Aging and the Emerging Role of Cellular Senescence in Osteoarthritis

Abstract

Objective.

The correlation between age and incidence of osteoarthritis (OA) is well known but the causal mechanisms involved are not completely understood. This narrative review summarizes selected key findings from the past 30 years that have elucidated key aspects of the relationship between aging and OA.

Methods.

The peer-reviewed English language literature was searched on PubMed using key words including senescence, aging, cartilage, and osteoarthritis, for original studies and reviews published from 1993-2023 with a major focus on more recent studies. Manuscripts most relevant to aging and OA that examined one or more of the hallmarks of aging were selected for further review.

Results.

All proposed hallmarks of aging have been observed in articular cartilage and some have also been described in other joint tissues. Hallmarks include genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, altered intercellular communication, disabled macroautophagy, chronic inflammation, and dysbiosis. There is evidence that these age-related changes contribute to the development of OA in part by promoting cellular senescence. Senescence may therefore serve as a downstream mediator that connects numerous aging hallmarks to OA, likely through the senescence-associated secretory phenotype (SASP) that is characterized by increased production of proinflammatory cytokines and matrix metalloproteinases.

Conclusions.

Progress over the past 30 years has provided the foundation for emerging therapies, such as senolytics and senomorphics, that hold promise for OA disease modification. Mechanistic studies utilizing physiologically-aged animals and cadaveric human joint tissues will be important for continued progress.

Introduction

The increasing incidence of osteoarthritis (OA) with age, coupled with the lack of effective non-surgical treatments to mitigate disease progression, make OA one of the most common causes of chronic pain and disability in older adults^{1, 2}. Similar to other chronic age-related conditions, OA is not an inevitable aspect of “normal” aging. For example, a study of 90-year-old participants in the Leiden 85-plus Study found 16% of the 82 individuals examined did not exhibit radiographic OA in their hands, hips, or knees³. Aging itself does not cause OA, but instead the changes that occur with age in cells, tissues, and organ systems contribute to the development of OA in conjunction with other risk factors such as prior joint injury, obesity, biological sex, and genetics. As an example of these complex interactions, Magnusson and colleagues used a large twin study to illustrate that the genetic contribution to knee replacement risk is consistent from age 50 to death in women, whereas heritability increased in importance for men starting at age 68⁴.

Over the past 30 years since Osteoarthritis and Cartilage was first published, the knowledge gained from studies in the basic science of aging, as well as studies specific to joint tissues, have significantly advanced our understanding of how aging contributes to the development and progression of OA. Thirty years ago, in a review of basic mechanisms of aging and OA written by David Hamerman⁵, much of the focus was on age-related changes in the extracellular matrix of articular cartilage. As an example, a study published in the first volume of Osteoarthritis and Cartilage in 1993 reported on age-related decreases in the content of aggrecan molecules containing the epidermal growth factor-like (EGF-like) domain⁶. Amongst the numerous changes to the cartilage extra-cellular matrix, the accumulation of advanced glycation end-products (AGEs) has been one of the most consistent age-related changes noted in cartilage. AGE accumulation alters the biomechanical properties of the tissue⁷ and is associated with reduced chondrocyte anabolic activity⁸.

More recent studies have broadened the scope of investigation into age-related changes present in other tissues and cell types of the joint and have incorporated a wide range of tools to investigate causal mechanisms. Over this time frame the terminology describing OA has changed to reflect a more nuanced understanding of the disease and these shifts in language closely parallel the emergence of similar concepts in aging research. For example, “degenerative joint disease” has fallen out of favor because it evokes a primarily passive mechanical concept of “wear and tear” without sufficient recognition of the biological processes that actively drive OA. In the aging field, the “geroscience hypothesis” proposes that there are a common set of active processes underlying most (if not all) age-related diseases that are consistent enough across tissue types and organ systems to be modifiable at an organism-wide scale for improved healthspan⁹. Similarly, the more consistent use of “osteoarthritis” in place of “osteoarthrosis” emphasizes that inflammation plays an important role in the disease. Heightened baseline inflammation has been recognized as one of the most consistent features of aging and the tight linkage of these processes has been termed “inflammaging”^10,11.

In this narrative review, we will examine the intersection of the aging “hallmarks” with features that emerge in joint tissues during OA. Of note, in part due to the widespread use of surgical waste tissue from end-stage OA, it is often challenging to parse whether these hallmarks emerge before initiation of OA or at some point during OA progression. We will then use cellular senescence, a key hallmark of aging that is driven by many of the other hallmarks, as a lens to examine how a better understanding of aging may translate into improved therapies for OA. Senescent cells exhibit the senescence-associated secretory phenotype (SASP). SASP factors include a host of proinflammatory cytokines and matrix degrading enzymes that are found in OA joints and directly contribute to OA pathogenesis. Finally, we will highlight several topics and experimental approaches that are likely to enhance rigor and accelerate research in this field.

Hallmarks of aging and the relation to the development of OA

The original nine hallmarks of aging described in 2013 included genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, and altered intercellular communication¹². These were revised in 2022 to add disabled macroautophagy, chronic inflammation, and dysbiosis¹³. There have been numerous studies published over the past 30 years that support the presence of these hallmarks in aging and/or OA joints. As consistent with the geroscience framework¹⁴, we propose that these hallmarks are more than just correlative but instead represent fundamental changes that enable a higher likelihood of OA onset and progression. The various hallmarks do not exist independently but rather are quite interconnected, with senescence likely positioned as a central mediator between processes that lead to initial cellular dysfunction and those that directly lead to OA pathology (Figure 1).

Figure 1.

The contribution of the hallmarks of aging to cellular senescence and the development of osteoarthritis (OA). Changes that occur within joint tissues during aging include the hallmarks of aging that lead to cellular senescence and the senescence-associated secretory phenotype (SASP). SASP factors promote chronic inflammation and degradation of joint tissues resulting in OA. Graphic created with Biorender.com.

Telomere attrition is a classic cause of in vitro senescence. One of the early studies on senescence in human chondrocytes showed that mean telomere length shortens with age and that shortened telomeres correlate positively to senescence-associated beta galactosidase¹⁵. A follow-up study demonstrated that telomerase overexpression can help avoid chondrocyte senescence, but only if cell cycle inhibitors are also inactivated¹⁶. Genomic instability in terms of mutational burden has not been extensively investigated in joint tissues, but chondrocytes did show higher chromosomal abnormalities with age and in OA tissue as compared to normal, with more abnormalities than peripheral blood leukocytes from the same subjects¹⁷. DNA damage burden is related to genomic instability, and both aged cadaveric donors and end-stage OA samples harbor DNA damage at levels high enough to induce senescence in chondrocytes^18–20.

Mitochondrial function is critical for healthy aging of joint tissues. Variations in mitochondrial DNA that can affect mitochondrial function have been associated with increased risk of OA. In order to determine a causal link between mitochondrial DNA haplogroups and the development of OA, conplastic mice were generated that had high risk mitochondrial DNA variants swapped with the low risk mitochondrial DNA variants and vice versa²¹. The mice generated with the low mitochondrial DNA risk variants developed less severe OA using a surgical model (destabilized medial meniscus, DMM) than the mice with the high-risk variants, providing strong evidence that differences in mitochondrial DNA that had been associated with changes in mitochondrial function can affect OA development. A consequence of mitochondrial dysfunction, which occurs in OA chondrocytes²², is increased levels of mitochondrial reactive oxygen species that can contribute to chondrocyte senescence as can other causes of redox imbalance²³. Selenoproteins are important intracellular antioxidants and a decrease in their levels seen in mice with genetic deficiency of selenophosphate synthetase 1 resulted in more severe age-related and surgically-induced OA that was associated with chondrocyte DNA damage, senescence, and increased levels of SASP factors²⁴.

Various epigenetic alterations, including those associated with aging, have been observed in OA²⁵. Sirt6 is a nuclear histone deacetylase that also functions in DNA repair. Loss of Sirt6 in mice results in a premature aging phenotype²⁶ while transgenic overexpression of Sirt6 promotes longevity²⁷. Recent studies have shown that Sirt6 loss in chondrocytes results in cell senescence and premature OA in mice^28,29. This was associated with increased STAT5 activity, which promotes proinflammatory signaling²⁸, and with decreased activity of the IGF-1-Akt pathway, which reduces anabolic activity²⁹. Age-related epigenetic changes may also interact with other OA risk factors such as genetics, for example by methylating promoters or increasing the accessibility of regulatory regions³⁰.

There is extensive evidence that a decline in autophagy occurs in aged and OA cartilage³¹. A recent study linked loss of the chaperone protein HSP90A to disrupted proteostasis and disabled macroautophagy³². FOXO proteins are important regulators of autophagy and their deletion in mice results in more severe age-related OA³³ while promotion of FOXO activity by TGF-β improves autophagy and reduces the susceptibility of chondrocytes to oxidative stress³⁴. There is also evidence for deregulated nutrient sensing in aging and OA including reduced activity of the nutrient sensor AMPK³⁵. Treatment of mice with metformin activates AMPK and is being studied for its ability to protect from certain age-related diseases in humans³⁶. Metformin reduced the severity of surgically-induced OA in mice³⁷ and could potentially serve as a disease-modifying drug in humans with OA³⁸.

Stem cell exhaustion is another hallmark of aging that may contribute to OA development. There is evidence to suggest that mesenchymal stromal cells can become senescent in OA and contribute to the pro-inflammatory and matrix destructive environment through production of SASP factors^{39, 40}. SASP factors promote chronic inflammation which is another hallmark of aging clearly associated with OA¹¹. Altered intercellular communication is hallmark of aging that has been studied at the systemic level through parabiosis experiments where pairs of young and old mice share circulatory factors through parabiosis surgery. OA was less severe in old mice paired with young mice compared to old mice paired with old mice and the reduction in OA severity was attributed to GDF-11 present in the circulation of the younger mice⁴¹. Another example of intercellular communication related to aging and OA is the promotion of cell senescence by extracellular vesicles produced by senescent chondrocytes⁴². Finally, dysbiosis is one of the new proposed hallmarks of aging based on a growing body of literature demonstrating that alterations in the gut microbiota contribute to aging and age-related conditions¹³. A role for gut microbial dysbiosis in the development of OA is not clear⁴³ although there is some evidence that increased intestinal permeability resulting in higher systemic endotoxin (lipopolysaccharide) may be involved⁴⁴.

Cellular senescence and the SASP

A primary beneficial function of senescent cells in tumor prevention is to elicit a robust inflammatory response by the release of SASP factors to clear persistently damaged cells. However, with increased rates of senescence induction and reduced clearance efficiency during aging, the continuous secretion of these same SASP factors becomes detrimental⁴⁵. There is a very strong correlation between the pro-inflammatory cytokines, chemokines, and matrix degrading enzymes found in the OA joint and those that constitute the SASP²³ (Table 1). In addition to cartilage, joint tissues such as synovium⁴⁶, bone⁴⁷, and meniscus⁴⁸ also harbor senescent cells. Approaches such as single-cell RNA-seq have started to unravel the cell types that are responsible for secreting factors into the OA joint⁴⁹. Overlapping this approach with rigorous definitions of senescence will likely enable a better understanding of the various sources of the SASP in joint tissues.

Table 1.

Selected SASP Factors Relevant to OA^*

Group	SASP Factors	Potential Role in OA
Pro-inflammatory cytokines	IL-1(α,β), IL-6, IL-7, IL-15, IL-17	Promote inflammation and expression of matrix degrading enzymes
Anti-inflammatory cytokines	IL-13	Reduce inflammation and are protective in OA
Chemokines	IL-8, GRO(α,β,δ), MIF, MIP(1α,3α), MCP(1,2,3,4), ENA-78, CCL5, SDF1	Promote OA by attracting inflammatory cells into the joint or through direct action on joint tissue cells
Growth factors and growth factor regulators	EGF, bFGF, HGF, TGF-β, IGF-2, IGFBP(2,3,4,6), LTBP-1, NGF, VEGF, CTGF, GDF15	Can promote OA or be protective depending on the cellular context including receptor expression
Matrix degrading enzymes and inhibitors	MMP(1,3,10,13,14), TIMP(1,2), PAI-1, tPA, uPA, cathepsin(B,D)	Degradative enzymes promote OA while TIMPs and PAI-1 may be protective.
Matrix proteins	Fibronectin, collagen (VI), tenascin, periostin, biglycan	Degradation and loss of matrix proteins occurs in OA. Periostin and certain matrix fragments, such as fibronectin fragments, can promote OA.
Other soluble factors	GM-CSF, HMGB-1, PGE2, HSP-90, 14-3-3 protein epsilon	Can promote OA (GM-CSF, HMGB-1, HSP-90, PGE2) or be protective (14-3-3 protein epsilon)

^*SASP factors selected from Coppe’ et al⁸⁰ and Basisty⁸¹.

bFGF, basic fibroblast growth factor; CCL, CC chemokine ligand; CTGF, connective tissue growth factor; EGF, endothelial growth factor; ENA, epithelial neutrophil-activating protein, GDF, growth and differentiation factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; GRO, growth-related oncogene; HGF, hepatocyte growth factor; HMGB, high mobility group box; HSP, heat shock protein; IGF, insulin-like growth factor; IGFBP, IGF binding protein; IL, interleukin; LTBP, latent transforming growth factor binding protein; MCP, membrane cofactor protein; MIF, macrophage inhibitory factor; MIP, macrophage inflammatory protein, MMP, matrix metalloproteinase; NGF, nerve growth factor; PAI, plasminogen activator inhibitor; PGE2, prostaglandin E2; SDF, stromal cell derived factor; t-PA, tissue-type plasminogen activator; TGF, transforming growth factor; TIMP, tissue inhibitor of metalloproteinases; u-PA, urokinase-type plasminogen activator; uPAR, u-PA receptor; VEGF, vascular endothelial growth factor.

Senescent cells are challenging to define, especially in vivo, due to the insufficient specificity of any particular senescence marker⁵⁰. The overuse of senescence-associated β-galactosidase (SA-β-gal) as a single marker has been a particular problem due to its lack of specificity, particularly in cultured cells⁵¹. One approach is to simultaneously analyze a large number of putative senescence markers and identify the sub-populations that are enriched with aging and depleted with the targeted removal of senescent cells; this paradigm was applied to bone and led to the prioritization of p16-positive; Ki-67 negative; and Bcl2-positive cells⁵². There is evidence that chondrocytes expressing senescence markers likely represent a heterogeneous population in that 5 of the 25 clusters identified by single-cell mass cytometry in OA samples contain p16-positive cells⁵³. A cell surface marker that holds promise for the identification of senescent chondrocytes is dipeptidylpeptidase-4 (CD26) which was associated with increased p16 and production of SASP factors⁵⁴. The recent “algorithmic assessment” paradigm posits that after an initial characterization, analyzing the SASP is likely to be the most informative for identifying the specific type of senescence present in a given biological sample⁵⁵.

Another challenge in identifying senescent cells is their dynamic nature and therefore the specific markers that are positive above background levels may change depending on the “stage” or “depth” of senescence. Using both physiological aging and “aging” through repeated passaging or extended time after a senescence-inducing stimulus, it was shown that some features of senescence such as activation of interferon signaling occur relatively late as compared to more commonly used markers such as p21⁵⁶. The widespread use of acute damage models to induce senescence and short time points (less than a week) to monitor cellular changes may provide a bias towards markers and states of senescence that are somewhat transient. While these model systems are valuable, care must be taken to distinguish an acute-phase response from a phenotype (senescence) that is defined at least in part by its persistence. For example, IL-1 has been used in many studies to stimulate a senescent-like phenotype in chondrocytes but this was found to be less potent as compared to direct DNA damage in causing sustained growth arrest and increased p21⁵⁷. Given that characterizing senescence under controlled in vitro conditions has limitations, it is no surprise that the complexity of physiological aging is even more challenging. A goal of the Cellular Senescence Network (SenNet) program funded by the US National Institutes of Health is to develop atlases of senescent cells from various tissues and disease states.

Targeting aging as therapeutic strategy for OA

All twelve of the aging hallmarks could be viewed as potential targets for the development of OA therapies, but over the last 10-15 years much of the translational focus has been on senescence. The burgeoning interest in cellular senescence as a phenotype can be traced to the intriguing finding that elimination of these cells in genetically-engineered mouse models can increase longevity and healthspan⁵⁸. This also supports the notion that the hallmarks of aging drive age-related conditions by promoting cellular senescence. Fortunately, the idea of targeting senescent cells for therapy has spurred more detailed and mechanistic investigations about the phenotype itself as a way to improve the likelihood of clinical success. Identifying small molecule interventions to mimic the results of the engineered mice became a major focus for aging researchers, as well as those seeking novel treatments for age-related diseases such as OA. Proposed senescence-targeting drugs and their molecular mechanisms have recently been reviewed in the context of OA therapeutics⁵⁹ and so will only be summarized here.

Senolytics are designed to cause selective cell death and remove the cells from tissues where they have accumulated. Senolytics typically inhibit specific cell survival proteins that are upregulated in senescent cells. By killing senescent cells, SASP factor production is reduced which is a major contributor to tissue dysfunction. In pre-clinical studies using the mouse ACLT model, intra-articular (IA) delivery of the senolytic UBX0101, which blocks the interaction of p53 and MDM2 allowing p53 to induce apoptosis, significantly reduced histologic features of OA, as well as pain behavior, in young 10-week-old animals⁶⁰. This intervention was less effective in 72-week old mice, but significant mitigation of post-traumatic OA was achieved using a combination of IA UBX0101 plus systemic delivery of Navitoclax as a second senolytic⁶¹. This latter finding may explain, at least in part, why IA UBX0101 given to older adult humans with symptomatic knee OA in a Phase 2 clinical trial was not effective in reducing pain or improving function after 12 weeks⁶².

Fisetin is a flavonoid that has antioxidant and anti-inflammatory properties that also appears to have senolytic activity through interference with cell survival pathways⁶³. Fisetin has been examined in preclinical models of osteoporosis and OA⁶⁴ and was found to reduce the severity of DMM-induced OA in mice⁶⁵. Fisetin, given as an oral supplement, was recently tested in humans in a phase 1/2 clinical trial for symptomatic knee OA but the results have not yet been reported (ClinicalTrials.Gov NCT04210986).

Rather than killing senescent cells, senomorphics target SASP factors such as cytokines or the pathways that promote SASP factor production such as JAK/STAT signaling or mTOR⁵⁹. In the mouse anterior ligament transection model, IL-17 was found to contribute to the development of senescence and the SASP⁶¹. In that study, intra-articular injection of an IL-17 antibody reduced senescence accompanied by a reduction in the severity of OA, suggesting that IL-17 inhibition could be considered as a senomorphic. Other senomorphics include the repurposed drugs metformin, which activates AMPK, and rapamycin, which targets mTOR. Metformin has been shown in pre-clinical models to reduce OA severity and is beginning to be examined as an OA therapeutic in humans³⁸. A retrospective study of individuals with type II diabetes found that those treated with metformin had a lower risk of incident OA, but not of joint replacement, compared to those treated with a sulfonylurea⁶⁶.

Because the toxicity of rapamycin limits its usefulness as a senomorphic and could make OA worse⁶⁷, rapamycin analogs (“rapalogs”) with less toxicity are being developed⁶⁸. Modulating levels of NAD⁺ is an intriguing approach for a senomorphic that would inhibit several aging pathways that promote OA (oxidative stress, DNA damage, cell senescence) and would promote activity of the sirtuins, including Sirt6⁶⁹. In an OA pre-clinical model, NAD⁺ levels were increased by inhibition of CD38, which mediates NAD degradation, resulting in a reduction in OA severity⁷⁰. As we learn more details of the mechanisms by which aging contributes to OA, additional therapies will be developed that hold promise for reducing OA progression in humans.

Challenges and promising areas for future research

In the 30 years since Osteoarthritis & Cartilage was first published, the global prevalence of osteoarthritis has more than doubled⁷¹. This has coincided with an increase in individuals within older age brackets: in 1993 only 3.9% of the population was ≥ 70 years old whereas this increased to 6.4% for 2023 and is estimated to jump to 12.3% in 2053⁷². With these demographic trends, it is imperative that age-related changes are considered across all aspects of OA research. One area that needs particular attention is the more consistent use of aged mice in analysis of genetically-engineered models or therapeutic interventions. Despite the many benefits of joint injury models such as the DMM model, there are numerous examples of divergent findings compared to OA that develops with physiological aging⁷³. One challenge is that the variability of spontaneous OA in aged mice requires large cohorts, which may be accentuated in genetically-mixed backgrounds such as those used to test treatments for healthspan effects in the National Institute on Aging Interventions Testing Program⁷⁴.

For studies with human joint tissue, increased use of cadaveric samples from wide age ranges are a valuable complement to the more widely accessible waste tissue from joint replacement surgeries. This facilitates the investigation of the potential initiating events that lead to OA by identifying changes that occur with age, such as increased DNA damage²⁰, before any macroscopic tissue damage is present. A challenge is the extensive overlap between SASP factors and the pro-inflammatory mediators and matrix degrading enzymes produced by cells within the OA joint that are not senescent. It will be important to make a distinction between senescent and non-senescent joint tissue cells in order to study the causal mechanisms distinct to the different ways that cells become dysfunctional and contribute to OA.

The role of senescence and SASP factors in mediating OA pain is another promising area for future studies. There is recent evidence that senolytics can reduce pain associated with OA in aged mice⁷⁵. SASP factors include the chemokine MCP-1 which is a mediator of OA pain in mice through binding to its receptor CCR2^76,77. It will also be important to determine if targeting senescence locally within the joint will be sufficient for treatment of OA in humans or if systemic targeting will be required. The answer may depend on the OA phenotype. A young to middle-aged adult with post-traumatic OA may only have senescent cells producing SASP factors in the affected joint while older adults with OA may have a greater contribution from senescence in multiple tissues outside of the joint.

While not a major focus of current translational research, the next 30 years could welcome more sophisticated genetic approaches to ameliorate or reverse age-related cellular changes. One example is the burgeoning field of partial reprogramming, which seeks to use temporary expression of rejuvenating factors to jump start the typically dormant regenerative potential of differentiated cells. Overexpression of six transcription factors for three days helped chondrocytes from older end-stage OA patients achieve a more “youthful” phenotype⁷⁸. Similarly, SOX5 was identified as a factor that overcomes senescence through the use of a CRISPR activation screen, and further experiments demonstrated that lentiviral overexpression of SOX5 in the joints of aged mice increased cartilage volume and decreased the level of catabolic features such as MMP-13⁷⁹.

With the progress made to date and the promising research focused on causal mechanisms underlying the link between aging and OA, it should not take another 30 years before we see the development of successful therapeutics that will help patients live more active and pain free lives.