Shedding of Kidney Injury Molecule-1, a Putative Adhesion Protein Involved in Renal Regeneration*

KIM-1 (kidney injury molecule-1) is a type I transmembrane glycoprotein expressed on dedifferentiated renal proximal tubule epithelial cells undergoing regeneration after toxic or ischemic injury. The extracellular domain of KIM-1 is composed of an immunoglobulin-like domain topping a long mucin-like domain, a structure that points to a possible role in cell adhesion by homology to several known adhesion proteins. Two splice variants (a and b), of the human KIM-1 having identical extracellular domains, differ in their cytoplasmic domains and tissue distributions. In this study, we report that the KIM-1b transcript is expressed predominantly in adult human kidney. We describe the generation of 10 monoclonal antibodies against the extracellular domain of human KIM-1, the mapping of their binding sites, and their use in identifying various forms of the protein. We show that human KIM-1b is expressed in adult kidney cell lines, and we demonstrate that a soluble form of KIM-1 is shed constitutively into the culture medium of the cell lines expressing endogenous or recombinant KIM-1b by membrane-proximal cleavage. A monoclonal antibody that binds at or close to the proteolytic site can partially block the shedding of KIM-1. Release of soluble KIM-1 is enhanced by activating the cells with phorbol 12-myristate 13-acetate and can be inhibited with two metalloproteinase inhibitors, BB-94 (Batimastat) and GM6001 (Ilomastat), suggesting that the cleavage is mediated by a metalloproteinase. We propose that the shedding of KIM-1 in the kidney undergoing regeneration constitutes an active mechanism allowing dedifferentiated regenerating cells to scatter on denuded patches of the basement membrane and reconstitute a continuous epithelial layer.

Renal epithelial cell injury is a feature of many acute and chronic renal diseases. Morphologic characteristics of injury to the proximal tubule epithelial cell include loss of proximal tubular brush border, loss of cellular polarity, dedifferentiation, and apoptosis. With advanced injury, viable and necrotic tubular epithelial cells detach from the basement membrane and contribute to intraluminal obstruction. Surviving dedifferentiated cells spread over the denuded basement membrane, undergo mitogenesis, and ultimately redifferentiate and reestablish normal epithelial polarity resulting in a normal functional epithelium (1,2). Although these processes are well described at the histopathological level, very little is known about the molecular factors that regulate these events. One of the genes identified from postischemic rat kidney by representational difference analysis (3) was designated kidney injury molecule-1 (KIM-1) 1 (4). This gene encodes a type I cell membrane glycoprotein containing, in its extracellular portion, a six-cysteine immunoglobulin-like domain and a Thr/Ser-Prorich domain characteristic of mucin-like O-glycosylated proteins. Immunoglobulin-like domains have been widely implicated in mediating protein-protein interaction (5) in particular at the cell surface where they are responsible for cell-cell and cell-extracellular matrix interactions. The mucin domain, which extends the Ig-like domain away from the cell surface like a stalk, could play a dual role of configuration and protection (6) as well as being involved in cell adhesion. Mucosal addressin cell adhesion molecule-1, for example, binds to Lselectin (CD62 ligand) via its mucin domain and interacts with ␣ 4 ␤ 7 integrin via its Ig domain (7). The cytoplasmic domain of KIM-1 is relatively short and possesses a potential phosphorylation site, indicating that KIM-1 may be a signaling molecule.
Rat KIM-1 mRNA and protein levels are dramatically upregulated in the postischemic kidney. In situ hybridization and immunohistochemistry revealed that KIM-1 is expressed in dedifferentiated proximal tubular epithelial cells in damaged regions, especially in the S3 segment of the proximal tubule in the outer stripe of the outer medulla, a region that is highly susceptible to injury as a result of ischemia or toxins (4). Because KIM-1 colocalizes with markers of proliferation, we propose that KIM-1 plays a role in the regeneration process.
The first identified homolog of KIM-1 was an African green monkey protein cloned as the receptor for hepatitis A virus (HAVcr-1) (8). Two human homologs of KIM-1 were subsequently cloned from kidney and liver. One was cloned as the homolog of KIM-1 (4), and the other one was cloned as the homolog of HAVcr-1 (9). We refer to them as KIM-1a and KIM-1b, respectively. These two human homologs are identical except for the C-terminal portion of their cytoplasmic domain (Fig. 1A). Analysis of their genomic structure and cDNA prod-ucts indicates that they are splice variants and reveals that KIM-1a is the major form in liver, and KIM-1b is predominant in kidney. 2 The murine KIM-1 homolog was identified recently and shown to belong to a family of three highly related genes expressed on activated T cells and located on chromosome 11 in a locus genetically linked to airway hyperreactivity. This region of the mouse genome is homologous to a region of the human chromosome 5q which includes the human KIM-1 gene and has also been linked to atopy and asthma (10,11). So, in addition to participating in restoring a functional renal epithelium, KIM-1 might play an important role in the immune response.
A number of membrane proteins have soluble forms that are released into the extracellular space. Although these soluble forms can result from alternative splicing, they more often derive from proteolysis of the membrane form. The cleavage occurs close to the transmembrane domain releasing physiologically active protein. These proteins are diverse in structure and function and include cytokines, growth factors, receptors, cell adhesion molecules, and enzymes (12)(13)(14). In many reported cases of shedding, inhibition by hydroxamic acid-based zinc metalloproteinase inhibitors has indicated that the cleaving enzymes or "sheddases" could be matrix metalloproteinases (MMPs) (12) or members of the ADAM (a desintegrin and metalloproteinase) family (15,16). One of the most notorious and best characterized sheddases to date is the tumor necrosis factor-␣-converting enzyme (TACE), an ADAM family member (ADAM17). Genetic deletion of TACE in mouse confers embryonic lethality, revealing the importance of protein shedding in vivo. The lack of functional TACE abolishes the shedding of not only tumor necrosis factor-␣ but also L-selectin, transforming growth factor-␣, tumor necrosis factor receptor, the amyloid precursor protein, fractalkine, and others, indicating that TACE is a sheddase with multiple substrates (17)(18)(19).
In this paper, we report the creation of a number of mAbs specific to the human KIM-1 extracellular domain and characterize the region to which they bind. Using these antibodies, we show that cells expressing endogenous or recombinant human KIM-1b constitutively shed the KIM-1 ectodomain into the extracellular milieu and that the release of soluble KIM-1 can be blocked with two different MMP inhibitors or with an antibody that binds to the proteolytic site. We also show that shedding of KIM-1 is increased in the presence of PMA and that both basal and PMA-activated shedding of KIM-1 are inhibited by BB-94. The possible relevance of KIM-1 shedding in kidney disease, including renal carcinoma, is discussed.

EXPERIMENTAL PROCEDURES
Materials-PMA (Sigma) was diluted into fresh medium from a stock solution at 1 mg/ml made in dimethyl sulfoxide. GM6001 (Chemicon International) and BB-94 (British Biotech Pharmaceuticals, Ltd.) also know as Ilomastat and Batimastat, respectively, were diluted from 2.5 mM stock solutions in dimethyl sulfoxide into fresh medium to concentrations ranging from 16 to 0.5 M. The 18-mer peptides used for the mapping of the mAb binding sites were purchased from Research Genetics and resuspended to a concentration of 5 mg/ml (stock solutions).
Recombinant KIM-1 Proteins-To obtain soluble forms of human KIM-1 protein, a construct (KIM1-Ig) was made in which the extracellular domain of human KIM-1 (residues 1-290) was attached to the Fc portion of human IgG1 (hinge, CH2 ϩ CH3 domains) and cloned into pEAG347, a BIOGEN mammalian expression plasmid containing a tandem promoter (SV40 early/adenovirus major late) for constitutive expression and the dihydrofolate reductase gene for methotrexate selection of stably expressing cell lines. Transfected Chinese hamster ovary cell lines expressing the fusion proteins were selected, adapted in suspension, and grown in fermentors. Another soluble form of KIM-1 (hKIM1(mucin⌬)-Ig) was made in which only the Ig-like domain (residues 1-135) was fused to the Fc portion of human IgG1. This fusion protein was transiently expressed in COS-7 cells. The KIM1-Ig fusion proteins were purified from conditioned media by chromatography on protein A-Sepharose (Amersham Biosciences). A human KIM-1b fulllength cDNA was obtained by reverse transcription-PCR using mRNA from the human carcinoma cell line 769-P and primers designed from the published DNA sequence (9). The sequence of the cDNA obtained was identical to that of the HAVcr-1 cDNA obtained from human kidney and liver (9).
Murine Monoclonal Antibodies against Human KIM-1-Four mice were immunized with human KIM1-Ig. The increase of the antibody titer against human KIM-1 was monitored by ELISA. The mouse that showed the best response was selected and boosted with the protein prior to the fusion. Dissociated spleen cells were fused with FL653 myeloma cells and plated into 96-well tissue culture plates in selection medium. Wells positive for growth were screened by ELISA for expression of antibody against human KIM-1. Ten hybridoma clones were selected and subcloned. Four of them (ABE3, ACA12, AKG7, and ARD5) were grown in the peritoneal cavity of mice. The ascitic fluid was collected, and each antibody was purified by chromatography using protein A-Sepharose. Biotinylated AKG7 and ARD5 were prepared by direct binding of amino-reactive sulfo-NHS-LC-biotin (Pierce). The antibodies were isotyped using the IsoStrip mouse mAb isotyping kit.
Rabbit Polyclonal Antibodies against KIM-1-A synthetic peptide (CKEVQAEDNIYIENSLYATD; Research Genetics) corresponding to the last 19 amino acid residues at the C terminus of the human KIM-1b (Fig. 1A), plus a cysteine residue for the conjugation, was used to raise polyclonal antibodies. The peptide was conjugated to maleimide-activated keyhole limpet hemocyanin (Pierce), and the conjugate was used to immunize a rabbit. Antisera were collected after several immunizations.
Cell Lines and Cell Culture-Cell lines listed hereafter were obtained from the American Tissue Culture Collection and grown as prescribed: 769-P (human renal cell adenocarcinoma; CRL-1933); HK-2 (human kidney proximal tubular cells immortalized by transduction with HPV-16; CRL-2190); 293 (embryonic kidney cells transformed with adenovirus; CRL-1573); HepG2 (hepatocellular carcinoma; HB-8065). 769-P cells were grown in RPMI medium supplemented with 10% fetal bovine serum, 10 mM HEPES, and 1 mM sodium pyruvate (complete medium). The serum was omitted in some short term shedding experiments to allow the concentration of the conditioned medium and subsequent analysis by Western blotting. For transient expression of human recombinant KIM-1b, COS-7 cells were transfected by electroporation with 10 g of plasmid DNA/10 6 cells. Transfected cells were plated and grown in Dulbecco's modified Eagle's medium with 4 mM glutamine and 10% fetal bovine serum. After 4 h of incubation to allow the cells to attach, the medium was replaced. Cell confluence was then ϳ20%.
Cytotoxicity Assay-Live cells were stained with MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) that was reduced in the mitochondria to a colored formazan product. The staining was done by adding 100 l of 5 mg/ml MTT/ml of cell culture and incubating for 3 h at 37°C. The formazan crystals were dissolved by adding 1 ml of 10% SDS, 10 mM HCl and incubating at room temperature with gentle agitation for 24 h. Absorbance at 570 nm was quantitated.
FACS Analyses-To assess the binding of anti-human KIM-1 mAbs, ARD5 and ABE3, to cell surface KIM-1, growing 769-P cells were lifted from a flask with EDTA, washed with FACS buffer (0.5% bovine serum albumin, 0.1% NaN 3 in PBS), distributed in tubes (0.5 ϫ 10 6 /tube), and pelleted (5 min at 200 ϫ g). The cells were then successively resuspended in 100 l of FACS buffer, incubated 30 min on ice with anti-KIM-1 mAb (1-125 g/ml in FACS buffer), washed twice with 2 ml of FACS buffer, resuspended in 100 l of Alexa 488-conjugated goat antimurine IgG, Fab'2 (Molecular Probes) diluted 200-fold and incubated again for 30 min on ice. After one final wash, the cells were resuspended in 50 l of FACS buffer and 50 l of 4% paraformaldehyde in PBS (for fixation). Cell sorting was done on 10,000 cells with a FACSCalibur machine (Becton Dickinson) and the data was processed with the Cellquest analysis software (Becton Dickinson).
To compare the relative densities of KIM-1 at the surface of 769-P cells growing in wells and treated with various concentrations of ABE3 or a control antibody, cells were lifted with EDTA and pelleted by centrifugation (5 min at 200 ϫ g). Cells were then successfully resuspended in 100 l of FACS buffer containing 5 g/ml of biotinylated ARD5, incubated 30 min on ice, washed with 2 ml of FACS buffer, resuspended in 100 l of phycoerythrin-streptavidin conjugate diluted 200-fold in FACS buffer, and incubated again 30 min on ice. After one final wash, the cells were resuspended in 50 l of FACS buffer. Cell sorting was done as described previously.
Immunofluorescence Microscopy-Cells were grown in the wells of eight-chamber slides until they reached ϳ30% confluence. They were washed with medium, then incubated at 37°C for 30 min with ABE3 at 50 g/ml or ARD5 at 2 g/ml in serum-containing medium. After one wash with medium followed by two washes with PBS to remove unbound antibodies, the cells were fixed with 4% paraformaldehyde (10 min at room temperature). The cells were washed again twice with PBS and once with a solution of 2% goat serum in PBS, incubated 10 min at room temperature in that same last solution and then 30 min with Alexa 488-conjugated goat anti-murine IgG, FabЈ2 (diluted 200-fold). After three washes with the 2% goat serum solution, the chambers were peeled off, and the slides were mounted with Vectashield (Vector Laboratories).
Cell Surface Biotinylation and Affinity Purification-769-P cells were grown in two flasks until 30% confluence. The cells were washed twice with PBS and incubated for 30 min at room temperature with 1 mg/ml sulfo-NHS-LC-biotin. After three washes with 10 mM glycine in PBS, medium was added to one flask for an additional day of culture while the cells of the other flask were lysed in lysis buffer (25 mM HEPES, 0.15 M NaCl, 5 mM EDTA, 1% Triton X-100) with a mixture of protease inhibitors (Complete; Roche Molecular Biochemicals). The conditioned medium from the reincubated flask was collected, and the cells were washed and lysed as described. The media and the lysates were clarified by centrifugation, and biotinylated proteins were extracted with streptavidin-agarose beads (Novagen). After overnight reaction at 4°C, the beads were washed twice with the lysis buffer, and the bound proteins were eluted with SDS-PAGE loading buffer. Aliquots of the lysates and the streptavidin-extracted proteins were analyzed by Western blotting as described below.
Metabolic Labeling and Immunoprecipitation-769-P cells were seeded in 6-cm tissue culture dishes and cultured in complete medium (RPMI supplemented with 10% of fetal bovine serum) until cells reached 80 -90% confluence. Cells were washed twice with pulse labeling medium (cysteine-deficient RPMI supplemented with 25 mM HEPES, pH 7.4, and 10% dialyzed fetal bovine serum) and incubated in this medium for 1 h at 37°C. The cells were then labeled for 2 h with 200 Ci/ml [ 35 S]cysteine in pulse labeling medium and chased for the indicated times in complete medium (2 ml/dish). The conditioned media were collected at the end of the chase, and the cells were lysed in 1 ml of cold PBS containing 1% Nonidet P-40, 5 mM EDTA, 400 mM phenylmethylsulfonyl fluoride, and 2 M leupeptin. One-ml aliquots of the The class, subclass, and light chain type of each antibody were determined using an isotyping kit. The specificity of the antibodies for the Ig-like domain or the mucin domain was determined by ELISA and Western blotting using Ig fusion proteins in which either the entire ectodomain (hKIM1-Ig) or the Ig-like domain alone (hKIM1(mucin⌬)-Ig) is fused to the Fc fragment of human IgG1. For Western blot analysis, the proteins were heat denatured in SDS-PAGE loading buffer with or without ␤-mercaptoethanol, a disulfide bond-reducing agent used to allow complete unfolding of the protein.

Clone and isotype
Bind Ig-like domain Bind mucin domain Detection of hKIM1-Ig by ELISA Detection of nonreduced hKIM1-Ig by Western blotting conditioned media were incubated for 1 h on ice with 1.5 g of AKG7 mAb then mixed with protein G-agarose to collect the immune complexes. After a preadsorption step on normal rabbit IgG-agarose beads, the cell lysates were incubated with anti-KIM-1b C-terminal peptide rabbit polyclonal antibodies for 1 h on ice, followed by separation with protein A-agarose. The 35 S-labeled proteins were fractionated by SDS-PAGE and visualized by autoradiography. Cell Extract and Conditioned Media Western Blot Analysis-Conditioned media samples were clarified by centrifugation (5 min at 16,000 ϫ g). Cells were rinsed twice with ice-cold PBS, and extracts were prepared either by direct in situ lysis of the cells grown in wells or by lysis of the cells that had been scraped from the surface of a flask with a rubber policeman and pelleted in a tube. In situ lysis was done by adding 250 l of lysis buffer to each well of a six-well plate. Cells from flasks were harvested with a rubber policeman in ice-cold PBS containing 5 mM EDTA and spun down at 2,000 ϫ g. The cell pellet was weighed and resuspended in lysis buffer (20 l of lysis solution/mg of cell pellet). After 5 min of lysis on ice, the extracts were clarified by centrifugation (5 min at 16,000 ϫ g). Samples (extracts and conditioned media) were mixed with reducing loading buffer and heated for 5 min at 95°C. Reduced denatured proteins were then separated by SDS-PAGE on 4 -20% polyacrylamide gels. The proteins were transferred onto a nitrocellulose sheet. The blot was blocked with a solution of 5% non-fat dry milk (Carnation) in PBST (PBS ϩ 0.05% Tween 20) and probed in FIG. 2. Mapping of ABE3, AKG7, and ACA12 binding sites. A, the binding of ACA12, AKG7, and ABE3 to eight 18-mer overlapping synthetic peptides (see sequences in B) was assessed by ELISA using 96-well plates coated with the peptides and horseradish peroxidase-conjugated goat anti-murine IgG. The absorbance values at 450 nm is a measure of the amount of bound antibodies. B, alignment of a section of human KIM-1 mucin domain, its predicted secondary structure (NNPREDICT program), and the 18-mer overlapping synthetic peptides used to map the binding epitopes of ACA12, AKG7, and ABE3. The regions of the binding sites deduced from the experiment described in A are highlighted in gray. Putative N-glycosylation sites are underlined. E, H, and dashes mark the positions of the amino acid residues in predicted ␤ strands, helices, or turn elements, respectively. the same solution with the murine mAb AKG7 or ABE3 (at 1 g/ml) or with the rabbit polyclonal antiserum raised against the C-terminal peptide of KIM-1b (diluted 1,000-fold), followed by either goat antimurine or goat anti-rabbit antibodies conjugated to horseradish peroxidase. Washes between the steps were done with PBST. Reactive bands were revealed by chemiluminescence.
ELISAs-To characterize the binding of anti-KIM-1 mAbs, wells were coated by incubation overnight at 4°C with 100 l of antigen (peptide or hKIM1-Ig fusion protein) in 50 mM sodium carbonate, pH 9.6. Potential remaining adsorption sites were blocked by incubation for 1 h at room temperature with 400 l of PBS containing 1% bovine serum albumin (blocking buffer). The wells were first incubated with the mAbs at various concentrations (from 1 g/ml to 1 ng/ml) then with horseradish peroxidase-conjugated goat polyclonal antibodies against mouse IgG (Jackson ImmunoResearch Laboratories) both diluted in blocking buffer. Plates were washed four times with PBST after each reaction step. The color reaction was carried out with tetramethylbenzidine and stopped with H 2 SO 4 .
To measure the concentration of soluble KIM-1 in conditioned medium, wells coated with ARD5 were incubated successively with the medium samples or KIM1-Ig standard diluted in blocking buffer, biotinylated AKG7, and then horseradish peroxidase-conjugated streptavidin. Plates were washed four times with PBST after each reaction step. The color reaction was carried out with tetramethylbenzidine and stopped with H 2 SO 4 .
Shedding Inhibition Studies-The experiment of KIM-1 shedding inhibition by ABE3 was done with 769-P cells grown in six-well plates to about 80% confluence. Cells were incubated in serum-containing medium supplemented with various concentrations of ABE3 or mouse IgG (Sigma). After 6 h of incubation, the conditioned media were col-lected, clarified by centrifugation (5 min at 16,000 ϫ g), and soluble KIM-1 was quantified by ELISA. The cells were lifted with EDTA and analyzed by FACS to compare the relative densities of cell surface KIM-1. Shedding inhibition with BB-94 and GM6001 was evaluated with 769-P cells grown in six-well plates to about 50% confluence. After two washes with fresh medium, the culture was continued in medium containing various concentrations of BB-94 or GM6001 (0.25-16 M) for the indicated times. Serum-containing medium was used for the 18-h inhibition study, and soluble KIM-1 released into the conditioned media was quantified by ELISA. Serum-free medium was used for the 1-h inhibition study. The conditioned media samples, supplemented with protease inhibitors and concentrated 12-fold (Microcon 30; Millipore), and the cell lysates were analyzed by Western blotting.
N-terminal Sequencing of Shed KIM-1-A 50% ammonium sulfate precipitation was used to concentrate and partially purify the recombinant human KIM-1 shed into the culture medium. The protein pellet was resuspended in PBS containing a mixture of protease inhibitors and dialyzed against PBS. The sample clarified by centrifugation was mixed overnight at 4°C with ARD5 and goat anti-mouse IgG beads (Sigma). The beads were drained and washed successively with PBS, PBS containing 0.1% Tween 80, and 0.1 M MES, pH 5.1. KIM-1 was then eluted with 0.1 M glycine, pH 2.5. The proteins were precipitated with trichloroacetic acid. After one wash with acetone, the protein pellet was dried and resuspended in SDS-PAGE loading buffer. The sample was run by SDS-PAGE and transferred to a polyvinylidene difluoride membrane; proteins were stained with Coomassie Blue, and the band corresponding to shed KIM-1 was excised. The membrane strip was loaded onto a gas phase sequencer (Applied Biosystems 470). The phenylthiohydantoin (PTH) amino acids were analyzed on-line with a PTH Analyzer (Applied Biosystems 120A) equipped with a PTH C18 column.

FIG. 3. Expression and shedding of native KIM-1.
A-C, 293, HK2, 769-P, and HepG2 cells were grown in the medium prescribed by ATCC for each cell line. After 3 days in culture, the conditioned media were removed, the cells were collected with a rubber policeman into ice-cold PBS and lysed with detergent. Lysates and conditioned media were analyzed by Western blotting using either ABE3, AKG7, or rabbit polyclonal antibodies against the human KIM-1b C terminus. D, 769-P cells in culture were incubated with sulfo-NHS-LC-biotin in PBS for 30 min at room temperature for biotinylation of cell surface proteins. Cells were either lysed directly after the biotinylation or put back in culture for an additional day after which the conditioned medium was collected, and the cells were lysed. Biotinylated proteins were purified on streptavidin-agarose (Novagen). The lysate (cell extract) and the proteins bound to streptavidin-agarose from the first extract (Ext.t 0 ), the second extract (Ext.t 1 ), and the conditioned media (CM.t 1 ) were analyzed by Western blotting. E, 769-P cells were pulse labeled with [ 35 S]cysteine in cysteine-free medium for 2 h and chased for the indicated times in complete medium at 37°C. Conditioned media were collected, and soluble KIM-1 was immunoprecipitated with AKG7. Cells were lysed, and the KIM-1 proteins were immunoprecipitated with anti-KIM-1b C-terminal peptide rabbit polyclonal antibodies. The immune complexes were collected with protein G-or A-agarose beads, separated by SDS-PAGE, and visualized by autoradiography.

RESULTS
Monoclonal Antibodies against KIM-1-To obtain highly specific antibodies against the human KIM-1 protein, murine mAbs were generated against the extracellular domain and characterized by ELISA and Western blot ( Table I). Four of the mAbs (ABE3, AKG7, ACA12, and ATE11) failed to bind the KIM-1 fusion protein lacking the mucin domain (hKIM1(mucin⌬)-Ig), indicating that their respective binding epitopes should be at least partly in the mucin domain. In addition, these four antibodies are the only ones that react with the reduced denatured hKIM1-Ig on Western blot, suggesting that their respective epitopes might correspond to a single stretch of amino acid residues in the protein primary structure. We tested by ELISA the binding of AKG7, ACA12, and ABE3 to eight overlapping synthetic 18-mer peptides starting at KIM-1 residue 210 in the mucin domain and ending at residue 290, the last residue of the extracellular domain (Fig. 2, A and B). Both AKG7 and ACA12 were found to bind to peptides 45 and 46, and although they show some difference in their respective affinities, one can conclude that both must bind an epitope within the 9-residue sequence stretch common to peptide 45 and 46 (Fig. 2B). This peptide stretch LQGAIRREP does not carry any putative sites for either N-or O-glycosylation, which suggests that ACA12 and AKG7 will thus be able to recognize the unglycosylated and the various glycosylated forms of the protein. ABE3, on the other hand, was found to bind to peptide 49 but not to overlapping peptides 48 and 50. Knowing that the longest linear peptide an antibody could possibly bind is approximately seven amino acid residue long, this suggests that the ABE3 binding site is within the sequence stretch centered on peptide 49 (Fig. 2B). This sequence stretch, GL-WNNNQTQLFL, in KIM-1 could potentially be N-glycosylated on the last asparagine residue. However because ABE3 binds to various glycosylated forms of KIM-1 as well as to the synthetic peptide, we can assume that the binding relies at least primarily on the interaction with the peptide component of the protein.
Expression and Shedding of Endogenous and Recombinant Human KIM-1-Four human cell lines, three from kidney (293, HK2, 769-P) and one from liver (HepG2), were analyzed for expression of KIM-1 (Fig. 3). Western blot analysis of detergent cell extracts using ABE3 mAb (Fig. 3A) revealed that KIM-1 is detectable in 769-P (third lane) as well as in HK2 (second lane) but not in 293 (first lane) or in HepG2 (fourth lane); one major band at about 100 kDa can be detected as well as two other bands at about 70 and 50 kDa (bands marked with arrows). This is consistent with what we have observed previously for the rat KIM-1 protein, which also appears as three distinct bands after SDS-PAGE (4). The same three bands were also detected with a polyclonal antiserum raised against a synthetic peptide corresponding to the C terminus of human KIM-1b (Fig. 3B). Although the expected size for the human KIM-1 polypeptide is 36 kDa, one can expect the protein band at a much higher apparent molecular mass because it contains four potential sites for N-glycosylation and multiple O-glycosylation sites (Fig. 1, A and B). Cell surface biotinylation of 769-P cells revealed that the 100 kDa band is the cell surface (mature) form of KIM-1 because it is the only one that is biotinylated (Fig. 3D, second lane). Analysis of the conditioned culture medium using AKG7 (Fig. 3C)  KIM-1 protein migrating about 10 kDa smaller than the mature KIM-1 band. The soluble KIM-1 protein released by surface-biotinylated 769-P cells is itself biotinylated, and its appearance in the medium is concomitant with a decrease of the biotinylated cell surface KIM-1, indicating that soluble KIM-1 originates from the cell surface (mature) KIM-1 (Fig. 3D). Similarly, 35 S-labeled soluble KIM-1 appears in the medium of 769-P cells metabolically labeled with [ 35 S]cysteine; this occurs at the apparent expense of the cell surface form, which disappears with a half-life of about 6 h (Fig. 3E). The soluble form of KIM-1 produced by HK2 and 769-P cells was also detected with ACA12 mAb but not with ABE3 mAb or the anti-KIM-1b Cterminal polyclonal antibodies. These results indicate that the soluble KIM-1 protein released in the medium must consist of the ectodomain excluding the ABE3 binding site. Interestingly, a protein band migrating close to the gel front was detected with the anti-KIM-1b C terminus antibodies but not with any of the anti-ectodomain mAbs (Fig. 3B). None of the four KIM-1 forms found in extracts of detergent-lysed 769-P cells is detected in hydrophilic extracts of sheared cells, indicating that all of them are membrane-associated. These results fit the model of a surface proteolytic cleavage of a transmembrane protein releasing a soluble form into the extracellular milieu and leaving a cell-associated C-terminal stalk. The possibility that soluble human KIM-1 might arise from alternative splicing of the hRNA was excluded when we observed that the soluble human KIM-1 was also produced when recombinant human KIM-1b was expressed from a cDNA construct in COS-7 cells (Fig. 4). Like the native mature KIM-1, the mature recom-binant KIM-1b made in COS-7 cells runs in the gel at about 100 kDa, and a small product running at about 14 kDa can also be detected exclusively with the anti-human KIM-1b C-terminal polyclonal antibodies. Analysis of cell culture supernatant using AKG7 showed very clearly that a soluble form of KIM-1 is released and accumulates in the cell culture medium (Fig. 4A). Like soluble endogenous KIM-1, the released recombinant KIM-1 is not detected with the anti-hKIM-1b C-terminal antibodies (Fig. 4C). It is also undetected with ABE3 (Fig. 4B), although a faint band is sometimes seen at high concentration of the protein, suggesting that part of the epitope is left after cleavage, allowing some weak binding of the antibody. A similar release of soluble KIM-1 is also observed when the hKIM-1a variant was expressed in COS-7 cells (data not shown), indicating that both variants are equivalent substrates for the protease. Together, these data strongly suggest that soluble KIM-1 is released into the extracellular milieu by proteolytic cleavage at a site proximal to the transmembrane domain and overlapping with the ABE3 binding epitope (Fig.  4D). N-terminal sequence analysis by Edman degradation of the soluble KIM-1b isolated by immunoprecipitation and SDS-PAGE identified the sequence SVKVGGEAGP, confirming the signal sequence prediction determined by the method of von Heijne and co-workers (20) using the SignalP Prediction program.
Inhibition of KIM-1 Shedding with ABE3 mAb-The fact that ABE3 does not bind to either of the two KIM-1 shedding products (soluble form and C-terminal fragment) suggested that the ABE3 binding site had been destroyed by the cleavage and therefore that the proteolytic nick resides within the ABE3 binding site. This also suggested to us that if ABE3 was capable of binding to native cell surface KIM-1, it could perhaps block the protease by steric hindrance of its KIM-1 substrate and thus inhibit the release of the soluble form of KIM-1. The binding of ABE3 to 769-P cells was tested by FACS (Fig. 5A) and by immunofluorescence microscopy (Fig. 5B). Results showed that ABE3 binds to native cell surface KIM-1, although poorly compared with another anti-hKIM-1 mAb, ARD5, which binds to the Ig-like domain of hKIM-1. This observation was not unexpected, as one could anticipate a poor accessibility of the ABE3 epitope because it lies in the glycan-protected mucin domain. Interestingly, inefficient binding to cell surface KIM-1 was also observed with AKG7, another mAb binding to the mucin domain (data not shown). Despite the poor binding of ABE3 to native KIM-1, its addition to the culture medium of 769-P cells results in a significant reduction of the soluble KIM-1 released from the cell surface (Fig. 6A) without chang-ing significantly the amount of KIM-1 at the surface of the cells (Fig. 6B). Taken together, these results indicate that ABE3 reduces shedding by directly inhibiting the proteolytic cleavage of KIM-1 rather than by lowering the concentration of KIM-1 at the cell surface as it would occur if ABE3 was inducing the internalization of KIM-1. The inhibitory effect of ABE3 on KIM-1 shedding from either 769-P cells or KIM-1-expressing COS-7 cells could be measured even after 2 days of culture without appreciable change in cell proliferation (data not shown). These results confirmed that the KIM-1 cleavage site lies at, or close to, the binding site of ABE3 and raises the possibility of using this antibody to test the functional role of KIM-1 shedding in various cellular assays. A single chain variable fragment form of ABE3, which should filter through the glomeruli and reach the renal tubular epithelial cells, may allow us to test the biological function of the KIM-1 shedding in vivo.

Inhibition of the Shedding with Metalloproteinase Inhibitors-An increasing number of reports have implicated MMPs
and ADAMs in the specific cleavage of cell surface proteins. BB-94 and GM6001 are two broad spectrum hydroxamic acidbased zinc metalloproteinase inhibitors that inhibit several metalloproteinases and have been used in many studies (21)(22)(23)(24). In particular, BB-94 has proved to be a potent inhibitor of TACE (23).
The inhibitory activity of various concentrations of BB-94 and GM6001 on the shedding of KIM-1 from 769-P cells was evaluated in two different experimental protocols. In the first protocol, the amount of soluble KIM-1 released into the conditioned media after 18 h of incubation in serum-containing medium was measured by ELISA (Fig. 7B), and the cells were stained with the mitochondria dye MTT to evaluate the cytotoxicity of the two compounds (25,26) (Fig. 7A). In the second protocol, soluble KIM-1 was detected by Western blot analysis of the conditioned medium after 1 h of culture in serum-free medium (Fig. 7C).
Although there was no apparent cytotoxicity with GM6001 at concentrations ranging from 0.5 to to 16 M, a small decrease of the MTT staining was observed with BB-94 at concentrations higher than 1 M, indicating that its activity can ultimately result in a decrease of cell number (Fig. 7A). An almost complete inhibition of KIM-1 shedding was achieved in the presence of either MMP inhibitor at 16 M (Fig. 7B). BB-94 showed a slightly higher potency than GM6001 with respective IC 50 values of about 1 M and 4 M. These results indicate that the cleavage of KIM-1 is mediated by a metalloproteinase, possibly a member of the MMP or the ADAM family. A very clear inhibition of KIM-1 shedding was also observed in serumfree medium (Fig. 7C). Quantitation of the soluble KIM-1 bands (media) and the mature KIM-1 (cell extracts) revealed that in the absence of MMP inhibitor, the equivalent of about 2% of the mature KIM-1 protein was released in 1 h of incubation at 37°C from the cells incubated in serum-free medium.
Activation of KIM-1 Shedding with PMA-Phorbol ester compounds such as PMA, which are known to activate protein kinase C, have been shown to induce the shedding of many proteins mediated by metalloproteinases of the ADAM family such as TACE. We tested the induction of the release of KIM-1 in serum-free medium from 769-P cells upon stimulation with 100 ng/ml PMA (Fig. 8A). Western blot analysis of the concentrated conditioned media showed that PMA increases the shedding of KIM-1 by only 2-fold (as evaluated by densitometry of the protein bands). Although modest, this increase was ob-served consistently in several experiments. The PMA-induced shedding of KIM-1 also appeared to be inhibited by BB-94 to the same extent as the constitutive shedding, indicating that the same metalloproteinase might act constitutively and after stimulation with PMA. DISCUSSION Changes in cell state during development, regeneration, or immune response require a rapid modulation of both surface protein expression and release, resulting in changes in the repertoire of proteins expressed at the cell surface (18,27). After acute injury to the kidney, the regeneration of the proximal tubule epithelium requires dedifferentiation and proliferation of the viable cells bordering the damaged areas to reconstitute an intact functional epithelial layer. This transition from normal epithelial cells to dedifferentiated cells is associated with a dramatic up-regulation of KIM-1 expression (4). Cell adhesion molecules have multiple roles that relate in important ways to epithelial function. In addition to tethering cells to the extracellular matrix and interconnecting cells to one another, they are also involved in cell locomotion, proliferation, and differentiation (28 -30). Preliminary studies have shown that recombinant KIM-1 expression results in an altered cell motility and scattering, indicating that KIM-1 could be involved in the migration of the dedifferentiated cells, facilitating the reconstitution of a continuous epithelial layer. The present paper describes the constitutive release of the ectodomain of KIM-1 by a metalloproteinase-dependent cleavage from cells in vitro. KIM-1 is expressed in the proximal tubule of humans with acute tubular necrosis, and soluble KIM-1 is detected in urine of patients with kidney disease (31).
Ectodomain shedding has been reported for a number of adhesion molecules including VCAM-1 (32), ICAM-1 (33), Lselectin (34), NCAM (35), E-cadherin (36), and L1 adhesion molecule (37). Although the cleavage of adhesive molecules can antagonize adhesion directly by down-regulation (36), or indirectly by competition (33,34), shedding can also promote adhesion when the membrane-released ectodomain binds to two cell surface or matrix components. In the case of L1, it has been observed that the shed soluble form binds to extracellular matrix and can support integrin-mediated cell adhesion and migration (14,16).
Integrins play an important role in the polarization of tubular epithelial cells. As tubular cells regenerate, they pass through a stage at which certain integrins are expressed in a nonpolarized fashion on plasma membranes. Subsequently as the cells differentiate and become polarized, these integrins become expressed only on the basal membrane. Integrin depolarization may result in the exfoliation of epithelial cells in the tubular lumen during acute renal failure (28,38).
Although the identity of the receptor(s) to which KIM-1 binds is still unknown, we have observed that soluble KIM-1 binds to various cell types in vitro and that the Ig-like domain is sufficient for its binding. 3 It is tempting to speculate that the Ig-like domain of KIM-1 could bind to one or several integrins and that its role would be to quench some apically located integrins, thus helping to avoid the undesirable attachment of exfoliated cells to one another and fibronectin, which contributes to cast formation and tubular obstruction (28).
In addition, the mucin domain of KIM-1, which would thus be oriented toward the lumen of the tubule, could provide a protective and antiadhesive sheet for the regenerating proliferating tubular cells. Indeed, in addition to their well known cytoprotective role of epithelia, mucins have been shown to have antiadhesive properties by disrupting molecular interactions between cells and between cells and the extracellular matrix (39 -43). The overexpression of antiadhesive mucins y carcinoma cells could also potentiate metastasis as described recently for MUC4 (44). We have observed that the kidney carcinoma cell line 769-P shows a very high level of expression of KIM-1. Preliminary data showed that KIM-1 is expressed on kidney tumor cell lines. Furthermore, immunohistochemical studies of kidney biopsies from patients with renal cell carcinomas show a clear overexpression of KIM-1 in the cancerous tissue (45). These observations suggest that KIM-1 and its soluble isoform may play a role in tumor progression and metastasis.
Although small molecule inhibitors can be used to inhibit metalloproteinases, a therapeutic intervention that would rely on a mAb to inhibit the shedding of a specific cell surface protein would offer the advantage of a unique and precise target. Herceptin (Transtuzumab), a humanized anti-HER2 receptor mAb commercialized for the treatment of breast cancer, was shown to prevent the release of HER2 ectodomain (46).
Early detection of tubular damage is likely to increase the chance of timely reversal of precipitants of injury and prevention of acute renal failure. Because KIM-1 up-regulation occurs very rapidly after tubular injury (31) and KIM-1 is shed into the urine (31), it is possible that detection of urinary KIM-1 could be used as a noninvasive test for early assessment of tubular integrity. Urinary KIM-1 might also be useful as a biomarker for renal cancer.
In summary, KIM-1 sheds its ectodomain by a metalloproteinase-dependent process. Shedding occurs constitutively and is increased to a modest extent by phorbol esters. The shedding occurs in vivo, with the ectodomain appearing in the urine of patients with acute tubular necrosis. A panel of antibodies has been raised to the molecule and the epitopes mapped. One of these antibodies is effective in blocking the cleavage of the ectodomain. Recognition that the ectodomain of KIM-1 is released in a regulated manner will help lead to an understanding of the functional role of this membrane glycoprotein. Understanding the role played by the metalloproteinase(s) and how antibodies can alter the shedding will provide insights into possible therapeutic strategies, which may alter the function of KIM-1 and related molecules.