The need for stable social attachments represents a ubiquitous phenomenon within social groups. Social affiliation is a fundamental requirement that provides security and mental health and thus, affiliative and reciprocal sociality is observable across species (Williams, 2006). Social rejection, the experience of being excluded or ostracized, can be detected in almost all social species and has been observed in humans across cultures and various developmental stages (Williams, 2006). Rejection by group members is only one form of negative social interaction, but since it threatens the need for social inclusion, and thus survival, individuals are more sensitive to rejection as compared to other types of aversive social interaction (Goodall, 1986). In humans, experience of rejection has been reported to interfere with pain processing (Bernstein and Claypool, 2012) and is involved in the development of neuropsychiatric disorders such as depression, borderline personality disorder (BPD) and anxiety disorders (Hawthorne, 2008; Stanley and Siever, 2010).

Adequate peer-contact gains particular significance during adolescence. From an evolutionary perspective, adolescence is crucial for the maturation of social skills, which are essential for successful social inclusion and reproduction and enable an individual to slowly separate from the protective influence of the family (Spear, 2000). This individuation from parents is accompanied by a deepening of relationships and intimacy with peers, especially among girls (Larson and Richards, 1991). Teenagers spend more time interacting with peers than during any other developmental period and are more susceptible to peer-influence. Sensitivity towards social rejection has been reported to be higher during adolescence than in adulthood, with girls being particularly sensitive to the perception of peer-rejection (Sebastian et al., 2010). Adolescent social rejection elicits strong feelings of distress, has been found to negatively affect psychological well-being and can result in severe adverse health outcomes (Lev-Wiesel et al., 2006), whereas adequate social support has been linked to positive effects on mental health (Kawachi and Berkman, 2001). Surprisingly, the neurobiological mechanisms mediating the consequences of social rejection remain largely unknown. At this point valid animal models represent a crucial link to increase our insight into mechanistic processes underlying social rejection and its consequences.

We previously reported on a unique experimental approach to model social rejection in rats, by modulating the ability of adolescent female animals to engage in social play behavior either in acute social encounters or over longer periods throughout adolescence (Schneider et al., 2014, 2016). Peer-directed playful activities have a considerable incentive value during adolescence and are crucial for the development of social competence (Trezza et al., 2010). Furthermore, previous studies found that depriving adolescent animals from play had detrimental effects on social behavior later on (van den Berg et al., 1999; Burleson et al., 2016) and interfered with the neuronal development of the fronto-cortical regions (Bell et al., 2010), but that providing brief daily social contact could partially alleviate effects of isolation rearing (Einon et al., 1978). In particular the reciprocity in a play fight, i.e., the active engagement of the play partner, has been identified as the main rewarding component of these social interactions (Pellis and McKenna, 1995). In our experimental approach, adolescent Wistar animals are deprived of these reciprocal social experiences by being subjected to playful encounters with rats from the inbred Fischer 344 strain. We could recently show that Fischer animals differ profoundly in their quality of social play from outbred strains, such as Wistar or Sprague Dawley. Fischer rats show a diminished responsiveness upon playful attacks, employ different defensive strategies and show increased locomotor play. Hence, during cross-strain playful encounters female adolescent Wistar rats receive only few playful responses from their Fischer partner while increasing their own attack frequency (Schneider et al., 2016). Therefore, Fischer partners do not adequately accommodate the need for social play in Wistar rats, since they negate the most rewarding aspects of social play (Pellis and McKenna, 1995).

In our previously used model, pairing of one female Wistar rat with either one same strain partner or one inadequate Fischer play partner throughout adolescence, induced lasting changes in pain sensitivity and CB1 receptor (CB1R) protein levels, implicating the endocannabinoid (eCB) system as a potential modulator of social rejection experiences (Schneider et al., 2014). However, social skills remained unaffected in adulthood. Based on these previous findings, we refined our initial approach: instead of rearing in pairs, adolescent female Wistar rats were now reared within a group of Fischer animals (inadequate social rearing condition, ISR), or adequately within a group of same-strain partners (control). In adulthood all Wistar rats were then screened for changes in social behavior, pain perception, and alterations in the eCB system in brain regions that are part of the so called social brain (amygdala, striatum and prefrontal cortex; O’Connell and Hofmann, 2011) and were previously found to be affected (Schneider et al., 2014).

In the present study, we refined our initial approach to model long-term peer-rejection in adolescent female Wistar rats. The present results demonstrate that the inability to adequately engage in playful peer-interaction during adolescence results in lasting impairments in pain sensitivity and a variety of social abilities in adulthood by decreasing social interaction behavior, social memory and the processing of socially transmitted information. No lasting effects were observed on object recognition memory, locomotor activity and anxiety-related behaviors. Although, increased closed arm entries were observed in ISR rats in the EPM, this effect appeared transient since no increased activity in a novel environment was detected in the open field and the EMT. Molecular analysis revealed a significantly affected central eCB system in adult ISR animals, with changes in the amygdala, striatum, and in the thalamus.

We previously employed a dyadic design to model long-term peer-rejection through inadequate playful interaction during adolescence (Schneider et al., 2014), which in contrast to the presently used group design, did not yield any lasting impairments on social interaction behavior, although minor changes in the eCB system and decreased pain sensitivity were detected. Here, female Wistar animals were reared within adequate/inadequate groups, accommodating the apparent need in female rats for housing in groups vs. housing in pairs (Patterson-Kane et al., 2004). Our previous dyadic design may therefore have already affected social behavior in control animals through the lack of a sufficiently social environment and in turn explain why no differences were observed in social skills between ISR and control rats (Schneider et al., 2014). Allowing for social group living normalizes the social behavior in controls while also increasing the “outsider” position of the lone Wistar female. The analysis of social play behavior, assessed during the rearing condition on pd 30 and 35 revealed distinct differences in the quality of playful interactions between the adolescent Wistar ISR and control females, similar to the behavioral findings observed recently in acute inadequate social encounters (Schneider et al., 2016). In particular, only half of the nape attacks initiated by an ISR Wistar rat resulted in a defensive playful response by the Fischer play partners, whereas in the control condition almost all attacks resulted in a defensive, play appropriate response. This is of interest because it is commonly understood that play solicitation (i.e., nape attacks) alone, in the absence of defense by the recipient, does not serve as a sufficient reinforcer for playful interactions and therefore the responsiveness of the play partner is considered crucial for the rewarding nature of social play (Pellis and McKenna, 1995). No differences were observed for attacks received and pinning indicating that the Fischer rats do initiate playful interactions, while the number of attacks initiated was substantially higher in Wistar rats of the ISR condition on both days. This enhanced attack frequency in ISR Wistar rats appears to be directly linked to the non-responsiveness of their Fischer play partners. In order to achieve a similar number of rewarding social play interactions as in the control condition, ISR Wistar rats initiated twice as many attacks. This specific lack of an appropriate and active response of the Fischer play partners upon play solicitation may constitute a major part of the aversiveness during the rearing in our social rejection model which lasts through the main period of playfulness during adolescence (Pellis and Pellis, 1990). This hypothesis is further supported by the pronounced increase in locomotor play in ISR Wistar rats on pd 30, implicating a shift toward playful activities perceived as more rewarding than social play with the Fischer partners. However, on pd 35 differences in locomotor play were no longer observed indicating that behavioral adaptations take place throughout the rearing experience.

No group differences were observed in plasma corticosterone levels during the rearing condition, indicating that the general stress load did not induce a pronounced physiological stress-response during adolescence. Hence, the persistent and profound behavioral and molecular effects found in females of the ISR condition may not be mediated by increased stress levels due to the inadequate group housing over adolescence.

Different lasting behavioral alterations of the adolescent rejection experiences were observed in adulthood. During social interaction testing ISR animals showed a decrease in anogenital exploration and approach behavior compared to controls. Both variables can be linked to a more active and motivational aspect of social exploration while non-anogenital exploration, which has been argued to take less effort (Hamilton et al., 2010), did not differ between groups. No further differences were detected for contact behavior, evasion or self-grooming. Decreased social interaction has been reported as a prominent finding in animal models of social deprivation. Social interaction deficits were mainly reported in male rats, while the isolation condition still lasted (Ferdman et al., 2007; Meng et al., 2010), however isolation rearing in females was found to induce either no (Ferdman et al., 2007) or only minor changes (Hermes et al., 2011). In contrast, in the present study female rats from the ISR condition showed a persistent decrease in social interaction and social short-term memory that was measured around 1 month after termination of the inadequate housing conditions. Similar lasting deficits in social short-term memory have been reported in studies of adverse early-life experiences (e.g., maternal separation, pre-weaning ethanol exposure; Kelly and Tran, 1997; Lévy et al., 2003; Lukas et al., 2011). Notably, the deficit in memory processing in ISR rats was specific to the social domain, since object recognition memory was unaffected. Likewise, short isolation periods in adult rats were shown to specifically affect social recognition but not object recognition memory (Shahar-Gold et al., 2013). Rearing with inadequate play partners also strongly affected the ability to process socially transmitted information in a social food transfer task. ISR rats significantly less preferred the socially transmitted food than controls. Although it could be argued that the deficits in social interaction observed in adult ISR rats may underlie the lack of social learning, it has previously been shown that this paradigm does not depend on the familiarity of the donor and that a brief investigation of the snout area is sufficient to promote the preference for the novel food source (Galef et al., 1988). Similar to the acquisition of social recognition memory, the initial processing of the transmitted odor during social food transfer testing in rodents depends on the amygdala, in particular the basolateral region (Wang et al., 2006; Carballo-Márquez et al., 2009). Finally, animals reared within a group of inadequate social play partners showed a persistent decrease in pain sensitivity, similar to the findings observed in our previous dyadic model (Schneider et al., 2014). ISR rats displayed a higher latency for responding to a thermal stimulus compared to controls, indicating that the inability to adequately engage in social play behavior during adolescence induced hypoalgesia in adulthood.

The eCB system is an ubiquitous neuromodulatory system which modulates various processes of neuroplasticity (Katona and Freund, 2012), social behavior (Schneider et al., 2008; Klugmann et al., 2011) and pain processing (Guindon and Hohmann, 2009). Our findings indicate a lasting increase in eCB signaling in ISR animals, mainly within the amygdala, the striatum and thalamus. We observed enhanced CB1R protein levels and receptor/G protein coupling with concomitant increased levels of AEA in the amygdala of ISR animals compared to controls, while protein levels of the AEA degrading enzyme FAAH were decreased. A trend for increased CB1R levels with accompanied increase in receptor coupling was also detected in the thalamus, whereas no group differences for CB1R were observed in the striatum and the mPFC. Two-AG levels were elevated in the striatum, whereas protein levels of the degrading enzyme MAGL were decreased. The described changes were present 100 days after the cessation of adolescent peer-rejection and the results are partially in line with our previous findings from the dyadic model (Schneider et al., 2014), where we detected an up-regulation of thalamic CB1R and increased FAAH levels in the amygdala. The group design employed in the present study therefore exerts a stronger, albeit similar, lasting effect on various components of the eCB system than rearing in inadequate pairs.

The amygdala is part of the social behavior network (O’Connell and Hofmann, 2011) and eCB signaling within the amygdala is well known to modulate social behavior (Trezza et al., 2012; Seillier et al., 2013) and social memory (Segev and Akirav, 2011). Since we observed profound changes in amygdalar eCB signaling in ISR rats, these region-specific alterations may have contributed to the social learning deficits observed in these animals. Although it has been shown that positive social experiences correlate with an acute elevation of amygdalar AEA levels (Trezza et al., 2012), a pharmacological inhibition of FAAH, resulting in an elevation of AEA, was found to attenuate social interaction (Seillier et al., 2013). We report here that adverse peer-experiences induce a lasting increase in amygdalar CB1R/AEA levels and impair social skills. The additionally found increase in agonist stimulated activity of CB1R in the amygdala and thalamus further indicates a profoundly dysregulated eCB system in ISR females. So far, investigations on the impact of aversive early life experiences on CB1R concentrate on isolation rearing and are often inconsistent (Malone et al., 2008; Robinson et al., 2010; Sciolino et al., 2010), although results indicate that receptor density and activity in amygdala and thalamus remain unaffected by isolation rearing while CB1R coupling is downregulated in the striatum (Zamberletti et al., 2012). The results therefore appear to differ from our peer-rejection model where no changes were detected in striatal CB1R expression and binding. However, we observed alterations in striatal eCB content. Although, not primarily involved in the social behavior network, striatal regions have been implicated in social reward processing (O’Connell and Hofmann, 2011). In human adolescents, increased activity in the striatum and thalamus was also observed following acceptance-feedback from peers (Guyer et al., 2012).

Environmental stressors can activate descending pain-inhibitory systems which suppress pain by inhibiting transmission from nociceptors to the central nervous system. This stress-induced analgesia has been shown to be mediated by AEA signaling in the basolateral amygdala (BLA; Connell et al., 2006; Rea et al., 2013), where CB1R and FAAH immunoreactivity are dense (Gulyas et al., 2004). Increased availability of AEA in the BLA has been suggested to suppress pain perception by activation of CB1R on GABAergic interneurons which would disinhibit output neurons and thereby lead to activation of the descending inhibitory-pain pathway (Connell et al., 2006; Rea et al., 2013). The persistent increase in amygdalar AEA levels in our ISR rats may therefore underlie the decrease in thermal pain sensitivity observed in these animals. Additionally, activation of thalamic CB1R has been shown to contribute to the antinociceptive efficacy of cannabinoids (Martin et al., 1996), which is in line with our findings on enhanced thalamic CB1R levels and activity in ISR animals. Experimentally induced social rejection has also been shown to acutely alter nociception in humans (Borsook and MacDonald, 2010; DeWall et al., 2010; Bernstein and Claypool, 2012). Additionally, altered pain perception has been reported in neuropsychiatric disorders that are linked to early experiences of social rejection, such as BPD (Schmahl et al., 2006) and post-traumatic stress disorder (PTSD; Strigo et al., 2010) and appear to be associated with dysregulated amygdala activity (Kraus et al., 2009).

Recent studies in mice indicated that reducing levels of 2-AG by genetic inactivation of the main 2-AG synthesizing enzyme DAGLα lead to a phenotype of increased anxiety and depressive-like behavior. Furthermore, AEA levels were reduced in a region specific manner, indicating possible crosstalk through a most likely indirect mechanism (Jenniches et al., 2016). Interestingly, social behavior and pain sensitivity (hot plate) in these knock-out mice appeared unaffected. Opposite patterns in region specific (cortical vs. sub-cortical) up- and down-regulation of AEA and 2-AG, respectively, but not necessarily FAAH and MAGL protein levels, were observed in repeated homotypic restrained stress, interpreted as an adaptive mechanism to counteract repeated stress responses (reviewed in Patel and Hillard, 2008). These changes in eCB levels did not affect CB1R density or affinity. In the present study changes were region specific but not complementary, i.e., decrease of 2-AG did not result in increase of AEA in the same region, and anxiety behavior was unaffected. Furthermore, a study comparing the sociability enhancing properties of acetaminophen in three mouse strains found similar CB1R densities but increased stimulated binding in a GTPγS assay in the frontal cortex of strains known to be deficient in social behavior (Gould et al., 2012). Hence, further studies are needed to clarify whether the presently found pattern in complementary up and downregulated ligands and catabolic enzymes are functionally independent as they appear to be brain region specific adaptations of the eCB system, or whether they result from regional crosstalk and interactions with other neurotransmitter systems involved in social motivation and social reward behavior such as the oxytocin or opioid system (Gordon et al., 2011; McCall and Singer, 2012; Smith et al., 2015).

Altered eCB signaling in the amygdala (and to a lesser extent in the striatum and thalamus) may have occurred during the inadequate peer-experience in adolescent rats in order to counteract with this socially aversive environment and subsequently lead to persistent adaptational changes within the eCB system into adulthood (Schneider et al., 2014). Furthermore, we could show recently that pre-treatment with the CB1 antagonist/inverse agonist SR141716 prevented changes in pain sensitivity during acute social rejection encounters, thereby underlining the involvement of the eCB system in social rejection (Schneider et al., 2016). A rapid loss of amygdalar AEA signaling, possibly through a rapid hydrolysis by FAAH, is involved in the reactivity toward different stressors (Hill and Tasker, 2012) hence, enhanced CB1R expression and decreased FAAH levels may occur in the inadequate peer-condition as a long-term consequence to counteract the continuous decreased AEA tone. Notably, a recent study reported alterations of peripheral AEA and 2-AG levels in BPD patients, which indicates a modulatory role of the eCB system in this disorder (Schaefer et al., 2014). Since BPD is characterized inter alia by a heightened sensitivity to social rejection, disturbances in social skills, altered pain perception and emotional dysregulation (Schmahl et al., 2006; Bungert et al., 2015), these processes may require intact eCB signaling for normal function. Future experiments will be needed to extend our initial findings onto further neuronal systems involved in social abilities and social rejection. Potentially interesting targets in this context are the endogenous opioid and the central oxytocin system, since both systems are known to functionally interact with the eCB system (Cota et al., 2006; Russo et al., 2012), take part in the regulation of social behavior (Insel and Fernald, 2004; MacDonald and Leary, 2005) and have also been implicated in BPD (Stanley and Siever, 2010; Roepke et al., 2012).

In summary, prolonged periods of inadequate peer-interactions, social rejection or ostracism during adolescence have been shown to have profound negative effects on mental and physical health in humans (Lev-Wiesel et al., 2006; Dempsey and Storch, 2008; Copeland et al., 2013). However, the underlying neurobiological mechanisms remain to be characterized. We here report on persistent impairments in social skills, pain sensitivity, and considerable alterations in eCB signaling in adult female rats that were deprived of adequate playful peer-interactions during adolescence which may therefore provide a suitable approach to model social peer-rejection in laboratory rodents. Our refined inadequate rearing model may also have the potential for a more broadened use in investigating basic behavioral and neurobiological consequences of modulations of social play, a key component of early life social experiences and social development. Finally, the present data contribute to a better understanding of the neurobiological underpinnings of social rejection and, thus increase our insights in the neuropathology of deficiencies in social competence and psychosocial functioning in neuropsychiatric disorders.

PS and MS designed the study and analyzed the behavioral data. PS performed experimental work and statistical analysis. PS and LB performed molecular analyses. PS, MS and RS wrote the manuscript. BL, CS, MB and AM-L were involved in the conception and critical revision of the work. All authors contributed to and have approved the final manuscript.

This work was supported by the Deutsche Forschungsgemeinschaft (DFG) FOR926, (SP9 (SCHN958/4-1) to MS; to central project CP1 BL) and KFO256 (SCHN958/5-1; IP7; to MS and RS), and AERIAL (BMBF01EE1406C; to RS); the funding entities had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.