Induction of p16 at sites of cartilage invasion in the SCID mouse coimplantation model of rheumatoid arthritis


Rheumatoid arthritis (RA) is the most common type of chronic inflammatory arthritis in humans. It is characterized by synovial hyperplasia and pathologic immunologic phenomena mediating progressive joint destruction (1). There is increasing evidence that activated RA synovial fibroblasts (RASF) are key players in the process of joint destruction (2). However, the mechanisms of activation of RASF are not completely understood. The tumor suppressor protein p16 inhibits the binding of cyclin-dependent kinases 4 and 6 with the D cyclins. Thereby, it blocks the phosphorylation of the retinoblastoma (Rb) protein. Unphosphorylated Rb does not release E2F and does not induce G1 entry into the cell cycle (3). Therefore, p16 is a cell cycle inhibitor mediating G1 arrest.

Since it is encoded on human chromosome 9p in the INK4a/ARF locus, p16 is also called p16INK4a. It is worth noting that this genetic region is frequently mutated in patients with cancer (4). Interestingly, Taniguchi et al demonstrated that RASF were different from osteoarthritis synovial fibroblasts (OASF), normal synovial fibroblasts (NSF), and skin fibroblasts in terms of regulation of the p16INK4a tumor suppressor gene (5). Synthesis of p16INK4a was enhanced by serum starvation, γ-irradiation, and induction of senescence and resulted in irreversible growth arrest. In contrast, expression of p16INK4a was not increased by these treatments, nor was recovery of proliferation after addition of serum-containing medium prevented in OASF or embryonic lung fibroblasts. Intriguingly, administering p16 by adenoviral gene transfer to rats with adjuvant-induced arthritis and to mice with collagen-induced arthritis resulted in reduction of synovial cell hyperplasia and of mononuclear cell infiltration (5, 6).

The role of the tumor suppressor protein p16INK4a in RA is not well defined. Questions remain regarding the expression pattern of p16INK4a in RA tissues, which cell types express it, and the role of p16INK4a in joint destruction, specifically at sites of cartilage invasion. Therefore, in the present study we investigated the expression of p16INK4a in RA synovial tissues and in control tissues by immunohistochemistry analysis. Levels of p16 messenger RNA (mRNA) and protein in cultured RASF were investigated by real-time polymerase chain reaction (PCR) and by immunofluorescence, respectively. Finally, we explored whether RASF express p16INK4a at sites of cartilage invasion in the SCID mouse coimplantation model of RA.

Synovial tissue specimens were obtained from RA (n = 13), OA (n = 2), and normal synovia (n = 1) at the time of synovectomy, joint replacement, or trauma surgery. Samples were fixed in 4% neutral buffered formalin for 6–8 hours, dehydrated, and then embedded in paraffin using an automated tissue processor. All RA patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for diagnosis of the disease (7).

Synovial fibroblasts were obtained by enzymatic digestion of synovial tissue from RA patients (n = 10) and normal synovial tissue (n = 1) and cultured as previously described (8). HeLa cells were used as p16-positive control cells. Cultured synovial fibroblasts were detached with EDTA and stained with anti–HLA–DR (BD PharMingen, Basel, Switzerland) and anti–Thy-1 (clone ASO2; Dianova, Hamburg, Germany). Cells were analyzed on a FACSCalibur flow cytometer and data were processed using CellQuest software (Becton Dickinson, San Jose, CA).

For immunohistochemistry analysis, tissue sections were dewaxed in xylol and rehydrated in decreasing concentrations of ethanol, pretreated by microwave heating in 0.01M citrate buffer (pH 6.0), and kept at 70°C for 30 minutes in a heat incubator for antigen retrieval. Slides were blocked at room temperature with blocking solutions A and B (Vector, Burlingame, CA), and afterwards with 2% horse serum in 4% nonfat milk/Tris HCl (0.1M, pH 7.6). Primary anti-p16 antibodies (clone G175-1239; diluted 1:100, final concentration 5 μg/ml) (BD PharMingen) were applied at 4°C overnight. Mouse isotype-matched IgG sera were used as negative controls. The streptavidin–biotin detection system was used for amplification of the signal, and nitroblue tetrazolium/BCIP (Roche, Rotkreuz, Switzerland) served as color reagent. Tissue sections from normal uterus and pancreas were used as positive controls. For each patient, the percentage of positive cells was evaluated on at least 10 high-power fields (400× magnification). All immunohistochemical reactions were repeated at least twice.

To determine the cell types expressing p16, parallel sections of synovial tissue specimens were stained with antibodies against CD68 (dilution 1:60, final concentration 6 μg/ml) (Dako, Glostrup, Denmark). Specimens from SCID mice were analyzed for the proliferation marker Ki67 with anti-Ki67 antibodies (dilution 1:100, final concentration 1 μg/ml) (Dako). Incubation was performed overnight at 4°C. To amplify the signal, for the anti-CD68 and anti-Ki67 antibodies, the same streptavidin–biotin detection system was applied. In addition, SCID mouse sections were investigated with rabbit anti-active caspase 3 antibodies (dilution 1:100, final concentration 0.5 μg/ml) (Cell Signaling Technology [Frankfurt, Germany], distributed by Bioconcept, [Allschwil, Switzerland]) using microwave pretreatment, the streptavidin–biotin detection system, and diaminobenzidine as color reagent (Vector). Tissue sections from juvenile thymus, breast carcinoma, and a neurologic tumor were used as positive controls.

Immunofluorescence was also used for detection of p16INK4a. RASF (n = 10) or NSF (n = 1) from passages 3–7, as well as HeLa cells, were cultured on chamber slides (Labtec II; Nunc, Basel, Switzerland) and fixed with methanol/acetone (1:1) for 10 minutes at −20°C. After rehydration in buffer (0.1M Tris HCl [pH 7.6]), slides were blocked with 2% horse serum in 4% nonfat milk. Slides were incubated for 30 minutes with the above-mentioned anti-p16 antibodies (dilution 1:500, final concentration 1 μg/ml). Mouse isotype-matched IgG served as a negative control. For detection, Cy3-conjugated sheep anti-mouse serum (dilution 1:1,000, final concentration 0.14 μg/ml) (Dianova, Hamburg, Germany) was applied for 30 minutes. After mounting, slides were analyzed with a Zeiss immunofluorescence microscope and positive cells were counted.

For quantification of p16 mRNA by real-time PCR, RNA was extracted from RASF (n = 8), NSF (n = 1), and HeLa cells using the RNA Miniprep Kit (Stratagene Europe, Amsterdam, The Netherlands), with DNase treatment. After reverse transcription (RT) (MultiScribe; PE Applied Biosystems, Rotkreuz, Switzerland), the generated complementary DNA was amplified by quantitative real-time PCR using specific primers and a TAMRA/FAM label. Primers and probe for p16 were checked for specificity by GenBank analysis. Their sequences were as follows: forward primer 5′-CCA-ACG-CAC-CGA-ATA-GTT-ACG-3′, reverse primer 5′-GGG-CGC-TGC-CCA-TCA-3′, probe 5′-CAT-GAC-CTG-GAT-CGG-CCT-CCG-A-3′. Expression of 18S ribosomal RNA using predeveloped primer/probes (PE Applied Biosystems) served as internal standard. DNA contamination was evaluated by using the mRNA sample (non-RT control) as reaction template.

SCID mice were obtained from Charles River (Sulzfeld, Germany) and kept permanently under sterile conditions. Implantation of RASF (n = 5 patients, 20 animals) and NSF (n = 1 patient, 9 animals) together with normal human cartilage was performed as previously described (9). After 60 days, mice were killed, and the implants fixed in 4% buffered formalin and embedded in paraffin according to standard procedures. Paraffin-embedded sections were stained with hematoxylin and eosin or analyzed for expression of p16 by immunohistochemistry. Invasion into cartilage was quantified according to a semiquantitative score ranging from 0 to 4, referring to the number of invading cells and the number of affected cartilage sites.

For statistical analysis, the Mann-Whitney U test was used. P values less than 0.05 were considered significant. Data were expressed as the mean ± SEM.

In the experiments using immunohistochemistry and specific anti-p16 antibodies, p16INK4a was found to be expressed in the synovial lining as well as in the sublining layer of RA synovial tissues (Figure 1). No p16INK4a staining occurred in normal synovium (results not shown), and there was only limited expression in OA synovial tissues. A mean ± SEM of 6.5 ± 2.0% of all cells in RA synovial tissue expressed p16INK4a, and this expression was significantly increased when compared with OA and normal synovial tissues (combined mean ± SEM 0.25 ± 0.25%; P = 0.014). Moreover, in a subgroup of patients with RA (3 of 13), higher p16INK4a expression, ranging from 14% to 23%, was detected. However, the staining patterns of p16 and CD68 were different. In some regions CD68 and p16 colocalized, whereas in others CD68-negative areas were clearly positive for p16, indicating that CD68-negative fibroblast-like cells also expressed p16.

Figure 1.

Expression of p16 by immunohistochemistry. Only limited p16 expression occurred in osteoarthritis (OA) synovium, but strong expression was observed in uterus glands (positive control). Expression of p16 was detected in the synovial sublining of rheumatoid arthritis patients (RA1) as well as in the lining layer in these patients (RA2) (original magnification × 100 for OA; × 200 for RA1; × 400 for uterus and RA2).

In vitro, >98% of the investigated synovial cells were positive for the fibroblast marker Thy-1, and <1% stained positive for type II major histocompatibility complex molecules. As determined by immunofluorescence, the constitutive expression of p16INK4a protein was 6 ± 0.5% positive cells in untreated RASF and 4% in NSF, whereas all HeLa cells exhibited staining for p16INK4a (results not shown). Accordingly, the levels of expression of p16INK4a mRNA as quantified by real-time PCR were 100 times lower in RASF than in HeLa cells.

Cultured RASF from 5 patients, showing a mean level of 6% p16-positive cells, as well as NSF from 1 patient were engrafted under the renal capsule of SCID mice. In accordance with our previous findings (9), RASF invaded into the coimplanted cartilage significantly more strongly than did NSF (mean ± SEM invasion score 3.0 ± 0.18 for RASF, 1.2 ± 0.18 for NSF; P < 0.01). Most intriguingly, RASF expressed the tumor suppressor protein p16 in 40% of the cells at sites of cartilage invasion (Figure 2A). Implanted NSF demonstrated only minimal cartilage invasion, and <2% of the cells adjacent to the cartilage were p16 positive (Figure 2A). In addition, parallel SCID mouse sections were immunohistochemically stained for the proliferation marker Ki67 and for the cleaved fragment of caspase 3, indicating apoptosis. In both staining reactions, only negligible signals were detected (results not shown), suggesting neither proliferation nor the occurrence of apoptosis. Of note, control tissues were positive for the determined markers.

Figure 2.

A, Expression of p16 in the SCID mouse coimplantation model of rheumatoid arthritis (RA). Limited cartilage invasion of normal synovial fibroblasts (NSF) and only few p16-positive cells were detected. In RASF, p16 was induced at sites of cartilage invasion, since in vitro almost all of the same cells were p16 negative (magnification ×400). B, Quantification of p16-positive cells at sites of cartilage invasion. Values are the mean and SEM. The expression of p16 was significantly higher in RASF than in NSF (P < 0.05 by Mann-Whitney U test).

The present study reveals that p16 is induced in vivo in RASF at sites of cartilage invasion in the SCID mouse coimplantation model of RA. The in vivo induction mechanism is as yet unclear, but in vitro, Taniguchi et al (5) demonstrated induction of p16 in RASF by ionizing irradiation, low serum concentration, or sensescence. They proposed a novel treatment of RA, since in rat adjuvant arthritis and mouse collagen-induced arthritis they could reduce synovial hyperplasia and mononuclear cell infiltration by gene transfer of p16 (5, 6).

Herein we report that the cartilage invasion process appears to be associated with the expression of p16. We suggest that p16 could mediate arrest of the cell cycle at sites of cartilage invasion, since the cells were also negative for the proliferation marker Ki67. This notion is supported by the results of investigations by Jung et al (10) demonstrating low proliferation in human colorectal adenocarcinomas at sites of tumor invasion, in association with the expression of p16. Furthermore, in tumor cells the transfection of p16 also resulted in apoptosis (11). However, in our in vivo study the expression of p16 at sites of invasion was not correlated with the detection of apoptotic cells. This result is in accordance with the findings of other studies (5) demonstrating that in vitro p16 gene transfer of RASF inhibited only the cell cycle, but did not effect apoptosis. The finding that p16 is expressed in RA tissues is in contrast with the results of Taniguchi et al (5), who reported no p16 expression in 3 RA tissues, studied using Western blot analysis. However, in the present study, immunohistochemistry analysis clearly showed positive staining for p16 in ∼6% of the cells. Therefore, we hypothesize that in the study by Taniguchi and colleagues, the 3 RA samples investigated might not be representative for the expression of p16 in all subsets of RA tissues, at least in samples derived from Europe.

In summary, we have shown that p16 was induced at sites of cartilage invasion, most likely mediating cell cycle arrest, but not apoptosis. From these data we propose a role for the tumor suppressor protein p16 in RA, specifically in the process of cartilage invasion. We assume that p16-expressing and low-proliferating RASF, which have been shown previously to be positive for the apoptosis inhibitor sentrin (12) and negative for the tumor suppressor PTEN (13) and to strongly express matrix-degrading enzymes (14), represent a phenotype associated with the arrest of the cell cycle and thereby maintain the invasive phenotype of cells mediating joint destruction in RA.


We thank Maria Comazzi and Ferenc Pataky for their excellent technical assistance, and Diego Kyburz and Janine Rethage for help in the FACS analysis. Drs. Kuchen and R̂ihoŝková's work was supported by the EMDO Foundation. Dr. Seemayer's work was supported by the Swiss National Science Foundation and the EMDO Foundation.