Advanced science (Weinheim, Baden-Wurttemberg, Germany)

DEL-1 is a natural protein that may reduce bone loss linked to aging cells

Updated

Abstract

Essence

DEL-1 functions as a secreted senolytic signal that may limit senescent stromal cells and senescence-associated bone loss.

Evidence

This preclinical cell and mouse study examined DEL-1 in senescent BM-MSCs, DEL-1-deficient and endothelial-overexpression mice, CD73-deficient models, and DEL-1 or macrolide treatment.

Caveat

The evidence is mechanistic and animal-based, with no human clinical test showing DEL-1 prevents senescence-associated bone loss.

Simplified

Full Text

Introduction

Cellular senescence is a complex biological process associated with aging, characterized by irreversible cell‐cycle arrest in response to various stress stimuli, including DNA damage, telomere shortening, oncogenic mutations, metabolic and mitochondrial dysfunction, as well as inflammation.[1, 2] Senescent cells display several distinctive features, such as, an enlarged and flattened cell morphology,[3] enhanced senescence‐associated β‐galactosidase (SA‐β‐gal) activity,[4] increased expression of senescence markers (e.g., the cyclin‐dependent kinase inhibitors p16[5] and p21 and the tumor‐suppressor p53 protein[6]) along with a senescence‐associated secretory phenotype (SASP). SASP is characterized by the release of high levels of pro‐inflammatory cytokines, chemokines, and extracellular matrix‐degrading enzymes, which via paracrine effects can spread senescence in neighboring cells.[7, 8, 9]

Senescent cells can in principle be removed by immune‐mediated clearance via macrophage phagocytosis[10, 11] or by killing via degranulation of natural killer cells.[12, 13] However, senescent cells are generally resistant to macrophage phagocytosis or killing, in part due to elevated expression of the “don't eat me” signal CD47 on their surface.[14] Moreover, with advancing age, the mechanisms for the removal of senescent cells from the tissues become even more ineffective.[12, 15] This aging‐related decline of the immune system's efficiency in removing senescent cells contributes to their accumulation within tissues, ultimately leading to impaired tissue function and age‐related pathologies.[12, 15] Another factor contributing to the persistence of senescent cells involves the activation of specific pathways known as senescent cell anti‐apoptotic pathways (SCAPs).[16] For instance, increased expression of anti‐apoptotic proteins, such as B‐cell lymphoma‐2 (BCL‐2), enables senescent cells to resist apoptosis and thus persist in tissues during aging or chronic inflammation.[16] Senolytic treatment with BCL2‐family inhibitors, including ABT‐263 and ABT‐737, has been proven effective in eliminating senescent cells.[17, 18]

Accumulation of senescent cells in the bone microenvironment has been linked to age‐related bone loss.[19] Mesenchymal stromal cells (MSCs) in the bone marrow stroma are multipotent cells that can differentiate along the osteogenic lineage and facilitate bone regeneration.[20] However, in aging, the function of bone marrow MSCs is remarkably impaired.[20, 21] For instance, the osteogenic differentiation potency of MSCs diminishes over time and, moreover, MSCs can undergo senescence as the bone ages.[22] Senescent MSCs suppress osteogenesis via paracrine effects[23]; conversely, their removal can enhance the osteogenic activity and bone‐forming potential of healthy MSCs, thereby preventing age‐related bone loss.[24, 25, 26, 27]

Developmental endothelial locus‐1 (DEL‐1) is a multi‐functional protein secreted by various stromal cells that comprises three N‐terminal epidermal growth factor (EGF)‐like repeats and two C‐terminal discoidin I‐like domains 3 (hence also known as EDIL3).[28] Its second EGF‐like repeat (E2) contains an Arg‐Gly‐Asp (RGD) motif, which mediates binding to β3 integrins, while its discoidin I‐like domains interact with phosphatidylserine (PS).[29, 30, 31] In addition, DEL‐1 binds β2 integrins, thereby regulating neutrophil recruitment to inflammatory sites.[32, 33] The ability of DEL‐1 to bind both αvβ3 integrin on macrophages and the “eat‐me” signal PS on apoptotic cells enables the interaction of apoptotic cells with macrophages, thereby promoting apoptotic cell phagocytosis (efferocytosis) and resolution of inflammation.[29] DEL‐1 also contributes to the regulation of myelopoiesis in the bone marrow[34] and stimulates osteogenic differentiation[35] via interaction with β3 integrins on hematopoietic stem cells and osteoprogenitors, respectively. By virtue of its anti‐inflammatory and pro‐resolving properties, DEL‐1 provides protection against various inflammation‐associated pathologies, including experimental autoimmune encephalomyelitis as well as inflammatory bone loss in periodontitis and arthritis.[32, 33, 36, 37, 38, 39, 40] The expression of DEL‐1 in stromal cells (e.g., endothelial cells, perivascular stromal cells, and MSCs) helps maintain tissue homeostasis.[33, 36, 38, 40] However, DEL‐1 expression declines severely with aging in both mice and humans and this aging‐related deficiency of DEL‐1 is associated with enhanced inflammation and disease susceptibility.[28, 33]

In this study, we investigated whether DEL‐1 plays a role in the association between MSC senescence and bone loss during aging or after chemotherapy. By using chemotherapy‐induced and aging‐related senescence models along with multiple complementary genetic approaches, we demonstrated that DEL‐1 exhibits senolytic properties on senescent bone‐marrow derived MSCs (BM‐MSCs) in a β3 integrin‐dependent manner, thereby promoting their clearance by macrophages. The DEL‐1‐mediated senolytic activity mainly targeted ecto‐5′‐nucleotidase (CD73)‐expressing PDGFRα+Sca1+(PαS) BM‐MSCs, which exhibited increased expression of β3 integrin compared to other subsets of BM‐MSCs. Whereas DEL‐1 overexpression in endothelial cells markedly reduced senescent BM‐MSC numbers and senescence‐associated bone loss, this activity was abrogated in the genetic absence of CD73. Mechanistically, the senolytic effect of DEL‐1 was dependent on a CD73/Adenosine/p38 mitogen‐activated protein kinase (p38 MAPK)/BCL‐2 pathway. Taken together, we demonstrated for the first time that DEL‐1 is an endogenously expressed senolytic protein in the bone marrow niche that mediates senescent cell clearance and prevents senescence‐associated bone loss.

Results

DEL‐1 Drives Macrophage‐Mediated Clearance of Senescent BM‐MSCs by Acting as a Senolytic Agent

The accumulation of senescent MSCs in aged tissues has been implicated in age‐related bone loss pathologies including osteoarthritis[41] and osteoporosis,[27] whereas clearance of senescent cells is essential for inflammation resolution and tissue regeneration.[24, 27] We previously demonstrated that full‐length DEL‐1 (i.e., containing both the N‐terminal EGF‐like repeats and the C‐terminal discoidin‐like domains) acts as a molecular bridge that promotes macrophage phagocytosis of apoptotic neutrophils, thereby promoting inflammation resolution.[29] This finding prompted us to investigate whether DEL‐1 could also contribute to macrophage clearance of senescent BM‐MSCs. BM‐MSC senescence was induced by exposure to an oxidative stressor, H2O2. The senescent state of the cells was confirmed by real‐time quantitative PCR (qPCR) for p16 and p21 expression (Figure S1A, Supporting Information), immunocytochemistry (SA‐β‐gal staining using X‐gal chromogenic substrate) and by flow cytometric analysis (using the fluorescent β‐galactosidase substrate C12FDG) (Figure1A); the latter analysis showed that BM‐MSCs became fully senescent at day 8 following H2O2 treatment (Figure 1A, right). To explore whether DEL‐1 can promote macrophage phagocytosis of senescent BM‐MSCs, we first conducted an in vitro macrophage phagocytosis assay by co‐culturing bone marrow‐derived macrophages with H2O2‐induced senescent BM‐MSCs pre‐stained with a pH‐sensitive dye, CypHer5E, in the presence of recombinant DEL‐1 (expressed as an Fc‐fusion protein; DEL‐1‐Fc) or Fc control. The percentage of phagocytic (CypHer5E+) macrophages in total CD45+F4/80+ macrophages was significantly increased in the presence of DEL‐1‐Fc as compared to Fc control treatment (Figure 1B). To determine the in vivo relevance of this observation, we performed thioglycolate (TG)‐induced peritonitis, which is conducive for investigating quantitative aspects of macrophage phagocytosis.[42] To this end, we isolated BM‐MSCs from human ubiquitin C promoter‐green fluorescent protein (UBC‐GFP) mice and induced senescence by H2O2, as confirmed by significantly elevated p16 and p21 expression (Figure S1B, Supporting Information). At 72 h after TG injection, cultured GFP‐expressing CypHer5E+ senescent BM‐MSCs were administered together with DEL‐1‐Fc or Fc control into naive wild‐type (WT) mice and peritoneal macrophage phagocytic activity was assessed 3 h later by flow cytometry (Figure 1C). Mice that received senescent GFP+BM‐MSCs with DEL‐1‐Fc displayed a higher percentage of CypHer5E+ peritoneal macrophages than those that received senescent BM‐MSCs with Fc control (Figure 1C, middle), along with a concomitant reduction in the number of remaining senescent GFP+BM‐MSCs (Figure 1C, right), suggesting that DEL‐1 promotes macrophage phagocytosis of senescent BM‐MSCs. To better understand the mechanisms underlying DEL‐1‐mediated macrophage phagocytosis of senescent BM‐MSCs, we set out to examine whether DEL‐1 modulates the expression of surface molecules on senescent BM‐MSCs that could potentially affect their uptake by macrophages. In this regard, the “don't eat me” signal CD47 appears to increase on senescent cells, which can thereby suppress their clearance by macrophages.[14] However, DEL‐1‐Fc treatment failed to alter CD47 expression on senescent BM‐MSCs as compared to treatment with Fc control (Figure S2, Supporting Information). We next determined the presence of the “eat me signal” PS on DEL‐1‐Fc‐treated and Fc control‐treated senescent BM‐MSCs by annexin V staining. Flow cytometric analysis showed that DEL‐1‐Fc significantly increased the percentage of annexin V+ senescent BM‐MSCs, relative to Fc control (Figure 1D). In stark contrast, DEL‐1‐Fc had no effect on PS surface abundance when a similar annexin V flow cytometric analysis was performed using healthy, non‐treated BM‐MSCs (Figure 1E). Given that annexin V staining is indicative of apoptosis, these data suggest that DEL‐1 promotes apoptosis selectively on senescent BM‐MSCs and thus DEL‐1 could have senolytic properties. The senolytic effect of DEL‐1 on senescent BM‐MSCs, as assessed by induction of apoptotic cell death monitored by increased annexin V staining, was reproducible when BM‐MSC senescence was induced by additional stimuli, such as, palbociclib (cyclin D kinase inhibitor) or doxorubicin (genotoxic agent) (Figure S3A,B, Supporting Information, respectively).

Given that DEL‐1 promotes apoptosis of senescent BM‐MSCs, we postulated that the in vivo ability of DEL‐1 to enhance BM‐MSC uptake by macrophages in the TG‐induced peritonitis model (Figure 1C) may be attributed to its senolytic effect. To address this question, we first determined whether DEL‐1 can induce apoptosis of senescent BM‐MSCs in vivo. Mice were i.p. injected with GFP‐expressing senescent BM‐MSCs together with DEL‐1‐Fc or Fc control. At 3 h post‐injection, the peritoneal contents were aspirated and apoptosis of senescent GFP‐BM‐MSCs was determined by flow cytometric analysis. The presence of DEL‐1‐Fc caused a significant increase in the percentage of annexin V+ cells and a concomitant significant decrease in the percentage of annexin V–/7‐AAD– live cells as compared to Fc control (Figure 1F), suggesting that DEL‐1 induces apoptosis of senescent BM‐MSCs in vivo. Moreover, in vitro, DEL‐1‐Fc decreased the abundance of senescent BM‐MSCs with larger size (Figure S4, Supporting Information), which may in turn facilitate their phagocytosis by macrophages given that the relatively large size of senescent cells constitutes a natural obstacle to phagocytosis.[14, 43]

We next explored whether the ability of DEL‐1 to induce apoptosis of senescent BM‐MSCs is sufficient to promote their clearance by macrophages. To this end, we incubated senescent BM‐MSCs with DEL‐1‐Fc or Fc control for 3 days, and upon washing, the cells were stained with CypHer5E and injected into mice that were subjected to TG‐induced peritonitis 72 h earlier. At 3 h post‐injection (75‐h time point), the peritoneal contents were aspirated and macrophages that phagocytosed senescent BM‐MSCs were identified by flow cytometry as live CD45+F4/80+CypHer5E+ cells. We observed that the percentage of phagocytic macrophages (CypHer5E+ cells in live CD45+F4/80+ cells) was increased when the senescent BM‐MSCs were pre‐treated, ex vivo, with DEL‐1‐Fc as compared to pretreatment with Fc control (Figure 1G). A similar experiment was performed in which full‐length DEL‐1‐Fc was replaced by a truncated version of the molecule that contains only the N‐terminal EGF‐like repeats (designated DEL‐1[E1‐E3]‐Fc) and yielded similar results (Figure 1H). Moreover, DEL‐1[E1‐E3]‐Fc significantly increased the percentage of annexin V+ senescent BM‐MSCs (relative to Fc control), comparably to full‐length DEL‐1‐Fc (Figure 1I), indicating that the N‐terminal EGF‐like repeats region of DEL‐1 is sufficient to promote the apoptosis of senescent BM‐MSCs and the subsequent clearance thereof by macrophages. Taken together, our findings thus far show conclusively that DEL‐1 promotes macrophage clearance of senescent BM‐MSCs through its senolytic effect.

DEL‐1 promotes macrophage phagocytosis of senescent BM‐MSCs through its senolytic effect. A–I) BM‐MSCs were induced to become senescent in the presence of HOfor 2 h and further cultured for 8 days. (A) HO‐induced senescence of BM‐MSCs was confirmed by SA‐β‐gal staining (left) and by CFDG staining (right), assessed by microscopy and flow cytometry, respectively. Shown are the representative images (left) and the corresponding FACS histogram (right). Scale bar, 500 µm. (B) CypHer5E‐stained senescent BM‐MSCs were co‐cultured with bone marrow‐derived macrophages, and macrophage phagocytosis was allowed for 2 h in the presence of DEL‐1‐Fc (2 µg mL) or an equal molar concentration of Fc control. The percentage of phagocytic macrophages (CypHer5Ecells in CD45F4/80macrophages) was determined by flow cytometry. (C) Phagocytosis of senescent BM‐MSCs in a thioglycollate (TG)‐induced peritonitis model. 9‐week‐old mice were injected intraperitoneally with 2% w/v TG, and after 72 h, the mice received CypHer5E‐stained GFP‐expressing senescent BM‐MSCs together with 10 µg of DEL‐1‐Fc or an equal molar amount of Fc control. Three hours later, the mice were sacrificed, and the percentage of phagocytic macrophages (CypHer5Ecells in live CD45F4/80macrophages), and the number of remaining GFPBM‐MSCs in the peritoneal exudates was determined by flow cytometric analysis. Cell numbers were normalized, and the average cell number of the Fc control group was set to 1. (D,E) Senescent BM‐MSCs (D) or healthy BM‐MSCs (E) were cultured in the presence of DEL‐1‐Fc (10 µg mL) or an equal molar concentration of Fc control for 3 days. Apoptotic cells positive for Annexin V‐APC staining in the total single‐cell population were assessed by flow cytometric analysis. (F) Senescent GFP‐BM‐MSCs together with 10 µg of DEL‐1‐Fc or an equal molar amount of Fc control were intraperitoneally injected into 8‐week‐old healthy mice. 3 h later, peritoneal cells were collected, and apoptosis of injected senescent GFP‐BM‐MSCs in the peritoneal exudates was analyzed by flow cytometry. Shown are representative FACS plots (left) and quantitative bar graphs (middle and right) of the percentage of apoptotic cells (Annexin V, middle) and live cells (Annexin V7‐AAD, right) in single GFPcells. (G,H) Senescent BM‐MSCs were incubated in the presence of DEL‐1‐Fc (10 µg mL) or an equal molar concentration of DEL‐1‐[E1‐E3]‐Fc or Fc control for 3 days (G, Fc vs DEL‐1‐Fc; H, Fc vs DEL‐1‐[E1‐E3]‐Fc]). The cells were then washed and pre‐stained with the fluorescent dye CypHer5E. CypHer5E‐labeled senescent BM‐MSCs were injected into WT mice 72 h after inducing peritonitis with 2% TG. Mice were killed 3 h later, and the percentage of phagocytic macrophages (CypHer5Ecells in live CD45F4/80macrophages) in the peritoneal exudates was determined by flow cytometric analysis. (I) Senescent BM‐MSCs were cultured for 3 days in the presence of DEL‐1‐Fc (10 µg mL), equal molar concentration of DEL‐1[E1‐E3]‐Fc or Fc control. Apoptotic cells positive for Annexin V‐APC staining in the single‐cell population were assessed by flow cytometry. Shown are representative FACS plots (left) and quantitative bar graphs (right). Data are means ± SD (B:=6‐7 mice/group; C:=6 mice/group; D,E,I:=4 cell cultures/group; F: 7‐8 mice/group; G:=3‐4 mice/group; and H:=5 mice/group). Data in B and F are pooled from 2 experiments.0.05;0.01;0.0001; NS, not significant. Unpaired Student'stest, except for C (Mann Whitney U test) and I (One‐way ANOVA and Dunnett's multiple‐comparisons test). Macs, macrophages. 2 2 2 2 12 −1 + + + + + + + −1 + – – + −1 + + + −1 * ** **** n n n n n p < p < p < t

The Senolytic Activity of DEL‐1 on Senescent BM‐MSCs is Mediated via Interaction with β3 Integrin and a CD73/Adenosine/p38/BCL‐2 Pathway

The second EGF‐like domain of DEL‐1 contains an integrin‐binding RGD motif and many of the known biological functions of DEL‐1 involve interactions with β3 integrin.[29, 34, 44] We thus examined whether the senolytic activity of DEL‐1 is dependent on interaction with β3 integrin expressed by BM‐MSCs, as a first approach to start understanding the underlying mechanism. To this end, we determined whether DEL‐1[RGE]‐Fc, a point‐mutant of DEL‐1 that does not bind β3 integrin due to Glu‐for‐Asp substitution in the RGD motif of DEL‐1,[34, 35, 37] can exert senolytic actions and induce apoptosis in senescent BM‐MSCs. By assessing the activity of caspase‐3/7 as a marker of apoptosis, we showed that DEL‐1[RGE]‐Fc failed to significantly increase caspase‐3/7 activity (relative to Fc control), in contrast to WT DEL‐1‐Fc, which significantly elevated caspase‐3/7 activity (Figure S5A,B, Supporting Information). Moreover, whereas WT DEL‐1‐Fc promoted macrophage phagocytosis of senescent BM‐MSCs in vitro, DEL‐1[RGE]‐Fc failed to promote this function (relative to Fc control) (Figure S5C, Supporting Information). These findings suggest that the integrin‐binding RGD motif of DEL‐1 is required for its senolytic activity. We next investigated potential DEL‐1 receptors on senescent BM‐MSCs involved in apoptosis induction. Given that DEL‐1 can bind the RGD‐recognizing β1 and β3 integrins, which are expressed on MSCs,[35, 45] we first tested the ability of DEL‐1 to induce apoptosis in senescent BM‐MSCs that lack β3 integrin. To this end, BM‐MSCs were isolated from mice with MSC‐specific deletion of the β3 integrin gene (Prx1Cre+β3flox/flox mice). Specific deletion of the β3 integrin‐encoding gene Itgb3 in sorted bone marrow stromal cells from Prx1Cre+β3flox/flox mice was evidenced by a remarkable reduction in Itgb3 integrin mRNA expression by qPCR (Figure S6, Supporting Information). Although DEL‐1‐Fc induced apoptosis in β3 integrin‐expressing senescent BM‐MSCs, as assessed by annexin V staining and flow cytometric analysis, this effect of DEL‐1‐Fc was abrogated in β3 integrin‐deficient senescent BM‐MSCs (Figure2A). In contrast, DEL‐1‐Fc retained its ability to induce apoptosis in β3 integrin‐expressing senescent BM‐MSCs that were treated with a neutralizing antibody to β1 integrin (Figure S7, Supporting Information). Taken together, β3 integrin is required for the senolytic effect of DEL‐1 on senescent BM‐MSCs.

Since β3 integrin mediates DEL‐1‐induced apoptosis and clearance of senescent BM‐MSCs, we assessed β3 integrin expression in various subsets of MSCs in the mouse bone marrow. Flow cytometry revealed that CD73+ MSCs[46] exhibited the highest expression of β3 integrin compared to other subsets of MSCs in the bone marrow (Figure S8, Supporting Information), including PDGFRα+Sca1+(PαS),[47, 48] Sca1+CD29+,[49] and LepR+ MSCs.[50] Based on these data and the fact that CD73 (Ecto‐5′‐nucleotidase) regulates the conversion of adenosine monophosphate to adenosine that can induce apoptosis (including in senescent cells) by binding to adenosine receptors,[51, 52, 53] we focused on the expression and function of CD73 on BM‐MSCs for further mechanistic studies.

First, we showed that DEL‐1‐Fc, but not DEL‐1[RGE]‐Fc, significantly increased CD73 expression on senescent BM‐MSCs (Figure 2B) and, consistently, augmented adenosine concentrations in the culture supernatants of senescent BM‐MSCs (Figure 2C), as compared to Fc control. To determine whether the senolytic effect of DEL‐1 on senescent BM‐MSCs is dependent on CD73 activity, we used α, β methylene adenosine 5′‐diphosphate (APCP), a specific CD73 inhibitor. Treatment with APCP abrogated DEL‐1‐induced apoptosis of senescent BM‐MSCs (Figure 2D; Figure S9A, Supporting Information) by restraining adenosine production (Figure 2E). Furthermore, a pan‐adenosine receptor antagonist blocked DEL‐1‐induced apoptosis (Figure 2F; Figure S9B, Supporting Information), suggesting that the senolytic effect of DEL‐1 on senescent BM‐MSCs is mediated via the CD73/adenosine/adenosine receptor (AR) pathway. Adenosine was previously shown to promote apoptosis of endothelial cells by inhibiting p38 MAPK activity.[54] Consistently, p38 MAPK inhibition promotes cisplatin‐induced apoptosis of cancer cells[55] and the senolytic activity of the SYK inhibitor R406 depends on its ability to inhibit p38 MAPK activity.[56] To explore whether DEL‐1‐induced apoptosis of senescent BM‐MSC is associated with inhibition of p38 activation, we examined p38 phosphorylation (p‐p38) in senescent BM‐MSCs treated with DEL‐1‐Fc in vitro. DEL‐1‐Fc significantly downregulated the intensity of p‐p38 as compared to Fc control treatment, as shown by flow cytometry (Figure 2G) and confirmed by western blot analysis (Figure 2H). Treatment of senescent BM‐MSCs with a pan‐adenosine receptor antagonist abrogated the DEL‐1‐Fc‐induced reduction in p‐p38 (Figure 2I). Moreover, inhibition of p38 activity in senescent BM‐MSCs using the p38 antagonist SB203580 resulted in dose‐dependent induction of apoptosis (Figure 2J), thus mimicking the pro‐apoptotic effect of DEL‐1‐Fc (Figure 1D). BCL‐2 is an anti‐apoptotic molecule and one of the hallmarks of cellular senescence. Targeting the elevated BCL‐2 level in senescent cells with senolytics, such as ABT‐263 and ABT‐ 737, is a potential strategy for eliminating senescent cells.[18, 26] To determine if BCL‐2 is a downstream target of p38 inhibition‐mediated apoptosis in senescent BM‐MSCs, we assessed its expression following DEL‐1 treatment. Intriguingly, we observed that DEL‐1 significantly decreased the expression of BCL‐2 in senescent BM‐MSCs, at both the mRNA and protein level, as assessed by quantitative PCR and flow cytometry, respectively (Figure 2K,L). Furthermore, the expression of other anti‐apoptotic BCL‐2 family members, B‐cell lymphoma‐extra large (BCL‐XL) and myeloid cell leukemia‐1 (MCL‐1), was also significantly diminished by DEL‐1 treatment (Figure S10, Supporting Information). Subsequently, flow cytometric analysis showed that BCL‐2 expression was inhibited by the p38 inhibitor, SB203580 in a dose‐dependent manner (Figure 2M), indicating that BCL‐2 is a downstream target of p38. Importantly, a pan‐adenosine receptor antagonist abrogated the DEL‐1‐mediated inhibition of BCL‐2 expression (Figure 2N). Taken together, our findings suggest that DEL‐1 binds β3 integrin on senescent BM‐MSCs and promotes their apoptosis via a CD73/adenosine receptor/p38 MAPK/BCL‐2 pathway (Figure S11, Supporting Information).

The senolytic effect of DEL‐1 is mediated through a β3 integrin/CD73/adenosine/p38/BCL‐2 pathway. A) Senescent BM‐MSCs from 7‐week‐oldandmice were incubated in the presence of DEL‐1‐Fc (10 µg mL) or an equal molar concentration of Fc control for 3 days. Representative FACS plot (left) and a quantitative bar graph (right) of the percentage of Annexin Vcells in senescent BM‐MSCs. B,C) Senescent BM‐MSCs were incubated for 3 days with either DEL‐1‐Fc (10 µg mL), DEL‐1[RGE]‐Fc (10 µg mL) or an equal molar concentration of Fc control. MFI of CD73 expression in senescent BM‐MSCs was assessed by flow cytometry (after normalization, the MFI of the Fc control group was set as 1) (B) and adenosine production was measured in the culture supernatants (C). D,E) Senescent BM‐MSCs were pre‐treated with a CD73 inhibitor (APCP, α,β‐methylene ADP; 10 µ) overnight. The cells were then further incubated with DEL‐1‐Fc (10 µg mL) or an equal molar concentration of Fc control for 3 days. Shown is the percentage of apoptotic cells positive for Annexin V‐APC staining in the total single‐cell population (D). Adenosine production was measured in the culture supernatants (E). F) Senescent BM‐MSCs were pre‐treated with adenosine receptor (AR) antagonist (; 1 µ) or vehicle control (0.004% DMSO, Ctrl) for 30 min and then incubated for 3 days in the presence of DEL‐1‐Fc (10 µg mL) or equal molar concentration of Fc control. Shown is the percentage of apoptotic cells positive for Annexin V‐APC staining in the total single‐cell population. G) Senescent BM‐MSCs were cultured in the presence of DEL‐1‐Fc (10 µg mL) or an equal molecular concentration of Fc control for 3 days. Representative FACS histogram (left) and MFI of phosphorylated p38 (p‐p38) MAPK expression in senescent BM‐MSCs (right). H) Representative immunoblot images of phosphorylated and total p38 in senescent BM‐MSCs treated with DEL‐1‐Fc (10 µg mL) or an equal molar concentration of Fc control for 3 days. I) Senescent BM‐MSCs were pre‐treated with adenosine receptor (AR) antagonist (; 1 µ) or vehicle control (0.004% DMSO) for 30 mins, followed by 3‐day incubation with DEL‐1‐Fc (10 µg mL) or equal molar concentration of Fc control, with the antagonist present in the culture during the entire incubation. Shown are the representative FACS histogram (left) and the MFI of phosphorylated p38 (p‐p38) MAPK expression in senescent BM‐MSCs. J) Senescent BM‐MSCs were incubated in the presence of vehicle (0.02% DMSO) or p38 inhibitor (SB203058; 0.5, 1, or 2 µ) for 3 days. Cells were harvested and analyzed for Annexin Vapoptosis by flow cytometry. Shown is the percentage of Annexin Vcells in senescent BM‐MSCs. K) Senescent BM‐MSCs were cultured in the presence of DEL‐1‐Fc (10 µg mL) or an equal molar concentration of Fc control for 3 days. Shown is the quantification ofexpression in the mRNA level.was used for normalization of mRNA expression and the gene expression of the Fc control was assigned an average value of 1. L) Healthy or senescent BM‐MSCs were cultured with DEL‐1‐Fc (10 µg mL) or an equimolar concentration of Fc control for 3 days. Representative flow cytometry histograms (left) and mean fluorescence intensity (MFI) of BCL‐2 expression in senescent BM‐MSCs (right) are shown. M) MFI of BCL‐2 expression in senescent BM‐MSCs treated with p38 inhibitor (SB203580; 0, 0.5, 1 µ) for 3 days. N) Senescent BM‐MSCs were pre‐treated with adenosine receptor (AR) antagonist (; 1 µ) or vehicle control (0.004% DMSO) for 30 min, followed by 3‐day incubation with DEL‐1‐Fc (10 µg mL) or equal molar concentration of Fc control. Representative flow cytometry histograms (left) and MFI of BCL‐2 expression in senescent BM‐MSCs (right) are shown. Data are means ± SD (A,F,G,I,J,L,M,N:= 5 cell cultures/group; B:= 8 cell cultures/group, pooled from 2 experiments; C,D,E,K:=4 cultures/group).0.05;0.01;0.001;<0.0001; NS, not significant. B,C,F: One‐way ANOVA with Tukey's multiple comparison test; A,D,E,G,I,K,L,N: Unpaired Student'stest; J,M: One‐way ANOVA with Dunnett's multiple comparisons test. Prx1 β3 Prx1 β3 Bcl‐2 Gapdh n n n p < p < p < p t Cre+ flox/flox Cre– flox/flox −1 + −1 −1 −1 −1 −1 −1 −1 + + −1 −1 −1 * ** *** **** m m m m m m CGS15943 CGS15943 CGS15943

DEL‐1 Deficiency Aggravates Senescence‐Associated Bone Loss by Reducing Apoptosis and Increasing the Numbers of β3 Integrin‐Expressing CD73Senescent MSCs +

To investigate whether DEL‐1 exerts senolytic action also in vivo and understand the underlying mechanisms, we employed a doxorubicin‐induced senescence and bone loss model[24] in DEL‐1‐deficient (Del1KO) mice and WT littermate controls. Doxorubicin, a potent inducer of DNA damage,[57, 58] triggers cellular senescence (including in stromal cells) and the accompanying SASP, thereby leading to bone loss that is counteracted by senescent cell clearance.[24]Del1KO and WT mice were administered a biologically relevant dose of doxorubicin (5 mg kg−1, body weight)[59] once a week for 4 consecutive weeks (Figure3A). This allowed us to examine cellular senescence and associated gene expression in the bone marrow as well as bone mass. Microcomputed tomography (µCT) imaging revealed significant femoral bone loss in both the cortical and trabecular compartments in Del1KO mice as compared to WT controls. We observed that the cortical thickness (Ct.Th) in the mid‐diaphysis in De1KO mice was significantly lower than in WT mice (Figure 3B). In the trabecular compartment, Del1KO mice displayed decreased trabecular volume fraction (Tb.BV/TV), trabecular number (Tb.N), thickness (Tb.Th), but increased separation (Tb.Sp), as compared to WT control (Figure 3C). Given our previously published observations that Del1KO mice and WT littermate controls have comparable bone densities (in the femur and spine) under steady‐state conditions (i.e., without inflammatory or other challenge),[33] the exacerbated bone loss observed in doxorubicin‐treated Del1KO mice (relative to WT littermates) is likely attributed exclusively to the doxorubicin challenge.

Next, we determined whether the increased bone loss in DEL‐1 deficiency results from reduced bone formation and/or enhanced bone resorption. To this end, we used ELISA to measure the serum levels of osteocalcin, as a marker of bone formation, as well as the serum levels collagen type I cross‐linked C‐telopeptide (CTX1) and tartrate‐resistant acid phosphatase 5b (TRAcP 5b), both being markers of bone resorption. In comparison to WT controls, Del1KO mice exhibited reduced osteocalcin concentrations (Figure 3D) and increased concentrations of CTX1 and TRAcP 5b (Figure 3E), consistent with osteocalcin and TRAP staining, respectively, as shown in Figure 3F.

Moreover, qPCR analysis indicated a significant increase in the expression of the senescence‐related genes p21 and p53 (but not of p16 in the bone marrow of Del‐1KO mice as compared to WT mice (Figure 3G). Among SASP genes investigated in the same assay, Cxcl2 was significantly upregulated in Del1KO mice, whereas a trend for increased expression of Il6 and Tgfβ1 did not reach statistical significance (Figure 3G). This increase in SASP was further confirmed at the protein level, as TGFβ1 and CXCL2 concentrations were significantly elevated in the bone marrow fluid of Del‐1KO mice (Figure 3H). Moreover, histological analysis revealed an abundance of p21+ cells in the bone marrow compartment of Del1KO mice (Figure 3I). These findings underscore the ability of endogenous DEL‐1 to prevent senescence‐associated bone loss while downregulating the expression of key senescent markers in the bone marrow.

Given that CD45–CD31– stromal cells are the primary targets of doxorubicin‐induced senescence,[24] we examined various subsets of MSC populations, including PDGFRα+Sca1+CD45–CD31–Ter119– (PαS), Sca1+CD29+CD45–CD31–Ter119–, and CD73+CD45–CD31–Ter119– cells in the bone marrow post‐doxorubicin treatment. Whereas no difference was observed between WT and Del1KO mice with regard to the numbers of PαS and Sca1+CD29+ MSCs (Figure S12, Supporting Information), we observed a substantial increase in the number of CD73+ stromal cells in the bone marrow of Del1KO mice, as compared to WT littermates, at day 7 following a single dose of doxorubicin (10 mg kg−1) (Figure4A) and at day 28 following doxorubicin treatment (5 mg kg−1) once weekly for 4 weeks (Figure 4A, right). This increase in the numbers of CD73+ stromal cells in DEL‐1 deficiency was not observed in the absence of doxorubicin treatment (data not shown). Further characterization of CD73+ bone marrow stromal cells revealed that this subset had a higher frequency of γH2AX expression (marker of DNA damage[60]) in Del1KO than in WT mice (Figure 4B), along with reduced percentage of Ki67‐positive (proliferative) cells (Figure 4C), indicating a heightened senescent phenotype of these cells under the condition of in vivo DEL‐1 deficiency. These findings suggest that DEL‐1 deficiency promotes senescence selectively in the CD73+ MSCs population and leads to their accumulation in the bone marrow after doxorubicin administration. Consistently, live CD73+CD45–CD31–Ter119– MSCs from Del1KO mice displayed increased expression of senescence‐related genes (p21, Il6, Cxcl2) (Figure 4D) and decreased annexin V‐determined apoptosis (Figure 4E), as compared to WT controls at day 7 following doxorubicin treatment. At the same time point, moreover, the bone marrow fluid of Del1KO mice contained decreased concentration of adenosine (relative to WT controls; Figure 4F), a key component of the DEL‐1‐induced pro‐apoptotic pathway in senescent BM‐MSCs (Figure 2; Figure S11, Supporting Information). Consistent with the importance of β3 integrin in mediating DEL‐1‐induced apoptosis that proceeds via a CD73/adenosine pathway (Figure 2; Figure S11, Supporting Information), we showed that β3 integrin expression was significantly higher in CD73+PαS than in CD73–PαS MSCs (Figure S13, Supporting Information). Consistently moreover, DEL‐1 deficiency resulted in increased frequency of γH2AX+ cells selectively within the CD73+PαS population (Figure 4G), since γH2AX expression in CD73–PαS cells was not affected by the presence or absence of DEL‐1 (Figure 4H), indicating an increased senescent phenotype in the β3 integrin‐rich CD73+PαS population. This was confirmed by subsequent analyses demonstrating an increased frequency of p21+ cells in the CD73+PαS population (Figure 4I), but not in the CD73–PαS population (Figure 4J) in DEL‐1‐deficient mice.

These in vivo data are consistent with the in vitro dissected mechanism of the DEL‐1 senolytic action and collectively suggest that β3 integrin‐expressing CD73+ senescent MSCs accumulate in the DEL‐1‐deficient bone marrow. At least in part, this is attributed to reduced apoptosis and hence impaired senolytic clearance of CD73+ senescent MSCs in DEL‐1 deficiency, which would otherwise occur in the presence of DEL‐1.

DEL‐1 deficiency exacerbates chemotherapy‐induced bone loss. Doxorubicin was administered to 9‐week‐old WT andmice at a dose of 5 mg kgper week via intraperitoneal injection for 4 weeks. After 4 weeks, femurs from both WT andmice were extracted and analyzed using microcomputed tomography to assess bone loss. A) Schematic representation of the experimental timeline for doxorubicin administration. B) Representative µCT images of bone microarchitecture at the sagittal plane (left) and the mid‐diaphysis of the femur (middle). Quantification of µCT‐derived cortical thickness (Ct.Th(mm), right). C) Representative µCT images of bone microarchitecture at the distal femoral metaphysis (left). Quantification of µCT–derived trabecular bone parameters, including bone volume fraction (Tb.BV/TV(%)), trabecular number (Tb.N(mm)), trabecular thickness (Tb.Th(mm)), and trabecular space (Tb.Sp(mm)) (right). D,E) Fasting serum levels of bone formation marker osteocalcin (D), bone resorption markers collagen type I cross‐linked C‐telopeptide (CTX1) and tartrate‐resistant acid phosphatase 5b (TRAcP 5b) (E). F) Histology, immunostaining, and TRAP staining of femurs. Representative images of H&E staining (scale bar, 500 µm), TRAP‐positive osteoclasts (purple; scale bar, 100 µm), and osteocalcin (OCN)‐positive osteoblasts (green; scale bar, 100 µm) in a boxed area on the H&E staining panels. The nuclei were counterstained using DAPI (blue). G) Expression of senescence‐related genes,, in the bone marrow cells assessed by qPCR.was used for normalization of mRNA expression and the gene expression of WT mice was assigned an average value of 1. H) Levels of IL‐6, TGFβ1, and CXCL2 in bone marrow fluid, measured by ELISA. I) Representative fluorescence images of p21⁺ cells (purple) in the bone marrow compartment of femoral bone sections. The nuclei were counterstained with DAPI (blue). Scale bar, 100 µm. Data are means ± SD (B,C:=4‐5 mice/group; D,E:=5 mice/group; G:=3–4 mice/group; H:=6–7 mice/group).0.05;0.01;<0.001; NS, not significant. Unpaired Student's‐test, except for “CTX1” in panel E (Mann‐Whitneytest). Del1 Del1 p21, p53, Il6, Cxcl2, Tgfβ1, and p16 Gapdh n n n n p < p < p t U KO −1 KO −1 * ** ***

DEL‐1 deficiency increases the abundance of senescent CD73MSCs in the bone marrow of doxorubicin‐treated mice. Doxorubicin (10 mg kggiven once for 7 days or 5 mg kgper week for 28 days) was administered to 8‐week‐old WT andmice, and bone marrow cells were analyzed by flow cytometry at day 7 (single 10 mg kgdose) or day 28 (5 mg kgper week). A) Representative FACS plot (left, day 7) and quantitative bar graphs (right) of the number of CD73CD45CD31Ter119MSCs at day 7 and at day 28. B,C) Representative FACS plots (left) and quantitative bar graphs (right) of (B) the percentage of γHAXcells in CD73CD45CD31Ter119MSCs or (C) the percentage of Ki67cells in CD73CD45CD31Ter119MSCs at day 7. D) CD73CD45CD31TER119MSCs in the bone marrow were sorted and analyzed for senescence‐related genes by qPCR at day 7 after doxorubicin treatment. Senescence‐related gene expression () determined by qPCR.was used for normalization of mRNA expression and the gene expression of WT mice was assigned an average value of 1. E) Representative FACS plots (left) and quantitative bar graphs (right) of the percentage of Annexincells in CD73CD45CD31Ter119MSCs at day 7. F) Adenosine level in the bone marrow fluid of doxorubicin‐treated WT ormice at day 7. G,H) Representative FACS plots (left) and quantitative bar graphs (right) of the percentage of γHAXcells in (G) CD73PαS (PDGFRαSca1CD45CD31Ter119), or (H) CD73PαS (PDGFRαSca1CD45CD31Ter119) at day 7. Representative FACS plots (left) and quantitative bar graphs (right) of the percentage of p21cells in I) CD73PαS (PDGFRαSca1CD45CD31Ter119), or J) CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs at day 7. Data are means ± SD (A:= 4 mice/group; B:= 5 mice/group; C,G,H:=3 mice/group; D:=4‐5 mice/group; E:=9‐10 mice/group; and F:= 8–9 mice/group; I,J:=6 mice/group). Data in panels E and F are pooled from 2 experiments.0.05;0.01;0.0001; NS, not significant. Unpaired Student's‐test, except for the “D7” graph in panel A, andin panel D (Mann‐Whitney U test). + −1 −1 KO −1 −1 + – – – + + – – – + + – – – + – – – + + – – – KO + + + + – – – – + + – – – + + + + – – – – + + – – – * ** **** Del1 p21, Il6, and Cxcl2 Gapdh Del1 n n n n n n n p < p < p < t Il6 2 2

Endothelial DEL‐1 Overexpression Inhibits Senescence‐Associated Bone Loss In Vivo and Reduces Senescent Cell Numbers in a CD73‐Dependent Manner

We next determined the cellular source of DEL‐1 responsible for its senolytic effect on senescent MSCs in the bone marrow. Based on the bone marrow expression profile of DEL‐1,[34] we reasoned that DEL‐1 could be derived from endothelial cells and/or from MSCs. By crossing Del1flox/flox with VE‐CadCre+ or Prx1Cre+ mice, we generated mice with specific deletion of DEL‐1 in vascular endothelial cells or in MSCs, respectively. Specific deletion of the DEL‐1‐encoding Edil3 gene in sorted bone marrow stromal cells from Prx1Cre+Del1flox/flox mice (Del1MSCKO) and in sorted lung endothelial cells from VE‐CadCre+Del1flox/flox mice (Del1ECKO) was suggested by the strongly reduced Edil3 mRNA expression by qPCR (Figure S14, Supporting Information). We first administered doxorubicin to both VE‐CadCre+Del1flox/flox mice (Del1ECKO) and VE‐CadCre–Del1flox/flox controls (Del1ECWT) and examined the number of senescent CD73+PαS MSCs in the bone marrow at day 7. Strikingly, the absence of DEL‐1 from endothelial cells resulted in a significant increase in the number of senescent MSCs (p21+CD73+PαS or C12FDG+CD73+PαS) in the bone marrow (Figure5A,B). In contrast, no discernible difference was observed in the number of senescent CD73+PαS MSCs in the bone marrow of Prx1Cre+Del1flox/flox (Del1MSCKO) and Prx1Cre–Del1flox/flox (Del1MSCWT) mice (Figure S15, Supporting Information). These results suggest that endothelial cells are a major cellular source of DEL‐1 responsible for its senolytic activity on senescent MSCs in the bone marrow of doxorubicin‐treated mice, implying that secreted DEL‐1 acts in a paracrine manner (Figure S11, Supporting Information).

To further corroborate the importance of endothelial DEL‐1 in promoting apoptosis of senescent MSCs in the doxorubicin‐induced bone loss model, we employed mice with endothelial cell‐specific overexpression of DEL‐1 (EC‐Del1 mice).[29, 38, 61] Microcomputed tomography analysis revealed a significant increase in trabecular volume fraction (Tb.BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th) and a decrease in trabecular separation (Tb.Sp) in EC‐Del1 mice compared to WT mice (Figure 5D), whereas there was no significant difference between the two groups with regard to cortical thickness (Ct.Th) (Figure 5C). This bone‐protective effect was associated with a significant reduction in the number of senescent p21+CD73+PαS or C12FDG+CD73+PαS MSCs in the bone marrow of EC‐Del1 mice as compared to WT littermates (Figure 5E,F). In contrast, no difference was observed in the numbers of CD73‐negative senescent MSCs (p21+CD73–PαS cells or C12FDG+CD73–PαS cells) between WT and EC‐Del1 mice (Figure 5G; Figure S16, Supporting Information), substantiating our earlier findings that DEL‐1 selectively targets CD73‐positive senescent MSCs (Figure 5E–G; Figure S16, Supporting Information). Importantly, the percentage of apoptotic (caspase‐3‐positive) cells in senescent p21+CD73+PαS MSCs was significantly enhanced in the bone marrow of EC‐Del1 mice as compared to WT littermates (Figure 5H). In contrast, no such difference in the frequency of caspase‐3‐positive cells in p21–CD73+PαS MSCs cells was observed (Figure 5I), indicating that DEL‐1 selectively promotes apoptosis in senescent MSCs, as we also showed in vitro (Figure 1D,E).

Whereas CD73 is important for the senolytic activity of DEL‐1 in vitro (Figure 2), such a requirement has not been addressed yet in vivo. To determine the role of CD73 in the senolytic effect mediated by DEL‐1 on senescent MSCs in the doxorubicin‐induced bone loss model, we employed CD73‐deficient (Cd73KO) mice and mice with endothelial‐specific overexpression of DEL‐1 in the CD73‐deficient background (Cd73KO/EC‐Del1 mice), which we generated by crossing Cd73KO mice and EC‐Del1 mice. As expected from our earlier experiment (Figure 5E,F), 7 days after doxorubicin administration, there was a significant reduction in the number of senescent MSCs (p21+PαS or C12FDG+PαS) in the bone marrow of EC‐Del1 mice, as compared to their WT littermates (Figure 5J,K). In contrast, in the CD73‐deficient background, the endothelial cell‐specific overexpression of DEL‐1 had no effect on the number of senescent PαS cells (p21+PαS or C12FDG+PαS), since the numbers of senescent PαS cells were comparable between the bone marrow of Cd73KO and Cd73KO/EC‐Del1 mice (Figure 5L,M). Furthermore, no difference in doxorubicin‐mediated bone mass was observed between WT and EC‐Del1 mice in Cd73KO background (Figure S17, Supporting Information). Taken together, these findings demonstrate that endothelial cell‐derived DEL‐1 mediates bone protection and clearance of senescent MSCs in a CD73‐depenent manner during chemotherapy‐induced senescence‐associated bone loss.

Endothelial‐specific DEL‐1 overexpression decreases the number of senescent CD73PαSMSCs and increases their apoptosis in doxorubicin‐treated mice. A,B) Doxorubicin (10 mg kg) was administered once to 10‐week‐old() and() mice and MSCs in the bone marrow were analyzed by flow cytometry at day 7. The number of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (A) and the number of CFDGCD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (B) are shown. C,D) Doxorubicin was administered to WT and EC‐Del1 mice at a dose of 5 mg kgper week via intraperitoneal injection. After 4 weeks, femurs from both WT and EC‐Del1 mice were extracted and analyzed using microCT to assess bone loss. (C) Representative µCT‐derived images of bone microarchitecture (left) and quantification of µCT‐derived cortical thickness (Ct.Th(mm)) (right) at the mid‐diaphysis of the femur. (D) Representative µCT‐derived images of bone microarchitecture (left) and quantification of trabecular bone parameters at the distal femoral metaphysis (right), including bone volume fraction (Tb.BV/TV(%)), trabecular number (Tb.N(mm)), trabecular thickness (Tb.Th(mm)), and trabecular space (Tb.Sp(mm)). E–M) Doxorubicin (10 mg kg) was administered once to mice (E–K, WT and EC‐Del1 mice; L,M,and/EC‐Del1 mice) and MSCs in the bone marrow were analyzed by flow cytometry at day 7. Representative FACS plots (left) and quantitative bar graphs (right) of the number of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (E); representative FACS plots (left) and quantitative bar graphs (right) of the number of CFDGCD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (F); and representative FACS plots (left) and quantitative bar graphs (right) of the number of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (G). (H) Representative FACS plots (left) and quantitative bar graphs (right) of the percentage of caspase3cells in senescent p21CD73PαS (PDGFRαSca1CD45CD31TER119) MSCs and (I) the percentage of caspase3cells in p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs. (J,L) The number of p21PαS (PDGFRαSca1CD45CD31Ter119) MSCs. (K,M) The number of CFDGPαS (PDGFRαSca1CD45CD31Ter119) MSCs. Data are means ± SD (A,B,H,I:=6‐7 mice/group; C,D:=8 mice/group; E,G,J:=9‐10 mice/group; F,K:=4‐5 mice/group, and L,M:=5‐6 mice/group). Data in panels C,D,E,G,J are pooled from 2 experiments.<0.05;<0.01; NS, not significant. Unpaired Student's‐test, except for “Tb.BV/TV” plus “Tb.Th” graphs in panel D, and panel M (Mann–Whitneytest). + + −1 Cre− flox/flox WT Cre+ flox/flox ECKO + + + + – – – + + + + – – – −1 −1 −1 KO KO + + + + – – – + + + + – – – + – + + – – – + + + + + – – – + – + + + – – – + + + – – – + + + – – – * ** VE‐Cad Del1 Del1 VE‐Cad Del1 Del1 Cd73 Cd73 n n n n n p p t U 12 12 12

DEL‐1 Promotes Clearance of Senescent BM‐MSCs in Aged Mice

Aging‐related cellular senescence has been established as a major contributing factor to bone loss in old mice, reminiscent of osteoporosis in elderly humans.[62] DEL‐1 expression is diminished in mucosal tissue in old age in both mice and humans.[33, 63] Correspondingly, we observed a significant reduction (>60%) in DEL‐1 expression in the bone marrow of aged mice (22‐month‐old) as compared to their younger counterparts (3‐month‐old) (Figure6A, left). Conversely, the gene expression of the senescence‐associated genes p21 and p53 were found to be significantly higher in the bone marrow of aged mice as compared to their young counterparts (Figure 6A, middle and right). Consistently, we observed that the numbers of CD73+PαS MSCs (Figure S18, Supporting Information, left) and senescent p21+CD73+ MSCs (Figure 6B) exhibited a significant increase in the bone marrow of aged mice (21‐month‐old) compared to young mice (2‐month‐old). Therefore, there is a reverse relationship between MSC senescence and DEL‐1 expression in the bone marrow microenvironment.

Further investigations revealed that 22‐month‐old Del1KO mice displayed a decrease in cortical thickness of the femur and notable damage in trabecular bone microarchitecture, as opposed to age‐matched WT littermate controls (Figure 6C,D). Consistent with doxorubicin‐induced bone loss, TRAP and osteocalcin immunostaining suggested the presence of increased osteoclasts and reduced osteoblasts (Figure S19, Supporting Information). Moreover, in their bone marrow, 22‐month‐old Del1KO mice displayed elevated mRNA expression of p21, p53, Il6, Cxcl1, and Tgfβ1 (Figure 6E) as well as increased numbers of senescent p21+CD73+PαS cells (Figure 6F) when compared to WT littermates. In contrast, the numbers of senescent p21+CD73–PαS cells were comparable between 22‐month‐old WT and Del1KO mice (Figure 6G). Conversely, but consistently, 22‐23‐month‐old EC‐Del1 mice exhibited a significant reduction in the number of senescent p21+CD73+PαS, as well C12FDG+CD73+PαS, cell populations relative to WT littermates (Figure 6J,K), accompanied by an increase in trabecular bone microarchitectural parameters (Figure 6I). In contrast, no significant difference was observed in cortical bone thickness (Figure 6H). In a subsequent TRAP and osteocalcin immunostaining analysis, we observed that EC‐Del1 mice exhibited overall increased bone formation (Figure S20, Supporting Information). These findings from an aging‐driven bone marrow senescence model reinforce the notion derived from the chemotherapy‐induced senescence model that the senolytic action of DEL‐1 targets a specific (CD73‐positive) subset of senescent PαS MSCs.

DEL‐1 deficiency aggravates age‐related bone loss, whereas DEL‐1 overexpression alleviates this effect. A), andgene expression of the bone marrow of 3‐month‐old and 22‐month‐old WT mice determined by qPCR.was used for normalization of mRNA expression and the gene expression of 3‐month‐old WT mice was assigned an average value of 1. B) Representative FACS histogram (left) and quantitative bar graph of the number of p21CD73MSCs (CD45CD31TER119cells) in the bone marrow of healthy young (2‐month‐old) and old mice (21‐month‐old). C) Representative µCT images (from 22‐month‐old WT andmice) of bone microarchitecture at the sagittal plane (upper left) and the mid‐diaphysis of the femur (lower left), and quantification of cortical thickness (Ct.Th(mm)) (right) at the mid‐diaphysis of the femur. D) Representative µCT images (from 22‐month‐old WT andmice) of bone microarchitecture (left) and quantification of trabecular bone parameters, including bone volume fraction (BV/TV(%)), trabecular number (Tb.N(mm)), trabecular thickness (Tb.Th(mm)), and trabecular space (Tb.Sp(mm)) (right) at the distal femoral metaphysis. E) Senescence‐related gene expression () in the bone marrow of 22‐month‐old WT andmice, determined by qPCR.was used for normalization of mRNA expression and the gene expression of WT mice was assigned an average value of 1. F,G) Representative FACS plots (left) and quantitative bar graphs (right) of the number of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (F) and of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (G) in the bone marrow of 22‐month‐old WT andmice are shown. H) Representative µCT images (22‐month‐old WT and EC‐Del1 mice) of bone microarchitecture (left) and quantification of cortical thickness (Ct.Th(mm)) (right) at the mid‐diaphysis of the femur. I) Representative µCT images (22‐month‐old WT and EC‐Del1 mice) of bone microarchitecture (left) and quantification of trabecular bone parameters, including bone volume fraction (Tb.BV/TV(%)), trabecular number (Tb.N(mm)), trabecular thickness (Tb.Th(mm)), and trabecular space (Tb.Sp(mm)) (right) at the distal femoral metaphysis. J,K) Representative FACS plots (left) or quantitative bar graphs (right) of the number of p21CD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (J) and of CFDGCD73PαS (PDGFRαSca1CD45CD31Ter119) MSCs (K) in the bone marrow of 22–23‐month‐old WT and EC‐Del1 mice, assessed by flow cytometry. Data are means ± SD (A:= 3 mice/group, B:= 5 mice/group; C,D:= 5–6 mice/group; E:= 5–7 mice/group, F,G:= 7–8 mice/group; H,I,K:= 4 mice/group; J:= 6 mice/group).<0.05;<0.01;<0.001; NS, not significant. Unpaired Student's‐test, except for “Tb.BV/TV” in panel D, and panel G (Mann–Whitneytest). Edil3, p21 p53 Gapdh Del1 Del1 p21, p53, Il6, Cxcl1, and Tgfβ1 Del1 Gapdh Del1 n n n n n n n p p p t U + + – – – KO KO −1 KO + + + + – – – + – + + – – – KO −1 + + + + – – – + + + + – – – * ** *** 12

Administration of DEL‐1 or DEL‐1‐Inducing Macrolide Diminishes Senescent Cells and Reverses Periodontal Bone Loss in Aged Mice

Senescent cells also accumulate in the periodontal tissues in old age and are thought to exacerbate local inflammation as well as tissue destruction including alveolar bone loss, in part through the release of SASP proteins.[64, 65, 66] In aged mice, senescent cells are particularly abundant in the periodontal ligament (PDL), the structure that connects the root of the tooth to the alveolar bone.[65] The PDL is of central importance to the inflammatory mechanisms that drive bone loss in periodontitis[65, 67, 68] but is also a rich source of MSCs that are responsible for periodontal soft tissue and bone regeneration.[69, 70, 71, 72] We determined whether local administration of DEL‐1‐Fc (or Fc control) or systemic administration of erythromycin (ERM), a macrolide antibiotic which upregulates periodontal tissue expression of DEL‐1,[73] can reduce the number of senescent cells within the PDL space of aged mice. To this end, we used an aging model of naturally occurring periodontitis.[74] As a negative (vehicle) control for systemic administration, we used ethanol (EtOH), and for a positive control we used Bis‐2‐(5‐phenylacetamido‐1,3,4‐thiadiazol‐2‐yl)ethyl sulfide (BPTES), a selective inhibitor of glutaminase‐1, which causes senolysis and attenuates aging‐related organ dysfunction.[75] 18‐month‐old mice were treated with the above mentioned experimental and control treatments for 9 days, based on our recent study where systemic administration of ERM promoted alveolar bone regeneration in mice of the same age within 9 days, in a manner dependent on DEL‐1 upregulation.[76] Nine days post‐treatment, we performed SA‐β‐gal staining on maxillary sections to quantify senescent (SA‐β‐gal+) cells within the PDL space. Senescent cells were readily detected in the PDL space of untreated Fc control‐treated old mice but not in untreated young mice (Figure7A). The number of SA‐β‐gal+ cells, however, were significantly decreased in mice treated with DEL‐1‐Fc, ERM or BPTES (Figure 7A,B). Similar results were obtained when immunofluorescence histochemistry was used to quantify the numbers of p16+ (Figure 7C, top) and p21+ cells (Figure 7C, bottom) within the PDL space of treated and control aged mice. Specifically, the numbers of cells identified by these two senescent markers were significantly reduced by DEL‐1‐Fc, ERM or BPTES (Figure 7C,D). Furthermore, evaluation of mRNA expression of p16 and p21 by PDL cells isolated from the various treated and control groups of mice was consistent with the microscopic analysis; indeed, treatments with DEL‐1‐Fc, ERM or BPTES caused a significant decrease in p16 and p21 mRNA levels in aged mouse PDL cells (Figure 7E). The same experimental treatments (i.e., DEL‐1‐Fc, ERM and BPTES) also resulted in bone regeneration relative to the negative controls (Fc or EtOH) (Figure 7F), indicating that a reduction in the number of senescent cells in the PDL (Figure 7A–E) is associated with reversal of bone loss. Moreover, we have shown that in aged periodontal tissues, ERM administration enhanced the clearance of terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)‐positive cells, while concurrently reducing p21 levels. The combined decrease in p21 and enhanced apoptotic cell clearance suggests that ERM promotes the elimination of senescent cells in the mesenchymal stromal cell‐rich periodontal ligament (Figure S21, Supporting Information). These findings align with the role of DEL‐1 in modulating senescence‐associated pathways in periodontal tissue, as seen in the femoral bone.

Treatment with DEL‐1 or DEL‐1‐inducing macrolide diminishes senescent cells and reverses periodontal bone loss in aged mice. Aged (18‐month‐old) C57BL/6 mice were administered vehicle control (ethanol), erythromycin (ERM; 100 mg kg), or BPTES (12.5 mg kg), intraperitoneally or were locally injected with DEL‐1‐Fc (1 𝜇g) or an equal molar amount of Fc control daily for nine days. A) Senescent cell staining using a SA‐β‐gal staining kit was performed on maxillary cryosections. Representative optical microscopy images from young (2‐month‐old) mice, which remained untreated, and aged (18‐month‐old) mice that were treated as described above. Arrowheads indicate SA‐β‐galcells in the PDL space. T: Tooth; PDL: periodontal ligament; B: alveolar bone. Scale bars, 100 µm. B) The number of SA‐β‐galcells within the PDL space were counted from 4 random coronal sections and averaged per mouse (= 9 mice/group). C) Frozen maxillae sections were stained for cellular senescence markers p16 and p21 (Green). Representative images of maxillary sections obtained by fluorescence microscope are shown. The nuclei were counterstained using DAPI (blue). T: Tooth; PDL: periodontal ligament; B: alveolar bone. Scale bars, 100 µm. D) The number of p16and p21cells within the PDL space were counted from 4 random coronal sections and averaged per mouse (= 6 mice/group). E) Gene expression analysis ofandof PDL cells isolated from the aged mice treated as described above.was used for normalization of mRNA expression and the gene expression of controls (Fc or EtOH) was assigned an average value of 1 (= 5 mice/group). F) The distance from the cementoenamel junction (CEJ) to the alveolar bone crest (ABC) was measured in aged mice treated as described above. The CEJ‐ABC data were transformed to show bone changes relative to the respective controls (DEL‐1‐Fc vs Fc control and ERM or BPTES vs EtOH control), which served as the baseline. Positive values (in mm) indicate bone gain relative to the baseline control (= 8 mice/group). Data are means ± SD.<0.001;<0.0001. One‐way ANOVA and Tukey's multiple‐comparisons test. −1 −1 + + + + *** **** n n p16 p21 Actb n n p p

Discussion

While the emergence of extrinsic senolytic drugs shows promise in preventing senescence and revitalizing aged tissues,[77, 78] the precise endogenous mechanism(s) through which the local tissue microenvironment controls the survival and buildup of senescent cells remains incompletely understood. Here, we present evidence that DEL‐1 serves as an endogenous senolytic protein, which effectively promotes the apoptosis of senescent MSCs in the bone marrow and prevents bone loss, as shown in three independent senescence‐associated bone loss models. The fact that the expression of DEL‐1 declines gradually with increasing age[33, 79] suggests that this endogenous senolytic mechanism is progressively compromised, thereby further accelerating the accumulation of senescent cells. To the best of our knowledge, no other endogenous protein with senolytic properties has been previously described.

Mechanistically, the senolytic action of DEL‐1 was shown to be dependent on β3 integrin, one of its known cellular receptors,[28] and the ecto‐5′‐nucleotidase CD73, the expression of which is upregulated by DEL‐1. The expression and functional significance of CD73 in senescent MSCs in the bone marrow has remained largely unexplored. Here we showed that DEL‐1 increases the levels of adenosine in the culture supernatants of senescent BM‐MSCs, in a CD73‐dependent manner, leading to their apoptosis. Indeed, inhibiting CD73 or adenosine receptors effectively blocked DEL‐1‐induced apoptosis of senescent BM‐MSCs, which also involved attenuation of p38 MAPK activity and downstream reduction in BCL‐2 expression, ultimately leading to apoptosis (see model in Figure S11, Supporting Information).

Consistently, we noted an inverse correlation between MSC senescence and DEL‐1 expression within the bone marrow of aged mice as well as mice subjected to chemotherapy. For instance, deficiency in DEL‐1 contributed to heightened BM‐MSC senescence in both the aging‐ and chemotherapy‐induced senescence models. Endothelial (but not MSC)‐specific deficiency of DEL‐1 resulted in higher BM‐MSC senescence, implying that endothelial‐derived secreted DEL‐1 acts as a paracrine senolytic factor (Figure, Supporting Information). Consistently, the overexpression of DEL‐1 in endothelial cells mitigated cellular senescence in the bone marrow. These observations underscore the pivotal role of endothelium‐derived DEL‐1 as a key regulator of MSC senescence in the bone marrow. S11

In line with the critical importance of a CD73/adenosine pathway in mediating DEL‐1‐induced apoptosis of senescent cells, DEL‐1 was shown to selectively target CD73‐positive senescent MSCs, namely the CD73+PαS stromal cells in the bone marrow, in two different in vivo models (chemotherapy‐induced and aging‐induced senescence). CD73+PαS MSCs exhibit elevated expression of β3 integrin in comparison to other subsets of BM‐MSCs, including Sca1+CD29+ MSCs, CD73–PαS, and LepR+ MSCs. This explains, at least in part, why the β3 integrin‐mediated senolytic action of DEL‐1 preferentially targets senescent CD73+PαS MSCs. Importantly, a recent investigation has shown that CD73+MSCs exhibit heightened proliferative activity and a superior capacity for osteogenic differentiation, thereby promoting bone regeneration more effectively than CD73–MSCs.[46] This finding implies that the accumulation of senescent CD73+MSCs in the bone marrow may lead to severe impairment of bone homeostasis.

Of note, in the bone microenvironment, the CD73+MSCs exhibit a distinct perivascular distribution.[46] This localization is particularly intriguing, given that DEL‐1 is also prominently localized in the perivascular area in the bone marrow,[34] suggesting a potential close interaction between DEL‐1 and CD73+ MSCs. Consistently, we found that vascular endothelial overexpression of DEL‐1 had a pronounced effect on the regulation of the senescent CD73+PαS (but not of CD73–PαS) MSC numbers in the bone marrow. Our current study therefore implicates CD73+PαS+ MSCs as major targets of endothelial DEL‐1‐mediated senolysis and explains why this subset can accumulate in old age, when the bone marrow levels of DEL‐1 decline severely.

Previously, we showed that DEL‐1 binds to PS through its discoidin I‐like domains, thereby enabling the recognition of apoptotic cells.[29] Moreover, the concomitant interaction of DEL‐1 with β3 integrin on macrophages, through an RGD motif in its second EGF‐like repeat, promotes the efferocytosis of apoptotic cells. In the present study, DEL‐1 was shown to depend on its RGD motif and on β3 integrin expression in order to induce apoptosis of senescent BM‐MSCs and thereby promote their phagocytosis by macrophages. In particular, a truncated version of DEL‐1 containing the N‐terminal EGF‐like repeats region but lacking the discoidin I‐like domains was sufficient to reproduce the senolytic effect of the parent molecule. This finding suggests that the senolytic effect of DEL‐1 is distinct from its efferocytosis‐promoting effect.

Our present work also revealed that DEL‐1 inhibited the activation/phosphorylation of p38 MAPK and expression of BCL‐2, thereby promoting apoptosis in senescent BM‐MSCs. This finding aligns with the mechanism of action of the senolytic agent R406 (SYK inhibitor), which exerts its senolytic effect through dual inhibition of FAK and p38 MAPK.[56] Inhibition of p38 MAPK in our senescent BM‐MSC model caused apoptosis, in agreement with an earlier study where p38 MAPK inhibition promoted tumor apoptosis.[55] It should be noted, however, that the activation of p38 MAPK can exhibit either pro‐ or anti‐apoptotic effects, depending on the specific cellular environment.[80, 81, 82, 83] Several established senolytic agents directly target anti‐apoptotic BCL‐2 family proteins, including ABT‐263, ABT‐737, and ABT‐199.[17, 18, 84] Importantly, we demonstrated that DEL‐1 inhibits p‐p38 MAPK, leading to a reduction in BCL‐2 levels, indicating BCL‐2 acts downstream of p‐p38 MAPK to regulate apoptosis, consistent with previous findings.[85] The downregulation of BCL‐2, a key anti‐apoptotic protein and hallmark of cellular senescence, lowers the apoptotic resistance threshold in senescent cells and appears to be a critical mediator of DEL‐1‐induced senolysis.

Senescent osteogenic progenitors, including MSCs, play a crucial role in driving skeletal senescence and bone loss.[27, 41] However, the extent to which the elimination of senescent MSCs can prevent senescence‐induced or age‐related bone loss remains incompletely understood. Recent studies suggest that senescent MSCs also affect bone formation.[26, 27, 86] Notably, increased miR‐29 expression has been observed in senescent MSCs, and targeting miR‐29a in these cells impedes skeletal senescence and enhances bone mass by promoting osteoblast bone‐forming activity.[86] Furthermore, senolytic agents have been shown to boost osteogenic activity of aged, but not young, BM‐MSCs in vitro and in vivo, suggesting that the removal of senescent BM‐MSCs enhances bone‐forming potential.[26] Additionally, targeting senescent BM‐MSCs by treatment of quercetin efficiently eliminates senescent cells in bone, restores the function of BM‐MSCs, and promotes bone formation in aged mouse models.[27] Consistent with these findings, doxorubicin‐challenged Del1KO mice displayed decreased concentrations of osteocalcin (a marker of bone formation) and increased concentrations of CTX1 and TRAcP 5b (markers of bone resorption), along with osteocalcin and TRAP immunostaining, suggesting that both impaired bone formation and elevated bone resorption contributed to their aggravated bone loss as compared to their DEL‐1‐sufficient littermate controls. Stated another way, the treatment with DEL‐1 mitigates doxorubicin‐induced bone loss by promoting both bone formation and inhibiting bone resorption.

Given that DEL‐1 mediates its senolytic effect in a β3 integrin‐dependent manner and recent findings that β3 integrin is significantly elevated in senescent cells and aged tissues,[87] we can infer that DEL‐1 can preferentially target senescent, rather than healthy, MSCs. Although DEL‐1 cannot specifically identify senescent MSCs (healthy cells may also express β3 integrin), healthy cells are not susceptible to the apoptosis‐inducing effect of DEL‐1, as shown in the present study. Therefore, by targeting β3 integrin‐expressing cells, DEL‐1 can efficiently exert a beneficial action (senolysis) without potential adverse effects on healthy cells.

Interestingly, even a relatively modest proportion (≈10–15%) of senescent cells in old primates can lead to significant tissue dysfunction.[88] The hereby established senolytic effect of DEL‐1 renders this protein a promising therapeutic approach to counteract cellular senescence and its pathologic sequalae. In this regard, administration of endogenous proteins as replacement therapies (due to aging‐related decline in their expression as is the case with DEL‐1) are likely to exhibit minor, if any, potential safety issues and have increased probability of FDA approval as compared with extrinsic biologics.[89, 90, 91] Moreover, expression of DEL‐1 in old age can be reinstated by administration of DEL‐1‐inducing agents such as macrolide antibiotics, including synthetic derivatives free of antibiotic action.[76] Given our present findings that endothelial cells constitute a major cellular source of DEL‐1 that is responsible for its senolytic effect, it is of particular significance to note that macrolides were also shown to upregulate the expression of DEL‐1 in human endothelial cells.[73] As a proof‐of‐concept, here we showed that local administration of DEL‐1 or systemic administration of a macrolide (ERM) reduced the numbers of senescent cells within the PDL space of aged mice, which are known to express very low levels of DEL‐1 mRNA and protein.[33, 76] Although the ability of DEL‐1 and ERM to reverse bone loss in aged mice was also shown in our previous publication, we did not previously implicate senolysis as a contributing mechanism.[76] Importantly, in our present work the senolytic and bone protective effects of DEL‐1 and ERM were comparable to those of BPTES, an established senolytic that ameliorates senescence‐associated pathology in different organs.[75]

In summary, our comprehensive findings establish DEL‐1 as a novel endogenously expressed senolytic protein, which promotes the removal of accumulated senescent MSCs in the bone marrow, particularly the CD73+PαS stromal cell subset, as shown using both chemotherapy‐induced and aging‐related cellular senescence models. Moreover, DEL‐1 caused a reduction of senescent cells in the periphery, where its senolytic activity was associated with reversal of aging‐related periodontal bone loss. Senolytic drugs have limitations in effectively eliminating senescent cells due to the dynamic and highly heterogeneous nature of various senescence programs. This results in different types of senescent cells showing varying levels of sensitivity to current senolytic treatments.[92, 93, 94] To overcome these challenges and advance anti‐aging interventions, there is a strong demand for the development of new approaches that can selectively target and remove senescent cells in various cell types and tissues. In this regard, DEL‐1 displays selectivity for senescent CD73+PαS stem cells, which in the bone marrow express high levels of the DEL‐1 receptor β3‐integrin. Thus, our findings may provide a potential strategy to enhance the senolytic efficacy against a specific subset of senescent MSCs, which promote loss of skeletal bone in old age and other contexts such as chemotherapy.

Experimental Section

Sex as a Biological Variable

Sex as a biological variable was considered by making use of both male and female animals. As there were no significant differences in the results obtained with male and female mice, the respective data were pooled per treatment group.

Mice

Mice were maintained in individually ventilated cages under specific pathogen‐free conditions on a standard 12‐h light/dark cycle. Food and water were provided ad libitum. 6–10‐week‐old and 18–23‐month‐old mice were used for experiments. 6–7‐week‐old mice were used for bone marrow‐derived MSC isolation in the in vitro experiments, and 8–10‐week‐old mice were used for the in vivo experiments. 6–10‐week‐old mice were purchased from Jackson Laboratories or from Charles River Laboratories (Japan, Inc., Kanagawa, Japan). 21–23‐month‐old C57BL/6 mice were kindly provided from the National Institute of Aging and 18‐month‐old C57BL/6 mice were purchased from Charles River Laboratories (Japan, Inc., Kanagawa, Japan). In old mice, all comparisons were made within age‐matched groups (i.e., WT vs Del1KO or WT vs EC‐Del1 at the same age), and not across 18— and 21–23—month‐old mice ensuring internal consistency within each experimental set. Because there were no significant differences in the results obtained with male and female mice, the respective data were pooled per treatment group. C57BL/6 Edil3–/– (Del1KO) mice were crossed with WT C57BL/6 (The Jackson Laboratory) mice to generate heterozygous breeders that in turn were crossed to generate KO and WT littermates. C57BL/6 mice overexpressing DEL‐1 in the endothelium (EC‐Del1) were previously described.[61, 95] B6.129S1‐Nt5e1Lft/J (Cd73KO, Stock No: 018986), B6.Cg‐Tg(Prrx1‐Cre)1Cjt/J (Prx1Cre), C57BL/6‐Itgb31.1Wlbcr/J (β3flox/flox), B6.FVB‐Tg(Cdh5‐cre)7Mlia/J (VE‐CadCre), and C57BL/6‐Tg (UBC‐GFP)30Scha/J (UBC‐GFP) transgenic mice were purchased from The Jackson Laboratory. Prx1Cre+β3flox/flox mice were generated by crossing Prx1Cre+ mice with β3flox/flox mice. The Del1flox/flox mice were developed by genOway (Lyon, France). The Edil3 gene targeting vector was constructed from genomic C57BL/6N mouse strain DNA. LoxP sites were inserted upstream and downstream of the promoter region and exon 1. The integrity of the targeting vector was assessed by full sequencing. Linearized targeting vector was transfected into C57BL/6 ES cells according to genOway's electroporation procedures. G‐418 resistant ES cell clones were subsequently validated by PCR, using primers hybridizing within and outside the targeted locus, and whole locus sequencing to confirm the absence of any genetic alteration. Recombined ES cell clones were microinjected into C57BL/6 blastocysts, and gave rise to male chimeras with a significant ES cell contribution. Breeding was established with C57BL/6N mice to produce the heterozygous Edil3 floxed line. Heterozygous animals were genotyped by PCR, and further validated by full sequencing of the targeted locus, including the floxed region, the homology arms, and a few kbs upstream and downstream of the homology arms. Prx1Cre+Del1flox/flox mice were generated by crossing Prx1Cre+ mice with Del1flox/flox mice. VE‐CadCre+Del1flox/flox mice were generated by crossing VE‐CadCre+ mice with Del1flox/flox mice. Cd73KO mice were crossed with EC‐Del1 mice to generate Cd73KO/EC‐Del1 offspring.

DEL‐1 Protein and Other Reagents

Full‐length human DEL‐1 was generated and purified as a fusion protein with the human IgG1‐Fc fragment (DEL‐1‐Fc) as were previously described.[34, 37] A truncated version of DEL‐1 that contains only the EGF‐like repeats (DEL‐1[E1‐E3]‐Fc) and a DEL‐1 point mutant in which Glu was substituted for Asp in the RGD motif of the second EGF‐like repeat (DEL‐1[RGE]‐Fc) were prepared as were previously described.[37] Recombinant human IgG1 Fc protein was purchased from R & D systems. As human and mouse DEL‐1 share 96% amino acid sequence and the two proteins have similar functions,[29, 32, 33, 37] human DEL‐1 was used in mouse experimental systems in the current study, as in previous publications.[29, 34, 35, 37] For signaling pathway assays, the following reagents were used; CGS15943, a potent adenosine receptor nonselective antagonist (1 µm); α,β‐methylene ADP (APCP), an ADP analog acting as CD73 inhibitor (10 µm); SB203850, a p38 MAPK inhibitor (0.5–5 µm). For the β1 integrin blocking experiment, anti‐β1 integrin neutralizing antibody (1 µg ml−1, catalogue no. 16‐0291‐85, Invitrogen) or Armenian Hamster IgG isotype control (catalogue no. 16‐4888‐81, Invitrogen) was used.

Culture of Bone Marrow‐Derived Mesenchymal Stromal Cells and Senescence Induction

Mouse femurs and tibias were flushed with PBS containing 2% FBS. Resulting cell suspensions devoid of RBC were cultured in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin at 37 °C in a 5% CO2 atmosphere. Fresh culture media was changed every 3 days, and adherent cells were harvested and sub‐cultured once the plate was ≈80% confluent. BM‐MSCs were used at passage 3 to 5 in all experiments.[96] BM‐MSCs were characterized by positive expression of CD44, Sca1, Thy7.2 (CD90) and negative expression for CD31, CD34, CD45, CD11b, MHC class II antigens by flow cytometry analysis.[96] Three different stressors were used for senescence induction. BM‐MSCs were treated with H2O2 (100 µm) for 2 h and culture supernatant was removed, followed by replacement with fresh culture media and further culture for 8‐10 days.[97] Moreover, BM‐MSCs were incubated with doxorubicin (100 nm) or palbociclib (0.5 µm) overnight and washed and further incubated for 8 or 10 days, respectively, until analysis. H2O2‐induced senescent BM‐MSCs were used in the figures unless otherwise stated. Flow cytometric analysis (using the fluorescent β‐galactosidase substrate C12FDG) was performed to confirm that the cells were senesced.

Quantitative Real‐Time PCR

Total RNA was extracted using an RNeasy plus kit (QIAGEN) or Direct‐zol RNA miniprep kit (Zymo Research), and cDNA was generated with an oligo(dT) primer and the iScript cDNA kit and then the resulting cDNAs were subjected to real‐time PCR amplification with Taqman PCR mater kit (Applied Biosystems) or the SsoFast EvaGreen Supermix (BioRad, Cat.1725201) by using the Applied Biosystems 7500 Fast Real‐Time PCR System or QuantStudio 3 Real‐Time PCR System (Thermo Fisher Scientific) or a CFX384 real‐time PCR (BioRad).[98, 99] Data were analyzed using the comparative (ΔΔCt) method. TaqMan probes and gene‐specific primers for detection and quantification of murine genes investigated in this study were purchased from Thermo‐Fisher Scientific and included: Il6 (Mm00446190_m1), Tgfβ1 (Mm01178820_m1), p16 (Mm00494449_m1), p21 (Mm00432448_m1), p53 (Mm01731290_g1), Gapdh (Mm99999915_g1), Cxcl1 (Mm04207460_m1), Cxcl2 (Mm00436450_m1), Edil3 (Mm01291247_m1), and Mmp12 (Mm00500554_m1), Bcl‐2 (Mm00477631_m1), Itgb3 (Mm00443980_m1).

Primers used for mRNA expression analysis in PDL cells or lung endothelial cells were as follows:

p16‐F CGCAGGTTCTTGGTCACTGT, p16‐R TGTTCACGAAAGCCAGAGCG; p21‐F CCTGGTGATGTCCGACCTG, p21‐R CCATGAGCGCATCGCAATC; Actb‐F GGCTGTATTCCCCTCCATCG, Actb‐R CCAGTTGGTAACAATGCCATGT. Edil3‐F CCTGTGAGATAAGCGAAGC, Edil3‐R GAGCTCGGTGAGTAGATG; 18S‐F GTTCCGACCATAAACGATGCC, 18S‐R TGGTGGTGCCCTTCCGTCAAT.

Flow Cytometry and Cell Sorting

Bone marrow cells were obtained by flushing with 2% FBS in PBS and stained with antibodies. For surface marker staining, the cells were pretreated with Fc blocker (2.4G2) for 10 mins at 4 °C and further incubated with antibodies against CD45 (clone 30‐F11, catalogue no. 103130 or 103107, Biolegend), CD31 (clone 390, catalogue no. 102405, 102407 or 102439, Biolegend), Ter119 (clone TER‐119, catalogue no. 116205, 116207 or 116223), Sca1 (clone E13‐161.7, catalogue no. 122513, Biolegend), PDGFRα (clone APA5, catalogue no. 135923 or 135907, Biolegend), CD29 (clone HMβ1‐1, catalogue no. 102225, Biolegend), CD73 (clone TY/11.8, catalogue no. 127206, Biolegend), β3 integrin (clone 2C9.G2, catalogue no. 562917, BD Biosciences), LepR‐biotin (catalogue no. BAF497, R&D Systems) for 30 mins at 4 °C. Cells were incubated with Anti‐biotin conjugated with streptavidin (catalogue no. 405206, Biolegend) for 30 mins at 4 °C to detect LepR. For intracellular staining, the cells were fixed and permeabilized using Cytofix/Cytoperm solution (BD) or Foxp3/Transcription Factor Staining Buffer Set (eBioscience) for 1 h at 4 °C followed by staining with antibodies against Ki67 (Clone 16A8, catalogue no. 652421, Biolegend or clone B56, catalogue no. 563756, BD Biosciences), γH2AX (clone 2F3, catalogue no. 613419, Biolegend or clone N1‐431, catalogue no. 564720, BD Biosciences), p21 (clone EPR18021, catalogue no. ab237265, Biolegend), p‐p38 (clone A16016A, catalogue no. 690203, Biolegend), BCL‐2 (clone BCL/10C4, catalogue no. 633508, Biolegend) for 30 mins at 4 °C. For cell size analysis, a size reference kit containing small (10.1–16.8 µm) and large diameter (>16.8 µm) beads was used for size reference. For in vitro BM‐MSC analysis, cultured BM‐MSCs were stained with antibodies against CD31 (clone 390, catalogue no. 102405, Biolegend), CD34 (clone RAM34, catalogue no. 11‐0341‐82, eBioscience), CD44 (clone IM7, catalogue no. 560569), CD90 (clone 30‐H12, catalogue no. 105311, Biolegend), Sca1 (clone E13‐161.7, catalogue no. 122513, Biolegend), CD29 (clone HMβ1‐1, catalogue no. 102225, Biolegend), CD45 (clone 30‐F11, catalogue no. 103130, Biolegend), CD11b (clone M1/70, catalogue no. 101215, Biolegend), I‐Ab (clone AF6‐120.1, catalogue no. BDB562011, BD Biosciences), CD47 (clone miap301, catalogue no. 127525, Biolegend). Cell acquisition was performed on a Novocyte flow cytometer (ACEA Biosciences). Data analysis was carried out with NovoExpress software (ACEA Biosciences).

For cell sorting, BM cells were collected from mice and stained with antibodies against CD45, Ter119, CD31, and CD73. Live CD45–Ter119–CD31– or CD73+CD45–Ter119–CD31– stromal cell population was sorted by an Aria II sorter (BD Biosciences). Over 95% purity was confirmed by the Aria II sorter. To isolate lung endothelial cells, lung tissues were minced and digested with collagenase D (1.5 mg mL−1, Sigma, Cat. COLLD‐RO Roche) and DNase I (5 unit mL−1, Sigma, catalogue no. 04716728001 Roche) in DMEM (Gibco, catalogue no. 31966) containing 0.5% bovine serum albumin (Sigma, catalogue no. A7030) for 60 min at 37 °C with shaking. Digested lung tissues were then filtered through a 100‐µm cell strainer and centrifuged at 600 g for 7 min. Red blood cells were removed by RBC lysis buffer and cells were resuspended with PBS containing 0.5% BSA and 2 mm EDTA. Cells were incubated with anti‐mouse CD31 (clone MEC 13.3, BD Bioscience, catalogue no. 553373) and anti‐mouse CD45 (clone 30‐F11, Biolegend, catalogue no.103112, 103130) after incubation with purified rat anti‐mouse CD16/CD32 (clone 2.4G2, BD Bioscience, catalogue no. 553141) for 20 min at 4 °C. After adding Hoechst 33258 (Invitrogen, catalogue no. H1398) to define live cells, live CD45–CD31+ cells were sorted as lung endothelial cells by BD FACSMelody Cell Sorter (BD Bioscience) with BD FACSChorus Software.

PS expression on apoptotic cells was detected using fluorescein‐labeled Annexin V/7‐Aminoactinomycin D (7‐AAD), according to the manufacturer's instruction (APC Annexin V Apoptosis Detection Kit with 7‐AAD, catalogue no. 640930, Biolegend). Cells were stained with annexin V in annexin‐binding buffer for 15 min at room temperature followed by FACS analysis. Caspase‐3 expression in senescent BM‐MSCs was quantified by intracellular staining using the caspase‐3 apoptosis kit (catalogue no. 550480, BD Biosciences) followed by FACS analysis. To quantify SA‐β‐gal activity, flow cytometric analysis was performed using fluorescent β‐galactosidase substrate C12FDG (Invitrogen).[100, 101]

Immunoblotting Analysis

Cell lysates were prepared using RIPA lysis buffer (catalogue no. sc‐24948A, Santa Cruz Biotechnology). Denatured protein samples (10 µg) were separated by SDS‐PAGE and transferred to a PVDF membrane. The membranes were incubated in blocking buffer (5% nonfat dry milk, 10 mm Tris [pH 7.5], 100 mm NaCl, and 0.05% Tween 20), followed by probing with a primary antibody against phospho‐p38 (catalogue no. GTX133460, GeneTex). After antibody removal using Restore Western Blot Stripping Buffer (catalogue no. 21059, Thermo Fisher Scientific), it was reprobed with an antibody against total p38 (catalogue no. 9212S, Cell Signaling). Protein signals were visualized with chemiluminescence using the Millipore ECL system following incubation with HRP‐conjugated secondary Ab. Images were captured using iBright 1500 (Invitrogen).

SA‐β‐gal Staining

Senescence‐associated β‐galactosidase staining was performed in senescent cell cultures. The cells were fixed with 4% PFA for 10 min, followed by washing with PBS, and then stained with X‐gal solution as per the manufacturer's instruction (Senescence β‐Galactosidase Staining Kit, catalogue no. 9860S, Cell Signaling). Cells were analyzed under a light microscope (at 200x magnification) and the representative images were acquired by a digital camera connected to a bright‐field microscope. The Cellular Senescence Assay Kit (Cell Biolabs USA, catalog no. CBA‐230) was used to detect senescent cells in PDL space (SA‐β‐gal+ PDL cells). The SA‐β‐gal+ PDL cells (blue‐stained) were photographed, and the number of positively stained cells per area were counted. The histological images were obtained using a BZ‐9000 Bio Revo all‐in‐one microscope.

Caspase‐3/7 Activity

Caspase‐3/7 activity was measured using the Caspase‐Glo3/7 Assay Kit (catalogue no. G8090, Promega). The addition of the Caspase‐Glo 3/7 reagent directly to the assay plate well results in cell lysis, followed by caspase cleavage of the tetrapeptide sequence DEVD substrate and generation of luminescence. Luminescence readings (RLU, relative light units), representing caspase‐3/7 activity, were determined using a plate‐reading luminometer according to the manufacturer's instructions.

Doxorubicin‐Induced Bone Loss and microCT Analysis

Doxorubicin (catalogue no. 2252, Tocris) was prepared in sterile water and administered (5 mg kg−1) to mice via intraperitoneal injection once weekly for 4 weeks. In certain experiments, mice were injected once with 10 mg kg−1 of doxorubicin. Femurs were isolated from doxorubicin‐administered mice and fixed with 10% formalin overnight. The bones were washed in PBS and stored in 70% ethanol until analysis. The femoral samples were scanned using microCT (MicroCT 35, Scanco Medical AG) at a resolution of 6 µm isotropic voxel size, 55 kVp energy, 145 µA intensity. At the femoral midshaft, the end of the growth plate region was used as a landmark to establish a consistent location for starting analysis, and the next 100 slices were acquired for trabecular bone and cortical bone analyses by visually drawing the volume of interest (VOI) separately. The trabecular bone tissue and cortical bone tissue within the VOI were segmented from soft tissue using a threshold of 272 mgHA/cm3 and a Gaussian noise filter (sigma = 1.2, support = 2.0). 3D standard microstructural analysis was performed to determine the geometric trabecular bone volume over total volume (Tb.BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation (Tb.Sp). For analysis of cortical bone, cortical thickness was recorded.[102]

Adenosine Quantification

Adenosine was measured in bone marrow fluid or cell culture supernatant. The femoral bone was flushed with 1 mL cold PBS and the bone marrow fluid was removed after centrifuge. Culture supernatant of senescent BM‐MSCs was harvested and clear supernatant was collected after centrifugation. Adenosine production in the clear samples was quantified using an adenosine detection kit (catalogue no. ab211094, Abcam) and an ELISA plate reader, following the manufacturer's protocol.

Macrophage Phagocytosis Assays—In Vivo Experiments

WT mice were injected intraperitoneally with 2% w v−1 thioglycolate (Sigma–Aldrich). After 72 h, the mice received intraperitoneally 3 × 105 senescent BM‐MSCs that were pre‐stained with 1 µm CypHer5E (a pH‐sensitive dye for which the fluorescence increases after uptake into the acidic phagolysosome; catalogue no. PA15401V, VWR International) together with 10 µg of DEL‐1‐Fc or an equal molar amount of Fc control. In other experiments, senescent BM‐MSCs were pretreated for 3 days in vitro with DEL‐1‐Fc (10 µg ml−1), or an equal molar concentration of Fc control, and after washing with PBS and staining with CypHer5E, the cells were injected to WT mice. 3 h post‐injection, the mice were euthanized and the peritoneal contents were harvested. Macrophage phagocytosis of CypHer5E‐labeled senescent BM‐MSCs was analyzed with flow cytometry. Phagocytic macrophages were identified as live CyHer5E+CD45+F4/80+ cells.

Macrophage Phagocytosis Assays—In Vitro Experiments

Senescent BM‐MSCs were stained with 1 µm CypHer5E. Afterward, WT bone marrow‐derived macrophages were co‐cultured with CypHer5E+ senescent BM‐MSCs, in a 1:1 ratio of senescent BM‐MSCs to macrophages, in the presence of DEL‐1‐Fc, DEL‐1[RGE]‐Fc or Fc control protein for 2 h. Cells were then washed, trypsinized, harvested and stained with antibodies. This was followed by flow cytometry analysis to evaluate macrophage phagocytosis. Phagocytic macrophages were identified as live CypHer5E+CD45+F4/80+ cells.

Femoral Bone Histology

Femoral bone tissues were fixed in 10% neutral‐buffered formalin at 4 °C overnight, decalcified in formic acid for 10 days, and then immersed in 30% sucrose in PBS before embedding in optimal cutting temperature (OCT) compound. Frozen sections (8 µm thick) of the distal femur were prepared using a cryostat and used for hematoxylin and eosin (H&E) staining, tartrate‐resistant acid phosphatase (TRAP) staining, and immunostaining for osteocalcin or p21 for senescence.

Immunohistochemistry and Immunofluorescence Staining

TrueBlack Lipofuscin Autofluorescence Quencher was used to reduce the lipofuscin autofluorescence in maxillary sections from aged mice. The quenched maxillary sections were later incubated with anti‐p16 antibody (catalogue no. ab211542, Abcam), anti‐p21 antibody (catalogue no. ab188224, Abcam), and followed by TUNEL assay (catalogue no. C10247, Thermo Fisher Scientific). Secondary antibody was donkey anti‐rabbit IgG conjugated with Alexa Fluor 488 (catalogue no. A‐21206, Thermo Fisher Scientific). All sections were counterstained with ProLong Diamond Antifade Mountant with DAPI (catalogue no. P36962, Thermo Fisher Scientific). The fluorescent images were obtained using a ZEISS Axio Imager 2 microscope and processed with ZEN Blue imaging software. For femoral bone staining, tissue sections were incubated with Alexa Fluor 647‐conjugated anti‐p21 (catalogue no. ab237265, Biolegend) or rabbit anti‐mouse osteocalcin (catalogue no. ab93876, Abcam), followed by Alexa Fluor 488‐conjugated goat anti‐rabbit IgG (catalog no. A‐11034, Invitrogen). Nuclei were counterstained with DAPI, and images were captured using a Nikon Eclipse Ni‐E automated fluorescent microscope. To visualize osteoclasts, TRAP staining was performed using a TRAP staining kit (catalogue no., K2606 APExBio), according to the manufacturer's instructions.

ELISA

Whole blood was collected from mice after overnight fasting and allowed to coagulate at room temperature for 30 min. It was then centrifuged at 2000 g for 10 min at 4 °C, and the supernatant was collected. The supernatant was immediately stored at −80 °C. Serum levels of CTX1 (catalogue no. AC‐06F1, Immunodiagnostic Systems), TRAcP 5b (catalogue no. SB‐TR103, Immunodiagnostic Systems), and osteocalcin (catalogue no. EEL003, Thermo Fisher Scientific), were measured using ELISA kits, following the manufacturer's protocol. IL‐6, TGFβ1, and CXCL2 in bone marrow fluid was measured by ELISA kit (catalogue no. 88‐7064‐88, catalogue no.88‐8350‐88, and catalogue no. EMCXCL2, respectively; Invitrogen).

Murine Maxillary Sections

Maxillary tissue sections used for immunofluorescence or SA‐β‐gal staining were prepared as were described previously.[76] In brief, the maxillae were fixed overnight in 4% paraformaldehyde phosphate buffer solution, then decalcified in ethylenediaminetetraacetic acid (EDTA) solution for one week at 4 °C. Using a cryostat, tissues were cut in coronal sections (diameter = 12 µm per slide).

Primary mouse PDL CELLS

Primary mouse PDL cells were harvested and maintained as previously described.[76, 103] In brief, after euthanizing mice, the maxilla and mandible were carefully separated. The surrounding gingiva was removed prior to the extraction of the teeth. A Leica (EZ4W) stereoscopic microscope (Leica Microsystems) was utilized during the extraction procedures. The collected teeth were washed with PBS and transferred to fresh wells. Liberase solution (Catalogue no. 05401119001, Merck) was used at a final concentration of 2 units mL−1. The teeth were incubated in a digestion buffer inside 15 mL low‐binding tubes (STEMFULL; Sumitomo Bakelite) and placed in an incubator at 37 °C with shaking at 200 rpm. Subsequently, 10 mL of PBS with 2% fetal bovine serum (FBS) as a wash buffer was added, and the cells were centrifuged at 400 g for 8 min at 4 °C. The resulting pellet was resuspended in a washing buffer and passed through a 70‐µm cell strainer. Following a second centrifugation at 300 g for 5 min at 4 °C, the cells were resuspended in a washing buffer. Then, the cells were seeded to PDL complete medium (SCGM Stromal Cell Bullet Kit; Lonza) to adjust the cell concentration at 104 cells mL−1. The cells were subsequently cultured for 1 week, changing medium every 2 days, and when 80% confluent, the cells were ready for subsequent experiments.

Alveolar Bone Analysis

Alveolar bone changes were evaluated morphometrically in maxillae by stereoscopic microscopy (Leica Microsystems, Wetzlar, Germany or Nikon Instruments, Melville, New York, USA) as were previously described.[73, 76] The distance from the cementoenamel junction to the alveolar bone crest (CEJ‐ABC) was measured at 14 pre‐established sites from the first molar to the third molar on both sides of the maxilla. Bone changes were calculated by subtracting the 14‐site total CEJ‐ABC distance for each mouse from the mean CEJ‐ABC distance of the control group. Bone changes with positive values indicate bone growth relative to control‐treated mice.

Statistical Analysis

After confirming normality, data were analyzed with a two‐tailed unpaired Student's t‐test (comparison of only two groups) or a one‐way ANOVA followed by Dunnett's or Tukey's multiple‐comparisons test (when comparing more than two groups). In certain experiments (comparison of only two groups) where the data did not follow normal distribution, the non‐parametric two‐tailed Mann‐Whitney U‐test was used. All statistical analysis was performed using GraphPad Prism software (GraphPad). p‐values <0.05 were considered to be statistically significant.

Study Approval

Animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pennsylvania (approval no. 805528), or the IACUC of Niigata University (approval no. SA00960), as well as by the Landesdirektion Sachsen (approval no. 25‐5131/522/3) and were performed in compliance with institutional, state, and federal policies.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

T.C. and G.H. Equal contribution as senior authors. J.‐H.L. conceived designed and performed research, analyzed and interpreted data, and co‐wrote the manuscript. K.S. and M.D.C.S. performed experiments and interpreted data. H.W., G.T., K.J.C., J.M.K., T.K., and X.L. interpreted data and edited the manuscript. T.M. designed and performed research, analyzed and interpreted data, and edited the manuscript. T.C. conceived the study, designed the research, interpreted the data and edited the manuscript. G.H. conceived and designed the study, supervised research, interpreted data, and wrote the manuscript. All authors read and approved the manuscript.

Supporting information

Acknowledgements

This work was supported by grants from the NIH (DE028561, DE026152 and DE029436 to G.H.), the JSPS KAKENHI (23KK0160 and 23K24525 to T.M.), the Mochida Memorial Foundation (J23G0194 to T.M.), AMED (JP23ym0126801 and JP24ym0126801 to T.M.), and the Deutsche Forschungsgemeinschaft (LI 3493/1‐1 to J.‐H.L and the Collaborative Research Centre TRR369, project C03 to T.C.) and the German Center for Child and Adolescent Health to T.C. The cartoon in Figure(Supporting Information) was created using BioRender.com. The authors thank Sylvia Grossklaus (Institute for Clinical Chemistry and Laboratory Medicine, University Hospital and Faculty of Medicine, Technische Universität Dresden, Dresden, Germany) for technical assistance. S11

Contributor Information

Jong‐Hyung Lim, Email: limhyung@upenn.edu.

George Hajishengallis, Email: geoh@upenn.edu.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

Associated Data

Supplementary Materials

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Funding

Competing interests

0 of 12
authors report competing interests
12 report none
PubMed

What Lands in Your Inbox Each Week:

  • 📚7 fresh studies
  • 📝plain-language summaries
  • ✅direct links to original studies
  • 🏅top journal indicators
  • 📅weekly delivery
  • 🧘‍♂️always free