CCN4/WISP1 Promotes Migration of Human Primary Osteoarthritic Chondrocytes

Abstract

Objectives

Previously, we have shown the involvement of cellular communication network factor 4/Wnt-activated protein Wnt-1-induced signaling protein 1 (CCN4/WISP1) in osteoarthritic (OA) cartilage and its detrimental effects on cartilage. Here, we investigated characteristics of CCN4 in chondrocyte biology by exploring correlations of CCN4 with genes expressed in human OA cartilage with functional follow-up.

Design

Spearman correlation analysis was performed for genes correlating with CCN4 using our previously established RNA sequencing dataset of human preserved OA cartilage of the RAAK study, followed by a pathway enrichment analysis for genes with ρ ≥|0.6.| Chondrocyte migration in the absence or presence of CCN4 was determined in a scratch assay, measuring scratch size using a live cell imager for up to 36 h. Changes in expression levels of 12 genes, correlating with CCN4 and involved in migratory processes, were determined with reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

Results

Correlation of CCN4 with ρ ≥|0.6| was found for 58 genes in preserved human OA cartilage. Pathway analysis revealed “neural crest cell migration” as most significant enriched pathway, containing among others CORO1C, SEMA3C, and SMO. Addition of CCN4 to primary chondrocytes significantly enhance chondrocyte migration as demonstrated by reduced scratch size over the course of 36 h, but at the timepoints measured no effect was observed on mRNA expression of the 12 genes.

Conclusion

CCN4 increases cell migration of human primary OA chondrocytes. Since WISP1 expression is known to be increased in OA cartilage, this may serve to direct chondrocytes toward cartilage defects and orchestrate repair.

Introduction

A prominent hallmark of osteoarthritis (OA) is progressive degradation of articular cartilage.¹ The articular cartilage is a highly specialized connective tissue required for smooth movement of the joints, with chondrocytes being the sole cell type present.² Cartilage has limited regenerative capacity after incurred damage. In the usual process of wound healing, neighboring cells proliferate and migrate toward the wound to initiate repair.³ In cartilage, however, damaged sites are not healed effectively, which is likely due to the lack of neighboring chondrocytes. Due to the vast density and pressure of the extracellular matrix (ECM), and the presence of a unique pericellular matrix (PCM) surrounding the chondrocyte, chondrocytes are very limited in their ability to migrate.⁴ That chondrocytes are able to migrate after damage occurs has been demonstrated in in vitro and ex vivo studies.^{5 -7} In OA cartilage, chondrocyte clusters are observed near lesioned sites, but are thought to be mainly a result of proliferation, rather than migration.⁴ Therefore, promoting chondrocyte migration in damaged articular cartilage toward cartilage lesions could be beneficial to cartilage regeneration and remodeling.

Wnt signaling is an important pathway that can regulate cell migration.^8,9 This pathway is involved in almost every aspect of embryonic development and controls tissue homeostasis later in life.¹⁰ Dysregulation of Wnt signaling has been associated with severe skeletal disorders. Over the years, it has become clear that dysregulated Wnt signaling also plays a prominent role in OA development and progression.^{11 -14} In several experimental mouse models of OA, we observed strong upregulation of several Wnt ligands in the synovial tissue.¹⁵ Likewise, we further demonstrated that expression of the canonical cellular communication network factor 4/Wnt-activated protein Wnt-1-induced signaling protein 1 (CCN4/WISP1) was upregulated in both the synovium and the articular cartilage of mice that developed OA.¹⁵ In macroscopically lesioned areas compared to preserved areas of articular cartilage from human OA joints, CCN4 was upregulated, positively correlating with the Mankin score.¹⁴ Moreover, we demonstrated that in humans, CCN4 expression was epigenetically regulated via DNA methylation, and that high CCN4 levels were detrimental for cartilage integrity.¹⁴ Furthermore, CCN4 has been recognized as an oncogene, controlling cell survival, migration, and proliferation in tumorigenic cells.¹⁶

In the present study, we aimed to further characterize functions of CCN4, relevant to OA. Thereto we explored RNA sequencing data on preserved and lesioned OA articular cartilage¹⁷ to identify genes correlating with CCN4 expression in preserved OA cartilage, on which we performed pathway analysis. This revealed enrichment for genes involved in cell migration, which was then tested in a scratch assay with human primary articular chondrocytes.

Methods

Correlation Analysis with CCN4 in Preserved OA Articular Cartilage

Genome-wide correlations (ρ ≥ 0.6) between mRNA levels of CCN4 and other genes were calculated in preserved human OA cartilage in our previously established RNA sequencing dataset on human OA cartilage derived from knee and hip joints of patients that are enrolled in the research osteoarthritis and articular cartilage (RAAK) study.¹⁷ The online tool DAVID was used to perform pathway enrichment on the genes correlating with CCN4 with genes expressed in human OA cartilage (RAAK) as a background, selecting the gene ontology (GO) terms Biological Processes (GOTERM_BP_DIRECT), Cellular Component (GOTERM_CC_DIRECT), and Molecular Function (GOTERM_MF_DIRECT).^17,18

Cell migration Assay

To investigate the effects of CCN4 on chondrocyte migration, primary chondrocytes were isolated from preserved regions of articular cartilage from human OA knee joints, and digested O/N at 37 °C in 1.5 mg/ml collagenase B (Roche, Basel, Switzerland) dissolved in DMEM GlutaMax™ (Gibco, Waltham, MA), supplemented with 100 mg/l sodium pyruvate (Invitrogen, Carlsbad, CA), 100 U/ml penicillin and 100 mg/ml streptomycin (pen/strep, Lonza, Basel, Switzerland). Cells were expanded for two passages in medium supplemented with 10% FCS (Sigma, St. Louis, MO), sodium pyruvate, and pen/strep, as described above. After two passages, cells were seeded in 24-well plates (N = 6 donors) with a density of 8.75 x 10⁴ cells/well in medium with 0.1% FCS, sodium pyruvate and pen/strep. The next day, a thin scratch was made using a sterile 200 µl pipette tip. Three groups were compared: (1) Control medium, (2) medium with 0.1 µg/ml rhCCN4 (Peprotech, London, UK), and (3) medium with 0.5 µg/ml rhCCN4. The migration of cells was recorded by time-lapse imaging for 36 h using a Zeiss Axiovert 200M microscope (Boston Microscopes, Wilmington, MA) with a stage incubator (Okolab, Ottaviano, Italy). The distance between the edges of the scratch was measured at four timepoints using Fiji-ImageJ v1.52. Chondrocyte numbers were measured at endpoint by fixing the cells in 4% paraformaldehyde and staining them with crystal violet. Crystal violet stained cells were dissolved in 10 v/v % acetic acid and absorbance was read at 590 nm using a CLARIOstar Plus plate reader (BMG Labtech, Ortenberg, Germany).

Reverse transcription quantitative real-time PCR (RT-qPCR)

Chondrocytes (N = 9 donors) were collected in TRIzol reagent (Thermo Fisher Scientific, Waltham, MA) 48 h after addition of CCN4, and RNA was extracted according to manufacturer’s protocol. DNA contamination was removed using DNase I (Thermo Fisher Scientific, Waltham, MA) and RNA was reverse transcribed into cDNA using 1.9 μL ultra-pure water, 2.4 μL 10 × DNAse buffer, 2.0 μL 0.1 M DTT, 0.8 μL 25 mM dNTP, 0.4 μg oligo dT primer, and 1 μL 200 U/μL M-MLV reverse transcriptase (Thermo Fisher Scientific, Waltham, MA), and 0.5 μL 40 U/μL RNAsin (Promega, Madison, WI). cDNA was amplified using Power SYBR™ Green PCR Master Mix (Applied Biosystems™, Foster City, CA) and mRNA expression was measured using the Quantstudio™ 1 Real-Time PCR System (Applied Biosystems™, Foster City, CA). -ΔCt expression levels were calculated using two housekeeping genes GAPDH and SDHA, with the following formula: ΔCt = Ct_{gene of interest}—Ct_{average housekeeping genes}. Fold changes were calculated using the 2^-ΔΔCt method with ΔΔCt = ΔCt_treated—ΔCt_Control. Primer sequences are listed in Table 1.

Table 1.

Primer Sequences.

Gene	Forward sequence	Reverse sequence
GAPDH	5’-TGCCATGTAGACCCCTTGAAG-3’	5’-ATGGTACATGACAAGGTGCGG-3’
SDHA	5’-TGGAGCTGCAGAACCTGATG-3’	5’-TGTAGTCTTCCCTGGCATGC-3’
CCND1	5’-ATCAAGTGTGACCCGGACTG-3’	5’-CTTGGGGTCCATGTTCTGCT-3’
CLMN	5’-TCATCGGGCAGATTAGCGAC-3’	5’-AGAGGTGGGTTGCACTTTTCT-3’
COL6A3	5’-ATGAACAAGCAGGACGTCGT-3’	5’-GCAGAGCTGACCAGGATGTT-3’
CORO1C	5’-GGACTGCACGGTCATGGTAT-3’	5’-CCACGATGCCGACTCTCTTT-3’
EML1	5’-AGAGGAAACCGGAATCGCAC-3’	5’-ACATAGCCTTCTTCTGCACTG-3’
MXRA5	5’-ATGTTGCAGAAGGTCGCAGA-3’	5’-TTTTCCCACGGACTTTGGCT-3’
PLXNA2	5’-GGAAAAGCAACTGCCTCCCT-3’	5’-ATGCTGCCCTTGTGGATGTC-3’
PTPRK	5’-CGATGAGCAGTGTGGAGAAGG-3’	5’-GTTCTTCTGTTGCTGCTGCTTTTG-3’
RAI14	5’-GAGGGCAAGACCGCTTTCCA-3’	5’-GTCCGGTAGTATCTTGGGCTG-3’
SEMA3C	5’-CAAAGATCCCACACACGGCT-3’	5’-ACTTGGTCCTCTGATCTCCTCC-3’
SMO	5’-CTGCTCATCTGGAGGCGTAC-3’	5’-GCTTAGAGAAGGCCTTGGCA-3’
TNFAIP6	5’-TCCATATGGCTTGAACGAGCA-3’	5’-GCCTTAGCTTCTGCGTAGGT-3’

Statistical Analysis

Transcriptome-wide correlations with WISP1 in preserved OA articular cartilage were calculated using a Spearman correlation with a Benjamini-Hochberg test to correct for multiple testing, and were carried out using R. Statistical differences of the distance between the edges of the scratch over time were tested using a two-way analysis of variance (ANOVA) with a Dunnett’s test to correct for multiple comparisons. Differences in scratch closure rate, chondrocyte numbers and mRNA expression levels were tested with a one-way ANOVA and are described using the mean ± difference between 95% confidence interval (CI) and the mean. Statistics and creation of the graphs were carried out in GraphPad Prism 9.0.0. P-values ≤ 0.05 were considered statistically significant.

Results

Correlation Analysis for CCN4/WISP1 in Preserved Human OA Cartilage

To characterize new functions of CCN4 in human cartilage, we explored our previously established RNA sequencing dataset of macroscopically preserved OA articular cartilage¹⁷ to identify genes correlating with CCN4. We identified 58 genes with a ρ ≥ 0.6 that significantly correlated with CCN4 (Table 2). Echinoderm microtubule-associated protein-like 1 (EML1, ρ = 0.7521, P = 1.90 x 10^-8) showed the strongest and most significant correlation with CCN4. Other notable correlations were with Collagen type VI alpha 3 (COL6A3, ρ = 0.6574, P = 1.78 x 10^-5), a collagen exclusively found in the pericellular matrix of articular cartilage¹⁹ and Smoothened (SMO, ρ = -0.664, P = 1.24 x 10^-5), a part of Hedgehog signaling which is implicated in the pathogenesis of OA.²⁰

Table 2.

Genes Correlating with CCN4 (ρ ≥|0.6|) in Preserved OA Articular Cartilage.

Ensembl ID	Gene Name	ρa	P-valueb
ENSG00000066629	EML1	0.7667	7.87 x 10-8
ENSG00000131389	SLC6A6	0.7521	1.90 x 10-7
ENSG00000184905	TCEAL2	–0.6796	7.15 x 10-6
ENSG00000152894	PTPRK	0.6793	7.24 x 10-6
ENSG00000110880	CORO1C	0.6782	7.59 x 10-6
ENSG00000134531	EMP1	0.6756	8.45 x 10-6
ENSG00000149948	HMGA2	0.6752	8.62 x 10-6
ENSG00000169439	SDC2	–0.6751	8.66 x 10-6
ENSG00000213694	S1PR3	0.6744	8.90 x 10-6
ENSG00000136052	SLC41A2	0.6737	9.18 x 10-6
ENSG00000075223	SEMA3C	0.6723	9.74 x 10-6
ENSG00000142149	HUNK	0.6716	1.00 x 10-5
ENSG00000079156	OSBPL6	0.6699	1.07 x 10-5
ENSG00000120278	PLEKHG1	0.6695	1.09 x 10-5
ENSG00000165959	CLMN	0.6666	1.23 x 10-5
ENSG00000128602	SMO	–0.6664	1.24 x 10-5
ENSG00000267100	ILF3-DT	–0.6653	1.30 x 10-5
ENSG00000070882	OSBPL3	0.6616	1.51 x 10-5
ENSG00000038945	MSR1	0.6598	1.62 x 10-5
ENSG00000133816	MICAL2	0.6580	1.74 x 10-5
ENSG00000163359	COL6A3	0.6574	1.78 x 10-5
ENSG00000185989	RASA3	0.6560	1.88 x 10-5
ENSG00000145685	LHFPL2	0.6555	1.92 x 10-5
ENSG00000175352	NRIP3	0.6526	2.15 x 10-5
ENSG00000164237	CMBL	–0.6496	2.42 x 10-5
ENSG00000253293	HOXA10	–0.6485	2.52 x 10-5
ENSG00000130508	PXDN	0.6471	2.66 x 10-5
ENSG00000106351	AGFG2	–0.6445	2.93 x 10-5
ENSG00000211445	GPX3	–0.6412	3.32 x 10-5
ENSG00000118257	NRP2	0.6409	3.36 x 10-5
ENSG00000134369	NAV1	0.6384	3.69 x 10-5
ENSG00000110092	CCND1	0.6373	3.84 x 10-5
ENSG00000076356	PLXNA2	0.6361	4.01 x 10-5
ENSG00000272168	CASC15	0.6352	4.15 x 10-5
ENSG00000107249	GLIS3	0.6317	4.71 x 10-5
ENSG00000244682	FCGR2C	0.6295	5.10 x 10-5
ENSG00000171017	LRRC8E	0.6288	5.23 x 10-5
ENSG00000260314	MRC1	0.6277	5.43 x 10-5
ENSG00000237452	BHMG1	–0.6256	5.86 x 10-5
ENSG00000167191	GPRC5B	–0.6216	6.74 x 10-5
ENSG00000163517	HDAC11	–0.6216	6.74 x 10-5
ENSG00000146411	SLC2A12	0.6212	6.83 x 10-5
ENSG00000171488	LRRC8C	0.6207	6.94 x 10-5
ENSG00000236609	ZNF853	–0.6207	6.94 x 10-5
ENSG00000156273	BACH1	0.6171	7.87 x 10-5
ENSG00000115468	EFHD1	–0.6165	8.02 x 10-5
ENSG00000170275	CRTAP	–0.6162	8.10 x 10-5
ENSG00000101825	MXRA5	0.6126	9.17 x 10-5
ENSG00000123610	TNFAIP6	0.6098	1.01 x 10-4
ENSG00000124731	TREM1	0.6097	1.01 x 10-4
ENSG00000168913	ENHO	–0.6087	1.05 x 10-4
ENSG00000039560	RAI14	0.6034	1.25 x 10-4
ENSG00000136010	ALDH1L2	0.6031	1.26 x 10-4
ENSG00000109436	TBC1D9	0.6031	1.26 x 10-4
ENSG00000176723	ZNF843	–0.6031	1.26 x 10-4
ENSG00000123329	ARHGAP9	0.6027	1.27 x 10-4
ENSG00000170190	SLC16A5	0.6020	1.31 x 10-4
ENSG00000167642	SPINT2	–0.6008	1.35 x 10-4

CCN4 = cellular communication network factor 4; OA = osteoarthritic.

^aTranscriptome wide correlations were calculated using a Spearman correlation.

^bP-values were corrected for multiple testing with the Benjamini-Hochberg method.

Analysis for these 58 genes demonstrated enrichment for particular pathways. Most significant enrichment was the GO term “neural crest cell migration” (GO 0001755; P = 7.03 x 10^-3), containing Coronin-1C (CORO1C), Semaphorin-3C (SEMA3C), and SMO (Table 3). In addition to these genes, we observed several other genes correlating to CCN4, such as the aforementioned COL6A3 and Plexin-A2 (PLXNA2), that are also involved in migratory processes.

Table 3.

Gene Enrichment Analysis for 58 Genes Correlating with CCN₄/WISP₁ in Preserved OA Articular Cartilage.

Categorya	Term	Count	%b	Genes	P-value
GOTERM_BP_DIRECT	GO:0001755~neural crest cell migration	3	5.5	SEMA3C, SMO, CORO1C	7.03 x 10-3
GOTERM_BP_DIRECT	GO:0045444~fat cell differentiation	3	5.5	CCND1, LRRC8C, HMGA2	1.78 x 10-2
GOTERM_BP_DIRECT	GO:0031175~neuron projection development	3	5.5	CLMN, EFHD1, PTPRK	3.19 x 10-2
GOTERM_BP_DIRECT	GO:0007165~signal transduction	8	14.5	ARHGAP9, RASA3, TNFAIP6, MRC1, HUNK, HMGA2, PTPRK, CORO1C	4.13 x 10-2
GOTERM_CC_DIRECT	GO:0005887~integral component of plasma membrane	10	18.2	MSR1, SLC6A6, MRC1, SLC2A12, PLXNA2, LRRC8C, S1PR3, PTPRK, LRRC8E, SLC16A5	1.21 x 10-2
GOTERM_CC_DIRECT	GO:0005886~plasma membrane	19	34.5	MSR1, NRP2, OSBPL6, SLC41A2, OSBPL3, SDC2, SLC2A12, HDAC11, EMP1, SPINT2, TREM1, SLC6A6, GPRC5B, SMO, MRC1, LRRC8C, PLXNA2, S1PR3, LRRC8E	1.76 x 10-2
GOTERM_MF_DIRECT	GO:0005096~GTPase activator activity	4	7.3	ARHGAP9, RASA3, TBC1D9, AGFG2	4.48 x 10-2
GOTERM_MF_DIRECT	GO:0008134~transcription factor binding	4	7.3	CCND1, GPX3, HDAC11, HMGA2	4.68 x 10-2

CCN4/WISP1 = cellular communication network factor 4/Wnt-activated protein Wnt-1-induced signaling protein 1; OA = osteoarthritic; GO = gene ontology; GOTERM_BP_DIRECT = gene ontology terms Biological Processes; GOTERM_CC_DIRECT = gene ontology terms Cellular Component; GOTERM_MF_DIRECT = gene ontology terms Molecular Function.

^aGene enrichment was carried out using the online functional annotation tool DAVID, with gene ontology (GO) terms Biological Processes (GOTERM_BP_DIRECT), Cellular Component (GOTERM_CC_DIRECT), and Molecular Function (GOTERM_MF_DIRECT).

^bPercentage involved genes/total genes.

CCN4/WISP1 Addition Increases Cell Migration in Primary Chondrocytes

Since we found cell migration as most significant enriched pathway, we determined whether CCN4 indeed affected migration of chondrocytes. To this end, we performed a scratch assay to measure chondrocyte migration at four different timepoints with and without addition of recombinant human CCN4. After 36 h, we observed that in the absence of CCN4, the distance between the edges of the scratch in untreated chondrocytes was reduced by 73.7% ± 4.4%, whereas in the 0.1 µg/ml CCN4 treated chondrocytes, it was reduced by 87 ± 2.4% (P = 4.00 x 10^-3, Figure 1A and 1D, Supplementary Table S1). Additionally, we observed that the scratch closure rate was highest in the first 12 h, and increased directly after exposure to CCN4 (Figure 1B, Control = 15.9 ± 2.5 µm/h vs 0.1 µg/ml rhCCN4 = 22.1 ± 4.2 µm/h, P = 2.32 x 10^-2). To rule out that the increased scratch closure was the result of enhanced chondrocyte proliferation as a result of the CCN4 exposure, we conducted a crystal violet staining. As shown in Figure 1C, we did not observe any effect of CCN4 on crystal violet staining (Control = 0.228 ± 0.037 A.U. vs 0.1 µg/ml CCN4 = 0.216 ± 0.085 A.U., P = 5.93 x 10^-1). Of note, while the 0.5 µg/ml CCN4 treated chondrocytes did show a similar trend on migration compared to the 0.1 µg/ml group, the difference with controls was not significant due to larger variation.

Figure 1.

CCN4 addition increases chondrocyte migration. Scratch assay in human primary chondrocytes without and with addition of 0.1 µg/ml or 0.5 µg/ml recombinant human CCN4. (A) The distance between the edges of the scratch at four timepoints in percentage relative to T = 0 (N = 6 donors). (B) Scratch closure rate in the scratch assay in the first 12 h. (C) Cell proliferation determined by measuring absorbance at 595 nm of crystal violet stained chondrocytes dissolved in acetic acid (N = 3 donors). (D) Representative images of the scratch assay at T = 0 and T = 36. Percentages of the wound closed are presented in a continuous line graph, depicting the mean and 95% confidence intervals. Wound closure rates and crystal violet absorbances are presented in a scatter dot plot, depicting the mean and 95% confidence intervals, and each dot representing a single donor. *P ≤ 0.05, ** P ≤ 0.01 for 0.1 µg/ml CCN4. Scale bar = 100 µm A.U. = Absorbance units. CCN4 = cellular communication network factor 4.

Finally, we investigated whether the genes involved in migration, matrix composition, and cytoskeletal reorganization that we identified to correlate with CCN₄ in preserved OA cartilage, could be downstream of CCN4 signaling in chondrocytes. We therefore determined mRNA levels of the three genes in the most significantly enriched pathway (e.g., CORO1C, SEMA3C, and SMO) and of nine other genes (e.g., CCND1, CLMN, COL6A3, EML1, MXRA5, PLXNA2, PTPRK, RAI14, and TNFAIP6) involved in aforementioned processes. However, no significant changes in expression of these 12 genes were detected after exposure to CCN4 for 48 h (Figure 2).

Figure 2.

Gene expression in primary chondrocytes after exposure to CCN4. Reverse transcription quantitative real-time PCR (RT-qPCR) analysis of 12 genes correlating with CCN₄, involved in migration, matrix composition and cytoskeletal reorganization (CCND1, CLMN, COL6A3, CORO1C, EML1, MXRA5, PLXNA2, PTPRK, RAI14, SEMA3C, SMO, and TNFAIP6 after 48 h treatment with 0.1 µg/ml and 0.5 µg/ml CCN4 and their controls (N = 9 donors). Percentages of the wound closed are presented in a continuous line graph, depicting the mean and 95% confidence intervals. -ΔCt expression levels are presented in a scatter dot plot, depicting the mean and 95% confidence intervals, and each dot representing a single donor. CCN4 = cellular communication network factor 4.

Discussion

In this study, we further characterized functions of CCN4/WISP1 on chondrocyte biology, related to OA pathophysiology. Correlation analysis on CCN4 in OA cartilage with subsequent pathway enrichment analysis revealed “neural crest cell migration” as most significant enriched pathway. In line with this, we showed that addition of CCN4 to human primary chondrocytes promoted migration in a scratch assay.

Given the limited healing capacity of cartilage, it was notable to observe that CCN4 correlated with genes that are involved in migration. Promoting cell migration could assist in recruitment of chondrocytes to defect sites to initiate cartilage repair.⁵ The effects of CCN4 on cell migration have been demonstrated in other cell types, especially tumorigenic cells, where it promotes migration, but also proliferation and cell survival.¹⁶ In chondrocytes, CCN4 may induce proliferation.²¹ Here, we repressed proliferation in our wound healing assay by seeding chondrocytes in low serum levels, and demonstrated that CCN4 promoted chondrocyte migration. By promoting both proliferation and migration in chondrocytes, CCN4 could aid in stimulating cartilage repair. Previously, we observed that CCN4 is highly expressed in chondrocyte clusters near cartilage lesions.¹⁴ Since proteoglycan content and expression of matrix components were decreased in the presence of high CCN4 levels, we suggested that increased CCN4 was detrimental to cartilage ECM.¹⁴ In light of the current findings, it is tempting to speculate, however, that this reduction in matrix deposition allows for the orchestration of chondrocytes toward damaged areas for its subsequent repair.

Pathway enrichment analysis in preserved OA cartilage revealed neural crest cell migration as most significant pathway containing these three genes, CORO1C, SEMA3C, and SMO. SEMA3C is part of a wide family of semaphorins, that are secreted factors that guide cell migration, while SMO is a part of Hedgehog signaling.^22,23 CCN4/Canonical Wnt signaling has been shown to interact with both semaphorins and Hedgehog signaling.^12,24 We investigated whether CCN4 affected transcription of these three genes, and of nine other genes correlating with CCN4, which are all involved in controlling cell shape and migration. However, we did not find any direct effects of CCN4 on expression of these genes at the 48-hour endpoint. Possibly, these genes are not directly downstream of CCN4, or expression changes occur within 48 h. In different types of cancer cells, by activating the αvβ3 integrin receptor, CCN4 can promote migration through the nuclear factor κB (NF-κB), activator protein 1 (AP-1) and intercellular adhesion molecule 1 (ICAM1), and epidermal growth factor receptor (EGFR)/extracellular signal-regulated kinase (ERK) pathways.²⁵ It would be interesting to investigate whether this also occurs in chondrocytes. Notably, chondrocytes treated with higher concentration of CCN4 did not show significant increased migration, but had overall same direction of effects. Since the variability in this group was also larger, likely this is due to the relatively small number of samples.

In conclusion, we here demonstrated that CCN4 induces migration of human primary OA chondrocytes. This could aid in attracting chondrocytes to cartilage defects and repopulating cartilage scaffolds to stimulate cartilage regeneration.