Amygdalo‐nigral circuit mediates stress‐induced vulnerability to the parkinsonian toxin MPTP

Abstract Aims The aim was to investigate the effect of mood disorders on parkinsonian toxin 1‐methyl‐4‐phenyl‐1,2,3,6‐tetrahydropyridine (MPTP)‐induced motor disability, substantia nigra pars compacta (SNc) dopaminergic (DA) neurons loss. Also, the neural circuit mechanism was elucidated. Methods The depression‐like (physical stress, PS) and anxiety‐like (emotional stress, ES) mouse models were established by the three‐chamber social defeat stress (SDS). The features of Parkinson's disease were reproduced by MPTP injection. Viral‐based whole‐brain mapping was utilized to resolve the stress‐induced global changes in direct inputs onto SNc DA neurons. Calcium imaging and chemogenetic techniques were applied to verify the function of the related neural pathway. Results We found that PS mice, but not ES mice, showed worse movement performance and more SNc DA neuronal loss than control mice after MPTP administration. The projection from the central amygdala (CeA) to the SNcDA was significantly increased in PS mice. The activity of SNc‐projected CeA neurons was enhanced in PS mice. Activating or inhibiting the CeA‐SNcDA pathway could mimic or block PS‐induced vulnerability to MPTP. Conclusions These results indicated that projections from CeA to SNc DA neurons contribute to SDS‐induced vulnerability to MPTP in mice.


| INTRODUC TI ON
Parkinson's disease (PD) is a degenerative disease of the central nervous system, and the most obvious symptoms are movement disorders such as shaking, rigidity, slowness of movement, and difficulty walking and gait. Later, it can cause neuropsychiatric disturbances that range from mild to severe, including disorders of speech, cognition, thought, and mood. 1 As PD advances, these nonmotor symptoms can be predominant. 2 In the experiment, MPTP is one of the most accepted neurotoxins widely utilized to induce the PD mouse model nowadays. 3 Both acute and chronic stress play an important role in PD. 4 There is a great deal of clinical evidence that stress can increase the symptoms of PD. 5 It has been reported that PD patients may find that their tremor is worsened when they become anxious or angry. 6 Stress-induced toxicity has been shown in many diseases, including schizophrenia, stroke, Alzheimer's disease, anxiety, and depression. 7 More recently, several authors have speculated about the role of emotional stress (ES) in PD. [8][9][10] A review revealed that enhanced psychological stress during a pandemic can disrupt several motor functions. 11 Besides, a survey of 5000 patients revealed that PD patients experience greater levels of stress than controls and that stress worsens both motor and nonmotor symptoms. 12 On the basis of these theories, we hypothesize that stress-induced mood disorders contribute to the progression of PD.
Stress-induced depressive disorders and anxiety disorders are common neuropsychiatric diseases worldwide that limit psychosocial functioning and diminish the quality of life. Mood disorders and PD share some common pathophysiological features. In our previously published study, 13 we observed that the mice in the physical stress (PS) group showed depression-like behavior, and others in the ES group manifested anxiety-like phenotype after treating with the three-chamber social defeat stress (SDS) procedures. Thus, this model provides the opportunity for studying simultaneously the impact of stress-induced depressive disorders and anxiety disorders on PD. The basal ganglia, a group of brain structures innervated by the dopaminergic system, are the most seriously affected brain areas in PD. 14 The main pathological characteristic of PD is selective loss of dopaminergic (DA) neurons in the substantia nigra pars compacta (SNc), 15,16 affecting up to 30% of the DA cell bodies and approximately 50% of the DA fibers by the time of death. 17 A hypothesis of the cause of depression was a deficiency in monoamine neurotransmitters, which consist of serotonin, norepinephrine, and dopamine.
It has been implicated in both major depressive disorder 18 and PD. 19 Studies have also revealed early-phase dysfunction of SNc DA neurons and impairment of cortico-striatal long-term depression in PD transgenic mice. 20 Besides SNc, it was reported that the spread of pathologic α-synclein was also observed in amygdala. 21 Amygdala is a region of the brain primarily associated with emotional processes.
The previous study has found amygdala changes in emotionality in people suffering from PD. 22 It is generally considered that the central nucleus of amygdala (CeA) functions as a major output of the amygdala by converging inputs from the basolateral nucleus and other amygdalar subregions during the information processing after stress. 23 Thus, we wonder if amygdalar subregion, such as CeA, was also involved in the effect of stress-induced mood disorders on PD.
These findings show the interdependence between mood disorders and advancing PD. However, it remains unclear how stress-induced mood disorders affect PD.
The aim of this study was to characterize the role of stressinduced mood disorders in the progression of PD and the neural mechanism. First, we utilized the three-chamber SDS procedures to establish depression-like (PS group) and anxiety-like (ES group) mouse models. Then, a series of behavioral tests were operated to measure motor disability caused by MPTP. Meanwhile, immunofluorescence staining was used to detect DA neurons death in SNc.
Next, whole-brain retrograde virus tracing and the GCaMP6s technique were applied to probe potential neural circuits underlying stress-induced vulnerability to the parkinsonian toxin MPTP at the anatomical and functional levels. Finally, we adopted chemogenetic manipulation in vivo to further verified the adequacy and necessity of this neural pathway.

| Three-chambersocialdefeatstress mouse model
SDS protocol was carried out according to Golden et al. 24 and Guangjian Qi et al. 13 Prior to experiments, CD1 were selected as aggressors based on two criteria: CD1 attack at least two consecutive sessions in 180 s during 3 consecutive days; the latency to initiate aggression must be less than 60 s. Then, SDS experiment was administrated as follows: (i) in the three-chamber cage, CD1 was placed in the middle chamber, between the PS mouse and PS mouse, separated with a perforated transparent partition and (ii) exposing intruder ES mouse to CD1 aggressor in the middle chamber. After 5 min of social defeat, transfer the ES mouse back to the left chamber for the remainder of 24 h. The ES mouse was staying in the right chamber and observed the whole procedure without receiving direct physical contact; and (iii) repeating SDS for successive 10 days.
For each subsequent daily 5 min defeat, CD1 aggressors were not removed from their chamber, but PS and ES mice were alternated daily to prevent habituation to the same CD1. Without any physical or emotional stimuli, the control mouse (CON) was housed in the left or right chamber with CD1 in the middle chamber for 10 days.

| MPTPmousemodel
The MPTP model was carried out on the basis of the previous study. 25 According to the paradigm, mice received MPTP (25 mg/kg, i.p.) or equal amounts of 0.9% saline (NS, i.p.) injection for 7 consecutive days. Behavioral performances were assessed in the locomotion test, rotarod test, pole test, and edge test. Neuronal death was detected by immunofluorescence staining 7 days after MPTP injection. Inferior colliculus (IC), Parabrachial nucleus (PB), and Interposed nucleus (IP), which was aligned by the mouse brain map of Allen Brain Atlas.

| Invivofiberphotometry
In the in vivo fiber photometry experiment, we first injected 300 nL mixed helper virus of AAV9-TH-Cre, AAV9-DIO-G, and AAV9-DIO-TVA-His-EYFP into bilateral SNc (150 nL/unilateral). After SDS administration, an equal amount of RV-EnVA-ΔG-GCaMP6s-dsRed was injected into the same site. Meanwhile, optical fibers were planted 0.5 mm above the coordinate of bilateral CeA (from bregma: AP ±2.85 mm, ML −1.4 mm, DV -4.23 mm). After 10 days, the retrograde virus was fully expressed. When mice undergo the SI test, the calcium activity of neurons in CeA that project to SNc TH + cells was synchronously recorded through the fiber-photometry system (Inper Technology Co., Ltd). The raw Ca 2+ fluorescence data was coupled with the track of mice in the SI test and then normalized by z-score analysis. Finally, the images and results were expressed through MATLAB.

| Invivochemogeneticmanipulation
For the hM4Di-inhibited manipulation experiment, mice were in-  Supplementary Table S1. Meanwhile, the detailed methods are provided in the Supplementary Appendix S1.

| PSaggravatesMPTP-inducedmovement disorders and SNc DA neuronal loss
To explore the potential effects of mood disorders on MPTPinduced damage, we employed a 10-day consecutive threechamber SDS mouse model ( Figure 1A). The PS mouse in the left chamber was introduced to the middle chamber to physically contact the aggressive CD1 mouse for 5 min and go through attack.
Without direct physical stimuli, the ES mouse in the right chamber was subjected to scene-dependent emotional stimulation by ob-

VTA SNc
vs. ES, p = 1.000) levels similar to those of CON mice (Figure 1 E,F).
Together, these results demonstrated that three-chamber SDS induced reliable depressive-like behaviors and typical anxiety-like behaviors.
Next, we wanted to determine whether SDS influenced the motor disability induced by MPTP. Following the forced swim-  Figure 1K,M). Surprisingly, ES mice showed no significant changes between the NS group and MPTP group through the above behavioral tests and cell calculations. This result indicated that animals with anxiety were not susceptible to MPTP. Taken together, these results suggested that PS could aggravate movement disorders and TH + neuron death caused by MPTP.

| MappingglobalchangesininputstoSNcDA neurons in depressed mice
Given that depression was susceptible to MPTP, we aimed to elucidate the neural mechanism of this phenomenon. Studies have revealed that the SNc receives efferent projections from other regions around the whole brain; thus, we utilized a brain-wide viral tracing tool to detect changes in upstream inputs to SNc TH +

| ActivityofSNc DA -projecting neurons in the CeA was increased in depressed mice when interacting with CD1
Emotional deficits are known to be associated with PD, and a key structure of emotional processing is the amygdala. In the above results, we found that projections from the CeA to the SNc DA were definitely changed in PS mice ( Figure 3A,B). Thus, we next GCaMP6s was expressed in the CeA, a social interaction test was performed on day 18 ( Figure 3C,D). The dynamics of SNc DAprojecting neurons in the CeA during social avoidance behaviors were recorded in real time ( Figure 3E). By analysis of average Ca 2+ activity in every position of the travel route, we found that the activity of SNc DA -projecting neurons in the CeA was increased (CON vs. PS, p = 0.0002) in depressed mice when the aggressive CD1 mouse entered the interaction zone ( Figure 3F). ( Figure 4F,G). We observed that activating CeA-SNc DA circuits could amplify dyskinesia induced by MPTP injection. Moreover, by brain tissue immunofluorescence staining, we found that the loss of TH + cells was increased in the SNc (mCherry + CNO vs. hM3Dq + CNO, p = 0.0276; hM3Dq + NS vs. hM3Dq + CNO, p = 0.0456) of hM3Dq + CNO mice but not in the VTA (Figure 4H-J). These results suggested that activating CeA-SNc DA circuits mimicked the vulnerability to MPTP as physical stress.

| InhibitionoftheCeA-SNc DA circuit reversed the MPTP vulnerability caused by depression
To determine whether the CeA-SNc DA circuit is required for depression-aggravating damage caused by MPTP, we injected inhibitory AAV-DIO-hM4Di-mCherry or AAV-DIO-mCherry as a control virus into the SNc, while AAV1-TH-Cre was injected into the CeA ( Figure 5A-C). After viral expression, mice received 10 days of SDS to establish a depressive-like model and CNO/saline injection daily. Then, all mice were subjected to MPTP treatment for a week. Next, the same series of behavioral tests were carried out.
Finally, the mice were sacrificed to detect TH + cells ( Figure 5A).
According to the results of behavioral tests, we found that the distance traveled in the locomotion test was not different among all mice ( Figure 5D), but mice injected with both hM4Di and CNO expressed more latency time (mCherry + CNO vs. hM4Di + CNO, p = 0.0001; hM4Di + NS vs. hM4Di + CNO, p = 0.0003) to fall off the rotarod than the other groups ( Figure 5E). This result implied that inhibition of the CeA-SNc DA circuit is beneficial for improving the ability but not the willingness of movement in mice. In the pole test (mCherry + CNO vs. hM4Di + CNO, p = 0.0020; hM4Di + NS vs. hM4Di + CNO, p = 0.0040) and ledge test (mCherry + CNO vs. hM4Di + CNO, p = 0.021; hM4Di + NS vs. hM4Di + CNO, p = 0.003), mice in the hM4Di + CNO group displayed better performance than the other groups ( Figure 5F,G). Through brain tissue immunofluorescence staining ( Figure 5H), we found that the number of TH + cells in the VTA was not different among the four groups ( Figure 5I). This phenomenon suggested that the VTA is not an impressionable tar-

| DISCUSS ION
In this exploratory study, we investigated whether stress-induced mood disorders contribute to sensitivity to the parkinsonian toxin MPTP and which neural circuit was involved in this process. There are three main findings. First, mice with PS, but not ES, were more susceptible to MPTP with worse dyskinesia and more DA neuron death in the SNc (Figure 1). Furthermore, the projection from the CeA to SNc DA was increased (Figure 2), and the activity of this pathway was enhanced in mice with PS during interaction with aggressive CD1 mice (Figure 3). In line with these results, we also found that activating the CeA-SNc DA circuit could mimic the vulnerability to MPTP (Figure 4) and that inhibiting this pathway could reverse vulnerability to MPTP caused by PS ( Figure 5).
Notably, in this study, we found that PS but not ES led to susceptibility to MPTP, although anxiety disorder also commonly occurred in PD as one of the nonmotor symptoms. 29 Previous studies have shown that MPTP-treated mice reproduce the features of anxiety, 30 and it was reported that MPTP-induced dopamine depletion in the basolateral amygdala led to anxiety-like behaviors. 31 However, studies on the influence of anxiety on PD are relatively rare. In our study, mice with ES showed neuropathology caused by MPTP similar to that of control mice. Moreover, ES mice with MPTP treatment showed no significant change in motor ability or neuronal death compared with ES mice without MPTP administration. This result indicated that ES-induced anxiety may not lead to the vulnerability of DA neurons to the parkinsonian toxin MPTP. The different impact between PS and ES on the movement performance and SNc DA neurons death may be attributed to two aspects. Firstly, the PS mice had worse performance and more neuron death than ES mice, which may be ascribed to the PS group undergoing severer stress than ES in the three-chamber SDS modeling. A previously published study has also suggested that ES-exposed mice, compared with the PS-exposed group, exhibited a much lesser degree of social avoidant behavior. 32 Secondly, the structural connectivity of the SNc DA neurons may be also involved in the difference between these groups. Along with the whole brain mapping of the direct inputs of SNc DA neurons in PS mice, the number of labeled CeA neurons in the PS group was approximately twice as many as that in controls ( Figure 2E).
Undoubtedly, the comparison of brain structural changes between ES mice and control mice is necessary for clarifying the reason why ES mice have no influence on the reactivity of SNc DA neurons to MPTP. Thus, this result and the potential mechanism remain to be further researched, which is a limitation of this study.
The CeA is a critical nucleus of the striatum involved in aversive stress processing. 33 For example, during fear conditioning, conditioned stimulus and unconditioned stimulus are relayed to the lateral nucleus of the amygdala from thalamic and cortical regions of auditory and somatosensory systems, respectively. [34][35][36] Conditioned stimulus information is then transmitted through the lateral amygdala and surrounding nuclei to the CeA to mediate fear conditioning. Unconditioned stimulus inputs to the CeA could be involved in higher-order integration. In addition, the CeA projects to brainstem areas that control the expression of stress responses. For example, damage to the lateral hypothalamus affects blood pressure, damage to periaqueductal gray interferes with freezing, 37 and damage to the bed nucleus disrupts the conditioned release of pituitary-adrenal stress hormones. 38 In our study, we extended the current knowledge of the function of the CeA and revealed enhanced activity of CeA in promoting the advance of PD.
In addition to the CeA, we also detected apparent changes in projections from SNr and ZI to SNc DA neurons. Whether these two nuclei were related to susceptibility to MPTP remained unclear in this study. It was reported that ZI plays a role in the initiation and control of movement, 39 but near-infrared light treatment protected SNc DA neurons instead of ZI from MPTP. 40 SNr, a major basal ganglia output nucleus, plays a vital role in movement execution. 41 It was reported that SNr displayed abnormalities in neural discharge caused by MPTP. 42 The dopamine transporter and TH in the mouse SNr were also reduced due to MPTP. 43 In addition, studies have revealed that the SNr-thalamus projections support the continuity of motor patterns, while SNr-SNc projections modulate the immediate motor drive behind them. 44 CNO, the most widely used synthetic DREADD ligand, is able to back metabolized to clozapine (CLZ). CLZ accumulated in brain tissue can occupy dopamine and serotonin 5HT receptors in the living brain. It is unknown Whether CLZ has DREADD-independent effects on behavior. In our study, CNO were delivered 3 mg/kg daily and behavioral tests were measured more than 24 h after the last dose. It was reported that CNO itself can enter the brain, and free cerebrospinal fluid levels were within the range to activate DREADD directly, while CLZ level remained below the detection limit, and the half maximal effective concentration (EC50) of CLZ was substantially lower (0.42 nM) than CNO (8.1 nM). 45 Furthermore, there was a study finding a CNO dose-dependent effect on the availability of both neuroreceptor sites, 46 and it was 5, but not 3 mg/kg, a dose of CNO reduced cocaine-induced locomotion. 47 For the chronic applications, non-DREADD-expressing mice that received i.p. vehicle, CNO, or compound 21 (C21) 5 days/week for 16 weeks showed largely lack of behavioral effects. 48 Thus, studies utilizing DREADDs to probe behaviors should consider this limitation when choosing a dose of CNO and include non-DREADD CNO controls.
The current treatments for PD are involved in various molecular targets, including α-synuclein, transcription factor EB, 49 striatal metabotropic glutamate receptor type 5, and autophagic pathways. 50 Meanwhile, the mucuna pruriens, 51