This research complies with all relevant ethical regulations and was approved by the relevant institutional review boards and ethics committees. All animal work performed in this study was approved by the Institutional Animal Care and Use Committee at The Jackson Laboratory (Bar Harbor, Maine, USA) (protocol #12004-01). Collection and use of fresh femoral head specimens from individuals undergoing elective total hip replacement was approved by the Research Ethics Board of the University Health Network (UHN) (Toronto, Ontario, Canada) (protocol #REB 22-5692). Collection and use of fresh femoral head specimens from individuals undergoing total hip arthroplasty was approved by the ethics board at the Technical University of Munich (Munich, Germany) (protocol #399/21).
Mice
All Mus musculus strains were obtained from The Jackson Laboratory. C57BL/6J (RRID:IMSR_JAX:000664) mice, B6.SJLPtprcaPepcb/BoyJ (RRID:IMSR JAX:002014, referred to as CD45.1+) mice and B6.Cg-Tg(Cdkn2a/luc/RFP/TK)1Cmps/J (RRID:IMSR JAX:03704542, referred to as p16-3MR) mice were obtained from and housed within The Jackson Laboratory. B6(Cg)-Dnmt3atm1Trow/J (RRID:IMSR JAX:03228913, referred to as Dnmt3afl-R878H/+) mice were crossed to B6.Cg-Tg(Mx-cre)1Cgn/J (RRID:IMSR JAX:00355662, referred to as Mx-Cre) mice or the combination of C57BL/6N-Fgd5tm3(cre/ERT2)Djr/J (RRID:IMSR JAX:02778963, referred to as Fgd5-CreER) and B6.Cg-Gt(ROSA)26Sortm14(CAG−tdTomato)Hze/J (RRID:IMSR JAX:00791464, referred to as LSL-tdTomato) mice. Dnmt3afl-R878H/+ Mx-Cre mice were crossed to C57BL/6J-Ptprcem6Lutzy/J (RRID:IMSR JAX:033076, referred to as JAXBoy) mice to generate CD45.1+ Dnmt3afl-R878H/+ Mx-Cre mice. B6;129S-Tet2tm1.1Iaai/J (RRID:IMSR JAX:01757331, referred to as Tet2KO) mice were crossed to C57BL/6-Tg(Tal1−cre/ERT)42-056Jrg/J (RRID:IMSR JAX:03746665, referred to as Scl-CreER) mice and LSL-tdTomato mice. To study leukaemic transformation, Dnmt3afl-R878H/+ Mx-cre mice were crossed with C57BL/6-Npm1tm1Trow/J (RRID:IMSR JAX:03316413, referred to as Npm1frt−cA/+) mice and B6N.129S6(Cg)-Gt(ROSA)26Sortm3(CAG−flpo/ERT2)Alj/J (RRID:IMSR JAX:01901666, referred to as R26FlpoER) mice. For AML models, Cre-inducible Npm1c mice67 were crossed with Cre-inducible Flt3GL−ITD mice47 and B6.Cg-Ndor1Tg(UBC−cre/ERT2)1Ejb/1J (RRID:IMSR JAX:00700168, referred to as UBC-Cre-ERT2) mice, or Cre-inducible Idh1R132H mice69 were crossed with Cre-inducible Npm1c, Cre-inducible C57BL/6-Nrastm1Tyj/J (RRID:IMSR JAX:00830470, referred to as NrasG12D) and UBC-Cre-ERT2 mice. Mice carrying the Mx-Cre allele were given 15 mg kg−1 polyinosinic–polycytidylic acid (poly(I:C), InvivoGen) by intraperitoneal (i.p.) injection every other day for a total of five doses between two and four months of age. After pIpC, genomic DNA was extracted from peripheral blood cells for recombination polymerase chain reaction (PCR). Using Mx-Cre, we consistently observe recombination efficiencies of >90% by PCR (Extended Data Fig. 1a). Mice carrying the Fgd5-CreER allele were given 125 mg kg−1 d−1 TAM by oral gavage for three days. Mice carrying the Scl-CreER or UBC-Cre-ERT2 alleles were given 100 mg kg−1 d−1 TAM by oral gavage for two days. Mice carrying the R26FlpoER were given 125 mg kg−1 d−1 TAM by oral gavage for three days. Following poly(I:C) or TAM treatment, mice recovered for >5 weeks before use in experiments. Mice were housed in cages under specific pathogen-free conditions in a 12/12-h light/dark environment, with temperatures of 65–75 °F with 40–60% humidity, and food and water provided ad libitum. All mice were used at 2–4 months of age except where specified in the figures, and both males and females were used for experiments except for transplantation experiments, where female mice were used.
Human subjects
For human scRNA-seq, fresh femoral head specimens were collected from individuals undergoing elective total hip replacement at Toronto Western Hospital (University Health Network, UHN). Written informed consent was obtained from all participants according to procedures approved by the Research Ethics Board of the UHN. Participants were not financially compensated. Donors were between 52 and 81 years of age and had no known autoimmune conditions, cancers or haematologic disorders. Samples from both males and females were included. Sample details are provided in Supplementary Table 3.
For bulk RNA-seq, co-culture and cytokine culture analyses, human BM samples were obtained from femoral heads of haematologically healthy individuals undergoing total hip arthroplasty, and mutational analysis was performed to assess CH as previously described71. All tissue was taken following informed consent and under ethics approval from TU Munich 339/21. Participants were not compensated financially. Donors were between 18 and 80 years of age. Samples from both males and females were included. Sample details are provided in Supplementary Table 3.
BM transplantation
For the transplantation experiment for scRNA-seq, 3 × 106 BM cells from Dnmt3a+/+ Fgd5-CreER LSL-tdTomato or Dnmt3aR878H/+ Fgd5-CreER LSL-tdTomato donor mice were transplanted by retroorbital (RO) injection into nine-month-old CD45.1+ recipient mice treated with busulfan (20 mg kg−1 d−1 i.p. for three days) prior to transplant. For busulfan-conditioned transplant experiments into young CD45.1+ recipient mice, 3 × 106 BM cells from Dnmt3a+/+ Mx-Cre or Dnmt3a+/+ Fgd5-CreER or Dnmt3aR878H/+ Mx-Cre or Dnmt3aR878H/+ Fgd5-CreER donor mice were transplanted by RO injection into young (3 months) CD45.1+ recipient mice treated with busulfan (20 mg kg−1 d−1 i.p. for three days) before transplant. For non-conditioned transplant experiments, 5 × 106 BM cells from Dnmt3a+/+ Mx-Cre or Dnmt3aR878H/+ Mx-Cre mice were transplanted by RO injection into young (2 months) or middle-aged (13 months) CD45.1+ recipient mice over three consecutive days. For transplant experiments into young p16-3MR mice, 3 × 106 BM cells from CD45.1+ Dnmt3a+/+ Mx-Cre or CD45.1+ Dnmt3aR878H/+ Mx-Cre mice were transplanted by RO injection into young (3 months) CD45.2+ recipient mice treated with busulfan (20 mg kg−1 d−1 i.p. for three days) before transplant. For transplant experiments assessing AML transformation, 3 × 106 BM cells from Dnmt3a+/+ Mx-Cre Npm1+/+ R26FlpoER or Dnmt3aR878H/+ Mx-Cre Npm1frt−cA/+ R26FlpoER mice were transplanted by RO injection into young (3 months) CD45.1+ recipient mice treated with busulfan (20 mg kg−1 d−1 i.p. for three days) before transplant. For secondary BM transplant, 3 × 106 BM cells from primary transplant recipients were transplanted by RO injection into young (3 months) CD45.1+ recipient mice treated with irradiation (10 Gy) prior to transplant. For transplant of AML, 5 × 105 BM cells from Npm1/Flt3ITD mice or Idh1/Npm1/NrasG12D mice were transplanted by RO injection into young (3 months) CD45.1+ recipient mice treated with busulfan (20 mg kg−1 d−1 i.p. for three days) or irradiation (10 Gy).
BM isolation and cell sorting for scRNA-seq
For mouse haematopoietic cell isolation, single-cell suspensions of BM were prepared by filtering crushed femurs, tibias and iliac crests from each mouse. Red cells were lysed using 1X red blood cell lysis buffer (ThermoFisher). Cells were then stained with a combination of fluorochrome-conjugated antibodies: CD45.1 BV650 (clone A20, BioLegend cat. no. 110736, dilution 1:200), CD45.2 A700 (clone 104, BioLegend cat. no. 109822, dilution 1:50), c-Kit APC (clone 2B8, BD Biosciences cat. no. 553356, dilution 1:200), Sca-1 BV510 (clone D7, BioLegend cat. no. 108129, dilution 1:200), CD150 BV785 (clone TC15-12F12.2, BioLegend cat. no. 115937, dilution 1:200), CD48 PE-Cy7 (clone HM48-1, BioLegend cat. no. 103424, dilution 1:200), Flt3/CD135 BV421 (clone A2F10, BioLegend cat. no. 135313, dilution 1:50), CD34 FITC (clone RAM34, BD Biosciences cat. no. 553733, dilution 1:50), FcgR BV711 (clone 93, BioLegend cat. no. 101337, dilution 1:200), CD11b BV605 (clone M1/70, BioLegend cat. no. 101257, dilution 1:200), a mature lineage (Lin) marker mix including B220 PE-Cy5 (clone RA3-6B2, BioLegend cat. no. 103210, dilution 1:200), CD4 PE-Cy5 (clone RM4-5, BioLegend cat. no. 100514, dilution 1:200), CD5 PE-Cy5 (clone 53-7.3, BioLegend cat. no. 100610, dilution 1:200), CD8a PE-Cy5 (clone 53-6.7, BioLegend cat. no. 100710, dilution 1:200), Ter119 PE-Cy5 (clone Ter119, BioLegend cat. no. 116210, dilution 1:200) and Gr-1 PE-Cy5 (clone RB6-8C5, Biolegend cat. no. 108410, dilution 1:200), and a viability stain. For isolation of non-haematopoietic cells, femurs, tibias and iliac crests were cut in half then spun into 100 μl of phosphate-buffered saline (PBS). The cell pellet was resuspended in 5 ml of digest solution (2 mg ml−1 collagenase IV and 1 mg ml−1 dispase in 1X Hank’s balanced salt solution (HBSS)) in one well of a six-well plate. The bone was chopped into bone chips and placed into 5 ml of digest solution in one well of a six-well plate. The plate was incubated at 37 °C in 5% CO2 for 10 min. Bone chips were mixed, digest solution was removed from the BM plug and added to ice-cold fluorescence-activated cell sorting (FACS) buffer (3% FBS in PBS) and replaced with fresh 5 ml of digest solution. This was repeated three times, and all digest buffer was pooled into ice-cold FACS buffer and filtered. Cells were then stained with a combination of fluorochrome-conjugated antibodies: CD45 BV650 (clone 30-F11, BD Biosciences cat. no. 563410, dilution 1:100), Ter119 PE-Cy5 (clone Ter119, BioLegend cat. no. 116210, dilution 1:200), CD41 FITC (clone MWReg 30, BioLegend cat. no. 133904, dilution 1:200), CD31 APC (clone MEC13.3, BioLegend cat. no. 102510, dilution 1:200), CD71 PE (clone R17217, Invitrogen cat. no. 12-0711-82, dilution 1:200), CD51 BV711 (clone RMV-7, BD Biosciences cat. no. 740755, dilution 1:100) and a viability stain. From each individual mouse, haematopoietic and non-haematopoietic cell fractions were kept separate and serially sorted into the same collection tube. First, haematopoietic cell populations were sorted on a FACSAria II (BD): 5,000 mature haematopoietic cells (CD45.2+ tdTomato+ Lin+), 6,000 committed haematopoietic progenitors (CD45.2+ tdTomato+ Lin− c-Kit+ Sca1−) and 7,000 haematopoietic stem and multipotent progenitor cells (CD45.2+ tdTomato+ Lin− c-Kit+ Sca1+) into a 1.5-ml tube with 10 μl of Iscove’s modified Dulbecco’s medium (IMDM) + 10% FBS at 4 °C. Next, this tube was transferred to sort non-haematopoietic cell populations using a FACSymphony S6 instrument (BD): 2,000 endothelial cells (CD45.2− Ter119− CD41− CD31+) and 8,000 MSCs (CD45.2− Ter119− CD41− CD31− CD71− CD51+). Cells were immediately counted on a Countess II automated cell counter (ThermoFisher) and 12,000 cells were loaded on to one lane of a 10X Chromium microfluidic chip (10X Genomics).
For human scRNA-seq, targeted sequencing of matched BM and peripheral blood was carried out by the Advanced Molecular Diagnostics Laboratory at the Princess Margaret Cancer Centre using a single-molecule molecular inversion probe (SmMIP) panel encompassing 36 genes frequently mutated in myeloid clonal haematopoiesis72. Low quality (based on assay parameters), common or germline (based on MAF), and likely benign (based on variant type and Combined Annotation-Dependent Depletion (CADD) score) variants were filtered out. Individuals were designated as CHIP-positive when at least one high-confidence variant with VAF ≥ 2% was detected. CHIP samples had mutations in DNMT3A, TET2, ASXL1 and/or CCAR1. Processing of femoral head tissue was performed following previously reported procedures19. BM cells were flow-sorted to exclude haematopoietic (CD45+), erythroid (GlyA+) and plasma (CD38+) populations. The enriched non-haematopoietic cell fraction was used for scRNA-seq.
scRNA-seq
For mouse scRNA-seq, single-cell capture, barcoding and library preparation were performed using 10X Chromium version 3.1 NEXTGEN chemistry according to the manufacturer’s protocol. Before sequencing, quality control of the cDNA libraries was assessed using an Agilent 4200 Tapestation set-up and quantified by KAPA qPCR. Libraries were sequenced using an Illumina NovaSeq 6000 S4 flow cell lane targeting 6,000 barcoded cells with an average sequencing depth of 75,000 reads per cell. Illumina base call files for all libraries were demultiplexed and converted to FASTQs using Bcl2fastq v2.20.0.422. The Cell Ranger pipeline (10X Genomics) was used to align reads to the mouse reference GRCm38.p93 (mm10 10X Genomics reference 2020-A), deduplicate reads, call cells, and generate cell-by-gene digital count matrices. Matrices were imported into PartekFlow (version 10.0.23.0720) for downstream analysis including log transformation, principal component analysis (PCA), graph-based clustering using the Louvain algorithm from the top 18 principal components, t-distributed stochastic neighbour embedding (t-SNE) visualization, and pathway enrichment analysis. The top 25 marker genes per cluster (Supplementary Table 1) were cross-referenced with published scRNA-seq profiles to annotate each population. Wild-type (+/+) and mutant (R878H) Dnmt3a transcripts were identified using samtools v1.13 mpileup73, and cells were labelled with mutation status using base functions in R v4.2.2. The MSC cluster was further subclustered to remove residual haematopoietic cells using Seurat v5.3.074 in R. Differential gene expression (DGE) analysis by genotype within each cluster was performed using the Hurdle model in Partek, equivalent to MAST75.
For human scRNA-seq, single-cell transcriptomic libraries were generated using either the Chromium Next GEM Single Cell 3′ or Chromium GEM-X 3′ kit (10X Genomics) and sequenced according to the manufacturer’s protocols. Reads were aligned to the GRCh38 reference genome with Cell Ranger. Cells containing fewer than 300 detected genes, greater than 10% mitochondrial gene content, or identified as doublets via scDblFinder were excluded. scRNA-seq analysis was performed using a standard Seurat workflow: gene expression was normalized (LogNormalize), the top 2,000 variable genes were identified (FindVariableFeatures), expression was scaled (ScaleData), and PCA was performed (RunPCA) and used for batch correction (RunHarmony) based on sequencing chemistry. Uniform manifold approximation and projection (UMAP) dimensionality reduction (RunUMAP) and clustering (FindNeighbors, FindClusters) were performed using the first 20 dimensions of the batch-corrected PCA. Cluster annotation used presto for DGE analysis combined with assessment of top marker genes. MSCs were identified using previously published signatures19 and characterized based on high expression of NGFR, PDGFRA, CXCL12, LEPR, PPARG, LPL, APOE, SPP1, NCAM1 and WIF1. For DGE analysis between CHIP and non-CHIP samples, a pseudobulk approach was applied; samples with fewer than 50 cells per population were excluded. DESeq2 was used to model CHIP status while controlling for age, obesity, sex and sequencing chemistry76. Variance-stabilized (VST) normalized counts were further adjusted for chemistry differences using the limma function removeBatchEffect.
GSEA
GSEA with GO77, Hallmark78, Reactome79 and other curated gene sets listed in Supplementary Table 2 was performed using GSEA software79,80. A separate pre-ranked GSEA was performed to test specific SASP and inflammatory signatures, as described in the following.
Senescence score computation
A composite senescence score was computed using the FRIDMAN_SENESCENCE_UP gene set18 (77 genes upregulated during cellular senescence). For mouse and human scRNA-seq data, per-cell module scores were computed using the AddModuleScore function in Seurat74 with default parameters, which calculates the average expression of the signature genes relative to a set of randomly sampled control genes matched for expression level. For bulk human MSC RNA-seq data from co-culture experiments, per-sample scores were calculated as the mean expression value across all signature genes detected in the dataset, without prior per-gene standardization.
GO term biological category classification
To reduce redundancy in the GO enrichment output and facilitate cross-cell-type comparison, all significantly enriched GO terms (FDR < 0.05) were assigned to one of 17 predefined biological categories using a keyword-based classification applied to GO term descriptions. Categories were defined as: General Metabolism & Regulation, Development & Differentiation, Inflammation & Immune Response, Transport & Localization, Transcription & Epigenetics, Vesicle & Secretory Trafficking, Cell Cycle & Proliferation, Apoptosis & Cell Death, Cytoskeleton & Motility, Signalling Pathways, Translation & Ribosome, Energy & Mitochondrial Metabolism, Proteostasis, Lipid Metabolism, DNA Damage & Repair, Oxidative Stress & ROS, and ECM & Cell Adhesion. Each term was assigned to the first matching category in a fixed priority order using a curated keyword list matched against the lower-cased GO term description; terms matching no category were retained as unclassified. All GO terms passing the FDR threshold were included; none were excluded after categorization. Six categories were designated senescence-relevant based on their established roles in senescence and SASP: Senescence & Ageing, Cell Cycle & Proliferation, DNA Damage & Repair, Oxidative Stress & ROS, ECM & Cell Adhesion, and Inflammation & Immune Response. Full category assignments and enrichment statistics are provided in the source data.
Pre-ranked GSEA for SASP and inflammatory MSC signatures
To test whether SASP and inflammatory gene signatures are differentially enriched in MSCs from mice with Dnmt3a-mutant versus control haematopoiesis, pre-ranked GSEA was performed using the GSEA desktop application (Broad Institute, v4.3.2) in GseaPreranked mode. All expressed genes were ranked by log2(fold change) derived from the Hurdle model output, from most upregulated in mutant to most upregulated in control, without filtering for significance. Five gene sets were tested: (1) a curated SASP signature derived from published secretory phenotype gene lists; (2) an inflammatory MSC (iMSC) gene signature adapted from Wang et al.22 comprising genes upregulated in BM MSCs of multiple myeloma patients compared to healthy donors, with mouse orthologues identified using published one-to-one orthologue mappings; (3) Hallmark IFN-α response; (4) Hallmark inflammatory response; and (5) Hallmark IFN-γ response (gene sets iii–v from MSigDB v7.575). Normalized enrichment scores (NES) and FDR q-values were calculated using 1,000 permutations; gene sets with fewer than ten members detected in the ranked list were excluded.
Leading edge gene extraction
Leading edge subsets were extracted from the GSEA results (.edb) output files for each gene set. For positively enriched gene sets (enrichment score > 0), the leading edge comprises genes ranked above the peak enrichment score position; for negatively enriched gene sets (enrichment score < 0), it comprises genes ranked below that position. The ranking score shown in Extended Data Fig. 1k is the log2(fold change) from the pre-ranked input list, with positive values indicating higher expression in mutant MSCs and negative values indicating higher expression in control MSCs. Genes appearing in the leading edge of more than one downregulated gene set are annotated as ‘Multiple’ in the figure.
Ranked scatter plot of all enriched GO terms
All GO terms significantly enriched in MSC DEGs (n = 3,227 terms at FDR < 0.05) were plotted in rank order by enrichment score on a log-scale y axis. Each point represents one GO term, colour-coded by biological category as defined above; terms not assigned to a senescence-relevant category are shown in grey.
Human gene-set z-score analysis
To assess whether the dissociation between SASP and inflammatory/iMSC signature enrichment observed in mouse MSCs is conserved in human CHIP, gene-set z-scores were calculated from VST-normalized, chemistry-corrected pseudobulk expression data generated from primary human MSCs from CHIP carriers (n = 6) and non-carrier controls (n = 7). For each gene in each signature, expression values were standardized across all samples by subtracting the gene mean and dividing by the gene standard deviation. The per-sample gene-set score was then calculated as the mean standardized value across all genes in the signature that were detected in the dataset. Differences between groups were tested using an unpaired two-sample t test.
MSC quantification
We evaluated staining of common markers used to define MSCs, including CD51 BV711 (clone RMV-7, BD Biosciences cat. no. 740755, dilution 1:100), PDGFRα APC/eFluor780 (clone APA5, Invitrogen cat. no. 47-1401-82, dilution 1:50) and biotinylated LepR (R&D Systems cat. no. BAF497, dilution 1:200) with streptavidin-BV421 (BioLegend cat. no. 405226, dilution 1:200) (Extended Data Fig. 2b–f). Co-expression of PDGFRα and LepR defined a population of CD51+ stromal cells as MSCs and PDGFRα+ LepR+ cells were not present in gated CD31+ ECs within the same stained sample. Thus, we used the EC population as an internal negative gating control within each sample. PDGFRα+ stromal cells are strongly enriched for LepR+ cells, whereas LepR+ stromal cells are strongly enriched for PDGFRα+ cells. This is highly consistent with the original report by Zhou et al. showing that nearly all LepR+ CD45− Ter119− CD31− BM stromal cells were positive for PDGFRα and nearly all PDGFRα+ CD45− Ter119− CD31− BM cells were LepR, and these MSC populations make up ~0.3% of the whole bone marrow (WBM)81.
MSC senescence quantification
Individual senescence readouts have limited sensitivity, so we applied rigorous complementary analyses to validate MSC senescence. We first developed our staining strategy for detecting senescent MSCs using wild-type young and old C57BL/6J mice (Extended Data Fig. 2g–k). To assess marker concordance, we examined co-expression patterns within PDGFRα+ LepR+ MSCs from old mice. Bcl-2+ Bcl-xL+ MSCs were markedly enriched for SA-β-galhi cells, and SA-β-galhi MSCs showed increased representation of Bcl-2+ Bcl-xL+ cells. Correlation analyses across all PDGFRα+ LepR+ MSCs from young and old mice revealed significant positive correlations between SA-β-gal mean fluorescence intensity (MFI), frequency of Bcl-xL+, frequency of Bcl-2+, and median SSC, which was independent of MSC abundance. Gene expression analysis on sorted MSCs from young and old mice further demonstrated that SA-β-gal MFI strongly correlated with transcript levels of Cdkn2a (p16), Cdkn1a (p21) and Trp53 (p53). Taken together, these data confirm that SA-β-gal, Bcl-2, Bcl-xL, Cdkn2a, Cdkn1a and Trp53 collectively identify a coherent senescent MSC population.
As senescent mesenchymal cells marked by p16 and BCL-2 have been demonstrated to increase with ageing26,27,28, we further validated our approach of assessing MSC senescence using wild-type young and old C57BL/6J mice (Extended Data Fig. 3a–n). In old mice, PDGFRα+ LepR+ MSCs showed increased SA-β-gal MFI, increased frequency of Bcl-xL+ Bcl-2+ MSCs and a higher proportion of senescent (SA-β-galhi Bcl-xL+ Bcl-2+) MSCs compared to young mice. Consistent with established features of senescence, senescent MSCs (SA-β-galhi Bcl-xL+ Bcl-2+) also exhibited increased forward scatter (size) and side scatter (granularity) compared to non-senescent MSCs (SA-β-gallo Bcl-xL⁻ Bcl-2⁻) within the same mice. We next validated these markers using an irradiation-induced DNA damage model. Irradiated (5 Gy) young p16-reporter mice displayed increased SA-β-gal levels in PDGFRα+ LepR+ MSCs, a higher frequency of Bcl-xL+ Bcl-2+ MSCs, and an increase in senescent MSCs (SA-β-galhi Bcl-xL+ Bcl-2+) compared to non-irradiated controls. Irradiation also induced an increase in p16-mRFP reporter signal in MSCs. SA-β-galhi MSCs were enriched for p16-mRFPhi cells, and SA-β-galhi p16-mRFPhi MSCs increased after irradiation. These SA-β-galhi p16-mRFPhi senescent MSCs exhibited increased granularity and size compared to SA-β-gallo p16-mRFPlo MSCs, mirroring the phenotype observed using SA-β-gal, Bcl-2 and Bcl-xL markers.
Human MSC and HSPC co-culture
Trabecular bone was minced, and mononuclear cells were extracted by gradient separation using Ficoll. MSCs were further isolated by plastic adherence and cryopreserved until use. CD34+ HSPCs were isolated from cryopreserved MNCs using the human CD34 MicroBead Kit (Miltenyi). For RNA-seq experiments, a total of 0.5 × 105 CD34+ cells ml−1 were seeded on top of a human MSC layer at a 1:2 ratio in serum free medium (80% IMDM with 20% BIT9500), supplemented with 10 µM β-mercaptoethanol, 0.4% ciprofloxacin, 2% FCS-MSC graded, and human stem cell factor (SCF; 10 ng ml−1), Flt3L (10 ng ml−1) and thrombopoietin (TPO) (2.5 ng ml−1). After 42 h of culture at 37 °C and 5% CO2, cells were collected and sorted by FACS to isolate pure HSPC (CD45+ CD34+) and MSC (CD45− CD90+) populations. MSCs were used as input into the Prime-seq protocol82 to prepare the libraries, which were sequenced using a NovaSeq6000 instrument. GSEA was performed with counts per million (c.p.m.) values and the SAUL_SEN_MAYO dataset33. For SA-β-gal staining experiments, a total of 0.1 × 105 CD34+ cells ml−1 were seeded on top of a human MSC layer at a 1:1 ratio in serum free medium (80% IMDM with 20% BIT9500), supplemented with 10 µM β-mercaptoethanol, 0.4% ciprofloxacin and human SCF (100 ng ml−1), Flt3L (100 ng ml−1) and TPO (25 ng ml−1). After six days of culture at 37 °C and 5% CO2, MSCs were collected and stained for MSC surface markers and β-galactosidase by flow cytometry. For cytokine treatment experiments, 3 × 104 hMSCs were seeded in 24-well plates and cultured for two days with αMEM supplemented with 1% Pen/Strep, 10 U ml−1 heparin, 1% L-glutamine and 10% human platelet lysate. Cells were then treated with 5 ng ml−1 recombinant human TNF-α (PeproTech), 50 ng ml−1 recombinant human IL-6 (PeproTech) or 10 ng ml−1 recombinant human IL-1β (PeproTech) for four days. Cells were washed with PBS and stained with 25 μl of 1,000× diluted Zombie-NIR (BioLegend), followed by 15-min incubation at room temperature in the dark. Cells were then washed with FACS buffer and subjected to surface marker antibody staining using CD105 FITC (clone SN6/N1-3A1, Ancell cat. no. 326-040, dilution 1:100), CD90 PE (clone 5E10, BioLegend cat. no. 328110, dilution 1:100) or CD90 APC (clone 5E10, BioLegend cat. no. 328114, dilution 1:100), CD73 PE-Cy7 (clone AD2, BioLegend cat. no. 344010, dilution 1:100) and CD45 eFluor450 (clone HI30, ThermoFisher Scientific cat. no. 48-0459-42, dilution 1:100). The following panel was used for MSCs: CD45− CD73+ CD90+. Cells were stained with the antibody cocktail for 20 min at 4 °C in the dark. Subsequently, the Cell Event Senescence Kit (ThermoFisher) was used for β-galactosidase staining. Cells were fixed in 2% paraformaldehyde (PFA), incubated for 10 min at room temperature, and stained in 0.25× staining mix solution for 1 h at 37 °C without CO2. Samples were analysed using the CytoFLEX S flow cytometer (Beckman Coulter). FlowJo V10 was used for data analysis.
Flow cytometry of mouse cells
The following surface marker combinations, using the antibodies described above, were used for cell sorting and/or analysis: HSCs (Lin− Sca-1+ c-Kit+ Flt3− CD150+ CD48−), MPPG/M (Lin− Sca-1+ c-Kit+ Flt3− CD150− CD48+), LSK cells (Lin− Sca-1+ c-Kit+), post-culture HSPCs (Lin− Sca-1+ Flt3− CD150+ CD48−), GMPs (Lin− Sca-1− c-Kit+ CD16/32+ CD34+), B cells (B220+ CD11b− CD3e−), T cells (CD3e+ B220− CD11b−), myeloid cells (CD11b+ B220− CD3−), granulocytes (CD11b+ B220− CD3− Ly6g+ Ly6c+), in vivo MSCs (CD45− CD41− Ter119− CD31− CD51+ PDGFRα+ LepR+), ex vivo MSCs (CD45− CD41− Ter119− CD31− CD51+ PDGFRα+) and endothelial cells (CD45− CD41− Ter119− CD31+). Additional antibodies included CD126 (IL6R) BUV563 (clone D7715A7, BD Biosciences cat. no. 741296, dilution 1:50), CD120a (TNFR Type 1/p55) PE (clone 55R-286, BioLegend cat. no. 113003, dilution 1:50) and CD120b (TNFRII) PE-Vio770 (clone REA1, Miltenyi Biotec cat. no. 130-104-746, dilution 1:50). Antibody cocktails were incubated with BM cells for >30 min at 4 °C. For secondary antibody staining, cells were stained first with cell-surface antibodies as described above. The cells were then washed and stained with streptavidin-BV421 at 4 °C for 30 min. For β-galactosidase staining, cells were incubated with C12FDG (5-dodecanoylaminofluorescein di-β-D-galactopyranoside, ThermoFisher) for 2 h at 37 °C and washed twice before staining with surface marker antibodies. For the CellROX Green Reagent stain (ThermoFisher), cells were incubated with CellROX Green for 30 min at 37 °C and washed three times before staining with surface marker antibodies. For the cell tracer experiments cells were incubated with CellTrace Yellow (ThermoFisher) for 20 min at 37 °C and washed twice before the addition of conditioned media. For experiments using Bcl-xL A700 (clone 54H6, Cell Signaling Technology cat. no. 48382S, dilution 1:100) or Bcl-xL AF488 (clone 54H6, Cell Signaling Technology cat. no. 2767S, dilution 1:100), Bcl-2 PE-Vio770 (clone REA356, Miltenyi Biotec cat. no. 130-105-475, dilution 1:20), pStat3 (Tyr705) AF488 (clone D3A7, Cell Signaling Technology cat. no. 4323S, dilution 1:100), phospho-p44/42 MAPK (Thr202/Tyr204) Alexa Fluor 647 (clone E10, Cell Signaling Technology cat. no. 4375S, dilution 1:100), phospho-NF-κB p65 (Ser536) PE (clone 93H1, Cell Signaling Technology cat. no. 5733S, dilution 1:100) or phospho-AKT (Ser473) PE-CF574 (clone M89-61, BD Biosciences cat. no. 562465, dilution 1:100), BM cells were first stained with cell-surface antibodies, then fixed and permeabilized using the FIX & PERM cell permeabilization kit (ThermoFisher), as per the manufacturer’s instructions, before intracellular staining. Stained cells were sorted using a FACSAria II or a FACSymphony S6 Sorter, or analysed on a FACSymphony A5 instrument. FlowJo V10 was used for data analysis.
Real-time PCR for senescence genes
RNeasy Micro Kit (Qiagen) was used to extract whole-cell RNA. A qRT–PCR assay was performed using Power SYBR Green PCR Master Mix (Applied Biosystems). PCRs had a pre-amplification (50 °C/120 s, 95 °C/10 min) and were then amplified for 40 cycles (95 °C/15 s, 60 °C/60 s), and the final step was a melt curve (95 °C/15 s, 60 °C/60 s 95 °C/15 s) on a QuantStudio 7 Flex set-up (ThermoFisher). Messenger RNA (mRNA) expression was normalized against hypoxanthine guanine phosphoribosyltransferase (Hprt) using the comparative cycle threshold method. Custom primers were designed and purchased from Integrated DNA Technologies; sequences are shown in Supplementary Table 4.
Mouse peripheral blood analysis
Peripheral blood was collected from mice via retro-orbital bleed. Red blood cells were lysed, and cells were stained with CD45.1 BV650 (clone A20, BioLegend cat. no. 110736, dilution 1:200), CD45.2 A700 (clone 104, BioLegend cat. no. 109822, dilution 1:50), B220 BUV496 (clone RA3-6B2, BD Biosciences cat. no. 612950, dilution 1:200), CD3e PerCP-Cy5.5 (clone 145-2C11, BioLegend cat. no. 100328, dilution 1:200), CD11b APC-Cy7 (clone M1/70, BioLegend cat. no. 101266, dilution 1:200), Ly6g APC (clone 1A8, BioLegend cat. no. 127614, dilution 1:200), Ly6c BV605 (clone HK1.4, BioLegend cat. no. 128035, dilution 1:200), F4/80 PE-Cy7 (clone BM8, Invitrogen cat. no. 25-4801−82, dilution 1:200) and Ter119 BV421 (clone Ter119, BioLegend cat. no. 116233, dilution 1:50). Data were captured using an LSRII instrument (BD) and analysed using FlowJo V10. Complete blood counts were performed on whole blood using an Advia 120 Hematology Analyzer (Siemens).
Luminex analysis
For multiplexed ELISA (Luminex) experiments following transplant with Dnmt3aR878H/+ Mx-Cre BM, one femur and one tibia were cut in half then spun into 100 μl of PBS. For experiments using conditioned media, BM was prepared by isolating, filtering and crushed pooled tibias, femurs and iliac crests of each mouse. Haematopoietic stem and progenitor cells were isolated using CD117 MicroBeads, mouse (Miltenyi), and cultured in StemSpan SFEM II (StemCell Technologies) and 10 ng ml−1 recombinant murine SCF (StemCell Technologies) for seven days. Cells were centrifuged at 300g for 5 min at 4 °C, then the supernatant was collected and spun again at 1,300g for 10 min at 4 °C to isolate conditioned medium. For SASP analysis, HSPC CM was collected as described above and placed on 75,000 MSCs in a 24-well plate for seven days. The HSPC CM was then washed off and replaced with normal MSC growth medium for three days. The medium was collected and centrifuged at 300g for 5 min at 4 °C and the supernatant was collected and spun again at 1,300g for 10 min at 4 °C before analysis. Custom Luminex assays were used to assess analyte levels including mouse GDF-15, IL-6, TNF-α, IL-1α, IL-1β, MMP3, OPN, G-CSF and CXCL12 (R&D Systems).
Mouse cell line and primary cell culture
Cell culture was carried out in 5% CO2 in a humidified incubator at 37 °C. MSCs were isolated from mouse BM by adherence to tissue culture plastic and expanded in minimum essential media (MEM) containing 20% FBS plus 1% Pen/Strep as described previously83. MSC marker expression (CD45− Ter119− CD51+ CD31−) was confirmed by flow cytometry. A BM endothelial cell line (BMEC, obtained from J. Butler) was cultured in low-glucose DMEM and Ham’s F-12 (1:1 ratio), supplemented with 20% heat-inactivated FBS, 1% Pen/Strep, 1% nonessential amino acids (Gibco), 10 mM Hepes (Gibco), 100 µg ml−1 heparin (Sigma-Aldrich) and 50 µg ml−1 endothelial growth factor (Sigma-Aldrich), as previously described32. HSPCs were isolated using CD117 MicroBeads, mouse (Miltenyi). MSCs or BMECs were seeded at a density of 75,000 cells in a 24-well plate in normal growth medium. Once 70% confluent, 0.1 × 106 HSPCs were then co-cultured with the MSCs or BMECs for seven days. Following this co-culture, haematopoietic cells were washed from MSCs and BMECs prior to analysis, and cell surface marker staining of CD45 was incorporated to exclude any residual haematopoietic cells. HSPCs were removed and analysed by flow cytometry. MSCs were analysed by flow cytometry, or RNA was extracted for qPCR. For transwell assays, MSCs were seeded at a density of 75,000 cells in a 24-well plate in normal growth medium. Once 70% confluent, 0.1 × 106 HSPCs were seeded in transwells on top of the MSCs for seven days. MSCs were then analysed by flow cytometry, or RNA was extracted for qPCR. For conditioned media cultures, HSPCs were cultured in StemSpan SFEM II (StemCell Technologies) and 10 ng ml−1 recombinant murine SCF (StemCell Technologies) for seven days. Cells were centrifuged at 300g for 5 min at 4 °C, then the supernatant was collected and spun again at 1,300g for 10 min at 4 °C to isolate conditioned medium (HSPC CM). HSPC CM was placed on 75,000 MSCs in a 24-well plate for seven days and analysed by flow cytometry and qPCR. For cell tracer experiments, cells were seeded at a density of 50,000 cells in a 24-well plate in normal growth medium. After 24 h, cells were stained with CellTrace Yellow, the stain was washed off, and HSPC CM was added for seven days. MSCs were analysed by flow cytometry. For cytokine treatment, MSCs were seeded at a density of 75,000 cells in a 24-well plate in normal growth medium. MSCs were then treated with 1 ng ml−1 or 5 ng ml−1 recombinant murine TNF-α (PeproTech), 50 ng ml−1 recombinant murine IL-6 (StemCell Technologies), 3 ng ml−1 recombinant murine IL-1α (PeproTech) or 3 ng ml−1 recombinant murine IL-1β (PeproTech) for seven days and analysed by flow cytometry. For inhibitor assays, MSCs were cultured with HSPC CM for five days with 10 µg ml−1 elsilimomab (MedChemExpress), 1 µg ml−1 etanercept (Millipore Sigma) or 5 µM Stattic (MedChemExpress). MSCs were then analysed by flow cytometry.
Senescence β-galactosidase assay
A senescence β-galactosidase staining kit (Cell Signaling Technology) was used to detect senescence-associated β-galactosidase (SA-β-gal) in MSCs. MSCs were seeded at a density of 75,000 cells in a 24-well plate. HSPC CM was placed on for five days and stained using the SA-β-gal staining kit following the manufacturer’s instructions. Imaging data were collected using a Leica Dmi1 microscope at ×20 magnification. Five randomly selected non-overlapping optical fields per well were analysed for SA-β-gal + MSC area using ImageJ, following an established protocol84.
Irradiation for senescence induction
p16-3MR mice received a single dose of sublethal irradiation (5 Gy). Control non-irradiated p16-3MR mice were placed in the irradiator for the same period without irradiation. Mice were housed for eight weeks post-irradiation before BM harvest and analysis by flow cytometry.
Targeting senescent cells in vivo
In the p16-3MR model, 25 mg kg−1 d−1 ganciclovir (MedChemExpress; in PBS) was administered by i.p. injection for five consecutive days to deplete p16-expressing cells. ABT-263 (Selleck; in ethanol:polyethylene glycol 400:Phosal 50 PG at a ratio of 10:30:60) was administered to mice by oral gavage at 50 mg kg−1 d−1 for seven days followed by 14 days off, repeated twice, then 5 mg kg−1 d−1 dasatinib (Selleck) and 50 mg kg−1 d−1 quercetin (Selleck) (D + Q) (in 10% PEG 400/saline) or vehicle control was administered to mice by oral gavage for three days followed by four days off, repeated for four weeks.
CFU assay
BM cells or HSPCs (ckit+) were plated in MethoCult GF M3434 medium (StemCell Technologies) and cultured at 37 °C and 5% CO2. Colonies were scored 7–10 days post plating using a Nikon Eclipse TS100 inverted microscope. For CFU replating assay, colonies were harvested, and 1.5 × 104 cells were replated in MethoCult GF M3434.
Statistics and reproducibility
No statistical method was used to predetermine sample size. No data were excluded from the analyses. The experiments were not randomized except where noted above. The investigators were not blinded to allocation during the experiments and outcome assessment except where noted above. For ex vivo imaging experiments, investigators were blinded to mouse genotypes prior to and during data analysis. For mouse experiments, animals were randomly assigned to experimental groups. All statistical tests were performed using Prism 9 software (GraphPad) as described in the figure legends. Data distribution was assumed to be normal, but this was not formally tested. Differences among group means were considered significant when the probability value was <0.05. Sample size (n) represents number of biological replicates.
Reporting Summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this Article.
