Collagen type II suppresses articular chondrocyte hypertrophy and osteoarthritis progression by promoting integrin β1−SMAD1 interaction

Hypertrophic differentiation is not only the terminal process of endochondral ossification in the growth plate but is also an important pathological change in osteoarthritic cartilage. Collagen type II (COL2A1) was previously considered to be only a structural component of the cartilage matrix, but recently, it has been revealed to be an extracellular signaling molecule that can significantly suppress chondrocyte hypertrophy. However, the mechanisms by which COL2A1 regulates hypertrophic differentiation remain unclear. In our study, a Col2a1 p.Gly1170Ser mutant mouse model was constructed, and Col2a1 loss was demonstrated in homozygotes. Loss of Col2a1 was found to accelerate chondrocyte hypertrophy through the bone morphogenetic protein (BMP)-SMAD1 pathway. Upon interacting with COL2A1, integrin β1 (ITGB1), the major receptor for COL2A1, competed with BMP receptors for binding to SMAD1 and then inhibited SMAD1 activation and nuclear import. COL2A1 could also activate ITGB1-induced ERK1/2 phosphorylation and, through ERK1/2-SMAD1 interaction, it further repressed SMAD1 activation, thus inhibiting BMP-SMAD1-mediated chondrocyte hypertrophy. Moreover, COL2A1 expression was downregulated, while chondrocyte hypertrophic markers and BMP-SMAD1 signaling activity were upregulated in degenerative human articular cartilage. Our study reveals novel mechanisms for the inhibition of chondrocyte hypertrophy by COL2A1 and suggests that the degradation and decrease in COL2A1 might initiate and promote osteoarthritis progression.


INTRODUCTION
Hypertrophic differentiation of chondrocytes is the terminal stage of endochondral ossification in the growth plate. 1,2 Hypertrophic differentiated chondrocytes are characterized by an enlarged size, high expression of collagen type X (COL10A1), runt-related transcription factor 2 (RUNX2) and matrix metalloproteinase (MMP), and low expression of cartilaginous specific markers, such as collagen type II (COL2A1) and SRY-box 9 (SOX9). Hypertrophic differentiated chondrocytes eventually undergo apoptosis and are replaced by calcification. 1,2 However, chondrocyte hypertrophy is also found in degenerative articular cartilage, such as in osteoarthritis (OA). In contrast to endochondral ossification, which involves a programmed process of chondrocyte hypertrophy, healthy mature articular chondrocytes (ACs) remain in a postmitotic quiescent state and resist proliferation and hypertrophy through unknown mechanisms. However, in osteoarthritic cartilage, hypertrophic differentiation, combined with apoptosis and calcification, can be observed in degenerative chondrocytes. [3][4][5] Chondrocyte hypertrophy in OA disturbs cartilage homeostasis and is thought to be responsible for OA development. 5,6 If the underlying mechanisms of the induction of AC hypertrophy can be elucidated, it may be possible to treat OA by inhibiting these pathological changes. [5][6][7] COL2A1 is the major component of the cartilage matrix, and together with other proteins and proteoglycans, COL2A1 can form complex extracellular scaffolds to bear mechanical forces, maintain physiological homeostasis, and provide anchoring sites for chondrocytes, extracellular matrix (ECM) molecules, and growth factors. 8 Degradation and reduction of COL2A1, which are regarded as typical pathological changes, are frequently observed in OA cartilage. 9 Chondrocyte hypertrophy in OA cartilage is thought to contribute to the COL2A1 decrease. 3,5 In addition to its structural function, COL2A1 is an important extracellular signaling molecule that can regulate chondrocyte proliferation, metabolism, and differentiation, similar to soluble signals. [10][11][12][13] In this study, Col2a1 loss was demonstrated in a Col2a1 p.Gly1170Ser mutant mouse, and the decrease in Col2a1 accelerated chondrocyte hypertrophy, indicating that Col2a1 could regulate chondrocyte differentiation as a signaling molecule. However, the specific mechanisms remain unclear.
Integrins are major surface receptors of chondrocytes, and integrin β1 (ITGB1) is considered to be the most common and important receptor for COL2A1. 10,[13][14][15] In addition to connecting chondrocytes and the ECM, integrins play a vital role in signal transduction, and mitogen-activated protein kinases (MAPKs) act as typical downstream effectors of integrin signaling. 10,13,16,17 The conditional deletion of Itgb1 in early limb mesenchyme resulted in accelerated chondrocyte hypertrophy and defects in chondrocyte proliferation, 18 suggesting that ITGB1 is indispensable for chondrocyte differentiation; however, the molecular mechanisms remain unknown. Itgb1 gene-deficient mice [18][19][20] showed similar cartilage deformities to those with collagen type II alpha1 (Col2a1) gene mutations, 21,22 and in vitro studies have revealed that blockage of ITGB1 can inhibit the influence of COL2A1 on chondrocyte metabolism. 10,13 However, whether ITGB1 mediates the effect of COL2A1 on chondrocyte hypertrophy deserves further research. Moreover, as reported, the integrin family is involved in OA progression, [23][24][25] and integrin α1β1 is able to protect against posttraumatic OA-induced cartilage degradation. 23 The detailed mechanisms underlying the effect of ITGB1 on OA progression remain unclear.
Transformation growth factor-β (TGFβ)/small mother against decapentaplegic (SMAD) superfamily plays an indispensable role in various biological processes of chondrocytes, including hypertrophy. 7,[26][27][28][29] Canonical TGFβ/SMAD signaling can be transduced by the formation of complexes of TGFβ superfamily ligands with type II and I serine/threonine kinase receptors. The binding of ligands allows the constitutively active type II receptor kinases to phosphorylate the dormant type I receptor kinases and then enables the phosphorylation of receptor-regulated SMADs (R-SMADs, including SMAD1/2/3/5/8). R-SMADs bind with common-SMAD (co-SMAD, that is, SMAD4) and translocate into the nucleus to regulate target gene transcription. 26,30 In addition to canonical signaling pathways, the TGFβ superfamily can extensively interact with many other pathways, and MAPK has been widely reported to be able to phosphorylate the linker region of R-SMADs to inhibit the nuclear accumulation of R-SMADs and reduce their transcriptional activity. [31][32][33][34][35] The TGFβ/SMAD superfamily can be divided into two branches: the TGFβ-SMAD2/3 branch and the BMP-SMAD1/5/8 branch. BMP-SMAD1/5/8 has a stimulatory effect on chondrocyte hypertrophy, while TGFβ-SMAD2/3 has the opposite effect. 26 The TGFβ and BMP pathways collaborate in regulating the differentiation state of chondrocytes and maintaining cartilage homeostasis. 26,36 During OA, the balance of TGFβ/BMP signaling is disturbed and contributes to AC hypertrophy and OA development. 7,26,29,36,37 However, the detailed mechanism underlying the imbalance between TGFβ and BMP activity in OA cartilage remains unclear.
The current study focused on the signaling function of COL2A1 and demonstrated that COL2A1 inhibited chondrocyte hypertrophy and OA progression via the suppression of BMP-SMAD1 pathway activity. Furthermore, the specific mechanisms by which COL2A1 influences BMP-SMAD1 signaling are illustrated.

RESULTS
Chondrocytes from Col2a1 mutant mice underwent enhanced hypertrophic differentiation due to Col2a1 loss The generation of Col2a1 p.Gly1170Ser knockin mice and the detailed phenotypes of all genotypes, that is, wild types (WT), heterozygotes (Hetero), and homozygotes (Homo), have been previously described. 21 To evaluate the influence of the Col2a1 mutation on endochondral ossification-related pathways, a qPCR analysis was conducted to profile the expression of 84 pathway genes in chondrocytes from all genotypes (the complete data are provided in Supplementary Table 1 and Supplementary Fig. 1a). Differentially expressed genes are listed in Fig. 1a and divided into four groups in Supplementary Table 2. There were no significant differences between the wild types and the heterozygotes, but 26 genes were upregulated in the homozygotes. These 26 genes were submitted to gene ontology annotations of biological processes to determine the most significantly overrepresented biological process. After manually removing redundant terms, "ossification", "bone development", "cartilage development", "positive regulation of developmental process", and "positive regulation of cell differentiation" were identified (Fig. 1b). Together with the significantly upregulated hypertrophic markers of Col10a1, Col1a1, Runx2, and Dmp1 (Supplementary Table 2), these alterations indicated accelerated chondrocyte hypertrophy. Immunoblotting confirmed that the expression levels of Runx2 and Col10a1 were upregulated in homozygotes, while no obvious difference was found between wild types and heterozygotes (Fig. 1c). Immunohistochemical (IHC) staining showed that the normal architecture of the growth plate disappeared in homozygotes, while the heterozygous growth plate remained almost normal. In the homozygous growth plate, the regular alignment of proliferative and hypertrophic cells was disturbed. Resting and hypertrophic-like chondrocytes could still be found, while proliferating chondrocytes became fusiform and almost disappeared. The expression levels of Runx2 and Col10a1 were much higher in homozygotes, especially in hypertrophic-like chondrocytes (Fig. 1d).
The Col2a1 p.Gly1170Ser mutation can cause an endoplasmic reticulum stress (ERS)-unfolding protein reaction (UPR)-apoptosis cascade. 21 In homozygotes, misfolded Col2a1 is retained and degraded in the ER, leading to Col2a1 loss. qPCR demonstrated no significant difference in Col2a1 mRNA levels among the three genotypes (Fig. 1e). Immunoblotting and IHC assays showed a sharply reduced Col2a1 protein level in homozygotes, while there was no obvious difference between wild types and heterozygotes (Fig. 1c, f). Transmission electron microscopy analysis revealed an obvious reduction in collagen fibers in the homozygous cartilage matrix throughout the entire growth plate ( Supplementary Fig. 1b). The above data showed decreased Col2a1 protein in homozygotes, which was caused by its degradation.
To verify that the loss of Col2a1 caused enhanced hypertrophy, a rescue experiment was performed in which purified COL2A1 was added to a culture of homozygous chondrocytes. COL2A1 supplementation partially downregulated the expression of hypertrophic markers, which was highly elevated in homozygotes (Fig. 1g, h). Col2a1 silencing upregulated hypertrophic markers to a level similar to that found in the homozygotes (Fig. 1i, j; Supplementary Fig. 1c, d).
COL2A1 exerted suppression on chondrocyte hypertrophy and on the production of matrix-degrading enzymes COL2A1 expression in SW1353 (a human chondrosarcoma cell line), Hs819.T (a human chondrosarcoma cell line), and 293T cells was much lower than that in human ACs, while Col2a1 expression in ATDC5 cells (a mouse chondrogenic cell line) was similar to that found in mouse primary ACs ( Supplementary Fig. 2a−d).
SW1353 and Hs819.T cells underwent an additional COL2A1 treatment, 10 and the treatment suppressed both mRNA and protein expression of the hypertrophic markers (Fig. 2a, b). A TUNEL assay was then conducted to indicate the late hypertrophic status (i.e., apoptosis). Cells were treated with IL-1β to both induce OA-like pathological changes and increase the apoptotic rate, 38,39 and simultaneous COL2A1 stimulation reduced apoptotic rates ( Supplementary Fig. 2e). Silencing COL2A1 (or Col2a1) not only upregulated the expression of hypertrophic markers (Fig. 2c, Fig. 1 Chondrocytes from Col2a1 mutant mice underwent enhanced hypertrophic differentiation due to Col2a1 loss. a Col2a1 p.Gly1170Ser knockin mice were constructed, and the primary chondrocytes isolated from the articular cartilage of embryos of all three genotypes were cultured for 7 d. Then, we performed a qPCR array analysis (the experiments were replicated with four different litters of embryos). Differentially expressed genes (P < 0.05, with a fold change of >2 or a fold change of <0.5) in homozygotes are shown in the histogram. b Gene ontology annotation analysis of the biological processes of the differentially expressed genes from the qPCR array assay. Enrichment scores corresponding to each pathway provided by the DAVID annotation tool are displayed as −log(P values). A term was considered to be significantly enriched only if it passed the count threshold of six genes per annotation term and presented an EASE score, with the Benjamini −Hochberg correction set to 0.05. In the DAVID database, the EASE score is a modified Fisher exact P value used for enrichment analysis within gene lists, with P value = 0 representing perfect enrichment.  (Fig. 2e). In histological studies, the COL2A1 group exhibited a more homogeneous hyaline cartilage-like morphology, with fewer hypertrophic chondrocytes (Fig. 2e). Moreover, qPCR and immunoblotting assays confirmed that, in the COL2A1 group, the expression of hypertrophic markers was significantly downregulated (Fig. 2f, g). The effect of COL2A1 on the production of matrix-degrading enzymes was assessed. COL2A1 treatment in human ACs downregulated the expression of MMP9, MMP13, ADAMTS4, and ADAMTS5, while COL2A1 silencing exerted the opposite effect ( Supplementary Fig. 2n, o), suggesting that COL2A1 not only inhibited chondrocyte hypertrophy but also suppressed the production of catabolic enzymes. COL2A1 suppressed chondrocyte hypertrophy through regulation of the BMP-SMAD1 pathway To determine which signaling pathways are involved in the transduction of COL2A1's effect on chondrocyte hypertrophy, the differentially expressed genes obtained from the qPCR array were submitted to functional annotations from Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. After manually removing the redundant terms, "TGFβ signaling pathway" and "ECM−receptor interaction" were considered to be significantly enriched (Fig. 3a). Considering the majority of differentially expressed genes belonging to the TGFβ/BMP superfamily Hs819.T  Table 2), the activities of both TGFβ-SMAD2/3 and BMP-SMAD1/5/8 branches were examined in Col2a1 mutant mouse chondrocytes. Similar to the qPCR array analysis, no obvious differences were found between heterozygotes and wild types. However, in homozygotes, Smad1 phosphorylation (Ser463/Ser465) and its typical early-response genes (Id1, Id2, and Dlx5) were upregulated, while p-Smad2/3 and their typical early-response genes (Acvr1, Stat1, and Serpine1) were slightly downregulated ( Fig. 3b−e). Additional COL2A1 treatment downregulated p-SMAD1 S463/465 and its typical early-response genes but upregulated the phosphorylation of both SMAD2 and SMAD3 and their typical early-response genes (Fig. 3f, g). Silencing COL2A1 (or Col2a1) induced the opposite effect ( Supplementary  Fig. 3a-d).
Whether BMP-SMAD1 or TGFβ-SMAD2/3 mediated the effect of COL2A1 was also studied. High-dose BMP2 treatment, which activated SMAD1, was demonstrated to be able to release the inhibitory effect of COL2A1 on chondrocyte hypertrophy (Fig. 3h). In addition, COL2A1 was shown to be able to suppress BMP2induced chondrocyte hypertrophy (Fig. 3i), suggesting that COL2A1 inhibited chondrocyte hypertrophy through BMP-SMAD1 signaling. However, SB505124, which blocked SMAD2/3 activation, could not reverse the repression of COL2A1 on chondrocyte hypertrophy ( Supplementary Fig. 3e). Confocal laser scanning showed that COL2A1 treatment inhibited the nuclear import of SMAD1 (Fig. 3j), while Col2a1 silencing facilitated this process ( Supplementary Fig. 3f).
To confirm the clinical relevance and validity of the current study, the expression of COL2A1, p-SMAD1 S463/465 , and COL10A1 and their correlations were examined in cartilage samples derived from 50 patients with knee OA. IHC and semiquantitative assays were performed. The specimens were divided into high COL2A1 protein and low COL2A1 protein expression groups based on the median as a cutoff value. The COL2A1 high expression group was composed of 7 males and 19 females with a mean age of (68.04 ± 8.41) years. The COL2A1 low expression group was composed of 7 males and 17 females with a mean age of (67.38 ± 6.83) years. The gender and age distributions were not significantly different between the two groups. Lower expression levels of p-SMAD1 S463/465 and COL10A1 were found in the COL2A1 high expression group (Fig. 3k−m). ITGB1 receptor mediated the effect of COL2A1 on the BMP pathway and chondrocyte hypertrophy To determine how COL2A1 represses BMP-SMAD1 signaling, the ability of COL2A1 to directly act on bone morphogenetic protein receptors (BMPRs) was first studied. Serine/threonine phosphorylation of BMPR1A and BMPR1B was assessed to indicate the activity of the BMP receptor complex. There were no observable differences in the phosphorylation of either receptor, regardless of COL2A1 treatment (Fig. 4a, b), suggesting that COL2A1 has no direct effect on BMPRs.
COL2A1 signaling was associated with OA cartilage degeneration and pathological changes of the subchondral bone Safranin O/Fast green staining, HE staining, and the Mankin score were evaluated in both injured cartilage (mostly from the rim of the ulcer, comprising~3-4 mm of the surrounding tissue) and paired macroscopically unaffected cartilage distal to the damaged zone from ten patients with knee OA (Fig. 7a,  and p-ITGB1 were dramatically decreased in the injured cartilage, while the expression levels of COL10A1, RUNX2, ID1, and p-SMAD1 S463/465 were significantly elevated (Fig. 7c, d; Supplementary Fig. 7a, b).
Moreover, we graded the pathological changes of the subchondral bone in 50 OA cartilage samples according to previous studies 40   Immunoprecipitation was performed on 293T extracts with anti-SMAD1 antibody or anti-ITGB1 antibody, followed by immunoblotting with anti-SMAD1 antibody and anti-ITGB1 antibody. c 293T cells were transfected with SMAD1-Flag, ITGB1-HA or vector-Flag plasmids, and immunoprecipitation was carried out with anti-Flag antibody followed by immunoblotting with anti-HA antibody and anti-Flag antibody. d 293T cells were transfected with SMAD1-Flag, ITGB1-HA, or vector-HA plasmids, and immunoprecipitation was carried out with anti-HA antibody, followed by immunoblotting with anti-HA antibody and anti-Flag antibody. e SMAD1 protein was purified from 293T cell extracts expressing SMAD1-Flag and identified by Coomassie blue staining (left panel) and mass spectrometry (right panel). f Purified SMAD1-Flag protein was incubated with the product expressing ITGB1-HA, which had been linked to HA affinity agarose beads for 4 h at 4°C, followed by bead washing and immunoblotting. g, h Immunoprecipitation was performed on 293T cells treated with 100 μg·mL −1 COL2A1 or vehicle (0.05 mol·L −1 acetic acid) for 1 h with anti-SMAD1 antibody (g) or anti-ITGB1 antibody (h), followed by immunoblotting with anti-SMAD1 antibody and anti-ITGB1 antibody. i, j Immunoprecipitation was performed on 293T cells treated with COL2A1 or vehicle with anti-SMAD1 antibody (i) or anti-BMPR1A antibody (j), followed by immunoblotting with anti-SMAD1 antibody and anti-BMPR1A antibody. k, l Immunoprecipitation was performed with anti-SMAD1 antibody (k) or anti-BMPR1B antibody (l) on 293T cells treated with COL2A1 or vehicle followed by immunoblotting with anti-SMAD1 antibody and anti-BMPR1B antibody. m, n 293T cells were pretreated with 10 μg·mL −1 ITGB1 blocking antibody or anti-human IgG antibody for 1 h and then treated with COL2A1 or vehicle for 1 h. Immunoprecipitation with anti-SMAD1 antibody (m) or anti-ITGB1 antibody (n) was conducted, followed by immunoblotting with anti-SMAD1 antibody and anti-ITGB1 antibody. Blk ab, blocking antibody    levels of these molecules were detected, and the results are shown in Fig. 3k−m). We found that cartilage with more severe pathological changes in the subchondral bone displayed lower expression of COL2A1 and higher expression of COL10A1 and p-SMAD1 S463/465 ( Supplementary Fig. 7c-e), suggesting that the decrease in COL2A1 might also play a role in the subchondral bone damage of OA joints.

DISCUSSION
In our previous study, the Col2a1 p.Gly1170Ser homozygous mutation was shown to cause chondrodysplasia by activating the ERS-UPR-apoptosis cascade. 21 In addition to causing the ERS-UPRapoptosis cascade, misfolded type II procollagen was restrained and degraded in the ER, which is in accordance with many other studies of the COL2A1 (or Col2a1) mutation. [41][42][43][44] Thus, Col2a1 secretion and assembly decreased sharply and led to a situation in which Col2a1 was nearly absent from the homozygous cartilage matrix. 21 Because purified COL2A1, the soluble COL2A1 peptide, which has no structural function, can still inhibit chondrocyte hypertrophy in vitro, we suspected that the signaling activity of COL2A1 played a predominant role in regulating chondrocyte hypertrophy. This study illustrates the novel role of COL2A1 in regulating the chondrocyte transition between the quiescent state and hypertrophic differentiation. In healthy articular cartilage, chondrocytes remain in a postmitotic quiescent state, while in OA cartilage, some chondrocytes around OA cartilage lesions can be activated and undergo hypertrophy during the whole disease process. 4,5 In the current study, we demonstrate that COL2A1 acts as an extracellular signaling molecule to inhibit chondrocyte hypertrophy. Upon interaction with COL2A1, ITGB1 receptors compete with BMP receptors to bind with SMAD1 and then inhibit SMAD1 activation and nuclear import. COL2A1 also activated ITGB1induced ERK1/2 phosphorylation, and through ERK1/2−SMAD1 interaction, it further repressed SMAD1 activation, thus inhibiting BMP-SMAD1-mediated hypertrophic differentiation (Fig. 8). COL2A1 loss led to the release of BMP-SMAD1 signaling suppression, resulting in chondrocyte hypertrophy and the promotion of OA progression, while supplementation with COL2A1 reversed chondrocyte hypertrophy. COL2A1 is suggested to be crucial in the chondrocyte quiescent state-hypertrophic differentiation-transition: COL2A1 maintains the chondrocyte quiescent state, while loss of COL2A1 leads to hypertrophic differentiation. We revealed the detailed mechanisms underlying  Fig. 8 Functional mechanisms of COL2A1 on the BMP-SMAD1 pathway. The canonical BMP-SMAD1 pathway is initiated with the oligomerization of BMPR1 and BMPR2 induced by BMPs. BMPR2 phosphorylates BMPR1, and BMPR1, in turn, phosphorylates the C-terminus of SMAD1 (Ser463/465). C-terminal phosphorylated SMAD1 proteins partner with SMAD4 and translocate into the nucleus to initiate target gene transcription. ITGB1 is the major receptor for COL2A1, and upon its interaction with COL2A1, ITGB1 competes with BMPR1 for binding to SMAD1, thus inhibiting SMAD1 S463/465 phosphorylation and nuclear import. COL2A1 can also regulate SMAD1 activity via ITGB1-induced ERK1/2 phosphorylation. Activated ERK1/2 phosphorylates the linker region (Ser206) of both unphosphorylated and C-terminal phosphorylated SMAD1. The linker region-phosphorylated SMAD1 is resistant to phosphorylation by BMPR1 at the C-terminal, while the interaction between the C-terminal-phosphorylated SMAD1 and SMAD4 can also be interrupted by a second phosphorylation of the linker region of SMAD1. As a result, SMAD1 activation can be repressed via various signaling pathways specifically in response to COL2A1. BMP, bone morphogenetic protein Collagen type II inhibits chondrocyte hypertrophy C Lian et al.
COL2A1's regulation of the cartilaginous phenotype, in which ITGB1 and BMP-SMAD1 are the critical downstream effectors. Thus, our findings provide new insight into the regulation of cartilaginous phenotype and OA progression.
COL2A1 is the major component of the cartilage matrix, and its degradation is considered a typical pathological process in OA development. 3,9,45 However, the current study suggests that COL2A1 degradation and reduction can initiate and promote OA progression by accelerating chondrocyte hypertrophy. We now introduce a disease-amplifying loop model to explain the effect of decreased COL2A1 on OA development. During the early stage of OA, once focal cartilage lesions develop, the proliferation and anabolism of surrounding chondrocytes are activated to repair the damage, during which BMP-SMAD1 is simultaneously activated and plays a critical role. However, the continuous activation of BMP-SMAD1 signaling also induces a portion of the surrounding chondrocytes to undergo hypertrophic differentiation, downregulating COL2A1 expression and upregulating MMPs that can break down the existing COL2A1. COL2A1 reduction prevents its suppression of the BMP-SMAD1 pathway and further accelerates chondrocyte hypertrophy, leading to the formation of the selfperpetuating disease-amplifying loop. Through this self-enhancing loop, chondrocyte hypertrophy and cartilage matrix destruction could propagate from focal damage to neighboring cartilage. In light of this disease model, we propose that the COL2A1 decrease plays a central role in OA development by acting as a main factor driving the aforementioned disease-amplifying loop. Targeting the key factors in this disease-amplifying loop may be a promising alternative OA therapy, considering that supplementation with functional COL2A1 directly into an OA joint remains difficult. The COL2A1 p.G1170S mutation was demonstrated to be able to cause early-onset OA in a five-generation family, 46 and several other COL2A1 mutations have also been reported to be related to OA. [47][48][49] Although COL2A1 mutation is not frequently observed in human OA in the general population, considering the central role of cartilage matrix in OA development, studies of the effect of COL2A1 mutation on OA could provide insight into the regulation of cartilage homeostasis and shed light on potential essential molecular mechanisms in OA. Mice with inactivation of one allele of the Col2a1 gene showed softer articular cartilage, disturbed collagenous network, reduced subchondral bone volume, and increased susceptibility to OA, which adds credibility to the role of the COL2A1 decrease in OA development. 50 Although ITGB1 signaling has been proven to suppress chondrocyte hypertrophy by both transgenic mouse models and in vitro studies, the detailed molecular mechanisms are unclear. 13,18 The current study showed that the canonical BMP-SMAD1 pathway may mediate the effect of ITGB1 signaling on chondrocyte hypertrophy. The direct interaction between ITGB1 and SMAD1 was identified for the first time. Further study showed that COL2A1 facilitated ITGB1−SMAD1 interaction while simultaneously attenuating the BMPR1A/B−SMAD1 interaction. In this way, ITGB1 inhibited SMAD1 S463/465 phosphorylation and nuclear translocation in response to COL2A1. We also confirmed that COL2A1 could downregulate SMAD1 activation via the ITGB1-ERK1/2 axis-induced SMAD1 S206 phosphorylation. The current study not only reveals the mechanisms underlying ITGB1 regulation on chondrocyte hypertrophy but also partly explains how ITGB1 affects OA progression.
A new negative regulation mechanism on BMP-SMAD1 signaling was illustrated in this study: SMAD1 activity can be negatively regulated directly by the ITGB1 receptor in response to COL2A1. It has been well established that COL2A1 activates the TGFβ-SMAD2/3 pathway through a membrane signaling complex formed by type I TGFβ receptors and integrins. 10 Although TGFβ-SMAD2/3 plays a role in inhibiting chondrocyte hypertrophy, TGFβ signaling alone seems insufficient for suppressing the onset of hypertrophy. 51,52 Moreover, a shift in signaling dominance from TGFβ-SMAD2/3 to BMP-SMAD1/5/8 could induce AC hypertrophy. 53 The current study also confirmed that the inhibitory effect of COL2A1 on chondrocyte hypertrophy was mainly mediated by the BMP-SMAD1 branch. In physiological conditions, upon interacting with ITGB1, the tuning modulation of COL2A1 acts to optimize the intensity and duration of both the BMP-SMAD1 signal and the TGFβ-SMAD2/3 signal and maintains the balance between them. Under pathological conditions, the tuning effect of COL2A1 is disturbed due to its decrease, BMP-SMAD1 signaling prevails, and chondrocytes undergo hypertrophy.
In conclusion, the current study demonstrated that COL2A1 acts as a potent suppressor for chondrocyte hypertrophy and OA progression through the negative regulation of SMAD1 activity.

Animals
Detailed procedures for the construction of Col2a1 p.Gly1170Ser knockin mice have been previously described. 21 Animals were housed in a temperature-and humidity-controlled room under a 12-h light-and-dark cycle with food and water provided ad libitum.
Source of human cartilage Human cartilage samples were obtained from 60 OA patients classified as grades 3 and 4 according to the Kellgren and Lawrence osteoarthritis grading system. The patients underwent total knee arthroplasty at Sun Yat-sen Memorial Hospital of Sun Yat-sen University between 2014 and 2015. Cartilage samples were taken from ten patients (seven women and three men with a mean age of (68.10 ± 5.55) years) from both the injured area (mostly from the rim of the ulcer, comprising~3-4 mm of the surrounding tissue) and the macroscopically unaffected area distal to the damaged zone. Cartilage samples were taken from another 50 patients (36 women and 14 men with a mean age of (67.72 ± 7.62) years) from only the injured area. Measurements of mRNA expression and histology were performed. The samples were kept frozen at −80°C until use or placed in paraformaldehyde for the histological study.
Blocking antibodies against human integrin beta1, recombinant human interleukin-1 beta, and recombinant human BMP2 were obtained from R&D systems. Purified COL2A1 was purchased from Chondrex, Inc. (Redmond, WA, USA). U0126 and SP600125 were from CST. SB505124 was obtained from MedChemExpress (Monmouth Junction, NJ, USA). Isolation and culture of human bone marrow-derived MSCs MSCs were isolated and purified from bone marrow obtained from healthy volunteer donors using density-gradient centrifugation, as described previously. 54 Cells were resuspended in low-glucose DMEM (Gibco, Waltham, MA, USA) with 10% FBS, seeded and incubated at 37°C/5% CO 2 . After 48 h, nonadherent cells were removed by replacing the medium. Then, the medium was replaced every 3 d. When the cells reached (80-90)% confluence, they were trypsinized, counted, and reseeded as the first passage. Cells from passages 3-6 were used for subsequent experiments.
Chondrogenic and hypertrophic differentiation of human MSCs A high-density pellet culture system was applied for the chondrogenic differentiation of human MSCs, as described previously. 54  To further induce hypertrophic differentiation, chondrogenic differentiated pellets were exposed to hypertrophic differentiation medium consisting of high-glucose DMEM with 50 μg·mL −1 ascorbate acid 2-phosphate, 40 μg·mL −1 proline, 1 nmol·L −1 dexamethasone, 1% ITS Universal Culture Supplement Premix, and 1 nmol·L −1 triiodothyronine (Sigma-Aldrich) for 14 d as described previously. 52 The medium was changed every 3 d.
Real-time RT-PCR assay Real-time PCR was performed on a Roche LightCycler 480 System (Roche, Basel, Switzerland) using SYBR Green Real-time PCR Master Mix (TOYOBO, Osaka, Japan). The primer sequences used in this study are listed in Supplementary Table 4. Each reaction was processed in triplicate, and an average ΔCt value from the whole group was taken. The relative expression levels of each gene were obtained using the 2 −ΔΔCt method.
Real-time PCR array Total RNA from cultured mouse primary chondrocytes was extracted using an RNeasy plus micro kit (Qiagen, Duesseldorf, Germany). The high-quality RNA from 12 samples (3 genotypes × 4 different litters) was converted into cDNA with an RT2 First Strand cDNA Kit (SABiosciences, Duesseldorf, Germany). Osteogenesisrelated gene expression was determined by using the PCR Array of PAMM-026 (SABiosciences; the list of genes is available on its website) and the 7500 Real-Time PCR system (Applied Biosystems, Waltham, USA) according to the manufacturer's instructions. Statistical analysis was performed using paired-samples tests, and the fold changes were calculated using the 2 −ΔΔCt method with the Web-Based PCR Array Data Analysis system (SABiosciences). Differentially expressed genes were identified by a fold change of >2 and a change in P value < 0.05 for increased expression or a fold change of <0.5 and a change in P value < 0.05 for decreased expression. The Database for Annotation, Visualization, and Integrated Discovery (DAVID), Bioinformatics Resources 6.7 (https://david.ncifcrf.gov/home.jsp) was also used for data mining.
Concentration and quality score measurement of RNA samples Total RNA was isolated from cells using RNAiso Plus reagent (TaKaRa, Kusatsu, Japan) and detected by RNA LabChip (Perki-nElmer, Inc., Waltham, MA, USA) and an RNA Reagent kit (PerkinElmer, Inc.) on a LabChip GX Touch HT Bioanalyzer (PerkinElmer, Inc.) according to the manufacturer's protocols. The concentration and RNA quality score (RQS) of all RNA samples are listed in Supplementary Table 5.
Preparation of nuclear and cytoplasmic extracts Nuclear and cytoplasmic extracts were prepared using nuclear and cytoplasmic extraction reagents (Thermo Scientific), following the manufacturer's recommended protocol.
Immunoblotting analysis Cells were washed twice with ice-cold phosphate buffer saline, harvested, and resuspended in RIPA Lysis Buffer (Beyotime, Shanghai, China) plus protease inhibitor cocktail (MedChemExpress). Cell lysates were obtained by centrifugation at 12 000 r·min −1 for 10 min at 4°C. Equal amounts of each sample were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to polyvinylidene fluoride membranes (Millipore, Boston, MA, USA). Membranes were blocked with 5% nonfat dry milk for 1 h at room temperature and then incubated with the designated antibodies. Antibody-specific labeling was revealed by incubation with HRP-conjugated secondary antibodies for 1 h and visualized with an electrochemiluminescence (ECL) kit (Millipore). Images were captured and analyzed with an ImageQuant Las 4000mini imaging system (GE Healthcare Life Science, Chicago, IL, USA). All immunoblotting assays were conducted with three biological replicates, and representative images are shown.
Collagen type II inhibits chondrocyte hypertrophy C Lian et al.
Using lysis buffer, lysates were prepared from cells transfected with plasmids during exogenous immunoprecipitation. The same amount of protein from each sample was incubated with 20 μL anti-Flag affinity agarose or anti-HA affinity agarose (Biotool) overnight at 4°C. The agarose-bound immunoprecipitates were washed seven times with immunoprecipitation wash buffer and collected by centrifugation. Samples were then subjected to immunoblotting analysis after the addition of 30 μL of sample buffer and denaturation.
TUNEL labeling assay A TUNEL assay was performed according to the manufacturer's instructions (MBL, Nagoya, Japan), and photographs were captured using a Leica DMI4000B microscope (Leica, Wetzlar, Germany). The percentages of TUNEL-positive cells relative to 4',6diamidino-2-phenylindole (DAPI)-stained cells were calculated. Three independent experiments were conducted and calculated for each experimental group.
Immunofluorescence and confocal microscopy analysis Cells were treated with purified collagen type II for 24 h, fixed with 4% paraformaldehyde, pretreated with 1% Triton X-100 and 0.5% bovine serum albumin (BSA) in phosphate buffer saline tween-20 (PBST), and blocked with 10% BSA in PBST for 30 min at room temperature. Cells were labeled with anti-SMAD1 antibody and treated with a secondary antibody, Alexa Fluor 555-conjugated goat anti-rabbit IgG, and then stained with DAPI. Signals were captured with a Zeiss LSM780 confocal laser scanning microscope (Zeiss, Jena, Germany).

Statistical analysis
The results are given as the mean ± standard deviation (SD). Statistical analysis was performed using the two-tailed independent Student's t test for comparisons of two independent groups, two-tailed paired-sample t test for comparisons of two matched groups, and one-way ANOVA followed by Dunnett's post hoc test for multiple comparisons. In all cases, a P value less than 0.05 was considered statistically significant. All statistical analyses were conducted with the SPSS 13.0 statistical software package.
Study approval All experimental procedures involving animals met the relevant guidelines for the humane care of laboratory animals and were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University. For human studies, prior patients' consent and approval from the Institutional Research Ethics Committee of Sun Yat-sen University were obtained.