Impact of housing conditions on social behavior, neuroimmune markers, and oxytocin receptor expression in aged male and female Fischer 344 rats.

Abstract

Aging is associated with a substantial decline in social behavior, whereas positive social interaction can improve overall health in aged individuals. In laboratory rodents, manipulations of the social environment across the lifespan have been shown to affect social behavior. Therefore, we examined the effects of long-term (5-6 weeks) housing conditions (alone, with one adult, or with two adults) on social behavior and the expression of neuroinflammation-related genes as well as oxytocin receptor (OXTR) gene expression in brain areas associated with social behavior regulation in aged male and female Fischer (F) 344 rats. Single-housed males and females exhibited increased social investigation, relative to pair-housed rats (one aged and one adult). Triple-housed (one aged and two adults) aged males exhibited lower levels of social investigation, relative to triple-housed aged females. Aged females were more socially active that their male counterparts. Although social housing condition significantly affected social behavior in males, it had no impact on cytokine gene expression in the paraventricular nucleus of hypothalamus (PVN), bed nucleus of the stria terminalis (BNST) or medial amygdala (MeA). However, in triple-housed aged females, who exhibited social behavior comparable to their single- and pair-housed counterparts, there was a significant increase in the expression of IL-1β and IL-6 mRNA in the MeA. No changes in cytokine gene expression were observed in the PVN or BNST, indicating that the increased expression of cytokines in the MeA was not a result of a generalized increase in neuroinflammation. Single-housed males and females exhibited elevated OXTR gene expression in the BNST. Taken together, these data indicate that manipulations of the social environment in late aging significantly influenced social interactions with a novel partner and gene expression in social behavior circuits and that these effects are sex-specific.

1.0. Introduction

In humans, aging is accompanied by a decline in the expression of social behavior [1,2], with similar social alterations evident in rodent models of aging [3-9]. Perturbations of the social environment may underlie some of the negative consequences of aging, given that social isolation is closely related to age-related debilitation [10-12]. On the other hand, the presence of strong social ties can enhance functioning even in late aging [12-17]. Interventions designed to enhance social interaction in aged individuals have been shown to increase well-being [18], life satisfaction [18], and overall health [19-21]. Even social interaction via telephone increases well-being, providing social support for individuals who may have age-related debilitation that prevents travel [22]. Thus, understanding the mechanisms of aging-related declines in social behavior is important for devising strategies to enhance social behavior in elderly individuals.

Rodent models have proven extremely useful in assessing social behavioral alterations associated with aging, although the mechanisms involved in age-related decline of social behavior are still not fully understood. Studies have manipulated several factors that influence social interaction, including the age of the social partner and housing conditions prior to testing. For example, relative to adult males, aged males exhibit reduced investigation of juvenile [3,4], adult [3], and age-matched conspecifics [3,8,9]. Similarly, in aged females, social behavior directed toward adult [3] and age-matched social partners is lower than in their adult counterparts [5], although to our knowledge social behavior directed toward sex-matched juveniles has not been assessed. These data demonstrate a robust decline in social interaction in aged males and females that is largely unaffected by the age of the social partner.

Short-term social isolation (e.g. 1-5 days) is often used as a strategy to increase social interaction in rodents. However, long-term social isolation (e.g., 2-6 weeks) produces substantial behavioral disruptions, particularly when imposed during critical developmental periods such as adolescence [23]. Post-weaning social isolation (e.g. isolation rearing) increases aggression [24,25], novelty-induced hyperactivity [26], and anxiety-like behaviors [26]. In addition, isolation rearing produces alterations in social behavior [9,27], impaired social recognition [28], and attenuates immediate early gene expression in the medial prefrontal cortex (mPFC) following social interaction with a novel conspecific [25]. Although environmental enrichment attenuated some of the behavioral (e.g., anxiety-like behavior, spatial learning alterations) consequences of isolation rearing [29], the impact of environmental enrichment on isolation-induced social deficits has yet to be determined. In addition, few studies have examined the impact of housing conditions on social behavior in aged rats. One study found that 2 weeks, but not 4 weeks, of social isolation increased sniffing duration in aged male F344/N rats [9]. It was recently reported that post-weaning group housing rescued cognitive deficits and decreased anxiety-like behavior in aged males [30]. However, it has yet to be determined how long-term manipulations of housing conditions impact behavior in aged females.

Oxytocin (OXT) is a neuropeptide that plays an important role in complex social behaviors [31,32]. Neurons that produce OXT can be found in the paraventricular nucleus of the hypothalamus (PVN) and supraoptic nucleus (SON) [33] and OXT release is elevated during parturition, suckling, and social defeat [34]. Furthermore, OXT release within the bed nucleus of the stria terminalis (BNST) is elevated in males, but not females, during social recognition [35]. Thus, there is a clear role for OXT in several aspects of social behavior, although the role of OXT in social behavior in late aging is poorly understood. Studies have demonstrated age-related increases in the size of nucleoli of OXT neurosecretory cells [36], suggesting increased production and/or trafficking of OXT. In aged rats, forced swim blunted OXT release in PVN and SON [37]. However, to our knowledge, no study has examined how manipulation of the social environment influences the OXT brain system in aged rats.

Social housing conditions can influence neuroinflammation in adult and aged rodents. In adult mice, 2 weeks of social isolation was sufficient to exacerbate ischemia-induced neuronal cell death and inflammation [38]. Social isolation for 2 weeks also increased infarct size after middle cerebral artery occlusion (MCAO). This was accompanied by increased striatal glial-fibrillary acid protein (GFAP) and decreased interleukin-6 (IL-6) gene expression [39]. In aged rats, social isolation for eight weeks increased anxiety-like behavior, as indexed by decreased time spent in open arms in an elevated plus maze, in addition to reducing splenic immune activity [40]. In contrast, long-term environmental enrichment has been shown to delay the expression of several behavioral and neural markers of aging [41]. However, these studies typically employ environmental enrichment from the age of about 3 months until senescence (e.g. ~ 20 months).

Therefore, the present study was designed to determine whether a shorter period (6 weeks) of social isolation versus social housing with 1 or 2 young adult conspecifics would impact social behavior and inflammation in aged male and female rats. While housing with young adults is not considered environmental enrichment, per se, it is certainly a form of social enrichment. In contrast, prolonged social isolation produces moderate signs of distress in young rodents [24-26], though the influence of prolonged social isolation has not been systematically evaluated among well-socialized aged rats of both sexes. The present study had several objectives: (i) examine whether long-term social isolation produced significant disruptions in social behavior in aged rodents, (ii) determine whether housing aged rodents with young conspecifics facilitated social interaction with a novel conspecific, and (iii) determine how housing condition altered expression of several neuroimmune related as well as OXT receptor (OXTR) genes in the brain.

2.0. Material & Methods

2.1. Subjects

Eighteen-month-old male and female F344 rats (n = 8-10/group) were acquired from the NIA colony at Charles River Laboratories. F344 is a commonly used strain in aging studies, evidenced by the fact that F344 is one of only a few selected rat lines maintained by the National Institute on Aging (NIA) for use in NIA-funded projects, with a growing body of literature on social behavior processes in F344 rats emerging in recent years [3,42,43]. Male and female housing partners were ordered from a separate Charles River Facility and were 60-70 days old at the time of arrival. Colony conditions were maintained at 22 ± 1°C; 12:12 light:dark cycle with lights on at 0700. Animals were housed in standard polycarbonate cages. Animals were treated in accordance with guidelines set forth by the Institute of Laboratory Animal Resources, (1996) and with the protocol approved by the IACUC at Binghamton University.

2.2. Housing Conditions

Upon arrival (Week 0), experimental subjects were randomly assigned to one of three housing conditions: single-, pair-, or triple-housed; conditions were maintained for 5-6 weeks (Figure 1A). Subjects in the pair- and triple-housed conditions were housed with 1 or 2 young adult sex-matched conspecifics, respectively. We chose to use young conspecifics as social housing partners for two reasons: (1) there are limits in the number of animals that can be obtained from NIA colonies and (2) this approach had the advantage of reducing the potential for aggression toward cage mates, particularly in males. Housing partners were used as social partners during social interaction testing, but were not tested with their cage mates, and the housing condition of the social partner was counterbalanced across experimental subjects. Weights were obtained upon arrival and weekly throughout the study to monitor health status.

Figure 1.

(A) Schematic representation of the experimental timeline. Aged rats were re-housed upon arrival and those housing conditions were maintained for 5-6 weeks, at which time social interaction was recorded with a novel young adult partner. Aged rats were returned to their original housing conditions that were maintained for 1 week before tissue collection for analysis of gene expression in socially-relevant brain structures. Social investigation (B), social contact (C) and social preference/avoidance (D) of the aged experimental subjects. Individual housing enhanced social investigation in aged males and females relative to their pair-housed counterparts (#, p < 0.05 vs. pair-housed). Triple-housed aged females showed higher levels of social investigation than triple-housed aged males (*, p < 0.05 vs. males). Aged females exhibited higher levels of social contact behavior and social preference (^, p < 0.05), relative to aged males. Social investigation (E), social contact (F), and social preference/avoidance (G) of adult partners. Sex differences in contact behavior were observed in rats that were pair- or triple-housed (^, p <0.05), but not in those that were single-housed. Data are expressed as mean ± SEM.

2.3. Social Interaction Testing

2.3a. Apparatus

Behavioral testing was conducted between 0800 and 1200 h under dim light (15-20 lux) as previously described [3,42]. Briefly, males and females were tested in separate, but adjacent, rooms. Testing was conducted across several days to accommodate the large number of animals being tested, and the room in which males and females were tested was counterbalanced across squads of animals. As in our previous studies [3,42,43], rooms and chambers were cleaned thoroughly between squads of animals to remove any odor cues that may influence behavioral testing. Animals were placed in Plexiglas® chambers (Binghamton Plate Glass, Binghamton, NY) that measured 45 × 30 × 30 cm. Each test chamber was divided into two equally sized compartments by a clear Plexiglas® partition containing an aperture (9 × 7 cm) that allowed for movement of animals between the compartments. Clean wood shavings were placed in the bottom of the apparatuses. After testing of each subject, the soiled wood shavings were removed, chambers were cleaned with water, and clean shavings were added for the next subject. At the end of each day of testing, chambers were cleaned with a 3% hydrogen peroxide solution to remove any odors. All behavioral testing was video recorded for later scoring by a trained observer without knowledge of the housing condition of any given animal.

2.3b. Behavioral Measures

Social investigation frequency was defined and scored as the sniffing of any part of the body of the partner, whereas frequency of contact behavior was scored as the sum of crawling over and under the partner and social grooming. The total number of crossovers (movement through the aperture) demonstrated by the experimental subjects was also recorded for each session. Each crossover was coded as toward or away from the social partner. These data were used to calculate a coefficient of social approach/avoidance [44]. This coefficient was calculated as: (crossovers to the partner – crossovers away from the partner) / (total number of crossovers both to and away from the partner) × 100. Total crossovers were used as an index of general locomotor activity under social test circumstances [3,42].

2.4. Experimental Design

Housing conditions were maintained for 5-6 weeks before behavioral testing occurred. Testing occurred across two days. On the first day, animals were pre-exposed to the testing apparatus for 30 minutes to acclimate them to the novel environment. The following day, each experimental subject was placed alone into the apparatus for a 20-minute period [45],, after which a novel adult sex-matched conspecific was placed into the apparatus for a 10-minute social interaction test. Following testing, animals were returned to their home cages and housed with their cage mates for ~1 week before tissue collection.

2.5. Tissue Collection and Processing

Animals were taken from their home cages and rapidly decapitated under low stress conditions. Trunk blood was collected into EDTA-coated glass tubes (BC Vacutainers, VWR Cat. No. VT6450, Radnor, PA). Blood samples were centrifuged at 3300 RPM (4° C) and plasma was separated and frozen at −80° C. Brains were removed, bisected, and one hemisphere was flash-frozen in methylbutane (EMD Millipore, Darmstadt, Germany; cat. no. MX0760-1) on dry ice and stored at −80° C until processing. To isolate regions of interest (ROI), brains were sectioned on a cryostat and ROIs were obtained using biopsy punches (BNST: 2 mm × 1 mm; PVN: 1.2 mm × 1 mm; MeA: 1.2 mm × 1 mm) according to the stereotaxic rat brain atlas [46] and collected into 2.0 mL RNase-free tubes (Eppendorf; Cat. No. 022363352). Tissue was homogenized using a TissueLyser (Qiagen, Valencia, CA) and RNA was extracted using an RNeasy Mini Kit (Qiagen; Cat. No. 74106) as previously described [42]. The quantity and purity of RNA was assessed with a Nanodrop spectrophotometer (Nanodrop 2000, Thermo Scientific, Wilmington, DE) and stored at −80° C until cDNA was synthesized.

2.6. Reverse-Transcription Polymerase Chain Reaction

cDNA was synthesized using a QuantiTect reverse transcription kit (Qiagen) and RT-PCR was performed using procedures described elsewhere [42,47]. All RT-PCR was run with a CFX384 real-time PCR detection system (Bio-Rad). Primer sequences and accession numbers can be found in Table 1. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a reference gene in all analyses. Expression was calculated using the 2^ΔΔCT method as previously described [48].

Table 1.

Primer sequences and accession numbers used in RT-PCR

Gene Target	Accession #	RT-PCR primer sequences	Function
IL-1βa	NM_031512.2	Forward:5’- AGGACCCAAGCACCTTCTTT-3’ Reverse:5’- AGACAGCACGAGGCATTTTT-3’	Pro-inflammatory cytokine involved in sickness behavior and fever induction
IL-6b	NM_012589.2	Forward:5’- TAGTCCTTCCTACCCCAACTTCC-3’ Reverse:5’ TTGGTCCTTAGCCACTCCTTC-3’	Cytokine produced in response to injury or infection; can exhibit both pro- and anti-inflammatory properties
TNFαc	NM_012675.3	Forward:5’- AACCACCAAGCAGAGGAGCA-3’ Reverse:5’- ATGGCAAATCGGCTGACGGT-3’	Pro-inflammatory cytokine involved in activation of immune cells during immune response
CD14d	NM_021744	Forward:5’- AACAACGGATACCTGGCTCG-3’ Reverse:5’- GTCCTTTCTCGCCCAACTCA -3’	Co-receptor with TLR4e. Recognizes pathogen-associated molecular patterns.
OXYRf	NM_012871	Forward:5’ - AAGGAAGCTTCTGCCTTCATCA-3’ Reverse:5’-TCTTGCTGACACTCGTCTCG – 3’	Receptor for oxytocin, a neuropeptide with a major role in social behavior
GAPDHg	NM_017008	Forward:5’- GTGCCAGCCTCGTCTCATAG-3’ Reverse:5’- AGAGAAGGCAGCCCTGGTAA-3’	Used as a stable reference gene for expression of target genes; critical for cellular metabolic function

^aInterleukin-1 beta;

^bInterleukin-6;

^cTumor necrosis factor alpha;

^dCluster of differentiation 14;

^eToll-like receptor 4;

^fOxytocin receptor;

^gglyceraldehyde-3-phosphate dehydrogenase

2.6. Data Analyses

Data were analyzed with Statistica Version 12 (StatSoft, Inc., Tulsa, OK). One single-housed female was removed from all analysis of behavior and gene expression due to abnormal social behavior (> 2 SD from the mean) on multiple dependent measures. Several animals were removed from the study due to poor health (single-housed male=2, pair-housed male=2, triple-housed male=1, triple-housed female=1). Body weight upon arrival was assessed with a 3 (Housing Condition) × 2 (Sex) ANOVA to ensure that there were no differences between groups at the beginning of the experiment. Body weight changes across the experiment (Week 0-6) were assessed using a 3 (Housing Condition) × 2 (Sex) × 7 (Week) mixed-factor ANOVA, with week treated as a repeated measure. Social behavior and gene expression data were assessed with (Housing Condition) × 2 (Sex) ANOVAs. Fisher’s Least Significant Difference (LSD) post-hoc tests were used in the case of significant main effects or interactions. A significance level of p < 0.05 was used for all statistical analyses.

3.0. Results

3.1. Body Weight

As expected, at the time of arrival (Week 0), males weighed more than females [F(1,47) = 797.18, p < 0.000] (see Table 2). There was no effect of housing condition on arrival weight, nor were there any interactions between sex and housing condition, indicating that there were no existing differences prior to assignment to housing conditions. There was a significant interaction between Week and Housing Condition in weights across the experiment [F(12, 282) = 3.70, p < 0.01]. Post-hoc analyses revealed that in single-housed animals, weights decreased slightly over the six-week housing period, whereas pair-housed animals gained a small amount of weight between Week 4 and 5 and lost a small amount between Weeks 5 and 6. In triple-housed animals, there were few weight differences from week to week, although there was a slight decrease in weight from Weeks 4 to 5. Overall, these changes in weight were very small, with no more than a 10% change in weight from arrival to the end of the experiment. There was also a significant interaction between Week and Sex [F(6,282) = 17.56, p < 0.0001]. Males lost more weight than females over the course of the experiment, although this change in weight was minimal. In males, the average weight (± SEM) in Week 1 was 437.80 ± 6.18 whereas the average weight in Week 6 was 425.00 ± 5.77 g. On the other hand, females lost weight at the beginning of the experiment, but weight was maintained after the first week. For example, in females, the average weight (± SEM) in Week 1 was 231.14 ± 3.48 g and in Week 6 it was 231.07 ± 3.15 g.

Table 2.

Mean (±SEM) body weight (g) for males and females in each housing condition over 6 weeks.

Sex	Housing Condition	Week 0	Week 1	Week 2	Week 3	Week 4	Week 5	Week 6
	Single	433.00 ± 4.86 a	429.25 ± 4.13	425.88 ± 4.43 c	429.63 ± 4.61	423.25 ± 5.15 bc	419.75 ± 4.75 bc	417.13 ± 4.06 bc
Male	Pair	453.25 ± 14.68 a	453.5 ± 14.81 c	451.25 ± 13.81	450.00 ± 14.69	445.88 ± 13.58 b	448.00 ± 13.33 b	442.00 ± 12.91 b
	Triple	427.78 ± 10.57 a	431.44 ± 9.68 c	428.89 ± 10.30	431.00 ± 9.82 c	426.89 ± 9.55 b	422.22 ± 9.21 b	416.89 ± 9.36 b
	Single	233.00 ± 4.38	232.11 ± 4.15 b	225.44 ± 3.44 c	230.67 ± 3.90 b	227.44 ± 4.12 c	228.22 ± 4.60 bc	229.22 ± 4.76 bc
Female	Pair	225.50 ± 9.31	231.70 ± 8.42 bc	228.70 ± 7.07	230.60 ± 7.19 b	229.40 ± 7.10	232.60 ± 6.69 b	231.10 ± 6.56 b
	Triple	227.11 ± 6.94	229.56 ± 4.53 bc	231.67 ± 4.27	233.00 ± 5.29 bc	232.67 ± 4.57	231.67 ± 5.41 b	232.89 ± 5.17 b

Weights were assessed weekly following arrival (Week 0).

^aindicates a main effect of sex in Week 0;

^bindicates a difference in weight from Week 0 in a given sex;

^cindicates a difference in weight from Week 0 in a given housing condition.

3.2. Social Behavior

Significant sex differences were evident for social investigation, with this form of behavior being higher in females than in males [F(1,47) = 18.17, p = 0.0001]. Social investigation behavior also varied significantly as a function of housing condition and sex [Housing Condition x Sex interaction F(2,47) = 3.82, p = 0.03]. Post-hoc tests revealed that single-housed males and females exhibited elevated social investigation relative to their pair-housed counterparts, with no significant differences evident between pair-housed and triple-housed animals within each sex. Significant sex differences were evident in triple-housed, but not single- or pair-housed age rats: aged females exhibited higher levels of social investigation relative to their male counterparts. Social contact behavior did not differ as a function of housing conditions and was significantly higher in aged females relative to their male counterparts [F(1,47) = 17.15, p = 0.0001] (Figure 1C). Significant sex differences were also evident for social preference/avoidance: aged males exhibited negative values of the coefficient indicating avoidance of the social partner, regardless of housing condition, whereas aged females exhibited social approach indexed via positive values of the coefficient, regardless of housing condition [F(1,45) = 22.16, p = 0.00002] (Figure 1D). The total number of crossovers, a measure of general locomotor activity during the test, also differed as a function of sex and was higher in females (mean ± SEM: 9.11 ± 0.70) than males (5.24 ± 0.78) [F(1,47) = 15.83, p = 0.0002]. In addition, single-housed rats (9.65 ± 0.92) exhibited more crossovers, relative to pair-(6.06 ± 0.96) and triple-housed (6.28 ± 0.95) aged rats [F(2,47) = 5.77, p = 0.006].

We also examined the behavior of the social partners since they were used as housing partners during the experiment. Although we counterbalanced the condition to which partners were assigned, it was possible that long-term housing conditions influenced the expression of social behavior in these adult animals. Sniffing was unaffected by the housing condition of the partner (Pair or Triple) or the condition of the aged subject with which they interacted (Single, Pair, or Triple) (Figure 1E). However, sex differences in contact behavior were apparent when partners were interacting with pair- and triple-housed subjects, but not when interacting with single-housed subjects [Condition of Aged Subject x Sex interaction: F(2,48) = 6.09, p = 0.004] (Figure 1F). No group differences were observed in social approach/avoidance of the partners (Figure 1G), with all animals demonstrating extremely low values of the coefficient, indicative of social indifference.

3.3. Gene Expression

RT-PCR was used to assess expression of several neuroinflammation-related genes (IL-1β, IL-6, TNFα, and CD-14) in addition to oxytocin receptor (OXTR) in the PVN, BNST, and MeA (Figs 2 and 3). In all ROIs, expression of GAPDH, used as a reference gene in all analyses, was not affected by housing condition or sex (all F’s < 1.50). In the PVN, expression of inflammation-related genes was not affected by housing condition, although there was a trend for increased IL-1β [F(1,44) = 3.59, p = 0.06] and IL-6 expression in aged males relative to aged females [F(1,43) = 4.06, p = 0.05] (Fig. 2). In the PVN, OXTR gene expression was unaffected by housing condition or sex (p’s > 0.25). Within the BNST, cytokine gene expression did not differ as a function of sex or housing condition (Figs. 2 and 3). For the most part, expression of neuroinflammation-related genes in the PVN and BNST was unaffected by sex or long-term housing manipulations. However, OXTR gene expression in the BNST was elevated in single-housed aged rats, relative to pair-housed aged rats, regardless of sex [F(2,44) = 3.91, p = 0.03] (Fig. 3E). By far, the MeA appeared to be the most responsive to long-term housing manipulations. Triple-housed aged females exhibited elevated IL-1β and IL-6 gene expression [Condition x Sex interaction, F(2,42) = 5.56, p = 0.007; F(2,40) = 3.46, p = 0.041, respectively] (Fig. 2C, F). CD14 expression was increased in triple-housed aged rats relative to single-housed animals, regardless of sex [F(2,41) = 3.87, p = 0.029] (Fig. 3C). In contrast to the BNST, OXTR gene expression in the MeA was unaffected by housing condition [F(2,41) = 1.96, p = 0.15].

Figure 2.

mRNA expression of IL-1β in PVN (A), BNST (B), and MEA (C). IL-6 mRNA expression in PVN (D), BNST (E), and MEA (F). TNFα mRNA expression in PVN (G), BNST (H), and MEA (I). (*) p < 0.05 vs. males within housing condition.

Figure 3.

mRNA expression of CD-14 in PVN (A), BNST (B), and MEA (C). OXTR mRNA expression in PVN (D), BNST (E), and MEA (F). ($) p < 0.05 vs. single-housed rats.

4.0. Discussion

The results of the present study showed that long-term manipulations of social housing conditions impacted subsequent social behavior and basal cytokine gene expression in a sex-dependent manner. Long-term social deprivation enhanced social investigation in aged animals regardless of sex, with this effect being more pronounced in aged males than in their female counterparts. Social contact behavior and social approach/avoidance were not affected by housing conditions. Sex differences were evident in contact behavior and social approach/avoidance: aged females demonstrated more contact behavior than their male counterparts, and social approach was evident in aged females, whereas aged males demonstrated avoidance of the social partner. Aged males housed with two adult males demonstrated substantially less social investigation than aged females housed under the same circumstances. Housing- and sex-related differences in social behavior of aged rats cannot be attributed to social behavior of their test partners, given that male and female pair- and triple-housed partners demonstrated similar levels of social investigation and values of the coefficient. Changes in cytokine gene expression as a result of long-term housing conditions were observed exclusively in the MeA of aged females. In contrast, social isolation enhanced OXTR expression in the BNST in both males and females.

It has been established that effects of social deprivation (i.e., isolation housing) on subsequent social interaction differ as a function of age. Long-term social isolation of juveniles and adolescents generally decreases adult social behavior [49,50], and similar social isolation of adults increases interaction with conspecifics [51,52]. Effects of social isolation in aged rats have been poorly investigated. To the best of our knowledge, only one study assessed the effects of long-term social isolation in aged (24-25 months of age) male F344/N rats [9]. In that study, social isolation for 4 weeks had no effect on social behavior of aged male rats. These findings are in contrast with our results, which clearly demonstrate the facilitating effects of social isolation on social investigation in aged male F344 rats. These discrepancies may be associated with a number of procedural differences between the two studies. In our study, rats were tested at 19-20 months of age with adult conspecifics (about 3.5-month-old), whereas in the Shoji & Mizoguchi study, social behavior was assessed in 25-month-old males exposed to an age-matched partner. It seems possible that the 19-20-month-old males show more adultlike responding to long-term social isolation [51,52] and, therefore, differ from 25-month-olds. Given that adult rats are more socially active than their aged counterparts [3], it is possible that socially deprived aged males found their younger partners more appealing than the same-age partners. Increases in social interaction following isolate housing are often viewed as increases in social motivation [9,52]. However, when social motivation was measured in the present study using the coefficient of social approach/avoidance [44], aged males demonstrated avoidance of a social partner regardless of housing conditions. It has been previously shown that, under familiar test circumstances used in the present study, group-housed adolescent and adult animals demonstrate social approach indexed via positive values of the coefficient, whereas social avoidance (i.e., negative values of the coefficient) is evident in these animals when tested under unfamiliar, anxiety-provoking test circumstances, reflecting social anxiety-like behavioral alterations [53,54]. However, even short-term social deprivation of adolescents and adults transforms social avoidance into social approach under unfamiliar test circumstances [54]. Taken together, these findings suggest that single-housed aged males, although demonstrating substantial increases in social investigation relative to aged males housed with an adult cagemate, showed rather high levels of anxiety-like behavior under social test circumstances, as evidenced by the emergence of social avoidance. The dissociation between increased social investigation and decreased social motivation in isolate-housed aged males is intriguing and needs more thorough investigation.

Aged males housed with two adult males for 5-6 weeks demonstrated reduced social interaction, relative to females housed under the same conditions. Although the goal of the present experimental manipulation was to provide social enrichment, it is possible that housing of aged males with adult aniimals was somewhat anxiogenic for these males. Indeed, previous studies have demonstrated sex-specificity in the effects of crowding on food intake. Specifically, adult males that were housed in groups of 10 exhibited reduced food and water intake, relative to females under the same conditions [55]. However, if group housing was highly aversive, we would have expected social avoidance to be exhibited specifically in pair- and triple-housed aged males. On the contrary, social avoidance was observed in aged males, regardless of housing conditions. In contrast to their male counterparts, aged females demonstrated social approach, suggesting that aging in males, but not in females, is associated with enhanced anxiety. This finding is in agreement with previous studies that demonstrated enhanced anxiety-like behavior in aged rats using non-social tests of anxiety [45,56], with these anxietylike aging-associated alterations being more pronounced in males than in females [57,58]. Furthermore, in general, aged females demonstrated higher levels of social interaction than their male counterparts. Sex differences in the expression of social behavior are common for adult rats, although the directionality of these differences is strain-dependent. For instance, Lister [45], Sprague-Dawley [59,60], and Wistar [61] male rats exhibit higher levels of social interaction than female rats. In contrast, isolate-housed adult [42] and pair-housed aged [43] Fisher 344 females are more socially active than their male counterparts. These previous findings and the results of the present study suggest that sex-related differences in Fisher 344 rats are evident regardless of age, with females being more socially active than their male counterparts.

Previous work from our lab has shown that pair housed Sprague Dawley rats display marked differences in basal expression of IL-1β in the hypothalamus, an effect that was strongly related to established dominance heirarchies (i.e., subordinate rats displayed heightened IL-1β; [62]. Since increased immune activation is evident in the brains of aged rodents [63] and social housing conditions can influence neuroinflammation [38], we assessed expression of multiple genes related to neuroinflammation (IL-1β, IL-6, TNFα, and CD-14) in brain regions with well-described roles in social behavior (PVN, BNST, and MEA). Contrary to the behavioral effects of long-term manipulations of housing conditions, few changes were observed in neuroimmune gene expression in aged males. CD14 gene expression was elevated in the MeA of triple-housed aged males and females, relative to single-housed males and females. Previous studies have demonstrated elevated expression of CD14 in the hypothalamus after brief footshock exposure [64], and on the surface of microglia following social defeat stress [65-67], an effect that was blocked by administration of propranolol, a β-adrenergic receptor antagonist [65]. Notably, stress-induced suppression of social investigation was reversed by icv administration of IL-1 receptor antagonist [68]. It is possible that group-housing aged rats with adult rats may have represented a mild social stressor, particularly in males, although it should be noted that CD14 gene expression was not elevated in PVN, a brain region that exhibited elevated IL-1β and CD14 gene expression following footshock stress [69] and increased microglia activation following repeated social defeat stress [65]. Acute social defeat had no effect on cytokine gene expression in PVN or BNST [70]. The lack of IL-1β gene expression in PVN and BNST suggests that any stress imposed by manipulations of housing conditions in aged male and female rats was likely mild, given that the PVN is especially sensitive to stress-induced expression of cytokines [47,71,72]. Thus, the present observation that triple-housed females displayed a substantial increase in expression of IL-1β and IL-6 in the MeA, with no changes evident in PVN or BNST, is particularly intriguing as this regionally-specific effect argues against group-housing causing a generalized increase in brain inflammation.

It should be noted that triple-housed females exhibited the greatest levels of social investigation and social approach and the highest expression of pro-inflammatory cytokines within the MeA. On its surface, this effect seems paradoxical, since heightened cytokine expression produced by bacterial mimetics [73] and stress challenges [68] typically reduce social behavior. Although it is premature to imply a causal relationship between high social activity and heighted cytokine expression in the MEA, one potential explanation is that local cytokine expression in the MeA may drive increased social behavior by altering synaptic function within the well-established MeA node of the social behavior circuitry. Alternatively, it is possible that chronic triple-housing in females produces heightened activity of the MeA across the 5-6 weeks of housing, thereby driving increased cytokine expression. This latter possibility is plausible since glial activity and cytokine expression is known to increase as a function of neuronal excitation, as glial activity is involved in maintenance functions within highly active circuitry. Regardless of the directionality, it is clear from these data that the MeA was most responsive to manipulations of long-term housing conditions, which is not surprising given its role in social behavior.

OXTR gene expression was elevated in the BNST of aged single-housed animals regardless of sex. This observation was rather surprising, given that OXT modulates social behavior in a sex-dependent manner [74]. For instance, in adult Wistar rats, females show less social approach and substantially lower OXTR binding densities than males in a number of brain regions, including the BNST [61]. However, the previously observed sex differences in the BNST OXT system and the lack of these differences in the present study are not surprising, given that sex differences in OXTR system are both brain region- and strain-specific [74]. It has been shown than aging in Sprague Dawley males is associated with a significant decrease in expression of OXTR in the BNST [75]. To the extent that OXTR gene expression in the BNST is involved in the modulation of social behavior and given that aging in rats is associated with substantial decreases in different forms of social behavior including social investigation, it is likely that the observed increases in social investigation in single-housed male and female F344 rats may be associated with increases in OXTR gene expression in the BNST.

Overall, the present study found several sex-specific effects of housing conditions among aged rats. Aged males displayed reduced investigation of a novel partner after extended living in a group (pair or triple-housed), whereas females displayed a high level of social interaction with a novel conspecific regardless of their recent housing history. Epidemiological data support the idea that strong social support has positive effects on mortality and that these effects were not sex specific [17]. In addition, interventions designed to promote well-being and increase social support have shown positive effects in the elderly, although in many cases, sex-specific effects were not examined [18,20,21]. The data from the current experiment suggest that there may indeed be sex-specific effects of social interventions on a variety of outcomes, including social behavior and neuroinflammatory markers. These findings provide important insight into not just how housing conditions in rodents impact subsequent social interaction but may have ultimately inform best practices for understanding and/or optimizing independent living conditions among aged humans.

Highlights.

• Group housing with young adults suppressed social behavior in aged males, but not females.

• Aged females were more socially active than their male counterparts.

• IL-1β and IL-6 gene expression was elevated in the medial amygdala of triple-housed females.

• OXTR gene expression was elevated in single-housed rats in BNST, relative to pair-housed rats.