Article Text

Differential upregulation of the three transforming growth factor beta isoforms in human osteoarthritic cartilage
  1. M Pombo-Suarez1,
  2. M T Castaño-Oreja2,
  3. M Calaza1,
  4. J Gomez-Reino1,3,
  5. A Gonzalez1
  1. 1
    Laboratorio Investigacion 2 and Rheumatology Unit, Hospital Clinico Universitario de Santiago, Santiago de Compostela, Spain
  2. 2
    Department of Morphological Science, University of Santiago de Compostela, Santiago de Compostela, Spain
  3. 3
    Department of Medicine, University of Santiago de Compostela, Santiago de Compostela, Spain
  1. Dr A Gonzalez, Laboratorio de Investigacion 2, Hospital Clinico Universitario de Santiago, Travesia de Choupana sn, 15706-Santiago de Compostela, Spain; Antonio.Gonzalez.Martinez.Pedrayo{at}; anlugon{at}


Objectives: Decreased levels of transforming growth factor beta (TGFβ) have been related to the failure of cartilage repair in experimental models of osteoarthritis. This study aimed to examine this aspect of osteoarthritis in human cartilage.

Methods: Cartilage samples were obtained from 11 patients with hip osteoarthritis and 11 patients with femoral neck fracture who were undergoing total hip replacement. Gene expression of the three TGFβ isoforms, collagen type II (COL2A1) and aggrecan (AGC1) was analysed by reverse transcription quantitative PCR and immunohistochemistry.

Results: Expression of the three TGFβ isoforms was increased in osteoarthritis cartilage. The upregulation was more marked for the TGFβ3 isoform (2.3-fold) than for TGFβ1 (1.6-fold) or TGFβ2 (1.7-fold). The messenger RNA levels of TGFβ1 and TGFβ2 were strongly correlated in osteoarthritis cartilage (rs  =  0.83, p = 0.002), but levels of TGFβ3 were uncorrelated with any of the two other TGFβ isoforms. Immunohistochemistry showed an extension of immunoreactivity for the three TGFβ isoforms to more chondrocytes and to deeper cartilage layers in the more severe osteoarthritis lesions. No correlation of TGFβ isoforms with COL2A1 or AGC1 expression levels was found.

Conclusions: The three isoforms of TGFβ were differentially upregulated in late osteoarthritis in relation to an increased percentage of TGFβ-positive chondrocytes. These results indicate that cartilage damage progresses in spite of the TGFβ stimulus for cartilage anabolism and that other causes of the failure to cope with the increased cartilage catabolism of osteoarthritis should be investigated.

Statistics from

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.

Primary osteoarthritis is the most common form of arthritis and is a leading cause of chronic disability. Its aetiology and pathogenesis are complex and are incompletely known. Current knowledge indicates that osteoarthritis lesions are the consequence of a failure of the anabolic response to cope with increased cartilage damage and degradation. Mechanical overload, chondrocyte senescence, inflammation, oxidative stress and other sources of chronic damage increase cartilage catabolism. The anabolic response includes chondrocyte activation, hypertrophy and dedifferentiation accompanied by an enhanced production of extracellular matrix components.

Chondrocyte culture studies and animal experiments have shown that transforming growth factor beta (TGFβ) is a potent inductor of the chondrocyte anabolic response and a strong antagonist of tumour necrosis factor alpha and IL-1-mediated cartilage catabolism.13 Therefore, TGFβ is considered pivotal in osteoarthritis cartilage maintenance. Consistent with this idea, TGFβ is upregulated in an experimental model of osteoarthritis, and blocking of TGFβ results in increased proteoglycan loss and reduced thickness of cartilage.4 Also, an allele of asporin, a small leucin-rich proteoglycan that sequesters TGFβ in the extracellular matrix, increases osteoarthritis susceptibility by enhanced binding to TGFβ.5 6 However, TGFβ is almost absent in the osteoarthritis cartilage of two mouse models7 and TGFβ1 is reduced in human osteoarthritis cartilage when analysed by immunofluorescence.8 Overall, the suggestion has been that decreased expression of TGFβ contributes to osteoarthritis by inefficient induction of the anabolic response.2 In the present study, we have addressed this question by analysing messenger RNA and protein expression of the three isoforms of TGFβ in human osteoarthritis cartilage. Contrary to the above hypothesis, the three TGFβ isoforms were overexpressed and differential regulation was also found.


Cartilage samples

Samples were obtained from the femoral heads of 11 osteoarthritis patients and 11 hip fracture patients (clinical and demographic data available in supplementary table 1, available online only). Primary osteoarthritis was diagnosed by anamnesis, clinical examination, radiographic findings and gross pathological findings at the time of joint removal. Patients with hip fracture were selected to exclude previous joint disease. This study was approved by the ethical committee for clinical research of Galicia and all participants gave their written informed consent.

Table 1 Extent of immunostaining for the three TGFβ isoforms in hip cartilage in function of osteoarthritis damage

Cartilage dissection and evaluation

Cartilage was graded by a modified macroscopic visual Collins’ scale that classifies lession from grade 0, without lesions, to grade 4, with gross changes, as described.9 Afterwards, cartilage was removed using a scalpel, chopped and snap-frozen in liquid nitrogen within 6 h of surgery. Care was taken to exclude fibrocartilage and areas surrounding osteophytes. Tissue pieces were stored at −80°C until further processing. A fragment was fixed in 10% formalin for histological study.

RNA extractions and reverse transcription quantitated PCR

RNA extractions from articular cartilage were performed with the addition of a DNase digestion step, as described.9 A total of 250 ng RNA was reverse transcribed to complimentary DNA using transcriptor reverse transcriptase (Roche, Barcelona, Spain) and random hexamers in a total volume of 20 μl. Each PCR was performed with 1 μl of the above reaction. Primer sequences and PCR conditions are specified in supplementary table 2 (available online only). All primers were in different exons to help prevent amplification of genomic DNA. PCR were performed using the DyNAmo SYBR Green qPCR kit (Finnzymes, Espoo, Finland) and the Chromo4 real-time PCR detection system (MJ Research, Basel, Switzerland). Reactions were run in duplicate and all samples were analysed in the same run. Duplicates showed coefficients of variation between 0.2% for TGFβ2 and 0.4% for TGFβ1. Control for genomic DNA contamination was performed in a well without reverse transcription for each gene-specific PCR. The specificity of amplification was confirmed by melting curve analysis and by DNA electrophoresis in agarose gels.

Data analysis

PCR efficiencies were over 95% for all primer sets. Raw Ct values were obtained from the logarithmic amplification phase using Opticon Monitor version 3.0 software. Ct values were adjusted for efficiency and normalised by the combined expression of three reference genes (see supplementary fig 1, available online only). The reference genes were TBP, RPL13A and B2M, which have been validated specifically for hip cartilage studies.9 Relative mRNA levels from osteoarthritis and control samples were compared using the delta-delta-Ct method. Normalisation by the reference genes and the determination of Ct values were performed using qBase software.10 Statistical analysis was done with the Mann–Whitney U test and the Spearman rank correlation test, performed with Statistica, version 7.

Figure 1

Increased expression of transforming growth factor beta (TGFβ) type 1 (A), TGFβ2 (B) and TGFβ3 (C) in osteoarthritis cartilage compared with controls without osteoarthritis. The expression quantitative PCR levels are represented for each sample. Values have been normalised to the combined expression of three reference genes, TBP, RPL13A and B2M, which have been specifically validated for hip osteoarthritis studies. Analysis was performed using qBASE software. Horizontal bars represent the geometric mean. OA, osteoarthritis.


Cartilage fragments with subchondral bone attached were fixed in 10% formalin for 24 h. Cartilage was detached from bone, dehydrated and embedded in paraffin wax. Perpendicular to articular surface sections (5 μm) were placed on microslides, deparaffinised, rehydrated and stained with haematoxylin, fast green and safranin O. Contiguous sections were deparaffinised, rehydrated in phosphate-buffered saline and incubated with monoclonal antibodies in a humid chamber overnight at 20–22°C. TGFβ1, TGFβ2 and TGFβ3 monoclonal antibodies MAB240, MAB612 and MAB643, respectively, from R&D Systems (Abingdon, UK) were selected for their low crossreactivity between the TGFβ isoforms. In particular, no crossreactivity between the three TGFβ isoforms has been detected by ELISA over 2% with these antibodies (data provided by R&D Systems), and was exceptionally over 6% in Western blots with some of these or several other commercial anti-TGFβ antibodies.11 A negative control staining is shown in supplementary fig 2 (available online only). Sections were treated with peroxidase-blocking reagent (Dako, Glostrup, Denmark) for 10 minutes, incubated with EnVision Detection Systems peroxidase/DAB, rabbit/mouse (Dako) for 30 minutes and finally with 3,3′-diaminobenzidine tetrahydrochloride (Dako) for 10 minutes. Controls to determine staining specificity were performed without primary antibodies. All these controls were negative. The extent of staining was scored semiquantitatively to reflect the frequency of chondrocytes showing immunoreactivity and their localisation. The score was as follows: + isolated positive chondrocytes in the I/II zone, ++ most of the chondrocytes in the I/II zone are positive, +++ positive chondrocytes in zone III and ++++ positive chondrocytes in all cartilage zones. The intensity of staining in chondrocytes was not considered.

Figure 2

Analysis of transforming growth factor beta (TGFβ) isoforms in cartilage sections in function of osteoarthritis severity. A progressive extension of immunoreactivity for the three TGFβ isoforms to a larger percentage of chondrocytes is apparent. First lane, undamaged cartilage; second lane, slight damage; third lane, moderate lesions; fourth lane, severely damaged cartilage. The first column shows haematoxylin, fast green and safranin O staining. The following columns show immunohistochemistry staining for TGFβ1, TGFβ2 and TGFβ3, respectively. The cartilage surface is oriented to the top of the figure. Objective magnification was 10× for safranin O staining and 20× for immunocytochemistry. Cartilage from osteoarthritis patients was not used for the analysis of undamaged sections.


Analysis of RNA expression in 11 osteoarthritis cartilage and 11 normal cartilage samples showed expression of the three TGFβ isoforms. The levels of expression were significantly higher in osteoarthritis cartilage than in normal cartilage (fig 1).9 10 TGFβ3 showed the largest increase, with expression 2.3 times that observed in control cartilage (p = 0.009). The two other TGFβ isoforms were also more abundant in osteoarthritis cartilage than in control cartilage, TGFβ2 by 1.7 times (p = 0.03) and TGFβ1 by 1.6 times (p = 0.045). The similar increase in the expression levels of TGFβ1 and TGFβ2 isoforms suggested their possible co-regulation. Consistent with this idea, their ranked levels showed a strong positive correlation in osteoarthritis cartilage (rs  =  0.83, p = 0.002). However, they were uncorrelated in normal cartilage. In contrast, levels of TGFβ3 were not correlated with levels of TGFβ1 or TGFβ2 in osteoarthritis cartilage or in normal cartilage.

Immunohistochemistry of the three TGFβ isoforms in cartilage sections was in agreement with the qPCR results (fig 2). In normal cartilage (including only fully preserved cartilage from hip fractures), the three isoforms were present in only a fraction of chondrocytes in the superficial cartilage layers. In contrast, osteoarthritis cartilage showed a more widespread expression. This change was progressive, following the severity of the lesions. There was an increase in the proportion of chondrocytes showing immunoreactivity for the three TGFβ isoforms and also an extension of positive cells to deeper cartilage layers (table 1). In the most severely affected cartilage sections immunoreactivity was present in most chondrocytes, even in the deepest cartilage layers. Chondrocytes showing staining for TGFβ1 were slightly more abundant throughout. Staining for TGFβ2 and TGFβ3 was similar, with a possible excess of chondrocytes positive for TGFβ2, at least in some subjects (table 1).

All the isoforms were preferentially detected in the cytoplasm of chondrocytes rather than in the extracellular matrix, which only showed diffuse and low staining (fig 2). We also examined whether there was a correlation between the COL2A1 or AGC1 mRNA levels and the expression of TGFβ isoforms, but none was found.


The quantitative evaluation of expression levels requires careful attention to technical details. We have used precise relative quantification of mRNA levels by qPCR by combining three reference genes along with other measures to improve the accuracy of results that we have previously identified.9 The three reference genes we have used, TBP, RPL13A and B2M, show stable expression in hip cartilage and the combined use of them provides a more accurate base for quantification.9 This has allowed us to identify a significant increase in the expression of the three TGFβ isoforms in cartilage from advanced hip osteoarthritis in comparison with normal cartilage. The changes were modest, approximately a 50% increase for TGFβ1 and TGFβ2 and slightly more than twice normal levels for TGFβ3. These changes probably pertained to the larger percentage of chondrocytes showing immunoreactivity for the three TGFβ isoforms in the histology study of osteoarthritis cartilage. Positive staining extended to the lower layers of cartilage in specimens with severe damage in the superficial layers. Therefore, the increased TGFβ expression was due to a larger percentage of cells expressing the TGFβ genes both in superficial and deep cartilage layers. It should be noted that some of our non-osteoarthritis samples could be from subjects affected by osteoporosis as they were obtained from hip fractures, but TGFβ levels in osteoporotic bone or serum have been described in most reports either as unchanged or as increased.12 13 Therefore, the increase we have found seems unrelated to this possibility.

An upregulation of TGFβ expression that persists to late phases of osteoarthritis is consistent with the role of TGFβ as an inducer of the anabolic response and inhibitor of the tumour necrosis factor alpha and IL-1-mediated cartilage catabolism.2 3 However, it is against the idea that TGFβ vanishes with osteoarthritis evolution and that this downmodulation contributes to cartilage lesions. This scenario was developed from data obtained with two murine models of osteoarthritis, in which TGFβ expression almost completely disappears before cartilage erosion.7 In contrast, a different mouse model shows a persistent upregulation of TGFβ with a larger increase of the TGFβ3 isoform.4 Previous studies in human osteoarthritis cartilage are scant and inconclusive. An immunofluorescence study found a slight decrease in the intensity of TGFβ1 staining and no appreciable changes in the other two TGFβ isoforms.8 In contrast, a recent publication showed overexpression of TGFβ1 and TGFβ3 mRNA in osteoarthritis cartilage, but only two control samples were used for comparison with six patient samples.6 The sustained overexpression of the three TGFβ isoforms we have observed indicates that stimulatory signals persist during osteoarthritis evolution and that chondrocytes remain able to respond to them. This is especially clear considering that overexpression required chondrocyte clonal proliferation and de novo expression in areas that do not express TGFβ isoforms in normal cartilage. However, these results do not necessarily imply that the increase in TGFβ is effective in cartilage protection. Available evidence on this point is inconclusive.7 8 1416 All in all, our data indicate that insufficient production of extracellular matrix components by chondrocytes in late osteoarthritis is unrelated to the regulation of TGFβ expression, although it is unclear if insensitivity to TGFβ signals is a contributing factor.

The three isoforms of TGFβ have overlapping but not identical activities in a variety of systems.1719 This probably explains how instances of differential expression of the three TGFβ isoforms abound.17 The marked and independent upregulation of TGFβ3 that we have found in osteoarthritis cartilage indicates that it responds to regulatory signals that are not shared with the two other isoforms. It also shows that it is inaccurate to study only TGFβ1 to define the role of TGFβ in osteoarthritis.

In conclusion, our results have shown that TGFβ isoforms are upregulated in osteoarthritis up to late phases of cartilage damage. This was reflected by the expression of these cytokines by more chondrocytes, including those from deeper cartilage layers. Two of the isoforms were co-regulated, TGFβ1 and TGFβ2, whereas TGFβ3 expression was independent and more markedly upregulated. This differential regulation can have unsuspected consequences in the osteoarthritis process as the isoforms are not completely interchangeable. These data indicate that studies in the pathogenesis of osteoarthritis need a more accurate definition of the role of TGFβ isoforms in cartilage maintenance.


The authors would like to thank sample donors for their generosity, Dr Fernando Baltar-Tojo for providing access to surgery samples, Isabel Castro-Perez and Maria Dolores Alvarez-Vilariño for collecting surgery material and Cristina Fernandez and Dolores Fernandez-Roel for outstanding technical support.


Supplementary materials


  • Competing interests: None.

  • Funding: This project was financed by the MMA Foundation (Madrid, Spain). MP-S received a bursary from the Fundacion Española de Reumatologia (Madrid) and a post-MIR contract from the Instituto de Salud Carlos III (Madrid).

  • Ethics approval: This study was approved by the ethical committee for clinical research of Galicia.

  • Patient consent: Obtained.

  • ▸ Additional tables and figures are published online only at