Influence of housing on the consequences of chronic mild stress in female rats

Abstract

The chronic mild stress (CMS) paradigm was developed to model anhedonia in animals. The repeated administration of a series of unpredictable, mild stressors attempts to mimic the daily stress associated with the onset of clinical depression in humans. Male animals are predominantly used in these investigations despite significant, well-documented sex differences in human depression. In this study, the CMS procedure was modified to be more ecologically relevant to female animals. The effects of stress on sucrose preference, social interaction, rate of weight gain, and regularity of the estrous cycle in female Sprague–Dawley (SD) rats were evaluated in both single- and paired-housed rats, during 3 weeks each of baseline, CMS, and post-CMS phases. The results indicate that only single-housed rats exposed to stressors have a reduced rate of weight gain, significantly attenuated sucrose preference levels, and increased social interaction scores during the CMS phase of the study. Housing condition more than exposure to stress appeared to contribute to the disruption of estrous cycling in some animals. These data suggest that housing affords some protection from the negative consequences of CMS, at least in female rats, and that lack of social interaction in the single-housing condition may render females more vulnerable to stress-related illnesses. The development of paradigms that model human depression should emphasize sex-specific differences.

• Anxiety
• chronic mild stress
• depression
• sucrose intake and preference
• social interaction test
• anhedonia

Introduction

In psychiatric disorders with stress and anxiety components, such as depression, there is a notable sex difference in the incidence, with females outnumbering males by two to one (APA 1994). This is primarily evident during the reproductive years, and the occurrence of depression increases dramatically between menses and menopause and declines rapidly thereafter (Wade et al. 2002). Animal studies designed to model depression have relied almost exclusively on the responses of male subjects or ovariectomized female animals in order to avoid the potential confound of hormone fluctuations on the measure of interest. While there are many social and environmental factors likely at play, the finding that sex differences in depression rates are quite stable across cultures suggest that biological differences, such as hormonal fluctuation, contribute significantly to this phenomenon (Gater et al. 1998). Intact female animals are, therefore, a necessary component of depression research.

The chronic mild stress (CMS) model is a validated and widely employed method of inducing a depressive state in rats similar to anhedonia, the core feature of the melancholic subtype of major depressive disorder (APA 1994). Anhedonia is defined as the inability to derive pleasure from events that in a non-depressed state would be enjoyed, and it is interpreted primarily from behavioural measures, such as the place preference test (Benelli et al. 1999a,b), rewarding brain stimulation (Morreau et al. 1992), and the consumption of a mildly sweet sucrose solution (Willner et al. 1987), with the latter being the most frequently employed.

Despite its widespread use, there is no convention for administering the sucrose test or determining hedonic status on the basis of its results. We have adapted a method similar to that described by Dunčko et al. (2003). It consists of a two bottle (0 and 1% sucrose) 3 h test, preceded by overnight water deprivation, from which both sucrose preference and intake are evaluated.

Although the CMS model has been validated in male rodents, reliability has been a problem. While many laboratories have been able to elicit the pattern that is indicative of anhedonia, that is, a significant decrease in consumption (Cheeta et al. 1994; Kim et al. 2003; Grønli et al. 2004) and preference (Willner et al. 1987; Dunčko et al. 2001a,b) of a 1% sucrose solution after CMS exposure, other investigators have been unable to replicate this finding (Matthews et al. 1995; Neilsen et al. 2000; Murison and Hansen 2001; Konkle et al. 2003). Dunčko et al. (2003) have suggested that such paradoxical results may reflect individual differences in susceptibility to stress within strain. For example, his group found about 30% of rats to be anhedonic at baseline, using a criterion of 60% sucrose preference to categorize rats as either hedonic (above 60%) or not.

While the results of behavioural tests are not always dependable in this context, metabolic, biochemical, and endocrine responses to chronic mild stressors are generally consistent with the expectations of a paradigm that models human depression (Harris et al. 1998; Dunčko et al. 2001a,b; Bielajew et al. 2002; Konkle et al. 2003). For example, in animals CMS exposure has been reported to cause sleep disturbances (Morreau et al. 1995; Grønli et al. 2004), cardiovascular changes (Grippo et al. 2002, 2003, 2004), decreased sexual behaviour (D'Aquila et al. 1994; Brotto et al. 2001), decreased locomotor activity (Harro et al. 1999), immune disturbances (Kubera et al. 2001; Siberman et al. 2004), and increased corticosterone secretion (Harris et al. 1998; Bielajew et al. 2002), all of which have been observed in the clinically depressed population.

Our laboratory has been generally unsuccessful in establishing sucrose intake or preference deficits following weeks of stressor challenge in male rats. We have, however, had limited success applying these procedures to female rats, more so with the Sprague–Dawley (S-D) than Long Evans strain. Recently, using a modification of the scheme of Dunčko et al. (2003) to identify what we termed reactive (significant difference in sucrose intake between baseline and CMS phases) and non-reactive (no significant difference between the two phases) female rats, we obtained the same strain difference pattern (Baker et al. 2006). Thus, S-D was the strain of choice in the current study.

Given the dearth of studies employing female rats as subjects in CMS designs, the adequacy of the paradigm as a model for studying depression in women has not been fully explored. There is ample behavioural evidence to suggest that there are sex differences in stress reactivity. For example, male and female (intact) rats react differently in response to inescapable foot shock (Leuner et al. 2004), tail shock (Wood et al. 2001), and some behaviours in the forced swim test (Drossopoulou et al. 2004), with females overall appearing to be more vulnerable to the effects of stress encounters. One concern with this paradigm is that the stress manipulations may have sex-specific consequences. The original CMS procedure employed pairing as one of the stressors, which is known to cause distress in male, but not female rats (Westenbroek et al. 2003a). Indeed, regular pairing may even provide a protective effect against stress in females (Westenbroek et al. 2003a,b). To examine this issue, we included two housing conditions in this study, single and paired. Another concern is that stressors such as food and water deprivation (Anderson et al. 1996; Tropp and Markus 2001) and overnight illumination (Sharp and LaRegina 1998; Anisimov et al. 2004) have each been associated with estrous cycle disruptions. Consequently, these were replaced by short exposures to stroboscopic light and white noise, which are stressors that have been included in other CMS reports (Neilsen et al. 2000; Konkle et al. 2003; Grippo et al. 2004, 2005; Baker et al. 2006). We reasoned that these modifications would be more effective in inducing a state of anhedonia in female rats and allow us to identify more clearly stress and estrous cycle interactions. To ensure that these modifications did not simply elicit a general anxiety-like response, exclusive of anhedonia, the anxiogenic profile was assessed using a test of social interaction, a naturalistic animal model of anxiety (File and Hyde 1978; File 1980).

Thus, this study examines the effects of social housing in female rats on the development of anhedonia after exposure to chronic mild stressors. The measures assessed were sucrose preference, social interaction, rate of weight gain, and regularity of the estrous cycle. We hypothesized that the presence of a cage mate would attenuate the effects of stress.

Methods

The use of animals was in accordance with the guidelines of the Canadian Council on Animal Care. All protocols received institutional approval.

Animals

A total of 43 S-D female rats (Charles River Laboratories, St-Constant, QC, Canada), ranging in weight from 258–390 g (average 320 g) upon arrival, were used. They were immediately randomly assigned to one of four groups, based on housing and stress conditions. The groups comprised the following: single-housed control group (n = 11), single-housed stress group (n = 10), paired-housed control group (n = 10), and paired-housed stress group (n = 12). Single-housed rats were individually maintained in standard size plastic cages (24 cm wide, 43 cm long, 20 cm high) and paired rats in larger cages (37 cm wide, 48 cm long, 20 cm high). All rats had free access to food and water, with one exception. Before each sucrose test, water was removed overnight to encourage intake during the morning test (approximately 16 h of deprivation/test). All rats received enrichment objects in the form of a black PVC tube, a Kong^® Toy, and Nestlets^® in the home cage except during stress exposure. Access to such objects has previously been associated with a reduction in chemical markers of stress that tend to increase as a result of single housing in female rats (Belz et al. 2003). A 12 h light/12 h dark cycle with lights on at 7:00 a.m. was maintained throughout the study. The estrous cycle was monitored daily via vaginal lavage taken between 9 and 11 a.m. using methods described previously (Marcondes et al. 2002). Estrus cycles were considered to be regular if the pattern of stages followed the standard criteria outlined by Long and Evans (1922) and if cycle lengths of individual animals were consistent with pre-baseline monitoring. Body weight was recorded weekly.

Procedures

Chronic mild stress

Table I provides an outline of the study phases and stress schedule delivery and description.

Table I.  Twenty days schedule of chronic mild stressors.

Label	Stressor	Definition	Time/week (hours)
Confine	Confinement with noise	Rats are individually placed in mouse cages while times were set to ring at random intervals.	5
Strobe	Stroboscopic light	While in home cage, rats are transferred to a dark room and exposed to 300 flashes of white light/minute.	6–7
Noise	White noise	An untuned radio is turned to max volume.	5
Cage tilt	Tilted cage	Cage is tilted horizontally 30° overnight.	2 × 17
Wet bed	Wet bedding	Warm water is mixed into home cage bedding so that all bedding material is wet.	2 × 17

Following three weeks of baseline monitoring, the CMS procedure was applied for 3 weeks. During this time, all enrichment objects were removed from animals in the stress groups. The CMS phase was immediately followed by a 3-week period during which baseline conditions were reinstated. Enrichment items were returned to the home cage and rats were handled only for daily vaginal lavages, weekly weight recording, and cage changes.

Sucrose preference test

Following a procedure similar to that previously described by Dunčko et al. (2003), a 3-h sucrose test was conducted on each of the last 4 days of every phase (baseline, CMS, and post-CMS) following overnight water deprivation. All tests were administered in the home cage and paired rats were separated from each other by a metal barrier with holes (rats were acclimatized to this separation before the baseline weeks began). The rats were exposed to two test bottles, one containing tap water and the other, a 1% sucrose solution, for 3 h beginning at 8:30 a.m. At 11:30 a.m., the volume in each bottle was measured. Preference for sucrose was calculated by dividing the amount of sucrose solution consumed by the total intake of fluid and converting this value into a percentage. After baseline sucrose preference had been calculated, rats were assigned to experimental groups. Those displaying at least a 60% sucrose preference were placed in the stress groups and those displaying a preference below 60% were placed in the control groups.

Social interaction test

During the last 2 days of each phase, the social behaviour towards a non-familiar stimulus rat during a 5 min test of social interaction was filmed and later scored by an observer blind to treatment. A second individual re-scored half of the data in order to determine inter-rater reliability.

Seven female rats were used as stimulus animals (average weight at each test: Baseline-281 g, CMS-308 g, post-CMS-311 g); they did not belong to any experimental group and did not receive any prior treatment with the exception of daily handling and vaginal lavage. The test arena was a novel environment (25 cm wide, 44 cm long, 28 cm high) to which all animals were acclimatized for 15 min approximately 30 min before the test. The social interaction tests were conducted during the dark/active phase (between 7 and 10 p.m.) in a room lit by ambient red light. The duration spent engaging in active social behaviour with the stimulus rat and the frequency of amicable, agonistic and nonsocial behaviours were recorded so that each test could be categorized in terms of active social time and the primary behaviour expressed. Figure 1 provides a detailed summary of the behaviours of interest recorded in the social interaction test. These have been described previously (Kiyokawa et al. 2004).

Figure 1 This figure shows a breakdown of the social interaction behaviours that were recorded during the last week of each phase; the behaviours were categorized as social and non-social. Their constituents include the following: Freezing: immobile. Threat posturing: arched back, all limbs extended with flanks turned toward opponent. Leaping: Jumping in air and landing on opponent. Boxing: upright posture facing opponent. Allogrooming: grooming stimulus rat. Anogenital contact: sniffing or nosing the anal and or genital area of stimulus rat. Lordosis: coccygeal region raised and the tail to one side (position which permits intromission and is stimulated by pressure of male's forelimbs on the flanks during mounting). Darting: rat in estrus runs a short distance from conspecific and pauses for mounting and then repeats.

Statistical analyses

All analyses were conducted on the raw values and the CMS and post-CMS data expressed as mean differences ± SEM from baseline. The weekly body weight data were analysed (SPSS 2005) via a mixed ANOVA design with group as the independent factor (four levels: single-housed control, single-housed stress, paired-housed control and paired-housed stress) and time as the repeated factor (nine levels of week). The significant interaction was followed-up by pair-wise post-hoc tests in each group in order to assess phase differences. A 1-way randomized group ANOVA was used to assess baseline differences and weight gain overall across the course of the study. Pair-wise tests were then conducted to further delineate group differences.

For each phase, the average sucrose preference score, based on three tests, was calculated for each rat and these data were evaluated using a mixed ANOVA design (four groups and three phases). Significant interaction was followed up with pair-wise post-hoc tests in each group. Note that the average value was calculated on fewer than three tests in some animals, due to spillage. Baseline differences were evaluated as described above for the weight data.

The social interaction data were evaluated via a 3 × 4 mixed ANOVA design, with three levels of phase (baseline, CMS and post-CMS) and four levels of group (single- and paired- housed control and stress groups). The significant interaction was followed up by pair-wise post hoc tests in each group.

Huynh-Feldt correction for violations of the assumption of sphericity was applied as required (Howell 2002) to all repeated variables having more than two levels. The adjusted degrees of freedom based on this correction are reported in the results section.

The estrous cycle data were subjected to a chi-square test of independence (3 (phases) × 4 (groups) contingency table). These data were based on the number of rats that generally displayed regular cycles during baseline, CMS and post-CMS weeks.

The alpha level was set at 0.05 for all omnibus analyses and a Bonferroni correction applied to all follow-up tests.

Results

The body weight data are shown in Figure 2. The boxed plot shows the average weight gain in each group over the course of the study (9 weeks) and includes the 3 weeks each of baseline, CMS, and post-CMS phases. The pattern suggests that control rats demonstrated a greater rate of weight gain over time while in comparison, rats in the stress groups gained less weight (paired-housed) or showed negligible weight gain (single-housed) during the CMS phase. Weight gain over the course of the study was significant across groups (F(3, 39) = 5.273, p = 0.004) due to pair-wise differences between the single-stressed and each paired group (p = 0.037, to paired stressed; p = 0.005, to paired control).

Figure 2 The difference in body weight from baseline at the CMS and post-CMS phases. Data are group means ± SEM. *p < 0.01 vs. baseline; §p < 0.01 vs. CMS. Within housing condition, significant differences between control and stressed groups are denoted by the symbol †. The boxed graph shows the percentage of average weight gain over the course of the study for each group. The * above each paired group indicates that they were both significantly different from the single-stressed group. The sample size was 11, 10, 10, and 12 rats in the single-housed control, single-housed stressed, paired-housed control, and paired-housed stressed groups, respectively.

The control and stress groups in each housing condition had similar baseline weights; however, the housing manipulation itself altered the baseline weights in that rats in the paired-housing condition had significantly lower body weights (F(1, 41) = 18.35, p = 0.0001), either due to higher activity levels or lower food intake than that associated with single-housed groups. The average group weight, as a function of housing condition was 280.5 ± 3.12 g in paired-housed rats and 315.5 ± 7.7 g in single-housed rats.

The mixed ANOVA conducted on the raw body weight values produced significant main effects of group (F(3,39) = 3.022, p = 0.041) and time (F(1.9, 75.6) = 40.621, p = 0.0001) and the interaction between the two (F(5.8, 75.6) = 5.426, p = 0.0001). Pair-wise post-hoc tests to examine differences in weight between phases in each group revealed no significant difference between any pair in the single-housed stressed group indicating negligible weight change in this group over the course of the study. In the remaining groups, CMS and post-CMS weight were significantly different from baseline weight (p < 0.0001). Furthermore, in the paired-housed groups, post-CMS weight was significantly greater than CMS weight (p < 0.0001).

Within housing conditions, there were significant differences from baseline in weight gain during the CMS phase between the control and stressed groups; this applies to both the single (p = 0.013) and paired-housed (p = 0.001) groups. At the post-CMS phase, these differences were not significant.

Figure 3 shows the average sucrose preference data (plotted as differences from baseline) for each group. The baseline data are shown at the bottom of the figure. Note that rats with low sucrose preference were assigned to the control groups and those with high preference, the stressed groups, thus forcing baseline group differences. An analysis of the baseline data produced an overall significant difference (F(3, 39) = 10.670, p = 0.0001). Pair-wise differences were found between the single-stressed and all other groups (p = 0.006, single control; p = 0.0001, paired control; p = 0.044, paired stressed). In addition, the comparison of the paired groups was significant (p = 0.028).

Figure 3 The percent difference in sucrose preference from baseline at CMS and post-CMS phases. The boxed graph shows the baseline values for each group. Data are group means ± SEM. Significant differences from baseline are indicated by an * above the relevant group. The single stressed group was significantly different from all other groups (*) and the paired stressed group was significantly different from its control counterpart (§) at baseline. The sample size was 11, 10, 10 and 12 rats in the single-housed control, single-housed stressed, paired-housed control, and paired-housed stressed groups, respectively.

The full analysis produced a significant interaction between group and phase (F(6, 54) = 2.7, p = 0.023). Follow-up post hoc tests indicated a difference in the preference pattern over phase in the single-housed stress group and paired-housed control groups. That is, the average preference during CMS was significantly reduced relative to its baseline value in the single-housed stressed group (p = 0.014). In contrast, the paired-housed control group showed elevated preference scores during both CMS (p = 0.007) and post-CMS phases (p = 0.007). The single-housed control and paired-housed stress groups maintained a fairly consistent level of preference from phase to phase.

The results of the social interaction test are shown in Figure 4 for each group with baseline data shown at the bottom of the figure. The inter-rater reliability, assessed using Pearson's r correlation, was 0.82 (p = 0.0001). The mixed ANOVA conducted on these data gave rise to significant main effects of group (F(2, 74) = 4.57, p = 0.013), phase (F(6, 74) = 3.80, p = 0.002), and their interaction (F(3, 74) = 5.12, p = 0.005). A significant pair-wise difference was found between baseline and post-CMS scores in the single-housed control group only (p < 0.0001); in this group, social interaction scores were also significantly elevated in the post-CMS phase compared to CMS values (p = 0.037). The behavioural phenotype did not differ between groups at any phase of the study, and no correlations were found between estrous cycle phase and interaction time, or test and stimulus animal weight.

Figure 4 The difference from baseline time spent in active social interaction at CMS and post-CMS phases. The boxed graph shows the baseline data for each group. Data are group means ± SEM. *p < 0.01 vs. baseline; §p < 0.01 vs. CMS. The sample size was 11, 10, 10 and 12 rats in the single-housed control, single-housed stressed, paired-housed control, and paired-housed stressed groups, respectively.

Figure 5 shows the percentage of rats in each group that displayed regular cycles. The units refer to decreases from baseline. The smallest change in this value occurred in the paired-housed control group. Irregular cycles were characterized by extended estrus or diestrus days, or an abnormal pattern of cyclicity. The chi-square test of independence that was performed on these data (all groups versus the three phases) was not significant. Goodness of fit tests to examine the effects of housing alone approached significance (p = 0.083), while the same test applied to the stressed versus non-stressed groups did not (p = 0.879).

Figure 5 The difference from baseline in the percentage of rats displaying regular estrous cycles at the CMS and post-CMS phases. Increasing values represent decreasing cyclicity. The boxed graph shows the baseline data for each group. The sample size was 11, 10, 10 and 12 rats in the single-housed control, single-housed stressed, paired-housed control, and paired-housed stressed groups, respectively.

Discussion

The present study was designed to evaluate the responses of female rats in different housing conditions to a regime of chronic mild stressors, excluding ones that independently influence the estrous cycle. The consequences of CMS were assessed via behavioural (sucrose preference and social interaction) and physiological (weight and estrous cycle regularity) measures. The rats were assigned to stress and non-stress groups at the baseline phase on the basis of sucrose preference. This was necessary to ensure that baseline preference levels in the stressed groups were sufficiently high to allow observable reductions following experimental manipulations but not so low as to suffer from floor effects.

The CMS procedure induced a mild anhedonic state only in the single-housed stressed rats as interpreted from the sucrose preference data. The paired-housed control group showed a surprising increase in sucrose preference during both the CMS and post-CMS phases. This might have been because some rats require longer periods of acclimatization to sucrose. Nonetheless, given the intentional group differences in baseline preference scores, the appropriate comparisons were within-group. Three weeks of CMS exposure reduced the preference for sucrose by 19.6% in the single-housed stressed group. During the post-CMS phase, when mild environmental enrichment was reinstated, sucrose preference values returned nearer to baseline levels. This overall pattern is typically found in studies of this nature (Willner et al. 1987; Dunčko et al. 2001a,b; Genedani et al. 2001; Grippo et al. 2002, 2004). A difference in sucrose preference between intact and ovariectomized female rats has been reported by one group who found a decrease in preference after CMS exposure in intact rats only (Dunčko et al. 2001, b), suggesting that hormonally unaltered females are more sensitive to the effects of CMS. Few CMS studies employ intact female rats, and of those that have, sucrose intake rather than preference was typically measured and shown to decrease after CMS exposure (Benelli et al. 1999a,b; Dunčko et al. 2001a,b; Konkle et al. 2003; Baker et al. 2006). A study employing a four week CMS regime demonstrated a decrease in both sucrose intake and preference in male and intact female rats (Grippo et al. 2004).

Similar to the sucrose data, the single-housed stressed rats showed the lowest rate of weight gain over the course of the study. In male rats, reduced weight gain following CMS has been frequently observed (Matthews et al. 1995; Dunčko et al. 2001, b; Neilsen et al. 2001; Bielajew et al. 2002). Our laboratory had previously reported similar findings in intact females (Konkle et al. 2003); however, others have also observed no change in weight gain in this context (Murison and Hansen 2001; Dunčko et al. 2001a).

Data pertaining to the estrous cycle were most surprising. Regardless of group, all rats showed a reduction in regular cycling following the baseline phase, an effect that persisted into the third week of the post-CMS phase. In our previous CMS studies using female rats, we observed a large proportion of acyclic rats after exposure to CMS (Konkle et al. 2003). Grippo et al. 2004 reported similar findings: the cycles of animals exposed to 4 weeks of CMS manipulations experienced a 40% lengthening of the estrous cycle relative to control groups. Although the stressors may influence estrous cycle regularity, the possibility that food and water deprivation or overnight illumination contribute to irregular cycles is equally viable (Sharp and La Regina 1998; Tropp and Markus 2001; Anisimov et al. 2004). In our study the modified CMS regime, which did not include these particular stressors, resulted in no difference between control and stressed rats in terms of the degree of cycle disruption. The graph (Figure 5) that shows plots of the estrous pattern in individual groups suggests that cycle regularity was more influenced by the housing condition than by the stressors. Almost all control rats in the paired-housed condition displayed regular cycles throughout the study, and the application of stressors was associated with some reduction in cyclicity in rats housed together. On the other hand, both groups of single-housed rats showed the greatest cycle disruptions during and post-CMS. We (Konkle et al. 2003; Baker et al. 2006) have observed abnormal cycle patterns in a high proportion of individually housed rats; in contrast, almost 100% of our group-housed rats displayed regular cycles over months of observation (Konkle et al. 2003).

Responses of all rats in the social interaction test suggested that CMS did not induce a general anxiety response as no group displayed decreased social interaction after exposure to the paradigm. This is consistent with responses of male rats (D'Aquila et al. 1994). A housing difference did emerge in our case, however. Regardless of stress experience, the paired rats maintained consistent times across phase (all within a range of 10 s). The pattern in Figure 4 suggests that single-housed rats were generally more variable in the time engaged in social interaction compared to their baseline performance level and that of paired rats during the CMS and post-CMS phases. Unexpectedly, the single-housed control rats displayed a significant spike in interaction during the post-CMS phase. None of these observations can be attributed to weight or cycle phase differences between the housing groups as neither variable correlated with social interaction time. It is possible that the manner in which the test was administered did not elicit much anxiety because the rats had been acclimatized to the test box prior to baseline tests, and tests were conducted in the dark; both of these factors have been found to increase exploration and social interaction (File and Hyde 1978). Furthermore, the social deprivation experienced by single-housed rats may have made them more motivated to seek out social interaction with the stimulus rat than their paired-housed counterparts, especially during subsequent tests when the presentation of an unfamiliar rat was less novel.

The social nature of rats naturally makes standard, single-housing practices highly stressful, particularly for female rats that display little to no aggression towards conspecifics regardless of familiarity. Housing female rats in pairs or groups tends to diminish stress effects that are normally observed in single-housed rats (Haller et al. 1999; Konkle et al. 2003; Belz et al. 2003; Westenbroek et al. 2003a,b). Presumably this is due to the lack of both social interaction, and the ability to display naturalistic behaviour in the impoverished environment of standard laboratory husbandry. Providing stimulating objects has been found to decrease hormonal indicators of stress in individually housed male rats (Belz et al. 2003). In our case, both single- and paired-housed groups were provided items in their home cage in an attempt to observe the effects of social housing more clearly. While the enrichment objects may have diminished some stress induced by living in typical ‘sterile’ housing conditions, they did not appear to alleviate the behavioural outcomes (notably sucrose preference) of stress associated with isolation or the exposure to CMS in this group. Indeed, the concept of social buffering, whereby anxiety-provoking stimuli elicit a diminished behavioural and hormonal response in animals that are kept in social groups (Kiyokawa et al. 2004), could be applied to explain the lack of CMS effect in behavioural tests in paired rats. This is consistent with clinical findings: the level of perceived social support is negatively related to the presence and severity of depressive symptoms (George et al. 1989).

Our modified version of the CMS paradigm, which excludes stressors that impact on the reproductive cycle, induces behavioural and body weight changes indicative of an anhedonic state but not anxiety. However, like humans, social support, via paired-housing, appears to provide some protection against the influence of the stressors, suggesting that single-housing, an unnatural living condition in female rats, induces a vulnerability to environmental stressors. These results highlight the importance of tailoring research tools to address gender differences and husbandry practices in animal models of anxiety and stress.

Acknowledgements

We are grateful to the Natural Sciences and Engineering Research Council of Canada for its support of this project.