Social and non-social mice had comparable baseline stress levels

Mice were conditioned to SI with a partner for four days in one compartment and were placed “alone” in the other compartment before being tested for expression of social CPP (Fig. 1a). After segregating the mice into social and non-social groups based on their preference score in the CPP test [unpaired two-tailed Student’s t-test, social vs. non-social, t (21) = 5.245; p < 0.0001 – Fig. 1b], CRF levels in the BNST were evaluated to investigate whether baseline stress levels would differ between the opposite social profiles. CRF levels in the BNST were similar in social and non-social mice (unpaired two-tailed Student’s t-test, social vs. non-social, t (21) = 1.125; p = 0.2731 - Fig. 1c). Additionally, no correlation was found between CRF expression in the BNST and the preference score of mice (p = 0.4335 - Fig. 1d).

Fig. 1

CRF levels in the BNST were similar in social and non-social mice. (a) Timeline. (b) SI CPP scores were used for segregating mice into social and non-social groups. Student’s t-test **** p < 0.0001, n = 9–14. (c) CRF levels in the BNST in social and non-social mice. Student’s t-test, p > 0.05, n = 9–14. (d) Correlation between social preference score and CRF levels in the BNST. p > 0.05

Inhibition of OX1R failed to shift the non-social phenotypes of mice

In this experiment, we investigated the effects of the orexin 1 receptor antagonist, SB334867, on social reward. After segregation into social and non-social profiles, based on the social preference score obtained in the CPP test 1 (T1), mice were randomly assigned to receive either SB334867 or a vehicle solution. The administration of SB334867 or the vehicle was performed 30 min prior to each SI conditioning session, with each mouse receiving a total of four i.p. injections. Following the conditioning period, mice were tested for social CPP in a second test (T2) (Fig. 2a).

At T1, social and non-social mice had different social phenotypes [three-way ANOVA, social phenotype effect, F (1, 43) = 33.13, p < 0.0001; treatment effect, F (1, 43) = 0.02299, p = 0.8802; test time effect, F (1, 43) = 1.753; p = 0.1925; social phenotype x treatment, F (1, 43) = 0.06163, p = 0.8051; social phenotype x test, F (1, 43) = 1.408, p = 0.2419; treatment x test, F (1, 43) = 0.03528, p = 0.8519; social phenotype x treatment x test time, F (1, 43) = 0.05141, p = 0.8217; Šídák’s multiple comparisons test: at T1 social pre-vehicle vs. non-social pre-vehicle, p < 0.01 and social pre-antagonist vs. non-social pre-antagonist, p < 0.01)]. At T2, social mice receiving vehicle injections during social conditioning remained in their social profile (Šídák’s multiple comparisons test, T1 pre-vehicle vs. T2 vehicle, p > 0.05). Furthermore, social mice receiving SB334867 kept their social profile (Šídák’s multiple comparisons test, T1 pre-antagonist vs. T2 antagonist, p > 0.05). Interestingly, from T1 to T2, four social mice reversed their profiles to non-social. Non-social mice receiving vehicle also kept their profile (Šídák’s multiple comparisons test, T1 pre-vehicle vs. T2 vehicle, p > 0.05). Three of these mice reversed their profiles from non-social to social. Contrary to expectations, treatment with the OX1R antagonist failed to shift the profile of non-social to social mice (Šídák’s multiple comparisons test: non-social: T1 pre-antagonist vs. T2 antagonist, p > 0.05). Indeed, only four mice out of fourteen shifted their profile from non-social to social after OX1R antagonist treatment (Fig. 2b).

In order to check for any possible effects of SB334867 on locomotion, the total distance traveled during T1 and T2 was evaluated in all groups of mice. A three-way ANOVA test revealed that treatment with OX1R antagonist yielded no effects on locomotion [treatment effect, F (1, 43) = 0.1674, p = 0.6844; social phenotype effect, F (1, 43) = 1.205, p = 0.2785; test time effect, F (1, 43) = 5.166, p = 0.0281; social phenotype x treatment, F (1, 43) = 0.3744, p = 0.5438; social phenotype x test time, F (1,43) = 1.424, p = 0.2392; treatment x test time, F (1, 43) = 0.01683, p = 0.8974; social phenotype x treatment x test time, F (1, 43) = 0.207; p = 0.6514] (Fig. 2c).

Fig. 2

OX1R antagonist (SB334867) did not shift the non-social phenotypes of mice. (a) Timeline. (b) Social preference score of social and non-social mice during the first (T1) and the second (T2) conditioned place preference test (i.e., before/after treatment). (c) Total distance traveled by the distinct groups during T1 and T2. Three-way ANOVA, followed by Šídák’s multiple comparisons test, ** p < 0.01, ns = non-significant, n = 11–14. FST = forced swim test, i.p = intraperitoneal

OX1R antagonist (SB334867) attenuated stress levels in non-social mice

To evaluate the effects of OX1R antagonist injections on stress, we assessed the percentage of incorrect transitions of cephalocaudal grooming in social and non-social mice at T2 (in a treatment-free state). In social mice, treatment with SB334867 did not alter the percentage of incorrect transitions of cephalocaudal grooming [two-way-ANOVA, treatment effect, F (1, 43) = 1.447; p = 0.2355, social phenotype effect, F (1, 43) = 2.156; p = 0.1493, treatment x social phenotype, F (1, 43) = 12.12; p = 0.0012; Šídák’s multiple comparisons test, social vehicle vs. antagonist, p > 0.05]. In non-social mice, SB334867 treatment significantly decreased the percentage of incorrect transitions of cephalocaudal grooming compared to vehicle-treated non-social mice (Šídák’s multiple comparisons test, non-social vehicle vs. antagonist, p < 0.01) and antagonist-treated social mice (Šídák’s multiple comparisons test, social antagonist vs. non-social antagonist, p < 0.01) (Fig. 3a).

One day after the second CPP (T2), mice underwent FST preceded by an injection of SB334867 or vehicle. In social mice, the immobility time in the FST was not affected by OX1R antagonist treatment [two-way ANOVA, treatment effect, F (1, 43) = 0.01109; p = 0.9166; social phenotype effect, F (1, 43) = 9.253; p = 0.0040; treatment x social phenotype, F (1, 43) = 1.624; p = 0.2094; Šídák’s multiple comparisons test, social vehicle vs. antagonist, p > 0.05]. In non-social mice, SB334867 treatment significantly reduced the immobility time compared to non-social mice that received vehicle (Šídák’s multiple comparisons test, non-social vehicle vs. antagonist, p < 0.05) (Fig. 3b). Yet, social mice and non-social mice receiving the antagonist treatment showed a comparable immobility time in the FST (Šídák’s multiple comparisons test, social antagonist vs. non-social antagonist, p > 0.05).

Fig. 3

OX1R blockade reduced stress levels only in non-social mice. (a) Percentage of incorrect transitions of cephalocaudal grooming during the second conditioned place preference test (T2). (b) Time spent immobile during the FST. Two-way ANOVA, followed by Šídák’s multiple comparisons test, * p < 0.05, ** p < 0.01, ns = non-significant, n = 11–14

OX1R expression in the BLA, a stress-related area, was higher in the non-social phenotype

In order to explore whether the selective effects of SB334867 treatment on stress-associated measures in non-social mice were due to different OX1R expression between social and non-social mice, OX1R-immunoreactive neurons were quantified in stress-related brain areas (supplementary Fig. 1) after being categorized based on their social preference [Two-tailed unpaired Student’s t-test, social vs. non-social: p < 0.0001, t (11) = 8.218] (Fig. 4a).

Non-social mice showed significantly higher OX1R expression in the BLA [Two-tailed unpaired Student’s t-test, social vs. non-social: t (9) = 3.185, p = 0.0111] but not in the CeA [Two-tailed unpaired Student’s t-test, social vs. non-social: t (10) = 1.365, p = 0.2021], in comparison to social mice. The difference in the BLA was maintained even after the application of Bonferroni correction (p < 0.05). Moreover, in the BNST, the expression of OX1R was higher in non-social mice compared to their counterparts [Two-tailed unpaired Student’s t-test, social vs. non-social: t (9) = 2.387, p = 0.0408]. However, after Bonferroni correction was applied, the result turned out to be non-significant (p = 0.1632). No significant difference in OX1R expression was observed in the PVT between social and non-social mice (Two-tailed unpaired Student’s t-test, social vs. non-social: t (10) = 1.707, p = 0.1185; ns) (Fig. 4b and c).

To further investigate whether any relationship exists between the social phenotype (i.e., preference score - PS) and the expression of OX1R, a correlation analysis was performed. A negative correlation between the PS and the number of OX1R-immunoreactive neurons was observed in the BLA (p = 0.0142, r = -0.7109) and the BNST (p = 0.0209, r = -0.6815). No correlation was found between the social preference score and the expression of OX1R in the CeA (p = 0.2013, r = -0.3970) or PVT (p = 0.5839, r = -0.1762) (Fig. 4d).

Fig. 4

Non-social mice express a significantly higher number of OX1R-immunoreactive neurons in the BLA. (a) Social preference scores, used to categorize the mice according to their social profiles. Student’s t-test, **** p < 0.0001. n = 6–7. (b) Representative image of OX1R-immunoreactive neurons (arrow); blue: DAPI staining (nuclei); red: OX1R. Scale bar: 50 μm (c) OX1R-immunoreactive neurons in social and non-social mice in stress-related areas of the brain, including the BNST, the PVT, the BLA, and the CeA. Student’s t-test, followed by Bonferroni correction. * p < 0.05, n = 4–7. (d) Correlation between social preference score and number of OX1R-immunoreactive neurons in the different stress-related brain regions. * p < 0.05