ARTICLE https://doi.org/10.1038/s41467-019-09491-5 OPEN Excessive mechanical loading promotes osteoarthritis through the gremlin-1–NF-κB pathway 1 1,2 1 1,3 1 1 Song Ho Chang , Daisuke Mori , Hiroshi Kobayashi , Yoshifumi Mori , Hideki Nakamoto , Keita Okada , 1 1 2 4 5 6 Yuki Taniguchi , Shurei Sugita , Fumiko Yano , Ung-il Chung , Joo-ri Kim-Kaneyama , Motoko Yanagita , 7 8 9 1 1,2 Aris Economides , Ernesto Canalis , Di Chen , Sakae Tanaka & Taku Saito Exposure of articular cartilage to excessive mechanical loading is deeply involved in the pathogenesis of osteoarthritis. Here, we identify gremlin-1 as a mechanical loading-inducible factor in chondrocytes, detected at high levels in middle and deep layers of cartilage after cyclic strain or hydrostatic pressure loading. Gremlin-1 activates nuclear factor-κB signalling, leading to subsequent induction of catabolic enzymes. In mice intra-articular administration of gremlin-1 antibody or chondrocyte-speciﬁc deletion of Gremlin-1 decelerates osteoarthritis development, while intra-articular administration of recombinant gremlin-1 exacerbates this process. Furthermore, ras-related C3 botulinum toxin substrate 1 activation induced by mechanical loading enhances reactive oxygen species (ROS) production. Amongst ROS-activating transcription factors, RelA/p65 induces Gremlin-1 transcription, which antagonizes induction of anabolic genes such as Sox9, Col2a1, and Acan by bone morpho- genetic proteins. Thus, gremlin-1 plays essential roles in cartilage degeneration by excessive mechanical loading. 1 2 Sensory and Motor System Medicine, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan. Bone and Cartilage Regenerative Medicine, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan. Division of Oral Anatomy, Department of Human Development and Fostering, Meikai University School of Dentistry, 1-1 Keyakidai, Sakado, Saitama 350-0283, Japan. Center for Disease Biology and Integrative Medicine, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, 5 6 Japan. Department of Biochemistry, Showa University School of Medicine, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8555, Japan. Department of Nephrology, Graduate School of Medicine, Kyoto University, 54 Shogoin Kawahara-cho, Sakyo-ku, Kyoto 606-8507, Japan. Regeneron Pharmaceuticals, Inc., 777 Old Saw Mill River Road, Tarrytown, NY 10591, USA. Departments of Orthopaedic Surgery and Medicine and the Musculoskeletal Institute, UConn Heath, Farmington, CT 06030, USA. Department of Orthopedic Surgery, Rush University Medical Center, Chicago, IL 60612, USA. Correspondence and requests for materials should be addressed to T.S. (email: firstname.lastname@example.org) NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 1 1234567890():,; ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 steoarthritis (OA) is the most prevalent joint disorder primary chondrocytes after 0.5 Hz, 10% cyclic tensile strain occurring with articular cartilage degradation. Since the loading for 30 min. Mmp13 mRNA expression was slightly Oestablishment of experimental mouse models with surgi- increased 1 h after loading, peaked by 12–24 h, and declined cally induced knee joint instability, a large number of studies have thereafter to baseline by 72 h after loading (Fig. 1a). To identify revealed the major molecules or signalling pathways responsible for target genes mediating Mmp13 induction by stress loading, we OA, such as a disintegrin-like and metallopeptidase with a throm- performed microarray analysis using mRNA samples of chon- bospondin type 1 motif 5 (Adamts5), matrix metalloproteinase-13 drocytes before and 24 h after loading. Abundantly expressed (Mmp13), hedgehog signalling, syndecan-4, Wnt signalling, and genes upregulated or downregulated more than two-fold are 1–13 hypoxia-inducible factor 2-alpha (HIF-2α) .Inparticular, shown in Supplementary Tables 1 and 2. Among upregulated and Mmp13 is responsible for degradation of type 2 collagen (Col2a1), a downregulated NF-κB-related genes identiﬁed by gene ontology major matrix protein component of articular cartilage, and plays analyses (Supplementary Tables 3 and 4), we focused on gremlin- 6,10 essential roles in OA development . 1 (Grem1), which showed abundant expression and was most The nuclear factor kappa-light-chain-enhancer of activated B highly upregulated by mechanical loading in the group (Supple- cells (NF-κB) protein complex plays essential roles in various mentary Table 3). Gremlin-1 is a secreted protein and known biological processes including cell survival, proliferation, differ- bone morphogenetic protein (BMP) antagonist . mRNA and entiation, apoptosis, aging, inﬂammation, and immune respon- protein levels of gremlin-1, as determined by real-time RT-qPCR 14–16 ses . It consists of v-rel reticuloendotheliosis viral oncogene and immunocytochemistry, increased after loading (Fig. 1b, c). homologue A (RelA, also known as p65), RelB, Rel, p105/p50, Next, we examined Gremlin-1 mRNA expression in mouse and p100/p52. These proteins form heterodimers to function as femoral heads under cyclic hydrostatic pressure loading by the transcriptional activators. Inhibitors of NF-κB(IκB) proteins, originally developed system (Supplementary Fig. 1a–e). Although including IκBα,IκBβ,IκBγ,IκBε,IκBζ and Bcl-3, bind and Gremlin-1 mRNA levels were unchanged by 5 MPa loading, they sequester NF-κB family members within the cytoplasm .In were signiﬁcantly increased by 20 MPa (Fig. 1d). Immuno- response to one of several signals, activation of IκB kinases (IKKs) ﬂuorescence determined that gremlin-1 protein was highly results in phosphorylated IκB proteins, which causes their increased within the deep zone of the femoral head after 20 MPa degradation to enable free NF-κB complexes to translocate from loading (Fig. 1e). We next examined localization of gremlin-1 the cytoplasm into the nucleus where they trigger target gene protein in the articular cartilage of a knee OA mouse model, transactivation . NF-κB signalling, which is widely involved in which involves surgical resection of the medial meniscus and OA pathophysiology through various effects, is activated in medial collateral ligament . In normal mouse articular cartilage, osteoarthritic chondrocytes during aging and inﬂammation . gremlin-1 protein was predominantly localized within the deep NF-κB signalling is essential to induce various inﬂammation- layer (Fig. 1f). Two weeks after surgical induction, gremlin-1 was related factors, including Mmp proteins, inducible nitric oxide highly detected in and around deep layer chondrocytes (Fig. 1f). synthase (iNOS), interleukin 1 beta (IL-1β), tumour necrosis Its expression was slightly increased in the superﬁcial zone after 8,9 factor alpha (TNF-α), and HIF-2α . HIF-2α further induces 4 weeks (Fig. 1f). We conﬁrmed increased gremlin-1 protein 8,9 various catabolic enzymes and OA-related genes . Recently, we levels in mouse primary chondrocytes, femoral heads, and showed that NF-κB signalling regulates articular cartilage articular cartilage using immunohistochemistry with a second homoeostasis and degeneration in a biphasic manner . Although primary antibody, and observed decreased protein levels using NF-κB signalling is inactivated in normal articular chondrocytes, siRNA against gremlin-1 as a negative control (Supplementary a small amount of intranuclear RelA is required for transcrip- Fig. 2a–d). We further examined gremlin-1 expression in human tional induction of anti-apoptotic genes that are indispensable articular cartilage surgical specimens. Similar to the expression for chondrocyte survival . Phosphorylated IκBα and increased pattern in mouse cartilage, gremlin-1 protein was localized within intranuclear RelA accompany cartilage degeneration, leading to the middle and deep layers of mildly degenerated human cartilage induction of catabolic and inﬂammatory molecules and accel- in the lateral tibial compartment (Supplementary Fig. 3). eration of OA development . Gremlin-1 was intensively detected in the clustered chondrocytes In addition to molecular biology research, clinical and epide- in the superﬁcial zone in severe human OA cartilage of the medial miologic studies previously revealed various factors to be asso- tibial compartment, as well as in and around deep layer chon- ciated with OA pathogenesis, including aging, obesity, joint drocytes (Supplementary Fig. 3). instability, trauma, and joint inﬂammation. Excessive mechanical loading is regarded as the essence of several of these factors. A Gremlin-1 exerts catabolic effects in vitro. To investigate the previous in vitro experiment using cell-stretcher systems showed effects of gremlin-1, which is increased by excessive mechanical induction of MMP13 by excessive mechanical loading . NF-κB loading and/or during OA development, we administered 19,22 signalling is also regulated by mechanical loading . However, recombinant human gremlin-1 (rhGREM1) protein to cultured molecular mechanisms underlying cartilage degeneration by chondrocytes. In mouse primary chondrocytes, catabolic marker excessive mechanical loading remain unknown. genes such as Mmp13 and Adamts5 were increased, while Sox9 Here, we describe a signalling pathway linking excessive (a master transcription factor for chondrocytes) and cartilage mechanical loading to cartilage degeneration. We perform a matrix genes Col2a1 and Acan were decreased by rhGREM1 in a screen for genes altered by mechanical loading, and focus on the dose-dependent manner, and time-dependent manner up to 24 h NF-κB-related gene Gremlin-1 among the mechanical stress- (Fig. 2a, and Supplementary Fig. 4). Similar results were obtained inducible candidates. We examine its expression in articular using rhGREM1 manufactured by a second company (Supple- cartilage, roles in in vitro and in vivo OA development, and mentary Fig. 5). Type 10 collagen (Col10a1), a hypertrophic further upstream and downstream pathways connecting excessive chondrocyte marker, was not changed by rhGREM1 (Supple- mechanical loading to cartilage degeneration. mentary Fig. 6). In ex vivo cultures of femoral head cartilage from wild-type (WT) mice, release of proteoglycans into the medium was increased by rhGREM1 treatment in a dose- Results dependent manner (Fig. 2b), while mRNA levels of marker genes Excessive stress loading induces gremlin-1 in chondrocytes.We within femoral heads were altered similarly to cultured chon- ﬁrst examined a time-course of Mmp13 mRNA levels in mouse drocytes (Fig. 2c). 2 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE a b Mmp13 Grem1 ## ## 6 Stress + Safranin O Grem1 Stress – ## 0 12 24 h 1 3 6 12 24 48 72 h Stress – Stress + Grem1 (5 MPa) Grem1 (20 MPa) ** 1.5 1.0 0.5 0 0 Stress – Stress + Stress – Stress + Stress – Stress + Fig. 1 Induction of gremlin-1 by excessive mechanical loading. a Time-course of Mmp13 mRNA levels in mouse primary chondrocytes after tensile stress loading (stress+), or without loading (stress −). Cells were cultured up to 72 h after uniaxial cyclic tensile strain (10%, 0.5 Hz, 30 min). P< 0.005, ## P< 0.001 versus stress—at each time-point (two-way ANOVA test). b Gremlin-1 mRNA levels in mouse primary chondrocytes after tensile stress loading. *P < 0.05 versus 0 h (one-way ANOVA test). c Gremlin-1 protein expression in mouse primary chondrocytes 24 h after tensile stress loading (stress+), or without loading (stress −). Nuclei were stained with DAPI (blue). Scale bars, 50 µm. d Gremlin-1 mRNA levels in mouse femoral head cartilage after hydrostatic pressure loading (5 or 20 MPa, 0.1 Hz, 30 min). P< 0.005 versus stress− (Student’s unpaired two-tailed t-test). e Gremlin-1 immunoﬂuorescence in mouse femoral head cartilage after 20 MPa hydrostatic pressure loading. Nuclei were stained with DAPI (blue). Scale bars, 100 µm. f Safranin O staining and gremlin-1 immunoﬂuorescence during the time-course of mouse osteoarthritis development after surgical induction. Inset boxes in safranin O staining indicate regions of immunoﬂuorescence. Scale bars, 100 µm and 50 µm, respectively. All data are expressed as mean ± SD of biologically independent three samples per group Gremlin-1 exerts catabolic effects on mouse cartilage. We next into the knee joints of OA model mice twice a week. Eight weeks examined the effects of gremlin-1 in vivo. The surgical OA model after surgery, OA progression was signiﬁcantly suppressed by was prepared by injecting rhGREM1 into the knee joints antibody injection (Fig. 3d). Moreover, chondrocyte apoptosis, of WT C57BL/6J male mice at 8 weeks of age, twice a week. Eight expression of catabolic marker proteins, and phosphorylated weeks after surgical induction, OA development was signiﬁcantly IκBα were also decreased (Fig. 3e, f). enhanced by intra-articular administration of rhGREM1 (Fig. 3a). Chondrocyte apoptosis determined by TUNEL staining was unchanged by rhGREM1 treatment (Fig. 3b). Mmp13, Adamts5, Conditional knockout of gremlin-1 suppresses OA develop- IκBα, phosphorylated IκBα, and HIF-2α were highly detected ment. To reveal the physiological roles of gremlin-1 in articular in the remnant cartilage of rhGREM1-treated groups (Fig. 3c). cartilage, we performed in vivo loss-of-function analyses using Moreover, intra-articular injection of rhGREM1-induced OA transgenic mice in which Cre recombinase was fused to a mutated development even in normal mouse knee joints (Supplementary ligand-binding domain of the human oestrogen receptor driven ERT2 Fig. 7). by the Col2a1 promoter (Col2a1-Cre ) to knockout gremlin-1 In a contrasting approach, we used an antibody against in adult articular cartilage after skeletal growth. Thus, adminis- gremlin-1. To conﬁrm the neutralizing effect of the commercially tration of the oestrogen antagonist tamoxifen resulted in trans- obtained antibody, we treated mouse primary chondrocytes with location of the fusion protein into nuclei causing gene targeting. rhGREM1 and gremlin-1 antibody. rhGREM1-induced increases We generated tamoxifen-inducible chondrocyte-speciﬁc homo- ERT2 in Mmp13 and Adamts5 were signiﬁcantly reduced by antibody zygous gremlin-1 knockout mice by mating Col2a1-Cre mice ﬂ/ﬂ ERT2 ﬂ/ﬂ ERT2 treatment (Supplementary Fig. 8). We then injected the antibody with Grem1 mice (Col2a1-Cre ;Grem1 ). Col2a1-Cre ; NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 3 0.25 0.5 Fold increase Fold increase Fold increase Fold increase 8 weeks 6 weeks 4 weeks 2 weeks 0 weeks ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 a b Mmp13 ## ** 01 0.1 10 Adamts5 μg per ml ## 2 * Mmp13 Sox9 Adamts5 6 # ** ## ## 0.5 ## ## Sox9 Col2a1 1 0.5 ## ## ## 0.5 Col2a1 ## 1.5 0.5 * Acan ** Acan ## ## 1 0.5 ## ## 0.5 01 0.1 0.5 310 0 μg per ml μg per ml Fig. 2 Catabolic effects of gremlin-1 in cultured chondrocytes. a mRNA levels of marker genes in primary mouse articular chondrocytes treated with recombinant human gremlin-1 (rhGREM1) for 24 h. b Amount of glycosaminoglycans (GAG) released into the culture medium determined by dimethylmethylene blue assay of wild-type mouse femoral heads cultured with various amounts of rhGREM1 for 3 days. c mRNA levels of marker genes in mouse femoral heads cultured with 10 µg/mL rhGREM1 for 3 days. All data are expressed as mean ± SD of biologically independent three samples # ## per group. *P < 0.05, **P < 0.01, P < 0.005, P < 0.001 versus 0 (vehicle) (one-way ANOVA test) ﬂ/ﬂ Grem1 (cKO) mice developed normally and displayed no phosphorylated IκBα were suppressed by gremlin-1 knockout, ﬂ/ﬂ skeletal abnormalities compared with Grem1 (Cntl) littermates similar to results observed in the surgical OA model (Fig. 4e, f). (Supplementary Fig. 9). Tamoxifen was injected into 7-week-old Efﬁcient knockout of gremlin-1 in articular chondrocytes was cKO and Cntl littermates daily for 5 days, and the surgical conﬁrmed at both mRNA and protein levels (Fig. 4c, f, OA model was induced 2 days after the last injection. Eight Supplementary Figs. 10, and 11). weeks after surgical induction, OA development was signiﬁcantly inhibited in cKO knee joints compared with Cntl joints (Fig. 4a). Chondrocyte apoptosis and expression of catabolic enzymes HIF- Gremlin-1 activates NF-κB signaling. We further analysed 2α and phosphorylated IκBα were suppressed by gremlin-1 molecular mechanisms underlying the catabolic effects of knockout (Fig. 4b, c). gremlin-1 in articular cartilage by performing a luciferase assay We further examined OA development with aging. Tamoxifen using reporter vectors containing response elements for repre- was injected into cKO and Cntl littermates daily for 5 days sentative signalling pathways. Among them, the NF-κB reporter at 8 weeks, 6 months, and 12 months. At 18 months of age, vector showed the highest transactivity induced by rhGREM1 cKO mice were signiﬁcantly resistant to cartilage degeneration (Fig. 5a). When various amounts of rhGREM1 were applied compared with control littermates (Fig. 4d), and showed no to cells transfected with the NF-κB reporter vector, transactivity skeletal abnormalities (Supplementary Fig. 9). Chondrocyte was increased in a dose-dependent manner (Fig. 5b). In addi- apoptosis and expression of catabolic enzymes HIF-2α and tion to rhGREM1, we next treated mouse femoral head cartilage 4 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications Fold increase μg per ml Fold increase NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE Safranin O TUNEL a b OARSI score n =8 n =8 S32/36 c Mmp13 Adamts5 p-IκBα Hif2a Safranin O IκBα ## % % % % % ** 60 60 60 60 60 Vehicle rhGREM1 30 30 30 30 0 0 0 0 0 de Safranin O TUNELI OARSI score 15 # ** n =8 n =10 S32/36 Safranin O Mmp13 Adamts5 IκBα p-IκBα Hif2a % % % % % 60 * 60 60 60 60 * * * Vehicle 30 30 30 30 30 Antibody 0 0 0 0 0 with the IKK inhibitor BMS-345541. Increased release of RelA-cKO femoral heads (Fig. 5d). Moreover, rhGREM1-induced proteoglycans into the medium by rhGREM1 treatment was increases in Mmp13 and Adamts5 were not observed in RelA- signiﬁcantly inhibited by BMS-345541 to a level similar to control cKO femoral heads, and decreases in anabolic marker genes ﬂ/ﬂ (Fig. 5c). Femoral heads from RelA (RelA-Cntl) and Col2a1- including Sox9, Col2a1, and Acan were not rescued (Fig. 5e). ﬂ/ﬂ Cre;RelA (RelA-cKO) mice were also treated with rhGREM1. These data indicate that gremlin-1 induces catabolic enzymes Release of proteoglycans was enhanced by rhGREM1 in through NF-κB signalling, but does not suppress anabolic genes RelA-Cntl femoral heads; however, it was not enhanced in in the same pathway. NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 5 Antibody Vehicle Antibody Vehicle rhGREM1 Vehicle rhGREM1 Vehicle Vehicle Vehicle rhGREM1 Antibody Antibody Vehicle rhGREM1 Vehicle Vehicle Vehicle Antibody rhGREM1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 Fig. 3 Effects of recombinant human gremlin-1 (rhGREM1) and gremlin-1 antibody in vivo. a Safranin O staining and OARSI scores of mouse knee joints after intra-articular administration (twice a week for 8 weeks) of 10 µLof10 µg/mL rhGREM1 or vehicle. Both experimental groups consist of n= 8 biologically independent animals. Inset boxes indicate regions of immunoﬂuorescence in (c). Scale bars, 100 µm. b TUNEL staining and rate of TUNEL- positive cells in mouse knee joints after intra-articular administration of rhGREM1 or vehicle. Nuclei were stained with DAPI (blue). Scale bars, 100 µm. n = 5 biologically independent experiments. c Safranin O staining and immunoﬂuorescence of Mmp13, Adamts5, IκBα, phosphorylated IκBα (dual Ser32/36), and HIF-2α proteins in mouse knee joints after intra-articular administration of rhGREM1 or vehicle. Scale bars, 50 µm. The percentage of positive cells in the immunoﬂuorescence is shown below. n = 5 biologically independent experiments. d Safranin O staining and OARSI scores of mouse knee joints after with intra-articular administration (twice a week for 8 weeks) of 10 µLof10 µg/mL gremlin-1 antibody or vehicle. Both experimental groups consist of biologically independent animals: vehicle n = 8, antibody n = 10. Inset boxes indicate regions of immunoﬂuorescence in (f). Scale bars, 100 µm. e TUNEL staining and rate of TUNEL-positive cells in mouse knee joints after intra-articular administration of gremlin-1 antibody or vehicle. Nuclei were stained with DAPI (blue). Scale bars, 100 µm. n= 5 biologically independent experiments. f Safranin O staining and immunoﬂuorescence of Mmp13, Adamts5, IκBα, phosphorylated IκBα (dual Ser32/36), and HIF-2α proteins in mouse knee joints after intra-articular administration of gremlin-1 antibody or vehicle. Scale bars, 50 µm. The percentage of positive cells in the immunoﬂuorescence is shown below. n = 5 biologically independent experiments. # ## All data are expressed as mean ± SD. *P < 0.05, **P < 0.01, P < 0.005, P < 0.001 versus vehicle (Student’s unpaired two-tailed t-test) We next examined which molecule functions as a receptor for Rac1 is widely known to act upstream of reactive oxygen gremlin-1 to stimulate NF-κB signalling. Among several candi- species (ROS) production in a variety of cell types . Thus, we dates from previous reports, we focused on vascular endothelial next examined in situ ROS production in mouse primary articular growth factor receptor 2 (VEGFR2), which is known to bind to chondrocytes after cyclic tensile strain loading. ROS was detected 24,25 gremlin-1 . Using the selective VEGFR2 inhibitor SU5416 in immediately after loading for 30 min, increased until 1 h, and was combination with rhGREM1, we found that increased Mmp13 diminished at 3 h (Fig. 8a). Several transcription factors including and Adamts5 by rhGREM1 were suppressed by SU5416 in a dose- NF-κB, HIF-1α, c-Jun, and nuclear factor erythroid 2-related dependent manner (Supplementary Fig. 12a). Moreover, SU5416 factor 2 (Nrf2) are activated by the Rac1-ROS axis . As such, we did not affect expression of Sox9, Col2a1,or Acan, similar to prepared a luciferase reporter vector containing a 5ʹ-end ﬂanking RelA-cKO (Fig. 5e, Supplementary Fig. 12a). The rate of p- region (from −1970 to +20 bp relative to the transcriptional start VEGFR2-positive cells increased 10 min after rhGREM1 treat- site) of human GREM1, and performed a luciferase assay by co- ment, while that of VEGFR2-positive cells was unaltered transfection with expression vectors for RelA, HIF-1α, c-Jun,or (Supplementary Fig. 12b). mRNA levels of VEGF family members Nrf2. Among the four transcription factors, only RelA markedly were not upregulated by rhGREM1 treatment in mouse primary enhanced activity of the GREM1 proximal promoter (Fig. 8b). chondrocytes (Supplementary Fig. 13). This region contains one NF-κB motif (−279 to −270 bp) and three sequences similar to the NF-κB motif (Fig. 8c). Deletion analysis by a series of 5ʹ-deletion constructs identiﬁed the core responsive element to RelA to be between −399 and −202 bp Gremlin-1 antagonizes the anabolic effects of BMPs. Gremlin-1 (Fig. 8c). Site-directed mutagenesis of the NF-κB motif in the treatment increased the expression of catabolic enzymes Mmp13 reporter construct containing the region between −399 to +20 bp and Adamts5, while decreasing the expression of cartilage matrix inhibited the enhanced activity by RelA, indicating it is the core genes Col2a1 and Acan (Fig. 2a). Among these alterations, inhi- responsive element of RelA (Fig. 8c). Furthermore, gremlin-1 bition of NF-κB signalling almost completely suppressed the expression was signiﬁcantly increased by adenoviral overexpres- increase of catabolic enzymes by gremlin-1 (Fig. 5c–e), but it sion of RelA in mouse primary chondrocytes (Fig. 8d). did not affect observed decreases of anabolic factors (Fig. 5e). Previously, we generated adult chondrocyte-speciﬁc RelA knock- We next examined antagonistic effects between gremlin-1 and out mice . In hetero-knockout and homo-knockout cartilage, BMPs. In mouse primary chondrocytes, rhBMP-2, rhBMP-4, gremlin-1 expression was decreased compared with respective and rhBMP-7 enhanced the expression of anabolic markers control cartilage (Supplementary Fig. 14). These data indicate that (Fig. 6). Anabolic effects elicited by rhBMPs were diminished by excessive mechanical loading enhances Gremlin-1 transcription rhGREM1 treatment (Fig. 6). through the Rac1–ROS–NF-κB pathway (Fig. 8e). Gremlin-1 is induced by mechanical loading through Rac1. Discussion Finally, we searched for an upstream pathway of gremlin-1 The present study identiﬁed gremlin-1 as a key factor mediating induction by mechanical loading using a screen of selective cartilage degeneration by excessive mechanical loading. Gremlin- inhibitors of mechanotransduction-related molecules, such as 1 activates NF-κB signalling, which results in induction of focal adhesion kinase (FAK), Rho-associated coiled-coil forming catabolic enzymes such as Mmp13 and Adamts5. The kinase (ROCK), and ras-related C3 botulinum toxin substrate 1 Rac1–ROS–NF-κB axis plays an essential role in gremlin-1 (Rac1). NSC23766, a speciﬁc inhibitor of the binding and acti- induction by excessive mechanical loading. Genetic deletion or vation of Rac guanosine triphosphate-binding protein (GTPase), neutralization of gremlin-1 by intra-articular injection of an decreased gremlin-1 induction by cyclic tensile strain loading, antibody suppressed OA development in mice. although a FAK inhibitor (PF-573228) and ROCK inhibitor Gremlin-1 is a well-known secreted BMP antagonist that was 11–13 (Y-27632) had no effect (Fig. 7a). EHT1864, another inhibitor formerly identiﬁed as a regulator of limb formation . of Rac family GTPases, also signiﬁcantly decreased gremlin-1 Gremlin-1 exerts potent inhibitory action via binding to and induction by loading, similar to NSC23766 (Fig. 7b). An active forming heterodimers with BMP-2, BMP-4, and BMP-7 . Rac1 pull-down assay conﬁrmed Rac1 activation by mechanical Binding to selective BMPs prevents ligand–receptor interactions loading (Fig. 7c). Moreover, gremlin-1 expression was sig- and subsequent downstream signalling . Indeed, gremlin-1 reg- niﬁcantly increased by adenoviral or lentiviral overexpression of ulates development of limb, lung, kidney and retina through 28–30 Rac1 in mouse primary chondrocytes or ATDC5 cells, respec- BMP inhibition . In addition to its inhibitory effect on tively (Fig. 7d, e). BMP, gremlin-1 exerts direct effects on cell function via 6 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE a Safranin O b TUNEL & DAPI OARSI score n = 9 n = 9 c S32/36 Safranin O IκBα p-IκBα Hif2a Grem1 Mmp13 Adamts5 % % % % % % 60 60 60 60 ## Cntl ## 30 30 30 30 cKO 30 0 0 0 0 0 0 d e Safranin O TUNEL & DAPI OARSI score n = 8 n = 8 f S32/36 Safranin O Mmp13 Adamts5 IκBα p-IκBα Hif2a Grem1 % % % % % 60 60 60 # * # Cntl 40 40 40 ## 30 30 30 cKO 20 20 20 0 0 0 0 0 BMP-independent mechanisms. For example, VEGFR2 was and cancer—some of which may be mediated by this axis . identiﬁed as a novel receptor for gremlin-1 . The gremlin- Considering that the induction of catabolic factors by gremlin-1 1–VEGFR2 axis exerts a proinﬂammatory effect , which is was diminished by inhibition of VEGFR2 or NF-κB in the present mediated by activation of NF-κB signalling . In addition to its data, the axis is also responsible for cartilage degeneration by roles in the development of various organs, gremlin-1 has been gremlin-1. However, we cannot exclude the possibility that carrier implicated in many diseases and pathogenic mechanisms, such as proteins in rhGREM1 solution may exert some catabolic effects in heart, lung and liver ﬁbrosis, as well as osteogenesis, angiogenesis intra-articular injection experiments. NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 7 cKO Cntl cKO Cntl cKO Cntl cKO Cntl Cntl Cntl cKO cKO cKO Cntl cKO Cntl Cntl Cntl cKO cKO ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ﬂ/ﬂ Fig. 4 Regulation of osteoarthritis development by gremlin-1. a Safranin O staining and OARSI scores of mouse knee joints of Grem1 (Cntl) and Col2a1- ERT2 ﬂ/ﬂ Cre ;Grem1 (cKO) mice 8 weeks after surgery. Tamoxifen induction was performed at 7 weeks. Both experimental groups consist of n = 9 biologically independent animals. Inset boxes indicate regions of immunoﬂuorescence in (c). Scale bars, 100 µm. b TUNEL staining and rate of TUNEL-positive cells in mouse knee joints of Cntl and cKO mice 8 weeks after surgery. Nuclei were stained with DAPI (blue). Scale bars, 100 µm. n = 5 biologically independent experiments. c Safranin O staining and immunoﬂuorescence of Mmp13, Adamts5, IκBα, phosphorylated IκBα (dual Ser32/36), HIF-2α, and gremlin-1 proteins in mouse knee joints of Cntl and cKO mice 8 weeks after surgery. Scale bars, 50 µm. The percentage of positive cells in the immunoﬂuorescence is shown below. n = 5 biologically independent experiments. d Safranin O staining and OARSI scores of mouse knee joints of Cntl and cKO mice at 18 months of age. Tamoxifen induction was performed for 5 days at 8 weeks, 6 months, and 12 months. Both experimental groups consist of n = 8 biologically independent animals. Inset boxes indicate regions of immunoﬂuorescence in (f). Scale bars, 100 µm. e TUNEL staining and rate of TUNEL-positive cells in mouse knee joints of Cntl and cKO mice at 18 months of age. Nuclei were stained with DAPI (blue). Scale bars, 100 µm. n = 5 biologically independent experiments. (f) Safranin O staining and immunoﬂuorescence of Mmp13, Adamts5, IκBα, phosphorylated IκBα (dual Ser32/36), HIF-2α, and gremlin-1 proteins in mouse knee joints of Cntl and cKO mice at 18 months of age. Scale bars, 50 µm. The percentage of positive cells in the immunoﬂuorescence is # ## shown below. n= 5 biologically independent experiments. All data are expressed as mean ± SD. *P < 0.05, **P < 0.01, P < 0.005, P < 0.001 versus Cntl (Student’s unpaired two-tailed t-test) a e * Mmp13 Vehicle ## ## 40 rhGREM1 Adamts5 ## ## 300 * Sox9 ## rhGREM1 0 0.1 1 10 μg per ml # # 0.5 Col2a1 2 ## rhGREM1 010 10 BMS-345541 005 μg per ml d ## ## Acan ## 0.5 0 0 Rela Rela Rela Rela Rela Rela Rela Rela –Cntl –cKO –Cntl –cKO –Cntl –cKO –Cntl –cKO Vehicle rhGREM1 Vehicle rhGREM1 Fig. 5 NF-κB signalling mediates the catabolic effects of gremlin-1. a Luciferase assay for screening of downstream signalling pathways of gremlin-1 using reporter vectors containing a response element for each pathway or transcription factor. *P < 0.05 versus each vehicle control (one-way ANOVA test). b Activities of luciferase reporter vectors containing an NF-κB motif with rhGREM1 treatment. *P < 0.05 versus 0 (vehicle) (one-way ANOVA test). c Amount of proteoglycan released into culture medium from wild-type mouse femoral heads cultured with or without rhGREM1 (10 µg/mL) and IKK # ﬂ/ﬂ inhibitor BMS-345541 (5 µM) for 3 days. P < 0.005 (one-way ANOVA test). d Amount of proteoglycan released into culture medium from RelA ﬂ/ﬂ ## (Rela-Cntl) and Col2a1-Cre;RelA (Rela-cKO) mice femoral heads cultured with or without 10 µg/mL rhGREM1. P < 0.001 (one-way ANOVA test). ## e mRNA levels of marker genes in RelA-Cntl and RelA-cKO femoral heads cultured with or without rhGREM1. P < 0.001 (one-way ANOVA test). All data are expressed as mean ± SD of biologically independent three samples per group In contrast to these catabolic effects, downregulation of carti- formation, and have been extensively used in cartilage regen- lage matrix gene expression by gremlin-1 was not dependent on erative research . Exogenous BMP-2 or BMP-4 enhance chon- the NF-κB pathway. Moreover, the present in vitro data revealed drocyte differentiation of stem cells and cartilage matrix 34–36 that gremlin-1 suppressed induction of Sox9, Col2a1, and Acan by production by chondrocytes . BMP-7 also exhibits chondro- rhBMPs (Fig. 6). BMPs play essential roles in bone and cartilage genic effects and the ability to repair cartilage and suppress 8 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications RLA μg per ml μg per ml RLA Mock WNT TGFβ AP1 HRE ATF6 p53 SRE NFκB CRE Fold increase NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE Sox9 Col2a1 Acan ## ## # ## ## ## ## ## ## ** * 1.5 0.5 0.5 0 0 0 ng per ml rhBMP2 010 010 010 100 0 10 100 100 0 10 100 100 0 10 100 μg per ml rhGREM1 00 0 1 1 1 00 0 1 1 1 00 0 1 1 1 Sox9 Col2a1 Acan ## ## ## ## ## # * * 1.5 1.5 1 1 0.5 0.5 0 0 0 rhBMP4 0 10 100 0 10 100 0 10 100 0 10 100 0 10100 0 10100 ng per ml rhGREM1 00 0 1 1 1 00 0 1 1 1 00 0 1 1 1 μg per ml Sox9 Col2a1 Acan ## # ## ## ## ## ## ** ## # 3 1.5 0.5 1 0 0 0 rhBMP7 0 10 100 0 10 100 0 10 100 0 10 100 0 10100 0 10100 ng per ml rhGREM1 00 0 1 1 1 00 0 1 1 1 00 0 1 1 1 μg per ml Fig. 6 Gremlin-1 antagonizes the anabolic effects of BMPs. mRNA levels of Sox9, Col2a1 and Acan in mouse primary chondrocytes treated with rhBMP2, 4, # ## or 7, and rhGREM1 for 24 h. *P < 0.05, **P < 0.01, P < 0.005, P < 0.001 versus vehicle (one-way ANOVA test). All data are expressed as means ± SD of biologically independent three samples per group 37,38 cartilage degeneration . Meanwhile, gremlin-1 consistently GTPases actively function with their downstream target proteins demonstrated a higher binding afﬁnity for BMP-2 > BMP-4 > when bound to GTP . Among various types of stimulants, BMP-7 . Although we could not determine speciﬁc targets of mechanical loading activates Rac1 through effects on the actin 26,44 gremlin-1 among BMP family members, gremlin-1 may affect cytoskeleton . Rac1 binds the complex of nicotinamide ade- articular cartilage homeostasis through downregulation of ana- nine dinucleotide phosphate (NADPH) oxidase enzymes to bolic genes by antagonizing BMPs. enhance ROS production in a variety of cell types in response to 26,45 BMPs are also involved in hypertrophic differentiation during mechanical loading . Activation of the Rac1-ROS pathway endochondral ossiﬁcation. BMP-2, which is highly expressed in enhances transcription of several molecules, such as NF-κB, HIF- the hypertrophic zone, promotes hypertrophic differentiation of 1α, c-Jun, and Nrf2 . Amongst upregulated transcription factors, chondrocytes in the proliferative zone . BMP-4 expression is RelA strongly induced gremlin-1 expression (Fig. 8b). Luciferase very low in the growth plate, while BMP-7 is more highly assays revealed functional NF-κB consensus motifs within the expressed in the proliferative zone than in the hypertrophic proximal 5ʹ-ﬂanking region of the GREM1 gene (Fig. 8c). In turn, zone . Hypertrophic differentiation is involved in OA develop- gremlin-1 exerts catabolic effects through the NF-κB pathway. In ment; however, it is not well understood how each endogenous fact, Mmp13 mRNA levels are slightly and transiently increased BMP produced from articular chondrocytes or neighbouring one hour after stress loading, and peak by 12–24 h after loading cells regulates their hypertrophic differentiation. In the present (Fig. 1a). Early and late increases of Mmp13 mRNA may result study, Col10a1 expression was decreased by cyclic tensile strain from activation of NF-κB signalling by Rac1-ROS and gremlin-1, loading (Supplementary Table 2), but not altered by rhGREM1 respectively (Fig. 8e). Thus, iterated intensive mechanical loading treatment in mouse primary chondrocytes (Supplementary causes gremlin-1 protein to excessively accumulate within the Fig. 6). Although we cannot determine the effect size, hyper- deep layer of articular cartilage, where it may function to amplify trophic differentiation seems to be less important in the regula- mechanical stress-induced NF-κB activation. However, it is dif- tion of OA development by gremlin-1, compared with the ﬁcult to discern the entire hierarchical relationship between these catabolic effects enacted through the gremlin-1–NF-κB axis. molecules in one experimental system. The hypothesis shown in The Rho family of small GTPases, such as Rac, Cdc42, and Fig. 8e should be examined by following studies in the future. 42,43 RhoG, are master regulators of many cellular activities . In conclusion, we found that induction of gremlin-1 by Conformationally regulated by the binding of GTP and GDP, excessive mechanical loading occurs via the Rac1-ROS-RelA/p65 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 9 Fold increase Fold increase Fold increase ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 a Grem1 b Grem1 1.5 2 # ## ## 0.5 Stress –+ + + + Stress–+ ++ + + NSC23766 0 0 10 100 0 0 μM Vehicle EHT1864 0 0 0 0 10 100 μM c d e GTP-Rac1 Grem1 Grem1 * * per total Rac1 * 2 30 kDa Total Rac1 20 kDa 0 0 30 kDa RAC1 RAC1 GTP- 20 kDa Rac1 Stress –+ Stress –+ 0 0 AxGFP AxRAC1 LxGFP LxRAC1 Fig. 7 Gremlin-1 induction by mechanical stress loading occurs through Rac1 activation. a Gremlin-1 mRNA levels in mouse primary chondrocytes treated with 10 µM inhibitors of FAK (PF-573228), ROCK (Y-27632), or RAC1 (NSC23766) 24 h after cyclic tensile strain loading. n = 3 biologically independent samples. *P < 0.05, P < 0.005 versus stress+, vehicle (one-way ANOVA test). b Gremlin-1 mRNA levels in mouse primary chondrocytes treated with Rac1 # ## inhibitors NSC23766 or EHT1864 at 24 h after cyclic tensile strain loading. n = 3 biologically independent samples. *P < 0.05, P < 0.005, P < 0.001 versus stress + , vehicle (one-way ANOVA test). c Rac1 pull-down activation assay using mouse primary chondrocytes with or without cyclic tensile strain loading. Quantiﬁcation of densitometry data are shown below, and ratios of positive Rac1 per total Rac1 are shown as fold-increase in the right graph. n = 5 biologically independent samples. *P < 0.05 versus stress– (Student’s unpaired two-tailed t-test). d Gremlin-1 mRNA levels in mouse primary chondrocytes transduced with an adenoviral vector containing Rac1 or GFP. n = 3 biologically independent samples. *P < 0.05 versus GFP (Student’s unpaired two-tailed t-test). e Gremlin-1 mRNA levels in ATDC5 cells lentivirally overexpressing Rac1 or GFP. n= 3 biologically independent samples. *P < 0.05 versus GFP (Student’s unpaired two-tailed t-test). All data are expressed as means ± SD axis, which accelerates OA development mainly through the M. Schmid (Technical University of Munich), respectively. To generate Col2a1-Cre; ﬂ/ﬂ ERT ﬂ/ﬂ ﬂ/ﬂ RelA and Col2a1-Cre ;RelA mice, RelA mice were mated with Col2a1- NF-κB pathway. Our ﬁndings contribute to understanding of ERT ﬂ/+ ERT Cre or Col2a1-Cre mice to obtain Col2a1-Cre;RelA and Col2a1-Cre ; the pathophysiology of various diseases in which mechanical ﬂ/+ ﬂ/ﬂ RelA mice, respectively, which were then mated with RelA mice. Col2a1- loading, gremlin-1, and NF-κB are involved, including OA. ERT2 ﬂ/ﬂ 50,51 Cre mice and Grem1 mice were generated as previously described .To ERT2 ﬂ/ﬂ ﬂ/ﬂ generate Col2a1-Cre ;Grem1 mice, Grem1 mice were mated with Col2a1- ERT2 ERT2 ﬂ/+ Cre mice to obtain Col2a1-Cre ; Grem1 mice, which were then mated Methods ﬂ/ﬂ with Grem1 mice. To conﬁrm knockout efﬁciency, we directly isolated mRNA ﬂ/ﬂ ERT2 ﬂ/ﬂ Cell cultures. Mouse chondrogenic ATDC5 cells (RIKEN Cell Bank) were grown from the articular cartilage of 16-week-old Grem1 and Col2a1-Cre ;Grem1 and maintained in Dulbecco’s Modiﬁed Eagle’s Media with F-12 supplement animals that had received tamoxifen administration at 7 weeks of age. For animal (DMEM/F-12, 1:1) with 5% fetal bovine serum (FBS). Primary articular chon- studies, no randomization was used. drocytes were isolated from 6-day-old C57BL/6J mice according to a standard protocol using collagenase D . Primary chondrocytes were cultured in DMEM Human samples. We obtained samples of human articular cartilage undergoing with 10% FBS. The medium was changed every 3 days. For most experiments, total knee arthroplasty. Written informed consent was obtained from all indivi- primary cells were transferred to serum-free DMEM for 24 h before being exposed duals, and approval was provided by the ethics committee of the University of to stimuli. In some experiments, cells were pretreated with recombinant human Tokyo. We have complied with all relevant ethical regulations. gremlin-1 (rhGREM1, #120-42, PeproTech for Figs. 2, 3, 5, 6, Supplementary Figs. 4, 6–8, 12, and 13; 5190-GR, R&D Systems for Supplementary Fig. 5), recombinant human BMP-2 (rhBMP2, #120-02, PeproTech), recombinant human Cyclic tensile strain loading of mouse primary chondrocytes. Mouse primary BMP-4 (rhBMP4, #120-05, PeproTech), recombinant human BMP-7 (rhBMP7, chondrocytes were seeded into silicon stretch chambers coated with ﬁbronectin at a #120-03, PeproTech), IKK inhibitor BMS-345541 (B9935, Sigma), selective 5 density of 1 × 10 cells/chamber; each chamber had a culture surface of 2 × 2 cm. VEGFR2 inhibitor SU5416 (ab145056, Abcam), FAK inhibitor PF-573228 (S2013, After 48 h, cyclic tensile strain (0.5 Hz, 10% elongation) was applied for 30 min Selleck Chemicals), ROCK inhibitor Y-27632 (257-00511, WAKO), Rac1 inhibitor using an STB-140 mechanical stretch system (STREX, Osaka, Japan) in a CO NSC23766 (SML0952, Sigma) and/or EHT1864 (S7482, Selleck Chemicals). For incubator . Control cells were seeded onto the same chambers and cultured gene silencing experiments, siRNA against mouse gremlin-1 (#1320001, Thermo- without cyclic tensile strain. Fisher Scientiﬁc) or negative control (#12935300, Thermo Fisher Scientiﬁc) were transfected into mouse primary chondrocytes using Lipofectamine 2000 reagent Cyclic hydrostatic pressure loading of mouse femoral head explants. The (Thermo Fisher Scientiﬁc) according to the manufacturer’s protocol. Cells were harvested 24 h after transfection. cyclic hydrostatic pressure loading system was developed by STREX. Femoral heads obtained from 3-week-old mice were put into the chamber, which consisted of gas permeable membranes with a metallic frame (Supplementary Fig. 1a); each Animals. All animal experiments were authorized by the Animal Care and Use chamber was ﬁlled with DMEM/F-12 (1:1) containing 5% FBS and set into a Committee of The University of Tokyo. We have complied with all relevant ethical cylindrical container (Supplementary Fig. 1a–c). Cyclic hydrostatic compression regulations. In each experiment, we compared the genotypes of littermates main- pressure was applied by the pumping machine connected to the cylindrical con- tained in a C57BL/6J background. Col2a1-Cre were purchased from Jackson tainers (Supplementary Fig. 1d, e). We employed 5 and 20 MPa for normal and ERT 48 ﬂ/ﬂ 49 Laboratory. Col2a1-Cre mice and RelA mice were generously provided by excessive stress loading, respectively, because maximum pressure to human knee 52–54 Professor Fanxin Long (Washington University, St. Louis) and Professor Roland cartilage ranges from 3 to 9 MPa during a normal gait . We applied 5 or 10 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications PF-573228 Y-27632 NSC23766 Fold increase Fold increase Fold increase Fold increase Fold increase NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE Stress (–) 0 min 30 min 1 h 2 h 3 h bc ## –1970 –740 –655 –399 –202 –92 +20 WT: ggggatttcc GFP mt: AATgatAtAc RELA 10 * 0 0 GFP RELA HIF1A Jun Nrf2 –1970 –740 –655 –399 –202 –92 –399mt de Grem1 Excessive GREM1 Chondrocyte mechanical loading VEGFR2 RAC1 NF-kB GREM1 ROS GREM1 Mmp13 NF-kB HIF2A GREM1 GREM1 MMPs GREM1 Cartilage BMPs matrix Chondrocyte Osteoarthritis GFP RELA Fig. 8 The ROS- NF-κB pathway enhances transcriptional induction of gremlin-1. a Fluorescence imaging time-course of ROS production in mouse primary chondrocytes after cyclic tensile strain loading. Nuclei were stained with DAPI (blue). Scale bars, 20 µm. b Luciferase assay using ATDC5 cells ## co-transfected with human GREM1 promoter (from −1970 to +20 bp relative to the transcriptional start site) and each expression vector. P < 0.001 versus GFP (one-way ANOVA test). c 5ʹ-deletion and mutation analyses of the luciferase assay. N: NF-κB motif, −399mt: reporter construct from −399 to +20 bp in which NF-κB motif is mutated. *P < 0.05, P < 0.005 (Student’s unpaired two-tailed t-test). d mRNA levels of Gremlin-1 and Mmp13 in mouse primary chondrocytes transfected with RelA or GFP.*P < 0.05, P < 0.005 versus GFP (Student’s unpaired two-tailed t-test). e Schematic diagram representing molecular pathways in which excessive mechanical loading induces osteoarthritis development through gremlin-1. All data are expressed as mean ± SD of biologically independent three samples per group 20 MPa hydrostatic pressure to mouse femoral heads for 30 min at a frequency of Cruz Biotechnology), and/or HIF-2α (1:200; NB100-122, Novus Biologicals). For 0.1 Hz. Control femoral heads were put into in the same chambers and cultured immunoﬂuorescence, a CSA II Biotin-Free Catalyzed Ampliﬁcation System (Agi- without cyclic hydrostatic pressure. lent Technologies) and Hoechst 33258 (Agilent Technologies) counterstain were used. TdT-mediated dUTP nick end labelling (TUNEL) staining was performed with an In Situ Cell Death Detection Kit (Roche) according to the manufacturer’s qRT-PCR. Total RNA was puriﬁed with an RNeasy Mini Kit (Qiagen). One instructions. Histological analyses were performed at least three times using 2–3 microgram of total RNA was reverse transcribed using ReverTraAce qPCR RT mice per group or genotype for conﬁrmation of results. Images were visualized Master Mix with gDNA Remover (Toyobo). Each PCR reaction contained 1 × under a ﬂuorescence microscope (BZ-X710, Keyence, Osaka, Japan). Counts of THUNDERBIRD SYBR qPCR Mix (Toyobo), 0.3 mM speciﬁc primers, and 20 ng immunoﬂuorescence-positive cells were calculated in ﬁve independent squares of cDNA. Copy number was normalized to rodent total RNA (Thermo-Fisher (100 × 100 µm) of articular cartilage harvested from the mouse medial tibial pla- Scientiﬁc), with rodent glyceraldehyde-3-phosphate dehydrogenase used as an teau, using BZ analyser software (Keyence). The same parameters were used for all internal control. All reactions were run in triplicate. Primer sequences are shown in acquisitions, and representative pictures are shown in the ﬁgures. Supplementary Table 5. Immunocytochemistry. Cells were ﬁxed with 4% paraformaldehyde buffered with Microarray analysis. Total RNA was isolated from mouse primary chondrocytes PBS (pH 7.4). After incubation for 10 min in PBS containing 0.1% Triton X-100 before and 24 h after cyclic tensile strain loading using an RNeasy Mini Kit (Sigma Aldrich), cells were rinsed in PBS, and treated with PBS containing 1% (Qiagen). Microarray experiments were performed using SurePrint G3 Mouse bovine serum albumin (BSA; Sigma-Aldrich). Samples were stained using primary Gene Expression 8 × 60 K (Agilent Technologies). Arrays were then scanned with antibodies against Grem1 (1:500, LS-C125371, LSBio for Fig. 1c; 1:250, AF956, an Agilent DNA Microarray Scanner (G2565CA), and acquired array images were R&D Systems for Supplementary Fig. 2a, b), VEGFR2 (1:500, 55B11, Cell Signaling analysed by Agilent Feature Extraction software (version 10.10.1.1). Technology for Supplementary Fig. 12b) and p-VEGFR2 (1:500, Tyr951, #2471, Cell Signaling Technology for Supplementary Fig. 12b) for 24 h in PBS containing Histological analyses. Tissue samples were ﬁxed in 4% paraformaldehyde buf- 1% BSA at 4 °C, followed by AlexaFluor 488-conjugated mouse and rabbit fered with phosphate-buffered saline (PBS, pH 7.4) at 4 °C for 1 day. Specimens secondary antibodies (1:500, Invitrogen) for 1 h. Slides were then treated with were decalciﬁed with 10% EDTA (pH 7.4) at 4 °C for 2 weeks, embedded in Vectashield Mounting Medium with DAPI (H-1200, Vector Laboratories). Images parafﬁn, and 4-µm thick sagittal sections were cut from specimens. Safranin O were visualized under a ﬂuorescence microscope (BZ-X710, Keyence, Osaka, staining was performed according to standard protocols. For immunohistochem- Japan). Counts of immunoﬂuorescence- positive cells were calculated in ﬁve istry, sections were incubated with antibodies against gremlin-1 (1:250; LS- independent squares (100 × 100 µm) using BZ analyser software (Keyence). C125371, LSbio for Figs. 1e, f and 4c, f, Supplementary Figs. 3 and 14: 1:250; AF956, R&D Systems for Supplementary Figs. 2c, 2d, and 11), Mmp13 (1:200; MAB13426, Chemicon), Adamts5 (1:500; H-200, Santa Cruz Biotechnology), IκBα OA experiment. For surgical model, tamoxifen (Sigma; 100 µg per g of body S32/36 ERT ﬂ/ﬂ (1:100; sc-371, Santa Cruz Biotechnology), p-IκBα (1:100; sc-101713, Santa weight) was intraperitoneally injected into 7-week-old Col2a1-Cre ;Grem1 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 11 Fold increase Fold increase Fold increase ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ﬂ/ﬂ and Grem1 mice daily for 5 days. A surgical procedure was then performed statistical analysis was performed by a two-way repeated measures for variance to establish an experimental OA model in 8-week-old male mice . Under (ANOVA) test for Fig. 1a and a one-way ANOVA followed by Tukey–Kramer general anaesthesia, resection of the medial collateral ligament and medial honest signiﬁcant difference comparison. Sample sizes were chosen based on prior meniscus was performed using a surgical microscope. Mice were analysed 8 weeks literature using similar methods and reporting moderate effect size. P values < 0.05 after surgery. We also analysed OA development with aging using 18-month-old were considered signiﬁcant. mice bred under physiological conditions. For the aging experiments, tamoxifen induction was performed for 5 days at 8 weeks, 6 months, and 12 months. All Reporting Summary. Further information on experimental design is available in mice were maintained under the same conditions (three mice per cage). OA the Nature Research Reporting Summary linked to this article. severity was quantiﬁed by the Osteoarthritis Research Society International (OARSI) system , which was assessed by two observers blinded to the experi- mental groups. Data availability The raw microarray data are deposited in the Gene Expression Omnibus (www.ncbi.nlm. Intra-articular administration of rhGREM1 or gremlin-1–neutralizing antibody. nih.gov/geo/) under accession no. GSE103159. All other data supporting the ﬁndings of this study are available within the article and the Supplementary Information ﬁle, or are We performed intra-articular administration of rhGREM1 or gremlin- 1–neutralizing antibody to C57BL/6J mice twice a week for 8 weeks after OA available from the corresponding author on reasonable request. surgery, or normal mice without surgery. For each administration, we injected 10 µL of 10 µg/mL rhGREM1 solution diluted in PBS, 10 µL of 10 µg/mL gremlin- Received: 8 January 2018 Accepted: 14 March 2019 1–neutralizing antibody (LS-C125371, LSbio) solution diluted in PBS, or 10 µL of PBS as a vehicle control. Proteoglycan release assay. Proteoglycan release from mouse femoral head explants was assessed according to the standard protocol . Brieﬂy, femoral heads ﬂ/ﬂ ﬂ/ﬂ were harvested from 3-week-old WT or RelA and Col2a1-Cre;RelA mice, and References cultured for 3 days with or without rhGREM1 (10 µg/mL) in DMEM containing 1. Glasson, S. S. et al. Deletion of active ADAMTS5 prevents cartilage 1% penicillin/streptomycin. Proteoglycan content within the medium was mea- degradation in a murine model of osteoarthritis. Nature 434, 644–648 sured as sulfated glycosaminoglycan by a colorimetric assay using dimethyl- (2005). methylene blue. 2. Stanton, H. et al. ADAMTS5 is the major aggrecanase in mouse cartilage in vivo and in vitro. Nature 434, 648–652 (2005). Luciferase assay. Reporter constructs containing the human GREM1 proximal 3. Zhu, M. et al. Inhibition of beta-catenin signaling in articular chondrocytes promoter region were prepared by PCR ampliﬁcation of a human genomic DNA results in articular cartilage destruction. Arthritis Rheum. 58, 2053–2064 template, which were cloned into the pGL4.10[luc2] vector (Promega). Deletion (2008). and mutation constructs were created by PCR. Transfection of ATDC5 cells was 4. Echtermeyer, F. et al. Syndecan-4 regulates ADAMTS-5 activation and performed in triplicate in 48-well plates using Lipofectamine 2000 transfection cartilage breakdown in osteoarthritis. Nat. Med. 15, 1072–1076 (2009). reagent (Thermo-Fisher). The reaction mixture consisted of Lipofectamine 2000 5. Lin, A. C. et al. Modulating hedgehog signaling can attenuate the severity with 400 ng of plasmid DNA, 200 ng of pGL4 reporter vector, 200 ng of effector of osteoarthritis. Nat. Med. 15, 1421–1425 (2009). vector, and 4 ng of pRL-TK internal control vector (Promega) per well. For 6. Little, C. B. et al. Matrix metalloproteinase 13-deﬁcient mice are resistant to screening of downstream gremlin-1 signalling, we purchased pGL4.49[luc2P/TCF- osteoarthritic cartilage erosion but not chondrocyte hypertrophy or LEF RE/Hygro], pGL4.48[luc2P/SBE/Hygro], pGL4.44[luc2P/AP1 RE/Hygro], osteophyte development. Arthritis Rheum. 60, 3723–3733 (2009). pGL4.42 [luc2P/HRE/Hygro], pGL4.39[luc2P/ATF6 RE/Hygro], pGL4.38[luc2P/ 7. Zhu, M. et al. Activation of beta-catenin signaling in articular chondrocytes p53 RE/Hygro], tpGL4.33[luc2P/SRE/Hygro], pGL4.32[luc2P/NF-κB-RE/Hygro], leads to osteoarthritis-like phenotype in adult beta-catenin conditional and pGL4.29[luc2P/CRE/Hygro] vectors (Promega). Transfections were performed activation mice. J. Bone Miner. Res. 24,12–21 (2009). as described above except with PBS and rhGREM1 (1 µg/mL) in lieu of effector 8. Saito, T. et al. Transcriptional regulation of endochondral ossiﬁcation by vectors. To evaluate the dose-response of NF-κB activity, we used 0.1, 1, and 10 µg/ HIF-2alpha during skeletal growth and osteoarthritis development. Nat. Med. mL of rhGREM1. Luciferase assays were performed with a PicaGene Dual Sea- 16, 678–686 (2010). Pansy Luminescence Kit (Toyo Ink) using a GloMax 96 Microplate Luminometer (Promega). Data are shown as the ratio of ﬁreﬂy activity to Renilla activity. 9. Yang, S. et al. Hypoxia-inducible factor-2alpha is a catabolic regulator of osteoarthritic cartilage destruction. Nat. Med. 16, 687–693 (2010). 10. Wang, M. et al. MMP13 is a critical target gene during the progression of Rac1 activation assay. To evaluate Rac1 activity, a Rac1 Pull-down Activation osteoarthritis. Arthritis Res. Ther. 15, R5 (2013). Assay Biochem Kit (Cytoskeleton) was employed. Mouse primary chondrocytes 11. Merino, R. et al. The BMP antagonist Gremlin regulates outgrowth, were plated at a density of 1 × 10 cells in 2 × 2 cm stretch chambers. After cyclic chondrogenesis and programmed cell death in the developing limb. tensile strain (0.5 Hz, 10% elongation) for 30 min, proteins were isolated from cells Development 126, 5515–5522 (1999). and 500 µg was used in a pull-down assay with p21-activated kinase–Rac1 p21 12. Khokha, M. K., Hsu, D., Brunet, L. J., Dionne, M. S. & Harland, R. M. Gremlin binding domain beads. The resulting pull-down product was then immunoblotted is the BMP antagonist required for maintenance of Shh and Fgf signals with Rac1 antibody according to the manufacturer’s instructions. His-tagged Rac1 during limb patterning. Nat. Genet. 34, 303–307 (2003). protein in the kit was used for control. The membranes were incubated with a 13. Wordinger, R. J., Zode, G. & Clark, A. F. Focus on molecules: gremlin. horseradish peroxidase-conjugated antibody (Promega), and visualized with ECL Exp. Eye Res. 87,78–79 (2008). prime (GE Healthcare) and an AE-6981 Light Capture II (ATTO, Tokyo, Japan). 14. Bonizzi, G. & Karin, M. The two NF-kappaB activation pathways and their Control cells were seeded onto identical chambers and cultured without cyclic role in innate and adaptive immunity. Trends Immunol. 25, 280–288 (2004). tensile strain. Original images of the assay were shown in Supplementary Fig. 15. 15. Li, Q., Withoff, S. & Verma, I. M. Inﬂammation-associated cancer: NF-kappaB is the lynchpin. Trends Immunol. 26, 318–325 (2005). Construction of expression vectors. Full-length human or mouse cDNA 16. Meffert, M. K. & Baltimore, D. Physiological functions for brain NF-kappaB. sequences of transcription factors were PCR-ampliﬁed and cloned into pCMV-HA Trends Neurosci. 28,37–43 (2005). or pShuttle vector (Clontech). We generated adenovirus vectors using an Adeno-X 17. Neumann, M. & Naumann, M. Beyond IkappaBs: alternative regulation of expression system (Clontech) and veriﬁed all vectors by DNA sequencing. NF-kappaB activity. FASEB J. 21, 2642–2654 (2007). 18. Hayden, M. S. & Ghosh, S. Signaling to NF-kappaB. Genes Dev. 18, 2195–2224 Fluorescence detection of ROS. Intracellular ROS accumulation was evaluated by (2004). ﬂuorescent microscopy using a Total ROS Detection Kit (ENZ-51010, Enzo Life 19. Marcu, K. B., Otero, M., Olivotto, E., Borzi, R. M. & Goldring, M. B. NF- Sciences). Mouse primary chondrocytes were plated in 2 × 2 cm stretch chambers kappaB signaling: multiple angles to target OA. Curr. Drug Targets 11, and cyclic tensile stress (0.5 Hz, 10% elongation) was applied for 30 min. Cells 599–613 (2010). were stained with 1 µM of ROS-sensitive ﬂuorescent dye at 37 °C for 30 min, and 20. Kobayashi, H. et al. Biphasic regulation of chondrocytes by Rela through then washed once with 1× wash buffer. After washing, cells were immediately induction of anti-apoptotic and catabolic target genes. Nat. Commun. 7, 13336 overlaid with a coverslip and observed under a BZ-X710 ﬂuorescence microscope (2016). (Keyence). Fluorescence signals were subsequently detected with excitation at 21. Tetsunaga, T. et al. Regulation of mechanical stress-induced MMP-13 and 490 nm and emission at 525 nm for detection of ROS. ADAMTS-5 expression by RUNX-2 transcriptional factor in SW1353 chondrocyte-like cells. Osteoarthr. Cartil. 19, 222–232 (2011). Statistical analyses. JMP Pro 13 was used for statistical analysis. Values are 22. Nam, J., Aguda, B. D., Rath, B. & Agarwal, S. Biomechanical thresholds expressed as mean ± standard deviation. For comparisons between two groups, regulate inﬂammation through the NF-kappaB pathway: experiments and Student’s unpaired two-tailed t-test was applied. For multiple comparisons, modeling. PLoS ONE 4, e5262 (2009). 12 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09491-5 ARTICLE 23. Kamekura, S. et al. Osteoarthritis development in novel experimental mouse 49. Algul, H. et al. Pancreas-speciﬁc RelA/p65 truncation increases susceptibility models induced by knee joint instability. Osteoarthr. Cartil. 13, 632–641 of acini to inﬂammation-associated cell death following cerulein pancreatitis. (2005). J. Clin. Invest. 117, 1490–1501 (2007). 24. Mitola, S. et al. Gremlin is a novel agonist of the major proangiogenic receptor 50. Chen, M. et al. Generation of a transgenic mouse model with chondrocyte- VEGFR2. Blood 116, 3677–3680 (2010). speciﬁc and tamoxifen-inducible expression of Cre recombinase. Genesis 45, 25. Corsini, M. et al. Cyclic adenosine monophosphate-response element-binding 44–50 (2007). protein mediates the proangiogenic or proinﬂammatory activity of gremlin. 51. Gazzerro, E. et al. Conditional deletion of gremlin causes a transient increase Arterioscler. Thromb. Vasc. Biol. 34, 136–145 (2014). in bone formation and bone mass. J. Biol. Chem. 282, 31549–31557 (2007). 26. Tzima, E. Role of small GTPases in endothelial cytoskeletal dynamics and 52. Fukubayashi, T. & Kurosawa, H. The contact area and pressure distribution the shear stress response. Circ. Res. 98, 176–185 (2006). pattern of the knee. A study of normal and osteoarthrotic knee joints. Acta 27. Gao, X. & Schottker, B. Reduction-oxidation pathways involved in cancer Orthop. Scand. 51, 871–879 (1980). development: a systematic review of literature reviews. Oncotarget 8, 53. Ahmed, A. M. & Burke, D. L. In-vitro measurement of static pressure 51888–51906 (2017). distribution in synovial joints--Part I: Tibial surface of the knee. J. Biomech. 28. Brazil, D. P., Church, R. H., Surae, S., Godson, C. & Martin, F. BMP Eng. 105, 216–225 (1983). signalling: agony and antagony in the family. Trends Cell Biol. 25, 249–264 54. Riegger-Krugh, C., Gerhart, T. N., Powers, W. R. & Hayes, W. C. Tibiofemoral (2015). contact pressures in degenerative joint disease. Clin. Orthop. Relat. Res. 348, 29. Dolan, V., Murphy, M., Alarcon, P., Brady, H. R. & Hensey, C. Gremlin - a 233–245 (1998). putative pathogenic player in progressive renal disease. Expert Opin. Ther. 55. Glasson, S. S., Chambers, M. G., Van Den Berg, W. B. & Little, C. B. The Targets 7, 523–526 (2003). OARSI histopathology initiative - recommendations for histological 30. Wordinger, R. J. & Clark, A. F. Bone morphogenetic proteins and their assessments of osteoarthritis in the mouse. Osteoarthr. Cartil. 18(Suppl 3), receptors in the eye. Exp. Biol. Med. 232, 979–992 (2007). S17–S23 (2010). 31. Lavoz, C. et al. Gremlin regulates renal inﬂammation via the vascular endothelial growth factor receptor 2 pathway. J. Pathol. 236, 407–420 (2015). Acknowledgements 32. Erdmann, R., Ozden, C., Weidmann, J. & Schultze, A. Targeting the Gremlin- We thank J. Sugita and K. Kaneko for technical assistance. This study was supported by VEGFR2 axis - a promising strategy for multiple diseases? J. Pathol. 236, JSPS KAKENHI Grant Numbers #25670634, #17H04310, and #17K19715, and by the 403–406 (2015). Advanced Research & Development Programs for Medical Innovation (PRIME) from 33. Deng, Z., Li, Y., Gao, X., Lei, G., Huard, J. Bone morphogenetic proteins Japan Agency for Medical Research and Development, AMED. We thank Amy Van for articular cartilage regeneration. Osteoarthritis Cartilage 26, 1153–1161 Deusen from Edanz Group (www.edanzediting.com/ac) for editing a draft of this (2018). manuscript. We thank Jay Mehta for helping to create an illustration. 34. Claus, S. et al. Chronic exposure of bone morphogenetic protein-2 favors chondrogenic expression in human articular chondrocytes ampliﬁed in monolayer cultures. J. Cell Biochem. 111, 1642–1651 (2010). Author contributions 35. Steinert, A. F. et al. Hypertrophy is induced during the in vitro chondrogenic T.S. designed and supervised the research. S.H.C., D.M., H.K., Y.M., H.N., K.O., Y.T., differentiation of human mesenchymal stem cells by bone morphogenetic S.S., and F.Y. performed research. U.C., J.K., M.Y., A.E., E.C., D.C., and S.T. contributed protein-2 and bone morphogenetic protein-4 gene transfer. Arthritis Res. materials and analytical tools. S.H.C. and T.S. analysed data and wrote the manuscript. Ther. 11, R148 (2009). 36. Kuroda, R. et al. Cartilage repair using bone morphogenetic protein 4 and muscle-derived stem cells. Arthritis Rheum. 54, 433–442 (2006). Additional information 37. Chubinskaya, S., Hurtig, M. & Rueger, D. C. OP-1/BMP-7 in cartilage repair. Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- Int. Orthop. 31, 773–781 (2007). 019-09491-5. 38. Sekiya, I. et al. Periodic knee injections of BMP-7 delay cartilage degeneration induced by excessive running in rats. J. Orthop. Res. 27, 1088–1092 (2009). Competing interests: The authors declare no competing interests. 39. Church, R. H. et al. Gremlin1 preferentially binds to bone morphogenetic protein-2 (BMP-2) and BMP-4 over BMP-7. Biochem. J. 466,55–68 (2015). Reprints and permission information is available online at http://npg.nature.com/ 40. Nilsson, O. et al. Gradients in bone morphogenetic protein-related gene reprintsandpermissions/ expression across the growth plate. J. Endocrinol. 193,75–84 (2007). 41. Kawaguchi, H. Endochondral ossiﬁcation signals in cartilage degradation Journal peer review information: Nature Communications thanks the anonymous during osteoarthritis progression in experimental mouse models. Mol. Cells reviewers for their contribution to the peer review of this work 25,1–6 (2008). 42. Etienne-Manneville, S. & Hall, A. Rho GTPases in cell biology. Nature 420, Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in 629–635 (2002). published maps and institutional afﬁliations. 43. Ridley, A. J. et al. Cell migration: integrating signals from front to back. Science 302, 1704–1709 (2003). 44. Ehrlicher, A. J., Nakamura, F., Hartwig, J. H., Weitz, D. A. & Stossel, T. P. Mechanical strain in actin networks regulates FilGAP and integrin binding Open Access This article is licensed under a Creative Commons to ﬁlamin A. Nature 478, 260–263 (2011). Attribution 4.0 International License, which permits use, sharing, 45. Trachootham, D., Alexandre, J. & Huang, P. Targeting cancer cells by ROS- adaptation, distribution and reproduction in any medium or format, as long as you give mediated mechanisms: a radical therapeutic approach? Nat. Rev. Drug Discov. appropriate credit to the original author(s) and the source, provide a link to the Creative 8, 579–591 (2009). Commons license, and indicate if changes were made. The images or other third party 46. Stanton, H. et al. Investigating ADAMTS-mediated aggrecanolysis in mouse material in this article are included in the article’s Creative Commons license, unless cartilage. Nat. Protoc. 6, 388–404 (2011). indicated otherwise in a credit line to the material. If material is not included in the 47. Ovchinnikov, D. A., Deng, J. M., Ogunrinu, G. & Behringer, R. R. Col2a1- article’s Creative Commons license and your intended use is not permitted by statutory directed expression of Cre recombinase in differentiating chondrocytes in regulation or exceeds the permitted use, you will need to obtain permission directly from transgenic mice. Genesis 26, 145–146 (2000). the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 48. Hilton, M. J., Tu, X. & Long, F. Tamoxifen-inducible gene deletion reveals a licenses/by/4.0/. distinct cell type associated with trabecular bone, and direct regulation of PTHrP expression and chondrocyte morphology by Ihh in growth region cartilage. Dev. Biol. 308,93–105 (2007). © The Author(s) 2019 NATURE COMMUNICATIONS | (2019) 10:1442 | https://doi.org/10.1038/s41467-019-09491-5 | www.nature.com/naturecommunications 13
Nature Communications – Springer Journals
Published: Mar 29, 2019
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera