New developments in chondrocyte ER stress and related diseases

Endoplasmic reticulum stress in rare genetic skeletal diseases due to misfolding of extracellular matrix proteins: recent advances in potential therapies

An extremely complex and diverse group of diseases, GSDs affect primarily the homeostasis and development of the skeleton. The severity of the more than 450 unique and well-characterised phenotypes range from relatively mild to severe and lethal forms. Individually, GSDs are rare but as a group of related orphan diseases they have an overall prevalence of at least 1 per 4000 children, representing a considerable unmet medical need⁹.

Over the last decade, ER stress has been identified as a core disease mechanism in a group of phenotypically unrelated GSDs, all of which result from dominant-negative mutations in a range of cartilage ECM structural proteins, particularly those GSDs resulting from mutations in the genes encoding cartilage oligomeric matrix protein (COMP), matrilin-3, and collagen types II and X¹⁰.

Archetypical examples of GSDs resulting from the accumulation of mutant cartilage proteins in the ER were first described several decades ago and include pseudoachondroplasia (PSACH), multiple epiphyseal dysplasia (MED) and metaphyseal chondrodysplasia type Schmid (MCDS)¹¹. More recently, our understanding of genotype-specific differences in ER stress pathways and the activation of different branches of the UPR has provided potential avenues for novel therapies. This is best exemplified by MCDS that is caused by missense or nonsense mutations in the gene encoding type X collagen (COL10A1).

Type X collagen and metaphyseal chondrodysplasia type Schmid. A study by Mullan et al. elegantly demonstrated that ER stress and the resulting growth plate pathology and reduced bone growth induced through the ER accumulation of mutant type X collagen can be corrected by the administration of carbamazepine (CBZ) in both cell and mouse models¹². Moreover, that study confirms that the mode of action of CBZ is to enhance mutant protein degradation by either the proteasome or autophagy in a mutation-dependent manner. In a follow-up study, it was demonstrated that an MCDS nonsense mutation in Col10a1 also caused mutant collagen X retention, ER stress and an UPR in hypertrophic chondrocytes¹³. Moreover, treatment of this mouse model of MCDS with CBZ reduced the disease severity in a manner similar to that previously described^12,13. Overall, these complementary studies provide confidence that CBZ is likely to be effective in an on-going clinical trial for patients with MCDS in which the disease is caused by nonsense or missense COL10A1 mutations (see https://mcds-therapy.eu).

Interestingly, whilst Xbp1 was alternatively spliced in the MCDS mouse models^14,15, the predominantly active pathways were the ATF6 and PERK branches of the UPR. Therefore, the role of the AFT6 branch of the UPR in relation to ER stress caused by the expression of mutant type X collagen has been studied in detail using cell and mouse models¹⁶. Previously, ATF6β has been overlooked as having little or no functional significance and is redundant to ATF6α. However, studies by Forouhan et al. confirmed that both isoforms have distinct functions under ER stress. Moreover, this study identified that the disease severity of MCDS is modulated by the opposing roles of ATF6α (chondroprotective) and ATF6β (detrimental). This brings a level of complexity to the ATF6 branch of the UPR, highlighting the functional significance of ATF6β under ER stress conditions and therefore outlining its possible therapeutic potential, which has been ignored in many diseases associated with ER stress.

Interestingly, it has been suggested that ATF4 expression (downstream of PERK) in the MCDS mouse model may in fact be pro-survival and involved in the de-differentiation pathways observed in the Col10a1 mutant chondrocytes. Indeed, ATF4 appeared to be upregulated in differentiating, proliferative chondrocytes, indicating zonal or differentiation state differences in the cartilage UPR responses. Overexpression of ATF4 in hypertrophic chondrocytes leads to a chondrodysplasia through reprogramming of hypertrophic cells towards Sox9 expression and a more juvenile phenotype¹⁵. Moreover, treating MCDS mice with the integrated stress response inhibitor (ISRIB) compound alleviates the MCDS phenotype¹⁵. ISRIB renders cells insensitive to eIF2α phosphorylation without affecting the phosphorylation process itself^17,18. Interestingly, treatment of MCDS mice with ISRIB restores bone growth and ablates the preferential translation of Atf4 and Chop whilst the Xbp1s signals remain unaffected.

Matrilin-3 and multiple epiphyseal dysplasia. Similar to MCDS, those forms of MED caused by MATN3 mutations (MATN3-MED) have ER stress and classic UPR as a core disease mechanism¹⁹; however, unlike MCDS, cell and mouse models of MATN3-MED do not respond to CBZ treatment (unpublished data). This dichotomy in drug response might be best explained by a recent study by Pirog et al., who demonstrated that different pathways of the UPR are activated depending on the differentiation state of growth plate chondrocytes²⁰. For example, growth plate pathology is most pronounced in the proliferative zone of the MATN3-MED mouse model, in contrast to the restricted hypertrophic zone pathology in both MCDS mouse models, and deletion of Xbp1 impacts on chondrocyte proliferation but not survival¹⁴. However, in both MCDS and MATN3-MED, the XBP1 branch of UPR is activated and there are clear similarities in gene expression changes. Therefore, it was unexpected that the genetic deletion of Xbp1 signalling in an MCDS mouse model resulted in no overt change in phenotype²¹ but that, in contrast, it caused a significant worsening of the phenotype in an MATN3-MED mouse model²⁰. Therefore, the authors concluded that the differentiation state of chondrocytes within the cartilage growth plate influences UPR signalling and that these differences could inform future therapies.

Cartilage oligomeric matrix protein and pseudoachondroplasia. ER stress is one of the stress pathways that form a network of interactions, together with oxidative stress, inflammation and autophagy. PSACH is another classic ER stress–related disease¹⁰; however, numerous previous studies using both in vitro^22,23 and in vivo^24,25 models have unequivocally demonstrated that ER retention of mutant COMP induces ER overload response (EROL)²⁶ that comprises oxidative stress and inflammation potentially mediated by nuclear factor kappa B (NF-κB) as opposed to a ‘classic’ UPR¹⁰. This observation identified new avenues for therapeutic intervention, including anti-inflammatory/antioxidant compounds such as aspirin and resveratrol²⁷. More recently, a role for mammalian target of rapamycin (mTOR) complex 1 in the pathogenesis of PSACH was demonstrated and suggested as a therapeutic target for COMPopathies²⁸.

Chondrocyte endoplasmic reticulum stress resulting from disrupted protein folding and trafficking

The correct structure and function of cartilage require the coordinated synthesis and secretion of a plethora of ECM components, including a broad range of macromolecules such as collagens, glycoproteins and proteoglycans. Chondrocytes, as professionally secreting cells, are particularly susceptible to ER stress and therefore the UPR plays a vital protective role. In addition to the genetic mutations described in the previous section, any disruption to normal protein folding or deletion of individual components of the UPR can have a profound detrimental effect. This has recently been illustrated by the engineered deletion in mouse chondrocytes of two important components of the protein-folding machine—mesencephalic astrocyte-derived neurotrophic factor (MANF)²⁹ and ER resident protein 57 (ERp57)³⁰—whereas compound heterozygosity for mutations in site 1 protease (S1P)³¹, a Golgi resident protease, causes a human chondrodysplasia.

Mesencephalic astrocyte-derived neurotrophic factor. MANF is an ER resident protein that is increased and secreted during ER stress due to an imperfect KDEL sequence. The global knockout of MANF (on C57BL/6) is lethal at birth because mice die of respiratory distress due to decreased saccular formation in the lungs²⁹. Interestingly, however, crossing MANF^−/− mice onto a Sv129 background alleviates the phenotype to the extent that mice survive to about 6 to 9 weeks of age. The cartilage-specific knockout of MANF is viable but causes a severe chondrodysplasia characterised by reduced long-bone lengths, confirming that endochondral ossification was disrupted²⁹. Despite morphologically normal growth plates, there was a concurrent reduction in chondrocyte proliferation in the cartilage-specific MANF null mice. This reduction in long-bone growth can be explained by the RNA sequencing (RNA-seq) analysis of chondrocytes from global and cartilage-specific knockout mice, which demonstrated increased expression of specific ER stress–related genes and a dysregulation of genes associated with chondrocyte proliferation, NF-κB signalling and terminal differentiation of chondrocytes.

Protein disulphide-isomerase A3/endoplasmic reticulum resident protein 57 (PDIA3/ERp57). The cartilage-specific knockout of ERp57 has previously been shown to cause ER stress and UPR, resulting in reduced chondrocyte proliferation and dysregulated apoptosis leading to abnormal bone growth³⁰. More recently, the administration of 4-phenylbutyric acid (4-PBA), a small chemical chaperone, to ERp57-deficient chondrocytes in micromass or explant cultures reduced ER stress³². Although these studies suggest that 4-PBA can alleviate ER stress in vitro, how effective it will be in vivo remains to be determined; indeed, previous studies in MATN3-MED mice have suggested limited efficacy¹⁹.

Site 1 protease. Mutations in S1P, a serine protease ubiquitously present in the Golgi apparatus and encoded by the MBTPS1 gene, cause a novel human recessive skeletal dysplasia³¹. Significantly, the mutations cause an almost complete loss of MBTPS1 mRNA transcripts that disrupts ER function, particularly in chondrocytes. More precisely, reduced levels of S1P protein impair the activation of the ER stress response gene BBF2H7, leading to the retention in the ER of type II collagen. This ultimately causes prolonged ER stress, resulting in increased chondrocyte apoptosis. Interestingly, the correction of the MBTPS1 mutations through antisense morpholino oligonucleotides or by the administration of PBA reduced collagen retention in the ER, suggesting a potential therapeutic avenue for this class of ER stress diseases.

Endoplasmic reticulum stress markers in osteoarthritis

Interestingly, the levels of ER stress markers BiP³⁵, CHOP and caspase-12 correlate and increase with OA severity³⁶, and OA-associated processes, such as biomechanical injury, age-related glycation, chronic inflammation, and oxidative stress, have been shown to trigger UPR responses in several studies^37–41. CHOP^36,42,43, alternatively spliced XBP1^39,44 and BiP^35,45 have been shown to be increased in multiple in vitro and in vivo OA studies, and it has been suggested that CHOP-mediated apoptosis is an important component of cartilage degeneration during OA progression. Moreover, interleukin 1 (IL-1) and IL-6 are important cytokine components in OA progression, and ER stress components have been shown to be upregulated by IL-1α treatment and to further sensitise the cells to IL-1β treatment^46–49.

Moreover, several ER stress markers, such as ERN1 (encoding IRE-1), PERK and CREB3L2 (encoding a transcription factor involved in the UPR), are decreased in OA chondrocytes⁴⁸. Whereas IL-1β treatment increases ERN1 and PERK expression in OA chondrocytes, platelet-derived growth factor (PDGF-BB) and IL-6 stimulation leads to a decreased expression of both but does not affect the levels of ATF6β, which are also comparable between the healthy and OA chondrocytes. Finally, small interfering RNA (siRNA) silencing of ERN1 led to decreased expression of COL2A1, MMP-13, and ADAMTS-4 and ADAMTS -5, and silencing of CREB3L2 showed a significant reduction of ADAMTS-5, which could be overcome by additional stimulation with IL-1β.

An adapted endoplasmic reticulum stress response could delay the onset of osteoarthritis

A recent study by Kung et al. employed the dislocation of the medial meniscus (DMM) mouse model to study trauma-induced OA⁵⁰. Interestingly, BiP expression was markedly increased after DMM surgery in wild-type mice, clearly demonstrating an involvement of ER stress and the UPR in the pathogenesis of OA. However, whilst BiP levels were also increased upon DMM surgery in the transgenic mice overexpressing a misfolding variant of thyroglobulin (Tg^cog) under the Col2a1 promoter (ColIITg^cog, a model of cartilage-specific ER stress⁵¹), the onset of OA was delayed, and a delay in cartilage apoptosis and damage was seen two weeks after surgery compared with wild-type controls. Interestingly, ColIITg^cog mice had elevated BiP protein levels prior to DMM surgery, and the pro-survival Xbp1-associated UPR signalling network was activated in ColIITg^cog mice after DMM, suggesting that pre-exposure to ER stress or activation of the specific pro-survival UPR responses such as the Xbp1s signalling could be chondroprotective.

ATF6α does not contribute to the capacity of the unfolded protein response in the context of osteoarthritis

Strikingly, although several studies suggested that ATF6 signalling may be responsible for the regulation of Xbp1 during OA progression, no difference in OA onset or progression could be seen in ATF6α-deficient mice after DMM surgery, suggesting that ATF6α may not be as relevant in this context as other branches of the UPR⁵⁰. Previous studies have shown a differential expression of ATF6α and ATF6β in a cartilage growth plate and a differentiation state–dependent modulation of ATF6 signalling¹⁶. Interestingly, silencing of ATF6β led to decreased ADAMTS-5 and ADAMTS-4 expression, indicating the importance of increasing our understanding of the UPR in the context of OA progression.

PERK: a major factor in osteoarthritis?

CHOP is a pro-apoptotic transcription factor downstream of the PERK and ATF6 signalling during the UPR. CHOP levels are increased in OA cartilage⁵² and in osteochondritis dissecans (OCD) cartilage degeneration in horses⁵³. It has been shown that CHOP can be upregulated as a result of inflammationor mechanical or oxidative stress in cartilage, all of which are important components of OA progression. Interestingly, chondrocyte apoptosis was decreased in CHOP null mice compared with wild-type controls, and OA-associated changes in expression of Col2a1 and Acan were suppressed in the absence of CHOP⁴². These data indicate that CHOP-mediated apoptosis may be an important component of OA progression.

A recent study by Hisanaga et al. investigated the role of PERK in ATDC5 cells and in primary chondrocytes from PERK-deficient mice⁵⁴. Interestingly, the absence of PERK led to a reduced secretion of collagen type II into the extracellular space in both models. Moreover, a decrease in PERK expression was detected in cultured OA chondrocytes, and siRNA silencing of PERK in chondrocytes led to a decrease in COL2A1 expression (further aggravated by IL-1β treatment) and an increase in COL1A1 expression. The switch from COL2A1 to COL1A1 expression is indicative of cartilage de-differentiation and fibrosis and is associated with later stages of OA⁵⁵; thus, these studies indicate that PERK modulation is a potential novel therapeutic target in OA.

The unfolded protein response in osteoarthritis: one of many therapeutic targets

The UPR signalling pathway is part of the complex crosstalk of signals comprising cellular response to stress, which include inflammation and autophagy^8,56. Autophagy is a process by which healthy chondrocytes remove their malfunctioning components and can be used to obtain energy during cellular stress, such as the UPR⁵⁷. Interestingly, pro-survival markers of autophagy microtubule-associated proteins 1A/1B light chain 3B (LC3) and beclin-1 are increased in early OA, but autophagy markedly decreases with OA severity^53,58,59. Moreover, deletion of mTOR from cartilage leads to an increase in autophagy and is chondroprotective⁶⁰.

Several pharmacological interventions in OA research also demonstrate the complexity of the crosstalk within the cellular stress pathways. Salubrinal (a chemical that elevates phosphorylation of eIF2α) is chondroprotective upon surgical induction of OA⁶¹; however, guanabenz, which also increases phosphorylation of eIF2α, does not have that effect. Interestingly, salubrinal can also reduce the phosphorylation of NF-κB and modulate matrix metalloproteinase (MMP) signalling by influencing the inflammation signalling pathway. IL-1β can suppress anti-inflammatory 5′ AMP protein kinase (AMPK) signalling in OA chondrocytes, and silencing of AMPK leads to increased levels of CHOP. 5-Aminoimidazole-4-carboxamide ribonucleotide (AICAR) is an AMP kinase signalling agonist. Interestingly, AICAR treatment of biomechanically injured alginate cultured bovine chondrocytes reduces the levels of CHOP⁴⁹. GLP-1R (glucagon-like peptide-1 receptor involved in glucose and energy homeostasis) can regulate ER stress-induced apoptosis and inflammatory response during OA. GLP-1R agonist has been shown to be chondroprotective in an anterior cruciate ligament (ACL) surgery model of OA in rat and decrease the levels of CHOP and active caspase-3 in cartilage via modulation of phosphoinositide 3-kinase/protein kinase B/mTOR (PI3K/AKT/mTOR) signalling pathway that is involved in autophagy⁶². H₂O₂ treatment of chondrocytes has been shown to trigger oxidative stress and to increase ER stress markers BiP, CHOP and caspase-12. Curcumin, an antioxidant and anti-inflammatory drug, can modulate the PERK/eIF2α/CHOP UPR axis through promoting sirtuin 1 (SIRT1) in chondrocytes and is chondroprotective in the ACL rat model of OA, leading to a decrease in CHOP and slower cartilage degradation⁴¹. All of the above further emphasise the complex role of ER stress in OA progression and the need to dissect the crosstalk of the cellular stress pathways in disease onset and progression.