TGF‐β1 promotes gap junctions formation in chondrocytes via Smad3/Smad4 signalling

Abstract Objectives Connexin‐mediated functional gap junction intercellular communication (GJIC) has a vital role in development, homeostasis and pathology. Transforming growth factor‐β1 (TGF‐β1), as one of the most vital factors in chondrocytes, promotes cartilage precursor cell differentiation and chondrocyte proliferation, migration and metabolism. However, how TGF‐β1 mediates GJIC in chondrocytes remains unclear. This study aims to determine the influence of TGF‐β1 on GJIC in mouse chondrocytes and its underlying mechanism. Methods qPCR and mRNA microarray were used to verify the expression of genes in the TGF‐β and connexin families in cartilage and chondrocytes. A scrape loading/dye transfer assay was performed to explore GJIC. Western blot analysis was used to detect connexin43 (Cx43) and Smad signalling components. Immunofluorescence staining was performed to characterize protein distribution. Results The TGF‐β1 mRNA was the highest expressed member of the TGFβ super family in cartilage. TGF‐β1 promoted functional GJIC through increased expression of Cx43. TGF‐β1‐mediated GJIC required the participation of TGF‐β type I receptor. TGF‐β1 activated Smad3 and Smad4 signalling to facilitate their nuclear translocation. The Smad3 and Smad4 signalling proteins bound to the promoter of Gja1 and thus initiated Cx43 gene expression. Conclusions For the first time, these results revealed a vital role of TGF‐β1 in cell‐cell communication in chondrocytes via gap junction formation. We describe the regulatory mechanism, the involvement of TGF‐β type I receptor and the nuclear translocation of Smad3/4.

hemichannels, which are single-membrane diffusion channels that allow passage of signalling molecules smaller than 1.2 kDa, such as metabolites, ATP, ions, second messengers and cyclic nucleotides. 2,3 Gap junctions are formed by connexin subunits. To date, there are at least 21 connexin family members in human and 20 in mice. 4 The expression patterns of connexins differ according to cell type and function. Among the connexin family members, connexin43 (Cx43), encoded by the Gja1 gene, is the main component of gap junctions in human tissue. It is widely expressed in bone cell types and is most abundant in osteoblastic cells, which are characterized by their distinct function in the skeletal maturation stage. 5,6 Cx43 has a substantial effect on chondrocyte mechanotransduction and cell cycle. 7,8 Accumulating evidence suggests that Cx43 hemichannels mediate homeostasis in cartilage and that impaired Cx43 hemichannels contribute to the aetiology of joint diseases. 9 The TGF-β super family is a large collection of secreted and structurally related proteins that are responsible for numerous cellular processes and have significant effects on the development and homeostasis in mammals. 10,11 This super family contains 33 members in the mammalian genome, which include the TGF-β subfamily, the activins and inhibins, the bone morphogenetic proteins (BMPs) and the growth and differentiation factors (GDFs). 12 Transforming growth factor-β1 (TGF-β1), a critical chondrocyte factor, plays a vital role in regulating cell proliferation, migration, differentiation, ECM formation and other developmental processes. [13][14][15] However, whether TGF-β1 induces cell-cell communication in chondrocytes needs to be further investigated.
In the current study, we used recombinant TGF-β1 to establish the relationship between TGF-β1 and gap junctions in chondrocytes. To investigate the effect of TGF-β1 on cell-cell communication, changes in cell migration and proliferation, gap junction formation and its regulator Cx43 were detected. Furthermore, we provided direct evidence that the TGF-β type I receptor and Smad3/4 signalling factors are involved in TGF-β1-mediated induction of Cx43. Articular cartilage tissue was collected from 8-week-old C57BL mice. Briefly, the articular cartilage was isolated, rinsed and immediately crushed into powder using liquid nitrogen. The fine residue was gathered for PCR assays. For chondrocytes, articular cartilage from newborn mice was shredded into small pieces and rinsed twice in 1 × PBS. Cartilage was trypsinized in 0.25% protease solution dissolved in Dulbecco's modified Eagle's medium (high-glucose DMEM, 0.1 mmol/L non-essential amino acids, 4 mmol/L l-glutamine, 1% penicillin-streptomycin solution, HyClone, Logan, UT, USA) for 30 minutes at 37°C and was transferred into 0.5% type II collagenase to digest overnight. Next, the type II collagenase-treated solution was neutralized 1:1 (v/v) with 10% heat-inactivated foetal bovine serum (FBS) in DMEM (high-glucose DMEM, 0.1 mmol/L non-essential amino acids, 4 mmol/L l-glutamine, 1% penicillin-streptomycin solution), and the suspension was centrifuged at 157 g (relative centrifugal force) for 5 minutes. The remaining chondrocytes were re-suspended in 10% FBS DMEM with 1% penicillin-streptomycin and then seeded into plates or T25 flasks and incubated in a 5% CO 2 incubator at 37°C. Passages 1-2 were used in the experiments.

| RNA extraction and quantitative real-time PCR
Cells were washed with cold PBS, and total RNA was extracted from

chondrocytes (cells) and joint residue (tissue) with the RNeasyPlus Mini
Kit (Qiagen, Shanghai, China) with a genomic DNA eliminator according to the manufacturer's instructions. The extracted RNA samples were dissolved in RNase-free water and quantified by a spectrophotometer instrument measuring absorbance at 260 nm. After treatment with DNase I, RNA (0.5-2 μg) was reverse-transcribed to cDNA with the first-strand cDNA synthesis kit (Mbi, Glen Burnie, MD, USA).
The sets of primers used to detect each gene in the TGF-β super family are listed in Table 1. The BLAST programme was used to determine all primer sequences, and all samples were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Quantitative real-time PCR (qPCR) was performed by amplifying a specific product with a reaction mixture containing cDNA, SYBR Green I PCR master mix (Takara, Tokyo, Japan) and primer pairs, and qPCR was run on an ABI 7300 instrument (Applied Biosystems, Shanghai, China) equipped with 96-well optical reaction plates. The reaction conditions are as follows: 95°C for 10 minutes followed by 40 cycles of 95°C for 5 seconds and 60°C for 30 seconds. Each gene was assayed in triplicate. The fold change calculations were performed with the 2 −ΔΔC t method.

| Protein extraction and Western blot
Cell lysates were collected in lysis buffer containing PMSF (P7626; Sigma) protease inhibitor. The protein concentration was evaluated with a BCA Protein Assay.
at room temperature for 2 hours. Immunoreactive blots were developed using the Super Signal reagents (Pierce, Rockford, IL, USA). ImageJ software (NIH, Bethesda, MD, USA) was used for densitometric analyses.

| Cell Counting Kit-8
Cells were seeded in a 96-well plate at a density of 1.5 × 10 3

| Scratch wound assay
We scratched a linear wound through cells in a 35-mm cell culture dish with a pipette tip and rinsed three times with 1 × PBS to remove cell debris. The culture medium was then replaced by media containing different concentrations of TGF-β1 to investigate its effects on chondrocyte migration. Images were captured at 0, 12 and 24 hours following scraping. The mobility ratio was evaluated by dividing the area of the migrated cell region by area of the scraped region.

| Scrape loading and dye transfer
The functional gap junctions between chondrocytes were evaluated with a scrape loading and Lucifer yellow transfer assay. Briefly, cells were seeded in dishes and grown to confluency. The medium was dis-

| Statistical analysis
All results are representative of at least three independent experi- the means of all groups. Data are presented as the mean ± SEM.
Significant differences were defined as P < 0.05.

| Expression of the TGF-β super family genes in chondrocytes
The TGF-β super family contains many secreted and structurally related proteins. The 33 members in the mammalian genome, including the TGF-β isoforms, activins and inhibins, GDFs and BMPs 16 were assessed in the current study. Based on our mRNA microarray data, we found that 28 members were expressed in chondrocytes. To confirm the mRNA expression levels of these genes, qPCR was used. The expression levels of 28 members in knee articular cartilage in vivo ( Figure 1A) and primary chondrocytes in vitro ( Figure 1B) are shown in descending order. The results indicated that TGF-β1 mRNA was the most abundant in cartilage (up to 310 × 10 −4 -fold change compared to the internal GAPDH control, Figure 1A). Even in the isolated primary were relatively low compared to the genes mentioned above.

| TGF-β1 promotes gap junction intercellular communication in chondrocytes
We first performed a scratch wound assay to determine the impact of TGF-β1 on the migration and proliferation of chondrocytes showed extensive Lucifer yellow transfer efficiency ( Figure 2C).
This increase in Lucifer yellow transfer efficiency was dose-dependent ( Figure S2). Moreover, the increase in Lucifer yellow transfer efficiency was cell density-dependent in both the control and the TGF-β1 group. The transmission speed of Lucifer yellow dye in the TGF-β1 group was 8-fold higher than that in the control group ( Figure 2D).

| TGF-β1 promotes gap junction formation through increased expression of Cx43 in chondrocytes
Gap junctions consist of two paired hemichannels with connexin subunits. 17 We found there were 17 connexin family members expressed in chondrocytes by mRNA microarray analysis ( Figure 3A). The mRNA level of Cx43 was the highest (up to 2970 × 10 −4 -fold change compared to the internal GAPDH control), followed by Cx45 (up to 174 × 10 −4 -fold change), and Cx29 (up to 110 × 10 −4 -fold change). The remaining connexin members all displayed a relatively low gene expression (<210 × 10 −4 -fold change).
To determine the effect of TGF-β1 on Cx43 expression, the primary chondrocytes were treated with recombinant TGF-β1 in different concentrations (1, 5 and 10 ng/mL). As shown in Figure 3B, the protein levels of Cx43 were significantly upregulated in a dosedependent manner both at the early stage (12 hours) and late stage (72 hours) as detected by Western blot assays ( Figure 3B). Using 10 ng/mL TGF-β1, we further showed that TGF-β1 increased the expression of Cx43 in a time-dependent manner ( Figure 3C). Cx43 in primary chondrocytes showed multiple bands, representing different phosphorylation states of Cx43, with approximate molecular masses approximately 43 kDa, in agreement with the findings of several labs. 18,19 In untreated chondrocytes, Cx43 is primarily distributed to the nucleus. Cx43 translocation from the nucleus to the cytoplasm

| TGF-β type I receptor is required for TGF-β1-modulated gap junction formation and Cx43 cytoplasmic translocation
Transforming growth factor-β1 signal transduction mainly relies on TGF-β type I receptor/ALK5 to activate the downstream signal of R-Smad phosphorylation. 20 To confirm the role of TGF-β type I receptor (TβRI) in TGF-β1-induced gap junctions in chondrocytes, we conducted an inhibition experiment using a specific ALK5 inhibitor, repsox, 21 Figure 4C, boxed area). These results were further confirmed with a fluorescence optical density assay ( Figure 4D).

F I G U R E 4
Repsox pre-incubation attenuated TGF-β1-induced gap junction formation and Cx43 expression. A, Repsox pre-incubation attenuated TGF-β1-induced gap junction formation. The concentration of TGF-β1 was 10 ng/mL. The repsox used was at 100 μmol/L for a 12 hours pre-incubation with primary chondrocytes. The images were based on three independent experiments observed with CLSM (n = 3). B, Quantitative analysis of the cell transmission speed of Lucifer yellow dye with respect to the control and TGF-β1-induced group (n = 3). *P < 0.05; **P < 0.01. C, Immunofluorescence staining of Cx43 (red) in chondrocytes after a 12-hour treatment with TGF-β1 (10 ng/ mL) in the presence or absence of repsox (100 μmol/L). The chondrocytes were counterstained with markers of the nucleus (DAPI, blue) and actin cytoskeleton (phalloidin, green). The images were based on three independent experiments observed by CLSM (n = 3). D, Fluorescence optical density (OD) was analysed in Image Pro Plus 6.0 to measure the specific distribution of Cx43 cross the cell body. Data analysis was performed on at least 40 cells per group. CLSM, confocal laser-scanning microscopy; Cx43, connexin43; TGF-β, Transforming growth factor-β1

| TGF-β1 modulates chondrocyte gap junctions through potential binding sites of downstream Smad signalling molecules in the promoter of Cx43
Following TGF-β type I receptor activation, we aimed to elucidate the inner mechanism of TGF-β1-modulated gap junction formation in chondrocytes. We first showed that TGF-β1 induced the expression of Smad3 and Smad4 by Western blotting (Figure 5A), and densitometry confirmed the increase ( Figure 5B). With the TβRI inhibitor, we next found that repsox reduced the expressions of Smad3, Smad4 and Cx43 ( Figure 5C). This synchronous change in Smad signalling components and Cx43 led us to consider their interactions. We found that TGF-β1 induces Smad3 and Smad4 accumulation in chondrocyte nuclei by CLSM ( Figure 5D). Using bioinformatics, we found a potential  Figure 5E). Importantly, the Smad4-binding sites were shared with Smad3 (same location and same sequences).
The results indicated that Smad3 and Smad4, which can form a protein complex in the cytoplasm, 22 translocate into the nucleus to synchronously initiate Gja1 expression at the same binding sites. Little is known about connexins in cartilage, but Cx43 was reported to be a highly expressed connexin in chondrocyte cells, while Cx45, Cx32

| D ISCUSS I ON
and Cx46 have also been detected. [27][28][29][30] In this study, we screened the connexin family and found that 17 connexin genes are expressed in mouse chondrocyte cells, with Cx43 being the most abundant.
F I G U R E 6 Schematic diagram illustrates the mechanism of TGF-β1 regulation of Cx43 in chondrocytes. The green lines point to Smaddependent pathways elucidated by the current study, and the blue dotted lines are potential mechanisms not included in the present study. TGF-β1 induced the canonical Smad signalling pathway, the activated TβRI phosphorylates the R-Smads (Smad2, Smad3), transform them into transcriptional co-regulator together with Smad4, and dock to potential binding site of promoter of Gja1 which encodes Cx43. As a result, Cx43 accumulated more on cell border and phosphorylated into functional gap junctions with larger size. Cx43, connexin43; TGF-β, Transforming growth factor-β1 A more detailed examination by confocal microscopy showed, in addition to the upregulation of Cx43 expression levels, TGF-β1 treatment altered the Cx43 distribution in sub-cellular structures of chondrocyte cells. Our study detected a dynamic and time-sensitive delivery of Cx43-containing connexins. Under TGF-β1 stimulation, changes in focal Cx43 patterns ( Figure 3D) were observed as early as 2 after treatment.
Initially, Cx43 foci accumulated outside of the nucleus and then translocated from cytoplasmic structures to the cell membrane and finally into the intermembrane cytoplasmic area between cells with protruding lamellipodia ready to dock onto the mature gap junction plaques in a time-dependent manner. This dynamic localization change also confirmed the known lifecycle of Cx43; Cx43 is first synthesized and transferred through the endoplasmic reticulum, and then it is oligomerized into connexin hemichannels in the trans-Golgi network, and finally, it is laterally accreted from the plasma membrane into the gap junctions with increasing abundance over time ( Figure 3D yellow arrow) in accordance with previous studies. 31,32 In the present study, cell proliferation was enhanced as TGF-β1 stimulated the Cx43 expression pattern.
We speculate that besides GJ formation, Cx43 may have a dual role in controlling cell proliferation. Lamiche et al 33 connected the divergent Cx43 functions with differential localizations at either the membrane or the cytoplasm, as examined by cell proliferation assays. Specifically, the C-terminal tail of Cx43 contains a nuclear targeting sequence that can directly mediate gene transcription of cell growth and cell death regulators. 34,35 Further studies are required to test this hypothesis.
Chondrocytes are the only cells found in healthy knee hyaline cartilage. Chondrocytes are found in small lacunae isolated from each other, and diffusion of substances may be the primary method of communication. 36 The present results show that under monoculture conditions, chondrocytes connect directly with each other by forming gap junctions, which was confirmed by LY uptake ( Figure 2B)  Figure 3D).
To understand the potential mechanism of TGF-β-induced Cx43 regulation, we examined possible signalling molecules in the TGF-β1 pathway. TGF-β1-mediated signalling is transduced at the surface through a complex of type I (TβRI) and type II (TβRII) receptors, which phosphorylate receptor-activated Smads (R-Smads) 3 and 4 and initiate Smad translocation into the nucleus where they ultimately regulate target genes such as transcriptional activators. 22 To date, besides the canonical Smad-dependent pathway, TGF-β signalling also participates in a Smad-independent signalling pathway that includes various branches of the mitogen-activated protein kinase (MAPK) pathway. 20,41 In the present study, pre-treatment with a strong TβRI/ALK5 inhibitor abolished the effect of TGF-β1 on Cx43 upregulation and gap junction formation, indicating the involvement of the ALK5 receptor. TGF-β1 significantly upregulated the expression of Smad3 and Smad4 as confirmed by Western blot analysis. In addition, TGF-β1 stimulation also affected the sub-cellular localization of Smad; immunofluorescence analysis detected Smad3 and Smad4 accumulation in the nucleus. Interesting, Smad4 seemed to co-localize the cytoskeleton. Further, a bioinformatics database revealed that Smad3 and Smad4 occupy two identical binding sites in the promoter of the Gja1 gene, and this binding may promote Cx43 expression in accordance with TGF-β1-induced Cx43 expression in chondrocytes. Thus, it is possible that TGF-β1mediated Cx43 signalling occurs via the canonical Smad pathway.
The mechanism of this pathway is illustrated in Figure 6.
In conclusion, the current study revealed the molecular mechanisms

ACK N OWLED G EM ENTS
This study was supported by NSFC grants (81600840, 81771047) to Jing Xie. We acknowledge Dr Chenghui Li of the Analytical and Testing Centre of Sichuan University for her support in CLSM imaging.

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