C57BL/6（wt），P16-3MR（由J. Campisi捐赠）18，营养不良MDX（DBA/2-靠背）和MDX/P16-3MR（肿瘤性MDX小鼠（p16-3mr鼠标）在Barcel housed and Antastial housed and and Artial of Bigigh and housital housed and housed housed and housed housed intartial housed and housed housed intartial housed inter。12 h – 12 h的灯光周期和随意以标准的食物饮食喂养。所有实验均遵循“三个RS”的原则 - 根据指令63/2010及其在成员国的实施。所有程序均获得PRBB动物研究伦理委员会（PRBB-CEEA）和地方政府（Generalitat de Catalunya）的授权，并根据欧洲指令2010/63/EU和西班牙法规RD 53/2013进行。除非另有说明，否则在每个实验中都使用雄性和雌性小鼠。根据杰克逊实验室的指南和协议，维护和基因分型实时菌落。这些小鼠被安置在一起，每天都对其健康进行疾病症状（与年龄相关的体重减轻等）进行监测，当兽医和生物服务人员建议时，它们在临床终点立即被安乐死。将小鼠随机分配给实验或治疗组。不使用盲目。没有使用统计方法来预先确定样本量。对于PCR基因分型，使用了以下引物：P16-3MR-1、5'-AACGCAAACGCATGATGATCACTG-3';和P16-3MR-2，5'-TCAGGGATGATGCATCTAGC-3'。阳性小鼠以202 bp的形式显示带。

　　如前报道，从欧盟/FP7肌肉联盟的生物库中获得了来自接受手术的患者的大量外侧肌肉的人类肌肉活检样本，如前报道5,6。在财团的五个研究中心中的每个中心中，都获得了当地伦理委员会的伦理批准。所有参与者均提供了书面知情同意书，并在参与前进行了医学筛查。生物领域的肌肉组织已直接在熔融等异戊烷中冷冻，并在-80°C下储存直到分析。根据存在浸润的单核细胞的存在，通过形态标准鉴定出受损的区域。数据来自69、82、80、89或85岁的女性患者；平均年龄为81±7.5岁。

　　槲皮素（USP，1592409; 50 mg kg-1）和dasatinib（LC Laboratories，d-3307; 5 mg kg-1）口服（通过饲料）。对照小鼠用等量的媒介物（10％乙醇，30％聚乙醇醇和60％的phosal）给予对照小鼠。腹膜内注射GCV（Sigma-Aldrich，G2536-100mg; 25 mg kg-1）（i.p.）。抗CD36抗体（Cayman Chemical，10009893; 10 µg或20 µg的年轻小鼠中的10 µg或20 µg在PBS中稀释的老鼠中的20 µg）通过I.P.施用，对照小鼠接受相等剂量的IgA对照抗体（Southern Biotech/Bionova，0106-14）。如图传说中所示，每天连续4-7天对使用GCV，塞溶剂和CD36进行处理。NAC（Sigma-Aldrich，A9165; 0.01 G ML-1）在肌肉受伤前1周加入饮用水中（每3天交换一次），并延长至安乐死。在PBS中，在PBS中，SIS3（SIGMA-ALDRICH，S0447-5MG; 10 mg kg-1）和BORTEZOMIB（TEVA，TEVA，TEVA，TEVA，THEVA，THEVA，TEVA，THEVA，THEVA，0.5 mg kg-1）在PBS中稀释的10％DMSO中。（从1到5 d.p.i.）。对于长期治疗，分别用D+Q或GCV施用3个月大的MDX和MDX/P16-3MR小鼠，每周两次，共2个月。

　　用氯胺酮 - 二甲苯（分别为80和10 mg kg -1; i.p.）或异氟烷对小鼠进行麻醉。如前所述51，通过肌内注射CTX（Latoxan，L8102; 10 µM）诱导骨骼肌的再生。在损伤后指示的时间，将小鼠安乐死并解剖肌肉，在液氮冷却的异戊烷中冷冻，并在-80°C下储存直至分析。

　　根据先前所述的方案52进行异源实验。简而言之，将EDL肌肉从p16-3MR或WT小鼠的解剖床中移除，并将其移植到p16-3MR或WT受体的TA肌肉表面上，或者WT受体鼠标，反之亦然。移植后第7天收集肌肉移植物。

　　如先前所述的53和300B设备（Aurora Scientific）评估了EDL肌肉的离体力测量值。通过将肌肉质量除以长度和肌肉密度（1.06 mg mm -3）的产物来确定的每个肌肉面积的归一化力，以计算特定力（Mn mm -2）。

　　在体内，在p16-3MR小鼠的TA，股四头肌和胃肌肌肉中测量了肾荧光素酶活性。将麻醉的小鼠注入肌内注射腔内H（Perkinelmer，760506），并使用IVIS Lumina III（Perkinelmer）系统立即测量荧光素酶活性。在体外，使用双酸酶酶报告基因测定套件（Promega，e1910），从冷冻保存的隔膜和TA肌肉中测量了肾上腺荧光素酶活性。使用Luminometer Centro LB 960（Berthold Technologies）测量信号，并将值标准化为使用Bradford方法（蛋白质测定，Bio-Rad，500-0006）测量的总蛋白质，并在造口毒素和伊植物（H＆E）染色后测量受损区域。

　　将肌肉机械分解并在Dulbecco修改后的Eagle培养基（DMEM）中孵育，其中包含Liberase（Roche，177246）和配置酶（Gibco，17105-041）在37°C下进行搅拌1-2小时。在需要时，在第二小时内添加了蜘蛛-β-GAL试剂（Dojindo，SG02; 1 µM）。然后将上清液过滤，并将细胞在裂解缓冲液（BD Pharm Lyse，555899）中孵育10分钟，在冰上孵育10分钟，重悬于胎儿牛血清（FBS）2.5％的PBS中并计数。BV711偶联的抗CD45（BD，563709; 1：200），APC-CY7偶联的抗F4/80（Biolegend，123118; 1：200），PE-偶联的抗-α77-辛辛蛋白17-0311-82; 1：200）和PE-CY7偶联的抗SCA1（Biolegend，108114; 1：200）抗体用于分离MCS（CD45+F4/80+）（SCA1+CD45-F4/80-α7-整合素-CD31-）。PE-CY7偶联的抗CD45抗体（Biolegend，103114）用于分离CD45阳性和CD45阴性种群（扩展数据图1H）。蜘蛛-β-GAL（蜘蛛）用于分离每种细胞类型的非年代细胞（蜘蛛）（蜘蛛+）（扩展数据图1H和4A）（补充表1）。使用FACS ARIA II（BD）系统对细胞进行分类。细胞谱系通过特异性细胞标记表达（扩展数据图4b – d）证实。分离的细胞用于RNA提取，细胞培养物，植入，增殖测定，或将其镀到载玻片上（Thermo Fisher Scientific，177402）进行免疫染色和SA-β-GAL分析。

　　为了隔离Roshigh和Roslow的种群，根据制造商的协议和PE-CY7偶联的抗CD45（Biolegend，Biolegend，103114; 1：200），PE-CY7-CY7-ConJUGET ATTI-COLETED，根据制造商的协议和PE-CY7偶联的抗CD45（Biolegend，1：200），用Cellrox绿色试剂（Invitrogen，C10444; 5 µm）染色。PE共轭抗α7-积聚蛋白（ABLAB，AB10STMW215; 1：200）和APC偶联的抗SCA1（Biolegend，108111; 1：200）抗体的抗体（α7-脑臂蛋白+CD45-CD31-）和FAPS（sca1+cda1+cda1+cda1+cda）抗体1）。使用FACS ARIA II（BD）系统对Cellroxhigh和Cellroxlow细胞进行分选。分离的细胞用于细胞培养和增殖测定。使用BD FACS Diva软件进行了收购。

　　按照改编的协议54如前所述进行细胞移植。收集FACS分离的蜘蛛+和蜘蛛细胞，重悬于20％FBS DMEM培养基中，并根据制造商的说明（Invitrogen，v22889）标记为Vybrant Dil细胞标记溶液（Invitrogen，V22889），并注入了使用五天的5天5天的ta肌肉中，并将其注射到受摄取的小鼠的TA肌肉中。在移植的蜘蛛+和蜘蛛种群中控制了MC，SC和FAP的细胞类型比例。每种TA肌肉都植入了10,000个细胞，除了每种衰老细胞类型分别移植时（图5K），在那里植入了5,000个细胞。收集植入的肌肉并在细胞移植后4天处理肌肉组织学。

　　Freshly sorted cells or C2C12 cells (ATCC, CRL-1772) were transfected with siRNA targeting Cd36 (On-Target plus SmartPool, Dharmacon, L-062017-00-0005; 5 nM) or unrelated sequence as control (On-Target plus non-targeting siRNA Pool, Dharmacon, D-001810-10-05; 5 nM) using theDharmafect协议（Dharmacon，T-2003-02）。CD36 siRNA的靶序列如下：5'-CCACAUAUCUACCAAAAAUU-3'，5'-GAAAGGAUAACAUAAGCAA-3'，5'-AUACAGAGAGUUCGUAUAUCUAUCUAUCUAUACUA-3’，5'-ggauuggauggagaggagguggugugugugugugugugugugauguaug-3’。与siRNAS孵育3小时后，新鲜分类的细胞洗涤并植入。

　　根据制造商的协议，使用了细胞因子抗体阵列（R＆D Systems，ARY028； ABCAM，AB193659）。对于细胞，在无血清DMEM中培养新鲜分类的细胞24小时。收集细胞培养的上清液，离心并与膜上的膜一起孵育。对于组织间质液，解剖小鼠的骨骼肌，并用完整的无小型EDTA蛋白酶抑制剂鸡尾酒（Roche，11836170001）缓慢注入PBS溶液。然后将PBS渗出物回收离心离心，并与伴有捕获的抗体的膜一起孵育。然后将膜与检测抗体，链霉亲和素HRP和Chemi试剂混合物一起孵育。使用Chemidoc MP成像系统（Bio-Rad）捕获并可视化免疫印迹图像，并使用公开可用的ImageJ软件分析了捕获图像中每个位置的强度。

　　为了评估体内的增殖，局部CTX注射会损伤肌肉，并用Ethynyl标记的脱氧尿苷（EDU，Invitrogen，A10044; 25.5 mg kg-1; i.p.; i.p.）2 h施用小鼠。收集并处理肌肉，以通过FACS在组织载玻片或细胞分离中进行免疫荧光染色。使用Click-IT EDU成像试剂盒（Invitrogen，C10086）检测到EDU标记的细胞。将EDU阳性细胞定量为分析的细胞总数的百分比。在新鲜排序的SC上定量体外增殖，并在20％FBS HAM的F10培养基中播种，并在胶原蛋白涂层的板上补充了B-FGF（Peprotech，100-18b-250ug; 2.5 ng ML-1）。经过3天的培养后，将SC用溴脱氧尿苷（Brdu，Sigma-Aldrich，B9285-1G；1.5μgml-1）脉冲标记1小时。通过使用大鼠抗BRDU抗体（ABCAM，AB6326，1：500）和一种特定的次级生物素化驴抗鼠抗体（Jackson Immunoresearch，712-066-150，1：250），使用大鼠抗BRDU抗体进行免疫染色（ABCAM，AB6326，1：500）检测BRDU标记的细胞。使用Vectastain Elite ABC试剂（Vector Laboratories，PK-6100）和3,3'-二氨基苯胺可视化抗体结合。将BRDU阳性细胞定量为分析的细胞总数的百分比。

　　SC在3 d.p.i.的再生肌肉组织中新鲜分离。并在20％FBS DMEM中镀上24孔板（Falcon，353047），并补充了B-FGF。随后，使用相同的培养基，将中等或新鲜的蜘蛛+和蜘蛛细胞群（图5i）或依托泊苷诱导的衰老C2C12细胞（图6i）使用相同的培养基播种在0.4 µm孔大小的细胞培养物（Falcon，353495）上。经过3天的培养，如上所述，对带有BRDU标记的SC进行了增殖测定。

　　Roshigh和Roslow SC和FAPS在受伤后24小时后新鲜分离出从再生肌肉中分离出来，在NAC（10 mm）或媒介物存在的情况下播种和培养3天。处理后，固定细胞并进一步加工以进行染色。用依托泊苷（Sigma-Aldrich，E1383，1 µM）处理维持在10％FBS DMEM中的C2C12细胞5天以诱导衰老，并收集用于RNA提取和RT-QPCR。用β-半乳糖苷酶染色试剂盒染色（如下所述），以确认其衰老状态。

　　根据制造商的说明，使用衰老的β-半乳糖苷酶染色试剂盒（Cell Signaling，9860）在新鲜分类的细胞和细胞培养物中检测到SA-β-半乳糖苷酶（SA-β-GAL）活性。根据制造商的说明，用油红O（Sigma-Aldrich，O0625）染色。根据指令，使用Cellrox绿色试剂（Invitrogen，C10444; 5 µM）通过免疫荧光测量ROS水平。使用原位细胞死亡检测试剂盒，荧光素（Roche，11684795910）进行TUNEL分析，根据制造商的描述，用DNase处理的细胞用作对染色的阳性对照。

　　将肌肉嵌入OCT溶液中（Tissuetek，4583），并用液氮冷却的异戊烷中，并储存在-80°C下，直至分析。收集肌肉冷冻（厚度为10μm），并染色用于SA-β-GAL（Applichem，A1007,0001），H＆E（Sigma-Aldrich，HHS80和45235），MyH3（MyH3），MyH3（DSHB，F1.652），Siius Red（Sigma-aldrich，Sigma-Aldrich，3655548），siif sigma-alder equere interfore in sigma-aldre supplue，3655548）。H＆E-E-和MYH3抗体染色的切片的CSA，使用Image J对Sirius红色染色阳性的肌肉面积阳性和SA-β-GAL+细胞的数量进行了定量。使用阳性和阴性对照组对每种初级抗体和二级抗体的顺序添加来进行双重免疫荧光。将切片在PBS上进行气干，固定，在PBS上洗涤，并根据标准方案与原代抗体一起在室温下用高蛋白质溶液在PBS中固定1小时。随后，将载玻片用PBS洗涤，并与适当的二抗和标记染料一起孵育。如前所述56，用端粒PNA探针（Panagene，F1002-5）进行γH2AX免疫荧光染色后进行端粒免疫鱼。

　　使用直立的DMR6000B显微镜（Leica）使用DFC550摄像头来获取数字图像，用于免疫组织化学颜色图片；a Thunder imager 3D live-cell microscope (Leica Microsystems) with hardware autofocus control and a Leica DFC9000 GTC sCMOS camera, using HC PL FLUOTAR ×10/0.32 PH1 ∞/0.17/ON257C and HC PL FLUOTAR ×20/0.4 CORR PH1 ∞/0-2/ON25/C objectives;具有×20和x40空气物镜的Zeiss Cell观察者HS和Zeiss Axiocam MRX相机；和Leica SP5共聚焦激光扫描显微镜，带有HCX PL Fluotar×40/0.75和×63/0.75目标。使用405、488、568和633 nm激发线激发不同的荧光团（三个或四个）。使用Leica应用程序（V.3.0）或LAS X（V.1.0）软件（Leica）或Zeiss LSM软件Zen 2 Blue进行采集。

　　使用mirnaeasy mini试剂盒（Qiagen，1038703）从frozen肌肉中分离总RNA。Picopure（Thermo Fisher Scientific，kit0204）用于从分类细胞中分离RNA。对于RT – QPCR实验，使用2 U DNase进行10 mg RNA的DNase消化（Qiagen，1010395）。使用上标III逆转录酶从总RNA合成cDNA（Invitrogen，18080-044）。对于新鲜排序的SC，FAP和MC中的基因表达分析，根据制造商的说明，使用SsoAdvanced Preamp Supermix（Bio-Rad，172-5160）预先放大cDNA。如前所述进行qPCR反应57。一式三份进行反应，并比较样品之间自动检测到的阈值周期值。RPL7管家基因的转录本被用作内源对照，每个未知样品都归一化为RPL7含量（补充表2中提供了本研究中使用的引物的列表）。

　　测序文库直接从裂解细胞中制备，而没有先前的RNA萃取步骤。使用Smart-Seq V4超低输入RNA试剂盒进行RNA逆转录和cDNA扩增，以从Clontech Takara进行测序。Illumina Nextera XT套件用于制备放大cDNA的文库。使用Illumina Hiseq 2500 Sequencer（51 bp读取长度，单端，约2000万读）对库进行测序。

　　使用NF核/RNASEQ（v.1.2）管道进行测序读数58。使用FASTQC（V.0.11.8）59评估阅读质量。修剪盛装（V.0.5.0）60用于修剪测序读数，消除了光明适配器的遗迹，并丢弃了短于20 bp的读数。使用HISAT2（v.2.1.0）62映射所得读数（GRCM38，ENSEMBL61版本81），并使用fartialurecounts（v.1.6.2）63进行量化。每千千次映射读数（RPKM）和每百万（TPM）基因表达值的读数是根据使用BioConductOctOr封装编码器（V.3.30.0）64和R（V.3.30）64和R（V.4.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.bess）的M-VALUES（TMM）平均计数的修剪平均值计算的。使用BioConductor封装DESEQ2（V.1.28.1）66进行差异基因表达分析和PCA。使用计数数据的方差稳定化转换用于可视化PCA中的样本到样本距离。如果表现出调整后的P，认为基因被认为是差异表达的< 0.05.

　　Functional enrichment analysis of the subsets of differentially expressed genes was performed using g:Profiler web server67 with the g:SCS significance threshold, ‘only annotated’ statistical domain scope, and canonical pathway KEGG68, Reactome69 and Wiki Pathways70 sets. For each gene subset, the top five significant gene sets were selected for representation.

　　The RPKM matrix after the removal of low-count genes (edgeR (v.3.30.0)64) was used as an input for the GSEA (v.4.0.3) software71. We used the signal-to-noise metric to rank the genes, 1,000 permutations with the gene set permutation type and weighted enrichment statistics. Gene set sizes were chosen as 15–500 for MSigDB 7.0 GO:BP and 10–1,000 for MSigDB 7.0 canonical pathways (BioCarta, KEGG, PID, Reactome and WikiPathways)72. Gene sets passing the FDR < 0.25 threshold were processed for further analysis. Network representation and clustering of GSEA results were performed using EnrichmentMap (v.3.2.1)73 and AutoAnnotate (v.1.3.2)74 for Cytoscape (v.3.7.2)75 with the Jaccard coefficient set to 0.25.

　　We checked whether upregulated genes (DESeq2 adjusted P < 0.05 and log2[fold change] > 0) from each Sen versus NSen comparison can be expressed in a form of secreted proteins by combining the evidence from multiple data sources: GO76 cellular component (GO:CC), Uniprot77, VerSeDa78, Human Protein Atlas79 and experimental data reporting SASP10,80. The genes encoding extracellular (GO:CC) and/or secreted (other sources) products, with evidence from at least one source, were included in the final list of SASP genes (1,912 in total). Functional enrichment analysis was performed using the g:Profiler web server67 with the g:SCS significance threshold, ‘only annotated’ statistical domain scope, and canonical pathway sets from KEGG, Reactome and Wiki Pathways. Gene sets passing the FDR < 0.05 threshold were processed for further analysis. Network representation and clustering of the g:Profiler results were performed using EnrichmentMap (v.3.2.1) and AutoAnnotate (v.1.3.2) for Cytoscape (v.3.7.2) with the Jaccard coefficient set to 0.25.

　　We used the minimum hypergeometric test implemented in the R package mHG (v.1.1)81 for the comparative enrichment analysis of senescent cells and previously published ageing datasets: mouse36, rat (Gene expression Omnibus (GEO): GSE53960), African turquoise killifish (GEO: GSE69122), and human (GTEx82 v6p). Data processing and analysis were performed as described previously36.

　　scRNA-seq was performed using the Chromium Single Cell 3′ GEM, Library & Gel Bead Kit v3, 16 rxns (10x Genomics, PN-1000075) according to the manufacturer’s instructions and targeting a recovery of 5,000 cells per dataset. Each dataset was obtained with a sample size of two mouse biological replicates. The libraries were constructed as instructed in the manufacturer’s protocol and sequenced using the MGI DNBSEQ-T7 sequencer platform. The average read depth across the samples was 15,551 per cell. Sequencing reads were processed with STARsolo (v.2.7.3a)83 using the mouse reference genome mm10 (GENCODE vM23 (ref. 84)).

　　From the filtered barcode and count matricesm, downstream analysis was performed using R (v.4.0.3). Quality control, filtering, data clustering, visualization and differential expression analysis were performed using the Seurat (v.4.0.3) and DoubletFinder (v.2.0) R packages85,86. Datasets were processed following Seurat standard integration protocol according to the tutorial instructions. Genes expressed in less than 3 cells and cells with fewer than 500 features, less than 2,000 transcripts and more than 20% reads mapping to mitochondrial genes as well as cells identified as doublets by DoubletFinder were removed. PCA was performed for dimensionality reduction and the first 30 components were used for UMAP embedding and clustering.

　　Omni-ATAC-seq was performed in freshly sorted cells as described previously87,88. After the transposition reaction and purification, the transposed fragments were amplified using 50 μl of PCR mix (20 µl of DNA, 2.5 µl of custom Nextera PCR primers 1 and 2, and 25 µl of KAPA HiFi HS Ready Mix for a total of 15 cycles). The PCR amplification conditions were as follows: 72 °C for 5 min; 95 °C for 30 s; 15 cycles of 95 °C for 10 s, 63 °C for 30 s and 72 °C for 60 s; and a final extension at 72 °C for 5 min. After PCR amplification, the libraries were purified, and the size was selected from 150 to 800 bp using AMPure XP beads. Paired-end sequencing was performed with 50 cycles on the Illumina NovaSeq 6000 platform.

　　Read quality was assessed using FastQC (v.0.11.8). All adaptors were removed using Fastp (v.0.21.0)89. The clean reads were then aligned to mm10 mouse genome assembly using Bowtie2 (v.2.2.5)90 with the settings ‘--very sensitive’. Low-mapping-quality reads were removed using samtools (v.1.3.1)91 with the settings ‘-q 30’. BigWig files were generated using deeptools (v.3.3.1)92 with the settings ‘-normalizeUsing CPM’. Peaks were called using Macs2 (v.2.1.0)93 with the options ‘--nomodel --keep-dup -q 0.01’. For differential accessibility analysis, union peak sets were created using Bedtools (v.2.29.2)94, reads corresponding to each region were assigned by FeatureCounts. Differentially accessible peaks were identified using DESeq2 (v.1.24.0) with the criteria of adjusted P < 0.1 and an absolute value of log2[fold change] > 1. Differentially accessible peaks were further annotated by HOMER (v.4.10.4)95, the associated motif enrichment analysis was performed by HOMER using the default settings.

　　An MA plot (log2-transformed fold change versus mean average) was used to visualize changes in chromatin accessibility for all peaks. As a peak score, we used an average of TPM-normalized read counts: (1) reads per kilobase were calculated by division of the read counts by the length of each peak in kilobases; (2) the per million scaling factor was calculated as a sum of all reads per kilobase for each sample; (3) reads per kilobase were divided by the per million scaling factor; (4) peaks with the ‘promoter-TSS’ annotation TSS ± 1kb were selected and the average was calculated for each group. For MA plots, we included only those peaks with an average normalized signal >5。具有log2转换的倍数更改为> 1或 <−1 was calculated. Normalized ATAC-seq signal profiles of proximal promoters were visualized for key genes using the Integrative Genomic Viewer (v.2.8.13)96.

　　For the analysis of transcription regulation, we combined the results of several methods: (1) motif enrichment analysis of differentially expressed genes with the TRANSFAC_and_JASPAR_PWMs and ENCODE_and_ChEA_Consensus_TFs_from_ChIP-X libraries using the R package EnrichR (v.2.1)97; (2) upstream regulator analysis of differentially expressed genes using the commercial Ingenuity Pathway Analysis (IPA, QIAGEN) software98; (3) analysis of transcription factor differential expression using DESeq2 (v.1.28.1); (4) motif enrichment analysis of differentially accessible regions using HOMER (v.4.10.4).

　　Potential regulators from EnrichR and IPA results passing the threshold of P < 0.05 were used to build a union set of transcription factors, which was further filtered to retain only the molecules with DESeq2 baseMean value > 0. For further validation of the activity status, transcription factors were matched to the known HOMER motifs passing the Benjamini Q < 0.05 threshold.

　　A discrete scoring scale (inhibited, possibly inhibited, unknown/contradictory, possibly activated, activated) was used to evaluate transcription factor activity based on combined evidence from the EnrichR, IPA, DESeq2 and HOMER results. We used z-score statistics to define the activity status of transcription factors from the EnrichR analysis results by matching the differential expression of target genes with activatory and inhibitory interactions from the Bioconductor package DoRothEA (v.1.0.0)99 and the web-based TRRUST v.2 database100. To define the activity status of transcription factors from IPA upstream regulators analysis results, IPA-calculated z-score and analysis bias was taken into account. Activity predictions were further corrected by differential expression of transcription factors using DESeq2. The expression z-score statistical value was calculated to functionally classify transcription factors as activators or repressors on the basis of the proportion of upregulated and downregulated target genes. We further calculated the chromatin accessibility z-score to estimate the prevalence of HOMER motif enrichment in open versus closed regions that together with the predicted transcription factor function enabled us to validate the RNA-seq activity predictions using ATAC-seq data.

　　To estimate the level of confidence, for each enrichment result, we calculated a discrete ‘trust’ score, with each point assigned for: (1) EnrichR adjusted P < 0.05; (2) IPA P < 0.05; (3) activity status ‘activated’ or ‘inhibited’; (4) unidirectional absolute z-scores of >2 from both the EnrichR and IPA results; (5) concordance between transcription factor differential expression and the prediction of its activity score; (6) activity validated by the analysis of ATAC-seq data. Transcription factors with average trust >1处理1以进行进一步分析。

　　For each transcription factor, we merged the target genes from EnrichR and IPA results, split them into upregulated and downregulated and processed them to functional enrichment analysis of canonical pathways (KEGG, Reactome) and GO:BP using R package gprofiler2 (v.0.1.9)101 with the following parameters: correction method ‘FDR’, ‘custom_annotated’ domain score consisting of target genes for all studied transcription因素。电子GO注释被排除在外。通过FDR的基因集< 0.05 threshold were processed for further analysis. For GO:BP, we selected the gene sets with a term size of >15和<500个基因。基于Gprofiler2结果的匹配项进一步映射转录因子，以从先前在GSEA/CYTOSCAPE分析中创建的基因集的主要功能簇。

　　处理了≥8（12中的）与NSEN比较的转录因子进行进一步过滤。在每种情况下，当以下任何一个属性在给定群集的上四分位数高于上四分位数的值时：比较数量，所有元素中的GSEA项的百分比，平均最小fdr和平均信任得分的-LOG10。此外，如果转录因子与文献中的衰老有关，我们得分为1分。For graphical representation, we selected examples of transcription factors and target genes based on literature research: 19 transcription factors (out of 29 with a score of ≥2) mapped to 9 clusters (matrix remodelling/fibrosis, interferon signalling, chemotaxis, lipid uptake, IGF regulation, detoxification, gene expression and protein translation, cell cycle, and DNA repair).

　　对于上调的每个转录因子，富集和IPA结果的靶基因合并并与SASP基因列表相交。对于SASP基因，我们提取了GO：MF项，将它们聚集在12类（粘附分子，趋化因子，补体成分，细胞因子，酶，酶调节剂，细胞外基质组成部分，生长因子，激素，激素，配体，蛋白酶和受体）和使用earlikent a a pronigners a a的extriquestry a i a a ifrinction a i a a ifrinction。使用Benjamini – Hochberg程序进行了多次比较的校正。

　　在同一GO中靶向富集的转录因子：≥8（满分12）sen与NSEN比较的MF簇进行了进一步的过滤。在每种情况下，我们在以下任何一个属性中的值都高于上四分位数的情况下，对于给定群集：目标之间的分泌蛋白的百分比，P <0.05的比较数量，调整后的P <0.05，平均P值的比较数量，平均信任分数，平均信任分数，平均信任分数。此外，在群集内至少一个SASP基因的启动子区域中存在转录因子基序中，我们使用ATAC-SEQ数据分析得分为1点。对于图形表示，我们选择了17个转录因子，分数为≥2，并且具有调整后P <0.05的比较≥3。它们与五个GO：MF类别（细胞外基质组成，细胞因子，趋化因子，补体成分和生长因子）相关，为此我们选择了最常见的靶基因。

　　For the analysis of lipid metabolism, we constructed a gene set using data from multiple sources: KEGG pathway maps (fatty acid degradation, cholesterol metabolism, regulation of lipolysis in adipocytes), WikiPathways (fatty acid oxidation, fatty acid beta oxidation, mitochondrial LC-fatty acid beta-oxidation, fatty acid omega oxidation,脂肪酸的生物合成，三酰基甘油酸合成，鞘脂代谢（一般概述），鞘脂代谢（综合途径），胆固醇代谢（包括Bloch和Kandutsch -russell Pathways）和胆固醇生物合成）和胆固醇生物合成）和文献研究104.102,102,104。我们通过过滤DESEQ2结果（在12个比较中的至少3分中调整了p <0.05）并提取了log2转换的倍数变化值来绘制衰老和非酸化细胞之间表达的差异，从而进一步估计了这些基因的表达。

　　为了重建细胞 - 细胞通信网络，我们修改了基于单细胞的方法FUNRES，以说明批量基因表达曲线39。简而言之，表达值超过1 tpm的转录因子被认为是表达的。使用信号转导的马尔可夫链模型检测到调节这些转录因子的受体，以检测高概率中间信号分子105。如果（1）表达受体并调节任何转录因子，则重建两个细胞群体之间的配体 - 受体相互作用，（2）表达配体，（3）受体 - 配体相互作用包含在细胞 - 细胞相互作用中。最后，通过将平均受体和配体表达在其各自的细胞群体中乘以分配分数。通过将细胞种群标签置于100次并重新计算置换数据集中的相互作用评分来评估显着性。如果相互作用至少为2 s.d，则认为相互作用很重要。大于排列的相互作用得分的平均值。在最终网络中只保留了重要的相互作用。

　　对于功能分析，我们在3 d.p.i的老鼠中选择了三个衰老细胞种群（SC，FAPS和MC）和非元素SC种群之间的配体 - 受体相互作用。我们使用的是bioconductor套件SPIA（v.2.40.0）40，并使用降低的非疾病kegg途径图来评估途径下游配体 - 受体受体相互作用的活性。对于每种相互作用，与衰老SC相比，非年代SC中差异表达的目标转录因子被分为上调并下调。作为一组基因，我们从研究的所有相互作用中列出了目标转录因子的列表。用2,000个排列进行了SPIA分析，并使用Fisher的产品方法合并了PPERT和PNDE。通过PGFDR <0.05阈值的途径被认为显着富集。对于每种途径，我们计算了激活或抑制分析相互作用总数的配体 - 受体相互作用的比率。对于结果表示，我们选择了八个激活和八个抑制途径，相互作用比最高。

　　每个实验组的样本量在相应的图标题中描述，除非另有说明，否则所有实验均以至少三个生物学重复进行。除测序数据分析外，所有统计分析都使用GraphPad Prism。以直方图显示的定量数据表示为平均值±S.E.M.（表示为错误条）。平均每个组的结果来计算描述性统计。除非另有说明，否则使用Mann – Whitney U检验（独立样本，两尾）进行比较。P <0.05被认为具有统计学意义。实验不是随机的。

　　有关研究设计的更多信息可在与本文有关的自然投资组合报告摘要中获得。