Enduring Deficits in Brain Reward Function after Chronic Social Defeat in Rats: Susceptibility, Resilience, and Antidepressant Response

INTRODUCTION

Mood disorders, like major depressive disorder (MDD), are characterized by deficits in reward processing (1). Clinically, these deficits are manifested in the symptom of anhedonia, or diminished interest or pleasure in rewarding stimuli and activities. Anhedonia is a trait feature of MDD (2) and may persist even after antidepressant treatment (3).

Exposure to stress precipitates the development of neuropsychiatric disorders characterized by anhedonia (4, 5). Studies in healthy humans and non-human animals demonstrated that stress induced anhedonia (6, 7). In humans, social loss, humiliation, entrapment and submissive behavior are associated with MDD (8–10). The social defeat procedure in rodents involves an ethological stressor whereby the experimental animal displays submissive behaviors when confronted by an aggressive, socially dominant conspecific (11). Thus, this procedure is considered analogous to psychosocial stressors in humans.

Repeated social defeat has been used to assess anhedonia-like behaviors in rodents, although almost exclusively with sucrose-based measures. Some studies demonstrated stress-induced decreases in preference for (12–15) and anticipation of (16) sucrose. Conversely, others reported unaffected sucrose consumption (16, 17) or preference (13, 18) after social defeat. Variability within experimental groups may partly explain these discrepancies. For example, rodents with high baseline exploratory activity (19) or low basal plasma corticosterone levels (20) displayed less sucrose preference after social defeat compared to animals with low exploratory behaviors and high corticosterone levels, respectively, suggesting intrinsic susceptibility and resilience to social defeat-induced anhedonia. Similar differential effects were observed in mice showing high or low social avoidance after chronic social defeat (15, 21), an effect mediated by the firing and neurotrophic signaling of ventral tegmental area (VTA) dopaminergic neurons (15, 22). Indeed, in humans, stress triggers depressive symptoms in vulnerable individuals, while resilient individuals adopt active coping strategies that promote healthy outcomes (23, 24). Thus, animal procedures modeling susceptibility and resilience to stress-induced anhedonia, like those described above, can provide insight into the factors that mediate stress-induced psychopathology.

Anhedonia reflects dysfunctional brain reward circuitries (25–27), yet there have been surprisingly few direct studies of the impact of social defeat on brain reward function. The goals of the present studies were to determine: 1) if chronic exposure to social defeat in rats produces deficits in brain reward function in all or only a subgroup of susceptible rats; and 2) whether repeated treatment with the selective serotonin reuptake inhibitor (SSRI) fluoxetine or the tricyclic antidepressant (TCA) desipramine would reverse such deficits. Brain reward function was assessed using the intracranial self-stimulation (ICSS) procedure that allows for daily assessment of reward thresholds to reliably quantify the development and reversal of anhedonia after chronic stress and antidepressant administration, respectively. Furthermore, we assessed whether stress-induced deficits in reward thresholds correlated with changes in brain-derived neurotrophic factor (BDNF) and related signaling molecules in depression-related brain areas, including the VTA, nucleus accumbens (NAc), central nucleus of the amygdala (CeA) and hippocampus. The forced swim test, a widely used assessment of antidepressant efficacy and a putative measure of behavioral despair, was also conducted after social defeat and fluoxetine treatment to determine whether brain reward function correlated with behavioral despair.

METHODS AND MATERIALS

RESULTS

DISCUSSION

Chronic exposure to social defeat immediately elevated ICSS reward thresholds in all rats, reflecting a stress-induced decrease in brain reward function. Thresholds remained elevated in a subset of susceptible rats, whereas thresholds in resilient rats were only acutely elevated during the initial four days of social defeat and were subsequently unaffected despite ongoing stress exposure. While stress-induced thresholds remained elevated in all susceptible rats, larger threshold elevations were associated with a diminished response to antidepressant treatment. Thus, the stress-induced anhedonia procedure presented here may provide a translational model with relevance to key features of MDD, including clinical observations that stress precipitates MDD in a subset of vulnerable individuals (23, 24), and increased anhedonia severity confers resistance to antidepressant treatment (36–38).

Susceptible rats displayed reward threshold elevations immediately and persistently throughout the 21-day stress and 21-day post-stress periods. Resilient rats also displayed an immediate elevation of reward thresholds, reflecting an acute stress response, but in contrast to susceptible rats, thresholds quickly returned to baseline despite continued social defeat. These divergent responses highlight one advantage of the ICSS procedure that is difficult to reproduce with other animal procedures; that is, observation of temporal changes in anhedonia and the development of susceptibility and resilience throughout the stress experience. Importantly, two independent statistical criteria were used to determine susceptibility and resilience, both of which classified rats identically across all three experiments. Furthermore, cluster analyses from all three experiments were unable to distinguish between resilient and control rats, suggesting that brain reward function was indeed unimpaired in resilient rats exposed to social defeat. Thus, defining susceptibility and resilience was not dependent on the statistical method used, as classification methods determined by the experimenter (two standard deviation Criterion 1) or completely blind to the experimenter (cluster analysis Criterion 2) yielded identical results.

Susceptibility and resilience is likely not due to differences in resident aggression, as susceptible and resilient rats had similar numbers of injuries and time spent in physical contact with the residents (see Supplemental Table S1). It should be noted that aggression was not quantified when intruders were housed behind the protective barrier, and likely varied across residents each day. However, intruder/resident pairings were alternated daily and each intruder (susceptible and resilient) was exposed to each resident only once throughout each experiment. Thus, it is unlikely that variation in aggression throughout each day contributed to overall susceptibility and resilience. Furthermore, both subgroups of stressed rats similarly lost weight during social defeat (see Supplemental Figures S2–S4), again suggesting similar stress experiences. Nonetheless, assessment of physiological stress markers is warranted to confirm stress severity. Similarities in the glucocorticoid response to social defeat between susceptible and resilient mice have been reported (15), although different hypothalamic-pituitary-adrenal axis responses were also observed in rats with different forced swim responses to social defeat (39).

Three weeks after termination of social defeat, susceptible rats had elevated pmTOR levels in the VTA relative to controls. mTOR is a signaling molecule that regulates cell metabolism, growth, proliferation and survival (40). Because stimulation of VTA dopaminergic neurons was shown to underlie susceptible behavior in socially defeated mice (15, 22) and stress increases synaptic strength at VTA dopaminergic neurons (41), chronically increased pmTOR signaling in the VTA may possibly facilitate the firing of dopaminergic neurons that contributes to susceptible behavior. However, because pmTOR was intermediately expressed in resilient rats compared to susceptible and control rats, further investigation may elucidate the exact role of this signaling molecule in regulating stress-induced brain reward deficits.

Interestingly, phosphorylation of AKT, an upstream regulator of mTOR, was also increased 3 weeks post-stress, albeit similarly between susceptible and resilient rats. Previous work in mice showed that chronic social defeat decreased pAKT in the VTA and that this decrease mediated susceptibility (42). This observation raises the possibility that the pAKT increase observed here in both susceptible and resilient subgroups represents an adaptive response. Similarly, BDNF levels were equally decreased in susceptible and resilient rats 3 weeks post-stress, consistent with a previous report that a similar stress procedure in rats decreased VTA BDNF levels (14). However, chronic social defeat in mice induced BDNF signaling from VTA to NAc in susceptible mice only (15, 43), suggesting that the decrease observed here, like the increase in pAKT, may be adaptive. Alternatively, BDNF and pAKT expression patterns in the present study may represent post-stress recovery-induced changes that are independent of any putative changes during stress exposure. Nonetheless, our results show that chronic adaptations in VTA BDNF-AKT signaling per se appear to be independent of persistent anhedonic behavior in our rat model. Moreover, chronic stress exposure had no long-term effects on IRS2 and pERK1,2, two intracellular signaling molecules involved in neurotrophin-mediated neuronal survival.

Clinically, the persistence of anhedonia makes it one of the most treatment-resistant symptoms of MDD (44). In humans, 47–52% of patients with MDD positively responded to an initial antidepressant medication (45, 46). In the present studies, antidepressant treatment was considered effective if reward thresholds returned to within two standard deviations of average pre-stress baseline levels (i.e., similar to resilient and control rats). In susceptible rats, thresholds returned to baseline in 44% treated with fluoxetine and 47% treated with desipramine, reflecting an antidepressant-like response. By contrast, thresholds remained elevated in vehicle-treated rats. Interestingly, the significant correlation between reward thresholds after stress and antidepressant treatment indicates that susceptible rats for which antidepressant treatment was effective had relatively lower stress-induced threshold elevations before antidepressant treatment compared to susceptible rats that were unaffected by antidepressant treatment. This pattern was observed for both fluoxetine and desipramine and is consistent with clinical observations that greater anhedonia severity predicts poorer antidepressant response (36–38, 47). Although the reported clinical studies did not differentiate between the different aspects of reward processing deficits that are often indiscriminately measured in humans, ICSS of the lateral hypothalamus of rats, which mediates motivated behaviors (48, 49), suggests that antidepressant efficacy may be associated with degree of avolition. Thus, consideration of motivational deficits may provide insights about the neurobiological factors underlying antidepressant resistance.

Finally, we assessed whether stress and antidepressant treatment altered forced swim test behavior. Originally developed to screen novel antidepressant medications, the forced swim test is widely used as a putative assessment of depression-like behavior, with increased immobility often interpreted as behavioral despair. The forced swim response to social defeat was assessed previously with mixed results. Some studies demonstrated increased immobility after social defeat (12, 13, 18), though others showed no effect (13, 15, 50) or even decreased immobility after stress (51). Here, fluoxetine decreased immobility in non-stressed rats compared to vehicle treatment, indicating that the swim test was effective in detecting SSRI activity, for which it was originally designed. However, immobility was also decreased in susceptible rats compared to controls that received vehicle treatment and was similar between susceptible and resilient rats. Because social defeat decreased immobility (an antidepressant-like effect), any potential further change in immobility in either direction with antidepressant treatment would be uninterpretable. Accordingly, the forced swim test was not repeated in the desipramine study. Interestingly, no correlation emerged between immobility and reward thresholds that were previously assessed. One explanation for this discrepancy is that the ICSS and forced swim test procedures may reflect two unique aspects of depression-like behavior – respectively, anhedonia and behavioral despair – that may not always correlate. Interestingly, chronic variable stress also decreased immobility in the forced swim test, an effect that was attributed to increased impulsivity (52, 53). Moreover, impulsivity correlates strongly with anhedonia and avolition in depressive disorders (54). Thus, increased impulsivity in susceptible rats may have contributed to decreased immobility, although this speculation requires further investigation.

Social stress is a ubiquitous construct that may or may not have detrimental consequences depending on individual characteristics. Thus, investigating the factors that underlie susceptibility and resilience to stress may provide a more promising means of understanding stress-induced psychopathology compared to investigating stress per se (55). Similarly, low response and remission rates of currently available antidepressant medications suggest that investigating the factors that underlie treatment response and resistance using procedures and methods of analyses as described above may facilitate the development of more effective treatment strategies than those currently available.

Supplementary Material