Research ArticlePhysiology

A dual role for NOTCH signaling in joint cartilage maintenance and osteoarthritis

See allHide authors and affiliations

Sci. Signal.  21 Jul 2015:
Vol. 8, Issue 386, pp. ra71
DOI: 10.1126/scisignal.aaa3792

Abstract

Loss of NOTCH signaling in postnatal murine joints results in osteoarthritis, indicating a requirement for NOTCH during maintenance of joint cartilage. However, NOTCH signaling components are substantially increased in abundance in posttraumatic osteoarthritis in humans and mice, suggesting either a reparative or a pathological role for NOTCH activation in osteoarthritis. We investigated a potential dual role for NOTCH in joint maintenance and osteoarthritis by generating two mouse models overexpressing the NOTCH1 intracellular domain (NICD) within postnatal joint cartilage. The first mouse model exhibited sustained NOTCH activation to resemble pathological NOTCH signaling, whereas the second model had transient NOTCH activation, which more closely reflected physiological NOTCH signaling. Sustained NOTCH signaling in joint cartilage led to an early and progressive osteoarthritic-like pathology, whereas transient NOTCH activation enhanced the synthesis of cartilage matrix and promoted joint maintenance under normal physiological conditions. Through RNA-sequencing, immunohistochemical, and biochemical approaches, we identified several targets that could be responsible for NOTCH-mediated cartilage degradation, fibrosis, and osteoarthritis progression. These targets included components of the interleukin-6 (IL-6)–signal transducer and activator of transcription 3 (STAT3) and mitogen-activated protein kinase signaling pathways, which may also contribute to the posttraumatic development of osteoarthritis. Together, these data suggest a dual role for the NOTCH pathway in joint cartilage, and they identify downstream effectors of NOTCH signaling as potential targets for disease-modifying osteoarthritis drugs.

INTRODUCTION

Osteoarthritis (OA) is the most common joint disorder observed worldwide and is a major cause of disability, which carries an extremely high socioeconomic burden (1). In the United States alone, nearly 25% of the population is expected to have physician-diagnosed OA by 2030 (2), which is estimated to cost between $100 billion and $200 billion annually (3, 4). OA is characterized clinically by fibrosis and degradation of joint cartilage, osteophyte formation, subchondral bone sclerosis, and synovial tissue hyperplasia (57). Most of these disease outcomes are determined by abnormal differentiation of chondrocytes coupled with an imbalance in the turnover of cartilaginous extracellular matrix (ECM). Under physiological conditions, articular and joint chondrocytes maintain a dynamic balance between the synthesis and degradation of ECM components. Normal joint cartilages are largely composed of an ECM rich in type II collagen (COL2A1) and the proteoglycan aggrecan (ACAN), which in the context of OA are degraded by specific collagenases [including matrix metalloproteinase 13 (MMP13), MMP9, and MMP3] and aggrecanases [including a disintegrin and metalloproteinase with thrombospondin motif 4 (ADAMTS4) and ADAMTS5] (810). The disease progression of OA can also be exacerbated by inflammatory cytokines, such as interleukin-1β (IL-1β), IL-6, and tumor necrosis factor–α (TNF-α) (1113), which both suppress matrix synthesis and promote matrix degradation. Although several factors have been implicated in the pathogenesis of OA, the genetic pathways that precisely regulate the catabolic (degradative) or anabolic (build up) cartilaginous response or the balance of these processes is just beginning to be understood (5, 14, 15). Understanding the molecular mechanisms that regulate cartilage anabolism and catabolism to maintain joint cartilage homeostasis will be extremely important in developing future disease-modifying OA drugs (DMOADs).

The NOTCH pathway was identified as a potential regulator of both catabolic and anabolic molecules in the cartilage ECM during development (1618). In mammals, NOTCH signaling is primarily initiated when a ligand of NOTCH interacts with one of the NOTCH family receptors, which leads to a series of receptor cleavage events that ultimately release the NOTCH intracellular domain (NICD) into the cytoplasm. The NICD then translocates to the nucleus and activates downstream target gene expression, including genes of the Hes/Hey families, through its interactions with the recombination signal binding protein for immunoglobulin κJ region and mastermind-like (RBPjκ-MAML) transcriptional complex (1921). NOTCH signaling components are found in the developing growth plate cartilage and in adult articular cartilage (22, 23), which suggests a functional role for NOTCH in regulating both cartilage development and homeostasis.

We previously found that loss of RBPjκ-dependent NOTCH signaling in all joint tissues, as well as in postnatal joint cartilages, results in an early and progressive OA-like pathology (24), indicating a requisite role for NOTCH in the maintenance of articular cartilage and joints. Studies have also demonstrated that the NOTCH pathway is highly activated in mouse and human joint tissues during posttraumatic OA (2527) and that temporary suppression of NOTCH signaling in murine joints leads to the delayed progression of OA (25). Together, these data suggest that physiological NOTCH signaling within joint tissues is essential for joint maintenance; however, when NOTCH signaling is abnormally activated, such as occurs during posttraumatic OA, temporary inhibition of the NOTCH pathway or its downstream effectors may provide a means for altering the progression of posttraumatic OA. Because the NOTCH pathway is just beginning to be understood in the context of OA and joint cartilage maintenance, identifying potential downstream effectors that may also serve as better drug targets will be crucial in both our understanding of the disease and the development of future therapeutics. To assess the potential dual role for NOTCH signaling in joint cartilage and to determine whether the amplitude, duration, or frequency of NOTCH activation influences cartilage physiology or pathology, we generated mouse models overexpressing NICD1 within postnatal joint cartilages in either a sustained or a transient manner. We further investigated the underlying mechanisms and downstream targets of NOTCH signaling during OA development and cartilage degeneration.

RESULTS

Sustained activation of NOTCH1 signaling in postnatal chondrocytes results in a progressive OA-like pathology

To determine whether NOTCH activation in OA was a reparative response or contributed to the pathology, we generated a NOTCH gain-of-function (GOF) genetic mouse model using the tetracycline-on (Tet-On) system in combination with the Cre-recombinase system: Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ (fig. S1A). Only in the presence of the reverse tetracycline transactivator (rtTA) and tetracycline [or its commercially available alternative doxycycline (DOX)] can the tetO promoter drive overexpression of NICD1 within cartilage. Therefore, we can control the robustness and duration of NOTCH signaling by adjusting the dose and frequency of DOX administration. We first generated and characterized the NOTCH1 overexpression profile in a sustained NOTCH GOF model with high doses and frequencies of DOX injections. Knee joints of Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ (sGOF NICD1) and Rosa-rtTAf/+ littermate controls (wild type) were injected with a single high dose of DOX (100 μg/g) and harvested at days 1 and 3 after injection. Immunofluorescence analysis revealed that NOTCH1 was increased in abundance in all zones of the articular cartilage in sGOF NICD1 mice at day 1 after injection, and maintained high abundance on day 3 after DOX delivery, suggesting sustained NOTCH1 activation (Fig. 1A).

Fig. 1 Sustained activation of NOTCH1 signaling in postnatal chondrocytes results in a progressive, OA-like pathology.

(A and B) Wild-type (WT) mice and Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice (sGOF NICD1) were subjected to high-dose administration of DOX (100 μg/g), and then sGOF NICD1 and WT samples were analyzed at the indicated times by (A) immunofluorescence staining for NOTCH1 and (B) real-time qPCR assay to determine the relative abundances of NICD1 and Hes1 mRNAs. All mRNA abundances were normalized to that of the gene encoding β-actin (Actb) and then were normalized to the controls. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test. Images are representative of at least three independent experiments. Scale bars, 50 μm. DAPI, 4′,6-diamidino-2-phenylindole. (C) ABH/OG staining and immunohistochemical analysis of COL2A1, SOX9, COL10A1, COL3A1, and MMP13 in knee sections from 2-month-old WT and sGOF NICD1 mice. High magnification of a centralized region of articular cartilage is shown in the yellow boxes. White arrows indicate cell clusters. Scale bars, 50 μm. Data are representative of at least three independent experiments. (D) Histomorphometry analysis of articular cartilage thickness, area, and chondrocyte number was performed on knee sections from 2-month-old WT and sGOF NICD1 mice. All results were normalized to WT controls, which were set at 1. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test. (E) Real-time qPCR analysis was performed to compare the relative expression of the indicated genes in articular chondrocytes isolated from 2-month-old WT and sGOF NICD1 mice. All mRNA abundances were normalized to that of Actb mRNA and then were normalized to the controls. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test.

To confirm these results, RNA was isolated from the articular chondrocytes of 1-month-old wild-type and sGOF NICD1 mice on day 0 and on days 1, 3, and 7 after receiving a single high-dose DOX injection. Real-time quantitative polymerase chain reaction (qPCR) analysis revealed that NICD1 expression peaked on day 1 after DOX injection (eightfold increased relative to wild type) and maintained a high level of expression at days 3 and 7 (four- and sixfold increased, respectively) (Fig. 1B). Expression of the NOTCH target gene Hes1 was also increased on days 1 and 3 (~1.6-fold increased relative to wild type at both time points), although expression returned to baseline by day 7 (Fig. 1B). Therefore, we injected 1-month-old sGOF NICD1 mice and control mice with high doses of DOX at high frequency (100 μg/g three times per week) to obtain sustained NOTCH activation (fig. S1B) and then harvested knee joints at 2 months of age. Histological analysis revealed the hallmark features of an OA-like pathology in the sGOF NICD1 mice, including (i) joint cartilage fibrosis and degeneration, (ii) meniscus degradation, and (iii) synovial tissue hyperplasia (Fig. 1C). Alcian blue, hematoxylin, and orange G (ABH/OG) staining demonstrated a loss of proteoglycan content in the articular cartilage of knee sections from the sGOF NICD1 mice, with the development of chondrocyte clusters in regions retaining some stain, a hallmark of early OA (Fig. 1C). Severe synovial hyperplasia was also observed in the sGOF NICD1 mice. Additionally, the growth plates often collapsed in sGOF NICD1 mice, which led to an altered architecture of the subchondral bone. Histomophometry performed on 2-month-old ABH/OG-stained knee sections established that articular cartilage thickness and area were substantially reduced (by ~30 and ~40%, respectively) in sGOF NICD1 mice compared to wild-type mice, and the number of chondrocytes was decreased by ~50% (Fig. 1D). Terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining revealed an almost 60% increase in TUNEL-positive cells in knee sections from sGOF NICD1 mice, especially in the regions in which chondrocytes formed cell clusters (fig. S2A). These data suggest that sustained NOTCH1 activation in joint cartilage led to early and severe cartilage loss, which was potentially mediated by the induction of apoptosis in articular chondrocytes.

To identify specific changes in the abundances of ECM-related molecules, we performed immunohistochemical analysis for COL2A1, COL10A1, SOX9, COL3A1, and MMP13 in knee sections from 2-month-old sGOF NICD1 mice and control mice (Fig. 1C). This analysis demonstrated that fibrotic cartilage regions exhibited enhanced COL3A1 abundance, which was suggestive of chondrocyte dedifferentiation and fibrosis. The abundance of COL2A1 was moderately reduced in the articular cartilage in sections from sGOF NICD1 mice, whereas COL10A1 abundance was increased in the articular cartilage deep zone. Note that SOX9 was largely absent from the superficial zones of sGOF NICD1 articular cartilage. In regions that had severe cartilage degeneration and fibrosis, SOX9 was completely absent (Fig. 1C). Consistent with articular cartilage loss, we detected an increase in the catabolic marker MMP13 within areas of cartilage degeneration.

Real-time qPCR analysis of RNA isolated from the articular chondrocytes of 2-month-old sGOF NICD1 mice and control mice (Fig. 1E) demonstrated that the expression of the anabolic gene Sox9 was decreased by ~2-fold in sGOF NICD1 mice, whereas the expression of the catabolic genes Mmp13 and Adamts5 was substantially increased (by 132- and 6-fold, respectively). Consistent with the immunohistochemical analysis, the dedifferentiation marker Col1a1 was increased ~35-fold in expression in sGOF NICD1 mice compared to that in control mice. Moderate increases in the expression of Col2a1 (~2-fold) and Acan (~4-fold) were also observed in sGOF NICD1 mice; however, the expression of Col10a1 was decreased (~3-fold), which was suggestive of abnormal synthesis of cartilage ECM (Fig. 1E). Together, these data suggest that sGOF NICD1 mice exhibited altered ECM synthesis, increased cartilage degradation, and cartilage fibrosis, consistent with a progressive OA-like pathology. These phenotypes were largely replicated at 2 and 8 months of age in another sustained GOF NICD1 genetic mouse model for which tamoxifen (1 mg/10 g for 5 days at 1 month of age) was used to decrease the cartilage-specific sustained expression of NICD1 from the Rosa26 locus, which was induced by the AcanCreERT2 driver (NICD1AcanTM) (fig. S3), further suggesting that enhanced NOTCH signaling specifically within chondrocytes was the driving force for OA development in these mice.

Transient activation of NOTCH1 signaling in postnatal chondrocytes results in increased synthesis of cartilage ECM and joint maintenance

Because sustained NOTCH activation caused by both low and high amounts of NICD1 resulted in an early and progressive OA-like pathology, we set out to determine whether transient NOTCH activation in joint cartilage led to a different bioactivity by more closely mimicking physiologic NOTCH signaling in immature cartilage during development. To generate a transient NOTCH GOF model, we injected the Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice (tGOF NICD1) and Rosa-rtTAf/+ littermate controls (wild type) with low doses of DOX at a low frequency (1 μg/g once per week for 1 month) at 1 month of age (fig. S1B) and then harvested knee joints at 2 and 4 months of age. Immunofluorescence and real-time qPCR analyses performed on 1-month-old tGOF NICD1 mice and control mice confirmed that a single low dose of DOX only transiently activated NOTCH1 signaling for 1 to 3 days in sparse chondrocytes of the articular cartilage (Fig. 2, A and B). Joint integrity analyses demonstrated that by 2 months of age, tGOF NICD1 mice that received a single injection of DOX per week exhibited normal knee joint architecture with increased synthesis of cartilage ECM (Fig. 2C). Histomophometry analysis revealed that articular cartilage thickness and area were increased by 23 and 31%, respectively, in tGOF NICD1 mice as compared to control mice (Fig. 2D). Note that the number of articular chondrocytes in the tGOF NICD1 mice was increased by 27% (Fig. 2D). By 4 months of age, the tGOF NICD1 mice exhibited the continuation of joint cartilage maintenance, with trends for increased cartilage area and chondrocyte number, as well as a 13% increase in articular cartilage thickness as compared to that in control mice (fig. S4, A and B). TUNEL staining revealed no substantial change in the extent of apoptosis in tGOF NICD1 mice at both 2 and 4 months of age (fig. S2, B and C).

Fig. 2 Transient activation of NOTCH1 signaling in postnatal chondrocytes results in increased synthesis of cartilage ECM and joint maintenance.

(A and B) WT mice and Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice (tGOF NICD1) were subjected to low-dose administration of DOX (1 μg/g), and then tGOF NICD1 and WT samples were analyzed at the indicated times by (A) immunofluorescence staining for NOTCH1 and (B) real-time qPCR assay to determine the relative abundances of NICD1 and Hes1 mRNAs. All mRNA abundances were normalized to that of Actb and then normalized to the controls. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test. Images are representative of at least three independent experiments. Scale bars, 50 μm. (C) ABH/OG staining and immunohistochemical analysis of COL2A1, SOX9, COL10A1, COL3A1, and MMP13 in knee sections of 2-month-old WT and tGOF NICD1 mice. High magnification of a centralized domain of articular cartilage is shown in the yellow boxes. Scale bars, 50 μm. Data are representative of at least three independent experiments. (D) Histomorphometry analysis of articular cartilage thickness, area, and chondrocyte number was performed on knee sections from 2-month-old WT and tGOF NICD1 mice. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test. (E) Real-time qPCR analysis was performed to compare the relative expression of the indicated genes in articular chondrocytes isolated from 2-month-old WT and sGOF NICD1 mice. All mRNA abundances were normalized to that of Actb mRNA and then were normalized to the controls. Data are means ± SD of at least three independent experiments. *P < 0.05 by two-tailed Student’s t test.

To identify specific changes in the abundances of ECM-related molecules, we performed immunohistochemical analysis of COL2A1, COL10A1, SOX9, COL3A1, and MMP13 in knee sections of tGOF NICD1 mice and control mice at 2 and 4 months of age (Fig. 2C and fig. S4A). By 2 months of age, COL2A1 abundance was substantially increased in the articular cartilage of knee sections of tGOF NICD1 mice. The number of chondrocytes expressing SOX9 was also increased in the superficial and tangential zones of tGOF NICD1 mouse knee sections, whereas a modest increase in COL10A1 abundance was observed strictly in the deep zone. The increased amounts of COL2A1 and SOX9 were also observed in 4-month-old tGOF NICD1 mice. Both COL3A1 and MMP13 were almost undetectable in tGOF NICD1 mice and control mice at 2 and 4 months of age (Fig. 2C and fig. S4A). RNA was isolated from the articular chondrocytes of 2- and 4-month-old tGOF NICD1 mice and control mice to access changes in the expression of ECM-related genes (Fig. 2E and fig. S4C). Real-time qPCR analysis demonstrated that by the age of 2 months, the expression of Sox9, Col2a1, and Acan was substantially increased (between two- and threefold for each) in tGOF NICD1 mutant mice, whereas the expression of Col10a1 was decreased by more than twofold. Moderate increases in the expression of Col1a1 and Mmp13 were also observed in the tGOF NICD1 mice at this time (by 3.5- and 6.2-fold, respectively) (Fig. 2E). We did not observe any substantial changes in the expression of Adamts4 and Adamts5 (Fig. 2E). The anabolic effect observed in 2-month-old tGOF NICD1 mice was maintained until 4 months of age, which was 2 months after the last administration of DOX. Gene expression analysis revealed the increased expression of Sox9, Col2a1, and Acan in the tGOF NICD1 mice (by 4-, 3.2-, and 3-fold, respectively), whereas the previously increased expression of Col1a1 and Mmp13 in 2-month-old tGOF NICD1 mice was now reduced by 2-fold in 4-month-old tGOF NICD1 mice as compared to age-matched wild-type mice (fig. S4C). Together, these data were consistent with the immunohistochemical analysis, and they demonstrated that transient activation of NOTCH1 signaling in adult joint cartilages enhanced the synthesis of articular cartilage ECM and promoted joint maintenance for at least 2 months after the last administration of DOX.

To investigate whether transient NOTCH1 activation protected joint integrity against trauma-induced OA, we performed a meniscal-ligamentous injury (MLI) surgery on 8-week-old tGOF NICD1 mice and control mice, which was followed by four low doses of DOX by injection (1 μg/g once per week). Knee joints were harvested 8 weeks after surgery, and joint integrity was analyzed by ABH/OG-stained histology (fig. S4D). Accelerated OA progression was observed in the tGOF NICD1 mice, including joint cartilage degeneration and clefting, meniscus degradation, osteophyte formation, and severe synovial tissue expansion (fig. S4D). These data suggest that transient NOTCH activation promoted cartilage and joint maintenance only under physiological conditions, but in pathological situations, such as the trauma-induced inflammatory environment, even transient NOTCH activation accelerated OA development and led to rapid joint degradation, which was potentially a result of synergistic effects with other proinflammatory factors within the injured joint environment.

Large-scale temporal gene expression profiling reveals potential NOTCH1 target genes responsible for cartilage fibrosis and degradation

We demonstrated that NOTCH signaling was a critical regulator of cartilage and joint maintenance under physiological conditions and that sustained activation of NOTCH in postnatal cartilages led to a progressive OA-like pathology; however, the downstream NOTCH targets responsible for OA development remained unclear. To address this issue, we developed an in vitro sustained NOTCH GOF model with costal chondrocytes isolated from Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice and control mice, and we performed RNA-sequencing (RNA-seq) analysis at 6 and 48 hours after the cells were treated with DOX in culture. These data revealed that NOTCH activation induced the expression of 451 genes at 6 hours and 1249 genes at 48 hours, such as NOTCH pathway genes (including HeyL, Notch3, and Jag1) and genes encoding cartilage-degradation enzymes (including Mmp13, Mmp9, Adamts4, and Adamts5), fibrous collagens (including Col3a1, Col4a1, and Col5a3), growth factors (including Pdgfb, Tgfb1, and Tgfb2), and specific cytokines and chemokines (including Il6, Ccl20, and Ccl17), whereas the expression of genes encoding other inflammatory factors often associated with OA remained relatively low (including Il1a, Il1b, and Tnf) (Fig. 3).

Fig. 3 Large-scale temporal gene expression profiling reveals potential NOTCH1 target genes responsible for cartilage fibrosis and degradation.

P2 costal chondrocytes isolated from control and Col2a1Cre; tetO-NICD1f/+; Rosa-rtTAf/+ mice were treated with DOX (10 μg/ml) for 6 or 48 hours in culture, after which RNA was collected from the chondrocytes for RNA-seq experiments. Each group contained three biological repeats. The log2 of the fold change in gene expression (Log2 FC) in the Col2a1Cre; tetO-NICD1f/+; Rosa-rtTAf/+ mice relative to that in the control mice of the indicated genes of interest at the indicated times are shown, including Notch pathway genes, typical chondrogenic and catabolic genes, genes encoding fibrous collagens, as well as genes whose products are involved in inflammatory signaling, tyrosine kinase signaling, GPCR signaling, and NO signaling. *P < 0.05 after Benjamini-Hochberg correction for multiple testing (Cufflinks version 2.0.2). Negative values represent a reduction in gene expression as compared to controls at each time indicated.

NOTCH activation also resulted in the decreased expression of 75 genes at 6 hours and 867 genes at 48 hours, such as those encoding cartilage-related collagens (including Col2a1, Col9a1, and Col11a1) and other chondrogenic factors (including Sox9, Sox5, Sox6, Acan, and Comp). In particular, we highlight here the profiles of NOTCH target genes, chondrogenic genes, catabolic genes, and genes encoding fibrous collagens and inflammatory factors that are likely associated with the NOTCH-induced progression of OA in vivo (Fig. 3). Most of the chondrogenic genes were decreased in expression, whereas the expression of Mmp and Adamts family genes was substantially increased at 48 hours after treatment with DOX. Furthermore, the expression of genes encoding collagens was differentially affected: the expression of genes encoding cartilage-related collagens, such as Col2a1, Col9a1, Col9a2, Col9a3, Col11a1, and Col11a2 was substantially reduced, whereas the expression of genes encoding fibrous collagens, such as Col3a1, Col4a1, Col4a2, Col5a3, Col6a2, and Col14a1, was markedly increased (Fig. 3).

These results identified previously uncharacterized NOTCH-regulated genes in chondrocyte cultures and demonstrated that sustained NOTCH signaling suppressed the chondrogenic phenotype and promoted cartilage fibrosis and degradation, implicating the NOTCH pathway as a critical regulator of the pathogenesis of OA. We also found that the expression of some genes encoding inflammatory factors was modestly increased, such as several cytokine-encoding genes. In particular, the expression of Il6 was 5- and 54-fold increased at 6 and 48 hours, respectively (Fig. 3), which was one of the most robustly responsive genes in DOX-treated chondrocytes from sGOF NICD1 mice. In addition, pathway analysis identified several clustered genes related to tyrosine kinase signaling, G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptor (GPCR) signaling, and nitric oxide (NO) signaling pathways as being substantially increased in expression in the DOX-treated chondrocytes from sGOF NICD1 mice, many of which have been implicated in OA, cartilage catabolism, or both (Fig. 3).

Sustained NOTCH1 signaling activates the IL-6–STAT3 pathway in OA cartilage

IL-6 is a proinflammatory cytokine that is produced by various types of cells and is associated with OA in humans (28). It is found in normal articular cartilage but is substantially increased in abundance in OA synovial membrane, cartilage, and synovial fluid, where it not only stimulates the production of the cartilage-degrading proteases but also suppresses the expression of anabolic genes, such as Sox9, Col2a1, and Acan (7, 9, 11). IL-6 signaling is initiated by ligand binding to a membrane-bound IL-6 receptor (IL-6R) or to a soluble form of IL-6R (sIL-6R) together with the co-receptor gp130, which ultimately leads to activation of Janus-activated kinase 1 (JAK1) and JAK2, as well as signals propagated through the extracellular signal–regulated kinase (ERK) pathway and the signal transducer and activator of transcription 1 (STAT1)–STAT3 pathway (2932). To better understand the role of IL-6 after NOTCH activation, we first confirmed the RNA-seq results with real-time qPCR analysis of postnatal day 2 (P2) costal chondrocytes, P21 articular chondrocytes, and the chondrogenic cell line ATDC5 (Fig. 4, A and B). We found that Il6 expression was markedly increased in all cell models in response to NOTCH activation.

Fig. 4 Sustained NOTCH1 signaling activates the IL-6–STAT3 pathway in cultured cells and OA cartilages.

(A) WT and Col2a1Cre; tetO-NICD1f/+; Rosa-rtTAf/+ mice (sGOF NICD1) were treated with DOX (10 μg/ml) for 6 or 48 hours. RNA was then collected from P2 costal chondrocyte cultures (top) or P21 articular chondrocyte cultures (bottom) isolated from the mice, which was subjected to real-time qPCR analysis to determine the relative abundance of Il6 mRNA. All mRNA abundances were normalized to that of Actb and then were normalized to the controls. Data are means ± SD of three independent experiments. *P < 0.05 by two-tailed Student’s t test. (B) ATDC5 cells were transfected with FLAG control plasmid or with plasmid encoding FLAG-NICD1 (top) or were separately treated with control diluents or rIL-6 protein (1 ng/ml) for 24 hours (bottom), and then the relative abundances of the indicated mRNAs were determined by real-time qPCR analysis. All mRNA abundances were normalized to that of Actb and then were normalized to the controls. Data are means ± SD of three independent experiments. *P < 0.05 by two-tailed Student’s t test. (C) ATDC5 cells that were left untreated or were treated with rIL-6 (100 ng/ml) for 30 or 60 min (top) and R26-NICD1f/+ primary chondrocytes that were infected with adenoviruses expressing either green fluorescent protein (Ad-GFP) or Cre (Ad-CRE) (bottom) were subjected to Western blotting analysis with antibodies specific for the indicated proteins. Western blots are representative of three independent experiments. (D) ABH/OG staining and immunohistochemical analysis of IL-6 and pSTAT3 in knee sections from 2-month-old WT and sGOF NICD1 mice. Red arrowheads indicate pSTAT3-positive cells. Scale bars, 50 μm. Data are representative of at least three independent experiments. (E) ATDC5 cells that were transfected with FLAG control plasmid or that were transfected with plasmid encoding FLAG-NICD1 and then were either left untreated or treated with IL-6–neutralizing antibody (anti–IL-6) for 24 hours were subjected to real-time qPCR analysis to determine the relative abundances of the indicated mRNAs. All mRNA abundances were normalized to that of Actb and then were normalized to the controls, which were set at 1. Data are means ± SD of three independent experiments. *P < 0.05 by one-way analysis of variance (ANOVA), followed by the Bonferroni method.

To better understand how IL-6 regulated ECM-related and OA-associated factors, we used an in vitro culture model of ATDC5 chondrogenic cells. Differentiated ATDC5 cells were transfected with plasmid encoding FLAG-tagged NICD1 or a control plasmid or separately were treated with recombinant IL-6 (rIL-6) proteins for 24 hours, and RNA was harvested for real-time qPCR analyses. Gene expression analysis revealed that rIL-6 suppressed the expression of Sox9, Cola2a1, and Acan, whereas it induced the expression of Mmp13 and Col3a1, which is consistent with the gene expression changes that were observed under sustained NOTCH activation (Fig. 4B). Because STAT3 is an important downstream mediator of IL-6 signaling and is the most abundant STAT molecule found in cartilage, we assessed the capacity of IL-6 to stimulate the generation of phosphorylated STAT3 (pSTAT3), the active form of the STAT3 protein, in chondrogenic ATDC5 cells. Western blotting analysis revealed that rIL-6 rapidly activated STAT3, as evidenced by a rapid increase in its phosphorylation at Tyr705 (Fig. 4C). Given that sustained NICD1 signals stimulate Il6 expression and that IL-6 activates STAT3 in chondrogenic cells, we predicted that STAT3 signaling would be enhanced in NICD1 GOF chondrocytes. Indeed, Western blotting analysis showed a substantial increase in the abundance of pSTAT3 in these cells 48 hours after NOTCH activation (Fig. 4C).

To determine whether NOTCH-induced IL-6–STAT3 signaling occurred within the joints of our NOTCH GOF model mice (sGOF NICD1 and NICD1AcanTM), we performed immunohistochemical analysis of joint sections from 2- and 8-month-old GOF mice and control mice. A low amount of IL-6 was found in normal articular cartilage and the synovial tissues of wild-type animals; however, IL-6 abundance was enhanced in the cartilage ECM and hyperplastic synovium of mice with sustained NOTCH signaling, especially in regions of severe fibrosis (Fig. 4D and fig. S3D). To confirm that increased IL-6 abundance was associated with enhanced STAT3 activation, we examined the amounts of pSTAT3 in both NOTCH GOF models. Immunohistochemical analysis revealed that pSTAT3 was undetectable in the knee joints of wild-type mice; however, a substantial increase in pSTAT3 abundance was detected in the fibrotic cartilages, degraded menisci, and hyperplastic synovial tissues of both sGOF NICD1 mice and NICD1AcanTM mice (Fig. 4D and fig. S3D).

To further demonstrate whether the effect observed under conditions of NOTCH activation was IL-6–dependent, we pretreated differentiated ATDC5 cells with IL-6–neutralizing antibodies for 24 hours and then transfected the cells with plasmid encoding FLAG-tagged NICD1 or a control plasmid. Real-time qPCR analysis revealed that neutralization of IL-6 signaling did not alleviate the NOTCH-mediated suppression of Sox9 expression but partially reduced the NOTCH-mediated induction of Mmp13 expression (Fig. 4E). Furthermore, neutralization of IL-6 signaling almost completely eliminated the NOTCH-mediated induction of Col3a1 expression. These data suggest that NOTCH-mediated Sox9 expression was IL-6–independent, whereas the NOTCH-stimulated expression of Mmp13 and Col3a1 occurred in an IL-6–dependent manner (Fig. 4E). Together, these data suggest that prolonged NOTCH activity induces the production of IL-6 in OA cartilage, which may directly result in the degeneration of joint cartilage and fibrosis through the IL-6–STAT3 signaling pathway.

We also confirmed the activation of the IL-6–STAT3 signaling pathway in a surgical-induced model of OA in which 2-month-old wild-type mice were subjected to an MLI surgery, which produces a progressive OA-like phenotype from 4 to 20 weeks after injury (33, 34). At 12 weeks after surgery, OA-like pathologies were observed in the MLI mice, including joint cartilage degeneration and fibrosis, osteophyte formation, and synovial hyperplasia (Fig. 5). Immunohistochemical analysis demonstrated increased IL-6 abundance in OA tissues, especially the fibrotic and degenerative cartilages and synovium, and pSTAT3 was detected in IL-6–positive regions (Fig. 5). These results suggest that injury-induced joint cartilage fibrosis and degeneration may also occur through inappropriate activation of the IL-6–STAT3 signaling pathway, a mechanism that may be shared with NOTCH-induced OA.

Fig. 5 The IL-6–STAT3 pathway is activated in a trauma-induced mouse model of OA.

ABH/OG staining and immunohistochemical analysis of IL-6 and pSTAT3 in knee sections from 5-month-old WT mice 12 weeks after they were subjected to sham treatment or MLI surgery. Red arrowheads indicate pSTAT3-positive cells. Scale bars, 50 μm. Data are representative of at least five independent experiments.

The ERK and p38 pathways likely contribute to the NOTCH1-induced OA-related phenotypes

In addition to STAT3, the ERK and p38 mitogen-activated protein kinase (MAPK) pathways are important mediators of pathological IL-6 signaling, as well as of numerous other cytokines and growth factors. To test whether IL-6 activated both pathways in chondrogenic cells, we examined the effects of rIL-6 on the phosphorylation (and thus activation) of ERK and p38 MAPK in ATDC5 cells. Western blotting analysis showed that the abundances of phosphorylated ERK1/2 (pERK1/2) and phosphorylated p38 (pp38) increased markedly within 30 min of treatment with rIL-6 and then rapidly declined, indicating that IL-6 can simultaneously activate both ERK and p38 MAPK pathways in ATDC5 cells (Fig. 6A). Consistent with the NOTCH-induced increase in IL-6 abundance in costal chondrocytes, the abundances of pERK1/2 and pp38 were also substantially increased in these NICD GOF primary cells (Fig. 6A). Similar to our analysis of pSTAT3 abundance, immunohistochemical analysis of articular cartilage sections from sGOF NICD1 mice further confirmed the increased abundance of pp38 in vivo (Fig. 6B). Unfortunately, the available anti-pERK1/2 antibodies were not compatible with our immunohistochemical analysis of paraffin sections of adult mouse knee joints. Therefore, further in vivo studies may be needed to elucidate the potential contributions of p38 and ERK1/2 to NOTCH-induced OA-like phenotypes.

Fig. 6 Sustained NOTCH1 signaling activates the ERK and p38 pathways in cultured cells and OA cartilages, leading to selected effects on the expression of catabolic, chondrogenic, and fibrous genes.

(A) ATDC5 cells that were left untreated or were treated with rIL-6 (100 ng/ml) for 30 or 60 min (left) and R26-NICD1f/+ primary chondrocytes that were infected with Ad-GFP or Ad-Cre viruses (right) were subjected to Western blotting analysis with antibodies specific for the indicated proteins. Western blots are representative of three independent experiments. (B) Immunohistochemical analysis of pERK in knee sections from 2-month-old WT and sGOF NICD1 mice. Red arrowheads indicate pERK-positive cells. Scale bars, 50 μm. Data are representative of three independent experiments. (C and D) Primary chondrocytes from R26-NICD1f/+ mice were infected with Ad-GFP virus or with Ad-Cre virus alone or in the presence of (C) the MAPK kinase (MEK) inhibitor U0126 or (D) the p28 MAPK inhibitor SB202190. Seventy-two hours later, RNA was isolated from the cells and was analyzed by real-time qPCR analysis to determine the relative abundances of the indicated mRNAs. All mRNA abundances were normalized to that of Actb and then were normalized to the control cells (Ad-GFP–infected cells; set at 1). Data are means ± SD of three independent experiments. *P < 0.05 by one-way ANOVA followed by the Bonferroni method.

To determine whether ERK and p38 MAPK contributed to the OA-associated phenotypes caused by sustained NICD1 signals in vitro, we evaluated the effects of the MEK inhibitor U0126 (which thus prevents the activation of ERK) and the p38 MAPK inhibitor SB202190 on the expression of anabolic (Sox9 and Acan), catabolic (Adamts5 and Mmp13), and fibrotic (Col3a1) genes. Real-time qPCR analysis revealed that both U0126 and SB202190 had no substantial effect on the NOTCH-mediated increase in Col3a1 expression or the suppression of Sox9 expression (Fig. 6, C and D). Whereas U0126 had no effect on the NOTCH-mediated suppression of Acan expression (Fig. 6C), SB202190 exacerbated the NOTCH-mediated suppression of Acan expression (Fig. 6D). Furthermore, both U0126 and SB202190 substantially reduced the NOTCH-induced expression of Mmp13 (Fig. 6, C and D). Moreover, SB202190 also partially reduced the NOTCH-mediated increase in Adamts4 expression (Fig. 6D). Together, these data suggest that the increased activation of ERK and p38 MAPK at least partially contributes to the NOTCH-mediated increase in cartilage catabolic activity, although inhibition of p38 MAPK activity may further inhibit some cartilage anabolic factors. Furthermore, inhibition of either ERK or p38 MAPK did not affect NOTCH-induced cartilage fibrotic gene expression, as indicated by the lack of a change in Col3a1 expression in the presence or absence of inhibitors after NOTCH activation (Fig. 6D).

Alternative signaling pathways are activated by sustained NOTCH1 signaling in chondrocytes

In addition to highlighting inflammatory signaling, the gene and pathway analysis of our RNA-seq data revealed that multiple alternative signaling pathways were markedly increased in extent in sGOF NICD1 chondrocytes compared to control chondrocytes. We highlight here the top three pathways in our analyses that have also been previously reported to be relevant in cartilage biology and arthritis-related pathologies. These include tyrosine kinase signaling, GPCR signaling, and NO signaling pathways (Fig. 3).

Many genes related to tyrosine kinase signaling were induced by sustained NOTCH1 signaling in chondrocytes (Fig. 3). Several of these, including Pgf, Flt1, and Ngf, are involved in the pathology of OA and cartilage catabolism (3539). To investigate whether tyrosine kinase signaling contributed to NOTCH1-induced, OA-related molecular phenotypes in vitro, we treated chondrocytes with genistein, a tyrosine kinase inhibitor. Our real-time qPCR analysis revealed that genistein did not alleviate the NOTCH1-dependent suppression of Sox9 and Acan expression; however, genistein substantially attenuated the NOTCH1-induced expression of catabolic (Adamts4 and Mmp13) and fibrotic (Col3a1) genes (fig. S5). These data indicate that tyrosine kinase signaling may partially mediate NOTCH1-induced catabolic and fibrotic processes.

Because several components of the GPCR signaling pathways were identified in our RNA-seq data set (Fig. 3) and because several are also implicated in cartilage biology and joint disease (40, 41), we next tested whether enhanced GPCR signaling contributed to the NOTCH1-induced, OA-associated molecular phenotypes. We individually inhibited three major downstream pathways of GPCR signaling, the Gαs–adenylate cyclase (AC) pathway, the Gαi pathway, and the Gαq/11–phospholipase Cβ (PLCβ) pathway, with the AC inhibitor SQ-22536 (fig. S6A), the Gαi inhibitor pertussis toxin (PTx) (fig. S6B), and the PLC inhibitor edelfosine (fig. S6C), respectively. Inhibition of any of these three pathways did not inhibit the NOTCH1-mediated induction of Adamts4 and Mmp13 or the NOTCH1-mediated suppression of Sox9 and Acan expression (fig. S6, A to C); however, both PTx and edelfosine slightly reduced the extent of the NOTCH1-dependent expression of Col3a1 (fig. S6, B and C), suggesting that GPCR signaling may partially mediate the NOTCH1-dependent induction of Col3a1 expression.

NO signaling is implicated in human OA and experimental models of OA (4245). The expression of a group of genes related to NO signaling, including genes encoding endothelin receptor A, soluble guanylyl cyclase (sGC; the only known receptor for NO), and some phosphodiesterases (PDEs), was increased by sustained NICD1 signals (Fig. 3). However, the endothelin receptor A antagonist BQ-123 and the sGC inhibitor ODQ only slightly inhibited the NOTCH1-mediated induction of Col3a1 expression (fig. S7, A and B). Similarly, the PDE inhibitor isobutylmethylxanthine (IBMX) only modestly inhibited the NOTCH1-mediated suppression of Sox9 and Acan expression (fig. S7C). In contrast, none of these inhibitors blocked the NOTCH1-dependent induction of Adamts4 and Mmp13 expression (fig. S7, A to C), suggesting that NO-related signaling is unlikely to mediate the effect of NOTCH1 activation on the increased expression of catabolic genes, but that it may have an effect on the NOTCH1-dependent anabolic and fibrotic responses in cartilage.

DISCUSSION

Here, we have provided genetic evidence that sustained versus transient NOTCH activation in postnatal joint cartilages leads to opposing effects on articular cartilage and joint maintenance. This study establishes that sustained NOTCH activation in adult joint cartilage results in a severe, early, and progressive OA-like pathology, whereas transient NOTCH activation results in increased synthesis of cartilage ECM and joint maintenance only under physiological conditions. In vitro and in vivo studies demonstrated the capability of NOTCH signaling to regulate the expression of genes required for anabolic, catabolic, and fibrotic processes, and RNA-seq experiments determined that sustained NOTCH activation suppressed the expression of chondrogenic genes but promoted the expression of genes encoding cartilage-related proteases (MMPs and ADAMTSs), fibrotic collagens, inflammatory factors (including IL-6), and components of a host of broad signaling pathways (tyrosine kinase, GPCR, and NO signaling) that affect cartilage biology and potentially contribute to OA. Together, these data suggest that NOTCH signaling is a critical pathway that regulates joint cartilage homeostasis, and suggest that in pathological situations of sustained signaling, NOTCH is involved in joint cartilage degradation and fibrosis, likely through activating at least the IL-6–STAT3, ERK, and p38 MAPK pathways, and potentially others. Therefore, an appropriate balance of NOTCH signaling must be achieved to maintain normal articular cartilage homeostasis and joint integrity.

NOTCH signaling components are widely found in human adult articular cartilage, indicating a potential role for NOTCH signaling in articular cartilage maintenance during adult life (46, 47). Furthermore, the abundances of NOTCH signaling components are substantially increased in OA cartilage compared to normal cartilage (2527, 46), indicating a role for NOTCH in the onset and progression of OA. We demonstrated here that sustained NOTCH activation in postnatal joint cartilages led to an early and progressive OA-like pathology, including joint cartilage degradation, fibrosis, and synovial expansion, which was likely a result of the marked increase in the abundances of cartilage-degrading enzymes and proinflammatory factors. Loss of or reductions in cartilage-specific NOTCH signaling are capable of reducing the abundance of MMP13 within murine joints, as well as delaying cartilage degeneration over the short term (8 weeks) (25). Our previous data demonstrated that although the decreased amounts of both anabolic and catabolic factors were observed with short-term inhibition of NOTCH signaling in joint cartilages, long-term (6- to 8-month) reduction in NOTCH signaling disrupts the normal physiology of the cartilage and ultimately results in cartilage degeneration (24). Together, these data suggest that NOTCH signaling may play a complex role in cartilage homeostasis, such that both sustained activation or permanent reduction of NOTCH signaling leads to cartilage degradation and joint failure.

NOTCH signaling is involved in both the anabolism and catabolism of cartilage. We and others previously showed that sustained NICD activation in committed growth plate chondrocytes (17, 48) and articular chondrocytes (49) stimulates the expression of Mmp13, whereas permanent inhibition of NOTCH signaling within the growth plate and primary articular chondrocyte cultures reduces Mmp13 expression (17, 48). Several lines of evidence have demonstrated that sustained NOTCH signaling in mesenchymal progenitors and growth plate chondrocytes suppresses the expression of Sox9, Col2a1, and Acan (17, 48, 50, 51). Data presented by Mead and Yutzey (18) indicated that loss of NOTCH signaling leads to the inappropriate expression of Sox9 in hypertrophic chondrocytes. Furthermore, we demonstrated here that sustained overexpression of NICD1 leads to the suppression of Sox9, Col2a1, and Acan expression in adult articular cartilage and chondrogenic cell cultures, whereas transient overexpression of NICD1 in adult joint cartilages promotes the expression of Sox9, Col2a1, and Acan in vivo. Other studies have demonstrated that short-term or transient NOTCH signaling promotes chondrocyte anabolism by inducing Sox9 expression in vitro (52), although the underlying mechanisms remain unclear. Together, these data suggest that transient or physiological NOTCH signaling in chondrocytes favors a balanced anabolic and catabolic cartilage-maintenance response, whereas sustained or enhanced NOTCH activity elicits a pathological response through the simultaneous suppression of chondrogenic genes and the induction of genes encoding catabolic factors.

All of the sustained NICD1 GOF mouse models exhibited robust increases in IL-6 and pSTAT3 signaling in degenerating and fibrotic regions of joint cartilages and synovial tissues. Previous studies identified a conserved RBPjκ-binding site within the promoter of Il6, which overlaps with a nuclear factor κB (NF-κB)–binding site (53). RBPjκ can directly bind to this site, and NOTCH signaling can induce Il6 transcription through interactions with the NF-κB pathway (53, 54). Consistent with these findings, we found that IL-6 production was increased in response to NOTCH activation in vivo and in vitro. RNA-seq analysis revealed that the abundance of Il6 mRNA was 5- and 54-fold increased after 6 and 48 hours of NOTCH1 activation, respectively, compared to that in control cells, which suggests that Il6 may be a direct NOTCH target gene. An in vitro study performed in chondrocyte cultures also suggested that the neutralization of IL-6 cannot modify the NOTCH-mediated suppression of Sox9 and Col2a1 expression, but can oppose the NOTCH-mediated induction of Mmp13 expression (55). Here, we further demonstrated that neutralization of IL-6 at least partially inhibited the NOTCH-dependent induction of Mmp13 expression and also abolished the NOTCH-mediated induction of Col3a1 expression. Furthermore, STAT3 activity can be modulated by NOTCH and the NOTCH effectors HES1 and HES5 in cultured neural cells in which both the NICD and the HES proteins associated with and facilitated complex formation between JAK2 and STAT3 (56). Therefore, these data suggest a potential role for NOTCH and HES proteins in promoting cartilage degeneration in both an IL-6–independent and an IL-6–dependent manner by regulating the phosphorylation and activity of STAT3.

We also determined that sustained NOTCH1 signaling activated the ERK and p38 MAPK pathways in chondrocytes both in vitro and in vivo. ERK and p38 MAPK can function as downstream effectors of IL-6 signaling, but they are also regulated independently of IL-6 through numerous pathways, including those mediated by other inflammatory cytokines, as well growth factors and growth factor receptor signaling pathways, including transforming growth factor–β (TGF-β) and TGF-β receptors, receptor tyrosine kinases (RTKs), and GPCRs. Although the mechanism by which NOTCH1 activates ERK and p38 MAPK in chondrocytes may occur through the induction of IL-6 signaling, we also demonstrated the ability of NOTCH1 to induce the expression of genes encoding other inflammatory cytokines (including Ccl20, Ccl17, and Cxcr4), RTK components (including Flt1, Pgf, Pdgfb, Ngf, and Ngfr), and GPCR family members (including Gpbar1, Rgs5, Gpr20, and Ednra). Several of these factors separately activate ERK and p38 MAPK signaling cascades that are also implicated in cartilage biology and joint disease (3541, 57). Attenuation of specific branches of these broad signaling pathways (RTKs and GPCRs), as well as others induced by NOTCH1 in chondrocytes (for example, NO), has demonstrated the complex nature by which NOTCH signaling regulates cartilage anabolism, catabolism, and fibrotic gene regulation. Future studies will be directed at dissecting this complex regulation to uncover the precise mechanism(s) involved in NOTCH-induced OA and to identify potential targets for the development of DMOADs.

MATERIALS AND METHODS

Mice

Animal studies were approved by the University of Rochester Committee on Animal Resources. All mouse strains, including AcanCreERT2 (58), Rosa-NICD1f/f (59), Col2a1Cre (60), tetO-NICD1 (61), and Rosa-rtTAf/+ (62) mice, were described previously. Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice and AcanCreERT2; Rosa-NICD1f/f (NICD1AcanTM) mice were viable and produced in Mendelian ratios. DOX was administrated to Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ mice and littermate controls with two strategies starting at 1 month of age: (i) 1 μg/g, once per week, or (ii) 100 μg/g, three times per week. Mice were harvested at 2 and 4 months of age. Tamoxifen (1 mg/10 g) was administered daily by intraperitoneal injection to all NICD1AcanTM mice and their littermate controls for five continuous days starting at 1 month of age. Mice were then harvested at 2 and 8 months of age.

Trauma-induced OA model and surgical procedures

We used a well-established mouse MLI model to mimic the clinical situation of trauma-induced OA (34). All experiments were performed according to the protocol approved by the Institutional Animal Care and Use Committee at the University of Rochester Medical Center (URMC). MLI surgery was performed on 8-week-old Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ (tGOF NICD1) mice and littermate controls. Briefly, the bilateral hindlimbs were shaved and prepared for aseptic surgery. The right knee joint was exposed, and the medial collateral ligament was transected. The joint space was then opened slightly, and the medial meniscus was detached from its anterior-medial tibial attachment through a microsurgical technique. The contralateral knee joint was sham-operated without any ligament transection or meniscus detachment. The skin incision was closed after the surgery. DOX was administered to both tGOF NICD1 mice and control mice through intraperitoneal injection (1 μg/g, once per week) for 4 weeks after MLI surgery. Knee joints were harvested 8 weeks after MLI surgery by the age of 4 months.

Analysis of mouse tissue sections

Knee joints were harvested and fixed in 10% neutral-buffered formalin for 3 days, decalcified in Formic Acid Bone Decalcifier (Immunocal, Decal Chemical Corp.) for 7 to 10 days, paraffin-processed, and embedded for sectioning. Tissues were sectioned at 5 μm and stained with ABH/OG. Immunohistochemical analyses were performed on sections with traditional antigen retrieval and colorimetric development methodologies with the following primary antibodies: anti-SOX9 (Santa Cruz Biotechnology), anti-COL2A1 (Thermo Scientific), anti-COL10A1 (Quartett), anti-COL1A1 (Abcam), anti-COL3A1 (Abcam), anti–MMP-13 (Thermo Scientific), anti–IL-6 (Abcam), and anti-pSTAT3 (Cell Signaling). The TUNEL cell death assay was performed on sections with the In Situ Cell Death Detection Kit, Fluorescein (Roche) according to the manufacturer’s instructions. Immunofluorescence analysis and β-galactosidase staining were performed on frozen sections. Knee joints were harvested and fixed in 4% paraformaldehyde for 2 hours at 4°C and decalcified with 14% EDTA at 4°C for 10 days. Tissues were washed in sucrose gradient, embedded with Tissue-Tek OCT medium, snap-frozen in liquid nitrogen, and sectioned at 10 μm with a Leica CM1850 cryotome. An anti-NOTCH1 primary antibody (Santa Cruz Biotechnology) was used for immunofluorescence analysis. β-Galactosidase staining was performed as previously described (63).

Murine costal chondrocyte isolation and RNA-seq

Murine costal chondrocytes were isolated as previously described (15), with modifications. Briefly, rib cages were dissected from 2-day-old Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ or control pups with the soft tissue removed and were plated in 1× phosphate-buffered saline (PBS). Rib cages were digested with Pronase (2 mg/ml, Roche) in 1× PBS for 1 hour in a 37°C shaking water bath and then were digested with collagenase D (3 mg/ml, Roche) in high-glucose Dulbecco’s modified Eagle’s medium (DMEM; Gibco) for 1 hour at 37°C. The rib cages were transferred into a petri dish and digested in collagenase D (3 mg/ml) in high-glucose DMEM for 4 to 6 hours. Murine costal chondrocyte cell suspensions were then filtered through 40-μm filters and seeded at a density of 500,000 cells per well in six-well tissue culture plates in high-glucose DMEM supplemented with 10% fetal bovine serum (FBS; Sigma) and 1% penicillin/streptomycin. Both sets of cell cultures were treated with DOX (10 μg/ml) for 6 or 48 hours. RNA was extracted with the RNeasy Mini Kit (Qiagen) and the RNase-Free DNase Set (Qiagen) according to the manufacturer’s instructions. Three biological replicates were prepared for both the Col2a1Cre; tetO-NICD1; Rosa-rtTAf/+ and control groups. One microgram of total RNA from each sample was sent to the Genomics Research Center (GRC) at URMC for mRNA sequencing and data processing. According to the GRC, RNA concentration was determined with a NanoDrop 1000 spectrophotometer, whereas RNA quality was assessed with an Agilent Bioanalyzer. The TruSeq RNA Sample Preparation Kit V2 (Illumina) was used for next-generation sequencing library construction according to the manufacturer’s protocols. Briefly, mRNA was purified from 100 ng of total RNA with oligo(dT) magnetic beads and then was fragmented. First-stand complementary DNA (cDNA) synthesis was performed with random hexamer priming followed by second-strand cDNA synthesis. End repair and 3′ adenylation were performed on the double-stranded cDNA. Illumina adaptors were ligated to both ends of the cDNA, which was then purified by gel electrophoresis and amplified with PCR primers specific to the adaptor sequences to generate amplicons of about 200 to 500 base pairs in size. The amplified libraries were hybridized to the Illumina single-end flow cell and amplified with the cBot (Illumina) at a concentration of 8 pmol per lane. Single-end reads of 100 nucleotides were generated for each sample and aligned to the organism-specific reference genome. Raw reads generated from the Illumina HiSeq2500 sequencer were de-multiplexed with configurebcl2fastq.pl version 1.8.3 software. Low-complexity reads and vector contamination were removed with sequence cleaner (“SeqClean”) and the National Center for Biotechnology Information (NCBI) UniVec database, respectively. The FASTX toolkit (fastq_quality_trimmer) was applied to remove bases with quality scores below Q = 13 from the end of each read. Processed reads were then mapped to the UCSC hg38 genome build with SHRiMP version 2.2.3, and differential expression analysis was performed with Cufflinks version 2.0.2 [specifically, cuffdiff2 and usage of the general transfer format (GTF) annotation file for the given reference genome].

ATDC5 cell culture and quantitative gene expression analyses

ATDC5 cells (RIKEN BioResource Center) were maintained in DMEM/F-12 (1:1) medium (Gibco) supplemented with 5% FBS and 1% penicillin/streptomycin. For ATDC5 cell differentiation studies, ATDC5 cells were cultured with DMEM/F-12 (1:1) medium supplemented with 5% FBS, 1% penicillin/streptomycin, and 0.1% insulin, transferrin, and sodium selenite premix (BD Biosciences). Differentiated ATDC5 chondrogenic cells cultured for 7 to 14 days were then cultured in serum-free DMEM/F-12 (1:1) medium for 6 hours, transfected with plasmid encoding FLAG-NICD1 or with the control FLAG plasmid with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions, or treated with vehicle control or recombinant mouse IL-6 protein (R&D Systems) for 24 hours. For acute signaling studies, differentiated ATDC5 cells were serum-starved overnight and then were treated with vehicle control or mouse rIL-6 protein (R&D Systems) for the times indicated in the figure legends. For experiments with the IL-6–neutralizing antibody (Abcam), differentiated ATDC5 cells were pretreated with the antibody for 24 hours and then were transfected with control plasmid or plasmid encoding FLAG-NICD1 with Lipofectamine 2000. After 24 hours, RNA was isolated with the RNeasy Mini Kit, and then cDNA synthesis and real-time qPCR analysis were performed as previously described (16). Sequences of primers specific for NICD1, Hes1, Sox9, Col2a1, Acan, Col10a1, Col3a1, Col1a1, Mmp13, Adamts4, Adamts5, and Il6 are available upon request.

Infection of cells with adenovirus

For experiments with inhibitors, primary costal chondrocytes were isolated from neonatal Rosa-NICD1f/+ pups and were cultured in DMEM containing 10% FBS and 1% penicillin/streptomycin as described earlier. Isolated chondrocytes were seeded in 12-well plates at 0.5 × 106 cells per well. After overnight culture, the cells were infected with Ad-GFP or Ad-CRE at a multiplicity of infection of 100 in the presence of polybrene (8 μg/ml). Twenty-four hours after infection, the cells were cultured either in fresh medium (for protein analysis) or in fresh medium containing vehicle or the appropriate inhibitors (for RNA analysis). Forty-eight hours later, cells were harvested for isolation of protein or RNA.

Inhibitors

The ERK1/2 inhibitor U0126 was obtained from Sigma and was used at a final concentration of 10 μM. The p38 MAPK inhibitor SB202190 (Sigma) was used at a final concentration of 2 μM. The protein tyrosine kinase inhibitor genistein was obtained from Tocris and was used at a final concentration of 60 μM. The PLC inhibitor edelfosine was purchased from Tocris and was used at a final concentration of 5 μM. The AC inhibitor SQ-22536 was obtained from Sigma and was used at a final concentration of 10 μM. PTx was obtained from Tocris and was used at a final concentration of 100 ng/ml. The NO-sensitive GC inhibitor ODQ was purchased from Sigma and was used at a final concentration of 10 μM. The endothelin receptor A antagonist BQ-123 was obtained from Sigma and was used at a final concentration of 5 μM. The PDE inhibitor IBMX was purchased from Sigma and was used at a final concentration of 65 μM.

Western blotting analysis

For Western blotting analysis, total proteins were extracted from cells with radioimmunoprecipitation assay buffer [20 mM tris-HCl (pH 8.0), 150 mM NaCl, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate] supplemented with protease and phosphatase inhibitors (Roche). Thirty micrograms of protein from each sample was resolved by 10% SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. Western blots were then blocked with 5% nonfat milk and incubated overnight with primary antibodies at a final dilution of 1:1000. Antibodies specific for STAT3, pSTAT3 (Tyr705), ERK1/2, pERK1/2 (Thr202/Tyr204), p38 MAPK, and pp38 MAPK (Thr180/Tyr182) were all purchased from Cell Signaling Technology.

Histomorphometry

Quantitative histomorphometry was performed on ABH/OG-stained sections with an OsteoMeasure analysis system (OsteoMetrics). Cartilage thickness was measured from the middle of the femoral and tibial condyles. Cartilage area was traced from both articular cartilage surfaces with the area tool in OsteoMeasure software. Three to five mice from each group were analyzed, and at least three slides were examined for each mouse.

Statistical analysis

Statistical analyses were performed using Student’s t test and one-way ANOVA followed by Bonferroni method as appropriate.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/8/386/ra71/DC1

Fig. S1. Development of a NOTCH1 activation model with the combined Tet-On/Cre system.

Fig. S2. Sustained activation of NOTCH1 signaling in postnatal chondrocytes results in increased chondrocyte apoptosis.

Fig. S3. Sustained activation of NOTCH1 signaling in postnatal chondrocytes results in a progressive, OA-like pathology.

Fig. S4. Transient activation of NOTCH1 signaling in postnatal chondrocytes results in increased synthesis of cartilage ECM and joint maintenance for as long as 3 months after injection with DOX, but does not protect from cartilage degradation after MLI surgery.

Fig. S5. Suppression of tyrosine kinase signaling leads to selected effects on NOTCH-regulated genes.

Fig. S6. Suppression of GPCR signaling leads to selected effects on NOTCH-regulated genes.

Fig. S7. Suppression of NO signaling leads to selected effects on NOTCH-regulated genes.

REFERENCES AND NOTES

Acknowledgments: We thank A. E. Kiernan for providing important mouse strains. We would like to gratefully acknowledge the technical expertise and assistance of S. Mack, K. Maltby, and A. Thomas within the Center for Musculoskeletal Research Histology, Biochemistry, and Molecular Imaging Core. We would also like to thank J. Ashton and colleagues within the GRC at URMC for RNA sequencing and data processing. Funding: This work was supported in part by the following NIH grants: R01 grants AR057022 and AR063071 to M.J.H.; R21 grant AR059733 to M.J.H.; P50 Center of Research Translation grant AR054041 to R.J.O.; and P30 Core Center grant AR061307, sub-award #7280 to M.J.H. Author contributions: M.J.H., Z.L., and J.C. conceived and designed experiments; Z.L., J.C., A.J.M., and C.W. performed experiments; Z.L., J.C., A.J.M., C.W., M.J.Z., R.J.O., and M.J.H. analyzed data and contributed to critical discussions; and Z.L., J.C., and M.J.H. wrote and edited the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article