The skeletal system is among the largest of the human organ systems, constituting up to 15% of the total human body weight (Su et al., 2019). It allows functional body movement, protects internal organs, and serves as reservoir for minerals (Mizokami et al., 2017; Su et al., 2019); bones also have extra-skeletal endocrine functions (Su et al., 2019) that are essential for overall body homeostasis and systemic health (Ambrosi et al., 2019). Skeletal system functions are affected by skeletal shape, strength, and stiffness, which substantially change with the stage of development and age (Sheehan et al., 2018).

Advancing age is a key risk factor for degenerative bone and cartilage disorders, such as osteoporosis and osteoarthritis (Raisz and Seeman, 2001; Su et al., 2019; Jeong and Park, 2020), which lead to decreased mobility and diminished quality of life (Lee et al., 2014). However, bone mass, strength, and vitality are affected by other factors aside from age (Demontiero et al., 2012; Nandiraju and Ahmed, 2019). Alterations in cellular components, hormonal, biochemical, and vasculature status, which can be brought about by metabolic disorders, are examples of intrinsic factors (Demontiero et al., 2012), whereas nutrition, physical activity, injury, and comorbidities are some of the contributing extrinsic factors (Demontiero et al., 2012; Sheehan et al., 2018). Congenital or acquired skeletal deformities also affect the form and function of the skeletal system due to geometric abnormalities of the bones and articulating surfaces (Sheehan et al., 2018).

Some age-related defects in bones and cartilage have been attributed to changes in the populations and functions of stem cells in skeletal tissues (Jeong and Park, 2020). These molecular and functional changes in skeletal stem/progenitor cells (SSPCs) lead to a negative bone balance with reduced bone remodeling coupled with continued, or even accelerated, bone resorption (Demontiero et al., 2012). By itself, stem cell regeneration of large skeletal defects is difficult and often lead to the delay or failure of bone repair (Vidal et al., 2020). Confounded by aging and age-related diseases (e.g., diabetes), incidence of bone fractures and failure of large bone defect repair is further amplified. (Clark et al., 2017; Wu et al., 2021).

Currently, the goals of therapies for degenerative bone conditions are fracture prevention and decreased bone resorption through antiresorptive agents. For the reconstruction of critical-sized bone defects, transplantation of an autologous free vascularized bone flap containing the patient’s cells, growth factors, and a vascularization bed is the current gold standard approach (Vidal et al., 2020). However, these vascularized bone flaps are of finite supply, and their harvest can result in significant morbidity and anatomical incompatibility. Prosthetic and bio-matrix materials are also unable to restore complex sensory and motor functions, do not expand with age, and present a risk of failure and infection (Borrelli et al., 2020; Tang et al., 2021). Recently, in the field of tissue engineering and regenerative medicine, the use of SSPCs in combination with scaffolds and growth factors has been introduced to facilitate bone regeneration (Miller et al., 2017; Borrelli et al., 2020; Tang et al., 2021). Therefore, a better understanding of the properties and regulation of SSPCs with respect to their locations and skeletal compartments, as well as the effects of age, can potentially facilitate the discovery of new approaches to bone defect reconstruction and degenerative bone disease therapy.

SSPCs are essential for skeletal development, bone remodeling, and bone repair and are characterized by the capacity for self-renewal and multipotency (Matsushita et al., 2020c). Traditionally, they have been known to reside within the bone marrow (BM), but recent scientific advancements identified distinct SSPC populations in various skeletal compartments such as the growth plate (GP), periosteum, and calvarial sutures (Matsushita et al., 2020c; Jeong and Park, 2020) (Figure 1). Adult SSPCs are heterogeneous, and each population potentially contributes to bone maintenance and regeneration in a different manner (Ortinau and Park, 2021). The cellular and matrix environment of each distinct SSPC population is also believed to regulate SSPC stemness and to direct its roles at different stages of development, homeostasis, and regeneration (Matsushita et al., 2020c). However, it is largely unknown how these SSPCs are regulated, and which specific roles they play in these biological processes (Iaquinta et al., 2019; Ortinau and Park, 2021). In this review, we present the different regulation mechanisms during the development and repair of the distinct SSPC populations in major compartments, namely, the suture, GP, periosteum, and BM. We also discuss the currently known changes that occur in the regulation pathways of SSPCs with aging.

The musculoskeletal system develops from various embryonic origins, including: 1) the paraxial mesoderm, 2) the parietal layer of the lateral plate mesoderm, and 3) the neural crest cells (NCCs), which undergo mesenchymal condensation to begin bone formation (Figures 2, 3) (Mitchell and Sharma, 2009). Most facial bones, the cranial vault, clavicle, and calvarial frontal bones originate from NCCs through the intramembranous ossification process. By contrast, most of the remaining bones in the skull and all perpendicular bones develop from mesoderm-derived (MDD) cells through the endochondral ossification process (Chung et al., 2009; Sadler and Langman, 2012; Schoenwolf et al., 2015; Moore et al., 2016). Some intriguing tissues include the clavicle originating from NCCs but forms through mixed intramembranous and endochondral ossification, and the calvarial parietal bones which originate from MDD but are formed through the intramembranous ossification process (Percival and Richtsmeier, 2013). Intramembranous bones develop via direct osteoblast differentiation within the mesenchyme, while endochondral bones develop with an intermediate cartilage structure before being replaced by or transformed into bones (Sadler and Langman, 2012; Schoenwolf et al., 2015; Moore et al., 2016; Galea et al., 2021; Shu et al., 2021). Furthermore, mesodermal cells from different embryonic origins show different transcriptomic signatures and differentiation potentials, suggesting that tissue-specific SSPCs with different embryonic origins are present in different bones and that they require differential regulation pathways for bone regeneration (Sacchetti et al., 2016).

Intramembranous bone formation begins with the expression of Runx2 driving SSPCs into the osteoblast lineage (Pazhanisamy, 2013) (Figure 2). This is followed by the sequential expression of osterix (Osx), type I collagen, Bglap (or osteocalcin), and Spp1 (or osteopontin), which are the core osteoblast differentiation factors. As osteoblasts become surrounded by bone matrix, they express late markers such as the dentin matrix protein 1 (Dmp1). Lastly, the expression of the osteocyte marker sclerostin is observed (Pazhanisamy, 2013; Galea et al., 2021). In endochondral bone formation, however, Sox9 initiates chondrocyte commitment in the pre-condensing mesenchyme. During early chondrocyte development, Sox9, Sox5, and Sox6 are highly expressed and subsequently drive the expression of early cartilaginous matrix components type II collagen (Col II) and aggrecan (Galea et al., 2021). During the terminal hypertrophic stage of endochondral ossification, co-expression of cartilaginous (type X collagen) and osteoblastic (Runx2, Osx, bone sialoprotein) markers is observed. Sox9 expression persists in early hypertrophic chondrocytes and induces the expression of type X collagen while inhibiting Runx2 activity. Later on, the degradation of Sox9 protein relieves the inhibition of Runx2, thus allowing for chondrocyte-osteoblast transformation and subsequent mineralization (Taher et al., 2011; Galea et al., 2021). Overall, endochondral and intramembranous ossification use distinct molecular signals responsible for the different types of bone formation.

The expression of regulatory genes in specific cell types and locations in the body may also account for the differences in SSPC functionality. For example, a mandibular injury site undergoes osteogenic regeneration through Homeobox non-expressing (Hox ⁻ ) neural crest-derived (NCD) cells, while tibial or long bone injuries ossify with Hox⁺ MDD cells (Lappin et al., 2006) (Figure 3). Hox genes encode the Hox proteins, which are master regulators of embryonic development, but these genes continue to be expressed throughout postnatal life. In humans, they control body proportions, vertebral segments, and proper spatial development of organs and limbs (Lappin et al., 2006; Rux and Wellik, 2017). Upon transplantation of MDD cells to a mandibular injury site, Hox⁺ MDD cells remain Hox⁺ , while inherently Hox ⁻ NCD cells transplanted to the tibial injury site become Hox⁺ . This indicates a sense of positional identity that is unchanged with transplantation, and this positional identity is also seen in facial bone development (Leucht et al., 2008). The first pharyngeal arch (PA) cells, which are Hox ⁻ , form most of the facial bones in an intramembranous manner. The second PA cells, which are Hox⁺, form the hyoid, styloid, and stapes bones in an endochondral manner (Weber et al., 2021). Inactivation of Hox genes in the second PA results in intramembranous ossification, while their overexpression in the first PA results in second PA-like elements (Bonaventure, 2001; Kitazawa et al., 2022). Overall, these findings suggest that embryonic cell origin may confer a differentiation bias to SSPCs.

Although some distinct SSPCs may come from the same embryonic cells, they undergo further development in their distinct skeletal compartments. With the complex development in each compartment, dynamic and specialized microenvironments are also formed (Kurenkova et al., 2020). Most likely, these microenvironments provide an additional layer of unique regulation to each SSPC population aside from what is offered by embryonic cell origin (Donsante et al., 2021) (Table 1). Parallel niches may therefore be progressively altered, explaining the different properties and functions of distinct SSPCs through time and condition, which will be discussed later.

Calvarial bone formation begins around the third week of gestation (Jha et al., 2018). At this stage, NCCs expand and form a condensed mesenchyme. Capillaries then begin to surround the mesenchymal condensation which may serve as a vehicle for nutrient supply, osteoblastic factor transport, and conduit of additional NCCs and SSPCs (Percival and Richtsmeier, 2013; van Gastel et al., 2020). Next, the cells in the mesenchyme center start to differentiate directly into osteoblastic cells and generate an osteoid (calvarial bone primordia) which later become mineralized bones (Ishii et al., 2015; Kenkre and Bassett, 2018). The Bone Morphogenic Protein (BMP 2/4/7) signaling pathway and its immediate-early effector homeodomain transcriptional factor (Msx2) play major roles in the early specification of the calvarial bone primordia from NCCs by positively controlling the expression of Runx2. Transcription factor Foxc1, on the other hand, negatively regulates Msx2 and Bmp2/4 and positively regulates Noggin to prevent premature differentiation of the frontal bone primordia, thus promoting apical migration of undifferentiated progenitor cells. Wnt signaling is a key regulator in the early specification of primordia that favors the osteogenic lineage. Wntless, a Wnt ligand transporter, found in the cranial surface ectoderm and underlying mesenchyme, promotes expression of the Wnt ligands (Wnt5a/11/4/3a/16) and secretion of Wnt protein, activating Twist1 and then β-catenin downstream. Β-catenin promotes the osteogenic lineage but represses the chondrogenic lineage in the cranial mesenchyme. Interestingly, a haploid deficiency of Fgfr1 in suture cells switches their fate to form ectopic chondrocytes in the suture mesenchyme, suggesting that local Fibroblast growth factor (FGF) signals are necessary for their direct intramembranous ossification (Maruyama et al., 2010).

Once the primordia are established, osteogenic precursors migrate, through the EphrinA (EphA) signaling, to the edge of the growing bone, where they contribute to the apical expansion of the calvarial rudiments. Wnt signaling is still a prerequisite at this point for the final phase of osteoblast differentiation; TGF-β signaling, on the other hand, is required to maintain a continuous osteogenic lineage commitment (Ishii et al., 2015). Between calvarial bones, cranial sutures develop while allowing calvarial expansion for brain growth (Sadler and Langman, 2012). A study by Deckelbaum et al. (2012) identified a group of Sonic hedgehog (Shh)-responsive cells in the head mesoderm as precursors of the coronal suture. These cells migrate first to the supraorbital ridge transiently expressing En1, a vertebrate homolog of the Drosophila transcription factor engrailed, before apically migrating together with the calvarial rudiments to form the coronal suture (Deckelbaum et al., 2012). Other embryonic origins of the suture precursor cells still need to be identified.

Long bone formation becomes visible by the end of the fourth week of gestation. Limbs initiate with small bud formation as outpocketing from the ventrolateral body wall. These limb buds generate a core of mesenchymal cells from the somatic layer of the lateral plate mesoderm covered by a layer of ectoderm. An apical ectodermal ridge (AER) is located at the distal end of the limb and induces rapid mesenchymal cell proliferation without differentiation. FGF signals in the so-called progress zone control proximal to distal limb growth (Bonaventure, 2001; Sadler and Langman, 2012; Schoenwolf et al., 2015; Moore et al., 2016). A unique feature of endochondral bone formation is the moment when the cells move further from the influence of the AER, causing local FGF levels to decrease and allowing the mesenchymal cells to differentiate into cartilage. This is where endochondral ossification begins, and skeletal compartments subsequently develop (Figure 2). GP formation starts off with early cartilage development through chondrocyte proliferation and differentiation of Fgfr3 ⁺ cells in the mesenchymal condensation (Ono and Kronenberg, 2016; Zieba et al., 2020; Matsushita et al., 2022). The proliferating chondrocytes become mature and later organize, through Notch signaling, at both sides of long bones as a tri-layer GP consisting of resting, proliferating, and hypertrophic chondrocyte zones (Ono and Kronenberg, 2016; Zieba et al., 2020). The remaining cells form the outer layer called the perichondrium. All of the mesenchymal condensations in the forming limbs still remain avascular at this point (Percival and Richtsmeier, 2013). The spatio-temporal differences on angiogenesis may also explain the unique ossification processes with avascular state limiting supply of osteogenic factors and SSPCs that would promote osteogenesis (Percival and Richtsmeier, 2013; van Gastel et al., 2020). Eventually, hypoxic condition in limb forming cells promotes vascular invasion to the perichondrium leading to osteoblast differentiation, and development of perichondrium to periosteum and articular soft tissues (Percival and Richtsmeier, 2013). The periosteum becomes a layer of connective tissue housing the proliferating progenitor cells with chondrogenic and osteogenic differentiation properties, while the osteoblasts differentiating mostly from Dlx5 ⁺ cells of the periosteum form a bony collar around the shaft of limb bones (Vanderbeck and Maillard, 2019; Sottoriva and Pajcini, 2021; Matsushita et al., 2022). Subsequently, the marrow cavity forms as long bones develop, and blood vessels invade the cartilage template from the osteogenic perichondrium, which are maintained through Notch signaling (Vanderbeck and Maillard, 2019; Sottoriva and Pajcini, 2021). Blood-borne hematopoietic progenitors and BM stromal cells then seed this environment. While most of the BM stromal cells originate from the outer perichondral Dlx5 ⁺ cells, a minimal contribution of inner perichondrial Osx ⁺ cells and cartilage Fgfr3 ⁺ cells implicate that BM stroma may have transitions from primitive progenitor cells in early postnatal development to definitive SSPCs in adult bone homeostasis, respectively. This is exemplified by the transition of fetal Osx ⁺ SSPCs to more long term postnatal Osx ⁺ BM SSPCs, and a shift from a more proliferative fetal Dlx5 ⁺ SSPCs to a more quiescent postnatal Dlx5 ⁺ BM-SSPCs. (Mizoguchi et al., 2014; Matsushita et al., 2022). With age, a subset of BM stoma cells further shifts towards adipocyte development (Taher et al., 2011; Bianco and Robey, 2015).

The difference between calvarial vs. long bone is apparently observed in bone injury healing. In general, all bones heal through three overlapping processes, namely, inflammation, bone formation, and bone remodeling (Sheen and Garla, 2021). Immediately after bone injury, a hematoma develops, leading to inflammation of the injury site. Inflammatory cells migrate into the injury site and secrete various cytokines and growth factors like tumor necrosis factor-α (TNF-α), BMPs, and interleukins, subsequently attracting more inflammatory and osteogenic progenitor cells (Wang et al., 2020; Sheen and Garla, 2021). Bone regeneration then begins with callus formation and a gradual decrease in inflammation. A unique process of long bone fracture healing is the fibrocartilaginous callus formation that first appears at days 5–11. During this process, vascular endothelial growth factor (VEGF) allows angiogenesis, and BMP drives the differentiation of SSPCs into chondroblasts, osteoblasts, and fibroblasts. At the same time, woven bones begin to appear adjacent to the periosteal layer (Wang et al., 2020; Sheen and Garla, 2021). Later on, on days 11–28, when Sox9 protein is degraded, inhibition of osteogenic Runx2 is relieved, and a bony callus then forms as chondrocytes calcify with calcium–phosphate crystals, which is followed by bone replacement by osteoblasts (Sadler and Langman, 2012; Schoenwolf et al., 2015; Moore et al., 2016; Galea et al., 2021). In contrast, calvarial injury repair is normally completed by repeated cycle of bone resorption by osteoclasts and bone formation by osteoblasts without forming fibrocartilaginous callus (Lim et al., 2013; Wang et al., 2020), although some recent studies show endochondral ossification upon scaffold induced calvarial injury repair (Ko and Sumner, 2021). Macrophage-colony stimulating factor and receptor activator of nuclear factor kappa-B ligand (RANKL) are two critical cytokines for osteoclast differentiation (Castillo et al., 2017; Wang et al., 2020). These factors recruit osteoclast precursors, activate their fusion to form multinucleated pre-osteoclasts, and induce downstream signaling molecules (e.g., mitogen-activated protein kinase, TNF-receptor-associated factor 6, NF-κB, and c-fos) and key transcription factors (e.g., nuclear factor of activated T-cells [NFATc1]) that regulate osteoclast gene expression (Kenkre and Bassett, 2018).

In summary, calvarial and long bone development and remodeling leads to different bone morphologies and histologic characteristics. Although not all the regulatory pathways involved are the same (Table 1), Notch, BMP, TGFβ, Hedgehog (Hh) and Wnt/β-catenin signaling pathways play a key role in embryogenesis and regulation within the different bone compartments (Sottoriva and Pajcini, 2021). In addition, the balance between proliferation and differentiation is important in the development and remodeling of these compartments.

The elderly population has poor capacity for skeletal regeneration and a limited physiologic SSPC reserve (Lee et al., 2014) leading to degenerative conditions (Jeong and Park, 2020). SSPCs from older individuals have similar clonogenicity, but impaired osteochondrogenic differentiation, as compared to younger individuals (Ambrosi et al., 2020). Changes in hormones and sustained pro-inflammatory stimuli in aging might alter epigenetics regulators (Beerman and Rossi, 2015; Josephson et al., 2019). A recent study showing the downregulation of histone deacetylase Sirtuin1 in aged human SSPCs supports this hypothesis (Ambrosi et al., 2020). Moreover, downregulation of osteogenic genes (e.g., Wnt signaling), and upregulation of fibroblast-like ECM- and cellular senescence-related genes, were seen in aged human SSPCs, suggesting skewing towards stromal/fibroblastic states (Ambrosi et al., 2020). Excess or continuous inflammation in the elderly, and low-grade chronic inflammation associated with degenerative and cardiometabolic diseases, are known to inhibit the regeneration of various tissues including bones (Franceschi et al., 2018; Josephson et al., 2019). Activation of Nuclear factor-kappa B (NF-κB), a regulator of innate immunity, resulted in increased expression of the senescence genes Cdkn1a and Cdkn2a, suggesting its central role as a mediator of a pro-inflammatory state and SSPC aging (Albensi, 2019) (Josephson et al., 2019). Extensive proliferation may also lead to cellular exhaustion. Sustained FGF2 signaling in SSPCs and increased neutrophil-related ROS in HSC cause loss of quiescence and impaired regenerative capacity of SSPCs (Owusu-Ansah and Banerjee, 2009; Chakkalakal et al., 2012).

With age, the rest of the BM undergoes adipocyte conversion where fat cells progressively increase in number. Adipocytes then inhibit BM-SSPC functions (Bianco, 2011). Normally, adipogenesis involves sequential expression of CCAAT enhancer-binding protein beta (C/EBPβ), gamma (C/EBPγ), alfa (C/EBPα), and finally peroxisome proliferator-activated receptor gamma (PPARγ) from progenitor cells. Several transcription factors direct age-related shifts in BM-SSPC differentiation. MAF bZIP transcription factor (MAF), a binding partner of Runx2, is increased in the young but decreased in the old. With age, reduced MAF promotes adipogenesis through upregulation of PPARγ and the suppression of osteogenesis (Nishikawa et al., 2010). Forkhead box P1 (FOXP1) also declines with age, losing its anti-adipocyte interaction with C/EBPβ and pro-osteogenic repression of Notch signaling pathway, all leading to bone loss (Li et al., 2017). Core-binding factor subunit beta (CBFβ) is another key co-factor of Runx2 that is reduced with aging. Normally, CBFβ inhibits adipogenic gene expression and enhances Wnt/β-catenin signaling (Wu et al., 2014; Wu et al., 2017). An increase of BM adipocytes with age is also associated with the gradual decrease of Adipoq ⁺ expression, potentially facilitating MALPs differentiation to adipocyte cells. Expression of Cxcl12 is also reduced, further leading to BM atrophy and adipogenesis (Zhang et al., 2021). Furthermore, BM adipocytes may also inhibit BM-SSPCs function by physically blocking blood flow through the sinusoid. The larger-sized adipocytes can compress the sinusoid, leading to its collapse (Bachman et al., 2002).

Unlike most bones, cranial bones are rarely affected by osteoporosis. However, the same mechanisms leading to this condition can also reduce cranial bone mass density and regeneration capacity (Cotofana et al., 2018; Hudieb et al., 2021). Radiographic and histologic studies showed a decreased computed volume of the calvaria and a lateral expansion of the skull, favoring a skeletonized facial appearance in elderly individuals (Cotofana et al., 2018). Increased soft tissue laxity and decreased fat (Cotofana et al., 2016) can contribute to increased bone resorption in elder individuals due to decreased mechanical loading. The tensile strength of the dura also decreases with age; alterations in collagen fiber organization may cause this change in dura properties, which ultimately affects the ECM of the tissue (Zwirner et al., 2019; Panteleichuk et al., 2021). The osteogenic activity of the dura also tends to be less active with age (Wan D. C. et al., 2008), probably due to the absence of skull growth-induced mechanical strain (Wang et al., 2020).

Further understanding of which cellular and molecular changes SSPCs undergo during stress, aging, and pro-inflammatory conditions, and which regulatory mechanisms control these changes, will offer new approaches to the treatment of bone diseases through the ages.

Calvarial and long bones are unique types of bone that are distinctively regulated but show subtle similarities in the involved pathways. In both types of bone, multiple types of distinct SSPCs are present and interact with each other to achieve skeletal development. Despite improvements in our understanding of SSPCs, the different functional responses and regulations of SSPCs in various locations, especially during injury, have not been thoroughly studied. Although some essential molecular regulators are shared by distinct SSPCs, their effect on differentiation, cell fate, or tissue type formation of distict SSPCs can be different. As such, local SSPCs contribute uniquely to their bone development, homeostasis, and regeneration. Different conditions (e.g., injury, stress, aging) result in different regulations as well. Studies looking into these differences are currently inadequate. Possible regulatory differences directing the rate of defect or injury healing in the craniofacial area and long bones have not been extensively investigated despite this long-observed difference.

Long bone is important for addressing the mechanical loading throughout life, while craniofacial deformities are important not only for the physiology of the organs in the craniofacial area but also for the quality of life of patients in general. Thus, studies pertinent to enhancing healing of both long bones and craniofacial bones may have to be given equal importance. The unique effect of the limited presence of ligaments, tendons, and muscles in the craniofacial area, as compared to long bones, is an interesting area of research. Characterization of the heterogenous SSPC population in the BM, and the regulatory mechanisms by which they contribute to BM maintenance, could be continued and expanded as a research initiative. Single-cell approaches together with in vivo, and ex vivo functional studies appear to be a powerful approach to facilitate SSPC characterization and biology. Additionally, the single-cell approach would allow further analyses on differential gene expression and the regulatory mechanisms established between cell populations, skeletal compartments, and cell conditions (development vs. injury vs. aging).

Advancements in SSPC research and interest in the aging bone have allowed the discovery of more unique populations such as those adipocyte marker-expressing cells that do not undergo adipogenesis, but rather unexpectedly remains undifferentiated in the marrow with marrow stroma and cortical bone maintenance roles. Further research could also be done to deepen our understanding of how each multiple types of SSPCs relate to each in the context of development, regeneration, and aging. These differences are necessary for designing specific tissue engineering and regenerative medicine therapies for bone repair.