Article Text


Extended report
Gene expression in human chondrocytes in late osteoarthritis is changed in both fibrillated and intact cartilage without evidence of generalised chondrocyte hypertrophy
  1. C J Brew1,
  2. P D Clegg1,2,
  3. R P Boot-Handford1,
  4. J G Andrew3,
  5. T Hardingham1
  1. 1
    Wellcome Trust Centre for Cell-Matrix Research and, Faculty of Life Sciences, University of Manchester, Manchester, UK
  2. 2
    Faculty of Veterinary Sciences, University of Liverpool Veterinary Teaching Hospital, Leahurst, Neston, UK
  3. 3
    Ysbyty Gwynedd, Penrhosgarnedd, Bangor, UK
  1. Correspondence to Professor P D Clegg, Faculty of Veterinary Sciences, University of Liverpool Veterinary Teaching Hospital, Leahurst, Neston CH64 7TE, UK; p.d.clegg{at}


Objectives: To investigate changes in gene expression in fibrillated and intact human osteoarthritis (OA) cartilage for evidence of an altered chondrocyte phenotype and hypertrophy.

Methods: Paired osteochondral samples were taken from a high-load site and a low-load site from 25 OA joints and were compared with eight similar paired samples from age-matched controls. Gene expression of key matrix and regulatory genes was analysed by quantitative real-time reverse transcription-polymerase chain reaction on total RNA extracted from the cartilage.

Results: There was a major change in chondrocyte gene expression in OA cartilage. SOX9 (38-fold) and aggrecan (4-fold) gene expression were both lower in OA (p<0.001), and collagen I (17-fold) and II (2.5-fold) gene expression were each increased in a subset of OA samples. The major changes in gene expression were similar at the fibrillated high-loaded site and the intact low-loaded site. There was no evidence of a generalised change in OA to proliferative or hypertrophic phenotype as seen in the growth plate, as genes associated with either stage of differentiation were unchanged (PTHrPR), or significantly downregulated (collagen X (14-fold, p<0.002), VEGF (23-fold, p<0.02), BCL-2 (5.6-fold, p<0.001), matrilin-1 (6.5-fold, p<0.001)). In contrast MMP-13 was significantly upregulated in the OA cartilage samples (5.3-fold, p<0.003).

Conclusions: The expression of key chondrocyte genes, including aggrecan and SOX9, was decreased in OA cartilage and the changes were similar in both fibrillated high-loaded and intact low-loaded cartilage on the same joint. However, there was no significant upregulation of type X collagen, and other genes associated with chondrocyte further differentiation and hypertrophy.

Statistics from

Osteoarthritis (OA) is a chronic disease affecting articular joints. It is a heterogeneous condition, which occurs as a consequence of many overlapping independent disease mechanisms that result in a common pathway leading to joint degeneration.1 2 Although many risk factors for developing OA have been identified, the mechanisms underlying OA pathogenesis are poorly understood.3 4 There has been considerable interest in changes in chondrocyte phenotype in OA,5 6 7 8 9 and based on studies using immunolocalisation and in situ hybridisation to demonstrate gene products, such as collagen X in samples of OA cartilage, chondrocyte hypertrophy has been proposed as a mechanism underlying cartilage failure.7 10 11 12 13 14 15 16 17 18 This concept was supported by the demonstration that OA-like lesions were produced in transgenic mice in which altered transforming growth factor β (TGFβ) signalling was driving articular chondrocytes to become hypertrophic.19 20 Together these observations led to the hypothesis that there was a causal link between chondrocyte hypertrophy and OA. However, it is not known whether chondrocyte hypertrophy is common in human OA, and if present, if it is active in all cartilage, or restricted to sites of cartilage fibrillation.

The rationale behind this study was that if chondrocyte hypertrophy was the common feature contributing to cartilage degeneration in human OA, then changes in gene expression associated with this process should be active in most OA cartilage, particularly at a late stage of the disease. The expression of selected genes strongly upregulated in growth plate chondrocytes at different stages of differentiation were therefore determined in samples of OA cartilage taken from knees at joint replacement and in age-matched control cartilage. Furthermore, we investigated two sites in each joint to determine whether evidence for chondrocyte gene expression changes correlated with the level of degenerative change at each site.

Materials and methods

Study design

Paired osteochondral samples were taken from the major loaded inferior area of the medial femoral condyle (MFC) and the lower loaded central site of the lateral posterior condyle (LPC) with ethical approval and fully informed consent. At inferior sites with erosions, samples adjoining the erosion were taken. Samples were from patients undergoing total knee arthroplasty for clinically and radiologically diagnosed osteoarthritis (OA)21 (n = 25; age range 60–84 years, mean 69.4; male:female ratio 1:1.5). Age-matched control articular cartilage was from patients undergoing above-knee amputation with no history of joint disease (n = 8; age range 56–82 years, mean 63.7 years; male:female ratio 1:1.6). The samples were harvested within 1 h of surgery, placed in RNAlater and transferred on ice to the laboratory. Full-thickness articular cartilage for mRNA analysis was removed from the selected sites with care to exclude the calcified cartilage zone, frozen in liquid nitrogen and stored at −80°C. Osteochondral samples adjacent to the selected sites were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 24 h at 4°C for histology and decalcified in 20% EDTA (pH 7.2) at 4°C. Care was taken that the adjacent sample for histology was grossly similar to that used for gene expression analysis.

Gene expression analysis

Total RNA was extracted from 200–300 mg articular cartilage with Trizol reagent after homogenisation with a Braun Mikrodismembanator, followed by RNA isolation using RNeasy minicolumns and reagents (Qiagen, Crawley, Surrey, UK). Any residual DNA present was removed by performing an on-column DNAse digestion using an RNase-Free DNAse kit (Qiagen). The quantity of the RNA eluted from the column was assessed by spectrophotometry. RNA samples were also analysed for quality using the RNA 6000 Nano LabChip Assay kit on an Agilent 2100 bioanalyser. cDNA was synthesised from approximately 100 ng of total RNA using global amplification methodology.22 Globally amplified cDNAs were diluted 1:1000, and 1 μl aliquots of the diluted cDNA were amplified by polymerase chain reaction in a 25 μl reaction volume in an ABI Prism 7700 Sequence Detection System (Perkin-Elmer, Applied Biosystems, Warrington, UK) with an SYBR Green Core Kit (Eurogentec, Seraing, Belgium), with gene specific primers (table 1) designed with a 3′ bias22 using ABI Primer Express software. Relative expression levels were normalised with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and calculated with the 2−ΔCT method.23

Table 1

Primer sequences used in real-time PCR relative quantification (all primers were from Invitrogen, Paisley, Scotland)


Decalcified osteochondral specimens in paraffin wax were cut (7 μm sections) and stained with haematoxylin and eosin and with 0.1% safranin-O and graded for OA damage using a standard score.24

Statistical analysis

Differences in histological grade between paired samples of cartilage from the same joint were tested using the Wilcoxon signed rank test. To identify the effect of disease state and site within the joint on gene expression we applied mixed effects linear regression models, and possible links between the expression of different genes were assessed using Pearson’s correlation. To verify potential groups of coexpressed genes principal components analysis was also performed.


Histological grading of the paired samples of cartilage from OA joints showed the difference in tissue damage between the medial inferior femoral cartilage and the lateral posterior femoral cartilage, representing high-loaded and low-loaded sites respectively (p<0.001). In contrast, in control joints there was no difference in grading between the corresponding sites. The results also showed that the high- and low-loaded sites from OA joints were both more damaged than site-matched cartilage from the control joints (p<0.001) (fig 1).

Figure 1

(A and B) Results of histological grading of age-matched control cartilage, n = 8 (A) and OA, n = 22 (B) articular cartilage samples from areas of low and high load. The histological grading was performed using a previously described classification.24 Grade 0, histologically normal cartilage; grade 1, cartilage surface irregularities and loss of metachromasia adjacent to superficial chondrocytes; grade 2, fibrillation of cartilage and the formation of some chondrocyte clusters, with minor loss of surface cartilage; grade 3, gross cartilage abnormalities including loss of superficial cartilage, extension of fissures close to subchondral bone and a large number of chondrocyte clusters.

Gene expression was investigated in the paired samples of articular cartilage from the selected sites on OA and age-matched control joints. The genes were selected for those highly expressed in normal articular chondrocytes and generally not expressed by other cell types (SOX9, collagen II (COL2A1) and aggrecan)5 and which had also previously been reported to change in OA.7 9 25 26 27 Collagen I (COL1A1) was also investigated, as it is frequently reported to be increased in OA.28 29 The differentiation of chondrocytes in OA was investigated by determining the expression of genes including matrilin-1, PTHrPR and Bcl-2, which are all expressed highly in the proliferative zone of growth plate30 31 32 and MMP-13, VEGF and collagen X, which are all expressed in terminally differentiated chondrocytes in the hypertrophic zone of the growth plate.14 33 34 Although none of these genes alone is definitive of the process, these six genes were selected as most would be predicted to be upregulated if chondrocyte further differentiation was a common factor in chondrocytes in late OA.

The expression of SOX9 (p<0.001) and aggrecan (p<0.001) were greatly reduced in OA cartilage in comparison with control tissue (control SOX9 38-fold higher and aggrecan fourfold higher than in OA), and they were downregulated at the intact, as well as at the damaged site (figs 2B and C). In contrast COL2A1 was strongly expressed in both OA and control cartilage, but with a wider range of gene expression in OA cartilage. Thus, while the mean showed a 2.5-fold increase in collagen II gene expression in OA, this was not significantly higher than in control cartilage (p<0.16) (fig 2A). However, paired cartilage samples from 11/25 joints showed COL2A1 expression more than two standard deviations (SD) above the age-matched control mean, which suggests that COL2A1 gene expression was significantly increased in a subset of patients. Again, there was no significant difference in gene expression between the paired samples from sites of high and low load in OA.

Figure 2

Gene expression analysis of collagen II (A), aggrecan (B), SOX9 (C) and collagen I (D) in age-matched control and osteoarthritis (OA) chondrocytes from different load-bearing areas of the knee (LPC, lateral posterior condyle; MFC, medial femoral condyle). These genes represent markers of a chondrogenic and fibrochondrogenic phenotype.

The COL1A1 gene was strongly expressed in the OA cartilage, and the mean expression was 17-fold higher than in the age-matched controls, but as there was a great variation in expression amongst the OA joints, this was not significantly higher (p<0.23) (fig 2D). However, paired samples from nine of 25 joints showed COL1A1 expression more than 2 SD above the age-matched control mean, which suggested that there was an increase in COL1A1 expression in a subset of patients, and in these patients the expression of COL1A1 approached a similar level to COL2A1 and was about 10% of the reference gene GAPDH. The results also suggested a trend (p<0.07) towards an increase in COL1A1 expression at the high-loaded compared with the low-loaded site of the OA joints. Interestingly, the patients with raised COL1A1 were not the same group as those with raised COL2A1, with only three patients having increases in both.

Of the genes known to be upregulated during differentiation in the proliferative zone of growth plate, matrilin-1 was expressed at a low level in control cartilage and it was even lower in OA (p<0.001) (control 6.5-fold higher than OA) and no difference between sites in either. Both Bcl-2 and PTHrPR were expressed at slightly higher levels than matrilin-1 and with no significant difference in PTHrPR expression between OA and control, while Bcl-2 expression was lower in OA (control 5.6-fold higher than OA, p<0.002) with no difference in expression between sites in OA and controls. There was a trend (p<0.07) for PTHrPR to be increased in the more highly loaded site in both OA and controls. The results thus showed that even though OA cartilage from sites of tissue damage contained many clusters of chondrocytes suggesting cell proliferation, there was no evidence of any increased expression of these selected genes, which are upregulated during the proliferative phase of differentiation of growth plate chondrocytes (figs 3A-C).

Figure 3

Gene expression analysis of matrilin-1 (A), Bcl-2 (B), PTHrPR (C), collagen X (D), MMP-13 (E) and VEGF (F) in age-matched control and osteoarthritis (OA) chondrocytes from different load-bearing areas of the knee (LPC, lateral posterior condyle; MFC, medial femoral condyle). These genes represent markers of a proliferative and hypertrophic chondrocyte phenotype.

The expression of VEGF and collagen X were determined as these are genes expressed strongly during chondrocyte hypertrophy (figs 3D–F).14 33 34 The analysis showed that both were expressed at a low level in control cartilage and were even lower in OA cartilage. VEGF expression in controls was 23-fold higher than in OA, (p<0.05) and collagen X expression in control was 14-fold higher than in OA, (p<0.002) and they were both similarly expressed at the different sites. In contrast, MMP-13 expression was significantly increased in OA cartilage (5.3-fold, p<0.003) compared with control tissue and the increase was at both sites. In addition to chondrocyte hypertrophy, MMP-13 is also expressed in articular chondrocytes in response to inflammatory mediators and during tissue remodelling in experimental OA.35 Thus although the results showed increased expression of MMP-13 in OA cartilage, this appeared not to be associated with hypertrophy, as there was no accompanying changes in the other genes active during chondrocyte hypertrophy in the growth plate.

To identify any correlation between the changes in gene expression the data were analysed using Pearson’s correlation (table 2). The results were also subject to principal components analysis, and a multivariate analysis was used to validate the findings of the correlation analysis. Most of the variation (43%) in the dataset was suggested to result from correlation between genes that were downregulated in OA. A smaller contribution to the variation (22%) was from genes that were either significantly upregulated, or had some trend towards upregulation in OA—namely, MMP-13, COL1A1 and COL2A1. The correlation analysis showed no link between the expression of genes associated with chondrocyte further differentiation and gene expression found in OA chondrocytes.

Table 2

Analysis of correlation of expression between genes


Previous studies on cartilage from OA joints have been reported to provide evidence for chondrocyte hypertrophy based on immunohistochemistry, or in situ hybridisation,10 12 13 in which genes, or proteins associated with growth plate chondrocytes, were detected. The evidence presented rarely gave an indication of how common the detected changes were across the joint surface, or how common the finding was in human OA. In this study, we examined quantitative changes in gene expression between control and OA knee cartilage to establish how frequent changes in gene expression were among different patients with OA. Furthermore, in order to distinguish changes at the sites of cartilage damage from those that were throughout all the cartilage in an OA joint, we took cartilage samples from the medial inferior femoral condyle, an area of high load, which was heavily fibrillated and from the lateral posterior femoral condyle, an area of low load, which was macroscopically intact in most OA knee joints.

The results of gene expression analysis showed major changes in OA cartilage, but based on the selected genes investigated did not show any pattern similar to growth plate-like differentiation. Importantly, elements of the changes were common across the 25 OA joints and therefore appeared to reflect features of a common OA process. However, the elevation of COL1A1 and COL2A1 expression in different subsets of patients showed there was heterogeneity in the detailed processes occurring in different OA joints. Changes in the expression of these major chondrocyte genes have been noted in previous studies of human OA cartilage.7 9 25 26 27 A detailed comparison between differently damaged sites in each OA joint has not been reported and it is therefore an important finding in this study that the gene expression profiles at both the damaged and intact sites of cartilage on an OA joint were similar to each other and significantly different from those of age-matched controls. This suggested that the expression changes in key genes in cartilage from late-stage OA joints were throughout the joint and were not localised to the sites of tissue damage. Some of the major changes in gene expression detected may therefore reflect a process that predisposes the cartilage to damage, rather than being a consequence of tissue damage.

Although our results showed, that for the genes we investigated, there was similar expression in fibrillated and intact cartilage within the same joint, this does not imply that there was no difference between all other genes expressed at the two sites. Recent studies have reported differences in gene expression in OA joints in cartilage samples harvested from minimally damaged and highly damaged sites within the same joint.36 37 However, in both these studies no comparison was made with expression in normal cartilage. Our study qualifies these findings to show that major differences in expression exist between age-matched control and OA cartilage irrespective of the extent of damage and that all cartilage on an OA joint whether fibrillated, or intact, is abnormal. The importance of our results is that they show that in most OA joints there was major downregulation of the expression of key chondrocyte genes, and this occurred in both fibrillated and intact cartilage. It requires further analyses to characterise this change fully and begin to understand its cause.

SOX9 was found to be more strongly downregulated in the OA samples than in some previous reports,26 but there was no correlation with the expression of collagen II, which has been shown to be SOX9 dependent.38 This lack of correlation between SOX9 and COL2A1 has been noted in previous studies6 26 and suggests that although SOX9 appears necessary for cartilage formation,39 there are other factors that regulate collagen II transcription, which are active in OA joints.

An important factor in this analysis is that it provides an assessment of gene expression in cartilage late in the OA process at joint replacement. As OA may develop over 10–20 years the changes detected may differ from those at the beginning of the disease. Other studies have investigated early stages of OA using experimental animal models, which result in OA-like degeneration.40 41 42 The early stages typically showed both enhanced matrix synthesis (collagen II and aggrecan) and increased matrix degradation with high levels of metalloproteinase expression.28 43 44 45 46 This analysis showed some interesting similarities with the results from early experimental OA and some notable differences. A clear similarity is that the biochemical and gene expression changes in early experimental OA have frequently been reported in both intact and fibrillated cartilage, just as noted here in late OA. So there is evidence in early OA of an active process affecting all cartilage in the joint. Other comparisons show that COL2A1 expression was increased in early experimental OA43 and in our study its expression was also increased in the OA cartilage of some patients with late OA. Aggrecan expression has also been reported to be elevated in early experimental OA,43 but clearly in the late OA samples analysed here its expression was suppressed. This suggests that its expression may change during progression of OA from initial enhancement, to later suppression. The expression of MMP-13 was greatly increased in the late OA cartilage and this has been similarly reported in early experimental OA cartilage.35

The gene expression changes in OA may thus reflect a pattern of failed remodelling of the joint, which at the late stage shows abnormal matrix production, both of collagen II and aggrecan, and potentially enhanced matrix turnover. There was also evidence of disease subsets, with either high expression of COL1A1 and COL2A1, but rarely both. These may reflect different stages, or different pathways of disease progression, such has been suggested for increased collagen II expression in hypertrophic OA47; this requires further analysis.

The gene expression of collagen X was low in control cartilage and even lower in late-stage OA, which appears to conflict with published evidence of collagen X in OA cartilage.11 12 The gene expression we measured reported the activity on the day of the operation, whereas immunolocalisation reports protein accumulated in the tissue over a period of possibly years before the tissue removal. The detection of type X in some cartilage sections may thus reflect its retention in the matrix following expression at some earlier stage in the disease. It is also possible that in some joints a proportion of cells early in OA, terminally differentiate, become hypertrophic and produce type X collagen, but undergo apoptosis and are no longer represented in the gene expression analysis of late OA tissue. However, this would also imply that this pathway does not affect a large proportion of chondrocytes that survive and remain viable within the tissue. These results may imply that chondrocyte terminal differentiation is not a common pathogenic mechanism in OA. It therefore remains important to establish how far the major changes in gene expression detected are reversible in the tissue, as there is now good evidence that provided with appropriate signals, OA chondrocytes,48 49 can reinitiate cartilage matrix production and this may provide a route to new strategies for initiating endogenous repair.


View Abstract


  • Funding Funding from Arthritis Research Campaign (Clinical Research Training Fellowship to CJB), the Wellcome Trust (Research Leave Fellowship GR067462MA to PDC) and the Wellcome Trust for support for The Wellcome Trust Centre for Cell-Matrix Research and the Research Councils (BBSRC, MRC, EPSRC) for support for UK Centre for Tissue Engineering.

  • Competing interests None.

  • Ethics approval Approval from the Greater Manchester Ethics Committee.

  • Patient consent Patient consent received.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.