Mitophagy in the basolateral amygdala mediates increased anxiety induced by aversive social experience

SUMMARY

Psychosocial stress is a common risk factor for anxiety disorders. The cellular mechanism for the anxiogenic effect of psychosocial stress is largely unclear. Here, we show that chronic social defeat (CSD) stress in mice causes mitochondrial impairment which triggers the PINK1-Parkin mitophagy pathway selectively in the amygdala. This mitophagy elevation causes excessive mitochondrial elimination and consequent mitochondrial deficiency. Mitochondrial deficiency in the basolateral amygdalae (BLA) causes weakening of synaptic transmission in the BLA-BNST (bed nucleus of the stria terminalis) anxiolytic pathway and increased anxiety. CSD-induced increase in anxiety-like behaviors is abolished in PINK1^−/− and Parkin⁻⁻ mice and alleviated by optogenetic activation of the BLA–BNST synapse. This study identifies an unsuspected role of mitophagy in psychogenetic stress-induced anxiety elevation, and reveals that mitochondrial deficiency is sufficient to increase anxiety and underlies psychosocial stress-induced anxiety increase. Mitochondria and mitophagy, therefore, can be potentially targeted to ameliorate anxiety.

Graphical Abstract

eTOC blurb

Duan et al. reveal a key role of mitophagy in anxiety induced by chronic psychogenic stress. Mitophagy is enhanced by stress to cause mitochondrial loss, leading to synaptic weakening in the anxiolytic pathway mediated by the extended amygdala. Mitochondria and mitophagy, therefore, may be targeted to ameliorate anxiety.

INTRODUCTION

Anxiety is sustained fear for potential future and anticipated threats. Occasional anxiety is a normal emotional state adaptive to environments. However, persistent, disruptive, or overwhelming anxiety like in anxiety disorders substantially affects life and work. At least one-third of individuals with anxiety disorders do not achieve sustained remission by using the current first-line medications (Pollack et al., 2008). Anxiety can be increased by chronic stress. A notable type of chronic stress in humans is chronic psychogenic stress, which is social in nature and caused, for example, by social defeat and abuse (Gross and Hen, 2004; Kessler, 1997; Lupien et al., 2009; Schmidt et al., 2008). The cellular mechanism by which chronic psychosocial stress influences anxiety is largely unclear.

Chronic stress has a significant impact on brain mitochondria in animals. For example, chronic mild stress in rodents causes changes in mitochondrial morphology and respiratory activity in the brain and depression-like behaviors (Gong et al., 2011; Madrigal et al., 2001; Rezin et al., 2008), suggesting a potential link between brain mitochondrial abnormalities and behavioral dysregulation. This notion is supported by evidence from both humans and animals. Human studies have detected altered mitochondrial morphology, function, and protein expression in the brains of individuals with psychiatric disorders (Manji et al., 2012; Marazziti et al., 2011; Mattson et al., 2008; Shao et al., 2008). The brains of high-anxious rodents display changes in the mitochondrial oxidative phosphorylation system, mitochondrial protein expression, and production of reactive oxygen species (Filiou and Sandi, 2019; Filiou et al., 2011; Hollis et al., 2015; Rammal et al., 2008). Deletion of BCL-2, a mitochondrial protein, increases anxiety-like behaviors in mice (Einat et al., 2005). mtDNA variations affect anxiety-like behavior in mice (Gimsa et al., 2009). Despite these connections between mitochondria, stress, and behavior, it has yet to be determined if the mitochondrial disturbance is a cause or consequence of behavioral alterations associated with stress, in particular psychosocial stress.

Mitochondria are essential for neural physiology. The subcellular distribution, biomass, and fission of mitochondria in dendrites determine synapse number (Li et al., 2004). Axonal mitochondria regulate short-term synaptic plasticity and neurotransmitter release (Guo et al., 2005; Kang et al., 2008; Tang and Zucker, 1997). Neurons, therefore, need to maintain proper mitochondrial quantity to support the structural and functional integrity of synapses. Mitochondrial quantity is governed by two opposing processes, mitochondrial biogenesis and mitophagy. PINK1 (PTEN-induced putative kinase protein 1) and Parkin (an E3 ubiquitin ligase) are two key proteins involved in mitophagy (Mishra and Chan, 2014; Youle and Narendra, 2011). Mutations in PINK1 and Parkin are associated with Parkinson’s disease (Hernandez et al., 2016; Lang and Lozano, 1998a, b; Pickrell and Youle, 2015). The role of mitophagy in psychiatric disorders is unexplored.

In this study, we investigate the role of mitochondria in CSD-induced behavioral alterations. We show that CSD causes mitochondrial impairments which lead to activation of the PINK1-Parkin mitophagy pathway in the amygdala. The consequent excessive elimination of mitochondria causes weakening of the BLA–adBNST synapses, thereby increasing anxiety-like behaviors. This study uncovers a previously unsuspected role of mitophagy in the anxiety increase associated with stress.

RESULTS

Mitochondria in the amygdala are altered by CSD

To test for the effect of psychosocial stress on mitochondria, we applied the CSD paradigm to C57BL/6 mice (8–9 weeks of age) (Golden et al., 2011). The behavior of defeated mice was assessed after 10 or 30 days of CSD exposure, both of which have been used in previous studies (Bondar et al., 2018; Bowens et al., 2012; Jianhua et al., 2017; Tallerova et al., 2014; Whalen et al., 2008) (Figure S1A). Based on previous studies (Krishnan et al., 2007), we divided defeated mice into two subgroups by social interaction score (SI, the ratio of time exploring social vs. non-social targets): susceptible group with SI ≤ 1 and resilient group with SI >1 (Figures S1E, S1H, S1L, and S1O). For both susceptible and resilient mice, 10 and 30 days of CSD had no effect on locomotion and increased anxiety-like behaviors as indicated by less time spent in the light compartment of the light/dark box, open arms of the elevated plus maze, and center of the open field (Figures S1B-S1D, S1G, S1I-S1K, and S1N). However, only 30-day CSD increased behavioral despair in the forced swim test, and this effect was limited to susceptible mice (Figures S1F and S1M). These findings are consistent with earlier reports (Bondar et al., 2018; Krishnan et al., 2007; Tallerova et al., 2014).

To examine mitochondria, we exposed transgenic mice expressing mitochondria-targeted YFP (mito-YFP) in neurons to CSD. Since the amygdala is a hub of encoding and processing fearful and threatening experiences, and the ventral hippocampus is involved in stress response and emotional reactions (Fanselow and Dong, 2010; Janak and Tye, 2015; Maddox et al., 2019), we analyzed mitochondria in these areas as CSD is a fearful, threatening, and stressful experience. Mito-YFP colocalized with the mitochondrial protein Tom20 (Figure S2A). We first analyzed mitochondria in mice defeated for 30 days as they had more behavioral alterations. CSD reduced mitochondrial size and total mitochondrial mass (indicated by mito-YFP intensity) in the basolateral amygdala and central nucleus of the amygdala (BLA and CeA; Figures 1A-1D, S2B, and S2C). These alterations were comparable in susceptible and resilient mice (Figures S2D-S2H). In the ventral hippocampus of defeated mice, mitochondria were smaller in the dentate gyrus (DG) but not in CA1 or CA3, while the total mitochondrial mass was unchanged in all three subregions (Figures 1E-1H and S2I-2K).

Figure 1. The size and mass of mitochondria decrease in the amygdala of defeated mice.

Mito-YFP mice were subjected to 30 days of CSD or untreated. (A, B, E, F) Representative images of mito-YFP in brain sections; scale bar: 150 μm in A; 500 μm in D; in B and E, 15 μm for images in the left two columns and 10 μm for images in the right column. (C, G) Quantification of average mitochondrial area in designated brain regions. (D, H) Quantification of integrated mito-YFP intensity in designated brain regions. The numbers below the bars indicate the number of brain sections from 3 control and 6 defeated mice. Data are presented as mean ± SEM. ** p < 0.01; *** p < 0.001. See also Figures S1 and S2, Table S1.

Next, we examined mitochondrial membrane potential (MMP), which drives ATP production (Mitchell, 1961), by using tetramethylrhodamine ethyl ester (TMRE) which accumulates in mitochondria proportionally to MMP. TMRE positive structures colocalized with mito-YFP (Figure S2L). TMRE uptake was reduced in the BLA and CeA of both susceptible and resilient mice (Figures 2A, 2B, and S2M-S2P). By contrast, TMRE uptake in CA1, CA3, and DG of the ventral hippocampus was unchanged in defeated mice (Figures 2A, 2B, and S2M). Hence, 30-day CSD causes MMP depolarization in the BLA and CeA but not in the ventral hippocampus.

Figure 2. MMP, COX activity, and the p-AMPK level are altered in the amygdala of defeated mice.

C57BL/6 mice were subjected to 30 days of CSD or untreated. (A) Representative images of acute brain slices stained with TMRE; 10x (for the amygdala) or 4x (for the hippocampus) objective was used for image acquisition; scale bar, 25 μm (amygdala) and 100 μm (hippocampus). (B) Quantification of the integrated TMRE intensity in designated brain areas; the number in the bar indicates the number of brain slices taken from 3 control and 6 defeated mice. (C) COX activity in total lysates; Amyg: amygdala; Hippo: hippocampus. (D) COX activity in the mitochondrial fraction. (E) Representative immunoblots. (F) Quantification for the ratio of pAMPK to AMPK. (G) Quantification for the ratio of pAMPK to GAPDH. The number in the bar indicates the number of animals in C, D, F, G. Data are presented as mean ± SEM. * p < 0.05, ** p < 0.01. See also Figures S3 and S4, Table S1.

MMP is generated when electrons move along the mitochondrial electron transport chain (METC) during oxidative phosphorylation (Lamond, 2002). To assess METC, we measured the activity of cytochrome c oxidase (COX, the last enzymatic complex of METC). Since the MMP alteration was comparable in susceptible and resilient mice, we pooled them in this analysis. COX activity was decreased in the amygdala but unchanged in the ventral hippocampus of defeated mice (Figures 2C and 2D), suggestive of METC impairments in the amygdala.

MMP and METC are coupled to ATP production. To assess cellular ATP, we analyzed phosphorylated AMP-activated protein kinase (p-AMPK), which inversely correlates with ATP levels (Hardie et al., 2012). p-AMPK increased in the amygdala but not in the ventral hippocampus of defeated mice (Figures 2E-2G). We also used an ATP sensor [iATPSnFR1.0 (Lobas et al., 2019)] to assess cellular ATP. The fluorescence intensity of iATPSnFR1.0 expressed by injected iATPSnFR1.0 AAV decreased in the BLA but not in the DG or CA3 of defeated mice (Figure S3), indicating a reduction of cellular ATP in the BLA.

10-day CSD did not affect mitochondrial size, total mitochondrial mass, or MMP in the amygdala of either susceptible or resilient mice (Figures S4A-S4F). Cellular ATP was unchanged by 10-day-CSD exposure (Figures S4G-S4I). Moreover, we assessed body weight, corticosterone, and cytokines. The mean body weight of defeated mice was lower (Figure S4J). Corticosterone, IL-1β, and IL-6 in the blood of defeated mice greatly increased at 2 hours after the 10-min direct interaction period on CSD day one and after 10 days of CSD, but returned to pre-CSD levels after 20 or 30 days of CSD (Figures S4K-S4M).

Taken together, these findings indicate that 30-day CSD alters multiple aspects of mitochondria in the amygdala.

The replication and mutation of mtDNA increase in the amygdala of defeated mice

The mitochondrial loss in mice exposed to 30-day CSD suggests a possible imbalance between mitochondrial biogenesis and mitochondrial elimination. Generation of new mitochondria involves the replication of mitochondrial DNA (mtDNA). To assess mtDNA replication, we used 5-ethynyl-2’-deoxyuridine (EdU) to label replicating mtDNA. The incorporation of EdU into replicating mitochondria was confirmed in cultured hippocampal neurons (Figure S5). We injected mice with EdU. EdU positive mitochondria increased in the BLA but not DG of defeated mice (Figures 3A-3D).

Figure 3. The effect of CSD on mtDNA.

C57BL/6 mice were subjected to 30 days of CSD or untreated. (A) Representative images of double labeling for EdU and Tom20 at the BLA in brain sections from mice injected with EdU; the yellow lines indicate the location of the orthogonal views along the XZ and YZ axis; scale bar, 5 μm for the left three images and 2 μm for the “XY” image. (B) Quantification for the proportion of Tom20⁺ structures containing EdU in the BLA. (C) Representative images of double labeling for EdU and Tom20 at the dentate gyrus in brain sections from mice injected with EdU; scale bar, 5 μm for the left three images and 2 μm for the “XY” image. (D) Quantification for the proportion of Tom20⁺ structures containing EdU in the dentate gyrus. (E) The ratio of mtDNA to nuclear DNA. (F, G) The average mtDNA mutation frequency per mtDNA site. (H) mRNA levels of PGC1α and TFAM in control and defeated mice. (I) Representative immunoblots of PGC1α and TFAM using lysates from control and defeated mice. (J) Quantification for PGC1α protein expression. (K) Quantification for TFAM protein expression. Amyg: amygdala; Hippo: hippocampus. The number in and below the bar indicates animal numbers in E-G, H, J, K, and the number of brain sections from 3 mice for each group in B and D. Data are presented as mean ± SEM. * p < 0.05; ** p < 0.01. See also Figure S5, Table S1.

Having found that CSD increases mtDNA replication in the amygdala, we analyzed mtDNA copy number (relative to nuclear DNA). 30-day CSD reduced the ratio of mitochondrially encoded cytochrome B gene to the nuclear-encoded actin gene in the amygdala (Figure 3E), indicating a reduction of mtDNA. This decrease was not likely to be offset by the mtDNA replication increase since it was much greater (49.01 ± 4.38%) than the fraction of mitochondria labeled by EdU (2.68 ± 0.56%).

mtDNA replication has a higher error rate than nuclear DNA and introduces transition mutations (A ↔ G and C ↔ T) (Johnson and Johnson, 2001; Longley et al., 2001; Spelbrink et al., 2000; Zheng et al., 2006). Another type of mtDNA mutation is transversion which is the substitution of A or G for C or T and vice versa. The G:C to T:A transversion is caused by the pairing of adenine with 8-hydroxyguanine which is an abundant base modification generated by reactive oxygen species, therefore a hallmark of oxidative mtDNA damage (Cheng et al., 1992; Dizdaroglu et al., 2002). We used mitoRCA-seq [an mtDNA sequencing method (Ni et al., 2015)] to analyze mtDNA mutation. In the amygdala of defeated mice, the frequency of all and transition mutations increased, while the transversion mutation frequency was unchanged (Figures 3F and 3G), suggesting that mtDNA replication may contribute to increased mtDNA mutation. We also analyzed transcriptional factors for mitochondrial genes including PGC-1α (peroxisome proliferator activated receptor gamma coactivator-1-alpha) and TFAM (transcription factor A). Both the protein and mRNA levels of PGC-1α and TFAM in the amygdala were unchanged by CSD (Figures 3H-3K).

Taken together, these findings indicate that 30-day CSD causes increases in mtDNA replication and mtDNA mutation and a decrease in mtDNA copy number in the amygdala.

CSD increases mitophagy in the amygdala

Next, we examined mitophagy, a cellular process eliminating mitochondria. We examined mitophagosomes by co-staining for Tom20 with the autophagosome marker LC3 (microtubule associated protein 1 light chain 3) or optineurin [an autophagy receptor for damaged mitochondria (Tanida and Waguri, 2010; Weidberg et al., 2011; Wong and Holzbaur, 2014)]. We first tested if LC3 staining can report autophagosomes in conditional Atg5 knockout mice obtained by crossing Atg5^flox/flox mice with CaMKIIα-Cre mice. Consistent with the reported fewer autophagosomes in Atg5 knockout mice (Kuma et al., 2004), LC3 puncta significantly decreased in knockout mice (Figures S6A-S6C).

In the BLA of defeated mice, the colocalization of Tom20 with LC3 or optineurin increased (Figures 4A-4D). We also examined mitochondria in lysosomes using KeimaRed (Katayama et al., 2011). We generated mitochondria-targeted KeimaRed (mtKeima) which localized on mitochondria (Figure S6D). We injected mtKeima AAV into the BLA before CSD and analyzed KeimaRed after CSD. The ratio of mtKeima in lysosomes (excited at 543 nm) to that in mitochondria (excited at 458 nm) increased in defeated mice (Figures 4E and 4F).

Figure 4. Mitophagy is increased in the amygdala of defeated mice.

Brain sections or brain lysates prepared from the amygdala of C57BL/6 mice subjected to 30 days of social defeat or untreated were used for immunostaining (A–D), EM (G–M), or immunoblotting (N–Q). In E and F, mice were injected with mtKeima AAV, subjected to 30 days of social defeat or untreated, then used for the preparation of acute brain slices and imaging. (A, C) Representative images of staining for Tom20 and LC3; the yellow lines indicate the location of the orthogonal views along the XZ and YZ axis; scale bar: 5 μm for the left 3 images, 2 μm for the “XY” images. (B, D) Quantification for the proportion of Tom20 positive structures containing LC3 or optineurin. (E) Representative images of mtKeima excited at 543 nm (red) and 458 nm (blue); scale bar, 10 μm. (F) Quantification for the ratio of mtKeima fluorescence excited at 543 nm to that excited at 458 nm. (G) Representative electron micrographs of mitophagosome-like structures; red arrows and red dash lines indicate mitophagosome-like structures; m: mitochondria; scale bar, 200 nm. (H) Quantification for the number of mitophagosome-like structures. (I) Representative electron micrographs of abnormal mitochondria indicated by red arrows; scale bar, 500 nm. (J) Quantification for mitochondrial number. (K) Quantification for the number of abnormal mitochondria. (L) Representative electron micrographs of autophagosome-like structures; arrows indicate autophagosome-like structures; scale bar, 500 nm. (M) Quantification for the number of autophagosome-like structures. (N) Representative immunoblots of LC3-II and SQSTM1/p62. (O) Quantification for the ratio of LC3-II to LC3-I. (P) Quantification for the ratio of LC3-II to tubulin. (Q) Quantification for the ratio of SQSTM1/p62 to GAPDH. The numbers in and below the bar and above the x-axis indicate the number of brain sections from 3-4 mice in B, D and F; the number of micrographs from 3 control and 6 defeated mice in H, J, K, M; animal numbers in O–Q. Data are presented as mean ± SEM in B, D, F, O–Q or violin plots (H, J, K, M) illustrating kernel probability density, i.e. the width of the outlined area represents the proportion of the data located there. * p < 0.05, ** p < 0.01, *** p < 0.001. See also Figure S6, Table S1.

Moreover, we assessed mitophagy by electron microscopy (EM). There were more mitophagosome-like structures, fewer mitochondria, and more mitochondria with partial loss of contents (abnormal mitochondria) in the BLA of defeated mice (Figures 4G-4K). We did not detect mitophagosome-like structures in glia within the BLA of either control or defeated mice. The number of glia and glial mitochondria was unchanged in the BLA of defeated mice (Figures S6E-S6G). 10-day CSD did not change the colocalization of Tom20 with LC3 or optineurin or mtDNA number (Figures S6H-S6K). General macroautophagy assessed from autophagosome-like structures, LC3, and SQSTM1/p62 was unchanged (Figures 4L-4Q).

Taken together, these findings indicate that mitophagy is elevated in the BLA following 30 days of CSD.

PINK1 and Parkin mediate CSD-induced mitophagy and mitochondrial loss

PINK1 and Parkin mediate a well-studied mitophagy pathway. To test for their roles in CSD-induced mitophagy, we applied 30-day CSD to PINK1 knockout (PINK^−/−) and Parkin knockout (Parkin^−/−) mice. The PINK1 and Parkin genes were mutated and their encoded proteins were deleted in knockout mice (Figures S7A-S7D). Unlike in WT mice, CSD had no effect on the colocalization of Tom 20 with LC3 or optineurin and increased total mitochondrial mass in the BLA of PINK1^−/− and Parkin^−/− mice (Figures 5A-5D). CSD-induced decrease in total TMRE uptake in the BLA was also blocked in PINK1^−/− and Parkin^−/− mice (Figures S7E and S7F). However, TMRE uptake by individual mitochondria was reduced in the BLA of defeated PINK1^−/− and Parkin^−/−mice (Figures S7E and S7G). PINK1 and Parkin increased in the BLA of defeated mice (Figures S7H-S7J). These findings support that damaged mitochondria were eliminated through PINK1 and Parkin-mediated mitophagy.

Figure 5. PINK1 and Parkin are required for CSD-induced mitophagy increase and increase in anxiety-like behaviors.

Fixed brain slices were prepared from WT, PINK1^−/−, and Parkin^−/− mice subjected to 30 days of CSD or untreated. (A) Representative images of staining for Tom20, LC3, and optineurin; the yellow lines indicate the location of the orthogonal views along the XZ and YZ axis; scale bar, 5 μm for images in the left 3 columns, and 2 μm for the “XY” images. (B) Quantification for the proportion of Tom20 positive structures containing LC3. (C) Quantification for the proportion of Tom20 positive structures containing optineurin. (D) Quantification for integrated Tom20 intensity. The number in and below the bar indicates the number of brain sections from 3 animals in B–D. (E) Schematic diagram of the experimental procedure. Defeated and control mice were divided into two groups for different testing schedules. (F–J) Quantification of the open field, light/dark box, social interaction, and forced swim tests after CSD in the same group of mice. (K and L) Quantification of the elevated plus maze and social interaction tests in the same group of mice. The number below the bar indicates the number of animals combined from multiple cohorts (For mice tested for open field, light/dark box and forced swim tests: WT, 4; PINK1 KO, 5; Parkin KO; 5. For mice tested in elevated plus maze: WT, 2; PINK1 KO, 3; Parkin KO, 3). Data are presented as mean ± SEM; * p < 0.05, ** p < 0.01; *** p < 0.001. See also Figures S7-S11, Table S1.

CSD-induced amygdala mtDNA decrease was blocked in PINK1^−/− and Parkin^−/− mice (Figure S8A), indicating that both PINK1 and Parkin are required for CSD-induced mtDNA reduction. MtDNA copy number, mtDNA transition mutation, and EdU incorporation into mitochondria were reduced in Parkin^−/− mice without CSD exposure, and CSD-induced increases in these parameters were blocked in Parkin^−/− mice (Figures S8B-S8E). These results suggest that Parkin is involved in both basal and CSD-induced mtDNA replication and mutation. This is consistent with previous reports that Parkin enhances mtDNA replication (Kuroda et al., 2006; Rothfuss et al., 2009). By contrast, mtDNA mutation and mitochondrial incorporation of EdU were comparable in WT and PINK1^−/− mice regardless of CSD exposure (Figures S8B-S8E), indicating that PINK1 is not involved in mtDNA replication. Hence, while both Parkin and PINK1 are required for CSD-induced mtDNA reduction, only Parkin contributes to CSD-induced alteration of mtDNA replication.

Taken together, these results show that PINK1 and Parkin-mediated mitophagy is involved in CSD-induced mitochondrial loss.

PINK1 and Parkin contribute to increased anxiety-like behaviors in defeated mice

To test if PINK1 and Parkin are involved in the behavioral sequelae of CSD, we analyzed the behavior of PINK1^−/− and Parkin^−/− mice exposed to CSD. While defeated PINK1^−/− and Parkin^−/− mice increased behavioral despair, their anxiety-like behaviors were unchanged (Figures 5E-5H, 5J, and 5K), indicating that PINK1 and Parkin contribute to CSD-induced increase in anxiety-like behaviors, but not in behavioral despair. CSD increased social avoidance in PINK1^−/− mice but not in Parkin^−/− mice (Figures 5I and 5L). The total cell number and the number of CaMKIIα⁺ and parvalbumin (PV)⁺ neurons which are abundant in the BLA were unchanged in defeated mice (Figure S9). To avoid potential effects on development by gene knockout, we knocked down PINK1 and Parkin with siRNAs. PINK1 and Parkin siRNAs greatly reduced the expression of their respective targets without changing Tom20 in the BLA (Figures S10A-S10J). PINK1 and Parkin siRNAs had no effect on TMRE uptake in transfected cells (Figures S10K and S10L).

We analyzed TMRE uptake in acute amygdala slices from mice injected with the siRNA virus. TMRE intensity in regions of BLA containing transduced cells (GFP⁺) was compared to those without viral transduction (GFP⁻). While total TMRE intensity in GFP⁻ regions was lower in defeated mice, it was unchanged in GFP⁺ regions (Figures S10M and S10N). However, TMRE intensity in individual mitochondria of defeated mice was reduced regardless of viral transduction (Figures S10M and S10O). Hence, PINK1 and Parkin siRNAs block CSD-induced decrease in total TMRE uptake but not MMP depolarization. PINK1 and Parkin siRNAs also blocked CSD-induced mtDNA loss (Figure S10P). PINK1 and Parkin siRNAs blocked the effect of CSD on anxiety-like behaviors but not on behavioral despair (Figure S11).

To test if mitochondrial deficiency in the amygdala is anxiogenic, we injected the BLA with AAV expressing mitochondria-targeted KillerRed [mtKillerRed, a fluorescent protein producing reactive oxygen species upon illumination to damage mitochondria without causing cell death (Williams et al., 2013)] and control AAV expressing mtKeima. mtKillerRed localized on mitochondria (Figure S12A). mtKillerRed⁺ and mtKeima⁺ neurons were readily detectable in the BLA (Figures S12B and S12C). The BLA was stimulated with light pulses (561 nm, 40 Hz, 10 ms pulse duration, 64 mW/mm² at the tip of the fiberoptic) for 50 min under isoflurane anesthesia or sham illuminated. While photostimulation increased anxiety-like behaviors in mtKillerRed AAV-injected mice, it had no effect on mtKeima AAV-injected mice (Figures 6A-6D). EM showed that illumination of mtKillerRed increased mitophagosome-like structures and abnormal mitochondria, reduced mitochondrial number, and didn't affect autophagosomes (Figures 6E-6K). Illumination didn't change the density of DAPI⁺, CaMKIIα⁺, or PV⁺ cells (Figure S12D-S12G). Hence, mitochondrial deficiency at the BLA can increase anxiety-like behaviors.

Figure 6. Illumination of mtKillerRed increases anxiety-like behaviors and mitophagy.

C57BL/6 mice were injected with mtKillerRed AAV or mtKeima AAV in the BLA, then tested for anxiety-like behavior. Animals were used for EM after behavioral testing. (A) Schematic diagram of the experimental procedure. (B) Representative maps of the time that an animal spent in the light box; color indicates total time spent at each location. (C) Quantification for the total time that animals spent in the light compartment of the light/dark box; the same animals are connected by lines. (D) Quantification for the total time that animals spent in the open arm of the elevated plus maze; the same animals are connected by lines. (E) Representative electron micrographs of mitochondria in photo-stimulated and sham-stimulated mice (m, mitochondria; arrows indicate abnormal mitochondria); scale bar, 500 nm. (F) Quantification for the number of abnormal mitochondria. (G) Representative electron micrographs of mitophagosome-like structures (indicated by the red arrow); scale bar, 500 nm for the left image and 100 nm for the enlarged image. (H) Quantification of the number of mitophagosome-like structures. (I) Quantification of the number of mitochondria. (J) Representative electron micrographs of autophagosome-like structures (indicated by the red arrow); scale bar, 500 nm for the left image and 100 nm for the enlarged image. (K) Quantification for the number of autophagosome-like structures. The numbers in and below the bar and above the x-axis indicate the animal number in C and D and the number of micrographs from 3 mice in F, H, I, K. Data are presented as mean ± SEM in C and D, or violin plots (F, H, I, K) illustrating kernel probability density, i.e. the width of the outlined area represents the proportion of the data located there. * p < 0.05, ** p < 0.01. See also Figure S12, Table S1.

Taken together, these findings indicate that the anxiogenic effect of CSD requires the PINK1-Parkin mitophagy pathway.

The BLA-BNST synaptic transmission is reduced in defeated mice

Given the importance of mitochondria for synaptic function (Hollenbeck, 2005; Sheng and Cai, 2012), we examined the effect of CSD on synaptic physiology. The ratio of excitatory to inhibitory synaptic currents (E/I), the frequency and amplitude of spontaneous excitatory and inhibitory postsynaptic currents (sEPSCs and sIPSCs), rheobase, firing rate, action potential half-width, action potential threshold, and resting membrane potentials were unchanged in BLA neurons of defeated mice (Figures S13A-S13J). Hence, CSD didn't affect the excitability, E/I balance, and postsynaptic response of BLA neurons.

We next examined synaptic transmission in the BLA–anterodorsal BNST (adBNST) pathway, which is anxiolytic (Kim et al., 2013), by stimulating it using optogenetics. We injected the BLA of mice with AAV expressing channelrhodopsin 2 (E123A) and EYFP fusion proteins (ChR2-EYFP) at 4 weeks before CSD, and prepared brain slices after CSD for electrophysiology. ChR2-expressing neurons were readily detected in the BLA (Figure S13K). Illumination with 473-nm light pulses induced action potentials and depolarizing photocurrents (Figures S13L and S13M).

Light-evoked EPSCs were recorded in adBNST neurons and slices were fixed for ChR2-EYFP imaging after electrophysiology. ChR2-EYFP expression was used to normalize EPSCs across slices. The input and output relationship (I/O) curve was shifted downwards in defeated mice (Figure 7A), indicating a weakening of synaptic transmission. To determine if the weakening occurs at pre- or postsynaptic sites, we analyzed paired-pulse ratio (PPR), the ratio of AMPA to NMDA receptor-mediated currents (EPSC_AMPA/EPSC_NMDA), and intrinsic biophysical properties of adBNST neurons.

Figure 7. BLA-adBNST synaptic transmission in defeated mice was reduced in a PINK1-dependent manner and optogenetic activation of the BLA-BNST projection abolishes the effect of CSD on anxiety-like behaviors.

The BLA was injected with ChR2 AAV (A–F) alone or along with lentivirus expressing siRNAs or scramble nucleotides (G–J) and subjected to 30 days of CSD or untreated. BNST slices were prepared after CSD, and EPSCs evoked by 473-nm light pulses were recorded in adBNST neurons using whole-cell patch-clamp. C57BL/6 mice (A, B, G–J), WT littermates of PINK1^−/− mice (C, D), and PINK1^−/− mice (E, F) were used. (A, C, E, G, I) EPSCs evoked by light pulses at various intensities. (B, D, F, H, J) Paired-pulse ratio analyzed by stimulating the BLA to BNST projections with pairs of light pulses (50 ms interpulse interval) at various intensities. N represents the number of neurons from 5 mice in A–J. (K–P) The BLA of C57BL/6 mice was injected with ChR2 AAV, tested for baseline anxiety-like behaviors, then subjected to 30 days of CSD or untreated. After CSD, mice were assessed for anxiety-like behaviors again in the absence or presence of 473-nm light pulses. (K) Schematic diagram of the experimental procedure. (L) Representative image of ChR2 expression at the BNST; the white bar indicates the implanted optic fiber. (M) The average time spent in the central arena during the 5-min block in the baseline test, and the time spent in the central arena during the 5-min block without or with light stimulation in the post-CSD test. (N) The average time spent in the open arms during the 5-min block in the baseline test, and the time spent in the open arms during the 5-min block without or with light stimulation in the post-CSD test. (O) Total distance traveled during the 5-min block in the baseline test, and total distance traveled during the 5-min block without or with light stimulation in the post-CSD test. (P) Total distance traveled in the elevated plus maze during the 5-min block in the baseline test, and total distance traveled in the elevated plus maze during the 5-min block without or with light stimulation in the post-CSD test. The number below the bar indicates the number of animals in K–P. The same animal in different test sessions is connected with lines in M–P. Data are presented as mean ± SEM; ** p < 0.01, *** p < 0.001.See also Figures S13-S15, Table S1.

As previously reported, light stimulation of the BLA input induced membrane depolarization and evoked synaptic currents in most adBNST neurons (Gungor et al., 2015; Kim et al., 2013; Lu et al., 2017) (Figures S14A-S14C). The firing rate, rheobase, action potential properties, resting membrane potential, and EPSC_AMPA/EPSC_NMDA were unchanged by CSD (Figures S14D-S14K). By contrast, PPR increased (Figure 7B), indicating that presynaptic release is reduced. Hence, CSD alters pre-, but not post-, synaptic properties of BLA–adBNST synapses.

Since PINK1 is specifically required for the effect of CSD on anxiety but not that on social avoidance or behavioral despair, we tested if PINK1 contributes to the synaptic effect of CSD. The BLA was injected with ChR2 AAV alone or along with PINK1 siRNA virus or scrambled oligonucleotide virus. Injected mice were exposed to CSD 6 weeks later, then used for the preparation of BNST slices. Light-evoked EPSCs were recorded in adBNST neurons. CSD induced a downward shift of the I/O curve and increased PPR in WT mice and mice injected with the scramble oligonucleotide virus (Figures 7C, 7D, 7G, and 7H). These effects were abolished in PINK1^−/− and PINK1 siRNA virus-injected mice (Figures 7E, 7F, 7I, and 7J). Moreover, we examined mitochondria in BLA axons projecting to the BNST by injecting the BLA with AAV expressing mCherry and mitochondria-targeted GFP (mtGFP). The area and density of mitochondria in BNST-projecting BLA axons were reduced (Figure S15).

To test if weakening of the BLA–adBNST synapse contributes to the anxiogenic effect of CSD, we stimulated the BLA to adBNST projection in defeated mice using optogenetics. WT mice were injected with ChR2 AAV at the BLA and implanted with optic fibers at the adBNST (Figure 7K). The BLA projections from cells transduced with ChR2-EYFP AAV were readily detected in the adBNST (Figure 7L). Without photostimulation, defeated mice displayed increased anxiety-like behaviors (Figures 7M-7P). Photostimulation reduced their anxiety-like behaviors to the pre-CSD baseline without affecting locomotion (Figures 7M-7P). These results indicate that weakening of the BLA–adBNST synapse contributes to increased anxiety-like behaviors in defeated mice.

Taken together, these findings indicate that CSD reduces the BLA–adBNST synaptic transmission, thereby increasing anxiety-like behaviors.

The mitophagy increase in defeated mice requires neural activation

As the amygdala is involved in emotion processing and social behavior (Phelps and LeDoux, 2005), mitophagy may be triggered by intense neural activities in the amygdala during CSD. We used chemogenetic inhibition of the amygdala to test this possibility. We injected the BLA with AAV expressing hM4Di (Gi-DREADD, a designer receptor exclusively activated by designer drugs) driven by the αCaMKII promotor to inhibit excitatory neurons (Figure 8A and 8B). Injected mice were exposed to CSD with or without treatment with clozapine-N-oxide (CNO, a hM4Di activator). CNO inhibited neural activation indicated by c-fos expression and had no effect on TMRE update (Figure S16). LC3-positive mitochondria increased in the BLA of defeated mice without CNO treatment (Figures 8C and 8D). CNO treatment blocked this increase as well as CSD-induced social avoidance and anxiety-like behaviors (Figures 8C-8I). Hence, BLA neural activity is required for CSD to enhance mitophagy and anxiety-like behaviors.

Figure 8. Chemogenetic inhibition of the amygdala blocks CSD-induced increases in mitophagy and anxiety-like behaviors.

C57BL/6 mice were injected with Gi-mCherry AAV in the BLA and subjected to 30 days of CSD or untreated, then used for immunostaining or behavioral tests. (A) Schematic diagram of the experimental procedure. (B) Representative image of Gi-mCherry in the BLA; scale bar: 100 μm. (C) Representative images of double staining for LC3 and Tom20 in the BLA; the yellow lines indicate the location of the orthogonal views along the XZ and YZ axis; scale bar, 5 μm for images in the left 3 columns, and 2 μm for the “XY” images. (D) Quantification for the proportion of Tom20 positive structures that contain LC3. (E) Time spent in the central arena of the open field box. (F) Total distance traveled in the open field box. (G) Time spent in the light compartment of the light/dark box. (H) Quantification for the social interaction test. (I) Time spent in the open arms of the elevated plus maze. The number in and below the bar indicates the number of brain sections from 4 mice for each group in D and animal number in E–I. Data are presented as mean ± SEM. * p < 0.05, ** p < 0.01. See also Figure S16, Table S1.

DISCUSSION

Anxiety disorders are the most common psychiatric disorders with a lifetime prevalence of ~28% (Kessler et al., 2005). A large proportion of people with anxiety disorders do not respond to antidepressants which are the current first-line medications. Comprehension of the cellular and molecular mechanisms underlying anxiety is crucial for the development of more efficacious pharmacotherapies. In this study, we uncover the role of mitophagy in anxiety associated with prolonged psychosocial stress.

Mitophagy is enhanced in the amygdala of mice exposed to 30-day CSD. This increase is likely due to mitochondrial loss and mitochondrial impairments such as MMP depolarization. It is possible that the prolonged and/or intensive neural activity in the amygdala during CSD causes recurrent mitochondrial impairment, therefore chronic mitophagy activation. This notion is supported by MMP depolarization in defeated PINK1^−/− and Parkin^−/− mice. Chronic mitophagy, in turn, causes excessive mitochondrial elimination. Given the importance of mitochondria for synapses, it is not surprising that CSD causes impairments of synaptic transmission between the BLA and adBNST. Hence, despite the beneficial effects of basal mitophagy, excessive mitophagy can cause a mitochondrial shortage and behavioral disturbances. Mitophagy, therefore, needs to be kept at appropriate levels for optimal synaptic and behavioral performance.

We found that CSD causes weakening of the BLA–adBNST pathway which is anxiolytic (Kim et al., 2013). This is due, at least partly, to decreased release of synaptic vesicles. This synaptic weakening requires PINK1 and contributes to increased anxiety. Unlike anxiety-like behaviors, CSD increases behavioral despair without involving PINK1 or Parkin, likely because behavioral despair is mainly modulated by the medial prefrontal cortex and hippocampus (Wang et al., 2015; Warden et al., 2012).

In the BLA of defeated PINK1^−/− and Parkin^−/− mice, although individual mitochondria are depolarized, this impairment appears to have been compensated by increased mitochondrial mass due to mitophagy inhibition, so that total TMRE uptake is unchanged. Hence, mitophagy inhibition in defeated PINK1^−/− and Parkin^−/− mice helps to preserve the total functional capacity of mitochondria, thereby antagonizing the adverse effect of mitochondrial impairments on behavior.

CSD also increases the replication and mutation of mtDNA in the amygdala. However, the mean mutation rate of individual mtDNA sites in defeated mice is still very low, and mutations altering gene expression or function are predicted to be rare. It is unlikely that mtDNA mutation contributes to the behavioral effect of CSD as defeated PINK1^−/− mice do not exhibit increased anxiety-like behaviors even they carry more mtDNA mutations. The mtDNA replication increase in defeated mice is via a Parkin-mediated, PINK1-independent mechanism. This is consistent with previous reports that Parkin promotes mtDNA replication (Kuroda et al., 2006; Rothfuss et al., 2009). In defeated mice, because the mtDNA replication rate is too low to compensate for mitochondrial elimination by mitophagy, mtDNA is reduced. Mitophagy, therefore, is the predominant cellular process determining mitochondrial amount in defeated mice.

CSD largely spares hippocampal mitochondria. This is likely because defeated mice experience threats eliciting fear and anxiety during agonistic interactions (Davis et al., 2010). As a hub of fear encoding and expression, the amygdala is activated during CSD. The resulting neural activity can put amygdala mitochondria at a greater risk than those in the hippocampus. This notion is consistent with our findings that chemogenetic inhibition of the amygdala mitigates the effects of CSD on mitophagy and anxiety.

In sum, our study demonstrates that mitophagy is an important cellular process that mediates increased anxiety-like behaviors impinged by chronic psychosocial stress and suggests that mitochondria and mitophagy can be potentially targeted to develop therapies to alleviate anxiety.

STAR☆METHODS

RESOURCE AVAILABILITY

LEAD CONTACT

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Zheng Li (lizheng2@mail.nih.gov).

MATERIALS AVAILABILITY

Plasmids generated in this study will be available upon request and MTA.

DATA AND CODE AVAILABILITY

• Original western blot images and microscopy data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

Male C57BL/6J and transgenic mice (7–24 weeks of age) were used. Male CD1 mice (16–24 weeks of age) were used as aggressors in chronic social defeat (CSD) treatment. Both male and female, 19-day-old Sprague Dawley rat embryos were used for neural cultures. All animal procedures followed the US National Institutes of Health Guidelines Using Animals in Intramural Research and were approved by the National Institute of Mental Health Animal Care and Use Committee. Mice were kept under a 12-hour light (9 pm–9 am)/dark (9 am–9 pm) cycle and had access to water and food ad libitum. Animals were randomly assigned to experimental groups.

Primary hippocampal neurons

The hippocampus of E19 rat embryos (both male and female littermates) was removed and dissociated using trypsin digestion. Primary hippocampal neurons were seeded on coverslips coated with poly-D-lysine (30 μg/ml) and laminin (5 μg/ml) at a density of 330 cells/mm², and grown in Neurobasal media supplemented with 2% B27, 1% penicillin-streptomycin and 1% glutamax (Thermo Fisher Scientific). Neurons were transfected with Lipofectamine 2000 (Thermo Fisher Scientific) following the manufacturer’s instructions.

METHOD DETAILS

Plasmid construction

The PINK1 siRNA, Parkin siRNA and scrambled siRNA constructs were generated by inserting annealed oligonucleotides (PINK1 siRNA: CTGTGTATGAAGCCACCAT; Parkin siRNA: GTCCCAACTCCCTGATTAA; Scramble siRNA: CTATCGCTAGTACTAGCGA) into the BglII/HindIII site of the pSuper vector (OligoEngine). The PINK1, Parkin and scramble siRNAs along with the promoter region in the pSuper vector were amplified by PCR and cloned to the ClaI site of the pRRLsin lentiviral vector. The mtKillerRed and mtKeima constructs were generated by inserting PCR amplified KillerRed (from pCS2-NXE+mem-KillerRed, Addgene #45761) and Keima (from mKeima-Red-N1, Addgene #54597) into the BglII/EcoRI site of the pGW1 vector containing the mitochondria-targeting sequence of Cox8 (ATGTCCGTCCTGACGCCGCTGCTGCTGCGGGGCTTGACAGGCTCGGCCCGGCGGCT CCC AGTGCCGCGCGCCAAGATCCATTCGTTG). The mtKillerRed and mtKeima sequences in the pGW1 vector were PCR amplified and ligated into the AgeI/EcoRI site of the pAAV vector (Addgene #26975).

Surgery

7-8-week-old mice were anesthetized by intraperitoneal injection of Ketamine/Xylazine (Ketamine: 80–120 mg/kg; Xylazine: 8–10 mg/kg) and kept under 2% isoflurane throughout surgery. Craniotomy was made, and 0.8-1 μl virus was injected into the basolateral amygdala bilaterally (BLA; AP: −1.4 mm, ML: ±3.4 mm, DV: −5.1 mm), ventral CA3 ipsilaterally (AP: −2.9; ML: +3.0; DV: −4.0), and dentate gyrus ipsilaterally (AP: −2.9; ML: −2.5; DV: −2.0) using a 5 μl gas-tight Hamilton Syringe (26 gauge, flat-tip needle) mounted on a stereotaxic device at a rate of 0.1 μl/min. After injection, the needle was left at the injection site for 5 min before withdrawal. For optogenetic experiments, the fiber optic was inserted into bilateral BLA (AP: +1.4 mm, ML: ±3.4 mm, DV: −5.0 mm) or unilateral adBNST (AP: +0.2 mm, ML: ±1.0 mm, DV: −3.9 mm).

Chronic social defeat

CD1 mice (16–24 weeks of age) were individually housed in a cage (35 cm x 32 cm x 17 cm) for one week. To screen for aggressive CD1 mice, a C57BL/6 male mouse (7–8 weeks of age) was placed in the home cage of a CD1 mouse for 5 min daily for three consecutive days. A new C57BL/6 mouse was used each day. Only the CD-1 mice initiating ≥2 attacks during all three testing sections were used as aggressors. During the CSD treatment, an 8-13-week-old male mouse and a CD1 mouse were housed on each side of a cage separated by a transparent, perforated divider placed in the middle of the cage. The divider was removed for 10 min every day to allow the mice to interact. If the CD1 mouse initiated ≥10 attacks during the first 5 min, the interaction session was terminated at 5 min to prevent serious injury. The defeated mouse was co-housed with a fresh CD1 mouse every day. Two control mice were housed in one cage on each side of the divider.

Behavioral tests

All behavioral tests were performed from 10 am to 6 pm. The mice were transported to the behavioral room at 1 hr before testing. The behavioral apparatus was cleaned with 70% ethanol between animals. Analysis of behavioral data was done blind to the treatment.

Open field test.

The test room was illuminated with 20 lux light. The mice were placed at the center of the test box (49 cm x 49 cm x 40 cm) at the beginning of the test and allowed to freely explore the box for 30 min. The central area of the box which accounts for 25% of the box was defined as the central arena. The behavioral test was video recorded and analyzed using automated behavioral tracking software (TopScan/ObjectScan; CleverSystems).

Light/dark box test.

The test room was illuminated with 20 lux light. Mice were placed in the light compartment of a light/dark box (46 cm x 27 cm x 30 cm; one-third of the box was dark and two-thirds were transparent) at the beginning of the test and allowed to freely explore the test box for 11 min. The behavioral test was video recorded and analyzed using automated behavioral tracking software (TopScan/ObjectScan; CleverSystems).

Social interaction test.

The test room was illuminated with 20 lux light. At the beginning of the test, mice were placed in the middle of a box (49 cm x 49 cm x 40 cm) containing two upside-down wire mesh pencil cups (10 cm in diameter), one empty and the other one enclosing the CD1 mouse that had defeated the experimental mouse. Mice were allowed to freely explore the box for 30 min. The behavioral test was video recorded and analyzed using automated behavioral tracking software (TopScan/ObjectScan; CleverSystems). The social interaction score was defined as the ratio of time that the experimental mouse spent within 5 cm of the cup containing the CD1 mouse to that spent within 5 cm of the empty cup.

Forced swim test.

The test room was illuminated with regular room light. A glass cylinder (20 cm in diameter) was filled with water (25°C) up to 30 cm in height. The mice were placed on the water and allowed to swim for 6 min. The behavioral test was video recorded and analyzed using automated behavioral tracking software (Forced Swim Scan; CleverSystems). Immobility during the last 5 min of testing was analyzed.

Elevated plus maze.

The test room was illuminated with room light (~200 lux). The maze was placed in the middle of the room with each arm pointing to the corner of the room. At the beginning of the test, a mouse was placed at the junction of the four arms and faced an open arm. The mouse was left in the maze for 5 min and video recorded. The duration and walking distance in each arm were analyzed using an automated behavioral tracking software: TopScan/ObjectScan from CleverSystems for testing without optogenetic stimulation and ANY-maze (Stoelting; IL, USA) for testing with optogenetic stimulation.

In vivo optogenetic stimulation

Mice were handled daily for 1 week before optogenetic stimulation. On the test day, the implanted fiber optic ferrule was connected to patch cords (0.22 NA, 200 μm diameter; Thorlabs). Mice were allowed to recover for 5 min in the home cage after connection. Light pulses were generated with a laser (CrystaLaser; NV, USA) controlled by OPTG4 software (Doric Lenses; QC, Canada). For optogenetic stimulation of mtKillerRed, 561-nm light pulses at 40 Hz (10-ms pulse duration, 64 mW/mm² at the tip of the fiberoptic) were bilaterally delivered to the BLA through two patch cords. For optogenetic stimulation of ChR2, 473-nm light pulses at 10 Hz (5-ms pulse duration, 159 mW/mm² at the tip of the fiber optic) were delivered to unilateral adBNST. The unilateral stimulation was balanced between the left and the right hemisphere. Optogenetic stimulation during the elevated plus maze test was initiated by ANY-maze software when the mouse reached the terminal of the closed arm.

CNO treatment

Clozapine N-oxide dihydrochloride (CNO, HelloBio, HB6149) was added to drinking water (12.5 mg/L final concentration) along with saccharine (0.1 g/L). The water bottle was wrapped with aluminum foil to avoid light and replaced every 3 days. Control mice were fed CNO-free water.

Acute brain slices

Mice were anesthetized with isoflurane and decapitated. For imaging experiments, brains were quickly removed and placed in ice-cold modified artificial CSF (ACSF) solution containing (in mM) 92 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 D-glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄ and 0.5 CaCl₂ (adjusted to pH 7.3 −7.4 with 10N NaOH, bubbled with 95% O₂/ 5% CO₂ for 1 hr). Coronal brain slices (320 μm) were cut using a vibratome (Leica VT 1000S). For whole-cell recording, coronal brain slices (350 μm) were cut using ice-cold NMDG-HEPES solution containing (in mM): 93 NMDG (N-Methyl-D-glucamine diatrizoate), 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 D-glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO₄ and 0.5 CaCl₂ (adjusted to pH 7.3 −7.4 with 10N HCl, bubbled with 95% O₂/ 5% CO₂). After cutting, slices were incubated in NMDG-HEPES solution (32°C) for 12-15 min, and then transferred to standard ACSF containing (in mM): 124 NaCl, 2.5 KCl, 1.2 NaH₂PO₄, 24 NaHCO₃, 5 HEPES, 12.5 D-glucose, 2 MgSO₄ and 2 CaCl₂ (adjusted to pH 7.3 −7.4, bubbled with 95% O₂/ 5% CO₂) at room temperature.

Isolation of mitochondria

Mice were anesthetized with isoflurane and decapitated. The amygdala and hippocampus were removed, immediately homogenized using a Dounce homogenizer (20 strokes) on ice in MB buffer comprised of (in mM): 210 mannitol, 70 sucrose, 1 EDTA, 10 HEPES, pH 7.5, 1 PMSF (Roche, 10837091001), leupeptin (1 μg/ml; Roche, 11017128001), and aprotinin (1 –g/ml; Roche, 10236624001). The homogenate was centrifuged at 2655 g for 10 min at 4°C. The supernatant was centrifuged using a TLA-120.2 rotor (Beckman Coulter) at 55000 rpm for 1 hr at 4°C. The pellet containing mitochondria was resuspended in cold MB buffer.

COX activity assay

The Cytochrome c Oxidase Assay kit (Sigma, CYTOCOX1) was used to measure COX activity following the manufacturer’s instruction. A spectrophotometer (Molecular Devices) was used to detect light absorption by cytochrome c at 550 nm. 1–2 μg protein was used for analysis and each sample was measured twice.

Assays for corticosterone, IL-1β, and IL-6

Blood was collected into EDTA coated tubes from the submandibular vein, then centrifuged at 1500g, 4°C for 10 min to separate blood cells and plasma. Plasma was used for the measurement of corticosterone, IL-1β, and IL-6 by ELISA according to manufacturer's instructions by using the Corticosterone Parameter Assay Kit, Mouse IL-6 Quantikine ELISA Kit, and Mouse IL-1 beta/IL-1F2 Quantikine ELISA Kit (R&D Systems).

mtDNA sequencing and data analysis

Genomic DNA was isolated from 5–10 mg brain tissues by using the DNeasy^® Blood & Tissue Kit (QIAGEN). The mtDNA sequencing library was prepared with the mitoRCA-seq procedure with a few modifications. Briefly, 50 ng genomic DNA was treated with Next Microbiome DNA Enrichment Kit (NEB) followed by rolling circle amplification (REPLI-g Mitochondrial DNA Kit, QIAGEN) to enrich mitochondrial DNA. The amplification product was used to construct sequencing libraries following the Nextera DNA library preparation procedure (Illumina). Sequencing libraries were pooled and sequenced with the Illumina Hiseq-2500 platform at the NHLBI DNA Sequencing and Genomics Core.

For data analysis, after de-multiplexing, raw sequencing reads were quality filtered and aligned to the reference mitochondrial genome (UCSC, mm10) by BWA using default parameters (http://bio-bwa.sourceforge.net). The uniquely mapped reads were used for downstream heteroplasmic mutation analysis. To minimize potential sequencing errors, the mapping quality score at the mutation sites was set to be greater than 30 (Q30), and the resulting base counts were used to compute mutation frequency. In addition, each mutation site was required to be supported by multiple independent reads to remove PCR amplification artifacts. To this end, a list of best-unique reads (BUR) was compiled, for which only one read of the best average quality score was kept among duplicated reads. The putative point mutation site was identified by a stringent cutoff of ≥10 BURs, which was adjusted based on average BUR coverage depth. The mtDNA total mutation load was calculated by summing the Q30 mutation frequency of all mutation sites with frequency of 0.25–15% derived from the Q30 method. The average mutation frequency per site is computed by dividing the total mutation load by the total base pair number of the mouse mtDNA.

Analysis of mtDNA copy number

The amygdala and ventral hippocampus were dissected out and immediately frozen on dry ice. DNA was extracted from 5–10 mg brain tissues using the DNeasy^® Blood & Tissue Kit (QIAGEN, 69506). The mitochondrial cytochrome b gene (cytob) and the nuclear β-actin gene were measured using quantitative PCR (qPCR) with the SYBR^® Green PCR Master Mix (Applied Biosystems, 4309155) and a real-time PCR machine (7900 HT, Applied Biosystems). The following primers were used for qPCR: cytob-F, cytob-R, β-actin-F, β-actin-R. The relative mtDNA copy number was derived from the following formula: 2^{(CTactin-CTcytob)}.

RNA isolation and quantitative reverse transcription PCR

Total RNA was extracted from brain tissues using Trizol (Ambion; 15596018) according to the manufacturer’s instructions. RNA was reversely transcribed into cDNA using a TransScript First-Strand cDNA Synthesis Kit (Invitrogen; 11752). Real-time PCR was performed using SYBR^® Green PCR Master Mix (Applied Biosystems, 4309155) on a real-time PCR instrument (7900 HT, Applied Biosystems). The following primers were used in PCR: PGC1α-F, PGC1α-R, TFAM-F, TFAM-R, TBP-F, TBP-R. The expression of PGC1α and TFAM was normalized to that of TATA-box binding protein (TBP) by using formula 2^{(CTTBP-CTPGC1α or TFAM)}.

Electron microscopy

Mice were anesthetized with isoflurane followed by intraperitoneal injection of ketamine and xylazine. Mice were perfused with 0.1 M phosphate buffer (pH 7.4), followed by 4% paraformaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4, pre-warmed to 37°C). Brains were removed and placed in fixative at 4°C overnight. The brain was washed in PBS 3 times at 4°C (10 min each), and then cut into 125-μm thick sections with a vibratome. Brain sections were washed with 0.1 M cacodylate buffer (pH 7.4) 3 times (5 min each) and stored at 4 °C overnight. The brain sections were incubated with 1% osmium tetroxide in 0.1 M cacodylate buffer for 30 min at room temperature (in the dark). After rinsing with cacodylate buffer 3 times (5 min each), sections were dehydrated with ethanol in the following order: 3×5 min in 50% ethanol, 1×10 min in 50% ethanol containing 1% uranyl acetate, 2×5 min in 75% ethanol, 1×10 min in 95% ethanol, 3×10 min in 100% ethanol. Residual ethanol was removed by rinsing with propylene oxide (PO) twice. The sections were embedded in Epon:PO (1:1) for 1 hr at room temperature, and then embedded in pure Epon overnight. The sections were transferred to fresh Epon for 1 day, and then baked in the oven for 2 days at 64°C. The basolateral amygdala area in the Epon block was removed and cut into thin sections (60 nm) using a Leica UC7 Ultramicrotome (Vienna). The thin sections were stained with lead citrate. A JEOL JEM-2100 transmission electron microscope (Peabody, MA) coupled with a Gatan digital camera (Pleasanton, CA) was used to take micrographs of randomly selected regions by an experimenter blind to the treatment of the animal.

Subcellular structures with electron-dense contents and bound by a double limiting membrane were identified as autophagosome-like structures. Of these structures, those containing discernible mitochondria, as indicated by the presence of double membranes and cristae, were identified as putative mitophagosome-like structures.

Electrophysiology

Brain slices were perfused with standard ACSF at a rate of 3 ml/min. Whole-cell patch-clamp recordings were performed at 30°C using Multiclamp 700B and Digidata 1550A (Molecular Devices). For recording of BNST neuron’s intrinsic properties and neuronal firing induced by light stimulation, recording pipettes (5–7 MΩ) were filled with the K-based intracellular solution (in mM: 130 K-gluconate, 10 KCl, 2 MgCl₂, 10 HEPES, 2 Mg-ATP, 0.2 Tris-GTP, adjusted to pH 7.2 with KOH, 280 mOsm). For other recordings, Cs-based intercellular buffer (in mM: 130 cesium methanesulfonate, 8 NaCl, 4 Mg-ATP, 0.3 Na-GTP, 0.5 EGTA, 10 HEPES, and 5 QX-314 at pH 7.3) was used. Series resistance (Rs) was monitored during recordings. Only neurons with a series resistance of < 30 MΩ and with a < 10% drift in Rs during the recording period were included in data analysis. For photostimulation, 473-nm light pulses generated with a laser (Crystalaser) were delivered to the slice through an optical fiber (200 μm in diameter). The laser was controlled by a pulse stimulator (Master-8, AMPI). Offline data analysis was performed using the Clampfit 9 software (Molecular Devices).

Image acquisition

For imaging TMRE, mtKeima, and ATP sensor, acute brain slices were incubated with modified ACSF (for mtKeima imaging) or modified ACSF containing 100 nM TMRE (for TMRE imaging) after cutting for 30 min at 33°C, and then recovered in standard ACSF (for mtKeima imaging) or standard ACSF containing 100 nM TMRE (for TMRE imaging) for 1 hr at room temperature before imaging. Brain slices were transferred to a chamber mounted on the sample stage of an Olympus Fluoview 1000 confocal microscope and imaged with a 4x (NA 0.133), 10x (NA 0.3) or 60x (NA 1.0) objective. Gain was set at 1.5 and PMT was set at 780. The thickness of optical sections for z-stack imaging was 10 μm for the 4x objective, 5 μm for the 10x objective, and 1 μm for the 60x objective. Slices were perfused throughout the imaging period with standard ACSF. For fixed brain sections and cultured neurons, z-stack images with an optical section thickness of 0.46 μm were acquired using a 63x objective (NA 1.4) and a Zeiss LSM 800 confocal microscope. For imaging EdU and Tom20 in brain sections, a 488-nm laser was used to scan the EdU signals (PMT: 730, gain: 1.0) and a 640-nm laser was used to scan the Tom20 signals (PMT: 720, gain: 1.0). For imaging LC3 and Tom20 in brain sections, a 488-nm laser was used to scan the EdU signals (PMT: 745, gain: 1.0) and a 640-nm laser was used to scan the Tom20 signals (PMT: 745, gain: 1.0). For imaging mito-YFP in brain sections, a 488-nm laser was used to scan the YFP signals (PMT: 720, gain: 1.0).

Image analysis

For the analysis of mitochondrial morphology, z-stack images (10 μm thickness) were collapsed using the “Max Intensity Z Projection” function and thresholded using the “Moments Thresholding” method in Image J. The area and integrated intensity of mitochondria ≥ 0.8 μm² were measured using the “Analyze Particles” function in ImageJ. For analysis of total TMRE fluorescence intensity, z-stack images were collapsed using the “Max Intensity Z Projection” function and measured in ImageJ. For analysis of TMRE intensity in individual mitochondria, 5 consecutive optical sections with the brightest signals from each brain slice were thresholded using the “Moments Thresholding” method and measured with the “Analyze Particles” function. The TMRE intensity of individual mitochondria in the 5 sections was averaged to obtain the average TMRE intensity per mitochondrion for each brain slice. To analyze the colocalization of LC3 and Tom20, Volocity software (PerkinElmer) was used to reconstruct 3D images from z-stack images. In the 3D image, the “Percentage Thresholding” method was used to threshold Tom20 signals, the “Intensity Thresholding” method was used to threshold LC3 signals, and the volume of the region doubly positive for LC3 and Tom20 was measured by using the “Intersect” function. To analyze EdU and Tom20 colocalization, z-stack images were deconvoluted using the “simple deconvolution” function in ZEN 2.3 software. Volocity software was used to reconstruct 3D images from deconvoluted images and measure the volume of the region doubly positive for EdU and Tom20 using the “Intersect” function. EdU signals were thresholded by using the “Intensity Thresholding” method. To analyze mtKeima images, z-stack images were collapsed using the “Max intensity z projection” function in ImageJ, and the integrated intensity of mtKeima in the Ex543 or Ex458 channels were measured. The experimenter was blind to the treatment during the analysis of confocal images. To analyze western blot images, the images were converted to greyscale in ImageJ. The bands were outlined and measured for mean grey value in Image J. For analysis of the ATP sensor, integrated intensity in the soma of transfected neurons was measured.

Immunohistochemistry

Mice were anesthetized by intraperitoneal injection of Ketamine/Xylazine. After adequate anesthesia, the thorax was cut open to expose the heart. The mice were perfused transcardially with 40 ml PBS (pH 7.4), followed by 45 ml ice-cold 4% paraformaldehyde (PFA) in PBS. After perfusion, brains were removed immediately, post-fixed in 4% PFA for 24 hr at 4°C, and then submerged in 30% sucrose for 48 hr at 4°C. Frozen brains were cut into 30-μm sections using a cryostat (Leica CM3050S). Free-floating sections were washed in PBS, permeabilized with 0.5% Triton X-100 in PBS for 1 hr at room temperature, blocked with 5% normal goat serum for 1 hr at room temperature, incubated with primary antibodies for 48 hr at 4°C, rinsed with PBS, incubated with secondary antibodies, rinsed with PBS, and then mounted with mounting media (Vectorlabs, H-1400).

EdU labeling

Mice were intraperitoneally injected with EdU (80 mg/kg) once a week for four times during CSD. After CSD treatment, mice were perfused, and brains were cut into 30-μm slices with a cryostat. Brain slices were heated in the antigen retrieval buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) at 95-100 °C for 20 min in a steamer, and then cooled down to room temperature. After antigen retrieval, the floating brain sections were washed in PBS (3x 10 min), permeabilized with PBS containing 0.5% Triton X-100 for 1 hr at room temperature, and then washed with PBS containing 3% BSA. The Click-iT^® EdU Alexa Fluor^® 488 Imaging Kit (Invitrogen, C10337) was used to stain EdU. 1× Click-iT^® reaction cocktail was prepared according to the manufacturer’s instructions and diluted (1:10) with the reaction buffer. The brain sections were incubated with the diluted reaction cocktail for 30 min at room temperature while being protected from light. After washing with PBS containing 3% BSA, the brain sections were incubated with the Tom20 antibody at 4 °C overnight, washed with PBS, incubated with the secondary antibody against Tom20 antibody, washed in PBS again, and then mounted in mounting media (Vectorlabs, H-1400).

For EdU labeling in primary hippocampal neurons, EdU was added to the neural medium (10 μM final concentration) at DIV 11. Neurons were fixed at DIV 14 in PBS containing 4% formaldehyde and 4% sucrose for 12 min at room temperature. After rinsing, neurons were stained by using the Click-iT^® EdU Alexa Fluor^® 488 Imaging Kit with 1:10 diluted Click-iT^® reaction cocktail. After EdU staining, neurons were incubated with the Tom20 antibody diluted in GDB buffer (0.1% gelatin, 0.3% Triton X-100, 16 mM sodium phosphate, 450 mM NaCl, pH 7.4) at 4 °C overnight, washed with PBS for three times, incubated with the secondary antibody at room temperature for 1 hour, washed with PBS three times again, and then mounted with mounting media (Vectorlabs, H-1000).

Immunoblotting

Brain tissues were dissected out and immediately frozen on dye ice. Frozen tissues were homogenized with a Dounce homogenizer (20 strokes) on ice in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF). The homogenate was incubated with RIPA buffer kept on ice for 1 hr, and then centrifuged at 13000 g for 10 min at 4°C. The supernatant was collected and measured for protein concentration. Protein samples were boiled and separated by gel electrophoresis (12–15% SDS-PAGE gel, 50 μg protein per sample, 100 V, 2 hr) and transferred to nitrocellulose blotting membranes (300 mA, 1 hr, 4°C). The membrane was blocked with 5% non-fat milk in TBST (50 mM Tris, 150 mM NaCl, 0.1 % Tween 20, pH 7.4) for 1 hr at room temperature, and incubated with primary antibodies diluted in TBST containing 5% BSA at 4°C overnight. After rinsing with TBST, the membrane was incubated with secondary antibodies diluted in TBST containing 5% non-fat milk for 1 hr at room temperature. For immunoblotting against PINK1, TBST containing 5% BSA was used during blocking and incubation with antibodies.

Viral production

Lentivirus.

HEK-293T cells (< 3 passages) were cultured on 15-cm plates coated with 0.2% gelatin in DMEM medium supplemented with 10% fetal bovine serum until reaching 80% confluence. The medium was replaced with fresh medium 2 hr before transfection. For transfection of each 15-cm plate, 22 μg pRRLsin lentiviral vector containing siRNA sequences, 15 μg psPAX2, 5 μg pMD2.G and 2 μg pAdVantage plasmids were added to 2 ml water containing 260 μl CaCl₂ (2 M). The DNA solution was added to 2 ml 2X HBSS (50 mM HEPES, 280 mM NaCl, 1.5 mM Na₂HPO₄, pH 7.05). After incubation at room temperature for 2 min, the mixture was added to the culture plate dropwise. The medium was replaced with 15 ml UltraCULTURE Media (UltraCULTURE, 1 mM sodium pyruvate, 0.075 % sodium bicarbonate, 1x glutamine) at 16 hr after transfection. The medium was collected at 48 hr after transfection and kept at 4 °C. 15 ml fresh UltraCULTURE medium was added to the plate and collected at 72 hr after transfection. The media collected at the two time points were combined, filtered with 0.45 μm filter bottles, and centrifuged at 25,000 rpm for 90 min at 4 °C (SW28 rotor, Beckman Coulter). The supernatant was removed and the pellet containing virus was dissolved by incubation with 100 μl 1x HBSS overnight at 4°C. For further purification of virus, the viral suspension was placed on the top of 1.5 ml 20% sucrose (in 1x HBSS) and centrifuged at 21000 rpm for 2 hr at 4°C (SW55 rotor, Beckman Coulter). The pellet was incubated with 100 μl 1x HBSS overnight at 4°C, aliquoted and stored at −80°C. The titer of purified virus was determined by transducing HEK-293T cells with a series of dilutions. The virus used for in vivo injection had a titer of 10⁹-10¹⁰ IU/ml.

AAV.

HEK-293T cells (< 3 passages) were cultured on the 0.2% gelatin-coated 15-cm plates until reaching 70–80 % confluence. For transfection of each 15-cm plate, 22.5 μg pAAV-vectors containing mtKillerRed or mtKeima, and 67.5 μg pDP1rs (pHelper) were added to 1.2 ml H₂O and 1.25 ml CaCl₂ (0.5 M). The mixture was added to 2.45 ml 2X HBSS and added to the culture plate dropwise after incubation at room temperature for 2 min. The medium was replaced with 15 ml fresh DMEM medium at 16 hr after transfection. To harvest AAV, at 48–72 hr after transfection, EDTA was added to the medium to a final concentration of 10 mM and incubated for 3 min at room temperature. The cells were pelleted by centrifugation at 1610 g for 5 min. The cell pellet was resuspended in 2.5 ml serum-free DMEM, subjected to four rounds of freeze/thaw cycles by alternating the tubes between a dry ice bath and a 37°C water bath, centrifuged at 10000 g for 10 min, and stored at −80°C. The virus was purified from the cell lysate by using the ViraBind™ AAV Purification Kit (VPK-140, Cell Biolabs) following the manufacturer’s instructions. The titer of purified virus was determined by transducing cultured hippocampal neurons with a series of dilutions. The virus used for in vivo injection had a titer of ~10¹³ IU/ml.

Quantification and statistical analysis

SAS software was used for statistical analysis. For the comparison of two groups, two-tailed Student’s t-test or paired Student’s t-test was used for data that passed the normality and equal variance tests, and Mann-Whitney rank sum test was used for data that failed the normality and equal variances tests. For comparison of ≥ 3 groups with one factor, one-way ANOVA or one-way RM ANOVA was used for data that passed the normality and equal variance tests, and Kruskal-Wallis one-way ANOVA on ranks was used for data that failed the normality and equal variances tests. For comparison of ≥ 3 groups with two factors, two-way ANOVA was used. The assumptions of normality and homogeneity of variance were evaluated, and then two-way ANOVA analysis was performed with SAS. If interaction was significant, one-way ANOVA with simple effects was used for post hoc analysis. If interaction was not significant, linear contrasts within the framework of a two-way analysis of variance was used for post hoc analysis. P < 0.05 was considered significant. The statistical results are described in Table S1.

Supplementary Material

Table S1

Table S1. Summary of statistical analysis, Related to Figure 1-8.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-LC3	MBL International	Cat# M152-3; RRID:AB_1279144
Rabbit monoclonal anti-Tom20	Santa Cruz Biotechnology	Cat# sc-11415; RRID: AB_2207533
Rabbit monoclonal anti-pAMPKα (clone 40H9)	Cell Signaling Technology	Cat# 2535S; RRID: AB_331250
Rabbit polyclonal anti-AMPKα	Cell Signaling Technology	Cat# 2532S; RRID: AB_330331
Rabbit polyclonal anti-p62	Cell Signaling Technology	Cat# 5114S; RRID:AB_10624872
Rabbit polyclonal anti-PGC-1α	Abcam	Cat# ab54481; RRID:AB_881987
Rabbit polyclonal anti-TFAM	Abcam	Cat# ab131607; RRID:AB_11154693
Rabbit polyclonal anti-PINK1	Novus Biologicals	Cat# BC100-494; RRID:AB_10127658
Rabbit polyclonal anti-Parkin	Cell Signaling Technology	Cat# 2132S; RRID:AB_10693040
Rabbit polyclonal anti-GAPDH	Millipore	Cat# ABS16; RRID:AB_10806772
Rabbit polyclonal anti-β-Tubulin	Invitrogen	Cat# 32-2600; RRID:AB_2180433
Goat anti-Rabbit HRP-conjugated antibody	Bio-Rad	Cat# 1706515; RRID:AB_11125142
Goat anti-Mouse HRP-conjugated antibody	Bio-Rad	Cat# 1706516; RRID:AB_11125547
Mouse monoclonal anti-Optineurin	BioLegend	Cat# 870102; RRID:AB_2820187
Mouse monoclonal anti-CaMKIIa (clone CB-a-2)	Innovative Research	Cat# 13-7300; RRID: AB_86627
Rabbit polyclonal anti-Parvalbumin	Swant	Cat# PV27; RRID: AB_2631173
Rabbit polyclonal anti-c-Fos	Abcam	Cat# ab190289; RRID: AB_2737414
Bacterial and virus strains
rAAV9.CaMKII.hChR2 (E123A)-eYFP.WPRE.hGH	Penn Vector Core	Lot# V3055TI-R
rAAV5-CaMKII-hM4D-mcherry	UNC Vector Core	Lot# AV6334D
rAAV-CaMKII-mtKillerRed	This paper	N/A
rAAV-CaMKII-mtKeimaRed	This paper	N/A
rAAV-Synapsin-cyto-mRuby3-iATPSnFR1.0	This paper	N/A
Lenti-PINK1 siRNA-GFP	This paper	N/A
Lenti-Parkin siRNA-GFP	This paper	N/A
Lenti-scramble siRNA-GFP	This paper	N/A
Chemicals, peptides, and recombinant proteins
tetramethylrhodamine ethyl ester (TMRE)	Invitrogen	Cat# T669
5-ethynyl-2-deoxyuridine (EdU)	Invitrogen	Cat# E10187
clozapine N-oxide dihydrochloride (CNO)	HelloBio	Cat# HB6149
Critical commercial assays
Cytochrome c Oxidase Assay kit	Sigma	Cat# CYTOCOX1
DNeasy® Blood & Tissue Kit	QIAGEN	Cat# 69506
SYBR® Green PCR Master Mix	Applied Biosystems	Cat# 4309155
Trizol	Ambion	Cat# 15596018
TransScript First-Strand cDNA Synthesis Kit	Invitrogen	Cat# 11752
Lipofectamine 2000	Tdermo Fisher Scientific	Cat# 11668-019
Click-iT® EdU Alexa Fluor® 488 Imaging Kit	Invitrogen	Cat# C10337
ViraBind™ AAV Purification Kit	Cell Biolabs	Cat# VPK-140
Corticosterone Parameter Assay Kit	R&R Systems	Cat# KGE009
Mouse IL-6 Quantikine ELISA Kit	R&R Systems	Cat# M6000B
Mouse IL-1 beta/IL-1F2 Quantikine ELISA Kit	R&R Systems	Cat# MLB00C
Experimental models: Organisms/strains
Mouse: C57BL/6J	Charles River	Strain code: 027
Mouse: CD1®MGS	Charles River	Strain code: 022
Mouse: B6.129S4-Pink1tm1Shn/J	Jackson Laboratory	Stock number 017946; RRID: IMSR_JAX:017946
Mouse: B6.129S4-Park2tm1Shn/J	Jackson Laboratory	Stock number 006582
Mouse: B6.Cg-Tg(Camk2a-cre)T29-1Stl/J	Jackson Laboratory	Stock number 005359; RRID: IMSR_JAX:005359
Mouse: C57BL/6J-Tg (Eno2-YFP/Cox8a)YRwb/J	Jackson Laboratory	Stock number 007857
Oligonucleotides
siRNA targeting sequence: PINK1:CTGTGTATGAAGCCACCAT	This paper	N/A
siRNA targeting sequence: Parkin: GTCCCAACTCCCTGATTAA	This paper	N/A
siRNA targeting sequence: Scramble: CTATCGCTAGTACTAGCGA	This paper	N/A
Primer: cytochrome b Forward: CTTCGCTTTCCACTTCATCTTACC	This paper	N/A
Primer: cytochrome b Reverse: TTGGGTTGTTTGATCCTGTTTCG	This paper	N/A
Primer: β-actin Forward: GCTCCTCCTGAGCGCAAGTACTC	This paper	N/A
Primer: β-actin Reverse: CTCATCGTACTCCTGCTTGCTG	This paper	N/A
Primers for PGC1α, TFAM, and TBP, see Table S2	See Table S2	N/A
Recombinant DNA
pCS2-NXE+mem-KillerRed	A gift from Michael Levin	Addgene plasmid #45761; RRID:Addgene_45761
Synapsin-cyto-mRuby3-iATPSnFR1.0	Lobas et al., 2019	Addgene plasmid #102557; RRID:Addgene_102557
mKeima-Red-N1	A gift from Michael Davidson	Addgene plasmid #54597; RRID:Addgene_54597
pAAV-CaMKIIa-hChR2(H134R)-mCherry	A gift from Karl Deisseroth	Addgene plasmid #26975; RRID:Addgene_26975
Software and algorithms
SPSS	IBM	https://www.ibm.com
R	N/A	https://www.r-project.org
MATLAB R2021a	MathWorks	https://www.mathworks.com
Clampfit 11.2	Molecular Devices	https://www.moleculardevices.com
TopScan/ObjectScan	CleverSystems	http://cleversysinc.com
ANY-maze	Stoelting	http://www.anymaze.co.uk
OPTG4	Doric Lenses	http://doriclenses.com
Volocity	PerkinElmer	http://www.perkinelemer.com

Highlights.

• Mitochondria in the amygdala are impaired by chronic social defeat stress

• Damaged mitochondria trigger mitophagy

• Elevated mitophagy causes excessive mitochondrial elimination

• Mitochondrial loss induces weakening of synapses in the BLA-BNST anxiolytic pathway