Identity of urinary trypsin inhibitor-binding protein with link protein

Urinary trypsin inhibitor (UTI), a Kunitz-type protease inhibitor, directly binds to some types of cells via cell-associated UTI-binding proteins (UTI-BPs). Here we report that the 40-kDa protein (UTI-BP 40 ) was purified from cultured human chondrosarcoma cell line HCS-2/8 by UTI affinity chromatography. The purified UTI-BP 40 was digested with trypsin, and the amino acid sequence of the peptide fragments was determined. The sequence of six tryptic fragments of UTI-BP 40 was identical with subsequences present in human link protein (LP). Authentic bovine LP and UTI-BP 40 displayed identical electrophoretic and chromatographic behavior. The UTI-binding properties of UTI-BP 40 and LP were indistinguishable. The direct binding and competition studies strongly demonstrated that the NH 2 -terminal fragment is the UTI binding part of the LP molecule, that the COOH-terminal UTI fragment (HI-8) failed to bind the NH 2 -terminal subdomain of the LP molecule, and that the LP and UTI-BP 40 exhibited significant hyaluronic acid binding. These results provide a demonstration that the UTI-BP 40 is identical to LP and that the NH 2 -terminal domain part of UTI is involved in interaction with the NH 2 -terminal fragment within the LP, which is bound to hyaluronic acid into the extracellular matrix.

Calibration of the gel filtration column was with high and low molecular weight standard (Cosmo Bio, Tokyo). Eluant was monitored at 280 nm. The eluted fractions were assayed for dot blot analysis using anti-LP, LP pep-N , LP pep-C , HABR-1, and HABR-2 (see below). The fractions containing Mr ~100 kDa UTI-BP (UTI-BP 100 ) was separated from the mixture of UTI-BP 45 and UTI-BP 40 . The mixture of UTI-BP 45 and UTI-BP 40 were further separated on reverse phase HPLC. This sample was applied to a 4.6 ∞ 250-mm Vydac C-18 column (Kanto Kagaku) (2). The C18 columns were packed for high performance and equilibrated with 5 % acetonitrile (AN)/0.1 % trifluoroacetic acid (TFA) before loading. The material was pumped directly onto the column. The column was eluted at 1.0 ml/min with a gradient from 5 % AN to 50 % AN over 90 min. Eluent was monitored at 214 nm and 280 nm. The fractions eluting between 26 % and 29 % AN (UTI-BP 45 ) and between 32 % and 34 % AN (UTI-BP 40 ) were pooled, dialyzed, and concentrated. The amount of protein in the soluble fraction was quantified in a Bradford assay (Bio-Rad) using BSA as a standard (32).

Purification of bovine link protein and hyaluronic acid-binding region (HA-BR) within aggrecan
The isolation of hyaluronic acid-binding protein (HA-BP) derived from bovine nasal cartilage has been described in detail elsewhere (22,33). HA-BP was purified by affinity chromatography on hyaluronic acid covalently coupled to Sepharose. A purified preparation of HA-BP was supplied by Chugai Pharmaceutical Co. Ltd., and Seikagaku Kogyo, Co. Ltd., Tokyo. Five mg of HABP was concentrated using a Centricon 10 ultrafiltration tube by centrifuging at 200 ∞g at 4 C and then further purified by gel filtration chromatography on a column of Sepharose CL-6B (2.5 ∞ 175 cm) equilibrated in 4 M guanidine HCl, 50 mM Tris-HCl, pH 7.4, as described by Tang et al (34). The crude hyaluronic acid-binding region (HA-BR) and LP peaks were fractionated by HPLC gel filtration using a SW3000 column (Kanto Kagaku, Tokyo). Aliquots of each fraction were tested for their immunoreactivity by a specific dot blot assay and a HA-BR peak (>100-kDa polydisperse band by Western blot) and a LP peak (~40-kDa band) were obtained. The LP purified from HA-BP does not contain HA-BR within aggrecan, which was confirmed by Western blot analysis with specific monoclonal antibodies raised against HA-BR within aggrecan (mAbs HABR-1 and HABR-2).

Preparations of polyclonal antibodies raised against UTI and its derivative as well as against LP and its synthetic peptides
Highly purified preparation of UTI was supplied by Mochida Pharmaceutical Co., Tokyo. Chondroitinase ABC (Sigma) was used for enzymatic deglycosylation. Briefly, 1 mg of the purified UTI was incubated with 1.0 µg of chondroitinase ABC for 24 h at 37 °C. The COOH-terminal fragment of UTI (HI-8; Mr 8 kDa) prepared by trypsin digestion was a gift of Dr. Dan Sugino (Nissin Food Products; Shiga, Japan). Polyclonal antibodies against UTI and HI-8 were prepared by intradermal injection of rabbits with 0.1 mg of purified proteins emulsified in Freund's adjuvant. The antiserum was specific for UTI and had a 50 % maximal binding at a dilution of 1/10,000 in an ELISA. The antisera to UTI and HI-8 were reactive with the 240-kDa IαI, the 120-kDa Preα−inhibitor (PαI), the 40-kDa UTI and the Mr 8-kDa HI-8 in ELISA or Western blotting assays. Affinity-purified IgG was prepared by mixing 3 ml of antiserum with 1 ml of UTI (or HI-8)-Sepharose overnight at 4 °C. Following washing, the IgG was eluted with 0.1 M glycine-HCl, pH 2.5. The pH of the eluted fractions was immediately raised, and the IgG was stored at -20 °C.
Antibodies against LP (pAb LP) purified from bovine cartilage were prepared in a similar manner. In addition, to generate anti-LP peptide antibodies, two synthetic oligopeptide sequences, 112 VFLKGGSDSDAS 123 (NH 2 -terminal fragment of LP) and 231 TVPGVRNYGFWDKDKS 246 (COOH-terminal fragment of LP), corresponding to the NH 2 -terminal domain and the COOH-terminal domain of human LP molecule, respectively, were selected. We searched for possible antigenic amino acid sequences on the LP molecule according to their predicted secondary structures and hydrophobicity. Each peptide was chosen based on their theoretical antigenic index and for specificity to the molecules. Antisera against LP synthetic oligopeptides were obtained from rabbits immunized four times with 0.2 mg peptide conjugated with keyhole limpet hemocyanin together with Freund's adjuvant. Titration of antisera was performed by an ELISA, with peptides used for immunization as antigen. When the antibody titer reached a plateau, blood was totally collected, and the serum was separated. Polyclonal antibodies against NH 2 -terminal fragment of LP (anti-LP pep-N ) and against COOH-terminal fragment of LP (anti-LP pep-C ) were prepared in a similar manner using the elute from protein A-Sepharose (Hitrap; Pharmacia, Uppsala, Sweden).

Production of monoclonal antibodies raised against UTI and hyaluronic acid-binding region within aggrecan
Male Balb/c mice were immunized at 14-day intervals by intraperitoneal injection of 20 µg affinity purified UTI. Three days after the last booster, spleen cells (1 ∞ 10 8 ) were fused with the mouse myeloma cell line NS-1 and seeded according to standard procedures (35). The antibodies were designated 2A6, 5C12, 4D1, and 8H11. 8H11 showed the strongest reactivity for UTI and reacted with the NH 2 -terminal domain of UTI. mAb 8H11 was isolated from ascites fluid by chromatography on a protein A-Sepharose column and used for ELISA.
Monoclonal antibodies raised against HA-BR within aggrecan were prepared in a similar manner. Two antibodies were selected and designated HABR-1 and HABR-2.
These mAbs were found to react with HA-BR of aggrecan, but not with LP. A list (antibody specificities and characterization) of the various mAbs and pAbs used in this study is shown in Table 1. A purified preparation of each antibody was biotinylated according to the method of Guesdon et al. (36), using N-hydroxysucciminidyl biotinamidocaproate (Sigma) using the manufacturer's suggested procedures.

SDS-PAGE and Western blot
The cell extracts, purified proteins, or tryptic fragments were dissolved in a sample buffer. The sample (20 µg protein/lane for cell extracts and 0.1~0.5 µg protein/lane for purified proteins) was processed for electrophoresis, using a SDS polyacrylamide gel under nonreducing conditions. The resulting gel was electrophoretically blotted onto PVDF membrane, which was blocked in TBS containing 2 % BSA, and then immunoblotted. The blot was subsequently processed for biotin-avidin-peroxidase method (40). Bands were visualized with the ECL detection system (Amersham, Tokyo). The membranes were then placed between two transparencies and exposed to Kodak film. In all experiments, some strips were incubated with non-immune rabbit (or mouse) IgG as a negative control.

Statistical analysis
The data presented are the mean of triplicate determinations in one representative experiment unless stated otherwise. Data are presented as mean ± standard deviation (SD). All statistical analysis was performed using StatView for Macintosh. The Mann-Whitney U test was used for the comparisons between different groups. P less than 0.05 was considered significant.

Determination of antibody specificity
Characterization of mAb 8H11: Clone 8H11 was produced by somatic cell fusion. The interaction of mAb 8H11 with UTI was evaluated by immunoblotting with UTI, chondroitinase ABC-treated UTI (deglycosylated UTI), HI-8, and with UTI reduced The ELISA data also confirmed that the 8H11 determinant was sequested in the NH 2terminal structure of UTI (data not shown).
Characterization of mAbs HABR-1 and HABR-2 and pAbs raised against LP and LP synthetic peptides as well as polyclonal antibodies raised against UTI-BP: By procedures described previously (2,4), the UTI-BP was purified from human HCS-2/8 cell lysates by UTI-coupled Sepharose 4B. As shown in Figure (Figure 3). This shows that the 40 kDa band does not contain more than LP. The interaction of anti-LP antibodies (pAb LP) with UTI-BPs was also evaluated by immunoblotting with UTI-BP 100 , UTI-BP 45 and UTI-BP 40 . pAb LP reacted with UTI-BP 40 and purified LP molecule, as well as with both LP-N and LP-C.
However, pAb LP failed to react with UTI-BP 100 or UTI-BP 45 . PAb LP pep-N , in which the epitope presents on the NH 2 -terminal domain of LP, reacted with UTI-BP 40 , LP, and LP-N, but not with LP-C, UTI-BP 100 , or UTI-BP 45 , whereas PAb LP pep-C , in which the epitope resides in the COOH-terminal domain of LP, reacted with UTI-BP 40 , LP, and LP-C, but not with LP-N, UTI-BP 100 , or UTI-BP 45 . Western blot analysis thus demonstrated that pAbs raised against LP synthetic peptides exclusively recognized both their respective domains of LP and UTI-BP 40 . It is unlikely that the UTI-BP 100 and UTI-BP 45 have antigenically cross-reactivity with LP. These results suggest that the UTI-binding sites purified from HCS-2/8 cells may contain other binding proteins or UTI receptors rather than LP.
MAbs HABR-1 and HABR-2 (data not shown here) react with UTI-BP 100 but not with UTI-BP 45 or UTI-BP 40 , suggesting that the UTI-BP 100 is comprised of HA-BR of aggrecan fragment.

Amino acid sequence of UTI-BP 40 tryptic fragments
The purified UTI-BP 40 was digested with trypsin, and the resultant peptides were purified by immunoblotting or reverse phase HPLC and identified by NH 2 -terminal sequencing (Figure 4). Aliquots of each blotting or each pool were analyzed by gasphase sequencing. A comparison with data in GenBank (Accession NM 001884) showed that the six tryptic peptides were identical with subsequences found in human LP. In every case, the UTI-BP 40 fragments corresponded to those expected from cleavage of LP at tryptic sites. The molecular mass of the six tryptic fragments was equivalent to 32.2 % of the mass of LP (354 aa). 40  showed no significant affinity for UTI, even if the concentration of biotinylated UTI was increased to 1 µmol/L. Although UTI binding to different ligands cannot be quantitatively compared by plate binding, our results indicate that the subdomain for UTI binding is the NH 2 -terminal domain within LP molecule and that UTI shows no significant affinity for HA-BR within aggrecan (UTI-BP 100 ). Our results support the hypothesis that HA-BR itself has an ability to bind UTI via the LP molecule, since HA-BR is known to directly and specifically interact with LP. To assure that the applied proteins stuck to the microtiter plate wells, we performed an immunodetection assay by using respective antibodies. The significant signals of absorbance at A 450 were obtained from these ligands tested, compared with those from the BSA control (data not shown). BSA failed to inhibit biotinylated UTI binding to immobilized UTI-BP 40 . In a parallel experiment, potent inhibition by UTI, UTI-BP 40 , LP, and LP-N was also observed in LP-coated microtiter plate wells (data not shown).

Reverse phase HPLC and SDS-polyacrylamide gel electrophoresis of UTI-BP
The identity of UTI-BP 40 with LP is directly provided by the following competition assays. First, studies on the binding of anti-UTI-BP antibodies to immobilized UTI-BP 40 were performed in the presence of LP (data not shown). We carried out an immunodetection assay by using biotinylated anti-rabbit IgG and avidinperoxidase. This experiment showed that LP (1 µM) was able to give ~90 % inhibition of anti-UTI-BP antibodies binding to UTI-BP 40 bound to a plate. Second, UTI-BP 40 (1 µM) almost completely blocked anti-LP antibodies binding to LP bound to a plate (data not shown). Thus, the antibodies were each blocked to greater than 90 % by the antigens indicated.

Biotinylated UTI binding to LP or UTI-BP 40 anchored via hyaluronic acid
We studied the interaction of LP or UTI-BP 40  showed no significant affinity for CS or HS. These results indicated that there is no significant difference in UTI binding activity between LP and UTI-BP 40 bound to immobilized HA. In addition, we added LP or UTI-BP 40 to the HA-PE-coated plate wells together with CS and HS to see if either could compete with LP or UTI-BP 40 binding. However, neither CS nor HS could compete with LP or UTI-BP 40 binding to HA. These results strongly indicated the specific interaction of both LP and UTI-BP 40 with HA, but not CS or HS. The direct binding studies strongly demonstrated UTI does not directly and specifically interact with HA, CS, or HS (data not shown). These results support the hypothesis that UTI has an ability to bind HA via the LP molecule or UTI-BP 40 . We confirmed again that the applied proteins (LP or UTI-BP 40 ) stuck to HAcoated microtiter plate wells (data not shown).

Discussion
Extracellular proteolysis is required in inflammation and in tumor processes where cell migration and invasion occurs (41)(42)(43)(44)(45). A growing body of evidence demonstrated that UTI effectively inhibits tumor cell invasion and metastasis. Tumor cell-associated plasmin, but not urokinase activity, was efficiently inhibited by UTI (46)(47)(48)(49). UTI interacts with a variety of cell types including neoplastic and non-neoplastic cells. The presence of UTI at the cell surface has been explained by the demonstration of UTI binding sites on the cell membranes. These UTI-binding protein (UTI-BP) and UTI receptor have been recently detected, but only a few have been isolated or extensively characterized (2)(3)(4)(5)(6)50). We initially reported the presence of proteins of the UTI-BP family on human choriocarcinoma SMT-cc1 cells and uterine fibroblasts (2,4). While the UTI-BP 40 was described as an UTI-binding site, it was able to bind hyaluronic acid as well and has been localized abundantly in cartilage and ovary in mice and rats (in submission). Therefore, in the present study we tried to isolate and characterize proteins of the UTI-BP family from human chondrosarcoma HCS-2/8 cells in a large scale, since by guest on March 23, 2020 http://www.jbc.org/ Downloaded from these cells expressed cartilage proteoglycans associated with hyaluronic acid to form proteoglycan aggregates (51,52).
Using the UTI-immunoaffinity beads, several proteins of the UTI-BP family were purified from HCS-2/8 cell extracts. The UTI-BP 40 was the major band consistently and specifically bound to UTI. The UTI-BP 100 and UTI-BP 45  However, the conclusion from the set of immunoprecipitation experiments is that anti-LP antibodies can immunoprecipitated approximately 60 % of UTI binding activity from the cell extracts (data not shown). Incomplete reduction after immunoprecipitation could be due to the presence of a heterogenous population of UTI binding proteins, since some fractions (UTI-BP 100 and UTI-BP 45 ) of which may be unable to react with the LP-related antibodies. It is likely that UTI can bind to components rather than members of the LP molecules. The UTI-BP 100 was identified immunologically as a HA-BR within aggrecan molecule. Therefore, UTI-BP 100 and aggrecan G1 domain share similar epitopes or that they are closely related if not identical molecules. It is unlikely that the HA-BR within aggrecan is another candidate for UTI-BP, since UTI does not directly bind to HA-BR. We have considered that the HA-BR may bind UTI via the LP molecule. The identity of UTI-BP 100 as aggrecan G1 domain should also be established by amino acid sequencing, tryptic maps, and specific binding and competition experiments.
In addition, the minor UTI-BP 45 was also specifically isolated. It is unlikely that UTI-BP 40 may represent a degradation product of UTI-BP 45 , since anti-LP antibodies did not cross-react with UTI-BP 45 . This may be a novel protein or may be a subunit of UTI-BP complex, each having the ability to bind UTI. Since the yield of UTI-BP 45 was small, studies on the specific binding of UTI to UTI-BP 45  The molecular weight of LP produced by HCS-2/8 cells is almost the same as that of UTI-BP 40 (Takigawa et al., 1999;unpublished data). LP is synthesized by the chondrosarcoma cells themselves and stabilizes the binding between proteoglycan subunits and hyaluronic acid (34,53,54). Since LP is found in the extracellular matrix, it is thought to be involved in the organization of an hyaluronic acid-rich matrix. We reported for the first time that LP directly binds UTI, which corresponds to a light chain of inter-α inhibitor. These data strongly demonstrate that a locally produced and expressed UTI-binding sites accumulate free UTI and/or inter-α inhibitor into the extracellular matrix of the chondrosarcoma cells.
It is possible that the UTI serves a number of different functions through the LP molecule. UTI could interact with LP anchored in hyaluronic acid-rich matrix on the surface of tumor cells. This may result in the effective inhibition or regulation of tumor cell-associated protease activity. Furthermore, our previous studies demonstrated the specific internalization of UTI by tumor cells (2,50,55). The process of UTI-BP-or UTI receptor-mediated endocytosis has been the subject of extensive study. Proteins of the UTI-BP family may be involved in the active endocytosis of UTI. Further research will reveal additional characteristics for the very interesting proteins of the UTI-BP family.
In summary, the present study characterizes the proteins of the UTI-BP family biochemically, immunologically and immunohistochemically, and identified UTI-BP 40 , which is identical with LP. Our results strongly support that UTI binding site is located     The upper row, aa sequence of LP; the lower row, aa sequence of tryptic fragments of UTI-BP 40 . X, unidentified residue.