Removing Aging Cells by Combining Asparaginase and Autophagy Blockers to Cut Asparagine Supply

ASNS, the sole enzyme catalyzing Asn biosynthesis, mediates the ATP‐dependent conversion of aspartate and glutamine into Asn and glutamate (Cai et al. 2022). To investigate whether SNCs exhibit dysregulated asparagine metabolism, we first analyzed ASNS expression across multiple senescence models. In human fetal lung diploid fibroblasts (2BS), ASNS mRNA levels were significantly reduced in SNCs induced by genotoxic stress (bleomycin treatment), ionizing radiation (IR), and replication stress, compared to young proliferating controls (Figure 1A). This downregulation was evolutionarily conserved, with mouse embryonic fibroblasts (MEF) showing similar ASNS suppression during senescence (Figure 1B). The phenomenon extended to diverse cell types, including human umbilical vein endothelial cells (HUVECs) and adult retinal pigment epithelial cells (ARPE‐19), which exhibited marked ASNS mRNA reduction upon senescence (Figure 1C,D). Western blot analysis confirmed corresponding decreases in ASNS protein levels across all tested SNC models (Figure 1E–H). Transcriptomic data from Aging Atlas (https://ngdc.cncb.ac.cn/aging/index↗) further validated ASNS downregulation in IR‐induced and replicative senescent human diploid fibroblasts (WI38, IMR90) and arterial endothelial cells (HAECs) (Figure S1A–C). Importantly, intracellular Asn levels were correspondingly depleted in these SNCs (Figure S2A–D), functionally linking reduced ASNS activity to asparagine deficiency.

To assess physiological relevance, we examined ASNS expression in aged mouse tissues. Liver, lung, kidney, intestine, and muscle from aged mice showed significantly increased P16 protein levels (Figure 1I–M), confirming elevated senescent cell burden. Strikingly, these tissues exhibited corresponding decreases in ASNS expression (Figure 1I–M), mirroring our in vitro findings. These results establish ASNS deficiency as a consistent feature of both cellular senescence and tissue aging.

Transcription factor ATF4, a known master regulator of ASNS, typically induces ASNS expression through the amino acid response (AAR) and unfolded protein response (UPR) pathways during nutrient stress (Lomelino et al. 2017). Surprisingly, despite the evidence of partial AAR/UPR pathway activation (elevated p‐GCN2 and p‐eIF2α; Figure S3A,B), both ATF4 mRNA and protein levels were significantly reduced in SNCs. ATF4 overexpression in senescent cells restored ASNS mRNA and protein expression (Figure S3D,E); furthermore, computational prediction using JASPAR identified two ATF4 binding sites within the ASNS promoter region, both exhibiting high‐affinity potential with relative profile scores exceeding 0.9 (Figure S3F,G), demonstrating that ATF4 deficiency is responsible for ASNS downregulation during senescence.

To elucidate the functional role of ASNS in cellular senescence, we first examined the consequences of ASNS depletion in young cells. Knockdown of ASNS significantly elevated senescence markers, including P16, P21, and P53 protein levels (Figure 2A), increased SA‐β‐gal activity (Figure 2B,C), and enhanced secretion of SASP factors IL1β and IL6 (Figure 2E,F), while concurrently suppressing cellular proliferation (Figure 2D). These results demonstrate that ASNS deficiency alone is sufficient to trigger premature senescence. Complementary to these loss‐of‐function experiments, we overexpressed ASNS in young 2BS fibroblasts prior to inducing senescence with bleomycin. Assessment at day 7 post‐treatment revealed that ASNS overexpression mitigated bleomycin‐induced increases in P16, P21, and P53 protein levels (Figure 2G), reduced the percentage of SA‐β‐gal positive cells (Figure 2H,I), and decreased IL1β expression (Figure 2J,K). While ASNS‐overexpressing cells maintained proliferative capacity for 7 days after bleomycin exposure, subsequent monitoring revealed proliferation arrest occurring at 72 h post this initial period (Figure 2L). This temporal pattern indicates that elevated ASNS expression can delay but not reverse the established senescence.

Given that Asn and Glu are direct catalytic products of ASNS activity, we next investigated whether the senescence phenotype induced by ASNS knockdown stems from metabolic perturbation. Intracellular metabolite analysis revealed a significant reduction in Asn levels in ASNS‐deficient cells, while Glu concentration remained unchanged (Figure ). To determine the functional specificity of these metabolites, we performed rescue experiments with exogenous supplementation. Notably, Glu supplementation showed no significant effects on ASNS knockdown‐induced senescence markers, including P16/P21/P53 protein levels (Figure ), proliferation rate (Figure ), or SA‐β‐gal positivity (Figure ). In contrast, Asn supplementation effectively normalized all these senescence parameters (Figure ), demonstrating the specific requirement for Asn in maintaining cellular homeostasis. S4A,B S4C S4D S4E,F S4C–F

To further define the temporal requirements for asparagine in senescence progression, we supplemented Asn at different time points following bleomycin treatment. Early Asn administration (day 0) significantly mitigated the upregulation of P16, P21, and P53 (Figure ), decreased SA‐β‐gal‐positive cells (Figure ), and maintained cell growth (Figure ). In contrast, delayed Asn supplementation (day 5) showed no protective effects (Figure ), suggesting that Asn primarily functions to slow senescence initiation rather than reverse its progression. S4H S4I,J S4K S4H–K

Collectively, these results establish ASNS as a critical metabolic regulator of cellular senescence through its control of asparagine biosynthesis. Our data demonstrate that ASNS deficiency drives premature senescence via Asn depletion, while its overexpression delays but cannot abrogate the senescent state. Mechanistically, this regulation depends specifically on ASNS‐derived asparagine production, highlighting a previously unrecognized metabolic checkpoint in senescence control.

Building upon our observation of reduced ASNS expression and intracellular asparagine (Asn) levels in senescent cells (Figure ), we sought to determine whether metabolic targeting of Asn could selectively eliminate senescent cells (SNCs). Treatment with asparaginase (ASNase) at concentrations ranging from 0.5 to 3 U/mL effectively depleted extracellular Asn (Figure ) and induced dose‐dependent apoptosis in SNCs, with increases ranging from 8% to 22% (Figure ). Notably, young cells exhibited minimal apoptosis except at the highest concentration tested (3 U/mL), suggesting a potential therapeutic window at lower doses. High concentrations of ASNase might cause nonspecific cell apoptosis. Thus, 0.5 U/mL ASNase was selected for the following experiments. However, residual intracellular Asn persisted in ASNase‐treated SNCs (Figure ), prompting investigation into alternative Asn sources that might sustain cell survival under conditions of extracellular deprivation. S2A–D S5A S5B,C S5D

In mammalian cells, cellular Asn pools are maintained through three primary pathways: extracellular uptake, de novo synthesis via ASNS, and protein degradation‐mediated recycling (Davidsen and Sullivan 2020; Hinze et al. 2019). Given the observed downregulation of ASNS in senescent cells and our pharmacological blockade of extracellular Asn with ASNase, we hypothesize that protein degradation might compensate to maintain Asn levels. Protein degradation in mammalian cells occurs primarily through two major pathways: the ubiquitin–proteasome system and the autophagy–lysosomal pathway. Our comparative analysis revealed that senescent cells exhibited decreased expression of proteasome subunits (PSMB5, PSMB6, PSMB7) but upregulated key autophagy regulators (BECN1, ATG7) and increased LC3BII/I ratios (Figure 3A). More importantly, following ASNase treatment, the expression of autophagy‐related proteins was elevated (Figure 3B). Electron microscopy revealed no obvious difference in autophagosome abundance between untreated young and senescent cells. However, ASNase treatment induced marked autophagosome accumulation in both young and senescent cells (Figure 3C). This suggests that under conditions of asparaginase‐induced decline in Asn levels, senescent cells may increase autophagic flux to maintain cellular Asn requirements, which is consistent with other reports (Gamerdinger et al. 2009; Wang et al. 2012; Young et al. 2009; Zhang et al. 2023).

To functionally assess the role of autophagy in senescent cell survival following ASNase treatment, we generated stable cell lines with knockdown of essential autophagy genes (BECN1/ATG7) (Figure S6A,B). Genetic inhibition of autophagy synergized with ASNase treatment to further deplete intracellular Asn levels (Figure S6C) and significantly enhance apoptosis compared to ASNase alone (Figure S6D,E). The specificity of this effect was confirmed by complete rescue with exogenous Asn supplementation (Figure S6C–E). We extended these findings using the clinically approved autophagy inhibitor hydroxychloroquine (HCQ) (Figure S7A–D), demonstrating that the ASNase + HCQ (AH) combination potently reduced intracellular Asn (Figure 3D) and increased senescent cell apoptosis by approximately 40% (Figure 3E–H). Strikingly, this effect was highly selective for senescent cells, whereas the established senolytic cocktail dasatinib + quercetin (DQ) induced comparable apoptosis in both young and senescent populations (Figure 3E–H). Mechanistically, analogous to tumors under nutrient stress (Darnal et al. 2024; Takahashi et al. 2017), AH triggers mitochondrial apoptotic cascades in senescent cells: immunoblotting detected decreased anti‐apoptotic BCL‐XL, increased pro‐apoptotic Bax, and elevated activated caspase‐9 and caspase‐3 (Figure S8A). This demonstrates that AH induces senescent cell death via the BCL‐2 family‐mediated mitochondrial apoptosis pathway.

The therapeutic potential of this approach was further validated across multiple senescence models, including replicative senescent HUVECs (Figure ), IR‐induced senescent 2BS cells (Figure ), replicative senescent 2BS cells (Figure ), bleomycin‐induced ARPE‐19 cells (Figure ), and replicative senescent MEFs (Figure ). In all models, AH treatment demonstrated consistent efficacy in eliminating SNCs while sparing proliferating counterparts. S10A–E S11A–C S11D–F S11G–I S11J–L

Prompted by these findings, we investigated AH's potential in eliminating therapy‐induced senescent (TIS) tumor cells—a state arising during chemo/radiotherapy that enables tumor survival under sublethal stress (Basu 2022). TIS is characterized by senescence phenotypes, including elevated levels of senescence‐associated proteins, increased secretion of SASP, and proliferation arrest. TIS is believed to help tumor cells evade death during sublethal drug therapy. Under certain conditions, TIS can reactivate proliferative potential and lead to cancer recurrence (Saleh et al. 2020). Therefore, eliminating TIS cells has become an important area in the field of cancer research.

We subsequently explored the translational relevance of these findings in TIS using gemcitabine‐treated pancreatic cancer cells (MiaPaCa‐2). Cells undergoing TIS exhibited characteristic features including P21 upregulation (Figure 4A), proliferation arrest (Figure 4B), increased SASP secretion (Figure 4C), and elevated SA‐β‐gal activity (Figure 4D,E). Mirroring our observations in normal SNCs, these senescent cancer cells showed significant ASNS downregulation at both mRNA (Figure 4F) and protein levels (Figure 4A), rendering them susceptible to AH‐mediated Asn depletion (Figure 4G) and subsequent apoptosis (Figure 4H,I). Parallel results in PANC‐1 cells (Figure S12A–I) confirmed the broad applicability of this metabolic vulnerability across cellular contexts.

To evaluate the therapeutic potential of AH in vivo, we treated 18‐month‐old C57BL/6J mice (a naturally aged mouse model) with ASNase (intraperitoneal injection) and HCQ (oral gavage) for 12 weeks (Figure 5A). Biochemical analysis confirmed that ASNase treatment effectively reduced serum Asn levels (Figure S14A). Immunohistochemical evaluation revealed that while HCQ monotherapy showed minimal effects on P16‐positive cell clearance, the ASNase or AH combination significantly reduced senescent cell burden across multiple tissues, including liver, lung, kidney, small intestine, and muscle (Figure 5B–F). This senolytic efficacy was further supported by complementary senescence markers: decreased SA‐β‐gal staining (Figure S13A–E) and increased LaminB1 expression (Figure S13F–J) in tissues from AH‐treated animals. These findings indicate that AH intervention effectively clears senescent cells in tissues. In addition, AH treatment reduced the serum levels of IL1β and IL6 (Figure 5G,H). Oxidative/antioxidative system imbalance has been widely used as a hallmark of aging (Cand and Verdetti 1989). After AH treatment, serum antioxidant markers superoxide dismutase (SOD) (Figure 5I) and glutathione peroxidase (GSH‐Px) (Figure 5J) significantly increased, with no notable changes observed in the lipid oxidation marker malondialdehyde (MDA) (Figure S14B). These results indicated that AH enhanced the antioxidative potential of aged mice.

Functional assessment demonstrated comprehensive physiological improvements in AH‐treated animals. These mice exhibited significantly enhanced motor performance in rotarod testing (Figure 5K), increased grip strength (Figure 5L, Figure S14C), and greater exploratory behavior in open field assays (Figure 5M,N, Figure S14D–G). The concordance between senescent cell clearance, reduced inflammation, improved oxidative stress markers, and enhanced physical function provides compelling evidence that AH combination therapy represents a promising strategy for mitigating multiple aspects of age‐related functional decline.

Given the established role of SNCs in age‐related pathologies, we evaluated the therapeutic potential of AH combination therapy in three prototypical senescence‐associated disease models: senile osteoporosis (SOP), non‐alcoholic fatty liver disease (NAFLD), and atherosclerosis.

SOP, characterized by progressive bone mass reduction and microstructural deterioration (Ali et al. 2024), represents a prototypical age‐related disorder where senescent cell accumulation contributes to pathogenesis (Farr et al. 2017; Guo et al. 2021). Using 4‐month‐old male SAM‐P/6 mice (Figure S15A)—a well‐characterized model of age‐related bone loss (Matsushita et al. 1986)—we evaluated the therapeutic potential of AH treatment. Three months of AH treatment significantly reduced P16‐positive cells in the bone marrow of the right femur compared to the control group (Figure S15B,C). Micro‐CT analysis of the left femur (Figure 6A) revealed differential treatment effects: while HCQ monotherapy showed minimal impact on bone mineral density (BMD), both ASNase (A) and AH combination therapy significantly increased BMD, with AH demonstrating superior efficacy (Figure 6B). The established senolytic cocktail DQ (dasatinib + quercetin) also enhanced BMD, though to a lesser degree than AH treatment. Comprehensive trabecular bone analysis showed that AH treatment improved multiple microstructural parameters, including increased bone volume/tissue volume ratio (BV/TV; Figure 6C), greater trabecular number (Tb.N; Figure 6D), enhanced trabecular thickness (Tb.Th; Figure 6E), and reduced trabecular spacing (Tb.Sp; Figure 6F). Biochemical assessment revealed decreased serum alkaline phosphatase levels in AH‐treated mice (Figure S15D), consistent with reduced bone turnover, while calcium and phosphorus homeostasis remained unaffected (Figure S15E,F). These findings demonstrate that AH combination therapy effectively reduces senescence cell burden in bone tissue and improves multiple parameters of bone quality, supporting its potential as a therapeutic strategy for senile osteoporosis.

To investigate AH's therapeutic effect on NAFLD progression, we employed 12‐week‐old ApoE(−/−) mice maintained on a high‐fat diet for 12 weeks (Figure S16A), a well‐established model for studying NAFLD pathogenesis. Research indicates that SNCs play a crucial role in the onset and progression of NAFLD (Engelmann & Tacke, 2022; Meijnikman et al. 2021). Following 12 weeks of AH treatment, quantitative analysis revealed significant reductions in hepatic P16‐positive cells (Figure 6G first row, H) and SA‐β‐gal activity (Figure 6G second row, I), demonstrating effective clearance of senescent cells in fatty liver tissue. Histopathological evaluation showed marked improvements in liver architecture with AH treatment. Compared to wild‐type controls, ApoE (−/−) mice exhibited characteristic NAFLD features including hepatocyte ballooning, disorganized parenchymal architecture, and cytoplasmic vacuolization (Figure 6G third row, HE staining). AH treatment substantially ameliorated these changes, with only minimal residual hepatocyte swelling observed (Figure 6G third row). Masson's trichrome staining demonstrated that AH treatment reduced the extensive collagen deposition, characteristic of diet‐induced hepatic fibrosis, in the ApoE (−/−) group (Figure 6G fourth row, J). Correspondingly, Oil Red O staining revealed significantly decreased lipid accumulation in AH‐treated animals (Figure 6G fifth row, K). Biochemical analysis showed that AH treatment normalized serum ALT and AST levels (Figure 6L,M), indicating improved hepatic function. Interestingly, while AH treatment improved liver histology and function, it did not significantly alter serum triglyceride (TG), total cholesterol (CHO), or LDL cholesterol levels (Figure 6N–P), suggesting that the therapeutic benefits occur independently of systemic lipid metabolism modulation. These findings collectively demonstrate that AH treatment effectively reduces senescent cell burden in NAFLD, leading to improved liver histopathology and function without affecting circulating lipid profiles.

Atherosclerosis represents a quintessential age‐related vascular pathology characterized by progressive plaque formation, wherein cellular senescence of endothelial cells, vascular smooth muscle cells, and macrophages plays a pivotal role (Xiang et al. 2022). These senescent cells exacerbate disease progression through sustained secretion of SASP factors, including pro‐inflammatory cytokines and matrix metalloproteinases (MMPs) (Sun et al. 2022). In our atherosclerosis model, immunofluorescence analysis of aortic root sections demonstrated that AH treatment significantly reduced P16‐positive senescent cells within plaques (Figure S16B second column, D). Subsequently, the impact of AH treatment on the progression of atherosclerosis was evaluated. Key characteristics of atherosclerosis include lesion area and plaque stability, with vulnerable plaques being prone to rupture and resulting in acute cardiovascular events (Pérez‐Medina et al. 2020). Compared to the control group, AH‐treated mice exhibited an increase in smooth muscle cells in the aortic root, which was presented by α‐SMA expression (Figure S16B sixth column, F), with no significant change in infiltrating macrophages, which was presented by F4/F8 expression (Figure S16B third column, E). Additionally, AH treatment resulted in an increase in collagen content in the aortic root (Figure 6Q,R), while significantly reducing the lesion area (Figure 6S) and necrotic core size (Figure 6 T). Moreover, oil red O staining of the aorta, extending from the aortic arch to the common iliac artery, revealed a marked reduction in the lesion area in AH‐treated mice compared to those in controls (Figure 6 U,V). These results collectively indicate that AH combination therapy (1) effectively eliminates senescent cells from atherosclerotic plaques, (2) limits plaque progression, and (3) enhances features of plaque stability—particularly through increased α‐SMA+ smooth muscle cell content and consequent collagen deposition—thereby potentially reducing risks of plaque rupture and acute cardiovascular events.

The translational potential of AH therapy necessitated comprehensive safety evaluation. In vitro assessment revealed minimal cytotoxicity in mouse mesenchymal stem cells (m‐MSCs), with only a marginal increase in apoptotic cells observed at therapeutic concentrations (Figure ). In vivo toxicity studies demonstrated excellent tolerability following 12‐week administration in mice. Serum biochemical analysis showed no significant alterations in hepatic (ALT, AST) or renal (BUN, creatine) function markers across treatment groups (Figure ). Complete blood count analysis confirmed all hematological parameters, including white blood cells, red blood cells, and platelet indices, remained within normal physiological ranges (Figure ). Histopathological examination of liver and kidney tissues revealed preserved organ architecture with no evidence of drug‐induced morphological abnormalities (Figure ). S17A S18A–D S18E–J S19A

The AH combination therapy exhibited a favorable safety profile across all evaluated parameters. Comprehensive toxicological assessment demonstrated: minimal in vitro cytotoxicity in stem cells; preservation of hepatic and renal function markers; maintenance of physiological hematological parameters; and absence of histopathological abnormalities in major organ systems. These collective findings substantiate that prolonged AH administration is not associated with significant systemic toxicity, thereby addressing a critical prerequisite for clinical development of this senolytic approach.

All animal experiments were approved by the Biomedical Ethics Committee of Peking University (approval no. LA2022296). All animal housing and experiments were conducted in strict accordance with the institutional guidelines for care and use of laboratory animals.

The antibodies used in this study are listed: ASNS (Proteintech, 14681‐1‐AP), P21 (Abcam, ab109520), P16 (Abcam, ab108349), P16 (Abcam, ab54210), P53 (Santa Cruz, sc126), ATF4 (ABclonal, A18687), LC3B (ABclonal, A5618), BECN1 (ABclonal, A11761), ATG7 (ABclonal, A19604), P62 (Proteintech,18,420–1‐AP), PSMB5 (ABclonal, A1975), PSMB6 (ABclonal, A4053), PSMB7(ABclonal, A14771), p‐GCN2(Thr899) (ABclonal, AP1428), GCN2(ABclonal, A2307), p‐PERK(T982) (ABclonal, AP1501), PERK (ABclonal, A21255), p‐eIF2α (S51) (ABclonal, AP0745), eIF2α (ABclonal, A21221), Bcl‐XL (Proteintech,26967‐1‐AP), Bax (Proteintech,60267–1‐lg), Caspase3(abcam, Ab32351), C‐Caspase3(CST,9664S), and C‐Caspase9(CST, 9505P).

The other compounds used in this study are listed along with their manufacturers' details: Gemcitabine (Selleck, S1714), Bleomycin (TargetMol, T6116), Dasatinib (Aladdin, D125110), Quercetin (Sigma, Q4951), L‐ASNase (MCE, HY‐P1923), Hydroxychloroquine (TargetMol, T9287), L‐Asn (Sigma, A0884), L‐Aspartic acid (Sigma, A6558), and glutamic acid (Sigma, G3640), Lys05 (TargetMol, T3437).

All cells were cultured in a 5% CO₂ humidified incubator at 37°C. The cells, namely, 293T, 2BS, MEFs, ARPE‐19, and human pancreatic ductal adenocarcinoma cell lines, viz. PANC‐1 and Mia Paca2, were routinely maintained in standard Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. The primary HUVECs were grown in endothelial cell medium supplemented with 5% FBS, 1% endothelial cell growth supplement, and 1% penicillin–streptomycin. MEFs were isolated and cultured as described (Tan and Lei 2019).

Lentiviral vectors for ASNS, BECN1, and ATG7 shRNA were constructed based on the pLKO.1 plasmid. The shRNA sequences for humans are listed.

shScramble‐F:5′‐CCGGCAACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTGTTTTT‐3′

shScramble‐R: 5′‐aattAAAAACAACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTG‐3′ shASNS‐F: 5′‐CCGGGCTCTGTTACAATGGTGAAATCTCGAGATTTCACCATTGTAACAGAGCTTTTTG‐3′, shASNS‐R: 5′‐aattCAAAAAGCTCTGTTACAATGGTGAAATCTCGAGATTTCACCATTGTAACAGAGC‐3′; shBECN1‐F: 5′‐CCGGGGTCTAAGACGTCCAACAACACTCGAGTGTTGTTGGACGTCTTAGACCTTTTTG‐3′, shBECN1‐R: 5′‐aattCAAAAAGGTCTAAGACGTCCAACAACACTCGAGTGTTGTTGGACGTCTTAGACC‐3′; and ShATG7‐F: 5′‐CCGGGCTGGTTTCCTTGCTTAAACACTCGAGTGTTTAAGCAAGGAAACCAGCTTTTTG‐3′, ShATG7‐R: 5′‐aattCAAAAAGCTGGTTTCCTTGCTTAAACACTCGAGTGTTTAAGCAAGGAAACCAGC‐3′.

Human ASNS coding sequences with flag‐tags and human ATF4 coding sequences, amplified from a cDNA library, were cloned into a PCDH vector. Sequences for cloning are listed. PCDH‐ASNS‐F:5′‐GTCAGATCCGCTAGCCACCATGTGTGGCATTTGGGCGCT‐3′; PCDH‐ASNS‐R:5′‐GTCACGCGTGAATTCAGCTTTGACAGCTGACTTGT‐3′ PCDH‐ATF4‐F: 5′‐GATTCTAGAGCTAGCCACCATGACCGAAATGAGCTTCCTGAG‐3′; PCDH‐ATF4‐R:5′‐CGCGGCCGCGGATCCCTAGGGGACCCTTTTCTTCC‐3′. Transfection was performed following Vigofect (Vigorous Biotechnology) instructions.

Bleomycin‐induced premature senescence was conducted with 60%–70% confluent cells treated with 50 μg/mL bleomycin. After 24 h, the complete culture medium was added, and the culture medium was changed every 2 d for 7 d.

IR‐induced premature senescence was conducted with 60%–70% confluent cells treated with 10 Gy IR. After 24 h, the complete culture medium was added, and the culture medium was changed every 2 d for 7 d.

Replicative senescence was conducted with 2BS cells that have been passaged 50–60 times, HUVECs that have been passaged 10–13 times, and MEFs that have been passaged 8–10 times.

The following were considered young cells: 2BS cells within 25 passages, HUVECs within 5 passages, and MEFs within 4 passages.

All animal experiments were performed according to the animal protection guidelines of Peking University Health Science Center, China. All animals were maintained in a specific pathogen‐free facility with a 12 light/12 dark cycle and free access to food and water. C57BL/6J, SAMP6, and ApoE^−/− mice were randomized into different groups.

All functional assays were conducted at least 5 d after the last drug dose was administered. Mice were trained at least two times on d 1 and 2 and tested on d 3, 4, and 5. Results presented are the average of three trials.

For rotarod test, mice were placed on the rotarod device (Unibiolab) with the following settings: an initial speed of 5 rpm, an acceleration rate of 0.1 rpm/s, a maximum speed of 30 rpm, and a cutoff time of 600 s. The device automatically recorded the speed and time at which the mice fell off the rod.

For grid test, mice were placed on the grip strength meter (Unibiolab), with all four limbs grasping the spring of the device. The mice were then gently pulled parallel to the device by their tails. The maximum and average grip strength (N) were recorded for five trials.

For open field test, after acclimating to the open field apparatus (Unibiolab) for 30 min, the mice were placed in a 40 cm square box. A camera recorded their activity for 10 min, capturing movement trajectories, speed, and other behaviors. The environment was kept bright and quiet, and the apparatus was cleaned to remove any odors.

For routine blood examination, 50 μL fresh blood was collected from each mouse and mixed with EDTA immediately. The blood samples were analyzed by an automatic blood cell analyzer (Mindray, BC‐2800vet).

For serum biochemical analysis, blood samples were collected, clotted for 45 min at 37°C and then overnight at 4°C, and later centrifuged (1000×g, 10 min) to obtain serum. Serum aliquots (200 μL) were analyzed for ALT, aspartate transaminase (AST), carbamide (UREA), and creatinine (CREA) by using a chemical analyzer (Mindray, Chemray 800).

Tissues were fixed overnight in 4% paraformaldehyde, embedded in paraffin, and cut into 5 μm sections. Sections were subjected to hematoxylin and eosin staining and Masson's trichrome staining for collagen detection. The collagen‐positive area was quantified with ImageJ. Immunohistochemical and immunofluorescence staining was performed following standard protocols. The following primary antibodies were used: ASNS (Proteintech, 14681‐1‐AP), P16 (Abcam, ab54210), Lamin B1(afantibody, AFRM0038).

SA‐beta‐gal activity was performed as described (Shlush and Selig 2013). Briefly, freshly frozen tissue sections or adherent cells were washed twice with cold PBS and fixed with 4% paraformaldehyde for 5 min at room temperature. After two cold PBS washes, the samples were incubated with freshly prepared SABG staining solution (containing 1 mM MgCl₂, 1 mg/mL X‐gal, 5 mM potassium ferricyanide, and 5 mM potassium ferrocyanide) overnight at 37°C. Staining was ended by washing three times with cold PBS.

Atherosclerotic plaque was detected with oil red O staining. Briefly, whole aortas were separated and fixed in 4% paraformaldehyde for 12 h at room temperature. Then, the fixed aortas were incubated in 0.5% oil red O solution prepared in 60% isopropyl alcohol for 15 min at 37°C. Subsequently, the samples were washed in 60% isopropyl alcohol solution for 1 min and washed with double distilled water. The oil red O‐stained aortas were dissected and pinned on foam flat. After the specimens were photographed, quantification of the plaque area was done with ImageJ.

Cells were seeded into 96‐well plates (3000 cells/well) with 100 μL medium containing 10% FBS. Next, the cells were treated with 100 μL DMEM and CCK‐8 (Dojindo, Japan) solution (90:10) at 37°C for 1 h; absorbance was measured at 450 nm using a 96‐well plate reader.

Mouse whole blood samples were stewed at 37°C for 1 h and overnight at 4°C. Then, the samples were centrifuged at 3000 rpm for 10 min at 4°C. The supernatant is serum. IL1β and IL6 in serum were measured by mouse interleukin 1 beta (IL1β) ELISA kit (Servicebio, GEM0002) and mouse interleukin 6 (IL6) ELISA kit (Servicebio, GEM0001), respectively. GSH‐PX, T‐SOD, and MDA were measured by glutathione peroxidase (GSH‐PX) (Njjcbio, A005), total superoxide dismutase (T‐SOD) (Njjcbio, A001‐1), and MDA (Njjcbio, A003‐1) assay kits, respectively, according to the manufacturer's protocols.

Cells were digested into single‐cell suspension by incubation with 0.25% trypsin at 37°C for 3 min. Cells were washed three times with PBS and stained with APC annexin V using the annexin V‐APC apoptosis detection kit (Elabscience, E‐CK‐A217) and propidium iodide according to the manufacturer's protocol. Cell apoptosis was detected by flow cytometry on an Accuri c6 Plus (BD). Data were analyzed using FlowJo software. Apoptotic cells were annexin V‐positive cells.

Total RNA was extracted from cells with TransZol reagent (TransGen Biotech, ET101‐01‐V2) and reverse transcribed into cDNA using TransScriptII all‐in‐one first strand cDNA synthesis supermix (TransGen Biotech, AH341‐01). The cDNA was quantitated by RT‐qPCR with PerfectStart green qPCR supermix (TransGen Biotech, AQ602‐01). Data were analyzed using the 2^(−ΔΔCt) method. GAPDH and ACTB served as endogenous normalization controls for mouse and human samples, respectively. The primers used are mentioned as follows.

Mouse primers:

GAPDH‐F: 5′‐TCACTGCCACCCAGAAGAC‐3′, GAPDH‐R: 5′‐TGTAGGCCATGAGGTCCAC‐3′; Il1b‐F: 5′‐TTGACAGTGATGAGAATGACC‐3′, Il1b‐R: 5′‐GCAGGTTATCATCATCATCC‐3′; Il6‐F: 5′‐ACCACTCCCAACAGACCTGTCTATACC‐3′, Il6‐R: 5′‐CTCCTTCTGTGACTCCAGCTTATCTGTTAG‐3′; MMP3‐F: 5′‐TCCTGATGTTGGTGGCTTCAG‐3′, MMP3‐R: 5′‐TGTCTTGGCAAATCCGGTGTA‐3′; Il8‐F: 5′‐TCGAGACCATTTACTGCAACAG‐3′, Il8‐R: 5′‐CATTGCCGGTGGAAATTCCTT‐3′; and ASNS‐F: 5′‐GCAGTGTCTGAGTGCGATGAA‐3′, ASNS‐R: 5´‐TCTTATCGGCTGCATTCCAAAC‐3′.

Human primers:

GAPDH‐F: 5′‐ACAACAGCCTCAAGATCATCAGCAAT‐3′, GAPDH‐R: 5′‐GTCCTTCCACGATACCAAAGTTGTCA‐3′; ASNS‐F: 5′‐GGAAGACAGCCCCGATTTACT‐3′, ASNS‐R: 5′‐AGCACGAACTGTTGTAATGTCA‐3′; ATG7‐F: 5′‐CAGTTTGCCCCTTTTAGTAGTGC‐3′, ATG7‐R: 5′‐CCAGCCGATACTCGTTCAGC‐3′; BECN1‐F: 5′‐GGTGTCTCTCGCAGATTCATC‐3′, BECN1‐R: 5′‐TCAGTCTTCGGCTGAGGTTCT‐3′; IL1B‐F: 5′‐ATGATGGCTTATTACAGTGGCAA‐3′, IL1B‐R: 5′‐GTCGGAGATTCGTAGCTGGA‐3′; IL6‐F: 5′‐ACTCACCTCTTCAGAACGAATTG‐3′, IL6‐R: 5′‐CCATCTTTGGAAGGTTCAGGTTG‐3′; IL8‐F: 5′‐ACTGAGAGTGATTGAGAGTGGAC‐3′, IL8‐R: 5′‐AACCCTCTGCACCCAGTTTTC‐3′; CXCL1‐F: 5′‐TTGCCTCAATCCTGCATCCC‐3′, CXCL1‐R: 5′‐AGTTGGATTTGTCACTGTTCAGC‐3′; MMP3‐F: 5′‐CGGTTCCGCCTGTCTCAAG‐3′, MMP3‐R: 5′‐CGCCAAAAGTGCCTGTCTT‐3′; ATF4‐F: 5′‐CCCTTCACCTTCTTACAACCTC‐3′, ATF4‐R: 5′‐TGCCCAGCTCTAAACTAAAGGA‐3′.

Cells were lysed by using RIPA buffer (Beyotime, P0013B) on ice for 10–20 min. Protein samples were quantified using BCA protein assay kit (Beyotime, P0010), boiled in 5× loading buffer. The samples containing 20–50 μg total protein were subjected to 6%–12% SDS‐PAGE and transferred to a 0.45 μm polyvinylidene fluoride membrane (Millipore). The membrane was blocked with 5% BSA (Biorigin, BN20800) and incubated overnight with indicated primary antibodies in TBS supplemented with 0.1% Tween 20 at 4°C. The membrane was then incubated with horseradish peroxidase (HRP)‐labeled IgG secondary antibodies at room temperature for 1 h, and the membrane was developed with chemiluminescent HRP substrate (Millipore, P90720) on a ChemiDoc imaging system (BioRad, USA). The band intensity was quantified with ImageJ software (https://imagej.en.softonic.com/↗).

Liquid chromatography–tandem mass spectrometry (LC–MS/MS) was employed for quantitative analysis of asparagine (Asn) and glutamic acid (Glu). For fluid samples, metabolites in 100 μL mouse serum or 100 μL supernatants collected from the culture medium were extracted by precooled 100% acetonitrile in a ratio of 1:5. For cell samples, adhesive cells were harvested and washed three times with normal saline; the viable cell count was recorded prior to metabolite extraction. Metabolites from cells cultured in a 6 mm dish were extracted by 500 μL precooled 100% acetonitrile. Then, the mixtures were vortexed for 5 min and centrifuged at 14,000 × g for 20 min at 4°C. Metabolites in these supernatants were collected and analyzed using a triple quadrupole LC–MS/MS system (Agilent, 1290/6460).

Following cervical dislocation euthanasia, left femurs were dissected free of connective tissues and fixed overnight in 4% paraformaldehyde at 4°C. After PBS rinses (3×), specimens were stored in 75% ethanol prior to micro‐CT scanning (SkyScan 1276, Bruker MicroCT) using standardized parameters: 50 kV source voltage, 200 μA current, and 6.054735 μm isotropic voxel size. Three‐dimensional reconstructions were generated via NRecon software (v1.7.4.2) with trabecular morphometric analysis—including bone mineral density (BMD) calibrated against hydroxyapatite phantoms and microarchitectural parameters (BV/TV, Tb.Th, Tb.Sp, Tb.N)—performed on distal femoral metaphyseal regions using CTAn software (v1.20.8.0).

Cells sample preparation commenced with primary fixation of cell pellets using 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 2 h at 4°C. Following triple PBS washes, secondary fixation was performed with 1% osmium tetroxide for 1.5 h prior to en bloc staining with 0.5% uranyl acetate. Specimens were dehydrated through an ethanol series (30%–100%) and acetone, then embedded in Epon‐Araldite resin via graded infiltration before polymerization at 60°C for 48 h. Ultrathin sections (70 nm nominal thickness) were cut using a Leica EM UC7 ultramicrotome equipped with a 45° diamond knife (Diatome), mounted on 200‐mesh copper grids, and post‐stained with 2% uranyl acetate and Reynolds' lead citrate. Imaging was conducted on a JEOL JEM‐1400 transmission electron microscope operated at 80 kV.

All statistical analyses were performed using SPSS 20 (IBM) and GraphPad Prism. Data distribution was assumed to be normal, but this was not tested. Data are presented as mean ± SEM. Unpaired two‐tailed t tests were used to estimate statistically significant differences between two groups. One‐way ANOVA test with Holm–Sidak multiple test correction was used for multiple comparisons. Two‐way ANOVA test with Tukey's post hoc comparison was used for CCK8 assay. *p < 0.05, **p < 0.01, ***p < 0.001, and ns, not significant.