Decision letter: Prefrontal-amygdalar oscillations related to social behavior in mice

Abstract

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The medial prefrontal cortex and amygdala are involved in the regulation of social behavior and associated with psychiatric diseases but their detailed neurophysiological mechanisms at a network level remain unclear. We recorded local field potentials (LFPs) from the dorsal medial prefrontal cortex (dmPFC) and basolateral amygdala (BLA) while male mice engaged on social behavior. We found that in wild-type mice, both the dmPFC and BLA increased 4–7 Hz oscillation power and decreased 30–60 Hz power when they needed to attend to another target mouse. In mouse models with reduced social interactions, dmPFC 4–7 Hz power further increased especially when they exhibited social avoidance behavior. In contrast, dmPFC and BLA decreased 4–7 Hz power when wild-type mice socially approached a target mouse. Frequency-specific optogenetic manipulations replicating social approach-related LFP patterns restored social interaction behavior in socially deficient mice. These results demonstrate a neurophysiological substrate of the prefrontal cortex and amygdala related to social behavior and provide a unified pathophysiological understanding of neuronal population dynamics underlying social behavioral deficits. Editor's evaluation This manuscript is of broad interest to readers studying the neural basis of sociability, social anxiety, and anxiety-like behaviors. The authors use mouse models to understand how electrical oscillations in two key parts of the brain (prefrontal cortex and amygdala) relate to social behavior. https://doi.org/10.7554/eLife.78428.sa0 Decision letter eLife's review process Introduction The medial prefrontal cortex (mPFC) plays a central role in social behavior (Duncan and Owen, 2000; Wood and Grafman, 2003; Bicks et al., 2015) through functional interactions with the amygdala (Kumar et al., 2014; Adhikari et al., 2015; Bukalo et al., 2015; Tovote et al., 2015), a region that is reciprocally connected with the mPFC (Vertes, 2004; Hoover and Vertes, 2007; Adhikari et al., 2015) and plays central roles in emotional responses such as fear and anxiety. A number of gene expression patterns and intracellular signaling pathways related to social behavior have been identified in the mPFC (Bicks et al., 2015; Schubert et al., 2015; Yan and Rein, 2022). Under pathological conditions, a number of studies have reported alterations in overall mPFC neuronal excitability and disruptions of mPFC-amygdala interactions in humans with psychiatric disorders with social behavior deficits such as autism spectrum disorders (ASD) and depression (Happé et al., 1996; Castelli et al., 2002; Pierce et al., 2004; Greicius et al., 2007; Drysdale et al., 2017) and animal models of these disorders (Brumback et al., 2018; Abe et al., 2019). A fundamental issue is how such molecular and cellular mechanisms are integrated to form organized mPFC and amygdalar neuronal population activity that cooperatively controls social behavior. Neurophysiological signatures representing neuronal population activity are local field potential (LFP) signals, consisting of diverse oscillatory patterns that dynamically vary with attentional, motivational, arousal states, and entrain synchronous rhythmic spikes (Buzsáki, 2006). Recently, LFP oscillations in the PFC have been shown to modulate social behavior. A PFC oscillation at a low gamma-range (20–50 Hz) band mediated by interneurons facilitates social interaction (Liu et al., 2020). Consistently, autism mouse models with social deficits exhibit impairments in PFC interneuronal activity (Han et al., 2012) and gamma oscillations (Cao et al., 2018). These studies suggest a key role of PFC gamma-range signals in the modulation of social behavior. On the other hand, several studies have shown that theta-range (4–12 Hz) LFP signals in the prefrontal-amygdalar circuit are associated with the expression of emotional behavior (Calıskan and Stork, 2019) such as fear retrieval (Seidenbecher et al., 2003; Likhtik et al., 2014; Stujenske et al., 2014; Dejean et al., 2016; Karalis et al., 2016; Ozawa et al., 2020) and anxiety (Adhikari et al., 2010; Likhtik et al., 2014). In addition, mPFC LFP power at a theta frequency (2–7 Hz) band influences oscillatory activity in the amygdala and ventral tegmental area during stress experiences (Hultman et al., 2016) and predicts vulnerability to mental stress in individual animals (Kumar et al., 2014). While these studies imply that social behavior is mediated by oscillatory signals at various frequency bands in the prefrontal-amygdalar circuit, their causal relationship and pathological changes remain fully elusive. Addressing these issues is critical for a unified understanding of neurophysiological mechanisms at a neuronal network level underlying social behavior and its deficits. In this study, we analyzed changes in LFP signals from the dorsal mPFC (dmPFC) and basolateral amygdala (BLA) among wild-type mice and mouse models with social behavioral deficits in a social interaction (SI) test. By extracting detailed animal’s behavioral patterns on a moment-to-moment basis that potentially reflect increased and decreased motivation for social behavior, we discovered prominent changes in dmPFC-BLA LFP signals that specifically varied with social behavior. Optogenetic experiments verified a causal relationship between these oscillatory signals and social behavior, highlighting the importance of frequency-specific manipulations of neuronal activity in the dmPFC-BLA circuit. Results Changes in dmPFC and BLA LFP power in a social interaction test Male C57BL/6 J mice were tested in a conventional SI test in which they freely interacted with an empty cage and the same cage containing a target CD-1 mouse for 150 s, termed a no target and a target session, respectively (Figure 1A). The degree of social interactions for each mouse was quantified as a SI ratio, which refers to the ratio of stay duration in an interaction zone (IZ) in a target session to that in a no target session. Consistent with previous observations (Golden et al., 2011; Venzala et al., 2012; Ramaker and Dulawa, 2017), the majority of wild-type mice (11 out of 14) exhibited SI ratios of more than 1 (Figure 1B), demonstrating their motivation for social interactions. From the mice performing the SI tests, LFP signals were simultaneously recorded from the dmPFC, corresponding to the prelimbic (PL) region, and the BLA using an electrode assembly (Figure 1C and D and Figure 1—figure supplement 1). The locations of individual electrodes were confirmed by a postmortem histological analysis. To compute an overall tendency of LFP power changes, a fourier transformation analysis was applied to LFP signals from each entire session. Absolute LFP power spectrums were variable across individual mice (Figure 1—figure supplement 2A), but their averages over all mice exhibited differences between the two sessions at relatively lower (below 10 Hz) and higher (10 Hz) frequency bands (Figure 1E, top; n=14 mice). To further examine these differences, two LFP spectrums from the two sessions were converted to a spectrum representing the ratios of LFP power at each frequency band in the target session relative to that in the no target session (Figure 1E, bottom). The spectrum revealed an increase and a decrease in dmPFC power at a frequency band of 4–7 Hz and 30–60 Hz, respectively, in a target session compared with a no target session. Overall, dmPFC 4–7 Hz and 30–60 Hz power in the target session was significantly increased to 113.1 ± 4.0% and decreased to 94.6 ± 1.6%, respectively (Figure 1F, n=14 mice, 4–7 Hz: t13=3.56, p=0.0070; 30–60 Hz: t13 = 2.99, p=0.021, paired t-test followed by Bonferroni correction), and BLA 4–7 Hz was significantly increased to 110.6 ± 2.7%, compared with that in the no target session (Figure 1G, n=6 mice, 4–7 Hz: t5=4.95, p=0.0086; 30–60 Hz: t5 = 2.13, p=0.17, paired t-test followed by Bonferroni correction). The same power analyses were applied to the same datasets at the other frequency bands, including 1–4 Hz, 7–10 Hz, 10–30 Hz, and 60–100 Hz bands, but no significant differences were found between the target and no target sessions (Figure 1—figure supplement 3A and B; p>0.05, paired t-test followed by Bonferroni correction at all the frequency bands). These results suggest that dmPFC-BLA 4–7 Hz and 30–60 Hz power specifically become higher and lower, respectively, when mice are exposed to an environment including the other target mouse. Unlike fear-related theta-gamma coupling reported previously (Stujenske et al., 2014), we found no pronounced phase-amplitude coupling between the 4–7 Hz and 30–60 Hz frequency bands in both the dmPFC and BLA LFP traces (a representative result shown in Figure 1—figure supplement 2B). In addition to the LFP power changes, dmPFC-BLA coherence at the 4–7 Hz band in the target session was significantly higher than that in the no target session (Figure 1H, n=6 mice, 4–7 Hz: t5=3.95, p=0.022; 30–60 Hz: t5 = 1.37, p=0.46, paired t-test followed by Bonferroni correction), confirming the coordination of dmPFC-BLA at the 4–7 Hz band. The granger causality spectrum exhibited a significantly higher granger causality index at the 4–7 Hz band for the direction from the dmPFC to BLA than that for the direction from the BLA to the dmPFC (Figure 1I, n=6 mice, 4–7 Hz: p=0.030; 30–60 Hz: p=0.26, Mann-Whitney U test followed by Bonferroni correction), possibly reflecting the preferential projection of the dmPFC to the BLA (Gabbott et al., 2005; Bukalo et al., 2015). Figure 1 with 3 supplements see all Download asset Open asset Changes in dorsal medial prefrontal cortex (dmPFC) and basolateral amygdala (BLA) Local field potential (LFP) signals in a social interaction (SI) test. (A) A SI test with an interaction zone (IZ; labeled in green). Movement trajectories (gray lines) of a wild-type mouse are superimposed. (B) (Left) Occupancy time in the IZ. Each line indicates an individual mouse (n = 14 wild-type mice). (Right) SI ratios computed from the occupancy time. Each dot represents an individual mouse. (C) (Left) LFPs were recorded from the dmPFC and BLA. (Right) Histological confirmation of electrode locations (arrows). The dotted boxes are magnified in the right panels. The green line shows the contour of the BLA. The details of electrode locations are shown in Figure 1—figure supplement 1. (D) Typical LFP signals from the dmPFC and BLA. (E) (Top) Comparison of dmPFC LFP power spectrograms between the target (red) and no target (black) sessions averaged over all mice (n = 14 mice). Original datasets from individual mice are shown in Figure 1—figure supplement 2A. Data are presented as the mean ± SEM. Cyan and magenta bars above represent 4–7 Hz and 30–60 Hz bands, respectively. (Bottom) The percentages of LFP power at individual frequency bands in the target session relative to those in the no target session. The percentages were computed in individual mice and were averaged over all mice. (F) The percentages of dmPFC 4–7 Hz (left) and 30–60 Hz (right) LFP power averaged over an entire period of the target session relative to those of the no target session (n = 14 mice). Data are presented as the mean ± SEM. Each gray dot represents an individual data points. * and # represent a significant increase and decrease in the target session, respectively (p<0.05, paired t-test vs no target). (G) Same as F but for the BLA (n = 6 mice). (H) Same as F but for dmPFC-BLA coherence (n = 6 mice). (I) Spectral granger causality averaged over dmPFC-BLA electrode pairs. (n = 6 mice). *p<0.05, Mann-Whitney U test followed by Bonferroni correction. (J, K) Same as F but when an unfamiliar C57BL/6J mouse was used as a target mouse (J) or a toy mouse was placed in the cage instead of a target mouse (K). Figure 1—source data 1 Individual data for Figure 1. https://cdn.elifesciences.org/articles/78428/elife-78428-fig1-data1-v1.xlsx Download elife-78428-fig1-data1-v1.xlsx As the percentages of the power changes were variable across the wild-type mice (Figure 1F), we examined whether the dmPFC LFP power changes in the target session in individual mice were related to their SI ratios (Figure 1—figure supplement 2C). However, we found no significant correlations between these two variables (n=14 mice, 4–7 Hz: R = –0.27, p=0.34; 30–60 Hz: R=0.40, p=0.15), demonstrating that individual differences in social behavior are not crucially associated with dmPFC power changes at least within the wild-type mouse group. A possible explanation for these power changes between the two sessions may be due to differences in running speed. To test this possibility, we compared moving speed between the two sessions and found that moving speed was significantly higher in the no target session than in the target session (Figure 1—figure supplement 2D; Z=20.20, p=9.0 × 10-89, Mann-Whitney U test). We then compared LFP power changes between running periods with a moving speed of more than 5 cm/s and stop periods with a moving speed of less than 1 cm/s, which occupied 20.2 and 40.6% of entire recording periods, respectively (Figure 1—figure supplement 2E and F). Both in the dmPFC and BLA, 4–7 Hz power during stop periods was significantly higher than that during running periods (dmPFC, Z=4.93, p=8.26 × 10–7; BLA, Z=2.42, p=0.016, Mann-Whitney U test), whereas 30–60 Hz power exhibited opposite changes (dmPFC, Z=2.13, p=0.033, p=0.19; BLA, Z=2.00, p=0.045), suggesting that locomotion is a crucial factor to affect these LFP power changes. We thus applied the same power analysis by specifically extracting stop and running periods. Similar to Figure 1F, significant increases in dmPFC LFP power in the target session were observed during stop periods but not running periods (Figure 1—figure supplement 2H, stop periods: n=14 mice, 4–7 Hz: t13=3.72, p=0.0052; 30–60 Hz: t13 = 3.05, p=0.019; Figure 1—figure supplement 2G, running periods: 4–7 Hz: t13=1.33, p=0.41; 30–60 Hz: t13 = 1.25, p=0.46, paired t-test followed by Bonferroni correction). These results confirm that while 4–7 Hz and 30–60 Hz power in the dmPFC and BLA is higher and lower, respectively, as moving speed is lower, the LFP power changes were still prominent in the entire target session, compared with the no target session, when mice stopped. In the SI test above, we utilized a CD-1 mouse as a target mouse in the cage that was substantially larger than the recorded C57BL/6 J mice. This recording condition may induce anxiety-related or fear-related behavior in the recorded mice. To reduce these emotional factors as possible, we performed a similar SI test using an unfamiliar C57BL/6 J mouse with a similar body size as a target mouse (Figure 1J). Similar to the results when the target CD-1 mice were used (Figure 1F), dmPFC 4–7 Hz power in the target session was significantly higher than that in the no target session (Figure 1J, n=6 mice, 4–7 Hz: t5=5.36, p=0.0060; 30–60 Hz: t5 = 1.04, p=0.70, paired t-test followed by Bonferroni correction). In addition, as a control experiment without social behavior, we performed a similar SI test by placing a plastic toy mouse in the cage as a novel object instead of a real mouse (Figure 1K). In this case, significant changes in dmPFC 4–7 Hz and 30–60 Hz power between the two sessions were not observed (Figure 1K, n=7 mice, 4–7 Hz: t6=0.51, p>0.99; 30–60 Hz: t6 = 0.60, p>0.99, paired t-test followed by Bonferroni correction). These results further confirm that the dmPFC power changes are induced specifically in a condition where mice exhibit social behavior while they are less associated with an object novelty or emotional components such as anxiety. While LFP power changes were observed in the target session with both a CD-1 mouse and a C57BL/6 J mouse as a target mouse (Figure 1F and J), there still remains a possibility that these changes are induced by increased anxiety and/or novelty against social interaction, as mice are generally anxious when encountering a novel (target) mouse. Previous studies have demonstrated that anxiety induces theta-range (4–10 Hz) power increases in the mPFC-BLA-ventral hippocampal circuit (Adhikari et al., 2010; Likhtik et al., 2014; Padilla-Coreano et al., 2019). We thus examined whether anxiogenic conditions could induce the similar changes in the dmPFC-BLA LFP signals observed in this study. The test box was divided into social avoidance zones (corners of the box opposing the target mice), peripheral areas (near the walls of the box), and a center area (Figure 2A). Based on the similarity of the no target session and conventional open field tests, mice are considered to more feel anxiety in the avoidance zones and peripheral areas, compared with the center area, in the no target session. To compare relative changes in LFP power across behavior and sessions, LFP power at each frequency band was z-scored based on the average and SD of LFP power at each frequency band in an entire period including the no target and target sessions. In the no target session, no significant differences in dmPFC and BLA 4–7 Hz and 30–60 Hz LFP power were observed among these areas (Figure 2B, dmPFC: n=7 mice, 4–7 Hz, F2,40 = 1.36, p=0.27; 30–60 Hz; F2,40 = 0.11, p=0.90; Figure 2C, BLA: n=6 mice, 4–7 Hz, F2,16 = 1.12, p=0.35; 30–60 Hz; F2,16 = 0.61, p=0.56, one-way ANOVA). On the other hand, mice are considered to most increase anxiety or most decrease motivation for social behavior in the avoidance zones (Golden et al., 2011) and more increase anxiety levels in the peripheral areas, compared with the center area, in the target session. In the target session, no significant differences in dmPFC and BLA 4–7 Hz and 30–60 Hz LFP power were observed among these areas (Figure 2B, dmPFC: n=14 mice, 4–7 Hz, F2,35 = 0.50, p=0.61; 30–60 Hz; F2,35 = 0.13, p=0.88; Figure 2C, BLA: n=6 mice, 4–7 Hz, F1,10 = 3.38, p=0.099; 30–60 Hz; F1,10 = 0.27, p=0.62, one-way ANOVA). These results suggest that anxiety-related environments are not crucially associated with 4–7 Hz and 30–60 Hz LFP power changes in the dmPFC and BLA observed in this study. Figure 2 with 1 supplement see all Download asset Open asset No pronounced changes in 4–7 Hz and 30–60 Hz power in the dorsal medial prefrontal cortex (dmPFC) and basolateral amygdala (BLA) in areas outside the interaction zone (IZ). (A) Schematic illustration showing social avoidance zones, peripheral areas, and a center area in the SI test. (B) Comparisons of dmPFC 4–7 Hz and 30–60 Hz power across avoidance zones, peripheral areas, and a center area (n = 14 mice). Data are presented as the mean ± SEM. Each line represents each mouse. p>0.05, Mann-Whitney U test followed by Bonferroni correction. (C) Same as B but for the BLA (n = 6 mice). The center area was removed from this analysis because of the limited number of samples. Figure 2—source data 1 Individual data for Figure 2. https://cdn.elifesciences.org/articles/78428/elife-78428-fig2-data1-v1.xlsx Download elife-78428-fig2-data1-v1.xlsx Increases in dmPFC 4–7 Hz power during social avoidance in socially deficient mouse models We next examined whether these LFP signals are altered in mice with reduced social interaction. The same tests were performed on Shank3 null mutant mice, termed Shank3 knockout (KO) mice, which have been reported to exhibit repetitive grooming behavior and social interaction deficits, mimicking symptoms associated with ASD (Peça et al., 2011; Mei et al., 2016). The SI ratios in 7 Shank3 KO mice were significantly lower than those in the 14 wild-type mice (Figure 3A, Z=2.65, p=0.016, Mann-Whitney U test followed by Bonferroni correction). We recorded LFP signals from the dmPFC and BLA of these Shank3 KO mice and found significant increases in dmPFC and BLA LFP power during the target session at the 4–7 Hz bands, similar to the wild-type mice (Figure 3B, dmPFC: n=7 mice, 4–7 Hz, t6=6.04, p=1.8 × 10–3; 30–60 Hz; t6=2.62, p=0.078; Figure 3C, BLA: n=7 mice, 4–7 Hz, t6=3.03, p=0.048; 30–60 Hz; t6=3.52, p=0.026, paired t-test followed by Bonferroni correction). Moreover, the dmPFC 4–7 Hz increases during the target session in the Shank3 KO mice were significantly larger than those observed in the wild-type mice (Figure 3B, F2,24=4.21, p=0.027, one-way ANOVA across wild-type, Shank3 KO, and defeated mouse groups; Z=2.36, p=0.046, Mann-Whitney U test followed by Bonferroni correction), whereas no differences were observed for the changes in dmPFC 30–60 Hz power (F2,24=1.55, p=0.23, one-way ANOVA; Z=1.23, p=0.44, Mann-Whitney U test followed by Bonferroni correction) and BLA power (Figure 3C, 4–7 Hz, F2,15=1.07, p=0.36, one-way ANOVA; p=0.90; 30–60 Hz, F2,15=1.01, p=0.39, one-way ANOVA; P>0.99, Mann-Whitney U test followed by Bonferroni correction). These results suggest that the increases in dmPFC 4–7 Hz power during a target session are more prominent in Shank3 KO mice, compared with wild-type mice. Figure 3 Download asset Open asset Further increases in dorsal medial prefrontal cortex (dmPFC) 4–7 Hz power in the target session in Shank3 knockout (KO) mice and defeated mice. (A) (Left) Movement trajectory of a Shank3 KO mouse and a defeated mouse in a target session. The orange areas represent the avoidance zones. (Right) SI ratios for Shank3 KO and defeated mice (n = 14 wild, 7 Shank3 KO, and 10 defeated mice). Each dot represents an individual animal. The data from wild-type mice similar to those shown in Figure 1B are presented for comparison. *p<0.05 versus wild, Mann-Whitney U test followed by Bonferroni correction. (B) The percentages of dmPFC 4–7 Hz (left) and 30–60 Hz (right) local field potential (LFP) power in the target session relative to those in the no target session (n = 14 wild, 7 Shank3 KO, and 6 defeated mice). Data are presented as the mean ± SEM. Each gray dot represents an individual data points. The data from wild-type mice similar to those shown in Figure 1F are presented for comparison. *p<0.05, versus wild, Mann-Whitney U test followed by Bonferroni correction. (C) Same as B but for the basolateral amygdala (BLA) (n = 6, 7, and 5 mice). Figure 3—source data 1 Individual data for Figure 3. https://cdn.elifesciences.org/articles/78428/elife-78428-fig3-data1-v1.xlsx Download elife-78428-fig3-data1-v1.xlsx During the target session, the Shank3 KO mice spent substantial (26.0 ± 6.8%) time in the avoidance zones (Figure 3A and Figure 2—figure supplement 1A). When the Shank3 KO mice stayed within the avoidance zone, dmPFC 4–7 Hz was significantly increased, compared with the other areas (Figure 4B, dmPFC: n=7 mice, 4–7 Hz, t5=4.39, p=0.014; 30–60 Hz; t5=0.33, p>0.99; BLA: n=7 mice, 4–7 Hz, t5=1.62, p=0.32; 30–60 Hz; t5=0.17, p>0.99, paired t-test followed by Bonferroni correction). Such significant changes were not observed in the wild-type mice (Figure 4A, dmPFC: n=13 mice that stayed in the avoidance zones, 4–7 Hz, t12=0.068, p>0.99; 30–60 Hz; t12=0.71, p=0.98; BLA: n=5 mice that stayed in the avoidance zones, 4–7 Hz, t4=0.49, p>0.99; 30–60 Hz; t4=0.27, p>0.99, paired t-test followed by Bonferroni correction). These results demonstrate that Shank3 KO mice specifically exhibit a dmPFC 4–7 Hz power increase during avoidance behavior. In the Shank3 KO mice, dmPFC-BLA coherence and the directionality between the dmPFC and BLA at the 4–7 Hz band were not prominent during social avoidance behavior (Figure 2—figure supplement 1B, n=6 mice, 4–7 Hz: t5=0.46, p>0.99; 30–60 Hz: t5 = 0.98, p=0.73, paired t-test followed by Bonferroni correction; Figure 2—figure supplement 1C, n=6 mice, 4–7 Hz: p=0.63; 30–60 Hz: p>0.99, Mann-Whitney U test followed by Bonferroni correction). Figure 4 Download asset Open asset Increases in dorsal medial prefrontal cortex (dmPFC) 4–7 Hz power during social avoidance in Shank3 knockout (KO) mice and defeated mice. (A) Comparisons of 4–7 Hz and 30–60 Hz power in the dmPFC and basolateral amygdala (BLA) between the avoidance zones and the other areas in the target session in wild-type mice (n = 14 and 6 mice). Data are presented as the mean ± SEM. Each gray line represents each mouse. p>0.05, paired t-test followed by Bonferroni correction. (B) Same as A but for Shank3 KO mice (n = 7 and 7 mice). *p<0.05, paired t-test followed by Bonferroni correction. (C) Same as A but for defeated mice (n = 6 and 5 mice). Figure 4—source data 1 Individual data for Figure 4. https://cdn.elifesciences.org/articles/78428/elife-78428-fig4-data1-v1.xlsx Download elife-78428-fig4-data1-v1.xlsx We next tested whether socially defeated mice with reduced social interaction exhibit similar LFP changes. Wild-type mice were exposed to social defeat stress for 10 consecutive days, termed defeated mice. SI ratios of the 6 defeated mice were significantly lower than those in the 14 wild-type mice (Figure 3A; Z=4.10, p=8.1 × 10–5, Mann-Whitney U test followed by Bonferroni correction). Similar to the wild-type and Shank3 KO mice, these defeated mice exhibited a significantly larger increase in dmPFC 4–7 Hz power during the target session than during the no target session (Figure 3B, dmPFC: n=6 mice, 4–7 Hz, t5=6.93, p=9.6 × 10–4; 30–60 Hz; t5=3.77, p=0.013; BLA: n=5 mice, 4–7 Hz: t4=1.88, p=0.14; 30–60 Hz: t4=0.33, p=0.76). In addition, similar to the Shank3 KO mice, the dmPFC 4–7 Hz increase in the defeated mice was significantly larger than that observed from the wild-type mice (Figure 3B, dmPFC: 4–7 Hz: Z=2.27, p=0.046; 30–60 Hz: Z=1.36, p=0.34; Figure 3C, BLA: 4–7 Hz: p>0.99; 30–60 Hz: p=0.50, Mann-Whitney U test followed by Bonferroni correction). Similar significant results were observed in the comparison between the defeated mice and defeated control mice (that were pair housed in the same cage with aggressor mice but not subject to physical contact) (Figure 2—figure supplement 1E and Figure 4F, 4–7 Hz: p=0.0087; 30–60 Hz: p=0.13, Mann-Whitney U test followed by Bonferroni correction). Furthermore, defeated mice spent 54.7 ± 9.5% of an entire recording time in the avoidance zones (Figure 3A and Figure 2—figure supplement 1A) and exhibited significant increases in dmPFC 4–7 Hz power in the avoidance zone (Figure 4C, dmPFC: n=6 mice, 4–7 Hz, t5=11.68, p=1.8 × 10–4; 30–60 Hz; t5=3.45, p=0.036; BLA: n=5 mice, 4–7 Hz, t4=0.47, p>0.99; 30–60 Hz; t4=0.46, p>0.99, paired t-test followed by Bonferroni correction). These results demonstrate that a dmPFC 4–7 Hz power increase during social avoidance also occurs in depression model mice, similar to Shank3 KO mice. Taken together, our results from the two mouse models suggest that dmPFC 4–7 Hz power increases during social avoidance behavior are a common hallmark across socially deficient mouse models. Changes in dmPFC LFP power during social approach behavior The social avoidance-related increases in dmPFC 4–7 Hz power implied that social interaction behavior may be associated with dmPFC 4–7 Hz power changes. To test this possibility, we first compared LFP power between when the wild-type normal mice stayed within and were outside the IZ as conventional measures in an SI test. However, no significant changes in 4–7 Hz and 30–60 Hz power in the dmPFC and BLA during the target session were detected between the IZ and the other areas (Figure 5—figure supplement 1; dmPFC, n=14 mice; 4–7 Hz: t13 = 0.76, p=0.46; 30–60 Hz: t13 = 1.38, p=0.19; BLA, n=6 mice; 4–7 Hz: t5 = 0.76, p=0.48; 30–60 Hz: t5 = 0.02, p=0.98, paired t-test followed by Bonferroni correction). While this analysis focused on the entire period during which the mice stayed in the IZ, their behavioral patterns within the IZ were not consistent across time; mice actively approach or interact with a target mouse in some periods, reflecting high motivation, whereas they occasionally turn around, move away from a target mouse, or continue to stay at a location in an IZ, possibly reflecting no strong motivation for social interaction. These behavioral observations indicate that animals’ motivation toward and salience regarding the other mouse are not equivalent even when they are similarly located in an IZ. The results suggest that changes in dmPFC-BLA oscillations are not simply explained by where the mice stayed in the SI test. We further analyzed how LFP patterns are associated with their instantaneous behavior every 1 s. We defined social approach behavior, potentially representing increased motivation for social interaction, as the time during which the mice approached the cage (within the half of the box containing the cage) with their cage-oriented moving directions θ less than 90° in the target session (Figure 5A–C). Assuming that mice potentially exhibited the highest motivation and salience during an initial bout of a social approach, this definition was restricted to the initial 5 s periods of social approach behavior. As a control behavior against approach behavior, we defined leaving behavior as the time during which the mice left from the cage with their cage-oriented moving directions θ more than 90° in the target session. No significant differences in the distributions of moving speed were found between the approach behavior and leaving behavior (Figure 5D; Z=0.85, p=0.39, Mann-Whitney U test). These results confirm that approach and leaving behavior is not explained by speed or locomotion itself, allowing us to compare LFP power between the two behavioral periods without being affected by moving speed. Both in the dmPFC and BLA, 4–7 Hz power was significantly decreased during the approach behavior compared wi