Enzymatic Machinery of Ubiquitin and Ubiquitin-Like Modification Systems in Chondrocyte Homeostasis and Osteoarthritis

Abstract

Purpose of Review

To date, a vast amount of information regarding ubiquitination (Ub) and ubiquitylation-like (Ubl) modification–related mechanisms has been reported in the context of skeletal cell homeostasis and diseases. In this review, we mainly focus on recent findings regarding the contribution of enzymatic machinery that directly adds or removes Ub and Ubl modifications from protein targets in chondrocyte homeostasis and osteoarthritis (OA) development.

Recent Findings

Mechanisms that promote homeostasis of articular chondrocytes are crucial for maintaining the integrity of articular joints to prevent osteoarthritis development. Articular chondrocytes are postmitotic cells that continuously produce and remodel cartilage matrix. In addition, the long lifespan of chondrocytes makes them susceptible to accumulating cellular damage. Ub and the evolutionarily conserved Ubl modifications, such as SUMOylation, ATGylation, and UFMylation, play important roles in promoting chondrocyte homeostasis, including regulating cell signaling and protein stability, resolving cellular stresses and inflammation, and maintaining differentiation and survival of chondrocytes.

Summary

Uncovering new components/functions of Ub/Ubl modification machinery may provide novel drug targets to treat OA.

Introduction

Osteoarthritis (OA) is one of the most prevalent diseases in the musculoskeletal system, and so far, it is incurable. Its pathology includes numerous structural changes in joint tissues, including articular cartilage degeneration and loss, synovial inflammation, thickening of subchondral bone, and ligament degeneration. Thus, OA is considered a disease of the joint as an organ system resulting in joint failure [1]. The hallmark feature of OA is the damage of articular cartilage, within which articular chondrocytes comprise the only cell type. Articular chondrocytes, albeit only take ~2% of total articular cartilage volume, are “professional” producers and remodelers of the cartilage extracellular matrix (ECM) [2]. In addition, being postmitotic cells, articular chondrocytes rarely replenish themselves, so their survival and ability to maintain the ECM are vitally important. The long lifespan of articular chondrocytes also means that they must be capable to effectively handle altered signaling from the microenvironment, including stimuli from apoptosis- or senescence-inducing stressors such as mechanical stress, reactive oxygen species (ROS), DNA damage, proteostasis stresses, and inflammation [3–7]. Ideally, alleviating the insults from the microenvironment or strengthening the stress-handling capacity of chondrocytes can be promising avenues for designing new OA treatments. Post-translational modifications (PTMs) are among the most crucial mechanisms for regulating cellular signaling and stress responses. Therefore, targeting PTMs may provide a new approach to prevent or intervene OA progression.

The human genome has about 30,000 protein-coding gene loci. PTMs can vastly expand the proteome size because they do not require de novo protein synthesis. Thus, they allow cells and organisms to regulate complex cellular processes and to respond to constant changes of the microenvironment in a dynamic and cost-effective fashion. PTMs participate in almost all aspects of normal cell physiology, and their dysregulations often contribute to disease development. Also, because of their reversible nature and dependence on enzymatic activity, PTM pathways are considered ideal drug targets for disease treatments. Ubiquitination (Ub) and ubiquitination-like (Ubl) PTMs, including SUMOylation, ATGylation (for autophagy), UFMylation, and Neddylation, represent a unique and evolutionarily conserved category of PTMs. Unlike the PTMs involving small chemical groups, such as phosphorylation, acetylation, and methylation, Ub or Ubl PTMs attach protein tags to their targets by several consecutive enzymatic steps and can occur as monomeric or polymeric chain modifications on the same lysine of the target proteins.

Our knowledge of Ub and Ubl PTMs has quickly expanded in various disease contexts. Here, we mainly focus on updating the recent findings regarding the contribution of the enzymatic machinery of Ub and Ubl modifications to chondrocyte homeostasis and OA development.

Ubiquitination

Ubiquitination is the most crucial PTM for precise and swift protein degradation relying on the sequential actions of the E1-E2-E3 enzyme cascade [8]. Ubiquitin (Ub) first undergoes a maturation process to expose its C-terminal di-glycine motif (GG), the latter glycine of which forms a covalent bond with the E1enzyme in an ATP-dependent manner. Next, this activated ubiquitin is transferred to E2’s catalytic cysteine via a thiol bond to form E2-Ub. Finally, the E3 enzyme recruits both E2-Ub and the target protein and facilitates the transfer of Ub onto a specific lysine of that protein (Fig. 1) [9]. Ubiquitin itself also contains lysins allowing for further ubiquitination, namely, polyubiquitination. Lys48 polyubiquitination is the most common type, which targets protein substrates to the proteosome-dependent degradation, while lys63 polyubiquitination regulates protein endocytosis or alters their biological functions. The ubiquitin modifications on protein targets can be reversed by specific deubiquitinases (DUBLs). In humans, there are two E1, roughly 40 E2, about 600 E3 enzymes, and even more DUBLs [10, 11]. This makes sense because E1 and E2 are common to all ubiquitination, whereas E3 and DUBLs confer substrate specificity [11]. Because of its high efficiency and specificity in protein degradation, ubiquitination is found to have a crucial role in modulating the majority of biological events and signaling pathways. For example, ubiquitination of cyclins, Rb, CDK inhibitors is the central mechanism of cell cycle progression. Ubiquitination of PCNA and P53 regulates sensing, repair, and/or tolerance of DNA damage [12]. Ubiquitination-dependent degradation of TGFβRs and SMADs, β-catenin, and IKKγ (NEMO) is an essential mechanism for the regulation of TGFβ/BMP signaling, Wnt signaling, and NF-κB pathway, respectively [13–17].

Fig. 1.

Ubiquitin and ubiquitin-like modifications in chondrocyte homeostatic defects. Various disruptions of Ub/Ubl modification systems contribute to the pathology of chondrocytes. Based on the published work included in this review, this scheme summarizes these disruptions and classifies them by stage (vertically) and type of modification (horizontally). The upper panel is a generalized schematic of Ub/Ubl modification and the lower panel shows the specific mechanism of these Ub/Ubl modification enzymatic components in regulating chondrocyte homeostasis and pathogenesis. Solid lines with arrows and blunt ends indicate positive and negative regulations, respectively; dotted lines represent tentative mechanisms

Studies have shown that the enzymatic machinery of the ubiquitination pathway plays a pivotal role in the regulation of cell proliferation, survival, and differentiation, and its disruption causes many human diseases. Human knee OA cartilage exhibits accumulation of proteins with K48-linked polyubiquitination in situ, suggesting proteasomal impairment [18]. In addition, ubiquitin-conjugating enzyme E2 M (UBE2M) is highly expressed in human OA cartilage tissue and positively correlates with chondrocyte apoptosis. Ubiquitination and destabilization of AXIN, a target of UBE2M, is enhanced in OA chondrocytes, suggesting that UBE2M promotes OA chondrocyte apoptosis and activates the Wnt/β-catenin pathway via downregulation of AXIN (Fig. 1) [16].

Ubiquitin E3 ligases have been widely studied and demonstrated to have an important role in skeletal homeostasis by regulating the stability and function of various signaling factors. For example, recent studies show that the NEDD4-like E3 ubiquitin protein ligase (NEDD4L) promotes skeletal stem cell (SSC) osteogenic differentiation by inducing K63-linked polyubiquitination and activating the AKT pathway [19]. SMAD-specific E3 ubiquitin protein ligase 1 (SMURF1) augments BMP signaling and osteogenic differentiation of skeletal stem cells [20]. PARKIN is an E3 ubiquitin ligase associated with Parkinson’s disease and aging. A recent study found that PARKIN is upregulated during osteoblast differentiation and enhances the expression of osteo-specific markers. In addition, upregulation of PARKIN increases β-catenin levels and promotes autophagy. Moreover, in a tibial fracture model, PARKIN overexpression can accelerate bone healing process [21]. Another study reported that PARKIN can degrade Mitofusin 2 (MFN2) through ubiquitination, thus inhibiting inflammation and aging-related metabolic change of articular chondrocytes and attenuating OA progress [22]. In addition, the declined PARKIN expression with age or OA may contribute to the accumulation of MFN2 in chondrocytes [22].

Ubiquitin E3 ligases play important roles in OA development. Expression of FBXO6, a component of a ubiquitin E3 ligase, is decreased in human OA cartilage and the mouse OA cartilage derived from the ACLT (anterior cruciate ligament transaction)-induced OA model or spontaneous OA model (STR/ort strain) [23]. Mechanistically, FBXO6 decreases MMP14 levels by promoting its ubiquitination, further attenuating the MMP14-dependent proteolytic activation of MMP13, which is a major cartilage matrix–degrading protease that is frequently overactivated in OA foci (Fig. 1) [23]. In addition, WW domain-containing protein 2 (WWP2) is a HECT-type E3 ubiquitin ligase abundantly expressed in articular cartilage. A study showed that mice lacking WWP2 or mice in which the WWP2 E3 enzymatic activity is interrupted (WWP2-C838A) exhibit accelerated progression of spontaneous or surgery-induced OA [24]. WWP2 maintains cartilage homeostasis by regulating the stability of ADAMTS5, which is a major aggrecanase in articular cartilage [24]. Another study reported that loss of DEPTOR (DEP domain-containing mTOR-interacting protein), known as an mTOR signaling inhibitor, promotes OA progression in a surgery-induced OA mouse model. DEPTOR interacts with and promotes the autoubiquitination and degradation of TRC8, a ubiquitin E3 ligase. When DEPTOR is decreased, TRC8 is stabilized thus promoting ER stress and OA progression in a manner independent of mTOR signaling [25].

The reversibility of ubiquitination is warranted by a group of specific deubiquitinases. Ubiquitin-specific peptidases (USPs) are the main members of the deubiquitinase family. USP49 attenuates OA progression by inhibiting the Wnt/β-catenin pathway. USP49 is lowly expressed in human OA chondrocytes and overexpression of USP49 in primary rat chondrocytes inhibits apoptosis by promoting AXIN deubiquitination and reducing the level of IL-1β, a marker for OA progression (Fig. 1) [26]. Moreover, USP3 can reverse the ubiquitination of TRAF6 (tumor necrosis factor-receptor-associated factor 6), which intermediates the signals from inflammatory cytokines to NF-κB activation, thus blocking IL-1β-induced chondrocyte apoptosis [27].

SUMOylation

SUMOylation is highly similar to ubiquitylation in terms of their enzymatic conjugation and deconjugation systems, but it uses small ubiquitin-like modifiers (SUMOs) as modification tags. SUMO1–3 are present in all cell types [28, 29]. SUMO2 and SUMO3 are about 95% identical, and SUMO1 shares ~50% homology with SUMO2 and 3. In general, SUMO2/3 modification is more dynamic and transient than SUMO1 and capable to form a polySUMO chain. Similar to ubiquitination, a SUMO first undergoes proteolytic maturation by a few SUMO-specific proteases (SENPs) to expose its C-terminal di-glycine motif. Next, it covalently bonds to SUMO-activating enzyme E1 (SAE1/UBA2) [30], and is then transferred to UBC9 (SUMO ubiquitin-conjugating enzyme 9), which is the only SUMO E2 enzyme. Finally, UBC9 confers this SUMO tag to a specific lysine of the target protein with the facilitation of SUMO E3 ligases, but in some cases, E3 ligases are dispensable. Reversibly, SUMO tags can be removed from target proteins by SENP family members (deSUMOylation) [31]. In humans, there are seven SENP proteins with distinct subcellular localization and specificity to SUMO1 and SUMO2/3. The effects of SUMO modifications on their substrate proteins are diverse, but generally, they mask or create new site(s) for protein-protein interaction(s). A large set of proteins contain the SUMOylation interactive motif (SIM), which specifically recognizes SUMOs. Through this, SUMO-SIM interaction facilitates a quick assemble of large protein complexes, which is crucial for cells to handle urgent cellular stresses or signaling changes. In addition, polySUMOylation also promotes protein degradation by recruiting Ub E3 ligase [32].

SUMOylation pathways have been shown to dynamically regulate numerous targets of chondrocyte homeostasis, including Sox9 [33–35]. It was reported that SOX9, a master regulator of chondrogenesis, interacts with PIAS SUMO E3 ligases (PIAS1, PIAS3, PIAXα, and PIAXβ) and can be modified by SUMO1 [36]. SUMOylation also increases SOX9 protein level and its transcription activity [36]. Furthermore, SOX9 is associated with SENP2 in the U2OS osteosarcoma cell line. SENP2 deSUMOylates SOX9 and promotes its ubiquitination and degradation [37] (Fig. 1). Another recent work reported that limb osteochondroprogenitor-specific loss of SHP2, a protein-tyrosine phosphatase that intermediates most receptor protein-tyrosine kinase signals to activate the RAS/ERK pathway [38, 39], can increase SOX9 protein in the perichondrium and chondrocytes. SHP2 regulates SOX9 abundance through increasing SOX9 phosphorylation and SUMOylation, partially in a PKA signaling–dependent manner [40]. In contrast, it was also reported that PIAS1 enhances SOX9 SUMOylation at K396; loss of this SUMOylation site alters the localization of SOX9 in the nucleus and increases its transcription activity (Fig. 1) [41]. SOX6 is a transcription factor in chondrogenesis acting downstream of SOX9 [34, 35]. SUMOylation leads to SOX6 sequestration to the PML nuclear bodies and the inability to exert its transcription activity. Mutations of two SUMOylation sites of SOX6, UBC9 deficiency, or SENP2 overexpression can promote SOX6 transcription activity [42].

In human primary articular chondrocytes, the SUMO E2 enzyme, UBC9, SUMOylates ETS-like-1 protein (ELK-1, a transcription factor). This abolishes the activity of ELK1 in regulating MMP13 gene transcription [43] (Fig. 1). Furthermore, in the ATDC5 chondrocyte cell line, HDAC9-dependent deacetylation of Nkx3.2, a transcription factor regulating chondrocyte differentiation and viability, can further trigger Nkx3.2 SUMOylation mediated by PIASy, a SUMOylation E3 ligase. Subsequently, SUMOylated Nkx3.2 can be degraded through the RNF4-dependent, SUMO-targeted ubiquitination pathway. This process promotes ATDC5 cell hypertrophy and apoptosis [44].

A large genome-wide association study found that a SNP (rs9350591 C/T) located at the upstream sequence of the SENP6 locus [45] is among the most significant SNPs associated with severe OA. A follow-up study reported that regardless of the presence of rs9350591 SNP, SENP6 expression was significantly decreased in osteoarthritic cartilage, suggesting that decreased SENP6 activity is a widespread phenomenon in OA [46]. Consistent with this, we recently reported that the loss of SENP6 postnatally in the whole body leads to premature skeletal aging in mice [47], and that osteochondroprogenitor (OCP)-specific Senp6 knockout mice show elevated apoptosis and cell senescence in OCPs and chondrocytes. Senp6 loss results in over-SUMOylation and destabilization of TRIM28 (tripartite motif-containing 28), which is a multifunctional protein regulating chromatin silencing and p53 inhibition. Thus, the loss of TRIM28 function leads to p53 hyperactivation that further causes OCP and chondrocyte death and senescence (Fig. 1). To what extent this mechanism contributes to SENP6 loss–induced OA is to be determined. Moreover, SENP2 expression was found to correlate with OA progression, and during inflammation-induced OA, SENP2 deSUMOylates aggrecan and type II collagen proteins. SENP2 overexpression or SUMO2/3 knockdown significantly enhances the rate of degradation of aggrecan and collagen type II in chondrocytes treated with IL-1β [48].

ATGylation

Autophagy is a protein degradation program induced by various cellular stresses including nutrient/energy starvation, hypoxia, ER stress, and organelle damage. It is crucial for maintaining cell homeostasis and preventing aging-related changes [49]. During the autophagic process, a double-membrane vesicle trafficking system (autophagosome) is formed in the cytoplasm to sequester unfolded/misfolded proteins, cellular organelles, and long-lived proteins to target them to lysosomes for degradation. Autophagy utilizes two consecutive Ubl modifications—ATG12 and ATG8 modification (ATGylation)—to facilitate autophagosome formation. ATG12 and ATG8 (LC3 in mammals) are Ubls, which have little sequence homology with ubiquitin, but share structural similarity [50]. First, ATG12 is covalently attached to ATG5, and they further complex with ATG16 to form an E3 ligase for the second ATGylation system, by which ATG8 is conjugated to phosphatidylethanolamine (PE, a type of lipid) on the surface of autophagosomes. The ATG8-modified PE provides a docking site for autophagy-related cargos or adaptor proteins and facilitating autophagosome fusion with the lysosome. ATG8 and ATG12 share the same E1 (ATG7) but use different E2s: ATG10, ATG3, and ATG10, respectively. After the fusion of the autophagosome with the lysosome, ATG8 can be either degraded with cargos or cleaved off by the peptidase ATG4 and recycled. Many studies have established that autophagy supports the survival of chondrocytes by relieving cellular damage caused by stress stimuli [51]. For example, cartilage-specific loss of mTOR, a key autophagy repressor, enhances autophagy and slows aging-induced OA [52], while Rapamycin, an mTOR inhibitor, attenuates glucocorticoid-induced chondrocyte apoptosis in OA cartilage by activating autophagy [53].

Cartilage-specific deletion of ATG5 (Col2-Cre; Atg5 ^fl/fl mice), a component of the E3 enzyme for ATG8 modification of PE, leads to an increase in the severity of age-related OA in male mice. However, the progression of post-traumatic OA is unaltered in these ATG5 conditional KO mice. Also, loss of ATG5 in cartilage impairs proliferation and survival of growth plate chondrocytes, and ultimately causes growth retardation. Likewise, mice with cartilage-specific deletion of ATG7, the E1 enzyme for either direct conjugation of ATG8 to PE or ATG12 modification of ATG5, have growth retardation with reduced proliferation and augmented apoptosis in growth plate chondrocytes (Fig. 1) [54, 55].

Activation of SIRT1, a class III histone deacetylase, promotes autophagy in chondrocytes of hip cartilage in mice. Also, SIRT1 can be co-immunoprecipitated with key autophagy-related proteins, including ATG5, ATG7, Beclin1, and LC3 (ATG8). Recent findings showed that SIRT1 is decreased with aging and in human OA cartilage. Activated SIRT1 can deacetylate autophagy-related proteins, including ATG5, ATG7, and LC3 in a manner independent of mTOR signaling [56, 57]. These studies suggest that SIRT1, a crucial anti-aging factor, utilizes an autophagy mechanism to attenuate OA development.

Kashin-Beck disease (KBD) is an endemic type of OA characterized by cartilage matrix degradation and chondrocyte death in growth plate and articular cartilage [58]. KBD chondrocytes exhibit defective autophagy [59]. A genome-wide association study showed that the ATG4C gene, which encodes a peptidase responsible for the maturation of ATG8, was strongly associated with KBD [60]. The protein and mRNA level of ATG4C was significantly decreased in KBD articular chondrocytes [60].

MicroRNAs play an important role in regulating OA progression via modulating autophagy. Some of them target autophagy via upstream pathways, such as mTOR signaling (mir155, mir20a) [61, 62], while others directly control the expression of ATGylation machinery. An association of miR128a expression level with poor autophagy was found in the articular chondrocytes from human patients with end-stage knee OA or from rats with anterior cruciate ligament transection (ACLT)–induced OA [63, 64]. miR128 targets the 3' untranslated region of Atg12. In vitro experiments showed that miR128 blocks ATG12 expression and ATG8-PE conjugation, and injection of an antisense oligonucleotide for miR128a maintains chondrocyte autophagy and alleviates ACLT-induced OA symptom. miR-20a, which was found to target ATG7 in macrophages, can block chondrogenesis in ADTC cells and has a negative correlation to autophagy during chondrogenesis. Inhibiting miR-20a leads to a significant elevation of ATG7 mRNA and protein levels and promotes chondrocyte differentiation [65].

UFMylation

UFM1 (ubiquitin-fold modifier 1) is a Ubl with less known functions. After the maturation process by cysteine proteases, UFSP1/2, UFM1 exposes its C-terminal glycine and then can be conjugated to target proteins facilitated by its E1 (UBA5), E2 (UFC1), and E3 (UFLs). It was reported that the autosomal dominant mutations (Y290H) of the UFSP2 gene lead to a Beukes hip dysplasia, which is an inherited skeletal condition resulting in dwarfism, premature hip degeneration, and OA. Two other groups reported the D426A and H428R mutations of the USFP2 gene are responsible for spondyloepimetaphyseal dysplasia (SEMD), which has similar symptoms to Beukes hip dysplasia [66–68]. Interestingly, a homozygous mutation of the DDRGK1 gene, also called UFBP1 (UFM1-binding protein 1 containing a PCI domain), was found to mediate the etiology of Shohat type SEMD. DDRGK1 forms a complex with UFM1 substrates and UFL1 (a UFM1 E3 ligase) and is required for UFMylation. DDRGK1 loss leads to SOX9 ubiquitination and degradation, thus reducing Col2a1 expression. One of the possibilities is that DDRGK1 facilitates SOX UFMylation, which may prevent SOX ubiquitination [69]. Another recent work reported that expression of UFL1 is decreased in human OA articular cartilage and in chondrocytes treated with IL-1β. Ectopic expression of UFL1 in chondrocytes inhibits IL-1β-induced NF-κB signaling and the production of cartilage matrix–degrading proteases, such as MMP3, MMP13, ADAMTS-4, and ADAMTS-5 (Fig. 1) [70]. These findings suggest that the UFM1 modification system may be targeted for OA therapy.

Other Ubl Modifications

To date, studies regarding the role of other Ubl PTMs have not yet been reported in the context of OA, but some recent work discovered their involvement in other inflammatory skeletal diseases. Interferon-stimulated gene 15 (Isg15) encodes a ubiquitin-like protein. Its expression is induced by interferon (IFN)-α and IFN-β, infection, DNA damage, and aging. ISG15 modification (ISGylation) has an important role in regulating processes such as DNA repair, protein translation, autophagy, and exosome secretion. [71]. Recent studies showed that ISGylation of NEMO (IKKγ), an inhibitor of NF-κB signaling, targets NEMO to autophagy-mediated lysosomal degradation, thus restraining osteoclastogenesis and inflammatory osteolysis [72]. Aberrant NEDD8 modification (Neddylation) is associated with neurodegenerative diseases, cancers, and inflammatory diseases. Recently, a study showed that Nedd8 is significantly upregulated in inflamed arthritic synovia. In fibroblast-like synovial cells, Neddylation of TRAF6 facilitates IL-17A-induced NF-κB signaling, which promotes the development of inflammatory arthritis [73].

Conclusions

Evolutionarily conserved Ub and Ubl modifications have many unique characteristics, commonalities, and functional connections. In this review, we summarized some of the most recent studies regarding this type of modification system in the context of chondrocyte homeostasis and OA, with the focus on enzymatic machinery that directly adds or removes Ub and Ubl modifications from protein targets. Some of these investigations provided interesting insights for treating OA by targeting components of these PTM systems. Compared to E1 and E2 enzymes, the E3 ligases (i.e., PARKIN, FBXO6, PIAS, ATG5, ATG16, and UFL1) and deconjugases (i.e., USP49, SENP2, SENP6, ATG4) are much more diverse and target-specific, thus conceivably more suitable for drug targets and further pre-clinical tests. Such ideas have been explored in other disease contexts. For example, therapeutic applications targeting protein-protein interactions have been used to design inhibitors of SKP2/SKP1 Ub E3 ligase complex formation for cancer treatments [74]. Moreover, OA is a degenerative disease of the whole joint as an organ system; other joint cells including synoviocytes, tenocytes, meniscus chondrocytes, and osteoblasts from subchondral bone also contribute to and/or are deteriorated during OA development. Hence, uncovering new Ub/Ubl E3 ligases and deconjugases or their novel functions in joint tissue homeostasis and OA development would be a promising direction for identifying suitable drug targets to treat OA, which is still an incurable disease.