Impacts of social isolation stress in safety learning and the structure of defensive behavior during a spatial-based learning task involving thermal threat

Abstract

Safety learning during threat and adversity is critical for behavioral adaptation, resiliency, and survival. Using a novel mouse paradigm involving thermal threat, we recently demonstrated that safety learning is highly susceptible to social isolation stress. Yet, our previous study primarily considered male mice and did not thoroughly scrutinize the relative impacts of stress on potentially distinct defensive mechanisms implemented by males and females during the thermal safety task. The present study assessed these issues while considering a variety of defensive behaviors related to safety-seeking, escape, coping, protection, ambivalence, and risk-taking. After a two-week social isolation stress period, mice were required to explore a box arena that had thermal threat and safety zones (5 vs. 30°C, respectively). Since visuospatial cues clearly differentiated the threat and safety zones, the majority of the no-stress controls (69-75%) in both sexes exhibited optimal memory formation for the safety zone. In contrast, the majority of the stress-exposed mice in both sexes (69-75%) exhibited robust impairment in memory formation for the safety zone. Furthermore, while the control groups exhibited many robust correlations among various defensive behaviors, the stress-exposed mice in both sexes exhibited disorganized behaviors. Thus, stress severely impaired the proper establishment of safety memory and the structure of defensive behavior, effects that primarily occurred in a sex-independent manner.

Keywords: anxiety; behavioral flexibility; conflict; decision making; fear; prefrontal cortex (PFC).

FIGURE 1

Social isolation stress led to robust impairment in safety memory, independent of mouse sex. (A) Depiction of a social isolation stress procedure. While controls remained in group-housing conditions, the experimental groups underwent single-housing for 14 days. After the isolation period, mice were returned to group-housing conditions with their former cagemates for a week, and then underwent training and testing in a safety learning task that involved a thermal threat. (B) Infrared image of the arena used for the thermal task, in which most quadrants had a noxious cold temperature (∼5°C, “threat zones”), while one quadrant had a pleasant warm temperature (∼30°C, “safety zone”). Visuospatial cues on the walls helped mice to differentiate the safety zone from the threat zones (e.g., plus symbols vs. vertical bars, counterbalanced across mice). (C) Examples of safety-seeking behavior. Software was used to track animals and quantify the time spent in the safety zone. (D) Experimental design to test for learning and memory. The training session lasted 10 min and included a warm safety zone, while the subsequent recall test lasted 5 min and no longer included the thermal reinforcer for the safety zone. (E) Average heatmaps for the male groups during the thermal task. (F) Quantifications of safety-seeking behavior for the male groups. Despite robust initial learning, many of the males that received the social isolation stress treatment exhibited significant memory impairment for the safety zone (**P = 0.0025). (G) Proportion of males that exhibited stress susceptibility or resiliency, based on memory recall. Consistent with the male distributions shown in the histogram insets, the cutoff was set to 50%. (H) Average heatmaps for the female groups during the thermal task. (I) Quantifications of safety-seeking behavior for the female groups. Similar to males, many of the females that received the social isolation stress treatment exhibited significant memory impairment for the safety zone (**P = 0.0086). (J) Proportion of females that exhibited stress susceptibility or resiliency, based on memory recall. Consistent with the female distributions shown in the histogram insets, the cutoff was set to 50% (N = 16 per group; CTL, no-stress control; SIS, social isolation stress; M, males; F, females).

FIGURE 2

Active defensive mechanisms related to escape were strongly exhibited during the thermal safety task. (A) Rearing behavior, quantified when mice adopted an upright standing posture with forelimbs touching the sidewalls of the box. (B) Rearing behavior did not differ between the control and stress groups. However, males exhibited significantly more rearing than females during the recall test (*P < 0.05). (C) Darting behavior, defined as events in which mice made sudden flight-like running movements in any direction. Representative events are shown as overlaid video frames. (D) The rate of darting behavior did not differ between the control and stress groups, or the male and female cohorts. (E) Jumping behavior, defined as events in which mice exhibited sudden hopping or leaping movements. Representative events are shown as overlaid video frames. (F) While only a handful of mice exhibited jumping behavior, no significant differences were detected when comparing the control and stress groups. However, males tended to do more jumping than females, particularly during the recall test (*P < 0.05) (N = 16 per group; CTL, no-stress control; SIS, social isolation stress; M, males; F, females).

FIGURE 3

Passive defensive mechanisms related to protection, coping, and ambivalence were also prominent during the thermal safety task. (A) Freezing behavior, defined as periods of minimal mobility. This behavior is often implemented as a protection mechanism against threat, pain, or punishment. (B) The rate of freezing behavior did not differ between the control and stress groups. However, males exhibited significantly more freezing than females during both sessions (*P < 0.05). (C) Crouching behavior, defined as periods in which mice adopted a hunched posture with the head and forelimbs rolled inward while supporting most of the body on the hindlimbs. This behavior is often implemented as a coping mechanism against cold temperatures. (D) The rate of crouching behavior did not differ between the control and stress groups. However, females exhibited significantly more crouching than males during the recall test (*P < 0.05). (E) Body stretching, defined as periods in which mice exhibited an elongated body posture. Stretched postures in rodents represent periods of ambivalence during threat, conflict, or uncertainty. (F) Body stretching was exhibited more prominently during the training session in all groups (****P < 0.0001). Yet, this behavior did not differ between the control and stress groups, or the male and female cohorts, during either training or recall (N = 16 per group; CTL, no-stress control; SIS, social isolation stress; M, males; F, females).

FIGURE 4

Other behaviors of interest during the thermal task. (A) Risk-taking behavior, defined as periods in which mice protruded their head into the cold zones while keeping the rest of the body inside the warm zone. (B) Risk-taking was exhibited more prominently during the training session in all groups (****P < 0.0001). While risk-taking did not differ between the control and stress groups, males exhibited significantly more risk-taking than females during the training session. (C) Grooming behavior as an index of stress during exposure to the cold. (D) Grooming rates did not differ between the control and stress groups. However, females exhibited a lot more grooming than males during the recall test (****P < 0.0001). (E) Motion speed as an index of general locomotion. (F) Although all the groups exhibited slightly faster average speeds during the training session than the recall test (*P < 0.05), the motion speeds did not differ between the control and stress groups, or the male and female cohorts (N = 16 per group; CTL, no-stress control; SIS, social isolation stress; M, males; F, females).

FIGURE 5

Linear regressions amongst the behaviors. (A–D) Regressions during the training session. (E–H) Regressions during the recall test. Correlations were computed for all the behavioral pairs, except for jumping behavior due to undersampling. These analyses considered Gaussian distributions, sixteen sample per correlation, and two-tailed P-values. Correlation coefficients greater than ±0.50, ±0.62, ±0.75, or ±0.82 corresponded to statistical significance levels of P < 0.05, P < 0.01, P < 0.001, or P < 0.0001, respectively. Graphical representations for all the linear regressions are shown in Supplementary Figures 2–4 (CTL, no-stress control; SIS, social isolation stress; M, males; F, females).