Losing a spouse or life partner is a deeply traumatic event that can have long-term repercussions. Given enough time, however, most surviving partners are able to process their grief. The neural processes that enable people to adapt to their loss remain unknown.

To explore this question, scientists often turn to animals that form long-term mating based pair bonds and can be raised in the laboratory. Monogamous prairie voles enter lifelong partnerships where the two individuals live together, prefer to cuddle with each other, and take care of their pups as a team. After having lost their mate, they show signs of distress that eventually subside with time.

Sadino et al. examined the biological impact of partner loss in these animals by focusing on the nucleus accumbens, a brain region important for social connections. This involved tracking gene expression – which genes were switched on and off in this area – as the voles established their pair bonds, and then at different time points after one of the partners had been removed.

The experiments revealed that establishing a relationship leads to a stable shift in nucleus accumbens gene expression, which may help maintain bonds over time. In particular, genes related to glia (the non-neuronal cells which assist neurons in their tasks) see their expression levels increase, indicating a previously undescribed role for this cell type in regulating pair bonding. Having their partner removed led to an erosion of the gene expression pattern that had emerged during pair bonding; this may help the remaining vole adapt to its loss and go on to form a new bond. In addition, Sadino et al. explored the gene expression of only neurons in the nucleus accumbens and uncovered biological processes distinct from those that occur in glia after partner separation. Together, these results shed light on the genetic and neuronal mechanisms which underlie adaptation to loss; this knowledge could one day inform how to better support individuals during this time.

The death of a romantic partner is cited as one of the most traumatic experiences in a person’s life and results in grief and corresponding indicators of mental and physiological distress (Keyes et al., 2014; Holmes and Rahe, 1967). However, for the majority of people acute grief subsides within approximately six months as the bereaved integrates and adapts to the loss (Shear et al., 2011). Most people will also eventually form a new pair bond, which provides a behavioral indicator of loss adaptation (Shear and Shair, 2005). The processes that enable such adaptation remain poorly understood but likely occur via the same neural systems that are uniquely engaged by pair bonding (Blocker and Ophir, 2016).

Socially monogamous prairie voles are a laboratory-amenable species that recapitulates many aspects of human social bonds, making them ideal for interrogating the neurobiology of bonding and loss. Pair bonding in this species results in a shift in sociobehavioral state where specific behaviors are only evident after the transition from living in a same-sex pair (i.e. family group) to an opposite-sex bonded pair. Pair bonded voles will prefer to affiliate with their partner, display selective aggression towards non-partner individuals, and exhibit robust and organized biparental care (Williams, 1992; Getz et al., 1981; Williams et al., 1992; Insel and Young, 2001; Carter et al., 1995; Winslow et al., 1993). Considering that other stable behavioral shifts, such as reproductive status and dominance hierarchies, are underwritten by changes in transcription, we aimed to determine if a mature pair bond—and the process of adapting to loss—are underwritten by stable changes in gene expression in the nucleus accumbens (Cardoso et al., 2015; Zayed and Robinson, 2012; Tripp et al., 2018).

While an extended network of brain regions mediates social processing and decision making, the nucleus accumbens (NAc)—a region important for reward, motivation, and action selection—is a critical hub that is engaged when forming a bond and is implicated in loss processing (Aragona and Wang, 2004; Lim and Young, 2004; Walum and Young, 2018). In humans, holding hands with a romantic partner enhanced blood oxygenation levels (BOLD) in the NAc and successful adaptation to spousal loss is associated with a reduced partner-associated BOLD signal in this region (O’Connor et al., 2008; Kreuder et al., 2017). Thus, the NAc contributes to pair bonding and loss and broad scale changes in this brain region potentially mediate shifts in transcription which may be required to adapt to partner loss. Similarly, in prairie voles, pair bond-associated behavioral changes are underwritten by several known neuromolecular changes within the NAc which maintain and reinforce pair bonds over time (Insel and Young, 2001; Aragona and Wang, 2004; Young and Wang, 2004). Further, when separated from their partner, prairie voles exhibit behavioral and physiological distress that mirrors what is seen upon bond disruption in humans and other species, such as titi monkeys (Carter et al., 1995). These include increased circulating glucocortioids levels, activation of the HPA axis, increased anxiety, decreased pain thresholds, and autonomic dysfunction (McNeal et al., 2014; Grippo et al., 2007; Bosch et al., 2016; Pohl et al., 2019; Osako et al., 2018). However, we have previously shown that if a pair bonded prairie vole loses their partner, given enough time, they will form a new bond that supersedes the original bond (Harbert et al., 2020; Tamarin et al., 1990). As in humans, the ability to form a new bond indicates that prairie voles can adapt to partner loss.

Here, we map the trajectory of the pair bond transcriptional profile by comparing opposite-sex paired males to same-sex sibling male pairs. In the wild, sexually naive male voles can cohabitate with other males, especially siblings, although these relationships do not persist after males form an opposite-sex pair bond and establish a nest/territory with their partner (Getz et al., 1981; Carter et al., 1995). By comparing opposite-sex pair bonded voles to their same-sex paired counterparts before and after partner separation, we have an ethologically relevant means to isolate the unique biology of bonding and loss independent of those that support general affiliative interactions (e.g. peer relationships) or the stressful effects of social isolation more broadly. Thus, we compared behavior and NAc transcriptional profiles of opposite-sex pair bonded and same-sex cohoused naïve male voles to explicitly define the pair bond both behaviorally and transcriptionally. Then we examined how affiliative pair bond characteristics were altered by partner separation to test the hypothesis that pair bond transcriptional signatures and bond-associated behaviors would erode as a function of time since partner separation. We reasoned that these changes represent key components of loss adaptation that, together, may prime the vole to be able to form a new bond.

We found that pairing induced a reliable affiliative preference for a peer or a pair bonded partner, and that this preference is remarkably stable, persisting even after four weeks of separation. We further show that pair bond-associated changes in accumbal gene expression were consistent at two and six weeks post-pairing. However, once opposite-sex pairs are separated, the pair bond transcriptional signature erodes as a function of separation time. To further home in on the transcriptional changes associated with partner separation specifically in NAc neurons, we pioneered translating ribosomal affinity purification in voles (vTRAP). Using vTRAP, we identified clusters of genes associated with dopaminergic signaling, mitochondrial organization, and steroid hormone signaling whose expression patterns are sensitive to acute pair bond disruption and loss adaptation. In sum, our parallel behavioral and transcriptional data suggests that erosion of pair bond transcriptional signatures in the NAc precedes changes in affiliative partner preference, providing insight into time-dependent neuromolecular changes that may contribute to loss adaptation.

We determined how bonding and extended separation affects social behavior and NAc transcription in opposite-sex and same-sex paired males. We employed timepoints that are experimentally validated and ethologically relevant (Figure 1A). We paired all study animals for 2 weeks, a duration that reliably produces mature pair bonds (Scribner et al., 2020; Brusman et al., 2021). For the separation timepoints, the ability to form a new bond after loss serves as a behavioral metric of loss adaptation. Prior work has shown that male voles are able to form a new pair bond 4 weeks post-separation, but not 2 weeks or earlier (Harbert et al., 2020).

Figure 1 with 3 supplements see all

Download asset Open asset

Consistent with our laboratory observations, wild male voles who lose a partner will remain at the nest for ~17 days (Tamarin et al., 1990). Thus, we assessed behavior and transcription in paired voles two days and four weeks after separation—timepoints before and after the animal can form a new bond.

The pair bond transcriptional signature is stable in intact bonds

Changes in behavioral states, such as a shift in dominance or reproductive status, are supported by stable changes in transcription, although this has not been examined in the context of pair bonds (Cardoso et al., 2015; Zayed and Robinson, 2012; Tripp et al., 2018). While prior work has demonstrated that mating and cohabitation in prairie voles result in transcriptional changes within the NAc, the consistency of these changes as long as the bond remains intact has yet to be assessed (Duclot et al., 2020; Tripp et al., 2021; Resendez et al., 2016; Resendez et al., 2012; Aragona et al., 2006). Thus, we compared NAc transcription in opposite- versus same-sex-paired voles following either 2 or 6 weeks of pairing/cohabitation (Figure 2A). By comparing opposite-sex to same-sex-paired animals, we identified transcripts specific to pair bonds compared with those associated with affiliative behavior more generally. For consistency, we limited transcriptional assessment to voles that had a baseline partner preference >50% (Figure 2B; Supplementary file 1; OS n=15; SS n=11. Excluded for PPT <50% OS n=3; SS n=8).

Figure 2 with 2 supplements see all

Download asset Open asset

We used DESeq2 to identify transcripts up- or down-regulated in opposite- relative to same-sex paired males after short-term and long-term pairing (Figure 3—figure supplement 2A, B; Love et al., 2014). When ordering transcripts based on their log₂FoldChange at the short-term timepoint, the global transcriptional differences for opposite- versus same-sex pairs were strikingly similar following either 2 or 6 weeks of pairing, suggesting stable pair bond associated transcription across these timepoints (Figure 2D; Spearman’s Rho = 0.38, p=2.2 × 10^–16). We further determined that the observed correlation across timepoints is stronger than what would be expected by chance. We shuffled each animal’s cohort identity at the long-term timepoint and calculated Rho values of the log₂FoldChange of differential gene expression between the observed short-term pair bond and the shuffled long-term pair bond over 1000 iterations (Figure 2—figure supplement 1A–D). The cohort identity is a combination of three variables: pairing type (opposite-sex or same-sex), separation condition (remain paired or separated), and timepoint (short- or long-term). For example, long-term separated opposite-sex animals is one cohort identity of 8 total. This approach effectively randomized timepoint, partner type, and pairing status without disrupting underlying structure that exists within our transcriptional dataset due to correlations in expression across genes. Our observed Rho value was greater than 100% of iterations, indicating that the similarity in gene expression across short and long-term pairing timepoints is greater than would be expected by chance (Figure 2—figure supplement 1D). Next, we asked whether our observed association across timepoints was driven by moderately but consistently expressed genes. Eliminating transcripts whose expression difference fell within the middle two quartiles in the short-term group resulted in a greater Rho value (R=0.44, p<2.2 × 10^–16), indicating that the correlation across short- and long-term timepoints is more strongly driven by transcripts with larger expression differences between opposite- and same-sex pairs. To further interrogate the biological function of altered transcripts, we set nominal thresholds of log₂FoldChange >0.30 or<–0.30, which are sufficient to identify replicable differences detectable via alternate methods (Walker et al., 2018), along with a p-value threshold of p<0.05 to identify up- and down-regulated gene lists (DEGs). This nominal p-value threshold was used with the aim of identifying gene families regulating molecular pathways responsive to bonding instead of specific gene candidates. Prior studies in animal brain tissue using this p-value have yielded biologically meaningful insights (Walker et al., 2018; Mukamel, 2022; Kronman et al., 2021; Walker et al., 2022; Labonté et al., 2017; Krishnan et al., 2007; Peña et al., 2017; Seney et al., 2018; DEGs based on these thresholds are listed in Supplementary file 2). There was significant overlap in the upregulated (104 shared, Fisher’s Exact Test: χ²=30.24, p=5.36 X 10^–31) and downregulated (46 shared, Fisher’s Exact Test: χ²=13.32, p=7.05 X 10^–14) transcripts across timepoints (Figure 3—figure supplement 2E; Wang et al., 2015).

To further examine similarity in transcriptional patterns between short-term and long-term pair bonds we employed Rank-Rank Hypergeometric Overlap (RRHO; Figure 2E; Plaisier et al., 2010; Cahill et al., 2018). This threshold free approach allows us to determine if similar global transcription patterns are observed between two comparisons. The resulting RRHO heatmap is arranged into quadrants based on the direction of gene expression and each point represents the significance derived from the number of overlapping genes via the hypergeometric distribution (Figure 2E). We determined how similar global transcription is between short-term and long-term pair bonds by comparing differential gene expression in the opposite- vs same-sex 2-week paired to opposite-sex vs same-sex 6-week paired groups. We observed extensive concordance of transcriptional patterns between 2- and 6-week paired voles, further supporting that the transcriptional profile of pair bonds is stable (Figure 2F). Confirming that our observed RRHO signal was unlikely to be attributable to chance, we randomly shuffled cohort identity (Figure 3—figure supplement 4E) and, separately, shuffled the gene ranking of the combined pair bond genes (Figure 3—figure supplement 4G), both of which ablated the RRHO signal. Since the pair bond transcriptional profiles are concordant at both time points, the 2-week and 6-week animals from each pairing type were pooled to create a single, well-powered opposite-sex- vs same-sex-paired comparison that defines the combined pair bond transcriptional signature (Figure 2G).

To determine which biological pathways might underlie pair bonding we employed Gene Ontology (GO) and Ingenuity Pathway Analysis (IPA) (Yu et al., 2012). Gene lists for GO terms or IPA were generated either from DEG lists or from genes in concordant (UU or DD) RRHO quadrants. GO terms that passed a false discovery rate of p<0.05 were retained while IPA terms that had activation scores of at least –2 or 2 and whose enrichment for genes predicted to be regulated was significant (p-value <0.05) were kept (Supplementary file 3). Combined pair bond GO terms associated with the upregulated DEGs and the quadrant UU genes strongly implicate changes in glial cells and extracellular matrix organization (Figure 2H), which is mirrored by activation of glioblastoma signaling via IPA (Figure 2—figure supplement 2A). IPA additionally identified upregulation of Synaptogenesis, CREB signaling, and Endocannabinoid Neuronal Synapse pathways. All these pathways are consistent with pair bonding as a form of complex learning that is mediated by neuromodulatory signaling (Walum and Young, 2018; Loth and Donaldson, 2021; Modi and Young, 2011; Johnson and Young, 2017). Predicted upstream regulators of these pathways broadly support a role for learning and neuromodulation, including Creb1, Estrogen receptor alpha (Esr1), and various growth-factor and developmental genes (Figure 2—figure supplement 2A). Combined pair bond GO terms associated with the downregulated DEGs and the quadrant DD genes are implicated in nucleoside synthesis and synaptic plasticity (Figure 2H). IPA indicates a suppression of corticotropin releasing hormone signaling, potentially reflecting the strong social buffering of stressors that occurs in pair bonded voles specifically (Bosch et al., 2016; Smith et al., 2013; Bosch et al., 2009; Smith and Wang, 2014; Burkett et al., 2016; Figure 2—figure supplement 2A). Finally, the strong correspondence in the significance of GO terms between the combined pair bond DEGs and the RRHO quadrants further validates that pooling the time points to represent a single pair bond transcriptional signature retains biologically relevant information (Figure 2H).

The pair bond transcriptional signature erodes following prolonged partner separation

We next tested the hypothesis that the pair bond transcriptional signature erodes following long-term partner separation. We paired opposite-sex and same-sex pairs for 2 weeks prior to a baseline partner preference test and then immediately separated males into new cages where they were singly housed for either 48 hours (short-term) or 4 weeks (long-term) prior to harvesting NAc tissue for RNAseq (Figure 3A). As before, we only included animals with a baseline partner preference >50% (Figure 3B; OS n=12; SS n=11. Excluded OS n=4; SS n=5). We determined the effect of separation on the pair bond gene signature by first defining differential gene expression of opposite- vs same-sex animals after short- and long-term separation and then used RRHO to examine genome-wide expression changes relative to the combined pair bond signature (Figure 3C).

Figure 3 with 5 supplements see all

Download asset Open asset

First, we performed differential expression analysis of opposite- vs same-sex-separated males at each separation time point (Figure 3—figure supplement 2C and D). There were significantly more shared DEGs between the combined pair bond group and short-term separation group than would be expected by chance, but this was not observed when comparing the shared genes between the combined pair bond and long-term separation group (Figure 3—figure supplement 2F Fisher’s exact test: pair bond:short-term sep χ²=31.86, p=6.78 X 10^–19; pair bond:long-term sep χ²=17.96, p=0.53). The difference in transcription between short- and long-term separation is particularly evident when comparing the global transcriptomes through the lens of pair bond transcription. Specifically, when genes are ordered based on their degree of down- or up-regulation in the pair bond group we found that gene expression was largely indistinguishable in short-term separated animals (Rho = 0.32, p<2.2 × 10^–16) and distinct from long-term separated animals (Rho = 0.048, p=1.6 × 10^–7), indicating that the pair bond signature is largely eroded after prolonged separation (Figure 3D). We further assessed whether these correlations differed from chance via the previously employed permutation analysis. We found that the pair bond:short-term Rho value indicated a stronger association than would be expected by chance (p=0.0362) while the pair bond:long term separation association value was highly likely to occur by chance (p=0.41). Together, these results suggest that extended separation erodes nucleus accumbens pair bond transcription, a potentially crucial molecular process needed to adapt to partner loss.

To more comprehensively assess genome-wide changes to the pair bond signature following extended partner separation we again employed RRHO (Figure 3E). We compared the ranked gene list of opposite- vs same-sex for the (1) combined pair bond gene signature (Figure 3D, top) versus either (2) short-term partner separation (Figure 3D, middle) or (3) long-term partner separation (Figure 3D, bottom). We found that the pair bond signature remains largely concordant after short-term separation as indicated by significant signal among upregulated genes (quadrant UU), and to a lesser extent, downregulated genes (quadrant DD; Figure 3E). However, following long-term separation, the pair bond gene signature is largely undetectable as indicated by a lack of significant signal throughout the RRHO plot. Plotting the long-term RRHO with an unadjusted p-value scale revealed few overlapping genes in the UU and DD quadrants (scale 1–16 Figure 3—figure supplement 4C rather than 1–300 as in Figure 3D). The low level of overlap indicates that there is a residual, albeit greatly reduced, pair bond signature that is still detectable after long-term separation (Figure 3—figure supplement 4C). Additionally, we used the same control analyses as previously employed (Figure 3—figure supplement 4G–I).

We next examined which genes are consistently present in the UU or DD quadrants between pair bonding and each separation timepoint. Specifically, we compared the UU and DD quadrants (inclusive of all transcripts in each quadrant) of the short-term (2 week) vs long-term (6 week) pair bond from Figure 2F, the combined pair bond vs short-term separation in the left side of Figure 3E, and the combined pair bond vs long-term separation on the right side of Figure 3E. The degree of overlap in the UU quadrants was greatest between the combined pair bond vs short-term separation and least between the combined pair bond vs long-term separation, a result consistent with the most extensive erosion of pair bond transcription occurring after long-term separation (Figure 3F: all stats in Supplementary file 1, genes in Supplementary file 2). The same trend was observed for transcripts in the DD quadrants (Figure 3G, all stats in Supplementary file 1, genes in Supplementary file 2). We next identified genes whose differential expression in pair bonded voles eroded as a function of separation time. We reasoned that such genes are involved in biological pathways that likely regulate adapting to partner loss and prime an animal to form a new pair bond. The genes that were no longer present in the pair bond:long-term separation intersection that were present in the pair bond:short-term separation intersection were deemed eroded genes. These eroded transcripts include 581 UU quadrant transcripts and 1,248 DD quadrant transcripts (Figure 3F and G). We next filtered the genome-wide expression data of Figure 3F by the eroded gene lists (Figure 3H,I). The eroded UU genes are strongly upregulated in a stable pair bond and show dramatic downregulation following long-term separation (Figure 3H). Thus, the eroded UU genes and their associated biological processes may represent key aspects of adapting to partner loss. There was also a corresponding reduction in DD gene overlap between a stable pair bond and following separation, further supporting that bond-related changes in gene expression erode after long-term separation regardless the initial direction of expression during bonding (Figure 3I).

We next determined how separation-induced transcriptional changes affected pair bond associated biological processes. We used the pair bond associated GO terms from Figure 2H as a reference point to determine if these terms were still significantly associated with the underlying transcriptional profile following partner separation (Figure 3J and K). We predicated two outcomes. First, separation would reduce the significance of the pair bond associated GO terms and second, pair bond genes whose expression eroded would be significantly represented by the original pair bond associated terms. The transcripts in the short-term separation UU quadrant were still significantly associated, though to a lesser degree, with the pair bond associated GO terms. After long-term separation, the underlying transcriptional signature is much less significantly associated with the pair bond GO terms. Further, the eroded transcripts are highly associated with the pair bond associated GO terms, including glial-associated terms. A potential role for glia was also implicated in the IPA identification of pathways for glioblastoma multiforme signaling and glioma signaling (Figure 2—figure supplement 2C). Upstream regulators inducing the erosion of these glial-associated genes include Sox2, estrogen receptor, Erbb3, and Ctnnb1 (Figure 3J, Figure 2—figure supplement 2C). We observe a similar trend in the DD quadrants where the pair bond associated GO terms become less representative of the underlying transcriptional signature following long-term separation but are associated with the eroded genes, although to a lesser degree (Figure 3K). IPA pathways activated in a stable pair bond (endocannabinoid signaling, synaptogenesis, and dopamine signaling) are significantly represented among the eroded DD genes (Figure 2—figure supplement 2C). Upstream regulators for the eroded DD transcripts include Hnf4a, Calmodulin, Hdac4, and Adora2a (Figure 3K, Figure 2—figure supplement 2C).

Finally, to gain more nuanced insight into the temporal trajectory of erosion, we also used RRHO to compare short-term to long-term partner separation (Figure 3—figure supplement 5). In contrast to the pattern of consistently upregulated genes between the pair bond and short-term separation, we observed high concordance of downregulated genes between separation timepoints (Figure 3—figure supplement 5A). We then identified transcripts in the UU and DD quadrants of the separation RRHO that exhibited the largest inversion of expression between pair bond and short-term separation (details in Materials and Methods). We reasoned that such transcripts represented potential gene sets that are rapidly eroded following partner separation and which remain eroded after long-term separation, representing a potential first wave transcriptional response to partner loss. We further asked which biological processes are represented by these consistently inverted genes via GO analysis (Figure 3—figure supplement 5B, C). We found that UU pair bond genes that are downregulated following pair bond separation are associated with dopaminergic processes and DD pair bond genes that are upregulated are associated with neuronal rearrangement processes. Thus, this RRHO analysis enriches our interpretation of transcriptional erosion of pair bond signatures following partner separation by revealing temporally complex shifts in gene expression resulting from partner loss.

Neuronal-specific interrogation reveals gene clusters associated with pair bond disruption and loss adaptation

Our tissue-level analysis of gene expression strongly implicated a role for glia in bond formation and loss, and we reasoned that relevant neuronal transcriptional changes may have been masked by these prominent glial transcriptional signals. Thus, we specifically examined separation-induced neuronal transcriptional changes by pioneering translating ribosome affinity purification in voles (vTRAP) (Heiman et al., 2014; Heiman et al., 2008). The vTRAP construct, DIO-eGFP-pvRPL10a, is a double-floxed inverse orientation eGFP-tagged ribosomal subunit that uses the prairie vole RPL10a gene sequence (Figure 4A; Figure 4—figure supplement 1A). The vTRAP construct is paired with Cre-recombinase (Cre) for inducible, cell-type specific expression. AAV-mediated delivery of hSyn-Cre, which drives neuron-specific Cre-recombinase expression, and hSyn-DIO-vTRAP vectors results in neuronal expression of GFP-tagged ribosomes. Subsequent immunoprecipitation of the tagged ribosomes provides a means to isolate neuron-specific, actively-translating mRNAs (Heiman et al., 2014; Heiman et al., 2008). vTRAP is Cre-dependent as confirmed by bilateral NAc injections of +/-Cre (Figure 4—figure supplement 1A). We performed vTRAP immunoprecipitation of 3 cohorts: (1) opposite-sex long-term separated, (2) opposite-sex long-term remain paired, and (3) same-sex long-term separated.

Figure 4 with 2 supplements see all

Download asset Open asset

We defined a set of neuronally enriched genes by using the intersection of threshold and threshold-free based analyses. We first identified neuronally-enriched transcripts by comparing the transcriptional profiles of the input fraction (equivalent to bulk RNA-seq) and pulldown fraction of the animals from the TRAP groups specified above. We predicted that the pulldown fraction would be enriched for neuronal markers and depleted in glial-associated transcripts. Differential expression analysis using the previously defined thresholds revealed that most transcripts are depleted in the pulldown fraction as expected (Figure 4B). GO terms support that hSyn-vTRAP successfully isolates neuronally-enriched transcripts and filters out genes associated with gliogenesis (Figure 4—figure supplement 2C). Second, we used weighted correlation network analysis (WGCNA) to identify gene modules significantly correlated with the input (negative correlation) or pulldown (positive correlation) fractions (Figure 4C, Figure 4—figure supplement 2B; Langfelder and Horvath, 2008). Of note, two of the most significantly positively correlated modules, blue and grey60, contain multiple neuronal genes of interest such as Drd1a, Drd2, Oprk1, Fosb, and Pdyn, which have previously been implicated in pair bonding and reward learning (Resendez et al., 2016; Resendez et al., 2012; Pitchers et al., 2010). Enrichr analysis for cell type also indicated that our identified neuronal enriched genes are predominantly associated with the nucleus accumbens and dorsal striatum (Figure 4D; Chen et al., 2013; Kuleshov et al., 2016; Xie et al., 2021). Finally, we validated that the gene list at the intersection of the DEGs and the WGCNA modules was predominately expressed in neurons and not glia using a publically available single nuclei NAc gene expression dataset generated from adult male rats (Savell et al., 2020). We filtered the dataset by our neuronal enriched gene list and plotted those gene’s expression in the rat nucleus accumbens, which confirmed that our genes are predominately expressed in neuronal cell types (Figure 4E).

We next queried potential neuronal transcriptional changes in our tissue-level RNAseq data. To ensure that the tissue-level expression is representative of neuron-specific expression we compared the expression patterns of the neuronal enriched genes in the input and pulldown fractions of opposite- vs same-sex long-term separated males (Figure 4—figure supplement 2D). The strong similarity in expression validates that bulk sequencing expression can be used to infer the neuronal expression of those same genes (Figure 4—figure supplement 2D). As such, we filtered the tissue-level RNAseq dataset from Figure 3D by the neuronal enriched gene list of Figure 4D to track neuronal gene expression throughout partner loss. We then performed unsupervised clustering of all genes (Figure 4F, top) and neuronal enriched genes (Figure 4F, bottom). We identified three neuronal-enriched gene clusters that had particularly interesting expression patterns. Cluster 1 consists of genes that were upregulated in opposite-sex paired males during pair bonding and short-term separation and downregulated following long-term separation. GO terms associated with these genes include adult walking behavior, a term that contains the autism-associated gene Shank3 (Figure 4G; Gong et al., 2015; Durand et al., 2007). Cluster 2 transcripts are downregulated in pair bonded males and then robustly upregulated after short-term partner loss and, to a lesser extent, at the long-term timepoint, indicating an acute transcriptional response to pair bond disruption. The GO terms associated with these genes suggest an involvement of dopaminergic systems (Drd1a and Drd2) and learning processes potentially engaged to adapt to loss (Figure 4G). Finally, Cluster 3 contains genes that were downregulated in pair bonded males and became upregulated only after long-term opposite-sex partner separation, representing transcripts that may help prime the vole to form a new bond. The GO terms associated with this cluster are primarily implicated in cellular metabolism (Figure 4G). IPA revealed that neuronal clusters 2 and 3 were also enriched for estrogen receptor signaling and oxidative phosphorylation pathways with cluster 2 also implicating opioid signaling pathways (Figure 2—figure supplement 2D). Steroid and opioid signaling have been extensively implicated in pair bonding (Resendez et al., 2016; Resendez et al., 2012; Loth and Donaldson, 2021; Inoue et al., 2013; Stetzik et al., 2018; Carter and Perkeybile, 2018; Lei et al., 2010; Cushing et al., 2008).

Our data provide the first comprehensive assessment of social behavior and accumbal transcription in pair bonded male prairie voles before and after partner loss, providing novel insight into the biological processes that may enable loss adaptation. Prairie voles, unlike laboratory mice and rats, form exclusive lifelong pair bonds that are distinct from affiliative same-sex relationships, making them ideal for studying the unique neurobiology of bonding and loss (Getz et al., 1981; Carter et al., 1995; Pohl et al., 2019; Osako et al., 2018; Lee et al., 2019; Goodwin et al., 2019; Lee and Beery, 2021). Additionally, the direct comparison of opposite- versus same-sex paired animals provides an important control for the effects of social context – whether living with another animal or alone (Arzate-Mejía et al., 2020; Wallace et al., 2009). By leveraging this comparison across our experiments, we demonstrated that pair bonding leads to consistent and persistent shifts in NAc transcription, which erode as a function of long-term partner separation. This suggests that substantial changes in accumbal gene expression may contribute to priming the vole to form a new bond but are not sufficient to erase an existing bond as measured by partner preference. In addition, we demonstrate the utility of vTRAP for interrogating neuron-specific responses to bonding and separation that are obscured when analyzing bulk sequencing results. Together, these results expand the utility of monogamous prairie voles for future behavioral and molecular-genetic studies of partner loss and loss adaptation.

All sequencing data is available on GEO (GSE192661). All lab specific code is available on the Donaldson Lab GitHub (https://github.com/donaldsonlab/separationRNAseq; copy archived at Sadino, 2023) and all metadata including comprehensive statistical analyses, animal information, differential gene expression, gene sets, and IPA analyses are available on the Donaldson lab Figshare. The vole optimized DIO-eGFP-pvRPL10a vector will be made available on Addgene.

1. 1. Sadino JM

   (2022) figshare

   Supplemental Data 1.

   https://doi.org/10.6084/m9.figshare.19911541
2. 1. Sadino JM

   (2022) figshare

   Supplemental Data 2.

   https://doi.org/10.6084/m9.figshare.19911577
3. 1. Sadino JM

   (2022) figshare

   Supplemental Data 3.

   https://doi.org/10.6084/m9.figshare.19911583
4. 1. Sadino JM

   (2023) NCBI Gene Expression Omnibus

   ID GSE192661. RNAseq data from: Prolonged partner separation erodes nucleus accumbens transcriptional signatures of pair bonding in male prairie voles.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE192661

1. 1. Ahern TH
   2. Modi ME
   3. Burkett JP
   4. Young LJ

   (2009) Evaluation of two automated metrics for analyzing partner preference tests

   Journal of Neuroscience Methods 182:180–188.

   https://doi.org/10.1016/j.jneumeth.2009.06.010

   • PubMed
   • Google Scholar
2. Software

   1. Allaire JJ
   2. Gandrud C
   3. Russell K
   4. Yetman CJ

   (2017) NetworkD3: D3 javascript network graphs from R, version 0.4

   CRAN.

   https://CRAN.R-project.org/package=networkD3
3. 1. Amadei EA
   2. Johnson ZV
   3. Jun Kwon Y
   4. Shpiner AC
   5. Saravanan V
   6. Mays WD
   7. Ryan SJ
   8. Walum H
   9. Rainnie DG
   10. Young LJ
   11. Liu RC

   (2017) Dynamic corticostriatal activity biases social bonding in monogamous female prairie voles

   Nature 546:297–301.

   https://doi.org/10.1038/nature22381

   • PubMed
   • Google Scholar
4. 1. Anders S
   2. Pyl PT
   3. Huber W

   (2015) HTSeq -- a python framework to work with high-throughput sequencing data

   Bioinformatics 31:166–169.

   https://doi.org/10.1093/bioinformatics/btu638

   • PubMed
   • Google Scholar
5. 1. Aragona BJ
   2. Liu Y
   3. Curtis JT
   4. Stephan FK
   5. Wang Z

   (2003) A critical role for nucleus accumbens dopamine in partner-preference formation in male prairie voles

   The Journal of Neuroscience 23:3483–3490.

   https://doi.org/10.1523/JNEUROSCI.23-08-03483.2003

   • PubMed
   • Google Scholar
6. 1. Aragona BJ
   2. Wang Z

   (2004) The prairie vole (Microtus ochrogaster): an animal model for behavioral neuroendocrine research on pair bonding

   ILAR Journal 45:35–45.

   https://doi.org/10.1093/ilar.45.1.35

   • PubMed
   • Google Scholar
7. 1. Aragona BJ
   2. Liu Y
   3. Yu YJ
   4. Curtis JT
   5. Detwiler JM
   6. Insel TR
   7. Wang Z

   (2006) Nucleus accumbens dopamine differentially mediates the formation and maintenance of monogamous pair bonds

   Nature Neuroscience 9:133–139.

   https://doi.org/10.1038/nn1613

   • PubMed
   • Google Scholar
8. 1. Arzate-Mejía RG
   2. Lottenbach Z
   3. Schindler V
   4. Jawaid A
   5. Mansuy IM

   (2020) Long-Term impact of social isolation and molecular underpinnings

   Frontiers in Genetics 11:589621.

   https://doi.org/10.3389/fgene.2020.589621

   • PubMed
   • Google Scholar
9. 1. Auguste YSS
   2. Ferro A
   3. Kahng JA
   4. Xavier AM
   5. Dixon JR
   6. Vrudhula U
   7. Nichitiu AS
   8. Rosado D
   9. Wee TL
   10. Pedmale UV
   11. Cheadle L

   (2022) Oligodendrocyte precursor cells engulf synapses during circuit remodeling in mice

   Nature Neuroscience 25:1273–1278.

   https://doi.org/10.1038/s41593-022-01170-x

   • PubMed
   • Google Scholar
10. 1. Bacmeister CM
    2. Huang R
    3. Osso LA
    4. Thornton MA
    5. Conant L
    6. Chavez AR
    7. Poleg-Polsky A
    8. Hughes EG

    (2022) Motor learning drives dynamic patterns of intermittent myelination on learning-activated axons

    Nature Neuroscience 25:1300–1313.

    https://doi.org/10.1038/s41593-022-01169-4

    • PubMed
    • Google Scholar
11. 1. Beery AK
    2. Lopez SA
    3. Blandino KL
    4. Lee NS
    5. Bourdon NS

    (2021) Social selectivity and social motivation in voles

    eLife 10:e72684.

    https://doi.org/10.7554/eLife.72684

    • PubMed
    • Google Scholar
12. 1. Berendzen KM
    2. Manoli DS

    (2022) Rethinking the architecture of attachment: new insights into the role for oxytocin signaling

    Affective Science 3:734–748.

    https://doi.org/10.1007/s42761-022-00142-5

    • PubMed
    • Google Scholar
13. Software

    1. Blighe K

    (2021) Kevinblighe/enhancedvolcano, version 7abca28

    Github.

    https://github.com/kevinblighe/EnhancedVolcano
14. 1. Blocker TD
    2. Ophir AG

    (2016) A preference to bond? male prairie voles form pair bonds even in the presence of multiple receptive females

    Animal Behaviour 122:89–97.

    https://doi.org/10.1016/j.anbehav.2016.10.007

    • PubMed
    • Google Scholar
15. 1. Bolger AM
    2. Lohse M
    3. Usadel B

    (2014) Trimmomatic: a flexible trimmer for illumina sequence data

    Bioinformatics 30:2114–2120.

    https://doi.org/10.1093/bioinformatics/btu170

    • PubMed
    • Google Scholar
16. 1. Bonnefil V
    2. Dietz K
    3. Amatruda M
    4. Wentling M
    5. Aubry AV
    6. Dupree JL
    7. Temple G
    8. Park HJ
    9. Burghardt NS
    10. Casaccia P
    11. Liu J

    (2019) Region-Specific myelin differences define behavioral consequences of chronic social defeat stress in mice

    eLife 8:e40855.

    https://doi.org/10.7554/eLife.40855

    • PubMed
    • Google Scholar
17. 1. Bosch OJ
    2. Nair HP
    3. Ahern TH
    4. Neumann ID
    5. Young LJ

    (2009) The CRF system mediates increased passive stress-coping behavior following the loss of a bonded partner in a monogamous rodent

    Neuropsychopharmacology 34:1406–1415.

    https://doi.org/10.1038/npp.2008.154

    • PubMed
    • Google Scholar
18. 1. Bosch OJ
    2. Dabrowska J
    3. Modi ME
    4. Johnson ZV
    5. Keebaugh AC
    6. Barrett CE
    7. Ahern TH
    8. Guo J
    9. Grinevich V
    10. Rainnie DG
    11. Neumann ID
    12. Young LJ

    (2016) Oxytocin in the nucleus accumbens shell reverses CRFR2-evoked passive stress-coping after partner loss in monogamous male prairie voles

    Psychoneuroendocrinology 64:66–78.

    https://doi.org/10.1016/j.psyneuen.2015.11.011

    • PubMed
    • Google Scholar
19. Preprint

    1. Brusman LE
    2. Protter DSW
    3. Fultz AC
    4. Paulson MU
    5. Chapel GD
    6. Elges IO
    7. Cameron RT
    8. Beery AK
    9. Donaldson ZR

    (2021) Emergent Intra-Pair Sex Differences and Behavioral Coordination in Pair Bonded Prairie Voles

    bioRxiv.

    https://doi.org/10.1101/2021.09.03.458892

    • Google Scholar
20. 1. Burkett JP
    2. Andari E
    3. Johnson ZV
    4. Curry DC
    5. de Waal FBM
    6. Young LJ

    (2016) Oxytocin-Dependent consolation behavior in rodents

    Science 351:375–378.

    https://doi.org/10.1126/science.aac4785

    • PubMed
    • Google Scholar
21. 1. Cahill KM
    2. Huo Z
    3. Tseng GC
    4. Logan RW
    5. Seney ML

    (2018) Improved identification of concordant and discordant gene expression signatures using an updated rank-rank hypergeometric overlap approach

    Scientific Reports 8:9588.

    https://doi.org/10.1038/s41598-018-27903-2

    • PubMed
    • Google Scholar
22. 1. Cardoso SD
    2. Teles MC
    3. Oliveira RF

    (2015) Neurogenomic mechanisms of social plasticity

    The Journal of Experimental Biology 218:140–149.

    https://doi.org/10.1242/jeb.106997

    • PubMed
    • Google Scholar
23. 1. Carter CS
    2. DeVries AC
    3. Getz LL

    (1995) Physiological substrates of mammalian monogamy: the prairie vole model

    Neuroscience and Biobehavioral Reviews 19:303–314.

    https://doi.org/10.1016/0149-7634(94)00070-h

    • PubMed
    • Google Scholar
24. 1. Carter CS
    2. Perkeybile AM

    (2018) The monogamy paradox: what do love and sex have to do with it?

    Frontiers in Ecology and Evolution 6:202.

    https://doi.org/10.3389/fevo.2018.00202

    • PubMed
    • Google Scholar
25. 1. Chen EY
    2. Tan CM
    3. Kou Y
    4. Duan Q
    5. Wang Z
    6. Meirelles GV
    7. Clark NR
    8. Ma’ayan A

    (2013) Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool

    BMC Bioinformatics 14:128.

    https://doi.org/10.1186/1471-2105-14-128

    • PubMed
    • Google Scholar
26. 1. Curtis JT

    (2010) Does fertility Trump monogamy?

    Animal Behaviour 80:319–328.

    https://doi.org/10.1016/j.anbehav.2010.05.014

    • PubMed
    • Google Scholar
27. 1. Cushing BS
    2. Perry A
    3. Musatov S
    4. Ogawa S
    5. Papademetriou E

    (2008) Estrogen receptors in the medial amygdala inhibit the expression of male prosocial behavior

    The Journal of Neuroscience 28:10399–10403.

    https://doi.org/10.1523/JNEUROSCI.1928-08.2008

    • PubMed
    • Google Scholar
28. 1. Danecek P
    2. Bonfield JK
    3. Liddle J
    4. Marshall J
    5. Ohan V
    6. Pollard MO
    7. Whitwham A
    8. Keane T
    9. McCarthy SA
    10. Davies RM
    11. Li H

    (2021) Twelve years of samtools and bcftools

    GigaScience 10:giab008.

    https://doi.org/10.1093/gigascience/giab008

    • PubMed
    • Google Scholar
29. 1. DeVries AC
    2. Johnson CL
    3. Carter CS

    (1997) Familiarity and gender influence social preferences in prairie voles ( microtus ochrogaster )

    Canadian Journal of Zoology 75:295–301.

    https://doi.org/10.1139/z97-037

    • Google Scholar
30. 1. Duclot F
    2. Sailer L
    3. Koutakis P
    4. Wang Z
    5. Kabbaj M

    (2020)

    Transcriptomic regulations underlying pair-bond formation and maintenance in the socially monogamous male and female prairie vole

    Biological Psychiatry 1:1–11.

    • Google Scholar
31. 1. Duncan GJ
    2. Emery B

    (2023) End of the road: astrocyte endfeet regulate OPC migration and myelination

    Neuron 111:139–141.

    https://doi.org/10.1016/j.neuron.2022.12.023

    • PubMed
    • Google Scholar
32. 1. Durand CM
    2. Betancur C
    3. Boeckers TM
    4. Bockmann J
    5. Chaste P
    6. Fauchereau F
    7. Nygren G
    8. Rastam M
    9. Gillberg IC
    10. Anckarsäter H
    11. Sponheim E
    12. Goubran-Botros H
    13. Delorme R
    14. Chabane N
    15. Mouren-Simeoni M-C
    16. de Mas P
    17. Bieth E
    18. Rogé B
    19. Héron D
    20. Burglen L
    21. Gillberg C
    22. Leboyer M
    23. Bourgeron T

    (2007) Mutations in the gene encoding the synaptic scaffolding protein SHANK3 are associated with autism spectrum disorders

    Nature Genetics 39:25–27.

    https://doi.org/10.1038/ng1933

    • PubMed
    • Google Scholar
33. 1. Getz LL
    2. Carter CS
    3. Gavish L

    (1981) The mating system of the prairie vole, Microtus ochrogaster: field and laboratory evidence for pair-bonding

    Behavioral Ecology and Sociobiology 8:189–194.

    https://doi.org/10.1007/BF00299829

    • Google Scholar
34. 1. Gobrogge KL
    2. Liu Y
    3. Young LJ
    4. Wang Z

    (2009) Anterior hypothalamic vasopressin regulates pair-bonding and drug-induced aggression in a monogamous rodent

    PNAS 106:19144–19149.

    https://doi.org/10.1073/pnas.0908620106

    • PubMed
    • Google Scholar
35. 1. Gong P
    2. Liu J
    3. Blue PR
    4. Li S
    5. Zhou X

    (2015) Serotonin receptor gene (HTR2A) T102C polymorphism modulates individuals’ perspective taking ability and autistic-like traits

    Frontiers in Human Neuroscience 9:575.

    https://doi.org/10.3389/fnhum.2015.00575

    • PubMed
    • Google Scholar
36. 1. Goodwin NL
    2. Lopez SA
    3. Lee NS
    4. Beery AK

    (2019) Comparative role of reward in long-term peer and mate relationships in voles

    Hormones and Behavior 111:70–77.

    https://doi.org/10.1016/j.yhbeh.2018.10.012

    • Google Scholar
37. 1. Grippo AJ
    2. Cushing BS
    3. Carter CS

    (2007) Depression-Like behavior and stressor-induced neuroendocrine activation in female prairie voles exposed to chronic social isolation

    Psychosomatic Medicine 69:149–157.

    https://doi.org/10.1097/PSY.0b013e31802f054b

    • PubMed
    • Google Scholar
38. 1. Harbert KJ
    2. Pellegrini M
    3. Gordon KM
    4. Donaldson ZR

    (2020) How prior pair-bonding experience affects future bonding behavior in monogamous prairie voles

    Hormones and Behavior 126:104847.

    https://doi.org/10.1016/j.yhbeh.2020.104847

    • PubMed
    • Google Scholar
39. 1. Heiman M
    2. Schaefer A
    3. Gong S
    4. Peterson JD
    5. Day M
    6. Ramsey KE
    7. Suárez-Fariñas M
    8. Schwarz C
    9. Stephan DA
    10. Surmeier DJ
    11. Greengard P
    12. Heintz N

    (2008) A translational profiling approach for the molecular characterization of CNS cell types

    Cell 135:738–748.

    https://doi.org/10.1016/j.cell.2008.10.028

    • PubMed
    • Google Scholar
40. 1. Heiman M
    2. Kulicke R
    3. Fenster RJ
    4. Greengard P
    5. Heintz N

    (2014) Cell type-specific mRNA purification by translating ribosome affinity purification (TRAP)

    Nature Protocols 9:1282–1291.

    https://doi.org/10.1038/nprot.2014.085

    • PubMed
    • Google Scholar
41. 1. Holmes TH
    2. Rahe RH

    (1967) The social readjustment rating scale

    Journal of Psychosomatic Research 11:213–218.

    https://doi.org/10.1016/0022-3999(67)90010-4

    • PubMed
    • Google Scholar
42. 1. Hughes EG
    2. Orthmann-Murphy JL
    3. Langseth AJ
    4. Bergles DE

    (2018) Myelin remodeling through experience-dependent oligodendrogenesis in the adult somatosensory cortex

    Nature Neuroscience 21:696–706.

    https://doi.org/10.1038/s41593-018-0121-5

    • PubMed
    • Google Scholar
43. 1. Inoue K
    2. Burkett JP
    3. Young LJ

    (2013) Neuroanatomical distribution of μ-opioid receptor mrna and binding in monogamous prairie voles (microtus ochrogaster) and non-monogamous meadow voles (microtus pennsylvanicus)

    Neuroscience 244:122–133.

    https://doi.org/10.1016/j.neuroscience.2013.03.035

    • PubMed
    • Google Scholar
44. 1. Insel TR
    2. Young LJ

    (2001) The neurobiology of attachment

    Nature Reviews. Neuroscience 2:129–136.

    https://doi.org/10.1038/35053579

    • PubMed
    • Google Scholar
45. 1. Johnson ZV
    2. Walum H
    3. Jamal YA
    4. Xiao Y
    5. Keebaugh AC
    6. Inoue K
    7. Young LJ

    (2016) Central oxytocin receptors mediate mating-induced partner preferences and enhance correlated activation across forebrain nuclei in male prairie voles

    Hormones and Behavior 79:8–17.

    https://doi.org/10.1016/j.yhbeh.2015.11.011

    • PubMed
    • Google Scholar
46. 1. Johnson ZV
    2. Young LJ

    (2017) Oxytocin and vasopressin neural networks: implications for social behavioral diversity and translational neuroscience

    Neuroscience and Biobehavioral Reviews 76:87–98.

    https://doi.org/10.1016/j.neubiorev.2017.01.034

    • PubMed
    • Google Scholar
47. 1. Keyes KM
    2. Pratt C
    3. Galea S
    4. McLaughlin KA
    5. Koenen KC
    6. Shear MK

    (2014) The burden of loss: unexpected death of a loved one and psychiatric disorders across the life course in a national study

    The American Journal of Psychiatry 171:864–871.

    https://doi.org/10.1176/appi.ajp.2014.13081132

    • PubMed
    • Google Scholar
48. 1. Kim D
    2. Langmead B
    3. Salzberg SL

    (2015) HISAT: a fast spliced aligner with low memory requirements

    Nature Methods 12:357–360.

    https://doi.org/10.1038/nmeth.3317

    • PubMed
    • Google Scholar
49. Preprint

    1. Kim J
    2. Sung Y
    3. Park H
    4. Choi DI
    5. Kim J
    6. Lee H
    7. Jung MK
    8. Noh S
    9. Ye S
    10. Lee J
    11. Islam MA
    12. Chun H
    13. Mun JY
    14. Kaang BK

    (2023) Astrocytic Connection to Engram Neurons Increased after Learning

    bioRxiv.

    https://doi.org/10.1101/2023.01.25.525617

    • Google Scholar
50. 1. Kogan JH
    2. Frankland PW
    3. Silva AJ

    (2000) Long-Term memory underlying hippocampus-dependent social recognition in mice

    Hippocampus 10:47–56.

    https://doi.org/10.1002/(SICI)1098-1063(2000)10:1<47::AID-HIPO5>3.0.CO;2-6

    • PubMed
    • Google Scholar
51. 1. Kopec AM
    2. Smith CJ
    3. Ayre NR
    4. Sweat SC
    5. Bilbo SD

    (2018) Microglial dopamine receptor elimination defines sex-specific nucleus accumbens development and social behavior in adolescent rats

    Nature Communications 9:3769.

    https://doi.org/10.1038/s41467-018-06118-z

    • PubMed
    • Google Scholar
52. 1. Kreuder AK
    2. Scheele D
    3. Wassermann L
    4. Wollseifer M
    5. Stoffel-Wagner B
    6. Lee MR
    7. Hennig J
    8. Maier W
    9. Hurlemann R

    (2017) How the brain codes intimacy: the neurobiological substrates of romantic touch

    Human Brain Mapping 38:4525–4534.

    https://doi.org/10.1002/hbm.23679

    • PubMed
    • Google Scholar
53. 1. Krishnan V
    2. Han M-H
    3. Graham DL
    4. Berton O
    5. Renthal W
    6. Russo SJ
    7. Laplant Q
    8. Graham A
    9. Lutter M
    10. Lagace DC
    11. Ghose S
    12. Reister R
    13. Tannous P
    14. Green TA
    15. Neve RL
    16. Chakravarty S
    17. Kumar A
    18. Eisch AJ
    19. Self DW
    20. Lee FS
    21. Tamminga CA
    22. Cooper DC
    23. Gershenfeld HK
    24. Nestler EJ

    (2007) Molecular adaptations underlying susceptibility and resistance to social defeat in brain reward regions

    Cell 131:391–404.

    https://doi.org/10.1016/j.cell.2007.09.018

    • PubMed
    • Google Scholar
54. 1. Kronman H
    2. Torres-Berrío A
    3. Sidoli S
    4. Issler O
    5. Godino A
    6. Ramakrishnan A
    7. Mews P
    8. Lardner CK
    9. Parise EM
    10. Walker DM
    11. van der Zee YY
    12. Browne CJ
    13. Boyce BF
    14. Neve R
    15. Garcia BA
    16. Shen L
    17. Peña CJ
    18. Nestler EJ

    (2021) Long-Term behavioral and cell-type-specific molecular effects of early life stress are mediated by h3k79me2 dynamics in medium spiny neurons

    Nature Neuroscience 24:667–676.

    https://doi.org/10.1038/s41593-021-00814-8

    • PubMed
    • Google Scholar
55. 1. Kuleshov MV
    2. Jones MR
    3. Rouillard AD
    4. Fernandez NF
    5. Duan Q
    6. Wang Z
    7. Koplev S
    8. Jenkins SL
    9. Jagodnik KM
    10. Lachmann A
    11. McDermott MG
    12. Monteiro CD
    13. Gundersen GW
    14. Ma’ayan A

    (2016) Enrichr: a comprehensive gene set enrichment analysis web server 2016 update

    Nucleic Acids Research 44:W90–W97.

    https://doi.org/10.1093/nar/gkw377

    • PubMed
    • Google Scholar
56. 1. Labonté B
    2. Engmann O
    3. Purushothaman I
    4. Menard C
    5. Wang J
    6. Tan C
    7. Scarpa JR
    8. Moy G
    9. Loh YHE
    10. Cahill M
    11. Lorsch ZS
    12. Hamilton PJ
    13. Calipari ES
    14. Hodes GE
    15. Issler O
    16. Kronman H
    17. Pfau M
    18. Obradovic ALJ
    19. Dong Y
    20. Neve RL
    21. Russo S
    22. Kazarskis A
    23. Tamminga C
    24. Mechawar N
    25. Turecki G
    26. Zhang B
    27. Shen L
    28. Nestler EJ

    (2017) Sex-Specific transcriptional signatures in human depression

    Nature Medicine 23:1102–1111.

    https://doi.org/10.1038/nm.4386

    • PubMed
    • Google Scholar
57. 1. Langfelder P
    2. Horvath S

    (2008) WGCNA: an R package for weighted correlation network analysis

    BMC Bioinformatics 9:559.

    https://doi.org/10.1186/1471-2105-9-559

    • PubMed
    • Google Scholar
58. 1. Lee NS
    2. Goodwin NL
    3. Freitas KE
    4. Beery AK

    (2019) Affiliation, aggression, and selectivity of peer relationships in meadow and prairie voles

    Frontiers in Behavioral Neuroscience 13:52.

    https://doi.org/10.3389/fnbeh.2019.00052

    • PubMed
    • Google Scholar
59. 1. Lee NS
    2. Beery AK

    (2021) The role of dopamine signaling in prairie vole peer relationships

    Hormones and Behavior 127:104876.

    https://doi.org/10.1016/j.yhbeh.2020.104876

    • PubMed
    • Google Scholar
60. 1. Lehmann ML
    2. Weigel TK
    3. Elkahloun AG
    4. Herkenham M

    (2017) Chronic social defeat reduces myelination in the mouse medial prefrontal cortex

    Scientific Reports 7:46548.

    https://doi.org/10.1038/srep46548

    • PubMed
    • Google Scholar
61. 1. Lei K
    2. Cushing BS
    3. Musatov S
    4. Ogawa S
    5. Kramer KM

    (2010) Estrogen receptor-alpha in the bed nucleus of the stria terminalis regulates social affiliation in male prairie voles (Microtus ochrogaster)

    PLOS ONE 5:e8931.

    https://doi.org/10.1371/journal.pone.0008931

    • PubMed
    • Google Scholar
62. 1. Lieberwirth C
    2. Wang Z

    (2016) The neurobiology of pair bond formation, bond disruption, and social buffering

    Current Opinion in Neurobiology 40:8–13.

    https://doi.org/10.1016/j.conb.2016.05.006

    • PubMed
    • Google Scholar
63. 1. Lim MM
    2. Young LJ

    (2004) Vasopressin-Dependent neural circuits underlying pair bond formation in the monogamous prairie vole

    Neuroscience 125:35–45.

    https://doi.org/10.1016/j.neuroscience.2003.12.008

    • PubMed
    • Google Scholar
64. 1. Liu J
    2. Dietz K
    3. DeLoyht JM
    4. Pedre X
    5. Kelkar D
    6. Kaur J
    7. Vialou V
    8. Lobo MK
    9. Dietz DM
    10. Nestler EJ
    11. Dupree J
    12. Casaccia P

    (2012) Impaired adult myelination in the prefrontal cortex of socially isolated mice

    Nature Neuroscience 15:1621–1623.

    https://doi.org/10.1038/nn.3263

    • PubMed
    • Google Scholar
65. 1. Loth MK
    2. Donaldson ZR

    (2021) Oxytocin, dopamine, and opioid interactions underlying pair bonding: highlighting a potential role for microglia

    Endocrinology 162:bqaa223.

    https://doi.org/10.1210/endocr/bqaa223

    • PubMed
    • Google Scholar
66. 1. Love MI
    2. Huber W
    3. Anders S

    (2014) Moderated estimation of fold change and dispersion for RNA-seq data with deseq2

    Genome Biology 15:550.

    https://doi.org/10.1186/s13059-014-0550-8

    • PubMed
    • Google Scholar
67. 1. Makinodan M
    2. Rosen KM
    3. Ito S
    4. Corfas G

    (2012) A critical period for social experience-dependent oligodendrocyte maturation and myelination

    Science 337:1357–1360.

    https://doi.org/10.1126/science.1220845

    • PubMed
    • Google Scholar
68. 1. McNeal N
    2. Scotti M-AL
    3. Wardwell J
    4. Chandler DL
    5. Bates SL
    6. Larocca M
    7. Trahanas DM
    8. Grippo AJ

    (2014) Disruption of social bonds induces behavioral and physiological dysregulation in male and female prairie voles

    Autonomic Neuroscience 180:9–16.

    https://doi.org/10.1016/j.autneu.2013.10.001

    • PubMed
    • Google Scholar
69. 1. Modi ME
    2. Young LJ

    (2011) D-Cycloserine facilitates socially reinforced learning in an animal model relevant to autism spectrum disorders

    Biological Psychiatry 70:298–304.

    https://doi.org/10.1016/j.biopsych.2011.01.026

    • PubMed
    • Google Scholar
70. 1. Mukamel EA

    (2022) Multiple comparisons and inappropriate statistical testing lead to spurious sex differences in gene expression

    Biological Psychiatry 91:e1–e2.

    https://doi.org/10.1016/j.biopsych.2021.06.026

    • PubMed
    • Google Scholar
71. 1. Nestler EJ
    2. Peña CJ
    3. Kundakovic M
    4. Mitchell A
    5. Akbarian S

    (2016) Epigenetic basis of mental illness

    The Neuroscientist 22:447–463.

    https://doi.org/10.1177/1073858415608147

    • PubMed
    • Google Scholar
72. 1. O’Connor M-F
    2. Wellisch DK
    3. Stanton AL
    4. Eisenberger NI
    5. Irwin MR
    6. Lieberman MD

    (2008) Craving love? enduring grief activates brain’s reward center

    NeuroImage 42:969–972.

    https://doi.org/10.1016/j.neuroimage.2008.04.256

    • PubMed
    • Google Scholar
73. 1. Okuyama T
    2. Kitamura T
    3. Roy DS
    4. Itohara S
    5. Tonegawa S

    (2016) Ventral CA1 neurons store social memory

    Science 353:1536–1541.

    https://doi.org/10.1126/science.aaf7003

    • Google Scholar
74. 1. Osako Y
    2. Nobuhara R
    3. Arai YCP
    4. Tanaka K
    5. Young LJ
    6. Nishihara M
    7. Mitsui S
    8. Yuri K

    (2018) Partner loss in monogamous rodents: modulation of pain and emotional behavior in male prairie voles

    Psychosomatic Medicine 80:62–68.

    https://doi.org/10.1097/PSY.0000000000000524

    • PubMed
    • Google Scholar
75. 1. Peña CJ
    2. Kronman HG
    3. Walker DM
    4. Cates HM
    5. Bagot RC
    6. Purushothaman I
    7. Issler O
    8. Loh YHE
    9. Leong T
    10. Kiraly DD
    11. Goodman E
    12. Neve RL
    13. Shen L
    14. Nestler EJ

    (2017) Early life stress confers lifelong stress susceptibility in mice via ventral tegmental area Otx2

    Science 356:1185–1188.

    https://doi.org/10.1126/science.aan4491

    • PubMed
    • Google Scholar
76. 1. Pitchers KK
    2. Frohmader KS
    3. Vialou V
    4. Mouzon E
    5. Nestler EJ
    6. Lehman MN
    7. Coolen LM

    (2010) ΔFosB in the nucleus accumbens is critical for reinforcing effects of sexual reward

    Genes, Brain, and Behavior 9:831–840.

    https://doi.org/10.1111/j.1601-183X.2010.00621.x

    • PubMed
    • Google Scholar
77. 1. Plaisier SB
    2. Taschereau R
    3. Wong JA
    4. Graeber TG

    (2010) Rank–rank hypergeometric overlap: identification of statistically significant overlap between gene-expression signatures

    Nucleic Acids Research 38:e169.

    https://doi.org/10.1093/nar/gkq636

    • PubMed
    • Google Scholar
78. 1. Pohl TT
    2. Young LJ
    3. Bosch OJ

    (2019) Lost connections: oxytocin and the neural, physiological, and behavioral consequences of disrupted relationships

    International Journal of Psychophysiology 136:54–63.

    https://doi.org/10.1016/j.ijpsycho.2017.12.011

    • PubMed
    • Google Scholar
79. 1. Pohl TT
    2. Jung O
    3. Di Benedetto B
    4. Young LJ
    5. Bosch OJ

    (2021) Microglia react to partner loss in a sex- and brain site-specific manner in prairie voles

    Brain, Behavior, and Immunity 96:168–186.

    https://doi.org/10.1016/j.bbi.2021.05.026

    • PubMed
    • Google Scholar
80. 1. Potretzke S
    2. Ryabinin AE

    (2019) The prairie vole model of pair-bonding and its sensitivity to addictive substances

    Frontiers in Psychology 10:2477.

    https://doi.org/10.3389/fpsyg.2019.02477

    • PubMed
    • Google Scholar
81. Software

    1. R:The R project for statistical computing

    (2022) The R project for statistical computing

    R Project.

    https://www.r-project.org/
82. 1. Resendez SL
    2. Kuhnmuench M
    3. Krzywosinski T
    4. Aragona BJ

    (2012) Κ-Opioid receptors within the nucleus accumbens shell mediate pair bond maintenance

    The Journal of Neuroscience 32:6771–6784.

    https://doi.org/10.1523/JNEUROSCI.5779-11.2012

    • PubMed
    • Google Scholar
83. 1. Resendez SL
    2. Keyes PC
    3. Day JJ
    4. Hambro C
    5. Austin CJ
    6. Maina FK
    7. Eidson LN
    8. Porter-Stransky KA
    9. Nevárez N
    10. McLean JW
    11. Kuhnmuench MA
    12. Murphy AZ
    13. Mathews TA
    14. Aragona BJ

    (2016) Dopamine and opioid systems interact within the nucleus accumbens to maintain monogamous pair bonds

    eLife 5:e15325.

    https://doi.org/10.7554/eLife.15325

    • PubMed
    • Google Scholar
84. Software

    1. Sadino J

    (2023) SeparationRNAseq, version swh:1:rev:45c0386e0655cc52f89e575668bf3f00ed907d7f

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:6aa42dbc34e9f11413e2eac6a5d26885b39ee0d0;origin=https://github.com/donaldsonlab/separationRNAseq;visit=swh:1:snp:b1f68d9e450e91713f4c797b3dda12c89d68ca23;anchor=swh:1:rev:45c0386e0655cc52f89e575668bf3f00ed907d7f
85. 1. Savell KE
    2. Tuscher JJ
    3. Zipperly ME
    4. Duke CG
    5. Phillips RA
    6. Bauman AJ
    7. Thukral S
    8. Sultan FA
    9. Goska NA
    10. Ianov L
    11. Day JJ

    (2020) A dopamine-induced gene expression signature regulates neuronal function and cocaine response

    Science Advances 6:eaba4221.

    https://doi.org/10.1126/sciadv.aba4221

    • PubMed
    • Google Scholar
86. 1. Scribner JL
    2. Vance EA
    3. Protter DSW
    4. Sheeran WM
    5. Saslow E
    6. Cameron RT
    7. Klein EM
    8. Jimenez JC
    9. Kheirbek MA
    10. Donaldson ZR

    (2020) A neuronal signature for monogamous reunion

    PNAS 117:11076–11084.

    https://doi.org/10.1073/pnas.1917287117

    • PubMed
    • Google Scholar
87. 1. Seney ML
    2. Huo Z
    3. Cahill K
    4. French L
    5. Puralewski R
    6. Zhang J
    7. Logan RW
    8. Tseng G
    9. Lewis DA
    10. Sibille E

    (2018) Opposite molecular signatures of depression in men and women

    Biological Psychiatry 84:18–27.

    https://doi.org/10.1016/j.biopsych.2018.01.017

    • PubMed
    • Google Scholar
88. 1. Shear K
    2. Shair H

    (2005) Attachment, loss, and complicated grief

    Developmental Psychobiology 47:253–267.

    https://doi.org/10.1002/dev.20091

    • PubMed
    • Google Scholar
89. 1. Shear MK
    2. Simon N
    3. Wall M
    4. Zisook S
    5. Neimeyer R
    6. Duan N
    7. Reynolds C
    8. Lebowitz B
    9. Sung S
    10. Ghesquiere A
    11. Gorscak B
    12. Clayton P
    13. Ito M
    14. Nakajima S
    15. Konishi T
    16. Melhem N
    17. Meert K
    18. Schiff M
    19. O’Connor M-F
    20. First M
    21. Sareen J
    22. Bolton J
    23. Skritskaya N
    24. Mancini AD
    25. Keshaviah A

    (2011) Complicated grief and related bereavement issues for DSM-5

    Depression and Anxiety 28:103–117.

    https://doi.org/10.1002/da.20780

    • PubMed
    • Google Scholar
90. 1. Smith AS
    2. Lieberwirth C
    3. Wang Z

    (2013) Behavioral and physiological responses of female prairie voles (microtus ochrogaster) to various stressful conditions

    Stress 16:531–539.

    https://doi.org/10.3109/10253890.2013.794449

    • PubMed
    • Google Scholar
91. 1. Smith AS
    2. Wang Z

    (2014) Hypothalamic oxytocin mediates social buffering of the stress response

    Biological Psychiatry 76:281–288.

    https://doi.org/10.1016/j.biopsych.2013.09.017

    • PubMed
    • Google Scholar
92. 1. Smith AS
    2. Williams Avram SK
    3. Cymerblit-Sabba A
    4. Song J
    5. Young WS

    (2016) Targeted activation of the hippocampal Ca2 area strongly enhances social memory

    Molecular Psychiatry 21:1137–1144.

    https://doi.org/10.1038/mp.2015.189

    • PubMed
    • Google Scholar
93. 1. Stetzik L
    2. Ganshevsky D
    3. Lende MN
    4. Roache LE
    5. Musatov S
    6. Cushing BS

    (2018) Inhibiting ERα expression in the medial amygdala increases prosocial behavior in male meadow voles (microtus pennsylvanicus)

    Behavioural Brain Research 351:42–48.

    https://doi.org/10.1016/j.bbr.2018.05.035

    • PubMed
    • Google Scholar
94. 1. Sun P
    2. Smith AS
    3. Lei K
    4. Liu Y
    5. Wang Z

    (2014) Breaking bonds in male prairie vole: long-term effects on emotional and social behavior, physiology, and neurochemistry

    Behavioural Brain Research 265:22–31.

    https://doi.org/10.1016/j.bbr.2014.02.016

    • PubMed
    • Google Scholar
95. Book

    1. Tamarin RH
    2. Ostfeld RS
    3. Pugh SR
    4. Bujalska G

    (1990) Social Systems and Population Cycles in Voles

    Springer.

    https://doi.org/10.1007/978-3-0348-6416-9

    • Google Scholar
96. 1. Thornton MA
    2. Hughes EG

    (2020) Neuron-oligodendroglia interactions: activity-dependent regulation of cellular signaling

    Neuroscience Letters 727:134916.

    https://doi.org/10.1016/j.neulet.2020.134916

    • PubMed
    • Google Scholar
97. 1. Tripp JA
    2. Feng NY
    3. Bass AH

    (2018) Behavioural tactic predicts preoptic-hypothalamic gene expression more strongly than developmental morph in fish with alternative reproductive tactics

    Proceedings. Biological Sciences 285:20172742.

    https://doi.org/10.1098/rspb.2017.2742

    • PubMed
    • Google Scholar
98. 1. Tripp JA
    2. Berrio A
    3. McGraw LA
    4. Matz MV
    5. Davis JK
    6. Inoue K
    7. Thomas JW
    8. Young LJ
    9. Phelps SM

    (2021) Comparative neurotranscriptomics reveal widespread species differences associated with bonding

    BMC Genomics 22:399.

    https://doi.org/10.1186/s12864-021-07720-0

    • PubMed
    • Google Scholar
99. 1. Tuesta LM
    2. Zhang Y

    (2014) Mechanisms of epigenetic memory and addiction

    The EMBO Journal 33:1091–1103.

    https://doi.org/10.1002/embj.201488106

    • PubMed
    • Google Scholar
100. 1. Ülgen DH
     2. Ruigrok SR
     3. Sandi C

     (2023) Powering the social brain: mitochondria in social behaviour

     Current Opinion in Neurobiology 79:102675.

     https://doi.org/10.1016/j.conb.2022.102675

     • PubMed
     • Google Scholar
101. 1. Vahaba DM
     2. Halstead ER
     3. Donaldson ZR
     4. Ahern TH
     5. Beery AK

     (2022) Sex differences in the reward value of familiar mates in prairie voles

     Genes, Brain, and Behavior 21:e12790.

     https://doi.org/10.1111/gbb.12790

     • PubMed
     • Google Scholar
102. 1. Walker DM
     2. Cates HM
     3. Loh YHE
     4. Purushothaman I
     5. Ramakrishnan A
     6. Cahill KM
     7. Lardner CK
     8. Godino A
     9. Kronman HG
     10. Rabkin J
     11. Lorsch ZS
     12. Mews P
     13. Doyle MA
     14. Feng J
     15. Labonté B
     16. Koo JW
     17. Bagot RC
     18. Logan RW
     19. Seney ML
     20. Calipari ES
     21. Shen L
     22. Nestler EJ

     (2018) Cocaine self-administration alters transcriptome-wide responses in the brain’s reward circuitry

     Biological Psychiatry 84:867–880.

     https://doi.org/10.1016/j.biopsych.2018.04.009

     • PubMed
     • Google Scholar
103. 1. Walker DM
     2. Zhou X
     3. Cunningham AM
     4. Lipschultz AP
     5. Ramakrishnan A
     6. Cates HM
     7. Bagot RC
     8. Shen L
     9. Zhang B
     10. Nestler EJ

     (2022) Sex-Specific transcriptional changes in response to adolescent social stress in the brain’s reward circuitry

     Biological Psychiatry 91:118–128.

     https://doi.org/10.1016/j.biopsych.2021.02.964

     • PubMed
     • Google Scholar
104. 1. Wallace DL
     2. Han M-H
     3. Graham DL
     4. Green TA
     5. Vialou V
     6. Iñiguez SD
     7. Cao J-L
     8. Kirk A
     9. Chakravarty S
     10. Kumar A
     11. Krishnan V
     12. Neve RL
     13. Cooper DC
     14. Bolaños CA
     15. Barrot M
     16. McClung CA
     17. Nestler EJ

     (2009) Creb regulation of nucleus accumbens excitability mediates social isolation-induced behavioral deficits

     Nature Neuroscience 12:200–209.

     https://doi.org/10.1038/nn.2257

     • PubMed
     • Google Scholar
105. 1. Walum H
     2. Young LJ

     (2018) The neural mechanisms and circuitry of the pair bond

     Nature Reviews. Neuroscience 19:643–654.

     https://doi.org/10.1038/s41583-018-0072-6

     • PubMed
     • Google Scholar
106. 1. Wang M
     2. Zhao Y
     3. Zhang B

     (2015) Efficient test and visualization of multi-set intersections

     Scientific Reports 5:16923.

     https://doi.org/10.1038/srep16923

     • PubMed
     • Google Scholar
107. Book

     1. Wickham H

     (2016) Ggplot2: Elegant Graphics for Data Analysis

     Springer International Publishing.

     https://doi.org/10.1007/978-3-319-24277-4

     • Google Scholar
108. 1. Williams J

     (1992) Development of partner preferences in female prairie voles (Microtus ochrogaster): the role of social and sexual experience

     Hormones and Behavior 26:339–349.

     https://doi.org/10.1016/0018-506X(92)90004-F

     • Google Scholar
109. 1. Williams JR
     2. Carter CS
     3. Insel T

     (1992) Partner preference development in female prairie voles is facilitated by mating or the central infusion of oxytocin

     Annals of the New York Academy of Sciences 652:487–489.

     https://doi.org/10.1111/j.1749-6632.1992.tb34393.x

     • PubMed
     • Google Scholar
110. 1. Winslow JT
     2. Hastings N
     3. Carter CS
     4. Harbaugh CR
     5. Insel TR

     (1993) A role for central vasopressin in pair bonding in monogamous prairie voles

     Nature 365:545–548.

     https://doi.org/10.1038/365545a0

     • PubMed
     • Google Scholar
111. 1. Xie Z
     2. Bailey A
     3. Kuleshov MV
     4. Clarke DJB
     5. Evangelista JE
     6. Jenkins SL
     7. Lachmann A
     8. Wojciechowicz ML
     9. Kropiwnicki E
     10. Jagodnik KM
     11. Jeon M
     12. Ma’ayan A

     (2021) Gene set knowledge discovery with enrichr

     Current Protocols 1:e90.

     https://doi.org/10.1002/cpz1.90

     • PubMed
     • Google Scholar
112. 1. Young LJ
     2. Wang Z

     (2004) The neurobiology of pair bonding

     Nature Neuroscience 7:1048–1054.

     https://doi.org/10.1038/nn1327

     • PubMed
     • Google Scholar
113. 1. Young KA
     2. Liu Y
     3. Wang Z

     (2008) The neurobiology of social attachment: A comparative approach to behavioral, neuroanatomical, and neurochemical studies

     Comparative Biochemistry and Physiology. Toxicology & Pharmacology 148:401–410.

     https://doi.org/10.1016/j.cbpc.2008.02.004

     • PubMed
     • Google Scholar
114. 1. Young RL
     2. Ferkin MH
     3. Ockendon-Powell NF
     4. Orr VN
     5. Phelps SM
     6. Pogány Á
     7. Richards-Zawacki CL
     8. Summers K
     9. Székely T
     10. Trainor BC
     11. Urrutia AO
     12. Zachar G
     13. O’Connell LA
     14. Hofmann HA

     (2019) Conserved transcriptomic profiles underpin monogamy across vertebrates

     PNAS 116:1331–1336.

     https://doi.org/10.1073/pnas.1813775116

     • PubMed
     • Google Scholar
115. 1. Yu G
     2. Wang L-G
     3. Han Y
     4. He Q-Y

     (2012) ClusterProfiler: an R package for comparing biological themes among gene clusters

     OMICS 16:284–287.

     https://doi.org/10.1089/omi.2011.0118

     • PubMed
     • Google Scholar
116. 1. Zayed A
     2. Robinson GE

     (2012) Understanding the relationship between brain gene expression and social behavior: lessons from the honey bee

     Annual Review of Genetics 46:591–615.

     https://doi.org/10.1146/annurev-genet-110711-155517

     • PubMed
     • Google Scholar
117. 1. Zhou B
     2. Zhu Z
     3. Ransom BR
     4. Tong X

     (2021) Oligodendrocyte lineage cells and depression

     Molecular Psychiatry 26:103–117.

     https://doi.org/10.1038/s41380-020-00930-0

     • PubMed
     • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Prolonged partner separation erodes nucleus accumbens transcriptional signatures of pair bonding in male prairie voles" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Margaret M McCarthy as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Catherine Dulac as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The disconnect between the behavioral responses in terms of endurance of the pair-bond, and the transcriptomics was of concern to the Reviewers. It was noted that there are subtle but perhaps important differences in the separation protocol used for the behavioral studies versus the transcriptomics. This combined with the behavioral data here not being consistent with the published literature undermines the current conclusions. The Reviewers felt that repeating the behavior but with the precise protocol used for the transcriptomics would allow for more direct comparison and strengthen the conclusions.

2) The value added by the use of TRAP was questionable. This approach should either be further validated or more strongly justified if it is to remain in the manuscript.

3) Some conclusions were overly anthropomorphized and an extensive literature on the roles of oxytocin and vasopressin in pair-bonding in this species was neglected. Changes in wording and emphasis combined with some new material in the Discussion could easily address this.

Reviewer #1 (Recommendations for the authors):

1) Prior work has implicated both the oxytocin and vasopressin systems in pair bonding, albeit in complicated ways, but there was both no discussion of that here nor an opportunity to provide a positive control since the nucleus accumbens does not include these systems. Perhaps the authors could discuss this as among the shortcomings.

2) Presumably the authors plan future studies using viral vectors to either over-express or repress candidate genes for manipulating the endurance of pair bonds. It is very unfortunate that the behavior did not correspond as expected. The authors' explanations for why this might be are legitimate but it does leave one wondering what to do next and highlights the challenge of such large sequencing studies.

Reviewer #2 (Recommendations for the authors):

This study was clearly a ton of work, and the authors marshall an impressive amount of data to address an important and understudied question. The combination of bioinformatics + bench work to get methods going in an emerging model is impressive.

It could help make the article more accessible to a wider readership if the authors could explain a little more clearly what vTRAP tells us that RNAseq doesn't. Why do we care about neuronal vs non-neuronal processes?

Some of the language about the transcriptomic results implies within-individual change although that can't be possible given that sampling is terminal, e.g. "…stably maintained while animals remained paired but eroded…" (abstract).

Interpretation of the transcriptional differences: clearly the transcriptional differences are not causal to the behavior that happened in the past. The decoupling of behavior and transcription offers some insights – the behavior remains the same (still shows behavioral signs of bonding) but the gene expression diverges; authors infer it has to do with the ability to establish a new bond but it's likely to be more heterogeneous? What about recovery from the loss, for example? Doesn't it seem likely that some of the DEGs reflect the ability to establish a new bond while others are probably about other processes, e.g. recovery?

Top left page 2: provide refs to support statements like "considering that other stable behavioral shifts…are underwritten by changes in transcription"

Why did the authors focus on males rather than females? The experimental design would be even better if they were brothers.

The sample size is ok with n=15, 16 pairs, but it would be helpful if the authors could provide some explicit acknowledgement of power (Figure 1).

R column, top of page 3: "latency to query"?

L column, page 3: need to explain why expected animals to have a preference for same-sex partners sooner before introducing the hypothesis.

L column, page 5: How to interpret opposite sex males initiated tumble fighting more quickly?

Page 5: please clearly state n for each treatment /time point group in the RNAseq analysis; presumably it's 50% of the n=15, 16 used in behavioral experiments?

Page 5, R column: which variables are being correlated here? "we calculated Rho values between the observed short-term pair bond and the shuffled…"

Page 5: how many individuals were excluded because they did not have baseline partner preference >50%?

Page 7: need to know experimental design for sequencing earlier, L column "on 8 groups of animals" – please remind us n for each group.

The major comparison is between same vs opposite sex housed pairs; therefore differences between treatment groups could reflect pair bonding OR they could reflect something about the sex of the partner. What about an additional control? They don't just differ in whether or not pair-bonded, but also in whether or not pair-bonded/housed/interacted with a male or a female

How did the authors choose which male member of the same-sex pair was used as the focal?

Page 8 L column: could this be correlated with the interesting behavioral results at the pair/individual level, i.e. pairs that were more bonded showed more of an effect than pairs that were less bonded?

Page 8, L column "the combined pair bond signature remains intact after short-term separation…" implies within-individual sampling.

Page 8, R column: this is a creative way to analyze RNAseq data!

vTRAP: Page 12, L column: It's neat that the authors were able to filter out genes so they could focus on genes expressed in neurons, but it seems like a lot of trouble to do all this work with cre and viruses – did they really learn that much more?

Page 12, R column: "the strong similarity in expression validates that bulk sequencing…" right – so did you really need to go to all the trouble of vTRAP?

Discussion on page 13 focuses on usual suspects/candidate genes: so why go to all this trouble to do this genomic approach if what really interests you are genes we already know are important? why not just have started with those?

Page 13: text here emphasizes that vTRAP didn't teach us anything particularly new we couldn't have figured out from the bulk RNASeq data alone. Here the authors emphasize that it's included in this paper to illustrate the utility of the method and make it available to the wider community of vole researchers, which is great but…. Does it belong in this paper?

Page 13 L column: the direct comparison of opposite versus same-sex paired animals provides a control for the effects of social context (alone vs with a pair) but it introduces another confound, namely the sex of the partner.

Page 13, R column: "correlation across partner preference behaviors…" seems like there could also be an opportunity to look for correlations between gene expression and behavior.

Page 14, L column: "transcripts associated with glia…" so can the effects of the experience of pair bonding vs pair bonding itself be teased apart? The experience of pair bonding could influence subsequent behavior via e.g learning -- there could be a transcriptional signal of that. There could also be a transcription signature of remembering that particular partner – can you differentiate between these two?

Sometimes the language goes a little too far "parsimonious with the human experience" page 15.

Methods: Tell us a little more about the colonies and genetic variation within them; how inbred are they?

Reviewer #3 (Recommendations for the authors):

1) The methods state that prairie voles were treated differently in the behavioral and transcriptomics studies. This might be why the authors do not see a strong behavioral degradation of pair bond-related behavior after long-term separation but do see a strong transcriptional signature. This reviewer understands that re-doing the behavioral study may not be feasible, this flaw should be addressed in the manuscript as a possible reason why the behavioral and transcriptomics results are not consistent. Further, the conclusions should be edited to fit this limitation.

2) While RRHO is helpful to assess overall patterns of transcriptional signatures across datasets, its utility for determining the exact transcripts is limited. For instance, even though the signal in the up-up quadrant in Figure 3F is stronger than the signal in the down-down quadrant in Figure 3G, there are more overlapping transcripts identified for a down-down. This is because of how RRHO determines the overlapping transcripts for its Venn diagram feature (by taking the point where the p-value is most significant and taking the list to the outside corner of that quadrant). It might be more helpful to use the RRHO quadrant information as a guide and then use traditional p-value cutoffs to look at overlapping transcripts. This would also reduce the number of transcripts fitting the desired pattern for follow-up experiments. The authors do this in the supplemental information which would be more helpful than the RRHO-identified overlapping transcripts presented in the main text. I suggest swapping these two approaches.

3) For the neuronal cluster analysis in Figure 4, the cluster 1 signal appears to be fairly weak, both in terms of the clustering in the neuronal enriched gene heatmap and the associated pathways. Why was this cluster selected? Further, it would be helpful to the reader if the authors included a description of each neuronal cluster in the figure (e.g., up in pair bond and ST separation, down in LT separation).

4) What is the signal for RRHO of short vs. long-term separation? Did I miss this? I feel that this is an important comparison to show.

5) In the volcano plots, it would be helpful to indicate the number of transcripts that are up or down-regulated on the figures.

6) Please include the coordinates for the NAc dissections.

7) The TRAP expression was verified in only one animal, which is not sufficient.

Essential revisions:

1) The disconnect between the behavioral responses in terms of endurance of the pair-bond, and the transcriptomics was of concern to the Reviewers. It was noted that there are subtle but perhaps important differences in the separation protocol used for the behavioral studies versus the transcriptomics. This combined with the behavioral data here not being consistent with the published literature undermines the current conclusions. The Reviewers felt that repeating the behavior but with the precise protocol used for the transcriptomics would allow for more direct comparison and strengthen the conclusions.

2) The value added by the use of TRAP was questionable. This approach should either be further validated or more strongly justified if it is to remain in the manuscript.

3) Some conclusions were overly anthropomorphized and an extensive literature on the roles of oxytocin and vasopressin in pair-bonding in this species was neglected. Changes in wording and emphasis combined with some new material in the Discussion could easily address this.

We thank the reviewers for their thoughtful and constructive comments which have helped us improve our manuscript. In our revised manuscript, have respond to three main weaknesses:

1. We addressed the inconsistency in the experimental design across the behavior and the transcription experiments by repeating the behavior with an experimental timeline that more exactly matches that of the animals used in transcriptional studies and in which separated animals were housed in separate colony rooms. Briefly, we replicated our initial results and found that same- and opposite-sex pairs display a partner preference even after prolonged partner separation. However, by including an additional control group that was never separated, we also found that although partner directed huddle increased in animals that remain paired from two to six weeks, this was not true for animals that were separated, suggesting that separation disrupts normal bond strengthening. Thus while the behavior seems surprisingly robust in light of the observed transcriptional changes, we are confident that our core finding – that voles can maintain a preference despite prolonged separation – is indeed a reliable finding.

2. We further validated and justified our use of TRAP and our focus on the NAc as the sole brain region of investigation as articulated in more detail in the manuscript and below in our response to reviewers.

3. We revised the language throughout the manuscript, especially in the discussion, to reduce anthropomorphizing our results and interpretations. Below we have provided responses to specific concerns articulated by each reviewer.

Reviewer #1 (Recommendations for the authors):

1) Prior work has implicated both the oxytocin and vasopressin systems in pair bonding, albeit in complicated ways, but there was both no discussion of that here nor an opportunity to provide a positive control since the nucleus accumbens does not include these systems. Perhaps the authors could discuss this as among the shortcomings.

A primary goal of our approach was to expand work in prairie voles beyond the narrow focus on nonapeptide systems. Justifying our approach to use RNA sequencing for an unbiased investigation, it was recently shown that the oxytocin receptor is not genetically required for pair bonding thus suggesting OXT-independent mechanisms of attachment in voles (20). We agree that an interesting area of future research will be to determine the contribution of nonapeptides to the observed pair bond transcriptional signature, but it was not a primary goal of the current study.

2) Presumably the authors plan future studies using viral vectors to either over-express or repress candidate genes for manipulating the endurance of pair bonds. It is very unfortunate that the behavior did not correspond as expected. The authors' explanations for why this might be are legitimate but it does leave one wondering what to do next and highlights the challenge of such large sequencing studies.

The opportunity to examine large biological datasets provides both challenges and opportunities. It this case, we have gained valuable timing information regarding the stability of pair bond transcription and the erosion of transcriptional signatures and robustness of partner preference following loss. Yet we are studying a

complicated system where we need additional information before confidently choosing and manipulating targets. In addition by repeating our behavioral experiments by including animals that remain paired we now know that separation impedes the strengthening of bonds providing us with an opportunity to potentially manipulate bond maturation. Such work is currently underway in our lab but beyond the scope of the current study.

Reviewer #2 (Recommendations for the authors):

This study was clearly a ton of work, and the authors marshall an impressive amount of data to address an important and understudied question. The combination of bioinformatics + bench work to get methods going in an emerging model is impressive.

Thank you

It could help make the article more accessible to a wider readership if the authors could explain a little more clearly what vTRAP tells us that RNAseq doesn't. Why do we care about neuronal vs non-neuronal processes?

We feel that inclusion of TRAP adds substantially to this manuscript and to our understanding of the neuromolecular underpinnings of bonding and loss in the NAc. The value of this experiment is twofold. As noted by Reviewer 3, “the TRAP approach in prairie voles is novel and will provide a great resource to the research community.” The prairie vole community has just developed its first transgenic Cre lines, which could be paired with vTRAP to query bond-associated gene expression changes exclusively in Cre-expressing neurons (12). Second, we noticed a puzzle in our tissue-level data. The majority of cells in the NAc are neurons (13, 14), and the vast majority of pair bonding studies of this region have focused on neuronal phenotypes, but our transcriptional signatures were linked to changes in glial populations. Ultimately, changes in glia are likely to act via their interactions with neurons, and vTRAP enables us to query specifically the neuronal transcriptional changes within our data. Supporting that this provides novel insights into our datasets, when we cluster transcripts based on their expression profiles following short and long-term separation, we predict different GO terms from the tissue level and neuronally-enriched gene sets. For instance, the GO terms resulting from cluster 2 for neuronal genes (Fig 4) includes “response to amphetamine” within the top 10 results, but the same cluster of genes from tissue level sequencing predicts this GO term as the 34th result.

Some of the language about the transcriptomic results implies within-individual change although that can't be possible given that sampling is terminal, e.g. "…stably maintained while animals remained paired but eroded…" (abstract).

Thank you. It was not our intent to imply within animal changes but rather to clarify potential transcriptional signatures that characterize bonding and loss where some language signifying continuity is unavoidable (e.g. you cannot experience loss without a bond to lose). We have updated the verbiage throughout to clarify that our measurements are not evidence of within-animal change.

Interpretation of the transcriptional differences: clearly the transcriptional differences are not causal to the behavior that happened in the past. The decoupling of behavior and transcription offers some insights – the behavior remains the same (still shows behavioral signs of bonding) but the gene expression diverges; authors infer it has to do with the ability to establish a new bond but it's likely to be more heterogeneous? What about recovery from the loss, for example? Doesn't it seem likely that some of the DEGs reflect the ability to establish a new bond while others are probably about other processes, e.g. recovery?

We agree that there appears to be a mismatch between the behavioral and transcriptional results. Because of this we repeated the behavior (please see public response to Reviewer 1) with separate cohorts of animals for each separation timepoint and the addition of a remain paired cohort that stayed with their partner throughout the six weeks of the experiment. When we compared opposite-sex paired males that remained with their partner to those that were separated from their partner we observed that separation impedes the continual strengthening of the pair bond. We interpreted this result to mean that upon separation the original pair bond is not lost per-say it is not being maintained in the way that it would be if the animals had stayed paired. Thus, the behavior now more closely resembles the transcriptional result of a residual pair bond signature that erodes over the course of separation. Future experiments will focus on a more nuanced understanding of how transcription can influence behavior to remain pliable in multiple social contexts. We addressed this comment in the discussion as follows:

“Pair bonding and loss are complex and multifaceted processes, engaging diverse neural circuits in temporal- and context-specific ways. Yet transcriptomics provides a single snapshot of gene expression at one point in time in one context. So while tissue-level transcriptional analysis is an excellent tool to capture an averaged biological state, one limitation of our approach is that it is not ideal for examining more temporally-dynamic processes, such as the rapid information-processing that occurs during social decision making.”

Top left page 2: provide refs to support statements like "considering that other stable behavioral shifts…are underwritten by changes in transcription"

Thank you for bringing this oversight to our attention. Three references have been added to this statement.

Why did the authors focus on males rather than females? The experimental design would be even better if they were brothers.

The methods state that the same-sex pairs were indeed siblings though we updated the language in the introduction to explicitly state that the same-sex pairs were siblings: “Here we map the trajectory of the pair bond transcriptional profile by comparing opposite-sex paired male to same-sex sibling pairs.” We plan to carry out similar studies in females, but colony size consideration and the complexity of the current dataset contributed to our decision to focus on males initially.

The sample size is ok with n=15, 16 pairs, but it would be helpful if the authors could provide some explicit acknowledgement of power (Figure 1).

We performed a power analysis prior to undertaking the study in order to estimate the group size needed to obtain significant partner preference at (α, Β blah blah). In general, it is not recommended to do posthoc power estimation. If a study was underpowered to detect the true effect size, a post-hoc power analysis of significant (p < 0.05) data is likely to subsequently overestimate power. Mathematically, the power at the observed effect is a function of the p-value. Instead, we have provided effect sizes for significant results, reported in the supplementary statistics table.

R column, top of page 3: "latency to query"?

We changed “query” to “determine” for clearer language.

L column, page 3: need to explain why expected animals to have a preference for same-sex partners sooner before introducing the hypothesis.

Thank you for pointing out our oversight in not including this information. We have added the following clause to the text:

“Similarly, we anticipated that same-sex sibling pairs would have a partner preference at baseline due to familiarity but would exhibit a faster and more robust loss of partner preference when separated.”

L column, page 5: How to interpret opposite sex males initiated tumble fighting more quickly?

We believed that repeating the partner preference test data was a crucial revision, and we correspondingly chose to avoid any potential confounds. Thus we did not to measure resident intruder due to concerns of that aggressive homecage interactions when the partner was removed could have unintended effects on affiliative behavior. All aggression data has been removed from the manuscript.

Page 5: please clearly state n for each treatment /time point group in the RNAseq analysis; presumably it's 50% of the n=15, 16 used in behavioral experiments?

Page 5: how many individuals were excluded because they did not have baseline partner preference >50%?

Group size and excluded animals for all transcription experiments are now clearly stated in the results of the manuscript. Additionally, cohort numbers can be found in the partner preference graphs of Figures 2 and 3.

In addition to listing the excluded animals in Supplemental Data S1, we have added the numbers of excluded animals that had a baseline partner preference <50% in the text as well. For the combined pair bond analysis the verbiage now reads as:

“For consistency, we limited transcriptional assessment to voles that had a baseline partner preference >50% (Figure 2B; Table S1) (OS n = 15; SS n = 11. Excluded for PPT <50% OS n = 3; SS n = 8).” For the separation analysis the text now reads “As before, we only included animals with a baseline partner preference >50% (Figure 3B) (OS n = 12; SS n = 11. Excluded for PPT <50% OS n = 4; SS n = 5).”

Page 5, R column: which variables are being correlated here? "we calculated Rho values between the observed short-term pair bond and the shuffled…"

The variable used to calculate the Rho values was the log2FoldChange as calculated by DESeq2 during differential gene expression analysis. We have revised the sentence to include this information:

“We shuffled each animal’s cohort identity at the long-term timepoint and calculated Rho values of the log₂FoldChange of differential gene expression between the observed short-term pair bond and the shuffled long-term pair bond over 1000 iterations”

Page 7: need to know experimental design for sequencing earlier, L column "on 8 groups of animals" – please remind us n for each group.

The experimental design is described in the first paragraph of the Results section and is graphically represented in Figures 2A and 3A. The verbiage in question was changed to “on all animals in this study (n = 49).”

The major comparison is between same vs opposite sex housed pairs; therefore differences between treatment groups could reflect pair bonding OR they could reflect something about the sex of the partner. What about an additional control? They don't just differ in whether or not pair-bonded, but also in whether or not pair-bonded/housed/interacted with a male or a female

Page 13 L column: the direct comparison of opposite versus same-sex paired animals provides a control for the effects of social context (alone vs with a pair) but it introduces another confound, namely the sex of the partner.

Please see our response to your request (point 1) in the public response.

How did the authors choose which male member of the same-sex pair was used as the focal?

The following sentence was added to methods under section RNA preparation and sequencing:

“The focal animal was chosen at random from each pair prior to the beginning of the experiment.”

Page 8 L column: could this be correlated with the interesting behavioral results at the pair/individual level, i.e. pairs that were more bonded showed more of an effect than pairs that were less bonded?

Page 13, R column: "correlation across partner preference behaviors…" seems like there could also be an opportunity to look for correlations between gene expression and behavior.

Please see our public response point R2.1.

Page 8, L column "the combined pair bond signature remains intact after short-term separation…" implies within-individual sampling.

We have updated the verbiage to make it clearer that we are examining across groups and not within animals. The text now reads:

“Specifically, when genes are ordered based on their degree of down- or up-regulation in the pair bond group we found that gene expression was largely indistinguishable in short-term separated animals but the pair bond signature(Rho = 0.32, p < 2.2X10^-16) is largely eroded in long-term separated animals (Rho = 0.048, p = 1.6X10^-7) (Figure 3D).”

Page 8, R column: this is a creative way to analyze RNAseq data!

Thank you!

vTRAP: Page 12, L column: It's neat that the authors were able to filter out genes so they could focus on genes expressed in neurons, but it seems like a lot of trouble to do all this work with cre and viruses – did they really learn that much more?

Page 12, R column: "the strong similarity in expression validates that bulk sequencing…" right – so did you really need to go to all the trouble of vTRAP?

Page 13: text here emphasizes that vTRAP didn't teach us anything particularly new we couldn't have figured out from the bulk RNASeq data alone. Here the authors emphasize that it's included in this paper to illustrate the utility of the method and make it available to the wider community of vole researchers, which is great but…. Does it belong in this paper?

We feel strongly that the vTRAP data adds substantially although we agree that this could have been more clearly articulated that it was in the prior version of the manuscript. The following is copied verbatim from our public response and is now included in the Discussion section: We feel that inclusion of TRAP adds substantially to this manuscript and to our understanding of the neuromolecular underpinnings of bonding and loss in the NAc. The value of this experiment is twofold. As noted by Reviewer 3, “the TRAP approach in prairie voles is novel and will provide a great resource to the research community.” The prairie vole community has just developed its first transgenic Cre lines, which could be paired with vTRAP to query bond-associated gene expression changes exclusively in Cre-expressing neurons (15). Second, we noticed a puzzle in our tissue-level data. The majority of cells in the NAc are neurons (16, 17), and the vast majority of pair bonding studies of this region have focused on neuronal phenotypes, but our transcriptional signatures were linked to changes in glial populations. Ultimately, changes in glia are likely to act via their interactions with neurons, and vTRAP enables us to query specifically the neuronal transcriptional changes within our data. Supporting that this provides novel insights into our datasets, when we cluster transcripts based on their expression profiles following short and long-term separation, we predict different GO terms from the tissue level and neuronally enriched gene sets. For instance, the GO terms resulting from cluster 2 for neuronal genes (Figure 4) includes “response to amphetamine” within the top 10 results, but the same cluster of genes from tissue level sequencing predicts this GO term as the 34th result.

Discussion on page 13 focuses on usual suspects/candidate genes: so why go to all this trouble to do this genomic approach if what really interests you are genes we already know are important? why not just have started with those?

We take a slightly different perspective. Our identification of such “usual suspects” demonstrates that our results are consistent with the known biology of pair bonding. This dually validates prior work and provides opportunities to expand our understanding by looking at other features of our dataset. In particular, our genomic approach has expanded our understanding via: (1) identification of transcriptional patterns important for regulating behavioral states, (2) identification of upstream regulators that modulate such multi-gene expression patterns, and (3) insight into potential biological processes not previously implicated in pair bonding, such as oligodendrogenesis/myelination, which would be unlikely to result from focusing on candidate genes alone.

Page 14, L column: "transcripts associated with glia…" so can the effects of the experience of pair bonding vs pair bonding itself be teased apart? The experience of pair bonding could influence subsequent behavior via e.g learning -- there could be a transcriptional signal of that. There could also be a transcription signature of remembering that particular partner – can you differentiate between these two?

We interpret this to comment to mean that (1) pair bond formation is distinct from bond maintenance, with the former likely reflecting complex learning processes and the latter evoking homeostatic mechanisms that reinforce the bond. And (2) bonding requires multiple component processes (social recognition memory, motivation, reward association, etc). Towards the former, we are currently investigating the hypothesis that gliogenesis/myelination may be required for bond formation/maturation, consistent with its role in other complex forms of learning and long-term memory (21–23). These results will be presented in a separate manuscript. Towards the latter, the behavioral test we employed (the partner preference test) is the gold standard to assessing pair bonds, but it cannot dissociate the different component processes that contribute to bonding. An important area of follow-up research will be to ask how transcriptional or other neurobiological changes contribute to the component processes and their integration in bonding. However, we feel that this question is beyond the scope of the current experiment.

Sometimes the language goes a little too far "parsimonious with the human experience" page 15.

We thank the reviewer for encouraging us to remain objective with our interpretations. We have addressed anthropomorphic wording in the manuscript with the following revision:

We changed “Such an explanation is parsimonious with the human experience.“ to

“Speculatively, this framework optimizes reproductive opportunities by dually enabling a new bond to form or resuming an existing bond. It is likewise tempting to extrapolate this result into the context of human experience. We do not forget our previous bonds or their emotional valence; rather we integrate the loss in order to continue on with life.”

Methods: Tell us a little more about the colonies and genetic variation within them; how inbred are they?

The current methods state that “Sexually naive adult prairie voles (Microtus ochrogaster, N = 219: 161M, 58F, P60 – P168 at experiment start) were bred in-house in a colony originating from a cross between voles obtained from colonies at Emory University and University of California Davis, both of which were established from wild animals collected in Illinois.”

We are careful to maintain an outbred colony. In particular, we maintain pedigrees, and we routinely breed in new, wild-derived voles when animals approach ~20% shared background among non-first-degree relations (approximately every 2 years). However, we have not performed genome sequencing to calculate their degree of relatedness.

Reviewer #3 (Recommendations for the authors):

1) The methods state that prairie voles were treated differently in the behavioral and transcriptomics studies. This might be why the authors do not see a strong behavioral degradation of pair bond-related behavior after long-term separation but do see a strong transcriptional signature. This reviewer understands that re-doing the behavioral study may not be feasible, this flaw should be addressed in the manuscript as a possible reason why the behavioral and transcriptomics results are not consistent. Further, the conclusions should be edited to fit this limitation.

The following is modified from the public response: The experimental designs employed to assess behavior and transcription differed, which may have contributed to the apparent mismatch in our results when comparing these two levels of biology. Thus we repeated our behavioral assessment using a design that directly matches the one we used in our transcriptional assessment. Specifically, we included a set of “remain paired” controls, separated partners were kept in separate colony rooms to ensure no possible access to partner-associated sensory cues (visual, auditory, olfactory), and separate cohorts of animals were used for short- and long-term separation. This design avoided partner re-introduction during the short-term partner preference test. In short, we replicated our initial findings; prairie voles will display a partner preference for same- or opposite-sex partner even after prolonged separation. Of note, animals that remained paired showed an increase in partner-directed affiliation (huddle time), which was not observed in separated animals suggesting that separation impairs the normal strengthening of bonds with time. Results associated with Figure 1 and the relevant discussion points have been updated in the manuscript.

2) While RRHO is helpful to assess overall patterns of transcriptional signatures across datasets, its utility for determining the exact transcripts is limited. For instance, even though the signal in the up-up quadrant in Figure 3F is stronger than the signal in the down-down quadrant in Figure 3G, there are more overlapping transcripts identified for a down-down. This is because of how RRHO determines the overlapping transcripts for its Venn diagram feature (by taking the point where the p-value is most significant and taking the list to the outside corner of that quadrant). It might be more helpful to use the RRHO quadrant information as a guide and then use traditional p-value cutoffs to look at overlapping transcripts. This would also reduce the number of transcripts fitting the desired pattern for follow-up experiments. The authors do this in the supplemental information which would be more helpful than the RRHO-identified overlapping transcripts presented in the main text. I suggest swapping these two approaches.

Please see our public response point R3.2.

3) For the neuronal cluster analysis in Figure 4, the cluster 1 signal appears to be fairly weak, both in terms of the clustering in the neuronal enriched gene heatmap and the associated pathways. Why was this cluster selected? Further, it would be helpful to the reader if the authors included a description of each neuronal cluster in the figure (e.g., up in pair bond and ST separation, down in LT separation).

Which clusters were spotlighted in Figure 4F was determined by a combination of transcriptional patterns due to separation, which genes are in the cluster, and the resulting GO terms for each cluster. We identified clusters that represented three interesting patterns: (1.) down only after long-term pair bond separation (cluster 1) (2.) up due to acute pair separation (cluster 2) and (3.) up only after long-term pair bond separation (cluster 3). While other clusters could have been used, we felt that these were important transcriptional patterns to highlight in the context of the rest of our results. Further, we noted that cluster 1 contains Shank3, a gene that has been linked to autism and for which mice containing the null allele show social behavior deficits (24). We reasoned that the expression of Shank3, and other genes with a similar transcriptional pattern after partner separation, may be of interest to a broad audience. Finally, we ran GO analysis on each cluster. Unfortunately, at this last step, none of the GO terms for cluster 1 were significant though this is likely due to the cluster’s small size. This is indicated by the light shading, but we felt it was worth including for the sake of transparency.

We have added the following description of cluster selection taken from this response has been added to the “Neuronal enriched genes throughout separation” methods:

“Which clusters were spotlighted in Figure 4F was determined by a combination of transcriptional patterns due to separation, which genes are in the cluster, and the resulting GO terms for each cluster. We identified clusters that represented three interesting patterns: 1.) down only after long-term pair bond separation (cluster

1) 2.) up due to acute pair separation (cluster 2) and 3.) up only after long-term pair bond separation (cluster 3).”

The following description of Cluster 1 is provided in the text:

“Cluster 1 consists of genes that were upregulated in opposite-sex paired males during pair bonding and short-term separation and downregulated following longterm separation.”

The following description of Cluster 2 is provided in the text:

“Cluster 2 transcripts are downregulated in pair bonded males and then robustly upregulated after short-term partner loss and, to a lesser extent, at the longterm timepoint indicating an acute transcriptional response to pair bond disruption.”

The following description of Cluster 3 is provided in the text:

“Finally, Cluster 3 contains genes that were downregulated in pair bonded males and became upregulated only after long-term opposite-sex partner separation, representing transcripts that may help prime the vole to form a new bond.”

Additionally, we updated cluster labels in Figure 4 to be more descriptive.

4) What is the signal for RRHO of short vs. long-term separation? Did I miss this? I feel that this is an important comparison to show.

We thank the reviewer for pointing this out. While we had looked at this in some of our initial analyses, we ultimately chose to focus on the question of what happens to the pair bond signature upon short- or long-term separation, thus omitting this secondary analysis. Based on the above question, we revisited this RRHO, which is now provided in Figure 3 —figure supplement 5. Notably, we see substantial overlap in the UU and DD quadrants in the short-term to long-term separation comparison. This pattern is particularly enriched for overlapping DD transcripts, which is the inverse of what we see in the pair bond::short term RRHO. One likely explanation for this potential overlap, especially in the DD quadrant, is that we are picking up on genes that have already eroded by the short-term separation timepoint and remain eroded after long-term separation. Supporting this interpretation, we took the UU and DD quadrant gene lists (referred to as stable separation genes) and filtered the DESeq2 results of the pair bond, short-term separation, and long-term separation (the same process as Figure 3H and I) for these genes. We identified an interesting pattern in that the genes that were most highly up- or downregulated in the pair bond showed a reversal of expression after short-term separation, correspondingly falling in the opposite UU or DD quadrant post-separation. Thus we filtered the stable separation genes for each condition by a log2FoldChange < -0.30 in the UU quadrant and > 0.30 in the DD quadrant and assessed the transcriptional state of these genes in the pair bond. Figure 3 —figure supplement 5B and C show that the genes that were most downregulated in the UU quadrant show a sharp inversion of expression after short-term separation that is maintained through long-term separation and vice-versa for the DD quadrant genes. Thus, there are subsets of transcripts that are up- or downregulated in pair bonded animals that are already exhibiting erosion after short-term separation. We further asked which biological processes are represented by the genes in Figure 3 —figure supplement 5D and E to investigate which processes are sensitive specifically to short-term separation. We found that UU genes that are then downregulated are associated with dopaminergic processes and DD genes that are then upregulated are associated with neuronal rearrangement processes. Thus, this RRHO analysis enriches our interpretation of transcriptional erosion of pair bond signatures following partner separation. Also related to point R3.2, overall we observed only slight to moderate erosion at the short-term timepoint, which was masked when relying exclusively on a strict p-value cutoff as in Supplemental Figure 6F. Rather, our rigorous experimental design, which has been noted as a strength of this manuscript, allows for more interrogation of the temporal complexity of these transcriptional signatures via threshold-free approaches.

5) In the volcano plots, it would be helpful to indicate the number of transcripts that are up or down-regulated on the figures.

We have added the number of upregulated and downregulated genes to each volcano plot. Additionally, gene lists and the associated differential expression values can be found in Supplemental Data 2.

6) Please include the coordinates for the NAc dissections.

The section “RNA preparation and sequencing” now includes the following description: “The NAc coordinates for dissections were: +1.7 AP, +/- 1 ML, -4.7 to -4.4 DV with the TRAP dissections additionally guided by GFP fluorescence visualized with a fluorescent dissecting scope.”

7) The TRAP expression was verified in only one animal, which is not sufficient.

We interpret this to mean that there are concerns that we only confirmed the Cre-dependence of the TRAP construct in a single animal as all other aspects of the TRAP work involved significantly more animals. We note that most labs rarely confirm Cre-dependence of vectors in more than one or two animals as the results, including those shown in Figure 4 —figure supplemental 2A, are typically definitive (i.e. no expression in the absence of Cre, expression in the presence of Cre). In addition to the images shown in Figure 4 —figure supplemental 2A, we used fluorescent guided dissection to harvest tissue/mRNA, serving as an additional visual confirmation of RPL10-GFP expression in the animals used to generate Figure 4. Additionally, a collaborating lab also confirmed that this vector also expresses in rats when Cre-recombinase is present. Finally, we performed two additional surgeries to confirm that TRAP is only expressed in the presence of Crerecombinase since the prior submission.

References

1. J. L. Scribner, E. A. Vance, D. S. W. Protter, W. M. Sheeran, E. Saslow, R. T. Cameron, E. M. Klein, J. C. Jimenez, M. A. Kheirbek, Z. R. Donaldson, A neuronal signature for monogamous reunion. Proceedings of the National Academy of Sciences. 117, 11076–11084 (2020).

2. A. M. Borie, S. Agezo, P. Lunsford, A. J. Boender, J.-D. Guo, H. Zhu, G. J. Berman, L. J. Young, R. C. Liu, Social experience alters oxytocinergic modulation in the nucleus accumbens of female prairie voles. Curr Biol. 32, 1026-1037.e4 (2022).

3. in vivo imaging identifies temporal signature of D1 and D2 medium spiny neurons in cocaine reward | PNAS, (available at https://www.pnas.org/doi/10.1073/pnas.1521238113).

4. M.-F. O’Connor, D. K. Wellisch, A. L. Stanton, N. I. Eisenberger, M. R. Irwin, M. D. Lieberman, Craving love? Enduring grief activates brain’s reward center. NeuroImage. 42, 969–972 (2008).

5. T. Hikida, S. Yao, T. Macpherson, A. Fukakusa, M. Morita, H. Kimura, K. Hirai, T. Ando, H. Toyoshiba, A. Sawa, Nucleus accumbens pathways control cell-specific gene expression in the medial prefrontal cortex. Sci Rep. 10, 1838 (2020).

6. C. Baimel, L. M. McGarry, A. G. Carter, The Projection Targets of Medium Spiny Neurons Govern Cocaine-Evoked Synaptic Plasticity in the Nucleus Accumbens. Cell Reports. 28, 2256-2263.e3 (2019).

7. N. S. Lee, N. L. Goodwin, K. E. Freitas, A. K. Beery, Affiliation, aggression, and selectivity of peer relationships in meadow and prairie voles. Frontiers in Behavioral Neuroscience. 13 (2019), doi:10.3389/fnbeh.2019.00052.

8. O. J. Bosch, H. P. Nair, T. H. Ahern, I. D. Neumann, L. J. Young, The CRF System Mediates Increased Passive Stress-Coping Behavior Following the Loss of a Bonded Partner in a Monogamous Rodent. Neuropsychopharmacology. 34, 1406–1415 (2009).

9. O. J. Bosch, J. Dabrowska, M. E. Modi, Z. V. Johnson, A. C. Keebaugh, C. E. Barrett, T. H. Ahern, J. Guo, V. Grinevich, D. G. Rainnie, I. D. Neumann, L. J. Young, Oxytocin in the nucleus accumbens shell reverses CRFR2-evoked passive stress-coping after partner loss in monogamous male prairie voles. Psychoneuroendocrinology. 64, 66–78 (2016).

10. A. J. Grippo, B. S. Cushing, C. S. Carter, Depression-like behavior and stressor-induced neuroendocrine activation in female prairie voles exposed to chronic social isolation. Psychosomatic Medicine. 69, 149– 157 (2007).

11. A. J. Grippo, D. Gerena, J. Huang, N. Kumar, M. Shah, R. Ughreja, C. Sue Carter, Social isolation induces behavioral and neuroendocrine disturbances relevant to depression in female and male prairie voles. Psychoneuroendocrinology (2007), doi:10.1016/j.psyneuen.2007.07.004.

12. K. Horie, K. Inoue, S. Suzuki, S. Adachi, S. Yada, T. Hirayama, S. Hidema, L. J. Young, K. Nishimori, Oxytocin receptor knockout prairie voles generated by CRISPR/Cas9 editing show reduced preference for social novelty and exaggerated repetitive behaviors. Horm Behav. 111, 60–69 (2019).

13. K. E. Savell, J. J. Tuscher, M. E. Zipperly, C. G. Duke, R. A. Phillips, A. J. Bauman, S. Thukral, F. A. Sultan, N. A. Goska, L. Ianov, J. J. Day, A dopamine-induced gene expression signature regulates neuronal function and cocaine response. Science Advances. 6, eaba4221 (2020).

14. Single-Cell RNA-Seq Uncovers a Robust Transcriptional Response to Morphine by Glia: Cell Reports, (available at https://www.cell.com/cell-reports/fulltext/S2211-1247(18)31384-6?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS2211124718313846%3 Fshowall%3Dtrue).

15. S. L. Fulton, S. Mitra, A. E. Lepack, J. A. Martin, A. F. Stewart, J. Converse, M. Hochstetler, D. M. Dietz, I. Maze, Histone H3 dopaminylation in ventral tegmental area underlies heroin-induced transcriptional and behavioral plasticity in male rats. Neuropsychopharmacology. 47, 1776 (2022).

16. S. G. Caradonna, T.-Y. Zhang, N. O’Toole, M.-J. Shen, H. Khalil, N. R. Einhorn, X. Wen, C. Parent, F. S. Lee, H. Akil, M. J. Meaney, B. S. McEwen, J. Marrocco, Genomic modules and intramodular network concordance in susceptible and resilient male mice across models of stress. Neuropsychopharmacol. 47, 987–999 (2022).

17. J. S. Wang, T. Kamath, C. M. Mazur, F. Mirzamohammadi, D. Rotter, H. Hojo, C. D. Castro, N. Tokavanich, R. Patel, N. Govea, T. Enishi, Y. Wu, J. da Silva Martins, M. Bruce, D. J. Brooks, M. L. Bouxsein, D. Tokarz, C. P. Lin, A. Abdul, E. Z. Macosko, M. Fiscaletti, C. F. Munns, P. Ryder, M. KostAlimova, P. Byrne, B. Cimini, M. Fujiwara, H. M. Kronenberg, M. N. Wein, Control of osteocyte dendrite formation by Sp7 and its target gene osteocrin. Nat Commun. 12, 6271 (2021).

18. D. A. Gallegos, M. Minto, F. Liu, M. F. Hazlett, S. Aryana Yousefzadeh, L. C. Bartelt, A. E. West, Cell-type specific transcriptional adaptations of nucleus accumbens interneurons to amphetamine. Mol Psychiatry, 1–15 (2022).

19. B. J. Hilton, A. Husch, B. Schaffran, T. Lin, E. R. Burnside, S. Dupraz, M. Schelski, J. Kim, J. A. Müller, S. Schoch, C. Imig, N. Brose, F. Bradke, An active vesicle priming machinery suppresses axon regeneration upon adult CNS injury. Neuron. 110, 51-69.e7 (2022).

20. K. M. Berendzen, D. S. Manoli, Rethinking the Architecture of Attachment: New Insights into the Role for Oxytocin Signaling. Affect Sci. 3, 734–748 (2022).

21. C. M. Bacmeister, R. Huang, L. A. Osso, M. A. Thornton, L. Conant, A. R. Chavez, A. Poleg-Polsky, E. G. Hughes, Motor learning drives dynamic patterns of intermittent myelination on learning-activated axons. Nat Neurosci. 25, 1300–1313 (2022).

22. E. G. Hughes, J. L. Orthmann-Murphy, A. J. Langseth, D. E. Bergles, Myelin remodeling through experience-dependent oligodendrogenesis in the adult somatosensory cortex. Nat Neurosci. 21, 696–706 (2018).

23. M. A. Thornton, E. G. Hughes, Neuron-oligodendroglia interactions: Activity-dependent regulation of cellular signaling. Neurosci Lett. 727, 134916 (2020).

24. C. M. Durand, C. Betancur, T. M. Boeckers, J. Bockmann, P. Chaste, F. Fauchereau, G. Nygren, M. Rastam, I. C. Gillberg, H. Anckarsäter, E. Sponheim, H. Goubran-Botros, R. Delorme, N. Chabane, M.-C. Mouren-Simeoni, P. de Mas, E. Bieth, B. Rogé, D. Héron, L. Burglen, C. Gillberg, M. Leboyer, T. Bourgeron, Mutations in the gene encoding the synaptic scaffolding protein SHANK3 are associated with autism spectrum disorders. Nat Genet. 39, 25–27 (2007).

Author details

1. Julie M Sadino

   Department of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2330-6198

2. Xander G Bradeen

   1. Department of Psychology and Neuroscience, University of Colorado Boulder, Boulder, United States
   2. Department of Adult Hematology, University of Colorado- Anschutz Medical Campus, Aurora, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

3. Conor J Kelly

   1. Department of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, United States
   2. BioFrontiers Institute, University of Colorado Boulder, Boulder, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2410-7892

4. Liza E Brusman

   Department of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, United States

   Contribution

   Data curation, Formal analysis, Visualization, Writing - review and editing

   Competing interests

   No competing interests declared

5. Deena M Walker

   Department of Behavioral Neuroscience, Oregon Health and Science University, School of Medicine, Portland, United States

   Contribution

   Data curation, Software, Formal analysis, Visualization, Writing - review and editing

   Competing interests

   No competing interests declared

6. Zoe R Donaldson

   1. Department of Molecular, Cellular, and Developmental Biology, University of Colorado Boulder, Boulder, United States
   2. Department of Psychology and Neuroscience, University of Colorado Boulder, Boulder, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   zoe.donaldson@colorado.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-6699-7905

• 1,067

  Page views

• 154

  Downloads

• 4

  Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.