Ferroptosis Promotes Microtubule-Associated Protein Tau Aggregation via GSK-3β Activition and Proteasome Inhibition

Ferroptosis is a form of regulated cell death resulting from iron accumulation and lipid peroxidation. In some particular brain regions, iron dyshomeostasis and peroxidation damage of neurons are closely related to a wide range of neurodegenerative diseases known as “tauopathies”, in which intracellular aggregation of microtubule-associated protein tau is the common neuropathological feature. However, the relationship between ferroptosis and tau aggregation is not well understood. The current study demonstrates that erastin-induced ferroptosis can promote tau hyperphosphorylation and aggregation in mouse neuroblastoma cells (N2a cells). Moreover, ferroptosis inhibitor ferrostatin-1 can alleviate tau aggregation effectively. In-depth mechanism research indicates that activated Glycogen synthase kinase-3β (GSK-3β) is responsible for abnormal hyperphosphorylation and accumulation. More importantly, proteasome inhibition can exacerbate the tau degradation obstacle and accelerate tau aggregation in the process of ferroptosis. Our results indicate that ferroptosis can lead to abnormal aggregation of tau protein and might be a promising therapeutic target of tauopathies.


Introduction
Ferroptosis is a recently discovered form of programmed cell death whose execution requires the accumulation of lipid peroxidation in an iron-dependent manner [1,2]. Ferroptosis is involved in diverse biological contexts and is genetically, biochemically, and morphologically distinct from apoptosis and necrosis. Interestingly, both the iron overload and lipid peroxidation in the brain are also closely related with the characteristic protein aggregation such as tau protein in Alzheimer's disease (AD) and αsynuclein in Parkinson's disease (PD), it highlights the associated research aiming at ferroptosis and neurodegenerative disease [3][4][5][6].
Tau protein, a highly soluble, microtubule-associated protein, whose primary function is to promote microtubule assem bly and stabilization, as well as maintain axonal transport and neuron survival [7,8]. As a phosphoprotein, phosphorylation modi cation can negatively regulate its ability to stimulate microtubule assembly [9]. Moreover, hyperphosphorylation is assumed to be critical for the formation of tau aggregation [10] and alter its degradation through the proteasome or autophagy [11,12]. Besides the role in stabilizing neuronal microtubules [13], promoting axonal outgrowth [14], and regulating axonal transport [15], tau also functions to facilitate neuronal iron e ux. Lei et al. have reported that tau de ciency (i.e., the loss of soluble tau) can impair APP-mediated iron export then lead to neurotoxic iron accumulation and neurodegeneration [16,17].
Moreover, the aggregated tau protein in neuro brillary tangles (NFTs) is associated with the induction of heme oxygenase-1 (HO-1) [18], which can oxidize cellular heme to generate biliverdin, free iron, and carbon monoxide (CO). The released ferrous iron subsequently catalyzes the Fenton reaction to generate high levels of endogenous reactive oxygen species (ROS), which perpetuate the intracellular oxidative stress of the brain [19,20]. These reports indicate that tau protein plays a crucial role in regulating cellular iron homeostasis, and tau abnormality is closely related to intracellular iron overload.
Iron is a crucial participant and regulator of neuronal physiological functions [21]. Iron is an indispensable cofactor for proteins and enzymes required to develop synapses, myelination, and neurotransmitter metabolism [22]. However, iron accumulates in the brain with age, associating with neurodegenerative disorders, including tauopathies such as AD [23]. Interestingly, it was also found that iron overload could induce tau hyperphosphorylation via the dysfunctional insulin signal in primary cultured neurons and mice [24]. Meanwhile, it has been extensively reported that iron overload generates redox-generated free radicals [25], which acts as a pro-oxidant to convert hydrogen peroxide into hydroxyl radicals through the Fenton reaction. The hydroxyl radical then attacks proteins or biological macromolecules [26], results in the formation of oligomeric tau via cysteine-cysteine binding [27] or via kinase pathways [28]. In addition, iron can also produce oligomeric tau by forming intermolecular coordination complexes mediated by phosphorylated amino acid residues [29,30].
All the above researches indicate that intracellular iron accumulation, lipid peroxidation, and tau aggregation are closely related, and a vicious cycle might be formed between iron overload and tau aggregation directly or indirectly by lipid peroxidation. It aroused our interest to unveil the relationship between ferroptosis, the iron-dependent form of regulated necrosis involved in lipid peroxidation, and tau aggregation. To address this directly, we treated N2a neuroblastoma (a neuron-like cell line) with erastin, which could inhibit the activity of cysteine-glutamate antiporter (system Xc -), leading to the depletion of the major cellular antioxidant glutathione (GSH), then decreased the activity of GPX and triggered ferroptosis [31,32]. The relationship between ferroptosis and tau aggregation was explored, with the associated mechanisms investigated in detail. Considering that intracellular iron overload and ROS accumulation play an essential role both in the process of ferroptotic cell death and in the pathogenesis of tau-related neurodegenerative diseases, our study indicates the possible new strategy for tauopathies therapy through the pharmacological interference with the ferroptosis processes of brain issues.

Cell culture and drug treatments
Mouse neuroblastoma Neuro2a (N2a) cells were cultured in a 5% CO 2 humidi ed incubator at 37 °C with a medium consisting of 45% Dulbecco's modi ed Eagle's medium (DMEM), 45% Opti-MEM supplemented with 10% (v/v) fetal bovine serum. For drug treatment, 5, 10 and 20 μM erastin were used to treat cells for 6 and 12 hr to induce ferroptosis. During the incubation with 10 μM erastin, cells were either treated or not treated with 1, 2 and 4 μM ferroptosis inhibitor Ferrostatin-1 for 6 and 12 hr. The cells were treated with 0.2% dimethyl sulfoxide (DMSO) as a control.

Cell viability analysis
Cell Counting Kit-8 (Beyotime Institute of Biotechnology, Jiangsu, China) was used to determine cell viability. In brief, cells were seeded in 96-cell micro-culture plate at an initial density of 5000 cells per well and allowed to grow for 48 h in an incubator maintained at 37 °C. The cells were treated with corresponding drugs for 6 and 12 hr respectively, then 10 μL reagent was added to each well, and cells were incubated in the incubator for 2 hr. Finally, the plates were scanned with a microplate reader (Bio-Rad, Hercules, CA, USA) at 450 nm to measure the absorbance.

Intracellular glutathione peroxidase (GPX), reactive oxygen species (ROS), and iron content detection
Following drug treatment, cells were washed once with cold PBS. They were lysed directly in an Eppendorf tube with lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 0.1% sodium dodecyl sulfate(SDS), 1% Nonidet P-40, 0.5% sodium deoxycholate,0.02% NaN 3 , 100 μg/ml phenylmethylsulfonyl uoride, 2 μg/ml Aprotinin) for 30 min and centrifugation for 15 min at 12000 ×g. The supernatant was removed, and protein concentration was determined. The relative indicators were then measured using commercial assay kits. In brief, the ROS levels were measured by adding 100 μL of protein supernatant and 100 μL of 2'7'-dichloro uorescein diacetate (DCFH-DA) to a 96-well plate, which was incubated for 10 min at 37 °C in the dark. Fluorescence detection was then carried out using an excitation wavelength of 488 nm and an emission wavelength of 525 nm. GPX and iron levels were assessed following the manufacturer's instruction (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), and the samples were incubated for 30 min at room temperature. The absorbance was measured at 340 nm to calculate the GPX level and calculate the iron content at 590 nm.

Cell morphology observation
N2a cells were seeded in the 24-well plate. When cells grew to 80% abundance, they were treated with erastin, erastin plus ferrostatin-1 for 6 and 12 hr, respectively. The morphology of the N2a cells with different treatments was observed using a phase-contrast microscope as in our previous study [33]. The obtained images were processed with image processing software, Image J, where neurites were manually traced with a Wacom tablet to measure the length. Average lengths of the neurites were then obtained by dividing the total length by the number of neurons measured.

Western Blot
Cells were seeded in six-well plates and treated as described above. The protein concentration was determined using the BCA protein assay. For Western blot analysis, twenty micrograms of total protein were separated using 10% SDS polyacrylamide gels and then transferred by electroblotting onto polyvinylidene di uoride membranes. After the membranes were blocked with 5% nonfat milk for 1 hr, washing the membranes three times with PBS (containing 0.1% Tween-20) for 10 min each time.
Membranes were probed overnight at 4 °C with antibodies. The immunoblots were washed and treated with the appropriate species of horseradish peroxidase (HRP)-conjugated secondary antibody at 37 °C for 1 hr and then wash the membranes three times with PBS. The immunoreactive bands were quantified using an Infrared Imaging System.

Immuno uorescence cytochemistry
N2a cells were mounted on coverslips in a 24-well plate. After drug treatment, cells were xed with Precooled methanol for 10 min at -20 °C. After washing with 0.2% Triton-PBS and 0.5% Triton-PBS, respectively for three times, cells were incubated for 1hr in blocking buffer (5% BSA), followed by incubation for 48 hours with primary antibodies at 4 °C (tau-5, 1:100; vimentin, 1:100). The coverslips were then washed three times with 0.1% Triton-PBS and incubated in the second antibody for 1 hr at 37°C . The coverslips were counterstained with Hoechst and washed three times with Triton-PBS. Finally, cells on the coverslips were observed and imaged using a uorescence microscope.

Measurement of 20S Proteasome Activity
According to the manufacturer's instructions and our previous literature [34], the 20S proteasome catalytic core chymotrypsin-like activity kit was measured using the 20S proteasome activity assay kit (Boston Biochem, Cambridge, MA, USA). In brief, following the above drug treatment, the cells were collected, resuspended with the 20S proteasome assay buffer, then centrifuged at 8000 ×g for 8 min, and the precipitation was collected. The precipitation was treated with the 20S proteasome lysis buffer for 30 min at room temperature. The cytosolic extract was subsequently collected as the supernatant following centrifugation for 12 min at 12000 ×g. For the proteasome activity assay, the cell lysates were incubated with chymotrypsin uorogenic substrate Suc-Leu-Leu-Val-Tyr-AMC. The chymotrypsin-like activity was determined to increase of the uorescence intensity in the reaction products by using the Synergy 2 Multi-Mode Microplate Reader (Bio Tek, Winooski, VT, USA) at 360 nm excitation and 480 nm emission wavelengths [35].

Immunoprecipitation
Cells were seeded in six-well plates and treated as described above. After one wash with cold PBS, the cells were lysed directly with RIPA lysis buffer (50 mM Tris-HCL, 150 mM NaCl, 5 mM EDTA, 1% Triton-X, 1% DOC, 0.1% SDS, 0.2 mM TPCK, 1 mM PMSF, 2 mM TLCK) for 30 min on ice, and centrifuged at 12000 ×g for 20 min subsequently. Add protein G to the supernatant at a ratio of 100:1 between the supernatant and protein G, and shake it slowly on a shaker at 4 °C for 30 min to remove non-speci c protein binding. After centrifugation at 12000 ×g for 15 min, an anti-tau-5 antibody was added to the sample according to the antibody ratio to sample at 1:100 and incubated overnight on a shaker at 4 °C. After incubating the primary antibody overnight, add protein G at a ratio of 5:1 between the supernatant and protein G, slowly incubate in a low-temperature studio for 2 hr and resuspend the pellet in loading buffer. After boiling for 10 min and centrifuging at 12000 ×g for 1 min, the protein concentration was determined using the BCA protein assay.

Statistics Analysis
All experiments were performed in triplicate. Data were expressed as mean±SD and analyzed using SPSS 27.0 statistic software. The One-Way ANOVA procedure followed by LSD's post hoc tests was used to determine the different means between groups (p<0.05).

Erastin decreases the viability of N2a cells
To determine the effect of the ferroptosis inducer erastin on the viability of N2a cells, we treated N2a cells with different concentrations of erastin for 6 and 12 hr. It was shown that the cell viability decreased in a concentration and time-dependent manner following erastin treatment. Compared to the control group, 5, 10 and 20 μM erastin treatment for 6 hr decreased the relative cellular viability by 7.35%, 12.72% and 12.24%, respectively. What's more, the relative cell viability decreased by 17.21%, 24.53% and 24.77%, respectively, when cells were treated with 5, 10 and 20 μM erastin for 12 hr (Fig. 1A). As there was no signi cant difference in the activity inhibition between the concentrations of 10 μM and 20 μM, we selected 10 μM erastin as the optimal treatment concentration in the subsequent experiments. We also found that 2 or 4 μM ferrostatin-1 (Fer-1) treatment, an effective inhibitor of ferroptosis, which can inhibit the accumulation of ROS then inhibit oxidative lipid damage and cell death [36], can effectively reverse erastin induced cell viability inhibition (Fig. 1B).

Erastin treatment induces intracellular ROS accumulation and iron overload in N2a cells
Erastin can induce system Xc − inhibition and cysteine deprivation, then lead to GSH deletion and GPX inactivation, nally initiating ferrotosis in many cell types. To con rm the sensitivity of N2a cells to erastin-induced ferroptosis, we assayed the intracellular GPX activity, ROS level and iron content after erastin treatment.
The results showed that compared with the control group, N2a cells treated with 10 μM erastin resulted in a signi cant decrease of intracellular GPX activity after 6 ( Fig. 2A) and 12 hr (Fig. 2B). Meanwhile, the ROS content (Fig. 2C, 2D) and iron levels increased signi cantly (Fig. 2E, 2F). The results also demonstrated that 2 μM ferrostatin-1 treatment could signi cantly reverse erastin induced GPX, ROS and iron imbalance. These results con rmed that erastin could induce ferroptotic cell death in N2a cells.

Erastin treatment caused morphological abnormality of N2a cells
The unique ability of neurons is to compute and allocate massive information, which relies on their polarized morphology. Damage factors such as peroxidation can cause microtubule depolymerization and neuron death [37]. Therefore, we observed the morphology changes of N2a cells following erastin exposure by using phase-contrast microscopy.
To the DMSO treated cells, the cell bodies showed tapered or with clear edges, and the neurites were stretched. However, after erastin treatment for 6 and 12 hr, the cell bodies became round, and the neurites retracted obviously (Fig. 3A). Compared with the erastin treatment group, the cell bodies of the erastin plus ferrostatin-1 co-treatment group recovered the polarity gradually. Most of the cell bodies were tapered, and the length of the neurites partially increased (Fig. 3B). The above results showed that erastin exposure could dramatically destroy the microtubule structure and inhibit the neurite outgrowth in N2a cells.

Erastin induced abnormal aggregation of tau protein
The microtubule is a major cytoskeletal component of neurites, and the regulation of microtubule stability is essential for neurite morphogenesis. Microtubule-associated protein tau plays a crucial role in establishing cell polarity and maintaining microtubule stability [38]. The dissociation of tau from microtubules would impede the microtubule assembly and disrupt the stability of the microtubule network, then destroy the process of axonal transport and synaptic plasticity. Meanwhile, the dissociated tau would self-aggregate and form intracellular inclusions. Based on the result that erastin treatment destroyed cytoskeleton structure signi cantly, to clarify whether the abnormality of neuronal morphology is related to tau protein, we double-stained the cells with antibody tau-5 (identi ed total tau) and vimentin (a known marker of aggregate-forming protein).
After N2a cells were treated with erastin for 6 and 12 hr, tau protein aggregated obviously around the nucleus ( Fig. 4A and 4B. e, red), the strongly positive staining of vimentin ( Fig. 4A and 4B. f, green) indicated that erastin could lead to the formation of intracellular protein aggregates. Furthermore, the colocalization of tau protein (red) and vimentin (green) around the nucleus suggested that the aggregates contain tau protein ( Fig. 4A and 4B. h, orange). It was also shown that the aggregated tau protein disappeared when cells were treated with erastin plus ferrostatin-1( Fig. 4A and 4B. l, orange). The above results demonstrate that erastin treatment would induce abnormal aggregation of tau protein in the cytoplasm.

Erastin remarkably increases the content of total tau and phosphorylated tau
Tau function and a nity for the microtubules mainly depend on its phosphorylation status. Hyperphosphorylation of tau protein at some key sites such as Ser396 and Thr231 can decrease the a nity between tau and microtubules, then cause the dissociation of tau from microtubules and the formation of intracellular tau aggregates [39,40]. In the present study, we detect the content of total tau and phosphorylated tau at Ser396 and Thr231 sites by Western blot (Fig. 5). It was shown that after erastin treatment for 6 and 12 hr, the intracellular total tau increased signi cantly. Meanwhile, the relative level of phosphorylated tau at Ser396 and Thr231 sites also upregulated dramatically, especially in the 6 hr erastin treatment group.
Interestingly, 2 μΜ ferrostatin-1 could effectively reverse erastin induced increase of total tau and phosphorylated tau. Statistical analysis revealed a positive correlation between the ROS content and the iron level (r=0.65), and we also found ROS induced a highly positive correlation with tau phosphorylation at Ser396 (r=0.82) and Thr231sites (r=0.76) (Fig. 6I). These results indicate that during neuronal ferroptosis, tau was hyperphosphorylated at Ser396 and Thr231 sites.

Erastin activated GSK-3β in N2a cells
Most studies on tau hyperphosphorylation explored the kinase-phosphatase imbalance and regarded GSK-3β as the signi cant kinase involved in tau pathology [41]. GSK-3β mediated tau phosphorylation is observed in areas that are in close proximity to the microtubule-binding domains and their amino acid residues [42]. Therefore, the hyperphosphorylated tau induced by GSK-3β is prone to dissociate from microtubules and self-aggregate in the cytoplasma [43]. Given that GSK-3β is one of the most important kinases that regulate tau phosphorylation at Ser396 and Thr231 sites [44], to clarify the underlying mechanisms of tau hyperphosphorylation in the process of erastin-induced ferroptosis, we detected the level of total GSK-3β and Ser-9-phosphorylated GSK-3β (negative related to GSK-3β activity). As shown in Figures 6A and 6C, after 6 and 12 hr erastin exposure, the total level of GSK-3β showed no signi cant change, while the relative level of Ser9-phosphorylated GSK-3β (p-GSK-3β) decreased with time.
Compared to the control group, the relative phosphorylated GSK-3β at Ser9 decreased to ~ 40% and 30%, respectively. Meanwhile, 2 μΜ ferrostatin-1 treatment effectively rescued erastin induced GSK-3β activity upregulation. These results suggest that GSK-3β activation is involved in the erastin induced tau hyperphosphorylation in N2a cells.

Effect of erastin on the content of Poly-Ub and Mono-Ub
Based on the above results that erastin treatment produces tau aggregates (Fig. 4) and elevates the total tau level (Fig. 5), we wonder whether degradation impairment contributes to erastin induced tau accumulation. Intracellular misfolded proteins that fail to be rescued by molecular chaperones are sent for degradation by proteasome and autophagy. The recognition signal for degradation of aggregated proteins by both the proteasome and autophagy is driven by polyubiquitin chain labeling of substrates [45] which depends on the ubiquitin pool homeostasis. To determine whether the metabolic abnormality of tau protein is associated with the ubiquitin cycle, we rst detected the content of polyubiquitin (Poly-Ub) and monomer ubiquitin (Mono-Ub) in N2a cells. It was shown that after the erastin treatment for 6 and 12 hr, the intracellular poly-Ub level increased signi cantly, while the level of mono-Ub decreased obviously (Fig. 7A, 7D). We also observed that ferrostatin-1 treatment could successfully reverse erastin induced ubiquitin cycle impairment. The above results indicate that during the process of erastin induced ferroptotic cell death, the degradation of polyubiquitin-labeled substrates was inhibited.

Page 9/29
3.8 Erastin induced abnormal ubiquitination modi cation of tau protein As pathologically modi ed proteins such as hyperphosphorylated tau can be degraded by ubiquitinproteasome system [46] and autophagy-lysosome pathway [47], and polyubiquitin label is the prerequisite of proteins degradation by these two systems. In view of the signi cant down-regulation of monomer ubiquitin and apparent up-regulation of poly-ubiquitin after erastin treatment, we further tested the effect of erastin on the ubiquitination of tau by immunoprecipitation. Following incubating with the tau-5 antibody (analyzed total tau proteins), the samples from different treatment groups were used for immunoblot analysis with tau-5 and anti-ubiquitin antibody (probed ubiquitination-conjugated proteins).
It was shown that compared to the control group, the ubiquitinated tau increased signi cantly, most so with 6 hr erastin treatment (Fig. 8). Moreover, ferrostatin-1 could rescue erastin induced association abnormality between ubiquitin and tau protein. The above results indicate that erastin can cause polyubiquitinated tau protein degradation impairment, which may be involved in the abnormal aggregation of tau.

Erastin treatment inhibited 20S proteasome activity in N2a cells
The proteasome is the main route for misfolded proteins degradation, and the 20S proteasome is a complex containing activities similar to trypsin, chymotrypsin, and caspase. As chymotrypsin-like activity inhibition can cause the protein degradation e ciency of the proteasome to be signi cantly reduced [48], we focused on the chymotrypsin-like activity alteration of the proteasome after erastin treatment. Figure 9 shown that chymotrypsin activity decreased signi cantly in a time-dependent manner after 6 and 12 hr erastin treatments, and 2 μM ferrostatin-1 dramatically mitigated erastin induced proteasome activity down-regulation. The above results indicate that the activity of the proteasome system is inhibited during ferroptosis and might be contributing the tau aggregation.

Erastin treatment activated the autophagy pathway
Considering that the autophagy-lysosomal system can also degrade abnormally aggregated tau protein and a lipidated form of LC3, LC3-II, is an autophagosomal marker in mammals, we further analyzed the relative ratio of LC3-/ in N2a cells after erastin treatment. It was shown that the intracellular LC3-/ ratio increased about 10% at 6 hr and 20% at 12 hr than the controls and showed clear time-dependence (Fig. 10). The above results indicated that in the rst 12 hours of erastin treatment, proteasome activity is inhibited. However, the autophagy-lysosome system is compensatively activated to maintain intracellular protein balance. But, obviously, the activation of the autophagy-lysosome system is insu cient to degrade the abnormally aggregated tau protein.

Discussion
Abnormal deposition of aggregated microtubule-associated protein tau is a common nal pathway of tau-related neurodegenerative diseases named "tauopathies", including Alzheimer's disease(AD), progressive supranuclear palsy (PSP), Huntington disease (HD), frontotemporal dementia with parkinsonism-17 (FTDP-17), and so on. It clari ed that the factors that promote tau aggregation are important to elucidate the pathogenesis and acquire a good curative effect of these diseases. Iron has been shown to promote aggregation of some neurodegenerative diseases characteristic aberrant proteins, such as tau [49,50] and β-amyloid [51]. Further support for the involvement of iron in the pathogenesis of tauopathies is provided by the recent discovery of a new form of cell death, ferroptosis. Arising from iron-dependent lipid peroxidation, ferroptosis is augmented in conditions of cysteine depletion and glutathione peroxidase (GPX) inactivation [20,52,53,26]. What's more, it has been widely observed that many neurodegenerative diseases such as AD are pathologically characterized by the accumulation of iron in the brain and peroxidation damage of neurons [54]. However, the detailed relation between ferroptosis and abnormal tau aggregation is still unclear.
In the present study, ferroptotic cell death was constructed by treating N2a cells with ferroptosis inducer erastin for 6 hr and 12 hr. It was shown that erastin exposure signi cantly inhibited the viability of N2a cells, prominently decreased the intracellular GPX activity (a key marker of ferroptosis),elevated ROS and iron content. What's more, 2 μM ferroptosis inhibitor ferrostatin-1 can effectively rescue the cytotoxic effect caused by erastin. The above results con rm that erastin can induce ferroptosis in N2a cells. Meanwhile, it was also found that erastin treatment caused noticeable morphology change of N2a cells, especially signi cantly shortened neurites, which means the neuronal skeletal system was impaired during ferroptosis. As tau protein is one of the most important microtubule-related proteins in neurons and plays a key role in maintaining the stability of the neuronal skeletal system [2, 55 ,56], the dissociation of tau from microtubules can induce not only the collapse of cytoskeletal structure but also the formation of intracellular tau aggregates. Naturally, the obvious neurites retraction aroused our interest to assay the status of tau protein during ferroptosis. Interestingly, the immuno uorescence cytochemistry results showed clear tau-positive aggregates in the cytoplasm after erastin treatment, and it directly demonstrated the causal relationship between ferroptosis and abnormal tau aggregation.
Based on the above results, we further explored the mechanisms involved in tau aggregation during ferroptosis. It was accepted widely that several post-translational modi cations were proposed to play important roles in tau aggregation [57]. Among them, phosphorylation modi cation can negatively regulate the ability of tau to stimulate microtubule assembly [9]. It was reported that phosphorylation of tau at the Thr231 site could lead to a trans-to-cis conformational change trigger the detachment of tau from microtubules [58]. Likewise, phosphorylation of tau at Ser396 reduces its a nity for MTs and its ability to stabilize MTs against depolymerization [59]. Thus, we assayed the phosphorylation pro le of Thr231 and Ser396 sites. It was found that compared to the control group, the phosphorylation modi cation levels of tau protein at Ser396 and Thr231 sites increased signi cantly after erastin treatment for 6 and 12 hr. Considering the equilibrium disruption between tau kinase and phosphatase activities is the major reason responsible for the hyperphosphorylation of tau protein, we further measured the content and activity of GSK-3β, a crucial tau kinase corresponding to 70% of tau pathological phosphorylation sites including Ser396 and Thr231 sites [60]. We found that the level of GSK-3β was not signi cantly changed after erastin treatment, but the reduction of phosphorylated GSK-3β at Ser9 suggested that erastin could induce GSK-3β activation.
Interestingly, Wu et al. also found that GSK-3β silence can block erastin-induced ferroptosis with less production of ROS, and GSK-3β overexpression can exacerbate erastin induced ferroptosis in breast cancer cells [45]. These results clearly demonstrate the relationship between ferroptosis and GSK-3β activation. Considering that GSK-3β dysregulation almost in uences all the major hallmarks of tauopathies, including tau hyperphosphorylation and aggregation [42,61,62], synaptic dysfunction, and memory impairment [63], clarifying its contribution to the ferroptosis will be the central objectives of further study and may provide a viable therapeutic approach to prevent or halt tauopathies.
Erastin induced tau aggregation strongly indicated a concomitant loss of proteostasis, which is maintained by several systems in the cell, including the ubiquitin proteasome system (UPS) and autophagy-lysosome system. Given the two protein degradation pathways both required ubiquitin label of substrates [64-67], we rst assayed the content change of poly-ubiquitin and mono-Ub after erastin treatment and found that intracellular poly-ubiquitin obviously increased, but the level of mono-Ub decreased signi cantly. Moreover, erastin treatment increased the amount of ubiquitinated tau protein signi cantly. These results suggested that ferroptosis would lead to tau degradation impairment.
As the ubiquitin-proteasome system is the principal route for the degradation of intracellular misfolded or damaged proteins [68], including tau protein [69], we further explored the mechanistic research and found that chymotrypsin-like activity of 20S proteasome, which directly affects the degradation e ciency of the proteasome [48], decreased in a time-dependent manner after erastin treatment. As proteasome cannot adequately handle misfolded protein aggregates [70], what's more, aggregates of some pathogenic proteins can directly inhibit proteasome activity [71]. These data together suggested that in the process of ferroptosis, hyperphosphorylated tau can't be effectively degraded by the proteasome system and even can induce proteasome activity inhibition.
In addition to the proteasome, autophagy also plays an important role in removing aggregated proteins [72,73], including tau proteins [74]. But whether autophagy is involved in tau clearance during erastin-induced ferroptosis is not demonstrated. LC3-I and LC3-II are two important biomarkers in autophagy, the conversion of LC3-I to LC3-II indicates autophagy activation [75]. In this study, the increased LC3-/ ratio suggested an enhanced autophagic ux following erastin administration. The founding is consistent with previous studies that erastin induced LC3 conversion and autophagy activation [76]. Since there is a contradiction between autophagy activation and tau accumulation in the process of erastin-induced ferroptosis, we consider that proteasome inhibition might play a more critical part than autophagy in the pathological aggregation of tau protein.
In conclusion, our data suggest that in the process of erastin-induced ferroptotic cell death, microtubuleassociated protein tau undergoes hyperphosphorylation and aggregation. The pathological changes of tau are related to the activation of GSK-3β and the inhibition of the proteasome system (Fig. 11). It provides a clear clue about the relationship between ferroptosis and tau aggregation, opens new therapeutic targets and strategies for tauopathies. Availability of data and materials The data are available upon reasonable request from the author for correspondence.    Erastin induced tau aggregation in N2a cells. The staining of tau protein (red) and vimentin (green) in N2a cells treated with erastin (10 μΜ) or erastin plus 2 μΜ ferrostatin-1 for 6 hr (A) and 12 hr ( B ). Red uorescence represents tau protein labeled with tau-5 antibody (a, e, i); green uorescence represents the vimentin (b, f, j); blue uorescence represents the nucleus of DAPI labeled cells (c, g, k), arrows represent tau protein aggregates. Scale bar = 20 μm. Erastin increased the level of total tau and phosphorylated tau in N2a cells. A and E, representative blots of total tau and phosphorylated tau-Ser396 and Thr231 after cells were treated with 10 μM erastin for 6 hr (A) and 12 hr (E) respectively. B-D, Quantitative analysis of the relative levels of tau-5, p-tau at Ser396, and p-tau at Thr231 after treating for 6 hr. F-H, Quantitative analysis of the relative levels of tau-5, p-tau at Ser396, and p-tau at Thr231 after treating for 12 hr. I, the correlation coe cients between ROS content, represented as mean ± SD. **p<0.01, as compared with the control group; ##p<0.01 as compared with the erastin treatment group.

Figure 7
The effects of erastin on ubiquitin levels in N2a cells. The intracellular Poly-Ub and Mono-Ub content changes were detected by Western blot after erastin, or erastin plus ferrostatin-1 treatment for 6 hr and 12 hr, respectively. A, B, and C, the relative levels of Poly-Ub and Mono-Ub after 6 hr. D, E, and F, the relative levels of Poly-Ub and Mono-Ub after 12 hr. The values were expressed as fold changes over control and are represented as mean ± SD (n=3). **p<0.01, as compared with the control group; ##p<0.01 as compared with the erastin treatment group.

Figure 8
Erastin treatment increased ubiquitinated tau in N2a cells. Extracts from cells were immunoprecipitated with anti-IgG control and anti-total tau antibody (tau-5). The immunoprecipitation was detected with an anti-ubiquitin antibody and tau-5. A and C, the input panel shows the presence of the studied proteins prior to immunoprecipitation in the extracts. B and D, quanti cation of the ratios of immunoprecipitated tau versus total input tau protein from three independent experiments. Values were normalized to those of the control group and represented the mean ± SD (n=3). * p<0.05, **p<0.01, as compared with the control group; #p<0.05, ##p<0.01 as compared with the erastin treatment group.  Erastin treatment activated autophagy in N2a cells. The changes of intracellular LC3-/ ratio in N2a cells were detected by Western blot after the treatment with erastin (10 μM) or erastin plus ferrostatin-1 for 6 hr (A and B) and 12 hr 6 hr (C and D). Data were expressed as the means ±SD (n=3). **p<0.01, as compared with the control group; ##p<0.01 as compared with the erastin treatment group.