We previously found that partner preference formation, following 1 week of cohabitation, can occur in the absence of Oxtr function⁵⁹. However, acute modulation of Oxtr signaling can alter patterns of social behaviors between individuals^29,50,66. These observations suggest that changes in Oxtr function may influence early social interactions that contribute to the trajectory of pair bond formation and/or specific aspects of attachment behaviors following the formation of partner preference. We first wished to determine whether Oxtr function facilitated pair bond formation by examining the display of partner preference following shorter periods of cohabitation. Consistent with previous studies showing that the cohabitation time required for the display of partner preference differs between males and females, we find that WT females demonstrate partner preference after only 6 hours of cohousing with a WT male partner, while WT males require 5 days following introduction to a potential WT mate (Fig. S1A-C)⁶⁷. We therefore cohoused WT or Oxtr^1-/- females with WT males for 6 hours, and WT or Oxtr^1-/- males with WT females for 5 days following introduction (short cohabitation), or in these combinations for 1 or 7 days (long cohabitation) for females or males, respectively (Fig. 1A, Fig. S1A).

Both males and females lacking Oxtr show an absence of partner preference following short cohabitation with WT partners (Fig. 1B). Oxtr^1-/- females, but not males, showed a decrease in the fraction of huddling time with their partner and an increase in the fraction of huddling time with a stranger compared to WT controls (Fig. 1C-D). Oxtr^1-/- males present an absence of partner preference (Fig. 1B), resulting from a trend towards decreased time with their cohoused partner and increased time with strangers (Fig. 1C-D). Consistent with our previous studies, we find no difference between WT and Oxtr^1-/-females when the pairs are cohoused for the longer duration (Fig. S1B-F).

To better understand the social interaction dynamics underlying differences in huddling behavior between WT animals and their siblings lacking Oxtr after short-term cohabitation, we quantified the cumulative time subjects spent huddling with either their cohoused partners or strangers across the assay.Oxtr^1-/- females show a significant decrease in the amount of time spent huddling with partners in the last 30 mins of the assay (150-180mins) (Fig. 1E). We also observe a significant increase in huddling time of Oxtr^1-/- males with strangers during the middle periods (75 – 130mins) (Fig. 1F), suggesting that they engage in more early prosocial behavior with strangers than WT males. Consistent with our previous work, we find no difference between cumulative huddling times with either partner or stranger between WTs and Oxtr^1-/- animals of both sexes after long cohabitation periods (Fig. 1G-H)⁴⁵. These observations demonstrate that Oxtr function changes the temporal dynamics of social interactions with both familiar and unfamiliar animal in a sex-specific manner. Taken together, our findings suggest that Oxtr function acts early in the process of pair bonding to facilitate the formation of the partner preference.

Oxtr influences the dynamics of interactions with partners and strangers differently between the sexes. We therefore wished to examine the detailed patterns of these interactions separately in females and males (Fig. 2A). We examined the distribution of huddles by WT vs Oxtr^1-/- females following short cohabitation as a measure of prosocial behavior displayed by these animals. We find that the top quartile of huddle durations (long huddles) from each subject animal make up an average of ∼90% of the total huddle duration, functionally separating these interactions from incidental side-by-side contact that may not constitute affiliative huddling (Fig. S2A Left). We classified long huddles as partner huddles or stranger huddles and discovered that, consistent with the overall time spent, the median huddle duration carried out by Oxtr^1-/- females is significantly shorter with partners and significantly longer with strangers, compared to WT females (Fig. 2B). The longest huddle displayed with partner is significantly longer for WT females compared to Oxtr^1-/- females though we find no difference in the length of the overall longest huddle. This is due to some Oxtr^1-/- females conducting their longest huddle with the stranger instead (Fig. S2B). Oxtr^1-/- females additionally executed a larger number of huddles only with the stranger, but not the partner, compared to WT controls (Fig. 2C). Taken together, these results suggest that Oxtr function in females both facilitates prosocial behavior with partners and suppresses such behavior with strangers during the early stages of pair bond formation.

We next wished to determine whether the loss of Oxtr function influences patterns of prosocial behaviors after the display of partner preference. We therefore examined patterns of huddling, in particular the duration and frequency of these interactions, following long cohabitation, a time point at which both WT and Oxtr^1-/- animals display robust preference for partners over strangers⁵⁹ (Fig. S1D). As with short cohabitation, the top quartile of huddles (by duration) for each subject animal following long cohabitation, described an average of 90% of total huddle time (Fig. S2A right). Following long cohabitation, preference and huddling behavior in Oxtr^1-/- females is indistinguishable from their WT siblings (Fig. 2D-E). Comparing changes in the frequency of top quartile huddles displayed by Oxtr^1-/- females after short vs long cohabitation, we find that both WT and Oxtr^1-/- females show decreased huddle frequency with stranger following long cohabitation (Fig. S2C-D). On comparing durations of long huddles, we find that Oxtr^1-/-females significantly increase their median huddle duration with partner post long cohousing (Fig. 2F-G). Thus, loss of Oxtr function in females reduces the duration of early prosocial interactions (short cohabitation) with partners and strangers while Oxtr-independent mechanisms eventually promote WT patterns of affiliative huddling and partner preference following longer periods of cohabitation.

Our results suggest that Oxtr function influences specific aspects of early pair bonding and attachment behaviors. We next wished to dissociate the influence of Oxtr on the formation and display of partner preference from the display of agonistic rejection behaviors towards novel strangers following cohabitation. We therefore developed a battery of assays to examine initial social interactions between mating partners at early and later stages of pair bonding, as well as stranger rejection following cohabitation (Fig.2A). We examined the behavior of subject females from each genotype with WT males when naive animals were first introduced to each other (Introduction), mating behaviors following estrus induction (Timed Mating), prosocial interactions following reunification with their bonded mate (Partner Reunification), and agonistic rejection of novel, opposite sex strangers (Stranger Rejection) (Fig. 2A see Methods). We find no differences in social behaviors during Introduction, Timed Mating and Partner Reunification (Fig. 2H-J). In contrast, during Stranger Rejection, Oxtr^1-/- females show more social behaviors towards novel naive males than WT females do in the absence of their partner (Fig. 2K). Taken together, these studies demonstrate that Oxtr function regulates specific components of pair bonding in females in a state-dependent manner, suppressing promiscuous social interactions with stranger males after cohabitation sufficient for the display of partner preference.

Given the Oxtr-dependent difference in huddling behavior between males and females after short cohabitation (Fig. 1C-D), we wished to assess sex differences in the patterns of behaviors displayed by Oxtr^1-/- animals and to further investigate Oxtr function in males (Fig. 3A). Examining long huddles (as above, Fig. 2), we find that such huddles by Oxtr^1-/- males with their partners following short cohabitation are significantly shorter when compared to WT siblings, contributing to diminished partner preference early in pair bonding (Fig. 3B Left). Interestingly, while Oxtr^1-/- males show no difference in the total amount of time spent huddling with strangers when compared to WT siblings (Fig. 1D), they display shorter and more frequent huddles towards them at this time (Fig. 3B, 3C Right). In males, Oxtr thus influences the pattern as well as amount of early affiliative behavior to distinct behavioral stimuli in different ways.

Oxtr^1-/- males show no difference in time spent with either partners or strangers after long cohabitation (Fig. S1E-F). However, these males show a persistent reduction in the duration of their long huddles with strangers at this time point (Fig. 3D). Following long cohabitation, WT and Oxtr^1-/- males show no difference in the duration of long huddles with partners and a reduction in the duration of long huddles with strangers (Fig. S3A-B). We find no difference in the frequency of long huddles between Oxtr^1-/- and WT males after long cohabitation (Fig. 3E). Consistently, comparing Oxtr^1-/- males’ behavior following short vs long cohabitation reveals that the frequency with which Oxtr^1-/- males huddle with strangers decreases following long cohabitation (Fig. 3F-G). Thus, specifically in males, loss of Oxtr function does not change the overall time spent huddling with strangers after short or long cohabitation but does cause a continuing disruption in the patterns of affiliative behavior that comprises huddling with strangers.

Our observations suggest that the loss of Oxtr disrupts certain characteristics of prosocial behaviors in males regardless of the time spent bonding with their partner. We next wished to dissect whether such loss of function resulted in atypical affiliative or agonistic behaviors across stages of bonding. We examined the behavior of Oxtr^1-/- males and their WT siblings in the same series of paradigms described previously to characterize the formation and maintenance of their pair bonds (Fig. 3A). Oxtr^1-/- males spend similar amounts of time as WT males engaging in prosocial behaviors with partners regardless of bonding state (Fig. 3H-J, S3E). Similar to our observations in Oxtr^1-/- females, Oxtr^1-/-males spend more time engaging in prosocial behaviors with strangers in the absence of their partners (Fig. 3K).

In contrast to time spent conducting mating behaviors with WT female partners (Fig. S3E), Oxtr^1-/- males demonstrate increased frequency of mating attempts when these females are in estrus (Fig. 3I). Examining all instances of mounting, we find that Oxtr^1-/- males carry out shorter mounting bouts more frequently than WT males (Fig. S3F). To determine whether the shorter bouts of mounting in Oxtr^1-/- males influence the receptivity of their partners, we compared the mount-to-intromission ratio for each animal⁶⁸ and find that females are equally receptive to Oxtr^1-/- males as their WT siblings (Fig. S3G). Thus, consistent with our previous observations, Oxtr influences patterns, but not the amount or success, of male mating behaviors, suggesting that such changes in males’ behaviors are equally able to elicit receptivity in females.

Oxtr function influences distinct components of males’ attachment-related social behaviors towards unfamiliar females at different stages of pair-bonding, lengthening mating bout durations when naïve and suppressing promiscuous prosocial behaviors upon bonding. Taken together our results suggest that Oxtr influences patterns of social behaviors with future partners during the early phases of pair-bonding differently in females and males, but upon bond formation, functions in both sexes to suppress promiscuous affiliative behaviors towards strangers (Fig. S3H).

Oxtr function influences patterns of social interactions between mates early during the formation of a pair-bond (Short Cohabitation in both sexes, Introduction and Timed Mating in males). We therefore investigated the effect of Oxtr loss on the choice made by a potential mate. We developed a variation of the PPT three-chamber assay to present a naive WT “chooser” with a choice between a naive opposite sex WT or naive opposite sex Oxtr^1-/- animal (Fig. 4A, Fig. S4A, see Methods). Using the “naïve-choice” paradigm we find that, over the course of 6 hours, WT choosers spend significantly more time with one of the two stimulus animals (winner) (Fig. 4B) and that WT and Oxtr^1-/- animals are chosen as the winner with equal frequency(Fig. 4C).

To examine whether Oxtr influences the dynamics of the choice made by a potential WT mate, we compared the time spent with the winner from a lane where the chooser had a choice between a naïve WT or Oxtr^1-/- animal (WTOxtr lanes) versus that of a chooser choosing between 2 naive WT animals (WTWT lanes) of the opposite sex (Fig. 4D). While naïve WT choosers show no difference in overall social choice between naive WT and Oxtr^1-/- opposite sex animals (Fig. 4C), WT female choosers show distinct behavioral dynamics in the presence of an Oxtr^1-/- male (WTOxtr lane). In the presence of an Oxtr mutant (WTOxtr lane), WT females spend significantly more time in contact with the winner, regardless of genotype, when compared to contact with the winner between two WT males (WTWT lane) (Fig. 4E). In contrast, we observe no difference in contact time with a winner when a naive WT male is choosing in WTOxtr versus WTWT lane (Fig. 4E).

We therefore manually examined the specific behaviors shown by female WT choosers during their interactions with both the winner and loser in WTWT and WTOxtr lanes. Preference indices using these metrics were highly concordant with automated scoring (Fig. S4A). The time spent by the female chooser displaying prosocial behaviors towards the winner and the time spent huddling with the winner in WTOxtr lanes was significantly greater than WTWT lanes (Fig. S4B) (Fig. 4E). The divergence between the amount of cumulative time spent huddling with the winner from WTOxtr versus WTWT lanes occurs after 5 hours have elapsed (313-360 mins, Fig. 4F), suggesting that complex interactions in the presence of a male lacking Oxtr influence how females “choose” a male partner regardless of whether they end up choosing a WT or Oxtr^1-/- winner.

We evaluated the duration of the longest huddles conducted by all WT female choosers, and find that the longest huddle is always conducted with the winner. The longest huddles conducted in the WTOxtr lanes, regardless of the genotype of the male winner, are greater than those from WTWT lanes (Fig. 4G). There is no difference in the latency to start the longest huddle between WTOxtr and WTWT lanes (Fig. S4C). Thus, the presence of the Oxtr^1-/- male in the lane leads a WT chooser female to execute the longest huddle for a greater amount of time compared to the longest huddle by the WT female chooser from a WTWT lane (Fig. 4G). These observations suggests either that naïve Oxtr^1-/- males elicit longer longest-huddle bouts from WT females, even when this huddle occurs with a WT male winner, or that the behavior of naïve WT males reduces the duration of WT females’ longest huddle - a reduction that is absent in the presence of a male lacking Oxtr.

To determine if specific patterns of behavior preceded the longest huddle with a winner, we aligned the onset of the longest huddle for all the WTOxtr and WTWT lanes and evaluated the behaviors performed 15 mins prior to its initiation. We find that the duration of sniffs (anogenital investigation, see methods) by WT female choosers in WTWT lanes is greater than those by choosers in WTOxtr lanes (Fig. 4H). We did not find any difference in the duration of sniffs specifically with the winner or the loser when comparing WTWT and WTOxtr lanes, or in the total time spent sniffing over the course of the assay (Fig S4D, E). Thus, the interactions with both winner and loser contribute to the increase in duration spent sniffing 15 minutes prior to the longest huddle in WTWT lanes. These observations may suggest that a female chooser from the WTWT lane continues to seek chemosensory input before the longest huddle and these huddles are shorter due to ambivalence in a choice between two WT males. In contrast, a WT female chooser in a WTOxtr lane may encounter larger differences between a WT versus Oxtr^1-/- male, leading to a display of more prosocial behavior towards the chosen male independent of the winner’s Oxtr function.

We next examined antagonistic behaviors displayed during the naive choice assay as such behaviors may reduce prosocial behaviors displayed by chooser females. We found no difference in the overall counts of aggressive behaviors between WTWT and WTOxtr lanes (Fig. S4F). However, WT female choosers attack the loser male in the WTWT lanes more compared to WTOxtr lanes (Fig. 4I, Fig. S4G). Thus, in WTWT lanes, winner selection may be influenced by the level of aggression elicited by naïve WT males, a factor that is lower in WTOxtr lanes. We find a correlation between the amount of time spent socializing (with both winner and loser) and the count of all the aggression during the assay (Fig. 4J). Female choosers from WTOxtr lanes tend to spend more time engaged in prosocial behaviors with the winner (Fig. 4E-F, S4B) and display inceased prosocial and reduced aggressive behaviors. Female choosers choosing between two WT males, however, display either high prosocial and low aggressive behaviors or low prosocial behavior and high aggression. These observations suggest that the presence of a second male may influence the behavioral state of a WT female. The presence of only WT males may induce aggression by females against eventual losers, while such displays are absent in the presence of a male lacking Oxtr regardless of a female’s choice.

Given the differences in the influence of the presence Oxtr^1-/- males on displays of aggression by WT females when choosing between males, we examined antagonistic behaviors exhibited during interactions between naive (Introduction) or bonded (Stranger Rejection) females with WT or Oxtr^1-/- males in the absence of a third animal. Strikingly, we find that counts of aggressive behaviors were elevated only when naive WT females encountered WT males compared to Oxtr^1-/- males, while this difference is lost following pair bonding (Fig. 4K). In contrast, we find no differences in aggressive behaviors when comparing naïve or bonded WT and Oxtr^1-/- females (Fig. S4H). Taken together, these data suggest sex- and state-specific roles for Oxtr function. Specifically, in naive males, Oxtr function induces displays of aggression by naive females, and the presence of a male lacking Oxtr diminishes a naïve WT female’s propensity to display agonistic behaviors even in the presence of a WT male. Following bonding, however, the absence of Oxtr function in males does not appear to influence the behavior of WT females despite the promiscuous prosocial displays such mutant males.

Finally, to determine if naive prairie voles lacking Oxtr display generalized changes to their behaviors independent of attachment-associated social behavior, we examined general pro-social, anxiety-related, and locomotor behaviors. With the exception of an increase in locomotion displayed by Oxtr^1-/- females when compared to WT females in an open field paradigm, we found no differences between Oxtr^1-/- and WT voles of either sex (Fig. S4I).

We have examined social interactions during the initial contact between mates, the early stages of pair bonding, and during social attachment behaviors following partner preference formation. Consistent with our findings and previous work, Oxtr function is not required for the eventual display of partner preference. Here, we demonstrate that Oxtr function sex-specifically facilitates early social interactions that contribute to the formation of partner preference, and controls promiscuity.

While partner preference formation can occur in the absence of Oxtr function, we find that it suppresses prosocial behaviors towards novel strangers following pair bonding. These results suggest that distinct components of pair bonding are differentially influenced by Oxtr function or compensated upon its loss. We next characterized the molecular signature associated with pair bonding to determine whether changes in gene expression following pair bond formation are controlled by Oxtr. Comparative studies demonstrate that Oxt binding in multiple brain regions differs significantly between promiscuous and monogamous rodent species. Oxt binding is particularly enriched in the NAc in prairie voles, while vasopressin binding is largely absent in this area^45,47. Thus, analyzing gene expression in the NAc allows us to identify bonding-related expression that is primarily due to Oxtr function and largely independent of the direct consequences of AVP signaling. Additionally, given the well-established role of the NAc in the association of reward or aversion with specific sensory and environmental cues^69–71 this region is well poised to play a role in the development of partner preference, even without Oxtr signaling, as well as stranger rejection, which appears to be dependent upon Oxtr function.

We performed bulk RNA-sequencing on NAc tissue from WT and Oxtr^1-/- animals of both sexes that were either: group housed with same-sex peers and sexually naïve (pre-pairing); or paired with WT opposite-sex partners for six days, using our timed mating paradigm (post-pairing) (Fig. 5A). PCA showed robust separation of WT and Oxtr^1-/-samples (Fig. S5A). Differential expression (DE) analysis of pairing condition, regardless of genotype or sex, identified 13 genes that show significant changes in expression post-pairing (Fig. S5B), all of which showed decreased expression (Fig. 5B). These genes include Agt, Sgk1, Ptgds, Fosb, Pdk4, and Acsm5 (Fig. 5C) and, while these genes have not previously been linked to shared gene regulatory or functional pathways, the predicted protein-protein interaction network for the 13 DE genes shows more connections than would be expected by chance (p=2.47 e⁻⁰⁵) (Fig. S5C). Intriguingly, even in the limited set of genes with altered expression in the context of pair bonding, changes are driven primarily by expression in WT females (Fig. 5C, S5D). Subgroup analyses by genetic background and sex identified additional genes including Adora2a, Fkbp5, and Spred3 in WT females and collagen-related genes, including Col4a1 and Col4a2, in Oxtr^1-/- females (Fig. 5D, S5D). Only one gene, Sema3b, is significantly different in Oxtr^1-/- males following pairing (Fig. 5D).

To further examine changes in patterns of gene expression in the NAc associated with pair bonding, we performed weighted gene coexpression network analysis (WGCNA)⁷² and calculated three sets of modules (M) from different sample subsets (modules^WT, modules^Mut, and modules^All) (Fig. S5F, S6D). Using the modules^WT subset, we identified 15 coexpression modules (Fig. S5F, Table 2) representing genes that share highly similar expression patterns within the NAc. We identified one module (Mpink^WT) that significantly and specifically correlated with pre-pairing status in WT animals (Pearson R correlation of 0.7, Bonferroni adjusted p-value (p_adj)=0.018) (Fig. 5E, S5E-F). Genes with high module membership for the Mpink^WT are also those with high gene significance for pairing condition supporting the module’s association with behavior condition (Fig. 5SG). Further, genes significantly differentially expressed with bonding are significantly enriched in Mpink^WT (Bonferroni p_adj=7.44 e⁻⁰⁸) and serve as hub genes within this module (Fig. S5E, G). These results suggest that the Mpink^WT module might be driving, or under the influence of pathways driving, the pre-pairing to post-pairing state change.

We asked whether any of the modules^WT identified were regulated by Oxtr and thus lost in Oxtr null animals. To address this, we assessed how well modules^WT are preserved in mutant samples. We calculated a Z_summary measure that combines module density and intramodular connectivity metrics into a composite statistic. We find that of the modules^WT, Mpink^WT had the lowest Z_summary (Z_summary = 2) in Oxtr^1-/- samples (Fig. 5F), indicating that dynamics of gene expression that correlate with bonding states are lost in the absence of Oxtr. Finally, the module eigengene (ME) for Mpink^WT when calculated for all 30 samples and compared between pre- and post-pairing, was significant only for WT females (Wilcoxon rank sum, Benjamini-Hochberg p_adj=0.03) and not for WT males or Oxtr^1-/-animals of either sex (Fig. 5G). These observations suggest there is sex-specific, coordinated regulation of gene expression in the NAc following pairing of mates that may be associated with pair bond associated behaviors.

We next wished to determine how Oxtr signaling contributes to changes in gene expression in the NAc associated with the naive state or the process of pair bonding (Fig. 6A) by comparing the transcriptional profile of WT and Oxtr^1-/- animals of each sex. Given that our lines are maintained on an outbred background, in order to minimize sample variability, tissue from three animals was pooled for each sample. We observed significant genotype specific differences in gene expression in animals from both sexes, with 466 genes showing differences between WT and Oxtr^1-/- females, 270 between WT and Oxtr^1-/- males, and 1,014 genes when samples from both sexes were combined (using a p_adj<0.05 and absolute (log₂(FC)) > 0.25 as criteria). Of the combined DE gene set, 370 genes are upregulated and 644 are downregulated (Fig. 6B, S6A-B). Genes with the greatest fold change across both sexes include Slc2a3, Kcna1, Rgs8, Cnksr2, Acvr, and Cnot6 (Fig. 6B). Genes specifically downregulated in Oxtr^1-/- females with log₂(FC) less than 1 include Abcb11 (a gene recently associated with treatment resistant schizophrenia⁷³), Ngfr, Pon3, and Shh. While there was no significant difference in Oxtr transcript levels by genotype (p_adj = 0.753)—consistent with minimal nonsense-mediated decay despite a premature stop codon—we have previously shown that no functional protein is produced in Oxtr^1-/- animals via autoradiography⁵⁹. V1aR showed a modest decrease in expression in Oxtr^1-/- animals (p_adj=0.03) (Fig. S6C), though this is likely influenced by variability in WT samples. WGCNA of all samples identified two modules (Mbrown^All and Mturquoise^All) that were significantly positively correlated with the WT genotype compared to mutant (Pearson R = 0.55, p_adj = 0.011 for brown and Pearson R = 0.8, p_adj = 6.6e-7 for turquoise) (Fig. S6D). None of the identified modules showed correlation with pairing condition (Fig. S6D). Our results thus reveal a dramatic change in patterns of gene expression in the NAc with global loss of Oxtr signaling, independent of bonding state.

Examination of the sets of genes whose expression is altered by the loss of Oxtr, regardless of bonding state or sex, revealed several distinct cellular processes regulated by Oxtr signaling. These processes comprise larger domains of cell function including extracellular channel binding activity, axoneme assembly complex, axon glial ensheathment development, and metal cation transport (Fig. 6C), as well as pathways for G protein signaling and dopaminergic neurogenesis (Fig. S6E). When analyzed by sex, more of the identified processes were enriched in female Oxtr gene sets than in male ones (female: 19/62, male: 6/62). To further link differentially expressed genes to behavioral traits, we asked whether genes altered by loss of Oxtr are selectively enriched for high-confidence associations with autism spectrum disorders (ASD) and other neurodevelopmental diseases (NDD) affecting social attachment in humans as compared to epilepsy, a disorder that does not disrupt social attachment^74–77. Consistently, we find that the genes upregulated with loss of Oxtr (p_adj<0.01, log₂(FC)>0) are specifically enriched for ASD- and NDD-associations but not epilepsy associations (Fig. 6D “Joint”). This enrichment is again primarily driven by genes upregulated in females and not males (Fig. 6D). Thus, similar to the factors that influence gene expression changes associated with pair bonding, many of the changes we identify in the NAc resulting from loss of Oxtr are primarily driven by female-specific effects that may play a conserved role in mediating social attachment in humans.

Given the expression of Oxtr in regions of the brain known to mediate social behavior in voles and other species^78,79, we wished to determine if genes whose expression is altered in the NAc by the loss of Oxtr were expressed not only in specific populations of cells within this region, but in other regions of the prairie vole brain implicated in social and attachment behaviors. We therefore examined the expression of 15 genes that showed significant changes in expression in WT females when compared to those lacking Oxtr using in situ hybridization (ISH) to both validate our RNA-seq findings as well as examine the expression patterns of these genes. Several of these genes, such as phosphodiesterase 10A (Pde10A) and diacylglycerol kinase beta (Dgkb), show robust expression throughout the striatum (data not shown), while genes such as the calcitonin receptor (Calcr) and delta like non-canonical notch ligand 1 (Dlk-1) are expressed in restricted populations of cells in the NAc (Fig. 6F-H, top panels). The expression of these genes is significantly reduced in Oxtr^1-/- females (Fig. 6E-H), supporting the specificity and resolution of our RNA-seq findings.

In parallel to our examination of gene expression in the NAc, our ISH revealed that loss of Oxtr in females altered the expression of the genes we examined in other regions of the prairie vole brain. We examined the expression of these genes in the lateral septum (LS), a region that has been implicated in agonistic and attachment behaviors in prairie voles and other rodents and shows species dimorphic patterns of Oxt and Avp binding. We observe that both Calcr and Dlk-1 are decreased in the absence of Oxtr (Fig. 6G, H middle panel) in the LS. We also find that expression of Calcr and Dlk-1 is decreased in the paraventricular nucleus of the hypothalamus (PVN), a principal source of Oxt and Avp in the rodent brain which sends projections to the NAc (Fig. 6G-H bottom panels)^80–82. These findings suggest that Oxtr function regulates patterns of gene expression in multiple regions of the brain implicated in social and attachment behaviors in a coordinated and region-specific manner.

We find here that loss of Oxtr results in coordinated changes in gene expression in multiple neural populations, including the PVN. We therefore wished to determine if the signatures of gene expression regulated by Oxtr in the NAc in prairie voles (ONP set), regardless of sex and bonding state, were enriched for patterns of gene expression associated with distinct neural populations in other regions of the brain, particularly those implicated in social and attachment behaviors. We compared the ONP gene set to the molecular signatures of 33 hypothalamic subtypes previously identified by single-cell RNA sequencing in mice⁸³. Of the 33 clusters examined, the four with highest overlap with the ONP gene set represent subsets of PVN neurons, in particular those that express Oxt and AVP (Fig. 7A, clusters 43 and 26), as well as PVN neurons that express Dlk1 and Calcr (cluster 29), both of which show reduced expression in the NAc and PVN in females lacking Oxtr (Fig. 6G-H bottom panel).

We next sought to determine if the overlap between Oxtr regulated genes and gene expression patterns of neural populations in the PVN correlated with changes in the expression of neuropeptides within PVN itself. We examined patterns of Oxt and Avp expression using immunohistochemistry in naive adults and compared patterns of expression between WT animals of both sexes and those lacking Oxtr (Fig. 7B, Fig. S7A-B). We find that the density of both Oxt and Avp producing neurons is lower in Oxtr^1-/- animals (Fig. 7C-E). To determine if the loss of Oxt or Avp expression was restricted to specific populations of PVN neurons, we examined the expression of these peptides across the anterior-posterior axis of the PVN in both WT and Oxtr^1-/- males and females. We find that animals of both sexes show a preferential loss of Oxt and Avp expressing neurons in the anterior PVN (Fig. 7F-G). Finally, we examined patterns of Oxt and Avp over the course of development to determine when loss of Oxtr function influences the development of their expression. We find that the density of Oxt neurons at P0 (Fig. S7C-F) and both Oxt and Avp neurons at P21 is lower in Oxtr^1-/- females compared to age matched WT females (Fig. S7G-J). In contrast, we find that Oxt and Avp neuron density loss occurs only in adult Oxtr^1-/- males (Fig. 7D-G, Fig. S7B). These observations suggest the coordinated, region-, and sex-specific influence of Oxtr function on the development of neurons expressing Oxt (and Avp) within the PVN, consistent with a trophic role for this pathway identified in mice^63,65,84,85.