SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Objective

Human articular chondrocytes do not express COL10A1 and do not undergo hypertrophy except in close vicinity to subchondral bone. In contrast, chondrocytes produced in vitro from mesenchymal stem cells (MSCs) show premature COL10A1 expression and cannot form stable ectopic cartilage transplants, which indicates that they may be phenotypically unstable and not suitable for treatment of articular cartilage lesions. CpG methylation established during natural development may play a role in suppression of COL10A1 expression and hypertrophy in human articular chondrocytes. This study was undertaken to compare gene methylation patterns and expression of COL10A1 and COL2A1 in chondrocyte and MSC populations, in order to determine whether failed genomic methylation patterns correlate with an unstable chondrocyte phenotype after chondrogenesis of MSCs.

Methods

COL10A1 and COL2A1 regulatory gene regions were computationally searched for CpG-rich regions. CpG methylation of genomic DNA from human articular chondrocytes, MSCs, and MSC-derived chondrocytes was analyzed by Combined Bisulfite Restriction Analysis and by sequencing of polymerase chain reaction fragments amplified from bisulfite-treated genomic DNA.

Results

The CpG island around the transcription start site of COL2A1 was unmethylated in all cell groups independent of COL2A1 expression, while 9 tested CpG sites in the sparse CpG promoter of COL10A1 were consistently methylated in human articular chondrocytes. Induction of COL10A1 expression during chondrogenesis of MSCs correlated with demethylation of 2 CpG sites in the COL10A1 promoter.

Conclusion

Our findings indicate that methylation-based COL10A1 gene silencing is established in cartilage tissue and human articular chondrocytes. Altered methylation levels at 2 CpG sites of COL10A1 in MSCs and their demethylation during chondrogenesis may facilitate induction of COL10A1 as observed during in vitro chondrogenesis of MSCs.

DNA methylation is a common regulatory mechanism responsible for locking DNA regions in an inactive state, thereby permanently blocking expression of the affected genes in a certain cell type. It has been suggested that genomic DNA methylation plays a crucial role in the generation of tissue-specific expression profiles, and methylation-based analysis has therefore been proposed as a promising tool for the characterization of cell types (1). Aberrant methylation patterns are often found in neoplastic cells, causing the modulation of transcription of linked loci (2). Differences in DNA methylation patterns of matrix metalloproteinases and ADAMTS-4 and correlation with protein levels have been reported for normal and osteoarthritic cartilage. However, a correlation between altered CpG methylation and direct changes in gene expression was not established (3).

For therapeutic use, phenotypic stability and functional suitability of cells and tissue engineering products are mandatory. State-of-the-art treatment for focal cartilage defects is the transplantation of primary autologous chondrocytes either in suspension or in combination with distinct matrices. Disadvantages of this method include the necessity of generating new defects in healthy cartilage in order to harvest human articular chondrocytes for expansion, and the dedifferentiation of these cells during monolayer expansion. The use of mesenchymal stem cells (MSCs) as an alternative source of chondrocytes seems attractive, since MSCs can easily be obtained from various mesenchymal tissues, such as bone marrow, adipose tissue, synovial tissue, and others (4, 5).

Our group has recently demonstrated that current in vitro protocols for production of chondrocytes from MSCs lead to premature induction of hypertrophic molecules, such as type X collagen, and to up-regulation of alkaline phosphatase enzyme activity, which is crucial for matrix calcification. Ectopic transplants composed of MSC-derived chondrocytes underwent vascular invasion, matrix calcification, and microossicle formation in SCID mice, rather than adopting a stable chondrocyte phenotype (6). Notably, human articular chondrocytes were resistant to hypertrophy under similar conditions, remained negative for COL10A1 expression, and formed stable uncalcified ectopic cartilage transplants. This indicated that human articular chondrocytes “remembered” their history as stable articular chondrocytes during culture expansion and acquired an intrinsic developmental arrest before hypertrophy that was independent of systemic factors. Epigenetic regulation mechanisms like DNA methylation are one way to establish such developmental memory in cells (7).

In order to investigate whether aberrant or missing genomic methylation patterns may be correlated with the unstable chondrocyte phenotype observed after chondrogenesis of MSCs, in the present study we compared the gene methylation patterns and expression of COL10A1 and COL2A1 in human articular chondrocytes, MSCs, and chondrocytes produced from MSCs in vitro. We tested whether silencing of the COL10A1 gene in human articular chondrocytes was correlated with DNA methylation while corresponding CpG sites remained unmethylated and open for access in MSCs. Beyond providing further insight into molecular mechanisms of chondrocyte differentiation, our analysis could help to improve cartilage regeneration from MSCs by elucidating possible means to prevent premature induction of hypertrophy in MSCs.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Donors.

MSCs were expanded from adipose tissue liposuction samples and bone marrow aspirates in order to exclude an influence of contaminating cell populations such as osteoprogenitor cells. After informed consent was obtained from the subjects, adipose tissue–derived MSCs were obtained from 3 women (ages 23–32 years) undergoing liposuction. Bone marrow–derived stromal cells were obtained from 3 women and 1 man (ages 61–69 years) undergoing hip replacement surgery. Articular cartilage from the knee was obtained from 3 men (ages 54–62 years) with knee osteoarthritis, from regions having no macroscopically evident degeneration, and from 2 women and 1 man (ages 39–52 years) with no cartilage abnormalities after leg amputation. Studies were approved by the local ethics committee.

Cell isolation and cultivation.

Adipose tissue and bone marrow–derived MSCs were isolated as previously described (8). Cells were seeded in culture flasks and maintained at 37°C in 6% CO2 and a humidified atmosphere. MSC expansion culture medium consisted of high-glucose Dulbecco's modified Eagle's medium (DMEM) with 40% MCDB 201 medium (Sigma-Aldrich, Deisenhofen, Germany), 2% fetal calf serum (FCS), 2 × 10−8M dexamethasone, 10−4M ascorbic acid 2-phosphate, 10 μg/ml insulin, 10 μg/ml transferrin, 10 ng/ml selenous acid, 100 units/ml penicillin, 100 μg/ml streptomycin, 10 ng/ml recombinant human epidermal growth factor (Strathmann Biotech, Hamburg, Germany), and 10 ng/ml recombinant platelet-derived growth factor BB (Strathmann Biotech). During expansion, medium was replaced twice a week. Cells were used after 2–3 passages. Spheroids of 4–5 × 105 MSCs or chondrocytes were formed, and chondrogenic differentiation was induced by 6-week treatment with a basal medium consisting of high-glucose DMEM supplemented with 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenous acid, 0.1 μM dexamethasone, 0.17 mM ascorbic acid 2-phosphate, 1 mM sodium pyruvate, 0.35 mM proline, 1.25 mg/ml bovine serum albumin, 10 ng/ml transforming growth factor β3, and, in the case of adipose tissue–derived cells, 10 ng/ml bone morphogenetic protein 6 (R&D Systems, Minneapolis, MN).

Macroscopically unaffected regions of knee cartilage tissue from patients with knee osteoarthritis were cut into small pieces and digested with collagenase B (1.5 mg/ml) and hyaluronidase (0.1 mg/ml) (Roche Diagnostics, Mannheim, Germany) in culture medium overnight, as previously described (9). Cells were cultured for 2–3 passages in DMEM, 10% FCS, 100 units/ml of penicillin, and 100 μg/ml of streptomycin at 37°C in 6% CO2, or, when subjected to 3-dimensional culture, in the same medium as MSCs.

Histology.

Pellets were fixed and stained according to standard procedures, as previously described (6), using Alcian blue (1%; Chroma, Köngen, Germany), a mouse anti-human type II collagen monoclonal antibody (clones I-8H5 and II-4C11; ICN Biomedicals, Aurora, OH), and a mouse anti-human type X collagen monoclonal antibody (X-34 and X-53) (10).

RNA isolation.

Total RNA was isolated by guanidinium thiocyanate–phenol extraction (peqGOLD Trifast; Peqlab, Erlangen, Germany) from 1 gm of normal cartilage pulverized in liquid nitrogen using a freezer mill (Mikro-Dismembrator S; B. Braun Biotech International, Melsungen, Germany), from 1 × 106 cultured cells, or from up to 8 parallel chondrogenic pellets after tissue homogenization in a Polytron homogenizer (Kinematica, Luzern, Switzerland). Polyadenylated messenger RNA (mRNA) was isolated from total RNA with oligo(dT)-coupled magnetic beads, according to the recommendations of the manufacturer (Dynabeads; Dynal, Oslo, Norway). The supernatant was used for DNA isolation.

Real-time reverse transcriptase–polymerase chain reaction (RT-PCR).

Quantification of mRNA levels was performed with a LightCycler (Roche Diagnostics), as previously described (6). The following primers were used for amplification: for β-actin, forward 5′-CTCTTCCAGCCTTCCTTCCT-3′ (on exon 4) and reverse 5′-CGATCCACACGGAGTACTTG-3′ (on exon 6); for COL2A1, forward 5′-TGGCCTGAGACAGCATGAC-3′ (on exon 51) and reverse 5′-AGTGTTGGGAGCCAGATTGT-3′ (on exon 53); for COL10A1, forward 5′-CCCTTTTTGCTGCTAGTATCC-3′ (on exon 2) and reverse 5′-CTGTTGTCCAGGTTTTCCTGGCAC-3′ (on exon 3). The β-actin signal was determined for each complementary DNA (cDNA) sample, and signals for all other genes were normalized to β-actin. Negative controls were added in each RT-PCR using water instead of cDNA to exclude contamination. Primer pairs were intron spanning and thus enabled us to exclude genomic contamination. For β-actin, an additional control reaction with non–reverse-transcribed mRNA was performed to assure the absence of significant contamination through genomic DNA.

DNA isolation.

Genomic DNA was isolated from 1 × 106 cultured cells or from up to 8 pellets cultivated in parallel with the samples used for RNA isolation, using the DNeasy tissue kit (Qiagen, Hilden, Germany). DNA from human articular cartilage was isolated from the supernatant obtained by RNA isolation, using ethanol precipitation and purification with the DNeasy tissue kit.

DNA methylation analysis.

Genomic DNA (500 ng) from distinct donors was treated with bisulfite using the EpiTect bisulfite kit (Qiagen). The regions of interest in COL10A1 and COL2A1 were amplified by PCR using bisulfite-treated DNA as a template and the primers listed in Table 1. Amplification was performed with 30–100 ng DNA, 10 pmoles primers, 0.5 mM dNTPs, 1 unit HotStar Taq DNA polymerase, 2–3 mM MgCl2, and buffers, according to the recommendations of the polymerase supplier (Qiagen). The PCR conditions were as follows: 94°C for 15 minutes, 40 cycles at 94°C for 30 seconds, 50–54°C for 30 seconds, and 72°C for 30 seconds, and 1 cycle at 72°C for 5 minutes. DNA fragments CX-2, 3, 5, 7, and 9 for COL10A1 and CII-1 and CII-2 for COL2A1 were analyzed by Combined Bisulfite Restriction Analysis (COBRA) (11). The purified PCR product (200 ng) (Ultrafree-DA; Millipore, Eschborn, Germany) was digested with 10 units of the respective restriction enzymes Tai I (ACGT), Taq I (TCGA), Bsh 1236I (CGCG), and Bsp 143I (GATC) (Fermentas, St. Leon-Rot, Germany) for 2 hours at the appropriate temperature. Fragments were analyzed on 10% polyacrylamide gels after staining with SYBR Green I (Sigma-Aldrich). All fragments were analyzed by complete DNA sequencing (MWG Biotech, Ebersberg, Germany) of the purified PCR products with the same primers used for amplification. Average ratios of methylation were estimated according to signal peaks for C and T in CpG sites analyzed on the sense strand, or for G and A in CpG sites analyzed on the antisense strand. CpG sites with methylation rates <20% were considered unmethylated, and sites with rates >80% were considered methylated. CpG sites with methylation rates between 20% and 80% were considered partially methylated.

Table 1. Primer pairs and restriction enzymes used for DNA methylation analysis*
Amplicon ID (amplicon length, bp)Primer (sequence)/location relative to transcription start siteEnzymes discriminating methylation sites after bisulfite treatment
  • *

    F = forward; R = reverse.

COL10A1 gene  
 CX-1 (308)CX-P1-F (AAAATAAGGAAAAGGAGAGTTT)/−164Taq I at 26; Tai I at 239, 253; Rsa I at 253; Bsp 143I at 28
  CX-P1-R (AAATCTTTAAAATTTCTTTCCAAAC)/+144 
 CX-2 (276)CX-P2-F (TAGATGATGGTTTAAATTTAGTATT)/−416 
  CX-P2-R (CTAAACTCTCCTTTTCCTTATTTTC)/−141 
 CX-3 (304)CX-E3-1F (TTGGTTTATTAGGTTTTGTTGGATT)/+4,203Taq I at 26; Bsp 143I at 261
  CX-E3-1R (AATCCCTATTATCCAAATTTTCCTAA)/+4,506 
 CX-4 (309)CX-E3-2F (AGTTTTAGGATTTAGGGGTTTTTT)/+4,513 
  CX-E3-2R (ACTCCTAACTTTCCAATACCTTCTAA)/+4,821 
 CX-5 (334)CX-E3-3F (GATTTTTTGGGAATATGGGATTTT)/+5,031Taq I at 175
  CX-E3-3R (CCTAATATTCCAAAAACACCTCTTAA)/+5,364 
 CX-6 (249)CX-E3-4F (GTTTAGTAGGAGTAAAGGGAATGTT)/+5,292 
  CX-E3-4R (CTAAAAAACCAAACTCTCCAAAATAA)/+5,540 
 CX-7 (313)CX-E3-5F (TAATAGTATTATGATTTAAGGATTGGAATT)/+5,753Taq I at 191, 236; Tai I at 74
  CX-E3-5R (ATACTCACATTAAAACCACTAAA)/+6,065 
 CX-8 (181)CX-P8-F (AAGAGGTTGAGAATTATTGTG)/−2,569 
  CX-P8-R (TCACATCTAACAATTAACTACT)/−2,389 
 CX-9 (101)CX-P9-F (AGTATTATGTATTAAGAGTTGAG)/−1,718Tai I at 38
  CX-P9-R (ACACACATACTACTATTTACTTA)/−1,618 
 CX-10 (254)CX-P10-F (TGTGTTTATGTTTGTGTATGG)/−1,011 
  CX-P10-R (TCCCAATTAAAAACACTTTCC)/−758 
 CX-11 (251)CX-P11-F (GTGTTTTTAATTGGGATGATTT)/−773 
  CX-P11-R (TAAATCCCTTATTCACAACAC)/−523 
COL2A1 gene  
 CII-1 (216)C2-P-F (GGGTATTGGTAGGGTTTAGG)/−139Bsh 1236I at 78, 134; Taq I at 103; Tai I at 149
  C2-P-R (CCTCATACAAAAAACCCTTAAAAC)/+77 
 CII-2 (372)C2-E1-F (TTGTTTTAAGGGTTTTTTGTATGAG)/+52Bsh 1236I at 29, 48, 105, 288, 324; Taq I at 328; Tai I at 140, 155, 346
  C2-E1-R (CAAAACTCTTCTTAATAAACTTCTAC)/+423 
 CII-3 (378)C2-P-3F (AGGGGGTGTTTGTAGAGG)/−343 
  C2-P-3R (CACACCAACTCTTTTCCTAACT)/−720 
 CII-4 (260)C2-I1-F (TGGTTTGGGATTTTAGTTTATT)/+813 
  C2-I1-R (AAACTCTTTTCCACCAACTTTA)/+554 

Sequence analysis.

DNA sequences for COL2A1 (ENSG00000139219) and COL10A1 (ENSG00000123500) were retrieved from the Ensembl database (http://www.ensembl.org). CpG sites were identified, and CpG islands were searched bioinformatically (12). The Database of Single-Nucleotide Polymorphisms (dbSNP) (http://www.ncbi.nlm.nih.gov/projects/SNP) was searched for SNPs. For COL10A1, no polymorphism coinciding with a CpG site was found. For COL2A1, 7 polymorphisms were found on possible methylation sites or on restriction sites used in COBRA, and the possibility of occurrence of minor alleles among samples was thoroughly examined. No evidence was obtained regarding these polymorphisms that needed to be considered. Sequencing of unconverted DNA fragments corresponding to CX-9 confirmed that no polymorphisms were present at the corresponding CpG sites. Primers were designed using PRIDE software, version 1.2 (http://pride.molgen.mpg.de/).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Analysis of the methylation pattern of COL10A1 and COL2A1 regulatory gene regions.

Promoters vary considerably with regard to the number of available CpG sites, and promoters with few CpG sites are called “sparse CpG promoters” (3). The size of the human type X collagen promoter is unclear, but it has been shown that the 2,400-bp region upstream of the transcription start site allows COL10A1 mRNA transcription (13). This region contains only 9 CpG sites and includes a known enhancer element of 0.6 kb, which is free of CpG sites. The COL10A1 promoter is therefore a sparse CpG promoter. The 9 CpG sites within the promoter region 2.5 kb upstream of the transcription start site were all analyzed in this study. Downstream of the start codon, in the coding region of COL10A1, 21 CpG sites were found in exon 3. Sixteen of those sites, along with 3 additional CpG sites located close to the transcription start site, were analyzed in the present study (Figure 1A). In contrast, the COL2A1 promoter region around the transcription start site is rich in CpG sites and contains a distinct CpG island (14) with 60 CpG dinucleotides within a 500-bp region (Figure 1B). In this study, 74 CpG sites from a 1,500-bp region around the transcription start site were analyzed.

thumbnail image

Figure 1. Methylation analysis of COL10A1 and COL2A1 sequences, using Combined Bisulfite Restriction Analysis (COBRA). A and B, Location of analyzed CpG sites in the genomic DNA regions of COL10A1 (A) and COL2A1 (B). Lower rows show higher-resolution views of regions of interest for DNA methylation analysis. Solid boxes represent exons (E), open boxes represent noncoding regions, and shaded boxes represent known regulatory elements in the COL10A1 promoter (P). Hatched boxes represent the TATA box. Horizontal lines represent the polymerase chain reaction amplicons used in the methylation analysis (amplicons CX-1 to CX-11 for COL10A1 and CII-1 to CII-4 for COL2A1), and vertical lines represent CpG dinucleotides. Arrows indicate the restriction enzyme recognition sites evaluated by COBRA. The restriction enzymes used were Tai I (T), Taq I (Q), Bsp 143I (P), and Bsh 1236I (B). I2 = intron 2. C, Detection of new restriction sites after bisulfite treatment. COL10A1 and COL2A1 DNA fragments were amplified after bisulfite treatment, and potential new restriction sites indicating CpG methylation were detected. Representative results obtained using amplicons CX-7 and CII-2 for adipose tissue–derived mesenchymal stem cells (M), the same mesenchymal stem cells after chondrogenic induction (Mi), and expanded human articular chondrocytes (C), separated by 10% polyacrylamide gel electrophoresis after digestion with the indicated enzymes, are shown. Undigested fragments (no equation image) and fragments resulting from in vitro methylation of DNA by treatment with Sss I methylase (S) were used as controls.

Download figure to PowerPoint

Identical genomic DNA samples derived from human articular chondrocytes, MSCs, and MSCs after 6 weeks of chondrogenesis were used for the COL10A1 and COL2A1 methylation analysis after bisulfite conversion. COL10A1-derived fragments downstream of the transcription start site were all almost completely digested, independent of cell type (Figure 1C), indicating that the corresponding CpG dinucleotides were all methylated in the original DNA templates. In contrast, COL2A1-derived fragments remained uncut for all 3 cell types (Figure 1C), indicating that all CpG dinucleotides analyzed in the promoter region of COL2A1 were unmethylated in the original DNA templates.

Notably, the only CpG site within the COL10A1 promoter region testable by COBRA, which was located in the CX-9 fragment, showed a differential restriction pattern between the cell types (Figure 2). High levels of digestion, indicative of high methylation levels, were observed in human articular chondrocytes. Intermediate levels, indicative of lower methylation levels, were seen in MSCs, and only very faint bands consistent with very low methylation levels were seen in MSCs after 6 weeks of chondrogenesis, for all donors tested. This indicated that while the tested COL2A1 DNA region was open for access in all cell types, the analyzed DNA of the COL10A2 coding regions was always inactivated. One CpG site in the COL10A1 promoter region testable by this method, however, seemed to undergo demethylation during chondrogenesis of MSCs.

thumbnail image

Figure 2. Methylation analysis of the amplicon CX-9 in the COL10A1 upstream genomic region, using Combined Bisulfite Restriction Analysis. COL10A1 DNA fragments were amplified after bisulfite treatment, and a potential new Tai I restriction site, indicating CpG methylation, was detected. Human articular chondrocytes (C) from donors 1–3 and adipose tissue–derived mesenchymal stem cells from donors 4–6 after expansion (M) and after chondrogenic induction (Mi) were analyzed by 10% polyacrylamide gel electrophoresis after restriction. Undigested fragments were used as controls for all samples, and representative results obtained using mesenchymal stem cells from donor 4 after expansion (no equation image) are shown. The length of the undigested fragment is 101 bp; restriction by Tai I results in fragments 38 bp and 63 bp long.

Download figure to PowerPoint

DNA methylation analysis of COL10A1 and COL2A1 by bisulfite sequencing.

Since COBRA can only detect those CpG sites that appear as part of a restriction enzyme recognition sequence, all COL10A1 and COL2A1 PCR fragments were also analyzed by complete DNA sequencing. Consistent with the data obtained by COBRA, complete DNA sequencing showed completely methylated sequences for COL10A1 coding regions and almost completely unmethylated DNA sequences for COL2A1 for all donors (n = 3) and all cell types tested (Figure 3). In the COL10A1 regulatory region upstream of the transcription start site, 7 of 9 CpG sites were fully methylated in all cell groups, while 2 CpG sites lying close together in the CX-9 amplicon showed differential methylation patterns between cells (Figure 4).

thumbnail image

Figure 3. Sequencing of COL2A1 and COL10A1 DNA fragments after bisulfite treatment. Methylation rates were calculated for each sample, and boxes represent the mean methylation rates observed in mesenchymal stem cells (M) from 3 donors, mesenchymal stem cells after chondrogenic induction (Mi) from 3 donors, and human articular chondrocytes (C) from 3 donors. Open boxes represent unmethylated cytosines, green boxes represent partially methylated cytosines, with mean methylation rates between 20% and 80%, and blue boxes represent methylated cytosines, with mean methylation rates >80%. Chromatograms of representative sequencing profiles indicating homogeneous peaks are shown for the underlined boxes.

Download figure to PowerPoint

thumbnail image

Figure 4. Sequencing of DNA fragments from the upstream genomic region of COL10A1 after bisulfite treatment. A, Sequence analysis of the antisense strand of amplicon CX-9. Chromatograms of samples of human articular chondrocytes (C), adipose tissue–derived mesenchymal stem cells (MSCs) (M), MSCs after chondrogenic induction (Mi), and healthy human articular cartilage (CART) from 3 donors per cell type are shown. Arrows indicate the relevant CpG dinucleotides. Sequencing was performed with the CX-P9-R reverse primer. The expected sequences of methylated and unmethylated DNA after conversion are shown. Unmethylated, and thus converted, CpG sites of the sense strand appear as CA on the antisense strand, and methylated CpG sites appear as CG. Note the A and G double peaks in each CpG site in MSC samples and MSC samples after chondrogenic induction from all donors. B, Summary of the sequence analysis of CpG sites in the COL10A1 upstream region (CX-8–CX-11). Rates of methylation in human articular chondrocytes and adipose tissue–derived MSCs were estimated for each donor sample (n = 3 per group) and are represented schematically by color coding. Results of the analysis of MSCs were confirmed by results obtained with samples from 2 additional adipose tissue donors and 1 bone marrow donor (results not shown).

Download figure to PowerPoint

While DNA from expanded human articular chondrocytes provided signals consistent with full methylation at both sites, partial methylation, estimated between 60% and 80%, was evident in MSCs from all 3 donors. Methylation levels were consistently reduced to <50% at both sites when the same MSCs were analyzed 6 weeks after induction of chondrogenesis. Similar methylation patterns for COL10A1 upstream genomic regions were obtained in human articular chondrocytes and healthy cartilage tissue (Figure 4A). In the analysis of COL2A1, some minor parallel peaks indicated partial methylation (<50%) for a few CpG sites in the COL2A1 amplicons CII-3 and CII-4 in MSCs from all donors, as well as for 1 CpG site in chondrocytes (Figure 3). In summary, no significant levels of CpG methylation were found for COL2A1 in any of the samples, while all but 2 of the sparse CpG sites in COL10A1 were always fully methylated.

Correlation of COL10A1 mRNA expression with demethylation of 2 CpG sites.

Successful chondrogenesis of the analyzed MSCs from all donors was evident from proteoglycan deposition and formation of a type II collagen–rich extracellular matrix in the spheroids (Figure 5A). In order to correlate the COL2A1 and COL10A1 methylation patterns in the samples with their gene expression, COL2A1 and COL10A1 mRNA levels were quantified by real-time RT-PCR. While neither COL2A1 nor COL10A1 was expressed in expanded MSCs, both transcripts were strongly induced after 6 weeks of chondrogenesis (Figure 5B). Human articular chondrocytes expanded for 2–3 passages and cartilage expressed COL2A1 at lower levels, which is consistent with dedifferentiation of chondrocytes in culture and low type II collagen turnover in established tissue. COL10A1 mRNA was not expressed in human articular chondrocytes, and its level was <1% of β-actin expression in articular cartilage (Figure 5B). In human articular chondrocytes that were exposed for 6 weeks to chondrogenic induction conditions similar to MCSs (Ci in Figure 5), COL2A1 expression was induced while COL10A1 mRNA remained below cartilage levels, confirming their known resistance to hypertrophy under these conditions (6). The strong induction of COL10A1 mRNA during chondrogenic differentiation of MSCs was thus correlated with loss of methylation at 2 partially methylated CpG sites around −1.68 kb from the transcription start site. There was no evidence of epigenetic regulation of COL2A1 transcription at the level of DNA methylation.

thumbnail image

Figure 5. Gene expression analysis. A, Production of type II collagen and type X collagen in mesenchymal stem cells (MSCs) (Mi) and chondrocytes (Ci) 6 weeks after chondrogenic induction. Representative spheroids from 1 MSC sample and 1 human articular chondrocyte sample are shown. Immunohistochemistry was used to determine type II collagen protein deposition. All MSCs and human articular chondrocytes produced type II collagen at 6 weeks, while type X collagen was deposited in MSC spheroids only. Alcian blue staining of spheroids 6 weeks after chondrogenic induction showed a proteoglycan-rich extracellular matrix (insets). B, Levels of COL2A1 and COL10A1 mRNA, quantified by real-time reverse transcriptase–polymerase chain reaction and displayed relative to β-actin levels, in MSCs from adipose tissue (M), the same MSCs after chondrogenic induction (Mi), expanded human articular chondrocytes (C), chondrocytes after chondrogenic induction (Ci), and healthy human cartilage tissue (CART).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

This study showed that the core promoter region of COL10A1, spanning 2.5 kb upstream of the transcription start site (13), corresponded to a sparse CpG promoter in which 2 neighboring CpG sites demonstrated differential methylation in chondrocytes and MSCs. While all of the other CpG sites analyzed in the COL10A1 genomic sequence were completely methylated in both cell types, this pair of CpG sites was partially unmethylated in MSCs, in which COL10A1 expression can be induced during chondrogenic differentiation. These CpG sites were consistently methylated in chondrocytes, in which COL10A1 was not expressed, even when chondrogenic conditions were used. Notably, chondrogenic differentiation of MSCs was associated with a reduction in methylation rates to <50% at these 2 CpG sites, while parallel transcription of COL10A1 was strongly induced. The demethylated state of these 2 cytosines in the COL10A1 promoter may thus correspond to a permissive state of the COL10A1 promoter, in which it may be accessible to transcription factors, while methylation of these cytosines would block COL10A1 expression in chondrocytes. In articular cartilage and in 1 human vascular endothelial cell sample in which COL10A1 was not expressed, the same 2 CpG sites were fully methylated. This strengthens the association observed between the methylation status of these CpG sites and gene expression, and supports the hypothesis that this pair of CpG sites may play an important role in epigenetic regulation of COL10A1 gene expression.

To our knowledge, this is the first study to show a change in methylation patterns during in vitro chondrogenic differentiation of MSCs. The methylation status of CpG sites was assessed by 2 independent methods, COBRA and bisulfite sequencing. COBRA was used to detect the first indication of possibly methylated CpG sites. This method was not optimized to achieve complete digestion and thus was not used in the present study for quantification of results. Bisulfite sequencing was used as the method of choice to cover all CpG sites of the analyzed regions and for quantification of methylation results.

The differentially methylated cytosines in the COL10A1 promoter are located at positions −1,680 bases and −1,674 bases from the transcription start site. The 2 CpG sites are separated from each other by only 4 bases and reside in a region between −2.4 kb and −0.9 kb, which was previously found to be crucial for COL10A1 expression (13). The CpG sites do not, however, reside within an enhancer element located between −2.4 kb and −1.8 kb, which acts independently of location (15). This suggests that the CpG sites identified in this study could play a regulatory role by affecting the binding of transcription factors to these genomic regions.

Interestingly, one of the CpG sites is part of a Pax8 and a N-myc transcription factor DNA binding motif. The analysis of the methylation status of the chondrocyte-specific gene chondromodulin has shown that methylation of one CpG site coinciding with a binding site of Sp-1 and Sp-3 correlated with chondromodulin gene expression and that methylation affected binding of Sp-3 to this site (16). A developmental association between Pax8 or N-myc and COL10A1 expression has not yet been established, but should be investigated in more detail. Analysis of the chondromodulin promoter has shown that DNA methylation and histone acetylation are both involved in the regulation of transcription (17). Histone acetylation has also been implicated in the regulation of chondrocyte hypertrophy (18) and may thus play a role in the regulation of COL10A1 expression as well.

The finding of particular methylation patterns at only 2 CpG sites on the COL10A1 promoter was surprising, since we initially expected that most or all of the present CpG sites may be involved in unsilencing COL10A1 in differentiating MSCs. It has been shown that the efficiency of repression by DNA methylation may in some cases depend on the total number of methylated CpG sites, but in other cases only a single CpG site may be crucial (19). Several studies have clearly shown that the demethylation of single CpG sites may be decisive for permitting gene expression (16, 20, 21). The molecular machinery for such silencing based on a single CpG site exists. Binding to one CpG site is sufficient for methyl-CpG binding protein 2 to promote chromatin condensation (22). Our results thus suggest that in sparse CpG promoters a complete demethylation of genomic regions is not necessary for allowing transcription, and that the methylation status of single CpG sites might be decisive for permitting gene expression.

The association between methylation status of CpG sites and gene expression described in this report does not demonstrate any causal relationship. Yet it suggests that in MSCs differentiating toward the chondrogenic lineage, COL10A1 expression may be allowed by 2 unmethylated cytosines in the promoter of the gene. Further studies are, however, needed to investigate this possibility.

The level of methylation, which was between 60% and 80% in expanded MSCs, decreased further during chondrogenic differentiation. The fact that only partial methylation patterns were observed in expanded MSCs may be due to the heterogeneity of expanded MSC populations, which is well established. Since it is now known that CD146 can be used as a marker to distinguish MSCs of distinct functional potential (23), further studies are needed to determine whether a correlation between distinct subpopulations and methylation can be established. Loss of methylation during chondrogenesis may either be caused by inhibition of methylation of the daughter DNA strand during replication and thus be related to cell growth, or be due to a loss of the methylated cell population during differentiation and thus correspond to a selective advantage of cells with demethylated CpG sites during chondrogenesis. Further studies are necessary to determine which of these mechanisms applies during chondrogenesis of MSCs.

Analysis of a population of hypertrophic chondrocytes from the growth plate is needed to determine conclusively whether activation of COL10A1 expression in hypertrophic chondrocytes is generally associated with unsilencing by loss of DNA methylation of 2 CpG sites in the COL10A1 promoter. It is not possible to conduct such an analysis in human cells, due to a shortage of available donor cells. Although COL10A1-expressing chondrocyte-like cells may be produced from MSCs by culture under chondrogenic conditions, their in vitro development may not be representative of the tightly controlled program of endochondral bone formation. Still, our results suggest that methylation patterns may play a possible role in unsilencing COL10A1 gene expression.

No evidence of epigenetic regulation of COL2A1 at the level of DNA methylation was found. COL2A1 contained a CpG island around the transcription start site, and with the exception of a few CpG sites in the first intron of COL2A1, which showed a low rate of methylation in MSCs and chondrocytes, the selected regulatory COL2A1 promoter regions were unmethylated. Thus, these promoter regions were open for access by DNA binding proteins and transcription factors independent of whether COL2A1 was expressed (i.e., in human articular chondrocytes, cartilage, and induced MSCs) or not (i.e., in expanded MSCs). This is consistent with global findings from large-scale DNA methylation profiling studies. CpG island promoter regions were found to be mostly unmethylated (87.9%), while almost 50% of sparse CpG promoters were hypermethylated (24).

The COL2A1 promoter in this respect seems similar to the aggrecan promoter, which also contains a distinctive CpG island and has been shown to be unmethylated in normal, aged, and osteoarthritic chondrocytes, with no evidence of regulation of aggrecan expression by DNA methylation (25). No case of unsilencing by complete demethylation of a CpG island is known so far. Therefore, methylation of the CpG island in the COL2A1 promoter was not expected in uninduced MSCs. Fernandez and colleagues (26) found different methylation profiles in chicken COL2A1 in 5 different cell lines. However, the CpG sites were located in the 3′ region of COL2A1, and therefore at a great distance from known regulatory elements controlling transcription.

Though the promoter and enhancer regions of COL10A1 and COL2A1 cover relatively large DNA regions of ≥4.6 kb for COL10A1 (13, 15, 27, 28) and >9 kb for COL2A1 (29, 30), we chose only limited parts for methylation analysis. The selected regions included CpG islands as defined by prediction programs (12), CpG dinucleotides associated with the RNA polymerase binding site or with known regulatory elements for COL10A1, and regions with a CpG content significantly higher than that of surrounding sequences. In particular, palindromic CpG sequences, which are thought to be most important for DNA methylation, were included (31). A regulatory function of CpG sites situated far from the transcription start site may exist, but this has never been proven in DNA methylation analysis. For these reasons, we consider our analysis of COL10A1 and COL2A1 regulatory gene regions to be relevant and comprehensive, although complete coverage of all CpG sites was not attained.

In summary, COL10A1 gene expression was activated during chondrogenesis in cells with partially unmethylated cytosines at positions −1,680 bases and −1,674 bases from the transcription start site, whereas it was absent in cells with fully methylated cytosines at these positions. DNA methylation may thus be at least one possible mechanism for the COL10A1 silencing and suppression of hypertrophy observed in articular chondrocytes, which could explain why developmental memory exists in articular chondrocytes but not in MSCs. Further analysis of the potential regulatory role of this pair of CpG sites and of the mechanisms underlying this regulation may allow future optimization of chondrogenic protocols to recruit MSCs for persistent and stable cartilage repair.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Dr. Richter had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Zimmermann, Boeuf, Richter.

Acquisition of data. Zimmermann, Boeuf, Dickhut, Boehmer, Olek, Richter.

Analysis and interpretation of data. Zimmermann, Boeuf, Olek, Richter.

Manuscript preparation. Zimmermann, Boeuf, Olek, Richter.

Statistical analysis. Boeuf.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

The authors thank Regina Foehr and Katrin Goetzke for excellent technical assistance and Proaesthetic (Heidelberg, Germany) for providing adipose tissue samples.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES