Differential paraventricular nucleus activation and behavioral responses to social isolation in prairie voles following environmental enrichment with and without physical exercise

Abstract

Social stressors produce neurobiological and emotional consequences in social species. Environmental interventions, such as environmental enrichment and exercise, may modulate physiological and behavioral stress responses. The present study investigated the benefits of environmental enrichment and exercise against social stress in the socially monogamous prairie vole. Female prairie voles remained paired with a sibling (control) or were isolated from a sibling for 4 weeks. The isolated groups were separated into isolated sedentary, isolated with environmental enrichment, and isolated with both enrichment and exercise conditions. Behaviors related to depression, anxiety, and sociality were investigated using the forced swim test (FST), elevated plus maze (EPM), and a social crowding stressor (SCS), respectively. cFos expression was evaluated in stress-related circuitry following the SCS. Both enrichment and enrichment with exercise protected against depression-relevant behaviors in the FST and social behavioral disruptions in the SCS, but only enrichment with exercise protected against anxiety-related behaviors in the EPM and altered cFos expression in the hypothalamic paraventricular nucleus in isolated prairie voles. Enrichment may alleviate emotion-related and social behaviors, however physical exercise may be an important component of environmental strategies for protecting against anxiety-related behaviors and reducing neural activation as a function of social stress.

Introduction

As social animals, humans benefit from the development and maintenance of social bonds. Companionship in the form of a friend, relative, or significant other can promote both psychological and physiological health. In contrast, social stressors such as loneliness and social isolation have negative consequences on physical health, psychological and emotional well-being, neurological function, and behaviors (Donovan et al., 2017; Richard et al., 2017). For instance, feelings of loneliness accelerate cognitive decline (Donovan et al., 2017), decrease health-promoting behaviors, and increase physical and mental health problems (Richard et al., 2017). On a neuroanatomical level, social stressors elicit changes in stress-related circuitry, such as activation of the hypothalamic-pituitary-adrenal (HPA) axis, amygdala, and hindbrain regions (Cacioppo et al., 2015; Kamal et al., 2014). This neural activation is associated with alterations in other stress-relevant brain regions such as the hippocampus and prefrontal cortex, along with peripheral systems such as the cardiovascular, endocrine, and digestive systems (Chrousos, 2009).

Studies involving animal models demonstrate that psychosocial stressors disrupt emotion-related behaviors, impair cardiovascular function, and alter stress reactivity (Carnevali et al., 2012; Finnell et al., 2017; Koolhaus et al., 2013; Mumtaz et al., 2018; Shively et al, 2009; Takatsu-Coleman et al., 2013). One valuable rodent model for investigating these interactions is the socially monogamous prairie vole. This rodent species displays a unique social structure by maintaining long-term opposite-sex pair bonds and family bonds (Carter, et al., 1995; Getz et al., 1993; McGuire & Getz, 1991), and experiences stress responses and behavioral disturbances similar to humans following the disruption of social bonds (Lieberwirth et al., 2012; Sun et al., 2014). For example, social isolation and the disruption of established social bonds in prairie voles are associated with depression- and anxiety-related behaviors, altered cognition, disrupted sexual behavior, elevated HPA axis functions, impaired autonomic regulation of the heart, and altered function in central stress circuitry (Bosch et al., 2009; Grippo et al., 2007b; McNeal et al., 2014, 2019; Pohl et al., 2019). These translational characteristics make the prairie vole a valuable model for investigating the impact of social environmental disruptions on behaviors, physiological systems, and stress-relevant neurobiological processes.

Given the influence of social stressors on psychological and physical well-being, it is critical to investigate potential prevention, mitigation, and treatment strategies against the negative consequences of social stress. Engaging in physically- and cognitively-stimulating activities improves both behavior and physiology, for example by promoting healthy aging, preventing neurodegenerative diseases, and improving hippocampal health following traumatic brain injury (Milgram et al., 2006; Miller et al., 2013). These environmental strategies may also protect against consequences of psychosocial stressors such as loneliness and depression (Schloesser et al., 2010). Animal studies have similarly employed environmental enrichment (EE) paradigms involving cognitively- and physically-enriching activities. EE in rodents often includes items that provide tactile stimulation, physical exercise, and/or increased social exposure (Crofton et al., 2015; Mesa-Gresa et al., 2013). Similar to human studies, research with animal models supports the use of enriching activities to protect against the effects of stress on several neurobiological functions, including amygdala function (Koe et al., 2016), peptide function in the hypothalamus (Costa et al., 2021), cognitive impairments (Morse et al., 2015), and dysfunction due to seizure disorders (Fares et al., 2013). Exposure to EE in rodents of various ages also improves spatial memory, stress reactivity, depression- and anxiety-relevant behaviors, ethanol-induced reward responses, and cardiovascular function in rodents, including prairie voles exposed to social stressors (Bahi, 2017; Costa et al., 2021; Fares et al., 2013; Grippo et al., 2014; Mora-Gallegos and Fornaguera, 2019; Normann et al., 2018). EE promotes hippocampal neural plasticity following exposure to social stress (Schloesser et al., 2010), suggesting that EE has neuroprotective effects against some of the negative consequences of social stress.

Previous studies have reported numerous benefits of EE, which may be attributed to the complexity of activities available in various EE paradigms. However, EE paradigms often include an exercise component, which may complicate the interpretation of the benefits of EE. Indeed, investigations of the benefits of exercise alone have revealed results comparable to those of EE on several psychological and physiological outcome measures. In humans, exercise is a well-established paradigm to improve cardiovascular fitness (Gielen et al., 2001), immune health (Jonsdottir, 2000), and anxiety and depression symptoms (Byrne & Byrne, 1993; Salmon, 2001). Research in rodent models indicates that exercise acts on the HPA axis and adrenal glands to attenuate the stress response to a psychological stressor (novelty) more so than to a physical stressor (forced swimming) (Droste et al., 2007). Further, exercise protects against depression- and anxiety-related behaviors in mice (Duman et al., 2008) and prairie voles (Grippo et al., 2014; Watanasryiakul et al., 2018). Some behavioral benefits of physical exercise on stress reactivity may be long-lasting. For instance, exercise-induced improvements in escape behaviors during a shuttle box stressor persist for two weeks following the removal of exercise access in rats (Greenwood et al., 2012). The influence of exercise on behaviors and physiological functions may be mediated by several neural mechanisms. For example, exercise promotes neurogenesis and cell survival, alters HPA axis activity, and prevents dysfunction in multiple stress-relevant brain regions (Hostinar et al., 2014; Klaissle et al., 2012; Koe et al., 2016; Lin et al., 2015; Olson et al., 2006; Sanders et al., 2019; van Praag et al., 1999; Zheng et al, 2012).

Although previous studies support exercise as a useful strategy for combatting the consequences of stress, it is important to address the issue that access or ability to exercise may be limited for certain populations. Individuals with physical limitations due to advanced age or disability, or financial or logistical limitations due to socioeconomic status, may be more vulnerable to the negative effects of social stress than the general population, and may be unable to exercise at levels that provide the benefits previously described. Therefore, it is important to determine whether exercise is a necessary component of EE paradigms to protect against social stress. The present study was designed to address whether the exercise component of an EE paradigm is responsible for behavioral and neurobiological benefits of EE against social stress in prairie voles, or whether an EE design without an exercise component would also produce benefits on behavioral and neurobiological outcomes. To investigate these research questions, behaviors of socially isolated prairie voles were compared in operational measures of depression, anxiety, and social interactions following either a full EE paradigm with an exercise component or an EE paradigm without an exercise component. Neural activation also was investigated in stress-relevant circuitry following a social behavioral task. This research design allowed for a direct comparison between the two EE paradigms relative to social isolation alone and a socially paired control condition in prairie voles. Specifically, it was hypothesized that access to exercise would provide more robust behavioral and neurobiological benefits against social isolation (relative to the lack of access to exercise), but that EE without an exercise component would also display some protective effects against behavior and neurobiological consequences of social stress. The results of this study may inform treatment and prevention strategies against social stressors for individuals for whom physical exercise is not a feasible or desirable option.

Materials and Methods

Animals

Forty-four pairs of adult (60–90 days) female prairie voles (25–50 grams), descendants of a wild stock caught from United States prairielands, were maintained on a 14/10 hour light/dark cycle (lights on at 0630 h) prior to and during the present study. The housing rooms were maintained at 20–23°C and relative humidity of 40–50%. Animals were allowed ad libitum access to food (Purina rabbit chow) and water. Offspring were weaned from the family group at 21 days of age and housed in same-sex sibling pairs until the beginning of the experiments during adulthood. One animal from each sibling pair was studied. Handling, cage changing, and measuring of body weight were standardized across the experimental timeline and groups. All procedures were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals, approved by Northern Illinois University’s Institutional Animal Care and Use Committee, and followed all mandatory safety and health guidelines.

Females were selected as subjects for the present study for several reasons. First, the use of females allows for comparisons with previously published data from prairie voles focused on the interactions of EE and exercise with social stressors, behaviors, and neuroendocrine functioning, including studies that have previously focused on the direct comparison of EE with exercise alone (Grippo et al., 2014; Jarcho et al., 2019; Watanasriyakul et al., 2018; Watanasriyakul et al., 2019). Additionally, females remain an understudied population in animal models and can provide insight into sex-specific responses to social stressors (APA, 2013; Beery & Zucker, 2011; Prendergast, Onishi, & Zucker, 2014).

Housing Conditions

Prior to the beginning of the experiment, all animals were housed with a same-sex sibling. For the purpose of the study, one animal from each sibling pair was randomly assigned to one of four possible housing conditions for 4 weeks. All groups were equally provided with food, water, and cotton nesting material. The paired control group (n=10) remained housed with a same-sex sibling in a standard-sized cage (12×18×28cm) without any additional manipulation. All isolated animals were separated into three additional conditions, and were housed without visual, olfactory, or auditory cues from the respective siblings. The isolated sedentary group (n=10) was housed alone in a standard-sized cage (12×18×28cm) without any additional manipulations. The isolated enriched group (n=10) was housed alone in a larger cage (25×45×60cm) with access to a variety of enriching and stimulating objects (described further below). Finally, the isolated enriched-exercise group (n=13) was housed in a larger cage (25×45×60cm) with access to the same enrichment items as the isolated-enriched group, as well as with access to a running wheel for physical exercise.

Enrichment and Physical Exercise

Both isolated enriched and isolated enriched-exercise conditions received access to the following items 24 hours per day, for the entire 4-week isolation period: two small wooden cylinders, two small rubber dice, a square of woven straw, a cardboard toilet paper roll, a tin foil ball, a plastic bowl with small food pellets, two plastic toys, 2 marbles, and a small plastic igloo house. Animals in the isolated enriched-exercise condition were additionally provided with a running wheel (4in diameter; Super Pet Mouse Silent Spinner Mini Exercise Wheel, #100079369, Kaytee Products, Inc., Elk Grove Village, IL), equipped with a bike odometer (Bell F12 Cyclocomputer, #7001115, Bell Sports, Van Nuys, CA) using custom-designed procedures (Northern Illinois University, College of Liberal Arts and Sciences Technical Services Shop). The odometer recorded distance traveled (km/day) and maximum speed (km/h). All items were purchased from standard retail vendors, such as local pet stores and online vendors. Enrichment items were sanitized with a diluted bleach solution, or replaced, once per week when the cages were changed.

Behavioral Measures

Time Course and Test Order

Behavioral testing commenced 48 hours after the end of the isolation/pairing phase, with each behavioral test separated by 48 hours, in the following order for all groups: EPM, FST, SCS. Previous experiments and preliminary validation protocols from prairie voles indicate that: (a) these short-term behavioral tests can be used to successfully measure emotion-related and social behaviors without adverse effects; (b) the order of tests does not influence the outcome of subsequent behavioral measures or associated physiological or neural measures; and (c) a separation of at least 24 hours between tests is a sufficient amount of time to ensure a full recovery of behavioral and physiological reactivity after the test, without carry-over effects on additional behaviors, autonomic, cardiovascular, and neuroendocrine dependent measures (Grippo et al, 2007b; Grippo et al., 2008; Grippo et al., 2014; McNeal et al., 2017).

Elevated Plus Maze

The elevated plus maze (EPM) was conducted 48 hours after the 4-week isolation/pairing phase during the light period, as an operational index of anxiety-related behavior (Pellow et al., 1985; Walf & Frye, 2007). The maze apparatus was elevated 57cm off the ground and consisted of two open arms of clear Plexiglass (49.5×10cm), two opposite closed arms of black Plexiglass (49.5×10×30.5cm) with high enclosed walls without a roof, and a center square section of clear Plexiglass (10×10cm) connecting the arms in the shape of a plus sign. The animal was placed in the center square of the maze and allowed to freely explore the maze for five minutes in a brightly lit room (room, 348 lux; open arms, 261 lux; center section, 217 lux; closed arms, 22 lux; all values relative to calibration level of 0 lux with lights off). Each animal was returned to its home cage (in the previous experimental condition) immediately following the test. The maze was sanitized with a diluted bleach solution following each trial.

Each trial was recorded using a digital video camera for later offline analysis. Behaviors were manually scored by experimentally-blind observers, trained to an inter-rater reliability level of at least 90%, using the Observer XT v. 8.0 (Noldus Information Technology, Leesburg, VA) video analysis software. The following behaviors were recorded: (a) time spent in the closed arms; (b) time spent in the open arms; (c) time spent in the center; and (d) number of center crossings. The animal was considered to be in a specific zone of the maze once all four paws crossed into the area. A reduced amount of time spent in the open arms of the apparatus is hypothesized to represent an anxiety-like response, and the total number of crossings into the center section is used as a measure of general locomotor activity (Walf & Frye, 2007). The durations of each behavior were averaged between the raters.

Forced Swim Test

The forced swim test (FST) was conducted 48 hours after the EPM during the light period, as an operational index of helpless behavior in an inescapable task (Slattery & Cryan, 2012). The apparatus consisted of a clear Plexiglass cylinder (height 46cm; diameter 20cm) filled to a height of 18cm with clean, room temperature water (approximately 21–25°C). The animal was gently placed into the tank of water for a five-minute trial. After each trial the animal was removed from the water and replaced in the home cage (in the previous experimental condition). A corner of the cage - approximately < 25% of the total area of the cage - was placed under a heat lamp to help the wet animal thermoregulate for 10–15 minutes. The tank was sanitized with diluted bleach solution, and the water was replaced, after each trial.

Each trial was recorded using a digital video camera for later offline analysis. Video analysis was performed by at least two trained and experimentally-blind observers, trained to a level of at least 90% inter-rater reliability. The behaviors in the FST included measures of active versus passive responses in response to a short-term stressor in rodents, including prairie voles (Slattery & Cryan, 2012; Grippo et al., 2008; Grippo et al., 2012). In the present study, the behaviors were categorized as the following: (a) immobility (passive response), defined as floating without any limb movement or with just enough movement to remain afloat; (b) swimming (active response), defined as the animal moving its fore- and hindlimbs in a coordinated manner without breaking the surface of the water; (c) struggling (active response), defined as moving the forelimbs and breaking the surface of water in the middle of the tank; and (d) climbing (active response), defined as the animal scratching at, or attempting to climb, the walls of the apparatus. The durations of each behavior were averaged between the observers.

Social Crowding Stressor

The social crowding stressor (SCS) test was administered 48 hours following the FST. Each animal was exposed to a 10-minute period of social crowding, allowing for the investigation of social behaviors during an acutely crowded environment (Djordjevic et al., 2005; Grippo et al., 2010). The experimental animal was placed into a clean standard-sized cage (12×18×28cm) without bedding containing three unrelated and unfamiliar female prairie voles of approximately the same age, size, and weight. Following the test, the experimental animal was removed from the group, replaced in its home cage (in the previous housing condition), and left undisturbed for 2 hours.

Each trial was recorded using a digital video camera for subsequent offline behavioral analysis. Behaviors were scored by at least two trained and experimentally-blind observers, trained to a level of at least 90% inter-rater reliability. The behaviors of interest included the following: (a) crowd investigating, when all animals were engaged in investigative behaviors such as sniffing or grooming each other; (b) crowd sitting quietly, when the animals were not moving and were sitting quietly in a group with their bodies touching; (c) crowd freezing behavior, when all four animals stopped their behaviors and froze, (d) crowd aggressive behavior, when at least one animal engaged in an aggressive behavior including one or more of swatting, lunging, wrestling, pinning, or biting another animal, and (e) other various and non-social behaviors, including one or more periods of time during which the animals were not interacting with each other, or were engaged in individual, non-social behaviors (such as exploring the cage). The durations of each behavior were averaged between the observers.

Neurobiological Measures

Tissue Collection

Two hours after the end of the SCS all experimental subjects were anesthetized by a subcutaneous injection of ketamine (67 mg/kg, sc; NLS Animal Health Owings, MD) and xylazine (13.33 mg/kg, sc; NLS Animal Health, Owings Mills, MD), and then euthanized via cervical dislocation. The brains were carefully removed. Brains were passively perfused in a fixative solution of 4% paraformaldehyde and 5% acrolein (Sigma Aldrich, St. Louis, MO) and gently agitated for 4 hours (Cushing, Yamamoto, Hoffman, & Carter, 2003). The tissue was then postfixed in 4% paraformaldehyde for 24 hours, followed by immersion in a 25% sucrose solution, and stored at 4°C. Brain tissue was sectioned into 40μm slices on a cryostat. Tissue was stored in well plates in a cryoprotectant antifreeze solution at −20°C, which is a buffered solution that prevents the tissue from freezing and prevents deterioration, demonstrated to protect brain tissue for several months when stored at −20°C (Grippo et al., 2007a; Watanasriyakul et al., 2019).

Immunohistochemistry and Analysis

Serial brain slices (40μm) were assayed for the cFos protein using a validated method described by Watanasriyakul et al. (2018), and using commercially available materials. Sections were rinsed in potassium phosphate buffered saline (KPBS) six times over the course of 1 hour, incubated in a 0.01% solution of sodium borohydride in KPBS for 20 minutes, and then repeatedly rinsed in KPBS for at least one hour. The tissue was then incubated in 0.014% phenylhydrazine dissolved in KPBS for 15 minutes before undergoing 6 washes in KPBS over one hour. The tissue was incubated in the primary cFos antibody (Catalog# PC38, anti-cFos, generated in rabbit; EMD Millipore, Billerica, MA) at a concentration of 1:100,000, diluted in a 4% TritonX-100 in KPBS solution. The tissue incubated at room temperature for 1 hour before incubating at 4°C for an additional 48 hours. The tissue then underwent 6 washes in KPBS before being incubated in a solution of biotin-goat-antirabbit IgG (Catalog# BA-1000; Vector Laboratories, Burlingame, CA; 1:600) in 4% TritonX-100 + KPBS for one hour. After 5 washes in KPBS, the tissue was then incubated in an A/B solution (Vectastain Elite, Catalog# PK-6100; Vector Laboratories, Burlingame, CA; 45 μL A, 45 μL B per 10 mL of 4% TritonX-100 + KPBS) for 1 hour. The tissue was washed three times in KPBS, followed by three times in 0.175M sodium acetate, before incubating in a solution of 3–3’-diaminobenzadine, nickel sulfate, 3% hydrogen peroxide, and sodium acetate for 12–15 minutes. The tissue was then rinsed three times in sodium acetate, followed by 3 times in KPBS, before being mounted on electrostatically charged microscope slides.

Stained and mounted tissue slides were left to air dry. Once dry, the tissues were dehydrated in a series of ethanol dilutions, cleared in Histoclear (National Diagnostics, Atlanta, GA), and the slides were coverslipped with Histomount (National Diagnostics, Atlanta, GA).

General Image Processing Procedures

All brain images were captured using a Nikon Eclipse E 800 microscope equipped with a Sensi-cam camera on a computer using IPLab software (Scanalytics Inc., Fairfax, VA). Paxinos and Watson’s (2006) rat atlas was used to determine the locations of all brain regions of interest. Cells were considered to be cFos-positive if they exhibited black and round/oval shape characteristics demonstrated in previous studies (Watanasriyakul et al., 2018). ImageJ (National Institutes of Health, Bethesda, MD) was used for all cFos-related quantification, using similar procedures previously described in prairie voles (Watanasriyakul et al., 2018, 2019). Briefly, all images were converted to 8-bit, and standardized shapes were created to determine cFos-immunoreactive density/cells within each brain region.

Manual quantification was used to quantify cFos-immunoreactivity in the PVN and amygdala. Raters clicked on cFos-positive cells within the standardized areas, with ImageJ keeping count of the final value for each section. Mean optical density was used to quantify cFos-immunoreactivity in all hippocampal subregions using method previously described (Watanasriyakul et al., 2019). Briefly, corpus callosum readings were used as background, and these numbers were subtracted from the density readings obtained from each hippocampal subregion to determine the final cFos-immunoreactivity value. The optical density values were relative to the gray values associated with the sample area (white: 0; black: 255). For both methods, three to four brain sections were analyzed from each animal, and values were averaged between hemispheres. Quantifications were averaged across multiple brain slices, hemispheres, and raters to provide an accurate estimation of cell count/optical density for all brain regions. Damaged sections were excluded from the analyses.

Region-Specific Image Processing Procedures

Hypothalamic Paraventricular Nucleus (PVN): The PVN is approximately Bregma −1.56 to −1.80 mm and was further characterized by the medial-lateral position of the fornix (relative to the third ventricle) and medial and dorsal location of the optic tract (relative to more central and ventral position in more rostral sections). Images of the PVN were captured at 40x magnification. A standardized square (area = 90,256 pixels/hemisphere) was used for quantification in the PVN.

Dorsal Hippocampus Subregions: The location of the dorsal hippocampus was approximately Bregma −2.64 to 3.00 mm—more specifically, the immediate area ventral to the corpus callosum. Regions of interest within the dorsal hippocampus included CA1, CA3, dorsal dentate gyrus (dDG), and ventral dentate gyrus (vDG) subregions. Images of all hippocampus subregions were captured at 10x magnification. An average reading of three standardized squares (50 × 50 pixels; area: 2500 pixels) were used to determine the density for each subregion.

Central and Basolateral Amygdala: For the current study, only sections in which both the central (CeA) and basolateral amygdala (BLA) subregions could be clearly identified were included in the analyses; these were approximately Bregma −2.64 to −3.00 mm. The BLA subregion is located adjacent to the external capsule, with a distinctive “tear-drop” shape. On the other side of the BLA amygdala, towards the medial portion of the section, the CeA subregion is located, with a distinctive oblong shape. Images of all amygdala subregions were captured at 10x magnification. Four standardized circles (area = 19,353 pixels/circle/hemisphere) were used for quantification in each amygdala subregion. Each subregion was counted individually, and the values from all four standardized circles from each subregion were summed.

Corpus Callosum: The location of the corpus callosum was approximately Bregma −2.64 to −3.00; specifically, the immediate area dorsal to the hippocampus and ventral to the retrosplenial dysgranular cortex. Images of the corpus callosum were captured at 10x magnification. An average reading of three standardized squares (50 × 50 pixels; area: 2500 pixels) were used to determine the density of the corpus callosum.

Statistical Analyses

Given 4 independent groups in the present design, without fully crossed conditions, behavioral and neurobiological data were analyzed with single-factor analyses of variance (ANOVA), followed by hypothesis-driven pairwise comparisons with independent groups Student’s t-tests (t-tests assuming unequal variances were used for all comparisons when the homogeneity of the variance assumption was violated, as noted in the results section). Effect sizes were calculated using Cohen’s d values (Cohen, 1988), using the following general categories to discuss strength of the effects: d = 0.0, negligible effect; d = 0.2, small effect; d = 0.5, medium effect; d = 0.8, large effect; d = 1.3 or larger, very large effect. Body weight data were analyzed using a mixed-design ANOVA, with time as the repeated factor.

Data are presented as means and standard error of the mean (SEM) in the results, tables, and figures. A p-value less than 0.05 was considered to be statistically significant for single analyses, assuming a 2-tailed distribution. A Bonferroni correction was applied to multiple comparisons; statistical significance was noted when the value exceeded the adjusted probability level.

To gain a more comprehensive understanding of the relationships between distance traveled and maximum speed reached in the running wheel with other dependent measures in the present study design, correlations were computed in the isolated enriched-exercise condition using Pearson’s r correlation coefficients. Given the small sample size in the present study, statistical significance of correlations was not computed. Rather, based on Cohen’s (1988) suggestions, the following general categories were used for discussing strength of the correlations: r = 0.1, weak correlation; r = 03, moderate correlation; r = 0.5, strong correlation; r = 0.8, very strong correlation.

Results

Behavioral Measures

Elevated Plus Maze

A single-factor ANOVA tested whether the duration of time spent exploring the open arms of the EPM was significantly different among the four housing conditions (Figure 1 top). A significant main effect of housing condition on the duration of time spent exploring the open arms of the EPM was observed [F(3,43) = 6.0, P < 0.002]. Animals in the isolated sedentary condition spent less time exploring the open arms of the EPM relative to both the paired control [t(9) = 3.8, P < 0.002; t-test assuming unequal variances; Cohen’s d = 2.0] and the isolated enriched-exercise condition [t(18) = 3.4, P < 0.004; Cohen’s d = 1.7]. However, the open arm exploration duration did not differ significantly between the isolated sedentary and isolated enriched conditions (P > 0.05). The duration of open arm exploration was slightly, but non-significantly, lower than the paired control value [t(18) = 1.18, P = 0.09]. These values did not significantly differ between paired control and isolated enriched-exercise conditions (P > 0.05), nor did they differ between the isolated enriched and isolated enriched-exercise conditions (P > 0.05).

Figure 1.

Mean (+ SEM) duration spent in the open arms (Panel A) and number of crosses into the center section (Panel B) of a 5-minute elevated plus maze in prairie voles following 4 weeks of paired control (n = 10), isolated sedentary (n = 10), isolated enriched (n = 10), or isolated enriched-exercise conditions (n = 13). Black dots indicate individual data points. *P < 0.05 vs. isolated sedentary condition.

A single-factor ANOVA tested whether the number of crosses into the center section of the EPM was significantly different among the four housing conditions (Figure 1 bottom). There was no significant main effect of housing condition on the number of crosses into the center section of the EPM (P > 0.05). No follow-up tests were conducted.

Forced Swim Test

A single-factor ANOVA evaluated whether the duration of immobility in the FST was significantly different among the four housing conditions (Figure 2). The test revealed a significant main effect of housing condition [F(3,43) = 5.8, p < 0.003]. The isolated sedentary condition exhibited greater levels of immobility relative to paired control [t(18) = 3.3, P < 0.005; Cohen’s d = 1.5], isolated enriched [t(18) = 3.0, P < 0.008; Cohen’s d = 1.4], and isolated enriched-exercise conditions [t(21) = 3.3, P < 0.004; Cohen’s d = 1.4]. The duration of immobility did not significantly differ among the paired control, isolated enriched, or isolated enriched-exercise conditions (P > 0.05 for all comparisons). The duration of time spent engaging in specific active behaviors (swimming, struggling, and climbing) did not significantly differ among the housing conditions and were summed to provide one index of active behaviors [these behaviors comprise the remainder of 300 seconds (total FST time) for each animal, after calculating the duration of immobility; data not shown].

Figure 2.

Mean (+ SEM) duration of immobility in a 5-minute forced swim test in prairie voles following 4 weeks of paired control (n = 10), isolated sedentary (n = 10), isolated enriched (n = 10), or isolated enriched-exercise conditions (n = 10). Note: the remainder of 5 minutes is comprised of active behavioral responses (swimming, struggling, and climbing). Black dots indicate individual data points. *P < 0.05 vs. isolated sedentary condition.

Social Crowding Stressor

Single-factor ANOVAs were used to evaluate the following categories of behaviors during the SCS: (a) investigative behaviors; (b) aggressive behaviors; (c) freezing; (d) sitting quietly together as a group; and (e) individual, non-social behaviors (Table 1). Significant main effects of housing condition were observed for aggressive behaviors [F(3,43) = 4.2, P < 0.01] freezing [F(3,43) = 16.0, P < 0.0001], and sitting quietly [F(3,43) = 3.18, P < 0.03]. No significant main effects of housing condition were observed for investigative or individual/non-social behaviors (P > 0.05 for both comparisons; no follow-up tests were conducted).

Table 1.

Mean (± SEM) social and non-social behaviors during a 10-minute social crowding stressor in prairie voles following 4 weeks of paired control (n=10), isolated sedentary (n=10), isolated enriched n=10), or isolated enriched-exercise conditions (n=13).

	Investigative Behaviors (duration, sec)	Aggressive Behaviors (instances, number)	Freezing (duration, sec)	Sitting Quietly with Bodies Touching (duration, sec)	Individual, Non-Social Behaviors (duration, sec)
Paired Control	67.8 ± 13.3	4.8 ± 1.6#	11.8 ± 3.1*	25.2 ± 3.3*	423.8 ± 30.3
Isolation Sedentary	71.1 ± 20.1	21.4 ± 9.0	43.5 ± 5.0	5.2 ± 2.0	279.2 ± 51.2
Isolated Enriched	63.5 ± 16.5	4.8 ± 2.3#	13.9 ± 3.7*	21.7 ± 3.6*	414.5 ± 38.3
Isolated Enriched-Exercise	61.1 ± 14.5	3.5 ± 1.3#	11.4 ± 3.8*	25.0 ± 3.0*	405.9 ± 41.1

^*P < 0.05 vs. isolated sedentary condition.

^#P < 0.1 vs. isolated sedentary condition (t-test assuming unequal variances).

Pairwise comparisons of aggressive behaviors indicated that the isolated sedentary condition demonstrated slightly greater number of aggressive behaviors relative to the other three conditions, however these differences were not statistically significant after applying the Bonferroni correction (P > 0.05 for all comparisons; t-tests assuming unequal variances).

Pairwise comparisons of freezing indicated that the isolated sedentary condition demonstrated significantly greater duration of freezing relative to the paired control [t(18) = 5.9, P < 0.0001; Cohen’s d = 2.5], isolated enriched [t(18) = 5.1, P < 0.0001; Cohen’s d = 2.1], and isolated enriched-exercise conditions [t(21) = 5.5, P < 0.0001; Cohen’s d = 2.2]. The duration of freezing did not differ significantly among the paired control, isolated enriched, or isolated enriched-exercise conditions (P > 0.05 for all comparisons).

Pairwise comparisons of duration of time spent sitting quietly together as a group, with bodies touching, indicated that the isolated sedentary condition demonstrated significantly lower duration spent sitting quietly relative to the paired control [t(18) = 4.9, P < 0.001; Cohen’s d = 2.4], isolated enriched [t(18) = 3.8, P < 0.001; Cohen’s d = 1.8], and isolated enriched-exercise conditions [t(21) = 3.1, P < 0.005; Cohen’s d = 2.2]. Duration of sitting quietly did not significantly differ among the paired control, isolated enriched, or isolated enriched-exercise conditions (P > 0.05 for all comparisons).

Neural Immunoreactivity

Single-factor ANOVAs were used to evaluate potential immunoreactivity differences among four housing conditions following the social crowding stressor in the following brain areas: (a) paraventricular nucleus (PVN); (b) hippocampal subregions (CA1, CA3, vDG, dDG); and (c) amygdala subregions (basolateral and central). A significant main effect was observed for the PVN [F(3,43) = 4.8, P < 0.006] (Figure 3), but not for any hippocampal or amygdala regions (P > 0.05, no follow-up comparisons were conducted; Tables 2 and 3, respectively).

Figure 3.

Example raw images showing sampling area in the hypothalamic paraventricular nucleus 2 hours after a 10-minute social crowding stressor in prairie voles following 4 weeks of paired control (Panel A; n = 10), isolated sedentary (Panel B; n = 10), isolated enriched (Panel C; n = 10), or isolated enriched-exercise conditions (Panel D; n = 10); and summary data showing mean (+ SEM) density (number of cells per sampling area) of cFos immunoreactivity for each group (Panel E). Scale bar (shown on Panel A) = 100 μm for all raw images in Panels A-D. Black dots indicate individual data points in Panel E. *P < 0.05 vs. isolated sedentary condition; ^P < 0.05 vs. isolated enriched condition.

Table 2.

Mean (± SEM) density (pixels) in hippocampal subregions 2 hours after a 10-minute social crowding stressor in prairie voles following 4 weeks of paired control (n=10), isolated sedentary (n=10), isolated enriched n=10), or isolated enriched-exercise conditions (n=13).

	CA1	CA3	dDG	vDG
Paired Control	16.2 ± 4.0	9.2 ± 2.0	14.8 ± 4.1	16.9 ± 4.7
Isolation Sedentary	17.3 ± 3.3	8.7 ± 2.2	14.4 ± 4.4	15.4 ± 3.8
Isolated Enriched	13.6 ± 1.7	7.3 ± 1.0	11.0 ± 1.5	11.3 ± 1.5
Isolated Enriched-Exercise	16.1 ± 2.8	8.9 ± 1.6	14.5 ± 2.6	15.2 ± 3.4

Table 3.

Mean (± SEM) density (number of cells per sampling area) in amygdala subregions 2 hours after a 10-minute social crowding stressor in prairie voles following 4 weeks of paired control (n=10), isolated sedentary (n=10), isolated enriched n=10), or isolated enriched-exercise conditions (n=13).

	CeA	BLA
Paired Control	9.2 ± 1.1	9.4 ± 0.9
Isolation Sedentary	11.8 ± 1.2	11.4 ± 0.7
Isolated Enriched	10.0 ± 1.0	10.6 ± 0.9
Isolated Enriched-Exercise	10.3 ± 0.8	10.2 ± 0.8

Pairwise comparisons conducted on PVN immunoreactivity as a function of housing condition indicated that the isolated sedentary condition exhibited greater immunoreactivity relative to the paired control condition [t(18) = 3.1, P < 0.007; Cohen’s d = 1.3] and the isolated enriched-exercise condition [t(21) = 3.9, P < 0.0008; Cohen’s d = 1.5], but not the isolated enriched condition (P > 0.05). The isolated enriched immunoreactivity levels were significantly greater than both the paired control [t(21) = 3.1, P < 0.007; Cohen’s d = 1.3] and the isolated enriched-exercise conditions [t(21) = 2.5, P < 0.05; Cohen’s d = 0.8]. Immunoreactivity in the paired control and isolated enriched-exercise conditions did not differ significantly (P > 0.05).

Body Weight

A mixed-design ANOVA tested whether body weight differences existed among the housing conditions, with time as the repeated factor. No significant main effects or interactions were observed (P > 0.05 for all comparisons; data not shown; no follow-up comparisons were conducted).

Running

Daily distance traveled and daily maximum speed were recorded via the running wheel odometer in the isolated enriched-exercise condition. Mean daily distance traveled was 1.40 ± 0.50 km/day (range 0.01 – 6.09 km/day). Mean daily maximum speed was 1.67 ± 0.28 km/hr (range 0.12 – 3.51 km/hr). These values were highly correlated at r = 0.99 (strong correlation).

Mean daily distance traveled and mean daily maximum speed reached were correlated with the outcome measures that yielded statistically significant group differences in the above-referenced sections; as well as with both baseline and final body weight. Given the highly correlated nature of mean daily distance traveled and mean daily maximum speed reached, Pearson’s r correlation coefficients of 0.3 (moderate correlation) or larger between mean daily distance traveled and other outcome measures are reported here (r values varied by 0.01 – 0.04 for all correlations between these two running wheel variables and the other outcome measures). Mean daily distance traveled was moderately positively correlated with duration of time spent exploring the open arms of the EPM (r = 0.44); and was moderately negatively correlated with instances of aggression (r = −0.33) and strongly negatively correlated with duration of freezing (r = −0.72) during the SCS. Mean daily distance traveled was moderately negatively correlated with baseline body weight (r = −0.34) as well as final body weight (r = −0.44). Mean daily distance traveled did not demonstrate meaningful correlations with the following outcome measures: duration of immobility during the FST, sitting quietly together as a group during the social crowding stressor, or cFos immunoreactivity in the PVN (all r values below 0.25).

Discussion

The present study used the prairie vole model to investigate the ability of two forms of EE – one that included access to physical exercise versus one that lacked an exercise component – to protect against behavioral and neural consequences of social isolation. Behavioral and immunohistochemical analyses were selected to assess whether physical exercise is critical to the benefits of EE in the context of social stress. It was hypothesized that, while the inclusion of a physical exercise component would provide additional buffering against the stress of social isolation compared to the lack of physical exercise, EE without physical exercise would also serve a protective role against the consequences of social isolation. The present data indicate differential neural and behavioral benefits of EE as a function of the presence of physical exercise, with strong to very strong effects demonstrated for the outcome measures. Exercise may serve a critical role in protecting against anxiety-related behaviors and neural activation in the hypothalamic PVN in response to social stress, however EE both with and without an exercise component are equally effective at protecting against depression-related behaviors and social behavioral changes during a social stressor.

The current study employed the EPM and FST to operationalize anxiety- and depression-like behaviors, respectively, in prairie voles exposed to social isolation (Slattery & Cryan, 2012; Walf & Frye, 2007). Previous research involving the prairie vole model has demonstrated that these behavioral tests are valid operational measures of maladaptive emotion-related behaviors (Grippo et al., 2008, Bosch et al., 2009, Grippo et al., 2012, Sun et al., 2014). As expected, the isolated sedentary condition displayed the highest levels of anxiety-related behavior in the EPM, whereas the paired control condition displayed the lowest levels of anxiety-related behavior, measured by duration of time spent exploring the open arms of the maze. These data support anxiety-related consequences demonstrated in previous female prairie voles exposed to social isolation (Grippo et al., 2007b; Lieberwirth et al., 2012). EE and exercise may be protective against this anxiety-relevant behavior; the isolated enriched-exercise condition spent a significantly greater amount of time in the open arms compared to the isolated sedentary condition. These data support previous observations of anxiolytic effects of EE paradigms in models of stress in rats and mice (Bahi, 2017; Costa et al., 2021; Mora-Gallegos and Formaguera, 2019), and anxiolytic consequences as a function of both EE and physical exercise in socially isolated prairie voles (Grippo et al., 2014). The present findings also suggest that the addition of an exercise component to an EE paradigm provides a more robust buffer against the display of anxiety-related behaviors in response to social stress (relative to EE without an exercise component). This interpretation is further supported by the positive correlation between distance traveled in the running wheel and duration of open arm exploration in the EPM in the isolated enriched-exercise condition.

In contrast to anxiety-related behaviors, the inclusion of an exercise component in the EE paradigm did not outperform EE without exercise in its ability to protect against depression-related behaviors in the FST. The paired control, isolated enriched, and isolated enriched-exercise groups displayed similar levels of immobility in the FST, which were significantly lower than those of the isolated sedentary condition. This pattern of responses suggests that both forms of enrichment can reduce social isolation-induced depression-related behaviors to a level that is comparable to paired housing conditions. The similarity of the protective effect on immobility duration between the EE housing conditions is supported by the lack of a correlation between daily distance traveled in the running wheel and immobility duration in the FST in the isolated enriched-exercise condition. The present data agree with previous studies focused on the benefits of EE and physical exercise in socially isolated prairie voles (Grippo et al., 2014; Normann et al., 2018), and in other rodent species (Costa et al., 2021; Duman et al., 2008). Similarly, EE and exercise have been suggested to reduce the severity of mood disorder symptoms and feelings of loneliness in humans (Byrne & Byrne, 1993; Salmon, 2001).

Social behaviors during the SCS followed a similar pattern to those observed in the FST. The SCS was used here to investigate the ability of EE and physical exercise to protect against social behavioral consequences of isolation during a short-term crowded environment (Djordjevic et al., 2005; Grippo et al., 2010). Social isolation was associated with a reduction in adaptive social behaviors and an increase in maladaptive behaviors, evidenced by a reduced duration of time sitting quietly in side-by-side contact with other animals, and increased instances of freezing and aggressive behaviors, relative to paired control conditions and both EE conditions. Side-by-side contact and aggressive behaviors have been previously reported to represent positive and negative social behaviors, respectively (Koolhaas et al., 2013; Sun et al., 2014; Lee & Beery, 2021). Freezing in rodents has been associated with anxiety-related behaviors and stress reactivity in previous tests of exploration (Diaz-Moran et al., 2012; Wardwell et al., 2020). However, maladaptive social behaviors observed in isolated prairie voles may depend on the social behavioral assessment used, the duration of social isolation, and sex of the subjects; as some previous isolation paradigms have produced a reduction in prosocial behaviors (Sun et al., 2014), some have produced an increase in prosocial behaviors (Lieberwirth et al., 2012; Perry et al., 2016; Sun et al., 2014), whereas others have yielded no significant change in social behaviors (Grippo et al., 2007a; Grippo et al., 2010). The present study design extends previous findings of social behavioral consequences in prairie voles in an acutely stressful environment by demonstrating that EE both with and without a physical exercise component may similarly protect against altered positive and negative social behaviors when data are considered at a group level. However, when considering potential individual differences, the pattern of correlations indicates that amount of physical activity may relate to the display of negative social behaviors during the SCS – demonstrated by the moderate negative relationship between daily distance traveled and aggressive behavior and the strong negative relationship between daily distance traveled and freezing behavior. In contrast, amount of physical activity may be less related to the display of positive social behaviors during the SCS, as a lack of correlation between daily distance traveled and duration of side-by-side contact was noted.

Acute social crowding was used as a social stressor in the present research design. Neural activation following the SCS was quantified via reactivity of the immediate early gene cFos in several brain regions including the PVN, hippocampus, and amygdala. Social isolation has been linked to neurological consequences in the hippocampus and PVN (Cacioppo et al., 2015; Kamal et al., 2014), and altered amygdala function has been implicated in treatment-resistant depression (Ferri et al., 2017). Further, exercise and enrichment have both been associated with increased hippocampal health and reduced sensitivity to stress in humans (Miller et al., 2013) and rodents (Schloesser et al., 2010). In the present study, cFos density differences were observed in the PVN, but not in the hippocampus or amygdala. Social isolation was associated with increased cFos activation in the PVN relative to social pairing, which is consistent with previous short-term social stress paradigms in isolated prairie voles (Grippo et al., 2007a; Grippo et al., 2010). EE with an exercise component attenuated PVN cFos activation, however the same benefit was not observed with EE when the exercise component was absent. Notably, EE alone protected against social behavioral disruptions during the SCS, but not SCS-induced immunoreactivity in the PVN, indicating a dissociation between the behavioral and neural benefits of EE alone.

The present results are limited in that they included a comparison between EE alone (without exercise) and EE with an exercise component – but not a direct comparison of these housing conditions with exercise alone. Therefore, although the present pattern of responding suggests that EE with a physical exercise component may be more effective at attenuating PVN activation than EE alone, these data do not completely support previous findings from isolated prairie voles exposed to other environmental interventions. For instance, exercise alone did not alter short-term neural activation in the PVN relative to sedentary conditions in female socially isolated prairie voles (Watanasriyakul et al., 2018). However, exercise alone attenuated both the duration of immobility during a 5-minute FST and corticosterone reactivity following the FST in socially isolated prairie voles (relative to isolated sedentary conditions) (Watanasriyakul et al., 2018). An investigation of long-term neural activity as a function of EE (with exercise) versus exercise alone in female socially isolated prairie voles again demonstrated that neither environmental condition altered neural activity in the PVN, but both conditions reduced basal corticosterone levels in socially isolated animals relative to isolated sedentary conditions (and corticosterone levels in both environmental conditions were comparable to paired housing) (Watanasriyakul et al., 2019). Considered together with these previous findings, the reduction in PVN activation observed here in the isolated enriched-exercise condition might suggest that a combination of environmental manipulations has a stronger effect on some neurobiological changes associated with social stress than either exercise or EE alone.

Both EE and physical exercise have been associated with positive neurological, physiological, behavioral, and psychological outcomes in human and rodent models (Gielen et al., 2001; Jonsdottir 2000; Milgram et al., 2006; Miller et al., 2013; Droste et al., 2007; Duman et al., 2008; Sanders et al., 2019). The present results support these previous studies and provide additional insight into the benefits of EE and exercise in the context of social stress. The present study may support a growing body of literature focused on ideal environmental conditions for rodents used for translational research, such as recent discussions about the benefits of EE and considerations of environmental temperature (Bailoo et al., 2018; Raun et al., 2020). Further, the present study – similar to these recent studies from Bailoo et al. (2018) and Raun et al. (2020) – draws attention to the importance of research using female rodents. Parallel studies in males as well as specific sex comparisons will further enhance our understanding of the benefits of environmental interventions on responses to social stressors.

In conclusion, social isolation is associated with robust emotion-related and social behavioral deficits, coupled with increased neural activation in the PVN in female prairie voles. The inclusion of an exercise component in an EE paradigm protects against anxiety-related behaviors and PVN activation to a greater extent than EE alone. However, EE both with and without an exercise component have similar protective effects against depressive behaviors and some social behavioral disruptions as a function of social isolation. The differential behavioral and neural benefits shown here indicate that some behavioral interventions might be more effective for certain consequences of isolation over other interventions. Exercise alone has been reported to reduce symptoms of anxiety and depression (Byrne & Byrne, 1993; Salmon, 2001) and provides many physiological benefits (Gielen et al., 2001; Jonsdottir et al., 2000), while enrichment protects against consequences of psychosocial stressors (Schloesser et al., 2010) and against several neurological problems (Milgram et al., 2006; Miller et al., 2013). Coupled with these previous findings, the results of the present study are uniquely relevant given the increased social stress and loneliness experienced by so many individuals during the past year (Luchetti et al., 2020; Palgi et al., 2020; American Psychological Association, 2021). Collectively, the patterns displayed in the present study indicate that the inclusion of exercise is not necessarily a critical component for buffering against some consequences of social isolation, which is beneficial for a growing population of older individuals and those who are less able to engage in exercise for physical, logistical, or personal reasons (Rugbeer et al., 2017; Silva et al., 2019). A continued focus on behavioral and neural benefits of environmental interventions using valid and reliable animal models, such as that described here, will provide valuable insight into behavioral strategies to protect against social stress and loneliness in humans.