Introduction

Osteoarthritis (OA) is a degenerative joint disease mainly characterized by low-grade inflammation, cartilage degradation, and subchondral bone remodeling, which together lead to pain and disability.1 Due to population aging, the incidence of symptomatic OA has increased markedly over the past few decades. According to a Global Burden of Disease study from 2017, among 354 other diseases and injuries, OA is the fifth leading cause of disability worldwide and is therefore the cause of large social health and economic burdens.2 Total knee arthroplasty (TKA) is a very successful surgical treatment to reduce knee pain and improve mechanical function for patients with end-stage OA, but it can also be accompanied by several complications, including aseptic loosening, prosthetic joint infection, and periprosthetic fracture. No treatments have been established to slow or reverse OA progression, often resulting in years of increasing pain and disability until TKA is indicated.

There are several risk factors involved in the pathogenesis of OA, such as aging, genetics, obesity, and injury, but biomechanical overload is one of the primary risk factors for OA.3,4 Previous research generally supports the conclusion that exercise of moderate intensity and/or most sporting activities that do not cause traumatic injury do not substantially increase the risk of OA.5,6 Additionally, clinical trials and in vivo animal studies have shown that proper exercise interventions truly relieve OA pain and improve joint function;7,8 therefore, structured land-based exercises are currently recommended as core treatments for knee OA.9 Notably, the effects of mechanical force on articular cartilage and chondrocytes depend on the force type, magnitude, frequency, duration, and orientation.

Mechanistically, physical forces are converted into biochemical or electrical signals to induce a cellular response. This special cellular process is called mechanotransduction. New discoveries of putative mechanotransduction elements are quickly expanding our understanding of this unique cell signaling process.10 For instance, estrogen receptor α (ERα) has recently been found to play a critical role in governing chondrocyte responses to compressive loading.11 Previously, ERα had been shown to be an important mechanical force -mediator in bone formation via the action of a ligand-independent pathway.12 Therefore, the novel function of ERα in converting forces into biochemical changes in chondrocytes needs be further investigated. In this review, we first briefly summarize the current understanding of mechanotransduction in chondrocytes and then explore the potential interaction of ERα with other molecules in regulating the chondrocyte response to loading. Finally, we propose several future research directions that can further our understanding of the role of ERα in mediating the cellular response to biomechanical cues under specific physiological and pathological conditions.

Three-tiered cascade of mechanotransduction pathways in chondrocytes

Previous studies have led to the identification of many cellular structures and molecules that mediate chondrocyte responses to force, including primary cilia, mitochondria, ion channel proteins, transcription factors, inflammatory cytokines, microRNAs, etc.13,14 Organizing these molecules based on their major functions would be helpful. To this end, some studies have provided a rough classification of mechanosensitive structures and molecules based on their roles in the mechanoresponses. For example, the nuclear localization of Yes-associated protein (YAP) is influenced by the ion channel protein Piezo1 in neural stem/progenitor cells and osteoblasts, and activated YAP promotes the expression of downstream targets, such as collagen-II.15,16 Moreover, recent discoveries are converging on a model in which multiple types of mechanical inputs in a variety of cellular settings influence the conformation and tension of the F-actin cytoskeleton, and then, a change in the cytoskeleton activates transcription factors, such as YAP, to generate downstream biological effects.17 Inspired by these studies, we propose three categories based on the principal role of the elements involved in mechanotransduction signaling, including sensors, transducers, and impactors (Fig. 1). Specifically, mechanosensors are defined as the elements that directly respond to mechanical loading by converting mechanical signals into biochemical signals and are primarily composed of cellular structures, including ion channels, cytoskeleton, primary cilia, etc. Mechanoimpactors are the final outputs of mechanical stimulation and can lead to changes cell phenotypes/functions and the extracellular matrix (ECM) compositions and structures; mechanoimpactors include proinflammatory cytokines, energy metabolism factors, and cell cycle regulators.18,19,20 Due to the widespread range of mechanoimpactor effects on cellular activity, mechanoimpactors are not discussed separately herein. In contrast, they are described in the context of specific sensors or transducers. Mechanotransducers are molecules for which clear upstream signaling leads to their activation and can also alter specific downstream molecules. Mechanotransducers connect the actions of mechanosensors and mechanoimpactors; they include kinases and transcription factors. A representative example demonstrating a three-tiered cascade is the Piezo protein–YAP–collagen axis. Piezo proteins (Piezo1 and Piezo2) are nonselective Ca2+-permeable cation channels that are also mechanosensors in various cell types.21,22 As a transcription factor, YAP has been shown to be a major downstream mechanotransducer activated by Piezos in osteoblasts, in which it regulates the expression of type II and type IX collagen in osteoblasts.15 Despite the conceptual simplicity of the three-tiered cascade in chondrocyte mechanotransduction, certain elements may exhibit more than one function. For example, mitochondria can be sensors, mediators, or impactors in mechanotransduction pathways.

Fig. 1
figure 1

The three-tiered cascade of mechanotransduction pathways in cells. The first structures influenced by mechanical stimulation are different types of mechanosensors, including pericellular matrix (PCM) and extracellular matrix (ECM) components, mechanically gated ion channels and porins (such as TRPV4 and Piezo), integrins, mitochondria, cytoskeleton, and primary cilia. Mechanotransducers include multiple types of factors, such as kinases (ROCK, PI3K-AKT, MAPK, etc.), growth factors (IGF, TGF-β, etc.), transcription factors (YAP, GREB1, etc.), secretory proteins (RCN2, etc.), and other molecules, which are activated by mechanosensors and then alter specific downstream molecules. Mechanoimpactors are the final outputs of mechanical stimulation and cause changes in cell phenotypes and behaviors, such as cell migration, survival, growth, energy metabolism, and inflammation

Full size image

Mechanosensors of chondrocytes and their roles in OA pathogenesis

Mechanosensors respond to mechanical stimulation that alter the molecular/cellular structures to induce electrochemical and biological signals. Since mechanosensors are the initiators of signaling cascades, we summarize current mechanotransduction pathways based on mechanosensing effects (Fig. 2). In addition, the related mechanotransducers and mechanoimpactors in each pathway are discussed in the following sections. The functions of mechanotransduction-relevant molecules in maintaining cartilage health and promoting OA pathogenesis are detailed in Table 1.

Fig. 2
figure 2

Different mechanosensors and their downstream pathways in articular chondrocytes. Pointed arrows indicate stimulatory effects, whereas blocked lines indicate inhibitory effects. IGF-1: insulin-like growth factor-1, TRPV4: transient receptor potential channel subfamily V member 4, GSK1016790A: TRPV4 agonist, GSK2193874: TRPV4 antagonist, GSK-3β: glycogen synthase kinase 3beta, HDAC6: histone deacetylase 6, PGE2: prostaglandin E2, NOS2: nitric oxide synthase, ACAN: aggrecan, ADAMT: a disintegrin and metalloproteinase with thrombospondin motifs, TGF-β: transforming growth factor-beta, GAG: glycosaminoglycan, IL: interleukin, NO: nitric oxide, COL2: type II collagen, GsMTX4: selective Piezo-inhibiting peptide, NF-κB: nuclear factor κB, F-actin: filamentous actin, MAPK: mitogen-activated protein kinase, ERK: extracellular signal-regulated kinase, PCM: pericellular matrix, CTGF: connective tissue growth factor, FGF: fibroblast growth factor, BMP: bone morphogenic protein, FAK: focal adhesion kinase, PI3K: phosphatidylinositol 3-kinase, JNK: c-Jun amino-terminal kinase, MMP: matrix metalloproteinase, IFT80/88: intraflagellar transport 80/88, COLX: type X collagen, RUNX2: Runt-related transcription factor 2, and AMPK: AMP-activated protein kinase

Full size image
Table 1 List of known mechanosensors in chondrocytes and their potential downstream genes and pathways
Full size table

Mechanically gated ion channels and porin-mediated pathways

Mechanically gated ion channels and porins are the most best characterized mechanosensors in chondrocytes. In general, these transmembrane proteins undergo structural changes upon mechanical loading and subsequently alter calcium flux. Calcium is second messenger that further influences intracellular signaling pathways.23

Transient receptor potential cation channel subfamily V member 4 (TRPV4), a Ca2+-permeable osmomechano-TRP channel, is highly expressed in articular chondrocytes and mediates the cell response to mechanical loading.24 Activation of TRPV4, via moderate mechanical loading or selective chemicals, has been found to enhance anabolic and suppress catabolic gene expression in cultured chondrocytes.19 In contrast, inhibiting TRPV4 reduces the anabolic response to mechanical loading. Clark et al. discovered that TRPV4-deficient mice (TRPV4−/−) exhibited severe osteoarthritic changes.25 Interestingly, the role of TRPV4 was affected by sex and age. For example, male TRPV4−/− mice presented with a greater calcified meniscal volume and cartilage damage than age-matched female mice. To investigate the mechanisms involved in these outcomes, Fu et al. used multiple stimuli, including mechanical loading (10% cyclic tensile strain, 0.33 Hz), osmotic stimulation, and chemical agonist stimulation to activate TRPV4 in primary chondrocytes and articular cartilage explants in vitro. They found that TRPV4 could inhibit interleukin (IL)-1β-mediated nitric oxide (NO) and prostaglandin E2 (PGE2) release and prevent cartilage degradation.26 The results strongly suggested that TRPV4 slows OA progression by inhibiting inflammatory pathways. However, contradictory results have also been reported. Specifically, Agarwal et al. found that treating chondrocytes with GSK101, a small-molecule agonist of TRPV4, increased the levels of intracellular Ca2+ and phosphorylated GSK3β, leading to a significant decrease in the gene expression of collagen type II and an increase in the levels of proinflammatory cytokines.27 Notably, the conclusion was drawn based on the use of a synthesized small molecule, which may not be an exclusive agonist for TRPV4. The potential influence of this inhibitor on other molecules may have confounded its effects on TRPV4 activation. Another possible reason for the discrepant findings, as reported in the same study, involves disruption to the TRPV4–GSK3β molecular axis in OA chondrocytes. In summary, it is likely that activating TRPV4 may impact normal and OA chondrocytes differently, a possibility that needs to be further investigated.

Piezo1 and Piezo2 are two important ion channels in chondrocytes. In contrast to TRPV4, which is primarily thought to play a protective role in cartilage, recent evidence has suggested that Piezo 1 and 2 are activated when injurious forces are applied. For example, a recent study by Lee et al. showed that activation of Piezo1 and Piezo2 after detrimental loading induced cartilage degradation and inflammation, and these effects were blocked by Piezo-specific blocking peptides and Piezo1- or Piezo2-siRNA.28 Servin-Vences et al. performed an RT‒qPCR analysis to determine the number of Piezo1, Piezo2, and TRPV4 transcripts in chondrocytes and found high expression levels of TRPV4 and Piezo1 but not Piezo2. Interestingly, both TRPV4 and Piezo1 contribute to currents activated by stimulation applied at cell–substrate contact sites, but only Piezo1 mediated stretching-activated current.29 These studies suggested that the exact roles of Piezo1 in chondrocytes need to be further elucidated, and they also highlight the overlapping function of ion channels in transducing mechanical signals.

Furthermore, inflammatory cytokines, such as IL-1α, have been shown to promote the expression of Piezo1, which results in excessive steady-state calcium levels and hypersensitivity of chondrocytes to mechanical loading.30 Given the connection between activation of Piezo 1/2 and OA progression, cytokines serve as rational targets for developing disease-modifying drugs.

Although TRPV4 and Piezo1 represent the two most important ion channels involved in mechanotransduction, their interaction in chondrocytes is, surprisingly, not fully understood. Du et al. indicated that TRPV4-mediated calcium signaling plays a central role in the chondrocyte response to physiological strain levels (3% and 8%), while Piezo1/2-mediated calcium signaling was critical for injury caused by various strain levels (13%–18%).31 In the same study, knockout of TRPV4 significantly suppressed stretching (3% strain)-induced calcium flux but did not influence the cell response to injurious stretching (>13% strain). Servin-Vences et al. reported similar results: depletion of TRPV4 did not affect the Piezo1-dependent response of murine chondrocytes to stretching.29 Therefore, even though these ion channels all function through calcium flux, they can act independently. In fact, in a study with human periodontal ligament cells (hPDLCs),32 Shen et al. found that extracellular signal-regulated kinase (ERK) participated in mechanotransduction mediated by the Piezo1 channel but not that mediated by TRPV4. Therefore, understanding the specific downstream pathways activated by these proteins is critical to tease out the different roles played by TRPV4 and Piezo1/2 in mediating the chondrocyte response to various mechanical stimuli.

Cilia-mediated pathways

Cilia are microtubule-containing structures that extend from the cell surface into the extracellular space. They can directly participate in cell sensing of different stimuli, including physical force. Thompson et al. revealed that intact cilia activated the hedgehog signaling pathway and induced the expression of a disintegrin and metalloproteinase with thrombospondin motifs-5 (ADAMTS-5) under 10% cyclic tensile strain (CTS).33 However, this response was lost when the strain magnitude was increased up to 20% due to cilia disassembly due to excessive stretching of chondrocytes. Interestingly, the inhibition of histone deacetylase 6 (HDAC6) prevented cilia disassembly and restored mechanosensitive hedgehog pathway function under 20% CTS.33 Moreover, chondrocytes in OA cartilage carry cilia with different characteristics than chondrocytes om healthy cartilage.34 For example, interleukin-1-treated primary chondrocytes exhibited a 50% increase in cilia length after 3 h of exposure.35 However, how this elongation of cilia affects the chondrocyte response to force needs to be further investigated.

ED-rich tail 2 (CITED2) is another target that is regulated by primary cilia when chondrocytes are subjected to load strain. Abolishing primary cilia by knockdown of intraflagellar transport protein (IFT88) blocked the transactivation of CITED2 and upregulated the expression of metalloproteinase (MMP)-1 and MMP-13 (ref. 36). Juhász et al. applied a uniaxial intermittent cyclic load to generate hydrostatic pressure and fluid shear stress on chondroprogenitor cells.37 The results showed that the activity of protein kinase A (PKA) was enhanced along with its downstream transcription factor CREB, together activating the chondrogenic transcription factor Sox9 to promote the generation of collagen II and other extracellular matrix material.

In summary, primary cilia function as mechanosensors, and their activity is mediated through two mechanisms, namely, as classical cell sensors of fluid flow and pressure38 and modulators of the PKA pathway, which depends on primary cilia length in chondrocytes.35 Moderate mechanical stimuli activate primary cilia and promote the secretion of extracellular matrix components, while excessive loading can induce cilia disassembly and result in injuries to cells.

Integrins and ECM-mediated pathways

Integrin family members are widely expressed in articular chondrocytes as important cell adhesion molecules. Various types of integrins are expressed in chondrocytes, including α1β1, α3β1, α5β1, α10β1, αVβ1, αVβ3, and αVβ5.39,40,41 They play critical roles in the interactions between chondrocytes and the ECM. Integrins are mechanosensors in various cell types and are present in chondrocyte cilia. Zhou et al. incubated C-28/I2 and C-20/A4 human chondrocyte cell lines with an anti-β(1) integrin (CD29) function-blocking antibody and then applied mechanical stimulation to the monolayer cell culture. The results showed that the mechanical stimulus activated Jun N-terminal kinase (JNK) and increased proteoglycan synthesis, which was partially dependent on integrin.42 Another study proved that hydrostatic compressive forces (HCFs) significantly increased the expression levels of integrin α2, integrin α5, and integrin β1 and thus increased the phosphorylation levels of FAK, ERK1/2, and PI3K in a dose-dependent manner.43 It also revealed that HCFs enhanced the viability of chondrocytes by inhibiting apoptosis through the integrin-FAK-ERK/PI3K pathway. Additionally, previous studies had demonstrated that integrin-mediated chondrocyte responses to mechanical stimulation promoted the secretion of interleukin 4 (IL-4), which resulted in membrane hyperpolarization.44 Studies of cytokine effects on chondrocytes in vitro have suggested that IL-4 alters the ratios of MMP to TIMP, favoring TIMPs by suppressing IL-1–stimulated MMP3 production.45,46 This chondroprotective response is absent in chondrocytes from OA cartilage.47 Activation of integrins by mechanical loading has also been shown to elevate the gene expression of aggrecan and decrease the expression of MMP-3. Interestingly, these changes are blocked by anti-IL-4 antibodies, highlighting the chondroprotective role of IL-4 in connecting integrins and mechanoimpactors.47

In hyaline cartilage, chondrocytes are surrounded by a dense extracellular matrix. Based on the distance to the chondrocyte membrane, cartilage ECM is divided into three parts: the pericellular, interterritorial, and territorial matrices. Specifically, the pericellular matrix, which is primarily composed of collagens (e.g., VI, IX, X, and XI), noncollagenous proteins, and proteoglycans,48 forms the immediate microenvironment of chondrocytes. Since all physical forces influence cells only through the ECM, ECM components are considered to be mechanosensors.49 However, the downstream targets of these ECM-mediated effects of force in cells are not clear. Given that the ECM mostly functions primarily through binding integrins, the deformation of the ECM may alter ECM component interactions with integrins, thereby changing integrin-relevant pathways. In addition, ECM components may regulate the level and state of the ion channels described above. For instance, Zelenski et al. showed that knockout of type VI collagen increased chondrocyte swelling, which may be mediated by elevating osmosis-induced TRPV4 signaling.50

In summary, the integrin family plays a critical role in mediating interactions between cells and the extracellular matrix, but the precise function of each integrin as a mechanosensor in chondrocytes has not been fully elucidated. ECM components can couple with multiple exoproteins, such as integrin and ion channels, to form conjugated mechanosensing complexes. Decoding the pathways involved in their conjugation and the mechanosensing modes of these complexes is essential for understanding the behaviors of chondrocytes under mechanical stimulation. Notably, the microfibril alignment and composition in the ECM are variable extensively in different cartilage layers.51 For example, the superficial layer contains a relatively small amount of proteoglycan, and the fibers are arranged parallel to the joint surface.51 Farther into the articular cartilage, the collagen fibers are oriented perpendicular to the joint surface. Moreover, collagen fibrils in the middle layer exhibit a larger diameter than those in the superficial layer, with a higher concentration of proteoglycans and lower amounts of water and collagen in the former. In the deep zone, collagen fibrils with the largest diameter, the highest concentration of proteoglycans, and the lowest concentration of water are found. The biomechanics of articular cartilage leverage these microanatomical features to reduce the forces of friction across the joint to extremely low values.51 We believe that these structural and compositional differences in the ECM in different zones influence the mechanical response of resident chondrocytes, which deserves further investigation.

Cytoskeleton- and nucleoskeleton-mediated pathways

Since cells are not rigid, mechanical loading of sufficient intensity can directly cause cell deformation, which results in dynamic changes to the cytoskeletal and nucleoskeletal structure.52 The cytoskeleton is a complex three-dimensional network consisting of cell microtubules, microfilaments, and intermediate fibers. The nucleoskeleton is also a three-dimensional network structure within the nuclear envelope, and it includes the nuclear matrix, nuclear lamina and pore complex, and chromosomal skeleton. Notably, the cytoskeleton is structurally integrated with the nucleoskeleton, and these integrated areas are called linker of the nucleus to cytoskeleton (LINC) complexes, and they enable the synergistic effects induced by mechanical stimuli.53 Notably, a dynamic change in skeleton structures can facilitate the movement of intracellular organelles, such as cell nuclei and mitochondria,54,55 and further influence chondrocyte DNA concentration and gene expression through signalling mediated by the nucleoskeleton.56,57 Reynolds et al. found that the disruption of the nuclear envelope associated with lamin A/C depletion significantly increases nuclear strain in regions of low DNA concentration. The increased nuclear peak strain shifts the phenotype of chondrocytes to a fibroblast-like phenotype by increasing the contractility of the actin cytoskeleton.57

In addition to the mechanosensing and mechanotransduction roles of the cytoskeleton, mechanical loading can alter the cytoskeletal structure. In applying dynamic osmotic loading to chondrocytes and using phalloidin staining to observe changes to the actin cytoskeleton, Chao et al. found that dynamic osmotic loading resulted in a more uniformly distributed actin cytoskeleton.58 Erickson et al. showed that osmotic stress-induced transient increases in Ca2+ levels and simultaneously caused a reorganization of intracellular actin through a mechanism that required Ca2+ in the extracellular medium.59 The results also indicated that cytoskeletal reorganization was partially mediated by Ca2+ flux. Overall, there is consensus indicating that the cytoskeleton and nucleoskeleton are vital structures through which cells sense mechanical stimuli, but the precise effects exhibited by these elements in diverse biological activities have not been explored.

In summary, mechanical loading on the cytoskeletal structure not only can influence organelles but can also directly affect cytoskeleton-associated pathways and molecules. For example, previous research revealed that mechanical force-related actin depolymerization “frees up” ROCK to mediate an increase in Col2 expression.60 Zhen et al. found that mechanical overload (via cellular contractile forces >40 pN) activated TGFβ mediated through actin filaments, and overexpressed TGFβ induced cartilage degeneration in OA.41 In the future, researchers should continue to investigate these mechanosensitive cytoskeleton-associated pathways and molecules and clarify their roles in the progression of OA.

Mitochondria

Previous research has suggested that mitochondria act as mechanosensors and impact a series of downstream signaling pathways by mediating cellular responses to mechanical loading.61,62 Through mitochondrial links to the cytoskeleton, mechanical loading may directly cause mitochondrial deformation. An instance of high kinetic energy produced by mechanical loading induced chondrocyte injury and caused acute mitochondrial dysfunction and altered cellular respiration. Sauter et al. applied two agents (cytochalasin B and nocodazole) that promoted the dissolution of microfilaments and microtubules in chondrocytes and found that both agents significantly reduced the impact-induced oxidant release from mitochondria, thereby preventing cell death.63 In accordance with previous results, Bartell et al. tracked the structural changes in mitochondria after mechanical impact by using electron microscopy to reveal mitochondrial loss of fission and cristae structures within a few minutes of mechanical injury.64 The Szeto–Schiller (SS) peptide, a newly developed mitochondrial-protecting drug that selectively targets the inner mitochondrial membrane, has been shown to protect the mitochondrial structure and preserve mitochondrial function in chondrocytes that had been subjected to mechanical overload.64 These results strongly indicate that mitochondria are important mechanosensors and play essential roles in the mitochondrial-cytoskeletal linkage that is involved in mechanically induced cartilage degeneration.

Respiratory chain complexes are parts of the mitochondrial biophysical structure, which plays a central role in energy metabolism and adenosine triphosphate (ATP) production. Moreover, the respiratory chain is also the primary source of superoxide radicals in various cell types, including chondrocytes. Through dynamic loading of bovine osteochondral explants, Wolff et al. revealed that physiological loading stimulated ATP production and the release of reactive oxygen species (ROS) and that these effects were significantly suppressed by the mitochondrial ROS scavenger MitoQ.54 A small amount of ROS generated under physiological loading are involved in intracellular signaling, which promote the phosphorylation of AMP-activated protein kinase (AMPK)to regulate chondrocyte metabolic homeostasis.61 The processes that produce and eliminate ROS in cells are maintained in dynamic equilibrium. For example, superoxide dismutase 2 (SOD2) metabolizes ROS in mitochondria and attenuates their harmful effects in cells. However, excessive mechanical loading impairs respiratory chain complex function, leading to the release of large amounts of ROS from mitochondria and downregulated SOD2, causing the breakdown of the SOD2-centric oxidation‒reduction balance and leading to chondrocyte apoptosis and cartilage degeneration.65,66

In recent studies, considerable attention has been focused on mitochondrial dysfunction induced by mechanical loading in OA, but the underlying mechanisms are often overlooked. Importantly, as highly dynamic organelles, mitochondria remain functional via the balance of fission and fusion. Mitochondrial biogenesis and mitophagy are two additional processes that contribute to their dynamic balance.61 Investigating the effects of mechanical stimulation on these mitochondrion-related processes is important for understanding the mechanosensing role of mitochondria in chondrocytes.

Transcription factors in mechanotransduction pathways in chondrocytes

Transcription factors (TFs) are proteins that bind to the cis-acting promoter element to convert or transcribe DNA into RNA, thus playing important roles in transducing the signals from mechanosensors into cell nuclei, in turn amplifying the effects of mechanical stimulation. In Fig. 3, we summarize representative transcription factors that are known to mediate chondrocyte responses to mechanical stimuli. Their functions are detailed in Table 2. Notably, the molecules listed as mechanosensitive TFs also respond to other factors or cellular activities. To the best of our knowledge, there is no TF that is exclusively activated by mechanical stimulation.

Fig. 3
figure 3

Representative transcription factors that mediate chondrocyte responses to mechanical stimulation. Different colors indicate different pathways. Red indicates a supportive effect, pointed arrows indicate stimulatory effects, and blocked lines indicate inhibitory effects. YAP: Yes-Associated Protein, SAV: salvador, MST1/2: serine/threonine kinases STK4 (MST1) and STK3 (MST2), LATS1/2: large tumor suppressor 1 and 2, MOB1 A/B: MOB kinase activator 1, RhoA: Ras homolog gene family member A, TAZ: transcriptional coactivator with PDZ-binding motif, TAK1: Tgf-β-activated kinase1, IL1β: interleukin-1β, VEGF: vascular endothelial growth factor, HIF-1α: hypoxia-inducible factor 1-α, COL2: collagen type II, PRG4: human proteoglycan 4, ACAN: aggrecan, SOX9: SRY-related high-mobility-group-box gene 9, CREB1: cAMP-response element-binding protein, CCN2: cellular communication network factor 2, TGF-β: transforming growth factor-β, Smad2/3: SMAD family member 2/3, RUNX2: Runt-related transcription factor 2, NRF2: nuclear factor erythroid2-related factor 2, AMPK: AMP-activated protein kinase, IkB-α: inhibitor of NF-κB-α, ROS: reactive oxygen species, Rac-1: Ras-related C3 botulinum toxin substrate 1 (Rac1), GAG: glycosaminoglycan, TNFα: tumor necrosis factor α, iNOS: inducible nitric oxide synthase, HIF-2α: hypoxia-inducible factor-2α, NLRP3: nucleotide-binding oligomerization domain, leucine-rich repeat and pyrin domain-containing 3, CARD: caspase raise structural domain, MMP13: matrix metalloproteinase 13, ADAMT5: a disintegrin and metalloproteinase with thrombospondin motifs-5 (ADAMT-5), EBPβ: CCATT enhancer-binding protein β (C/EBPβ), AP-1: activator protein 1, JNK: c-Jun N-terminal kinase, ERK: extracellular signal-regulated kinase, MAPK: mitogen-activated protein kinase, MKK7: mitogen-activated protein kinase kinase 7

Full size image
Table 2 Mechanosensitive transcription factors (TFs) in chondrocytes
Full size table

Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)

The nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) family includes five members: RelA (p65), RelB, c-Rel, NF-κB1 (p50), and NF-κB2 (p52).67 The most common form of NF-κB is a heterodimer of the p50 and p65/RelA proteins, which can activate the expression of TNF-α, IL-1β, IL-6, iNOS, COX2, chemokines, and adhesion molecules, among other proteins, and participate in inflammatory and immune responses in joints.68 In its inactivated state, the heterodimer comprising p50 and p65 usually combines with inhibitors of NF-κB (IκB). IκB can be phosphorylated by inhibitor of kappa B kinase (IκK) and then degraded via proteases to release the heterodimer of NF-κB, which plays roles with TFs. NF-κB has been identified as a mediator of mechanotransduction in several cell types. For example, an increase in the translocation of activated NF-κB to the nucleus has been observed in a number of cell types under mechanical stimuli, including chondrocytes,69 osteoblasts,70 and vascular endothelial cells.71 Chang et al. revealed that ras-related C3 botulinum toxin substrate 1 (Rac1) was overexpressed under excessive mechanical loading to promote the generation of ROS. Then, ROS phosphorylated IκBα (dual Ser32/36) to release the p65/p50 dimer, thus leading to joint inflammation and cartilage degradation.4 The activation of Rac1 has been shown to associate with alpha6/beta4 integrin.72 Changes in the formation or dissolution of actin filaments and microtubules can also affect the nuclear localization of NF-κB. In particular, the nuclear localization of activated NF-κB is correlated with the microtubule-based motor protein dynein.73 Mikenberg et al. used colchicine and vincristine to disrupt cellular microtubules and found ectopic accumulation of NF-κB in the nucleus and reduced NF-κB-dependent transcriptional activity.74 Moreover, the treatment of cells with low concentrations of TN16 (25 and 50 nmol·L−1), which suppressed microtubule dynamics without visibly affecting microtubule organization, enhanced the association of NF-κB with microtubules and facilitated the nuclear translocation of NF-κB.75,76 Interestingly, moderate mechanical stimulation inhibited the nuclear translocation of NF-κB and maintained cartilage health. For example, Yang et al. showed that cyclic tensile strain alleviated the chondrocyte damage induced by IL-1β by activating AMP-activated protein kinase (AMPK) phosphorylation and suppressing the nuclear translocation of nuclear factor (NF)-κB p65 (ref. 77). Additionally, moderate-intensity treadmill exercise inhibited the nuclear translocation of NF-κB p65 and the formation of nucleotide-binding and oligomerization domain-like receptor-containing protein 3 (NLRP3) by downregulating tumor necrosis factor-related apoptosis-induced ligand (TRAIL).78

Yes-associated protein (YAP)

Previous research has strongly suggested that YAP is a downstream molecule of the ion channel protein Piezo1, and it plays crucial mechanotransduction functions in Piezo1-activated pathways.15,16 However, few studies have reported on the relationship between YAP and ion channel proteins in chondrocytes. Xie et al. embedded chondrocytes into a biomaterial scaffold to create a cartilage-like hydrogel and exposed this construct to mechanical stimuli. The results showed that the expression levels of YAP, Piezo1, and TRPV4 increased with increasing dynamic loading time.79 Another study reported that G protein-coupled estrogen receptor (GPER) promoted the expression and nuclear localization of YAP under mechanical stress, but it inversely inhibited the expression of Piezo1 in chondrocytes.80 The same study also showed that YAP was significantly decreased in degenerated cartilage, and silencing YAP caused chondrocyte apoptosis. However, contradictory results of YAP function in OA were reported in other studies. Zhang et al. constructed YAP conditional knockout (cKO) mice and found that cKO-YAP mice presented with an unchanged level of collagen type II and with cartilage that had been protected from degeneration in an OA model. Furthermore, intra-articular injection of the YAP-selective inhibitor verteporfin maintained cartilage homeostasis to a significant level in a mouse OA model.81 Therefore, the specific role and functions of YAP in chondrocytes exposed to mechanical stimuli still need to be further investigated.

Other TFs function as mechanotransducers

Various TFs can function as mechanotransducers. For instance, mechanical loading upregulated Creb1 by increasing the cAMP level or activating WNT/β-catenin signaling. Mechanical stimulation of chondroprogenitor cells resulted in increased cAMP levels, significantly enhanced protein kinase A (PKA) activity and increased signal intensity of nuclear CREB, the archetypal downstream transcription factor mediating PKA signaling.37 By generating Prg4-CreERT2;Ctnnb1fl/fl and Prg4-CreERT2;Ctnnb1-ex3fl/wt mice with loss- and gain-of-function Wnt/β-catenin signaling in the cartilage superficial zone (SFZ), Xuan et al. revealed that OA development was significantly accelerated, and this increased rate was accompanied by decreased Prg4 expression and SFZ destruction in SFZ-specific β-catenin-knockout mice.82 Recently, our research team revealed that estrogen receptor α (ERα) is another TF that can function as a mechanotransducer in chondrocytes by mediating the expression of various OA-related genes.11 For example, damaging mechanical loads inhibited the expression of ERα, which in turn downregulated chondrogenic genes (COL2 and SOX9) and upregulated degenerative genes (MMP13 and ADAMTS4). When ERα was eliminated, mechanical loading continuously induced ECM turnover and led to cartilage ossification. Moreover, as described above, ERα participates in mechanotransduction in other cell types. Therefore, studies on ERα to date provide a solid starting point to further elucidate the roles played by ERα in mediating chondrocyte responses to various forces.

With the development of large-scale gene sequencing and bioinformatic analytical tools, opportunity to identify mechanosensitive TFs in cartilage is increased. By comparing the transcriptome differences in cartilage isolated from damaged distal medial condyle and intact posterior lateral condyle of femurs, Dunn et al. identified several key TFs (FOSL1, AHR, E2F1, and FOXM1) with links to a large proportion of differentially expressed genes in damaged cartilage. FOSL1 was shown to be activated in loaded mandibular cartilage, suggesting that these four TFs may also participate in mechanotransduction processes in chondrocytes.83,84 More in-depth studies are required to identify additional TFs that participate in mechanotransduction.

To date, relevant studies to fully reveal the relationship between these TFs and the mechanosensors that we have listed above are lacking. We tentatively ascribe mechanotransduction activity to these TFs to be the expectation that future research will reveal additional comprehensive molecular pathways, which may help in better understanding the process of mechanoresponsiveness in chondrocytes. Notably, the abovementioned TFs are not exclusively activated in response to mechanical loading. When studying their mechanics-related function and interpreting experimental results, it is critical to consider the influence of other factors. For example, NF-κB and YAP are also regulated by inflammation and mitogenic growth factors, respectively.

Role of ERα in regulating chondrocyte phenotype and responses to mechanical force

ERα activation

The structure of ERα is composed of five domains, which include a DNA-binding domain, a ligand-binding domain, a DNA hinge, and two domains of activation functions (AF1 and AF2); AF1 is located at the N-terminus, and AF2 is adjacent to the ligand-binding domain.85 ERα is a well characterized receptor of estrogens, such as the sexual hormone 17β‐Estradiol (E2), but it can also be activated in a steroid-independent manner, including in a certain growth factor-dependent manner and ligand-independent manner (Fig. 4). Estrogen-related drugs for OA management and steroid-dependent functions of ERα in chondrocytes have were summarized by Xiao et al.86,87,88 In the present review, we discuss the potential of direct phosphorylation of ERα in a ligand-independent mechanism, specifically a mechanism triggered by mechanical load.12

Fig. 4
figure 4

Known functions of ERα. a–d Ligand-dependent regulatory pathways. ERα binds with ligands such as E2, which can (a) directly activate MAPK signaling, b bind with DNA to function as a transcription factor (classic signaling factor), c bind with other transcription factors to influence gene expression (tethered signaling), and d modulate mRNA splicing or translational processes by binding to mRNA. e Steroid-independent regulatory pathways. ERα can be activated by mechanical load or several growth factors, E: estrogen, ERα: estrogen receptor α, MAPK: mitogen-activated protein kinase, ERK: extracellular signal-regulated kinase, P: phosphorylation, ERE: estrogen response element, TF: transcription factor, AP1: activating protein-1, Sp1: specificity protein 1, and GFR: growth factor receptor

Full size image

Expression of ERα in normal and OA cartilage

While female animals present with clearly higher levels of circulating estrogen-related ligands, recent studies have revealed that ERα is highly expressed in articular cartilage in animals of both sexes. Claassen et al. compared ERα protein levels in articular cartilage from cows, pigs, and humans after immunohistochemical staining and revealed that ERα was expressed in chondrocytes from different species.89 Oshima et al. also found that ERα was expressed in the articular cartilage and subchondral bone layers of adult rats of both sexes.90 Importantly, ERα was highly expressed in immature rats and expressed at lower levels in ovariectomized rats as well as aged rats, highlighting that ERα level changes were influenced by estrogen levels and aging.

Genetic analyses have revealed associations between alterations in ESR1, the gene encoding ERα, and osteoarthritis. For example, Ryder et al. conducted a systematic review of genetic association studies on OA. They found that ESR1 and AGC1, among other genes, showed a significant association with OA in 14 independent studies.91 Moreover, polymorphisms considered to be haplotype alleles of ESR1 [px (54%), PX (34%), and Px (12%)] were shown to be associated with an increased prevalence of radiographic knee OA, in particular in the presence of osteophytes.92,93 Another case‒control study investigated the association of ERα gene polymorphisms in patients with primary knee OA and healthy controls in the Korean population and found that the allele frequency for the exon 8G/A BtgI polymorphism (codon 594) was significantly different between patients with primary knee OA and individuals serving as controls (odds ratio = 1.38, 95% confidence interval = 1.01–1.88; P = 0.044).94 Moreover, the ESR1 gene SNP rs2234693 was significantly associated with KOA in a dominant genetic model (TT + TC vs. CC) (odds ratio (OR) = 1.30; 95% confidence interval (CI) = 1.02–1.66; P = 0.03), and the T-allele frequency was also higher than that of allele C (OR = 1.38; 95% CI = 1.06–1.80; P = 0.02) in a 1953 subjects (1033 OA cases and 920 controls) from the Chinese Han population that were analyzed.95 Hence, genetic variations in the ERα gene have been associated with OA, especially in the presence of osteophytes.

Studies on the association between ERα levels and OA severity remain inconclusive. Ushiyama et al. extracted normal chondrocytes from the hip joint of femoral neck fracture patients and OA chondrocytes from the knee joint of TKA patients and compared the gene expression level of ESR1 between these two groups via qRT‒PCR. No significant difference in ESR1 gene expression was found between the cells from these joints.96 Notably, findings from this study may be of limited generalizability, as normal cartilage was harvested from a hip joint but not a knee joint, and the size of the healthy normal cartilage sample set was small. Several studies have reported RNA sequencing results used to compare the transcriptome differences in chondrocytes from normal cartilage (isolated from the knee joint of cadavers), mildly damaged cartilage (isolated from non-load-bearing cartilage areas in OA knees), and severely damaged cartilage (isolated from load-bearing cartilage areas in OA knees). ESR1 showed the highest expression level in the normal cartilage and the lowest expression in OA cartilage from the load-bearing area (i.e., damaged cartilage).84,97,98 In particular, Wang et al. tested the protein expression level of ERα in preserved and damaged cartilage from human knee joints, as well as that from OA mouse models. The results showed that ERα was significantly decreased in damaged human cartilage and OA mouse knee joints, suggesting that ESR1 expression may be inhibited by mechanical overload and may be associated with OA pathology.11

The function of ERα in chondrocytes has also been studied through loss-of-function analysis. Sniekers et al. revealed that ERα−/−β−/− mice presented with increased osteophytosis compared with that of their wild-type littermates, and deletion of either ERα or ERβ failed to induce cartilage damage or obvious osteophytosis.99 Wang et al. found loss of proteoglycan staining in ERα−/− mice; however, these mice showed a degree of cartilage integrity maintenance that exceeded that in the controls.11 After applying cyclic mechanical loading to female osteoblast-specific-ERα-knockout (pOC-ERαKO) mice, Ziemian et al. found that the pOC-ERαKO mice developed more severe cartilage damage, larger osteophytes, and greater joint capsule fibrosis than the littermate controls, suggesting that loss of ERα in subchondral bone may be a potential mechanism of OA pathology.100 Engdahl et al. investigated the role of estrogen and ERα in a rheumatoid arthritis animal model established with cartilage-specific knockout (Col2a1-ERαKO) mice. They used antigen-induced arthritis to establish the model and analyzed knee joint synovitis and cartilage destruction. The results showed that ERα expression in cartilage was not required for estrogen-induced amelioration of joint destruction, but ERα expression in cartilage conferred protective effects against synovitis. These results suggest that ERα may play a role in inflammatory joint disease.101

To date, the mechanisms that result in the reduction of ERα levels in OA have not been fully determined. Our recent study demonstrated that mechanical overload caused decreases in ERα levels,11 implying that ERα may be highly responsive to stress. In fact, Guo et al. showed that IL-1β treatment induced a reduction in ERα levels, which was accompanied by elevated levels of miR-203 (ref. 102). Moreover, inhibiting miR-203 has been shown to restore ERα levels. Tian et al. performed bilateral ovariectomy and intra-articular monosodium iodoacetate injections (MIA) to establish a postmenopausal OA model in rats and found that the level of miR-203 was higher when the level of ERα was lower in OA knee joint cartilage. Similar to the in vitro study mentioned above, intra-articular injection of a miR-203 inhibitor attenuated cartilage degradation by elevating ERα levels in vivo.103 Overall, these results suggest that different OA-inducing stressors may reduce ERα levels in chondrocytes, and this effect is partially mediated by upregulated miR-203 activity.

The ligand-independent mechanotransduction function of ERα

Our recent study revealed a ligand-independent mechanotransduction function for ERα in chondrocytes,11 and similar observations with other cell types has been previously reported. Rooney et al. reviewed mouse models used to study the role of ERα in bone and found that after ERα was depleted in differentiated osteoblasts, osteocytes, and osteoclasts in female mice, cancellous bone mass diminished.104 Zhang et al. found that mechanical loading stimulated cell proliferation and enhanced alkaline phosphatase activity in mesenchymal stem cells, noting that the effect was enhanced or inhibited by overexpressed and blocked ERα expression, respectively.105 Swift et al. revealed that mechanical loading significantly enhanced ERα expression in osteoblasts.106 However, according to the results from a study performed by Ehrlich et al., high mechanical strain decreased ERα expression.107 Overall, physiological loading appears to maintain a stable level of ERα, while damaging loads inhibit ERα expression.

Multiple independent studies have shown that the mechanotransduction role of ERα is at least partially estrogen independent. By stimulating the lower limbs of ovariectomized rats by mechanical overloading, the expression of ERα was significantly inhibited in bone, an effect that was independent of the estrogen concentration.108 Windahl et al. developed ERα AF1- or AF2-specific knockout mice and inserted a luciferase reporter into the estrogen response element (ERE) to determine whether the mechanotransduction of ERα in bone relied on estrogen.12 They found that the depletion of ERα AF1 blocked the effects of mechanical load on bone, but ERα AF2 knockout did not. Importantly, the luciferase reporter showed that ERα did not combine with the ERE to mediate the response to mechanical signals. In another study, Aguirre et al. first depleted ERα in osteocytes and osteoblasts, which impaired the cell response to stretching, and then transfected cells with an ERα mutant that could not bind estrogen. The results indicated that mutant ERα was sufficient to restore mechanoresponsiveness in these cells.109 In the same study, the researchers also demonstrated that the ligand-independent function necessarily involved ERα targeted to the plasma membrane, not to the nucleus. Zaman et al. also revealed specific strain-induced membrane localization of ERα in bone cells.110 Notably, the mechanoresponsiveness and ligand-independent function of ERα involved translocation from the cytoplasm to the nucleus, similar to the translocation of E2. Taken together, these data suggest that ERα exhibits a ligand-independent mechanotransduction function that requires its localization to the intact cell plasma membrane or cytoplasm, not the nucleus.

Recent findings have redefined the traditional concept of sex hormones as the main regulators of skeletal sexual dimorphism. For example, after aromatization-base transformation into estrogens, androgens activate not only the androgen receptor (AR) but also the ER. Sex hormone receptors exert an impact on the mechanical sensitivity of the growing skeleton.111 Therefore, even though sufficient evidence has shown that the mechanics-associated functions of ERα do not require classic ligands, very few studies have examined the effect when ligands are present. This effect is particularly important to women undergoing perimenopause. Although a simple experiment seems to be indicated, dissecting the functional changes of ERα based on the influence of decreased ERα levels after binding to estrogen is a challenge. Another important area for study is identification of the difference in DNA-binding sites in ERα in the presence and absence of estrogen. Potential ERα competition for classic estrogen responsive elements and other unknown responsive elements may ultimately affect the cell response to mechanical force.

Downstream targets that are regulated by ERα in mechanotransduction pathways

To date, the targets of ERα activity contributing to mechanotransduction in chondrocytes have not been explored. Manolagas et al. reviewed the role played estrogen and androgen receptors in bone, suggesting that the most relevant downstream pathway of ERα involves WNT signaling, as ERα can sense mechanical loading and activate the WNT pathway to promote osteogenesis.112 Armstrong et al. also found that a single short period of dynamic mechanical strain increased the nuclear accumulation of activated beta-catenin. This effect was absent in primary osteoblast cultures from mice lacking ERα, suggesting that Wnt/beta-catenin signaling contributes to early responses to mechanical strain in cells, at least partially, through functional ERα.113 Under the stimulation of estrogen or loads, ERK-1, which belongs to the MAPK family, was activated after ERα was phosphorylated, demonstrating that ERα activates the MAPK/ERK-1 signaling pathway in response to mechanical loading.114 Another MAPK family member, JNK-1/2, is also involved in ERα signaling. Zhao et al. found that hydrostatic pressure was mediated through ERα to activate JNK-1/2 and in turn promoted the chondrogenic differentiation of MSCs.115 Last, Sunters et al. revealed that ERα facilitated the mechanical loading-activated signaling cascade in bone, which included insulin-like growth factor I receptor (IGF-IR), phosphatidylinositol 3-kinase-mediated phosphorylation of AKT, and then inhibited of GSK-3beta activity.116

Considering these findings in chondrocytes and other cell types, we propose a network that is regulated by ERα (Fig. 5), and this proposed pathway needs to be validated in future studies.

Fig. 5
figure 5

Potential targets that are regulated by ERα mediating chondrocyte responses to mechanical stimulation. Pointed arrows indicate stimulatory effects, and blocked lines indicate inhibitory effects. Some proposed pathways and interactions in chondrocytes remain to be determined. ERα: estrogen receptor α, MAPK: mitogen-activated protein kinase, ERK: extracellular signal-regulated kinase, JNK: c-Jun N-terminal kinase, GFR: growth factor receptor, PI3K: phosphatidylinositol-3-kinase, AKT: protein kinase B, and GSK-3β: glycogen synthase kinase 3 beta

Full size image

Conclusion

Mechanical loading plays a fundamental role in the health and disease of articular cartilage. Physiological loading maintains the integrity of cartilage, while insufficient or excessive physical force can lead to its degradation. At present, a full molecular map by which physical forces are translated into biochemical responses remains to be established. Here, we summarize the current knowledge on the mechanotransduction pathways in articular chondrocytes. Inspired by previous studies, we propose a three-tiered cascade of mechanotransduction pathways to better characterize the role of these mechanoresponsive molecules and structures in chondrocytes. We believe that targeting mechanics-associated factor(s) is an important direction that will lead OA treatment because its converts the damaging effects of physical force and to be beneficial effects. Moreover, modulating these specific molecules to simulate mechanical signaling may not only allow physically disabled individuals to maintain healthy cartilage but also potentially regenerate/preserve cartilage in patients who have joint pain or restricted mobility; this possibility is particularly promising in the context of emerging concomitant cell therapies. In addition, by temporarily suppressing or activating key factors in the mechanotransduction pathway, it may also be possible to protect cartilage from damage due to short-term but high-intensity loading. For example, recent studies on TRPV4 and Piezo1/2 have indicated these possible outcomes, but the identification of the most efficacious targets is needed, and a greater number of more specific and efficacious small-molecule modulators are needed.

Notably, mechanical stimulation of articular cartilage induced by the movement of a joint involves very complex processes, including compression, shear stress, tension, hydrostatic pressure, and fluid movement.117 Moreover, the distribution and organization of cartilage components, including chondrocytes, various types of collagens, proteoglycans, noncollagenous proteins, and water, varies among the four distinct cartilage layers: The superficial, middle, deep, and calcified zones. All these objective reasons indicate that different mechanical effects on chondrocytes are mediated differentially by different layers.51 Due to the limitation in simulating the complexity of the native environment, most research has explored one type of mechanical stimulus without including the variation inherent in different cartilage zones. Although the simplified conditions have helped us understand mechanotransduction in chondrocytes, more physiologically relevant conditions are needed to investigate how chondrocytes sense different types of mechanical stimulus signals and convert them into biochemical changes. Moreover, as described above, most currently identified mechanically responsive transcription factors also participate in other cellular activities. Therefore, the degree of mechanotransduction pathway uniqueness in chondrocytes needs to be further investigated, which may include identifying unique transcription factors and kinases, discovering special molecular interactions that are observed only when a mechanical signal is applied, and identifying factors that can exclusively lead to an outcome on the basis of mechanical stimulation. This information will be critical to developing mechanics-regulated biomaterials and drugs in the future.10

An important topic that we need to specifically address in the future is sex-dependent differences in mechanotransduction pathways, which have often been overlooked. Using bovine explants, Hernandez showed that male and female cartilage tissues exhibit different mechanical properties, which are associated with different pericellular matrix compositions.118 Yu et al. compared the ERα level in articular chondrocytes of male and female rats at different ages. The results revealed that the ERα expression level and the number of ERα-positive chondrocytes remained stable in 4-weeks-old male rats until they were 12 months old, yet the aforementioned index significantly decreased with age in female rats.119 This study proved that the decrease in the level of ERα under the influence of aging was more pronounced in females. Whether a sex-dependent difference with the decrease ERα level results in a different responses to mechanical force between old male and female chondrocytes, however, remains to be reported. Melville et al. previously discovered that ERα in mature osteoblasts differentially regulated bone mass in males and females.120 It is therefore suggested that chondrocytes and animals from both sexes be included in future studies to determine the potential sex-dependent difference in mechanotransduction pathways, which will be critical to developing disease-modifying drugs. Overall, current evidence suggests that sexual dimorphism is not just a result of the discrepancy in sex steroid secretions between the sexes; in contrast, it depends on sex-dependent differences in the cellular response to mechanical forces.

In our previous study, we discovered that ERα is a negative regulator of p16INK4a, a marker of senescence, and that restoring the ERα level directly reduced the p16INK4a level.11 Given that different stressors, such as mechanical overload, DNA damage, and inflammation, cause senescence, we speculate that ERα levels are influenced by these stressors and mediate the development of senescence in chondrocytes. In fact, our previous data showed that mechanical overload, one type of stressor that damages cartilage, led to decreased ERα levels.11 Therefore, ERα is thought to be a hub connecting senescence and mechanotransduction pathways. Future studies are needed to position ERα and its interactions in the context of other factors in these pathways.

Last, our current studies suggest that the role of ERα in influencing mechanotransduction in chondrocytes does not require the presence of estrogen. However, it is not clear whether the presence of estrogen affects the ligand-independent functions of ERα. If it does, how do chondrocytes interpret and integrate these two activation mechanisms? Estrogen therapy has been evaluated in human clinical trials to treat osteoarthritis. As discussed above, the binding of estrogen or selective estrogen receptor modulators resulted in the rapid degradation of ERα. If OA chondrocytes lose the capacity to sufficiently produce new ERα, they may undergo senescence conversion due to ligand-induced decreases in ERα levels. Therefore, it is important to delineate the ligand-dependent and ligand-independent functions of ERα in chondrocytes.