Here, we found that prolonged social isolation of adult animals caused reward-related deficits and an anxiety-like phenotype. Both abnormalities were reversed by chronic, but not acute, imipramine treatment. These findings establish adult social isolation as an animal model of depression- and anxiety-like behavior that responds uniquely to chronic antidepressant administration. The generation of animal models of mood disorders that respond, similar to the human conditions, to chronic antidepressant treatment has been a major goal for this field of research1,35,36
. Recent research has shown that novelty-suppressed feeding37
and social defeat3
also show unique responses to chronic administration of antidepressants and both of these models involve a combination of depression- and anxiety-like symptoms as well. Social isolation, as described here, represents an additional model; one that is uniquely related to passive as opposed to active stress and is therefore potentially related to subtypes of human depression and anxiety syndromes that are related to isolation. Use of these various models may lead to better therapeutics for the different subtypes of depression and anxiety caused by active versus passive forms of stress.
The results of our study provide information about the molecular basis of the anxiety-like symptoms induced by social isolation. We found that social isolation decreased CREB activity in the NAcSh, an effect that was reversed by chronic imipramine, and that reduced levels of CREB in social isolation mediated the anxiety-like, but not the anhedonia-like, behaviors seen in the isolation model. Thus, overexpression of CREB in the NAcSh of isolated animals reverses only the anxiety-like phenotype and does not affect the deficits seen in reward-related behavior. Conversely, overexpression of the dominant-negative mutant mCREB in non-isolated animals caused only anxiety-like behavior and did not result in anhedonia-like symptoms. In fact, previous work has shown that mCREB overexpression in the NAcSh of non-isolated animals increases responses to rewarding stimuli and acts similar to an antidepressant, whereas CREB over-expression decreases reward responses and induces depression-like behaviors2,16,17,20
. In addition, these overexpression systems modulate anxiety-like behavior in non-isolated animals, with mCREB increasing anxiety and CREB decreasing it2
. Our working hypothesis is that CREB activity in the NAcSh during social isolation is downregulated to off-set depressive-like molecular mechanisms; however, this decrease in CREB in the NAcSh comes at the cost of a profound anxiety-like effect. Furthermore, our findings support a model in which social isolation decreases CREB activity in the NAcSh, which mediates the increase in anxiety-like behavior, whereas the restoration of CREB activity induced by chronic antidepressant treatment reverses those anxiety symptoms seen in social isolation.
In contrast, the anhedonia-like phenotype induced by social isolation, which is also reversed by antidepressant treatment, is not mediated by reduced CREB activity in the NAcSh and is presumably mediated by other molecular pathways in this brain region or by other neural circuits. For example, antidepressants have been shown to increase CREB activity in the hippocampus20,38
, with CREB over-expression in this region producing antidepressant-like effects39
. Therefore, perhaps the anhedonia-like phenotype and the reversal of depressive-like behaviors by imipramine that are observed in the social isolation model are achieved in part through a hippocampal CREB-dependent mechanism. CREB function is also implicated in regulating emotional behavior in several additional circuits, such as amygdala and prefrontal cortex15,20,40
, which further underscores the complex mechanisms that are probably involved in anxiety and depression syndromes.
Although we interpreted the increase in ejaculation latency observed in the social isolation model as a deficit in sexual reward, it has not been previously reported as a depression-like phenotype per se
. However, other animal models that produce depression-like phenotypes, such as lipopolysaccharide injections or methylphenidate treatment as pups, also increase ejaculation latency24,25
. Notably, we found that chronic imipramine administration to non-isolated control animals increased ejaculation latency, similar to the phenotype seen in drug-free isolated animals. This is consistent with both human reports and rodent findings of sexual side effects of antidepressant medications in normal animals and nondepressed humans on antidepressant medications21–23
. Together, our findings parallel differences in antidepressant responses seen in normal humans and those with some stress-related disorder. Thus, antidepressants do not produce mood elevating or anxiolytic effects in normal people and sexual deficits caused by depression or anxiety can be alleviated by these treatments in some affected individuals. Likewise, chronic imipramine decreased CREB activity in the NAcSh of normal animals, but increased CREB activity in socially isolated animals. It will be interesting to learn in future studies the molecular basis by which chronic imipramine treatment has opposite effects on CREB activity in the NAcSh, depending on the behavioral history of the animal, as such information could provide insight into the unique effects of long-term antidepressant administration in human populations.
Our microarray findings also led to the investigation of the electrophysiological properties of NAcSh neurons after social isolation. Prolonged isolation increased expression of seven K+
channels, including five voltage-gated K+
channels (Kcna1, Kcnd2 and3, Kcnq3 and Kcns2), an inwardly-rectifying K+
channel (Kcnj4) and a large-conductance calcium-activated K+
channel (Kcnma1). Functionally, these K+
channels contribute to action potential shape (Kcna1, Kcnd3 and Kcnq3), membrane potential hyperpolarization (Kcnj4) and neuronal excitability (Kcnma1). For example, Kcnma1 (also known as SLO1 BK channel) is important in mediating neuronal intrinsic excitability41,42
. We found it interesting to note that adaptive changes at the channel level are primarily mediated by K+
channels. No other channels were changed by social isolation with the exception of one Cl−
channel. This is similar to findings from the social defeat stress model, in which four K+
channels were upregulated by social defeat in the ventral tegmental area and regulation was not found for any other channels26
. These findings may suggest that K+
channels are the major ion channels mediating neuronal plasticity in the brain in both active and passive stress models.
On the basis of the upregulation of several K+
channels in this brain region in socially isolated animals and the reversal of this upregulation by either CREB or imipramine, we hypothesized that dampening the electrical excitability of NAcSh neurons may be a crucial step in the molecular events by which social isolation induces anxiety-like, but not anhedonia-like, symptoms. Indeed, we found that social isolation reduced the excitability of NAcSh neurons, consistent with an upregulation of K+
channel expression. These findings are consistent with our recent demonstration that downregulation of CREB per se
reduces NAc neuronal excitability34
. Although we did not observe an effect of CREB downregulation on resting membrane potential in this earlier study, this could be a result of the fact that the prior study examined NAc neurons from early postnatal animals and examined short-term effects of CREB downregulation, whereas the social isolation model involved adult animals and prolonged inhibition of CREB activity. Notably, we also found that chronic antidepressant treatment restored the observed deficits in both neuronal excitability and resting membrane potential that were induced by social isolation. This is one of the first studies to investigate electrophysiological mechanisms underlying deficits produced in adulthood social isolation and their reversal by chronic imipramine treatment.
By artificially mimicking the effect of reduced excitability via viral-mediated overexpression of a K+ channel in this region in non-isolated animals, we were able to mimic the increase in anxiety-like, but not anhedonia-like, symptoms induced by social isolation. We then found that downregulation of CREB activity in the NAcSh via overexpression of mCREB was sufficient to increase the expression levels of some of these K+ channels that are upregulated by social isolation. Although this is presumably not the only pathway underlying these behavior abnormalities, our results suggest a molecular pathological pathway by which social isolation, via downregulation of CREB activity and upregulation of K+ channel expression in the NAcSh, induces anxiety-like symptoms. This is of particular interest because CREB is traditionally viewed as a transcriptional activator. In this case, however, it would appear that CREB is acting as a repressor of certain genes, and when its expression was decreased, by either social isolation or mCREB overexpression, an increase in particular K+ channels was observed. Further work is needed to determine whether CREB serves as a direct transcriptional repressor for these genes or regulates them indirectly via other transcription factors. In addition, our microarray analysis of isolated and control animals under CREB and chronic antidepressant conditions identified numerous molecular pathways that may also contribute to the behavioral phenomena observed here and can be investigated in future studies.
In summary, our data provide new insight into the role of CREB in the NAcSh as a crucial regulator of responses to emotional stimuli (Supplementary Fig. 2
online). Prior work has demonstrated that drugs of abuse and active stress induce CREB activity in this region and that this ‘high CREB’ state is associated with a general blunting of emotional responding, including anhedonia-like symptoms and reduced anxiety-like behavior. In contrast, prolonged social isolation during adulthood decreases CREB activity in the NAcSh and this ‘low CREB’ state is associated with emotional hyper-reactivity, including profound anxiety. This role for CREB in the NAcSh as an ‘emotional rheostat’ is consistent with CREB’s role as an electrophysiological rheostat of NAc neuronal excitability, with increased CREB increasing NAc excitability and decreased CREB decreasing it, as has been shown both previously34
and here. Notably, chronic administration of imipramine normalized CREB activity at both extremes: preventing an increase in CREB activity from active stress (data not shown) and increasing it when in the low state (as shown here). In this way, we hypothesize that antidepressants may reverse anhedonia-like symptoms by preventing the induction of CREB activity in the NAcSh during periods of active stress and reversing anxiety-like symptoms when CREB activity is low. Clearly, the situation is far more complex, with anxiety and depression symptoms, as well as sexual behavior, being regulated by numerous molecular mediators not only in the NAc, but in several other brain regions as well1,15,20,27,28,38,39,43–45
. Nonetheless, our findings further characterize the long-term sequelae of social isolation and provide insight into the neural and molecular mechanisms that distinguish anxiety and depression symptoms in a chronic stress model.