Realgar-Induced Neurotoxicity: Crosstalk Between the Autophagic Flux and the p62-NRF2 Feedback Loop Mediates p62 Accumulation to Promote Apoptosis

Realgar is a traditional Chinese medicine that contains arsenic. It has been reported that the abuse of medicine-containing realgar has potential central nervous system (CNS) toxicity, but the toxicity mechanism has not been elucidated. In this study, we established an in vivo realgar exposure model and selected the end product of realgar metabolism, DMA, to treat SH-SY5Y cells in vitro. Many assays, including behavioral, analytical chemistry, and molecular biology, were used to elucidate the roles of the autophagic flux and the p62-NRF2 feedback loop in realgar-induced neurotoxicity. The results showed that arsenic could accumulate in the brain, causing cognitive impairment and anxiety-like behavior. Realgar impairs the ultrastructure of neurons, promotes apoptosis, perturbs autophagic flux homeostasis, amplifies the p62-NRF2 feedback loop, and leads to p62 accumulation. Further analysis showed that realgar promotes the formation of the Beclin1-Vps34 complex by activating JNK/c-Jun to induce autophagy and recruit p62. Meanwhile, realgar inhibits the activities of CTSB and CTSD and changes the acidity of lysosomes, leading to the inhibition of p62 degradation and p62 accumulation. Moreover, the amplified p62-NRF2 feedback loop is involved in the accumulation of p62. Its accumulation promotes neuronal apoptosis by upregulating the expression levels of Bax and cleaved caspase-9, resulting in neurotoxicity. Taken together, these data suggest that realgar can perturb the crosstalk between the autophagic flux and the p62-NRF2 feedback loop to mediate p62 accumulation, promote apoptosis, and induce neurotoxicity. Realgar promotes p62 accumulation to produce neurotoxicity by perturbing the autophagic flux and p62-NRF2 feedback loop crosstalk


Introduction
Realgar, a traditional Chinese medicine whose key ingredient is arsenic sulfide (As 2 S 2 or As 4 s 4 ), has been used as a medicine for thousands of years in China and is widely used in the clinic [1,2]. Some classical prescriptions, such as An-Gong-Niu-Huang-Wan, and Niu-Huang-Jie-Du-Pian, contain 6-12% realgar [1]. However, under the influence of the concept of nontoxic side effects of traditional Chinese medicine, drug-induced arsenic poisoning events caused by long-term and excessive use of realgar alone or its compound preparations have been reported [3]. Epidemiological and animal experiments have verified that long-term exposure to arsenic not only leads to skin damage, digestive system damage, and tumor occurrence but also causes CNS damage [4][5][6][7][8]. The effects of arsenic exposure and arsenic accumulation induced by realgar as a drug on human health have aroused widespread public concern. The brain is one of the main target organs of arsenic toxicity [9]. Clarifying the neurotoxic mechanism of realgar and discovering sensitive molecular targets of early changes have important theoretical and practical significance for early effective measures to prevent the occurrence of drug-induced arsenic poisoning and reasonably guide the clinical use of drugs.
Our previous study found that the multifunctional protein p62/SQSTM1, which is induced by stress and acts on selective autophagy, is continuously highly expressed in the cortex after realgar exposure [10]. However, its molecular mechanism in realgar-induced central nervous system toxicity is not very clear. p62 is a ubiquitin-binding protein encoded by SQSTM1. Its amino acid sequence has multiple functional domains, including the PB1, TB, KIR, and LIR domains [11]. The LIR domain can interact with LC3 and participate in the process of autophagy. It is also known as a marker protein of autophagy degradation [12]. In addition, p62 competes with nuclear factor erythroid 2-related factor 2 (NRF2) to bind KEAP1 (Kelch-like ECH associated protein 1) through the KIR domain, promoting the dissociation of NRF2-KEAP1 and activating NRF2 [13]. NRF2, as a key transcription factor involved in anti-apoptosis and antioxidant defense [14], can directly induce p62 transcription and promote p62 protein expression after activation, thus forming a p62-NRF2 feedback loop [11,15]. According to a previous study, p62 accumulation induces apoptosis by activating the caspase-8 inflammatory cascade and increasing caspase-9 cleavage [16]. However, the role of the p62-NRF2 feedback loop in realgar-induced neurotoxicity remains to be further explored.
Autophagy is an important regulatory mechanism maintaining cellular homeostasis. In neural cells, disordered autophagy can promote the accumulation of protein aggregates, disrupt cellular homeostasis, and lead to neuronal apoptosis and cognitive dysfunction [17,18]. The autophagic flux is a dynamic process that includes autophagy induction, autophagosome formation, and autophagy degradation. Damage to any of these processes disturbs autophagic flux homeostasis [19]. Studies have shown that enhanced autophagy induction can promote the transport of p62 to lysosomes, thereby promoting p62 degradation [20]. The c-Jun N-terminal kinase (JNK) family, as a member of mitogen-activated protein kinase (MAPK), is involved in cell differentiation, apoptosis, and autophagy [21,22]. The JNK/c-Jun signaling pathway can upregulate the expression of the autophagy-induced related proteins Beclin1 and Vps34 after being activated by cytokines, stress, inflammatory factors, and other factors, and then, the ECD of Beclin1 combines with the C2 of Vps34 to promote the formation of the Beclin1/Vps34 core complex and activate autophagy induction [23]. LC3 is hydrolyzed to LC3I, and then, LC3I is converted to LC3II, which recruits p62 through the LIR and binds together to the autophagosome membrane to participate in autophagosome formation, and is eventually degraded in the lysosome [24,25]. However, if autophagy is overactivated, the subsequent highly efficient degradation is important for maintaining cellular homeostasis. Autophagy degradation depends on lysosomal hydrolysis enzymes and the acidic environment of the lysosome [26,27]. When autophagy degradation is dysfunctional, the degradation of p62 is blocked, and p62 begins to accumulate. Excessive p62 further enhances the p62-NRF2 feedback loop, which promotes p62 accumulation and induces neuronal apoptosis in a crosstalk manner.
Therefore, our research explores whether realgar can promote the accumulation of p62, aggravate neuronal apoptosis, and lead to CNS toxicity by perturbing the crosstalk between the autophagy flow and the p62 NRF2 feedback loop in vivo and in vitro. This study could provide new ideas and basic experimental data for the study of the CNS toxicity of realgar.

Animals and Reagents
Female Sprague-Dawley rats aged 3 weeks were purchased from the Laboratory Animal Center of China Medical University (Shengyang, China). The rats were fed in an environment under a 12-h light-dark cycle at 21-24 °C. All animals received food and water ad libitum. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of China Medical University (ethics committee approval numbers: CMU2019109 and CMU2021120). Realgar (> 90.0% As2S2) was purchased from Henan Sanmenxia Yuhuangshan Pharmaceutical Co., Ltd. (Henan, China; lot number: 180428). The information of all reagents is listed in Table S1.

Experimental Protocol
We investigated realgar-induced neurotoxicity and disruptions in the autophagic flux in vivo by arbitrarily dividing 3-week-old rats into 3 groups as follows: (1) a control group of rats, (2) a group of rats receiving 0.3-g/kg realgar, and (3) a group of rats receiving 0.9-g/kg realgar. The animals in the control group received 0.5% (w/v) sodium carboxymethyl cellulose (CMC-Na) by oral administration once daily for 8 weeks, while those in the realgar groups received realgar suspensions at different dosages.

Detection of the Metabolic Products of Realgar in Rats
The contents of arsenic species were determined using liquid chromatography-atomic fluorescence spectrometry (LC-AFS). Initially, rat urine was centrifuged at 7500 × g for 5 min at 4 °C and diluted 30-fold before detection. Then, blood was digested with H 2 SO 4 and diluted 30-fold before detection. Finally, a 50-mg cortex sample was homogenized in water, and the supernatant was digested in H 2 SO 4 for detection. The LC-AFS6500 parameters were set as follows: the element of the detection tunnel was set to arsenic, the negative high voltage was 300 V, the total current was 90 mA, and the carrier gas flow rate was set to 5 mL/min.

Behavioral Tests
In the novel object recognition test, the rats were first placed in an open field arena containing two identical objects (10 cm width × 10 cm length × 10 cm tall) and allowed to freely explore for 3 min. After 24 h, one object was replaced with a novel object (different shape and color than the old object), and the rats were allowed to explore for another 3 min. The time that the rats spent interacting with the novel and old objects was recorded. The cognitive index was calculated as follows: time spent exploring new or old object/Total time. The open-field test was performed in a large acrylic cube 50 cm in height and 100 cm in width with a black bottom. Briefly, the rats were individually placed near the wall side and allowed to freely move for 5 min to assess their exploration ability. The time spent in the center zone (30 × 30 cm imaginary square), velocity and distance traveled were evaluated.

SH-SY5Y Cell Culture and Treatment
As previously described, SH-SY5Y cells were cultured in Roswell Park Memorial Institute (RPMI 1640) medium supplemented with 12% fetal bovine serum (FBS) at 37 °C in a humidified 5% CO 2 atmosphere. For the detection of the autophagic flux, SP600125, rapamycin, and chloroquine were added to the medium before stimulation with DMA. p62 siRNA or NRF2 siRNA was added to the medium 24 h before the cell stimulation.

Electron Microscopy
After gradient dehydration with ethanol solutions, the cerebral cortex was incubated with diluted acetone solutions, cleared with propylene oxide, and infiltrated overnight, followed by polymerization at 70 °C overnight with a heating polymerization apparatus. Transverse cerebral cortical Sects. (50 nm) were acquired using an ultrathin slicing instrument. The tissue samples were stained with uranyl acetate and a 50% ethanol saturated solution for 15 min to 1 h. Finally, the samples were observed and photographed under a HITACHI H-7650 transmission electron microscope.

Western Blot Analysis
Cerebral cortical tissues or treated SH-SY5Y cells were homogenized in RIPA buffer containing protease inhibitors and phosphate inhibitors. The protein concentration was quantified with a BCA protein assay kit. The supernatant was separated on a 10-12% SDS-polyacrylamide gel and then wet electrotransferred to a PVDF membrane. After blocking with 10% evaporated skim milk at room temperature for 1 h on a rocker, the PVDF membrane was incubated overnight at 4 °C with the corresponding primary antibodies. The blots were rinsed with TBST four times for 5 min each and incubated with a secondary antibody conjugated with horseradish peroxidase for 1 h. An ECL detection system was used to measure the levels of the target proteins. The signal intensity was detected with ImageJ software.

Coimmunoprecipitation Assay
Cortical tissue weighing 100 mg or treated SH-SY5Y cells were homogenized, incubated for 30 min, and centrifuged at 4000 rpm for 10 min. The supernatants were incubated with specific primary antibodies, followed by the addition of 30 μL of PureProteome™ Protein A/G Mix Magnetic Beads and incubation under gentle rotation. The agarose beads were collected, washed four times with lysis buffer, and resuspended in loading buffer.

Analysis of mRNA Levels
The total RNA was extracted from cerebral cortical tissues or SH-SY5Y cells with TRIzol reagent, and the RNA concentration was measured spectrophotometrically. The cDNA templates were synthesized using a cDNA synthesis kit according to the manufacturer's instructions. The analysis was performed using RT-PCR with gene-specific primers (Table S2) on a QuantStudio™6 Flex real-time fluorescence quantification system with TB Green Real-time PCR Master Mix. The amplification of the target cDNA was normalized to GAPDH or ACTB expression. The relative levels of the target mRNA expression were calculated using the 2 −ΔΔCt method.

Fluorescence and Confocal Microscopy Analyses
Sections of the cortex were mounted on glass slides. The sections were washed with PBS and incubated with the following specific primary antibodies: RBFOX3 (BM4354, 1:100) obtained from Boster Biological Technology (China), 8OHdG (sc-393871, 1:200) obtained from Santa Cruz (USA), and Beclin1 (11,306-1-AP, 1:100) obtained from Proteintech Group (Wuhan, China). SH-SY5Y cells were grown on glass coverslips. After presage processing, the cells were fixed with 4% PFA in PBS, followed by permeabilization with 0.2% Triton X-100 in PBS. Then, the cells were blocked with 5% BSA. The fixed cells were incubated with primary antibodies. The slides were then washed and incubated with a fluorescent secondary antibody. DAPI staining solution was used for nuclear counterstaining. The stained samples were examined under an AI + confocal microscope.

RFP-GFP-LC3B Assay
Briefly, 1.5 × 10 5 SH-SY5Y cells were grown on glass-bottom dishes and infected with adenovirus for 24 h. Then, the SH-SY5Y cells were treated with DMA for an additional 6 h. All samples were examined under an AI + confocal microscope equipped with a 60 × oil immersion objective.

LysoSensor Green DND-189 Staining
The lysosomal pH was quantified using LysoSensor Green DND-189. Briefly, SH-SY5Y cells were loaded with Lys-oSensor Green DND-189 in prewarmed basal medium for 30 min at 37 °C. Then, the cells were washed three times with PBS. After the treatment, the fluorescence intensity of the cells was quantified using an H1MD Multifunctional Microplate Reader with 443-nm excitation and 505-nm emission filters.

CTSD and CTSB Activity Assays
A cathepsin D activity fluorometric assay kit and a cathepsin B activity fluorometric assay kit were used to detect CTSD and CTSB catalytic activity. Briefly, the cortex was dissected from the brain and ground in CD/CB cell lysis buffer with a tissue lyser II, followed by incubation on ice.
After centrifugation, the supernatant was transferred to a 96-well plate, and a master assay mix containing CD/CB reaction buffer and CB substrate was added to each sample. The samples were incubated and read with an HMID Multifunctional Microplate Reader with 328-nm/400-nm excitation and 460-nm/505-nm emission filters.

Proximity ligation assay
The protein interactions in SH-SY5Y cells were detected using a Duolink® PLA assay kit following the manufacturer's protocol. After the treatments, the cells were fixed with 4% PFA and permeabilized with 0.3% Triton X-100. A blocking solution was added to the cells, which were incubated. Then, the slides were incubated with a primary antibody and then a PLA probe solution. After being washed, the slides were incubated with the amplification solution protected from light. Finally, the cell nuclei were stained with DAPI, and the slides were imaged under an AI + confocal microscope.

Statistical Analysis
The data are presented as the X ± SD. All data were tested for normality by the graphical method and were in line with a normal distribution. The statistical comparisons were performed using Student's t tests or one-way analysis of variance (ANOVA), two-way ANOVA, and two-way repeatedmeasures ANOVA, followed by Dennett's post hoc test. P < 0.05 was considered statistically significant. Origin 2018 and GraphPad Prism software version 8.00 were used for the statistical analyses and graphic production.

Content of Arsenic Species in the Cortex, Blood and Urine After Realgar Exposure
First, we performed quantitative analyses of arsenite (As(III)), dimethylarsinic acid (DMA), monomethylarsonic acid (MMA), and arsenate (As(V)) in the cortex, blood, and urine after realgar exposure by using LC-AFS. The results indicated that only DMA was detected in the cortex and blood, and As(III), MMA, and As(V) were not detected ( Fig. 1A). As(III), MMA, and DMA were detected in urine, and As(V) was not detected (Fig. 1a). In the cortex, the content of DMA was significantly increased in a dose-dependent manner in the 0.9-g/kg realgar group (P < 0.05) and was 4.6fold higher than that in the 0.3-g/kg realgar group (Fig. 1b). In blood, compared with the 0.3-g/kg realgar group, the content of DMA in the 0.9-g/kg realgar group did not significantly change, suggesting that the arsenic concentration reached a plateau value (Fig. 1c). In urine, the contents of As(III), MMA, and DMA were significantly increased in a dose-dependent manner in the 0.9-g/kg realgar group (Fig. 1d) (P < 0.05) and were 1.73-fold, 1.48-fold, and 2.12fold higher than those in the 0.3-g/kg realgar group, respectively. These data show that the metabolites of arsenic in realgar can enter and accumulate in the brain and are mainly dominated by DMA.

Effects of Realgar on the Neurobehavior of Rats
We evaluated the impacts of realgar on rat neurobehavior by performing a novel object recognition test and an open-field test. The novel object recognition test evaluates cognitive function by recording the time spent exploring new and old objects and calculating the recognition index [28]. The rats were first placed in an open field containing two identical objects and allowed to freely explore for 3 min during the training stage. After 24 h, one object was replaced with a novel object, and the above steps were repeated during the testing stage. The results indicate that there were no significant differences in the time spent exploring each object and the corresponding recognition index among the control group, the 0.3-g/kg realgar group, and the 0.9-g/kg realgar group ( Fig. 2a-b). In the testing stage, the control group spent more time exploring and in proximity to the new object, and the recognition index was increased (P < 0.05), but the 0.3-g/kg and 0.9-g/kg realgar group did not differ ( Fig. 2c-d). These results suggest that realgar can impair cognitive function.
An open-field test was used to analyze the changes in exploratory ability and the emotions of rats (as reflected by the time spent in the center zone, the number of times the rats entered the center zone, and the distance and velocity traveled) [29]. The results indicate that the 0.3-g/kg realgar group did not significant differ from the control group in the time spent in the center zone, but the 0.9-g/kg realgar group spent significantly less time in the center zone ( Fig. 2e-f) (P < 0.05). Meanwhile, we found that the 0.3-g/kg and 0.9-g/kg realgar groups exhibited significantly fewer entries into the center of the open field than the control group (Fig. 2e, g) (P < 0.05). Moreover, we found that the 0.9-g/kg realgar group had a significantly decreased distance and velocity traveled compared with the control group and the 0.3-g/kg realgar group (Fig. 2h-i) (P < 0.05). Based on these results, realgar can decrease exploratory ability and induce anxietylike behavior in rats.

Effects of Realgar on the Ultrastructure and Apoptosis of Neurons
We first observed ultrastructural changes in neurons by using electron microscopy. In the control group, the nuclear membrane was smooth and complete, the nucleus was large and round, chromatin was evenly distributed, the structure of the Golgi complex and endoplasmic reticulum in the cytoplasm was clear and neatly distributed, the mitochondria were round or oval, and obvious mitochondrial cristae were visible. In the 0.3-g/kg realgar group, the neurons had large and rounded nuclei, but the cytoplasmic Golgi complex and endoplasmic reticulum were blurred, and mitochondrial cristae were broken. In the 0.9-g/kg realgar group, the neurons exhibited an unclear boundary of the nuclear membrane, . h Travel distance of the rats (n = 10). i Travel velocity of the rats (n = 10). Compared with the control group, *P < 0.05. Compared with the 0.3-g/kg realgar group, # P < 0.05 disorder of the Golgi complex and endoplasmic reticulum structure and distribution and expansion of mitochondria (Fig. 3a). These findings suggest that realgar can damage the ultrastructure of cortical neurons.
Next, we explored the effect of realgar exposure on cortex cell apoptosis. Compared with the control group, the percentage of apoptotic cells in the 0.3-g/kg and 0.9-g/kg realgar groups significantly increased in a dose-dependent manner, i.e., 17.02% and 34.22%, respectively (Fig. 3b) (P < 0.05). The Western blot analysis showed that Bax/ Bcl-2 and cleaved caspase-9/caspase-9 in the 0.3-g/kg realgar group did not significantly change, but their ratio was significantly increased in the 0.9-g/kg realgar group (Fig. 3c) (P < 0.05) and was 1.4-fold and 1.3-fold that in the control group, respectively. These results indicate that realgar can promote apoptosis in cortical cells. Taken together, these results suggest that realgar has CNS toxicity.

Realgar-Mediated p62 to Promote Nerve Apoptosis by Perturbing Autophagic Flux Homeostasis
Autophagy, an important self-regulation mechanism, eliminates aggregated and misfolded proteins and impaired organelles [30]. Autophagic flux homeostasis is an important way to evaluate autophagy function [31]. The results show that the ratios of LC3II and LC3I in the 0.3-g/kg and 0.9-g/kg realgar groups were significantly higher than those in the control group ( Fig. 4a-b) (P < 0.05), i.e., 1.3-fold and 1.6-fold, respectively. In the 0.3-g/kg realgar group, the expression level of p62 protein was increased, but there was no significant difference, and the p62 protein expression level in the 0.9-g/kg realgar group was significantly increased (Fig. 4a, c) (P < 0.05) and was 1.7 times that in the control group. Meanwhile, we also observed that the number of autophagosomes in the SH-SY5Y cells were significantly increased by transmission electron microscopy after treating SH-SY5Y cells with DMA for 4 or 6 h (Fig. 4d-e) (P < 0.01). These findings suggest that realgar perturbs cortical autophagic flux homeostasis, but the increase in p62 indicates that this may be related to impaired autophagic degradation.p62 is not only a marker protein of autophagy degradation but also a key protein tightly linked to autophagy and apoptosis [16]. The results show that the cell viability significantly decreased after the treatment with different concentrations of DMA for 24 h, i.e., 74.4% at 5 mM (Fig. 4f) (P < 0.05) and 55.6% at 10 mM (Fig. 4f) (P < 0.01). In addition, compared with the control group, the p62 protein expression level did not significantly change following the treatment with 2.5 mM DMA. However, the expression level of p62 was significantly increased following the treatment with 5 mM and 10 mM DMA (Fig. 4g-h) (P < 0.01), i.e., 1.6-fold and 1.9-fold, respectively. Interestingly, the expression level . c Apoptosis rate of rat cortical cells (n = 3). d Protein expression of Bax, Bcl-2, caspase-9, and cleaved-caspase-9. e Ratio of Bax and Bcl-2 (n = 6). f Ratio of cleaved caspase-9 to caspase-9 (n = 6). Compared with the control group, *P < 0.05 1 3 of p62 was significantly decreased when the SH-SY5Y cells were treated with DMA at 1 h (P < 0.01) but significantly increased at 12-24 h (Fig. 4i-j) (P < 0.01). This result suggests that DMA could promote p62 protein expression levels in SH-SY5Y cells. Next, we silenced p62, and the silencing efficiency was identified as 72% (Fig. 4k) (P < 0.01). Compared with the DMA group, Bax/Bcl-2 and cleaved-caspase-9/caspase-9 were significantly decreased in the DMA + p62 siRNA group (Fig. 4l-n) (P < 0.01), i.e., 63% and 85%, respectively. Jun, Beclin1, Vps34, p62, and LC3II/LC3I protein expression. i Relative expression levels of p-c-Jun, Beclin1, Vps34, p62, and LC3II/ LC3I (n = 6). j Binding of the Beclin1-Vps34 complex was detected by co-IP after SP600125 intervention. k Detection of Beclin1-Vps34 complex formation by PLA. Compared with the control group, *P < 0.05. Compared with the 0.9-g/kg realgar group, # P < 0.05 Taken together, these results suggest that p62 is involved in the DMA-induced apoptosis of SH-SY5Y cells.

Realgar activated Autophagy Through the JNK/ Vps34 Complex Pathway to Promote p62 Aggregation
Under stress conditions, the expression of Beclin1 was upregulated, and Beclin1 formed a complex with Vps34 [32]. The Beclin1/Vps34 complex plays an important role in the induction of autophagy. To elucidate the molecular mechanism by which autophagy was activated by realgar, we first detected the expression of the Beclin1/Vps34 complex. Compared with the control group, the results show that the expression levels of the Beclin1 protein in the 0.3-g/kg and 0.9-g/kg realgar groups were significantly increased (Fig. 5a-b) (P < 0.05), i.e., 1.2-fold and 1.3-fold, respectively. The expression level of the Vps34 protein in the 0.3-g/kg realgar group was increased, but there was no significant difference, while the expression level of the Vps34 protein in the 0.9-g/kg realgar group was significantly increased (Fig. 5a, c) (P < 0.05), i.e., 1.4fold of that in the control group. Moreover, co-IP showed that the binding capacity of the Beclin1/Vps34 complex in SH-SY5Y cells was enhanced after the DMA treatment (Fig. 5d), suggesting that realgar exposure promoted the formation of the Beclin1/Vps34 complex and activated autophagy induction.
The JNK/Vps34 complex pathway plays an important role in the induction of autophagy [33]. The results show that compared with the control group, the levels of p-JNK/ JNK and p-c-Jun/c-Jun in the 0.3-g/kg realgar group were increased, but there was no significant difference. However, the levels of p-JNK/JNK and p-c-Jun/c-Jun were significantly increased in the 0.9-g/kg realgar group (Fig. 5e-g) (P < 0.05) and were 1.3-fold and 2.0-fold that in the control . k After silencing p62, the relative expression levels of p62 mRNA (n = 3). l After silencing p62, protein expression images of Bax, Bcl-2, caspase-9, and cleavedcaspase-9. m Ratio of Bax and Bcl-2 (n = 3). n Ratio of cleaved caspase-9 to caspase-9 (n = 3). Compared with the control group, *P < 0.05, **P < 0.01, Compared with the DMA group, ## P < 0.01 group, respectively, suggesting that realgar activates the JNK/c-Jun signaling pathway. Next, we used SP600125 (JNK-specific inhibitor) as an intervention. The Western blot analysis showed that compared with 0.9-g/kg realgar, SP600125 not only inhibited the protein expression level of p-c-Jun but also significantly decreased the protein expression levels of Beclin1 and Vps34 (Fig. 5h-i) (P < 0.05) by 23% and 20%, respectively. Co-IP also showed that the binding ability of Beclin1 to Vps34 was weakened after inhibiting the JNK pathway (Fig. 5j). The PLA indicated that the formation of Beclin1/Vps34 complexes (red dots) was increased after the DMA treatment in SH-SY5Y cells compared with the that in the control group and was decreased in the DMA + SP600125 group compared with that in the DMA group (Fig. 5k). These data suggest that realgar promotes Beclin1/Vps34 complex production to induce autophagy by activating JNK/c-Jun.
After autophagy induction, LC3 recruits a large amount of p62 to participate in autophagosome formation and promote the autophagic flux [34]. We found that compared with the control group, the protein expression levels of LC3II/LC3I and p62 in the 0.9-g/kg realgar group were significantly increased (Fig. 5h-i) (P < 0.05), indicating that realgar promotes the formation of autophagosomes, which is consistent with the above results (Fig. 3a). After using SP600125 to inhibit the induction of autophagy, the LC3II/LC3I ratio in the realgar + SP group was significantly decreased compared with that in the 0.9-g/kg realgar group (Fig. 5h-i) (P < 0.05), which was 73% of the value. However, although the expression level of the p62 protein decreased, there was no significant difference (Fig. 5h-i). These data further suggest that autophagy degradation is impaired in the presence of autophagic activation, which promotes p62 aggregation.

Realgar Impairs Autophagy Degradation by Inhibiting Lysosomal Hydrolase Activity but not Autophagosome-Lysosomal Fusion
After the formation of autophagosomes, a large amount of p62 accumulates in autophagosomes and is degraded by migrating lysosomes [35]. The degradation mainly depends on the fusion of autophagosomes and lysosomes and the activity of lysosomal hydrolase [36]. We employed mRFP-GFP-LC3, a tool used for the detection of autophagosome and lysosome fusions, to trace autophagolysosome formation. In SH-SY5Y cells, LC3 (green dots) and . Compared with the control group, *P < 0.05, **P < 0.01, ***P < 0.001. Compared with the DMA group, # P < 0.05 autophagosomes (yellow dots) were significantly increased in both the DMA group and the CQ group (chloroquine, a specific inhibitor of autophagosome-lysosome fusion) compared with those in the control group (Fig. 6a-b) (P < 0.05). However, compared with the CQ group, the number of autophagolysosomes in the DMA group was significantly increased, and the number of autophagosomes was significantly decreased (Fig. 6a, c) (P < 0.05). This result suggests that DMA does not affect the fusion of autophagosomes and lysosomes.
The function of lysosomal hydrolases plays an important role in autophagy degradation. We found that compared with the control group, the catalytic activities of CTSB and CTSD in the lysosome were decreased in the 0.3-g/kg realgar group, but the difference was not significant. However, the activities of CTSB and CTSD were significantly decreased in the 0.9-g/kg realgar group (Fig. 6d-e) (P < 0.05), i.e., 79% and 74% of the control group, respectively. Next, we used LysoSensor Green DND-189 to detect the lysosomal acidic environment. The results show that after DMA exposure, the green fluorescence intensity in SH-SY5Y cells was significantly reduced in a dose-dependent manner (Fig. 6f) (P < 0.05). These results suggest that DMA can alter the lysosomal acidic environment of SH-SY5Y cells. Therefore, we speculate that realgar impairs autophagic degradation by inhibiting the activity of lysosomal hydrolase, resulting in the accumulation of p62.

Amplified p62-NRF2 Feedback Loop Is Involved in Realgar-Induced p62 Accumulation
NRF2 is a key transcription factor in antioxidant defense [37]. The activation of the NRF2-ARE signaling pathway enhances p62 transcriptional activity and promotes its production [38]. 8-Hydroxydeoxyguanosine (8-OHdG) labeling of oxidized DNA can reflect the ROS levels in cells [39]. In this study, 8-OHdG colocalized with the RBFOX3/ NeuN antibody. We found that the ROS levels were significantly increased in the 0.3-g/kg and 0.9-g/kg realgar groups ( Fig. 7a-b) (P < 0.05), i.e., 2.7-fold and 6.8-fold of that in the control group, respectively. Meanwhile, compared with the control group, the protein expression levels of NRF2 and p62 were increased in the 0.3-g/kg realgar group, but there was no significant difference. In the 0.9-g/kg realgar group, the levels of NRF2 and p62 were significantly increased ( Fig. 7c and Fig. 3c) (P < 0.05). In addition, both the ROS levels and NRF2 protein levels were dose-dependently increased after the DMA treatment in SH-SY5Y cells (Fig. 7d-e) (P < 0.001). After silencing NRF2, the expression level of the p62 protein was significantly decreased in the DMA + NRF2 siRNA group compared with that in the DMA group (Fig. 7f-g) (P < 0.01). These findings suggest that DMA activates the NRF2 signaling pathway and promotes the expression of p62.p62 can also compete with NRF2 for binding to KEAP1, forming a p62-NRF2 positive feedback loop [40]. The Co-IP results revealed that compared with the control group, DMA increased the formation of the KEAP1-p62 complex in SH-SY5Y cells (Fig. 7h) and decreased the formation of the NRF2-KEAP1 complex (Fig. 7i). After silencing p62, the NRF2 protein level in the DMA + p62 siRNA group was significantly lower than that in the DMA group (Fig. 7j-k) (P < 0.05), suggesting that DMA activates NRF2 in a p62-dependent manner. Taken together, these data suggest that realgar promotes p62 accumulation by amplifying the p62-NRF2 feedback loop.

Discussion
Realgar, a mineral Chinese medicine containing arsenic, has been historically used for 2500 years in China [1]. The Chinese Pharmacopoeia stipulates that the main functions of realgar are detoxification, insecticide, anticonvulsant, and malaria treatment, and its compound preparations are used for the treatment of encephalitis, cerebral palsy, heat clearing, and detoxification. Some classic prescriptions, such as An-Gong-Niu-Huang-Wan used to treat stroke, Niu-Huang-Jie-Du-Pian used to clear heat and detoxification, and Xiao-Er-Zhi-Bao-Dan used to anticonvulsant, all contain realgar. Researchers have analyzed 191 Chinese patent medicines containing realgar and found that approximately 86% of them are designed for oral administration with a daily intake of arsenic greater than 10 mg/kg [41], which is much more than the 15 g/day PDE limit specified by ICH [42], suggesting that a potential health hazard exists if people abuse realgar-containing medicines. The clinical long-term abuse of realgar or realgar-containing compound preparations could cause side effects manifesting as multisystem damage, including skin damage, digestive system damage, and nervous system damage [1]. The clinical manifestations of nervous system damage mainly include dizziness, headache, fatigue, limb numbness, insomnia, dreaminess, and delirium [43]. In a previous study, realgar was orally administered to mice for 8 weeks, and the dose-dependent increase in the total arsenic levels in the brain resulted in CNS toxicity, which manifested as cognitive impairment [44]. Therefore, our study used newly weaned rats as subjects to explore the toxic mechanism of realgar in the CNS and provide a theoretical basis and experimental data to guide the rational use of drugs in clinical practice and protect people's health. Studies have shown that after taking realgar or realgar-containing medicine, most realgar is excreted through feces [45], and only a trace amount of arsenic enters the body and accumulates in brain tissue, damaging the CNS of humans and animals [44,46]. Arsenic is metabolized in the liver [47]. Briefly, iAs is methylated by arsenic-3-methyltransferase . h Binding of the p62-NRF2 complex was detected by Co-IP. i Binding of the NRF2-KEAP1 complex was detected by co-IP. j Protein expression image of NRF2 after silencing p62. k Relative NRF2 protein expression level after silencing p62 (n = 3). Compared with the control group, *P < 0.05, **P < 0.01, ***P < 0.001. Compared with the DMA group, # P < 0.05, ## P < 0.05 (AS3MT) [48] using S-adenosylmethionine (SAM) as the methyl donor to form monomethylarsonic acid (MMA) [49][50][51] and dimethylarsinic acid (DMA) [52]. DMA is an end product of the metabolism of arsenic in realgar [53]. In this study, the LC-AFS method was used to detect the levels of iAs(III), MMA, DMA, and iAs(V) in rat urine, blood, and cortex. iAs(III), MMA and DMA were detected in urine, their contents exhibited a dose-dependent relationship, and iAs(V) was not detected. Only DMA was detected in blood, and its contents did not significantly differ between the 0.3-g/kg and 0.9-g/kg realgar group, indicating that the rate of metabolic production of DMA was consistent with its elimination rate. Only DMA was detected in the cortex, and its content exhibited a dose-dependent relationship. Notably, we previously detected iAs and MMA in blood and MMA in the brain using hydride generation-cold trapping-atomic fluorescence spectrometry (HG-CT-AAS). However, in this study, As(III), MMA, and As(V) were not detected in the cortex or blood. Is the detection sensitivity of LC-AFS low? Do the pretreatment conditions need to be further optimized? The specific reasons for this discrepancy need to be further explored. Undeniably, the results of the two different instruments prove that arsenic in realgar can enter and accumulate in the brain, and it is mainly DMA. Therefore, we used DMA as a treatment factor to explore the effect of realgar on nerve cells in vitro.
The effect of realgar on CNS function was investigated. In this study, a novel object recognition test and open field test were used to observe the effect of realgar on the neurobehavior of rats. Based on our results, after the realgar exposure, the rats showed a decreased ability to recognize new objects, produced anxiety-like behaviors, and exhibited a decreased exploratory ability. Our results suggest that realgar caused a blurred neuronal membrane structure, reduced mitochondrial cristae and vacuolization, damaged the double membrane of some mitochondria, and disrupted the Golgi. In particular, mitochondrial damage is very serious. Meanwhile, after exposure to realgar, the number of TUNEL-positive cells was increased, and the apoptosis rate was significantly increased. The apoptosis-related proteins Bax/Bcl-2 and cleaved-caspase-9/caspase-9 in the 0.9-g/ kg realgar group were significantly increased. This finding shows that realgar promotes apoptosis in cortical cells and has CNS toxicity, but its specific mechanism still needs to be further explored.
Previous studies have shown that realgar disrupts autophagy and can lead to CNS toxicity [10]. Autophagy is an important regulatory mechanism maintaining cellular homeostasis [54], and CNS injury induced by an imbalance of autophagic flux homeostasis is related to neurodegenerative diseases [55]. In vivo, as the level of realgar increased, the protein expression levels of LC3II/LC3I and p62, which are important indicators used to monitor the formation of autophagosomes, were significantly increased in the cortex. The accumulation of autophagosomes was observed in SH-SY5Y cells treated with DMA for 4 h and 6 h, indicating that realgar disturbed autophagic flux homeostasis in the cortex. In vivo, when SH-SY5Y cells were treated with different concentrations of DMA, the cell viability was significantly decreased. At 10 mM, the cell viability was 55.6%, and the expression level of p62 increased in a dose-dependent manner. In the time analysis, the results showed that the expression level of p62 was significantly decreased after 1 h of the DMA treatment but was significantly increased at 24 h. This finding further confirms that realgar can promote the expression of p62 by disturbing the homeostasis of the autophagic flux. Recent studies have shown that p62, a marker protein of autophagic degradation, can activate the apoptotic pathway through the caspase-8 cascade and increase cleavedcaspase-9 cleavage. p62 plays an important role in regulating the process of cell death or survival [56]. After silencing p62 in SH-SY5Y cells, DMA-induced Bax/Bcl-2 and cleavedcaspase-9/caspase-9 were significantly reduced, indicating that DMA can promote neuronal apoptosis by promoting p62 overexpression. Taken together, these results suggest that realgar can perturb autophagic flux homeostasis and mediate p62 overexpression to promote neuronal apoptosis.
Autophagy is a highly dynamic process that includes autophagy induction, autophagosome formation, autophagosome fusion with lysosomes, and the degradation stage in lysosomes [57,58]. Autophagy induction is often controlled by the mTOR-dependent PI3K/Akt pathway [59] and mTOR-independent JNK/Vps34 pathway [60,61]. Li found that realgar induces oxidative stress in liver tissue, activates the JNK/c-Jun signaling pathway, and then regulates the release of inflammatory factors [62]. Our results show that as the level of realgar increases, the protein expression levels of p-JNK, p-c-Jun, Beclin1, and Vps34 in the cortex were significantly increased. The binding capacity of the Beclin1/Vps34 complex was enhanced in the DMA-treated SH-SY5Y cells at 24 h. These results suggest that realgar exposure activates the JNK/c-Jun signaling pathway and upregulates the Beclin1/Vps34 complex. After the intervention with the JNK-specific inhibitor SP600125, the protein expression levels of p-c-Jun, Beclin1, Vps34, and LC3II induced by realgar were significantly decreased, the interaction ability of Beclin1/Vps34 complexes was significantly weakened in vivo, and the formation of Bec-lin1/Vps34 complexes was significantly reduced after the DMA treatment for 24 h in vitro. This finding indicates that realgar activates autophagy induction by activating the JNK/Vps34 complex pathway. However, after the SP600125 intervention, the expression level of p62 induced by realgar decreased with no significant difference. This finding further suggests that in the case of realgar-activated autophagy induction, autophagic degradation is impaired, thereby perturbing autophagic flux homeostasis and promoting p62 aggregation.
Efficient autophagy degradation is critical for maintaining autophagic flux homeostasis in the presence of enhanced autophagy induction. The following two types of deficits in autophagy-mediated degradation have been identified: (i) inhibition of autophagosome and lysosome fusion [63] and (ii) disruption of lysosomal degradation function [64]. mRFP-GFP-LC3B is a tool used to detect autophagosome and lysosome fusion. CQ is a specific inhibitor that inhibits autophagosome and lysosome fusion. We transfected GFP-mCherry-LC3 and used CQ as a positive control. The results show that CQ significantly reduced the number of autophagolysosomes, while DMA significantly increased the number of autophagolysosomes, indicating that DMA did not affect the fusion of autophagosomes and lysosomes.
Lysosomes contain a variety of acid hydrolases that can degrade a variety of endogenous and exogenous macromolecules and directly respond dynamically to autophagy [65]. The function of the lysosome is mainly determined by the activity of lysosomal hydrolase and the acidic environment of the lysosome [66]. CTSB and CTSD play important roles in the cathepsin family [67,68]. Lysosomal dysfunction due to lysosomal hydrolase dysfunction is strongly linked to neuropathology [27]. As the level of realgar increases, the activity of CTSB and CTSD in the cortex significantly decreased in vivo, and the lysosomal acidic environment was significantly altered. Taken together, these results suggest that realgar impairs autophagic degradation by inhibiting lysosomal hydrolase activity rather than autophagosome-lysosome fusion, leading to p62 accumulation.p62 has multiple functional domains and can interact with other proteins to mediate a variety of cellular functions. In addition to autophagy, p62 plays an important role in antioxidant responses [40,69]. The KIR domain of p62 interacts with KEAP1 to regulate the expression of NRF2, a key antioxidant factor. Activated NRF2 upregulates the level of p62 mRNA and promotes the synthesis of the p62 protein to form a p62-NRF2 feedback loop [40]. Our results show that with increasing doses of realgar exposure, both the ROS levels and NRF2 protein expression levels in the cortex were significantly increased. Similarly, the ROS levels and NRF2 protein expression levels in SH-SY5Y cells were also significantly increased. Meanwhile, the expression level of p62 was significantly reduced after silencing NRF2 in SH-SY5Y cells. This finding indicates that realgar induces NRF2 activation and promotes p62 production. In addition, the binding of the p62-KEAP1 complex was enhanced, and the binding of the NRF2-KEAP1 complex was weakened after the DMA treatment of SH-SY5Y cells. The expression level of NRF2 was significantly reduced after silencing p62 in SH-SY5Y cells. These results suggest that NRF2 is activated in a p62-dependent manner and that NRF2 activation indirectly increases p62 transcriptional activity, activating the p62-NRF2 feedback loop, which, in turn, promotes p62 accumulation in DMA-exposed SH-SY5Y cells. However, the increase in the ROS levels may be directly induced by realgar exposure or may be closely related to realgar-induced protein accumulation, autophagosome accumulation, or mitochondrial damage, which still needs to be further explored.
In summary, our results suggest that realgar promotes the accumulation of p62 by disturbing the autophagic flow and crosstalk of the p62-NRF2 feedback loop, leading to apoptosis and CNS toxicity. On the one hand, realgar induces autophagy by activating the JNK/Vps34 complex pathway and recruits a large amount of p62 aggregates; on the other hand, realgar impairs autophagic degradation by reducing the activity of lysosomal hydrolase and changing its acidic environment, reducing the degradation of p62 and leading to the excessive accumulation of p62. Moreover, the amplified p62-NRF2 feedback loop promotes the accumulation of p62 in this process. The accumulation of p62 eventually induces neuronal apoptosis. These results provide a new perspective and experimental data for the study of the mechanism of realgar-induced CNS toxicity and are also of great significance for clinical guidance of the rational use of realgar.