have shown that OXT increases activation of brain regions responsible for self-control (e.g., anterior cingulate cortex, prefrontal cortex, supplementary motor area) (178, 219, 224). Therefore, OXT may reduce caloric intake in humans via effects on multiple neurobiological pathways, including homeostatic, hedonic and impulse control circuitry.

Based on data in animal models showing that OXT reduces body fat and induces weight loss in part via increased fat utilization and energy expenditure (previously discussed in section 3.1), human studies have recently examined whether IN OXT has similar effects in men. A single dose of 24 IU IN OXT reduced respiratory quotient in men, indicating increased fat utilization (105), but had no impact on resting energy expenditure (105, 239) or diet-induced thermogenesis (239). Only one small pilot study to date has investigated the chronic effects of IN OXT on BW in humans with overweight/obesity (277). In this randomized, placebo-controlled trial of 20 men and women, nine of whom completed OXT treatment, eight weeks of 24 IU IN OXT before meals and at bedtime resulted in approximately 20 lbs of weight loss and 10 cm waist circumference reduction (277). The effects of chronically administered OXT on eating behavior, fat metabolism and energy expenditure were not assessed. Currently, an NIH-funded study of 8 weeks of IN OXT for weight loss is underway in 60 adults with obesity ( NCT03043053). This study, which includes an equal number of males and females, will investigate the efficacy, safety and underlying mechanisms of chronic IN OXT (24 IU before meals and at bedtime) as a weight loss therapy.

As in animal models (see section 3.6), the effects of IN OXT on glucose homeostasis are inconsistent in human studies. A single dose of 24 IU IN OXT in men across the weight spectrum reduced fasting insulin without a change in glucose levels, consistent with improved insulin sensitivity (105); after a meal, OXT resulted in reduced glucose levels without a change in insulin, independent of caloric intake (165, 239). Interestingly, OXT attenuated the increase in insulin and glucose after a 75 gram oral glucose tolerance test in men of normal-weight (92) but not in those with obesity and insulin resistance (23), suggesting that the effects of exogenous OXT on glucose homeostasis may be impacted by an individual’s adiposity and/or glucose regulation status. This is supported by the finding that 8 weeks of IN OXT failed to improve fasting or postprandial levels of insulin or glucose after a 100 g standard meal despite weight loss in a small study of prediabetic men with overweight/obesity (277).

While the glucose regulatory effects of OXT may be reduced in the context of obesity (23), there is evidence that OXT effects in modulating homeostatic caloric intake (239) and weight (277) may be stronger in obesity. In a study of 18 men with obesity and 20 men with normal weight, a single dose of 24 IU IN OXT vs. placebo reduced caloric intake in the fasting state only in those with obesity, while reducing postprandial snacking in both groups (239). Further, in the pilot study of chronic IN OXT for weight loss in adults, those with obesity (BMI approximately 34–44 kg/m²) appeared to have greater weight reduction than those with overweight (277). Future studies will be important to improve our understanding of the impact of adiposity and metabolic status on OXT effects in humans. More investigation of women will also be critical as OXT is known to have sex-specific effects.

7. Therapeutic implications

Based on the effects of OXT in reducing caloric intake, increasing energy expenditure, decreasing metabolically unfavorable fat depots (e.g., visceral and possibly liver fat), and potentially improving glucose homeostasis, OXT-based therapeutics are under active investigation for weight loss in humans (102). Intravenous/intramuscular OXT is FDA-approved and has a long track record of clinical use in parturition (244). Risks in the setting of parturition include arrhythmias, changes in blood pressure (see section 7.1 below) and rarely hyponatremia (244). IN OXT was previously FDA-approved (at lower doses than studied for weight loss) for lactation in women and was discontinued due to business -- not safety -- concerns. It continues to be used outside of the U.S. and is generally considered to be safe and well-tolerated. However, safety of IN OXT in individuals with obesity at the doses required for weight loss will require systematic investigation. Rigorous randomized controlled trials will also be necessary to determine the efficacy and underlying mechanisms of OXT actions. Currently proof of concept trials of IN OXT in adults with obesity ( NCT03043053) and children/young adults with hypothalamic obesity due to brain tumors ( NCT02849743) are underway.

7.1. Effects of OXT treatment on blood pressure and heart rate in animals and humans

Anorexigenic agents that that are thought to activate SNS outflow, including MC3/4R ligands which act, in part, through downstream CNS OXT circuits (see section 2), may adversely affect cardiac function in rodents (100, 213), nonhuman primates (88) and humans (57). This is a potential concern given the large number of overweight and obese humans with hypertension (145). Current data suggest that PVN OXT neurons are anatomically positioned to control SNS outflow (77, 151, 208, 215, 228) and OXT null mice appear to have reductions in SNS tone (27). However, animals studies have provided mixed results demonstrating an increase, decrease or no change in heart rate (65, 118, 121, 142, 173, 177, 266, 269) and blood pressure (118, 121, 142, 173, 271) depending on species and route of OXT administration [for review see (58, 172)]. More recently, acute systemic injections of OXT at doses that suppressed food intake (3 mg/kg, IP), resulted in an increase in blood pressure in DIO Long-Evans rats (1 and 3 mg/kg, IP) while a more selective OXT analogue (OXT^GLY) had no effect (216). Studies using IN OXT in humans have not reported adverse side effects on heart rate (26, 91, 105) or blood pressure (91, 105) in men (26, 91, 105, 277) and non-pregnant women (91, 277). While the data are mixed in animal models (depending on species and route of administration) and IV/IM OXT in the context of childbirth may have effects on blood pressure and/or heart rate, preliminary studies in overweight/obese humans using IN OXT, at doses shown to reduce caloric intake (105) and BW (277), have not been shown to have an effect on either. It will be important to confirm in future studies that chronic administration of OXT-based therapeutics is not associated with adverse effects on blood pressure and heart rate in obese nonhuman primates and humans.

7.2. Remaining challenges

One limitation to the use of OXT as a therapeutic target to treat obesity is the short half-life in the periphery [T_1/2=3.2–6 min (39, 247)] relative to the CNS [_T1/2=19 min (130)]. There is a focus on developing novel therapeutics to target the OXT system as a way to circumvent the short half-life and prolong duration of action in the periphery. In some cases, modifications have been made to the disulfide bond (140) [essential for biological activity (140, 225)] to alter its activity and stability. Recent findings indicate that when encapsulated into a nanoparticle formulation, OXT may have improved stability and passage across the blood brain barrier (163, 274). One other modification has been to replace the Pro⁷ of OXT with either N-(p-fluorobenzyl)glycine or N-(3-hydroxypropyl)glycine which yielded long-lasting effects in an animal model of social behavior (71). Carbetocin is another analogue with modifications that include N-terminal desamination and replacement of the 1–6 disulfide bridge with a methylene group (70) and a reported half-life of 17.2 min in horses (201) and 41 min in non-pregnant women (232). Altirriba and colleagues have demonstrated that carbetocin reduces weight gain over a 10-day period in ob/ob mice (1), but it has not yet been examined in DIO rodent models. Novel OXT analogues have recently been produced by attaching a fatty acid to the peptide (acylated OXT) or replaced proline (position 7) with glycine in the acylated OXT to generate acylated OXT^Gly. Chronic systemic treatment (1× daily) with acylated OXT (sc, 2 μmol/kg) and acylated OXT^Gly (sc, 2 μmol/kg) over a 14-day period reduced food intake and BW in DIO C57BL/6 mice. As expected, these effects were also associated with reductions in fat mass. While acylated OXT^Gly exerted a protective effect on lean mass, acylated OXT actually produced a relative reduction in lean mass (216) (also see section 3.3). Snider also noted improvements in glucose tolerance following acute administration of another analogue, OXT^Gly, in both lean and DIO C57BL/6 mice (IP, 2 mg/kg) (also see section 3.6). In contrast, they found that chronic 14-day treatment with acylated OXT or acylated OXT^Gly failed to improve glucose tolerance in DIO C57BL/6 mice and, in fact, acylated OXT^Gly may have actually worsened glucose tolerance (216). Consistent with this finding, fasting plasma glucose tended to be higher following chronic treatment with both analogues (not significant) while fasting plasma insulin was significantly higher in the group treated with acylated OXT^Gly. Zhang and Cai found that 2× daily 3V administration of other analogues, [Ser4, Ile8]-OT (isotocin) and [Asu1,6]-OT, resulted in an improvement in glucose tolerance and fasting blood insulin levels that were independent of changes in BW in HFD-fed C57Bl/6J mice and streptozotocin-treated mice with diabetes (277). Importantly, from a translational perspective, the improvements in glucose tolerance independent of changes in BW were also observed following peripheral administration of [Ser4,Ile8]-OT (277). It will be helpful to include weight-restricted controls and examine the effects of these analogues on glucose tolerance following a more extended treatment at doses that are not associated with adverse side effects in both genetically obese and DIO rodent and DIO nonhuman primate models.

OXT is a pulsatile hormone (10, 47, 146), but the secretory patterns in humans are not well defined. Thus far, OXT administration multiple times per day, continuously, or during the early part of the light cycle to coincide with diurnal peaks of circulating OXT in rodent models (corresponds to time when rodents do not normally eat) appears to maximize its effects (16, 276). Further research into the secretory dynamics of OXT in normal human physiology may help guide optimal drug development and dosing. It would also be useful to examine if intermittent or pulsatile administration (30–32, 62, 278) of OXT in animal models will enhance its ability to promote weight loss and potentially overcome the reductions in OXTR binding that develop with chronic CNS infusions (43, 72, 171). However, further studies will be required to determine at what point reductions in OXTR binding occur following continuous OXT infusions and whether this precedes or coincides with the waning of OT’s effects on food intake. Mechanistic studies delineating central vs. peripheral effects of OXT will also provide important information. In vitro data indicate that OXT has a 150-, 1250- and 4375-fold greater affinity for the human OXTR relative to the vasopressin (V1aR), vasopressin 1B receptor (V1bR) and vasopressin 2 receptor (V2R), respectively (126). Similar specificity has also been observed for the rat and mouse species (personal communication with Dr. Gilles Guillon). While recent findings indicate that OXT failed to reduce food intake in OXT null mice (75), previous in vitro data indicate that OXT may also cross-react with the V2R receptor (216) and others have reported that OXT is about 10-fold more selective for the OXTR relative to VP (54, 179). Given the mixed results and the possibility for off-target effects, OXT agonists that are more selective to the OXTR could reduce the potential for side effects due to cross-reactivity with VP receptors. Ultimately using OXT-based therapeutics given alone or in combination with other interventions targeting other pathways (69) may be useful in improving efficacy while minimizing side effects.

8. Conclusions

Taken together, preclinical data indicate that OXT is involved in the regulation of eating behavior, energy expenditure, lipid metabolism, and glucose homeostasis. The anorexigenic properties of OXT appear to be mediated by homeostatic and reward-related neural pathways (4, 5, 15, 17, 64, 132, 138, 152, 153, 156, 205). Studies in humans support these mechanisms, and also indicate that OXT may act via modulating impulse control circuitry (178, 219, 245). While studies in animal models show that acute or chronic administration of OXT increases energy expenditure (16, 147, 275, 276), single dose IN OXT does not appear to impact energy expenditure in men (105, 239). The relevance of proposed mechanisms for OXT effects on energy expenditure in animal models (via sympathetic activation of BAT and WAT) may be less relevant in humans, and the question of whether chronic administration of OXT increases energy expenditure in humans is an important area for future investigation. OXT increases markers of lipolysis and reduces fat mass, including metabolically unfavorable visceral fat depots, in animals (1, 18, 34, 74, 120, 121, 137, 177, 191, 198, 216, 268, 276). In humans, IN has been shown to increase fat utilization when given as a single dose (105), and reduce waist circumference (a marker for visceral fat) when given chronically over eight weeks (277). Effects of OXT on glucose homeostasis are mixed in preclinical and human studies, and may depend on dose, delivery, and/or metabolic state. In humans, several studies have shown improvements in glucose homeostasis with a single dose of IN OXT (92, 105, 165, 239), however one investigation in obese men with insulin resistance did not demonstrate OXT effects in response to an oral glucose tolerance test (23). Further, a pilot study of eight weeks of IN OXT in obese adults showed no improvement in glucose homeostasis despite significant weight loss (277).

Based on translational data demonstrating that OXT induces weight loss, OXT-based therapeutics are considered promising for the treatment of obesity. Safety, efficacy and underlying mechanisms of these drugs will be important to investigate in patients with obesity.