Social experience alters oxytocinergic modulation in the nucleus accumbens of female prairie voles

Summary

Social relationships are dynamic and evolve with shared and personal experiences. Whether the functional role of social neuromodulators also evolves with experience to shape the trajectory of relationships is unknown. We utilized pair bonding in the socially monogamous prairie vole as an example of socio-sexual experience that dramatically alters behaviors displayed toward other individuals. We investigated oxytocin-dependent modulation of excitatory synaptic transmission in the nucleus accumbens as a function of pair bonding status. We found that an oxytocin receptor agonist decreases the amplitude of spontaneous Excitatory Postsynaptic Currents (EPSCs) in sexually naive virgin, but not pair-bonded, female voles, while it increases the amplitude of electrically evoked EPSCs in paired voles, but not in virgins. This oxytocin-induced potentiation of synaptic transmission relies on the de novo coupling between oxytocin receptor signaling and endocannabinoid CB1 receptor signaling in pair bonded voles. Blocking CB1 receptors after pair bond formation increases the occurrence of a specific form of social rejection – defensive upright response – that is displayed towards the partner but not towards a novel individual. Altogether, our results demonstrate that oxytocin’s action in the nucleus accumbens is changed through social experience in a way that regulates the trajectory of social interactions as the relationship with the partner unfolds, potentially promoting the maintenance of a pair bond by inhibiting aggressive responses. These results provide a mechanism by which social experience and context shift oxytocinergic signaling to impact neural and behavioral responses to social cues.

eTOC Blurb

Borie et al use slice physiology and behavior in the prairie vole to uncover how glutamatergic signaling in the nucleus accumbens is differentially modulated by an oxytocin receptor agonist depending on prior social experience. De novo coupling between the oxytocin and endocannabinoid systems in accumbens after bonding lessens partner rejection.

Graphical Abstract

Introduction

Social relationships evolve over time through interactions – both positive and negative – that reshape how one subsequently perceives ¹, engages with ² and feels ³ about another individual. Thus, the brain mechanisms acting when relationships are formed and expressed are likely to change over the course of subsequent encounters between two individuals. While considerable work has helped elucidate the neural mechanisms underlying the formation of social relationships ^2,4,5, much less is known about what modulates their expression or maintenance. In particular, although neuromodulators are well documented to act centrally in the brain to modulate social behaviors ^6–8, here we investigated whether that action can change over the course of a social relationship.

Oxytocin (OXT) is a key neuromodulator for the regulation of social behaviors ^9,10, and a target for the treatment of neuropsychiatric disorders ^11–13. OXT is produced mainly by neurons located in the supra-optic and paraventricular nuclei of the hypothalamus. OXT is released within the brain as relationships form and it facilitates experience-dependent social adaptations including the formation of social memories ^14,15, the establishment of mating-induced pair bonds ² and maternal responsiveness ¹⁶. OXT action results in modulation of neuronal activity in brain areas regulating social behaviors ¹⁷ to increase the salience and reinforcing value of social stimuli ^18,19. Furthermore, OXT neurons are activated in response to social stimuli including social touch ²⁰, infant cries ²¹ and mating. As a relationship develops, these events offer an opportunity for the OXT system to modulate the expression of the social bonds.

Elucidating whether OXT’s actions may change over the course of a social bond requires animals that form long term relationships. Upon social interaction and mating, prairie voles (Microtus ochrogaster) form a lifelong pair bond resulting in a dramatic change in the way they interact with other individuals ²². These behavioral shifts are accompanied by concomitant shifts in neural dynamics within brain networks controlling social behaviors ^4,23, particularly involving a brain area at the intersection between the social neural network and the mesolimbic reward system: the nucleus accumbens (NAc). Prairie voles and other monogamous species, including marmosets and humans express high levels of oxytocin receptors (OXTR) in the NAc while non-monogamous species, such as laboratory mice and rhesus macaques, do not ^24–26. Furthermore, OXTR signaling in the NAc plays a critical role for the formation of a preference for an individual, and the density of NAc OXTR correlates with such preferences ^27–29. Thus, OXTR signaling in the NAc may provide mechanistic insights into experience and context dependent modulation of social behavior.

Here, we studied OXT-induced modulation of NAc core activity both in sexually naïve animals and in pair bonded females to determine whether OXT may have different physiological effects in the context of pair bond formation versus expression. Using a combination of slice physiology, pharmacology, CRISPR/Cas9 genome editing and behavior, we provide evidence that OXTR agonists may reduce synaptic noise in the NAc of virgin animals, potentially facilitating bonding, yet potentiates glutamatergic transmission in pair bonded females by coupling to an endocannabinoid receptor (CB1) dependent, presynaptic mechanism. Furthermore, by blocking CB1 in vivo, we demonstrate that this mechanism specifically represses defensive responses in the partner’s presence, presumably helping maintain a trajectory to consolidate the bond. Altogether, we demonstrate that the action of neuromodulators such as OXT are context-dependent and should be considered in light of the specific social history of individuals.

Results

The effect of an OXTR agonist on spontaneous and evoked synaptic activity depends on social experience

We sought to examine the electrophysiological effects of OXTR signaling on NAc core medium spiny neurons (MSNs) and whether those effects are influenced by social experience. We ovariectomized sexually naive female subjects and cohabitated them with either a same sex littermate (Virgin group, green in figures) or a male (Paired group, blue in figures) for 24h. We performed slice electrophysiology recordings of MSNs in whole cell configuration 48h later (Figure 1A). Basal properties of NAc MSNs remained the same in both Virgin and Paired groups (Figure S1) and were consistent with a prior study in voles ³⁰. We then applied in the bath a high specificity OXTR agonist, TGOT (10⁻⁷M) ³¹, for 10-min. TGOT did not influence the frequency of spontaneous Excitatory PostSynaptic Currents (sEPSC) in either Virgin or Paired females (Figure 1B). However, TGOT induced a significant decrease of the sEPSCs amplitude in Virgin but not in Paired females (Figure 1C–D), potentially decreasing the effect of spontaneous excitation in Virgins.

Figure 1: Pair bonding differentially alters spontaneous and evoked synaptic activity in response to TGOT.

(A1) Experimental timeline (left) and schematic showing placement of stimulation and recording electrodes (right). (A2) Representative examples of spontaneous EPSC recordings (top, arrows are synaptic events) and evoked EPSCs (bottom, dotted arrows are stimulation artifacts and grey double arrowhead indicates the measured amplitude). Basal properties of NAc MSNs remained the same in both Virgin and Paired groups (Figure S1). Effect of 10min bath application of TGOT on spontaneous EPSC (B) frequency and (C) amplitude. (D) Average over a 15min period of the effect of 10min bath application of TGOT on spontaneous EPSC amplitude. E) Amplitude of the evoked EPSC as a function of stimulation intensity. Data expressed as % relative to the maximum amplitude. (F) Effect of 10min bath application of TGOT on evoked EPSC amplitude. (G) Average over a 15min period of the effect of 10min bath application of TGOT on evoked EPSC amplitude. Number of cells as indicated in each panel. Data (mean ± SEM) expressed as % relative to baseline in all figures, unless specified otherwise. See Figure S5 for individual data. Statistics reported in Table S1.

To further examine TGOT’s effects on NAc neurotransmission, we electrically evoked EPSCs (eEPSC) in whole cell recordings. The relationship between stimulation intensity and eEPSC amplitude was not influenced by prior social experience (Figure 1E), indicating a similar sensitivity of neurons across groups to the electrical stimulation. Moreover, the raw amplitudes of the eEPSCs were not different between groups, suggesting that social experience did not generally increase the basal excitation of MSNs (Figure S1H). Strikingly, there was a significant interaction between social experience and time (Figure 1F). TGOT produced a more potentiated response in Paired compared to Virgin females (Figure 1G), potentially increasing excitatory neurotransmission in Paired females.

Across individuals, the amplitude of TGOT-induced potentiation depends on socio-sexual experience

Because the effect of TGOT on evoked glutamatergic activity depends on social experience, we next investigated whether the strength of the post-cohabitation partner preference might be associated with the amplitude of this electrophysiological response. We performed modified partner preference tests – social preference tests (SPT) – before the slice electrophysiology experiment to evaluate the pair bond strength (Figure 2A–B). Males were placed under cups, restricting physical contact with the experimental female. We assessed the time spent near either male. Before the 24h cohabitation, there was no consistent preference in either the Virgin or Paired group. After cohabitation, Virgin females still showed no systematic preference, but those cohabitated with a male spent significantly more time near that individual (Figure 2C). Hence, like the traditional partner preference test, the SPT reflects the strength of the animal’s pair bond.

Figure 2: The amplitude of oxytocin-induced potentiation in field recordings correlates with the strength of the social preference.

(A1) Experimental timeline (left) and schematic showing placement of stimulation and recording electrodes (right). (A2) Representative example of evoked EPSP (right, dotted arrows are stimulation artifacts and grey double arrowhead indicates the measured amplitude). (B) Experimental protocol –Social preference test (SPT). (C) Duration of the interactions with male A and male B in SPT1 and SPT2 for Virgin and Paired individuals. See Figure S6 for individual data. (D) Effect of DNQX on the evoked EPSP amplitude. (E) Effect of 10min bath application of TGOT on evoked EPSP amplitude. (F) Average over a 15min period of the effect of 10min bath application of TGOT on evoked EPSP amplitude. (G) Amplitude of the effect of TGOT on the evoked EPSP amplitude during the 2 last minutes of TGOT application plotted against that animal’s preference for the partner. Animals showing a preference for the partner (i.e. preference >50%) are shown as upward triangles; for stranger, downward triangles. A significant correlation was found specifically for Paired animals preferring the partner. See Figure S5 for individual data. Statistics reported in Table S2.

To relate this behavioral preference to the electrophysiological response to TGOT, we turned to field recordings. As a reflection of activity in a larger population of neurons, the evoked Excitatory Postsynaptic Potential in the field (eEPSP_f) is less sensitive to cell-to-cell variability inherent in whole-cell recordings. The eEPSP_f was blocked by DNQX, thus demonstrating its glutamatergic nature (Figure 2D). Consistent with the results obtained in whole cell recording, TGOT application differentially modulated the eEPSP_f in Virgin and Paired females, as reflected in a significant interaction between group and time (Figure 2E–F). Specifically, the amplitude of the eEPSP_f was significantly increased following TGOT application in Paired females, whereas in Virgin females it remained unchanged (Figure 2F). The time course of the potentiation in field recordings from Paired animals appears sooner after TGOT application than what was observed in whole cell recordings, potentially due to cell type variation or depth of the recording. Nevertheless, specifically for partner-preferring Paired females, the amplitude of the field response correlated with our measure for partner preference, defined as the fraction of time spent near the partner relative to the total time spent near either animal during the second SPT (Figure 2G). Therefore, how the NAc responds to OXTR agonists depends on the effect of the vole’s prior social experience, pointing to a behavioral relevance of the plasticity in the OXT system’s physiological effects. We then focused on this robust experience-dependent potentiation in Paired females to decipher the underlying mechanism.

TGOT-induced potentiation in Paired female voles relies on postsynaptic OXTR and a putative presynaptic mechanism

Since the TGOT-induced potentiation in whole cell recordings was observed in Paired animals for the evoked, but not the spontaneous EPSC amplitude, even when both responses were measured from the same MSN, we suspected the involvement of a pre-synaptic, rather than a post-synaptic, mechanism. To further evaluate this possibility, we measured the effect of TGOT on the Paired pulse ratio (PPR) in evoked EPSC and EPSP_f, from whole cell (Figure 3B–C) and field recording (Figure 3D–E) configurations, respectively. The PPR is sensitive to the probability of presynaptic vesicle release ³². In Paired animals, the stronger the evoked, TGOT-modulated postsynaptic response to the first stimulation pulse in a pair, the smaller the second response and PPR, indicating that TGOT facilitated a larger initial presynaptic release of glutamate, In Virgin animals where there was no systematic TGOT-mediated potentiation, these values were not correlated. Hence, pair bonding experience leads to a significant correlation between TGOT-induced change in EPSC amplitude and TGOT-induced change in PPR suggesting that in Paired females, TGOT influences presynaptic glutamate release in the NAc, which leads to a potentiation of the glutamatergic transmission.

Figure 3: Oxytocin-dependent potentiation in Paired prairie voles relies on a putative presynaptic mechanism but involves OXTR in NAc cells.

(A) Representative example of evoked EPSPs (right, dotted arrows are stimulation artifacts and grey double arrowhead indicates the measured amplitude) and formula for Paired Pulse Ratio (PPR) calculation. Correlation between EPSC amplitude and PPR in whole cell recordings (WCR) from (B) pair bonded voles and (C) Virgin animals. Correlation between EPSP amplitude and PPR in field recording from (D) pair bonded animals and (E) Virgin animals (B) to (E) Each dot represents the average over a 5min period (10–14min, 15–19min, 20–24min or 25–29min – timing as indicated in figure 1E or 2E) for one cell. The insets represent the absolute value of baseline PPR for each cell. In Paired animals, TGOT application significantly decreased the PPR in both WCR and field recordings (Figure S2). (F) Viral-vector mediated CRISPR/Cas9 strategy (top) and timeline of the experiment (bottom). (G) Representative examples of the infection visible through the green fluorescence (left) and corresponding autoradiograms (right) showing that the AAV-mediated CRISPR/Cas9 strategy is efficient to reduce OXTR binding in the NAc specifically on the side infected with the sg-RNA targeting OXTR. (H) Effect of a 10min TGOT application on the evoked EPSP amplitude in NAc knocked down for OXTR and in the contralateral control side (field recordings). (I) Average over a 5min period of the effect of a 10min TGOT application on the evoked EPSP amplitude in NAc knocked down for OXTR and in the contralateral control side (field recordings). See Figure S5 for individual data. Statistics reported in Table S3.

These results might suggest the involvement of presynaptic OXTR rather than OXTRs located on cells in the NAc. To explicitly test the necessity of NAc OXTR for the OXT-evoked potentiation observed in Paired animals, we used an AAV-mediated CRISPR/Cas9 strategy to knock down OXTR expression in cells from the NAc. If OXTR required for the potentiation are located on the pre-synapse, knocking down OXTR expression in NAc cells should not affect the amplitude of the potentiation. Females were administered a combination of AAV-vectors driving expression of spCas-9 and an sgRNA targeting OXTR in one hemisphere’s NAc and vectors enabling for expression of spCas-9 and a control sgRNA targeting LacZ in the contralateral NAc (Figure 3F), as previously described ³³. After allowing at least 4 weeks for expression and genome editing, animals were cohabitated with a male for 24h to allow for the development of a pair bond, and their brains were extracted and sliced to perform slice electrophysiology in field recording configuration.

Expression of the AAV coding for the sg-RNA was evaluated (Green fluorescence) and found comparable on both sides (Figure 3G1). Furthermore, autoradiograms performed using I¹²⁵-OVTA show that this strategy was efficient to knock down OXTR binding in the NAc infected with the sg-RNA targeting OXTR, but not on the contralateral side infected with the sg-RNA targeting Lac Z (Figure 3G2). The LacZ hemisphere responded electrophysiologically to TGOT application with a potentiation of the eEPSP_f amplitude, as expected. The response of the OXTR knocked-down hemisphere was significantly reduced, producing a significant interaction (Figure 3H–I). Altogether, these data show that the activation of OXTR located on NAc cells is necessary for the TGOT-induced potentiation of glutamatergic transmission into NAc, yet this potentiation likely arises from increased presynaptic release. This surprising result is in contrast to a previously described presynaptic mechanism for oxytocinergic modulation in mice ³⁴. Thus, we hypothesized that in pair bonded voles specifically, a retrograde signal may serve to link postsynaptic OXTR within the NAc to presynaptic modulation of excitatory transmission.

CB1 receptor activation is necessary for TGOT-induced potentiation in pair bonded females

Endocannabinoids (eCB) are retrograde signaling molecules modulating the synaptic strength between cells ³⁵ and have been implicated in the electrophysiological ^36–38 and behavioral ³⁹ effects of OXT. Indeed, OXTR can be coupled to the intracellular signaling G-protein, Gq ⁴⁰, leading to phospholipase C activation, a mechanism that can promote eCB production ⁴¹. To evaluate whether the OXT- dependent potentiation in Paired females relies on the activation of the eCB system, we performed electrophysiological experiments in field configuration and measured the effect of TGOT bath application in presence or absence of a CB1 receptor antagonist (AM4113, 10⁻⁶M). Using separate slices from the same animals as a control, we showed that AM4113 application prevented the TGOT-induced potentiation in Paired females (Figure 4A). Interestingly, this application also prevented the correlation between the effect of TGOT on the PPR and on the eEPSP_f amplitude (Figure 4B–C). Hence, the activation of the eCB system is necessary for the TGOT-induced potentiation observed in Paired females.

Figure 4: CB1 receptor activation is necessary for the TGOT-dependent potentiation.

(A) Effect of 10min bath application of TGOT on evoked EPSP amplitude in pair bonded females in presence or absence of AM4113 (field recording). One slice per condition for each animal and data paired by experimental animal. Correlation between EPSP amplitude and PPR in field recordings from pair bonded animals after treatment with TGOT in (B) absence or (C) presence of AM4113. (D) Effect of 10min bath application of ACEA on evoked EPSP amplitude in Virgin females (field recording). One slice per condition for each animal. One sample Wilcoxon test (compared to 100%), p<0.05 at min 2, 3, 4, 5, 8, 9, 11, 12, 13, 14, 27, 28, 30, 32 and 40. (E) Average over a 5min period (10 to 15min after beginning the application) of the effect of a 10min ACEA application on the evoked EPSP magnitude (absolute value of the amplitude) in NAc of Virgin and Paired females (Figure S3). (F) Correlation between EPSP amplitude and PPR in field recordings from Virgin animals after treatment with ACEA. (G) Average over a 5min period (10 to 15min after beginning the application) of the effect of 10min bath application of TGOT or ACEA on evoked EPSP amplitude (field recording). See Figure S5 for individual data. Statistics reported in Table S4.

Pair bonding couples inherent eCB potentiation to OXTR signaling

Importantly, we did not investigate a role for eCB in TGOT-induced potentiation in Virgins because there was no systematic modulation of glutamatergic signaling in this group. Nevertheless, CB1 receptors are expressed in the NAc of sexually naive females ⁴². The application of a CB1 receptor agonist, arachidonyl-2’-chloroethylamide (ACEA), produces a potentiation of the post-synaptic response to an electrical stimulation of NAc core in Virgin females (Figure 4D and E) and in Paired females (Figure S3). The amplitude of the effect of ACEA on the eEPSP_f correlated with its effect on the PPR (Figure 4F), confirming a presynaptic function. Interestingly, ACEA’s effects in Virgin females were comparable to the amplitude of the response to TGOT in Paired females (Figure 4G). Thus, together our results suggest that in Virgin animals, OXTR activation is not coupled to a functioning CB1-mediated potentiation of excitatory transmission in the NAc, but pair bonding establishes this link between these neuromodulatory systems.

Blocking CB1 receptor activation in the NAc of Paired females selectively increases partner rejection

The eCB-mediated electrophysiological effect of TGOT emerges only after socio-sexual experience. Although the OXTR agonist presumably has many different effects, we wanted to determine whether disrupting the ability for eCB to induce this potentiation in NAc has any behavioral consequence for social interactions. We therefore implanted bilateral cannulas in the NAc core of ovariectomized females to locally deliver AM4113 (Figure 5A, Figure S4A–D). We cohabitated estradiol-primed subjects for 24h with a male and then conducted a first SPT the next day. Experiments described in this section were thus exclusively performed in pair bonded females. One day later, we administered subjects with AM4113 (or vehicle) in the NAc and conducted a second SPT (SPT2). During SPT2, irrespective of their treatment, subjects spent more time on average near the partner than near the stranger (Figure 5B).

Figure 5: Blocking CB1 receptor activation in the nucleus accumbens selectively increases partner rejection but not stranger rejection.

(A) Experimental timeline. Cannula locations in the NAc core shown in Figure S4. (B) Duration of the interactions with partner (P) and stranger (S) during SPT1 and SPT2 in Paired females administered with vehicle or AM4113 in the NAc. See Figure S6 for individual data. (C) Experimental protocol - intruder test. Duration of the affiliative behaviors displayed toward (D) the novel male or (E) the partner, in voles administered with vehicle or AM4113 in the NAc. Duration of the aggressive/defensive behaviors displayed toward (F) the novel male or (G) the partner, in voles administered vehicle or AM4113 in the NAc. Examples of the defensive upright behavior may be seen in Video S1. Data are (mean ± SEM). Statistics reported in Table S5.

The electrophysiological experiments above were performed after the pair bond was established, indicating that the eCB-dependent potentiation may naturally occur after pair bond formation and influence pair bond expression even in the absence of an effect on pair bond formation ⁴³. Furthermore, the NAc plays a role in action selection, so mechanisms acting at this time point may affect the nature of social interactions. To assess this, we performed intruder tests after SPT2. The female was first placed alone in her home-cage for 10 minutes, before a novel stranger male was introduced for 10 minute and then replaced by the partner for the final 10 minutes (Figure 5C). Since side by side contact is not influenced by CB1 receptor antagonist ⁴³, we hypothesized that a CB1-dependent change in the quality of the social interactions would likely occur through a change in agonistic – i.e. aggressive/defensive - interactions rather than a change in affiliative interactions.

Affiliative and aggressive/defensive behaviors were scored blind to treatment group. No effect of treatment was observed in affiliative behaviors performed by the female (Figure 5D–E) whether the female interacted with a stranger male or with her partner. For aggressive/defensive behaviors, females showed similar rejection responses towards the stranger irrespective of the modulation of the eCB system (Figure 5F). However, in the presence of the partner, though vehicle-treated females tended to reject their partner less, animals given AM4113 spent about twice as much time in a defensive upright posture when the partner was present, as compared to vehicle treated animals (Figure 5G), despite showing no difference in overall rearing behavior within the cage (Figure S4E–G).

Hence, CB1 receptor activation occurs during a female’s social interaction with her partner to decrease the expression of a rejection-like, upright rearing response when the partner is present. Thus, this eCB-dependent mechanism would play an important role in defining the behavioral trajectory of social interactions by reinforcing a pair bond through a reduction of partner rejection.

Discussion

Our study demonstrates that adult social experience, modeled here by pair bonding in female prairie voles, can alter the electrophysiological action of neuropeptides on a brain area, the NAc, important for social information processing and social salience. Activating NAc OXTRs induces divergent electrophysiological responses in virgin versus pair-bonded female voles, even when the OXTR agonist is delivered in the same way (e.g., concentration, timing, application mode). While pair bonding may change the level of OXTR expression in prairie vole NAc ⁴⁴, we show here that an OXTR agonist decreases the amplitude of spontaneous EPSCs of virgin animals but increases the amplitude of the evoked activity of paired animals, indicating that simply expressing more OXTR is not sufficient to explain our results. Instead, pair bonding experience leads to distinct mechanisms for OXTR agonists to modify the signal-to-noise ratio in excitatory neurotransmission. In paired prairie voles, they act on receptors located on NAc cells and trigger an eCB-mediated synaptic potentiation involving a putative presynaptic mechanism. Notably, even in virgin animals, TGOT significantly decreases the PPR (data not shown) but, this change is not correlated to the change in EPSC amplitude, suggesting the lack of a consistent relationship with the strength of glutamatergic synaptic transmission. Further studies would be needed to understand the function of such a change. Blocking CB1 receptors after the pair bond is established increases a partner-directed rejection behavior without affecting stranger-directed rejection. Hence, the socio-sexual experience-contingent eCB-dependent oxytocinergic potentiation of NAc MSNs described here could influence the trajectory of social behavior and promote the expression of a bond.

To our knowledge, this is the first study showing that acute oxytocinergic modulation of excitatory transmission can be different depending on an adult’s social history. Altered connectivity within neural networks is usually attributed to synaptic plasticity ^45,46 or state-dependent neuromodulation (e.g., release or receptor expression) ^16,47, rather than changes in how neurotransmitters themselves work within the network. In fact, most studies of how OXTR agonists facilitate social behavior assume its mechanistic stability over state and time in adults ^16,39,48. Variability in the effects of OXTR agonists across individuals are presumed to arise from individual differences in how much OXT is released ^49,50, how many OXTR there are (King et al. 2016; Ross et al. 2009) or methylation status of the OXTR gene ^53–55. Here, we showed that the way OXTR agonists engage cellular pathways to affect glutamatergic neurotransmission is context-dependent, even when the average strength of postsynaptic responses is the same.

We found that the electrophysiological effect of OXTR agonist application depends on social experience, with a potentiation of excitation mediated through a coupling between OXTR and CB1 receptor activation emerging in paired animals. The potentiation we observed was unexpected, since OXTR agonists in the NAc core ³⁴, ventral tegmental area ³⁸ or prefrontal cortex ³⁶ typically induces long-term depression (LTD) of glutamatergic transmission. CB1 receptors, found presynaptically ³⁵ and expressed in the vole NAc ⁴², are implicated in such OXT-induced depression of excitatory transmission ^36–38. On the other hand, OXT-induced potentiation, while observable when combined with an electrical induction paradigm ^36,56,57, has not previously been linked to the eCB system. Even though eCB is usually associated with synaptic depression, its effects can be bi-directional depending on the concentration of eCB, the type of presynaptic neuron targeted (GABAergic vs glutamatergic), involvement of astrocytes ⁵⁸, and the precise timing of presynaptic-to-postsynaptic firing ⁵⁹. Thus, our finding of a pharmacologically induced, eCB-mediated, OXT-induced potentiation in pair bonded females illustrates a new way for these neuromodulators to work together.

The physiological effect of an OXT-eCB-based potentiation could allow context-dependent, upstream inputs to gain enhanced salience by amplifying excitatory transmission. The eCB system is generally engaged in naturally rewarding contexts ⁶⁰, such as food ingestion ⁶¹, physical exercise ⁶², and sexual activity ⁶³. Given OXT’s well-documented involvement in social-specific behaviors ^19,46, the coupling in paired voles between OXTR and the eCB system may therefore reinforce the value of social interactions ³⁹. This value is likely different for partners and strangers, though our fixed instead of random presentation order in the intruder tests precludes a rigorous conclusion about AM4113’s effects in these two social contexts. Nevertheless, vehicle (but not AM4113) treated animals spent more time in the defensive upright position against the stranger than against their partner (Wilcoxon signed rank, p=0.0098 for vehicle, p=0.95 for AM4113), hinting that eCB’s normal actions decrease rejection specifically toward a valued partner. The potentiation of glutamatergic signaling carrying partner-specific information into the NAc core might then promote further positively valanced social interactions. When this may first happen after a virgin meets a novel male remains unclear, since we manipulated the eCB system only after animals showed a robust partner preference. Perhaps the initial pair bond formation, potentially when mating releases OXT ⁶⁴, triggers the association between both systems. Notably, one study did not find an impact of CB1 blockade on pair bond formation ⁴³. Nevertheless, future studies investigating a virgin’s initial encounter with a male should explore whether subtle behaviors – increasingly recognized for their biological relevance ^65,66 – like “rejection” might be limited by CB1 activation – with or without the involvement of OXTR. Intriguingly, TGOT-induced potentiation may also simply be delayed in virgin females (Figure 2E), which could suggest an oxytocin-induced priming of oxytocin sensitivity.

Our conclusions fit within a working model (Figure 6) for how OXT, eCB and glutamatergic inputs into NAc may interact in vivo to adapt social responses to the context based on one’s social experience. In virgins, social interactions facilitating future pair bonding (e.g. mating), would release OXT, resulting in decreased amplitude of sEPSCs in NAc MSNs, which may serve to decrease noise, facilitating linking the cues of the future partner with the reward system ² (Figure 6 left). In females that are already pair bonded though, a potentiation of the glutamatergic transmission would arise from a coincidence of events: the activation of the OXT-eCB pathway along with the activity of upstream neurons reflecting the familiar partner (Figure 6 right).

Figure 6: Working model.

Model for the mechanism of OXT action on synaptic transmission in the nucleus accumbens of female prairie voles. AEA = anandamide, PVN = paraventricular nucleus, glu.= glutamate

Here, we used the pair-bonding process to model social experience. Indeed, pair-bonding depends on the duration of cohabitation and constitutes a shift over time in the behavioral repertoire of adult animals ⁴. Pair-bonds depend on the extent and nature of the social interactions ⁶⁷, can be broken, as well as reformed with new individuals, and are thus dynamic ⁶⁸. Rather than being a deterministic program in these animals, pair bonding in voles involves cognitive resources to associate cues from the partner with the reward system, and leads to a dramatic change of the behavior displayed both in the presence of the partner and novel conspecifics. Future studies should investigate whether peer-based social experience such as competition or cooperation, also influence the mode of action of social neuromodulators. Importantly, oxytocin and the reward systems are involved in various types of social relationships and social-decision making in a large range of species ^69,70, suggesting that the findings of this paper may not be limited to pair bonding.

Two non-exclusive mechanisms may contribute to explain the context-dependent effect of the CB1 antagonist in pair bonded females in response to a novel and partner male: OXT release could be higher in response to the familiar male (Ross et al. 2009; Lukas et al. 2013), or interacting with a familiar versus a novel conspecific could differentially activate regions upstream of the NAc ⁷². Importantly, in our slice electrophysiology experiments, we controlled for both OXT concentration and presynaptic stimulation intensity. We presume that for interactions with the partner but not novel male, OXTR activation triggers the local production of anandamide, a potent agonist of CB1 receptors. This facilitates the release of more glutamate upon activation of the pre-synaptic neurons, while leaving spontaneous synaptic transmission. Such an OXT-induced mechanism could help filter salient inputs to the NAc – similar to how OXT enhances the signal-to-noise ratio in other brain areas ^9,73,74 by a mechanism involving the eCB system.

In vivo, the putative mechanism described above would potentiate the effect of socially induced presynaptic activity, inhibiting rejection behavior when the partner is present. In virgins, this inhibition would not happen and interactions would be more agonistic ^75,76. This working model may help to conceptualize the mechanisms underlying context-dependent behavioral differences resulting from exogenous OXT administration in humans ⁴⁸. Indeed, the effect of intranasal OXT depends on the partner status of the subject: it keeps men in a monogamous relationship from being too close to a new woman without affecting single men ⁷⁷. OXT’s effects further depend on the interrelationship between a subject and the social stimuli: it increases the perceived attractiveness of one’s partner but not of other women⁷⁸. Interestingly, OXT increases NAc’s response to the partner pictures compared to other women.

Finally, it has long been believed that OXT acts in the NAc core to first establish the bond ^2,17, and then to increase the relative value of the partner, thus promoting the expression of the bond. Our results suggest that in both aspects of pair bonding, OXT could act through a modulation of the signal-to-noise ratio. Importantly, socio-sexual experience appears to condition the reinforcing properties of the OXT system through its association with the eCB system. This could sustain behavioral changes happening as the social relationship unfolds, and alterations in the described mechanism could negatively affect the trajectory of relationships. By showing that context influences OXT’s mode of action in the NAc, our study offers a potential new explanation for inconsistent outcomes of OXT’s effects in human ⁷⁹. Indeed, while inter-individual differences were identified as the likely source of variability in clinical trial outcome, it is usually attributed to variation in the intrinsic properties of the OXT system ^80,81. Here, we point out that a complementary explanation – social experience influencing the mode of action of OXT – should motivate the factoring in of social history when evaluating the effect of OXT treatments.

STAR Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Robert Liu (robert.liu@emory.edu).

Materials availability

pAAV-U6-sgRNA(Oxtr)-CMV-eGFP and pAAV-U6-sgRNA(LacZ)-CMV-eGFP plasmid will be made freely available upon written request to LJY.

Data and code availability

Data used in each figure panel and custom R code for analyzing the vole location in the SPT have been deposited in GitHub and are publicly available as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

All experiments were performed in accordance to the guidelines and approved by the Emory University Institutional Animal Care and Use Committee. Prairie voles (microtus ochrogaster) bred at Emory University originated from field caught specimens from Illinois, USA. Weaning was performed at postnatal day 20–23 and animals were house-grouped (2 to 3 voles per cage) in same-sex sibling cages.

Prairie voles between 2 and 6 months old were used for all the experiments. Animals were housed under a 14:10 hour light/dark cycle, temperature was maintained at 68–72 degrees Fahrenheit and humidity at 40–60%. Food (Lab Rabbit Diet HF #5326, LabDiet) and water were given ad libitum. Enrichment consisting of cotton pieces used to build a nest were added to cages. When animals were isolated, supplemental enrichment material (Nylabones, Bioserve) was added to the cage. All females used were ovariectomized prior to the beginning of the experiments.

METHOD DETAIL

Slice electrophysiology

Slicing

After isoflurane anesthesia followed by quick decapitation, brains were extracted and 300μm thick coronal slices were sliced using a Leica VTS-1000 vibrating-blade microtome in ice-cold slicing solution (in mM: 130 NaCl, 3.5 KCl, 1.1 KH₂PO₄, 6 MgCl₂, 1 CaCl₂, 10 glucose, 2 kynurenate, 30 NaHCO₃, bubbled with 95% O₂ and 5%CO₂). Slices were transferred to a 32˚C slicing solution for 1h and then to room-temperature artificial cerebro-spinal fluid (ACSF) (in mM: 130 NaCl, 3.5 KCl, 1.1 KH₂PO₄, 1.3 MgCl₂, 2.5 CaCl₂, 10 glucose, 0.4 ascorbate, 0.8 thiourea, 2 Na pyruvate, 30 NaHCO₃, 95% O₂ and 5%CO₂).

Recordings

Slices were transferred to the recording chamber mounted on the fixed stage of a Leica DM6000 FS microscope (Leica Miscrosystems Inc. Bannackburn, IL) and perfused at ≈2mL/min with ACSF maintained at 32°C. Recordings were performed using MultiClamp700B amplifier in conjunction with pClamp 10.2 software and a DigiData 1322A (Axon Instruments).

Borosilicate glass patch electrodes (WPI, Sarasota, FL, USA) were used to acquire whole-cell patch-clamp or field recordings in the NAc core. For whole cell recordings, electrodes had a resistance of 4–7 MΩ, were filled with a “patch solution” (in mM: 130 K-gluconate, 2 KCl, 10 HEPES, 3 MgCl2, 2 K-ATP, 0.2 Na-GTP, 5 phosphocreatine, adjusted to pH 7.3 with KOH). Pipette offset and capacitance were automatically compensated for using MultiClamp command software (Molecular Devices). For field recordings, electrodes were filled with ACSF. See Figure S5 for individual data for all slice electrophysiology figure panels.

A concentric bipolar stimulation electrode was placed near the recording electrode and moved inside the NAc core until a signal was perceived by the recording electrode. The intensity of the stimulation was defined at the beginning of each recording such as it induces 50% of the maximum evoked response. A pair of stimulations with 50ms interval were applied every 30s during the recording. The values for eEPSC and eEPSP_f were obtained in response to the first stimulation. The influence of TGOT on spontaneous EPSCs was measured between 1 and 5s after the stimulation.

Recordings from fast spiking interneurons were excluded based on their electrophysiological properties so that the results presented here were obtained exclusively from MSNs.

Pharmacology

Pharmacological compounds were diluted in ACSF and bath applied. The concentration used are the following: TGOT 10⁻⁷M (Sigma-Aldrich), AM4113 10⁻⁶M (Sigma-Aldrich), DNQX 2×10⁻⁵M (Sigma-Aldrich), ACEA 10⁻⁷M (Tocris). TGOT was applied for 10min. AM4113 was applied for at least 10min before the application of TGOT and was maintained for the duration of the recording. DNQX application lasted for 10min and was performed at the end of the recording.

Behavior: Social preference test (SPT)

All females were ovariectomized and primed for 3 days with subcutaneous administration of estradiol benzoate (17-β-Estradiol-3-Benzoate, Fisher Scientific, 2 μg dissolved in sesame oil starting 3 days prior to experiments). The next day, a SPT could be performed before animals were cohabitated with a same sex littermate or with one of the males used in the SPT for 24h. Previous studies showed that 24 h cohabitation with a male induces a strong pair bond as indicated by more time spent huddling with the partner compared to the stranger in a partner preference test ⁶⁷. Animals were then isolated until euthanasia and brain extraction which occurred 2 days later except for a potential second SPT test which would be performed on the day prior to euthanasia.

For the SPT, two males of similar age and weight were used as stimuli. One of them was randomly assigned as the future partner and was cohabitated with the experimental female after the test. The experimental arena consisted of a rectangular arena (30×8 inches). Two metallic cups (4 inches diameter and 4 inches height) constituted of metal bars were placed on each sides of the arena. Males were placed under these cups shortly before the female was introduced in the middle of the arena. The female was given 2h to freely explore the arena and/or the cups. When 2 SPT were performed, the same males were used as stimuli. For experiments comparing virgin and paired voles, 2 littermate females were used for each experiment and were tested with the same 2 males.

IdTracker.ai was used to track the center of mass of the experimental subjects ⁸² . A custom R code was then used to calculate the duration spent in a circle of a radius equal to 3 times the radius of the cup (i.e. within 4 inches from the edge of the cup) for each cup thus defining the time spent in the “partner zone” and the time spent in the “stranger zone”. The time spent on top of one of the cups was counted as time near the stimulus animal located under the cup. Preference for the partner expressed as a percentage was calculated as follow: time in the partner zone (in s) / [time in the stranger zone (in s)+time in the partner zone (in s)] * 100.

Behavior: Intruder test

For experiments testing the behavioral effects of blocking CB1 receptors, after the SPT, females were placed in their home cage for 10minutes. A stranger male that the female did not encounter before was then introduced with her for 10minutes. The stranger male was removed and the partner was introduced with the female for another 10minutes. Sideview videos were recorded and used for behavioral annotations (The Observer XT, Noldus).

Surgeries

Anesthesia induction was achieved with 3% Isoflurane and maintained with 1–3% Isoflurane during the surgery. Meloxicam 2mg/kg was subcutaneously administered before the surgery begins and for 2 days following the surgery. Animals were monitored daily and were given at least 1 week to recover after surgery before experiments began. Anterior/posterior coordinates were referenced to Bregma, and dorsal/ventral coordinates were referenced to the top of the skull. Animals were placed in a stereotaxic frame and ear bars coated with Lidocaine were used to stabilize the head.

Virus injection

500nL of viruses were injected in the NAc core (10˚ angle, +1.9mm AP, +/−2.6mm ML, −4.5mm DV) using a 10μL syringe (nanofil, World Precision Instrument, USA) with NanoFil Needle (NF33BV-2, World Precision Instruments, USA) at a rate of 100nL/min (Micro4 pump, World Precision Instruments, USA).

Cannula implantations

Bilateral guide cannulas (26G - P1 Technologies, USA) heading to the NAc core were implanted (AP +1.5mm, ML +/−1.5cm, DV −4.3cm). Dental cement (Glass Ionomere cement –Harvard Apparatus, USA) and sutures (Ethicon J493G –ShopMedVet, USA) were used to secure the implant.

Intra-cerebral infusions

AM4113 was diluted in ACSF and used at a final concentration of 20μM with 0.1% DMSO. The vehicle was ACSF with 0.1% DMSO.

Internal cannulas (P1 Technologies, USA) with a projection of 0.2mm were used for the injections. They were connected through 2 different tubing to two 1 μL Hamilton syringes (P1 Technologies, USA) controlled by a microinjector pumps (micro 4, World Precision Instrument, USA). Injections of 500nL/hemisphere were performed at 100 nL/min in isoflurane anesthetized voles. Internal cannulas were left in place 5min after the end of the injection to allow for diffusion. Experiments were started 1h after the beginning of the injection. Positions of cannulas were verified post mortem.

AAV-mediated CRISPR/Cas9 Knock-down of OXTR

AAV9-particles were synthesized using pAAV-U6-sgRNA-CMV-eGFP and pAAV-RSV-spCas9 (Addgene plasmids #85451 and 85450, kindly gifted by Hetian Lei), pAAV9-SPAKFA (Penn Vector Core, PA, USA) and pAAV/Ad (ATCC, VA, USA). sgRNA sequences cloned into pAAV-U6-sgRNA-CMV–eGFP were the following: sgRNA(OXTR): 5’-GCTGCGGTGGCCCGGCTGTG-3’, sgRNA(LacZ): 5’-GTGAGCGAGTAACAACCCGT-3’. AAV9-particles were produced in HEK293T cells and purified with AVB-affinity chromatography ⁸³. 3 virus were generated: AAV9-U6-sgRNA(Oxtr)-CMV-eGFP (3.0×10^10 genomic copies/μl), AAV9-U6-sgRNA(LacZ)-CMV-eGFP (3.0×10^10 genomic copies/μl) and AAV9-RSV-spCas9 (1.5×10^10 genomic copies/μl) and used as previously described ³³. A minimum of 4 weeks was allowed for virus expression before the experiments started.

OXTR autoradiography

Autoradiography was performed as previously described (Ross et al. 2009). Fresh frozen brains were sectioned on a cryostat at 20 uM and mounted on Fisher Superfrost plus slides and stored at −80C. The slides were thawed and fixed for two min with 0.1% paraformaldehyde in PBS at RT and then incubated in 50 pM 125I-OVTA (2200 Ci/mmol; PerkinElmer; Boston, MA), a selective, radiolabeled OTR ligand, for one hour. Unbound 125I-OVTA was then removed with Tris-MgCl2 buffer and sections were allowed to dry. Sections were exposed to BioMax MR film.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data are shown as means±standard error of the mean (SEM). In testing statistical significance, α was set at 0.05. The n for each data set is indicated in figures and/or legends. All data were averaged per experimental groups. Data were considered non-parametric when sample size was moderate. We used factorial Analysis of Variance (ANOVA) to compare multiple groups followed by post hoc comparisons with Sidak tests. All statistical analysis was performed with GraphPad Prism 9.

Unless otherwise noted, for all figures, all available data was used. In the case of the correlation in Fig. 2G, because of technical issues (e.g. poor slice quality preventing physiology; or camera/computer memory filling up during a video recording preventing behavioral analysis), we had less data for the correlation between behavior and physiology (Fig. 2G) than we did for behavior alone (Fig. 2C) or physiology alone (Fig. 2E and 2F). Furthermore, for this figure, only animals presenting a preference for the partner (9/15) were considered.

Supplementary Material

Video S1

Video S1. Defensive upright response to partner (2 clips) and stranger (2 clips). Related to Figure 5.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
125I-OVTA	Bachem	4031339.005
Bacterial and virus strains
AAV9-U6-sgRNA(Oxtr)-CMV-eGFP	33 and This paper	N/A
AAV9-U6-sgRNA(LacZ)-CMV-eGFP	33 and This paper	N/A
AAV9-RSV-spCas9	33 and This paper	N/A
Biological samples
Chemicals, peptides, and recombinant proteins
TGOT	Bachem	4013837.005
AM4113	Sigma-Aldrich	SML1804
DNQX	Sigma	CAS-2379–57-9
ACEA	Sigma	CAS-220556–69-4
17-β-Estradiol-3-Benzoate	Sigma-Aldrich	E8515–200MG
NaCl	Fisher	S271–500
KCl	Fluka	60128
KHPO	Sigma	P5655–500G
MgCl	Sigma-Aldrich	M0250–500G
CaCl	Sigma-Aldrich	C8106–500G
Glucose	Sigma	G-8270
Kynurenate	Sigma	K3375–5G
NaHCO	Sigma	S233–500
Ascorbate	Sigma	A4034–100G
Thiourea	Sigma-Aldrich	T8656–50G
Na Pyruvate	Sigma-Aldrich	P2256–100G
HEPES	Sigma	H3375–250G
DMSO	Sigma	D-5879
Critical commercial assays
Deposited data
Data for figure panels	This paper	https://github.com/rcbliu/voleOTephys
Experimental models: Cell lines
Experimental models: Organisms/strains
Prairie voles (microtus ochrogaster)	Emory Yerkes colony	N/A
Oligonucleotides
Recombinant DNA
Software and algorithms
IdTracker.ai	82	https://idtrackerai.readthedocs.io/en/latest/
Custom R code for SPT analysis	This paper	https://github.com/rcbliu/voleOTephys
Observer XT, Noldus	Noldus	https://www.noldus.com/observer-xt/
Other

Highlights.

• TGOT-induced spontaneous and evoked activity depends on social experience in voles

• TGOT-induced potentiation of NAc is associated with social preference

• NAcc OXTRs potentiate excitatory transmission via an eCB mechanism in paired females

• Blocking CB1 receptors in NAc of paired females increases partner rejection