Melanocortin agonism in a social context selectively activates nucleus accumbens in an oxytocin-dependent manner

Abstract

Social deficits are debilitating features of many psychiatric disorders, including autism. While time-intensive behavioral therapy is moderately effective, there are no pharmacological interventions for social deficits in autism. Many studies have attempted to treat social deficits using the neuropeptide oxytocin for its powerful neuromodulatory abilities and influence on social behaviors and cognition. However, clinical trials utilizing supplementation paradigms in which exogenous oxytocin is chronically administered independent of context have failed. An alternative treatment paradigm suggests pharmacologically activating the endogenous oxytocin system during behavioral therapy to enhance the efficacy of therapy by facilitating social learning. To this end, melanocortin receptor agonists like Melanotan II (MTII), which induces central oxytocin release and accelerates formation of partner preference, a form of social learning, in prairie voles, are promising pharmacological tools. To model pharmacological activation of the endogenous oxytocin system during behavioral therapy, we administered MTII prior to social interactions between male and female voles. We assessed its effect on oxytocin-dependent activity in brain regions subserving social learning using Fos expression as a proxy for neuronal activation. In non-social contexts, MTII only activated hypothalamic paraventricular nucleus, a primary site of oxytocin synthesis. However, during social interactions, MTII selectively increased oxytocin-dependent activation of nucleus accumbens, a site critical for social learning. These results suggest a mechanism for the MTII-induced acceleration of partner preference formation observed in previous studies. Moreover, they are consistent with the hypothesis that pharmacologically activating the endogenous oxytocin system with a melanocortin agonist during behavioral therapy has potential to facilitate social learning.

Graphical Abstract

1. Introduction

Deficits in social behavior and cognition are debilitating features of many psychiatric disorders, including autism spectrum disorder (ASD), schizophrenia, post-traumatic stress disorder (PTSD), and social anxiety disorder. One potential therapeutic target for treating these social deficits is the oxytocin system, which plays myriad roles in social behavior. Numerous studies have demonstrated prosocial effects of exogenous oxytocin administration, including increasing trust (Kosfeld et al., 2005), inclusion (Riem et al., 2013), communicational reciprocity (Spengler et al., 2017), empathy (Hurlemann et al., 2010), eye gaze (Guastella et al., 2008), social approach (Preckel et al., 2014), and altruism (Marsh et al., 2015). Although many oxytocin studies are underpowered (Quintana et al., 2021), show small effect sizes (Walum et al., 2016), fail to replicate (e.g., increasing trust; Lane et al., 2015; Declerck et al., 2020), or show context-dependent effects (Shamay-Tsoory & Abu-Akel, 2016), collectively, they are consistent with animal research demonstrating that oxytocin is a powerful modulator of the social brain (Froemke & Young, 2021; Marsh et al., 2021).

The promising preclinical oxytocin literature has prompted many clinical trials, most commonly using intranasal oxytocin in patients with ASD. While small, early trials were inconclusive (Anagnostou et al., 2012; Parker et al., 2017; Yamasue & Domes, 2018; Wang et al., 2019; Yamasue et al., 2020), a recent Phase II clinical trial definitively demonstrated that daily oxytocin administration has no positive effect on social behavior in children and adolescents with ASD (Sikich et al., 2021). One plausible reason this oxytocin supplementation paradigm failed is that while oxytocin enhances the salience of social information (Shamay-Tsoory & Abu-Akel, 2016; Xue et al., 2020) and facilitates social learning (Hurlemann et al., 2010; Eckstein et al., 2015), it does not necessarily cause prosocial behavior (Shamay-Tsoory et al., 2009; De Dreu et al., 2010 & 2011; Bartz et al., 2011; Ne’eman et al., 2016). That is, the potential therapeutic effects of oxytocin are context-dependent, which might explain some of the inter-study variability in the outcomes of both preclinical and clinical oxytocin research (Ford & Young, 2022).

A potentially more effective clinical application would be to pair oxytocin administration with a positive social learning experience like Applied Behavior Analysis (ABA) therapy (Meyer-Lindenberg et al., 2011; Ford & Young, 2022). ABA is the gold standard for treating social deficits in ASD, but it can require upwards of 40 hours per week to be effective (Eldevik et al., 2010; Linstead et al., 2017). Administering oxytocin immediately prior to ABA might improve the efficacy and efficiency of therapy by enhancing the salience of the therapeutic social information (Gordon et al., 2013; Gordon et al., 2016; Shamay-Tsoory & Abu-Akel, 2016; Xue et al., 2020), increasing the signal-to-noise ratio of that information as it is processed in the brain (Oettl et al., 2016; Froemke & Young, 2021), and accelerating learning by promoting synaptic plasticity at a molecular level (Dölen et al., 2013; Rajamani et al., 2018; Nardou et al., 2019; Froemke & Young, 2021). Notably, the goal of oxytocin-enhanced ABA therapy would not be to improve social behavior transiently through direct and immediate pharmacological activity, but rather to improve behavior more permanently by using oxytocin to enhance the learning and neural rewiring that occurs during behavioral therapy (Ford & Young, 2021 & 2022). One recent clinical trial demonstrated the potential of this therapeutic paradigm by showing a positive and enduring effect on social behavior in children with ASD when oxytocin was administered immediately prior to positive social interactions with their caregivers (Le et al., 2022).

Despite the potential of pairing oxytocin with a therapeutic context, the intranasal administration of exogenous oxytocin has unavoidable limitations. Oxytocin has a short half-life and it is unclear how long intranasal oxytocin remains active, how much enters the brain, and how far it diffuses (Quintana et al., 2021). Oxytocin receptors (OXTR) are widespread in brain (Quintana et al., 2019; Inoue et al., 2022), so exogenous administration likely has less-targeted effects than endogenous release from oxytocinergic neurons. For these reasons, using blood-brain-barrier-penetrant pharmaceuticals that activate the endogenous oxytocin system might be an improvement over using exogenous oxytocin (Modi & Young, 2012; Young & Barrett, 2015). More specifically, administering a pharmacological agent that selectively amplifies the physiological activity of oxytocinergic neurons during social interactions could allow for more targeted activation and therapeutic rewiring of social circuits and brain regions compared to activating OXTR everywhere exogenous oxytocin diffuses. One candidate pharmacological mechanism for activating the endogenous oxytocin system in this fashion is agonism of melanocortin receptors, specifically melanocortin 4 receptors (MC4R) (Modi et al., 2015; Peñagarikano et al., 2015; Mastinu et al., 2018). MC4R are expressed on oxytocinergic cells in the hypothalamic paraventricular nucleus (PVN) and supraoptic nucleus (SON) (Siljee et al., 2013), and stimulating MC4R activates oxytocin neurons (Kublaoui et al., 2008), increases their firing rate (Paiva et al., 2017), and induces central oxytocin release (Sabatier et al., 2003; Sabatier, 2006). Critically, MC4R agonism appears safe and well tolerated in humans, with melanocortin agonist Bremelanotide approved to treat hypoactive sexual desire disorder (Dhillon & Keam, 2019).

The nonspecific melanocortin receptor agonist and alpha-melanocyte stimulating hormone (α-MSH) analog Melanotan II (MTII) is a commercially-available metabolic precursor of Bremelanotide. MTII improves social behavior in a mouse model of ASD (Minakova et al., 2019), protects against social deficits caused by neonatal isolation (Barrett et al., 2015), and accelerates the formation of partner preference, a type of social learning indicative of monogamous pair bond formation, in prairie voles (Barrett et al., 2014; Modi et al., 2015). MTII can also increase the amount of oxytocin released in the nucleus accumbens (NAc), an area critical for social learning, but only when oxytocin release is triggered by a physiological stimulus (Modi et al., 2015). It is unknown, however, if MTII can increase neuronal activity in social brain regions. Here, in order to model oxytocin-enhanced therapy via melanocortin receptor agonism, we administered MTII to prairie voles in social and non-social contexts using Oxtr-knockout (Oxtr-KO) voles (Horie et al., 2019) and OXTR antagonist to identify oxytocin-dependent effects. We then assessed the ability of melanocortin receptor agonism in these contexts to enhance activity in brain regions that subserve social learning by quantifying the immediate early gene product Fos, a marker of neuronal activation, in PVN, NAc, basolateral amygdala (BLA), prelimbic cortex (PLC), and dorsal CA2 of hippocampus (CA2).

2. Methods

2.1. Subjects

All experiments used adult (75–186 days of age), sexually-naïve male (M) and female (F) prairie voles (Microtus ochrogaster) from our outbred colony at Emory University, which was originally derived from field-caught voles in Champaign, Illinois. Oxtr-KO voles and their wildtype (WT) littermates were generated from heterozygous breeders and genotyped as previously described (Horie et al., 2019). There were no significant group differences in the age of animals in any experiment, and age did not correlate with Fos in any analyzed brain region or behavior in the experiment using peripheral administration of MTII in a social context (sections 2.3 and 3.2). Voles were group-housed in ventilated 26 x 18 x 19 cm Plexiglass cages with Bedo’cobbs Laboratory Animal Bedding (The Andersons, Maumee, Ohio) and ad libitum access to food (Lab Rabbit Diet HF #5326, LabDiet) and water, and were maintained at 22 °C with a 14:10 hour light/dark cycle. All procedures were performed in accordance with the Institutional Animal Care and Use Committee at Emory University.

2.2. Peripheral MTII administration in a non-social context

Three days prior to experimentation, subjects were removed from group housing, weighed, placed in cages by themselves, and randomly assigned to either the MTII (n = 12 (6 M, 6 F)) or saline group (n = 12 (6 M, 6 F)). On the day of the experiment, voles received intraperitoneal (i.p.) injections of either 0.3 mL 0.9% saline (Pfizer, New York, NY) or 10 mg/kg MTII (Toronto Research Chemicals, Toronto, ON) dissolved in 0.9% saline and were then returned, alone, to their home cages. The 10 mg/kg dose was chosen for the effects it demonstrated in prior studies (Barrett et al., 2014; Modi et al., 2015). Two hours and 15 minutes after injection, voles were anesthetized with an overdose of isoflurane and transcardially perfused.

2.3. Peripheral MTII administration in a social context

Three days prior to experimentation, WT and Oxtr-KO littermates were removed from group housing, weighed, placed in cages by themselves, and randomly assigned to receive either saline (WT n = 14 (6 M, 8 F); Oxtr-KO n = 12 (6 M, 6 F)) or MTII (WT n = 13 (6 M, 7 F); Oxtr-KO n = 15 (7 M, 8 F)). On the day of the experiment, voles received i.p. injections of either 0.3 mL 0.9% saline or 10 mg/kg MTII and were then placed in clean test cages. Forty-five minutes after injection, a novel, opposite-sex stimulus animal (intact males, ovariectomized females) was placed in the test cage. The social interaction was videotaped (HF R800, Canon, Melville, NY) and later reviewed to ensure no fighting or mating occurred. Using BORIS software (version 7.13.9; Friard & Gamba, 2016) on a MacBook Pro running macOS (version 12.3.1, Apple, Cupertino, CA), the behavior of each experimental animal was scored for the amount of time it spent engaged in social investigation and autogrooming. Increased autogrooming, stretching, and yawning are behavioral side effects of melanocortin receptor agonism in many animals (Ferrari, 1958; Adan et al., 1999; Bertolini et al., 2009), which may be mediated in part by oxytocin or vasopressin signaling at vasopressin-1A receptors (Schorscher-Petcu et al., 2010). Thirty minutes after the stimulus animal was introduced, both animals were removed and the experimental animal was returned to its home cage. The experimental vole remained undisturbed for 1 hour, at which time it was anesthetized with an overdose of isoflurane and transcardially perfused.

2.4. Central MTII administration dose-response

To determine the optimal MTII dose for central administration, we examined the dose-response relationship for MTII-induced behavioral side effects. WT voles were anesthetized with isoflurane and 26-gauge guide cannulas (Plastics One, Roanoke, VA) were implanted targeting the right lateral ventricle (from Bregma: −0.4 mm A/P, +1.0 mm M/L, −2.5 mm D/V with 1 mm internal cannula protrusion) and secured with glass ionomer cement (Harvard Apparatus, Holliston, MA). After surgery, voles were singly housed. One week after cannula implantation, voles were lightly anesthetized with isoflurane and received an intracerebroventricular (i.c.v.) injection of 2 μL aCSF (artificial cerebrospinal fluid) (vehicle; Tocris Bio-Techne, Minneapolis, MN), 0.001 nmol MTII, 0.01 nmol MTII, 0.1 nmol MTII, or 1 nmol MTII dissolved in 2 μL aCSF (n = 6 (3 M, 3 F) per group). Following i.c.v. injection, voles were placed in clean test cages by themselves. From 45–60 minutes post-injection, their behavior was videotaped. Videos were scored for the time each animal spent ambulating, autogrooming, and stretching using Noldus Observer XT software (version 14, Noldus, Wageningen, Netherlands) on a Dell Precision 3630 computer running Windows 10 (Dell, Round Rock, TX). The time animals spent engaged in each behavior was graphed as a percent of the total time scored (Supplemental Figure 1A). Although autogrooming and stretching induced by melanocortin receptor agonism do not disrupt an organism’s ability to respond to social or environmental stimuli (Bertolini et al., 2009), spontaneous ambulation was almost entirely eliminated at the 1.0 nmol dose (Supplemental Figure 1B). Based on these data, 0.1 nmol MTII was selected for subsequent experiments with central administration as the highest dose at which the subject’s ability to engage socially with a stimulus animal was unlikely to be impaired.

2.5. Central MTII administration in a social context

WT voles were surgically implanted with guide cannula targeting the lateral ventricle as previously described and were singly housed thereafter. One week after surgery, subjects were lightly anesthetized with isoflurane and received i.c.v. injections of 2 μL aCSF (n = 20 (10 M, 10 F)), 0.1 nmol MTII dissolved in 2 μL aCSF (n = 19 (10 M, 9 F)), or 0.1 nmol MTII and 5 ng of the highly selective oxytocin antagonist (d(CH₂)₅¹,Tyr(Me)²,Thr⁴,Orn⁸,des-Gly-NH₂⁹)-Vasotocin (OTA; Bachem, Bubendorf, Switzerland) dissolved in 2 μL aCSF (n = 20 (10 M, 10 F)). Five ng of this OTA inhibits consoling behavior and mating-induced partner preference formation in prairie voles (Burkett et al., 2016; Johnson et al., 2016). Subjects were then placed in a clean test cage and, 45 minutes later, a novel, opposite-sex stimulus animal (ovariectomized females, intact males) was placed in the test cage. The social interaction was videotaped and later reviewed to ensure no fighting or mating occurred. Thirty minutes after the stimulus animal was introduced, the stimulus animal was removed and the experimental animal was left in the test cage, where it remained undisturbed for 1 hour. It was then anesthetized with an overdose of isoflurane and transcardially perfused.

2.6. Central MTII administration in a novel context

WT voles were surgically implanted with guide cannula targeting the lateral ventricle as previously described and were singly housed thereafter. One week after surgery, subjects were lightly anesthetized with isoflurane and received i.c.v. injections of either 2 μL aCSF (n = 10 (5 M, 5 F)) or 0.1 nmol MTII dissolved in 2μL aCSF (n = 15 (7 M, 8 F)). Subjects were then placed in a clean test cage and, after 45 minutes, a novel rodent figurine was placed in the cage. After 30 minutes, the figurine was removed, and the vole was left undisturbed for 1 hour. It was then anesthetized with an overdose of isoflurane and transcardially perfused.

2.7. Perfusion and sectioning

Immediately following isoflurane overdose, all subjects were transcardially perfused with 30–40 mL phosphate-buffered saline (PBS, pH 7.4; Teknova, Hollister, CA) followed by 30–40 mL 4% paraformaldehyde (Polysciences, Warrington, PA) in PBS at a rate of 4–8 mL per minute, depending on vole size, using a peristaltic pump (Easy-Load II Masterflex, Cole-Palmer, Vernon Hills, IL). Brains were extracted and post-fixed in 4% paraformaldehyde in PBS for 24 hours before being placed in 30% sucrose and 0.01% azide in PBS for at least one week prior to sectioning. Brains were then sliced into 40 μm coronal sections using a freezing microtome (HM 440 E, Microm, Walldorf, Germany) and stored in PBS with 0.01% azide until staining.

2.8. Fos immunohistochemistry

Every third section underwent immunohistochemical labeling of Fos at RT in custom 144- or 64-well staining trays on a rotator set to 100 revolutions per minute. Sections were washed three times in PBS, then incubated in a peroxidase blocking solution for 10 minutes (Bloxall, Vector Laboratories, Newark, CA), washed twice in PBS, and incubated for 20 minutes in 0.05% Triton X-100 (Sigma-Aldrich, St. Louis, MO) and 2.5% horse serum (ThermoFischer Scientific, Waltham, MA) in PBS (PBSTN). Sections were then incubated in an avidin-blocking solution (Vector Laboratories) for 15 minutes, washed twice in PBS, incubated in a biotin-blocking solution (Vector Laboratories) for 15 minutes, washed twice in PBS, and incubated for 30 minutes in primary rabbit monoclonal anti-c-Fos antibody (EPR21930-238, Abcam, Cambridge, UK) diluted to 1:6000 in PBSTN. After incubation in primary antibody, sections were washed seven times in PBS before being incubated for 30 minutes in biotinylated goat anti-rabbit IgG antibody (Vectastain Elite ABC-HRP Kit, Vector Laboratories) in PBS with 0.05% Triton X-100 and 1.5% horse serum according to the manufacturer’s instructions. Following secondary antibody incubation, sections were washed seven times in PBS and incubated in an avidin-biotin peroxidase system (Vectastain Elite ABC-HRP Kit, Vector Laboratories) for 30 minutes. Then, sections were washed seven times in PBS, stained with a nickel-DAB peroxidase substrate kit (SK-4100, Vector Laboratories), again washed seven times in PBS, and mounted onto Superfrost Plus slides (Fisher Scientific, Waltham, MA) in dH₂O. After drying, slides were coverslipped with VectaMount permanent mounting medium (Vector Laboratories). Due to slight variations in staining with each batch of immunohistochemistry, Fos scores from different experiments should not be compared to one another.

2.9. Nissl staining

Every third section underwent Nissl staining to facilitate identification of brain regions in Fos-labeled sections and to confirm that cannulas successfully targeted the lateral ventricle. Sections were mounted onto Superfrost Plus slides and allowed to dry before being submerged in xylene for 5 minutes, 95% ethanol for 3 minutes, 70% ethanol for 3 minutes, and dH₂O for 3 minutes. Sections were then stained for 12 minutes in dH₂O containing 0.01% cresyl-violet, 0.5% acetic acid, and 0.07% sodium acetate. After staining, sections were submerged in 70% ethanol for 3 minutes, 95% ethanol for 2 minutes, and dipped once in 100% ethanol before being submerged in xylene for 5 minutes. While wet with xylene, slides were coverslipped with Eukitt mounting medium (Sigma-Aldrich) and then allowed to dry.

2.10. Imaging

Slides were scanned with a NanoZoomer 2.0-HT (Hamamatsu Photonics, Shizuoka, Japan) using NDP.scan software (version 3.3, Hamamatsu Photonics), and images were captured with NDP.view2 software (Hamamatsu Photonics). Using Nissl-stained sections and a rat brain atlas (Paxinos & Watson, 2013) as references to determine anatomical boundaries, a blinded experimenter outlined PVN, NAc, BLA, PLC, and CA2 bilaterally in three consecutive Fos-labeled sections from each subject with ImageJ (version 1.52; Schindelin et al., 2012) on a MacBook Pro running macOS (version 12.3.1, Apple, Cupertino, CA). These regions were chosen for their high levels of Oxtr expression (Inoue et al., 2022) and their roles in social reward, learning, and memory (Walum & Young, 2018; Watarai et al., 2021). PVN, BLA, PLC, and anterodorsal CA2 were outlined in their entirety; for NAc, a medial sample consisting primarily of NAc shell was analyzed due to its dynamic role in social learning and pair bonding (Aragona et al., 2006; Resendez et al., 2012; Dreyer et al., 2016). Fos-positive nuclei were quantified using the ImageJ “Analyze Particles” function with the threshold held constant across subjects for each brain region in each experiment. For each section, the Fos-positive nuclei count was divided by the area of the region analyzed in mm², and the resulting values were averaged for each subject to yield a single value representing Fos-positive cell density in a given brain region. The Fos-labeled sections displayed in figures have Fos-positive cell densities that approximate the means for their respective groups. Damaged sections were excluded, as were subjects with less than two intact sections for a given region.

2.11. Statistics

All statistical analyses were performed in GraphPad Prism (version 9.4.1, GraphPad Software, San Diego, CA) using raw values for Fos-positive cell density (Fos-positive cell count per mm²). However, cell density varies widely between brain regions, so to facilitate visualization of the data, graphs with data from multiple brain regions present the fold difference between the control group (saline- or aCSF-injected animals) and experimental groups for each brain region. Similarly, due to inherent differences in cell density between brain regions, there was a significant effect of brain region, which is not reported, in all experiments. The ability of MTII to activate oxytocinergic PVN cells via melanocortin receptor agonism is well established in the literature (Kublaoui et al., 2006 & 2008; Barrett et al., 2014; Modi et al., 2015), so the PVN served as a positive control and was analyzed separately from the four oxytocin-sensitive target areas (Inoue et al., 2022). ANOVAs were performed for all experiments unless a value was missing from the dataset due to tissue damage, in which case a mixed-effects analysis was performed instead. Initial experiments in both social and non-social conditions found no sex effect or interaction, so data were collapsed across sex for those initial experiments and sexes were pooled for subsequent experiments. Oxytocin-sensitive target brain regions (NAc, BLA, PLC, CA2) were treated as repeated measures because Fos expression was highly correlated between brain regions within subjects (Supplemental Figure 2). These Pearson’s correlation coefficients and the resulting correlation matrices for Fos expression between brain regions within subjects were generated from z-scores with means and standard deviations pooled within each brain region. Behaviors (social investigation, autogrooming) during social interactions were analyzed with ANOVAs using the percent of time spent engaged in each behavior rather than raw values for time because some subjects were occluded from view for parts of the social interaction, and thus the total amount of time scored varied slightly between subjects. Pearson’s correlation coefficients were calculated to examine the relationship between NAc Fos expression and time spent engaged in social investigation.

3. Results

3.1. Melanocortin receptor agonism in a non-social context

To determine how a peripherally administered melanocortin receptor agonist affects the activation of social brain areas in a non-social context, we injected voles with 10 mg/kg MTII (n = 12 (6 M, 6F)) or saline (n = 12 (6 M, 6 F)) i.p. and returned them, alone, to their home cages. A one-tailed Welch’s t test comparing saline-injected and MTII-injected voles found a significant increase in Fos-positive cell density in the PVN of MTII-treated animals (M = 1889, SEM = 298) compared to saline-treated animals (M = 526, SEM = 83.8; t(12.7) = 4.41, p = .0004) (Figure 1A), as expected (Kublaoui et al., 2006 & 2008; Barrett et al., 2014; Modi et al., 2015). We compared Fos-positive cell density in NAc, BLA, PLC, and CA2 between MTII- and saline-injected groups using a three-way ANOVA with Geisser-Greenhouse correction, brain region as a within-subjects repeated measure, and sex and treatment as between-subjects factors. This analysis found no effect of sex (F (1, 20) = 0.084, p = .77) and no sex-brain region interaction (F (3, 60) = 0.18, p = .91), sex-treatment interaction (F (1, 20) = 0.63, p = .44), or sex-treatment-brain region interaction (F (3, 60) = 0.015, p > .99). Consequently, treatment was collapsed across sex and sexes were pooled in subsequent experiments with non-social contexts. The resulting two-way, repeated-measures ANOVA found neither a treatment effect (F (1, 22) = 1.17, p = .29) nor a treatment-brain region interaction (F (3, 66) = 0.63, p = .60), indicating that while peripheral MTII administration activates PVN cells, it does not significantly alter Fos immunoreactivity in the NAc, BLA, PLC, or CA2 (Figure 1B). Representative images of sections with Fos-positive cell densities that approximate their groups means are shown in Figure 2.

Figure 1. Peripheral MTII in a non-social context increases Fos in PVN, but not in other regions.

Voles were injected i.p. with either saline (n = 12 (6 M, 6 F)) or MTII (n = 12 (6 M, 6 F)) and returned to their home cages alone. A one-tailed Welch’s t test found a significant increase in Fos-positive cell density in MTII-injected voles (M = 1889, SEM = 298) compared to saline-injected voles (M = 526, SEM = 83.8; t(12.7) = 4.41, p = .0004) (A). In NAc, BLA, PLC, and CA2, a two-way, repeated-measures ANOVA failed to detect a treatment effect (F (1, 22) = 1.17, p = .29) or a treatment-brain region interaction (F (3, 66) = 0.63, p = .56) (B). Error bars show standard error of the mean. MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; i.p. = intraperitoneal; M = male; F = female; *** = p < .001.

Figure 2. Representative images of Fos-labeled brain regions after a non-social context.

Voles were injected i.p. with either saline or MTII and returned to their home cages alone. Representative sections of PVN, NAc, BLA, PLC, and CA2 with Fos-positive cell densities that approximate their group means are shown here. All scale bars = 250 μm. MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; L.V. = lateral ventricle; A.C. = anterior commissure; i.p. = intraperitoneal.

3.2. Melanocortin receptor agonism in a social context

Next, we injected 10 mg/kg MTII (n = 13 (6 M, 7 F)) or saline (n = 14 (6 M, 8 F)) i.p. and allowed voles to interact with a novel, opposite-sex vole for 30 minutes. A one-tailed Welch’s t test comparing saline- and MTII-injected voles again found a significant increase in Fos-positive cell density in the PVN of MTII-treated animals (M = 2917, SEM = 231) compared to saline-treated animals (M = 1713, SEM = 225; t(24.9) = 3.74, p = .0005) (Figure 3A). We compared Fos-positive cell density in NAc, BLA, PLC, and CA2 between MTII- and saline-injected groups using a three-way ANOVA with Geisser-Greenhouse correction, brain region as a within-subjects repeated measure, and sex and treatment as between-subjects factors. This analysis found no effect of sex (F (1, 23) = 0.13, p = .72) and no sex-brain region interaction (F (3, 69) = 0.34, p = .79), sex-treatment interaction (F (1, 23) = 2.08, p = .16), or sex-treatment-brain region interaction (F (3, 69) = 2.05, p = .12). Consequently, treatment was collapsed across sex and sexes were pooled in subsequent experiments with social contexts. The resulting two-way, repeated-measures ANOVA found no main effect of treatment (F (1, 25) = 0.37, p = .55), but in contrast to the experiment with a non-social context, this ANOVA found a treatment-brain region interaction (F (3, 75) = 10.6, p < .0001). A Holm-Šídák post hoc test was used to compare Fos-positive cell density within brain regions (Figure 3B) and revealed that in a social context, MTII significantly increases Fos immunoreactivity in NAc (p = .0034) and decreases Fos immunoreactivity in BLA (p = .021) compared to saline.

Figure 3. Peripheral MTII in a social context increases Fos in NAc in WT, but not Oxtr-KO, subjects.

WT and Oxtr-KO voles were injected i.p. with either saline (WT n = 14 (6 M, 8 F); Oxtr-KO n = 12 (6 M, 6 F)) or MTII (WT n = 13 (6 M, 7 F); Oxtr-KO n = 15 (7 M, 8 F)) and allowed to interact socially with a novel, opposite-sex vole for 30 minutes. A one-tailed Welch’s t tests found significant increases in Fos-positive cell density in the PVN of MTII-injected animals of both WT (M = 2917, SEM = 231; t(24.9) = 3.74, p = .0005) and Oxtr-KO (M = 2825, SEM = 305; t(18.2) = 5.44, p < .0001) genotypes compared to saline-injected animals (WT M = 1713, SEM = 225; Oxtr-KO M = 1038, SEM = 122) (A,C). A two-way, repeated-measures ANOVA analyzing NAc, BLA, PLC, and CA2 revealed a significant treatment-brain region interaction (F (3, 75) = 10.6, p < .0001) in WT voles, and a Holm-Šídák post hoc test found a significant increase in Fos immunoreactivity in NAc (p = .0034) and a significant decrease in Fos immunoreactivity in BLA (p = .021) in the MTII group compared to the saline group (B). In Oxtr-KO voles, a mixed-effects model found neither a treatment effect (F (1, 25) = 0.87, p = .36) nor a treatment-brain region interaction (F (3, 74) = 0.60, p = .61) (D). Error bars show standard error of the mean. MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; WT = wildtype; KO = knockout; i.p. = intraperitoneal; M = male; F = female; * = p < .05; ** = p < .01; *** = p < .001; **** = p < .0001.

Previous research with MTII demonstrated behavioral effects that were dependent on oxytocin signaling (Modi et al., 2015), so to explore if these social context-dependent neural effects were mediated by oxytocin, we repeated this experiment with Oxtr-KO voles from the same litters. A one-tailed Welch’s t test indicated PVN Fos immunoreactivity was significantly increased in MTII-injected voles (n = 15 (7 M, 8 F); M = 2825, SEM = 305) compared to saline-injected voles (n = 12 (6 M, 6 F); M = 1038, SEM = 122; t(18.2) = 5.44, p < .0001) (Figure 3C). Fos immunoreactivity in NAc, BLA, PLC, and CA2 were compared using a mixed-effects model with Geisser-Greenhouse correction, brain region as a within-subjects repeated measure, and treatment as a between-subjects factor. This analysis found neither a treatment effect (F (1, 25) = 0.87, p = .36) nor a treatment-brain region interaction (F (3, 74) = 0.60, p = .61) (Figure 3D).

To determine if the differences in NAc and BLA Fos immunoreactivity between WT and Oxtr-KO voles were mediated by oxytocin-dependent effects of MTII, we performed separate two-way ANOVAs for NAc and BLA with genotype and treatment as between-subjects factors. In NAc, there was a significant treatment-genotype interaction (F (1, 50) = 9.34, p = .0036) in addition to significant main effects of treatment (F (1, 50) = 8.32, p = .0058) and genotype (F (1, 50) = 4.05, p = .0496). A Holm-Šídák post hoc test found significantly increased Fos-positive cell density in MTII-injected WT animals compared to saline-injected WT animals (p = .0006), saline-injected Oxtr-KO animals (p = .0063), and MTII-injected Oxtr-KO animals (p = .0031), indicating that the MTII-induced increase in NAc Fos immunoreactivity observed only in WT animals depends on OXTR (Figure 4A). In BLA, the ANOVA found a main effect of treatment (F (1, 50) = 12.6, p = .0009) but no effect of genotype (F (1, 50) = 0.010, p = .92) or treatment-genotype interaction (F (1, 50) = 0.11, p = .75). A Holm-Šídák post hoc test found significant decreases in Fos-positive cell density in MTII-injected voles compared to saline-injected voles with both WT (p = .017) and Oxtr-KO (p = .028) genotypes, indicating that the MTII-induced decreases in BLA Fos immunoreactivity do not depend on OXTR (Figure 4B).

Figure 4. Peripheral MTII in a social context increases NAc Fos only in WT subjects without altering behavior compared to Oxtr-KO subjects.

WT and Oxtr-KO voles were injected i.p. with either saline (WT n = 14 (6 M, 8 F); Oxtr-KO n = 12 (6 M, 6 F)) or MTII (WT n = 13 (6, 7 F); Oxtr-KO n = 15 (7 M, 8 F)) and allowed to interact socially with a novel, opposite-sex vole for 30 minutes. A two-way ANOVA of NAc Fos-positive cell density data from both genotypes found a treatment-genotype interaction (F (1, 50) = 9.34, p = .0036), and a Holm-Šídák post hoc test found Fos immunoreactivity significantly increased in the MTII-injected WT group compared to the saline-injected WT (p = .0006), saline-injected Oxtr-KO (p = .0063), and MTII-injected Oxtr-KO (p = .0031) groups (A). A two-way ANOVA of BLA data from both genotypes found only a main effect of treatment (F (1, 50) = 12.6, p = .0009), and a Holm-Šídák post hoc test found Fos immunoreactivity decreased significantly both in the MTII-injected WT group compared to the saline-injected WT group (p = .017) and in the MTII-injected Oxtr-KO group compared to the saline-injected Oxtr-KO group (p = .028) (B). Two-way ANOVAs of subjects’ behavior during the social interaction revealed main effects of MTII treatment both on the time spent investigating the stimulus animal (F (1, 50) = 10.0, p = .0027) (C) and on the time spent autogrooming (F (1, 50) = 23.9, p < .0001) (D). Holm-Šídák post hoc tests revealed that in both genotypes, MTII reduced social investigation (WT, p = .047; Oxtr-KO, p = .047) and increased grooming (WT, p = .0020; Oxtr-KO, p = .0013) compared to saline. There was no main effect of genotype (social investigation: F (1, 50) < 0.0001, p = .99; autogrooming: F (1, 50) = 0.0012, p = .97) or a treatment-genotype interaction (social investigation: F (1, 50) = 0.015, p = .90; autogrooming: F (1, 50) = 0.08, p = .78) for either behavior. Error bars show standard error of the mean. MTII = Melanotan II; NAc = nucleus accumbens; BLA = basolateral amygdala; WT = wildtype; KO = knockout; i.p. = intraperitoneal; M = male; F = female; * = p < .05; ** = p < .01; *** = p < .001.

To determine if the differences in Fos expression could be attributed to differences in behavior between groups, we compared the amount of time subjects spent investigating the stimulus animal as well as the amount of time subjects spent autogrooming during the social interaction. Separate two-way ANOVAs with genotype and treatment as between-subjects factors found significant main effects of MTII treatment on both the percent of time subjects spent engaged in social investigation (F (1, 50) = 10.0, p = .0027) (Figure 4C) and the percent of time subjects spent autogrooming (F (1, 50) = 23.9, p < .0001) (Figure 4D). Neither ANOVA found a main effect of genotype (social investigation: F (1, 50) < 0.0001, p = .99; autogrooming: F (1, 50) = 0.0012, p = .97) or a treatment-genotype interaction (social investigation: F (1, 50) = 0.015, p = .90; autogrooming: F (1, 50) = 0.08, p = .78). Holm-Šídák post hoc tests found that in both genotypes, the average percent of time subjects spent engaged in social investigation significantly decreased from 47.3% (WT) and 46.7% (Oxtr-KO) with saline to 32.5% (WT) and 33% (Oxtr-KO) with MTII administration (WT, p = .047; Oxtr-KO, p = .047), while the average percent of time spent autogrooming significantly increased from 4.3% (WT) and 3.4% (Oxtr-KO) with saline to 21.9% (WT) and 23.2% (Oxtr-KO) with MTII administration (WT, p = .0020; Oxtr-KO, p = .0013). The similarity of the MTII treatment effects on behavior in both genotypes suggests that behavioral differences did not drive the MTII-induced increase in NAc Fos that occurred only in WT animals (Figure 4A). Furthermore, there was no significant correlation between NAc Fos and social investigation within any group (WT, saline: r(12) = .049, p = .87; WT, MTII: r(11) = −.18, p = .55; Oxtr-KO, saline: r(10) = −.040, p = .90; Oxtr-KO, MTII: r(13) = .24, p = .38) or within a pooled dataset containing all subjects (r(52) = −.14, p = .32) (Supplemental Figure 3). Representative images of sections with Fos-positive cell densities that approximate their groups means are shown in Figure 5.

Figure 5. Representative images of Fos-labeled brain regions after a social context.

WT and Oxtr-KO voles were injected i.p. with either saline or MTII and allowed to interact with a novel, opposite-sex vole for 30 minutes. Representative sections of PVN, NAc, and BLA with Fos-positive cell densities that approximate their group means are shown here. All scale bars = 250 μm. WT = wildtype; KO = knockout; MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; L.V. = lateral ventricle; A.C. = anterior commissure; i.p. = intraperitoneal.

3.3. Melanocortin receptor agonism in a social context with central administration

Next, we sought to replicate the social context-dependent effects of melanocortin receptor agonism with i.c.v. administration of aCSF (n = 20 (10 M, 10 F)), 0.1 nmol MTII (n = 19 (10 M, 9 F)), or 0.1 nmol MTII combined with OTA (MTII+OTA) dissolved in aCSF (n = 20 (10 M, 10 F)). There were two purposes of this experiment. The first was to determine if the changes in Fos expression observed with i.p. administration were due to indirect effects of MTII or oxytocin acting in the periphery rather than in the brain. For example, there is some evidence that MTII activation of peripheral receptors increases Fos expression in the nucleus of the solitary tract, which projects to and may regulate the activity of hypothalamic oxytocin neurons (Paiva et al., 2017). Second, we wished to confirm that the absence of a NAc Fos increase in Oxtr-KO voles reflected the oxytocin-dependence of this effect rather than abnormal physiology arising from congenital Oxtr knockout. Brown-Forsythe and Welch ANOVA tests found a significant effect of treatment on Fos-positive cell density in PVN (W (2.0, 29.8) = 11.5, p = .0032), and Dunnett’s T3 multiple comparisons test found significant increases in both the MTII group (p = .0056) and MTII+OTA group (p = .0040) compared to the aCSF group (Figure 6A). We compared Fos-positive cell density in NAc, BLA, PLC, and CA2 using a mixed-effects model with Geisser-Greenhouse correction, brain region as a within-subjects repeated measure, and treatment as a between-subjects factor. This revealed a significant treatment-brain region interaction (F (6, 163) = 3.58, p = .0023) as well as a significant treatment effect (F (2, 56) = 5.03, p = .0098). A Holm-Šídák post hoc test revealed that Fos immunoreactivity in NAc was significantly increased in the MTII group compared to both the aCSF (p = .0029) and MTII+OTA (p = .011) groups, but no significant differences were found in other brain regions (Figure 6B). Although BLA Fos immunoreactivity was lower in both the MTII and MTII+OTA groups compared to the aCSF group, the differences were not statistically significant (aCSF vs. MTII, p = .18; aCSF vs. MTII+OTA, p = .36). These results are consistent with those of the peripheral-injection experiments and confirm that the MTII-induced increase in NAc Fos immunoreactivity that occurs in a social context requires oxytocin signaling. Representative images of sections with Fos-positive cell densities that approximate their groups means are shown in Figure 7.

Figure 6. Central MTII in a social context induces an oxytocin-dependent increase in NAc Fos.

Voles were injected i.c.v. with either aCSF (n = 20 (10 M, 10 F)), MTII (n = 19 (10 M, 9 F)), or MTII combined with OTA (MTII+OTA; n = 20 (10 M, 10 F)) and allowed to interact socially with a novel, opposite-sex vole for 30 minutes. Brown-Forsythe and Welch ANOVA tests found a significant treatment effect in PVN (W (2.0, 29.8) = 11.5, p = .0032), and Dunnett’s T3 multiple comparisons test revealed significant increases in Fos immunoreactivity in both the MTII (p = .0056) and MTII+OTA (p = .0040) groups compared to the aCSF group (A). A repeated-measures mixed-effects model analyzing NAc, BLA, PLC, and CA2 found a significant treatment-brain region interaction (F (6, 163) = 3.58, p = .0023), and a Holm-Šídák post hoc test indicated Fos immunoreactivity in NAc was significantly increased in the MTII group compared to both aCSF (p = .0029) and MTII+OTA (p = .011) groups (B). Error bars show standard error of the mean. aCSF = artificial cerebrospinal fluid; MTII = Melanotan II; OTA = oxytocin antagonist; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; i.c.v. = intracerebroventricular; M = male; F = female; * = p < .05; ** = p < .01.

Figure 7. Representative images of Fos-labeled brain regions after a social context and i.c.v. injection.

Voles were injected i.c.v. with either aCSF, MTII, or MTII+OTA and allowed to interact with a novel, opposite-sex vole for 30 minutes. Representative sections of PVN and NAc with Fos-positive cell densities that approximate their group means are shown here. All scale bars = 250 μm. aCSF = artificial cerebrospinal fluid; MTII = Melanotan II; OTA = oxytocin antagonist; PVN = paraventricular nucleus; NAc = nucleus accumbens; L.V. = lateral ventricle; A.C. = anterior commissure; i.c.v. = intracerebroventricular.

3.4. Melanocortin receptor agonism in a novel context

To address the possibility that the context-dependent effects of MTII were due to the novelty of the social stimulus rather than its sociality, we allowed voles to interact for 30 minutes with a novel rodent figurine rather than a novel conspecific after administering i.c.v. either 0.1 nmol MTII (n = 15 (7 M, 8 F)) or aCSF (n = 10 (5 M, 5 F)). A one-tailed Welch’s t test again indicated PVN Fos immunoreactivity was significantly increased in MTII-injected voles (M = 829, SEM = 168) compared to aCSF-injected voles (M = 436, SEM = 125; t(22.9) = 1.88 p = .037) (Figure 8A). We compared Fos-positive cell density in NAc, BLA, PLC, and CA2 using a mixed-effects model with Geisser-Greenhouse correction, brain region as a within-subjects repeated measure, and treatment as a between-subjects factor. This analysis found neither a treatment effect (F (1, 23) = 0.81, p = .38) nor a treatment-brain region interaction (F (3, 65) = 0.13, p = .94) (Figure 8B). These data are consistent with the hypothesis that the context-dependent effects of MTII observed in previous experiments require a social stimulus rather than simply a novel stimulus. Representative images of sections with Fos-positive cell densities that approximate their groups means are shown in Figure 9.

Figure 8. Central MTII in a novel context does not increase NAc Fos.

Voles were injected i.c.v. with either aCSF (n = 10 (5 M, 5 F)) or MTII (n = 15 (7 M, 8 F)) and exposed to a novel rodent figurine for 30 minutes. A one-tailed Welch’s t test indicated PVN Fos immunoreactivity was significantly increased in the MTII group (M = 829, SEM = 168) compared to the aCSF group (M = 436, SEM = 125; t(22.9) = 1.88, p = .037) (A). In NAc, BLA, PLC, and CA2, a mixed-effects model failed to detect a treatment effect (F (1, 23) = 0.81, p = .38) or a treatment-brain region interaction (F (3, 65) = 0.13, p = .94) (B). Error bars show standard error of the mean. aCSF = artificial cerebrospinal fluid; MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; i.c.v. = intracerebroventricular; M = male; F = female; * = p < .05.

Figure 9. Representative images of Fos-labeled brain regions after a novel context and i.c.v. injection.

Voles were injected i.c.v. with either aCSF or MTII and allowed to interact with a novel rodent figurine for 30 minutes. Representative sections of PVN and NAc with Fos-positive cell densities that approximate their group means are shown here. All scale bars = 250 μm. aCSF = artificial cerebrospinal fluid; MTII = Melanotan II; PVN = paraventricular nucleus; NAc = nucleus accumbens; L.V. = lateral ventricle; A.C. = anterior commissure; i.c.v. = intracerebroventricular.

4. Discussion

Our data are the first to demonstrate that combining melanocortin agonism with a social context induces a synergistic, selective, oxytocin-mediated increase in NAc activity. Consistent with previous research on MTII (Barrett et al., 2014; Modi et al., 2015) and other MC4R agonists (Kublaoui et al., 2008), MTII administration increased Fos expression in PVN, which sends dense oxytocinergic projections to NAc (Ross et al., 2009; Ross & Young, 2009). However, MTII only increased NAc Fos in a social context. This increase required oxytocin signaling, but it was not driven by changes in social behavior. MTII may also have decreased BLA Fos in a social context independent of oxytocin signaling, though the effect size was small and not statistically significant in some experiments.

Previously, Modi et al. (2015) showed that MTII can accelerate partner preference formation, a form of social learning indicative of pair bond formation, in monogamous prairie voles. They further showed that while MTII administration alone does not cause oxytocin release in NAc, MTII increases the amount of oxytocin released in the NAc when release is triggered by a physiological stimulus. A wide range of social stimuli are also believed to induce oxytocin release (Uvnäs-Moberg, 1998; Dobloyi et al., 2018), including social touch (Tang et al., 2020) and mutual eye gaze (Nagasawa et al., 2015). In mice, a five-minute exposure to an anesthetized juvenile is sufficient to activate oxytocin neurons in PVN (Resendez et al., 2020), and in voles, oxytocin release in NAc increases during both sexual and nonsexual social interactions (Ross et al., 2009). Therefore, it is plausible that the 30-minute social interactions in our experiments triggered oxytocin release in NAc, and that MTII increased the amount of oxytocin released by this social stimulus. This mechanism likely mediates the oxytocin- and social context-dependent increases in NAc activation we observed in the present study. Furthermore, this increase in NAc activation may represent a neural correlate of the MTII-induced acceleration of social learning reported by Modi et al. (2015). Indeed, Modi et al. found that a microinfusion of OTA into the NAc prevented partner preference formation following peripheral MTII injection.

The enhancement of oxytocin-dependent NAc activity has important translational implications due to the critical role the NAc and oxytocin play in social and reward learning in both humans (Cohen et al., 2009; Gordon et al., 2013) and voles (Keebaugh et al., 2015). Many forms of social learning appear to be mediated by dynamic NAc physiology involving oxytocin and its interaction with other neurotransmitter systems. Oxytocin and serotonin signaling interact in NAc to induce synaptic plasticity and confer rewarding properties to social interactions (Dölen et al., 2013). Furthermore, synaptic plasticity in NAc mediated by oxytocin-serotonin interactions can re-open a critical period for social reward learning in adult mice (Nardou et al., 2019). In prairie voles, concurrent activation of NAc OXTR and D2-type dopamine receptors is necessary for pair bond formation (Liu & Wang, 2003). This pair bonding then causes an upregulation of NAc D1-type dopamine receptors, which mediate selective aggression toward non-partner voles to help maintain the pair bond (Aragona et al., 2006). Additionally, pair bonding reverses the electrophysiological effect of OXTR agonism on NAc medium spiny neurons; OXTR agonism decreases the amplitude of excitatory postsynaptic currents in virgins but increases their amplitude in pair-bonded voles (Borie et al., 2022b). This experience-induced change in NAc physiology is mediated by the coupling of OXTR and endocannabinoid receptor signaling, and it appears to facilitate the subsequent expression of behaviors that maintain the pair bond (Borie et al., 2022a). Collectively, these findings indicate that the enhancement of oxytocin-dependent NAc activity is an ideal target for translational interventions seeking to facilitate social learning, rewire neural circuitry, and effect lasting behavioral changes. This is particularly true of social learning intended to meliorate the social deficits of ASD, in which decreased social reward may be a core etiological component (Dölen, 2015; Clements et al., 2018; DeMayo et al., 2019).

It is crucial to note that the oxytocin- and context-dependent neural effects of MTII were not the result of behavioral changes. When MTII or saline was administered peripherally in a social context, there were no behavioral differences between WT and Oxtr-KO subjects in either the MTII or saline treatment groups. MTII administration increased autogrooming at the expense of social investigation equally in both WT and Oxtr-KO subjects, but only WT animals that received MTII had an increase in NAc Fos expression. These findings indicate that the increase in NAc Fos was not the result of behavioral differences between genotypes, nor was it the result of MTII-induced changes in behavior. Rather, it was most likely caused by MTII-induced changes in oxytocin-dependent neural activity. This is consistent with the hypothesis that melanocortin receptor agonism increases OXTR engagement during social encounters, thus modulating neural circuits to enhance the salience of social stimuli (Modi et al., 2015; Young & Barrett, 2015). The resulting increase in NAc activation for a given level of social stimulation is a plausible underlying mechanism for the accelerated pair bonding Modi et al. (2015) observed. If pharmacological melanocortin agonism during a social interaction does indeed accelerate social learning by amplifying oxytocin-dependent NAc activation, it would suggest a method by which behavioral or cognitive therapies for social deficits could be enhanced (Young & Barrett, 2015; Ford & Young, 2022).

The selectivity with which MTII enhanced NAc activation without affecting Fos expression in PLC or CA2 also carries important translational implications. First, a therapeutic intervention that causes targeted release of endogenous oxytocin would minimize the risk of receptor desensitization from chronic administration of exogenous oxytocin (Rajagopal & Shenoy, 2018; Freeman et al., 2018; Le et al., 2022). Second, interventions with more targeted neural effects may be less likely to have undesired off-target effects (Ford & Young, 2021). OXTR are widespread throughout the brain; in prairie voles, Oxtr mRNA is expressed in over 250 distinct brain regions (Inoue et al., 2022). In humans, the spatial distribution of OXTR mRNA, which includes NAc, is highly correlated with the fMRI-based localization of diverse cognitive processes ranging from aversive states like anxiety, stress, and fear, to appetitive states involving sexual and food stimuli, and anticipatory states involving motivation, incentive, and reward (Quintana et al., 2019). The widespread distribution of OXTR and the disparate behavioral and cognitive functions they subserve may contribute to the divergent and at times contradictory effects of intranasal oxytocin. Although most studies have focused on its prosocial effects, intranasal oxytocin can also have antisocial effects, like increasing feelings of envy and schadenfreude (taking pleasure in the misfortune of others) in a game of chance (Shamay-Tsoory et al., 2009) and increasing aggressive behavior in which participants in a monetary game choose to hurt their opponent rather than help themselves (Ne’eman et al., 2016). Intranasal oxytocin can also promote ethnocentric bias by increasing favoritism for one’s ethnic in-group and derogation of an ethnic out-group (De Dreu et al., 2011), and it can decrease trust and cooperation in people with borderline personality disorder (Bartz et al., 2011). Although the importance of controlling context in oxytocin-based therapeutic interventions cannot be overstated (Hurlemann, 2017; Ford & Young, 2022), an intervention that selectively activates a specific brain area via the endogenous oxytocin system may have more reliably therapeutic effects than intranasal administration of exogenous oxytocin.

There are, however, some important limitations to our study. First, we did not test the ability of MTII to enhance social learning. As such, we cannot conclude that the neural effects we observed would correlate with enhanced social learning, though the similarities between our study and that of Modi et al. (2015) implicate our findings as a plausible mechanism underlying the MTII-induced acceleration of partner preference formation Modi et al. demonstrated. Furthermore, while partner preference and pair bond formation in voles are instances of social learning with potential translational utility for ASD (Modi & Young, 2011), they might not be valid models for certain forms of social learning in humans. Nevertheless, the social learning that occurs in pair bonding requires many oxytocin-dependent processes, including attending to social cues, determining the valence of social stimuli, and forming social memories (Keebaugh et al., 2015; Walum & Young, 2018; Rigney et al., 2022). As such, partner preference formation is a useful behavioral readout for assessing the ability of pharmacological manipulations to activate the endogenous oxytocin system, independent of its translational utility as a model for any particular form of social learning (Modi & Young, 2012).

Second, the doses of MTII we used caused behavioral side effects, specifically stretching stereotypies and increased autogrooming. The time MTII-treated subjects spent engaged in these behaviors was most likely responsible for the reduction in time spent interacting socially with stimulus animals (Figure 4C,D). Although these effects are undesirable from a translational perspective, the goal of melanocortin-assisted behavioral therapy would not be to enhance social behavior during therapy, but rather to enhance oxytocin-dependent social learning during therapy so as to improve social functioning outside of therapy and in the absence of the drug (Ford & Young, 2022). To that end, it is encouraging that we saw increased oxytocin-dependent NAc activation despite decreased time engaged in social investigation, in addition to prior evidence that this dose accelerates the social learning of pair bond formation (Modi et al., 2015). Nevertheless, future studies should seek to identify a more translationally applicable dose of MTII that minimizes behavioral side effects while preserving its ability to accelerate social learning (Modi et al., 2015) and increase oxytocin-dependent NAc activity in a social context.

Third, our quantification of Fos was agnostic to cell type. While we can make inferences based on the general functions of anatomical brain regions, we cannot make inferences about the specific functions of activated cells or how they might affect the activity of the brain regions in which they were located. For example, it is possible that the increase in Fos-positive NAc cells was driven disproportionately by activation of D1-type dopamine receptor-expressing cells rather than those expressing D2-type dopamine receptors, or vice versa. Such a finding would have important behavioral implications. Similarly, it is also possible that the increase in NAc Fos was driven by the activation of many GABAergic interneurons, which could have resulted in net inhibition of NAc activity despite an increase in the number of Fos-positive cells. However, this particular scenario is unlikely given that GABAergic interneurons comprise less than five percent of NAc neurons (Tepper et al., 2018). Conversely, if MTII caused an important physiological change by activating a small number of highly influential cells, our Fos quantification would not have detected such an effect.

There are also important limitations to the translational potential of MTII stemming primarily from its promiscuity as a ligand and the widespread distribution and diverse functions of melanocortin receptors. MTII binds to four of the five melanocortin receptors: MC1R, MC4R, MC3R, and MC5R in order of decreasing affinity. MC1R are primarily expressed in peripheral melanocytes (Mountjoy, 2010). MC5R are mostly located in the periphery with some expression in the brain, including PVN, and they contribute to immune system modulation, exocrine function, and metabolism (Shukla et al., 2012; Xu et al., 2020). MC3R, which play a role in feeding and metabolism, are found in many brain areas but are particularly concentrated in the thalamus, the ventral tegmental area, and several nuclei of both the amygdala and hypothalamus (Bedenbaugh et al., 2022). They are expressed at relatively low levels in the BLA and PVN, although the PVN receives abundant projections from MC3R-expressing neurons (Bedenbaugh et al., 2022). MC4R are involved in feeding, metabolism, and reproduction and are even more widely expressed in the brain than MC3R, though they are most highly concentrated in the brainstem, numerous hypothalamic nuclei including PVN, and several amygdala nuclei including BLA (Gelez et al., 2010; Mountjoy, 2010; Modi et al., 2015). Both MC3R and MC4R are expressed in NAc, where their activation appears to suppress appetite and food intake (Mountjoy, 2010; Lim et al., 2012; Modi et al., 2015; Pandit et al., 2015; Eliason et al., 2022). Although the prairie vole NAc contains MC4R (Modi et al., 2015), NAc MC4R are unlikely to be responsible for the NAc Fos increases observed in the present study because NAc Fos did not increase when MTII was administered in a nonsocial context, nor did it increase when MTII was administered and oxytocin signaling was prevented by congenital Oxtr knockout or pharmacological OXTR antagonism.

As a consequence of the expansive distribution and functionality of melanocortin receptors, melanocortin receptor agonism has the potential for numerous off-target effects. Melanocortin receptor agonism has been reported to: increase skin pigmentation, sexual arousal, stretching, and yawning (Bertolini et al., 2009); reduce appetite (Bertolini et al., 2009); elevate blood pressure and heart rate, which can lead to nausea and flushing (Li et al., 2013; Dhillon & Keam, 2019); induce histamine release and hypothermia, though this has only been reported in mice (Jain et al., 2018); possibly increase the risk of more serious conditions including cutaneous (Habbema et al., 2017) and renal (Peters et al., 2020) complications, though the current evidence is anecdotal and inconclusive. Despite these concerns, off-target effects may be greatly reduced or even eliminated with careful dose calibration and intermittent (e.g., before ABA therapy sessions) rather than chronic administration. For example, Dhillon & Keam (2019) explain that in developing Bremelanotide, an active metabolite of MTII, “some patients reported an increase in blood pressure, which was attributed to variability in drug uptake with intranasal administration.” Switching to subcutaneous administration to achieve more consistent dosing resolved this problem. Indeed, the recent FDA approval of Bremelanotide to treat hypoactive sexual desire disorder indicates that melanocortin agonists like MTII can be safe and well-tolerated in humans with careful dose calibration (Dhillon & Keam, 2019). Of course, whether MTII would be safe and well-tolerated at efficacious doses for the enhancement of behavioral or cognitive therapy remains to be seen.

Future translational studies utilizing melanocortin agonists should, if possible, consider using Bremelanotide given the relative ease of repurposing drugs that have already been approved by the FDA. Additionally, to further minimize off-target effects, the possibility of using or developing more selective MC4R agonists should be explored. It is also important to note that the mechanism by which melanocortin agonism increases oxytocin release in NAc only when release is triggered by another stimulus (Modi et al., 2015) remains to be elucidated. It has been suggested that MC4R agonism causes somatodendritic release of oxytocin (Sabatier et al., 2003), which acts in an autocrine and paracrine manner to synchronize oxytocin neurons and prime oxytocin vesicles for increased activity-dependent release when subsequent physiological stimulation occurs (Ludwig & Leng, 2006; Ludwig & Stern, 2015; Modi et al., 2015; Ludwig et al., 2016). However, the ability of melanocortin agonism to trigger somatodendritic release of oxytocin in vivo has been inconsistent; in one study, neither systemic (intravenous) nor i.c.v. administration of MTII caused somatodendritic oxytocin release in SON, but retrodialysis of α-MSH into SON did induce somatodendritic oxytocin release (Paiva et al., 2017). Notably, no studies have examined melanocortin-induced somatodendritic release of oxytocin in PVN, and the sparse expression of Oxtr mRNA in vole PVN further decreases the likelihood of this autocrine priming mechanism (Inoue et al., 2022). Focal knock-down of OXTR in PVN and SON would help elucidate the role of autocrine and paracrine oxytocin signaling in the effects of melanocortin receptor agonism (Boender & Young, 2020).

5. Conclusions

Despite lingering questions regarding the underlying mechanism, our results provide a proof of concept that combining a pharmaceutical manipulation with a particular environmental stimulus can influence neuronal activity in ways that neither the pharmaceutical nor the environmental stimulus can alone. That is, pharmacology and environmental stimuli can interact synergistically. Thus, this study provides additional support for the argument that oxytocin therapies should be paired with a therapeutic social context (Young & Barrett, 2015; Shamay-Tsoory & Abu-Akel, 2016; Hurlemann, 2017; Ford & Young, 2022; Le et al., 2022). Critically, we demonstrated that such multimodal interventions can have targeted effects restricted to specific brain areas. Developing more targeted treatments is a key goal for translational neuroscience (Ford & Young, 2022), and as such, other possible synergies between pharmacology and environmental stimuli should be explored. Arguably the most successful combination of pharmacology and environmental stimuli to date is 3,4-methylenedioxymethamphetamine (MDMA)-assisted psychotherapy for PTSD, which received Breakthrough Therapy Designation from the FDA in 2019 (Feduccia et al., 2019) and has since passed a Phase III clinical trial with very promising results recently published in Nature Medicine (Mitchell et al., 2021). It should be noted that the authors of this trial suggest the remarkable efficacy of MDMA-assisted therapy may be mediated by MDMA “reopening an oxytocin-dependent critical period of neuroplasticity” (Mitchell et al., 2021), which can occur in NAc for critical periods of social learning (Nardou et al., 2019). Given the ability of MTII to accelerate the social learning of pair bond formation (Modi et al., 2015) and enhance oxytocin-dependent NAc activity in a social context, melanocortin agonist-assisted behavioral therapy could be a viable therapeutic paradigm for treating social deficits in ASD and other disorders.

Supplementary Material

3

Supplemental Figure 1. Dose-response of MTII and behavioral side effects. Voles were injected i.c.v. with 2 μL of aCSF, 0.001 nmol MTII, 0.01 nmol MTII, 0.1 nmol MTII, or 1 nmol MTII dissolved in aCSF (n = 6 (3 M, 3 F) per group). Their behavior was filmed for 15 minutes and scored for the time they spent ambulating, autogrooming, and stretching. The mean percent of time each group spent engaged in these behaviors is shown (A), as is the percent of time each individual subject spent ambulating (B). Notably, ambulation was greatly reduced at the 1.0 nmol MTII dose. Error bars show standard error of the mean. MTII = Melanotan II; aCSF = artificial cerebrospinal fluid; i.c.v. = intracerebroventricular; M = male; F = female.

4

Supplemental Figure 2. Correlations of Fos expression between brain regions within subjects. Fos expression levels were highly correlated (red) between target brain regions (NAc, BLA, PLC, CA2) within each subject regardless of context (non-social A,B; social C,D) or drug treatment (saline A,C; MTII B,D). While PVN Fos demonstrated less correlation with other regions, Fos expression levels in target brain areas were not independent of one another. Consequently, data from target areas were analyzed as repeated measures in ANOVAs. Numbers shown are Pearson’s correlation coefficients. PVN = paraventricular nucleus; NAc = nucleus accumbens; BLA = basolateral amygdala; PLC = prelimbic cortex; CA2 = CA2 of hippocampus; MTII = Melanotan II.

Highlights.

• Melanocortin agonist Melanotan II activates hypothalamic paraventricular nucleus

• Melanotan II activates nucleus accumbens only when paired with a social context

• This context-dependent nucleus accumbens activation requires oxytocin signaling

• Combining pharmacology and context had selective, synergistic neural effects

• This pharmacology-context synergy has therapeutic potential for social deficits