The socially enriched environment test: a new approach to evaluate social behavior in a mouse model of social anxiety disorder

Abstract

Social anxiety disorder (SAD) is a common anxiety disorder characterized by a marked fear of social situations. Treatments for SAD, including exposure therapy and medication, are not satisfactory for all patients. This has led to the development of several paradigms to study social fear in rodents. However, there are still some social impairments observed in SAD patients that have never been examined in rodent models. Indeed, social situations avoided by SAD patients include not only social interactions but also public performances and being observed by others. Nevertheless, tests used to assess sociability in rodents evaluate mostly social interaction in pairs. Thus, we developed a new test—a socially enriched environment test—that evaluates sociability within a group of three unfamiliar conspecifics in an enriched environment. In this study, we induced a SAD-like behavior (i.e., social fear) in male mice using social fear conditioning (SFC) and then tested social fear using the socially enriched environment test and the three-chamber test. Finally, we tested the effects of fear extinction and acute diazepam treatment in reversing social fear. Results revealed, in conditioned mice, decreased object exploration in proximity to conspecifics, social interaction, and mouse-like object exploration. Extinction training, but not acute diazepam treatment, reversed SFC-induced behavioral changes. These findings demonstrate that the socially enriched environment test provides an appropriate behavioral approach to better understand the etiology of SAD. This test may also have important implications in the exploration of new treatments.

Fear is a natural, powerful, and primitive emotion. Although fear is a survival mechanism, it can also lead to the development of mental health issues such as social anxiety disorder (SAD), which is characterized by a marked fear of social situations (American Psychiatric Association 2013). Because fear conditioning may play a significant role in the development of SAD (Öhman and Mineka 2001), a fear extinction-based approach, exposure therapy, is often used to reduce symptoms (Iverach and Rapee 2014; Mayo-Wilson et al. 2014). Medication originally designed for depression or generalized anxiety is also used [antidepressants (Mayo-Wilson et al. 2014) and anxiolytics (Helmus et al. 2005), respectively]. Unfortunately, these treatment options (exposure therapy and medication) are not entirely satisfactory.

To acquire information that will help improve SAD treatments, rodent studies may be useful (Lukas et al. 2011; Toth et al. 2012). For this purpose, several paradigms, including maternal separation (Franklin et al. 2011), social defeat stress (Huhman 2006), and footshock exposure (Haller and Bakos 2002; Toth et al. 2012) have been tested. However, the main SAD-like symptom (i.e., social fear) is primarily assessed from the social behavior of animals interacting in pairs (e.g. social preference test and three-chamber test) (Landauer and Balster 1982). Very few studies have been designed to assess the behavior of rodents in a complex environment with the presence of many peers (Shemesh et al. 2013; Houwing et al. 2019), but not yet to assess social fear.

To overcome this limitation, we developed a socially enriched environment (SEE) test designed to evaluate social behavior of mice in an environment containing several objects and a group of three unfamiliar conspecifics, which allowed us to evaluate collective and individual object exploration. We hypothesized that the increase or decrease in novelty-seeking behavior in this enriched environment would be strongly correlated not only with the presence of social stimuli in the same behavioral apparatus but also with the proximity of these social stimuli and the type of social interactions that can be engaged in.

To induce social fear (which was hypothesized to be characterized by reduced object exploration and increased freezing behavior in the presence of conspecifics, as well as reduced mouse-like object exploration), a SAD mouse model based on social fear conditioning (SFC) was used (Toth et al. 2012). We also performed a commonly used sociability test, the three-chamber test, for comparison. To suppress social fear, two different treatments—fear extinction and diazepam treatment—were tested.

Results

Experiment 1: behavioral effects of SFC

Effects of social fear in the SEE test

The behavioral schedule is outlined in Figure 1. Object exploration outside the proximity zone (OPZ) was recorded when no stimulus mouse entered a zone of 10 cm, counting from the head or the back of the experimental mouse (Fig. 2A). When the head of a stimulus mouse entered the proximity zone, the type of object exploration was considered as inside the proximity zone (IPZ) (Fig. 2B). SFC reduced IPZ object exploration, while the OPZ object exploration time remained unchanged, as compared with mice that were not subjected to SFC (Fig. 3A). A two-way ANOVA performed on SFC and exploration type data showed a significant effect of SFC (F_(1,20) = 13.350, P = 0.001) but not of exploration type (F_(1,20) = 0.180, P = 0.675). However, their interaction (SFC × exploration type) was significant (F_(1,20) = 4.643, P = 0.043). Holm–Sidak post-hoc analysis showed a significantly lower OPZ object exploration displayed by SFC⁺ mice compared with SFC⁻ mice (t = 4.108, P = 0.003). The two groups did not differ from each other regarding OPZ object exploration time (t = 1.060, P = 0.415).

Figure 1.

Time line of the three experimental procedures. Experiment 1: behavioral effects of SFC. Experiment 2: behavioral effects of acute diazepam treatment. Experiment 3: behavioral effects of social fear extinction. (SFC) Social fear conditioning, (SEE) socially enriched environment, (EPM) elevated plus maze.

Figure 2.

Socially enriched environment test setup. Schematic illustration depicting OPZ object exploration (A) and IPZ object exploration (B) of an experimental mouse (in blue), with the presence of three unfamiliar conspecifics (in white). (OPZ) Out of the proximity zone, (IPZ) inside the proximity zone.

Figure 3.

Behavioral responses to SFC in the SEE test and in the three-chamber test during experiment 1. (A) OPZ and IPZ object exploration time in the SEE test. (B) Preference for OPZ object exploration. (C) Social interaction time. (D) Freezing duration. (E) White ball exploration duration. (F) Schematic illustration of sociability and social novelty sessions in the three-chamber test. (G) Approach to cages time during sociability session. (H) Social preference index during sociability session. (I) Approach to cages time during social novelty session. (J) Social novelty index during social novelty session. Results are expressed as mean ± SEM. SEE test: n(SFC⁻) = 6, n(SFC⁺) = 6; three-chamber test: n(SFC⁻) = 7, n(SFC⁺) = 5. (***) P < 0.001, (**) P < 0.01, (*) P < 0.05: significant differences compared with SFC⁻ group. (£££) P < 0.001, (££) P < 0.01: significant differences compared with stranger 1. (OPZ) Out of the proximity zone, (IPZ) inside the proximity zone.

To better illustrate the preference for OPZ object exploration over IPZ object exploration, a percentage of this preference was calculated (Fig. 3B). We found that preference for OPZ object exploration was significantly higher in the SFC⁺ group than in the SFC⁻ group (unpaired t-test: t = 3.450, P = 0.006).

Data also showed that SFC reduced social interaction (Fig. 3C). Indeed, social interaction time differed significantly between SFC⁺ and SFC⁻ groups (unpaired t-test: t = 2.928, P = 0.015). Furthermore, SFC increased freezing duration in the SFC⁺ group (unpaired t-test: t = 3.155, P = 0.010) (Fig. 3D).

Finally, when exposed to white balls (Fig. 3E), SFC⁺ mice spent less time exploring them compared with the SFC⁻ group (unpaired t-test: t = 3.079, P = 0.011).

Effects of social fear in the three-chamber test

During the sociability session, a two-way ANOVA performed on data on time spent near to the cage (Fig. 3G) demonstrated no significant effect of cage (F_(1,20) = 1.303, P = 0.267). However, there were significant effects of SFC (F_(1,20) = 31.300, P < 0.001) and of the SFC × cage interaction (F_(1,20) = 16.560, P < 0.001). Holm–Sidak post-hoc analysis revealed that SFC⁻ mice spent significantly more time near the cage with stranger 1 than near the empty cage (t = 4.037, P = 0.001), while SFC⁺ mice avoided approaching the cage with stranger 1 and differed significantly from SFC⁻ mice (t = 4.538, P < 0.001).

Moreover, the analysis of the social preference index confirmed the avoidance of stranger 1 induced by SFC (Fig. 3H). Mann–Whitney test indicated a significant difference between the two groups [U (10) = 0, P = 0.002].

In the social novelty session (Fig. 3I), a two-way ANOVA analysis performed on the time spent near each cage data demonstrated significant effects of SFC (F_(1,20) = 42.300, P < 0.001), of cage (F_(1,20) = 13.350, P = 0.016), and of SFC × cage interaction (F_(1,20) = 13.470, P = 0.001). Holm–Sidak post-hoc analysis showed that SFC⁻ mice spent significantly more time near the cage with stranger 2 than near the one with stranger 1 (t = 5.673, P < 0.001). However, all SFC⁺ mice approached neither stranger 1 nor stranger 2. They significantly differed from SFC⁻ mice in approaching the cage with stranger 2 (t = 7.194, P < 0.001). This was confirmed with the social novelty index (Fig. 3J), which demonstrated a significant difference between SFC⁺ and SFC⁻ mice according to the Mann–Whitney test (U (10) = 0, P = 0.003).

Effects on anxiety-like and depressive-like behaviors

The unpaired t-test performed on the anxiety index (Fig. 4A) showed no significant difference between the SFC⁺ and SFC⁻ groups (t = 0.458, P = 0.651). Similarly, the unpaired t-test done on grooming time data (Fig. 4B) revealed no significant difference between the two groups (t = 0.829, P = 0.415).

Figure 4.

Effects of SFC on anxiety-like and depressive-like behaviors during experiment 1. (A) Anxiety index measured in the EPM test. (B) Grooming time measured in the splash test. Data are presented as mean ± SEM. n(SFC⁻) = 13, n(SFC⁺) = 11.

Experiment 2: the effect of acute diazepam treatment on social fear induced by SFC

Effect on social fear in the SEE test

Data showed that acute diazepam treatment failed to reverse the avoidance of IPZ object exploration induced by SFC (Fig. 5A). A three-way ANOVA analysis (SFC × exploration type × treatment) revealed a significant effect of treatment (F_(1,44) = 8.068, P = 0.007), but there was not a significant effect of SFC (F_(1,44) = 1.474, P = 0.231) or of exploration type (F_(1,44) = 2.313; P = 0.135). However, the SFC × exploration type interaction was significant (F_(1,44) = 4.087, P = 0.049). The remaining interactions did not reach significance (SFC × treatment: F_(1,44) = 1.611, P = 0.211; treatment × exploration type: F_(1,44) = 0.005, P = 0.981). Holm–Sidak's multiple comparisons did not reveal any significant difference between the groups.

Figure 5.

Behavioral responses to acute diazepam treatment in the SEE test and the three-chamber test during experiment 2. (A) OPZ and IPZ object exploration time in the SEE test. (B) Preference for OPZ object exploration. (C) Social interaction time. (D) Freezing duration. (E) White ball exploration duration. (F) Schematic illustration of sociability and social novelty sessions in the three-chamber test. (G) Approach to cages time during sociability session. (H) Social preference index during sociability session. (I) Approach to cages time during social novelty session. (J) Social novelty index during social novelty session. Results are expressed as mean ± SEM. SEE test: n(SFC⁻/Veh) = 6, n(SFC⁻/Dia) = 7, n(SFC⁺/Veh) = 6, n(SFC⁺/Dia) = 7; three-chamber test: n(SFC⁻/Veh) = 5, n(SFC⁻/Dia) = 6, n(SFC⁺/Veh) = 6, n(SFC⁺/Dia) = 6. (*) P < 0.05: significant differences compared with OPZ object exploration. (£££) P < 0.001: significant differences compared with stranger 1. (¤¤¤) P < 0.001, (¤¤) P < 0.01, (¤) P < 0.05: significant differences compared with the SFC⁻/Veh group. (###) P < 0.001, (##) P < 0.01, (#) P < 0.05: significant differences compared with the SFC⁻/Dia group. ($$) P < 0.01: significant differences compared with the SFC⁺/Veh group. (OPZ) Out of the proximity zone, (IPZ) inside the proximity zone.

To better evaluate the effect of treatment on object exploration, we performed a two-way ANOVA analysis on each of the conditioned and unconditioned groups separately. On the one hand, the analysis performed on SFC⁻ groups revealed a significant effect of treatment (F_(1,22) = 12.14, P = 0.002). However, neither the exploration type nor the treatment × exploration type interaction reached significance (F_(1,22) = 0.18, P = 0.675; and F_(1,22) = 0.136, P = 0.715, respectively). According to Holm–Sidak's multiple comparisons, the SFC⁻/Dia group exhibited both higher OPZ and IPZ object exploration times than the SFC⁻/Veh group (t = 2.725, P = 0.024; and t = 2.203, P = 0.038, respectively).

On the other hand, the two-way ANOVA analysis performed on SFC⁺ groups did not show any effect of treatment (F_(1,22) = 0.946, P = 0.341). There was a significant effect of the exploration type (F_(1,22) = 4.811, P = 0.039) but not of the treatment × exploration type interaction (F_(1,22) = 0.089, P = 0.767). Holm–Sidak's post-hoc analysis performed on object exploration time did not show any significant difference between the SFC⁺/Veh and the SFC⁺/Dia groups.

Furthermore, to evaluate the effect of SFC on exploration time, we performed a two-way ANOVA analysis on each of the vehicle and diazepam-treated group. The two-way ANOVA analysis performed on vehicle-treated groups did not show a significant effect of SFC (F_(1,20) = 0.003, P = 0.953) or of exploration type (F_(1,20) = 2.690, P = 0.116). However, the SFC × exploration type interaction was significant (F_(1,20) = 6.94, P = 0.015). Holm–Sidak's post-hoc analysis revealed that SFC⁺/Veh mice displayed a significantly higher OPZ object exploration time compared with IPZ object exploration time (t = 3.023, P = 0.013). Regarding the diazepam-treated group, two-way ANOVA analysis demonstrated no significant effect of SFC (F_(1,24) = 2.242, P = 0.147), of exploration type (F_(1,24) = 0.814, P = 0.375), or of the interaction SFC × exploration type (F_(1,24) = 0.886, P = 0.355). As a result, Holm–Sidak's post-hoc analysis tests did not show any significant difference between OPZ and IPZ object exploration within the SFC⁺/Dia group.

Analysis of the preference for OPZ object exploration (Fig. 5B) showed that both the SFC⁺/Veh and SFC⁺/Dia groups preferred OPZ object exploration rather than IPZ object exploration (67.09% ± 5.213% and 63.44% ± 5.483%, respectively). A two-way ANOVA analysis on these data revealed a significant effect of SFC (F_(1,22) = 16.940, P < 0.001). However, there was not a significant effect of treatment (F_(1,22) = 0.007, P = 0.931) or of SFC × treatment interaction (F_(1,22) = 0.753, P = 0.394). Post-hoc analysis using Holm–Sidak's test indicated that the SFC⁺/Veh group exhibited a significant preference for OPZ object exploration compared with the SFC⁻/Veh group (t = 3.396, P = 0.015). Similarly, SFC⁺/Dia mice displayed a higher preference for OPZ object exploration compared with the SFC⁻/Dia group (t = 2.390, P = 0.025).

Regarding social interaction results (Fig. 5C), a two-way ANOVA analysis revealed significant effects of SFC (F_(1,22) = 15.930, P < 0.001) and of treatment (F_(1,22) = 13.290, P = 0.001). However, the SFC × treatment interaction was not significant (F_(1,22) = 1.552, P = 0.225). Holm–Sidak's post-hoc comparisons showed that SFC⁻/Dia mice expressed a significantly increased social interaction time compared with the SFC⁻/Veh group (t = 3.459, P = 0.008), while social interaction time was significantly lower in the SFC⁺/Dia group compared with the SFC−/Dia group (t = 3.854, P = 0.004). However, no significant difference was observed between SFC⁺/Dia and SFC⁺/Veh (t = 1.697, P = 0.103).

A two-way ANOVA analysis on freezing duration (Fig. 5D) revealed significant effects of SFC (F_(1,22) = 57.880, P < 0.001), of treatment (F_(1,22) = 10.280, P = 0.004), and of SFC × treatment interaction (F_(1,22) = 6.411, P = 0.019). Holm–Sidak's post-hoc analysis showed that the SFC⁺/Veh and SFC⁺/Dia groups expressed a significantly increased freezing time compared with the SFC⁻/Veh group (t = 6.909, P < 0.001) and with the SFC⁻/Dia group (t = 3.736, P = 0.003), respectively. Furthermore, the SFC⁺/Dia group froze significantly less than the SFC⁺/Veh group (t = 4.058, P = 0.002).

Finally, a two-way ANOVA analysis performed on data of white ball exploration (Fig. 5E), showed a significant effect of SFC (F_(1,22) = 21.660, P < 0.001) but not of treatment (F_(1,22) = 0.118, P = 0.734). However, a significant interaction between SFC and treatment was found (F_(1,22) = 12.940, P = 0.001). Holm–Sidak's post-hoc test revealed that white ball exploration time was significantly increased in the SFC⁻/Dia group compared with the SFC⁻/Veh group (t = 2.787, P = 0.021) and with the SFC⁺/Dia group (t = 6.073, P < 0.001).

Effects on social fear in the three-chamber test

During the sociability session, SFC induced a significant avoidance of stranger 1, and diazepam treatment failed to reverse this social avoidance (Fig. 5G). A three-way ANOVA analysis on data on the time spent near the cages revealed a significant effect of each of the three factors (SFC: F_(1,38) = 162.500, P < 0.001; cage: F_(1,38) = 55.100, P < 0.001; and treatment: F_(1,38) = 37.110, P < 0.001). Both SFC × cage and SFC × treatment interactions were significant (F_(1,38) = 168.1, P < 0.001; and F_(1,38) = 6.140, P = 0.017, respectively), while the cage × treatment interaction was not significant (F_(1,38) = 1.208, P = 0.278). Holm–Sidak's post-hoc analysis revealed that SFC⁻/Veh and SFC⁻/Dia groups spent more time exploring stranger 1 compared with the empty cage (t = 6.850, P < 0.001; and t = 13.390, P < 0.001, respectively). In addition, SFC⁻/Dia mice spent significantly more time exploring stranger 1 compared with SFC⁻/Veh mice (t = 6.992, P < 0.001). However, both the SFC⁺/Veh and SFC⁺/Dia groups avoided exploring stranger 1, which resulted in a significant decrease of the approach to stranger 1 compared with the SFC⁻ groups (SFC⁺/Veh vs. SFC⁻/Veh group: t = 9.082, P < 0.001; SFC⁺/Dia vs. SFC⁻/Dia group: t = 16.820, P < 0.001). Moreover, SFC⁺/Dia mice spent significantly more time near the empty cage compared with stranger 1 (t = 4.657, P < 0.001) and compared with SFC⁺/Veh mice (t = 3.668, P = 0.009).

A two-way ANOVA analysis on the social preference index data (Fig. 5H) showed significant effects of SFC (F_(1,19) = 400.4, P < 0.001) and of treatment (F_(1,19) = 6.329, P = 0.021), while their interaction was not significant (F_(1,19) = 3.318, P = 0.084). Holm–Sidak's multiple comparisons test revealed that SFC⁻/Dia mice displayed a significantly higher social preference index than SFC⁻/Veh mice (t = 2.996, P = 0.014). In addition, SFC⁺/Veh mice expressed a significantly lower social preference index compared with SFC⁻/Veh mice (t = 12.570, P < 0.001). The same result was observed for SFC⁺/Dia mice compared with SFC⁻/Dia mice (t = 15.820, P < 0.001).

During the social novelty session, data showed that diazepam treatment failed to reverse the avoidance of social stimuli (Fig. 5I). A three-way ANOVA analysis on these data indicated a significant effect of each of the three factors (SFC: F_(1,38) = 100.2, P < 0.001; cage: F_(1,38) = 8.447, P = 0.006; and treatment: F_(1,38) = 13.80, P < 0.001). The SFC × cage and SFC × treatment interactions were significant (F_(1,38) = 11.53, P = 0.001; and F_(1,38) = 7.027, P = 0.011, respectively). However, the cage × treatment interaction was not significant (F_(1,38) = 1.666, P = 0.204).

Holm–Sidak's post-hoc comparisons showed for SFC⁻/Dia mice a significantly elevated approach to stranger 2 compared with stranger 1 (t = 4.393, P < 0.001) and compared with SFC⁻/Veh mice exploring stranger 2 (t = 4.220, P = 0.001). SFC⁺/Veh mice expressed a reduced but not significant approach to stranger 1 and a significantly lower approach to stranger 2 compared with SFC⁻/Veh mice (t = 2.415, P = 0.136; and t = 4.774, P < 0.001, respectively). Moreover, SFC⁺/Dia mice displayed a significantly reduced approach to both stranger 1 and stranger 2 compared with SFC⁻/Dia mice (t = 4.243, P = 0.001; and t = 8.729, P < 0.001, respectively).

Regarding the social novelty index (Fig. 5J), a two-way ANOVA analysis revealed a significant effect of SFC (F_(1,19) = 10.69, P = 0.004) without any significant effect of treatment (F_(1,19) = 3.196, P = 0.089). Also, the SFC × treatment interaction was not significant (F_(1,19) = 2.133, P = 0.160). Further analysis using Holm–Sidak's multiple comparisons showed that SFC⁺/Veh mice expressed a significantly lower social novelty index compared with SFC⁻/Veh mice (t = 3.268, P = 0.008). An increased social novelty index was obtained in SFC⁺/Dia mice, but this result is due to the absence of preference between stranger 1 and stranger 2 and does not reflect a reversal of social fear.

Experiment 3: effects of extinction on social fear

Effects of extinction training

During the first day of extinction training, a two-way repeated measures ANOVA analysis on data on the nonsocial exposure sessions (Fig. 6A) revealed significant effects of SFC (F_(1,13) = 8.782, P = 0.011) and of session (F_{(1.987,25.83)} = 7.511, P = 0.002). However, the SFC × session interaction did not reach significance (F_(2,26) = 0.366, P = 0.696). Holm–Sidak's post-hoc analysis showed that SFC⁺/Ext mice displayed a significantly lower nonsocial investigation time during the first exposure compared with SFC⁻/Ext mice (ns1: t = 3.042, P = 0.03). In contrast, the two groups did not differ from each other during the following nonsocial exposures (ns2: t = 2.597, P = 0.055; ns3: t = 1.122, P = 0.091).

Figure 6.

Behavioral responses to fear extinction during experiment 3. (A) Investigation of nonsocial stimuli (ns1–ns3) during extinction training 1 d following SFC. (B) Investigation of social stimuli (s1–s6) during extinction training 1 d following SFC. (C) Investigation of nonsocial stimuli (ns1–ns3) during extinction training 2 d following SFC. (D) Investigation of social stimuli (s1–s6) during extinction training 2 d following SFC. (E) OPZ and IPZ object exploration time in the SEE test 1 d after fear extinction. (F) Preference for OPZ object exploration. (G) Social interaction time. (H) Freezing duration. (I) White ball exploration time. Results are expressed as mean ± SEM. n(SFC⁻/Ext) = 8, n(SFC⁺/Ext) = 7. (**) P < 0.01, (*) P < 0.05: significant differences compared with the SFC⁻/Ext group. (###) P < 0.001: significant differences compared with OPZ object exploration duration. (OPZ) Out of the proximity zone, (IPZ) inside the proximity zone.

During social investigation sessions (Fig. 6B), a two-way repeated measures ANOVA analysis demonstrated a significant effect of SFC (F_(1,13) = 27.48, P < 0.001) without any significant effect of session (F_{(1.541,20.04)} = 1.161, P = 0.320). The SFC × session interaction was not significant (F_(5,65) = 0.517, P = 0.761). Holm–Sidak's post-hoc comparisons revealed a significantly lower investigation time of SFC⁺/Ext mice during the six sessions compared with the SFC⁻/Ext group (s1: t = 4,759, P = 0.008; s2: t = 4.633, P = 0.008; s3: t = 3.575, P = 0.008; s4: t = 4.953, P = 0.005; s5: t = 5.099, P = 0.006; and s6: t = 3.851, P = 0.008).

During the second day of extinction training, a two-way repeated measures ANOVA analysis performed on nonsocial investigation data (Fig. 6C) revealed a significant effect of session (F_{(1.865,24.24)} = 5.107, P = 0.015), but no significant effect of SFC was found (F_(1,13) = 4.599, P = 0.051). In addition, the SFC × session interaction was not significant (F_(2,26) = 0.159, P = 0.853). The Holm–Sidak's multiple comparisons test indicated no significant difference in nonsocial investigation between SFC⁺/Ext and SFC⁻/Ext mice during the three exposures (ns1: t = 1.756, P = 0.186; ns2: t = 2.021, P = 0.181; and ns3: t = 1.804, P = 0.186).

In the social exposures (Fig. 6D), a two-way repeated measures ANOVA analysis revealed a significant effect of SFC (F_(1,13) = 6.401, P = 0.025), but no significant effect of session was observed (F_{(2.370,30.81)} = 1.307, P = 0.288). However, the SFC × session interaction was significant (F_(5,65) = 2.561, P = 0.035). The Holm–Sidak's multiple comparisons test showed a significantly lower social investigation displayed by the SFC⁺/Ext group compared with the SFC⁻/Ext group during only the first exposure (s1: t = 6,792, P < 001, s2: t = 2,226, P = 0,229; s3: t = 2,175, P = 0,238; s4: t = 0,670, P = 0,519; s5: t = 1,238, P = 0,427; and s6: t = 1,874, P = 0,241).

Effects of extinction on social fear in the SEE test

Social fear extinction reversed the avoidance of IPZ object exploration induced by SFC (Fig. 6E). Indeed, a two-way ANOVA analysis performed on object exploration duration revealed a significant effect of SFC (F_(1,26) = 9.475, P = 0.004) and of the exploration type (F_(1,26) = 48.82, P < 0.001). However, the SFC × exploration type interaction was not significant (F_(1,26) = 0.045, P = 0.833). Holm–Sidak's post-hoc analysis showed no significant difference regarding both OPZ and IPZ object exploration between SFC⁺/Ext and SFC⁻/Ext mice (t = 2,027, P = 0,055; and t = 2,327, P = 0,055, respectively). Moreover, a significantly higher IPZ object exploration time was observed as compared with OPZ object exploration time within both SFC⁻/Ext and SFC⁺/Ext groups (t = 5,269, P < 0.001; and t = 4.638, P < 0.001, respectively). Also, no preference for OPZ object exploration was exhibited by SFC⁺/Ext mice compared with SFC⁻/Ext mice (unpaired t-test: t = 0.445, P = 0,663) (Fig. 6F).

Regarding social interaction, data indicated that no social avoidance was expressed by SFC⁺/Ext mice compared with SFC⁻/Ext mice (unpaired t-test: t = 1.078, P = 0.300) (Fig. 6G). Moreover, no significant difference of freezing responses was displayed by the SFC⁺/Ext group compared with the SFC⁻/Ext group (t = 1.949, P = 0.073) (Fig. 6H).

Finally, unpaired t-tests performed on white balls exploration time indicated no significant difference between SFC⁻/Ext and SFC⁺/Ext groups (t = 1.783, P = 0.097) (Fig. 6I).

Discussion

In the present study, we developed a new sociability test that enabled us to evaluate social behavior of mice exploring an enriched environment with the presence of a group of unfamiliar conspecifics (the SEE test). SFC reduced IPZ object exploration, decreased social interaction, and increased freezing behavior. These behaviors, which were better captured using the SEE test than the three-chamber test, were reversed with extinction training but not with acute diazepam treatment.

As to assessing sociability in animals, to date, two types of approaches have been used. Collective social behavior, which is used in large groups such as flocks of birds (Cavagna et al. 2010) and fish schools (Sumpter et al. 2008) to study the dynamic of the group (Shemesh et al. 2013), and individual social behavior, which is used in pairs in laboratory animals to study the preference to approach a social stimulus rather than avoiding it or exploring a nonsocial stimulus (Toth and Neumann 2013). In this last case, social fear is exhibited by less social investigation of unfamiliar conspecifics, and fear responses such as freezing, defensive attacks, or negative ultrasonic vocalizations are also displayed (Blanchard et al. 2005). One inconvenience is that the social stimulus is often confined to a restricted area while the experimental animal remains free to roam (Haller and Bakos 2002; Lai and Johnston 2002; Nadler et al. 2004). This may bias the social dynamic existing between the experimental animal and the social stimulus, such as dominance versus submission relationships.

Given that the laboratory setting is an artificial and deficient environment, the inclusion of enriched environments with other socially relevant parameters (such as sniffing, walking, rearing, digging, and dragging objects) (Renner and Seltzer 1991; Blois-Heulin and Belzung 1995; Belzung 1999) is necessary for a more reliable assessment of social behavior.

In the SEE test, conditioned mice avoided exploring nonsocial stimuli in the proximity of social stimuli. This result is in accordance with the symptoms observed in SAD patients who fear taking part in activities when other people are around (American Psychiatric Association 2013). However, the object exploration of conditioned mice while being away from stimulus mice (outside the proximity zone) did not change after social fear conditioning. Thus, SFC did not impact the motivation of mice to explore nonsocial stimuli in the absence of social stimuli, providing evidence that SFC does not impact general exploration of nonsocial stimuli. This can be explained by the specificity of SFC to induce fear toward social stimuli only (Toth et al. 2012). In contrast, the decrease of social interaction in conditioned mice can be explained by social fear (Toth et al. 2012; Kornhuber et al. 2019). Furthermore, using the same SAD model as in our study, a study by Kornhuber et al. (2019) found elevated corticosterone levels in conditioned mice only when they were confronted with social stimuli. Most tests rely on the evaluation of social approach versus social avoidance. However, social avoidance does not necessarily reflect social fear and might be the result of a decrease in social motivation (Toth and Neumann 2013), which can bias the evaluation of social fear. Thus, it is important to explore additional fear responses associated with social fear, such as freezing (Blanchard et al. 2005). In our study, the increase of freezing duration confirms high levels of fear associated exclusively with the presence of unfamiliar social stimuli in the same environment.

We also found that investigation of nonsocial stimuli that resemble social stimuli in size, shape, and color was decreased in conditioned mice. It can be suggested that in addition to olfactory cues, visual cues can also play an important role in distinguishing between neutral and aversive stimuli and therefore between nonsocial and social stimuli (Winslow 2003).

Concerning the three-chamber test, it is commonly used to evaluate social deficits in animal models of psychiatric disorders (Rein et al. 2020). It enables the quantification of a wider variety of parameters compared with other tests, such as preference for social stimulus over nonsocial stimulus and preference for novel social stimulus versus familiar social stimulus (Landauer and Balster 1982).

Data of the three-chamber test revealed that conditioned mice displayed an increased fear toward the unfamiliar social stimulus. In the social novelty session, these mice avoided approaching both familiar and novel social stimuli. This result reveals that SFC induced a consistent fear of social stimuli in both sessions of the test, as expressed by other animal models of dysfunctional social behavior (Chang et al. 2017).

Both SEE and three-chamber tests present advantageous features as well as limitations. The SEE test appears to mimic the dynamics of human social interactions better than the three-chamber test. Indeed, the SEE test evaluates social interaction while the experimental animal is free to explore a new environment with the presence of other peers free to explore as well, whereas in the three-chamber test, the stimulus mouse is locked in a cage and the experimental mouse is free to approach it. Moreover, social behavior in rodents encompasses other components such as social dominance and hierarchy, which play a significant role in guiding and modulating social interactions (Kunkel and Wang 2018). Hence, submitting mice to the three-chamber test could impact the expression of innate social behavior and therefore interfere with the evaluation of social behavior.

Regarding nonsocial stimulus investigation, the SEE test not only evaluates how experimental mice interact with the unfamiliar conspecifics but also investigates whether, in the presence of congeners, experimental mice express fear generalization (e.g., fear of white balls) and a change in motivation to explore neutral objects (e.g., toys). In contrast, the three-chamber test does not evaluate the mouse's preference for exploring a social stimulus over a nonsocial stimulus. In this regard, several variants of this test have been developed to improve its sensitivity in detecting social deficits (Rein et al. 2020).

The SEE test also allows the evaluation of social interactions in an enriched new environment, which helps to reduce fear and to motivate the mice to interact, as enriched environments usually facilitate social responses (Shemesh et al. 2013; Neal et al. 2018; Houwing et al. 2019). Regarding motivation, studies have suggested that novelty plays an important role in motivating animals to explore. In fact, novelty seeking has a reward value, and novel stimuli trigger reward processing in the dopaminergic midbrain through the activation of the substantia nigra/ventral tegmental area (Koster et al. 2016). In the context of the SEE test, enhancing the motivation of the animal to explore a new enriched environment can be important not only in reducing stress related to being in this environment but also in enhancing the exploratory behavior of the animal toward both social and nonsocial stimuli.

Because behavioral responses such as fear and anxiety are shared symptoms of all anxiety disorders (American Psychiatric Association 2013), we performed the EPM and the splash tests to evaluate the effect of SFC on general anxiety and depressive-like behavior, respectively. Results showed that neither general anxiety nor depressive-like behavior was induced by SFC. It is worth mentioning that social fear and instability generally lead to an increased anxiety and a depressive-like behavior (Öhman and Mineka 2001; Iverach and Rapee 2014). In addition, a variety of stressors intended to induce social fear in animals do not specifically induce social fear but generate other confounding behaviors such as general anxiety and depressive-like behavior (Helmus et al. 2005; Mayo-Wilson et al. 2014). In contrast, SFC is more a social fear-specific paradigm (Lukas et al. 2011; Toth et al. 2012).

Benzodiazepines have been used to reduce social impairments in SAD patients (Franklin et al. 2011) and to screen their pharmacological effects in some animal models of SAD (Huhman 2006). While their exact mechanisms remain unknown, increased inhibitory neurotransmission through positive allosteric modulation of postsynaptic GABAA receptors has been suggested (Haller and Bakos 2002). In support of this, an animal study has shown that increasing GABAergic neurotransmission with a benzodiazepine treatment reduces social defeat stress (Landauer and Balster 1982).

In a rat model of stress-induced anxiety, diazepam abolished shock-induced social avoidance (Shemesh et al. 2013). Similarly, diazepam has been shown to reverse short-term social fear in an animal model of SAD (Huhman 2006), probably through lateral septal neuronal activation (Houwing et al. 2019).

Here, using acute diazepam treatment, our results indicated that unconditioned diazepam-treated mice displayed increased OPZ and IPZ object exploration. However, the treatment failed to reverse avoidance of collective object exploration induced by SFC. Furthermore, diazepam treatment increased social interaction and white ball investigation in unconditioned mice while failing to improve these behaviors in conditioned mice. Finally, diazepam treatment decreased freezing duration in conditioned mice, which could be explained by the anxiolytic effect of this compound (Park et al. 2018). One can suggest that acute diazepam treatment failed to restore both the natural preference of mice for IPZ object exploration and the social fear induced by SFC.

In the three-chamber test, diazepam treatment significantly increased social investigation during both sociability and social novelty sessions in unconditioned groups. This result supports the increase of sociability with this treatment, as reported elsewhere (Toth et al. 2012; Shemesh et al. 2013; Houwing et al. 2019). Regarding conditioned mice, diazepam failed to reverse social fear. This finding is not in agreement with a previous study based on a similar SAD mouse model (Toth et al. 2012). This discrepancy may be explained by the way in which social behavior was measured—during social investigation while submitting mice to SFC extinction (Toth et al. 2012) versus while using a sociability test, as we did. In addition, the reported effect was not significant in the first extinction session, indicating no treatment effect regardless of fear extinction (Rao and Sadananda 2016).

Exposure therapy, which is based on fear extinction procedure, can help overcome SAD (Isingrini et al. 2010). Here, we also used extinction to examine its fear reduction effect with our SAD mouse model. To this end, mice were subjected twice (1 d apart) to a nonsocial extinction session followed by a social extinction session. The data show a significant reduced nonsocial investigation time in conditioned mice during the first trial. The extinction occurred from the second trial and was maintained during the last trial on the first day and during the three trials of the second day. However, conditioned social fear remained significantly elevated during the first six trials on the first day. Extinction was observed from the second trial on the second day and was maintained until the last trial.

In addition, on the third day, using the SEE test, we found that conditioned mice subjected to extinction and unconditioned mice behaved similarly, confirming a complete extinction of social fear, as reported in previous studies (Cavagna et al. 2010; Toth et al. 2012).

Regarding the underlying mechanisms of fear extinction, several theories suggest that fear extinction does not erase fear memories but instead generates a new memory in competition with the pre-existing one (Blanchard et al. 2005; Sumpter et al. 2008; Toth and Neumann 2013). Recently, evidence has shown that fear extinction also relies on the recruitment of the reward circuit given that the absence of the fear condition stimulus is a form of reward (Nadler et al. 2004).

Taken together, our newly developed SEE test has more translational validity given that environmental influences on the expression of social behavior are considered instead of only relying on simplified rodent setups like the three-chamber test. We believe that future work with the SEE test could pave the way for exploring the etiology, neurobiological mechanisms, and novel treatment approaches of SAD.

Materials and Methods

Animals

Experiments were performed on 162 male Swiss mice (10–12 wk old, weighing 30–35 g) raised in the central animal care facilities of the Cadi Ayyad University, Marrakesh, Morocco. Animals were housed in groups of three to five in transparent Plexiglas cages (26 × 27 × 15 cm) with wood chip bedding under controlled environmental conditions (12:12 light/dark cycle at 22°C ± 2°C) and standard diet and water ad libitum. Behavioral experiments were performed between 9:00 a.m. and 3:00 p.m. All animal procedures were in strict accordance with the ethical standards and guidelines of the European Council Directive (EU2010/63) and the Council Committee of research laboratories of the Cadi Ayyad University of Marrakesh, Morocco.

Experimental design

Social fear conditioning paradigm

To induce social fear, the SFC paradigm was performed according to Toth et al. (2012). The paradigm consisted of the administration of mild electric footshocks to an experimental mouse while exploring an unfamiliar stimulus mouse, resulting in a specific social fear lasting for at least 15 d.

The experiment was designed in three phases: social isolation, acclimatization, and SFC (Fig. 1).

In the first phase, experimental mice were housed individually 3 d prior to SFC and remained isolated throughout all of the behavioral procedures. Social isolation was used to enhance social investigation during SFC (Niesink and van Ree 1982) and to prevent the extinction of social fear in the case of group housing conditions (Cherng et al. 2010; Ruis et al. 1999).

In the second phase, social stimulus mice with identical corresponding weights and ages were acclimatized to stimulus cages for 10 min each day during 3 d before SFC. The stimulus cages consisted of iron cylinders (diameter: 7 cm, height: 7 cm) composed of vertical iron bars evenly spaced (permitting a direct contact between mice), removable lids, and a floor made of cork for electrical insulation. The cages allowed visual, olfactory, and tactile contacts between mice.

One day before SFC, experimental mice were acclimatized during 30 min to the Skinner box made of a stainless steel grid floor (20 × 20 × 25 cm; Bioseb LE918), with an empty cage (nonsocial stimulus) placed in one of the corners of the Skinner box.

During the SFC phase, an experimental mouse was placed inside the conditioning box for a 30-sec habituation time. Thereafter, a nonsocial stimulus was placed in the box, and the experimental mouse was free to explore the apparatus for 3 min. The nonsocial stimulus was then replaced by a social stimulus (cage containing a stimulus mouse) during 3 min. Conditioned mice (SFC⁺) were first allowed to explore the social stimulus for 2–3 sec and then received an electric footshock for 1 sec at 0.7 mA (shock generator; Bioseb LE10026) each time they made a direct contact with the social stimulus. To perform a reliable conditioning, SFC⁺ mice received between two and five shocks. When no further social contact was made for 2 min, mice were returned to their home cage. Unconditioned mice (SFC⁻) were submitted to the same regimen without any footshock.

Between each conditioning session, the Skinner box and stimulus cages were cleaned with 70% ethanol. Cages used as nonsocial stimuli were different from those used as social stimuli to avoid odor cues from previous subjects. Experimental animals that were not being conditioned remained in a neighboring room to prevent them hearing vocalizations emitted during SFC.

Behavioral tests

Social testing

To evaluate the effect of SFC on social behavior, the SEE test and the three-chamber test (Fig. 1) were conducted 24 h after SFC on two different groups.

Socially enriched environment (SEE) test

This test consisted of the evaluation of a mouse social behavior in a new enriched environment containing different objects and three freely moving unfamiliar conspecifics during 30 min (Fig. 2).

The apparatus consisted of a transparent rectangular arena (40 × 60 × 25 cm) containing clean mouse bedding and eight unfamiliar manufactured objects. These objects included two cubes (5 × 5 × 5 cm), one hollow cylinder (length: 9 cm, diameter: 3.5 cm), two small white ping-pong balls glued together (diameter: 4 cm), and three irregular objects (height: 6–9 cm). Six of these objects were placed along the arena walls, while two objects remained in the center so that all objects could be spaced by at least 20 cm. Objects were made of different materials with different textures as well as shapes and sizes to enhance the exploratory behavior of mice. Between each trial, the whole apparatus was cleaned with 70% ethanol, and mouse bedding was changed with clean bedding to prevent olfactory cue bias. Room lighting was 23 lux to avoid the anxiogenic effect of bright light on mice. The experiment was videorecorded by Debut video capture software v4.0 (NCH Software), and parameters were manually calculated.

The experimental mice and the stimulus mice were brought into the experimental room 30 min before testing for room acclimatization. Each experimental mouse was tested with a group of three unfamiliar stimulus mice with identical weights that were housed collectively under the same conditions. For each group of experimental mice, two groups of three stimulus mice were used. Between each trial, stimulus mice were transferred into their home cage and replaced by another group. Also, the objects’ disposition was changed to enhance their exploratory activity.

The test started by placing the experimental mouse and the three stimulus mice in the center of the apparatus and allowing them to freely explore the arena for 30 min. The parameters considered were as follows:

1. Object exploration time outside or inside a proximity zone to conspecifics was recorded when no stimulus mouse entered a zone of 10 cm (OPZ) (Fig. 2A) or when the head of a stimulus mouse entered the proximity zone (IPZ) (Fig. 2B), respectively. The perimeter of 10 cm was determined after many observations as being approximately the maximum distance of approach, after which the experimental mouse moved away or froze.

2. Preference for OPZ object exploration, which is expressed as a percentage and was calculated as the time of OPZ object exploration divided by the total time of OPZ and IPZ object exploration.

3. Social interaction time, which was measured as the total time spent by the experimental mouse exploring a stimulus mouse by making a direct contact (including whole-body sniffing, ano–genital sniffing, nose-to-nose sniffing, and the time spent following a stimulus mouse while moving in the apparatus). Social interaction must have been initiated by the experimental mouse to measure its sociability. When a stimulus mouse approached the experimental mouse and made direct contact with the experimental mouse, this contact was not scored.

4. Freezing time, which is defined as the complete absence of body movements except for respiratory-related movements (Fanselow 1990). It was measured using a 1-sec time-sampling technique.

5. White balls exploration time, which was calculated as the total time spent by the experimental mouse exploring the two white balls. It indicates whether conditioned mice would display a specific fear for social stimuli or for nonsocial stimuli similar in size and shape to a mouse. In other words, this parameter enabled us to explore whether visual cues alone can induce fear in conditioned mice.

Three-chamber test

This test was assessed to evaluate general sociability and interest in social novelty (Landauer and Balster 1982; Hanks et al. 2013). The apparatus (Figs. 3F, 5F) consisted of a transparent glass arena (40 × 60 × 25 cm) with a black mat floor divided into three equal chambers by two removable transparent glass walls with a central aperture (8 × 5 cm), allowing mice to transition between chambers. Two inverted wire cups (diameter: 7 cm, height: 10 cm) composed of thin vertical iron bars evenly spaced by 1 cm (permitting direct contact between mice) were placed in the center of the two distal chambers. These inverted cages were different from those used during social fear conditioning.

The experimental procedure consisted in three sessions: habituation, sociability, and social novelty. Each one lasted 10 min with an intertrial interval of 3 min. In the habituation session, the experimental mouse was first placed in the central chamber and was then allowed to explore freely the whole apparatus with the two cylindrical cages left empty. During the sociability session, the experimental mouse remained in the apparatus, and an unfamiliar stimulus mouse (stranger 1) was placed inside one of the two cages while the opposite one remained empty. In the social novelty trial, a second unfamiliar stimulus mouse (stranger 2) was placed inside the cage that was left empty during the sociability trial. The parameters considered were as follows:

1. Time spent in chamber, which was measured as the total time spent by experimental mice within each of the three chambers during the three sessions.

2. Approach to cage duration, which was calculated as the total time spent by experimental mice near each cage (with or without a stimulus mouse) during the three sessions in a proximity zone of 2 cm around the cages.

3. Social preference index, which is expressed as a percentage and was calculated by dividing the time spent by the experimental mouse in the chamber containing stranger 1 during the sociability session by the total time (10 min) (Park et al. 2018).

4. Social novelty index, which is expressed as a percentage and was calculated by dividing the time spent by the experimental mouse in the chamber containing stranger 2 during the social novelty session by the total time (10 min) (Park et al. 2018).

These parameters were automatically recorded using a video-tracking system (EthoVision XT8.5).

Specificity of the induced social fear

To verify that the SFC procedure did not induce anxiety-like and depressive-like behaviors, common comorbidities of SAD (Chartier et al. 2003), the elevated plus maze and the splash test were conducted 2 d and 3 d after SFC, respectively (Fig. 1).

Elevated plus maze

To assess anxiety-like behavior of conditioned mice, the elevated plus maze (EPM) test was performed 1 d after SFC (Fig. 1) according to Handley and Mithani (1984) and Lapiz-Bluhm et al. (2008). The apparatus consisted of two open arms (50 × 5 cm) and two closed arms (50 × 5× 15 cm) connected to a central area (5 × 5 cm) to form a plus sign. The maze was raised 50 cm above the ground and was lit with an approximate intensity of 200 lux. The mouse was placed at the central area of the maze facing an open arm and was left to explore the maze for a single 5-min recorded session. A video tracking system (EthoVision XT8.5, Noldus) automatically recorded the number of arm entries and the time spent in each arm during 5 min. An anxiety index was calculated as 1 − [(open arm time/total time) + (open arm entries/total entries)]/2 (Rao and Sadananda 2016).

Splash test

To assess depressive-like behavior of conditioned mice, a splash test was conducted as described by Isingrini et al. (2010) 2 d following SFC (Fig. 1). A 10% sucrose solution was squirted on the mouse's dorsal coat in its home cage, and grooming behavior (licking and scrubbing) was videorecorded by Debut video capture software v4.0 for 5 min. The total duration of grooming was manually measured as an index of self-care and motivational behavior, which is considered to be an indicator of depression (Engel et al. 2016).

Drug treatment

Acute diazepam treatment has been shown to have a reversal effect of short-term social fear induced by SFC (Toth et al. 2012). To test the effect of acute diazepam treatment on social fear during the SEE test and the three-chamber test, diazepam solution (10 mg/2mL Diapharm , Pharma 5) was administered intraperitoneally at a volume of 5 mL/kg and a dose of 0.5 mg/kg (Toth et al. 2012) 30 min before social behavior testing (Fig. 1). Vehicle groups received the same volume of saline solution (0.9%).

Extinction

Extinction training

To reduce the conditioned fear responses induced by SFC, extinction training was conducted 24 and 48 h after SFC (Fig. 1) in the home cage of the experimental mice and was performed in two sessions according to Toth et al. (2012). The first session aimed to explore nonsocial investigation and consisted of exposing the experimental mouse to three nonsocial stimuli (empty cages like those used in SFC) put in one of the corners of the home cage for 3 min with an interexposure interval of 3 min. The second session was assessed to explore social investigation by exposing the experimental mouse to six different social stimuli (unfamiliar male mice with similar weight and age confined in the cages) for 3 min with an interexposure interval of 3 min. The same procedure was repeated the following day. Parameters considered were the time spent exploring nonsocial and social stimuli.

Postextinction SEE test

To investigate the complete inhibition of social fear induced by SFC, the SEE test was assessed 24 h following the extinction training (Fig. 1).

Animal groups

To perform the SFC paradigm, 89 experimental mice were used. The study was organized in three experiments (Fig. 1). Experiment 1 included two groups: unconditioned mice (SFC⁻; n = 13) and conditioned mice (SFC⁺; n = 11). Experiment 2 included four groups: unconditioned vehicle-treated mice (SFC⁻/Veh; n = 11), unconditioned diazepam-treated mice (SFC⁻/Dia; n = 13), conditioned vehicle-treated mice (SFC⁺/Veh; n = 12), and conditioned diazepam-treated mice (SFC⁺/Dia; n = 13). The six groups were then split into 12 subgroups according to the test conducted. Experiment 3 comprised two groups: unconditioned mice (SFC⁻; n = 8) and conditioned mice (SFC⁺; n = 8). In total, 73 stimulus mice were used in the three experiments.

The behavioral scoring was done by the experimenters. One of the observers assigned a specific number to each mouse, while the other observer (the scorer) analyzed the videos, blinded to each mouse's treatment. At the end of all analyses, each mouse was renamed according to its group.

Statistical analysis

Statistical analyses were performed using Prism 9 (Graphpad Software). For statistical evaluations, a Shapiro–Wilk's normality test was run on data sets, and a parametric test (t-test) was used whenever the normality or lognormality test was passed while a nonparametric test (Mann–Whitney) was used in case of normality and lognormality failure. When evaluating multiple parameters, either two-way ANOVA with or without repeated measures or three-way ANOVA was used followed by Holm–Sidak's post-hoc for multiple comparisons. A difference was considered statistically significant at P < 0.05.

Competing interest statement

The authors declare no competing interests.