Isolation, Cloning, and Sequence Analysis of the Integrin Subunit α10, a β1-associated Collagen Binding Integrin Expressed on Chondrocytes*

We have found that chondrocytes express a novel collagen type II-binding integrin, a new member of the β1-integrin family. The integrin α subunit, which has aM r of 160 kDa reduced, was isolated from bovine chondrocytes by collagen type II affinity purification. The human homologue was obtained by screening a human chondrocyte library with a bovine cDNA probe. Cloning and cDNA sequence analysis of the human integrin α subunit designated α10 show that it shares the general structure of other integrin α subunits. The predicted amino acid sequence consists of a 1167-amino acid mature protein, including a signal peptide (22 amino acids), a long extracellular domain (1098 amino acids), a transmembrane domain (25 amino acids), and a short cytoplasmic domain (22 amino acids). The extracellular part contains a 7-fold repeated sequence, an I-domain (199 amino acids) and three putative divalent cation-binding sites. The deduced amino acid sequence of α10 is 35% identical to the integrin subunit α2 and 37% identical to the integrin subunit α1. Northern blot analysis shows a single mRNA of 5.4 kilobases in chondrocytes. A peptide antibody against the predicted sequence of the cytoplasmic domain of α10 immunoprecipitated two proteins with masses of 125 and 160 kDa from chondrocyte lysates under reducing conditions. The peptide antibody specifically stained chondrocytes in tissue sections of human articular cartilage, showing that α10β1 is expressed in cartilage tissue.

shown to associate with 10 different ␣ subunits, ␣1-␣9 and ␣v and to mediate interactions with extracellular matrix proteins such as collagens, laminins, and fibronectin. The major collagen binding integrins are ␣1␤1 and ␣2␤1 (22)(23)(24)(25). The integrins ␣3␤1 and ␣9␤1 have also been reported to interact with collagen (26,27), although this interaction is not well understood (28). The extracellular N-terminal regions of the ␣ and ␤ integrin subunits are important in the binding of ligands (29,30). The N-terminal region of the ␣ subunits is composed of a 7-fold repeated sequence (12,31) containing FG and GAP consensus sequences. The repeats are predicted to fold into a ␤-propeller domain (32), with the last three or four repeats containing putative divalent cation binding sites. The ␣-integrin subunits ␣1, ␣2, ␣D, ␣E, ␣L, ␣M, and ␣X contain an ϳ200 amino acid inserted domain, the I-domain (A-domain), that shows similarity to sequences in von Willebrand factor, cartilage matrix protein, and complement factors C2 and B (33,34). The I-domain is localized between the second and third FG-GAP repeats; it contains a metal ion-dependent adhesion site (MI-DAS), and it is involved in binding of ligands (35)(36)(37)(38).
Chondrocytes, the only type of cells in cartilage, express a number of different integrins including ␣1␤1, ␣2␤1, ␣3␤1, ␣5␤1, ␣6␤1, ␣v␤3, and ␣v␤5 (39 -41). We have shown that ␣1␤1 and ␣2␤1 mediate chondrocyte interactions with collagen type II (25), which is one of the major components in cartilage. We have also shown that ␣2␤1 is a receptor for the cartilage matrix protein chondroadherin (42). In the present study we have isolated a novel collagen type II binding integrin, ␣10␤1, from bovine articular chondrocytes. Cloning and sequence analysis of the human homologue is described, and expression of ␣10 on chondrocytes is examined.
Cell Isolation and Culture-Bovine chondrocytes were isolated by digestion of articular cartilage from 4 -6-month-old calves with collagenase (CLS1; Worthington Biochemical Corp., Lakewood, NJ) as described elsewhere (45 (46). The cells were filtered and washed as described above. Human chondrocytes were cultured in Dulbecco's minimum essential medium and F-12 (1:1) supplemented with 10% fetal calf serum, 25 g/ml ascorbic acid, 50 IU of penicillin, and 50 g/ml streptomycin (Life Technologies, Inc.). To harvest cells, the culture dish was washed three times with Ca 2ϩ /Mg 2ϩfree PBS, and the cells were incubated with 0.5% trypsin and 1 mM EDTA (Life Technologies, Inc.) in Ca 2ϩ /Mg 2ϩ -free PBS for 5 min. Detached cells were suspended in medium containing 10% fetal calf serum or in PBS containing 1 mg/ml trypsin inhibitor (Sigma) and then washed in PBS.
Coupling of Affinity Columns-Collagen type II isolated from nasal cartilage by pepsin digestion (47) was coupled to CNBr-Sepharose 4B (Pharmacia Biotech, Uppsala, Sweden) according to the published procedure (25). A control column was produced by treating CNBr-Sepharose 4B in a similar manner but in the absence of protein. Bovine fibronectin (Sigma) was coupled to CNBr-Sepharose 4B according to instructions from the manufacturer. After blocking, the fibronectin-Sepharose was washed three times with PBS.
Western Blot-Human chondrocyte membrane proteins immunoprecipitated with polyclonal antibodies against ␣10 (10 g/ml affinitypurified IgG) or ␤1 (100 g/ml IgG) were separated by 8% SDS-PAGE and transferred to a nitrocellulose membrane essentially as described by Towbin et al. (48). The membrane was blocked with 3% dried milk in 10 mM Tris-HCl, pH 7.4, 0.15 M NaCl, and 0.2% Tween (blocking buffer) and then incubated with the ␤1 antibody (20 g/ml) in blocking buffer containing 1% dried milk. The ␤1 subunit was detected after incubation with a secondary antibody conjugated with horseradish by chemiluminescence using the ECL system (Amersham Pharmacia Biotech).

Isolation of Internal Peptides by In-gel Digestion and Peptide
Sequencing-Affinity-purified proteins were concentrated by precipitation using the methanol/chloroform protocol (49). After reduction/alkylation with dithiothreitol/iodoacetamide (50), the precipitated proteins were subjected to SDS-PAGE on a 4 -12% polyacrylamide gel, and protein bands were visualized by Coomassie staining. The 160-kDa protein band was excised from the gel and prepared for in-gel digestion (51). Briefly, the gel slice was washed extensively to remove SDS and the dye, and after complete drying, protease was forced into the gel by rehydration with a solution of modified trypsin (Promega, Madison, WI) in 0.2 M NH 4 HCO 3 buffer. After an overnight incubation, peptides were extracted and then isolated by narrow bore reversed phase liquid chromatography on a RPC C2/C18 stainless steel 2.1/10 column operated in a SMART System (Amersham). Several peptides were analyzed by Edman degradation in a Perkin-Elmer Applied Biosystem Model 476 sequencer operated according to the manufacturer's instructions.
mRNA Purification and cDNA Synthesis-mRNA from bovine or human chondrocytes were isolated using a QuickPrep ® Micro mRNA purification kit (Pharmacia). cDNA was synthesized at 42°C for 1 h using the Superscript TM II RNase H Ϫ Reverse Transcriptase cDNA Synthesis system (Life Technologies, Inc.) random DNA hexamers and oligo(dT) (Promega, Madison, WI).
PCR Amplification-PCR reactions were performed in 50-l reaction volumes and contained 1 ϫ Taq polymerase buffer (Life Technologies, Inc.), 1.5 mM MgCl 2 , 1 M of each primer, 0.025 units/l Taq polymerase, 1 l of DNA template (bovine chondrocyte cDNA), and 0.1 mM each of dATP, dGTP, dCTP, and dTTP (Boehringer Mannheim). PCR samples were heated to 94°C for 5 min in a thermocycler and then subjected to 35 cycles consisting of 30 s at 94°C (denaturation), 30 s at 48 or 52°C (annealing) and 3 min at 72°C (extension). The PCR products were re-amplified using 1 l of each product for an additional 35 cycles. Amplified DNA was analyzed by 1% agarose gel electrophoresis. Small DNA fragments were analyzed using 4% MethaPhore TM -agarose (FMC BioProducts, Rockland, ME).
The degenerate primers GAY AAY ACI GCI CAR AC (DNTAQT, forward) and TIA TIS WRT GRT GIG GYT (EPHHSI, reverse) were used in PCR to amplify the nucleotide sequence corresponding to the bovine peptide 1 (Table I). A 900 base pair PCR fragment was then amplified from bovine cDNA using an internal specific primer TCA GCC TAC AT- A Triton X-100 lysate of bovine chondrocytes (2.5 ϫ 10 9 cells) was applied to a fibronectin-Sepharose precolumn followed by a collagen type II-Sepharose column. The lanes show EDTAeluted proteins from the fibronectin-Sepharose (A), flow-through from the collagen type II-Sepharose column (B), and EDTA-eluted proteins from the collagen type II-Sepharose (C). The eluted proteins were precipitated by methanol/chloroform, separated by SDS-PAGE (4 -12%) under reducing conditions, and stained with Coomassie Blue. The 160-kDa protein with affinity for collagen type II is indicated with an arrow.
T CAG TAT (SAYIQY, forward) corresponding to the cloned nucleotide sequence of peptide 1 together with the degenerate primer ICK RTC-CCA RTG ICC IGG (PGHWDR, reverse) corresponding to the bovine peptide 2 (Table I). Mixed bases were used in positions that were 2-fold degenerate, and inosines were used in positions that were 3-or 4-fold degenerate.
To obtain cDNA that encoded the 5Ј end of ␣10, we designed the primer AAC TCG TCT TCC AGT GCC ATT CGT GGG (reverse; residues 1254 -1280 in ␣10 cDNA) and used it for rapid amplification of the cDNA 5Ј end (RACE) as described in the Marathon TM cDNA amplification kit (CLONTECH INC., Palo Alto, CA).
Cloning and Sequencing of cDNA-PCR fragments were isolated and  Library Screening-The cloned 900-base pair PCR fragment corresponding to bovine ␣10-integrin was digoxigenin-labeled according to the DIG DNA labeling kit (Boehringer Mannheim) and used as a probe for screening of a human articular chondrocyte ZapII cDNA library (provided by Michael Bayliss, The Royal Veterinary Basic Sciences, London, UK) (52). Positive clones containing the pBluescript SK ϩ plasmid with the cDNA insert were rescued from the ZAP vector by in vivo excision as described in the ZAP-cDNA ® synthesis kit (Stratagene). Selected plasmids were purified and sequenced as described earlier using T3, T7, and internal specific primers.
Northern Blot Analysis-Bovine chondrocyte mRNA was purified using a QuickPrep ® Micro mRNA purification kit (Amersham), separated on a 1% agarose formaldehyde gel, transferred to nylon membranes, and immobilized by UV cross-linking. cDNA probes were 32 Plabeled with Random Primed DNA labeling kit (Boehringer Mannheim). Filters were prehybridized for 2-4 h at 42°C in 5ϫ SSE (20 ϫ SSC, 3M NaCl, 0.3 M trisodium citrate⅐2H 2 O, pH adjusted to 7.0 with 1 M HCl), 5ϫ Denhardt's solution, 0.1% SDS, 50 g/ml salmon sperm DNA, and 50% formamide and then hybridized overnight at 42°C with the same solution containing the specific probe (0.5-1 ϫ 10 6 cpm/ml). Specifically bound cDNA probes were analyzed using the phosphoimaging system (Fuji). Filters were stripped by washing in 0.1% SDS for 1 h at 80°C before reprobing. The ␣10-integrin cDNA probe was isolated from the race1-containing plasmid using the restriction enzymes BamHI (Life Technologies, Inc.) and NcoI (Boehringer Mannheim). The rat ␤1-integrin cDNA probe was a kind gift from Staffan Johansson, Uppsala, Sweden (25).
Tissue Staining-Human cartilage from the trochlear groove, obtained during surgery, was provided by Anders Lindahl, Sahlgrenska University Hospital, Gothenburg, Sweden. Frozen sections of cartilage tissue were fixed in acetone at Ϫ18°C for 5 min, washed in PBS, and then treated with 2 mg/ml hyaluronidase (Sigma) in PBS, pH 5.0, for 15 min at 37°C. After washing with PBS, sections were blocked for 15 min at room temperature in 0.1% H 2 O 2 in PBS to remove endogenous peroxidase activity. Sections were then washed in PBS, blocked with 0.5% casein and 0.05% thimerosal in PBS (blocking buffer) for 15 min at room temperature, and then incubated overnight at 4°C with the affinity-purified antibodies against the integrin subunits ␣9 or ␣10 (5 g/ml in blocking buffer). For control, the ␣10 antibody was preincubated with the ␣10 peptide (0.1 mg/ml) for 30 min at 4°C. After washing in PBS, sections were incubated with biotinylated goat anti-rabbit secondary antibody (Vector Laboratories Inc; diluted 1:200 in blocking buffer) at room temperature for 60 min. Washed sections were then incubated with VECTASTAIN ABC reagent (Vector Laboratories, Inc. Burlingame, CA) for 1 h at room temperature and washed, and the color was developed using 1 mg/ml diaminobenzidine, 0.02% H 2 O 2 and 0.1 M Tris-HCl, pH 7.2. Sections were rinsed in water for 5 min followed by 75, 95, and 99.5% ethanol for 5 min each and then three times in xylene for 3 min at room temperature. Samples were mounted in Pertex (Histolab Products AB, Gothenburg, Sweden) and examined by light microscopy.

Identification and Isolation of the Chondrocyte ␣10 Integrin
Subunit-Affinity purification of 125 I-labeled membrane proteins from human chondrocytes on collagen type II-Sepharose followed by immunoprecipitation showed that these cells, in addition to ␣1␤1 and ␣2␤1, express an unidentified ␤1-related ␣ subunit (Fig. 1). This integrin subunit had an apparent molecular mass of approximately 160 kDa under reducing condition and was slightly larger than the ␣2 integrin subunit. This finding is in agreement with a previous study from our group showing that bovine chondrocytes also express an unidentified collagen binding ␤1-associated ␣ subunit of similar molecular mass (25). To isolate this protein, we affinity-purified collagen type II-binding proteins from bovine chondrocytes. The chondrocyte lysate was first applied to a fibronectin-Sepharose precolumn, and the flow-through was then applied to a collagen type II-Sepharose column. As shown in Fig. 2, a number of proteins were eluted from the affinity columns. A protein with molecular mass of approximately 160 kDa was specifically eluted with EDTA from the collagen column but not from the fibronectin column. The molecular mass of this protein corresponded with the molecular mass of the unidentified ␤1-related integrin subunit (Fig. 1). The 160-kDa protein band was excised from the SDS-PAGE gel and digested with trypsin, and several of the isolated peptides were analyzed. Table I shows the amino acid sequence of six individual peptides.
Cloning and Sequencing of the Human Integrin ␣-Subunit Homologue-The nucleotide sequence corresponding to peptide 1 (Table I) was obtained by PCR amplification, cloning, and sequencing of bovine cDNA. From this nucleotide sequence an exact primer was designed and applied in PCR amplification with degenerate primers corresponding to peptides 2-6 ( Table  I). Primers corresponding to peptides 1 and 2 amplified a 900base pair PCR fragment from bovine cDNA that was cloned, sequenced, and used for screening of a human articular chondrocyte ZapII cDNA library to obtain the human integrin ␣-subunit homologue. Two overlapping clones, hc1 and hc2 (Fig. 3), were isolated, subcloned, and sequenced. These clones

Comparison of the cytoplasmic tails of I-domain-containing integrin ␣ subunits
The underlined sequence in ␣10 represents the peptide that was used for antibody production. contained 2 ⁄3 of the nucleotide sequence, including the 3Ј end of the cDNA. A third clone (Race1; Fig. 3), which contained the 5Јend of the ␣10 cDNA, was obtained using the RACE technique. From these three overlapping clones of ␣10 cDNA, 3884 nucleotides were sequenced (Fig. 4). The sequence contains a 3504-nucleotide open reading frame that is predicted to encode a 1167 amino acid mature protein. The predicted sequence included a signal peptide (22 amino acids), a long extracellular domain (1098 amino acids), a transmembrane domain (25 amino acids), and a short cytoplasmic domain (22 amino acids). Sequence analysis of the 160-kDa protein sequence showed that it was a member of the integrin ␣-subunit family, and the subunit was named ␣10.
Comparison of ␣10 Integrin Subunit with Other ␣ Subunits-Analysis of ␣10 with known ␣ subunits showed that its struc-ture follows the conserved pattern of integrin ␣ subunits (Fig.  5). The extracellular domain contains a 7-fold repeated sequence including FG and GAP consensus sequences, three putative divalent cation binding sites (DXD/NXD/NXXXD), and an I domain of 199 amino acids. The protein contains 10 potential N-linked glycosylation sites (NX(T/S)). The calculated molecular mass is 153 kDa if carbohydrate chains with an average molecular weight of 2.5 kDa are assumed to attach to all 10 putative glycosylation sites. This is in agreement with the molecular mass of ␣10 as judged by SDS-PAGE where the molecular mass was estimated to approximately 160 kDa.
Expression of the ␣10 Integrin Subunit on Chondrocytes-Northern blot analysis of mRNA from bovine chondrocytes showed that a human ␣10 cDNA probe hybridized with a single mRNA of approximately 5.4 kilobases (Fig. 7). As a comparison, a cDNA probe corresponding to the integrin subunit ␤1 was used. This cDNA probe hybridized a mRNA band of approximately 3.5 kilobases on the same filter. Translation of the ␣10 nucleotide sequence revealed an open reading frame of 3504 nucleotides (Fig. 4), which indicates that around 2000 nucleotides in the mRNA is not translated.
To study expression of ␣10 at the protein level, 125 I-labeled membrane proteins from human chondrocytes were immunoprecipitated with polyclonal antibodies against the integrin subunits ␤1, ␣1, ␣2, ␣3, and ␣10 (Fig. 8). A polyclonal peptide antibody raised against the cytoplasmic domain of ␣10 precipitated two protein bands with molecular masses of approximately 160 and 125 kDa under reducing conditions. The ␣10associated ␤-chain migrated as the ␤1 integrin subunit both under reducing and nonreducing conditions (Figs. 8, a and b). To verify that the ␣10-associated ␤-chain indeed is ␤1, chondrocyte lysates were immunoprecipitated with antibodies against ␣10 or ␤1 followed by Western blot using antibodies against the ␤1 subunit (Fig. 8c). These results clearly demonstrated that ␣10 is a member of the ␤1-integrin family.
Expression of ␣10 in cartilage was examined by immunostaining of human articular cartilage from the trochlear groove with the polyclonal ␣10 antibody. As shown in Fig. 9, this antibody specifically stained the chondrocytes in the cartilage tissue sections. The staining was completely abolished when the antibody was preincubated with the ␣10 peptide. A control antibody against the ␣9 integrin subunit did not stain chondrocytes in the tissue sections (Fig. 9). DISCUSSION The present study demonstrated that human chondrocytes express a novel, collagen type II-binding integrin in the ␤1 family. We have, in an earlier study, presented some evidence for that bovine chondrocytes and human chondrosarcoma cells also express this integrin (25). Because bovine chondrocytes are readily available in large amounts, we used these cells in the isolation of the integrin subunit ␣10. As shown in Fig. 2, several proteins were eluted from the columns in the affinity purification experiments. It was difficult to interpret the pro-tein pattern in the eluate because typical integrin bands were not clearly distinguished on the SDS-PAGE gel. This may be explained by partial protein degradation, although a mixture of protease inhibitors were included in the lysate buffer. Based upon the finding that the ␤1 antibody immunoprecipitated an unknown collagen-binding integrin ␣ subunit with a moleculare mass of 160 kDa (Fig. 1), a protein with similar molecular mass that was specifically eluted with EDTA from the collagen type II column was excised from the gel and used for peptide sequencing. This 160-kDa protein was not eluted from the fibronectin-Sepharose, indicating that fibronectin is not a ligand for ␣10␤1. However, this will be investigated in cell adhesion experiments using cells transfected with the ␣10 subunit.
The immunoprecipitation experiments showed that ␣2 and ␣10 integrin subunit have similar molecular masses under reducing conditions (Fig. 1). To avoid contamination of ␣2, the 160-kDa protein was excised from the SDS-PAGE gel as a very narrow band. This was apparently successful since human homologues to all six bovine peptides ( Table I) that were isolated from the 160-kDa protein were found in the predicted amino acid sequence of human ␣10 subunit (Fig. 4).
The deduced amino acid sequence of ␣10 was found to share the general structure of the integrin ␣ subunits described in previously published reports (6 -21). The large extracellular N-terminal part of ␣10 contains a 7-fold repeated sequence that was recently predicted to fold into a ␤-propeller domain (32). The integrin subunit ␣10 contains three putative divalent cation binding sites (DXD/NXD/NXXXD) (53), a single spanning transmembrane domain, and a short cytoplasmic domain. In contrast to most ␣-integrin subunits, the cytoplasmic domain of ␣10 does not contain the conserved sequence KXGFF(R/K)R. The predicted amino acid sequence in ␣10 is KLGFFAH. Several reports indicate that the integrin cytoplasmic domains are crucial in signal transduction (54) and that membrane-proximal regions of both ␣and ␤-integrin cytoplasmic domains are involved in modulating conformation and affinity state of integrins (55)(56)(57). It is suggested that the GFFKR motif in ␣-chains are important for association of integrin subunits and for transport of the integrin to the plasma membrane (58). The KXG-FFKR domain has been shown to interact with the intracellular protein calreticulin (59), and interestingly, calreticulin-null embryonic stem cells are deficient in integrin-mediated cell adhesion (60). It is, in this context, tempting to speculate that the sequence KLGFFAH in ␣10 may have a key function in regulating the affinity between ␣10␤1 and collagen.
Integrin ␣ subunits are known to share an overall identity of 20 -40% (61). Sequence analysis showed that the ␣10 subunit is most closely related to the I domain-containing ␣ subunits ( Fig. 6) with the highest identity to ␣1 (37%) and ␣2 (35%). The integrins ␣1␤1 and ␣2␤1 are known receptors for both collagens and laminins (24,62,63), and we have also recently demonstrated that ␣2␤1 interacts with the cartilage matrix protein chondroadherin (42). Since ␣10␤1 was isolated on a collagen type II-Sepharose, we know that collagen type II is a ligand for ␣10␤1. We have also shown by affinity purification experiments that ␣10␤1 interacts with collagen type I (data not shown), but it remains to be seen whether laminin or chondroadherin are also ligands for this integrin.
The peptide antibody that we raised against the cytoplasmic domain of ␣10 immunoprecipitated two proteins from human chondrocytes with molecular masses of approximately 125 and 160 kDa. The molecular mass of 160 kDa correlates with the unidentified ␤1-associated ␣ subunit that was affinity-purified on collagen type II-Sepharose. The 125-kDa protein was in Western blot recognized by an antibody to the ␤1 subunit. This, FIG. 9. Immunostaining of human articular cartilage. An antibody raised against the cytoplasmic domain of ␣10 (see Table II) stained the chondrocytes in tissue sections of human articular cartilage (A). The staining was depleted when the antibody was preincubated with the ␣10 peptide (B). A control antibody recognizing the ␣9 integrin subunit did not bind to the chondrocytes (C). together with previous findings that ␣1␤1 and ␣2␤1 are present on isolated chondrocytes demonstrate that chondrocytes express at least three collagen-binding integrins in the ␤1 family (25). Further studies will answer the question whether these integrins have similar or different functions in cartilage.
Immunohistochemistry using the ␣10 antibody showed staining of the chondrocytes in tissue sections of human articular cartilage. The antibody staining was clearly specific because preincubation of the antibody with the ␣10 peptide completely abolished the staining. An antibody against the integrin subunit ␣9 did not stain the chondrocytes (6). This integrin is a receptor for tenascin C (64) and is not known to be present in cartilage.
Taken together, we have isolated and characterized a novel collagen type II-binding integrin designated ␣10␤1. The ␣10 subunit was isolated from bovine chondrocytes, and the human homologue was cloned and sequenced. Antibodies against the ␣10-integrin subunit stained chondrocytes in tissue sections of articular cartilage, indicating that ␣10␤1 indeed is expressed in cartilage. Further investigations including ligand interactions, tissue distribution, signal transduction, and knockout mutation will demonstrate the function of the integrin ␣10␤1.