Osteoblasts impair cholesterol synthesis in chondrocytes via Notch1 signalling

Abstract Objectives Previous reports have proposed the importance of signalling and material exchange between cartilage and subchondral bone. However, the specific experimental evidence is still insufficient to support the effect of this interdependent relationship on mutual cell behaviours. In this study, we aimed to investigate cellular lipid metabolism in chondrocytes induced by osteoblasts. Methods Osteoblast‐induced chondrocytes were established in a Transwell chamber. A cholesterol detection kit was used to detect cholesterol contents. RNA sequencing and qPCR were performed to assess changes in mRNA expression. Western blot analysis was performed to detect protein expression. Immunofluorescence staining was conducted to show the cellular distribution of proteins. Results Cholesterol levels were significantly decreased in chondrocytes induced by osteoblasts. Osteoblasts reduced cholesterol synthesis in chondrocytes by reducing the expression of a series of synthetases, including Fdft1, Sqle, Lss, Cyp51, Msmo1, Nsdhl, Sc5d, Dhcr24 and Dhcr7. This modulatory process involves Notch1 signalling. The expression of ncstn and hey1, an activator and a specific downstream target of Notch signalling, respectively, were decreased in chondrocytes induced by osteoblasts. Conclusions For the first time, we elucidated that communication with osteoblasts reduces cholesterol synthesis in chondrocytes through Notch1 signalling. This result may provide a better understanding of the effect of subchondral bone signalling on chondrocytes.

regulators of lipid tissue, and mammals have developed complex mechanisms to maintain intracellular cholesterol levels within a narrow range. 8 When these homeostatic mechanisms are overwhelmed, various diseases, such as atherosclerosis, occur. The role of lipids in various diseases has been studied for decades. 9 The association between lipids and chondrocytes has become an increasing focus of research interest in recent years. The lipidic microenvironment in human chondrocytes induces oxidative stress and elicits a proinflammatory response, which might be reflected in joint diseases. 10 Shabana et al. reported that the activation of HH signalling in chondrocytes induces cholesterol accumulation, which increases the severity of osteoarthritis (OA) in transgenic mice. 11 The Notch signalling pathway is crucial for the determination of cell-to-cell communication and cell fate during development and is necessary for tissue homeostasis. In skeletal muscle and brain, activation of the Notch signalling pathway enforces the resting state of local adult stem cells, thereby limiting their tissue repair potential and subsequently affecting metabolism. 12,13 In the immune system, activation of Notch signalling promotes the polarization of M1 macrophages, induces a systemic low-grade inflammatory state, and aggravates insulin resistance in peripheral tissues. 14,15 In addition, Notch stabilizes and activates mTorc1, 16 which plays a central role in lipid metabolism, 17  is inseparable for adipogenesis. 17 However, Notch signalling has rarely been studied in chondrocytes.
In the current study, we aimed to study the effect of osteoblastinduced culture on chondrocyte lipid metabolism using osteoblastinduced chondrocytes and chondrocyte monoculture models. We found that osteoblasts reduced the cholesterol content and the expression of the cholesterol synthase seladin-1 by performing qPCR, WB and IF. Additionally, osteoblasts inhibited cholesterol synthesis in chondrocytes through Notch1 signalling. This study will help us further understand how osteoblast-derived signals regulate chondrocyte behaviour. Articular chondrocytes and primary calvarial osteoblasts were isolated from C57BL/6J mice 2-3 days after birth. Briefly, the articular cartilage and skull bone were dissected from the mice, cut into pieces, and enzymatically digested. First, the tissue was trypsinized with a 0.25% protease solution dissolved in Dulbecco's modified Eagle's medium (HyClone) for 30 min. Then, chondrocytes were digested with 0.003% type II collagenase (Sigma-Aldrich) for 12 h, and osteoblasts were digested with 0.006% type I collagenase (Sigma-Aldrich) for 12 h. After centrifugation at 1000 rpm for 8 min, the supernatant was removed, and complete supplemental medium composed of either DMEM, 10% FBS and 1% penicillin-streptomycin or α-MEM, 10% FBS and 1% penicillin-streptomycin was mixed with the two types of cells and tissues. Then, the primary chondrocytes and osteoblasts were seeded into a 25 cm 2 cell culture flask and cultivated at 37 °C with 5% CO 2 in a standard humidified atmosphere.

| Cell culture
Osteoblasts and chondrocytes were used at the second passage in the current study.
Chondrocytes were seeded into a six-well plate, and osteoblasts were seeded on a Transwell chamber with a 0.4 μm pore size to construct a coculture model of osteoblast-induced chondrocytes and chondrocyte monocultures. After the cells adhered, chondrocytes and osteoblasts were equilibrated with DMEM/α-MEM containing 10% FBS for 12 h, substituted with DMEM/α-MEM containing 2% FBS, and starved for 12 h. When finally changing the medium to DMEM/α-MEM containing 1% FBS, the Transwell chamber was placed in a six-well plate. After culture for 3 days, cell lysates (1000 ml) were collected. The cell confluence at the beginning of the coculture was approximately 60%, which was still in the logarithmic growth phase.

| ELISA
We collected chondrocyte culture medium from osteoblast-induced chondrocytes and chondrocyte monoculture models after 72 h.
ELISA kits (CSB-EL001918MO; CSB-EL001712MO, CUSABIO) were used to detect apolipoprotein B and angiopoietin-like 4 levels according to the instruction manual.

| RNA sequencing
Chondrocytes were cultured in the osteoblast induction system and the monoculture system for 12 h and then incubated with 2% FBS and 1% penicillin-streptomycin. The cells were starved in DMEM for 12 h. Then, the medium was changed to DMEM containing 1% FBS and 1% penicillin-streptomycin and cells were cultivated for 72 h. TRIzol was used to collect cell lysates (cell confluence rate of up to 95%). Three independent replicate experiments were performed using chondrocytes extracted from the knee joints of different postnatal C57 mice. The cell samples were sent to Shanghai Lifegenes Biotechnology Co., Ltd. for the transcriptome analysis.
Before transcriptome sequencing, we used the RNA Nano 6000 Assay Kit and the Bioanalyzer 2100 system to evaluate the RNA integrity. According to the manufacturer's instructions, each sample used 1.5 μg of RNA as the input material for RNA detection. A HiSeq 4000 PE Cluster Kit (Illumina) was used to cluster the index-coded samples. The original data were first processed with internal scripts to obtain clean data and then compared with the reference genome using HISAT2 v2.1.0. During data analysis, HTSeq v0.6.1 was employed to calculate the number of reads mapped to each gene. The method used to calculate gene FPKMs was to add the FPKMs of each genomic transcript. The GO enrichment analysis of differentially expressed genes was performed using the DAVID database.
GO terms with a p value less than 0.05 were considered observably enriched in differentially expressed genes. KEGG is a database and resource for the public to understand the advanced functions in biological systems. KOBAS v3.0 software was used to detect the statistically significant enrichment of differentially expressed genes in various KEGG pathways. Differentially expressed genes were significantly enriched for KEGG pathways with a p value less than 0.05 (adjusted p values were used to screen different candidate genes). In the differential expression analysis, p < 0.05 and |FoldChange| ≥ 1.2 were the thresholds for significantly different expression.

| RNA extraction and quantitative realtime PCR
The RNeasyPlus Mini Kit (Qiagen) was used to extract total RNA  Table 1. was incubated with the membrane for 2 h. The Immobilon ® Western (P90719, Millipore) kit was used to visualize immune complexes, and the protein expression levels were analysed with ImageJ software (NIH). In addition, β-actin was used as an internal control.

| Statistical analysis
The results of each analysis are presented as the means ± SD.
Experiments were performed in triplicate (n = 3). Statistical analyses were performed using one-way analysis of variance to determine differences between groups. Fisher's protected least effective difference test was performed as a post hoc analysis. In each analysis, the critical significance level was set to p < 0.05.

| Osteoblasts reduce the intracellular cholesterol content in chondrocytes
We established an osteoblast-induced chondrocyte coculture model and compared it to a normal chondrocyte monoculture model to study the effects of osteoblasts on chondrocytes ( Figure 1A). We first used ELISAs to detect the levels of apolipoprotein B (apoB), which represents the cholesterol content 18 and angiopoietin-like 4 (ANGPTL4), which inhibits lipid metabolism, to explore the effects of osteoblasts on chondrocyte lipid metabolism. 19,20 We observed a significant decrease in apoB and ANGPTL4 levels in chondrocytes induced by osteoblasts (apoB levels were decreased to 12.5% and ANGPTL4 levels were decreased to 65%) ( Figure 1B). We obtained intuitive evidence for the effect of osteoblasts on chondrocyte cholesterol levels by measuring the fluorescence of Filipin III to show the changes in the cholesterol content in chondrocytes. We detected a significantly decreased cholesterol content in chondrocytes induced by osteoblasts ( Figure 1C). The quantitative analysis of fluorescence further confirmed this decrease ( Figure 1D).

| Osteoblasts decrease the gene expression profile of the cholesterol metabolic pathway in chondrocytes
We performed RNA sequencing to precisely determine the gene changes (Dhcr7) and 24-dehydrocholesterol reductase (Dhcr24, also known as seladin-1) mRNAs were all consistent with those identified using RNA sequencing ( Figure 2C). We also showed the results for changes in the expression of genes in the brassinosteroid biosynthesis pathway and other steroid biosynthesis pathways ( Figure S1), but the expression of these genes was not significantly changed.

| Osteoblasts modulate the expression of the final synthetase, seladin-1, in cholesterol synthesis in chondrocytes
Seladin-1 is a key factor involved in the last step of cholesterol synthesis. 21 The results described above confirmed its expression at the mRNA level (Dhcr24, also known as seladin-1, Figure 2A Figure 3F).

| Osteoblasts downregulate Notch1 signalling to mediate cholesterol synthesis in chondrocytes
We analysed the RNA sequencing data and identified the downregulation of signalling pathways in chondrocytes induced by osteoblasts.
From the pheatmap, we detected changes in the top 16 signalling pathway candidates in chondrocytes induced by osteoblasts ( Figure 4A).
Among them, Notch1 and its downstream target gene, Hes-related family bHLH transcription factor with YRPW motif 1 (hey1), were significantly downregulated. The inhibition of canonical Notch signalling inhibits adipogenesis, 22 whereas activation of this pathway stimulates adipogenesis. 23 Thus, the Notch signalling pathway is considered an important regulator of chondrocyte cholesterol metabolism induced by osteoblasts. Using western blotting, we detected changes in the expression of Notch1 at the protein level, consistent with the results at the mRNA level ( Figure 4B). Based on the quantitative analysis, the level of the Notch1 protein in chondrocytes was confirmed to be significantly decreased after osteoblast induction. Notch1 expression in osteoblast-induced chondrocytes was reduced to as low as 50% relative to that in monocultured chondrocytes ( Figure 4C). Using immunofluorescence staining and CLSM, we observed a decreased distribution of Notch1 in osteoblast-induced chondrocytes ( Figure 4D).
Further quantification confirmed the total changes in Notch1 immunofluorescence in chondrocytes induced by osteoblasts ( Figure 4E).

| Osteoblasts decrease the expression of ncstn and hey1 to impair cholesterol synthesis in chondrocytes
We aimed to further determine the importance of Notch1 signal-  Figure 5D). According to the linear fluorescence quantification, we observed a dramatic reduction in ncstn levels in the nuclear region of chondrocytes induced by osteoblasts ( Figure 5E). At the same time, we detected hey1 levels using western blotting and observed a significant decrease in its levels in chondrocytes induced by osteoblasts ( Figure 5F,G). Using CLSM, we detected a substantial decrease in hey1 levels in the nuclear region ( Figure 5H). Total fluorescence quantification showed a decreased amount of hey1 in chondrocytes induced by osteoblasts ( Figure 5I). Based on the linear fluorescence quantification, we observed a dramatic reduction in hey1 levels in the nuclear region of chondrocytes induced by osteoblasts ( Figure 5J). Taken together, these results indicate the potential role of Notch1 signalling in mediating osteoblast-induced cholesterol synthesis in chondrocytes. ApoB, which is usually present in the form of apolipoprotein B100 and apolipoprotein B48, is a large amphiphilic protein. 27 It is located on the surface of low-density lipoprotein. ApoB mainly recognizes and promotes the uptake of LDL by cells, and thus it is an indispensable part of lipoprotein metabolism. Since LDL is a lipoprotein enriched in cholesterol, the apoB content potentially reflects the cholesterol content. 18 ANGPTL4 is also called peroxisome proliferator-activated receptorγ, hepatic fibrinogen or fasting-induced adipokine. 28 The main role of ANGPTL4 is to regulate lipid metabolism and angiogenesis. The role of ANGPTL4 in lipid metabolism is to inhibit lipoprotein lipase activity, increase plasma triacylglycerol and nonesterified fatty acid levels and then reduce adipose tissue reserves. 21 Another study found that ANGPTL4 is only expressed at low levels in normal cartilage, and the amount of ANGPTL4 also represents the number of cartilage cells to a certain extent. 28 In summary, we propose that the level of this protein may indicate a decrease in cholesterol levels. Therefore, we used Filipin III to specifically stain cholesterol and found that the cholesterol content in chondrocytes was indeed reduced after osteoblast induction.

| DISCUSS ION
Steroid compounds are widely found in animals and plants, and they play extremely important roles in regulating biological activities in the lives of animals and plants. 29 while the level of seladin-1 in the cytoplasm was significantly reduced ( Figure 3D,E). This change is similar to the phenomenon mentioned in a previous study: seladin-1, which is normally confined to the perinuclear cytoplasmic region, is localized in the nuclei after oxidative challenge. 36 40 Actinomycin D is a ribosomal stress-inducing agent, and ribosomal stress is related to the cell cycle, apoptosis and other processes. 41 We speculate that osteoblasts may exert this effect on chondrocytes. More detailed studies are needed to confirm this assumption.
In summary, our research is the first to reveal the molecular mechanism by which osteoblasts inhibit cholesterol synthesis in chondrocytes and confirms that osteoblasts play a regulatory role through Notch1 signalling and seladin-1. This research has improved our comprehension of the molecular mechanisms by which osteoblast-derived signals regulate chondrocyte behaviour.

ACK N OWLED G EM ENTS
We acknowledge financial support from the National Nature

CO N FLI C T O F I NTE R E S T
The authors have no competing interests to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
Any data generated in this study are available from the corresponding author upon request.