The same stress has divergent effects on social versus asocial manifestations of anxiety-like behavior over time

Abstract

Stress may lead to augmented anxiety, which may, with time culminate in some form of anxiety disorder. Behavioral alterations related to increased anxiety can be broadly classified into two types–social, affecting interactions between individuals, and self-oriented, affecting the anxious individual only. While a growing body of literature now exists describing the effects of stress-induced anxiety on self-oriented behavior in animal models of anxiety disorders, the effects of such aberrant anxiety on social behavior has largely remained uncharacterized in these models. This study aims to fill this gap in our understanding by examining changes in social behavior following a single 2-hour episode of immobilization stress, which has been shown to cause delayed structural and functional changes in the amygdala. To this end, we examined social behavior, measured as active social interactions, anogenital sniffing, nose-to-nose contacts, allogrooming, actively following and crawling under, as well as self-oriented asocial behavior, manifested as self-grooming and rearing, in adult male rats. Stressed animals showed reduced social interaction 1 day after immobilization stress and this decrease was persistent for at least 10 days after stress. In contrast, individualistic behaviors were impaired only 10 days, but not 1 day later. Together, these results not only show that the same single episode of stress can elicit divergent effects on social and asocial measures of anxiety in the same animal, but also suggest that enhanced social anxiety soon after stress may also serve as an early indicator of its delayed behavioral effects.

Keywords:

• Stress
• anxiety
• social interaction
• grooming
• rearing
• PTSD
• delayed effect

Introduction

Stressful experiences often lead to distinct effects on social versus subjective measures of anxiety (Collimore et al., 2010). Patients suffering from Post-Traumatic Stress Disorder (PTSD), for instance, show social avoidance manifested by enhanced gaze aversion, as well as increased subjectively-reported anxiety. Both these behavioral indices independently indicate an elevated anxious state. Interestingly, such observations have also been replicated in preclinical models of PTSD. Rats undergoing chronic exposure to stress show significantly decreased social interaction (Doremus-Fitzwater et al., 2009) and enhanced avoidance of the open arm on an elevated plus-maze (Mitra et al., 2005), each indicating stress-enhanced anxiety-like behavior. While the majority of such reports use repeated episodes of stress, a single stressful experience has also revealed similar findings. For example, rats exposed to an inescapable stressor show reduced social exploration of a juvenile up to 36 hours later (Christianson et al., 2008). On the other hand, asocial anxiety on the elevated plus-maze remains elevated even 10 days after a single, 2 hour-long episode of immobilization (Chakraborty & Chattarji, 2019; Mitra et al., 2005). However, studies in animal models of stress have often quantified these measures using separate tests and/or studies. Thus, to get a comprehensive picture of the behavioral deficits, it is important to understand how both social and asocial anxiety change over time in the same animal following stress exposure. This was the main goal of the present study. To this end, we first hypothesized that a single stress episode should affect the expression of both social interaction as well as asocial anxiety-like behaviors. Second, earlier studies reported that such a single stressor causes an impairment in social interaction 24 hours later (Christianson et al., 2008, 2009). However, increase in anxiety on the elevated plus-maze, an exploration-based individualistic behavior that does not involve social interaction, has typically been observed only a few days after stress (Chakraborty et al., 2020; Chakraborty & Chattarji, 2019; Mitra et al., 2005). Hence, we hypothesized that while a single episode of stress may affect social interaction at an earlier time point, asocial/individualistic measures of anxiety like rearing and grooming should only elicit a delayed enhancement. Together, these behavioral analyses will enable us to test whether certain signatures of stress-induced behavioral changes may become evident soon after a brief severe physical stressor such that it could serve as a marker for the eventual onset of long-term behavioral deficits.

In the present study, therefore, rats underwent a single 2-hour session of immobilization stress which has been previously shown to capture key behavioral, morphological and physiological correlates of PTSD. These include increased excitatory activity and dendritic spine density in the basolateral amygdala, as well as increase in anxiety-like behavior on the plus-maze 10 days after stress (Chakraborty et al., 2020; Chakraborty & Chattarji, 2019; Mitra et al., 2005; Yasmin et al., 2016). Social behaviors was quantified using an ethologically relevant social interaction paradigm (File & Seth, 2003) in freely interacting pairs of control or stressed animals, tested either 1 day or 10 days after 2-hour immobilization. In addition, we also simultaneously quantified exploratory rearing and self-grooming in these interacting pairs, which are asocial measures of rodent anxiety-like behavior (Kalueff et al., 2016; Sturman et al., 2018).

Methods

For the purpose of this study, foundation colonies of Sprague–Dawley rats were acquired from Charles River and subsequently bred and maintained in the animal facility at the National Center for Biological Sciences (NCBS), Bangalore (55–60 days old, ∼300 gm at the beginning of the experiment). Pair-housed animals were kept in temperature and lighting controlled conditions (14/10 h light/dark cycle) with food and water ad libitum. All procedures were approved by the Institutional Animal Ethics Committee, NCBS.

During acute immobilization stress (2 hours, between 10 am and 12 noon) (Chakraborty & Chattarji, 2019; Mitra et al., 2005), each animal was placed in a plastic immobilization cone with an apical perforation for breathing, without any access to food or water. Following a 2-hour immobilization period, they were returned to their home cages, and remained undisturbed until behavioral testing either 1 or 10 day later. Control animals (two groups tested, either 1 day or 10 days after stress) were unstressed and housed separately from stressed animals. Subsequently, pairs of both stressed and control animals underwent social interaction test either 1 day (Stress-1d) or 10 days after stress (Stress-10d), in different experimental cohorts. Control animals were paired with unfamiliar, non-littermate controls, and stressed animals were paired with unfamiliar, non-littermate stressed rats during the social interaction test (body weight difference between interactors ≤ 20 g). Specifically, for the purpose of this experiment 24 (12 pairs, 6 pairs for each time-point), 24 (12 pairs) and 26 (13 pairs) animals were tested in pairs, as a part of the control, stress-1d and stress-10d groups respectively.

During the social interaction test, both animals were introduced simultaneously into a uniformly lit, novel, square shaped arena (1 m × 1 m × 0.4 m), and were allowed to interact freely for 10 minutes (File & Seth, 2003). Each session was recorded with a digital video recorder (Samsung 470 D) and was analyzed offline by a researcher who was blind to the experimental groups. Active social interactions including anogenital sniffing, nose-to-nose contacts, allogrooming, actively following and crawling under were analyzed. Asocial behaviors quantified included exploratory rearing (standing up on hind limbs and exploration of surrounding) and self-grooming (natural tendency of rodents to repeatedly clean their fur using forelimbs). Rearing was further classified into supported and unsupported rearing (Crusio, 2001; Sturman et al., 2018). For all measures (except nose-to-nose contacts, given its transient nature), total duration and total number of behavioral bouts were quantified. Both interacting rats were analyzed for all the behaviors, across all experimental groups.

Graphical values are expressed as mean ± standard error of mean. Any data point beyond mean ± 2 standard deviations were considered as statistical outliers, and were excluded from analysis. Data were checked for normality using D’Agostino and Pearson’s normality test. Normally distributed datasets were compared using one-way ANOVA, followed by post-hoc Bonferroni’s test. Non-normally distributed datasets were compared using Kruskal-Wallis ANOVA, followed by Dunn’s test for post-hoc comparisons. All sample sizes indicate pairs of animals. Statistical analyses and plots were computed using GraphPad Prism software (GraphPad software Inc., La Jolla, California, USA, version 6). Statistical details for all tests have been summarized in Table 1, grey cells denote significant difference detected with ANOVA and post-hoc tests were perform only when the ANOVA results were significant. 

Table 1. Details of statistical analyses for all behavioral parameters analyzed.

Figure reference	Behavior type	Controls	Stress 1d	Stress 10d	One way ANOVA (F Stats )/ Kruskal-Wallis ANOVA (H Stats)	Post Hoc (Bonferroni's or Dunn’s with Adjusted p values)
		Mean ± SE	Mean ± SE	Mean ± SE		Control vs Stress 1d	Control vs Stress 10d
Figure 1(b)	Nose-to-nose contact	23.83 ± 1.89	26.92 ± 1.4	12 ± 0.76	F (2, 34) = 32.67, p < .0001	0.26	<0.0001
Figure 1(c)	Active Interaction (Time)	268.4 ± 11.52	174.9 ± 6.5	188.4 ± 13	F (2, 34) = 21.21, p < .0001	<0.0001	<0.0001
Figure 1(d)	Active Interaction (Number)	51.58 ± 2.73	36.83 ± 1	39.77 ± 1.6	F (2, 34) = 16.30, p < .0001	<0.0001	0.0002
Figure 1(e)	Active Following (Time)	71.71 ± 6.21	60.02 ± 7.21	53.67 ± 8.3	F (2, 34) = 1.54, p = .2277	N/A	N/A
Figure 1(f)	Active Following (Number)	18 ± 1.26	15.25 ± 1.9	14.86 ± 2.2	F (2, 34) = 0.83, p = .4409	N/A	N/A
Figure 1(g)	Anogenital Sniffing (Time)	78.75 ± 9.89	51.57 ± 7.4	40.54 ± 6	F (2, 34) = 6.32, p = .0046	0.0416	0.0028
Figure 1(h)	Anogenital Sniffing (Number)	28.58 ± 2.63	19.17 ± 1.2	15.46 ± 1.3	F (2, 34) = 13.87, p < .0001	0.0019	<0.0001
Figure 1(i)	Crawling Under (Time)	2.27 ± 0.81	2.2 ± 0.9	1.93 ± 1.2	F (2, 35) = 0.03, p = .9696	N/A	N/A
Figure 1(j)	Crawling Under (Number)	0.92 ± 0.34	0.83 ± 0.34	0.77 ± 0.4	F (2, 34) = 0.06, P = 0.9333	N/A	N/A
Figure 1(k)	Allogrooming (Time)	4.29 ± 1.82	0.56 ± 0.6	2.12 ± 0.9	F (2, 34) = 2.42, p = .1038	N/A	N/A
Figure 1(l)	Allogrooming (Number)	1.17 ± 0.52	0.08 ± 0.08	0.54 ± 0.24	F (2, 34) = 2.63, p = .0862	N/A	N/A
Figure 2(b)	Unsupported rearing (Time)	31.83 ± 4.92	23.7 ± 4.2	15.93 ± 1.8	F (2, 34) = 4.44, p = .0193	0.2882	0.0105
Figure 2(c)	Unsupported rearing (Number)	25.08 ± 2.3	20.76 ± 2.6	15.92 ± 1.6	F (2, 34) = 4.54, p = .0179	0.3436	0.01
Figure 2(d)	Supported rearing (Time)	50.01 ± 7.44	45.14 ± 4.3	45.21 ± 5.8	F (2, 34) = 0.2153, p = .8074	N/A	N/A
Figure 2(e)	Supported rearing (Number)	37.67 ± 3.06	33.83 ± 2.5	37.69 ± 2.9	H (2) = 0.88, p = .6412	N/A	N/A
Figure 2(f)	Self-grooming (Time)	36.33 ± 4.3	50.76 ± 6.7	80.69 ± 14.8	H (2)=12.21, p = .0022	0.2384	0.0014
Figure 2(g)	Self-grooming (Number)	14.08 ± 0.69	17.58 ± 1.6	17.54 ± 0.86	F (2, 34) = 3.080, p = .0590	N/A	N/A

Results

To assess temporal changes in anxiety-like behavior following 2-hours of immobilization, pairs of stressed animals were subjected to social interaction tests at either 1 day later or after a delay of 10 days (Figure 1(a), 2(a)). In our experiment, pairs of control animals were tested at the same time-point as pairs of stressed animals. Thus, two groups of controls animals were tested, either 1 day or 10 days after stress; and 6 pairs of control animals were used for each time-point. Upon comparing these two control groups, we found no difference between them for any of the social or asocial behaviors (Supplementary Figure 1). Therefore, all control animals were clubbed together into a single group for subsequent statistical analysis.

Figure 1. Stress causes a decrease in social interaction 1 day later that persists up to 10 days later. (a) Animals were subjected to 2-hours of immobilization stress, and social interaction was measured either 1 day or 10 days later, in separate groups of animals. (b) Stress reduces the number of nose-to-nose contacts (inset) 10 days later. (c) Stress leads to a significant reduction in time spent in active interactions (inset). (d) Stress also reduces number of interactions, both 1 day (Stress-1d) and 10 days (Stress-10d) later. (e) Time spent in active following (inset), or (f) number of active follows does not change either 1 day or 10 days after stress. (g) Total time spent in anogenital sniffing (inset) decreases significantly both 1 and 10 days after stress. (h) Total number of anogenital sniffs also decreases significantly both 1 day and 10 days later. (i) Time spent in and (j) number of crawling under (inset) episodes does not change at 1 or 10 days after stress. (k) Time spent is allogrooming (inset), and (l) number of allogrooming bouts also does not change 1 or 10 days after stress. (Control, N = 12 pairs, Stress-1d, N = 12 pairs, Stress-10d, N = 13 pairs). * indicates p < .05, ** indicates p < .01, *** indicates p < .001, **** indicates p < .0001 in post-hoc Tukey’s test.

Figure 2. Stress leads to only delayed changes in asocial anxiety-like behavior. (a) Animals were subjected to 2-hours of immobilization stress, and social interaction was measured either 1 day or 10 days later, in separate groups of animals. (b) Total time spent in unsupported rearing (inset) decreases significantly 10 days but not 1 day after stress. (c) Total number of unsupported rears decreases significantly 10 days after stress. (d) Total time spent in supported rearing (inset) and (e) number of supported rears remain unaffected by stress. (f) Total time spent in self-grooming (inset) shows a significant increase only 10 days after stress, but not 1 day later. (g) Number of self-grooming bouts remain unchanged. (Control, N = 12 pairs, Stress-1d, N = 12 pairs, Stress-10d, N = 13 pairs). * indicates p < .05 post-hoc Tukey’s test. ## indicates p < .01 in post-hoc Dunn’s test.

Firstly, upon approaching its partner rat head on, animals displayed nose-to-nose contacts (Figure 1(b)). This transient social behavior only showed a reduction in its number 10 days after stress, but not 1 day later (Control: 23.8 ± 1.9, Stress-1d: 26.9 ± 1.4, Stress-10d: 12.0 ± 0.76) (Figure 1(b)). Upon broadening our analysis to compare all such active social interactions, we observed that indeed stressed animals spent significantly less time interacting between them 1 day later, as compared to unstressed controls (Control: 268.4 ± 11.5 s, Stress-1d: 174.9 ± 6.5 s) (Figure 1(c)). Interestingly, such an impairment lasted even 10 days later (Control: 268.4 ± 11.5 s, Stress-10d: 188.4 ± 13 s) (Figure 1(c)). Stress also led to a significant decrease in number of interactions, both 1 day as well as 10 days after stress (Control: 51.6 ± 2.7, Stress-1d: 36.8 ± 1.0, Stress-10d: 39.77 ± 1.6) (Figure 1(d)). Comparing all three groups by a one-way ANOVA revealed significant effects for both these parameters (Table 1).

To better understand the change in social behavior, we next looked at the impact of stress on other behavioral sub-categories as well. We observed that animals actively followed their partners, although there was no effect of stress on either the time spent (Figure 1(e)) or the number of bouts of active following (Figure 1(f)) (Time: Control: 71.7 ± 6.2 s, Stress-1d: 60.0 ± 7.2 s, Stress-10d: 53.7 ± 8.3 s; Number: Control: 18.0 ± 1.3, Stress-1d: 15.3 ± 1.9, Stress-10d: 14.9 ± 2.2). Interestingly, active following most frequently terminated in anogenital sniffing of the partner rat, which is a predominant component of the rodent social behavior repertoire (Figure 1(g)). Time spent in anogenital sniffing showed a significant decrease both 1 and 10 days after stress (Control: 78.8 ± 9.9 s, Stress-1d: 51.6 ± 7.4 s, Stress-10d: 40.5 ± 6 s) (Figure 1(g)), as did the number of anogenital sniffing bouts (Control: 28.6 ± 2.6, Stress-1d: 19.2 ± 1.2, Stress-10d: 15.5 ± 1.3) (Figure 1(h)). Upon close physical contact, animals sometime crawled under its partner– the number and time spent in which did not differ either 1 or 10 days after stress (Time: Control: 2.3 ± 0.8 s, Stress-1d: 2.2 ± 0.9 s, Stress-10d: 1.9 ± 1.2 s; Number: Control: 0.92 ± 0.3, Stress-1d: 0.8 ± 0.3, Stress-10d: 0.8 ± 0.4) (Figure 1(i-j)). Rats also showed allogrooming, wherein one rat groomed its interacting partner. There was no overall difference in either time spent in or occurrence of this behavior (Time: Control: 4.3 ± 1.8 s, Stress-1d: 0.6 ± 0.6 s, Stress-10d: 2.1 ± 0.9 s; Number: Control: 1.2 ± 0.5, Stress-1d: 0.08 ± 0.08, Stress-10d: 0.54 ± 0.24) (Figure 1(k-l)). But we found a greater proportion rats showed allogrooming 10 days after stress compared to 1 day later (Supplementary Figure 2).

Next, we analyzed the effects of stress on exploratory rearing, an asocial behavior. When rearing occurred, it was either without any support with both forelimbs in the air (unsupported rearing) (Figure 2(b), inset), or with support of the arena walls (supported rearing) (Figure 2(d), inset). Thus, while “supported rearing” occurred at the less anxiogenic periphery, “unsupported rearing” occurred in the comparatively more anxiogenic regions of the arena, requiring them to be analyzed separately. In contrast to the reduction in social interaction seen both 1 day and 10 days later, stressed animals only exhibited a significant decrease in both time spent in (Control: 31.8 ± 4.9 s, Stress-1d: 23.7 ± 4.2 s, Stress-10d: 15.9 ± 1.8 s) (Figure 2(b)) as well as number of unsupported rears 10 days after stress, but not 1 day later (Control: 25.1 ± 2.3, Stress-1d: 20.8 ± 2.6, Stress-10d: 15.9 ± 1.6) (Figure 2(c)). Unlike unsupported rearing, there was no effect of stress on the time spent in (Figures 2(d)) as well as number of supported rears either 1 day or 10 days after stress (Figures 2(e)) (Time, Control: 50 ± 7.4 s, Stress-1d: 45.1 ± 4.3 s, Stress-10d: 45.2 ± 5.8 s; Number, Control: 37.7 ± 3.1, Stress-1d: 33.8 ± 2.5, Stress-10d: 37.7 ± 2.9).

Lastly, time spent in self-grooming (Figure 2(f), inset), another measure of asocial behavior, exhibited a significant increase 10 days after stress that was not evident 1 day later (Control: 36.3 ± 4.3 s, Stress-1d: 50.8 ± 6.7 s, Stress-10d: 80.7 ± 14.8 s) (Figure 2(f)). However, the number of self-grooming bouts was comparable across all three groups (Control: 14.1 ± 0.7, Stress-1d: 17.6 ± 1.6, Stress-10d: 17.5 ± 0.9) (Figure 2(g)).

Taken together, these results suggest that a single episode of immobilization stress has divergent temporal effects on social versus asocial measures of anxiety-like behavior. Comparing these parameters in the same animal (Figure 3(a)) revealed that while all animals (100%) interacted less than controls 1 day after stress, only about half of these animals (58%) showed decreased unsupported rearing at this time point (Figure 3(a)). However, following a delay of 10 days, almost the entire population of stressed animals showed both decreased interaction (92%) as well as decreased unsupported rearing (100%) (Figure 3(b)). This indicates that while stress causes a decrease in social interaction that lasts up to 10 days later, asocial anxiety-like behavior is only affected at a later time-point after stress.

Figure 3. Divergent patterns of social and asocial anxiety within the same animal after stress. Proportions of animals which lie either below (red) or above (white) the control mean level (100%) for (a) 1 day after stress (b) and 10 days after stress. Corresponding pie charts show that while 100% and 58% of animals show reduced social interaction and unsupported rearing respectively 1 day after stress, these proportions become 92% and 100%, respectively, 10 days later. All values indicate difference with respect to control mean (dotted line). (Control, N = 12 pairs, Stress-1d, N = 12 pairs, Stress-10d, N = 13 pairs).

Discussion

Here, we show that a 2-hour immobilization impairs social interaction 1 day after, an effect that persists even 10 days later. This extends previous reports of social deficits soon after stress (Christianson et al., 2008, 2009). While individualistic measures of anxiety, such unsupported rearing (Christianson et al., 2009) and self-grooming (Rojas-Carvajal & Brenes, 2020), are known to be affected immediately after a single episode of stress in an open-field test, we show that these deficits can occur alongside impaired social behaviors. Importantly, both asocial correlates of anxiety were only evident 10 days after stress, consistent with the delayed, temporal build-up in anxiety-like behavior seen on the elevated plus-maze after acute immobilization (Mitra et al., 2005). This unique temporal profile of the expression of social versus asocial deficits points to an important difference in their etiology, wherein the deficits in social behaviors seen soon after stress could serve as a potential marker for behavioral changes that emerge later on.

Upon micro-analysis of different social behavior sub-categories, we observed that the reduction in social interaction was caused by a significant decrease in both anogenital sniffing, a behavior that conveys the stressed state of an animal (Sterley et al., 2018), as well as nose-to-nose contacts. Episodes of active following being not different between groups, it indicates that most social follow-approaches ended in successful investigation in controls, but not in stressed rats. Furthermore, a decrease in nose-to-nose interaction was only seen at the delayed time-point, suggesting that the spectrum of social deficits broaden with time. 10 days after stress rats also show a reduction in exploratory rearing. Single stressor-induced reduction in unsupported rearing does indicate a state of heightened anxiety (Otabi et al., 2020; Sturman et al., 2018) when there is also an increase in self-grooming behavior (Kalueff et al., 2016), in agreement with our present observations. Interestingly, a higher proportion of stressed animals exhibited allogrooming 10 days after stress than 1 day later, possibly hinting at a divergent evolution of empathetic behaviors over time as well (Lu et al., 2018).

Interestingly, the same model of stress used here is also known to trigger physiological and structural plasticity in the amygdala, a brain area that regulates both social interaction and anxiety (Felix-Ortiz et al., 2016). Previous studies reported increases in the frequency of excitatory synaptic events and de novo protein synthesis in the basolateral amygdala (BLA) soon after 2-hour immobilization stress, which persists even after 10 days (Madan et al., 2018; Yasmin et al., 2019). However, enhanced dendritic spine-density in BLA principal neurons was only seen 10 days later (Mitra et al., 2005), suggesting that changes early on may pave the way for longer lasting modifications in plasticity later on. This might, in turn, also affect how the BLA interacts with other brain areas, thereby giving rise to a range of social and asocial behavioral deficits that vary over time. The present study only examined the effects of a single stressor in male rats. Therefore, in future studies it would be important to explore gender differences in the proximal versus delayed behavioral effects of single stressor, because different neural mechanisms are known to mediate social interaction in male versus female rodents (Stack et al., 2010).

In conclusion, while previous reports suggest a specific, delayed increase in anxiety on the plus-maze after a single episode of stress (Kim et al., 2014; Mitra et al., 2005), expanding our analysis revealed that stress effects on social correlates of anxiety are temporally different, setting in earlier in time. Such early deficits might serve as an important diagnostic marker for the delayed detrimental consequences of stress, which may help in predicting and, in turn, altering the etiology of PTSD and other stress disorders.

Author contributions

KS and SC contributed to the experimental design. KS and PC performed the experiments and analyzed the data. KS, PC and SC interpreted the results. PC and SC wrote the manuscript.

Acknowledgements

This study was funded in full or in parts by the Department of Biotechnology, Govt. of India. KS was supported by the DBT-RA scheme.

Disclosure statement

No potential conflict of interest was reported by the author(s).