Age-related alterations and senescence of mesenchymal stromal cells: Implications for regenerative treatments of bones and joints

The most common clinical manifestations of age-related musculoskeletal degeneration are osteoarthritis and osteoporosis, and these represent an enormous burden on modern society. Mesenchymal stromal cells (MSCs) have pivotal roles in musculoskeletal tissue development. In adult organisms, MSCs retain their ability to regenerate tissues following bone fractures, articular cartilage injuries, and other traumatic injuries of connective tissue. However, their remarkable regenerative ability appears to be impaired through aging, and in particular in age-related diseases of bones and joints. Here, we review age-related alterations of MSCs in musculoskeletal tissues, and address the underlying mechanisms of aging and senescence of MSCs. Furthermore, we focus on the properties of MSCs in osteoarthritis and osteoporosis, and how their changes contribute to onset and progression of these disorders. Finally, we consider current treatments that exploit the enormous potential of MSCs for tissue regeneration, as well as for innovative cell-free extracellular-vesicle-based and anti-aging treatment approaches.

The most common clinical manifestations of age-related musculoskeletal degeneration are osteoarthritis and osteoporosis, and these represent an enormous burden on modern society. Mesenchymal stromal cells (MSCs) have pivotal roles in musculoskeletal tissue development. In adult organisms, MSCs retain their ability to regenerate tissues following bone fractures, articular cartilage injuries, and other traumatic injuries of connective tissue. However, their remarkable regenerative ability appears to be impaired through aging, and in particular in age-related diseases of bones and joints. Here, we review age-related alterations of MSCs in musculoskeletal tissues, and address the underlying mechanisms of aging and senescence of MSCs. Furthermore, we focus on the properties of MSCs in osteoarthritis and osteoporosis, and how their changes contribute to onset and progression of these disorders. Finally, we consider current treatments that exploit the enormous potential of MSCs for tissue regeneration, as well as for innovative cell-free extracellular-vesicle-based and anti-aging treatment approaches.

Introduction: Aging and the burden of bone and joint degenerative disorders
Osteoarthritis, osteoporosis and systemic inflammatory conditions such as rheumatoid arthritis are among the most common and disabling of the musculoskeletal degenerative disorders. Osteoarthritis is an insidious and heterogeneous disease that affects the normal function of the synovial joints, and generally targets the knees, hands, hips and spine (Kloppenburg and Berenbaum, 2020). However, it is not a single disorder, but a group of disorders that are characterised by pain and loss of joint function, and which lead to impaired mobility of the patient. Osteoarthritis has a high economic burden on both individuals and society (Vincent, 2020). In 2017, osteoarthritis affected 303 million people globally, with 9.6 % of men and 18.0 % of women aged over 60 years showing symptomatic osteoarthritis. Aging and the increasing prevalence of obesity worldwide have now led to increased population risk of osteoarthritis, which is the fourth fastest increasing condition, behind diabetes, Alzheimer's disease and other musculoskeletal conditions (March et al., 2016).
Osteoporosis is a skeletal disorder that is characterised by compromised bone strength that predisposes a person to increased risk of bone fractures (Klibanski et al., 2001). It is the most common metabolic bone disease, and in the European Union, its estimated prevalence 27.6 million in 2010, with a 3− 4-fold higher risk in women than men (Hernlund et al., 2013). Based on a recently published report of the International Osteoporosis Foundation, the prevalence for osteoporosis in 2017 was 22.5 % for women and 6.8 % for men aged above 50 years in the EU6 countries, which includes the five largest European countries and Sweden (Borgström et al., 2020). Approximately 2.7 million fractures occurred in the EU6 countries that year (Borgström et al., 2020). Osteoporotic fractures are associated with pain, disability, increased risk of further fractures, and mortality (Haentjens et al., 2010). As well as the individual burden of disease, osteoporotic fractures result in enormous long-standing direct and indirect costs for healthcare systems (Tran et al., 2021). Indeed, higher costs and longer hospital stays were recently reported for elderly patients above the age of 65 years (Kim et al., 2021).
Mesenchymal stromal cells (MSCs) have remarkable regenerative abilities and have been proposed for use in a variety of therapeutic strategies for traumatic and degenerative defects of bones and joints. However, the irreversible termination of cell division and associated phenotypic changes associated with cellular aging and senescence can affect their regenerative properties. Here, we review the evidence relating to the age-related alterations of joint-tissue-resident MSCs, and discuss their possible contributions to the onset and progression of musculoskeletal tissue degeneration, with a focus on osteoarthritis and osteoporosis. We shed light on possible regenerative treatments of bones and joints using MSCs, taking into account age-related MSC changes, as well as novel cell-free and anti-aging therapies.

Risk factors for osteoarthritis and osteoporosis development and progression
Osteoarthritis is strongly linked to aging (Loeser, 2011). However, aging alone is not the only cause of osteoarthritis; rather, it can promote the development of osteoarthritis. Other risk factors include obesity and joint trauma (Fig. 1). Obesity is associated with the incidence and progression of osteoarthritis in both weight-bearing and non-weight-bearing joints (March et al., 2016). This becomes a vicious circle, where people with osteoarthritis avoid physical activity due to joint pain, which leads to further weight gain, loss of joint stability and muscle weakness (sarcopenia). Joint trauma has become particularly important over the past few decades. The combination of widespread sports and recreational activities in people of all ages and the aging of the population in developed countries has led to increased joint injuries (March et al., 2016;Roos, 2005). In association with joint structure, joint biomechanics also has an important role. Excessive loading of apparently normal joints can result in primary osteoarthritis, while normal loading in structurally abnormal joints results in secondary osteoarthritis. Other risk factors for osteoarthritis include sedentary lifestyle, chronic postural defects, low bone density and genetic risk factors (Di Nicola, 2020;Reynard and Barter, 2020).
Several aspects have been identified that affect bone quality and quantity, and that can result in osteoporosis and increased risk of fractures. Clinical risk factors include low body mass index, smoking, alcohol intake, and family history of bone fractures and hip fractures. Moreover, secondary causes for bone loss (e.g., glucocorticoid treatment) and co-morbidities (e.g., diabetes, inflammatory diseases) can lead to systemic bone loss and increased risk of osteoporotic fractures (Kanis et al., 2008). Furthermore, a clear association between aging and risk of fractures has been shown (Kanis et al., 2008).
Under the influence of local and systemic factors, bone can undergo age-dependent modifications, in terms of its structure, composition and function. Both women and men experience bone loss during aging, which is reflected in continuous decreases in bone mineral density and femoral strength (Keaveny et al., 2010). Bone remodelling shifts towards greater bone resorption and less bone formation. The typical changes in the bone microstructure during aging include loss of trabecular number, trabecular and cortical thinning, and increase in cortical porosity (Corrado et al., 2020). In this context, oestrogen positively affects bone metabolism in several ways. This sex hormone increases calcium resorption and decreases renal calcium excretion (Ji and Yu, 2015). Moreover, oestrogen directly affects bone via oestrogen receptor-alpha (ERα) and -beta (ERβ). The osteoprotegerin/ receptor activator of NF-кB ligand (OPG/RANKL) system appears to be one of the most important regulators of bone metabolism and postmenopausal bone loss. Oestrogen deficiency leads to an imbalance in the OPG/RANKL system and consequently to high bone resorption and bone loss (Streicher et al., 2017).
Moreover, osteosarcopenia, which defines the combination of bone and muscle loss, is a common phenomenon in geriatric patients, and it is associated to falls and the consequent osteoporotic fractures (Inoue et al., 2021). Thus, next to female gender and glucocorticoid treatment, age appears to be one of the most important risk factors for osteoporosis-related fractures.

The role of inflammation
Inflammation has been suggested as one of the seven mechanistic 'pillars of aging' and age-related diseases, with the other six being adaptation to stress, macromolecular damage, epigenetics, metabolism, proteostasis, and stem cells and regeneration (Kennedy et al., 2014). These six other pillars are interconnected, and they all converge on the seventh, inflammation. Age-associated low-grade inflammation is known as 'inflammaging' (Franceschi et al., 2000), which refers to the long-term effects of chronic physiological stimulation of the innate immune system, where macrophages have a major role (Franceschi et al., 2018). Inflammaging occurs in the absence of infection, and is primarily driven by endogenous signals (Franceschi et al., 2018).
Traditionally, osteoarthritis has been classified as a noninflammatory, or degenerative, arthritis, or simply as a 'wear and tear' disease. However, the non-inflammatory classification of osteoarthritis has become less clear (Sokolove and Lepus, 2013), as inflammaging has also been recognized in the pathophysiology of osteoarthritis (Berenbaum, 2013). In particular, synovitis (i.e., inflammation of the tiny membrane wrapped around the synovial joints) has been reported for a significant proportion of patients with primary osteoarthritis. Initial inflammation and local homeostatic imbalance between cartilage and the synovial membrane have key roles in early osteoarthritis. In clinical osteoarthritis, progressive joint failure results from the interplay between joint damage, which at some point fails to be resolved, and immune responses to the perceived damage, which persists as a state of chronic inflammation (Sokolove and Lepus, 2013).
Recently, Josephson et al. showed that rather than actual chronological aging, it is inflammaging that is responsible for the decline in the numbers and functions of skeletal stem cells (SSCs) and for delayed bone healing (Josephson et al., 2019). They first showed direct correlation between stem cell numbers and time to reach bone fracture union in a human patient cohort. In the animal model that reproduced this age-associated decline in bone healing, increased cellular senescence caused by systemic and local proinflammatory environments was identified as the major contributor to the decline. By decoupling age-associated systemic inflammation from chronological aging in transgenic nuclear factor kappa B1 (NF-κB) knock-out mice, it was demonstrated that the elevated inflammatory environment (rather than the chronological age) was responsible for the decrease in SSC numbers and functions. Moreover, they provided evidence of functional rejuvenation of aged SSCs by pharmacological inhibition of NF-κB (Josephson et al., 2019).

The interplay between joint tissues in degenerative disorders
Although the hallmark of osteoarthritis is breakdown of cartilage and remodelling of the underlying bone (Sacitharan and Vincent, 2016), it is a disease of the whole joint, and thus also involves the synovium, menisci, ligaments and neural tissue (Goldring and Goldring, 2016;Lories and Luyten, 2011). Indeed, it is still not clear which tissue is the initiator of age-related joint degeneration. Based on the roles of the synovium, muscles and ligaments in biomechanics and proprioception, their contributions to early osteoarthritis have probably been underestimated. The progression of osteoarthritis includes deterioration of several joint tissues, with inflammation of the capsule-bursa tissue, changes to the synovial fluid, erosion of the cartilage, osteochondral inflammatory damage, and bone distortion (Di Nicola, 2020).
The pathophysiology of osteoarthritis has been researched intensively over the last decade. Initially, these studies focused on the mechanisms within the cartilage that contribute to the wear and tear in osteoarthritis. The mechanisms of osteoarthritis development that have been proposed include inflammation (Attur et al., 2002;Bonnet and Walsh, 2005;Chen et al., 2017;Houard et al., 2013;Loeser, 2006), angiogenesis (Ashraf and Walsh, 2008;Bonnet and Walsh, 2005) and oxidative stress (Zahan et al., 2020). More recently, however, the focus has shifted to subchondral bone as the initiator of the osteoarthritis pathology (Goldring and Goldring, 2016;Lories and Luyten, 2011). In terms of associations with cartilage damage, these have been defined as changes that occur at the levels of bone remodelling (Day et al., 2001;Thambyah and Broom, 2009;Zupan et al., 2013), collagen type I synthesis (Blair-Levy et al., 2008), bone marrow lesions ) and bone attrition (i.e., specific type of subchondral bone loss that affects the shape of the bone, which results in flattening or depression of the surface) (Neogi et al., 2009). At the molecular level, two targets in particular have been suggested, based on their known roles in skeletal development: bone morphogenetic protein (BMP) (Neogi et al., 2009) and Wnt/β-catenin signalling (Luyten et al., 2009). The BMP and Wnt/β-catenin antagonists gremlin1 (GREM1), frizzled-related protein (FRZB) and Dickkopf 1 homolog (DKK1) have been identified as factors secreted by the articular cartilage that can inhibit chondrocyte hypertrophy, and their expression is inversely correlated with the levels of cartilage degeneration in osteoarthritis (Leijten et al., 2012).

Current treatment strategies for osteoarthritis and osteoporosis
Despite intense research into the pathophysiology of osteoarthritis, no concrete targets have been identified to date for the prevention of the degenerative changes, and no disease modifying drugs have been approved. In total joint arthroplasty, the affected joint is surgically replaced by an artificial prosthesis, and this is currently still the most effective osteoarthritis treatment for the relief of pain and to improve joint function (MacInnes et al., 2012). The main obstacles for non-surgical treatments of osteoarthritis are the difficulty in detection of the disease early after onset, the heterogeneity of the disease, and the complex overlap among different osteoarthritis phenotypes. Additionally, it is a challenge to develop a targeted drug with high activity, specificity, potency and bioavailability in the absence of toxicity for long-term use in predominantly older adults (Oo et al., 2018). In recent years, a particular focus has been for treatments that can relieve the pain and improve joint function following a patient-centred approach, while also taking into account co-morbidities (Bannuru et al., 2019).
In contrast to the pain that drives those with osteoarthritis to seek medical attention, people with osteoporosis often seek urgent medical attention due to falls and fragility fractures. Osteoporosis is the most common bone disease that predisposes to fragility fractures. However, fractures of the hip and vertebrae have also been shown to be prevalent, and as high as 33 % for older adults with osteoarthritis (Cevei et al., 2013). The management of osteoporosis-related fractures depends greatly on the anatomic region affected. For hip fractures, which are among the most common, early orthopaedic care and most often surgical intervention are crucial for the prevention of complications, and for the reduction of morbidity and mortality (Warriner et al., 2011). Although there is some debate as to which implant should be used for certain types of fractures, there is general agreement that fracture setting should be carried out rapidly to allow early mobilisation. Displaced femoral neck fractures are usually treated with total or partial joint replacement, undisplaced femoral neck fractures are treated with screws or sliding plates, and trochanteric fractures are treated with intramedullary devices or dynamic hip screws.
In contrast, the treatment of osteoporosis-related vertebral fractures remains somewhat controversial, and there is still no gold standard. Conservative treatment with pain medication, short-term bed rest, antiosteoporotic drugs, bracing and exercise appear to be the most common courses for primary management. There are numerous studies on surgical interventions, such as vertebroplasty and kyphoplasty, but to date there is still no convincing evidence to support routine surgical management (Warriner et al., 2011). However, around 15 %-35 % of patients have unstable fractures that lead to kyphosis or chronic non-union, or they develop neurological deficits, and consequently require surgical intervention. The implants and techniques used have evolved tremendously over the years, and these now allow minimal invasive and safe surgical stabilisation (Warriner et al., 2011). Despite this, prevention and management of osteoporosis are crucial to reduce the risk for osteoporosis-related vertebral fractures.
The management of osteoporosis includes physical exercise, prevention of falls, calcium and vitamin D substitution, reduction of risk factors, and pharmacological treatment to prevent osteoporotic fractures. All of the drugs available are known to reduce fracture risk at the lumbar spine, and most of them are also effective against peripheral fracture risk (Kocijan et al., 2020). Anti-osteoporotic drugs have also been reported to reduce mortality after hip fracture (Behanova et al., 2019). Current osteoporosis treatments are either anti-resorptive, osteo-anabolic or both. The anti-resorptive bisphosphonates were introduced many years ago, and to date they remain the first choice agents for treatment of osteoporosis (Kanis et al., 2019). Bisphosphonates are known to increase bone mineral density at the lumbar spine and hip, and to reduce the risk of peripheral and vertebral fractures. Similar efficacy for fracture risk reduction was shown for the anti-RANKL antibody denosumab, which also provides long-standing increases in bone mineral density and positive effects for bone microstructure (Kocijan et al., 2020). Teriparatide is a recombinant parathyroid hormone, and it is available for the therapy of severe osteoporosis. Due to its osteoanabolic activity, the beneficial effects of teriparatide on bone microstructure (Weigl et al., 2021), and thus the reduction in risk of vertebral fractures is higher compared to the use of anti-resorptive drugs (Kendler et al., 2018). The dual-acting anti-sclerostin antibody romosozumab was approved recently for treatment of severe, postmenopausal osteoporosis, which has combined osteo-anabolic and anti-resorptive activities (Kocijan et al., 2020).

Mesenchymal stromal cells: Chronological aging and the role of cell senescence
Mesenchymal stromal cells include multipotent progenitor cells that build up the joints during embryonic development and are retained in adult life in several connective tissues, including bone and bone marrow, skeletal muscle, adipose tissue, skin and others (Čamernik et al., 2018). By consensus, MSCs are defined as plastic-adherent cells that can undergo osteogenic, chondrogenic and adipogenic differentiation, and that express a defined pattern of surface antigen markers (Viswanathan et al., 2019). Mesenchymal stromal cells are isolated using various protocols, which can include enzymatic digestion of the tissue, density gradient centrifugation, cell isolation based on plastic adherence, and cell sorting based on specific surface antigen expression (e.g., CD73, CD90, CD105, CD164, CD271, podoplanin [PDPN], Stro-1 antigen) or lack of expression (e.g., CD14, CD19, CD45, CD146) (Čamernik et al., 2019;Chan et al., 2018;Lv et al., 2014) (Table 1). Mesenchymal stromal cells are heterogeneous in nature, whereby different markers can identify different subpopulations (Chan et al., 2018;Maijenburg et al., 2012;Zha et al., 2021); indeed, some studies have proposed the abandonment of the term 'MSC' in favour of the use of tissue-specific nomenclature (Bianco and Robey, 2015;Sipp et al., 2018). Furthermore, the properties of MSCs are known to vary between individual clones from the same donor, between different donors, and between different tissue sources (Elahi et al., 2016;Merryweather-Clarke et al., 2018;Mohamed-Ahmed et al., 2018;Muraglia et al., 2000;Reumann et al., 2018). Comparative studies have indicated specific differences in the developmental potential of MSCs from bone marrow, adipose tissue, synovium and perinatal tissues (Brennan et al., 2017;Huang et al., 2005;Mohamed-Ahmed et al., 2018;Topoluk et al., 2017). For instance, it has been shown that only bone-marrow MSCs have the potential to form a haematopoietic niche via a vascularised cartilage intermediate in vivo, as compared to MSCs from white adipose tissue, umbilical cord and skin (Reinisch et al., 2015).
As for any other cells in the body, MSCs are subject to age-related changes due to both cell-intrinsic and cell-extrinsic mechanisms of aging (Kassem and Marie, 2011;Marie, 2014). Intrinsic mechanisms of aging can cause cellular dysfunctions that are not tissue specific, which include telomere shortening, accumulation of oxidative damage, impaired DNA repair, altered gene expression profiles, and epigenetic alterations (Marie, 2014;Meng et al., 2020;Yu et al., 2018). Extrinsic aging mechanisms are related to changes in the systemic and local tissue microenvironments, and include changes in expression and signalling of hormones and growth factors (Nehlin et al., 2019), structure and composition of tissues (Boskey and Coleman, 2010;Burr, 2019), and intercellular communication (Genetos et al., 2012;Tiede-Lewis and Dallas, 2019). MSCs that reside in connective tissues other than bone and joint tissues, such as adipose tissue, can be differently affected by aging (Dufrane, 2017;Mantovani et al., 2014). Importantly, chronological aging of MSCs in vivo might have some features that overlap (i.e., are in common with) and some that do not overlap with in-vitro aging, which is often used as a study model for senescence-associated changes in MSCs (Ganguly et al., 2017;Wagner et al., 2009). In the next sections, we focus primarily on the evidence regarding the changes in bone-marrow MSCs that result from chronological aging.

Changes in MSC numbers, population structure and cellular 'fitness'
Overall, it is believed that the numbers, proliferation and regenerative capacities of MSCs are negatively affected by chronological age. Earlier studies of bone-marrow MSCs in rodents were reviewed by Kassem and Marie (2011), who reported variable findings with regards to the numbers of colony forming unit fibroblasts (CFU-F) that stained positive for alkaline phosphatase (as a surrogate measure for numbers of MSCs) with increasing chronological age. Some studies have reported age-related decreases in MSC numbers, while others have reported increases, or no evident age-related changes, or also bi-phasic responses, with an increase in MSCs numbers from childhood to middle age, followed by a decrease (Kassem and Marie, 2011). Similarly, earlier studies on human bone-marrow MSCs reported variable findings (D'Ippolito et al., 1999;Kassem and Marie, 2011;Stolzing et al., 2008), which might have been due to the different methods used for tissue harvesting and cell isolation. However, several later studies have supported a decrease in MSC numbers with increasing donor age (Siegel et al., 2013;Stolzing et al., 2008).
Furthermore, changes in the composition of bone-marrow MSC populations with increasing age have been reported. Maijenburg et al. isolated bone-marrow mononuclear cells by density gradient centrifugation, followed by expansion of plastic-adherent MSCs. They showed that the MSC subset distribution characterised by expression of CD271 and CD146 correlated with donor age, whereby CD271 + CD146-cells were most common in adults and CD271 + CD146 + cells were dominant in children; further, foetal CD271-CD146 + cells represented a third MSC subset with colony-forming activity (Maijenburg et al., 2012). Similarly, Siegel et al. reported that CD71 + , CD146 + and CD274 + bone-marrow MSCs isolated by density gradient centrifugation and plastic adherence correlated negatively with age of the bone-marrow donor (Siegel et al., 2013), and Josephson et al. reported a decline in CD271 + bone-marrow progenitors with increasing donor age in dissociated iliac crest bone graft samples (Josephson et al., 2019). Supporting such MSC population changes, Dusher et al. conducted single-cell transcriptional analysis to determine the effects of aging on mouse adipose MSC populations (Duscher et al., 2014). Adipose tissue was digested with collagenase, with the resulting stromal vascular fraction cultured, and the primary MSCs obtained sorted as single cells with the surface marker profile of CD45-CD31-CD34 + . They showed aged-related depletion of a subpopulation that was characterised by a pro-vascular transcriptional profile, as well as compromised ability of aged MSCs to support vascular network formation in vitro and in vivo.
In addition to the decline in MSC numbers and changes in MSC subpopulations, other studies have also reported declines in general MSC 'fitness' with increasing chronological age; this might have important implications for MSC-based therapies, especially when ex-vivo cell expansion is required (Ganguly et al., 2017). Stolzing et al. isolated human bone-marrow MSCs using density gradient centrifugation and plastic adherence. They showed a decline in proliferation capacity and stress responses, increased oxidative damage, and higher levels of reactive oxygen species (Stolzing et al., 2008). Similarly, Li et al. reported that aged rat adipose tissue MSCs isolated by collagenase digestion, followed by density gradient centrifugation and plastic adherence were more susceptible to damage by reactive oxygen species, which resulted in decreased cell adhesion and increased cell apoptosis in the aged MSCs (Li et al., 2014). These aged MSCs also had lower survival rates upon transplantation in vivo compared to young MSCs. Furthermore, in-vivo aging of rat bone-marrow MSCs was associated with significant reductions in cell migration (Geißler et al., 2012).

Gene expression changes
Baker et al. and Meng et al. reviewed changes in gene expression in MSCs with aging, and underlined the difficulty in drawing common conclusions (Baker et al., 2011;Meng et al., 2020). Wagner et al. used density gradient centrifugation and plastic adherence to isolate bone-marrow MSCs that they compared across 12 donors of different ages. They reported that 67 genes were up-regulated with age, with most of these related to regulation of the extracellular matrix, mesoderm development, synaptic vesicle endocytosis, and chemotaxis. They also identified 60 genes that were down-regulated with age, which were associated with DNA repair, mitosis and transcriptional regulation (Wagner et al., 2009). Yu et al. evaluated age-related changes in rhesus macaque bone-marrow MSCs that they isolated by density gradient centrifugation and plastic adherence, and they showed decreased heat shock protein expression with age, as well as for changed expression of various microRNAs (Yu et al., 2011). Wu et al. evaluated microarray data for bone-marrow MSCs isolated from femoral heads of four elderly and five middle-aged donors, and identified six hub genes and 11 transcription factors related to adherent junctions, DNA damage induced by oxidative stress, telomere attrition, MSC differentiation and epigenetic modulation that were key in aging (Wu et al., 2019). Alves et al. analysed bone-marrow MSCs isolated by plastic adherence from 61 donors to define a gene signature for donor age (Alves et al., 2012). They also noted considerable variations between the donors for the biological characteristics of the cells, which did not correlate with donor age. However, a number of genes that correlated with donor age were identified, with many of these related to neuronal functioning and growth factor activity in extracellular regions. Bustos et al. compared gene expression profiles of freshly sorted bone-marrow MSCs from young and aged mice, where the MSCs from aged mice showed decreased expression of several metalloproteinase genes, and genes involved in osteogenic differentiation, MSC activation, anti-inflammatory responses and cell migration, without any increase in markers of cell senescence (Bustos et al., 2014).
Alterations in the expression of non-coding RNAs with MSC senescence and aging have also been reported. Pandey et al. carried out microRNA profiling of bone-marrow and adipose MSCs from 16 donors, which were isolated from density-gradient-separated bone-marrow cells and collagenase-digested adipose tissue cells using plastic adherence (Pandey et al., 2011). They showed age-related differences for 45 microRNAs in bone-marrow MSCs and 14 microRNAs in adipose MSCs. For both of these cell sources, more microRNAs were down-regulated in the MSCs from the older donors. Their pathway analyses then suggested that these microRNAs are involved in the regulation of cell proliferation and inflammation, and significant age-related decreases in the expression of components of the MAPK/ERK and NF-κB pathways were seen. Similarly, Yoo et al. carried out in-vitro studies on bone-marrow MSCs from a commercial source, and identified 43 microRNAs that were regulated by senescence (Yoo et al., 2014). Davies et al. isolated extracellular vesicles from bone-marrow interstitial fluid of young and aged mice, and reported that specific aging-related microRNAs can affect the properties of the bone-marrow microenvironment through shuttling within extracellular vesicles (Davis et al., 2017). Li et al. analysed freshly sorted Sca-1 + CD29 + CD45-CD11b-mouse bone-marrow MSCs and identified 92 differentially expressed long non-coding RNAs between young and aged mice . They reported that one long non-coding RNA, known as Bmncr, is a key regulator of the age-related alterations to the osteogenic niche and the cell-fate switch of bone MSCs.
Bmncr maintains the expression of the extracellular matrix protein fibromodulin, and activates the BMP2 pathway. Restoring Bmncr levels in human bone-marrow MSCs reversed the age-related switch between osteoblast and adipocyte differentiation .

Epigenetic changes
Cakouros and Gronthos recently reviewed the epigenetic signatures associated with bone-marrow MSCs and haematopoietic stem-cell aging (Cakouros and Gronthos, 2019). They underlined some common changes between the different cell types, which included increases in histone acetylation, decreases in DNA methylation and hydroxymethylation, and genomic changes in H3K27me3 (i.e., the epigenetic modification that indicates the histone H3 protein trimethylated on lysine 27). In an earlier study, Bork et al. investigated differences in DNA methylation between bone-marrow MSCs isolated from young and old donors by density gradient centrifugation and culture on fibronectin-coated flasks, as well as long-term cultured bone-marrow MSCs (Bork et al., 2010). While the overall methylation patterns were maintained, they noted that upon aging, 295 CpG islands were hypermethylated and 349 CpG islands were hypomethylated. Gene ontology analysis revealed the hypermethylated genes were significantly over-represented in various signal transduction pathways and in limb morphogenesis. In contrast, the hypomethylated genes were mainly significantly over-represented in sequence-specific DNA binding and developmental processes. They also reported a correlation in the methylation changes between in-vivo aged MSCs and long-term cultured MSCs. Fernandez et al. studied the genome-wide DNA methylation status of bone-marrow MSCs obtained from donors aged between 2 years and 92 years (Fernández et al., 2015). These MSCs were either from a commercial source or were isolated by density gradient centrifugation and plastic adherence. They identified 18,735 hypermethylated and 45, 407 hypomethylated CpG islands associated with aging. As in differentiated cells, the hypermethylated sequences in the MSCs were enriched in chromatin-repressive marks. The hypomethylated CpG islands were strongly enriched in the active chromatin mark H3K4me1 in MSCs and differentiated cells, which suggested that this is a chromatin signature of DNA hypomethylation during aging that is independent of cell type. Further analyses also suggested that the dynamics of DNA methylation during aging depend on a complex mixture of factors that include the DNA sequence and the cell type and chromatin context involved, and that depending on the locus, the changes can be modulated by genetic and/or external factors.

Changes in differentiation potential and senescence-associated changes
These alterations described for gene expression patterns, epigenetic profiles and intracellular damage are reflected in changes in the regenerative functions of MSCs with aging. Several studies have shown decreased osteogenic and chondrogenic differentiation potentials with increased chronological age of MSC donors (Guang et al., 2013;Kanawa et al., 2013;Stolzing et al., 2008;Zhou et al., 2008). These findings are related to changed expression in the cell-surface receptors involved in osteogenic differentiation (Stolzing et al., 2008;Zhou et al., 2011), and to changed expression and signalling of growth factors (Marie, 2014;Moerman et al., 2004). Furthermore, during aging, there is a shift in MSC differentiation potential from osteogenesis to adipogenesis (Kim et al., 2012;Nehlin et al., 2019). Moerman et al. showed that aging activates adipogenic programmes and suppresses osteogenic programmes in mouse bone-marrow MSCs, via TGF-β/BMP signalling pathways, and increases expression of adipocyte-specific transcription factor PPAR-γ2 (Moerman et al., 2004). This shift in differentiation capacity is linked to increased oxidative stress (Almeida et al., 2009) and changes in MSC epigenetic status (Meyer et al., 2016). It contributes to decreased bone strength, and is implicated in multiple pathological processes that interfere with the appropriate maintenance of bone-tissue repair, possibly via stimulation of osteoclastogenesis (Takeshita et al., 2014).
In addition to changes in MSC differentiation potential, there have been reports of changes in expression of pro-inflammatory factors and diminished immunomodulatory potential (Gnani et al., 2019;Siegel et al., 2013), which have been associated with early senescence of MSCs. Cell senescence is a multifaceted process in which cells stop dividing and undergo distinctive phenotypic alterations, which include profound chromatin and secretome changes, and tumour-suppressor activation (Van Deursen, 2014). Some studies have shown the importance of cell senescence as a safeguard against cancer (Campisi, 2013), as well as in development, wound healing, tissue repair and aging/age-related disorders (Van Deursen, 2014). The senescent phenotype is characterised by an enlarged-cell morphology, reduced telomere length, increased activity of senescence-associated β-galactosidase, altered autophagy, increased G1 cell-cycle arrest, production of reactive oxygen species, and up-regulation of p21, p14 and p53 expression. In addition, senescent cells show altered protein expression, altered responses to growth factors, and altered secretion profiles, which are collectively known as the senescence-associated secretory phenotype (SASP). This SASP includes the secretion of proinflammatory cytokines, chemokines and extracellular matrix proteins, which together create a toxic microenvironment. The SASP not only drives responses that reinforce senescence in a cell-autonomous manner, but also promotes actions on neighbouring cells, via a paracrine mechanism, which thus contributes to increased cell senescence and tissue dysfunction (Nelson et al., 2012).
The accumulation of senescent cells in musculoskeletal tissues with aging can contribute to age-related pathologies, presumably through the secretion of SASP-associated factors (Jeon et al., 2017;Van Deursen, 2014). For instance, increased levels of SASP-associated factors such as interleukin 6 (IL-6), IL-8 and monocyte chemotactic protein 1 (MCP1) have been shown for conditioned media of MSCs from aged donors compared to those from young donors; these effects were exacerbated at the later passages (Gnani et al., 2019). In this study, the MSCs were isolated by density gradient centrifugation and plastic adherence of the CD34 − cell fraction. The factors secreted by the MSCs from aged donors were shown to activate pro-inflammatory gene expression in haematopoietic stem cells from young donors, and to decrease their clonogenic potential (Gnani et al., 2019). Accumulation of senescent cells in articular cartilage has been shown for osteoarthritis, as well as in the synovium following tearing of the anterior cruciate ligament (Jeon et al., 2017). A recent study also examined accumulation of senescent cells in the bone of mice during aging (Farr et al., 2016). Enriched osteoblast-progenitor, osteoblast and osteocyte populations from young and aged mice were analysed for expression of senescence-related genes and the SASP components. In all three of these cell populations, senescence markers were markedly increased with aging. In contrast, relatively few SASP-associated factors were altered in the osteoblast progenitors and the osteoblasts, while many of the SASP-associated factors were increased in osteocytes from the aged mice. In-vivo analyses of aged cortical bone samples further confirmed these increases in the numbers of senescent osteocytes.

The properties of mesenchymal stromal cells in osteoarthritis and osteoporosis
As described above, aging affects several processes within the musculoskeletal system at the cell and organ levels, which thus contribute to degenerative disorders. Moreover, the hormonal, immunological and metabolic (micro)environments present during tissue injury might also be linked to aging (Sacitharan and Vincent, 2016). These challenging environments can greatly influence the regenerative capacity of MSCs, and accentuate age-related tissue damage.
Stem cell 'exhaustion' has been suggested as one of the hallmarks of aging (Partridge et al., 2018), which might impair any of the mechanisms of MSC regenerative actions; i.e., their paracrine activity, immunomodulation and apoptosis, or changes in their differentiation into tissue progenitors, or different extracellular-vesicle-mediated effects (Fig. 1). The lack of these actions might contribute to the changes observed in the joint tissues of patients with osteoarthritis and osteoporosis.
For instance, the lack of paracrine or extracellular-vesicle-mediated MSC effects might contribute to impaired type II collagen synthesis, deposition and crosslinking, which would lead to increased stiffness of the extracellular matrix, reduced mechano-responsiveness of chondrocytes, and increased cartilage brittleness (Bank et al., 1998). The lack of the modulation of local inflammatory and immunological responses by joint-resident MSCs might contribute to chronic inflammation and osteoarthritis (Di Nicola, 2020). In osteoporosis, bone-marrow MSCs more readily differentiate into adipocytes than osteoblasts, which contributes to the 'obesity' of bones and to the lower bone density in these patients (Kawai et al., 2012;Kim et al., 2012;Pino et al., 2012;Rosen and Bouxsein, 2006). However, in-vitro data using vertebral bone-marrow-derived MSCs did not confirm this (Haddouti et al., 2020), so further elucidation of the precise role of MSCs in the pathogenesis of osteoporosis is still awaited.
An overview of prior studies that have demonstrated changes to MSCs in the joint-degenerative disorders of osteoarthritis and osteoporosis is given in Table 1. It is important to note that most of these studies did not include age-matched controls to discriminate between the properties of the MSCs that are age related and those that are disease related. This is particularly difficult, as osteoarthritis and osteoporosis are more common in the elderly. Thus, clearly defined criteria for disease diagnosis and age-matched control subjects are of vital importance to fully assess the differential contributions from age and disease (Brady et al., 2015). Also, the method of MSCs isolation is immensely important for the evaluation of the properties of MSCs in vitro. Most studies have taken advantage of the more feasible plastic adherence method for establishing primary MSCs in culture (Table 1). However, there are also some high impact studies where more complex cell sorting methods were used to study highly defined MSC populations (Murphy et al., 2020;Roelofs et al., 2020).
To clarify the osteoporosis-associated deficiencies in bone regeneration, Benisch et al. compared the transcriptomes of bone-marrow MSCs from patients with osteoporosis with those of elderly donors without osteoporosis (Benisch et al., 2012). They confirmed the increased expression of previously identified osteoporosis-associated genes, such as runt-related transcription factor 2 (RUNX2) and collagen type I alpha 1 (COL1A1), and genes that code for inhibitors of WNT and BMP signalling. They also newly identified a repressor of BMP-induced transcription, the MAB21L2 gene. Overall, the gene expression changes identified indicated intrinsic deficiencies in self-renewal and differentiation potential of MSCs from osteoporotic patients. Comparisons with MSCs from young donors have also indicated that transcriptional changes differ widely between MSCs from osteoporotic and non-osteoporotic donors.
A study of bone-marrow MSCs from patients with end-stage osteoarthritis who were undergoing joint arthroplasty suggested that these MSCs have altered features, such as reduced chondrogenesis and adipogenesis in vitro, in comparison with those from healthy donors with no signs of osteoarthritis (Murphy et al., 2002). Another study investigated the effects of age on MSCs, and reported that aging alone causes intrinsic alterations to MSCs that might contribute to the process of skeletal aging in humans (Zhou et al., 2008). Bone-marrow MSCs were isolated from the femoral heads of patients undergoing hip arthroplasty for primary osteoarthritis. These MSCs showed age-dependent decreases in cell proliferation and osteoblast differentiation, and an increase in senescent and apoptotic cells. It was suggested that up-regulation of the p53 pathway with age might have had a critical role in these alterations (Zhou et al., 2008).
Prior studies from our group have focused on the identification of biological changes in trabecular bone and skeletal muscle MSCs from patients with different types of osteoarthritis and from patients with femoral neck fractures due to osteoporosis (Čamernik et al., 2020a,b, 2019). Trabecular bone from the subchondral part of the femoral head and gluteus medius muscle were selected as the sources of the MSCs, due to their relative proximity to the hip joint that is commonly affected in patients with osteoarthritis and osteoporosis. From comparison of these muscle and bone MSCs from the patients with osteoarthritis, we showed that the muscle MSCs had improved viability ex vivo and greater clonogenicity, growth rate, and osteogenic and myogenic potential in vitro. The senescence rates and adipogenic and chondrogenic potentials were similar. Moreover, we also showed positive correlation between CD271 expression and adipogenesis, although this remains to be further examined to establish whether there are any contributions of CD271-expressing muscle MSCs to the muscle steatosis associated with osteoarthritis (Čamernik et al., 2019).
Next, we questioned whether bone MSCs from patients with primary osteoarthritis are 'exhausted' in terms of their ex-vivo and in-vitro properties, as compared with bone MSCs from patients with hip dysplasia, where their osteoarthritis might be the consequence solely of the abnormal hip biomechanics (Čamernik et al., 2020b). Interestingly, the freshly isolated cells from patients with primary osteoarthritis showed decreased viability ex vivo and reduced osteogenic and chondrogenic potentials in vitro, compared to the cells from patients with hip dysplasia. There were no differences in clonogenicity, growth kinetics, senescence, adipogenic potential and immunophenotype between these two patient groups. Gene expression profiling showed that expression of the leptin receptor, as a well-known marker of bone-marrow MSCs, was significantly lower in bone MSCs from patients with primary osteoarthritis. Leptin-receptor-positive MSCs were previously identified as the major MSC source of the bone and adipocytes in adult bone marrow (Zhou et al., 2014). Taken together, these results suggested that subchondral bone MSC exhaustion is implicated in the pathology of primary osteoarthritis.
We also obtained muscle and bone MSCs from post-mortem donors who had shown no age-related joint disorders (i.e., healthy donors), and compared these with muscle and bone MSCs from patients with osteoarthritis and osteoporosis, to investigate whether their regenerative properties were 'exhausted' (Čamernik et al., 2020a). Indeed, these data revealed reduced osteogenesis of the MSCs from both of these tissues from patients with osteoarthritis, compared to those from the healthy donors, and also reduced adipogenesis for the bone MSCs. As an indicator of chondrogenic potential, the chondrogenic pellet diameter was also reduced in the bone MSCs from patients with both osteoarthritis and osteoporosis, compared to the healthy donors. Similar to our previous study (Čamernik et al., 2019), significant positive correlation was seen between adipogenesis and CD271 expression in the muscle-derived MSCs. Moreover, there was again lower expression of the leptin receptor in bone MSCs from patients with osteoarthritis, in comparison with those of the patients with osteoporosis and with the healthy donors. Interestingly, the freshly isolated muscle cells from patients with osteoarthritis and osteoporosis showed higher clonogenicity compared to those from the healthy donors, which might indicate that they were activated and contributed to the disease phenotypes (Čamernik et al., 2020a). Muscles and their strength have important roles in the pathogenesis of bone and joint diseases. Further knowledge on how age-related changes in muscle stem cells can contribute to muscle aging has been reviewed elsewhere (Muñoz-Cánoves et al., 2020).
Zhen at al. investigated the role of transforming growth factor β1 (TGF-β1) in subchondral bone pathology and articular cartilage degeneration during the progression of osteoarthritis (Zhen et al., 2013).
Using an anterior cruciate ligament transection model of osteoarthritis in mice and rats, they showed that high concentrations of TGF-β1 induced formation of nestin + MSCs clusters. This led to aberrant bone formation, which was accompanied by increased angiogenesis. In contrast, inducible knockout of TGF-β1 receptor II in nestin + cells reduced the changes in subchondral bone and articular cartilage in the rat osteoarthritis model. As TGF-β1 levels are also known to be increased in human subchondral bone, the authors suggested that TGF-β1 inhibition represents a therapeutic approach for osteoarthritis (Zhen et al., 2013). Moreover, when subchondral bone without and with bone-marrow lesions was compared within the same patient with hip osteoarthritis, numerical and topographical changes in the native MSCs were seen (Campbell et al., 2016). Bone-marrow lesions adjacent to the cartilage defects and the areas of osteochondral angiogenesis were the site of CD271 + MSCs accumulation. These MSCs also showed lower proliferation and mineralisation capacities in vitro (Campbell et al., 2016). Kohno et al. compared synovium-derived MSCs from patients with osteoarthritis and rheumatoid arthritis ex vivo (Kohno et al., 2017). They showed that cell yields, surface markers and chondrogenic potential of the primary synovial MSCs in rheumatoid arthritis were comparable to those for osteoarthritis. Despite the common knowledge that synovitis is a hallmark of inflammatory arthritis such as rheumatoid arthritis, these authors suggested that the synovium derived from patients with rheumatoid arthritis might be the source of the MSCs for cartilage and meniscus regeneration (Kohno et al., 2017). Sanjujo-Rodriguez et al. recently clarified the beneficial effects of knee-joint distraction for cartilage regeneration via synovial-fluid-derived MSCs (Sanjurjo- Rodriguez et al., 2020). Following knee off-loading as early as 3 weeks, these MSCs showed significant increases in colony formation and showed transcriptional changes, including higher expression of the chondrogenic commitment markers GREM1 and growth differentiation factor 5 (GDF5) (Sanjurjo-Rodriguez et al., 2020). Watanabe et al. analysed MSCs in synovial fluid of patients with degenerative meniscus injury in the knee (Watanabe et al., 2020). They showed that synovial-fluid MSCs were scarce; however, following harvest of the synovium and meniscus repair, the numbers of synovial-fluid MSCs significantly increased, as shown by colony-forming assays in vitro (Watanabe et al., 2020). However, whether this increase is beneficial in terms of halting osteoarthritis onset or progression remains to be determined.
Although cartilage is recognized as a hypocellular tissue, cartilagederived MSCs have been identified and associated with osteoarthritis. Jayasuriya et al. showed the persistence of two distinct populations of MSCs in human osteoarthritis cartilage (Jayasuriya et al., 2018). While both of these osteoarthritis MSC populations were multipotent and contributed to the osteoarthritis phenotype and disease progression, one population was preferentially chondrogenic and the other was preferentially osteogenic. Furthermore, both of these osteoarthritis MSC populations showed significantly greater expression of the hypertrophic osteoarthritis cartilage markers COL10A1 and RUNX2, compared to osteoarthritis chondrocytes from the same tissues. Induction of chondrogenesis in these osteoarthritis MSCs further stimulated COL10A1 expression and MMP-13 release, which suggests that these MSCs contribute to the osteoarthritis phenotypes, and thus represent novel targets for osteoarthritis therapies (Jayasuriya et al., 2018).
A characteristic feature of osteoarthritis that can be observed via Xray imaging is the presence of osteocartilaginous formations; i.e., osteophytes (van der Kraan and van den Berg, 2007). Osteophytes are present in late-stage osteoarthritis at the joint margins, where the synovium attaches to the articular cartilage and merges with the periosteum. In a recent study, Roelofs et al. showed that the osteophytes are formed from platelet-derived growth factor α (PDGFα) + GDF5 + lineage progenitors (Roelofs et al., 2020). In the normal joint, these progenitors are present at the junction of the periosteum and synovium, near the articular cartilage, and they become activated in osteoarthritis. They were further divided into two subsets: the first subset was the proteoglycan 4 (or lubricin)-expressing progenitors that resided in the synovial lining and that provide chondrocytes to the cartilaginous part of the osteophyte; and the second subset were 'SRY-box transcription factor 9 ′ -expressing progenitors in the underlying periosteum, which give rise to the transient cartilage template that is then remodelled into bone. The authors suggested that both of these cell subsets are targets for the disease modification strategies of osteoarthritis (Roelofs et al., 2020).

Mesenchymal stromal cells for regenerative treatments of bones and joints
Cell therapies and tissue engineering approaches based on MSCs have been intensively studied to tackle traumatic and degenerative defects of articular cartilage and bone. The aim is to activate endogenous repair mechanisms towards functional regeneration of the lost tissues, preferably with no remaining implant materials. Following tissue damage, MSCs are actively involved in all phases of tissue healing. They promote immunosuppression in the first acute inflammatory phase, both directly via cell-to-cell interactions and indirectly via soluble paracrine factors that they produce. They can effectively suppress innate and adaptive immune responses (Zachar et al., 2016). Their actions involve T-cell suppression, macrophage activation, and potentially neutrophil recruitment via paracrine mechanisms (reviewed by DiMarino et al., 2013). The repair and remodelling that follow as the next steps in the tissue healing processes entail regulation of extracellular matrix deposition, collagen synthesis, fibroblast proliferation, platelet activation, fibrinolysis and angiogenesis (DiMarino et al., 2013). Importantly, the tissue-reparative effects of MSCs based on their immunomodulatory and trophic functions are substantially influenced by their tissue environment; i.e., by cells and various factors.
When MSCs are evaluated for therapeutic purposes, it is crucial to consider their heterogeneity and to describe their properties, as well as to use specific nomenclature with regards to their tissue origin and the isolation/ preparation procedures used (Bianco and Robey, 2015;Sipp et al., 2018). As mentioned in Section 4, these aspects significantly influence the MSC characteristics, and thus their therapeutic potential. It is possible that MSCs from certain tissues or their highly defined sub-populations will be better suited for specific clinical indications than others (e.g., due to accessibility, numbers available, biological properties, other factors), even when transplanted to a tissue different than their origin.

Pre-clinical evidence for regenerative actions of MSCs in bones and joints
Most of the knowledge of MSCs and their potential for tissue regeneration comes from animal models of fractures and cartilage injury Roelofs et al., 2020;Worthley et al., 2015;Zhou et al., 2014). The majority of these studies focused on the identification of bone-derived and bone-marrow-derived MSCs and their potential for tissue engineering and regeneration . For bone regeneration, Park et al. demonstrated that osteoblastic cells are non-replicative and short lived in the mouse system, and are replenished from the Mx1 + bone-marrow stromal cells with MSC features, thus 'feeding' the cell replacement demands of the adult skeleton .
Chan et al. provided a breakthrough through their identification of the ultimate skeletal stem cells in mice , and more recently, the same group provided information on the boosting of these endogenous MSCs for cartilage and skeletal regeneration using a microfracture surgery model (Murphy et al., 2020). They showed that localised acute injury and the cascade of events that triggers this are sufficient to 'wake up' the dormant resident stem cells, and to dramatically alter their behaviour in favour of tissue regeneration. However, tissue regeneration following microfracture procedures was shown to most commonly lead to fibrous tissue formation, which has inferior properties compared to hyaline cartilage (Mithoefer et al., 2009). In contrast, Murphy et al. showed that in a hydrogel localised co-delivery of BMP2 and soluble vascular endothelial growth factor receptor 1 (sVEGFR1), a VEGF receptor antagonist, skewed the differentiation of microfracture-activated SSCs towards articular cartilage (Murphy et al., 2020). These authors had previously shown the chondrogenic potential of treatments with BMP2 and sVEGFR1 in mice  and humans (Chan et al., 2018). As block of both of these factors is already in clinical use in the USA (i.e., BMP2 as Infuse; VEGF signalling as Avastin), they might be tested in combination with microfracture surgery in cases of early, mild osteoarthritis cartilage regeneration (Murphy et al., 2020). In humans, multipotent SSCs were only recently identified as PDPN + CD146-CD73 + CD164 + , and shown to undergo local expansion in response to acute skeletal injury, and to regenerate cartilage (Chan et al., 2018;Murphy et al., 2020). Further studies are needed to characterise the functional properties of SSCs in patients with osteoarthritis and osteoporosis, and to evaluate whether approaches for boosting endogenous SSCs can be developed, and how they compare to using the more heterogeneous MSC populations in cell therapies.
For cartilage therapies, synovium-derived MSCs underpin the synovial hyperplasia following joint injury, and are also a promising target (Hunter and Bierma-Zeinstra, 2019). The hyperplasia of the synovial membrane following joint injury has for a long time been attributed mainly to stromal cells, including type B synoviocytes (also called synovial fibroblasts), with infiltration of inflammatory and immune cells. Kurth et al. were the first to show that traumatic-injury-induced synovial hyperplasia is also sustained by proliferation and chondrogenic differentiation of synovial MSCs (Kurth et al., 2011). In adult mouse synovium, a subpopulation of synovial MSCs that are the progeny of GDF5-expressing joint interzone cells was identified using lineage tracing (Roelofs et al., 2017). Following joint surface injury, these GDF5 + progenitors underpin synovial hyperplasia through proliferation, and contribute to cartilage repair. Mechanistic studies have shown that the transcriptional co-factor Yes-associated protein (Yap) is up-regulated after injury in these cells, and its conditional ablation in cells of GDF5 lineage prevents synovial lining hyperplasia and decreases the contribution of these cells to cartilage repair. Moreover, intramuscular injection of adult human synovial MSCs that overexpress BMP7 resulted in formation of an ectopic joint-like structure, which provides scientific rationale for the use of stem cells from the adult synovium for joint regenerative therapies (Roelofs et al., 2017).
Boosting the immunosuppressive actions of MSCs to promote the resolution of inflammation within damaged joints might represent another therapeutic target for osteoarthritis. In the shift from inflammatory to non-inflammatory status, bioactive metabolomes such as those of resolvins, protectins and maresins have active roles (Serhan and Petasis, 2011). All three of these molecules stimulate self-limited innate responses, enhance innate microbial killing and clearance, and are organ protective (Serhan and Levy, 2018). Resolvins in particular have been shown to be efficient not only for resolution of acute inflammation, but also of chronic inflammation (Benabdoune et al., 2016;Serhan and Petasis, 2011). Resolvin D1 showed cytoprotective, anti-inflammatory, anti-catabolic and anti-apoptotic effects on osteoarthritis chondrocytes (Benabdoune et al., 2016). Direct evidence for the biosynthesis of these molecules was originally obtained in human periodontal ligament stem cells and in mouse bone-marrow MSCs (Romano et al., 2019). Given the limited evidence that MSCs themselves can generate these anti-inflammatory molecules and express their receptors, it would be interesting to see whether increasing the expression of these anti-inflammatory mediators in joint-resident stem cells represents a way to 'empower' endogenous MSCs for suppression of inflammation and acceleration of tissue repair.

Approaches and clinical evidence for MSC-based cartilage regeneration
From the view-point of natural history, articular cartilage disorders caused by trauma or joint disease ultimately progress to osteoarthritis. Traditionally, the management of focal articular cartilage lesions and osteoarthritis has been palliative. The discovery of the multilineage differentiation potential of MSCs (Pittenger et al., 1999) led to strong interest in orthopaedic society, and instantly popularised the idea that their chondrogenic capacity can be used to repair or regenerate articular cartilage (Wakitani et al., 1994). The results of this pronounced enthusiasm were a multitude of clinical studies and meta-analyses that attempted to provide conclusive evidence regarding the outcomes of treatments using MSCs for biological cartilage repair. The majority of clinical trials on MSCs for the treatment of articular cartilage have been conducted on the knees. These data can be divided into two groups according to the treatment approaches used, and these should be reviewed separately: a surgical approach with implantation of MSCs on scaffolds has been used for the treatment of focal articular cartilage lesions, and MSC injections have been used for the treatment of osteoarthritis.
One of important concerns about the data reported for osteoarthritis treatment is the variability of the possible concomitant treatments; i.e., combined injections of MSCs with platelet-rich plasma or hyaluronic acid, or surgery. Just recently, the conclusions of three systematic reviews and meta-analyses on short-term and medium-term safety and efficacy of intra-articular injection of MSCs without any adjuvant therapies for knee osteoarthritis were published (Kim et al., 2020a;Ma et al., 2020;Tan et al., 2021). In their study, Tan et al. included a total of 19 levels of evidence I and II studies with 440 osteoarthritic knees, and they concluded that injection of MSCs alone is safe, reduces pain, and improves knee function (Tan et al., 2021). Similar conclusions were arrived at by Ma et al. based on their analysis of 10 randomised controlled clinical trials from 2015 to 2019 that involved 335 patients with knee osteoarthritis who were treated with intra-articular injections of MSCs without adjuvant surgery . Furthermore, a few high-level clinical studies have suggested improvements in articular cartilage quality on magnetic resonance imaging after injectable MSCs treatments of knee osteoarthritis, but the evidence was relatively poor (Freitag et al., 2019;Khalifeh Soltani et al., 2019;Vega et al., 2015). Detailed reviews of the currently available literature reveal significant heterogeneity of clinical trials for their different cell sources, various methods of cell preparation, and diversity in the clinical assessment tools used. All of these differences make the evidence of the clinical efficacy of injectable MSCs for treatment of osteoarthritis further contentious.
The MSCs used in conjunction with scaffolds represent an alternative approach for the treatment of focal cartilage and osteochondral lesions. However, there is currently no evidence for their superiority over chondrocyte implantation or microfracturing (Deng et al., 2016). In a large retrospective study, Kim et al. evaluated 467 patients (as 483 knees) who underwent MSC implantation on a fibrin glue scaffold for knee osteoarthritis, with a minimum 5-years of follow-up (Kim et al., 2020b). They concluded that MSC implantation provided encouraging outcomes with acceptable duration of symptom relief in patients with focal knee cartilage lesions and early osteoarthritis. Veber et al. also reported encouraging clinical results for the treatment of knee osteochondral lesions in 15 patients using a combination of filtered bone-marrow aspirate and a biomimetic scaffold (Veber et al., 2020). In 7 of the 8 patients who underwent magnetic resonance imaging and arthroscopic evaluation after the treatment, nearly normal to normal lesion repair was seen, along with good integration of newly formed tissue with the surrounding cartilage.
The clinical efficacy of MSCs has been shown to depend on the cell doses used. Large doses (i.e., high numbers) of MSCs can be obtained by cultivation and expansion methods, although the effects should be balanced with the safety: larger doses have been associated with greater risk of post-treatment swelling and discomfort (Doyle et al., 2020;Gupta et al., 2016;Rock and Kono, 2008).
Although preclinical studies indicate differences in chondrogenic potential and clonogenicity between MSCs from different sources, there is still no agreement on whether this variable has any significant clinical impact. Some studies have reported significantly better outcomes using bone-marrow MSCs as compared to adipose-tissue MSCs, and using cultured MSCs as opposed to uncultured MSCs for injectable treatments of knee osteoarthritis (Tan et al., 2021). On the contrary, a meta-analysis by Han et al. focused on the impact of different sources of MSCs on the efficacy of treatment of osteoarthritis, and they suggested improved clinical results using adipose tissue MSCs compared to bone-marrow MSCs, although this was backed up with limited clinical evidence . Compared to autologous MSCs, allogeneic MSCs can induce adverse reactions or host immune rejection (Consentius et al., 2015).
With regards to age-related alterations and senescence of MSCs, it is relevant to evaluate how the clinical results for MSC treatments of cartilage have depended on the age of the patient. Patient age was shown not to influence the response to treatments with injectable MSCs for osteoarthritis in clinical praxis (Kim et al., 2020a;Ma et al., 2020;Tan et al., 2021). However, in contrast to injectable use of MSCs, patient age was identified as one of the independent factors associated with failure of treatment of focal cartilage lesions with MSCs on scaffolds (Kim et al., 2020b).
In conclusion, there is currently limited evidence on the clinical efficacy of MSCs used for the treatment of focal articular cartilage lesions and osteoarthritis, although preliminary data remain promising. It appears that contemporary MSC treatments offer symptomatic relief of these conditions, mainly based on the immunomodulatory capacity and paracrine action of these cells. However, there is still no clear clinical evidence that they can reverse or delay the progression of osteoarthritis. Standardisation of the processes is an urgent problem that needs to be addressed in the future. Finally, the results of clinical trials that have studied new strategies are awaited, such as those using synovial membrane as the source of MSCs, and those using extracellular vesicles for delivery Zupan et al., 2020a).

Approaches and clinical evidence for MSC-based bone regeneration
About 5 %-10 % of bone fractures result in delayed or failed healing, for which advanced patient age has been identified as one of the most important risk factors (Gruber et al., 2006;Zura et al., 2016). MSCs have been investigated in a variety of cell-therapy and tissue-engineering approaches, which have included various combinations of biomaterials, growth/ signalling factors, and pre-cultivation strategies (Frohlich et al., 2008). Bone-marrow MSCs and adipose-tissue MSCs have been among the most studied, due to their relevance for translation to the clinic (Marolt et al., 2010). Based on this preclinical work, a number of clinical studies have been initiated with MSCs, which have focused on enhanced healing of long-bone fractures and fracture non-unions, as well as regeneration of jaw bone defects and treatment of femoral-head osteonecrosis . With only a few studies completed and with published research findings, and due to the many variables in the protocols, it is challenging to draw any common conclusions. This limitation is reflected by the current lack of standard MSC-based bone regeneration treatments that have been translated to the clinic (Sallent et al., 2020).
Nevertheless, the data available suggest that MSC-based therapeutic approaches for bone regeneration are safe and can provide benefits for the patients. In their pivotal study, Hernigou et al. showed that a cell therapy approach that involved percutaneous injection of concentrated autologous bone-marrow mononuclear cells enhanced the healing of fracture non-unions (Hernigou et al., 2005). They evaluated the number of progenitor cells according to the CFU-F assay, and reported positive correlation between the total number and the concentration of transplanted CFUs, and the formation of mineralised callus. A negative correlation was also reported between the concentration of CFUs in the graft and the time to reach bone union. Age had no significant effects on the number of progenitor cells received by the male patients, but increasing age was shown to be associated with significant decreases in the total number of progenitors received by the female patients. Similarly, Le Nail et al. determined the threshold numbers of concentrated bone-marrow progenitors for successful healing of open tibia fractures and reported that the patient age did not influence the treatment results (Le Nail et al., 2014). Furthermore, the use of freshly isolated autologous bone-marrow MSCs together with autologous platelet-rich plasma and a demineralised bone matrix carrier resulted in shorter times to bone union in distal tibial fractures, compared to the no treatment control (Liebergall et al., 2013). For the treatment of mandibular fractures, application of autologous adipose MSCs 24 h after their isolation resulted in higher ossification values at 12 weeks compared to the no treatment control (Castillo-Cardiel et al., 2017). In these last two studies, the effects of patient age were not investigated. Taken together, these studies suggest that injection/ application of cell therapies can enhance the healing of bone fractures, and that the success of the therapy depends on the dose of transplanted osteoprogenitors. However, the potential impact of age-related changes/ senescence of MSCs were not investigated in detail.
Therapies with bone-marrow mononuclear cells have also been evaluated in several studies to treat osteonecrosis of the femoral head. Whereas Emadedin et al. reported improvements to the disease score, reduced joint injuries, and pain relief (Emadedin et al., 2019), Hauzeur et al. reported that the cell therapy did not improve stage three osteonecrosis (Hauzeur et al., 2018).
A clinical report by Quarto et al. described the first use cultured bone-marrow MSCs together with hydroxyapatite biomaterial for the treatment of large traumatic bone defects in three patients (Quarto et al., 2001). They reported good callus formation and implant integration at the interfaces with the host bones. Several registered clinical studies later investigated cultured MSCs from the bone marrow and adipose tissue, alone and in combination with biomaterials, to enhance the healing of delayed/ non-unions of long bones, and jaw bone defects. Emadedin et al. reported autologous bone-marrow MSC injections as safe for the reconstruction of lower limb long-bone atrophic non-unions, with three of five patients showing evidence of union at 6 and 12 months after the treatments (Emadedin et al., 2017). Gomez Barrena et al. conducted the first European multicentre, non-comparative trial, where bone-marrow MSCs associated with biphasic calcium phosphate biomaterial were used to treat non-unions in tibia, femur and humerus (Gómez-Barrena et al., 2019). They reported no severe adverse events with the treatment, and 26 of 28 patients showed radiological healing after 1 year. The effect of patient age and any potential age-related changes in MSC properties were not investigated. Khojasteh et al. used buccal fat pad MSCs that were cultured on bovine bone mineral in conjunction with either autologous spongy bone and collagen membrane, or autologous cortical bone, for the treatment of alveolar cleft defects (Khojasteh et al., 2017). Compared with the control group of autologous spongy bone with collagen membrane, the cell-therapy groups showed trends to higher bone formation 6 months after treatment. There were age differences between the patients of the different groups, but the age-related effects were not analysed further. Redondo et al. used bone-marrow MSCs that had been pre-differentiated on a cross-linked serum scaffold to treat cystic bone defects of the maxilla (Redondo et al., 2018). Their treatment was reportedly safe and significantly increased bone growth, as compared to the no treatment control of the contralateral bone area. Gjerde et al. treated mandibular bone atrophy with bone-marrow MSCs mixed with biphasic calcium phosphate, and reported that the treatment resulted in sufficient new bone formation to allow dental implant placement after 4-6 months (Gjerde et al., 2018).
Current clinical studies are focused on evaluation of the effects of different MSC doses, with comparisons of the effects of MSC therapies to autologous bone-graft controls, and with the outcomes compared between autologous and allogeneic MSCs . On the other hand, pre-clinical research is also focused on the engineering of viable personalised bone grafts to repair complex bone defects. For the repair of cranio-maxillofacial defects, Bhumiratana et al. cultured adipose-MSC-seeded scaffolds under dynamic conditions that supported direct bone formation through intramembranous ossification pathways (Bhumiratana et al., 2016). In contrast, a 'developmental engineering' approach that involves the construction of hypertrophic cartilaginous grafts has been suggested for the treatment of long-bone defects (Epple et al., 2019;Scotti et al., 2010). This approach might provide benefits in terms of enhanced graft survival and vascularisation, which will lead to better regeneration outcomes.
Systemic infusions of MSCs have been considered for the treatment of aging-related and disease-related bone loss. However, preclinical studies showed that these failed to promote any osteogenic response, as the MSCs did not 'home' to the bone surfaces; they stayed predominantly trapped in the lungs and the liver (Gao et al., 2001). Guan et al. reported on a hybrid compound that was composed of specific peptidomimetic ligands against α4β1 integrin on the MSCs. These were conjugated to the bisphosphonate alendronate, as a bone-seeking component, to 'direct' the MSCs and this hybrid compound to bone (Guan et al., 2012). This compound increased MSC migration and osteogenic differentiation in vitro. In a mouse xenotransplantation model, a single intravenous injection of this compound with human MSCs increased the MSC homing to and retention in the bone, and increased bone formation. Furthermore, in immunocompetent mice, this compound alone increased trabecular bone mass 12 weeks after treatment, and also prevented the trabecular bone losses induced by aging and oestrogen deficiency, possibly through increased migration of endogenous MSCs to the bone. In a follow-up study, Yao et al. reported that this compound restored both trabecular and cortical bone loss induced by oestrogen deficiency and advanced age in mice . More recently, the same group reported that this compound both prevented and rescued the osteoporotic phenotype in a mouse model of glucocorticoid-induced osteoporosis (Mohan et al., 2017). This compound is currently in a clinical trial to evaluate the safety and tolerability of intravenous treatments for osteopenia secondary to glucocorticoids ("Safety and tolerability of intravenous LLP2A-alendronate for osteopenia secondary to glucocorticoids -Full Text View -ClinicalTrials.gov, " n.d.).

Extracellular vesicle therapies
Despite the remarkable properties of MSCs that have been demonstrated in pre-clinical studies, and the promising results in some clinical studies, there has been limited market authorisation for MSC-based cell therapies. There is also no doubt that these cellular therapies are associated with relative high costs due to the complex manufacturing protocols, and substantial regulatory procedures and standardisation processes.
Mesenchymal stromal cells were initially expected to regenerate damaged tissues by local implantation and when under the influence of microenvironmental cues. However, their mode of action in vivo is still not clear, as they show rapid clearance of the transplanted bulk MSC population, and only very low levels of engraftment (typically 3 % or less). Galleu et al. have shown that systemically injected MSCs undergo apoptosis induced by recipient cytotoxic cells (Galleu et al., 2017). The phagocytes then engulf the apoptotic MSCs, and produce indoleamine 2, 3-dioxygenase, which triggers immunosuppression in graft versus host disease. These authors also suggested that the need for recipient cytotoxic cell activity can be replaced by infusion of apoptotic MSCs generated ex vivo.
An additional hypothesis on the therapeutic mode of action of MSCs centres around their secretome (Praveen et al., 2019). As well as a plethora of protein-based factors, one aspect of the MSC secretome that has gained a lot of attention recently is the extracellular vesicles. Extracellular vesicles modulate the immune system (Barry, 2019), and can enhance angiogenesis, with their further activities shown to include neurogenesis, tissue repair, wound healing, and anti-fibrotic and anti-tumour effects (Doeppner et al., 2015); indeed, similar to MSCs themselves. Several studies have compared transplantation of preparations enriched in MSC extracellular vesicles with transplantation of MSCs, with no large differences in the therapeutic effects seen for several different diseases (Madrigal et al., 2014). Furthermore, they have also been shown to promote musculoskeletal regeneration (Hofer and Tuan, 2016).
Although MSC-based therapies have been shown to be relatively safe from a clinical standpoint, the use of a cell-free secretome might indeed help to generate off-the-shelf complex biopharmaceuticals instead of medicinal products for advanced therapies. Using standardised and scalable bioprocesses might circumvent several obstacles of current MSC therapies, including donor variability and cell supply. This is especially the case where MSC-secretome therapies would be based on immortalized primary-like human MSC cell lines from different tissues that have different activities (Kehl et al., 2019). In addition, pre-conditioning of MSCs in bioreactors can be performed before their implantation, to potentially enhance their therapeutic efficiency (Najar and Fahmi, 2020).
With the MSC secretome composed of cytokines, chemokines, growth factors, proteins and extracellular vesicles, the identification and use of one of these factors individually might represent an additional alternative to the cell therapy (Eleuteri and Fierabracci, 2019). While MSCs might have their functions modified by the influence of the local tissue (Le Blanc and Davies, 2018), extracellular vesicle treatments might instead lead to more predictable effects and better control of treatment (reviewed by (Alcaraz et al., 2019)).

Elimination of senescent cellssenolytics
Interestingly, initial studies have shown that removal of senescent cells delays several age-related diseases, and increases the healthy lifespan in transgenic mouse models of aging (Baker et al., 2016(Baker et al., , 2011. In two studies, Baker et al. used an INK-ATTAC transgenic mouse model that expressed the FK506 binding protein-caspase 8 (FKBP-Casp8) fusion protein and green fluorescent protein (GFP) under the control of a p16Ink4a promoter fragment, which is transcriptionally active in senescent cells. In this mouse model, naturally occurring p16Ink4a-positive senescent cells were ablated upon administration of AP20187, a synthetic drug that induces dimerization of a membrane-bound myristoylated FKBP-Casp8 fusion protein, expressed specifically in p16Ink4a-positive senescent cells. This finding has resulted in an impressive body of literature that shows that agents that specifically eliminate senescent cells, known as senolytics, might alleviate various major age-associated conditions (Jeon et al., 2017). Senolytics are a class of drugs that range from those of the first generation (dasatinib, quercetin, fisetin, others) to recently developed experimental compounds that are being tested in ongoing clinical trials (Kirkland and Tchkonia, 2020). In the context of cartilage, intra-articular injection of a senolytic agent that selectively eliminate senescent cells (i.e., UBX0101) attenuated the development of post-traumatic osteoarthritis, reduced pain, and increased cartilage development (Jeon et al., 2017). In this way, interleukin 17 has a major role in driving the inflammatory conditions induced by senescent cells (Faust et al., 2020).
Consistent with these demonstrations that degeneration processes can be slowed down by removing senescent cells from aging tissues, Jayasuriya et al. showed that suppression or removal of a distinct population of MSCs, the so-called 'osteoarthritis MSCs', might represent a better strategy to preserve cartilage (Jayasuriya et al., 2018). These authors showed that osteoarthritis MSCs persist in cartilage of patients with osteoarthritis, to give rise to osteoarthritic tissue hallmarks, such as chondrocyte hypertrophy, matrix protease release, cell senescence, and osteophyte formation upon activation by mechanical, chemical or biological stress. They also suggested that once the articular cartilage reaches the osteoarthritic stage, the activation of MSCs would result in further progression of osteoarthritis, rather than cartilage regeneration (Jayasuriya et al., 2018).
Similarly, osteoporosis has been shown to be relieved by the use of senolytic treatments, whereby senescent cells were shown to inhibit osteogenic differentiation of MSCs (Farr et al., 2017), while targeting senescent osteoclast progenitors does not appear to affect age-associated loss of bone mass (Kim et al., 2019). Surprisingly, senescent-cell-induced osteoporosis is also independent of oestrogen deficiency in mouse models (Farr et al., 2019).

Conclusions
Aging has been shown to be associated with tissue degeneration. With osteoarthritis and osteoporosis-related bone fragility fractures at the forefront, the majority of chronic degenerative musculoskeletal diseases drive healthcare use that leads to significant costs to individuals and to society. Further research is of utmost importance for successful treatment and optimal administration strategies in clinical practice for both of these disorders. This needs to focus in particular on the identification of biomarkers to detect these disorders at early stages, on the discovery of the targets in cartilage and bone breakdown, and on the specific characterisation of osteoarthritis phenotypes.
Cell therapies with MSCs represent a strong tool for musculoskeletal tissue regeneration. However, very few MSC-based cell therapies have succeeded in gaining market authorisation, as there are still major obstacles that need to be addressed to exploit their full potential. The urgent matters to address are: phase 3 clinical studies with outcome measures that distinguish between symptom-modifying and diseasemodifying effects; clear criteria for the definition of 'MSCs'; development of manufacturing standards to produce consistent cell products; and further research into the mechanisms of action of MSCs (Barry, 2019;Galipeau and Sensébé, 2018).
However, we still lack the knowledge of the cellular and molecular mechanisms that underlie joint-tissue homeostasis and its remodelling and repair in health and disease. Further understanding of these processes will help in the pharmacological modulation of the MSC niche to boost endogenous MSCs and to achieve regeneration of connective tissues. The ultimate aim of these efforts would be to delay the onset of this musculoskeletal degeneration, and in particular the degeneration of cartilage in patients with osteoarthritis and of bone in patients with osteoporosis. Both of these patient groups would benefit from empowering muscle strength as a support for the joints and bones. In addition, further studies of MSCs biology in health and disease would provide better strategies for these cell and/or cell-free therapies.