A Standardized Protocol for the Induction of Specific Social Fear in Mice

Dear Editor,

Social anxiety disorder (SAD) is one of the most common psychiatric disorders. It is characterized by intense and persistent fear of one or more social situations in which one is exposed to possible scrutiny by others [1]. The lifetime prevalence of SAD ranges from 0.2% to 13%, with higher prevalence in high-income countries. Untreated SAD can last a lifetime and is often accompanied by impairments in social life, which severely compromises the quality of life. Furthermore, SAD is associated with a higher risk of other mental and physical disorders including depressive disorders, substance abuse disorders, general anxiety disorders, and cardiovascular diseases [2]. Despite the severe social-psychological burden caused by SAD, its pathological mechanisms are poorly understood. Currently, treatment options for SAD remain inadequate. Even worse, a large proportion of patients show little response to the available psychological and pharmacological therapies [3]. To reveal the key neural processes underlying SAD and hence to develop potential therapeutic interventions, the establishment of an appropriate animal model is essential. Here we describe an experimental protocol for social fear conditioning (SFC) that can reliably produce robust, persistent, and specific social fear in mice; it is based on operant conditioning by pairing social investigation with aversive stimuli.

The SFC protocol consisted of four stages: housing acclimation (5 days), conditioning apparatus habituation (1 day), conditioning (1 day), and behavioral testing (1 day) (Fig. 1).

Fig. 1.

Schematic of the SFC paradigm. A Schematic of the experimental design. B Housing acclimation stage. The experimental mice are separated from their cage mates and individually housed for 5 days. C Conditioning apparatus habituation stage. The experimental mouse is allowed to explore the conditioning apparatus for 10 min. D Conditioning stage. This stage constitutes 5-min acclimation to the conditioning apparatus, 2-min free social interaction with the stimulus mouse, and 20-min pairing of a social investigation with a foot-shock. E Criteria for ‘social investigation’ in the conditioning stage. F Behavioral testing stage. The expression of social fear responses is examined with the social preference-avoidance test or the three-chamber social interaction test.

Housing Acclimation

In this stage, the experimental mice were separated from their cage-mates for 5 days (Fig. 1B). Since the SFC paradigm induced fear responses to their conspecifics, the experimental mice needed to be individually housed after conditioning to avoid fear extinction. To keep the housing condition consistent throughout the entire experiment, the experimental mice were therefore acclimated to the single-housing condition beforehand. However, to avoid potential anxiogenic effects caused by long-term social isolation, we recommend keeping the experimental mice group-housed until the experiment begins. During the 5 days of housing acclimation, mice were handled by the experimenter for 3–5 min per day so that the experimental mice became familiar with the experimenter and hence overcame stress.

Conditioning Apparatus Habituation

The habituation stage aimed to familiarize the experimental mouse with the conditioning apparatus and to assure the mouse that the context was safe (Fig. 1C). Each experimental mouse was allowed to explore the conditioning apparatus for 10 min the day before the conditioning day. Habituation helped the establishment of fear specificity to the social stimulus but not to the context.

Conditioning

The conditioning stage aimed to establish a successful association between social investigation and foot-shock in the experimental mouse (Fig. 1D). In this stage, an experimental mouse was placed in the conditioning apparatus and allowed to acclimate to the apparatus for 5 min. Afterward, an unfamiliar stimulus mouse was randomly introduced into one of the two stimulus containers, and the experimental mouse was allowed to freely investigate the stimulus mouse for 2 min. Then, a mild foot shock (1 s, 0.6 mA) was automatically delivered to the experimental mouse each time it investigated the stimulus mouse in the following 20 min. Social investigation was detected by the conditioning system when the following two criteria were met (Fig. 1E):

1. The distance between the nose of the experimental mouse and the center of the stimulus mouse was no more than 4.5 cm.

2. The angle between the experimental mouse's head direction and the line connecting the center points of the two mice was no more than 45°.

Behavioral Testing

The behavioral testing stage aimed to examine the behavioral consequences in social fear conditioned mice. We use the open field arena social preference-avoidance test and the three-chamber social interaction test to assess social fear in the conditioned mice (Fig. 1F). Note that the test apparatuses were different from the conditioning apparatus to rule out context-dependent fear expression. Also, the test room was quiet and illuminated by dim light to avoid disturbance from the external environment, as loud noise and bright light can induce anxiogenic effects in mice. The overall activity of each mouse in the testing apparatus was automatically registered by a video camera and stored on a computer for post-hoc analysis.

The total time that the experimental mouse spent in each zone/chamber was analyzed using a behavioral analysis program. The number of stretched approaches, a behavioral indicator of an elevated fear state in rodents and social investigations were manually counted by viewing recorded videos.

For the social preference-avoidance test in the open field arena, we calculated the following parameters:

• (A)Mean duration of investigation: the total time in the social zone divided by the number of investigations.
• (B)Index (social/corner): the total time in the interaction zone divided by the total time in the corner zone.
• (C)Index (target/no target): the ratio of total time spent by the experimental mouse in the social zone before and after introducing the social stimulus.
• (D)Percentage of stretched approaches: the fraction of stretched approaches among total approaches.
• (E)Approach speed: the movement velocity of the experimental mouse when approaching the stimulus mouse.

For the three-chamber social interaction test, we also calculated:

• (A)Mean duration of investigation: the total time in the social zone divided by the number of investigations.
• (B)Index (chamber): the difference in the time spent in the social and neutral chambers divided by the sum of the time spent in both chambers.
• (C)Index (zone): the difference in the time spent in the social and neutral zones divided by the sum of the time spent in both zones.
• (D)Percentage of stretched approaches: the fraction of stretched approaches among total approaches.
• (E)Approach speed: the movement velocity of the experimental mouse when approaching the stimulus mouse.

In the social preference-avoidance test, compared with unconditioned control mice, social fear conditioned mice spent dramatically less time in the social zone, demonstrated significantly smaller social indexes, fewer social investigations, shorter mean duration of investigation, and slower approach speed, as well as a significant increase in the time spent in the corner zones, and the number of stretched approaches (Fig. 2A–H). Likewise, in the three-chamber social interaction test, social fear conditioned mice exhibited a dramatic decrease in the time spent in the social chamber/zone, smaller social indexes, fewer social investigations, shorter mean duration of investigation, and slower approach speed, as well as a significant increase in the number of stretched approaches (Fig. 2I–Q). In contrast, the conditioned mice did not exhibit fear responses to a novel object, indicating the specificity of fear responses to the social stimulus (Fig. S2A–H). Moreover, the induced social fear responses were long-lasting and stable for up to a month (data not shown).

Fig. 2.

Social fear expression assessed with the social preference-avoidance test and three-chamber social interaction test. A Representative movement traces of an unconditioned control mouse (upper) and a conditioned experimental mouse (lower) in the social preference-avoidance test. B Time spent by unconditioned (UC) and conditioned (C) mice in the social and corner zones. C Social index (time in social zone/time in corner zones). D Social index (time in social zone with target/time in social zone without target). E–H Statistic comparison of the number of social investigations (E), mean duration of investigation (F), percentage of stretched approaches (G), and approach speed (H) by unconditioned (UC) and conditioned (C) mice. I Representative movement traces of an unconditioned control mouse (upper) and a conditioned experimental mouse (lower) in the three-chamber social interaction test. J Time spent by the unconditioned (UC) and conditioned (C) mice in each chamber. K Social index (the difference in the time spent in the social chamber and neutral chamber divided by the total time spent in both chambers). L Time spent by the unconditioned (UC) and conditioned (C) mice in each zone. M Social index (the difference in the time spent in the social zone and neutral zone divided by the total time spent in both zones). N–Q Statistic comparison of the number of social investigations (N), mean duration of investigation (O), percentage of stretched approaches (P), and approach speed (Q) by unconditioned (UC) and conditioned (C) mice. *P <0.05; **P <0.01; ****P <0.0001; two-way ANOVA, Bonferroni multiple comparison post hoc tests for B, J, L; unpaired t test for C–H, K, M–Q. Numbers in columns indicate the number of mice. Error bars, SEM.

To examine changes in the non-social domains, we used an open field assay to test general anxiety and locomotion, and used the forced swim to test depressive-like behavior. In the open field test, we found that the conditioned mice spent comparable time in the center zone and there was no difference in total distance traveled compared with control mice, suggesting that SFC did not induce impairments in anxiety and locomotion (Fig. S2I–K). In the forced swim test, we found no difference between the two groups in the immobility time, indicating that SFC did not produce depressive-like behavior (Fig. S2L, M). Together, these results illustrate that this protocol induces robust and specific social fear without introducing changes in the non-social domains. Other examples of results obtained from our group can be found in reference [4].

At least three important criteria need to be taken into consideration when we develop specific animal models for social fear. First, animals should demonstrate not only reduced social investigation but also aversive responses towards their conspecifics. Second, animals should show fear responses selectively to social stimuli but not non-social stimuli. Third, behavioral changes should be limited to the social domain and should not manifest in non-social domains. In all these conditions, the SFC paradigm has clear advantages over other existing paradigms. First, in paradigms such as foot-shock exposure [5], maternal separation [6], juvenile social isolation [7], social instability [8], and chronic subordinate colony housing [9], the decreased social interaction time in the experimental mouse could merely reflect deficits in social motivation but not necessarily social fear. In contrast, mice that underwent SFC also exhibited slower approach speeds and stretched approaches towards a stimulus mouse. These aversive responses in confrontation with the stimulus mouse are good signs indicating successful induction of social fear. Second, a locomotor defect, general anxiety, or depressive-like behavior has been reported in mice that underwent paradigms such as foot-shock exposure [5], juvenile social isolation [7], social instability [8], social defeat [10], or maternal separation [11]. These non-social phenotypes add confounding factors in the interpretation of these animal models in the context of social fear. In comparison, the SFC paradigm does not induce impairments in these non-social domains and experimental mice that underwent SFC did not show fear responses to a non-social stimulus either. Third, unlike the social defeat paradigm in which the fear response of the experimental mouse is dependent to some extent on the aggressiveness of the aggressor mouse, the operation of SFC is well controlled and can be applied to individual experimental mice in a relatively constant way. Thus, the SFC paradigm outrivals other paradigms in several aspects and can efficiently induce robust and specific social fear in a well-controlled fashion.

There are also several limitations to the use of SFC to mimic the social fear aspect of SAD in mice. First, SAD has two subtypes: generalized SAD and specific SAD or performance-only SAD [1]. And human SAD is often irrational in the sense that affected individuals realize that their fears are not associated with danger and can even be elicited by photos of feared objects like crowds or angry faces [12]. However, such complexity of human social situations is difficult or even impossible to model in rodents. In the SFC paradigm, the conditioned mice show a general social fear of their conspecifics independent of complex social situations, and therefore should be used with care when considering human SAD. Second, according to studies involving twins, genetic factors contribute ~54% to SAD in children and 27% in adults [13]. However, social fear induced by the standard SFC protocol originates from environmental factors only and cannot reflect the genetic etiology. Nevertheless, with the discovery of genetic defects associated with a higher risk of SAD [14], the SFC paradigm could be extended to genetically-engineered animals to study the genetic and environmental interplay in the pathogenesis of SAD.

Taken together, the SFC paradigm is a reliable and effective means of establishing a mouse model with specific social fear.

Supplementary Information

Below is the link to the electronic supplementary material.

Conflict of interest

The authors claim that there are no conflicts of interest.