Oxylipin-PPARγ-initiated Adipocyte Senescence Propagates Secondary Senescence in the Bone Marrow

SUMMARY

Chronic use of glucocorticoids decrease bone mass and quality and increase bone marrow adiposity, but the underlying mechanisms remain unclear. Here, we show that bone marrow adipocyte (BMAd) lineage cells in adult mice undergo rapid cellular senescence upon glucocorticoid treatment. The senescent BMAds acquire a senescence-associated secretory phenotype, which spreads senescence in bone/bone marrow. Mechanistically, glucocorticoids increase the synthesis of oxylipins, such as 15d-PGJ2, for PPARγ activation. PPARγ stimulates the expression of key senescence genes and also promotes oxylipin synthesis in BMAds, forming a positive feedback loop. Transplanting senescent BMAds into bone marrow of healthy mice is sufficient to induce the secondary spread of senescent cells and bone loss phenotype, whereas transplanting BMAds harboring a p16INK4a deletion did not show such effects. Thus, glucocorticoid treatment induces a lipid metabolic circuit that robustly triggers the senescence of BMAd lineage cells that, in turn, act as the mediators of glucocorticoid-induced bone deterioration.

eTOC

Liu et al. reveal that glucocorticoids directly induce cellular senescence of bone marrow adipocytes (BMAds) through a prostaglandin-PPARγ-INK positive feedback loop. The senescent BMAds trigger a secondary senescence in the surrounding bone cells by producing senescence-associated secretory phenotype (SASP) factors, leading to bone deterioration.

Graphical Abstract

INTRODUCTION

Bone marrow adipose tissue mass increases under diverse physiological and clinical conditions, including aging, obesity, type 2 diabetes, anorexia, hyperlipidemia and osteoporosis, as well as in response to radiotherapy and treatment with thiazolidinediones and glucocorticoids¹. Many of these conditions are associated with decreased bone mass and increased fracture risk. It is now accepted that the increase in the number of bone marrow adipocytes (BMAds) during the above contexts regulate the activities of other cell populations within and outside the bone by secreting cytokines, chemokines, lipid species, and adipokines^2–5. However, these studies have been conducted primarily under culture conditions in vitro⁴. It remains unknown whether BMAds have a negative impact on bone remodeling via secretion of these factors in vivo under pathological conditions. The mechanisms by which BMAds acquire the secretory phenotype are also unclear.

Chronic glucocorticoid treatment is a common cause of secondary osteoporosis, leading to increased fracture risk^6–8. Glucocorticoids exert direct effects on many bone/bone marrow cells that express the nuclear glucocorticoid receptor, including osteoclasts^8–14, mesenchymal stem/progenitor cells^15,16, osteoblasts¹⁷, osteocytes^18–20, and adipocytes^21,22. Pharmacological doses of glucocorticoids impair osteogenesis by promoting osteoblast and osteocyte apoptosis⁹ and by inhibiting osteoblast differentiation¹⁷. Glucocorticoid treatment also increases osteoclast bone resorption by upregulating receptor activator of nuclear factor kappa-Β ligand (RANKL) expression and promoting the survival of osteoclasts^23–25. Glucocorticoids inhibit angiogenesis, and thus osteogenesis, by decreasing endothelial growth factors, such as vascular endothelial growth factor (VEGF), platelet-derived growth factor BB (PDGF-BB), and angiogenin (ANG)^26–28. Some of the adverse effects of glucocorticoids on bone may be mediated through a change in systemic metabolism because glucocorticoids have a major impact on glucose and lipid metabolism in the body^29,30. In addition, increased concentrations of glucocorticoids trigger adipogenesis both in vitro and in vivo by upregulating the expression of Pparg2 or Cebpa³¹. In the long bones of glucocorticoid-treated animals, bone marrow adiposity accrual is concomitant with (or even precedes) a decrease in bone mineral density, trabecular bone, or osteoblast number^32,33. The fate change of BMAds and their role in glucocorticoid-induced osteoporosis remain poorly understood.

Cellular senescence refers to a stable cell-cycle arrest process, which is mediated mainly by the activation of p53-p21- or p16-Rb-regulated pathways^34,35. In addition, senescent cells are metabolically active and develop other characteristics, including enlarged and flattened morphology, increased senescence-associated β-galactosidase (SA-βGal), loss of high mobility group box 1 (HMGB1), downregulation of Lamin B1, formation of senescence-associated distension of satellites (SADS), and most importantly development of the senescence-associated secretory phenotype (SASP)^34–42. The SASP factors include cytokines, chemokines, growth factors, proteases, and extracellular components that can affect biological signaling in an autocrine or paracrine manner^43,44. Recently, non-protein components in the SASP have also been found. For example, senescent cells secrete oxylipins, especially prostaglandin-related lipids, and the prostaglandins reinforce the senescence phenotype and SASP secretion⁴⁵. SASP factors have been shown to spread senescence to neighboring non-senescent cells through a paracrine manner or cell-cell contact⁴⁶. We recently identified that growing young mice accumulate vascular endothelial cell senescence in the metaphysis of long bones in response to glucocorticoid treatment²⁷. It remains unclear whether glucocorticoid treatment also induces cellular senescence in adult mice. Here, we show that, unlike bone vascular cell senescence in primary spongiosa of growing young mice, glucocorticoid treatment induces rapid cellular senescence in BMAd lineage cells. The senescence program is initiated in lineage committed adipocyte precursors and progresses into full cellular senescence in mature BMAds, which secrete a panel of SASP factors to drive secondary senescence within the bone environment. We also show, using in vitro and in vivo approaches, that PPARγ and oxylipins (e.g., prostaglandin D2 derivatives) form a vicious positive feedback loop to trigger and maintain glucocorticoid-induced BMAd senescence.

RESULTS

Glucocorticoid treatment induces rapid BMAd senescence in mice

We previously demonstrated that glucocorticoid treatment in growing young mice induces vascular endothelial cell senescence in the metaphysis of long bone²⁷. To examine whether glucocorticoid treatment also induces cellular senescence in the adult skeleton, we treated 3-month-old C57BL/6 mice with methylprednisolone (MPS), a commonly used synthetic glucocorticoid, for different durations. MPS treatment induced a progressive increase in bone marrow adipose tissue mass, as evidenced by the higher percentage of bone marrow adipose tissue volume per total volume (% AV/TV) in decalcified distal femurs stained with osmium tetroxide (OsO₄) in MPS-treated mice compared to vehicle-treated mice (Figure S1A–S1B). MPS treatment for 2 weeks induced an increase in the number of SA-βGal⁺ cells in both the distal femur and femoral head (Figure S1C–S1D). However, MPS treatment for 4 weeks resulted in many more SA-βGal⁺ cells in both bone regions compared with 2-week MPS treatment or vehicle treatment (Figure S1C–S1D). Moreover, the number of cells positive for HMGB1 (loss of nuclear expression of which is another feature of senescent cells) was moderately lower in bone tissue of 2-week MPS-treated mice relative to vehicle-treated mice (Figure S1E–S1F). However, the lower incidence of HMGB1⁺ cells in both the distal femur and femoral head was much greater in 4-week MPS-treated mice compared with vehicle-treated mice (Figure S1E–S1F). We also used a mouse senescence reporter strain harboring p16^tdTom, in which p16INK4a-activated cells (tdTom⁺) can be detected⁴⁷. In situ fluorescence analysis of the femoral bone tissue sections showed that MPS treatment induced a greater number of tdTom⁺ cells in both the distal femur and femoral head in a time-dependent manner (Figure 1A–1B). Consistent with these results, we also detected a progressive accumulation of tdTom⁺ senescent cells in the femoral bone in response to MPS treatment by flow cytometry analysis (Figures 1C–1E). Therefore, glucocorticoid treatment induces a gradual increase in senescent cells in the adult long bone.

Figure 1. Glucocorticoid treatment induces rapid senescence of BMAds in mice.

Three-month-old p16^tdTom mice were treated with methylprednisolone (MPS) at 10 mg/m²/day or vehicle by daily intraperitoneal injection for different time periods as indicated.

(A-B) Representative images of tdTom⁺ cells from proximal and distal femur in (A) and analysis of cell number per mm² tissue area (N. tdTom⁺ cells) in (B).

(C) Schematic diagram illustrating the experimental procedure. Femoral bone and bone marrow tissue was collected from mice and digested, and the isolated cells were subjected to flow cytometry analysis.

(D-E) Representative flow cytometry images of tdTom-expressing cells of the femoral bone in (D) and the percentages of tdTom⁺ cells out of total bone/bone marrow cells in (E).

(F-H) Immunofluorescence staining of perilipin in femoral bone sections. Representative images are shown in (F) Quantification of the percentage perilipin⁺ cells in total tdTom positive cell population (% perilipin⁺ cells/total tdTom⁺ cells) and the percentage of tdTom⁺ cells in total perilipin positive cell population (% tdTom⁺ cells/total perilipin⁺ cells) are shown in (G) and (H), respectively.

(I-J) Three-month-old C57BL/6 mice were treated with MPS or vehicle for 2 weeks. (I) Fluorescent in situ hybridization followed by immunofluorescence staining was performed to identify SADS-positive senescent cells (in red) and perilipin⁺ adipocytes (in green), respectively. (J) Double-immunofluorescence staining of γH2AX and perilipin in femoral bone sections.

(K-M) Three-month-old C57BL/6J mice were treated with MPS at 10 mg/m²/day or vehicle for different time periods as indicated. Double-immunofluorescence staining of γH2AX and perilipin in femoral bone sections. Representative images are shown in (K) Quantification of the number of perilipin-positive cells per mm² tissue area (N. Perilipin⁺ cells) and the number of γH2AX-positive cells per mm² tissue area (N. γH2AX⁺ cells) are shown in (L) and (M), respectively.

n=5 mice. Data are represented as mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001 as determined by two-tailed Student’s t-tests for 2 groups and one-way ANOVA for 3 or more groups.

We noticed that most of the tdTom⁺ cells at 2 weeks after MPS treatment were large, ring-like cells that were like adipocytes in appearance. Indeed, immunofluorescence staining of the bone tissue sections with the adipocyte marker perilipin showed that approximately 85.87±6.63% of the total tdTom⁺ cells were perilipin⁺ adipocytes after 2 weeks of MPS treatment (Figure 1F–1G), indicating that most bone marrow senescent cells were adipocytes during the early stage of glucocorticoid treatment. The percentage of adipocytes out of total senescent cells decreased to approximately 27.31±11.92% after 4 weeks of MPS treatment (Figure 1F–1G). Therefore, many other cell types, in addition to adipocytes, became senescent at a later stage of glucocorticoid treatment. Of note, the percentage of perilipin⁺ cells that became senescent (tdTom⁺) was approximately 23.87±8.29 and 24.53±6.66% after 2 and 4 weeks of MPS treatment, respectively (Figure 1H), indicating that only a portion of BMAds undergo cellular senescence in response to MPS treatment. We also detected the senescence-associated distension of satellites (SADS) and the DNA damage marker γH2AX, robust markers of senescent cells^48–51, and found that SADS⁺ and γH2AX⁺ signals were localized primarily in perilipin⁺ BMAds of MPS-treated mice (Figure 1I–1J). We also examined whether shorter periods of glucocorticoid treatment could induce cellular senescence. The number of perilipin⁺ BMAds started rising at 1 week after MPS treatment, whereas γH2AX⁺ senescent cells were higher only at 2 weeks but not at the earlier time points after MPS treatment relative to the vehicle treatment group (Figure 1K–1M). Therefore, senescence was not induced before the increase of bone marrow adiposity, further confirming that most of the senescent cells are BMAds. Similarly, co-staining of bone tissue sections with SA-βGal and Oil Red O, a marker of mature adipocytes, showed markedly greater SA-βGal/Oil Red O–double positive cells in MPS-treated mice relative to vehicle-treated mice (Figure S1G–S1H). Therefore, adipocytes are the first cell type to show a senescent phenotype in the bone/bone marrow microenvironment following glucocorticoid treatment.

Senescent vascular cells and osteoblasts accumulate in mouse bone/bone marrow in a late phase of glucocorticoid treatment

Because we observed many more senescent cells after 4 weeks (vs. 2 weeks) of MPS treatment, we attempted to identify the major cell type(s) that undergo cellular senescence during this later phase of MPS treatment. We previously found that bone vascular endothelial cells are a vulnerable cell type to glucocorticoid-induced cellular senescence²⁷. We examined whether 4-week MPS treatment also induces the accumulation of senescent vascular endothelial cells in adult mice. As expected, we found a much greater number of endomucin (Emcn)⁺ vascular endothelial cells that express tdTom in the distal femur of p16^tdTom mice after 4 weeks of MPS treatment (Figures 2A–2B and S2H). We noticed that some tdTom⁺ cells were osteoblast-like, which were present on the bone surface in a closely packed layer. Indeed, the number of tdTom-expressing osteocalcin (OCN)⁺ osteoblasts were markedly greater in the distal femur of mice with 4-week MPS treatment versus vehicle treatment (Figure 2C–2D). Notably, almost none of the tdTom-expressing cells were Emcn⁺ vascular endothelial cells or OCN⁺ osteoblasts after 2 weeks of MPS treatment (Figure 2A–2D). These results suggest that the accumulation of senescent vascular cells and osteoblasts is a later response to MPS treatment relative to the senescence of BMAds. Further, flow cytometric analysis of the femoral bone/bone marrow cells of p16^tdTom mice also showed a dramatically higher incidence of tdTom⁺ cells in the CD144⁺ cell population, which includes vascular endothelial cells and endothelial progenitor cells, in the 4-week MPS-treated group vs. the vehicle-treated group (Figure 2E–2F). Consistent with these findings, the total number of SA-βGal⁺ cells and the percentage of SA-βGal–expressing endothelial cells were also much higher in the 4-week MPS-treated mice vs. vehicle-treated mice (Figure S2A–S2C). We sorted the CD144⁺ bone vascular endothelial cells using fluorescence-activated cell sorting (FACS) followed by quantitative real-time polymerase chain reaction (qRT-PCR) analysis. CD144⁺ cells from MPS-treated mice (vs. vehicle-treated mice) had reduced expression of the cell proliferative marker Mki67 (Figure S2D) and higher expressions of Cdkn2a (Figure S2E) and Tp53 (Figure S2F), but not Cdkn1a (Figure S2G). Therefore, bone vascular cells and osteoblasts may undergo secondary senescence after the primary BMAd senescence induced by glucocorticoid treatment. To further validate this assumption, we tested whether senescent cells could still be detected in the bone tissue of mice after stopping glucocorticoid treatment. The number of perilipin⁺ BMAds was greater in 2-week MPS-treated mice (vs. vehicle group) when the mice were killed immediately after treatment. However, the number of perilipin⁺ BMAds were similar to the level of the vehicle-treated control group when the mice were killed a week after the termination of treatment (Figure 2G–2H), indicating a regression of BMAd accumulation. Oil Red O staining of the bone tissue sections showed a consistent result (Figure 2K and 2L). Importantly, the numbers of γH2AX⁺ (Figure 2I) and SA-βGal⁺ (Figure 2M) senescent cells remained higher in this group compared with the vehicle group. Most of the senescent cells were non-BMAds because the senescence markers γH2AX (Figure 2G and 2J) and SA-βGal (Figure 2K and 2N) were not overlapping with BMAd markers. Therefore, although both the expansion of bone marrow adipose tissue and the senescent BMAds regressed after termination of MPS treatment, senescence likely spread to other cell types through BMAd-produced SASP factors.

Figure 2. Senescent vascular cells and osteoblasts accumulate in murine bone/bone marrow in a late phase of glucocorticoid treatment.

Three-month-old p16^tdTom mice were treated with MPS at 10 mg/m²/day or vehicle for different time periods as indicated.

(A-D) Immunofluorescence staining of Endomucin (Emcn) (A) or Osteocalcin (Ocn) (C) in femoral bone sections. Yellow arrow heads, cells are tdTom- and Emcn-double positive in (A) or tdTom- and Ocn-double positive in (C). BM, bone marrow; BS, bone surface. Quantified numbers of Emcn- and tdTom-double positive cells per mm² tissue area (N. Emcn⁺tdTom⁺ cells) and Ocn- and tdTom-double positive cells per mm² tissue area (N. Ocn⁺tdTom⁺ cells) are shown in (B) and (D), respectively.

(E-F) Bone/bone marrow cells were isolated from femoral bone for flow cytometry analysis. Representative images are shown in (E). Percentages of tdTom-expressing cells in total CD144⁺ vascular cell population are shown in (F).

Three-month-old C57BL/6J mice were treated with MPS at 10 mg/m²/day or vehicle for different time periods as indicated.

(G-J) Double-immunofluorescence staining of γH2AX and perilipin in femoral bone sections. Representative images are shown in (G). Quantification of the number of perilipin-positive cells per mm² tissue area (N. Perilipin⁺ cells), the number of γH2AX-positive cells per mm² tissue area (N. γH2AX⁺ cells), and the percentage of γH2AX⁺ cells in total perilipin-positive cell population (% γH2AX⁺ cells/total BMAds) in (H), (I), and (J), respectively.

(K-N) Representative images of SA-βGal and Oil red staining of femoral bone sections are shown in (K). Quantification of the numbers of Oil Red⁺ cells per mm² tissue area, SA-βGal⁺ cells per mm² tissue area, and the percentage of SA-βGal⁺ cells in total Oil Red⁺ BMAds are shown in (L), (M), (N), respectively.

n=5 mice, Data are represented as mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001 as determined by two-tailed Student’s t-tests for 2 groups and one-way ANOVA for 3 groups.

Glucocorticoids directly activate a cellular senescence program in BMAd lineage cells

We then tested whether glucocorticoid treatment directly induces cellular senescence of BMAd lineage cells. Bone marrow mesenchymal stromal cells (BMSCs) were isolated from mice and cultured in various conditions where the cells underwent different stages of adipogenic differentiation. We first evaluated if glucocorticoid treatment induces senescence of BMSCs and early-stage adipocyte precursors during adipogenic differentiation. To do this, we cultured BMSCs in a growth medium, where BMSCs maintain their undifferentiated state, and dexamethasone (DEX) (a synthetic glucocorticoid) was added to the medium for different time periods (Figure S3A). mRNA expressions of key adipogenesis marker genes, including Pparg, Cebpa, and Adipoq, were elevated rapidly (1 or 3 days) after DEX treatment relative to vehicle-treatment controls (Figure S3C). However, perilipin⁺ mature adipocytes were almost undetectable at all time points after DEX treatment (Figure S3B). Thus, although the cells were committed/differentiated to the adipocyte lineage, most of them were premature adipocyte precursors even at 7 days after DEX treatment. We then tested if the expression of cellular senescence markers increased in the cells. Increased Cdkn2a (p16INK4A encoding gene) expression (Figure S3D) and SA-βGal⁺ cells (Figure S3E) were detected only at 7 days, but not at 1 and 3 days, after DEX treatment vs. the vehicle control group. Moreover, cells that were positive for γH2AX and that lost lamin-B1 expression, hallmark changes for cells that have progressed into full senescence^52,53, were not detected in all three treatment groups (Figure S3F and S3G). The results suggest that the cellular senescence program is initiated in the premature adipocyte precursors, but the full senescence phenotype only developed in mature BMAds. Our data also suggests that the senescence program was not induced in uncommitted BMSCs in response to glucocorticoid treatment. Supporting the conclusions, flow cytometry analysis showed that the percentage of p16-tdTom⁺ cells was elevated in bone marrow CD45⁻CD31⁻CD24⁻Sca1⁺ adipocyte precursors (Figure S3H–S3I) but not in the CD45⁻CD31⁻CD24⁺Sca1⁺ BMSC population^54,55 in p16^tdTom mice after 2 weeks MPS treatment (Figure S3J–S3K).

We then examined cellular senescence in the later stage of adipogenesis when most of the cells became mature adipocytes. BMSCs isolated from p16-3MR mice, in which the senescent cells can be eliminated by treatment with ganciclovir (GCV)⁵⁶, were first incubated with adipogenesis medium to induce their differentiation into pre-adipocytes and adipocytes. The cells were then treated with DEX with or without GCV (Figure 3A). Most of the cells became mature adipocytes as approximately 56.43±3.52% of the cells were perilipin⁺ (Figure 3B). DEX-treated cells had dramatically increased Cdkn2a expression (Figure 3C) and SA-βGal expression (Figure 3D) relative to vehicle-treated cells. Importantly, in contrast to premature adipocyte precursors (Figure S3F and S3G), adipocytes at the later stage of adipogenesis exhibited full senescence phenotype, including γH2AX expression and loss of lamin-B1 (Figure 3E and 3F). Of note, DEX-induced SA-βGal⁺ cell accumulation, γH2AX expression, and loss of lamin-B1 were all abolished by co-treatment with GCV (Figure 3D–3F), suggesting that they are senescence-dependent features. We also isolated BMAds directly from p16-3MR mice with an established centrifuge-based method and subsequently treated the adipocytes with DEX with or without GCV (Figure 3G)⁵⁶. Increased p16-RFP⁺ cells, which represent senescent cells in p16-3MR mice, were detected only in the DEX treatment group and not in the vehicle group or DEX and GCV co-treatment group (Figure 3H), confirming the glucocorticoid-induced senescence of BMAds. We also tested whether glucocorticoids can directly induce osteoblast and endothelial cell senescence in vitro by treating calvarium-originated osteoblasts or human umbilical vein endothelial cells (HUVECs) with DEX. Surprisingly, DEX failed to increase SA-βgal expression of these two cell types (Figure S3L–S3M). These results suggest that the bone marrow adipocyte lineage, but not osteoblasts and vascular endothelial cells, are the direct target of glucocorticoid-induced cellular senescence.

Figure 3. Glucocorticoids directly activate a cellular senescence program in cultured BMAd lineage cells.

(A-F) Bone marrow mesenchymal stromal cells (BMSCs) were isolated from 3-month-old p16-3MR mice and differentiated into bone marrow adipocytes/preadipocytes with adipogenesis medium for 10 days (A). The differentiated adipocytes/preadipocytes were then stimulated with vehicle, dexamethasone (DEX) or DEX + Ganciclovir (GCV) for 3 days. Cells were fixed and subjected to perilipin immunostaining (B), Cdkn2a (p16 encoding gene) mRNA measurement by qRT-PCR (C), SA-βGal staining (D), γH2AX immunostaining (E), and Lamin-B1 immunostaining (F), respectively.

(G-H) BMAds were isolated from 3-month-old p16-3MR mice as shown in (G) and were treated with vehicle, DEX or DEX + GCV for 3 days. The cells were then subjected to RFP staining with representative pictures in (H).

(I-M) BMSCs were isolated from C57BL/6 mice and differentiated into bone marrow adipocytes/preadipocytes with adipogenesis medium for 10 days (I). The differentiated adipocytes/preadipocytes were stimulated with vehicle or DEX and subjected for bulk RNA-Seq. Principal-component analysis (PCA) of the DEG expression profile of sequenced adipocyte samples was shown in (J). Venn map analyzing the number of DEGs overlap with previously identified aging and senescence ASIGs genes (K). Top enriched pathways of ASIGs genes significantly upregulated or downregulated in DEX-treated cells relative to vehicle-treated cells based on RNA-seq data (L). Heatmap showing increased expression of INK family genes in DEX-treated cells relative to vehicle-treated cells in (M).

In vitro experiments were repeated 3 times. Data are represented as mean ±s.e.m. ***p<0.001 as determined by two-tailed Student’s t-tests.

To further confirm that glucocorticoids directly induce the senescent phenotype of BMAd lineage cells and to determine the underlying molecular mechanisms, we conducted RNA-seq analysis of the in vitro differentiated adipocytes and pre-adipocytes (Figure 3I). Principal component analyses (PCA) revealed that mRNA expression profiles can effectively differentiate DEX-treated cells from the vehicle-treated controls (Figure 3J). After the screening of differentially expressed mRNAs by the filter criteria (p < 0.05), 6725 differentially expressed genes (DEGs) were identified in the DEX- vs. vehicle-treated cells (Table S1). Comparison of our data with previously defined aging/senescence-induced genes (ASIGs)⁵⁷ identified a total of 390 ASIGs in the DEX- vs. vehicle-treated cells (Figure 3K and Table S1). WikiPathway analysis showed that significantly differentially expressed ASIGs genes have influenced “senescence and autophagy in cancer,” “DNA damage repair,” “cell cycle,” and “TGF-β signaling pathway”, which were notable for their known links to cellular senescence (Figure 3L). The top significantly upregulated ASIGs in DEGs are listed in Figure S4. Of note, the key senescence genes Cdkn2a, Cdkn2c, and Cdkn2d, which are p16INK4a, p18INK4c, and p19INK4d encoding genes, respectively, are among the significantly upregulated genes (Figure 3M), supporting the notion that INK family proteins are major mediators of glucocorticoid induced BMAd senescence.

Glucocorticoids induce senescence of BMAd lineage cells by activating a prostaglandins-PPARγ-INK positive feedback loop

BMAds are a source of lipids, and they have an active lipid metabolism that regulates bone remodeling and, potentially, whole-body metabolism^5,58,59. A recent study revealed an increase in the syntheses of prostaglandin D2–related oxylipins, such as dihomo-15d-PGJ2, as a key mechanism promoting cellular senescence and the corresponding SASP in fibroblasts⁴⁵. Because glucocorticoids modulate lipid metabolism in adipocytes^22,31,60, we explored whether glucocorticoid treatment is also associated with prostaglandin generation. By analysis of the RNA-seq data, we found that Ptgds and Ptgis, which encode for prostaglandin D2 synthase and prostaglandin I2 synthase, respectively, were significantly upregulated in isolated BMAds (Figure 4A). The other two synthases in the family, Ptges and Tbxas1, which encode for prostaglandin E2 synthase and Thromboxane A synthase, respectively, were downregulated.

Figure 4. Glucocorticoids induce senescence of BMAd lineage cells by activating prostaglandin pathway.

(A) Heatmap of RNA-seq data showing expression changes encoding oxylipin synthesis genes in DEX-treated relative to vehicle-treated adipocytes/preadipocytes.

(B) Schematic diagram showing the procedure of the conditioned medium (CM) preparation from cultured BMAds.

(C) The concentrations of prostaglandins (PGJ2, PGD2, PGE2) in CM were measured by ELISA.

(D) BMAds were isolated and treated with Vehicle, DEX, or 15d-PGJ2 for 3 days. mRNA levels of Cdkn2a, Cdkn1a, and Cdkn1b in the cells were measured by qRT-PCR analysis.

(E) Adipocytes/preadipocytes differentiated from BMSCs were treated with Vehicle, DEX, or 15d-PGJ2 for 3 days. SA-βGal staining was performed.

(F) Adipocytes/preadipocytes differentiated from BMSCs were treated with Vehicle, DEX, or DEX plus celecoxib for 3 days. mRNA levels of Cdkn2a, Cdkn1a, and Cdkn1b in the cells were measured by qRT-PCR analysis.

In vitro experiments were repeated 3 times. Data are represented as mean ±s.e.m. **p<0.01, ***p<0.001 as determined by two-tailed Student’s t-tests for 2 groups and one-way ANOVA for 3 groups.

These results suggest that the production of PGD2 and PGI2 in BMAds may increase in response to glucocorticoid treatment. Thus, we used ELISAs to measure the concentrations of prostaglandins, including PGJ2 (a sum of 15d-PGJ2 and dihomo-15d-PGJ2), PGD2 (a precursor of 15d-PGJ2), and PGE2 in the conditioned medium in different treatment groups (Figure 4B). As we expected, DEX-treated cells showed elevated PGJ2 and PGD2 secretion but unchanged PGE2 secretion (Figure 4C). It has been reported that 15d-PGJ2 induces fibroblast senescence⁴⁵. Consistent with this finding, we found that 15d-PGJ2 exerted similar effects as DEX on adipocytes/pre-adipocytes in inducing upregulation of senescence-related genes, including Cdkn2a and Cdkn1b (p16INK4a and p27KIP1 encoding genes) (Figure 4D)⁴⁵. Cdkn1a (p21CIP1 encoding gene) level was not changed by DEX or 15d-PGJ2 treatment, which is consistent with the findings from Figure S3G.

We also examined whether a reduction in prostaglandin production occurred in response to inhibition of cyclooxygenase 2 (COX2), a key enzyme involved in prostaglandin production, and thus whether such inhibition could antagonize adipocyte/pre-adipocyte senescence^61–63. Whereas the adipocytes/pre-adipocytes treated with DEX alone had increased SA-βGal expression, the number of SA-βGal⁺ cells diminished in cells co-treated with DEX and celecoxib, a selective COX2 inhibitor (Figure 4E). We then measured the levels of key senescence-inducing genes, including Cdkn2a and Cdkn1b, and Cdkn1a, in the cells of different treatment groups^34,41. The expressions of Cdkn2a and Cdkn1b, but not Cdkn1a, were elevated in the DEX treatment group and decreased to basal levels in the DEX and celecoxib co-treatment group (Figure 4F). These results suggest that glucocorticoids increase the production of oxylipins, which, in turn, induce the senescence of BMAds. The results from Figure S2G, 4D, and 4F also consistently suggest that p21CIP1 is not involved in BMAd senescence induced by glucocorticoids.

As 15d-PGJ2 is a natural agonist of PPARγ^64–66, we postulated that PPARγ may be involved in this process. DEX and the PPARγ agonists 15d-PGJ2 and rosiglitazone (Rosi) induced Pparg gene expression in adipocytes/pre-adipocytes (Figure 5A). Importantly, co-treatment with celecoxib abolished this effect of DEX (Figure 5B), indicating that increased secretion of prostaglandin(s) mediates DEX-induced upregulation of Pparg expression.

Figure 5. An oxylipins-PPARγ positive feedback loop mediates glucocorticoid-induced BMAd senescence.

(A-E) Adipocytes/preadipocytes differentiated from murine BMSCs were stimulated with different treatments as indicated. mRNA levels of the indicated genes were measured by qRT-PCR analysis.

(F-G) Adipocytes/preadipocytes differentiated from murine bone marrow MSCs were treated with different reagents as indicated and subjected to SA-βGal staining.

(H) Heatmap of RNA-seq data showing the changes in INK family genes in cells with the indicated treatments.

(I-J) qRT- PCR analysis of Ptgds and Ptgis levels in BMAds with the indicated treatments. Experiments were repeated 3 times. Data are represented as mean ±s.e.m. *p< 0.05, **p<0.01, ***p<0.001, as determined by one-way ANOVA.

We then tested whether activated PPARγ leads to cellular senescence. We found that the PPARγ synthetic agonist Rosi induced dramatic upregulation of senescence marker genes, including Cdkn2a and Cdkn1b, but not Cdkn1a (Figure 5C)^34,41. Moreover, DEX- and 15d-PGJ2-induced upregulation of senescence marker genes were largely abrogated by co-treatment of the cells with either of the PPARγ antagonists T0070907 and GW9662 (Figure 5D and 5E). Consistent with the changes in senescence gene expression, SA-βGal expression was also greatly stimulated in the adipocytes/pre-adipocytes treated with PPARγ agonists 15d-PGJ2 and thiazolidinedione (TZD) compared with vehicle-treated cells (Figure 5F). In contrast, the effects of DEX and 15d-PGJ2 on inducing SA-βGal expression were almost abolished by the PPARγ antagonists T0070907 and GW9662 (Figure 5G). Likewise, RNA-seq data showed that the significantly upregulated INK family genes induced by DEX were effectively reversed by co-treatment with the PPARγ antagonist GW9662 (Figure 5H). Intriguingly, DEX-induced upregulation of Ptgds and Ptgis expression were both blunted by treatment of the cells with the PPARγ antagonists GW9662 (Figure 5I and 5J). Thus, PPARγ activation, in turn, promotes PGD2 and PGI2 synthesis in BMAds. Together, these results suggest that the reciprocal feedback loop of oxylipins and PPARγ signaling is crucial in initiating and maintaining the senescence program of BMAds.

Senescent BMAds acquire a SASP that spreads senescence to bone marrow vascular cells and osteoblasts

We next investigated whether senescent BMAds acquire a SASP. We conducted analysis of our RNA-seq data again and found a typical SASP expression profile in the DEX-treated vs. vehicle-treated adipocytes/preadipocytes, as evidenced by upregulation of multiple previously identified SASP factors⁴³, including inflammatory cytokines, chemokines, growth factors, secreted proteases and secreted ECM components (Figure 6A). Intriguingly, the effects of DEX on inducing the upregulation of these factors were fully antagonized by the PPARγ antagonist GW9662, indicating that PPARγ activation is a key contributor not only to senescence cell cycle arrest but also to SASP development of the BMAds.

Figure 6. Senescent BMAds acquire a SASP that spreads senescence to bone marrow vascular cells and osteoblasts.

(A) Heatmap of RNA-seq data showing SASP gene expression in adipocytes/preadipocytes treated with vehicle, DEX or DEX+GW9662.

(B-C) BMAds were isolated from 3-month-old p16-3MR mice, cultured in alpha-MEM for 1 day, and stimulated with vehicle, DEX, or DEX+GCV for 3 days. The secretion of cytokines in the CMs were measured (B). Heatmap of differentially expressed cytokines in CM was shown in (C).

(D) A schematic diagram illustrating the procedure of the CM-based senescence assays.

(E-F) Calvarium osteoblasts (OB) or HUVECs (EC) were cultured with CM prepared from (D). SA-βGal staining of the cells in (E). Western blot analysis of p16 protein expression in (F).

(G) Schematic diagram showing the experimental design for transplantation.

(H-L) Representative micro-CT images in (H) and quantitative analysis of trabecular bone volume (BV/TV) (I), trabecular thickness (Tb. Th) (J), trabecular number (Tb. N) (K), and trabecular separation (Tb. Sp) (L).

n=5 mice. Data are represented as mean ±s.e.m. ***p<0.001, as determined by two-tailed Student’s t-tests for 2 groups.

To substantiate the data from the transcriptomic analysis, we performed proteomics assays. Isolated BMAds from p16-3MR mice were treated with DEX with or without GCV using the same procedure as illustrated in Figure 3G. Conditioned media from different treatment groups were collected for an antibody array screening analysis to identify the secreted factors in the senescent BMAds. In this array, 111 candidate factors, including various types of cytokines and chemokines, were screened. As shown in Figure 6B and 6C, DEX-treated BMAds had higher expression of a broad range of the factors tested, whereas the expression of almost all these factors in GCV-co-treated cells became much lower relative to cells treated with DEX, and even reached the levels of the vehicle-treated cells in some cases, suggesting a diminished SASP of BMAds by senescent cell clearance. Of note, approximately half of these factors (labelled with red boxes in Figure 6C) were also identified in the RNA-seq analysis, further confirming them as SASP factors.

We then tested whether the SASP factors secreted by BMAds can act on other cell lineages in a paracrine manner. As we detected accumulated senescent vascular endothelial cells and osteoblasts in mouse bone tissue at a later stage (4 weeks) after MPS treatment in vivo (Figure 2), we assessed the effects of BMAd-secreted SASP factors on endothelial cell and osteoblast activities. To do so, we used the combination of Dasatinib and Quercetin (D+Q), which effectively eliminates senescent cells⁴⁸, and we prepared conditioned medium from vehicle-treated BMAds (Veh CM) and DEX-treated BMAds in the absence (DEX CM) or presence (DEX+DQ CM) of the senolytic drug D+Q. The conditioned media were added to the cultures of endothelial cells and osteoblasts (Figure 6D). Both endothelial cells and osteoblasts had significantly higher SA-βGal expression when incubated with DEX CM compared to Veh CM. Conditioned medium prepared from senescent cell–eliminated cells (DEX+DQ CM) failed to induce SA-βGal expression in the endothelial cell and osteoblast cultures (Figure 6E). Western blot analysis showed that DEX CM induced the expression of the senescence marker p16INK4a, whereas conditioned medium prepared from BMAds with DEX and senolytic drug co-treatment (DEX+DQ CM) failed to induce p16INK4a expression in endothelial cells and osteoblasts (Figure 6F). To test whether the senescence-associated BMAd secretome directly causes bone loss and widespread cellular senescence in bone and bone marrow, 3-month-old C57BL/6 mice were treated with vehicle and MPS for 2 weeks (Donors), and the mature BMAds and CD45⁻CD31⁻CD24⁻Sca1⁺ adipocyte precursor cells (APCs) were separately isolated from the mice and transplanted into the femoral bone marrow cavity of 3-month-old wild-type mice (Acceptors) (Figure 6G and S5A). Four weeks after transplantation, the bone phenotype and cellular senescence were examined. By μCT analyses, we found that the Acceptors transplanted with BMAds from MPS-treated donors (MPS-BMAd), compared with those transplanted with the same number of BMAds from vehicle-treated Donors (Veh-BMAd), exhibited a low bone mass phenotype (Figure 6H), reduced bone volume/tissue volume (BV/TV) (Figure 6I) and trabecular thickness (Tb.Th) (Figure 6J), unchanged trabecular number (Figure 6K), and increased trabecular separation (Tb.Sp) (Figure 6L). In contrast, the Acceptors transplanted with APCs isolated from MPS-treated donors (MPS-APC) did not show evident phenotypic bone changes relative to those transplanted with APCs from vehicle-treated donors (Veh-APC). Histological analyses showed dramatically increased SA-βGal⁺ senescent cells (Figure S5B–S5C) and loss of type-H vessels (Figure S5D–S5E) in bone and bone marrow of the MPS-BMAd (vs. Veh-BMAd) group. The impairment of bone angiogenesis as evidenced by the loss of type-H vessels was not observed in the MPS-APC (vs. Veh-APC) group although SA-βGal⁺ senescent cell number was a little higher in MPS-APC relative to Veh-APC group. The results suggest that senescent BMAds, but not the premature adipocyte precursors, are sufficient to cause secondary senescence in bone/bone marrow and the resultant bone deterioration.

Targeting senescent BMAds or the SASP attenuates glucocorticoid-induced bone loss

We asked whether inhibiting the initial senescence of BMAds could prevent glucocorticoid-induced secondary senescence of bone vascular cells and osteoblasts. To achieve this purpose, we generated conditional Adipoq-Cre::p16^flox/flox (p16-cKO) mice, in which p16INK4a is deleted selectively in adipocytes. Two weeks of MPS treatment in p16^flox/flox (wild-type [WT]) mice induced senescence of BMAds, as indicated by SA-βGal-expressing adipocytes that were also Oil Red O-positive. However, senescent SA-βGal⁺ BMAds were undetectable in p16-cKO mice treated with MPS for the same duration (Figure 7A and 7B), indicating a blockage of BMAd senescence by p16INK4a deletion from the cells. We then evaluated whether the secondary senescence induced by 4 weeks of MPS treatment can be prevented in p16-cKO mice. Indeed, 4-week MPS treatment in WT mice induced increases in the numbers of SA-βGal⁺Emcn⁺ double-positive cells (senescent vascular endothelial cells) (Figure S6A and S6B). However, the accumulation of these senescent cells was largely abrogated in p16-cKO mice treated with MPS for the same duration. As a result, the numbers of type-H vessels (Figure S6C and S6D) and Ocn⁺ mature osteoblasts (Figure S6E and S6F) were significantly increased in p16-KO mice compared with WT littermates after MPS treatment, suggesting improved angiogenesis and osteogenesis in p16-KO mice. By μCT analysis we found that the MPS-induced low-bone-mass phenotype (Figure 7C), characterized by lower bone volume/tissue volume (BV/TV) (Figure 7D) and trabecular thickness (Tb.Th) (Figure 7E), unchanged trabecular number (Tb. N) (Figure 7F), as well as an increase in trabecular separation (Tb.Sp) (Figure 7G), was partially rescued by p16 knockout in adipocytes. We are aware that Adipoq-Cre-driven p16 deletion may also inhibit adipocyte senescence in peripheral tissues. To further test if the SASP from peripheral adipose tissue may also be involved in glucocorticoid-induce bone defects, we transplanted BMAds isolated from MPS- or vehicle-treated Adipoq-Cre; p16^flox/flox mice (p16-cKO donors) into femoral bone marrow cavity of C57B/L6 mice (acceptors) (Figure 7H). BMAds from MPS-treated p16^flox/flox mice (WT donors) caused bone loss phenotype in acceptors relative to those transplanted with BMAds from vehicle-treated WT donors. Intriguingly, BMAds from MPS-treated p16-cKO donors failed to induce similar low bone mass phenotype in the acceptors (Figure 7I–7M). The results also suggest that the peripheral adipose tissue is not a major player in this process.

Figure 7. Targeting senescent BMAds rescues glucocorticoid-induced bone loss.

(A-G) Three-month-old AdipoQ-Cre::p16 flox/flox (p16 cKO) mice and p16 flox/flox (WT) mice were injected with daily MPS treatment at 10 mg/m²/day or vehicle for 2 weeks. Representative images of SA-βGal staining and Oil red staining in femoral bone tissue sections (A). Quantification of the number of SA-βGal-expressing BMAds per mm² tissue area (N. βGal⁺ BMAds) in (B). Representative micro-CT images of p16 cKO or WT littermates treated with vehicle or DEX are shown in (C) with quantitative analysis of trabecular bone volume (BV/TV) (D), trabecular thickness (Tb. Th) (E), trabecular number (Tb. N) (F), and trabecular separation (Tb. Sp) (G).

(H) Schematic diagram showing the experimental design for transplantation.

(I-M) Representative micro-CT images of WT mice transplanted with BMAds from p16 cKO or WT littermates treated with vehicle or DEX are shown in (I) with quantitative analysis of BV/TV (J), Tb. Th (K), Tb. N (L), and Tb. Sp (M).

(N) Diagram model illustrating the mechanisms of glucocorticoids induced BMAds senescence in bone and bone marrow.

n=5 mice. Data are represented as mean ±s.e.m. *p< 0.05, **p<0.01, ***p<0.001 as determined by two-way ANOVA.

Finally, to further validate the importance of BMAd-produced SASP factors in mediating the adverse effect of glucocorticoids on bone, we tested whether inhibiting the SASP using a pharmacological approach during the early-phase of MPS treatment might alleviate glucocorticoid-induced bone deficit. To do so, we used the JAK inhibitor (JAKi) ruxolitinib, which is known to inhibit the SASP of senescent cells^53,67. Mice were treated with vehicle, MPS, or MPS with JAKi, as illustrated in Figure S6G. Representative micro-CT images of distal femur micro-architecture showed gross observations of bone loss in the mice treated with MPS alone for 4 weeks, but much improved bone mass in those co-treated with JAKi for the first 2 weeks (Figure S6H). Quantification of the parameters in the trabecular bone of the distal femur showed significantly reduced BV/TV, Tb.Th, and Tb.N, but increased Tb.Sp, in the MPS group relative to the vehicle-treated group. These parameters were largely rectified by JAKi co-treatment (Figure S6I–S6L). Together, the in vivo data provides evidence to support the role of senescent BMAds in inducing bone deterioration through a paracrine effect.

DISCUSSION

It is now widely accepted that primary senescent cells induced by direct insults to the cells can induce secondary senescence in neighboring cells through SASP^68,69. Our study shows that glucocorticoid treatment induces primary senescence in a small number of BMAd lineage cells, which spread senescence to other bone and bone marrow cells by secreting SASP factors, leading to an accumulation of senescent cells and impaired bone microenvironment (Figure 7N). Moreover, we have identified a positive interacting feedback loop of 15d-PGJ2-PPARγ-INK signaling that triggers and maintains the primary senescence of the BMAd lineage cells. We also identified a panel of SASP factors from BMAds that induce the secondary senescence of bone vascular endothelial cells and osteoblasts. Inhibiting the primary BMAd senescence or suppressing the SASP alleviated glucocorticoid-induced bone loss. Hence, the senescent BMAd lineage cells may be a new biological target to treat glucocorticoid-associated bone disorders, such as osteoporosis and osteonecrosis.

Our data demonstrate that glucocorticoids induce primary senescence of BMAd lineage cells, characterized by multiple cellular senescence markers, including p16INK4a, SA-βGal, DNA damage marker γH2AX, and SADS, as well as loss of nuclear expression of HMGB1 and nuclear envelop protein lamin-B1. More importantly, out of the differentially expressed genes in the glucocorticoid- vs. vehicle-treated adipocytes/preadipocytes, 390 genes are aging- and cellular senescence-associated, strongly supporting the activation of a cellular senescence program in glucocorticoid-treated BMAd lineage cells. Of note, the INK family members Cdkn2a, Cdkn2c, and Cdkn2d, which encode for p16INK4a, p18INK4c, and p19INK4d, respectively, but not Cdkn1a, which encodes for p21CIP1, were upregulated in glucocorticoid treated BMAd lineage cells. It was recently identified that p16^high senescent cells and p21^high senescent cells in fat tissue are two distinct cell populations in the context of obesity⁷⁰. Our results suggest that INK family proteins, but not p21, are likely BMAd senescence-inducing factors that are upregulated in response to glucocorticoid treatment.

Our results showed that glucocorticoid-induced BMAd senescence is initiated in the committed adipocyte precursors as evidenced by a gradual increase in p16INK4a expression and the appearance of SA-βGal during the adipogenesis of BMSCs. However, our findings from the in vitro adipogenic differentiation culture and in vivo transplantation experiments strongly suggest that mature BMAds, but not adipocyte precursors, display the features of full senescence^52,53, such as dramatically high p16INK4a expression, DNA damage marker γH2AX expression, nuclear structural change represented by lamin-B1 loss, and the paracrine effects via SASP. Cellular senescence is a highly dynamic, multi-step process^52,53. It is generally accepted that the initiating step is the transition of temporal to stable cell-cycle arrest through activation of p53-p21 and/or p16INK4a pathway. The progression to full senescence is characterized by lamin-B1 reduction-associated nuclear deformation, chromatin remodeling, and the secretory phenotype (i.e., SASP). Therefore, glucocorticoids activate a progressive cellular senescence program in BMAd lineage cells, with mature BMAds having a full senescence phenotype. Of note, the elevation of adipogenesis marker genes, including Pparg, Cebpa, and Adipoq, occurred earlier than the appearance of the cellular senescence marker Cdkn2a in the BMSC culture assays. Moreover, glucocorticoid treatment did not induce increased expression of senescence markers in uncommitted BMSCs. Therefore, the senescence program is activated and completed much later during the process of BMSC adipogenesis.

We identified a molecular mechanism underlying glucocorticoid-induced senescence of BMAd lineage cells. MPS treatment induces elevated production of oxylipins, particularly PGD2 and its derivative 15d-PGJ2, in BMAds. Although we had technical difficulties in detecting intracellular dihomo-15d-PGJ2 level, our findings that 15d-PGJ2 is sufficient to induce BMAd senescence and that a selective COX2 inhibitor antagonized BMAd senescence induced by DEX agrees with a recent report that PGD2-related oxylipins, especially dihomo-15d-PGJ2, promote cellular senescence and the SASP⁴⁵. We further found that these oxylipins activate PPARγ, which is both necessary and sufficient to induce BMAd senescence. It was reported that PPARγ accelerates cellular senescence by inducing p16INK4a expression in human diploid fibroblasts ⁷¹. Consistent with this finding, we found that both the PPARγ endogenous agonist 15d-PGJ2 and PPARγ synthetic agonists induces dramatic upregulation of the key senescence-inducing factors p16INK4a and p27KIP1, whereas knockdown of PPARγ in BMAds or PPARγ antagonism inhibits the upregulation of these genes induced by DEX treatment. In fact, our RNA-seq result shows that the PPARγ antagonist GW9662 normalizes the expression levels of many more cellular senescence-associated genes that were upregulated in DEX-treated cells, further confirming the key role of PPARγ activation in mediating BMAd senescence. Moreover, we provide evidence that PPARγ activation is also necessary and sufficient to upregulate the gene expression of Ptgds and Ptgis in BMAds. Therefore, oxylipins and PPARγ form a strong positive feedback loop, contributing substantially to cellular senescence of BMAds. This finding suggests that even subtle alterations in this signaling circuit can be amplified, resulting in rapid cellular senescence.

Our in vitro and in vivo data suggest that glucocorticoid treatment directly activate a senescence-associated program in BMAd lineage cells but not in other cell types, such as uncommitted BMSCs, vascular endothelial cells, and osteoblasts. Moreover, our data suggest that the secondary senescence of bone vascular cells and osteoblasts is indirectly mediated through SASP factors secreted by the senescent BMAds. Our RNA-seq and protein array study consistently reveal broadly increased levels of growth factors, cytokines, and chemokines in DEX-treated BMAds, all of which were downregulated after clearance of the senescent cells. This finding suggests that the SASP of BMAds is an important contributor to the paracrine function of BMAds under pathological conditions. Indeed, our in vitro results show that CM prepared from the senescent BMAds induced the senescence of vascular endothelial cells and osteoblasts, whereas the senolytic drugs D+Q blunted the senescence of both cell types, validating the role of the SASP of BMAds in inducing the secondary senescence of the neighboring cells.

Our in vivo results further demonstrate that short-term administration (2 weeks) of the SASP inhibitor (JAKi) also largely rescues the low bone mass phenotype in mice treated with longer-term glucocorticoid (4 weeks). Moreover, blockade of primary BMAd senescence by deleting p16 selectively in BMAds almost fully abolished secondary senescence of the bone vascular endothelial cells and osteoblasts in glucocorticoid-treated mice. We are aware that the rescued bone phenotype may not be solely attributed to the inhibition of BMAd senescence as AdipoQ-Cre has been used to label not only mature adipocytes but also adipocyte precursors and multi-potent BMSCs^72–74. However, our in vitro and in vivo experiments showed that senescence was not induced in BMSCs in response to glucocorticoid treatment. More importantly, our transplantation experiments clearly showed that transplanting mature adipocytes, but not adipocyte precursor cells isolated from MPS-treated mice, results in secondary senescence in bone marrow and a bone loss phenotype. The results suggest that mature BMAds with full senescence features are sufficient to spread the senescence in bone and bone marrow microenvironment for bone loss, whereas immature adipocyte precursors do not have this capacity.

It appears that the primary bone/bone marrow cell types undergoing cellular senescence in response to glucocorticoid treatment are age- and location-dependent. We recently revealed that in growing young mice, glucocorticoid treatment induces vascular endothelial cell senescence in the metaphysis of long bone by reducing the number of osteoclasts and the secreted angiogenin²⁷. During childhood, the metaphysis of long bone has abundant osteogenesis-coupled type-H vessels with actively proliferative endothelial cells, whereas type-H vessels diminish gradually in adulthood⁷⁵. Conversely, there is much less bone marrow adipose tissue at young ages, as it accumulates with aging⁵⁴. It is not surprising that vascular cells in the metaphysis of growing bone are the primary cell type that becomes senescent, whereas BMAds undergo more repaid senescence than other cell types in adult bone in response to glucocorticoid treatment.

Limitations of the study

Several important questions arise from our work. For instance, accumulated senescent BMAds have been observed in old mice⁷⁶, but is oxylipin-PPARγ-INK signaling is also involved in the senescence of BMAds during aging? Bone marrow adipose tissue mass increases under many pathological conditions, such as obesity, type 2 diabetes, anorexia, hyperlipidemia. Are there senescent BMAd accumulation under these conditions? Does the identified mechanism, i.e. oxylipin-PPARγ-INK signaling, also mediate BMAd senescence in these conditions? Is BMAd senescence a key contributor to bone marrow adiposity? Finally, we are aware that it is technically challenging to fully exclude the influence of peripheral adipose tissue on glucocorticoid-induced bone remodeling changes. It is important to determine whether glucocorticoids also induce cellular senescence in peripheral adipose tissue and whether transplanting senescent peripheral adipose tissue from mice may have similar adverse bone effects. These experiments are important to define the distinct role of BMAds and peripheral adipose tissue in regulating bone/bone marrow in pathological settings. Recently, a BMAd-specific Cre mouse model was generated⁷⁷. This mouse line serves as a useful tool to study the specific contribution of BMAds to glucocorticoid-induced bone deterioration.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mei Wan (mwan4@jhmi.edu).

Materials availability

• This study did not generate new unique reagents.

Data and code availability

• RNA-Seq of differentiated BMAds data has been deposited in the NCBI GEO Database and is publicly available as of the date of publication with the accession number: GSE205732.

• Unprocessed source data underlying all blots and graphs is deposited in Data S1.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse models

AdipoQ-Cre mice (C57BL/6 background, stock no. 010803) were purchased from The Jackson Laboratory (Bar Harbor, ME). p16^flox/flox mice (C57BL/6 background) were generated by Dr. Gloria H. Su’s laboratory from the Department of Pathology, Columbia University Medical Center⁷⁸. p16^tdTom reporter mice (C57BL/6 background) were generated by Dr. Norman E. Sharpless’s laboratory, University of North Carolina School of Medicine (Chapel Hill, NC)⁴⁷.

We crossed the 3-month-old Adipoq-Cre mice with 3-month-old p16^flox/flox mice. The offspring were backcrossed with p16^flox/flox mice (both sexes) to generate Adipoq-Cre::p16^flox/flox (p16 cKO) and p16^flox/flox (WT) mice. 3-month-old male p16 cKO mice and WT mice were used for experiments. The genotypes of the mice were determined by PCR analyses of genomic DNA extracted from mouse-tail snips using the following primers: Adipoq-Cre forward, 5’- GGA TGT GCC ATG TGA GTC TG-3’ and reverse, 5’- ACG GAC AGA AGC ATT TTC CA −3’; loxP p16 allele forward, 5’-AGG AGT CCT GGC CCT AGA AA-3’ and reverse, 5’-CCA AAG GCA AAC TTC TCA GC-3’.

To detect bone phenotypes after glucocorticoid treatment, we injected 3-month-old male C57BL/6 mice with MPS (10 mg/m²/day) for different time periods (3 days, or 1, 2 or 4 weeks). Body surface area (BSA) was calculated using Meeh’s formula, with a k constant of 9.82 for mice × body weight (g) to the two-thirds power (BSA = kŴ2/3). For ruxolitinib (JAKi) treatment, JAKi was dissolved in 10% DMSO and administered by oral gavage at a dosage of 60 mg/kg daily for 2 weeks as previously described⁷⁶.

All mice were maintained at the animal facility of The Johns Hopkins University School of Medicine (Baltimore, Maryland) under protocol MO18M139 approved by the Institutional Animal Care and Use Committee of The Johns Hopkins University. Mice were housed in normal conditions with a 12h:12h light:dark cycle in a temperature-controlled room with food and water provided ad libitum.

Cell lines

In this study, human umbilical vein endothelial cells (HUVECs) were purchased from ATCC (Cat. No, PCS-100-013). Cells were cultured in Vascular Cell Basal Medium (PCS-100-030, ATCC) supplemented with Endothelial Cell Growth Kit-BBE (PCS-100-040, ATCC) and penicillin/streptomycin at 37°C.

Primary bone marrow stromal cells (BMSCs), BMAds and bone marrow adipocyte precursor cells (APC) were isolated from 3-month-old C57BL/6 mice. BMSCs were maintained in BMSCs growth media (MesenCult MSC Expansion Kit, Stem Cell Technologies, Cat #05513) with penicillin/streptomycin at 37°C. BMAds and adipocyte precursor cells (APC) were maintained in DMEM supplemented with 15% fetal bovine serum and penicillin/streptomycin at 37°C. Primary osteoblasts were isolated from young (2-week-old) C57BL/6 mice and maintained in DMEM supplemented with 15% fetal bovine serum and penicillin/streptomycin at 37°C. For primary cell isolation protocols see “method details” section.

METHOD DETAILS

μCT analysis of the femoral bone

Femurs were harvested from mice, dissected free of soft tissue, and fixed in 4% phosphate-buffered paraformaldehyde for 48 hours. For μCT analysis, mice femora were dissected free of soft tissue, fixed overnight in 10% formalin at 4°C and analyzed by high-resolution μCT (Skyscan 1172, Bruker MicroCT, Kontich, Belgium). The scanner was set at 65 kV, 153 μA and a resolution of 9.0 μm/pixel. We used NRecon image reconstruction software, version 1.6 (Bruker MicroCT), CTAn data-analysis software, version 1.9 (Bruker MicroCT), and CTVol 3-dimensional model visualization software, version 2.0 (Bruker MicroCT) to analyze parameters of the trabecular bone in the metaphysis. To perform 3-dimensional histomorphometry analysis of trabecular bone, we selected the regions of interest from 1mm below the distal epiphyseal growth plate(reference level) and extending toward the distal direction for proximally 2mm in length. Trabecular bone was analyzed to determine trabecular BV/TV, Tb.Th, Tb.N, and Tb.Sp. Femur length was analyzed using CTAn data-analysis software, version 1.9 (Bruker MicroCT).

Histological analysis

Femoral bones were dissected and freshly fixed in 4% paraformaldehyde overnight, followed by a 14-day decalcification in 0.5M EDTA (pH 7.4). The samples were then dehydrated in 20% sucrose plus 2% polyvinylpyrrolidone solution for 24 hours and embedded in OCT. Ten μm-thick coronal sections of the femurs were obtained for SA-βGal staining. Senescent cells were detected using a senescence βGal staining kit according to the manufacturer’s instructions (Cell Signaling Technology, Danvers, MA). For Oil Red O staining following SA-βGal staining, the slides were rinsed in PBS for 3 times, placed in 100% propylene glycol for 2 min, and stained in filtered 0.5% Oil Red O solution in propylene glycol for 30 min. The slides were transferred into an 85% propylene glycol solution for 1 min, rinsed in PBS 3 times, mounted, and sealed with nail polish. Representative images were acquired with a BX51 microscope (Olympus, Tokyo, Japan).

Twenty μm-thick coronal sections of the femurs were obtained for immunofluorescence staining. Bone sections were blocked in PBS with 3% BSA for 1 hour and then stained overnight (>8 hours) with a primary antibody: Emcn (Santa-Cruz, sc-65495, 1:50), CD31 (R&D Systems, FAB3628G, 1:100), γH2AX (Cell signaling, 20E3, 1:200), Perilipin (Sigma, P1873, 1:500), Perilipin (Cell signaling, 9349, 1:200), Osterix (Abcam, ab22552, 1:200), Osteocalcin (Takara, M188, 1:200), RFP (Rockland Immunochemicals, 600-401-379, 1:200) and HMGB1 (Abcam, ab18256, 1:300). Fluorescence-conjugated secondary antibodies (Jackson ImmunoResearch, 1:200) were used in immunofluorescence procedures to detect fluorescent signals. Slides were then mounted with anti-fade prolong gold (Invitrogen) mounting medium and sealed with nail polish, after which representative images were acquired using a Zeiss LSM780 confocal microscope or an Olympus BX51 microscope.

Senescence-associated distension of satellites (SADS) analysis

SADS assay, which is a fluorescence in situ hybridization (FISH) for detection of peri-centromeric satellite heterochromatin DNA, was performed following the protocol kindly provided by Drs. Sundeep Khosla and Joshua Farr⁷⁶. Briefly, paraffin-embedded femoral bone tissue sections were cross-linked with 4% paraformaldehyde and dehydrated in ethanol. The bone tissue sections were then denatured and hybridized in the buffer containing 1.0 μg/mL Cy3-labeled, CENPB-specific (ATTCGTTGGAAACGGGA) peptide nucleic acid FISH probe (Panagene Inc, Korea) for 2 hours at room temperature in the dark. After washing, immunofluorescence staining of the tissue sections was performed using primary antibody against perilipin (Novus Biologicals, NB100-60554, 1:500). Finally, the slides were mounted with DAPI-containing mounting media (Life Technologies). Images were acquired with a Zeiss LSM780 confocal microscope, and the intensity profiles were acquired with Image J software (NIH).

FACS sorting and flow cytometry

Flow cytometric analysis and FACS sorting of bone vascular cells from femur were performed as previously described^79,80 with modifications. Briefly, the epiphysis was removed from the distal femora and proximal tibia, and the trabecular bones were then crushed in ice-cold PBS with a mortar and pestle. Whole bone marrow was digested with collagenase A (Sigma, 11088793001, 2 mg/mL) and trypsin (Gibco, 27250018, 2.5 mg/mL) in PBS at 37°C for 30 min to obtain single-cell suspensions. Ammonium-chloride-potassium lysing buffer was used to remove red blood cells. For sorting of vascular cells, after washing and filtration, cells were incubated for 45 min at 4°C with CD144 antibody (BV421, Biolegend, 1:50) and sorted using BD FACS Aria II (BD Biosciences). p16^tdTom mice were used for the analyses of the incidence of tdTom⁺ cells in different bone marrow cell populations. The antibody used in this study to identify vascular endothelial cells was anti-CD144 (BV421, Biolegend, 1:50). The antibodies used to identify BMSCs and APCs were anti-CD45 (APC, Biolegend, 1:50), anti-CD31 (APC, Biolegend, 1:50), anti-Sca1 (APC-Cy7, Biolegend, 1:50) and anti-CD24 (BV421, Biolegend, 1:50). Flow cytometry cells were analyzed using a BD LSRII flow cytometer.

Isolation of BMAds and adipocyte precursor cells (APC)

Mature BMAds were isolated directly from bone marrow of mice according to previously described protocols (Scheller et al., 2015; Fan et al., 2017). Briefly, femurs and tibias were collected from mice, and the two ends of the bones were snipped. The bones were placed in a small microcentrifuge tube (0.6 mL) that was cut open at the bottom. The small tube with the bones were then placed into a bigger microcentrifuge tube (1.5 mL). Fresh bone marrow was spun out by quick centrifuge. Red blood cells were lysed using RBC lysing buffer (Sigma). After centrifugation (3000 rpm, 5 minutes, RT), floating adipocytes were collected from the top layer and washed with PBS for 3 times. To isolate APCs, cell pellet from the previous step was collected and washed with PBS for 3 times. The cells were then resuspended and incubated with CD45 (APC, Biolegend, 1:50), CD31 (APC, Biolegend, 1:50), Sca1 (APC-Cy7, Biolegend, 1:50) and CD24 (BV421, Biolegend, 1:50). FACS sorting of CD45⁻CD31⁻CD24⁻Sca1⁺ APCs was performed using BD FACS Aria II (BD Biosciences).

Cell Transplantation

To induce senescence of BMAds to be transplanted, donor mice were injected with MPS (10 mg/m²/day) for 2 weeks. BMAds and APCs were isolated, respectively as described in the above section. For intra-femoral transplantation, acceptor mice (C57BL/6) were anesthetized, and a longitudinal skin incision was made across the front of the right knee to visualize patellar tendon. A 27G needle was inserted through the patellar tendon, between the femur condyles, and the needle was advanced slowly in a twisting motion through the bone surface to 2–4mm deep. Successful penetration was confirmed by slow but constant bleeding. The cell suspensions of BMAds (3*10³) or APCs (2*10⁴) in 20 μl PBS were slowly injected into the medullary space of the femur. The hole was immediately sealed with bone wax, and the skin was sutured.

BMSCs isolation, culture, and adipogenesis differentiation

BMSCs isolation, culture, and adipogenesis differentiation were conducted according to previously described protocol with customization^81,82. Briefly, 3-month-old male C57BL/6 mice were euthanized, and femurs and tibias were dissected. Bone marrow cells were flushed by using 22-gauge syringes with PBS. The freshly isolated single-cell suspensions were plated at a density of 5 × 10⁴ /cm² in 6-well plates in BMSCs growth media (MesenCult MSC Expansion Kit, Stem Cell Technologies, Cat #05513). Cells were allowed to grow for 3 days before supernatant was aspirated, rinsed with PBS 3 times, and medium was changed 3 times a week. For adipogenesis differentiation, MSCs were cultured with adipogenesis media (StemPro Adipogenesis Differentiation kit, Gibco) for designated days with medium changed 3 times a week.

Primary osteoblast isolation

Primary osteoblasts were isolated from young (2-week-old) mice. Briefly, calvarium was aseptically dissected and processed by serial digestion using 300 active U/mL of collagenase type-IA (Sigma-Aldrich) dissolved in α-minimal essential medium (αMEM; Gibco). All steps of the digestion took place in a 6-well Petri dish, on a rotating shaker set to 200 RPM, in a 37°C and 5% CO₂ humidified incubator. Following each sequential digestion, the digest solution with suspended cells was removed from the bone fragments and kept. The bone fragments were rinsed with Hank’s balanced salt solution 3 times, and the rinse buffer was added to the digestion solution. The combined cell suspension solution was spun down at 200 × g for 5 min, followed by removal of supernatant from the cell pellet and resuspension of the cells in culture medium.

In vitro cell treatment and staining

For treatments, 10⁻⁶ M DEX was added to cells and cultured for designated times. Other drugs and compounds used in this study were as follows: ganciclovir (GCV, MilliporeSigma, 20*10⁻⁶ M), rosiglitazone (MilliporeSigma, 20*10⁻⁶ M), T0070907 (MilliporeSigma, 10⁻⁶ M), GW9662 (MilliporeSigma, 10⁻⁶ M), dasatibib (MilliporeSigma, 10⁻⁶ M) + quercitin (MilliporeSigma, 10*10⁻⁶ M), 15d-PGJ2 (MilliporeSigma, 10*10⁻⁶ M), and celecoxib (MilliporeSigma, 50*10⁻⁶ M). Dosages and time courses are noted in the corresponding text and figure legends.

For SA-βGal staining, cells were fixed and stained with a senescence βGal staining kit according to the manufacturer’s instructions (Cell Biolabs). For immunofluorescence staining, cells were washed with PBS, fixed in 10% formalin, and blocked with 30 mg/ml BSA/PBS at 37°C for 1 hour. Cells were stained with primary antibody at 4°C overnight. Primary antibody used were: γH2AX (Cell signaling, 20E3, 1:200), Perilipin (Cell signaling, 9349, 1:200) and Lamin-B1 (Santa cruz, sc-374015, 1:200). Fluorescence-conjugated secondary antibodies (Jackson ImmunoResearch, 1:200) were used in immunofluorescence procedures to detect fluorescent signals. Cells were then mounted with anti-fade prolong gold mounting medium (Invitrogen) and imaged with inverted fluorescence microscope (Olympus).

Proteomic profiling in conditioned medium (CM) collected from cell culture

CM was collected from each group after centrifugation at 700 × g for 10 min at 4°C and stored at −80°C for subsequent experiments. For cytokine measurement, conditioned medium was subjected to a proteome profiler array using a mouse XL cytokine array kit (R&D, ARY028), and the relative levels of 111 cytokines and chemokines were measured. The array procedure and data analysis were performed according to the manufacturer’s instructions.

For measurement of prostaglandins, CM were collected as described and analyzed according to the manufacturer’s instructions. The ELISA kit used for measurement were as follows: PGD2-MOX ELISA(Cayman Chemical company, 500151); PGJ2 ELISA(ENZO, ADI-900-023); PGE2ELISA(R&D systems, KGE004B).

Western blot analysis

Western blot analysis was conducted as previously described^83,84. Briefly, cells were lysed in RIPA lysis buffer and harvested with a rubber policeman. Samples (20 μg protein) were loaded into precast electrophoresis gels (Bio-Rad), separated by SDS-PAGE and electro-transferred onto a PVDF membrane. After blocking with 5% nonfat milk, the blots were incubated with primary antibodies overnight at 4°C and then with secondary antibodies at room temperature for 1 hour. Blots were developed with the SuperSignal West Femto chemiluminescence kit (Thermo Scientific) followed by autoradiography. The following antibodies were used: anti-mouse p16 (Abcam, 211542, 1:1000), anti-human p16 (Cell Signaling, 80772, 1:1000), anti-mouse p21 (Cell Signaling, 37543, 1:1000), anti-human p21 (Abcam, ab109520, 1:1000), and GAPDH (Cell Signaling, 2118, 1:1000).

qRT-PCR

Total RNA for qRT-PCR was purified from the cultured or sorting cells using RNeasy Mini Kit (QIAGEN) according to the manufacturer’s protocol. Complementary DNA (cDNA) was prepared with random primers using the SuperScript First-Strand Synthesis System (Invitrogen) and analyzed with SYBR GreenMaster Mix (QIAGEN) in the thermal cycler with 2 sets of primers specific for each targeted gene. Target-gene expression was normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) messenger RNA, and relative gene expression was assessed using the 2^−ΔΔCT method. Primers used for qRT-PCR were as follows: Mki67 (5′-ACCGTGGAGTAGTTTATCTGGG-3′) and (5′-TGTTTCCAGTCCGCTTACTTCT-3′); Cdkn2a (5′- GAAAGAGTTCGGGGCGTTG-3′) and (5′-GAGAGCCATCTGGAGCAGCAT-3′); Tp53 (5′-ATCGCCTTCGACATCATCGC-3′) and (5′-CCCCATGCGTACTCCATGAG-3′); Cdkn1a (5′-AGAAGGTACTTACGGTGTGGT-3′) and (5′-GAGAGATTTCCCGAATTGCAGT-3′); Cox-2 (5′- CCACTTCAAGGGAGTCTGGA-3′) and (5′- AGTCATCTGCTACGGGAGGA-3′); and Pparg 5′- GTACTGTCGGTTTCAGAAGTGCC-3′) and (5′- ATCTCCGCCAACAGCTTCTCCT-3′).

Preparation and sequencing of RNA-sequencing libraries

Bone marrow adipocytes were differentiated from isolated bone marrow stem cells with StemPro Adipogenesis Differentiation kit (Gibco) for 14 days and treated with 10⁻⁶ M DEX or 10⁻⁶ M DEX plus GW9662 (MilliporeSigma, 10⁻⁶ M) for 24 hours. The cells were lysed and total RNA was collected using RNeasy Mini Kit (QIAGEN, 74014) following manufacturer’s instruction. RNA integrity was assessed using Agilant RNA 6000 Nano kit (Agilent), and RNA with a RIN > 7 was used in sequencing. Libraries were prepared using the Smart-Seq 2 protocol with a template switching oligonucleotide, followed by PCR amplification and purification. The constructed cDNA libraries were sequenced using an Illumina HiSeq 2000. The RNA-seq data have been deposited to the GEO repository (GSE205732).

RNA-sequencing data analysis

Sample demultiplexing and conversion to FASTQ files was performed using Illumina’s fastq software with all default options. Reads from FASTQ files of the cultured cells were aligned to mouse genome using the STAR aligner (v2.4.2). Uniquely mapped reads were used for gene expression estimates as transcripts per million reads (TPMs). Kruskal–Wallis tests were used for identifying genes differentially expressed among sample groups. Pathway analysis was performed for functional annotation of the ASIGs in the dataset using established tools available online (EnrichR). Heatmap was formed using established online tool (Bioinfo intelligent cloud).

QUANTIFICATION AND STATISTICAL ANALYSIS

Data are presented as means ± standard errors of the mean. Unpaired, 2-tailed Student t-tests were used for comparisons between 2 groups. For multiple comparisons, 1-way or 2-way analysis of variance with Bonferroni post hoc test was used. All data were normally distributed and had similar variation between groups. Statistical analysis was performed using SPSS software, version 15.0 (IBM, Armonk, NY). p< 0.05 was deemed significant. All representative images of bones or cells were selected from at least 3 independent experiments with similar results unless indicated otherwise in the figure legend.

Supplementary Material

1

Data S1. Unprocessed source data underlying all blots and graphs. Related to Figures 1–7 and Supplemental Figures 1–6.

3

Table S1. Expression matrix of RNA-Seq. Related to Figures 3–6.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rat monoclonal anti-Emcn	Santa-Cruz	CAT# sc-65495
Mouse/Rat polyclonal anti-CD31	R&D Systems	CAT# FAB3628G
Rabbit monoclonal anti- γH2AX	Cell signaling Technology	CAT# 20E3
Rabbit polyclonal anti-Perilipin	Sigma	CAT# P1873
Rabbit polyclonal anti-Perilipin	Cell signaling Technology	CAT# 9349
Rabbit polyclonal anti-Osterix	Abcam	CAT# ab22552
Mouse monoclonal anti-Osteocalcin	Takara	CAT# M188
Rabbit polyclonal anti-RFP	Rockland Immunochemicals	CAT# 600-401-379
Rabbit polyclonal to HMGB1	Abcam	CAT# ab18256
Rat monoclonal anti-CD144	Biolegend	CAT# 138013
Rat monoclonal anti-CD45	Biolegend	CAT# 103111
Mouse monoclonal anti-CD31	Biolegend	CAT# 303115
Rat monoclonal anti-Sca1	Biolengend	CAT# 108125
Rat monoclonal anti-CD24	Biolegend	CAT# 101825
Mouse monoclonal anti-Lamin-B1	Santa cruz	CAT# sc-374015
Rabbit monoclonal anti-human p16	Abcam	CAT# 211542
Rabbit monoclonal anti-mouse p16	Cell signaling Technology	CAT# 80772
Rabbit monoclonal anti-mouse p21	Cell signaling Technology	CAT# 37543
Rabbit monoclonal anti-human p21	Abcam	CAT# ab109520
Rabbit monoclonal anti-GAPDH	Cell Signaling	CAT# 2118
Biological samples
Chemicals, peptides, and recombinant proteins
Collagenase A	Sigma	CAT# 11088793001
Trypsin	Gibco	CAT# 27250018
Ganciclovir	MilliporeSigma	CAT# 82410-32-0
Rosiglitazone	MilliporeSigma	CAT# 155141-29-0
T0070907	MilliporeSigma	CAT# 313516-66-4
GW9662	MilliporeSigma	CAT# 22978-25-2
Dasatibib	MilliporeSigma	CAT# 302962-49-8
Quercitin	MilliporeSigma	CAS# 117-39-5
15d-PGJ2	MilliporeSigma	CAS# 87893-55-8
Celecoxib	MilliporeSigma	CAS# PZ008
Critical commercial assays
Senescence βGal staining kit	Cell Signaling Technology	Cat# 9860S
Mouse XL cytokine array kit	R&D System	Cat# ARY028
PGD2-MOX ELISA kit	Cayman Chemical Company	Cat# 500151
PGJ2 ELISA kit	ENZO	Cat# ADI-900-023
PGE2 ELISA kit	R&D System	Cat# KGE004B
RNeasy Mini Kit	QIAGEN	Cat# 74014
BCA protein assay kits	Thermo Fisher Scientific	Cat# 23227
Deposited data
RNA-Seq of differentiated BMAds	GEO Database	GSE205732
Data S1. Unprocessed source data underlying all blots and graphs.	This paper
Experimental models: Cell lines
Human umbilical vein endothelial cells(HUVEC)	ATCC	Cat# PCS-100-013
Experimental models: Organisms/strains
Mouse: AdipoQ-Cre mice	The Jackson Laboratory	JAX: 010803
Mouse: C57BL/6 floxed p16	Qiu et al.	https://www.entcolumbia.org/research/research-labs/su-lab
Mouse: p16-tdTomreporter mice (C57BL/6 background)	Liu et al.	https://doi.org/10.1073/pnas.1818313116
Mouse: C57BL/6J	The Jackson Laboratory	JAX:000664
Oligonucleotides
PCR primer for genotype: Adipoq-Cre forward: 5’- GGA TGT GCC ATG TGA GTC TG -3’	The Jackson Laboratory	https://www.jax.org/Protocol?stockNumber=010803&protocolID=25488
PCR primer for genotype: Adipoq-Cre reverse: 5’- ACG GAC AGA AGC ATT TTC CA-3’	The Jackson Laboratory	https://www.jax.org/Protocol?stockNumber=010803&protocolID=25488
PCR primer for genotype: p16-floxed forward: 5’- AGG AGT CCT GGC CCT AGA AA-3’	Qiu et al.	https://doi.org/10.18632/oncotarget.357
PCR primer for genotype: p16-floxed reverse: 5’- CCA AAG GCA AAC TTC TCA GC-3’	Qiu et al.	https://doi.org/10.18632/oncotarget.357
RT-PCR primer: Mki67 forward: 5′-ACCGTGGAGTAGTTTATCTGGG-3′	Liu et al	N/A
RT-PCR primer: Mki67 Reverse: 5′-TGTTTCCAGTCCGCTTACTTCT-3′	Liu et al	N/A
RT-PCR primer: Cdkn2a Forward: 5′- GAAAGAGTTCGGGGCGTTG-3′	Liu et al	N/A
RT-PCR primer: Cdkn2a Reverse 5′-GAGAGCCATCTGGAGCAGCAT-3′	Liu et al	N/A
RT-PCR primer: Tp53 Forward: 5′-ATCGCCTTCGACATCATCGC-3′	Liu et al	N/A
RT-PCR primer: Tp53 Reverse: 5′-CCCCATGCGTACTCCATGAG-3′	Liu et al	N/A
RT-PCR primer: Cdkn1a Forward: 5′-AGAAGGTACTTACGGTGTGGT-3′	Liu et al	N/A
RT-PCR primer: Cdkn1a Reverse: 5′-GAGAGATTTCCCGAATTGCAGT-3′	Liu et al	N/A
RT-PCR primer: COX-2 Forward: 5′- CCACTTCAAGGGAGTCTGGA-3′	This paper	N/A
RT-PCR primer: COX-2 Reverse: 5′- AGTCATCTGCTACGGGAGGA-3′	This paper	N/A
RT-PCR primer: Pparg Forward: 5′- GTACTGTCGGTTTCAGAAGTGCC-3′	This paper	N/A
RT-PCR primer: Pparg Reverse: 5′- ATCTCCGCCAACAGCTTCTCCT-3’	This paper	N/A
Software and algorithms
NRecon image reconstruction software	Bruker MicroCT	Version 1.6
CTAn data-analysis software	Bruker MicroCT	Version 1.9
CTVol 3-dimensional model visualization software	Bruker MicroCT	Version 2.0
Prism 9	Graphpad software	http://www.graphpad.com/
Flowjo_v10.8	Flowjo LLC.	http://www.flowjo.com/
Zen black	Carl Zeiss Microscopy	https://www.zeiss.com/microscopy/
Other
Vascular Cell Basal Medium	ATCC	Cat# PCS-100-030
Endothelia Cell Growth Kit-BBE	ATCC	Cat# PCS-100-040
MesenCult MSC Expansion Kit	Stem Cell Technologies	Cat# 05513
StemPro Adipogenesis Differentiation kit	Gibco	Cat# A1007001

LIFE SCIENCE TABLE WITH EXAMPLES FOR AUTHOR REFERENCE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-Snail	Cell Signaling Technology	Cat#3879S; RRID: AB_2255011
Mouse monoclonal anti-Tubulin (clone DM1A)	Sigma-Aldrich	Cat#T9026; RRID: AB_477593
Rabbit polyclonal anti-BMAL1	This paper	N/A
Bacterial and virus strains
pAAV-hSyn-DIO-hM3D(Gq)-mCherry	Krashes et al.1	Addgene AAV5; 44361-AAV5
AAV5-EF1a-DIO-hChR2(H134R)-EYFP	Hope Center Viral Vectors Core	N/A
Cowpox virus Brighton Red	BEI Resources	NR-88
Zika-SMGC-1, GENBANK: KX266255	Isolated from patient (Wang et al.2)	N/A
Staphylococcus aureus	ATCC	ATCC 29213
Streptococcus pyogenes: M1 serotype strain: strain SF370; M1 GAS	ATCC	ATCC 700294
Biological samples
Healthy adult BA9 brain tissue	University of Maryland Brain & Tissue Bank; http://medschool.umaryland.edu/btbank/	Cat#UMB1455
Human hippocampal brain blocks	New York Brain Bank	http://nybb.hs.columbia.edu/
Patient-derived xenografts (PDX)	Children’s Oncology Group Cell Culture and Xenograft Repository	http://cogcell.org/
Chemicals, peptides, and recombinant proteins
MK-2206 AKT inhibitor	Selleck Chemicals	S1078; CAS: 1032350-13-2
SB-505124	Sigma-Aldrich	S4696; CAS: 694433-59-5 (free base)
Picrotoxin	Sigma-Aldrich	P1675; CAS: 124-87-8
Human TGF-β	R&D	240-B; GenPept: P01137
Activated S6K1	Millipore	Cat#14-486
GST-BMAL1	Novus	Cat#H00000406-P01
Critical commercial assays
EasyTag EXPRESS 35S Protein Labeling Kit	PerkinElmer	NEG772014MC
CaspaseGlo 3/7	Promega	G8090
TruSeq ChIP Sample Prep Kit	Illumina	IP-202-1012
Deposited data
Raw and analyzed data	This paper	GEO: GSE63473
B-RAF RBD (apo) structure	This paper	PDB: 5J17
Human reference genome NCBI build 37, GRCh37	Genome Reference Consortium	http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/human/
Nanog STILT inference	This paper; Mendeley Data	http://dx.doi.org/10.17632/wx6s4mj7s8.2
Affinity-based mass spectrometry performed with 57 genes	This paper; Mendeley Data	Table S8; http://dx.doi.org/10.17632/5hvpvspw82.1
Experimental models: Cell lines
Hamster: CHO cells	ATCC	CRL-11268
D. melanogaster: Cell line S2: S2-DRSC	Laboratory of Norbert Perrimon	FlyBase: FBtc0000181
Human: Passage 40 H9 ES cells	MSKCC stem cell core facility	N/A
Human: HUES 8 hESC line (NIH approval number NIHhESC-09-0021)	HSCI iPS Core	hES Cell Line: HUES-8
Experimental models: Organisms/strains
C. elegans: Strain BC4011: Strain BC4011: srl-1(s2500) II; dpy-18(e364) III; unc-46(e177)rol-3(s1040) V.	Caenorhabditis Genetics Center	WB Strain: BC4011; WormBase: WBVar00241916
D. melanogaster: RNAi of Sxl: y[1] sc[*] v[1]; P{TRiP.HMS00609}attP2	Bloomington Drosophila Stock Center	BDSC:34393; FlyBase: FBtp0064874
S. cerevisiae: Strain background: W303	ATCC	ATTC: 208353
Mouse: R6/2: B6CBA-Tg(HDexon1)62Gpb/3J	The Jackson Laboratory	JAX: 006494
Mouse: OXTRfl/fl: B6.129(SJL)-Oxtrtm1.1Wsy/J	The Jackson Laboratory	RRID: IMSR_JAX:008471
Zebrafish: Tg(Shha:GFP)t10: t10Tg	Neumann and Nuesslein-Volhard3	ZFIN: ZDB-GENO-060207-1
Arabidopsis: 35S::PIF4-YFP, BZR1-CFP	Wang et al.4	N/A
Arabidopsis: JYB1021.2: pS24(AT5G58010)::cS24:GFP(-G):NOS #1	NASC	NASC ID: N70450
Oligonucleotides
siRNA targeting sequence: PIP5K I alpha #1: ACACAGUACUCAGUUGAUA	This paper	N/A
Primers for XX, see Table SX	This paper	N/A
Primer: GFP/YFP/CFP Forward: GCACGACTTCTTCAAGTCCGCCATGCC	This paper	N/A
Morpholino: MO-pax2a GGTCTGCTTTGCAGTGAATATCCAT	Gene Tools	ZFIN: ZDB-MRPHLNO-061106-5
ACTB (hs01060665_g1)	Life Technologies	Cat#4331182
RNA sequence: hnRNPA1_ligand: UAGGGACUUAGGGUUCUCUCUAGGGACUUAGGGUUCUCUCUAGGGA	This paper	N/A
Recombinant DNA
pLVX-Tight-Puro (TetOn)	Clonetech	Cat#632162
Plasmid: GFP-Nito	This paper	N/A
cDNA GH111110	Drosophila Genomics Resource Center	DGRC:5666; FlyBase:FBcl0130415
AAV2/1-hsyn-GCaMP6- WPRE	Chen et al.5	N/A
Mouse raptor: pLKO mouse shRNA 1 raptor	Thoreen et al.6	Addgene Plasmid #21339
Software and algorithms
ImageJ	Schneider et al.7	https://imagej.nih.gov/ij/
Bowtie2	Langmead and Salzberg8	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Samtools	Li et al.9	http://samtools.sourceforge.net/
Weighted Maximal Information Component Analysis v0.9	Rau et al.10	https://github.com/ChristophRau/wMICA
ICS algorithm	This paper; Mendeley Data	http://dx.doi.org/10.17632/5hvpvspw82.1
Other
Sequence data, analyses, and resources related to the ultra-deep sequencing of the AML31 tumor, relapse, and matched normal	This paper	http://aml31.genome.wustl.edu
Resource website for the AML31 publication	This paper	https://github.com/chrisamiller/aml31SuppSite

PHYSICAL SCIENCE TABLE WITH EXAMPLES FOR AUTHOR REFERENCE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
QD605 streptavidin conjugated quantum dot	Thermo Fisher Scientific	Cat#Q10101MP
Platinum black	Sigma-Aldrich	Cat#205915
Sodium formate BioUltra, ≥99.0% (NT)	Sigma-Aldrich	Cat#71359
Chloramphenicol	Sigma-Aldrich	Cat#C0378
Carbon dioxide (13C, 99%) (<2% 18O)	Cambridge Isotope Laboratories	CLM-185-5
Poly(vinylidene fluoride-co-hexafluoropropylene)	Sigma-Aldrich	427179
PTFE Hydrophilic Membrane Filters, 0.22 μm, 90 mm	Scientificfilters.com/Tisch Scientific	SF13842
Critical commercial assays
Folic Acid (FA) ELISA kit	Alpha Diagnostic International	Cat# 0365-0B9
TMT10plex Isobaric Label Reagent Set	Thermo Fisher	A37725
Surface Plasmon Resonance CM5 kit	GE Healthcare	Cat#29104988
NanoBRET Target Engagement K-5 kit	Promega	Cat#N2500
Deposited data
B-RAF RBD (apo) structure	This paper	PDB: 5J17
Structure of compound 5	This paper; Cambridge Crystallographic Data Center	CCDC: 2016466
Code for constraints-based modeling and analysis of autotrophic E. coli	This paper	https://gitlab.com/elad.noor/sloppy/tree/master/rubisco
Software and algorithms
Gaussian09	Frish et al.1	https://gaussian.com
Python version 2.7	Python Software Foundation	https://www.python.org
ChemDraw Professional 18.0	PerkinElmer	https://www.perkinelmer.com/category/chemdraw
Weighted Maximal Information Component Analysis v0.9	Rau et al.2	https://github.com/ChristophRau/wMICA
Other
DASGIP MX4/4 Gas Mixing Module for 4 Vessels with a Mass Flow Controller	Eppendorf	Cat#76DGMX44
Agilent 1200 series HPLC	Agilent Technologies	https://www.agilent.com/en/products/liquid-chromatography
PHI Quantera II XPS	ULVAC-PHI, Inc.	https://www.ulvac-phi.com/en/products/xps/phi-quantera-ii/

Highlights.

• Bone marrow adipocytes (BMAd) undergo rapid senescence after glucocorticoid treatment.

• Glucocorticoids activate prostaglandin-PPARγ-INK signaling to initiate BMAd senescence.

• Senescent BMAds acquire a SASP, causing secondary senescence of surrounding bone.

• Blockage of BMAd senescence inhibits the secondary senescence, rescuing bone loss.

INCLUSION AND DIVERSITY STATEMENT

We worked to ensure sex balance in the selection of non-human subjects. One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in their field of research or within their geographical location. One or more of the authors of this paper self-identifies as a gender minority in their field of research. While citing references scientifically relevant for this work, we also actively worked to promote gender balance in our reference list.