Our results demonstrate differential effects of oxytocin receptor agonism on expression and acquisition of pair bonding in male prairie voles. Specifically, we show for the first time that administration of an OXTR-specific agonist after cohabitation does not facilitate and possibly disrupts the expression of preferential affiliative behavior of a male animal towards its partner. In contrast, this agonist's presence during the acquisition of pair bonding facilitates such preferential affiliative behavior. In conjunction, these two points indicate that effects of OXTR agonism on indices of social bonding are highly dependent on the social context of this treatment.
We point out that our analysis of huddling behavior was done using non-parametric tests that did not provide a statistical assessment of interactive effects between treatments (LIT versus vehicle) and stimulus animal (partner versus stranger), which would have been possible by ANOVA. Therefore, our findings of inhibitory effects of OXTR agonism on expression of pair bonding and facilitating effects of OXTR agonism on acquisition of pair bonding could be viewed as suggestive. However, the inhibitory effect on expression of pair bonding manifested itself not only in a significant difference in huddling of vehicle-injected males with their target females and lack of such significant difference in LIT-injected males (Fig. 2A), but also in a significant preferential huddling in vehicle-, but not in LIT-injected animals (Fig. 2B), and in significant difference in times spent in partner’s chamber versus stranger’s chamber in vehicle-, but not LIT-, injected voles (Fig. 2C). Therefore, we believe that this experiment provides sufficient evidence for the negative influence of OXTR agonism on the expression of pair bonding.
The inhibitory effects of OXTR stimulation on expression of pair bonding are in agreement with studies demonstrating that administration of OXT inhibits measures of distress and depression-like behaviors in socially-isolated animals, including male and female prairie voles that have been previously pair-bonded [58–60]. Moreover, administration of OXTR antagonist or viral knockdown of OXTR expression in nucleus accumbens produced depression-like behavior in male pair-bonded prairie voles after separation from their partner and even in the presence of their partner [60]. These studies suggested that activation of OXTR substitute for the social buffering effects of the presence of social partner, while blockade of OXTR activity counteracts such effects. However, interpretations of studies on effects of OXT separation-induced distress are difficult to make because OXT can also have anxiolytic effects in non-social paradigms [61, 62]. In the future, it would be beneficial to examine effects of LIT on expression of pair bonding via other indices of pair bonding besides partner preference, such as selective aggression, preferential mating, and separation distress. However, development of selective aggression and other indices of pair bonding, could require longer periods of cohabitation than employed in our study.
In fact, our study used shorter cohabitation times (4 and 2 hours) than a larger number of previous studies on pair bonding in prairie voles [14, 15]. Many laboratories use 6 hours as the shortest interval of cohabitation that does not lead to significant partner preference or other behaviors indicative of a pair bond, but allows for the observation of facilitation of pair bonding by various manipulations. These laboratories then use 24-hour or even longer intervals of cohabitation to allow studies on inhibition of pair bonding [9, 19, 69]. However, the length of cohabitation required to establish a pair bond frequently differs between laboratories. Factors contributing to this difference between laboratories are unclear, but are unlikely to involve the configuration of testing chambers [15, 70].
Many factors leading to differences in the time of cohabitation required for pair bonding remain unexplored. They could involve differences in diets and housing conditions (for example, ventilated versus non-ventilated cages). Environmental stress and resulting changes in glucocorticoid levels can enhance partner preference in male prairie voles [71, 72]. It is possible that levels of environmental stressors, due to differences in housing conditions, are different between laboratories or animal colonies. Effects of environmental conditions on pair bonding are worthy of systematic assessments. Importantly, different behaviors associated with pair bonding, including partner preference, selective aggression, preferential mating, and separation distress, could require different cohabitation times for their development. Moreover, even in this case, they would assess the process of pair bonding, but not the strength of the pair bond. The strength of the bond assumes its stability across time, which has been investigated in only a few studies [9, 69], and is deserving future experimental research. Nevertheless, a substantial number of other reports have successfully used cohabitation times as short as 1 hour to evaluate effects on pair bonding using the PPT [10, 16–18, 57, 73]. We note that these studies included experiments performed on animals only a few generations removed from wild-caught animals [10, 16]. In that sense, our observations of significant partner preference following 4 hours of cohabitation should not be considered unusual.
As expected with short cohabitation intervals, we did not see development of selective aggression in the male subjects in this study. In fact, the number of aggressive events in males were overall very low. For example, all aggressive events in the experiment testing effects of LIT on expression of pair bonding (Fig. 2D) were produced by stranger females and not by the LIT-injected focal male voles. In agreement with this, there was no difference in aggressive events during PPT when LIT was administered prior to cohabitation (Fig. 3D) and there was a tendency of decreased, rather than increased, aggressive encounters during cohabitation (Figure S4B). Therefore, the overall effect of LIT on pair bonding cannot be explained by its effects on aggression. Instead, we hypothesize that the specific activation of OXTR signals the presence of a social partner. Specifically, when LIT is administered in males that have established selective affiliation, this manipulation substitutes for the partner-induced activation of OXTRs resulting in indiscriminate affiliative behavior towards both partner and stranger female voles.
As mentioned previously, a facilitation of acquisition of pair bonding by OXTR agonism was an effect expected from the previous literature [9, 10, 16, 19, 57]. The ability to observe the facilitating effects of LIT were aided by the fact that partner preference in vehicle-injected voles was lower in the experiment testing the effects on acquisition (Fig. 2A) than in the experiment testing expression of pair bonding (Fig. 3A). This less robust partner preference in vehicle-injected voles was likely due to the fact that the cohabitation procedure was modified from the first experiment to allow observing animals (see Methods). This modification could have distracted the animals and contributed to weaker pair bonding. However, the difference in the strength of partner preference is not a factor contributing to opposite directions of LIT’s effects on acquisition versus expression of pair bonding because we did not observe any facilitating effects of this treatment on expression of pair bonding in the experiment with two hours of cohabitation (Fig. 4A).
The facilitating effects of LIT on the acquisition phase of pair bonding in our study manifested themselves in a significant difference in huddling time with partner versus stranger animals and in a significant preferential huddling in LIT-injected, but not vehicle-injected test animals during the entire PPT (Fig. 3A,B). These effects are also in agreement with the observed significant difference in time spent in partner’s chamber in LIT-injected voles, but not in the vehicle-injected voles in the second half of the PPT (Figure S3F). While these comparisons were also done using non-parametric tests, they are in overall agreement with publications showing that central administration of OXT can promote acquisition of pair bonding [9, 16, 19]. An early study demonstrated that acquisition of pair bonding in male prairie voles, measured by partner preference and selective aggression against strangers, was inhibited by central administration of a AVP receptor, but not an OXTR, antagonist, and that central administration of AVP, but not OXT, promoted acquisition of pair bonding [31]. Two subsequent studies in female prairie voles showed that central administration of OXT promoted acquisition of pair bonding, measured by partner preference, and that OXTR antagonist, but not AVP antagonist, inhibited this behavior [19, 68]. These findings gave rise to an early hypothesis that OXTR regulates pair bonding in females, whereas AVP receptors regulate pair bonding in males. This hypothesis was addressed by a direct comparison between males and females showing that central administration of either OXT or AVP can promote acquisition of pair bonding in both sexes, but that this behavior in males is sensitive to lower doses of AVP than OXT [16]. The same comparative study showed that the facilitating effects of OXT were attenuated by either administration of an OXT or an AVP receptor antagonist. Thus, both receptors are involved in facilitating effects of OXTR on acquisition of pair bonding with a potential additional reliance on AVP receptors in males. Subsequent studies using peripheral administration of peptides showed that OXT promoted acquisition of pair bonding in female, but not male, voles suggesting lesser reliance of this behavior on OXTR in males [10, 57]. On the other hand, administration of an OXTR antagonist into the nucleus accumbens of male prairie voles or into the caudate putamen of female prairie voles prior and during cohabitation inhibited subsequent partner preference [32, 74]. Experiments using viral overexpression and RNA interference in the nucleus accumbens definitively indicating the role of OXTR in formation of pair bonding did not clarify its role in male prairie voles, as they were performed only in females [34, 75, 76]. Thus, while the facilitating role of OXT in acquisition of pair bonding is well established, the specific role of OXTR in this behavior in males is not completely delineated. Our finding of facilitating effects of LIT on acquisition of pair bonding indicates the specific importance of OXTR for this behavior in males. This finding does not mean that AVP receptors are not involved in this behavior, only that activation of OXTRs is sufficient to facilitate acquisition of pair bonding.
Our need to rely on non-parametric tests in statistical analyses of behavior was driven by the lack of normal distribution of our data. This lack of normality is primarily due to presence of a substantial number of zero values. This distribution is expected when using short cohabitation intervals which may lead to low measures of partner preference, particularly huddling time, that can also be disrupted by treatments. As in other studies using either longer or shorter times intervals of cohabitation [73, 77], there is a substantial individual variability in the huddling times in vehicle-injected animals. This variability could be not only due to technical variation of sampling, but also due to biological factors, including genetic polymorphisms regulating regional distribution of relevant receptors [78–81] and epigenetic factors, for example differences in dominance status [82]. In turn, variations in these factors, could contribute to differences in sensitivity and effects of OXTR activation. It would be interesting to investigate the contribution of genetic and epigenetic factors to the effects of OXTR agonists on acquisition and expression of pair bonding.
Since LIT was administered systemically, it could be targeting multiple sites of OXTR expression. Based on previous demonstrations that central sites of OXTR expression are responsible for regulation of social behaviors, it is tempting to hypothesize which of the neurocircuits are involved in signaling the presence of social partner [83]. Earlier studies on viral manipulations of OXTR expression in female prairie voles or pharmacological manipulations of OXTR activity in male prairie voles implicated the nucleus accumbens in promoting pair bonding [32, 34, 75, 76]. Furthermore, chemogenetic inhibition of nucleus accumbens prior to PPT (testing expression of pair bonding) decreased affiliation of female prairie voles with their male partners, while chemogenetic activation of this brain region increased affiliative behaviors towards strangers [84]. Depression-like behavior following viral knockdown of OXTR expression in nucleus accumbens also suggests a potential role of this brain region in expression of other indices of pair bonding [60]. On the other hand, chemogenetic activation of the lateral septum in male prairie voles increased non-specific affiliative behavior and acquisition of pair bonding, but decreased selective aggression, an index of expression of pair bonding [85]. The latter observation suggest that varied brain regions could be involved in regulation of not only different phases of pair bonding but also different behaviors associated with pair bonding. Further assays modulating OXT and OXTR with pathway- and region-specificity will have to be conducted to deduce the exact neurocircuitry that signals the presence of a social partner.
Due to colony limitations, we have tested the effects of OXT agonism only in male animals. The generality of our observations should be also evaluated in female voles. Nevertheless, given the interest in manipulating the OXT system for the treatment of various mental conditions, our observations suggest important translational implications. Specifically, they suggest unpredictable consequences of targeting OXTR for established social affiliations which would depend on the history of the interactions and their social conditions. Here we suggest that potential social consequences of administration of OXTR agonist should be kept in mind. These consequences, however, would not preclude the potential benefit to target OXTR in the case of relevant psychiatric conditions.