Granulin-epithelin precursor binds directly to ADAMTS-7 and ADAMTS-12 and inhibits their degradation of cartilage oligomeric matrix protein

Authors


Abstract

Objective

To determine 1) whether a protein interaction network exists between granulin-epithelin precursor (GEP), ADAMTS-7/ADAMTS-12, and cartilage oligomeric matrix protein (COMP); 2) whether GEP interferes with the interactions between ADAMTS-7/ADAMTS-12 metalloproteinases and COMP substrate, including the cleavage of COMP; 3) whether GEP affects tumor necrosis factor α (TNFα)–mediated induction of ADAMTS-7/ADAMTS-12 expression and COMP degradation; and 4) whether GEP levels are altered during the progression of arthritis.

Methods

Yeast two-hybrid, in vitro glutathione S-transferase pull-down, and coimmunoprecipitation assays were used to 1) examine the interactions between GEP, ADAMTS-7/ADAMTS-12, and COMP, and 2) map the binding sites required for the interactions between GEP and ADAMTS-7/ADAMTS-12. Immunofluorescence cell staining was performed to visualize the subcellular localization of GEP and ADAMTS-7/ADAMTS-12. An in vitro digestion assay was employed to determine whether GEP inhibits ADAMTS-7/ADAMTS-12–mediated digestion of COMP. The role of GEP in inhibiting TNFα-induced ADAMTS-7/ADAMTS-12 expression and COMP degradation in cartilage explants was also analyzed.

Results

GEP bound directly to ADAMTS-7 and ADAMTS-12 in vitro and in chondrocytes, and the 4 C-terminal thrombospondin motifs of ADAMTS-7/ADAMTS-12 and each granulin unit of GEP mediated their interactions. Additionally, GEP colocalized with ADAMTS-7 and ADAMTS-12 on the cell surface of chondrocytes. More importantly, GEP inhibited COMP degradation by ADAMTS-7/ADAMTS-12 in a dose-dependent manner through 1) competitive inhibition through direct protein–protein interactions with ADAMTS-7/ADAMTS-12 and COMP, and 2) inhibition of TNFα-induced ADAMTS-7/ADAMTS-12 expression. Furthermore, GEP levels were significantly elevated in patients with either osteoarthritis or rheumatoid arthritis.

Conclusion

Our observations demonstrate a novel protein–protein interaction network between GEP, ADAMTS-7/ADAMTS-12, and COMP. Furthermore, GEP is a novel specific inhibitor of ADAMTS-7/ADAMTS-12–mediated COMP degradation and may play a significant role in preventing the destruction of joint cartilage in arthritis.

Arthritis is characterized by the breakdown of extracellular matrix (ECM) components and subsequent loss of articular cartilage. Destruction of articular cartilage ECM in arthritic joints is mediated by excessive proteolytic activity (1). Cartilage consists mainly of ECM with very few cells. The ECM is a network of proteins and macromolecules that provides both strength and nutrients for the cells. Cartilage oligomeric matrix protein (COMP), a prominent noncollagenous component of cartilage, accounts for ∼1% of the wet weight of articular tissue. COMP is a 524-kd pentameric, disulfide-bonded, multidomain glycoprotein composed of approximately equal subunits (∼110 kd each) (2). COMP fragments have been detected in the cartilage, synovial fluid, and serum of patients with knee injuries, osteoarthritis (OA), and rheumatoid arthritis (RA) (3). In previous studies to identify the physiologic enzymes responsible for COMP degradation, we performed a functional genetic screen, which led to the isolation of ADAMTS-7 and ADAMTS-12 as COMP-binding partners (4, 5). Subsequent studies showed that both ADAMTS-7 and ADAMTS-12 were able to digest COMP in vitro and that their levels were significantly elevated in arthritic cartilage and synovium compared with those in normal controls (3–5).

ADAMTS-7 and ADAMTS-12 belong to the ADAMTS metalloproteinase family. The ADAMTS family consists of secreted zinc metalloproteinases with a precisely ordered modular organization that includes at least 1 thrombospondin (TSP) type I repeat (6). Thus far, 19 members of this family have been cloned, and some of them have been implicated in disease (7). For instance, ADAMTS-13 mutants have a role in thrombotic thrombocytopenic purpura, which is characterized by reduced circulating platelets (8). Mutations in the ADAMTS-2 gene (procollagen I N-proteinase) cause Ehlers-Danlos syndrome Type VII C, a genetic condition characterized by defects in collagen synthesis (9). ADAMTS-2 is also mutated in bovine dermatosparaxis (9). A number of ADAMTS members have been implicated in the breakdown of cartilage in OA and RA, including ADAMTS-4 (aggrecanase 1), ADAMTS-5 (aggrecanase 2) (10), ADAMTS-7 (4), and ADAMTS-12 (5).

Granulin-epithelin precursor (GEP), also known as plasma cell–derived growth factor, progranulin, acrogranin, or GP80, was first purified as a growth factor from conditioned tissue culture media (11). GEP is a 593–amino acid secreted glycoprotein with an apparent molecular weight of 80 kd (12), which acts as an autocrine growth factor. GEP contains 7½ repeats of a cysteine-rich motif (CX5–6CX5CCX8CCX6CCXDX2HCCPX4CX5–6C) in the order P-G-F-B-A-C-D-E, where A-G are full repeats and P is a half-motif. The C-terminal region contains the conserved sequence CCXDX2HCCP, has a metal binding site, and may be involved in a regulatory capacity. Notably, GEP undergoes proteolytic processing with the liberation of small, ∼6-kd repeat units known as granulins (or epithelins), which retain biologic activity (13); the peptides are active in cell growth assays and may help to mediate inflammation (14). Increasing evidence has also implicated GEP in the regulation of development, differentiation, and disease. It has been isolated as a differentially expressed gene in mesothelial differentiation (15), sexual differentiation of the brain (16), and macrophage development (17), as well as in RA and OA (18). Mutations in GEP cause tau-negative frontotemporal dementia linked to chromosome 17 (19, 20). In addition, GEP has been shown to be a crucial mediator of wound response and tissue repair (21, 22).

Several GEP-associated proteins have been reported to affect the action of GEP in various processes (22). Among them, COMP associates with GEP and potentiates GEP-mediated chondrocyte proliferation (23). Recently, ADAMTS-7 was also found to interact with GEP and inactivate GEP-mediated chondrogenesis (24). We undertook the present study to determine 1) whether a protein interaction network exists between GEP, ADAMTS-7/ADAMTS-12, and COMP; 2) whether GEP affects the interaction between ADAMTS-7/ADAMTS-12 and COMP; 3) whether and how GEP inhibits ADAMTS-7/ADAMTS-12–mediated degradation of COMP; and 4) whether GEP levels are altered in arthritis.

MATERIALS AND METHODS

Plasmid constructs.

Yeast expression constructs.

Complementary DNAs encoding GEP (GenBank accession no. NM_002087) and each granulin were amplified using PfuTurbo DNA polymerase (Promega) with pcDNA3.1-GEP as a template. A summary of the GEP primers is available at http://mrc.med.nyu.edu/laboratories/chuanju-liu/projects.

The fusion vectors pDBleu and pPC86 (Life Technologies) link proteins to the Gal4 DNA binding domain and the VP16 transactivation domain, respectively. Complementary DNA inserts encoding GEP were cloned in-frame into the Sal I/Not I sites of pDBleu to generate pDB-GEP. This procedure was repeated for the granulin units. GEP fragments and constructs are shown below. The segments encoding series deletion mutants of ADAMTS-7/ADAMTS-12 were cloned in-frame into the Sal I/Not I sites of the pPC86 vector to generate the indicated plasmids (4, 5).

Glutathione S-transferase (GST) fusion protein.

The bacterial expression vector pGEX-3X (Life Technologies) was used to produce recombinant GST fusion proteins in Escherichia coli. Complementary DNA fragments encoding the EGF domain of COMP were inserted in-frame into the Bam HI/Eco RI sites of pGEX-3X to generate the plasmid pGEX-EGF (4, 5).

All constructs were verified by nucleic acid sequencing. Subsequent analysis was performed using Curatools (Curagen) and BLAST software (http://www.ncbi.nlm.nih.gov/BLAST).

Expression and purification of GST and His-tagged proteins.

For expression of GST fusion proteins, plasmids pGEX-3X and pGEX-EGF were transformed in DH5α E coli competent cells (Life Technologies). Fusion proteins were affinity-purified on glutathione–agarose beads, as described previously. His-GEP was purified by affinity chromatography using a HiTrap chelating column (Amersham Biosciences). Briefly, bacteria lysates supplemented with 20 mM HEPES (pH 7.5) and 0.5M NaCl were applied to a HiTrap chelating column; the column was washed with HEPES buffered saline (HBS) buffer (40 mM HEPES [pH 7.5], 1M NaCl, and 0.05% Brij-35) containing 10 mM imidazole. His-GEP was eluted with HBS buffer containing 300 mM imidazole.

Assay of protein–protein interactions using the yeast two-hybrid system.

Two independent yeast colonies were analyzed for the interaction of 2 proteins, one of which was fused to the Gal4 DNA binding domain and the other to the VP16 transactivation domain. The procedures of Hollenberg et al and Vojtek et al were followed for 1) growing and transforming the yeast strain MAV203 with the selected plasmids and 2) detecting β-galactosidase activity and growth phenotypes on selective SD-Leu–Trp–His–Ura–3AT+ plates (25, 26).

In vitro binding assay.

For examination of the binding of ADAMTS-12 to GEP in vitro, Ni2+–nitrilotriacetic acid–Sepharose beads were preincubated with either His or His-tagged GEP and then incubated with purified ADAMTS-12. Bound proteins were resolved by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and detected by Western blotting with anti–ADAMTS-12 antibodies (5, 27).

To examine whether GEP disrupts the interaction between ADAMTS-7/ADAMTS-12 and COMP, glutathione–Sepharose beads (50 μl) were preincubated with either purified GST (0.5 μg, serving as control) or the GST-fused EGF domain of COMP (GST-EGF), and then incubated with recombinant ADAMTS-7 or ADAMTS-12. Various amounts of purified GEP were then added into this system and incubated overnight. Bound proteins were resolved by 12% SDS-PAGE and detected by Western blotting with anti–ADAMTS-7 or anti–ADAMTS-12 antibodies (4, 5, 27).

Coimmunoprecipitation.

Approximately 500 μg of cell extracts prepared from isolated human chondrocytes were incubated with anti-GEP (25 μg/ml N-19 [acrogranulin]; Santa Cruz Biotechnology), anti–ADAMTS-12 (25 μg/ml) (4, 5, 27), or control rabbit IgG (25 μg/ml) antibodies for 1 hour, followed by incubation with 20 μl of protein A–agarose (Invitrogen) at 4°C overnight. After washing 5 times with immunoprecipitation buffer, bound proteins were released by boiling in 20 μl of 2× SDS loading buffer for 3 minutes. Released proteins were examined by Western blotting with anti–ADAMTS-12 or anti-GEP antibodies, using the ECL chemiluminescence system (Amersham Biosciences).

Indirect immunofluorescence cell staining.

Cultures of human C-28/I2 chondrocytes (kindly provided by Dr. Mary B. Goldring) were plated on glass coverslips coated with polylysine and grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Invitrogen) under an atmosphere of 5% CO2 at 37°C. After reaching 80% confluence, the cells were fixed with cold acetone:methanol (1:1) for 20 minutes and washed twice in 4°C phosphate buffered saline (PBS) for 5 minutes. The samples were then air-dried. Following rehydration in PBS and blocking with 20% goat serum in PBS for 30 minutes, the cells were incubated with primary antibodies (rabbit polyclonal anti–ADAMTS-7 or anti–ADAMTS-12 antibodies [4, 5, 27] diluted 1:50 and polyclonal goat anti-GEP antibodies diluted 1:50) (Santa Cruz Biotechnology) at room temperature for 1 hour. After being washed with PBS, the coverslips were incubated with secondary antibodies (fluorescein isothiocyanate–conjugated goat anti-rabbit IgG diluted 1:100 and rhodamine-conjugated chick anti-goat IgG diluted 1:100) (Santa Cruz Biotechnology) for 1 hour. The nuclei were stained with 4′,6-diamidino-2-phenylindole. The specimens were observed under a fluorescence microscope with the appropriate optical filters. Microscopic images were captured using the Image Program (Media Cybernetics) and an Olympus microscope.

In vitro digestion assay.

Purified human COMP was incubated with recombinant intact ADAMTS-7 (4) or ADAMTS-12 (5) in a digestion buffer (50 mM Tris HCl, 150 mM NaCl, 5 mM CaCl2, 2 mM ZnCl2, and 0.05% Brij-35 [pH 7.5]) (4, 5, 27). Various amounts of purified GEP were added to this system and incubated at 37°C for 12 hours. The digested products, including intact COMP and COMP fragments, were resolved by 10% nonreducing SDS-PAGE, and released proteins were examined by Western blotting with polyclonal anti-COMP antibodies, using the ECL chemiluminescence system (Amersham Biosciences).

Knockdown of GEP by specific small interfering RNA (siRNA).

The human chondrocyte cell line C-28/I2 was used as a model for analyzing the efficiency of knockdown by the siRNA and for determining the consequences of knockdown of GEP for COMP degradation by ADAMTS-7 and ADAMTS-12. The cells were maintained in DMEM supplemented with 10% FBS. The control vector pSUPER and the plasmid pSUPER-siGEP were cotransfected with the corresponding expression plasmid into C-28/I2 cells using Lipofectamine 2000 reagent (Invitrogen), and the levels of GEP were monitored using immunofluorescence cell staining. The data demonstrated that the siRNA 5′-GCCTATCCAAGAACTACAC-3′ was able to efficiently reduce the expression of human GEP. The C-28/I2 cells were then transfected with ADAMTS-7 or ADAMTS-12 expression plasmid, without or with pSUPER-siGEP, and cultured for 3 days. The media were collected and assayed by Western blotting with anti-COMP antibody.

Cartilage explants.

Human cartilage was cultured as described previously (27, 28) with approval of Institutional Review Boards (IRB# 12758). Briefly, human knee cartilage was dissected into fragments ∼4 mm in diameter by punches 1–2 mm in thickness. The cartilage was dispensed into tissue culture flasks (0.7 gm/flask) and incubated overnight in control, serum-free DMEM (Invitrogen) containing 25 mM HEPES, 2 mM glutamine, 100 μg/ml streptomycin, 100 IU/ml penicillin, 2.5 μg/ml gentamicin, and 40 units/ml nystatin. Fresh control medium (10 ml) with tumor necrosis factor α (TNFα) (5 ng/ml) (in triplicate for statistical analysis) was then added (day 0). On day 2, the supernatants were harvested for COMP degradation analysis by Western blotting, and RNA was extracted from the cartilage samples as described below. In some cultures, we added antibodies against ADAMTS-7 and/or ADAMTS-12 (5 μg/ml of anti–ADAMTS-7 and/or anti–ADAMTS-12 rabbit polyclonal antibodies) (4, 5) or 100 ng/ml of recombinant GEP. On day 7, culture supernatants were harvested and COMP degradative fragments were resolved using Western blotting.

Histology and immunohistochemistry.

The articular cartilage from OA and RA patients and normal controls was harvested and fixed in 4% PBS buffered paraformaldehyde at 4°C overnight and then decalcified in 10% EDTA for 4 weeks. After the tissue was dehydrated and embedded in paraffin, 6-μm sections were cut. Sections were stained with Safranin O–fast green to assess the presence of proteoglycan.

In the case of immunohistochemistry assay, tissue sections were deparaffinized by immersing into xylene, rehydrated by graded ethanol, then treated with 0.1% trypsin for 30 minutes at 37°C. After blocking in 20% goat serum for 60 minutes at room temperature, sections were incubated with anti–ADAMTS-7 polyclonal antibody (4) (1:200 dilution), anti–ADAMTS-12 polyclonal antibody (5) (1:200 dilution), rabbit anti-COMP polyclonal antibody (5) (1:100 dilution), or anti-GEP polyclonal antibody (23) (1:100 dilution; Santa Cruz Biotechnology) at 4°C overnight, followed by incubation with a horseradish peroxidase–conjugated secondary antibody for 60 minutes at room temperature. The signal was detected using the Vector Elite ABC Kit (Vectastain; Vector).

Statistical analysis.

A 2-sample Student's t-test was used to determine significant differences (P < 0.001) in the level of GEP between normal and arthritic cartilage.

RESULTS

Interaction between GEP and ADAMTS-12 in yeast.

A previous study revealed an association between ADAMTS-7 and GEP (24). Since ADAMTS-7 and ADAMTS-12 are members of the same family and share a similar domain organization and structure (29, 30), we first used a yeast two-hybrid assay to determine whether GEP also binds to ADAMTS-12. Briefly, a plasmid encoding GEP-Gal4DBD fusion protein and a plasmid encoding ADAMTS-12-VP16AD fusion protein were cotransformed into the yeast strain MAV203. Plasmid pairs encoding c-Jun/c-Fos and rb/lamin fusion proteins were used as positive and negative controls, respectively. The assay revealed a positive interaction between GEP and ADAMTS-12, as verified by positive β-galactosidase activity as well as a positive growth phenotype on plates lacking histidine and uracil and containing 3AT (Figure 1A).

Figure 1.

Granulin-epithelin precursor (GEP) associates with ADAMTS-12 (TS-12). A, GEP binds ADAMTS-12 in a yeast two-hybrid assay. The fusion vectors pPC86-ADAMTS-12 encoding ADAMTS-12 fused to VP16 (activation domain) in pPC86 and pDB-GEP encoding GEP fused to Gal4 (DNA-binding domain) in pDBleu were cotransfected into yeast strain MAV203 and selected on SD-Leu–Trp–His–Ura–3AT+ plates (left) and observed for β-galactosidase activity (right). The proteins c-Jun and c-Fos served as a positive control, whereas rb and lamin served as a negative control. B, GEP directly binds to ADAMTS-12. His-GEP fusion protein was immobilized on His beads, incubated with recombinant ADAMTS-12, and resolved by immunoblotting with anti–ADAMTS-12 antibodies. ADAMTS-12 and His beads were used as positive and negative controls, respectively. C and D, ADAMTS-12 and GEP coimmunoprecipitate from chondrocytes. Human chondrocyte extracts were incubated with control (Ctr) IgG or with anti-GEP or anti–ADAMTS-12 antibodies. Following precipitation, proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and detected with anti-GEP (C) or anti–ADAMTS-12 (D) antibodies. Cell lysates served as positive control. E, GEP colocalizes with ADAMTS-7 and ADAMTS-12 on the cell surface of chondrocytes. C-28/I2 human chondrocytes were stained with anti-GEP, anti–ADAMTS-7, or anti–ADAMTS-12 antibodies. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). (Original magnification × 40.)

Direct binding of GEP to ADAMTS-12 in vitro and in vivo.

To further verify the interaction between GEP and ADAMTS-12, an in vitro binding assay was performed (Figure 1B). His-GEP conjugated to beads, but not His alone, was able to pull down ADAMTS-12. Since only purified protein was used in these assays, the interaction between GEP and ADAMTS-12 was direct.

Previous studies involving coimmunoprecipitation assays have demonstrated that ADAMTS-7 binds to GEP in native human chondrocytes (24). We sought to determine whether this was also true for ADAMTS-12. Briefly, human chondrocyte extracts were incubated with anti-GEP antibody, anti–ADAMTS-12 antibody, or IgG (control). The immunoprecipitated complexes were subjected to reducing SDS-PAGE and detected using anti-GEP antibodies (Figure 1C) or anti–ADAMTS-12 antibodies (Figure 1D). GEP and ADAMTS-12 were detected in complexes immunoprecipitated by either anti–ADAMTS-12 or anti-GEP antibodies, but not control IgG, thus demonstrating that ADAMTS-12 binds to GEP in chondrocytes.

GEP colocalizes with ADAMTS-7 and ADAMTS-12 in the pericellular matrix of chondrocytes.

Given that both ADAMTS-7 and ADAMTS-12 interact with GEP, we sought to verify whether all 3 proteins are colocalized within the same subcellular compartment. Indirect immunofluorescence cell staining of primary human chondrocytes revealed that both ADAMTS-7 and ADAMTS-12 are colocalized with GEP in the pericellular matrix of isolated adult human chondrocytes (Figure 1E). These findings are consistent with a specific binding interaction between ADAMTS-7 (24), ADAMTS-12, and GEP.

Each granulin repeat unit is sufficient for binding to ADAMTS-7 and ADAMTS-12.

GEP undergoes proteolytic processing with the liberation of ∼6-kd repeating units known as granulins, which retain at least some of the biologic activity of full-length GEP (13). These peptides are active in cell growth assays and may be mediators of inflammation (14, 31). Since both ADAMTS-7 and ADAMTS-12 can bind to full-length GEP, we sought to determine whether this interaction also holds true for each repeat granulin using a filter-based β-galactosidase assay. Briefly, a plasmid encoding a partial or single granulin unit fused to Gal4DBD was cotransfected into yeast along with a plasmid encoding either ADAMTS-7-VP16AD (Figure 2B) or ADAMTS-12-VP16AD (Figure 2C) fusion protein. The associations between ADAMTS-7/ADAMTS-12 and various granulin units are summarized in Figure 2A. Our conclusion from this set of assays is that individual granulin units, such as granulin A, granulin B, granulin C, granulin D, granulin E, granulin F, and granulin G, but not partial granulin (P), are each sufficient to bind to ADAMTS-7 and ADAMTS-12.

Figure 2.

Each granulin unit of GEP is sufficient to bind to ADAMTS-7 and ADAMTS-12. A, Schematic structure of GEP constructs. The numbers refer to amino acid (aa) residues. The binding between the designated GEP granulin unit (A–G and P) and both ADAMTS-7 and ADAMTS-12 is indicated by + or –. B, Interaction between ADAMTS-7 and various GEP granulin units. Independent yeast colonies were obtained by cotransforming each with 2 plasmids, one encoding a GEP granulin unit fused to Gal4 and the other encoding ADAMTS-7 fused to VP16. The transformants were transferred onto a nitrocellulose membrane, and β-galactosidase (β-gal) activity was determined. Due to their known interaction, c-Jun and c-Fos served as a positive control, whereas unrelated proteins rb and lamin served as a negative control. Results are shown in duplicate. C, Interaction between ADAMTS-12 and various GEP granulin domains. Two plasmids encoding the GEP granulin and ADAMTS-12 fusion proteins were cotransfected into yeast. Protein–protein interactions were determined by assaying β-gal activity. See Figure 1 for other definitions. Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

Four C-terminal TSP motifs of ADAMTS-7 or ADAMTS-12 are necessary and sufficient for association with GEP.

To identify the specific motif(s) in ADAMTS-7 and ADAMTS-12 that are responsible for GEP binding, we generated multiple deletion constructs, each expressing a unique ADAMTS-7 (Figure 3A) or ADAMTS-12 (Figure 3C) deletion mutant. Each plasmid encoded a fragment of ADAMTS-7 or ADAMTS-12 fused to VP16AD. Each plasmid was then cotransfected with GEP-Gal4DBD into yeast; positive protein–protein interactions were indicated by the presence of β-galactosidase activity. For ADAMTS-7, only constructs containing the 4 C-terminal TSP repeats resulted in a positive interaction with GEP (Figure 3B). Interestingly, constructs containing only 2 of these C-terminal TSP repeats failed to interact with GEP. GEP interaction was still preserved in the construct containing no other domain except for the 4 TSP repeats. Similar findings were observed for ADAMTS-12 (Figure 3D). Taken together, these data suggest that the 4 C-terminal TSP repeats of ADAMTS-7 and ADAMTS-12 are both necessary and sufficient for their interaction with GEP.

Figure 3.

GEP selectively binds to 4 C-terminal thrombospondin (TSP) motifs of ADAMTS-7/ADAMTS-12. A, Schematic of structure of ADAMTS-7 constructs. Each label at left corresponds to a segment of ADAMTS-7. The numbers refer to amino acid residues. The binding between each ADAMTS-7 segment and GEP is indicated at right. B, Interaction between GEP and ADAMTS-7 domains. The fusion vectors pPC86-ADAMTS-7 and pDB-GEP were cotransformed into yeast, and β-galactosidase (β-gal) activity was determined (see Materials and Methods). Results are shown in duplicate. C, Schematic of structure of ADAMTS-12 constructs. Each label at left corresponds to a segment of ADAMTS-12. The numbers refer to amino acid residues. The binding between each ADAMTS-12 segment and GEP is indicated at right. D, Interaction between GEP and ADAMTS-12 domains. The fusion vectors pPC86-ADAMTS-12 and pDB-GEP were cotransformed into yeast, and β-gal activity was determined (see Materials and Methods). Results are shown in duplicate. See Figure 1 for other definitions. Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

GEP disrupts the interaction between ADAMTS-7/ADAMTS-12 and COMP.

Previous reports have demonstrated that the 4 C-terminal motifs of ADAMTS-7 and ADAMTS-12 are also responsible for their interactions with COMP (4, 5). In addition, both GEP and ADAMTS-7/ADAMTS-12 bind to COMP via the same EGF domain of COMP (4, 5, 23). These findings led us to investigate whether GEP interferes with the ability of ADAMTS-7 to bind and digest COMP. For this purpose, a GST pull-down assay was performed. Consistent with our previous report, GST-EGF, but not GST alone, efficiently pulled down ADAMTS-7 (lanes 1 and 2 in Figure 4A). However, the addition of GEP resulted in the dose-dependent inhibition of this interaction (Figure 4A). Similar inhibition by GEP was also observed for ADAMTS-12 and GST-EGF (Figure 4C). Taken together, these data indicate that GEP is able to disturb the interactions between ADAMTS-7/ADAMTS-12 and COMP in a dose-dependent manner.

Figure 4.

GEP disrupts the interaction between cartilage oligomeric matrix protein (COMP) and ADAMTS-7/ADAMTS-12. A and C, GEP prevents ADAMTS-7/ADAMTS-12 from binding COMP. The glutathione S-transferase (GST)–fused EGF domain of COMP (GST-EGF) was immobilized on glutathione–Sepharose beads and incubated with ADAMTS-7 (A) or ADAMTS-12 (C) in the presence of increasing amounts of GEP, as indicated. Bound proteins were examined by immunoblotting with anti–ADAMTS-7 (A) or anti–ADAMTS-12 (C) antibodies. B and D, GEP protects COMP against degradation by ADAMTS-7/ADAMTS-12. COMP was incubated with ADAMTS-7 (B) or ADAMTS-12 (D) along with various amounts of GEP, as indicated. COMP cleavage products were resolved using nonreducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and anti-COMP antiserum. Arrows indicate intact COMP protein; arrowheads indicate COMP degradative fragments. E, Small interfering RNAs (siRNA) repress GEP expression. Immortalized human chondrocytes, C-28/I2, transfected with either pSUPER (CTR) or pSUPER-GEP siRNA (siGEP), were stained with anti-GEP. The nuclei were stained with DAPI. (Original magnification × 40.). F, Knockdown of GEP increases COMP degradation. C-28/I2 cells were transfected with a plasmid encoding ADAMTS-7 or ADAMTS-12, and pSUPER-GEP siRNA, as indicated. After 3 days, the medium was analyzed by reducing SDS-PAGE using an anti-COMP antibody. See Figure 1 for other definitions.

GEP acts as a specific inhibitor of COMP degradation by ADAMTS-7 and ADAMTS-12.

Joint-destructive processes such as OA and RA are associated with the breakdown of cartilage ECM proteins such as COMP, a process that may be mediated by matrix metalloproteinases (MMPs) such as ADAMTS-7 and ADAMTS-12. The finding that GEP is able to disrupt the interaction between ADAMTS-7/ADAMTS-12 and COMP suggests that GEP may protect against COMP degradation by these metalloproteinases. GEP granulin units are already known to function as inhibitors of degradative enzymes, such as thrombin (22, 32). In order to determine whether GEP protects COMP from degradation by ADAMTS-7, we employed an in vitro digestion assay (Figure 4B). Purified human COMP was incubated with purified ADAMTS-7 in a specialized buffer (see Materials and Methods). In the absence of GEP, ADAMTS-7 successfully digested COMP into smaller fragments (lane 1 in Figure 4B). However, the addition of purified GEP inhibited this degradation in a dose-dependent manner, resulting in lower levels of COMP fragments and higher levels of intact COMP protein (Figure 4B). Similar results were observed for ADAMTS-12 (Figure 4D). Taken together, these data indicate that GEP is able to protect COMP from digestion by ADAMTS-7/ADAMTS-12 metalloproteinases.

To further verify the importance of GEP in inhibiting the degradation of COMP by ADAMTS-7 and ADAMTS-12, we sought to suppress GEP gene expression in human chondrocytes using the siRNA approach. We identified 19-nucleotide gene-specific sequences for GEP and then generated pSUPER-siGEP construct encoding siRNA targeting this specific gene sequence. Immunofluorescence cell staining with human C-28/I2 chondrocytes transfected with pSUPER-siGEP or pSUPER vector demonstrated that expression of the specific siRNA efficiently reduced the level of endogenous GEP protein (Figure 4E). Next, we examined whether the siRNA knockdown of GEP would affect COMP degradation. The C-28/I2 chondrocytes were transfected with ADAMTS-7 or ADAMTS-12 expression plasmid, without or with pSUPER-siGEP for 3 days. Western blotting with anti-COMP antibody (Figure 4F) showed that the intensity of the COMP fragment was increased in the media collected from the cells transfected with pSUPER-siGEP. These results further indicated that GEP was an inhibitor of COMP degradation.

GEP inhibits TNFα-induced ADAMTS-7/ADAMTS-12 expression.

TNFα is the central proinflammatory cytokine that plays a cardinal role in joint-destructive processes such as OA and especially RA. Previous studies have found that TNFα up-regulates the expression of cartilage matrix–degrading enzymes, such as ADAMTS-7 and ADAMTS-12 (27). We sought to examine whether GEP inhibits the TNFα-mediated induction of ADAMTS-7/ADAMTS-12 expression. To test this, human cartilage explants were cultured in the presence of TNFα and varying amounts of GEP for 2 days in serum-free medium. Real-time polymerase chain reaction (PCR) was then used to analyze ADAMTS-7 and ADAMTS-12 expression levels. The presence of GEP led to a significant decrease in TNFα-induced ADAMTS-7 and ADAMTS-12 expression (Figures 5A and B). In the presence of 100 ng/ml of GEP, ADAMTS-7 and ADAMTS-12 expression levels returned to near baseline.

Figure 5.

GEP inhibits tumor necrosis factor α (TNFα)–induced ADAMTS-7/ADAMTS-12 expression and cartilage oligomeric matrix protein (COMP) degradation. A and B, GEP inhibits TNFα-induced ADAMTS-7 (A) and ADAMTS-12 (B) mRNA expression. Human cartilage explants were cultured in the presence of 5 ng/ml of TNFα and various amounts of GEP for 2 days. ADAMTS-7 and ADAMTS-12 mRNA levels were determined by real-time polymerase chain reaction. Values are the mean ± SD arbitrary units, and the leftmost bar in each group indicates an mRNA level of 1. ∗ = P < 0.05 versus untreated controls. C, GEP inhibits TNFα-mediated COMP degradation. Osteoarthritic cartilage explants were cultured in the presence of 5 ng/ml of TNFα with either anti–ADAMTS-7/anti–ADAMTS-12 blocking antibodies or 100 ng/ml recombinant GEP, as indicated, for 7 days. The media were analyzed with nonreducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis and anti-COMP antibodies. Arrow indicates intact COMP protein; arrowhead indicates COMP degradative fragments. See Figure 1 for other definitions. Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

We then sought to determine whether GEP inhibits TNFα-induced COMP degradation in cartilage. To answer this, we cultured human cartilage explants for 7 days with TNFα in the presence of anti–ADAMTS-7 blocking antibodies, anti–ADAMTS-12 blocking antibodies, or recombinant GEP. COMP degradation was visualized with Western blotting using anti-COMP antibodies. In accordance with a previous report (27), TNFα treatment resulted in COMP degradation (lane 1 in Figure 5C), while anti–ADAMTS-7 and anti–ADAMTS-12 blocking antibodies inhibited COMP degradation (27), suggesting that both ADAMTS-7 and ADAMTS-12 are important for TNFα-mediated COMP degradation. Interestingly, the addition of recombinant GEP resulted in the complete inhibition of TNFα-induced COMP degradation in cartilage.

GEP levels are significantly elevated in diseased cartilage.

Although the data from our study seem to suggest that GEP protects COMP from digestion by ADAMTS-7 and ADAMTS-12, it is unclear whether this plays a role in human disease. To examine this question, we performed real-time PCR using articular cartilage extracts from human subjects. Samples of healthy adult articular cartilage were obtained from the knees of 4 deceased individuals who had had diseases unrelated to arthritis (from the Musculoskeletal Transplant Foundation). Arthritic cartilage was obtained from 12 patients undergoing elective total knee arthroplasty for end-stage arthritis. OA articular cartilage (Kellgren/Lawrence [33] grade 3 or 4) was obtained from the distal femora of 8 patients, and RA cartilage (American College of Rheumatology [ACR] stage III and IV disease) was obtained from the knees of 4 patients who fulfilled the 1987 revised criteria of the ACR (formerly, the American Rheumatism Association) for the classification of RA (34). GEP messenger RNA (mRNA) was significantly up-regulated in both OA and RA cartilage (P < 0.001 versus normal cartilage) (Figure 6A).

Figure 6.

Increased expression of GEP in arthritic cartilage. A, Expression of GEP mRNA in normal knee cartilage from 4 deceased individuals and in knee cartilage from 8 patients with osteoarthritis (OA) and 4 patients with rheumatoid arthritis (RA), assayed by real-time polymerase chain reaction. Values are the mean and SD arbitrary units, and the level of GEP mRNA expression for nonarthritic cartilage was set to 1. ∗∗∗ = P < 0.001 versus normal cartilage. B, Expression of GEP protein in normal, OA, and RA cartilage, assayed by Western blotting. Total cartilage extracts from 3 normal donors (N1–N3), 3 OA patients (O1–O3), and 3 RA patients (R1–R3) were resolved using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and probed with anti-GEP and antitubulin (internal control) antibodies. C, Histologic and immunostaining assays for ADAMTS-7, ADAMTS-12, cartilage oligomeric matrix protein (COMP), and GEP in normal, OA, and RA cartilage. The cartilage specimens were assayed using Safranin O–fast green staining (a, f, and k) and immunohistochemistry for ADAMTS-7 (b, g, and l), ADAMTS-12 (c, h, and m), COMP (d, i, and n), or GEP (e, j, and o). Insets show higher-magnification views of chondrocytes. The 4 molecules were present in the pericellular matrices of all zones in normal cartilage (be) and colocalized on the surface of articular chondrocytes (insets). Bars = 100 μm. See Figure 1 for other definitions.

To assess GEP protein expression in OA and RA cartilage, we next performed Western blot analysis. Total cartilage extracts from 3 normal donors and 3 OA and 3 RA patients were resolved using 10% SDS-PAGE and probed with anti-GEP and antitubulin (internal control) antibodies (Figure 6B). Consistent with the expression pattern of GEP mRNA, cartilage from arthritis patients (especially RA patients) contained elevated GEP levels as compared with normal cartilage.

To further verify these findings, we then performed histologic and immunohistochemical assays (Figure 6C). Safranin O–fast green staining showed that proteoglycan was greatly reduced in OA and moderately reduced in RA. Loss of articular cartilage layers and surface fibrillation could be seen in OA cartilage. We further examined the expression of ADAMTS-7, ADAMTS-12, COMP, and GEP using immunohistochemistry. These 4 molecules were present in the pericellular matrices of all zones in normal cartilage and colocalized on the surface of articular chondrocytes; however, their localizations were mainly in the territorial and interterritorial matrices of the superficial zone of the diseased cartilage. In addition, the levels of ADAMTS-7, ADAMTS-12, and GEP were moderately increased in OA cartilage and markedly increased in RA cartilage. COMP was degraded and its degradative fragments increased.

DISCUSSION

Arthritis is a disease characterized by the proteolytic degradation of cartilage ECM components with the subsequent loss of articular cartilage and bone. This process is regulated by a host of cytokines and mediators of inflammation. Our study involves one component of cartilage ECM, namely, COMP. Although the function of COMP is not completely understood, it appears to mediate chondrocyte attachment as well as stabilize the ECM of articular cartilage by interacting with matrix components, such as type II collagen, type IX collagen, aggrecan, and fibronectin (35, 36). Fragments of COMP, which are thought to result from the degradative action of proteolytic enzymes, have been detected in the diseased cartilage, synovial fluid, and serum of patients with posttraumatic knee injuries, primary OA, and RA, suggesting that COMP degradation may play a key role in human disease (3).

Given the potential importance of COMP degradation, attention has been turned to studying the enzymes that interact with COMP. Purified COMP is digested by several MMPs in vitro, including MMP-1, MMP-3, MMP-9, MMP-13, MMP-19, and MMP-20 (37). ADAMTS-4, ADAMTS-7, and ADAMTS-12 are members of the ADAMTS metalloproteinase family, and are able to digest COMP (4, 5, 38). Both ADAMTS-7 and ADAMTS-12 are elevated in arthritic cartilage (4, 5). Furthermore, the expression of both ADAMTS-7 and ADAMTS-12 is induced by the presence of inflammatory cytokines, such as TNFα (27). Concurrently, TNFα is also able to induce COMP degradation, suggesting that the TNF-induced digestion of COMP may be mediated, at least in part, by these 2 metalloproteinases (27).

Our finding that anti–ADAMTS-7 and anti–ADAMTS-12 blocking antibodies inhibit TNFα-induced COMP degradation (Figure 5C) (27) reveals that this is indeed the case. More interestingly, we observed that GEP inhibits the TNFα-induced expression of both ADAMTS-7 and ADAMTS-12, and in doing so, GEP was able to prevent COMP degradation. This supports the notion that GEP may act to protect cartilage from the destructive actions of ADAMTS-7 and ADAMTS-12. Note that data were generated based on human articular cartilage tissues at the end stage of the disease; it remains to be determined whether these results may represent and interpret the situation in the early stages of arthritis, particularly OA. A key observation from our data is that the fragments resulting from in vitro ADAMTS-7– and ADAMTS-12–mediated COMP degradation are similar in size to the fragments observed in OA as well as to the fragments observed in our cell-based assays (27). This serves as a point of correlation for our in vitro and in vivo data and suggests that our observations regarding ADAMTS-7– and ADAMTS-12–mediated COMP degradation may certainly be relevant to the pathogenesis of arthritis.

GEP is a secreted glycoprotein that acts as an autocrine growth factor and is highly expressed in chondrocytes (23). GEP undergoes proteolytic processing with the liberation of ∼6-kd repeating units known as granulins, which retain at least some of the biologic activity of full-length GEP (13). GEP is involved in various biologic and pathologic processes, including RA and OA (18), macrophage development (17), mesothelial differentiation (15), wound healing and tissue repair (21, 22), and cancer progression (13, 39). GEP is a potent stimulator of chondrocyte proliferation, a property that requires the binding of COMP (23). Our data demonstrate that the expression of GEP is significantly increased in OA cartilage. Additionally, GEP levels are even higher in RA (Figure 6). This strongly suggests that GEP may play a role in the inflammatory component of arthritis pathogenesis, and also supports the concept that arthritic chondrocytes may exhibit increased anabolic activity, including the release of growth factors (40, 41). Although expression levels of ADAMTS-7, ADAMTS-12, and GEP are all increased in arthritis (Figure 6C), it may be that elevated GEP levels are not sufficient to completely neutralize the presence of increased ADAMTS enzymes, which in turn leads to COMP degradation.

In this study we demonstrate that both ADAMTS-7 and ADAMTS-12 interact with GEP and its repeat granulin units, as evidenced by data from yeast two-hybrid, coimmunoprecipitation, and in vitro binding assays (Figures 1 and 2). This interaction occurs via the 4 C-terminal TSP motifs of ADAMTS-7 and ADAMTS-12 (Figure 3). ADAMTS-4 and ADAMTS-5 lacking this domain fail to interact with GEP in a yeast two-hybrid assay (not shown). We also found that GEP is able to disturb the binding of ADAMTS-7/ADAMTS-12 to COMP and thus inhibits the digestion of COMP by ADAMTS-7/ADAMTS-12. ADAMTS-7, ADAMTS-12, and GEP all bind to the same EGF domain of COMP (4, 5, 23), and both GEP and COMP bind to the 4 C-terminal TSP motifs of ADAMTS-7/ADAMTS-12 (4, 5) (Figure 3), suggesting that GEP competitively inhibits the binding of ADAMTS-7/ADAMTS-12 to COMP. In addition, GEP, ADAMTS-7, ADAMTS-12, and COMP colocalize in the pericellular matrices in human normal articular cartilage (Figure 6C). Collectively, these 4 molecules are intertwined in a protein–protein interaction network that could play a key role in the pathogenesis of arthritis.

Every granulin unit of GEP is able to efficiently bind to ADAMTS-7 and ADAMTS-12 (Figure 2). This is of particular significance since GEP granulins can retain biologic activity, are active in cell growth assays, and may play an active role in inflammation (13). GEP is composed of seven (A–G) and a half (P) tandem repeats of a granulin/epithelin motif containing 12 cysteines (42). In many cases, mammalian GEP is secreted as an intact mitogenic protein, with the component granulins released after proteolytic processing (43). Granulin domains A, B, C, and D were isolated from human inflammatory cells (44). The granulin/epithelin family displays a complex interplay of agonistic and antagonistic effects on mammalian cell growth in vitro. In some systems the granulin peptides stimulate cell proliferation, but in other systems they act as inhibitors of mitosis (45). At present, only granulin A/epithelin 1 and granulin B/epithelin 2 have displayed biologic activity. The actions of the other granulins (C, D, E, F, and G) are unknown.

In summary, we have established that GEP binds to both ADAMTS-7 and ADAMTS-12 in cartilage. Subsequent characterization of this novel association shows that GEP regulates ADAMTS-7 and ADAMTS-12 at the following 2 levels: 1) GEP inhibits TNFα-induced ADAMTS-7 and ADAMTS-12 expression, and 2) GEP disrupts the binding and cleavage of COMP by ADAMTS-7 and ADAMTS-12 via direct protein–protein interactions. These findings will not only lead to a better understanding of the actions of metalloproteinases, growth factors, and matrix proteins in cartilage, but may also provide us with promising therapeutic targets, including GEP or its derivatives, for preventing and treating diseases of cartilage destruction.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Liu had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Guo, Kong, Liu.

Acquisition of data. Guo, Lai, Tian, Kong.

Analysis and interpretation of data. Guo, Lai, Tian, Lin, Liu.

Acknowledgements

We thank Dr. Mary B. Goldring for providing human C-28/I2 chondrocytes.

Ancillary