Homocysteine-Potentiated Keap1 Promotes Neuronal Senescence via Inhibiting Ubiquitination of β-Catenin

Elevated serum homocysteine (Hcy) is an independent risk factor of Alzheimer’s disease (AD). It has been reported that Hcy dramatically accelerates the aging of endothelial progenitor cells or endothelial cells. However, whether and how Hcy produces neuronal senescence is largely unknown. Mouse neuroblastma 2a (N2a) cells were treated with Hcy, and senescence-associated β-galactosidase (SA-β-gal) staining was applied to assay senescence. Senescent markers and related proteins were examined by western blot, quantitative Polymerase Chain Reaction (qPCR), immunouorescence staining. Methylation of promoter was assay by bisulte sequencing PCR (BSP). Immunoprecipitation (IP) was applied to examine association between proteins. Rats were injected with homocysteine and examined neuronal senescence. like between β-catenin-WTX promoter’s CqG island to high β-catenin. Hcy-increased competed with β-catenin Knockdown β-catenin and Keap1 both attenuated Hcy-induced senescence of N2a Hcy-induced also showed along with elevated senescent markers.


Introduction
Homocysteine (Hcy) is an intermediate product of cysteine homologue and methionine metabolism and is considered a high risk factor for AD (Miller, 1999). Previous reports have shown that the AD-like animal model established by tail vein injection of Hcy almost completely characterizes the pathological characteristics of AD (Jiang et al., 2015). The animal model of Hcy reproduces the pathological features of AD, such as Tau hyperphosphorylation, β-amyloid plaque (Aβ) accumulation and neuron loss (Zhang et al., 2008;Zhang et al., 2009). In addition, it has been reported that the increase Hcy level in blood leading to hyperhomocysteinemia is related to vascular aging and accelerating the aging process, which also contributes to the progression and development of AD (Ostrakhovitch et al., 2019;Zhang et al., 2015).
Aging is a high risk for AD. The traditional view of cellular senescence is a phenotype characterized by a durable arrest of cell cycle and a collection of stress granules (Moreno-Blas et al., 2019). As a result, studies on brain cellular senescence mostly focused on glial cells (Chinta et al., 2015). Recent studies focus on cortical and Purkinje neurons found several features of senescence, such as senescenceassociated β-galactosidase (SA-β-gal) activity, γH2AX and macro-H2A foci, and lipofuscin accumulation all exist in a p21 CIP1/WAF1 -dependent manner (Kang et al., 2015). According to this criterion, it is found that human neurons will also become senescence, because p16/Cdkn2a is expressed in the prefrontal cortex pyramidal neurons in the brain of people over 77 years of age (Jurk et al., 2012). Long-term cultured rat primary cortical neurons shown the characteristics of cellular senescence before glial cells, and developed a functional senescence-related secretory phenotype, which can induce premature paracrine in mouse embryonic broblasts senescence (Ishikawa et al., 2020). While senescent cells in peripheral tissues induced by Hcy have been the focus of numerous reports, their participation in or contribution to cognitive decline with aging or neurodegenerative diseases remains largely unknown.
Particularly, Hcy could accelerate telomere attrition by increasing telomere loss per replication in vascular endothelial cells (Shin et al., 2016;Zhang et al., 2015), upregulation of p16, p21, and p53 (Curro et al., 2014), activation of β-galactosidase activity (Walsh et al., 2007), all of which are typical pathological characteristics of cellular senescence. Until now, most reports focus on homocysteine-induced cellular senescence in non-neuronal cells such as endothelial cells (McCully, 2018;Sun et al., 2019). In view of these facts, it is of great research value to study whether Hcy can induce neuronal senescence and its mechanism.
In general, the cellular Wnt/β-catenin signaling pathway is essential for regulating proliferation, cell cycle, apoptosis and axis polarity induction (Reya et al., 2005). Once activated by Wnt, the accumulated βcatenin is transferred to the nucleus, binds to transcription factors and changes the expression of downstream target genes, which will promote neuron survival and neurogenesis, and enhance synaptic plasticity (Jia et al., 2019). Previous studies indicated that the loss of function of Wnt/β-catenin signaling component promoted the occurrence and progression of AD-like symptoms, and recovery is a promising strategy to ameliorate these symptoms (De Ferrari et al., 2014;Jia et al., 2019). However, recent studies have revealed another function of β-catenin, indicating that abnormally activated β-catenin is related to the dysfunction of distal lung epithelial cells and lead to accelerated aging (Liu et al., 2007). Moreover, another study showed that intervention of the association of GSK3β reduced the phosphorylation level and degradation rate of β-catenin, and the accumulated β-catenin further induced neuronal senescence (Chow et al., 2019). In addition to GSK3β, Wilms tumor gene on X chromosome (WTX) is another factor inducing β-catenin's ubiquitination (Major et al., 2007). Besides, another study found that as blood Hcy levels increased, β-catenin accumulated in the nucleus of the vascular endothelium (Beard et al., 2012).
Since it is predicted that the senescence marker p53/p21 mediates the senescence of normal β-catenin cells, it is reasonable to propose that Hcy promotes cell senescence by activating β-catenin .
In present study, we rst investigated SA-β-gal activity, p16, p21 and p53 in mouse neuroblastoma N2a cells after induction by Hcy. Then β-catenin and related proteins were evaluated. Furthermore, WTX and Keap1 were investigated to reveal underlined mechanism of how β-catenin was activated.
Establishment of homocysteine rat model Two months' old male SD rats were purchased from the animal center of Tongji Medical College, Huazhong University of Science and Technology. All rats were kept at 23±2°C on daily 12-h light-dark cycles with ad libitum access to food and water. For establishment of rats' model, homocysteine (400 μg/kg/day) was injected via vena caudalis for 14 days. Senescence-associated β-galactosidase assay with immunostaining.
Senescence-associated β-galactosidase (SA-β-gal) staining was performed according to the manufacturer's protocol. In brief, 10,000 N2a cells seeded on 15-mm glass coverslips were washed with PBS and xed with 4% formaldehyde for 15 min at room temperature. Fixed cells were washed twice with PBS before incubation in SA-β-gal staining solution at 37 °C for 16 h. After incubation, samples were washed twice with PBS, and the numbers of blue SA-β-gal-positive cells were quanti ed using a microscope. For staining of tissue samples, frozen specimens sectioned at 30-µm thick were brie y xed in 4% formaldehyde for 5 min, then staining was performed using the same procedures as for the cultured cells. The exact timing was varied according to the thickness of the section and prior treatment of the tissue.
Co-immunostaining was performed after the SA-β-gal staining procedure. In brief, cells glass coverslips were washed twice in PBS after the blue color was developed, then they were permeabilized with 0.5% Triton X-100 in PBS for 5 min at room temperature. After that, blocking with 0.5% BSA in PBS was performed for 1 h, and primary antibodies (Keap1 or β-catenin) incubation in 0.5% BSA was allowed for overnight at 4°C. Once this incubation step was completed, samples were washed three times with PBS (3 min each), and incubated with secondary antibodies (Alexa Fluor 488 goat-anti-mouse or rabbit IgG (1:1000), and Alexa Fluor 594 goat anti-rabbit or mouse IgG (1:1000) in PBS with BSA) for 60 min at room temperature in 0.5% BSA. The washing step was then repeated, followed by nuclei staining with 1 µg ml 4,6-diamidino-2-phenylindole (DAPI) solution for 10 min. Samples were washed twice in PBS, mounted and observed under microscope. The coverslips were then rinsed and mounted onto glass slides with glycerol. Cells slice were examined using immuno uorescence microscopy (SV120, Olympus, Japan).

Protein degradation assay
For assay the degradation rate of β-catenin, N2a cells were added 100 μM MG132 and 100 μM cycloheximide after 72 hours' treatment with 100 uM Hcy, and cells were collected every hour until 5 hours. All cell samples were harvested and stored at -80°C for further analyzed by western blot.

Western blot and immunoprecipitation
Protein extracts were generated from rat brains or cell cultures, resolved by SDS-PAGE, and transferred onto PVDF membranes. After being blocked by 5% nonfat milk in TBS with 0.1% Tween-20, the membrane was incubated with the primary antibodies in 0.5% BSA overnight at 4°C, followed by peroxidase-conjugated goat anti-rabbit or mouse IgG (1:1000; Beyotime, China) for 60 min at room temperature. The immunoreactive bands were detected using the Enhanced Chemiluminescent Substrate in luminometer (ChemiScope 6000, Clinx, China). Band intensity was measured using ImageJ (NIH, USA).
To analyze ubiquitinated β catenin, we performed immunoprecipitation (IP) experiments using N2a cells lysates. Speci ed antibody (β catenin) and protein G agarose were incubated with the N2a cells lysates (100μg) overnight at 4°C. The resins were washed three times with PBS. After elution by 2×loading buffer, and boiled at 95°C for 10 min, the bound proteins were analyzed by western blotting using β catenin and ubiquitin antibodies.
Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) Total RNA was isolated from the primary astrocyte using TRIzol Reagent (15596018 ThermoFisher Scienti c, USA) according to the manufacturer's manual. Reverse transcription was carried out to synthesize cDNA using Hifair® 1st Strand cDNA Synthesis Kit (11121ES60, Yeasen, China) and qRT-PCR assays were performed using SYBR mix (11203ES03, Yeasen, China). Primer sequences for each gene are synthesized by AUGTC (Beijing, China) and listed as followed: p16 forward primer: AACTCTTTCGGTCGTACCCC, reward primer: GCGTGCTTGAGCTGAAGCTA. p21 forward primer: CCTGGTGATGTCCGACCTG; reward primer: CCATGAGCGCATCGCAATC. P53 forward primer: CTCTCCCCCGCAAAAGA AAAA, reward prime: CGGAACATCTCGAAGCGTTTA; β-catenin forward primer: ATGGAGCCGGACAGAAAAGC, reward primer: CTTGCCACTCAGGGAAGGA. Keap1 forward primer: GATGGGCAGGACCAGTTGAA, reward primer: CCGAGGACGTAGATCTTGCC. The speci city of the PCR product was con rmed by analyzing the melting curve. All PCRs were performed in Triplicate. Results were normalized to mRNA expression in primary astrocytes of sham-operated control.

Bisul te sequencing PCR (BSP) assay
The Keap1 promoter in N2a cell was determined to range from − 2000 to − 70 bp by the Transcriptional Regulatory Element Database from Cold Spring Harbor (http://rulai.cshl.edu/cgi-bin/TRED/tred.cgi? process=promInfo&pid=19717).The Keap1 promoter in mouse was searched in the Transcriptional Regulator Element Database (accession number 46672, NM 009741). The CpG island in the promoter (− 300 to 0 bp) was detected using the UCSC Genome Browser, and the methylation status was analyzed using BSP. Primers for BSP were designed through MethPrimer (http://www.urogene.org/methprimer/), and then were blasted and con rmed using meth BLAST. Keap1's primer as followed, F primer: AATTTTAGGGGATATTGTAT AGTTTA; R primer: ATTCCAACCCTTCCTACAAATAC. A Genomic DNA Extraction kit (Takara, Japan) was used to extract DNA from the cells. The bisul te conversion of DNA was performed with an EpiTect Bisul te kit (Qiagen), PCR was performed on the bisulfate-modi cation samples, and PCR products were cloned into plasmids, and the inserted sequences were sequencing analysis.

Statistical Analysis
Results were expressed as means ±SEM. Unpaired t test was used to assess statistical signi cance between two groups. With respect to multiple comparisons involving 3 or more groups, statistical signi cance was assessed by one-way ANOVA followed by post hoc test (Bonferroni's method). Statistics were computed with Graphpad Prism 6 (GraphPad Software). p<0.05 was considered as statistically signi cant.

Results
Hcy promoted N2a cells senescence in a time-dependent way.
Cell senescence was characterized by enhancement of β-galactosidase activity, upregulation of p16, p21 and p53, which are regarded as marker of senescence. To verify whether Hcy promote neuronal cell senescence, we treated with N2a cells using Hcy. In order to avoid its toxicity, the effect of different concentrations of Hcy (0 to 200 μM) on cell viability was tested. According to CCK8 result (Figure 1 A), Hcy didn't produce toxicity to N2a cells until 200 μM (p value =0.036), so we chose a concentration of 100 μM Hcy in the follow-up study. Next, in order to nd the appropriate time for inducing cell senescence, N2a cells was cultured with Hcy or DMSO and collected at different time point. As shown by the number of SA-β-gal positive staining cells, we observed that Hcy promoted the senescence of N2a cells in a timedependent manner (0 vs 48 h, p value=0.0088; 0 vs 72 h, p value =0.001) (Figure1 B and D), and 72 h of induction was considered the ideal time. In addition, after 72 h of Hcy treatment, Western blot and qPCR were used to detect cell senescence markers p16, p21, and p53 to con rm the senescence-promoting effect of Hcy. Compared with the normal control, the protein level of senescence markers p16 (p value=0.047), p21 (p value=0.019) and p53 (p value=0.0112) in Hcy-treated N2a cells increased signi cantly (Figure 1 C and E). In consistent with protein level, Hcy also enhance mRNA level of p16 (p value=0.0336), p21 (p value=0.047) and p53 (p value=0.0164), indicating that Hcy can effectively enhance the senescence of N2a cells.
Hcy induced cell senescence by enhancing protein level of β-catenin Previous reports indicated that non-degraded β-catenin is both necessary and su cient to drive the neuronal senescent-like phenotype (Chow et al., 2019). And hyperhomocysteinemia could potentiate Wnt/ β-catenin signaling in initiating cardiogenesis (Han et al., 2009). Then we assayed the expression of βcatenin in N2a cells after Hcy induction. Upregulation of β-catenin was observed after induction of Hcy (p value=0.001), without alternation of β-catenin' mRNA level (Figure2 A, B and C). Next, degradation rate of β-catenin was assayed after Hcy's treatment. After incubation of Hcy, MG132 and cycloheximide were added to avoid de novo expression. Compared to control, degradation rate was more slowly in the condition of Hcy treatment (3 h, p value=0.034; 4 h, p value=0.021; 5 h, p value=0.012) (Figure 2 D, E). To further verify the ubiquitination level, β-catenin was immunoprecipitated by its speci c antibody and examined using pan-ubiquitination antibody. As expected, the ubiquitination level of β-catenin was markedly reduced after Hcy's induction (p value=0.0167) (Figure 2 (Figure 2 M and N). These evidences suggested accumulation of β-catenin mediated N2a cellular senescence.

Keap1 inhibited β-catenin's binding to WTX
As reported by previous studies, β-catenin could be ubiquitinated by GSK3β and WTX (Chow et al., 2019; Major et al., 2007). Next, we examined both activity of GSK3β and expression of WTX in N2a cells after induction of Hcy. Unexpectedly, both activity of GSK3β, indicated by GSK3β pS9 (S9), and expression of WTX didn't change signi cantly after induction of Hcy (Figure 3 A and B). However, another WTX binding protein Keap1 was observed upregulated greatly after induction of Hcy (p value=0.001) (Figure 3 A and  B). This result leaded us to the hypothesis that the increase in Keap1 expression promoted an increase in its binding to WTX, thereby reducing the binding of β-catenin. Further analysis by IP revealed that after induction of Hcy, the association between WTX and Keap1 increased (p value=0.0167) along with decreased binding between WTX and β-catenin (p value=0.0196) (Figure 3 C, D and E), suggesting the increased binding of Keap1 to WTX is the cause of the decreased binding of β-catenin. Moreover, mRNA of Keap1 (p value=0.0163) (Figure 3 F) was enhanced after induction of Hcy further con rmed upregulation of Keap1. These results suggested Keap1 contributed to upregulation of β-catenin.

Hcy induced hypomethylation of Keap1's promoter and expression
Previous studies discovered Hcy induced demethylation of DNA, especially the promoter sequence , thus promoted mRNA expression of targeted gene and resulted in increased protein expression. Moreover, it's observed that protein and mRNA level of Keap1 was enhanced signi cantly after induction of Hcy compared to control. Next, DNA methylation transferase 1 (DNMT1), whose main function is the transfer of methyl groups, was tested by western blot after induction of Hcy. Compared to control, Hcy treatment signi cantly downregulated the expression of DNMT1 in N2a cells (p value =0.0475) (Figure 4 A and B). Then we used the BSP method to analyze the methylation level of the CpG island enrichment region of the Keap1 promoter, and observed that Keap1's CpG island methylation was reduced compared to the control (p value =0.0254) ( Keap1 and β-catenin were both enhanced in hippocampus of Hcy-induced rats To verify the above alternations in animal model, SD rats were given 400 μg/kg/day via tail intravenous injection for 14 days. Then rat's brain was sectioned and co-labelled with SA-β-gal staining. Microscopic image revealed that positive number of SA-β-gal staining increased after stimulation of Hcy in neuron of hippocampus compared to rats just received normal saline (p value=0.0003) ( Figure 5 A and B). Furthermore, protein level of senescent markers p16 (p value=0.011), p21 (p value=0.029) and p53 (p value=0.01) were upregulated by the administration of Hcy (Figure5 C, D) in the homogenate of brain, along with increased Keap1 (p value=0.015) and β-catenin (p value=0.023). In consistent with above results, mRNA level of p16 (p value=0.006), p21 (p value=0.0107), p53 (p value=0.0028) and Keap1 (p value=0.025) in Hcy rats' brain also enhanced greatly compared to control rats ( Figure 5 E), providing evidences that senescence was occurred in neurons. Besides, neuron loss was observed in region of CA1 (p value=0.021) and CA3 (p value=0.016) by Nissl staining in brain of rats compared to normal (Figure5 E and F). Based on above results, we proposed that neuron loss was probable due to senescence of neuron.

Discussion
Organ aging and age-related diseases are the main contributors that induce the accumulation of senescent cells in the brain of humans and animals (Baker et al., 2018). In this study, we observed that Hcy induces neuronal senescence in both in vivo and in vitro models, as evidenced by an increase in SAβ-gal positive staining cells and senescence markers. We next observed that β-catenin, an upstream of senescent markers, increased after Hcy treatment and co-localized with SA-β-gal-stained positive cells, whereas knockdown attenuated SA-β-gal-stained positive cells. The increase in β-catenin is due to decreased binding to WTX, resulting in increased protein accumulation caused by increased deubiquitination. On the other hand, we observed that Keap1, another protein that binds to WTX, was increased after Hcy treatment. IP assay showed that WTX binds to more Keap1 after Hcy treatment, while reducing β-catenin binding. Mechanistically, we found that Hcy induced a decrease in DNMT1 activity, leading to an increase in the level of Keap1 promoter CpG island demethylation and mRNA transcriptional activation, leading to an increase in Keap1 protein expression. Rat models con rmed that Hcy promoted cortical neuronal senescence as well as increases in senescent markers, Keap1 and β-catenin, which are thought to be contributing to neuronal loss. Our results uncover a novel function of Keap1 in neuronal senescence and highlight the Keap1/β-catenin pathway may be promising target for the treatment of AD.
In our previous study, elevated serum Hcy replicated many AD-like pathologies, such as neuron loss, synaptic impairment and hyperphosphorylated tau in neuron (Zhang et al., 2019), suggesting Hcy impaired neuron in many aspects. On the other hand, Hcy has been reported to induce cellular senescence, i.e. reducing telomerase activity and AKT phosphorylation (Zhu et al., 2006), enhancing SA-βgal activity (Xu et al., 2000) and DNA hypomethylation (Zhang et al., 2015) in endothelial progenitor cells or endothelial cells, and upregulation of p21 in hepatic HepG2  and p53 in neuron (Kruman et al., 2000), suggesting Hcy is an important causality in promoting cellular senescence. In addition, neurons with neuro brillary tangles in the brains of AD patients exhibit high expression of the senescent marker Cdkna2 mRNA (Musi et al., 2018). In light of this evidence, we propose the hypothesis that neurons develop a tendency toward senescence upon stimulation with Hcy. In this study, we observed that Hcy induced N2a cells toward senescence, as evidenced by that enhancing activity of βgalactosidase as well as senescent marker p16, p21 and p53. Meanwhile, Hcy treated rats also showed positive SA-β-gal staining in cortex neuron and enhanced senescent marker p16, p21 and p53 in hippocampus. These results lead us to believe that Hcy promotes not only somatic cell senescence, but also has the same effect on neurons, so that the elucidation of the mechanism of Hcy-promoted neuronal senescence providing new target for the prevention and treatment of AD.
Next, we explored how Hcy promotes the upregulation of these senescent markers. Aberrant activation of β-catenin has been reported linked to accelerated aging (Liu et al., 2007). It's reported that aberrant βcatenin activity could induce a parallel p53/p21-mediated senescence pathway (Chow et al., 2019). Besides, β-catenin translocated and increased nuclear localization in response to elevated homocysteine in vascular endothelial cadherin (Beard et al., 2012), suggesting a close between β-catenin and Hcy. Thus, we examined protein level of β-catenin in N2a cells after induction of Hcy. As expected, we observed that Hcy induces elevated β-catenin protein levels as determined by Western blot. Immuno uorescence further showed that SA-β-gal-stained positive cells co-located with high expression of β-catenin, whereas knockdown of β-catenin attenuated SA-β-gal-stained positive cells. However, mRNA of β-catenin didn't change by treatment of Hcy, suggesting it was not directly regulated by Hcy, as Hcy could induced hypomethylation of cytosines in cytosine-guanine dinucleotide (CpG) islands in promotor regions, thus promote gene's expression (Kato et al., 2009). Next, we observed that accumulation of β-catenin was resulted from slowdown of its degradation. It is reported that GSK3β could form a multimeric complex with β-catenin that induces N-terminal phosphorylation of β-catenin, leading to its ubiquitin/proteasomemediated degradation (Ji et al., 2015). However, we did not observe an increase in GSK3β activity, indicting the presence of another protein that deubiquitinates β-catenin. According to published literatures, WTX could bind to β-catenin and its E3 ubiquitin ligase aptamer β-transducin repeatcontaining protein (β-TrCP), promoting β-catenin degradation via ubiquitination (Major et al., 2007). Contrary to expectations, we found no signi cant change in WTX protein level after induction of Hcy. However, it has been discovered that WTX could interact with another ubiquitin ligase adaptor Keap1, which functions to regulate the ubiquitination of the transcription factor NRF2. After WTX's competing for binding to Keap1, less NRF2 was deubiquitinated and accumulated (Hast et al., 2013). Then we examined protein level of Keap1 in N2a cells after Hcy's treatment. Intriguingly, Keap1's protein level was observed to be elevated by Hcy, along with increasing mRNA level. As determined by immunoprecipitation assays, WTX showed increased binding to Keap1 and less binding to β-catenin. Immuno uorescence staining also con rmed that Hcy promoted co-localization of Keap1 with SA-β-gal staining positive cells, while knockdown of Keap1 expression attenuated the Hcy-induced SA-β-gal staining positive cells, suggesting Keap1 mediated β-catenin's accumulation by competitive combination to WTX.
It has been studied that methylation levels of gene KEAP1 CpGs at various promoter and intragenic locations showed a signi cant inverse correlation with the transcript levels (Fabrizio et al., 2020). In addition, genetic experiments have de ned the importance of the DNA methyltransferase DNMT1 for the maintenance of CpG island methylation in cells (Robert et al., 2003). In consistent with previous study, we found Hcy decreased the protein levels of DNMT1 compared to controls, along with the decreased CpG methylation of KEAP1 as measured by BSP, suggesting that the rise in Keap1 protein level is due to the elevation of CpG demethylation caused by the Hcy-induced decrease in DMNT1 activity, which in turn activates the transcription and translation of mRNA.
However, most of the experiments in this study were performed on in vitro N2a cells, and studies at the animal level as well as on transgenic animals are lacking. Whether removal of Hcy-induced senescent cells can reverse the learning memory impairment has also not been performed, which may be the next step to be focused on.

Conclusion
In summary, we found that Hcy enhanced upregulation Keap1 via demethylation its CpG island in promoter, resulting in activation of mRNA transcription and protein expression. More Keap1 accumulates with reduced β-catenin binding due to WTX binding, which in turn activates senescence markers p16, p21 and p53 with SA-β-gal activation, ultimately promoting cellular senescence. The discovery of this Keap1/ β-catenin pathway provides a new target for the prevention and treatment of AD.