Age-related inflammation triggers skeletal stem/progenitor cell dysfunction

Significance

As we age, our capacity for tissue repair and regeneration in response to injury declines. Accordingly, bone repair is delayed and impaired in older patients. At the cornerstone of bone healing is the skeletal stem/progenitor cell (SSPC), whose function and number diminishes with age. However, the mechanisms driving this decline remain unclear. Here, we identify age-associated inflammation (“inflamm-aging”) as the main culprit of SSPC dysfunction and provide support for a central role of NF-κB as a mediator of inflamm-aging. Our results show that modification of the inflammatory environment may be a translational approach to functionally rejuvenate the aged SSPC, thereby improving the regenerative capacity of the aged skeleton.

Abstract

Aging is associated with impaired tissue regeneration. Stem cell number and function have been identified as potential culprits. We first demonstrate a direct correlation between stem cell number and time to bone fracture union in a human patient cohort. We then devised an animal model recapitulating this age-associated decline in bone healing and identified increased cellular senescence caused by a systemic and local proinflammatory environment as the major contributor to the decline in skeletal stem/progenitor cell (SSPC) number and function. Decoupling age-associated systemic inflammation from chronological aging by using transgenic Nfkb1KO mice, we determined that the elevated inflammatory environment, and not chronological age, was responsible for the decrease in SSPC number and function. By using a pharmacological approach inhibiting NF-κB activation, we demonstrate a functional rejuvenation of aged SSPCs with decreased senescence, increased SSPC number, and increased osteogenic function. Unbiased, whole-genome RNA sequencing confirmed the reversal of the aging phenotype. Finally, in an ectopic model of bone healing, we demonstrate a functional restoration of regenerative potential in aged SSPCs. These data identify aging-associated inflammation as the cause of SSPC dysfunction and provide mechanistic insights into its reversal.

All tissues are affected by aging, but diseases that weaken the skeleton constitute the most prevalent chronic impairment in the United States (1). Although skeletal diseases and conditions are seldom fatal, they can significantly compromise function and diminish quality of life. Perhaps most importantly, age-related changes in skeletal health may be traced back to the skeletal stem cell. Like other stem-cell pools, skeletal stem/progenitors are impacted by aging. For example, skeletal stem cells from people older than age 65 y, even if they are healthy, make less bone than stem cells from younger individuals, irrespective of sex (2). Instead of becoming bone-producing osteoblasts, skeletal stem cells from older people differentiate into fat-producing adipocytes (3), and this may partly explain why bone-forming ability declines with increasing age (3, 4).

Chronic inflammation in the elderly (“inflamm-aging”) is thought to be a major contributor to the decline in the regenerative capacity of the skeleton (5). In contrast to a well-balanced inflammatory response after trauma, which is crucial for successful bone repair (6), chronic unbalanced elevation of proinflammatory cytokines inhibits regeneration in a variety of other tissues (7). Its effect on the skeletal stem/progenitor cell (SSPC) is yet unknown. To address this knowledge gap, we hypothesized that chronic inflammation mediated by NF-κB activation—irrespective of age—contributes to a deterioration of the regenerative function of the stem-cell pool by inducing cellular senescence and decreasing SSPC number and function. Our data provide convincing evidence that pharmacologic inhibition of NF-κB activation leads to a functional rejuvenation of the SSPC pool, resulting in bone regeneration equal to that seen in young animals.

Results

Skeletal Stem Cell Frequency Decreases with Aging.

To investigate a clinically relevant age-associated effect on skeletal stem cell frequency and function, we first examined iliac crest bone graft (ICBG) samples from 36 patients (20 male, 16 female) with ages ranging from 24 to 89 y who underwent operative fixation of an upper- or lower-extremity fracture. FACS with CD271 as a human skeletal stem cell marker (8–11) revealed that SSPC frequency significantly declined with increasing age (Fig. 1 A and B). It is generally well accepted among orthopedic surgeons that fractures in elderly subjects heal more slowly and less reliably, and therefore we asked whether SSPC frequency correlates with time to bony union. We prospectively evaluated clinical and radiographic fracture union in this cohort and discovered that a lower SSPC number was associated with longer time to fracture union (Fig. 1C). To identify the mechanism involved in this decline in SSPC number and function, we chose a mouse model to further investigate the process of skeletal stem cell aging.

Fig. 1.

Skeletal stem/progenitor cell frequency declines in the aging patient. (A) FACS analysis of ICBG samples from 36 patients (20 male and 16 female) of varying ages revealed a significant (P < 0.05) negative correlation between age and SSPC number. (B) SSPC frequency is significantly decreased in patients older than 50 y of age (P < 0.05). (C) SSPC number is negatively correlated with time to bony union (P < 0.05). Green dots identify fractures that healed clinically and radiographically within 6 mo. Red dots mark patients with fracture union after 6 mo.

Aging Impairs Bone Regeneration.

To evaluate the extent to which the process of aging affects bone healing, we first employed a standardized tibial monocortical defect model in young (12-wk-old) and middle-aged (52-wk-old) male C57BL/6 mice. We analyzed bone healing by using histology, histomorphometry, and micro-CT (μCT). Two weeks after surgery, the injury sites were analyzed by histology. Whereas injuries in the young animals showed abundant woven bone within the defect site (Fig. 2 A and C), the injuries in the middle-aged animals exhibited a smaller area of woven bone, with bone formation predominantly between the cortical edges (Fig. 2 B and D). μCT imaging and 3D rendering confirmed this finding (Fig. 2 E and F). Histomorphometry using μCT demonstrated a smaller callus volume [bone volume/total volume (BV/TV)], trabecular number (Tb.N), and trabecular thickness (Tb.Th) and increased trabecular spacing (Tb.Sp; Fig. 2G). This first experiment demonstrated that 52-wk-old WT mice exhibit a phenotype of age-related impaired bone regeneration. Thus, we elected to use this age group for the subsequent experiments aimed at understanding the effect of aging on bone healing and SSPC function.

Fig. 2.

Aging results in impaired bone healing. (A and B) Histological sections of tibial monocortical defects 14 d after injury in 12-wk-old and 52-wk-old WT mice stained with Movat’s pentachrome. (C and D) Aniline blue staining depicting bone matrix deposition within the cortical defect. (E and F) Three-dimensional μCT reconstructions with surface rendering of bony regenerate (red) showing smaller callus size in the defect site of middle-aged animals. (G) Analysis of μCT data showing BV/TV (n = 6, P < 0.001), Tb.N (n = 6, P < 0.001), Tb.Th (n = 6, P < 0.05), and Tb.Sp (n = 6, P < 0.001) at postoperative day (POD) 7 and POD 14 in young and middle-aged mice. bm, bone marrow; c, cortical bone; is, injury site.

Aging Leads to a Decrease in SSPC Number.

The key ingredient to successful bone regeneration is the SSPC. To determine whether a decline in SSPC number is responsible for the impaired regenerative capacity of the aging skeleton, as seen in our human cohort, we used FACS with the inclusive SSPC marker LepR (12). CD45⁻CD31⁻Ter-119⁻LepR⁺ cells (LepR⁺ cells) comprise a heterogeneous mix of Sca-1⁺, PDGFRα⁺, CD51⁺, and CD105⁺ SSPCs (SI Appendix, Fig. S1), and Morrison and coworkers (12) demonstrated that LepR⁺ cells make up 0.3% of bone marrow cells; they differentiate into bone, cartilage, and fat in vivo and in vitro and, most importantly, give rise to bone postnatally and in response to injury. Bones from middle-aged mice contained significantly fewer LepR⁺ cells compared with bones from young mice (Fig. 3 A and B), and cfu assays confirmed this finding (Fig. 3C).

Fig. 3.

SSPC frequency declines with aging. (A) Representative FACS plots of young and middle-aged skeletal elements showing decrease in LepR⁺ SSPCs with aging. (B) Summary plot of FACS data demonstrating decrease in SSPC frequency in 52-wk-old (wo) mice (n = 5, P < 0.01). (C) cfu assay of young and middle-aged bone marrow showing representative colony staining and graphical depiction of quantification (n = 3, P < 0.0001).

Circulating Systemic Factors Lead to Skeletal Stem Cell Aging.

Having now established that SSPC frequency declines in mice similarly to our observation in humans, we next sought to identify the cause for this decline in stem cell number. Cell senescence, an irreversible arrest in cell division, has been associated with stem cell attrition in a multitude of other aged tissues (reviewed in ref. 13). Cell senescence is accompanied by a senescence-associated secretory phenotype (SASP), a local proinflammatory microenvironment, which acts on surrounding cells and inhibits their proliferation and cellular function (14). This paracrine effect of the SASP then induces senescence in cells within the immediate vicinity, commencing a vicious cycle that results in a functional decline of the entire tissue and organ (14, 15). We hypothesized that serum from middle-aged mice contains proinflammatory SASP factors and that this cytokine milieu leads to a functional decline of the skeletal stem cell. SSPCs from young (12-wk-old) mice were exposed to sera from middle-aged (52-wk-old) mice in vitro (Fig. 4A). Compared with the homochronic control group (young serum/young cells), which demonstrated a linear increase in cell proliferation over a 7-d time course, the heterochronic group (middle-aged serum/young cells) exhibited a functional arrest in cell division (Fig. 4B), suggesting the presence of an “aging” factor within the serum. In line with this assumption, senescence-associated β-gal (SA-β-gal) staining revealed a significant increase in SSPC senescence after heterochronic serum treatment (Fig. 4C). Next, we sought to identify whether the SSPCs treated with middle-aged serum started to express SASP factors themselves. Quantitative RT-PCR for SASP markers demonstrated that, relative to the homochronic control, interleukin-1α (Il1a), tumor necrosis factor-α (Tnfa), and nuclear factor kappa-light-chain-enhancer of activated B cells p65 (Rela) significantly increased, consistent with a SASP (Fig. 4D). Thus, aging is associated with a systemic cytokine milieu that directly leads to an arrest of cell division, induces cell senescence, and results in expression of SASP factors. We confirmed these in vitro finding by using an SA-β-gal assay (16) on SSPCs freshly harvested from young and middle-aged mice, and we discovered that middle-aged SSPCs were four times more likely to be senescent compared with young SSPCs (Fig. 4 E–G). qRT-PCR of the whole bone marrow confirmed an increase in the senescence markers Cdkn2a (p16) and Cdkn1a (p21) (17–19) (Fig. 4H).

Fig. 4.

Systemic cytokines are responsible for aging phenotype. (A) Schematic illustration depicting homochronic (young serum/young cells) and heterochronic (serum from middle-aged mice/young cells) in vitro culture conditions. (B) Cell proliferation assay of young SSPCs treated with sera from young or middle-aged mice showing an inhibitory effect of middle-aged sera on mitotic activity (n = 5). (C) Young SSPCs subjected to middle-aged serum exhibit more cell senescence at 4 d (n = 3, P < 0.01) and 7 d (n = 3, P < 0.001) as measured by SA-β-gal staining. (D) qRT-PCR shows induction of the expression of senescence-associated genes Il1a, Tnfa, and Rela in cells subjected to sera from middle-aged mice (n = 3, P < 0.05). (E–G) SA-β-gal staining of SSPCs from 12- and 52-wk-old mice showing increased senescence (arrowheads) in the aging animal. Quantification reveals a significant increase in senescence at 52 wk of age (n = 4, P < 0.01). (H) Consistent with the SA-β-gal staining, expression levels of Cdkn2a (p16) and cdkn1a (p21) were elevated in the middle-aged bone compartment (n = 7, P < 0.05). (I) Western blot for p-p65 reveals increased NF-κB activation in the middle-aged SSPCs (n = 3, P < 0.05). (J) Immunofluorescence of NF-κBp65 in young and middle-aged SSPCs revealed nuclear localization of NF-κBp65 in the middle-aged cells. (K) Quantification of NF-κBp65 (p65) nuclear localization demonstrates increased NF-κB activation in 52-wk-old SSPCs (n = 3, P < 0.001).

In response to the heterochronic serum treatment, we observed an increase in Il1a and Tnfa expression in the young SSPCs (Fig. 4D). These two proinflammatory cytokines lead to activation of NF-κB, a key mediator of inflammation (20). Therefore, we surveyed young and middle-aged SSPCs for NF-κB activation. Because phosphorylated NF-κBp65 (p-p65) is a prerequisite for nuclear localization and thus NF-κB activation, we first examined NF-κB activation by using Western blot for p-p65. We observed an increased p-p65/p65 ratio in middle-aged SSPCs (Fig. 4I). We then confirmed increased nuclear localization of NF-κBp65 by using immunofluorescence and detected a twofold increase in middle-aged SSPCs (Fig. 4 J and K). Together, these data confirm activation of NF-κB as a key inflammatory mediator in middle-aged SSPCs.

These data strongly suggest that the age-associated decline in SSPC frequency is caused by a systemic proinflammatory environment mediated through NF-κB and likely leading to increased cellular senescence.

NF-κB–Mediated Inflammation Induces SASP in Young Skeletal Stem/Progenitor Cells.

The age-associated cytokine profile leads to a systemic proinflammatory environment, inducing and potentiating NF-κB activation (Fig. 4) (13), and we hypothesize that this inflammatory milieu is responsible for the decrease in SSPC number and function. NF-κB is the fundamental transcriptional regulator of inflammation and controls the expression of genes encoding for proinflammatory cytokines, chemokines, and adhesion molecules (20). Proinflammatory stress and cell senescence activate Nfkb expression (20). We postulate that, with aging, SSPC frequency and function declines, and that this decrease in SSPC number and function is caused by an increased inflammatory microenvironment. To experimentally separate inflammation from aging, we used the Nfkb1^−/− mouse model, which has served as a model organism for low-level chronic inflammation in a plethora of studies involving liver regeneration (21), memory loss (22), and stress response (23). Deletion of NF-κB1 results in the activation of the NF-κB, which in turn leads to increased senescence and accelerated aging. Nfkb1^−/− mice lack the expression of the p105 and p50 NF-κB protein. The lack of these two NF-κB subunits results in the inability to form p50:p50 homodimers (repressor of proinflammatory gene expression) while still being able to generate RelA-containing NF-κB dimers (activators of proinflammatory gene expression), which results in an enhanced response to inflammatory stimuli (24, 25). Young, 30-wk-old Nfkb1^−/− mice housed in a pathogen-free environment exhibit hallmarks of premature aging with ataxia, kyphosis, sarcopenia, cardiac hypertrophy, and many other age-associated conditions that are related to an activated chronic inflammatory state (21), thus offering a valuable model organism to study regeneration in a model of low-grade inflammation in the absence of chronological aging.

First, we had to confirm that, in fact, 30-wk-old Nfkb1^−/− mice exhibit a proinflammatory cytokine profile similar to the one observed in middle-aged WT mice. qRT-PCR confirmed the expression of a SASP-like phenotype with up-regulation of Rela (NF-κBp65), Cyclooxygenase 2 (Cox2), Il6, Il10, Tnfa, Il1b, and Cdkn2a (p16) in bone marrow stromal and bone-lining cells (Fig. 5A). FACS analysis of 30-wk-old, 52-wk-old Nfkb1^−/−, and age-matched WT mice demonstrated fewer LepR⁺ SSPCs in the Nfkb1^−/− mice (Fig. 5B and SI Appendix, Fig. S2A), further confirming that inflammation, not aging, is driving the decline in SSPC number. We next analyzed the expression of SASP factors within freshly isolated LepR-positive SSPCs and found significantly higher expression levels of Il1b, Il6, and Rela in the SSPC population of 30-wk-old Nfkb1^−/− mice compared with age-matched WT mice (SI Appendix, Fig. S2B). Gene-expression analysis of the microenvironment, here captured in Q1/2 (CD31, CD45, and Ter-119–positive cells), revealed no changes in comparison with a similar cell population in age-matched WT mice (SI Appendix, Fig. S2B). We then further investigated the microenvironment of Nfkb1^−/− mice and separated the myeloid and lymphoid compartments by using well-accepted surface markers. Within the lymphoid compartment, Il1b and Rela were down-regulated and Tnfa was up-regulated in Nfkb1^−/− mice, whereas there were no significant differences for these cytokines in the myeloid compartment (SI Appendix, Fig. S2). These data suggest a shared contribution of the bone marrow and the SSPC compartment to the proinflammatory milieu to which the SSPCs then respond with increased SASP expression (SI Appendix, Fig. S2). Next, we sought to test whether this proinflammatory environment in the Nfkb1^−/− mice inhibits cell division, similar to what we had observed in the middle-aged WT mice. We performed a cell proliferation assay and detected an absence of cell proliferation in SSPCs from 30-wk-old Nfkb1^−/− mice, whereas age-matched WT SSPCs exhibited a linear increase in cell proliferation (Fig. 5C). An absence of cell proliferation can be attributed to apoptosis, senescence, quiescence, and premature differentiation. We used CellTrace to identify cell-cycle activity of WT and Nfkb1^−/− cells and demonstrated that Nfkb1^−/− cells are low-cycling as they are preferentially found in generation 1 (SI Appendix, Fig. S3 A and B). We then further analyzed the cells within generation 1 and showed that fewer than 1% of cells were apoptotic, without difference between the two groups (SI Appendix, Fig. S3C). Last, premature differentiation was ruled out by using cell-surface markers characteristic for stem cells. We observed no difference in PDGFRα⁺ Sca-1⁺ cell number between the two cell populations (SI Appendix, Fig. S3C).

Fig. 5.

Young Nfkb1^−/− mice mimic aging phenotype of SSPCs. (A) qRT-PCR of bone tissue from 30-wk-old Nfkb1^−/− and age-matched WT mice revealing proinflammatory SASP-like phenotype in Nfkb1^−/− mice (n = 3, P < 0.05). (B) Nfkb1^−/− mice show characteristic decline of SSPC number at younger age (n = 4, P < 0.05). (C) Proliferation assay of WT and Nfkb1^−/− SSPCs showing linear increase in cell number for WT cells and a steady state for Nfkb1^−/− cells (n = 4). (D) Alizarin red staining of mineralization assay for WT and Nfkb1^−/− cells in osteogenic media (OM) and GM. (E) Quantification of Alizarin red and alkaline phosphatase staining showing reduced osteogenic differentiation of Nfkb1^−/− cells (n = 4, ***P < 0.001). (F) Expression analysis for osteogenic genes of WT and Nfkb1^−/− cells treated with osteogenic differentiation media (n = 4, **P < 0.01 and ***P < 0.001).

Thus far, we have shown that the proinflammatory environment in Nfkb1^−/− mice leads to a decrease in SSPC number, an increase in cell senescence, and a decrease in cell division. Next, we investigated the trilineage potential of Nfkb1^−/− SSPCs. First, we exposed 30-wk-old SSPCs to osteogenic and growth media (GM) and assessed mineral deposition. Whereas WT cells formed confluent Alizarin red-positive mineral, Nfkb1^−/− cells deposited only small islands of Alizarin red-positive bone matrix when subjected to osteogenic differentiation media (Fig. 5D). Quantification revealed diminished mineral deposition and alkaline phosphatase activity in Nfkb1^−/− cells (Fig. 5E). This decrease in osteogenesis of the Nfkb1^−/− cells was confirmed by qRT-PCR for Runx2, osterix (Osx), and alkaline phosphatase (Alk Phos; Fig. 5F). Second, we analyzed chondrogenic and adipogenic differentiation in Nfkb1^−/− cells. Chondrogenesis, assessed by using micromass cultures, revealed smaller and disorganized micromasses without the characteristic hypertrophic center surrounded by less differentiated chondrocytes in the periphery, as seen in the WT micromasses (SI Appendix, Fig. S4A). Adipogenic differentiation revealed an increase in Oil Red O-positive cells. In addition, we observed a greater number of adipocytes in the tibial bone marrow of Nfkb1^−/− mice (SI Appendix, Fig. S4B). Finally, to identify whether the Nfkb1^−/− phenotype is truly related to the proinflammatory environment, we treated young WT SSPCs with serum from young Nfkb1^−/− mice. Serum treatment resulted in a uniform increase in proinflammatory cytokine expression in the young cells (SI Appendix, Fig. S5). The previously described heterochronic serum experiments in WT mice (Fig. 4) demonstrated an arrest in proliferation in response to treatment with serum from middle-aged serum (Fig. 4B). We sought to determine whether this same effect on proliferation can be observed when cells were treated with serum from Nfkb1^−/− mice. We used proliferating cell nuclear antigen (Pcna) gene expression to evaluate cell division in this experiment and observed a decrease in Pcna expression in the cells treated with Nfkb1^−/− serum (SI Appendix, Fig. S5). Collectively, these data confirmed that heightened inflammation, and not chronological aging, is responsible for a decrease in number and function of SSPCs.

Inflammation-Associated SSPC Decline Is Reversible.

Increased cell senescence (26), predominant adipogenic differentiation of SSPCs (27), and amplified apoptosis (28) are all associated with aging and result in an increased inflammatory response known as inflamm-aging (29, 30). First, we sought to identify whether the inflamm-aging phenotype is detectable in middle-aged mice on a systemic level. We performed multiplex analysis on serum from young and middle-aged mice and detected increases of the proinflammatory cytokines IFN-γ, TNF-α, and IL-6 in the aging animals (Fig. 6A). We confirmed this finding by qRT-PCR, which revealed up-regulated expression of the proinflammatory cytokines Rela, Tnfa, Il6, Il1a, and Il1b in SSPCs from middle-aged animals (Fig. 6B). These experiments confirmed that 52-wk-old WT mice exhibit an inflammatory cytokine profile consistent with inflamm-aging. If, in fact, chronic low-grade inflammation is responsible for the decrease in SSPC number, increased senescence, and overall decreased regenerative potential of aged animals, treatment with an antiinflammatory drug may overcome these negative effects. We therefore treated 52-wk-old mice with sodium salicylate, a low-grade antiinflammatory agent proven to inhibit NF-κB pathway activation (31–33), for 8 wk, and then harvested SSPCs and analyzed the expression levels of the aforementioned pro- and antiinflammatory cytokines. Salicylate treatment resulted in a decrease of the proinflammatory cytokines Rela, Cox2, and Il1b (Fig. 6B). This confirmed that the treatment protocol successfully repressed aging-induced chronic inflammation, as the expression level returned to levels measured in young animals.

Fig. 6.

Antiinflammatory treatment reverts inflamm-aging phenotype and increases SSPC pool. (A) Serum cytokine levels of IFN-γ (P < 0.05), TNF-α (P < 0.01), and IL-6 (P < 0.01) in young and middle-aged WT mice (n = 10). (B) Antiinflammatory drug treatment with salicylate reverses inflamm-aging phenotype (n = 6, P < 0.05). (C) Immunofluorescence for NF-κBp65 reveals nuclear localization of NF-κBp65 after treatment with middle-aged serum and a decrease in nuclear localization of NF-κBp65 in cells subjected to serum from middle-aged NSAID-treated mice. Quantification confirms increase in NF-κB activation with aging and decrease to juvenile levels after NSAID treatment (n = 4, P < 0.01). (D) Western blot for p-p65 confirming the decreased NF-κB activation in response to NSAID treatment (n = 3, *P < 0.05 and **P < 0.01). (E) SA-β-gal staining of young, middle-aged, and salicylate-treated middle-aged SSPCs show reversal of senescence phenotype in aging animals (n = 3, P < 0.05). (F) Graph showing SSPC frequency in young, middle-aged, and salicylate-treated middle-aged mice (n = 11, **P < 0.01 and ***P < 0.001). (G) Cfu-forming assay confirms increase of SSPC frequency after salicylate treatment (n = 3, P < 0.001). tx, treated; wo, weeks old.

As we postulated that the age-associated elevation of inflammatory cytokines results in increased NF-κB activation, we wanted to determine whether the observed systemic NSAID-induced reduction in cytokine levels resulted in decreased NF-κB signaling. We again treated young SSPCs with serum from young, middle-aged, and middle-aged NSAID-treated mice in vitro. This experiment revealed that serum from middle-aged mice treated with sodium salicylate did not result in nuclear localization of NF-κBp65, as shown by immunofluorescence and quantification (Fig. 6C). We confirmed this observation by using Western blot for p-p65, which was increased in the cells treated with middle-aged serum and returned to juvenile levels when treated with serum from NSAID-treated middle-aged mice (Fig. 6D).

Previously, we had shown that the proinflammatory environment had a direct negative effect on cell division (Fig. 4B) and induced senescence (Fig. 4 C and E–G). If salicylate treatment reduces this chronic inflammatory milieu, theoretically, cell senescence should decrease as a result. We performed an SA-β-gal assay with cells from young, middle-aged, and middle-aged salicylate-treated mice. Similar to our previous results (Fig. 4G), with aging, the senescent cell fraction increased (Fig. 6E); however, after salicylate treatment, the percentage of SA-β-gal–positive cells significantly decreased (Fig. 6E). This was confirmed in the heterochronic serum assay. Young SSPCs treated with serum from middle-aged antiinflammatory-treated mice exhibited a senescence phenotype similar to the young homochronic group (SI Appendix, Fig. S6). If senescence is reduced in response to salicylate treatment, does this lead to an increase in the SSPC frequency within the bone marrow? By using FACS for LepR, we showed that aging resulted in a significant decrease in SSPC number and that salicylate treatment partially recovered this loss of SSPC number (Fig. 6F). Cfu-forming assay analysis confirmed the reversal of decreased cfu formation after salicylate treatment (Fig. 6G). These analyses further confirmed that it is the inflammatory component of inflamm-aging and not the chronological aging component that leads to a decrease in SSPC number.

RNA Sequencing Analysis Reveals Rejuvenation of the SSPC Pool After Antiinflammatory Treatment.

Number and function of SSPCs are critical for successful bone regeneration. Having established that SSPC number declines with age and that this decrease can be halted by modulating the age-associated proinflammatory environment, we next sought to understand the effect of changes in the immune environment on the transcriptome of SSPCs. We first used an unbiased sequencing approach to compare the transcriptome of young and middle-aged SSPCs. Hierarchical cluster analysis revealed a stark separation between young and middle-aged SSPCs (Fig. 7A). In line with our FACS analysis showing a rescue of SSPC frequency, we observed a shift of the transcriptional profile of middle-aged SSPCs to that of the young SSPCs after antiinflammatory treatment. We then used gene-set enrichment analysis (GSEA) to further understand this potential rejuvenation of the SSPC pool (Fig. 7B). SSPCs from young animals enriched for genes associated with stemness, as did SSPCs from middle-aged antiinflammatory-treated mice, again supporting a reversal of the aging phenotype. Similarly, genes associated with osteogenesis and decreased adipogenesis were enriched in the young SSPCs, and again immunomodulation with an antiinflammatory agent led to an increased enrichment for these gene sets in middle-aged SSPCs (Fig. 7B). From these data, we conclude that, on a transcriptional level, modulation of the proinflammatory environment in the aging animal leads to a reversal of the aging phenotype.

Fig. 7.

RNA-seq analysis reveals a shift toward increased stemness and osteogenesis and decreased adipogenesis in middle-aged antiinflammatory-treated mice. (A) Heat map representing gene-expression values of the top 133 genes with a false discovery rate-adjusted P value less than 0.01 (q > 0.01) across all samples. Hierarchical clustering of these genes reveals that LepR⁺ SSPCs isolated from 52-wk-old antiinflammatory-treated mice cluster with 12-wk-old LepR⁺ SSPCs. Columns indicate single samples, and rows indicate genes. (B) GSEA plots demonstrate that young SSPCs and middle-aged treated SSPCs positively correlate with gene sets for stemness (BOQUEST STEM CELL UP), osteogenesis (SKELETAL DEVELOPMENT), and decreased adipogenic potential (TSENG ADIPOGENIC POTENTIAL DN).

Antiinflammatory Drug Treatment Abrogates the Aging Phenotype of SSPCs and Restores Regenerative Potential.

The previous set of experiments using unbiased sequencing strongly suggest that SSPC aging can be reversed. Next, we set out to interrogate the cellular function of the rejuvenated SSPC in an in vitro and in vivo environment. First, we evaluated the bone marrow compartment as a whole. Gene expression levels of Osx and Osteocalcin (Oc) significantly decreased in response to aging (Fig. 8A). We next analyzed bone marrow cells from middle-aged mice treated with salicylate, and here Osx, Oc, and Alp significantly increased compared with middle-aged untreated animals, and even reached levels equal to or higher than cells from young animals (Fig. 8A). To test the functional differentiation capacity of SSPCs, we exposed SSPCs to osteogenic differentiation media in vitro and then quantified mineralization as a readout for osteogenic differentiation. Young SSPCs demonstrated robust mineralization, whereas middle-aged SSPCs showed only some isolated foci of mineralization (Fig. 8B). SSPCs harvested from salicylate-treated middle-aged mice recovered their osteogenic function, which resulted in mineralization comparable to that of young mice (Fig. 8B). Aging has been associated with fatty degeneration of the bone marrow compartment (27). To directly test whether inflammation associated with aging can be attributed to this fatty degeneration, we subjected young and middle-aged SSPCs to adipogenic differentiation media. As expected, middle-aged cells were more prone to differentiate into adipocytes, as shown by increased Pparg and Fabp4 expression, and this was reversed in cells from NSAID-treated mice (Fig. 8C). We confirmed this expression pattern by using a functional adipogenic differentiation assay and again demonstrated an increase in adipogenesis with aging (Fig. 8D). However, SSPCs from middle-aged salicylate-treated mice exhibited less adipogenic differentiation (Fig. 8D), suggesting that inflamm-aging plays an active role in fatty degeneration of the bone marrow. We then evaluated tissue sections of young, middle-aged, and middle-aged NSAID-treated mice and observed a significant increase in adipocyte number with aging, which was reversed in mice treated with sodium salicylate (Fig. 8E).

Fig. 8.

SSPC function is restored in aging animals after antiinflammatory treatment. (A) qRT-PCR of SSPCs shows increased osteogenic gene expression after repression of inflamm-aging (n = 3, *P < 0.05, **P < 0.01, and ***P < 0.001). (B) Representative images and quantification of osteogenic differentiation analyzed by Alizarin red staining (n = 3, P < 0.001). (C) Adipogenic differentiation analyzed by qRT-PCR for adipogenic markers (n = 3, P < 0.05) and (D) Oil Red O staining (representative images and quantification; n = 3, **P < 0.01 and ***P < 0.001). (E) Histomorphometry for adipocyte number in the tibial bone marrow of young, middle-aged, and middle-aged NSAID-treated mice (n = 5, *P < 0.05 and ***P < 0.001). (F and G) Pentachrome histology of renal capsule (rc) transplants of 52-wk-old (wo) untreated and salicylate-treated SSPCs into young WT host mice. (F) Histomorphometry shows a decrease in regenerate size in middle-aged mice and restoration of regenerative function after salicylate treatment (n = 4, P < 0.05). bg, bone graft; tx, treated.

Next, we investigated whether the suppression of low-grade inflammation had a direct effect on homeostasis of the skeleton and bone regeneration in vivo. First, we analyzed bone homeostasis and performed μCT analysis of lumbar spine segments and humeri of young, middle-aged, and middle-aged salicylate-treated mice. BV/TV, Tb.N, and Tb.Th were significantly reduced in aged animals compared with young animals, whereas Tb.Sp was increased, confirming an age-appropriate bone loss (SI Appendix, Fig. S7). Middle-aged animals treated with salicylate showed a skeletal phenotype comparable to the untreated middle-aged control animals, indicating that salicylate treatment over a 12-wk time course in this age group did not affect bone homeostasis.

Finally, we set out to examine whether the decrease in cell senescence, increase in SSPC number, and shift of the osteogenic/adipogenic balance toward osteogenesis in response to antiinflammatory treatment resulted in a measurable proregenerative effect in vivo. To avoid masking effects of an endogenous healing response in a skeletal injury, we elected to use an ectopic transplantation model, as this represents the most stringent readout of in vivo bone formation of SSPCs (34). Here, we used a subrenal capsule transplantation assay to test whether suppression of chronic inflammation in the middle-aged animal successfully restores osteogenic capacity of the SSPCs in an in vivo setting. The subrenal capsule assay represents an ideal functional assay because cells transplanted between the renal capsule and the parenchyma receive sufficient blood supply and nutrients while being devoid of proosteogenic or prochondrogenic stimuli that would confound the readout (35). SSPCs from young, middle-aged, and middle-aged salicylate-treated mice were transplanted under the renal capsule. After 3 wk, histological staining and histomorphometry revealed a decrease in bone formation in the group containing middle-aged SSPCs compared with young SSPCs (Fig. 8 F and H). In stark contrast, cells transplanted from salicylate-treated middle-aged mice exhibited a robust osteogenic response, similar to that observed with juvenile SSPCs (Fig. 8 G and H), indicating that suppression of chronic inflammation in the middle-aged animal can restore the regenerative capacity of SSPCs.

Discussion

Chronic inflammation in elderly subjects has been linked to a variety of diseases (20). Here, we provide evidence that SSPCs respond to elevated levels of proinflammatory cytokines with increased senescence, decreased stem/progenitor cell number, and decreased functionality. In a pharmacological rescue experiment, we show that reduction of age-related chronic inflammation leads to a functional restoration of bone regeneration through a decrease in stem/progenitor cell senescence, increase in stem/progenitor cell number, and osteogenic gene expression (Fig. 9).

Fig. 9.

Schematic illustration of the cellular mechanism leading to impaired bone regeneration in elderly subjects. During aging, senescent SSPCs secrete SASP factors, which result in activation of NF-κB in adjacent SSPCs, inducing them to undergo senescence. This accumulation of senescent SSPCs leads to decreased SSPC self-renewal and proliferation, resulting in a decrease in overall number. Together with a decline in osteogenic differentiation, this leads to impaired bone regeneration in elderly subjects. NSAID treatment inhibits NF-κB activation, thereby blocking the effect of the SASP factors on surrounding SSPCs, which leads to a decrease in SSPC senescence, increase in SSPC number, improved osteogenic differentiation, and, finally, enhanced bone regeneration.

Chronic inflammation in elderly subjects has been attributed to a constant decay of extracellular macromolecules and intracellular organelles, resulting in an initiation and maintenance of an immune response (29). Degeneration of aged cells leads to secretion of reactive molecules, proinflammatory enzymes, and mediators that in turn will lead to further deterioration of the tissue, starting a vicious cycle that is characteristic of this chronic condition. Although all cells are affected by this proinflammatory milieu, the impact on the resident stem/progenitor cell, responsible for regeneration in response to injury, is likely the most detrimental, as it jeopardizes maintenance of tissue integrity. The toxic mediators within the tissue environment result in DNA damage, protein degradation, and organelle injury of the stem/progenitor cell, causing cellular senescence, which in turn results in further stimulation of the chronic inflammatory status through the cell’s secretory phenotype known as the SASP (29, 36). Here, we aimed at disrupting this vicious cycle by modulating the proinflammatory environment by using a mild antiinflammatory drug. As shown by serum cytokine analysis and gene-expression analysis of bone marrow cells, the described approach resulted in an inflammatory milieu comparable to that of a young animal, thus allowing us to study whether the detrimental effects of inflamm-aging on SSPCs can be reversed.

The effect of inflamm-aging on SSPCs can be broken up into cell-intrinsic and cell-extrinsic changes. Whereas the proinflammatory environment of the aged skeleton exerts its negative effect on the stem/progenitor cell through cell-extrinsic mechanisms such as modulation of signaling pathways [Wnt, Notch, BMPs (37)], cell-intrinsic mechanisms include self-renewal defects and induction of stress-induced pathways that lead to cellular senescence (13). Whereas cell-intrinsic defects such as senescence and self-renewal defect are considered irreversible, extrinsically induced defects may be reparable. As shown here, modulation of the inflammatory milieu by using a pharmacological compound resulted in a restoration of regenerative capacity. Interestingly, we observed that a short exposure of middle-aged SSPCs to a young systemic environment, as seen in the renal capsule transplantation assay, did not lead to a functional restoration. This suggests that the brief exposure to the young environment alone is not sufficient to functionally rejuvenate SSPCs or that the molecular microenvironment repressed the properties of the resident stem/progenitor cells, and, when modulated with an antiinflammatory drug, this repression was reversed, resulting in return of regenerative potential. This finding is supported by heterochronic transplantation assays and parabiosis experiments, which showed that the aging of a variety of stem cells was largely driven by cell-extrinsic mechanisms of the surrounding environment (38). In these assays, introduction of circulation from young animals restored the regenerative potential of the aged skeleton (39), indicating that the detrimental effects of aging on the progenitor pool may be reversible (40). A prime target for the mechanism of action observed in the parabiosis model is the inflammatory cytokine environment. Here, we provide strong evidence that inflammation, not chronological aging, is the main driver of SSPC dysfunction, and we demonstrate that NF-κB activation in young animals can mimic the aging phenotype in respect to SSPC number and function. This is further supported by the expression of the senescence-associated genes Cdkn1a (p21) and Cdkn2a (p16). Recent data suggest distinct roles for Cdkn1a and Cdkn2a. Kim et al. (41) and Baker et al. (42) suggest that expression of Cdkn1a is associated with senescence induced by chronological aging, whereas Cdkn2a expression is associated with stress-induced senescence (43). In our experiments, dysfunctional SSPCs from middle-aged mice exhibit heightened Cdkn2a expression (Figs. 4H and 5A), possibly indicating a stress response to the increased inflammatory environment rather than an age effect.

Our findings provide strong support for a central role of NF-κB as a mediator of inflamm-aging. Chen et al. (44) previously demonstrated a link between aging and NF-κB activation in a progeroid mouse model during bone homeostasis, supporting our findings during bone regeneration. During homeostasis, NF-κB–mediated inflammation promotes osteoclastogenesis (44) in the aging animal, which may contribute to the osteoporotic phenotype characteristic of aging animals. Although osteoclasts play an important role during regeneration, their main function is centered around the remodeling phase of repair, which occurs at later stages when bone matrix deposition has been completed. Because our focus aimed at the early stages of regeneration rather than remodeling and homeostasis, we did not further examine the connection between NF-κB and osteoclastogenesis.

There are caveats to this work. Usually, 52-wk-old mice are not considered aged yet; however, as demonstrated here, they clearly exhibit an increased proinflammatory cytokine profile consistent with the process of aging. We selected this age group rather than studying older animals for two reasons. First, SSPC number is already significantly decreased at 1 y of age, rendering transcriptional analyses of this cell population difficult. At 2 y, when mice are considered aged, SSPC number is very low, making it impossible to study the effects of inflamm-aging on this cell population. Second, we chose this middle-aged group to better represent the human age cohort most often debilitated by orthopedic conditions, the baby-boomer generation (45). It is this population that would most benefit from a translational approach that improves bone regeneration. Septa- and octogenarians, on the contrary, often undergo replacement procedures (i.e., hip hemiarthroplasty, shoulder arthroplasty) rather than being subjected to procedures that rely on functional bone regeneration. Thus, our study was executed by using 52-wk-old mice in an attempt to simulate the physiology of middle-aged orthopedic patients.

Sodium salicylate, an antiinflammatory drug with similar antiinflammatory potency as aspirin, exerts its effect through indirect inhibition of prostaglandin biosynthesis (reviewed in ref. 46). Although this nonspecific inhibitory effect undoubtedly reversed the inflamm-aging phenotype of the SSPCs in these murine models, future work will focus on a more targeted suppression of inflamm-aging using specific Cox2 inhibitors, small-molecule antagonists against epigenetic regulators of the innate immune system, and modulators of NF-κB activity.

In summary, we present compelling data that age-associated inflammation, regulated by NF-κB and triggered by increased cell senescence, leads to a functional decline of SSPCs, which can be overcome and reversed by suppression of inflammation by using a low-grade antiinflammatory drug.

Materials and Methods

Patients and Specimens.

All experiments involving human subjects were approved by the New York University (NYU) School of Medicine Institutional Review Board. After informed consent was obtained, specimens were obtained during routine ICBG harvest. One cubic centimeter of ICBG was immediately transferred into a microcentrifuge tube and placed on ice. Samples were dissociated and stained with antibodies against CD45 and CD271. FACS analysis was performed by using a BD LSRII cell analyzer with high-throughput sampler. CD45-negative and CD271-positive SSPCs were displayed as percentages of total single cell number (8–11). Patient data, including radiographic and clinical time to union, were extracted from a prospective patient database at the NYU Langone Orthopedic Hospital.

Isolation of SSPCs.

Tibiae and femurs were harvested as previously described (47). Dissociated cell samples were stained with antibodies against CD31, CD45, Ter-119, and LepR for purification by flow cytometry (Moflo XDP; Beckman-Coulter). CD31⁻CD45⁻Ter-119⁻LepR⁺ cells were identified as SSPCs (12, 34).

Isolation and Culture of SSPCs.

For the in vitro experiments, tibial and femoral SSPCs were isolated by centrifugation (48). SSPCs were resuspended in GM (DMEM containing 10% FBS and 1% penicillin/streptomycin; Thermo Fisher Scientific) and then plated in 75-mL tissue culture flasks. Media was changed every 2 d. All cellular assays described were performed with SSPCs at passage 1 from at least three different mice in three technical replicates.

Statistical Analysis.

A priori power analysis to obtain statistical significance (P = 0.05, power 80%) resulted in n = 4 for each group after body-size adjustment, expecting a 25% difference between the groups. Prism 7 (GraphPad Software) was used for statistical computations. A Student’s t test was used for all comparisons in which there were two groups; ANOVA followed by Holm–Šidák correction for post hoc testing was applied for analyses in which there were two or more comparisons being made. Error bars in the figures represent SEMs. An asterisk denotes a P value <0.05 unless denoted otherwise in the figure legend.

Further standard materials and methods including animals, antiinflammatory treatment, monocortical defects, renal capsule transplants, histology and histomorphometry, μCT analyses, multiplex ELISA, SDS/PAGE and Western blots, cfu assay, proliferation assay, osteogenic and adipogenic differentiation, RNA isolation and quantitative real-time PCR, in vitro homochronic and heterochronic serum treatment, identification of senescent cells, cell-cyle analysis, and RNA sequencing (RNA-seq) analysis are described in the SI Appendix.