Abstract
In its tissue-specific function as a collagen receptor of chondrocytes, cartilage annexin V (anchorin CII) occupies a key position in the organization of the cell-extracellular matrix (ECM) junction for the tissue. The general role of annexin V (Anx V) in other tissues suggests involvement in cellular secretory processes and in regulation of apoptosis. Immunohistochemical analysis of Anx V in growth plate cartilage, confirmed by in situ hybridization, suggests that Anx V is prominently expressed and forms a major constituent of growth plate chondrocytes. Anx V epitopes are also located in the pericellular matrix of hypertrophic cartilage. In adult articular cartilage the expression is downregulated, with the highest levels of immunostaining found in the upper third of the articular cartilage layers and almost no antigen found in the deep layers. Osteoarthritic (OA) cartilage is characterized by a significant upregulation of message and protein throughout the entire depth of the tissue, an accumulation of cytoplasmic annexin V epitopes, and a release of epitopes into the pericellular and interterritorial matrix, in part co-localized with granular structures. Therefore, Anx V expression and tissue distribution may serve as a histological marker for metabolic alterations and for changes in the cellular phenotype associated with OA.
THE ANNEXINS (Anx) are a family of calcium and phospholipid binding proteins (for review see Moss 1992). At least one of the 20 members of this family thus far described is expressed in almost every eukaryotic cell type from mammals (Barton et al. 1991), insects (Johnston et al. 1990), plants (McClung et al. 1994), and yeast (Ernst 1991), to primitive eukaryotes (Morgan and Fernandez 1995). As common as these proteins are, no single clear function for all has been established.
One member, Anx V, Mr 34-36 kD (depending on species), contains four conserved 70-amino-acid repeat domains found in all annexins (Fernandez et al. 1988). The ammonium sulfate protein crystal suggests a disk-shaped structure with a convex and a concave face (Huber et al. 1992). It has been proposed that a stable membrane-associated variant of Anx V exists that differs from the cytosolic soluble variant in the three-dimensional arrangement of a tryptophan residue at position 187 (Sopkova et al. 1993). Many characteristics have been demonstrated for Anx V, such as the binding of collagen (Mollenhauer and von der Mark 1983; von der Mark et al. 1991; Kirsch and Pfäffle 1992; Turnay et al. 1995), formation of calcium channels (Rojas et al. 1990; Berendes et al. 1993), inhibition of protein kinase C (Schlaepfer et al. 1992; Raynal et al. 1993), promotion of urokinasetype plasminogen activator synthesis and migration of keratinocytes (Nakao et al. 1994), binding of human platelets (Andree et al. 1985; Tait et al. 1995), and binding to the small hepatitis B virus envelope protein (Hertogs et al. 1993). In addition, elevated levels of Anx V have been shown in the pathology of cystic fibrosis (Gaspera et al. 1995). Increased levels of anti-Anx V antibodies have been found in the sera of patients with lupus erythematosus (Matsuda et al. 1994) and rheumatoid arthritis (Dubois et al. 1995).
Anx V has been used as a marker for apoptosis, or controlled cell death (Reutelingsperger and van Heerde 1997). In addition, the same authors postulated that Anx V might be involved in regulation of apoptosis. One of the features of apoptotic cells is the partial reversion of the polarity of the plasma membrane for its lipids. Apoptotic cells contain phosphatidylserine (PS) on the outer leaflet. Fluorescence-conjugated Anx V can be used to demonstrate this reversion because it binds to the apoptotic but not to the normal cell, making it a (commercially available) marker for early apoptotic stages. This hypothesis on the involvement of Anx V in apoptosis is based on the following observations. Extracellular Anx V protects activated platelets with high PS in the outer membrane leaflet from coagulation (Bevers et al. 1983). Erythrocytes show an age-dependent accumulation of PS. Anx V in the blood plasma may be used to prevent premature removal of those cells by the reticuloendothelial system (Connor et al. 1994). Anx V inhibits phagocytosis of apoptotic bodies, although not completely (Reutelingsperger and van Heerde 1997).
In cartilage, annexins II, V, and VI have been detected (Mollenhauer and von der Mark 1983; Genge et al. 1992; Suarez et al. 1993; Böhm et al. 1994). Anx V, in particular, has been localized to the chondrocyte surface (Mollenhauer et al. 1984). It was first named anchorin CII because of its location on the cell surface and its proposed function of anchoring the chondrocyte to its extracellular matrix (ECM) via binding to collagen Type II, a protein that constitutes the major portion of all ECM molecules in most cartilage tissues. The protein and the cDNA of anchorin CII have since been sequenced and found to be 100% homologous to Anx V from other tissues.
Fetal chicken chondrocytes have high expression levels of Anx V mRNA (von der Mark et al. 1991), which is markedly upregulated in chondrocytes from the hypertrophic zones of tibial growth plate cartilage (Kirsch et al. 1996). As described in fetal chicken, Anx V is a major constituent of matrix vesicles, plasma membrane elements released by growth plate chondrocytes (Genge et al. 1992). Here, Anx V contributes to calcium uptake by the calcifying vesicles (Kirsch and Wuthier 1994). Evidence for these findings in human tissues has not yet been published.
Adult articular cartilage, a bradytrophic (slowly metabolizing) tissue, is characterized by a significantly downregulated metabolism. Cellular turnover of molecules from the ECM, in particular collagen Type II, is minimal. Osteoarthritis (OA) is characterized by an increased turnover of molecules from the ECM. An imbalance of synthesis and degradation of aggrecan and collagen Type II, the major matrix molecules, results in progressive destruction of articular cartilage. The tissue gradually degenerates with the development of fibrillation, fissures, ulceration, and full-thickness defects. In a number of patients, secondary ossification takes place with the formation of osteophytes. Mineralizing matrix vesicles are deposited in the ECM of OA cartilage and may serve as initiation sites for pathological mineral deposits (Derfus et al. 1992). Although initially a focal event, the disease spreads across the entire joint and involves neighboring tissues such as bone, ligaments, tendons, synovium, and peri-articular muscles.
In this study we compared the expression of Anx V in cartilage obtained from OA patients at the time of knee arthroplasty with normal cartilage obtained from adult donors (39-93 years) with no known history of joint disease. Two samples from newborn donors and one fetal sample were included to obtain information on growing articular (epiphyseal) and growth plate cartilage. The expression of Anx V was upregulated in the diseased cartilage samples to a degree comparable to those of the fetal and newborn growing cartilage, which displayed very intense signals. This was true both for protein, as detected by immunohistology, and for message, as detected by in situ hybridization.
Materials and Methods
Reagents
Chemicals, either reagent or molecular biology grade, were purchased from Sigma Chemical (St Louis, MO) unless otherwise noted. Chondroitin ABC lyase (EC 4.2.2.2), keratanase (EC 3.2.1.103), and keratanase II were obtained from Seikagaku America (Rockville, MD).
Tissue Acquisition
Normal human cartilage samples were obtained through the Regional Organ Bank of Illinois from 21 donors within 24 h of their death and from six OA patients through the Department of Orthopaedic Surgery, Rush-Presbyterian-St. Luke's Medical Center, Chicago, Illinois, according to the institutions' protocols and with institutional approval. Of the 21 donors, 18 were adults (8 women and 10 men aged 39-93 years). In addition, the growth plate of two fetuses and the epiphyseal cartilage of one newborn were obtained for analysis.
Recombinant Human Annexin V (rhanxV)
RhanxV was produced from a bacterial strain transfected with an expression vector including the cDNA from human placental Anx V (purchased from American Type Culture Collection, Rockville, MD; ATCC 67916) as described elsewhere (Kaplan et al. 1988). E. coli TG1 containing the expression vector pKK233-2 with a 1.446-KB insert for the complete cDNA of Anx V was grown in medium supple-mented with ampicillin, treated for 2 hr with thiogalactoside, and harvested by centrifugation. Extraction of Anx V was performed by sonication of the bacteria in extraction buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 2 mM MgCl2), followed by centrifugation to remove the bacterial fragments. The soluble supernatant with the extracted rhanxV was used in the quality control of the antibodies applied in the histology.
Anx V Antisera
Polyclonal serum #9757 was raised against a 20-amino-acid synthetic peptide specific for the N-terminal sequence of human Anx V (NH2-M A Q V L R G T V T D F P G F D E R A D-COOH). The sequence was tested for specificity by the BLASTP program (Altschul et al. 1990) using the SWISSPROT database provided by NCBI.NLM.NIH.GOV and was found to be unique for human Anx V. The synthetic N-terminal peptide was coupled with bis-(sulfosuccinimidyl)-suberate to keyhole limpet hemocyanin according to the manufacturer's protocols (Pierce; Rockford, IL). Rabbits were immunized with the conjugate in complete Freund's adjuvant. The resulting antiserum showed crossreactivity to bovine and rat Anx V (87% homology in the corresponding sequence) but not to chicken Anx V (65% homology) (King et al. 1997). In some cases, serum IgG was purified by affinity chromatography on a protein A column according to the manufacturer's protocol (Pharmacia Biotech; Piscataway, NJ) before its application in immunohistochemistry. The polyclonal antiserum #8958 was raised against chicken chondrocyte membrane Anx V (Böhm et al. 1994).
Histological Processing
Full-thickness articular cartilage without the calcified cartilage or subchondral bone was sliced into approximately 3 × 3 mm2 pieces. The cartilage was fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin under RNase-free conditions. Sections of 6 μm were cut and processed for histology, immunohistochemistry, or in situ hybridization. For histological grading, the sections were stained with Safranin O and Fast Green and then graded using a modification of the scale of Mankin (Mankin et al. 1971). All cartilage from adult donors was considered normal (Mankin grades 0-5). Cartilage from the OA patients received grades equal to or higher than 6.
Immunohistochemistry
Rabbit anti-human Anx V peptide antiserum (#9757) was used as the source of the primary antibodies for indirect immunostaining. To increase the penetration of antibodies into cartilage tissue, deparaffinized sections were digested with keratanase (10 mU/ml), keratanase II (0.1 mU/ml), and chondroitinase ABC (10 mU/ml) in 100 mM Tris/50 mM sodium acetate buffer (pH 6.5) at 37C for 90 min. Sections were then incubated with nonimmune goat serum in PBS (1:100 dilution) for 30 min to block nonspecific binding sites. Primary antibody was applied at a dilution of 1:500 in 1% bovine serum albumin (BSA) in PBS overnight at 4C. The bound antibody was detected by biotinylated second antibody and an avidin-biotin-alkaline phosphatase reaction kit according to the manufacturer's protocols, including the inhibition of tissue-endogenous alkaline phosphatase with levamisole (Pierce). Alternatively, bound #9757 antibodies were detected with rhodamine-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch; West Grove, PA) in a 1:100 dilution and a 1-hr incubation at room temperature (RT). Control reactions were performed by incubating sections with (a) no primary antibodies, (b) normal rabbit serum (1:250 dilution) followed by second antibody, or (c) #9757 antiserum (1:250 dilution) preabsorbed with synthetic peptide (M A Q V L R G T V T D F P G F D E R A D) conjugated to BSA and followed with alkaline phosphatase-labeled goat anti-rabbit IgG.
Western Blotting
Following the methods of Towbin et al. (1979), samples (50 μg protein) were separated by SDS gel electrophoresis and electroblotted to nitrocellulose (Böhm et al. 1994). The Anx V-specific bands were identified with antiserum #9757 (1:400 in 3% skim milk, overnight incubation at 4C). The bound primary antibodies were detected with peroxidase-conjugated goat anti-rabbit IgG (1:500 in 3% skim milk, 4 hr incubation at RT) and developed with 4-chloro-1-naph-thol in PBS, pH 6.0.
Isolation of Chicken and Human Chondrocyte Membranes
Chicken chondrocyte membranes were isolated from adult chicken sternal cartilage as described (Böhm et al. 1994). In brief, sternal tissue was homogenized and extracted with 8.5% sucrose, 0.05 M Tris-HCl, pH 7.5. After the removal of coarse ECM fragments by low-speed centrifugation at 1500 × g, the cell membranes were collected by high-speed centrifugation at 40,000 × g. An additional centrifugation of the resuspended pellet in a stepwise sucrose gradient at 100,000 × g (Mollenhauer and von der Mark 1983) resulted in the enriched membrane fraction at the 17%/40% sucrose interface. Aliquots of this fraction were submitted to gel electrophoresis and Western blotting. Human chondrocyte membranes were prepared starting from freshly isolated primary human articular chondrocytes (Aydelotte and Kuettner 1988). The cells were mechanically disrupted in the extraction buffer, then submitted to differential centrifugation as described for the chicken membranes.
In Situ Hybridization
A 29-mer anti-sense oligonucleotide probe was designed with a sequence (5′-CCAGGGAAGTCAGTCACAGTGCCTCTGAG-3′) complementary to bp 106-134 of human Anx V mRNA, exon 3 (accession number U05761). The specificities and sequence homologies of this probe were compared with sequence data available from the EMBL/GenBank DDB database. Gel-purified oligonucleotide probes were 3′-endlabeled with 5′-[α-thiol-35S]-dCTP using terminal deoxynucleotidyl transferase (New England Nuclear; Wilmington, DE). The radiolabeled probe was hybridized to cartilage sections as previously described (Sandell et al. 1991). Autoradiographs were prepared to visualize mRNA probe hybridization by exposing sections to photographic emulsion (Kodak NTB2; Eastman Kodak, Rochester, NY) for 3 days at 4C. Emulsion was developed in D19 solution (Eastman Kodak) diluted 1:1 with distilled water at 16C. Sections were counterstained with cresyl violet acetate and coverslipped. A cDNA probe for collagen Type I (a gift from Dr. L. Sandell, University of Washington, Seattle) was used as negative control on cartilage sections. Competitive inhibition controls were performed by mixing the radiolabeled probes with unlabeled probes in ratios of 1:1, 1:2, and 1:4.
Amplification of Annexin V mRNA
Polymerase chain reaction (PCR) was performed to validate the signal obtained by in situ hybridization, as described previously by King et al. (1997). In brief, the primers were chosen to match the upstream position targeted with the in situ probe (see above) and a downstream position of 545 bases from the coded N-terminus of the polypeptide. The primers were obtained through Gene Link (Thornwood, NY). PCR was performed on RNA extracted from human chondrocytes cultured in alginate beads for 14 days (Aydelotte and Kuettner 1988). The RNA was extracted and purified and PCR was performed with a kit obtained from Life Technologies (Gaithersburg, MD) according to the manufacturer's protocols, using the above described primer pair. Superscript RNase H-reverse transcriptase (Life Technologies) was used for first-strand cDNA synthesis. As a control, a commercially available PCR kit for glyceraldehyde phosphate dehydrogenase (GAPDH) was used. Thirty cycles of amplification were done at a melting temperature of 55C. The reaction product was then purified by ethanol precipitation. The samples were separated on a 4% agarose gel and detected by ethidium bromide fluorescence.
Digestion of Chicken Chondrocyte Membranes with Matrix Metalloproteinases (MMPs)
MMP-1 (fibroblast collagenase) and MMP-3 (stromelysin) were gifts from Dr. Elizabeth Arner (DuPont-Merck, Wilmington). MMP-8 (neutrophil collagenase) (Hasty et al. 1990) was a gift from Dr. Karen Hasty, University of Tennessee, Memphis). Chicken chondrocyte membranes (50 μg) were suspended in 0.05 M Tris-HCl, pH 7.5, 0.4 M NaCl, 0.01 M CaCl2, and digested for 16 hr at 37C at an enzyme:sub-strate ratio of 1:25. The reaction was stopped by denaturing the protein in SDS gel electrophoresis buffer. For analysis, the samples were separated in 15% polyacrylamide SDS gels, transferred to nitrocellulose, and Anx V epitopes were detected by antiserum #8958 (see above). Parallel samples were stained with Coomassie Blue after electrophoretic separation.
Results
Specificity of Antiserum #9757
The antiserum was tested by Western blotting using rhanxV and the peptide-BSA conjugate as target antigens (Figure 1). In addition, an aliquot of the antiserum was blocked before the incubation with the blotted proteins by incubation with the BSA-conjugated peptide. As shown in Figure 1, the antibodies recognized exclusively the recombinant Anx V in the bacterial extract, the BSA-conjugated peptide, and in the chondrocyte membrane extract, the Anx V polypeptide at about 34 kD. Pretreatment of the antiserum with the conjugated peptide to block the specific immune reactions completely eliminated the antibody reaction against the blotted antigens.
Reaction pattern of anti-Anx V antiserum #9757 with the synthetic peptide against the first 20 amino acids of rhanxV (A,C), with rhAnx V (B,D), and with human chondrocyte plasma membranes (E, Coomassie stained; F, Western blot). The test for the reaction with the peptide was the Western blot of the peptide-BSA conjugate (67-kD band, A,C) and rhanxV (34-kD band, B,D) as target antigens. The antiserum was preincubated overnight in the presence of either BSA alone (A,B) or peptide-conjugated BSA (C,D) to block the specific epitopes. After centrifugation of the sample to remove immunoprecipitates, the volume of the aliquot was expanded with 3% skim milk solution and the blots were incubated and further processed as described in Materials and Methods. Note the complete absence of antigen recognition in the serum after the treatment with peptide-BSA (C,D), and the selective recognition of Anx V in the preparation of human chondrocyte membranes.
Anx V Immunohistology in Human Growing Cartilage
The epiphyseal and growth plate cartilages that were used for the immunohistochemistry were obtained from two fetuses and one newborn. The cartilage sections were incubated with primary antibody #9757 followed by a secondary antibody conjugated to either avidin-biotin-alkaline phosphatase (Figures 2A–2E) or rhodamine (Figures 2F–2K). To more precisely define the location of chondrocytes within the epiphyses and growth plates, a low-power photomicrograph of a section stained with hematoxylin and eosin is included (Figure 2L). Four regions (I-IV) are marked to indicate the region of the cartilage from which the higher-power photomicrographs were taken. In this section from a fetal metatarsus, the epiphysis contains cartilage canals but no secondary center. Across the young cartilage there was a gradient of staining, with the least intense at the developing articular surface (Figure 2A, Zone I) and the most intense in the hypertrophic region of the growth plate (Figures 2D and 2I, Zone IV). Cells immediately adjacent to the joint space had no detectable staining for Anx V; cells located just beneath the articular surface varied in the intensity of staining. Some cells were intensely stained, whereas others were either very lightly stained or were negative. Cells located more centrally in the epiphysis and in the proliferative zone of the growth plate were more uniform in staining. In the hypertrophic chondrocytes there was intense intracellular staining as well as strong granular staining in the ECM surrounding the hypertrophic chondrocytes. A similar pattern was seen in the hypertrophic zones in the early secondary centers of ossification of the tissues (not shown), except that the cells were in a circular array around the invading blood capillaries. These staining patterns were detectable with both fluorescence and alkaline phosphatase detection kits and were abolished by blocking the antibody with the BSA-conjugated peptide.
Fetal human epiphyseal and growth plate cartilage from a metatarsus. (A-E) Alkaline phosphatase-conjugated secondary antibody. (F-K) Rhodamine immunofluorescence of adjacent sections. (L) Hematoxylin-eosin staining of the tissue (kindly provided by Dr. Thomas Aigner, Institute of Pathology, University of Erlangen, Germany). The roman numerals I-IV define the respective areas magnified in A-K. The sections in A-K are oriented with the articular surface to the left. Note the increase in cell size from top to bottom, the sporadic staining of cells in A and F, and the intense extracellular staining in D and I. E and K are peptide-blocked controls with pictures taken from region IV.
Chondrocytes from normal adult tissue stained most strongly in the areas of the plasma membranes and the lacunar walls. This staining pattern was persistent throughout the entire age range tested (39-93 years) in the normal joint cartilage (Figures 3A and 4A-4G). The staining patterns could be suppressed by preincubating the antiserum (#9757) with the peptide-conjugated BSA (Figure 5). The overall intensity of staining declined from the middle to the deep zone, although all the tissues were treated with chondroitinase ABC and keratanases to unblock epitopes. Staining was always present in the plasma membranes and the lacunar wall, indicating maintenance of the Anx V expression on the plasma membranes. Although the cells of the deeper zones of the adult articular cartilage were not always visibly stained in the plasma membranes after alkaline phosphatase staining, the presence of some plasma membrane/lacunar wall epitopes in these cells was revealed by staining with the fluorescent (rhodamine-conjugated) second antibody (not shown). In the deepest layers, in the vicinity of or just underneath the tidemark, occasional weak staining of the ECM in the pericellular matrix was seen (Figure 4G).
Anx V Immunohistology in Normal and Osteoarthritic Cartilage
In normal adult articular cartilage (Mankin grade 0), Anx V was localized in the upper layers of the cartilage, with no detectable staining in the deeper layers (Figure 3A). In cartilage (Mankin grade 5) with a damaged superficial layer, staining was intense along the outer surface and some chondrocytes within the deep zone were now positive.
The cartilage received from OA patients showed various degrees of damage, with fibrillation and/or fissures penetrating into the deep layers of the joint cartilage and clusters of cells (Mankin grades >5). In this cartilage, the intensity of stain was enhanced (Figures 3C–3E) even in areas that appeared undamaged (Figure 3E) compared to normal cartilage (Figure 3A).
Low-power magnifications (x 25) of articular cartilages from donors (A,B) and from OA patients (C-E). The articular surface is to the right. Note the confinement of the staining to cells in the upper third of the normal cartilage with the intact superficial layer (A) and the progression of the staining into deeper layers in the sample from a donor with a damaged superficial layer (B). The sections from OA patients in C and D show examples from areas with severe damage of the joints and in E from a “normal-looking” area peripheral to the damaged tissue.
The increased intensity of staining was more apparent at higher magnification (Figure 4). In the normal cartilage (Figures 3A–3G) there was cytoplasmic staining of the chondrocytes and the lacunar wall that was less intense than that in the OA cartilage (Figures 4H–4M, 5A, and 5C). The cells in clusters in the OA cartilage were always heavily stained (Figure 4K). As a particular feature of the OA cartilage, staining in the ECM beyond the lacunar wall was observed (Figures 4L–4O and 5C), quite similar to that in the zone of hypertrophic chondrocytes in the growth plates (Figure 2D and 2I). Frequently, the staining had a granular pattern, with the granules displaying very different sizes (Figures 4N and 4O). Both the cytoplasmic and matrix staining could be blocked with the peptide-conjugated BSA (Figures 5B and 5D), thus indicating a binding of the antibodies to Anx V-specific epitopes both inside and outside of the cells.
High-power magnifications (x 150) of cells from different layers of normal (A-G) and OA (H-O) knee cartilage, starting from the layer adjacent to the joint space in A and H, to the deep zone cells adjacent to calcified zone in G and O.
Inhibition of Anx V immunostaining by blocking of the antiserum with peptide-BSA. The samples were taken from OA cartilage. (A) Intracellular staining of chondrocytes from midzone cartilage. (B) Peptide-BSA control. (C) Staining of antigen deposits in the ECM and intracellular staining. (D) Peptide-BSA control.
In situ Hybridization of Anx V Message
To compare relative differences in Anx V mRNA between growth plate, normal adult, and OA cartilage, each in situ hybridization was performed using tissues for all three from multiple donors. The tissues were processed using standardized protocols to minimize differences in mRNA preservation. For each hybridization, all three types of cartilage were handled simultaneously with a single probe preparation, the same length of exposure, and the same conditions for developing the emulsion (Figure 6). In the young cartilage, Anx V mRNA was detectable at low levels in chondrocytes throughout the epiphysis. In the growth plates, chondrocytes in the resting and proliferative zones had levels of mRNA similar to those in the epiphyses. However, in the hypertrophic zone the expression of Anx V mRNA was higher. Overall detectable mRNA for Anx V was lower in the normal articular cartilage, with expression highest in the superficial layer and some chondrocytes in the middle layer. Chondrocytes in the deep layer were only weakly positive or negative. Expression was elevated in the chondrocytes from the OA cartilage, especially in those cells located in clusters. The signals could be competed with the same but nonradioactive oligonucleotide probe, indicating specificity of the probe. The specificity of the probe was further tested by using PCR with the probe being one of the two necessary primers and the other primer designed according to a site 545 BP downstream. As shown in Figure 7, only one species of cDNA was amplified, with a molecular weight matching the expected size of about 500 BP.
In situ hybridization with Anx V anti-sense probe. (A) Fetal growth plate, from the proliferative zone (top) to the hypertrophic region (bottom). (B) Donor (50 years old) cartilage, superficial to middle zone. (C) OA patient (60 years, female) cartilage, surface to deep zone. Note the increased accumulation of silver grains over the hypertrophic chondrocytes (A) and on the cell clusters (C).
Susceptibility of Anx V to Metalloproteinase Activity
The observation of Anx V-specific epitopes in the ECM of the OA cartilage prompted us to explore the basis for this staining pattern because annexins are not described as part of the ECM, with the exception of Anx V-containing mineralizing matrix vesicles released by chondrocytes of growth plates (Genge et al. 1992). Alternatively, the Anx V epitopes in the peri-cellular matrix of the chondrocytes from OA cartilage may have been released from the cell surfaces by the proteolytic activity of MMPs. One MMP, neutrophil collagenase (MMP-8), has been shown to be upregulated in damaged cartilage (Chubinskaya et al. 1996). To test this hypothesis, purified plasma membranes from chondrocytes were incubated with MMP-1, MMP-3, and MMP-8, and the resulting material was tested in Western blots. As is shown in Figure 8, all three MMPs tested were able to digest to a variable but low degree the membrane-attached Anx V. This partial digestion may release epitopes into the pericellular matrix where they could be detected by the antiserum (see Figures 4 and 5).
RNA from two independent samples of human chondrocytes cultured in alginate beads was extracted, reverse-transcribed, amplified by PCR, and separated by agarose gel electrophoresis. The left two lanes of the ethidium bromide-stained gel contain aliquots obtained using the Anx V-specific primer pair, each corresponding to the amplification of RNA from about 500 cells. The right two lanes contain aliquots of the same samples amplified with a GAPDH-specific primer pair. The approximate length of the amplified product is about 500 BP for Anx V and about 450 base-pairs for GAPDH. The total separation range for the gel is from about 1200 to about 200 BP. Arrow points to the expected product sizes.
Digestion of chondrocyte plasma membranes with matrix metalloproteinases (MMP-1, MMP-3, MMP-8); Coomassie stain and Western blot. Chicken chondrocyte membranes (50 μg protein) were incubated for 16 hr at 37C with the MMPs (for details see Materials and Methods). The reaction mixture was dissolved in electrophoresis sample buffer and separated in a 15% polyacrylamide gel, transferred to nitrocellulose, and immunostained with anti-chicken Anx V antiserum #8958. The lane with the Anx V standard contains purified chicken cartilage Anx V. Note the decrease in staining after treatment with MMP-1 and MMP-8 and the appearance of a doublet after treatment with MMP-3. The faint band at 67 kD in the untreated control and the Anx V standard is a dimer of the 34-kD monomeric Anx V.
Discussion
The purpose of this study was to determine the pattern of expression of Anx V in human fetal and adult normal cartilage and to compare these findings with those obtained from OA cartilage. The study was motivated by two particular functional properties of Anx V in cartilage: to serve as a collagen “receptor” on chondrocytes (Mollenhauer and von der Mark 1983; Mollenhauer et al. 1984) and as a calcium channel (Huber et al. 1992), in particular in mineralizing cartilage matrix vesicles of fetal growth plate cartilage (Genge et al. 1990,1992). Both the collagen and the calcium metabolism are central aspects of cartilage physiology in development, growth, and adulthood. Collagen and calcium processing are altered in OA, contributing to degenerative processes and to pathological tissue remodeling, e.g., the formation of atypical ECM (collagen Types X and III, characteristic for fetal cartilage) and of osteophytes (Schmid and Linsenmayer 1987; Sandell et al. 1991; von der Mark et al. 1992; Young et al. 1995; Vornehm et al. 1996).
Although Anx V is a constitutive component of the cytoplasm of many different cell types, its preferential location in chondrocytes is the outer surface of the plasma membranes (Mollenhauer et al. 1984; Böhm et al. 1994). In agreement with these findings from chicken cartilage, the present immunolocalization of Anx V in the newborn and normal adult human joint cartilage indicated a preferential expression in the vicinity of the plasma membranes and the lacunar walls, with some minor staining in the cytoplasm and, in general, no staining detectable in the ECM. Fixation artifacts, such as ruptures of the plasma membranes and shrinkage of the cells, do not allow a more precise localization at present. However, even in fetal cartilage, displaying high expression levels of mRNA shown by in situ hybridization, intense staining of the cytoplasm of normal cells was not observed except in late hypertrophic cells of the fetal growth plate and in developing secondary centers of ossification.
This strong staining of the cytoplasm of hyper-trophic chondrocytes of the calcifying portion of the growth plate suggests a profound difference in the metabolic processing and function of Anx V in these cells. A gradient of increasing mRNA expression of Anx V from the resting zone through the proliferating zone into the hypertrophic zone has also been recently described in chicken growth plates (Kirsch et al. 1996). The picture suggests that Anx V of the hyper-trophic portion is upregulated to provide large quantities of the protein for tissue mineralization, as proposed by Wuthier, on the basis of the group's observations in chicken embryos (Genge et al. 1990,1992).
As mentioned earlier, some investigators found parallels in the expression pattern of matrix proteins in growth plate cartilage and in OA (von der Mark et al. 1992; Aigner et al. 1995; Vornehm et al. 1996). Similarly, substantially increased cytoplasmic staining for Anx V was typical for most of the tissue samples from OA cartilage, even in regions without macroscopically visible tissue damage. In combination with the elevated signals detected by in situ hybridization, these findings suggest a massive upregulation of the production of Anx V in the diseased joints. Because the intensified staining of the cytoplasm was observed in “normal-looking” cartilage from OA patients and even in tissue samples from organ donors with minor damage to the joint surface, i.e., “healthy” individuals, the change in the expression pattern of Anx V may occur very early in the development of the disease.
The pericellular staining of the ECM is another feature of hypertrophic growth plate cartilage (Genge et al. 1990,1992) that is found again in OA cartilage. This staining may be associated with active mineralizing vesicles released by the diseased cells. Such vesicles are components of OA cartilage according to an earlier report by Derfus et al. (1992). A potential increase in the number of vesicles released and deposited by chondrocytes in OA cartilage might possibly account for the frequently observed pathological mineralization of OA cartilage.
Alternatively, the detection of Anx V epitopes by the antibodies could be the result of extracellular proteolytic cleavage of the surface-associated Anx V and the release of proteolytic fragments into the ECM. Employing recombinant MMP-1, MMP-3, and MMP-8, enzymes that are upregulated in damaged human cartilage (Cole and Kuettner 1995; Chubinskaya et al. 1996; Cole et al. 1996), we could demonstrate that Anx V is indeed cleaved by these enzymes and may be released into the ECM as fragments where we detected the immunoreactive epitopes. If these Anx V molecules or Anx V fragments retain some functional properties, such as collagen binding, then we can speculate that these fragments interfere with the regular function of the cell surface-located Anx V as a collagen receptor.
Whether due to the production of mineralizing vesicles or to the proteolytic release, a depletion of Anx V from the chondrocyte membranes might place additional demands on the biosynthetic capacity of chondrocytes and might explain the upregulation in message levels and cytoplasmic protein observed here. Because these features are also seen in hypertrophic fetal growth plate chondrocytes, they may indicate a switch to some embryogenetic differentiation patterns in OA cartilage. In that, our findings would coincide with those made for other proteins of the growth plates: the expression of the growth plate-specific collagen Type X (Aigner et al. 1993; Walker et al. 1995), the shift in the proportions of large vs small proteoglycans (Poole et al. 1993; Cs-Szabo et al. 1995), the activation of metalloproteinases (Chubinskaya et al. 1996; Cole et al. 1996), and finally even the complex processes of pathological mineralization and of osteophyte formation in OA cartilage.
Finally, the wide variation in size of the deposited granules suggests that at least some of the deposits are the remnants of dead chondrocytes, apoptotic bodies (Kerr and Harmon 1991). Unlike tissues that can be accessed by macrophages, the apoptotic bodies in cartilage cannot be removed, because this tissue is excluded from direct contact with the blood or lymph system. Consequently, the cell fragments will accumulate in the ECM. Unfortunately, very little is known about apoptosis in OA.
It is possible that apoptosis and hypertrophic differentiation are linked in growth plate cartilage to allow the controlled transition of the fetal cartilage to bone (Roach et al. 1995). The picture unraveled in OA may underscore the hypothesis of resumed embryogenetic patterns in this disease and warrants further investigation in this direction.
Footnotes
Acknowledgements
Supported in part by the NIH/NIAMS SCOR grant 2-P50-AR-39239 and by a local chapter grant from the Arthritis Foundation. HK is the recipient of a postdoctoral fellowship from the Merck Foundation.
We would like to acknowledge the skillful technical support of Larry Madsen and by Michele Healey. The collaboration of Dr. Allan Valdellon and his staff from the Regional Organ Bank of Illinois is gratefully acknowledged.