The Influence of Environmental Enrichment on Affective and Neural Consequences of Social Isolation Across Development

Abstract

Social stress is associated with depression and anxiety, physiological disruptions, and altered brain morphology in central stress circuitry across development. Environmental enrichment strategies may improve responses to social stress. Socially monogamous prairie voles exhibit analogous social and emotion-related behaviors to humans, with potential translational insight into interactions of social stress, age, and environmental enrichment. This study explored the effects of social isolation and environmental enrichment on behaviors related to depression and anxiety, physiological indicators of stress, and dendritic structural changes in amygdala and hippocampal subregions in young adult and aging prairie voles. Forty-nine male prairie voles were assigned to one of six groups divided by age (young adult vs. aging), social structure (paired vs. isolated), and housing environment (enriched vs. non-enriched). Following 4 weeks of these conditions, behaviors related to depression and anxiety were investigated in the forced swim test and elevated plus maze, body and adrenal weights were evaluated, and dendritic morphology analyses were conducted in hippocampus and amygdala subregions. Environmental enrichment decreased immobility duration in the forced swim test, increased open arm exploration in the elevated plus maze, and reduced adrenal/body weight ratio in aging and young adult prairie voles. Age and social isolation influenced dendritic morphology in the basolateral amygdala. Age, but not social isolation, influenced dendritic morphology in the hippocampal dentate gyrus. Environmental enrichment did not influence dendritic morphology in either brain region. These data may inform interventions to reduce the effects of social stressors and age-related central changes associated with affective behavioral consequences in humans.

Introduction

Social stressors contribute to affective disturbances including cognitive deficits, low self-efficacy beliefs, depression, anxiety, and social withdrawal, coupled with disrupted physiological processes indicative of increased stress reactivity and premature mortality (Cacioppo et al., 2015; Cohen-Mansfield et al., 2016; O’Súilleabháin et al., 2019; Steptoe et al., 2013; Xia & Li, 2018). The pandemic has highlighted the importance of adaptive social interactions for promoting psychological and physical health (Bland et al., 2022; Luchetti et al., 2020; Palgi et al., 2020; Park et al., 2021). Social stressors, such as long-term social isolation, also produce behaviors relevant to negative affect and emotion in operational measures in rodents (Brenes & Fornaguera, 2008; Ieraci et al., 2016; Koike et al., 2009). For example, socially isolated rats display altered immobility in a forced swim test (FST) and altered sucrose intake in a fluid consumption test, indicative of depression-relevant behaviors (Brenes Sáenz et al., 2006). Socially isolated mice exhibit altered behaviors in a social interaction test, anxiety-like behaviors in an elevated plus maze (EPM), and altered exploration in an open-field test, indicative of depression- and anxiety-relevant behaviors and disrupted social behaviors (Ieraci et al., 2016; Koike et al., 2009).

The negative affective consequences of social stress — including behaviors that represent anxiety, depression, fear, increased reactivity to stress, and altered social interactions — may change across developmental phases. For instance, aging individuals are vulnerable to negative affective consequences of loneliness due to losing loved ones, age-related declining physical health, and altered social roles (Cohen-Mansfield et al., 2016). Some aging adults report higher rates of anxiety and depression relative to younger adults; however, others report the opposite pattern (Prenderville et al., 2015; Volkert et al., 2013; Wuthrich et al., 2015). Aging adults may exhibit different symptoms of these negative affective pathologies relative to younger adults, and symptoms may be more prominent for men relative to women (Adams, 2001; Förster et al., 2019; Gum et al., 2010; Wuthrich et al., 2015). Similarly, differential anxiety- and depressive-like behaviors have been observed in different ages of rodents in response to stress (Ennaceur et al., 2008; Francia et al., 2006; Malatynska et al., 2012; Prenderville et al., 2015).

In addition to negative affective outcomes such as signs of depression and anxiety, changes in stress circuitry have been observed in aging humans and in animal models, including in regions that are associated with both stress activation and inhibition. For example, relative to adults ages 22–50, adults ages 65–84 display altered volumes and asymmetry indices in the hippocampus and amygdala (Malykhin et al., 2008). Aging rats display lower hippocampal volume and reduced neuronal density compared to younger and middle-aged rats (Driscoll et al., 2006). Further, aging is associated with increased dendritic arborization and length, but may not influence spine density or number of neurons in the amygdala in rats (Rubinow et al., 2009; Sotoudeh et al., 2020). These changes associated with both activation and inhibition mechanisms of stress may mediate affective behaviors in aging individuals, including symptoms of depression, anxiety, and stress in humans and behavioral disruptions that represent these constructs in animal models.

Research involving translational animal models will inform our understanding of behavioral and neural processes associated with negative affect and emotions, social stress, and aging. Prairie voles are socially monogamous rodents that cohabitate in pairs or family groups and form strong social bonds similar to humans, providing a valuable model for investigating behavioral and neurobiological consequences of social stress (Carter et al., 1995; Insel & Young, 2001; Kenkel et al., 2021; Lieberwirth et al., 2012; Young et al., 2011). Prairie voles exposed to social stressors display depression- and anxiety-related behaviors in operational behavioral tests, coupled with altered neural activation in stress circuitry including in the amygdala, hippocampus, hypothalamus, and frontal cortex (Bosch et al., 2009; Pan et al., 2009; Stowe et al., 2005; Sun et al., 2014; Watanasriyakul et al., 2018, 2019). For instance, social isolation is associated with increased neural activation as well as altered dendritic morphology and spine density, in the basolateral amygdala (BLA) in prairie voles, which may be associated with negative affective behaviors such as altered exploration in an EPM and increased immobility in a FST (Hylin et al., 2022; Watanasriyakul et al., 2019). Chronic stress is also associated with central nervous system disruptions such as reduced neuronal volume and decreased neurogenesis in the hippocampus (see McEwen, 2017); however, social isolation in prairie voles has not been associated with functional hippocampal changes in some previous studies (McNeal et al., 2018; Watanasriyakul et al., 2019). Considered together, the evidence from previous research indicates that the prairie vole has high translational value to inform our understanding of social bonding, stress, and negative affective behaviors that are relevant to depression and anxiety across development. However, specific neural changes, especially in areas that promote stress responses such as the amygdala, and in areas that have down-regulatory influences on stress responses such as the hippocampus, require further investigation in the context of social behavior and stress.

The widespread influence of social stressors on disrupted emotions, depressive and anxiety disorders, physiological health consequences, and increased neurobiological stress reactivity highlights the importance of developing effective and feasible treatment strategies to protect against these negative health consequences. Environmental enrichment (EE) is a paradigm that introduces physical, cognitive, and/or social activities to humans or rodents (including prairie voles), which can change structure and function in areas of the brain that promote adaptive neural communication and protect against stress (Clemenson et al., 2015; Clemenson & Stark, 2015; Flores-Gutiérrez et al., 2018; Li & Tang, 2005; Prounis et al., 2018; Reguilón et al., 2020; Seong et al., 2018; Woo & Leon, 2013). In rodent models of social stress, EE improves depression- and anxiety-related behaviors and increases social interactions and exploration (Brenes et al., 2009; Galani et al., 2007; Hüttenrauch et al., 2016; Peña et al., 2009; Simpson & Kelly, 2011). Similarly, EE that involves physical, cognitive, and sensory stimulation has been presented previously in place of social interactions to decrease anxiety- and depressive-like behaviors in socially isolated prairie voles (Grippo et al., 2014; Normann et al., 2018; Normann et al., 2021). EE paradigms that involve social, cognitive, sensory, and/or physical stimulation also influence the function or structure of brain regions associated with stress in rats, mice, and prairie voles — which may vary across development — such as altered neurotransmitters in the hippocampus and frontal cortex in rats and mice, decreased immediate early gene activity in the BLA in rats and prairie voles, reduced glucocorticoid receptor function in the BLA in rats and mice, altered neurogenesis in the hippocampus of aging rats, short-term increases in peptide function in cortical and hippocampal regions in young adult (but not in aging) rats, and differential spine density and dendritic length in cortical regions based on both age and sex in rats (Kolb et al., 2003; Novaes et al., 2017; Rapley et al., 2018; Segovia et al., 2006; Sztainberg et al., 2010; Watanasriyakul et al., 2019; also see for review Gubert & Hannan, 2019).

The interactions of stress and aging are important considerations when studying potential protective effects of EE. For example, behaviors that are relevant to negative affect, such as increased immobility duration in the FST, are positively correlated with age and extent of social monogamy in isolated prairie voles (Grippo et al., 2021); therefore, treatments for social stress may be influenced by age and social history. Cognitive, motor, sensory, and/or social activities offered through a variety of EE paradigms may promote cognition, reduce depression- and anxiety-related behaviors, alter social behaviors, and change brain communication associated with stress reactivity (Casemiro et al., 2016; Clemenson et al., 2018; Park et al., 2019; Prounis et al., 2018; Woo & Leon, 2013). However, EE differentially influences negative affective processes and emotional reactivity to stress as a function of age, sex, social environmental conditions, and timing of treatment in rodents. Rats exposed to EE at varying ages display lower anxiety- and depression-related behaviors (Brenes et al., 2008; Brenes & Fornaguera, 2008; Harati et al., 2013; Simpson & Kelly, 2011). Increased depression-related behaviors are observed in isolated young adult male rats and female prairie voles, and are reduced following EE (Brenes et al., 2008; Grippo et al., 2014; Normann et al., 2021). In contrast, presenting EE prior to a chronic stress paradigm in adolescent rats improves depression-related behaviors during adulthood in females, but not in males (Smith et al., 2018). These previous results suggest that age may interact with other variables (such as sex, species, or environmental conditions) to influence the effects of EE on behavior or neurobiological outcome measures; however, additional information regarding the potential benefits of EE in aging rodents, including prairie voles, is warranted.

Given the value of the prairie vole model for investigating translational research questions focused on social behavior and stress, integrative neurobiological and behavioral research with this species will contribute to our understanding of the benefits of EE for protecting against social stress. In particular, valuable knowledge may be gained by investigating the effects of social isolation and EE at multiple stages of development on behaviors related to negative affect (such as those relevant to depression and anxiety), as well as in brain regions that are specifically associated with activation and inhibition of stress responses. The present study therefore investigated the influence of social isolation and EE in young adult and aging prairie voles on negative affective behaviors relevant to depression and anxiety, physiological indicators of stress, and dendritic structure of hippocampal and amygdala subregions. We hypothesized the following: (a) social isolation would increase negative affective behaviors (both depression-related and anxiety-related behaviors using operational behavioral tests); (b) both age groups would display behavioral improvements as a function of EE; (c) physiological indicators of long-term stress, including adrenal and body weights, would be associated with social isolation and would vary by age and EE status; and (d) dendritic morphology in hippocampus and amygdala subregions would vary as a function of age, social isolation, and EE.

Method

Animals

Forty-nine adult, male prairie voles were bred at Northern Illinois University (NIU) and used in the experiments described here. Prairie voles were descendants of a wild stock of animals originating from the Champaign, IL region, that were systematically outbred to ensure an appropriate level of genetic diversity. Male prairie voles were specifically chosen for these experiments based on considerations of human aging and sex interactions in the context of social stress (Förster et al., 2019) and to allow for specific comparisons of these data to previous studies of social stress in young adult and aging male prairie voles (Grippo et al., 2021; Hylin et al., 2022). All animals were housed under standard environmental conditions, with a temperature of 20–21°C, a relative humidity of 40–50%, and a 14:10 light/dark cycle (with lights on from 6:30am–8:30pm). Offspring remained with the natal group until 21 days of age, at which time they were weaned and housed in sibling pairs until the commencement of the experiments at a specified age range. At the beginning of the experimental design (prior to any manipulations), young adult prairie voles (n = 23) were a mean of 2.5 months of age (2.5 ± 0.2 months, range 1.8–4.4 months; 76 ± 4.9 days, range 54–133 days), and aging prairie voles (n = 26) were a mean of 17.2 months of age (17.2 ± 0.6 months, range 13.5–21.4 months; 517 ± 17.1 days, range 404–642 days). All procedures were approved by the NIU Institutional Animal Care and Use Committee (IACUC), and followed all federal guidelines set forth in the Guide for the Care and Use of Laboratory Animals and the Three Rs guidelines for responsible conduct in animal research.

Groups and Conditions

For all procedures described here, only one animal from each sibling pair was studied. Each age group was randomly assigned to one of 3 conditions, for a total of 6 groups: (a) aging paired (n = 8), (b) aging isolated (n = 9), (c) aging isolated with EE (aging EE; n = 9), d) young adult paired (n = 8), (e) young adult isolated (n = 7), and (f) young adult isolated with EE (young adult EE; n = 8). The sample sizes listed here were determined from analyzing previous data generated across several study designs, which were used to conduct power analyses focused on behavioral, physiological, and neural outcome measures. For the analyses, Cohen’s d values (Cohen, 1992) were calculated for each outcome measure, and a desired statistical power of 0.8 was defined to minimize the chances of a type II error. A probability level of p < 0.05 was set a priori. The power analyses yielded predicted sample sizes of 6–8 prairie voles per condition as sufficient to detect statistically significant differences in the behavioral, physiological, and neural outcome measures used here. Actual sample sizes of n = 7–9 were used in the present research design, to account for potential attenuation of sample sizes due to multiple dependent measures conducted in the same animal.

All animals remained in the respective housing conditions for 4 weeks. Body weight was recorded prior to and following the 4-week period in all animals, and cage changes and handling were matched among all groups. Paired conditions were housed with a familiar male sibling in a standard-sized cage (12 × 18 × 28 cm), containing bedding and ad libitum access to food (Purina rabbit chow) and water. Isolated conditions were housed individually in the standard-sized cage, containing bedding and ad libitum access to food (Purina rabbit chow) and water, in a separate room, without visual, olfactory, or auditory cues from the unstudied sibling. Isolated conditions with access to EE were housed individually in a larger cage (25 × 45 × 60 cm) in a separate room, without visual, olfactory, or auditory cues from the unstudied sibling. In addition to bedding, food, and water, this group received the following items continuously in the home cage to provide cognitive, sensory, and physical stimulation: (a) a running wheel, (b) a wood block, (c) a tin foil ball, (d) a cardboard toilet paper roll, (e) a mini straw hat, (f) two plastic toys (one inside the cage and one hanging from the cage top), (g) a small bowl with food pellets, (h) a wood jack chew toy, (i) two flat-bottomed marbles, (j) a plastic igloo house, and (k) a square of cotton nesting material (Grippo et al., 2014). All EE items were introduced into the cage at the same time at the beginning of the experimental design, and were sanitized with a 10% bleach solution and dried, or replaced as necessary, once per week. Although EE paradigms in other models have included a social enrichment component (e.g., McQuaid et al., 2016; Prounis et al., 2018; van Praag et al., 2000), a social component was intentionally not used here because the EE paradigm was designed to be presented in lieu of social interactions in the isolated conditions.

Forced Swim Test

A single 5-min FST was used to evaluate helpless behavior. This test is hypothesized to serve as an operational index of helpless behavior in rodents, including prairie voles, representing a maladaptive depression-relevant behavior (Bosch et al., 2009; Cryan et al., 2005; Grippo et al., 2008; Grippo et al., 2012). Twenty-four hours following the final day of the 4-week housing condition period, each animal was placed into a clear, cylindrical Plexiglas tank (46-cm height; 20-cm diameter), filled to a depth of 18 cm with tap water (20–24° C), for 5 min. Prior to each trial, the tank was cleaned thoroughly with 10% bleach solution, rinsed with clean water, and filled with clean water. Immediately following each trial, animals were returned to the home cages, in the respective housing conditions, and provided access to a heat lamp for 15 min to ensure appropriate regulation of body temperature.

Each FST trial was recorded using a digital video camera and manually scored by 2–3 experimentally blind observers who were trained to an inter-rater reliability level of at least 90% using standardized training practices with previous video files and the present set of files. Durations of the following behaviors were coded for the entire 5-min test session: (a) swimming: active, coordinated movement of fore limbs and hind limbs without breaking the surface of water, (b) struggling: breaking the surface of water using fore limbs, (c) climbing: scratching at/climbing the walls of the cylinder, and (d) immobility: minimum to no movement, limb movement only to facilitate floatation. An increased duration of immobility is hypothesized to represent a depressive (helpless, maladaptive) behavioral response in prairie voles; increased durations of swimming, struggling, and climbing are hypothesized to represent active (adaptive) coping strategies (Grippo et al., 2008; Grippo et al., 2012).

Elevated Plus Maze

A single 5-min EPM was used to assess exploratory behavior. This test is hypothesized to serve as an operational index of anxiety-relevant behavior in rodents, including prairie voles (Grippo et al., 2014; Pellow et al., 1985). The maze apparatus consisted of two arms of clear Plexiglas (49.5 × 10 cm), and two opposing closed arms of black Plexiglas with walls and an open roof (49.5 × 10 × 30.5 cm). The 4 arms were connected at the center forming a square (10 × 10 cm). The apparatus was elevated off the ground at a height of 57 cm. Twenty-four hours following the end of the FST, each animal was placed into the center section of the EPM and was allowed to freely move within the maze for 5 minutes. Prior to introducing each animal, the maze was cleaned thoroughly with 10% bleach solution and wiped dry; and immediately after the test animals were returned to the home cages, in the respective housing conditions.

Each EPM trial was recorded using a digital video camera. The behaviors were coded manually by 2–3 experimentally blind observers who were trained to an inter-rater reliability level of at least 90% using standardized training practices with previous video files and the present set of files. Durations and frequencies of the following behaviors were coded for the entire 5-minute test session: (a) duration spent in each section of the maze (open, closed, and center), and (b) frequency of crosses into the center section. A reduced duration of exploration of the open arms is hypothesized to represent anxiety-related behavior; the frequency of center crosses was evaluated as an index of general physical activity (Grippo et al., 2014; Pellow et al., 1985).

Voluntary Exercise

Voluntary exercise was monitored in the EE treatment conditions to determine the level of long-term physical activity, given variations in EE paradigms used in previous rodent model studies as well as the hypothesis that physical activity may be an important factor in promoting adaptive behaviors and brain function (e.g., Kim et al., 2010; Kobilo et al., 2011; Normann et al., 2021). The daily distance traveled (km/day) and maximum speed during exercise (km/h) in the running wheel were recorded from an odometer adapted for the wheel (Bell F12 Cyclocomputer, Model #7001115, Bell Sports, Van Nuys, CA; Normann et al., 2021; Northern Illinois University, College of Liberal Arts and Sciences Technical Services Shop).

Collection of Brain Tissue and Adrenal Glands

All animals were anesthetized using ketamine (67mg/kg, sc; NLS Animal Health, Ownings Mills, MD) and xylazine (13.33 mg/kg, sc; NLS Animal Health, Owings Mills, MD), and transcardially perfused using phosphate buffered saline and 4% paraformaldehyde. Immediately following euthanasia, the brain was carefully removed and preserved using procedures described in the following section. The adrenal glands were dissected and weighed, as a physiological indicator of long-term social stress (Jarcho et al., 2019).

Processing of Brain Tissue for Golgi Staining

The Golgi-Cox staining method followed procedures outlined previously for evaluation of dendritic morphology in the BLA and hippocampal dentate gyrus (DG; Gibb & Kolb, 1998). Brains were placed in 20 ml of Golgi-Cox solution (Glaser & Van der Loos, 1981) and stored in the dark. The brains were stored in this solution for 14 days, after which time they were placed in 30% sucrose and again stored in a dark environment. The brains remained in the sucrose solution for a minimum of 2 days before being sectioned. Brains were sectioned into 200-µm-thick sections using a Vibratome 1500. Sections were placed on 2% gelatinized microscope slides. The slides were processed using a procedure outline previously (Gibb & Kolb, 1998).

Golgi Analysis Procedures

A Sholl analysis was used for the evaluation of neuronal morphology in the BLA and DG, using reliable, validated procedures described in previous studies, and including specific reliability and validity protocols conducted with prairie vole brain tissue (Gibb & Kolb, 1998; Hylin et al., 2022; Kolb et al., 1998; Reinhart et al., 2021; Zhong et al., 2019). The border of the BLA was identified using Figure 57 in Paxinos and Watson (Paxinos & Watson, 2006; Fig. 1), and the border of the hippocampal DG was identified using Figure 55 in Paxinos and Watson (Paxinos & Watson, 2006; Fig. 2). Interneurons from each the BLA and the DG of the hippocampus were selected manually to ensure that individual neurons within a specific brain region did not overlap with neighboring neurons in the same region. Interneurons were drawn by hand using an Olympus CX31 microscope. Ten neurons (five per hemisphere) were drawn per brain, and were used for the analysis of dendritic branching characteristics and spine density in each brain region as described below. All images were analyzed using ImageJ (National Institutes of Health, Bethesda, MD). Concentric circles were overlaid on the scanned image of the neuron drawing of the BLA (Fig. 3) and DG (Fig. 4), with a 10-μm separation between each circle.

Fig. 1.

Example image of a brain section from the basolateral amygdala (200-μm thickness, 2.5×, scale bar = 500 µm) showing Golgi-Cox-stained interneurons (a), and the relevant cropped section of the brain image from Figure 57 in Paxinos and Watson (2006) for reference (b). Abbreviations in (b) for the present purposes: BLA = basolateral amygdaloid nucleus, anterior; BLP = basolateral amygdaloid nucleus, posterior; additional abbreviations not used here can be found in Paxinos & Watson, 2006, Figure 57. b Republished with permission of Elsevier Science & Technology Journals, from The Rat Brain in Stereotaxic Coordinates, 6^th Edition, Paxinos, G., Watson C., 2006; permission conveyed through Copyright Clearance Center, Inc.

Fig. 2.

Example image of a brain section from the hippocampal dentate gyrus (200-μm thickness, 2.5 ×, scale bar = 500 µm) showing Golgi-Cox-stained interneurons (a), and the relevant cropped section of the brain image from Figure 55 in Paxinos and Watson (2006) for reference (b). Abbreviations in (b) for the present purposes: GrDG = granular layer dentate gyrus; MoDG = molecular layer dentate gyrus; PoDG = polymorph layer dentate gyrus; additional abbreviations not used here can be found in Paxinos & Watson, 2006, Figure 55. b Republished with permission of Elsevier Science & Technology Journals, from The Rat Brain in Stereotaxic Coordinates, 6^th Edition, Paxinos, G., Watson C., 2006; permission conveyed through Copyright Clearance Center, Inc.

Fig. 3.

Example image of a brain section from the basolateral amygdala (200-μm thickness, 40 ×, scale bar = 50 µm) showing Golgi-Cox-stained interneurons (a), and example drawing of an interneuron showing concentric circles overlaid on the image drawing with 10-µm separation between each circle (b)

Fig. 4.

Example image of a brain section from the hippocampal dentate gyrus (200-μm thickness, 40 ×, scale bar = 50 µm) showing Golgi-Cox-stained interneurons (a), and example drawing of an interneuron showing concentric circles overlaid on the image drawing with 10-µm separation between each circle (b)

Dendritic Branching Analysis

Dendritic branching analysis was conducted on interneuron images with overlaid concentric circles, and analyzed with an Olympus CX31 microscope. The amount of dendritic branching within each concentric circle was recorded to provide an estimation of branch extent, and the number of dendrites that crossed each concentric circle was recorded to provide an estimation of dendritic length. First-order branches were defined as branches originating from the soma.

Spine Analysis

Analysis of spine density was conducted using an Olympus BX51 microscope and Ocular software for picture capturing. The oil immersion technique was used for analysis following previously outlined procedures (Glaser & Van der Loos, 1981). Proximal and terminal spine densities were calculated from interneurons in the BLA and DG. Proximal spines were determined from non-terminating first- and second-order branches. Terminal spines were determined from the ends of higher-order branches (fourth ordered branches or higher), 30–40 μm from the center of the soma, that did not bifurcate further.

Data Analyses

Behavioral data and adrenal to body weight ratios were analyzed with two-factor independent groups analyses of variance (ANOVA) using age and housing condition as the independent factors. Absolute body weight was analyzed with a mixed-design ANOVA, with age and housing condition as independent factors, and time as the repeated factor.

Dendritic branching data in the BLA and DG were analyzed with three-factor multivariate analyses of variance (MANOVA) followed by simple main effects analyses, with age, social isolation status, and EE status the independent factors, and the following as dependent variables: (a) all branch orders, (b) total branching, (c) all branch crossings at each distance from the soma, and (d) total distance from the soma. Dendritic spine data in the BLA and DG were analyzed with a MANOVA and simple main effects analyses, using age, social isolation status, and EE status as the independent factors and proximal and terminal spine density as dependent measures.

Physical exercise was evaluated as a potential covariate in initial hypothesis testing, given previous hypotheses suggesting that physical exercise is an important component of EE paradigms, and may confer unique central nervous system benefits (Campeau et al., 2010; He et al., 2012; Kobilo et al., 2011; Normann et al., 2018; Normann et al., 2021). Daily distance traveled and maximum speed reached in the running wheels were used as covariates such these values were used for EE conditions, and zero values were used for all non-EE conditions.

Following initial hypothesis testing, multiple comparisons were conducted with independent-groups Student’s t-tests for statistically justified and hypothesis-driven research questions. The specific pairwise comparisons to be conducted with Student’s t-tests were determined a priori, based on the hypotheses of the study design. Only pairwise comparisons that were justified by initial group comparisons were conducted. A Bonferroni correction was applied to all multiple comparisons to reduce the likelihood of type II errors.

Exploratory, post-hoc analyses were conducted using Pearson’s r correlation coefficients for illustrative purposes (statistical significance is not included). Correlations were computed for each age group to compare the following variables: (a) immobility during the FST vs. open arm exploration duration during the EPM; (b) immobility during the FST vs. adrenal-body weight ratio; (c) open arm exploration duration during the EPM vs. adrenal-body weight ratio; and (d) daily mean distance traveled in the running wheel vs. adrenal-body weight ratio (in the EE conditions). Recommendations from Cohen (1992) were used to interpret a correlation of approximately 0.1 as a small effect, approximately 0.3 as a moderate effect, and approximately 0.5 as a large effect.

Data analyses were conducted using SPSS for Windows, version 27 (IBM, Armonk, NY) and StatPlus for Mac, version v6 (AnalystSoft Inc., Walnut, CA). The present data did not violate the assumptions of homogeneity of variance or independence. A probability value of p < 0.05 was considered to be statistically significant for all initial comparisons using 2-tailed tests; and adjusted probability levels, based on the number of comparisons, were used to determine statistical significance for all Bonferroni-corrected pairwise comparisons with t-tests. Cohen’s d values are reported for all pairwise comparisons (Cohen, 1992).

Results

Forced Swim Test

Behaviors during the FST were evaluated for maladaptive, helpless behaviors, relevant to the construct of depression. The duration of immobility in the FST was evaluated with a two-factor ANOVA, with age and housing condition as the independent factors (Fig. 5). The main effects of housing condition, F(2,48) = 26.40, p < .001, and age ,F(1,48) = 13.32, p < .001, were observed, with a lack of housing condition by age interaction (p > 0.05). Hypothesis-driven, statistically justified follow-up Student’s t-tests, with a Bonferroni correction for multiple comparisons, indicated that social isolation significantly increased immobility duration relative to social pairing for both aging, t(15) = 3.05, p < 0.008; Cohen’s d = 1.5, and young adult, t(13) = 4.61, p < 0.001; Cohen’s d = 2.3, age groups. EE significantly reduced immobility levels relative to social isolation in both aging, t(16) = 3.91, p < 0.001, Cohen’s d = 1.5, and young adult, t(13) = 5.80, p < 0.001, Cohen’s d = 2.3, age groups, such that the immobility levels in the paired and isolated EE groups were comparable for both age groups (p > 0.05 for both comparisons). The immobility levels did not significantly differ by age in the isolated condition (p > 0.05); however, aging prairie voles displayed slightly higher levels of immobility than young adult prairie voles in both the paired, t(14) = 2.27, p < 0.027, Cohen’s d = 1.3, and isolated EE, t(15) = 2.97, p < 0.005, Cohen’s d = 1.5, conditions.

Fig. 5.

Duration of immobility (mean ± SEM; a), and descriptive statistics (range and median; b) during a single 5-min forced swim test in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions. Note that the remainder of 5 min is comprised of active coping behaviors (swimming, struggling, and climbing). For the indicated comparisons: *p < 0.05; **p < 0.01; ***p < 0.001

Elevated Plus Maze

Behaviors during the EPM were evaluated for exploratory behaviors relevant to the construct of anxiety, as well as general motor activity. The duration spent in the open arms of the EPM was evaluated with a two-factor ANOVA, with age and housing condition as the independent factors (Fig. 6a, b). The main effects of housing condition, F(2,48) = 47.05, p < 0.001, and age, F(1,48) = 5.29, p < 0.026, were observed, with a lack of housing condition by age interaction (p > 0.05). Hypothesis-driven, statistically-justified follow-up Student’s t-tests, with a Bonferroni correction for multiple comparisons, indicated that social isolation significantly reduced open arm exploration time relative to social pairing in both aging, t(15) = 6.64, p < 0.001; Cohen’s d = 3.1, and young adult groups, t(13) = 7.03, p < 0.001; Cohen’s d = 1.9. EE significantly increased open arm exploration time relative to social isolation alone in both aging, t(16) = 2.43, p < 0.047; Cohen’s d = 2.8, and young adult groups, t(13) = 5.26, p < 0.001; Cohen’s d = 3.5; however, open arm duration in the isolation EE condition of aging animals was intermediate between pairing and isolation alone. This value was significantly lower than the paired condition in aging animals, t(15) = 4.28, p < 0.010; Cohen’s d = 1.5. By contrast, the duration of time spent in the open arms in the isolated EE condition did not significantly differ from the paired condition in young adult animals (p > 0.05). The open arm duration of the aging EE condition was significantly lower than the young adult EE condition, t(15) = 2.90, p < 0.006; Cohen’s d = 1.2.

Fig. 6.

Duration of exploration of open arms (mean ± SEM; a), descriptive statistics for open arm exploration (range and median; b), duration of exploration of closed arms (mean ± SEM; c), and descriptive statistics for closed arm exploration (range and median; d) during a single 5-min elevated plus maze in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions. For the indicated comparisons: *p < 0.05; **p < 0.01; ***p < 0.001

The duration spent in the closed arms of the EPM was evaluated with a two-factor ANOVA, with age and housing condition as the independent factors (Fig. 6c, d). The main effects of housing condition, F(2,48) = 14.29, p < 0.001 and age, F(1,48) = 7.56, p < 0.008 were observed, with a lack of housing condition by age interaction (p > 0.05). Hypothesis-driven, statistically-justified follow-up Student’s t-tests, with a Bonferroni correction for multiple comparisons, indicated that social isolation significantly increased closed arm exploration time versus social pairing in the aging, t(15) = 3.83, p < 0.001, Cohen’s d = 1.9 and young adult groups, t(13) = 3.73, p < 0.001; Cohen’s d = 1.6. Both EE conditions displayed intermediate values between pairing and isolated alone, such that the value did not differ significantly from either the paired or isolated conditions for either age group (p > 0.05 for all comparisons with a Bonferroni correction). Aging isolated EE animals spent slightly, but significantly, more time in the closed arms than young adult isolated EE animals, t(15) = 2.29, p < 0.031; Cohen’s d = 1.0.

The duration spent in the center section of the EPM was evaluated with a two-factor ANOVA, with age and housing condition as the independent factors (Table 1, left column). A main effect of age was observed, F(1, 48) = 10.89, p < 0.002, such that aging animals spent more time in the center section relative to the young adult animals. No main effect of housing condition and no interaction effect were observed (p > 0.05 for both conditions); no follow-up comparisons were conducted.

Table 1.

Center section exploration during a 5-min elevated plus maze (mean ± SEM), including duration spent in the center section and frequency of crosses through the center section, in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Center section duration (sec)	Center crosses (number)
Young Adult	Paired	102.2 ± 8.3	12.5 ± 2.0
	Isolated	90.5 ± 14.1	9.0 ± 2.2
	EE	68.9 ± 4.4	15.0 ± 2.1
Aging	Paired	68.8 ± 7.6a	15.1 ± 1.6
	Isolated	58.1 ± 13.0a	13.7 ± 2.3
	EE	54.8 ± 9.1a	12.4 ± 3.8

Data shown above are means ± SEM

^ap < .05 vs. young adult conditions for center section duration (main effect of age)

The number of crosses into the center section of the EPM was evaluated with a two-factor ANOVA, with age and housing condition as the independent factors (Table 1, right column). No main effects or interactions were observed (p > 0.05 for all comparisons); no follow-up comparisons were conducted.

Body and Adrenal Weights

A mixed-design ANOVA was conducted to compare body weight at baseline and on the final date of experimental procedures, as an index of physiological responsiveness to social stress. A main effect of age was observed, such that aging animals were significantly heavier than young adult animals, F(3,195) = 87.61, p < 0.001 (Table 2). Neither a main effect of housing condition nor an interaction effect was observed (p > 0.05 for all comparisons). No follow-up tests were conducted.

Table 2.

Body weight (mean ± SEM) in young adult and aging prairie voles at baseline and following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Baseline (g)	Final (g)
Young Adult	Paired	37.5 ± 1.7	39.4 ± 1.6
	Isolated	39.1 ± 2.2	39.9 ± 2.5
	EE	37.9 ± 2.4	38.7 ± 1.8
Aging	Paired	52.9 ± 1.3a	52.1 ± 1.2a
	Isolated	49.4 ± 1.2a	48.8 ± 1.6a
	EE	51.7 ± 1.6a	52.4 ± 1.4a

Data shown above are means ± SEM

^ap < .05 vs. young adult groups at the same time point (main effect of age)

An independent-groups ANOVA was conducted to compare adrenal-to-body weight ratio on the final date of experimental procedures (Fig. 7), as a physiological indicator of long-term adrenal responsiveness to social stress normalized to final body weight. This analysis yielded a main effect of housing condition, F(2,48) = 146.65, p < 0.001, a main effect of age, F(1,48) = 4.24, p < 0.045, and a housing condition by age interaction, F(2,48) = 1.91, p < 0.049. Hypothesis-driven, statistically justified follow-up Student’s t-tests, with a Bonferroni correction for multiple comparisons, indicated that social isolation significantly increased adrenal-to-body weight ratio relative to pairing in both aging, t(15) = 9.64, p < 0.001; Cohen’s d = 4.2 and young adult groups, t(13) = 10.98, p < 0.001; Cohen’s d = 3.0. EE significantly reduced adrenal-to-body weight ratio relative to social isolation for both aging, t(16) = 11.33, p < 0.001; Cohen’s d = 3.9 and young adult groups, t(13) = 10.98, p < 0.001; Cohen’s d = 3.3. For each age group, the ratio in the isolated EE condition did not significantly differ from that of the paired condition (p > 0.05 for both comparisons).

Fig. 7.

Adrenal-to-body weight ratio (mean ± SEM; a) and descriptive statistics (range and median; b) in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions. For the indicated comparisons: ***p < 0.001

Basolateral Amygdala Dendritic Morphology

A three-factor MANOVA conducted on dendritic branching characteristics in the BLA yielded a main effect of age, such that aging prairie voles (relative to young adult prairie voles) exhibited decreased first-order, main effect: F(1,50) = 6.23, p < 0.016], fourth-order, main effect: F(1,50) = 7.64, p < 0.008, and fifth-order dendritic branching, main effect: F(1,50) = 8.85, p < 005; and a main effect of social isolation, such that social isolation (relative to paired conditions) was associated with lower first-order dendritic branching, main effect: F(1,50) = 4.46, p < 0.041 (Table 3). Aging prairie voles (versus young adult prairie voles) also exhibited decreased total branching, main effect: F(1,50) = 10.68, p < 0.002; Fig. 8. Social isolation (versus paired conditions) was associated with fewer dendritic branches at a distance of 20 μm from the soma, main effect: F(1,50) = 5.41, p < 0.017; Table 4 and lower overall dendritic length indicated by number of concentric circles crossed at increasing distances from the soma, main effect: F(1,50) = 6.11, p < 0.017; Fig. 9. No main effects of EE status and no two or three-way interactions were observed (p > 0.05 for all comparisons). No follow-up comparisons were conducted on EE status.

Table 3.

Number of dendritic branches of neurons in the basolateral amygdala (mean ± SEM), organized by branch order, in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Branch order
		1st	2nd	3rd	4th	5th	6th
Young Adult	Paired	5.70 ± 0.20	3.21 ± 0.25	1.85 ± 0.21	0.65 ± 0.15	0.18 ± 0.05	0.10 ± 0.001
	Isolated	5.05 ± 0.15b	2.70 ± 0.26	1.75 ± 0.31	0.45 ± 0.12	0.13 ± 0.02	0.00 ± 0.00
	EE	5.25 ± 0.16b	3.05 ± 0.21	1.73 ± .030	0.51 ± 0.14	0.21 ± 0.05	0.11 ± 0.001
Aging	Paired	5.15 ± 0.20a	2.85 ± 0.23	1.50 ± 0.22	0.42 ± 0.05a	0.05 ± 0.01a	0.00 ± 0.00
	Isolated	4.81 ± 0.14ab	2.86 ± 0.22	1.52 ± 0.22	0.42 ± 0.08a	0.07 ± 0.02a	0.00 ± 0.00
	EE	4.70 ± 0.29ab	2.55 ± 0.19	1.40 ± 0.19	0.31 ± 0.04a	0.06 ± 0.02a	0.10 ± 0.009

Data shown above are means ± SEM

^ap < .05 vs. young adult animals for the same branch order (main effect of age)

^bp < .05 vs. paired animals for the same branch order (main effect of social isolation)

Fig. 8.

The total number of dendritic branches of neurons in the basolateral amygdala (mean ± SEM) in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions. For the indicated comparison: **p < 0.01 (main effect of age)

Table 4.

Number of dendrites of basolateral amygdala neurons (mean ± SEM) crossing concentric circles spaced 10 μm apart, originating from the soma of the neuron, in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Distance from soma (µm)
		10	20	30	40	50
Young Adult	Paired	8.15 ± 0.22	7.60 ± 0.30	2.63 ± 0.25	0.50 ± 0.10	0.06 ± 0.05
	Isolated	7.20 ± 0.25	5.59 ± 0.32b	2.01 ± 0.25	0.40 ± 0.11	0.05 ± 0.04
	EE	7.49 ± 0.23	6.79 ± 0.34b	2.28 ± 0.23	0.43 ± 0.09	0.06 ± 0.05
Aging	Paired	7.51 ± 0.29	6.82 ± 0.33	2.25 ± 0.40	0.51 ± 0.24	0.06 ± 0.05
	Isolated	7.40 ± 0.25	6.48 ± 0.32b	1.83 ± 0.32	0.22 ± 0.08	0.00 ± 0.00
	EE	7.41 ± 0.24	6.09 ± 0.20b	1.60 ± 0.23	0.14 ± 0.05	0.00 ± 0.00

Data shown above are means ± SEM

^bp < .05 vs. paired animals at the same distance from the soma (main effect of social isolation)

Fig. 9.

The total dendritic length of neurons in the basolateral amygdala (mean ± SEM), shown by number of dendrites crossing each concentric circle (each spaced 10 μm apart, originating from the soma of the neuron), in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE). For the indicated comparisons: **p < 0.01 (main effect of social isolation)

A MANOVA conducted on spine density in the BLA yielded a main effect of age. An increase in terminal spine density was observed in aging prairie voles, F(1,41) = 8.50, p < 0.006; Fig. 10; however, age did not influence proximal spine density (p > 0.05). The main effects of isolation status and EE status were not observed (p > 0.05 for all comparisons; no follow-up tests were conducted; data not shown). No two- or three-way interactions were observed (p > 0.05 for all interactions; no follow-up comparisons were conducted; data not shown).

Fig. 10.

Density of proximal and terminal spines in the basolateral amygdala (mean ± SEM) in young adult and aging prairie voles following 4 weeks of housing conditions (housing conditions pooled for each age group). For the indicated comparison: **p < 0.01 (main effect of age)

Hippocampal Dentate Gyrus Dendritic Morphology

A three-factor MANOVA conducted on dendritic branching characteristics in the DG yielded main effects of age, such that aging prairie voles exhibited decreased branching for fourth, main effect: F(1,48) = 6.68, p < 0.013, fifth, main effect: F(1,48) = 6.17, p < 0.017, and sixth order branches, main effect: F(1,48) = 4.4, p < 0.042 (Table 5), and decreased total branching, main effect: F(1,48) = 15.18, p < 0.001; Fig. 11. However, aging prairie voles exhibited increased branch length at 40 μm, main effect: F(1,48) = 36.37, p < 0.001, 50 μm, main effect: F(1,48) = 30.77, p < 0.001, and 60 μm from the soma, main effect: F(1,48) = 15.27, p < 0.001 (Table 6), and increased total branch length indicated by number of concentric circles crossed at increasing distances from the soma, main effect: F(1,48) = 15.29, p < 0.001; Figure 12. Neither isolation nor EE status yielded main effects, and no two- or three-way interactions were observed (p > 0.05 for all comparisons). No follow-up comparisons were conducted on EE status.

Table 5.

The number of dendritic branches of neurons in the hippocampal dentate gyrus (mean ± SEM), organized by branch order, in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Branch order
		1st	2nd	3rd	4th	5th	6th
Young Adult	Paired	2.00 ± 0.41	1.60 ± 0.21	1.78 ± 0.17	1.31 ± 0.10	0.45 ± 0.13	0.12 ± 0.09
	Isolated	2.66 ± 0.07	1.92 ± 0.08	1.70 ± 0.15	0.90 ± 0.34	0.33 ± 0.30	0.13 ± 0.10
	EE	2.60 ± 0.31	1.81 ± 0.15	1.55 ± 0.15	0.99 ± 0.21	0.31 ± 0.24	0.10 ± 0.11
Aging	Paired	2.23 ± 0.37	1.65 ± 0.14	1.45 ± 0.29	0.59 ± 0.33a	0.18 ± 0.22a	0.02 ± 0.03a
	Isolated	2.41 ± 0.25	1.77 ± 0.06	1.56 ± 0.16	0.72 ± 0.32a	0.25 ± 0.21a	0.03 ± 0.04a
	EE	2.29 ± 0.07	1.48 ± 0.21	1.72 ± 0.10	0.85 ± 0.19a	0.21 ± 0.24a	0.01 ± 0.01a

Data shown above are means ± SEM

^ap < .05 vs. young adult animals for the same branch order (main effect of age)

Fig. 11.

Total number of dendritic branches of neurons in the hippocampal dentate gyrus (mean ± SEM) in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions. For the indicated comparison: ***p < 0.001 (main effect of age)

Table 6.

The number of dendrites of hippocampal dentate gyrus neurons (mean ± SEM) crossing concentric circles spaced 10 μm apart, originating from the soma of the neuron, in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE) conditions

		Distance from soma (µmm)
		10	20	30	40	50	60
Young Adult	Paired	3.44 ± 0.05	6.01 ± 0.02	4.13 ± 0.86	1.32 ± 0.78	0.10 ± 0.90	0.01 ± 0.30
	Isolated	3.69 ± 0.33	5.92 ± 0.20	4.24 ± 0.25	1.04 ± 0.12	0.04 ± 0.03	0.00 ± 0.00
	EE	3.75 ± 0.40	5.90 ± 0.10	5.40 ± 0.26	2.21 ± 0.70	0.11 ± 0.66	0.01 ± 0.01
Aging	Paired	3.01 ± 0.24	5.12 ± 0.15	5.32 ± 0.10	4.52 ± 0.002a	2.11 ± 0.31a	0.51 ± 0.08a
	Isolated	3.02 ± 0.22	5.04 ± 0.16	5.38 ± 0.13	4.51 ± 0.03a	2.42 ± 0.10a	0.56 ± 0.11a
	EE	3.13 ± 0.16	5.25 ± 0.22	5.52 ± 0.12	4.51 ± 0.02a	2.89 ± 0.08a	0.50 ± 0.23a

Data shown above are means ± SEM

^ap < .05 vs. young adult animals at the same distance from the soma (main effect of age)

Fig. 12.

Total dendritic length of neurons in the hippocampal dentate gyrus (mean ± SEM), shown by number of dendrites crossing each concentric circle (each spaced 10 μm apart, originating from the soma of the neuron), in young adult and aging prairie voles following 4 weeks of paired, isolated alone, or isolated with environmental enrichment (EE). For the indicated comparison: ***p < 0.001 (main effect of age)

A MANOVA conducted on spine density in the DG yielded a main effect of age; a reduction in proximal spine density was observed in aging prairie voles, F(1,49) = 10.87, p < 0.002; Fig. 13; however, age did not influence terminal spine density (p > 0.05). No main effects of isolation or EE status were observed; and no two- or three-way interactions were observed (p > 0.05 for all comparisons; no follow-up comparisons were conducted; data not shown).

Fig. 13.

Density of proximal and terminal spines in the hippocampal dentate gyrus (mean ± SEM) in young adult and aging prairie voles following 4 weeks of housing conditions (housing conditions pooled for each age group). For the indicated comparison: **p < 0.01 (main effect of age)

Voluntary Exercise

Physical activity was evaluated in the EE conditions, given previous variations in EE paradigms and the potential for long-term physical activity to influence behavioral and neural responses to EE. Daily distance traveled and maximum speed on the running wheel were compared as a function of age with hypothesis-driven Student’s t-tests. Distance traveled did not differ (means ± SEM, aging = 1.2 ± 0.6 km/day; young adult = 1.8 ± 0.7 km/day; p > .05); however, the aging animals reached a lower maximum speed than the younger animals, means ± SEM, aging = 0.7 ± 0.2 km/h; young adult = 1.8 ± 0.3 km/h; t(15) = 2.9, p < .01, Cohen’s d = 1.7. Neither daily distance traveled nor maximum speed reached in the running wheels in the EE conditions (relative to zero values for conditions that did not include EE) statistically influenced any of the above dependent measures.

Exploratory Correlations by Age

To gain a better understanding of the relationships between specific behavioral and physiological variables in the present study, the following correlations were computed for each age group: (a) immobility during the FST vs. open arm exploration duration during the EPM (aging: r = −.43; young adult: r = −.61); (b) immobility during the FST vs. adrenal-body weight ratio (aging: r = .69; young adult: r = .78); (c) open arm exploration duration during the EPM vs. adrenal-body weight ratio (aging: r = −.49; young adult: r = −.83); and (d) daily mean distance traveled in the running wheel and adrenal-to-body weight ratio (EE groups only; aging: r = .03; young adult: r = −.88). All correlations were larger in the young adult animals relative to the aging animals.

Discussion

Given the interactions of age, social status, and stress, the prairie vole model was used in the present study to investigate negative consequences of social stress, as well as potential protective effects of EE, in two developmental age groups. Behavioral and physiological indicators of long-term social stress were observed in the present study, coupled with neural structural changes in regions of the brain associated with stress activation and inhibition, as a function of age and social isolation. Social isolation increased depression- and anxiety-related behaviors in young adult and aging prairie voles. Improvements in depression-related behaviors were observed in both age groups following EE; however, young adult prairie voles exhibited a greater reduction in anxiety-related behaviors after EE relative to aging prairie voles. A physiological indicator of stress — long-term adrenal function measured via adrenal-to-body weight ratios — indicated that social isolation increased physiological stress responsiveness in both age groups, and that EE similarly reduced adrenal-to-body weight ratio in both age groups. Dendritic morphology in the BLA was altered as a function of both social isolation and age, whereas dendritic morphology in the DG of the hippocampus was altered as a function of age only. However, EE did not influence BLA or DG morphology, suggesting that the emotion and stress-related benefits of EE may not be associated with dendritic structural changes in these subregions.

Negative affective behavioral measures relevant to emotional responsiveness and stress were operationalized via the FST and EPM, which are hypothesized to be valid and reliable measures of behaviors relevant to the constructs of depression and anxiety (Donovan et al., 2020; Grippo et al., 2012; Normann et al., 2021). Prairie voles exposed to social isolation displayed greater durations of immobility in the FST versus paired prairie voles, with no age difference. These results support previous data indicating that rodents exposed to social isolation display increased immobility in the FST, and contribute to a body of literature that discusses negative consequences of social isolation and long-term social stress on emotion-related behaviors in animal models, including in young adult and aging prairie voles (Grippo et al., 2012; Grippo et al., 2021; Normann et al., 2021; Panossian et al., 2020; Zanier-Gomes et al., 2015). Behavioral and physiological disruptions related to depression have been observed in rats and prairie voles of both sexes following long-term social isolation, including altered immobility in the FST, anhedonia in a sucrose consumption task, altered weight gain, disrupted corticosterone and altered negative feedback of the hypothalamic-pituitary-adrenal axis, disrupted cardiac structure and function, and altered circadian rhythmicity of behavioral and cardiovascular variables (Carnevali et al., 2012, 2020; de Jong et al., 2005; Serra et al., 2005; Sun et al., 2014; Zanier-Gomes et al., 2015). More specifically in prairie voles, social isolation and the disruption of social bonds influence social behaviors — which may be increased or decreased depending on the social history and experimental protocol — such as disrupting social bonds, altering affiliative behaviors, and increasing stress reactivity in social behavioral tasks (Donovan et al., 2020; Normann et al., 2021; Sun et al., 2014). The present data also mirror increased depression in socially isolated and aging humans (Luchetti et al., 2020; Palgi et al., 2020; Xia & Li, 2018).

Socially isolated prairie voles of both age groups (versus paired) displayed anxiety-related consequences indicated by decreased exploration of the open arms and increased exploration of the closed arms in the EPM, without a change in general physical activity shown via a lack of difference in crosses of the center section of the maze. These data are similar to anxiety-related consequences, shown via altered exploratory behavior in operational tasks of exploration, in young adult male and female prairie voles isolated from a same-sex sibling (Grippo, Wu, et al, 2008; Grippo et al., 2014; Hylin et al., 2022), male prairie voles isolated from a female partner (Sun et al., 2014), and male mice separated from a group (Ieraci et al., 2016). However, the present results are contrary to others demonstrating no effect of social isolation on EPM behaviors in aging (vs. young adult) mice, or in young adult isolated (vs. paired) prairie voles (Bosch et al., 2009; Panossian et al., 2020). The behavioral differences between the current and previous studies may be due to differential responses to social isolation as a function of species differences or duration of social isolation.

The present results further establish that social isolation produces negative affective behaviors that are relevant to depression and anxiety syndromes in humans. Social pairing was protective in the present study — which may be considered an ideal environmental condition for promoting emotional health (Normann et al., 2018). However, if an ideal social environment is not feasible, EE may serve a protective role against negative affective behaviors. EE reduced depression-related behaviors in both age groups (versus isolation alone), to levels that were comparable to paired conditions. These data are consistent with findings that EE protects against depression-related consequences of social isolation in young adult female prairie voles separated from a female sibling (Grippo et al., 2014) or a male partner (Normann et al., 2018). These findings also support other stress reduction paradigms in developmental rodent models (Brenes et al., 2009; Dong et al., 2018; Singhal et al., 2019; Torres-Lista & Giménez-Llort, 2015). However, aging paired and isolated prairie voles exposed to EE exhibited slightly higher levels of immobility relative to young adult prairie voles in the same environmental conditions (see Fig. 5). This pattern of responding may suggest that aging itself is associated with altered behavioral responses in the FST or physiological functions that mediate swimming ability; the lack of an age difference in immobility in aging and young adult isolated conditions may have been due to a ceiling effect in this variable. Further, EE may have differentially influenced aging and young adult prairie voles, given that aging prairie voles displayed greater body weights and slightly lower maximum speeds in the running wheels relative to young adult prairie voles. Therefore, aging prairie voles may not have received the same physical benefits from EE as the young adult prairie voles due in part to altered anatomical or physiological functions such as declining muscle health or endurance. Previous social isolation paradigms in young adult female prairie voles indicate that both EE with an exercise component and exercise alone reduce immobility in the FST to similar levels (Grippo et al., 2014), and that removal of exercise from the EE paradigm (relative to EE with an exercise component) does not interfere with the benefits of EE on improving immobility levels in the FST (Normann et al., 2021). These previous studies suggest that exercise may be an important (but perhaps not essential) environmental strategy for achieving benefits in depression-relevant behaviors. The exercise component alone was not evaluated directly in the present design; therefore, the behavioral effects of the physical, cognitive, and sensory activities cannot be separated.

In contrast to depression-related improvements across age groups, EE improved exploration time in the EPM in young adult prairie voles by increasing open arm duration and reducing closed arm duration to levels comparable to paired prairie voles, indicative of reduced anxiety-like behaviors. This pattern of improvement in the EPM is similar to the improvement as a function of EE (involving cognitive, physical, and social activities) in young adult mice exposed to an environmental stressor (Novaes et al., 2017), and the improvement as a function of EE (including cognitive, sensory, and physical activities) in young adult prairie voles exposed to social isolation (Grippo et al., 2014). However, aging prairie voles exposed to EE displayed intermediate exploration values between those of paired and socially isolated alone (see Figure 6). Differential correlations between behaviors in the FST and EPM were observed between the age groups with a stronger correlation in young adults, suggesting that the relationship between depression- and anxiety-related behaviors may vary across development. Additionally, it is possible that the greater intensity of exercise in young adult prairie voles mediated improvements in anxiety-related behaviors, perhaps by influencing neural or behavioral responses to a greater extent in this age group. The potential interaction between physical exercise and age observed here is consistent with previous partial benefits of EE involving only cognitive and sensory activities on anxiety-related behaviors. For example, EE with an exercise component increased open arm exploration duration in the EPM in young adult female socially isolated prairie voles (relative to social isolation without EE), but EE that lacked a physical exercise component did not produce the same improvement in this behavioral outcome measure (Normann et al., 2021). These previous data, coupled with the present set of results in the EPM, suggest that physical exercise may be a critical environmental strategy that is associated with improved anxiety-related behavioral outcomes. Consistent with this hypothesis, higher-intensity exercise may be more effective than lower-intensity exercise for improving anxiety in humans (Aylett et al., 2018).

Dendritic morphology was evaluated in the BLA and DG given the importance of these regions in the context of stress, social behavior, and age (Almaguer et al., 2002; Ashokan et al., 2016; Ehninger et al., 2011; Gualtieri et al., 2017; Lieberwirth et al., 2012). Altered dendritic branching in the BLA was observed as a function of both age and social isolation. Aging prairie voles displayed significantly reduced dendritic branching extent and length versus young adult prairie voles. Aging prairie voles also exhibited greater terminal spine density relative to young adult prairie voles. Further, isolated prairie voles displayed decreased branching compared to paired voles. The BLA morphological changes as a function of age and social isolation mirror previously observed changes in BLA morphology in isolated young adult male prairie voles relative to social pairing (Hylin et al., 2022). Considered together, the current data and those from Hylin et al. (2022) provide evidence that long-term social stress is associated with altered dendritic structure in the BLA, similar to the effects of acute stress previously observed in rats and mice (Qiao et al., 2016). Various stress-relevant brain regions communicate with the BLA, including the hippocampus and prefrontal cortex, and the BLA plays an important role in processing negative emotions such as fear and anxiety (Herman et al., 2005; Hostinar et al., 2014; McEwen, 2017; McKlveen et al., 2015). Therefore, it is possible that altered dendritic morphology and increased spine density in the BLA reflect differential age-related plasticity properties and increased stability of communication in this region (see for instance Forrest et al., 2018; Hercher et al., 2010). For instance, age-related differences in amygdala dendritic arborization and length have been observed in adolescent vs. adult rats exposed to social instability stress (Tsai et al., 2014). Changes in the responsiveness of the amygdala to stress-relevant information across development might be associated with impaired stress coping and increased negative emotions following social stressors.

Aside from the BLA, dendritic morphology in the DG was influenced primarily by age. Aging prairie voles displayed significantly reduced branching in the DG similar to the pattern observed in the BLA, but longer branch lengths compared to younger animals. Further, contrary to the BLA pattern, aging prairie voles exhibited decreased spine density relative to young adult prairie voles in the DG. These DG-specific alterations may indicate an inverse relationship between DG morphology and age, suggesting that age or developmental processes are associated with neurobiological changes in this region. For instance, aging itself is not only associated with general structural and functional changes in the hippocampus, but also involves neurodegenerative disruptions that may influence cognitive processes or responses to stress (Bartsch & Wulff, 2015; Fjell et al., 2014). Given the role of the hippocampus in negative feedback mechanisms associated with stress (Herman et al., 2005; Hostinar et al., 2014), it is possible that reduced dendritic branching and spine density in the DG observed in aging animals is correlated with emotional vulnerability to stressors or impaired stress-coping strategies. However, it is also possible that sex interacts with age-related changes in the brain. Male prairie voles were specifically used here based on considerations of possible emotional vulnerability in men as a function of social stress (Förster et al., 2019) and to contribute to a body of literature focused on the influence of social stress on behavioral responses as well as central nervous system morphology in young adult and aging male prairie voles (Grippo et al., 2021; Hylin et al., 2022). Therefore, the present specific results may not translate to female subjects. Sex differences exist in hippocampal structure and functions in both humans and animal models (McEwen, 2002; Yagi et al., 2020; Yagi & Galea, 2019); therefore, EE may influence DG structure or stress circuitry communication differentially in males and females. A more detailed analysis of functional and structural changes in stress circuitry in both sexes — not only in the amygdala and hippocampus but also in other stress-relevant brain regions such as the hypothalamus, bed nucleus of the stria terminalis, and frontal cortex — may further inform our understanding of central mechanisms underlying the influence of social isolation on behavior and emotions across developmental trajectories.

Contrary to the hypotheses of the present study, EE did not influence dendritic morphology in the BLA or DG. Despite human studies suggesting that EE strategies involving cognitive, social, and/or physical activities may improve cognitive function and emotional states in older individuals (Casemiro et al., 2016; Park et al., 2019), and rodent studies demonstrating benefits of EE paradigms with similar species-relevant cognitive, sensory, physical, and/or social activities on stress-related brain structure and function (Kolb et al., 2003; Novaes et al., 2017; Segovia et al., 2006; Sztainberg et al., 2010; Watanasriyakul et al., 2019), the present data do not support the hypothesis that improvements in depression- and anxiety-related behaviors and adrenal responsiveness in isolated prairie voles exposed to EE are associated with dendritic changes in the BLA or DG. Further research is necessary to determine structural or functional changes in stress-relevant brain regions that may underlie the benefits of EE in a social context, including specific causal experimental designs.

In conclusion, social isolation increases negative affective behaviors in both young adult and aging prairie voles, including contributing to altered depression- and anxiety-relevant behaviors in operational behavioral tests. These behavioral changes are supported by increased long-term adrenal responsiveness, suggesting that depressive- and anxiety-related consequences of social isolation are associated with similar adrenal responses in young adult and aging prairie voles. EE reduces depressive behaviors and adrenal-to-body weight ratios in both age groups, but may be slightly more effective at reducing anxiety-related behaviors in young adult vs. aging prairie voles. It is possible that aging prairie voles did not experience the same physical benefits of EE as young adult prairie voles or that EE was not as effective at changing neurobiological and behavioral responsiveness to stress due to age-related neural changes or neurodegenerative disruptions, potentially influencing both depression- and anxiety-related behaviors. However, inconsistent with the behavioral and adrenal improvements as a function of EE in socially isolated young adult and aging prairie voles, EE did not significantly influence measures of dendritic structure in the BLA or hippocampal DG. Additional research is necessary to understand central mechanisms underlying the protective effects of EE on behavioral and physiological consequences of social isolation. For example, future research may focus on structural and functional analyses of stress-related regions such as additional amygdala and hippocampal subregions, hypothalamus subregions, and cortical regions in both males and females. Causal research designs will also be important to determine specific mediation or moderation of behavioral and physiological alterations as a function of social stressors. The present study and continued research using valid and reliable animal models will inform strategies to protect against negative consequences of social stress humans.

Additional Information

Funding Information

This research was funded in part by National Institutes of Health grants HL112350 and HL147179. The funder had no role in the study design, data collection, data analysis and interpretation, writing of the report, or decision to submit the article for publication.

Conflicts of Interest

The authors declare no competing interests.

Data Availability

Data for the project can be obtained here: https://osf.io/d29zg/.

Author Contributions

All authors contributed significantly to the study conceptualization, study design, data collection, analyses, interpretation, and/or writing of the manuscript. All authors have approved the final version of the submitted manuscript.

Ethical Approval

All procedures described in this manuscript were approved by the Northern Illinois University Institutional Animal Care and Use Committee, and are in compliance with all federal guidelines. The methods conform to the Guide for the Care and Use of Laboratory Animals from the Institute of Laboratory Animal Research. All efforts were made to ensure that procedures were conducted in a responsible and ethical manner in the context of the Three Rs (replacement, reduction, and refinement).

Informed Consent

Not applicable.