Cell surface ectodomain cleavage of human amphiregulin precursor is sensitive to a metalloprotease inhibitor. Release of a predominant N-glycosylated 43-kDa soluble form.

Biosynthesis and processing of amphiregulin (AR) have been investigated in human colorectal (HCA-7, Caco-2) and mammary (MCF-7) cancer cell lines, as well as in Madin-Darby canine kidney cells stably expressing various human AR precursor (pro-AR) forms. Both cells expressing endogenous and transfected AR produce multiple cellular and soluble forms of AR with an N-glycosylated 50-kDa pro-AR form being predominant. Our results demonstrate that sequential proteolytic cleavage within the ectodomain of the 50-kDa pro-AR form leads to release of a predominant N-glycosylated 43-kDa soluble AR, as well as the appearance of other cellular and soluble AR forms. Cell surface biotinylation studies using a C-terminal epitope-tagged pro-AR indicate that all cell surface forms are membrane-anchored and support that AR is released by ectodomain cleavage of pro-AR at the plasma membrane. We also show that pro-AR ectodomain cleavage is a regulated process, which can be stimulated by phorbol 12-myristate 13-acetate and inhibited by the metalloprotease inhibitor, batimastat. In addition, we provide evidence that high molecular mass AR forms may retain the full-length N-terminal pro-region, which may influence the biological activities of these forms.

Amphiregulin (AR) 1 is one of six mammalian ligands that bind the epidermal growth factor receptor (EGFR/HER-1). AR, which was originally isolated from conditioned medium of a PMA-treated human breast carcinoma cell line, is a cell typedependent bifunctional modulator of cell growth (1). AR protein is synthesized as a 252-amino acid transmembrane glycopro-tein, and two major soluble forms of 78 and 84 amino acids, which migrate between 20 and 25 kDa on SDS-PAGE, have been characterized (1)(2)(3). In addition, soluble AR forms of 9.5-10, 35, and 55-60 kDa have been identified in several other cell types (4 -8). The variety of soluble AR forms implies intricacies in AR processing that have not yet been elucidated.
AR mRNA has been isolated from normal human tissues including ovary, placenta, testis, pancreas, spleen, kidney, lung, breast, and colon (3). For normal colonic mucosa, AR protein expression has been localized to more differentiated cells in the non-proliferative compartment (9). AR mRNA and protein have been detected in many colon cancer cell lines, primary colorectal tumors, and metastatic colorectal tumors (9 -12). Although AR is expressed in normal colonic mucosa, it is uniformly overexpressed in colon cancer (11). In addition, AR mRNA and protein have been detected in normal human mammary epithelial cell and normal human keratinocyte cultures (6,13,14). AR has been shown to act as an autocrine factor for human colon carcinoma cell lines, normal human mammary epithelial cell lines, and normal human keratinocytes (9,(13)(14)(15)(16)(17)(18). Recent studies indicate that AR may be involved in upregulation of matrix metalloproteases, invasion by tumor cells, as well as hyperproliferation and inflammation observed with psoriasis (19 -21). These findings suggest that AR is involved in both normal and aberrant cellular processes. AR elicits biological effects by binding directly to and activating EGFR/HER-1 (22). Heterodimerization of EGFR/HER-1 with other HER family members enables AR to activate other HER family members, such as HER-2 (23,24). Whereas initial receptor binding assays suggested that AR has a lower affinity for EGFR/HER-1 than EGF (2), recent studies indicate that the affinities of EGF and AR for EGFR/HER-1 may be more similar than previously observed (13,27). Studies of recombinant AR forms with extensions C-terminal to mature AR have indicated that these forms have increased biological activity (25)(26)(27), suggesting that differential processing of the C terminus of mature AR may be important in regulating the different biological responses elicited by AR.
To date, most of the biochemical characterization of AR has been performed under steady state conditions (1,3,(5)(6)(7)(8). The present studies provide a detailed analysis of the biosynthesis and processing of AR in epithelial cells, both endogenous AR in human colorectal and mammary cancer cells, as well as various human pro-AR forms stably expressed in MDCK cells. We show that a predominant 50-kDa pro-AR form can be detected at the cell surface and is preferentially cleaved at the distal site within the ectodomain to release a major 43-kDa AR form into the medium. Sequential ectodomain cleavage of 50-kDa cell surface pro-AR, however, can lead to the appearance of other cell surface pro-AR forms and to the release of several soluble AR forms. The distal cleavage event can be stimulated by PMA and inhibited by the metalloprotease inhibitor, batimastat.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Cell culture reagents were purchased from Life Technologies, Inc. Chemicals were purchased from Sigma, unless otherwise stated. Tran 35 S-label was purchased from ICN Biomedicals (Costa Mesa, CA). Sulfo-NHS-LC-biotin and protein A-agarose were purchased from Pierce. N-Glycosidase F, endoglycosidase H (endo H), O-glycosidase, and neuraminidase were purchased from Boehringer Mannheim. All electrophoresis reagents were purchased from Bio-Rad. Rainbow markers and ECL kit were purchased from Amersham Pharmacia Biotech. Long form of recombinant human AR was purchased from R & D Systems Inc. (Minneapolis, MN). The matrix metalloprotease inhibitor batimastat (BB94) (28) was kindly provided by Dr. Peter Brown (British Biotech, Oxford, UK).
Monoclonal antibodies to human AR (AR 6R1C2.4 (AR mAb), AR 4.14.18, and AR 12.38.4) were used in this study (6,7). Monoclonal antibodies to the 9E10 epitope of c-MYC (MYC mAb) and to heparan sulfate were obtained from American Type Culture Collection (Rockville, MD) and Seikagaku America, Inc. (Ijamsville, MD), respectively. Monoclonal antibody to human EGFR (mAb 528) was generously provided by Dr. Hideo Masui (Memorial Sloan-Kettering Cancer Center, NY). Affinity purified rabbit antisera to mouse immunoglobulin was purchased from Cappel Laboratories (Durham, NC). Peroxidase-conjugated donkey anti-mouse IgG (HRP) and peroxidase-conjugated streptavidin (SA-HRP) were purchased from Jackson ImmunoResearch.
For polarizing cell lines, integrity of tight junctions was assessed by measuring transepithelial resistance across the Transwell filter with a Millicell Electrical Resistance System (Millipore Corp., Bedford, MA). Experiments were performed when resistance was Ͼ200 ohms⅐cm 2 .
AR Constructs, Transfection, and Selection of Expressing Clones-Two human AR cDNAs encoding the wild type AR precursor were obtained from Dr. Gary Shipley (Oregon Health Sciences University, Portland, OR) (13) and Dr. Greg Plowman (Sugen, Redwood City, CA) (3) and were subcloned into the pCB6 CMV expression vector. Both human AR cDNAs were used in these studies and gave equivalent results. A secreted AR construct (ARsec184) which ends in . . . ER-CGEK 184 was generated by excising the EcoRI-NotI fragment (885 base pairs) from the full-length pro-AR cDNA, digesting with HphI to isolate a 576-base pair fragment, and ligating the HphI fragment with HphI/ BamHI annealed adapter oligonucleotides (5Ј-GGGAAAAGTGATA-AG-3Ј and 3Ј-CCCCTTTTCACTATTCCTAGG-5Ј) into Bluescript II (SKϪ). Subsequently, an XbaI-EcoRI fragment was excised and subcloned into pCB6. A secreted AR construct (ARsec190) which ends in . . . SMKTHS 190 was generated in the same manner as ARsec184 using annealed adapter oligonucleotides (5Ј-GGGAAAAGTCCATGAAAACT-CACAGCTGATAAG-3Ј and 3Ј-CCCCTTTTCAGGTACTTTTGAGTGT-CGACTATTCCTAGG-5Ј). The C-terminal MYC-tagged AR construct (AR/c-MYC) was generated by isolating an EcoRI-RsaI fragment and ligating the RsaI fragment with RsaI/BamHI-annealed oligonucleotides (5Ј-ACATGCTATAGCAGAGCAGAAGCTGATCTCCGAGGAGGACC-TTTAATGAG-3Ј and 3Ј-TGTACGATATCGTCTCGTCTTCGACTAGA-GGCTCCTCCTGGAAATTACTCCTAG-5Ј). All constructs were confirmed by DNA sequencing. Constructs were transfected into MDCK II and CHO cells by calci-um-phosphate precipitation method as described previously (31). Transfected MDCK II and CHO cells were selected in medium containing 500 and 1000 g/ml G418, respectively. Individual clones were initially screened by indirect immunofluorescence, and high expressing clones were subsequently screened by metabolic labeling combined with immunoprecipitation.
Metabolic Labeling-All cells were grown on 24-mm Transwell filters in the appropriate medium for 4 -6 days. To increase AR expression in transfected cell lines, cells were treated with 5 mM sodium butyrate in the appropriate medium for 16 -18 h before labeling. For pulse and 2-h labeling, cells were rinsed 2 times with serum-free, L-cysteine/L-methionine-free DMEM (DMEϪ) and incubated in DMEϪ for 30 min at 37°C. Cells were labeled with 1-2 mCi/ml Tran 35 S-label as described previously (31). For pulse-chase experiments, cells were labeled at 37°C for the indicated amount of time, rinsed once with 18°C chase medium (DMEM containing 10 times excess L-cysteine and L-methionine), and incubated at 37°C with chase medium for the specified times.
Cell Surface Biotinylation-Cells were grown on 24-mm Transwell filters as described above. All steps were performed on ice or at 4°C, unless otherwise stated. Prior to biotinylation, cells were rinsed 3 times with PBS containing 0.1 mM CaCl 2 and 1.0 mM MgCl 2 (PBS ϩ ). For polarizing cells, 1.5 mg/ml sulfo-NHS-LC-biotin in PBS ϩ was added to the basal compartment, and PBS ϩ was added to the apical compartment, and cells were incubated for 30 min. Non-polarizing cells were incubated with 1.5 mg/ml sulfo-NHS-LC-biotin in both the apical and basal compartments. Following biotinylation, cells were rinsed and washed 2 times with ice-cold 100 mM glycine in PBS ϩ and were rinsed 3 times with PBS ϩ containing 0.2% BSA (PBS/BSA).
For biotinylation-chase experiments, cells were rinsed once with 4°C serum-free DMEM and twice with PBS ϩ . Cells were biotinylated as described above. Following biotinylation, cells were rinsed twice with PBS ϩ , rinsed once with serum-free DMEM, and incubated at 37°C with the indicated chase medium for the specified time.
AR Immunoprecipitation-All immunoprecipitation protocols were performed at 4°C or on ice, unless otherwise stated. After final washes, filters were cut out, and cell lysates were prepared as described previously (31). Cell lysates were incubated overnight with AR mAb (0.2 g/ml). Specificity of AR mAb for the different forms of AR was determined by preincubating the antibody with an excess (10 -100 ng/sample) of recombinant human AR before immunoprecipitation. In order to precipitate antibody-AR complexes, affinity purified rabbit anti-mouse IgG antibody was added for 1 h followed by addition of a 50% slurry of protein A-agarose for 2 h. For immunoprecipitation from conditioned medium, phenylmethylsulfonyl fluoride (Boehringer Mannheim) was added to 2 mM, and the medium was precleared, filtered through a 0.2-m filter (Gelman Sciences, Ann Arbor, MI), and incubated overnight with AR mAb (0.05 g/ml). Protein A-agarose was pelleted and complexes were stringently washed. Immunoprecipitates were analyzed under reducing conditions on 12.5% SDS-PAGE, unless otherwise specified. For metabolically labeled samples, gels were fixed, treated with Amplify (Amersham Pharmacia Biotech) for 30 min, and dried, and fluorography was performed with BioMax MR film (Eastman Kodak Co.).
Enzymatic digestion of immunoprecipitates with endo H, N-glycosidase F, O-glycosidase, and neuraminidase was performed using the manufacturer's protocol.
AR Western Blotting-Biotinylated samples were separated by SDS-PAGE under reducing conditions, and non-labeled samples were separated under non-reducing conditions to eliminate confounding detection of heavy and light immunoglobulin chains. After separation by SDS-PAGE, immunoprecipitates for biotinylated or non-labeled samples were electrophoretically transferred overnight at 30 V to nitrocellulose (0.2 m, Bio-Rad). All steps were performed at room temperature. Membranes were rinsed twice in TBS-T (1ϫ TBS, Tween 20) and blocked for 30 min in TBS-T containing 3% BSA (TBS-T/BSA). Once blocked, membranes were incubated for 1 h with primary antibody in TBS-T/BSA (AR mAb at 5 g/ml or MYC mAb at 10 g/ml) and then rinsed and washed with TBS-T. After washing, membranes were incubated with secondary antibody in TBS-T/BSA (SA-HRP at 1:20,000 or donkey anti-mouse IgG-HRP at 1:5,000) for 30 min, washed extensively with TBS-T, and rinsed with TBS. Membranes incubated in SA-HRP received no primary antibody. The ECL Western blotting kit was used per manufacturer's directions, and fluorography was performed using BioMax MR film.
PMA and Batimastat Treatment-Cells seeded on 24-mm Transwell filters were biotinylated and washed for chase experiments as described above. For the first 5 min of chase, cells were incubated at 37°C in either serum-free DMEM containing Me 2 SO (control) or 5 M batimastat. After 5 min, the chase medium was changed to serum-free DMEM containing either Me 2 SO, 5 M batimastat, 1 M PMA, or a combination of batimastat and PMA, and incubated at 37°C for the indicated time. Cells and conditioned medium were collected, and AR immunoprecipitation combined with detection of biotinylated AR by Western blotting was performed as described previously.

Multiple Endogenous Soluble and Cellular Forms of AR Are
Detected for Epithelial Cells-To identify the forms of AR produced in epithelial cells under basal conditions, we examined newly synthesized AR in two human colon cancer cell lines, HCA-7 and Caco-2, as well as a human breast cancer cell line, MCF-7, that produce AR. Cell lysates and conditioned medium from cells metabolically labeled with Tran 35 S-label for 2 h were immunoprecipitated with AR mAb, and immunoprecipitates were analyzed by SDS-PAGE and fluorography. A predominant diffuse 43-kDa AR form, a less intense 19-and 21-kDa doublet, and weak immunoreactive form of Ͻ14.3 kDa running at the dye front were detected in conditioned medium (Fig. 1A). In separate experiments on higher percentage SDS-PAGE, the weak immunoreactive form was resolved as a 9-kDa band (data not shown). In addition, clonal MDCK cell lines stably transfected with a cDNA for wild type human pro-AR (MDCK-AR) released similar AR forms (Fig. 1A).
Analysis of total cell lysates of HCA-7 cells revealed 50-, 45-, 26-and 28-kDa doublet and 16-kDa forms of AR (Fig. 1B). All cellular AR bands had a diffuse appearance, except for 45-kDa AR which appeared as a very narrow band. A similar profile was observed for total cell lysates from MDCK-AR cells (Fig.  1B). The 50-kDa AR appeared to be the predominant cellular form. All AR forms were also detected with other AR antibodies to demonstrate AR specificity (data not shown), and AR Western blotting confirmed that each form contained AR immunoreactivity (data not shown). A summary of the biosynthesis and processing of pro-AR is shown in Fig. 8.
Biosynthesis and Processing of Wild Type Pro-AR in MDCK Cells-To delineate potential relationships among cellular and soluble AR forms, we performed pulse-chase experiments using MDCK-AR cells. Cells were metabolically labeled with Tran 35 -S-label for 20 min and chased for various times, and total cell lysates were analyzed. At 0 min of chase, a narrow band of 45 kDa and a diffuse band of 50 kDa were detected ( Fig. 2A). In independent pulse-chase experiments examining total cell lysates, only the 45-kDa band was detected by 0 min of chase after 5 min of metabolic labeling, indicating it was a more immature pro-AR form (data not shown). By 20 min of chase, the 45-kDa band had diminished, which coincided with an increase in intensity of the 50-kDa AR. Intensity of the 50-kDa AR peaked at 20 min of chase and decreased with a half-life of ϳ20 min. At 20 min of chase, a 26-and 28-kDa doublet and a 16-kDa AR form were also detected; however, intensities of these forms were significantly less than the 50-kDa form. Based on an independent pulse-chase experiment involving a 5-min metabolic labeling with Tran 35 S-label, increasing intensities of the 26-and 28-kDa doublet and the 16-kDa band were concurrent with loss of the 50-kDa AR through 40 min of chase, suggesting that the 26-and 28-kDa doublet may be derived from the 50-kDa AR (data not shown). By 4 h chase, no cellular AR forms were detectable.
To examine specifically cell surface AR forms, we performed pulse-chase experiments in combination with cell surface biotinylation. Cells were pulsed with Tran 35 S-label, chased for various periods, and cell surface-biotinylated at the end of each chase. AR immunoprecipitates from total cell lysates were eluted, re-immunoprecipitated with streptavidin-agarose, and analyzed by SDS-PAGE and fluorography. The 50-kDa, 26-and 28-kDa doublet, and 16-kDa forms were detected at the cell surface. The inability to detect 45-kDa pro-AR at the cell surface, together with its rapid kinetics of appearance in pulsechase experiments, support that it represents an immature FIG. 1. Multiple cellular and soluble AR forms are detected in epithelial cells. A, HCA-7, Caco-2, MCF-7, as well as MDCK cells expressing wild type human pro-AR (MDCK-AR) grown on Transwell filters were metabolically labeled with Tran 35 S-label for 2 h and chased for 2 h. Conditioned medium was collected and immunoprecipitated with AR mAb. B, HCA-7 or MDCK-AR cells grown on Transwell filters were metabolically labeled with Tran 35 S-label for 2 h. After labeling, total cell lysates were prepared, and AR immunoprecipitation was performed. All immunoprecipitates were analyzed by SDS-PAGE and fluorography. Arrowheads indicate the various AR forms, and molecular masses are indicated in kilodaltons.

FIG. 2. Biosynthesis of wild type pro-AR in polarized MDCK cells.
MDCK-AR cells grown on Transwell filters were metabolically labeled for 20 min and chased for various amounts of time. After chase, cell surface biotinylation was performed. AR immunoprecipitation was performed on total cell lysates (A) and conditioned medium (C). For cell surface AR (B), AR immunoprecipitates from total cell lysates were eluted and re-immunoprecipitated with streptavidin-agarose. All immunoprecipitates were analyzed by SDS-PAGE and fluorography. Arrowheads indicate the various AR forms, and molecular masses are indicated in kilodaltons. intracellular pro-AR form.
Several forms of soluble AR were released into the medium during the 4-h chase period (Fig. 2C). The 43-kDa, 19-and 21-kDa doublet, and 9-kDa-soluble AR forms were detected by 20 min of chase, with 43-kDa AR being the predominant form. Intensities of all soluble AR forms peaked by 1 h of chase. After 1 h of chase, the 43-kDa AR form gradually diminished through 4 h of chase. Both the 19-and 21-kDa doublet and 9-kDa AR forms persisted up to 2 h of chase and were diminished by 4 h. Interestingly, the increase in 43-kDa AR coincided with the loss of 50-kDa pro-AR from the cell surface, whereas increases in the 19-and 21-kDa AR doublet and 9-kDa forms more closely paralleled loss of the 26-and 28-kDa doublet and 16-kDa cell surface forms, respectively, suggesting a direct relationship between processing of cell surface forms and release of soluble forms (see Fig. 8). These data indicate that the predominant AR forms are a 50-kDa pro-AR and a 43-kDa soluble AR. The relative amounts of these forms were confirmed by metabolic labeling experiments using [ 35 S]cysteine (data not shown).

N-Glycosylation, but Not O-Glycosylation or GAG Incorporation, Contributes to the Size of the Predominant Cellular and
Soluble AR Forms-Previous studies suggest that the major species of soluble AR are low molecular mass forms of 18 -35 kDa (1, 6). One possibility to account for the high molecular mass AR forms detected in these studies is the presence of posttranslational modifications. Potential N-and O-linked glycosylation sites have been identified for pro-AR (3), and Nglycosylation of AR has been demonstrated in PMA-treated cells (1,5,8). To determine the extent of glycosylation of cellular and soluble AR forms, cell lysates and conditioned medium from MDCK-AR cells metabolically labeled with Tran 35 S-label for 2 h were immunoprecipitated with AR mAb and digested with different glycosidases. We first examined which forms were endo H-sensitive. As shown in Fig. 3A, only the 45-kDa cellular AR is endo H-sensitive, confirming that it represents an immature pro-AR form that has not been modified in the Golgi. In addition, tunicamycin treatment resulted in the appearance of an ϳ39-kDa AR form (data not shown). These data also suggest that unmodified pro-AR has an electrophoretic mobility of ϳ39-kDa, which is much slower than the predicted electrophoretic mobility of ϳ26-kDa for the full-length, unmodified AR precursor minus the signal peptide (3). A similar electrophoretic mobility for pro-AR has been proposed based on in vitro transcription and translation of the full-length AR cDNA (8).
Since all other cellular and soluble (data not shown) AR forms were endo H-resistant, we studied more complex modifications. Digestion of cellular AR immunoprecipitates with N-glycosidase F shifted the electrophoretic mobilities of all bands except the 16-kDa AR (Fig. 3B). When soluble AR forms were digested with N-glycosidase F, electrophoretic mobilities of the 43-kDa band and the 19-and 21-kDa doublet were shifted (Fig. 3B). Taken together, these results demonstrated that N-glycosylation contributes approximately 8-kDa to the molecular mass of the predominant 50-kDa cellular and 43-kDa soluble AR forms.
To address the possibility of O-glycosylation, we performed digestions with both N-glycosidase F, neuraminidase, and Oglycosidase. No additional shifts in the electrophoretic mobilities of either cellular or soluble AR forms were observed with N-glycosidase F and neuraminidase treatment, but digestion with these enzymes and O-glycosidase resulted in slight (Ͻ0.5-kDa) shifts in the electrophoretic mobilities of either cellular or soluble AR forms (Fig. 3B), which is in agreement with previously published data (22).
Since N-and O-linked glycosylation did not fully account for the increased molecular mass of predominant AR forms and the more diffuse, high molecular mass N-glycosylated forms of AR did not shift to a single, crisp band after deglycosylation, we examined possible GAG incorporation. Several potential sites for GAG incorporation have been identified for pro-AR (3). To address whether the predominant high molecular mass forms of AR had GAGs attached, we stably transfected wild type pro-AR into a parental CHO cell line (CHO-K1) and two CHO mutants deficient in GAG formation. The CHO-677 cell line is deficient in the production of heparan sulfate, whereas the CHO-745 cell line is deficient in the production of both heparan and chondroitin sulfates (29,30). Comparison of the AR profiles from parental and mutant CHO lines, as well as MDCK cells expressing wild type pro-AR, showed no differences in electrophoretic mobilities among cellular and soluble AR forms (Fig. 4,  A and B), indicating that heparan and chondroitin sulfate attachments did not contribute to the molecular masses of predominant AR forms. This was confirmed by heparatinase digestion of AR immunoprecipitates from total cell lysates and conditioned medium of MDCK-AR and CHO-K1-AR cells, in which heparatinase had no effect on electrophoretic mobilities of the various AR forms (Fig. 4C). These results differ from previous studies in which treatment of PMA-stimulated MCF-7 cells with 4-methylumbelliferyl ␤-D-xyloside, an inhibitor of GAG addition to core proteins, resulted in a slight decrease in electrophoretic mobility of a high molecular mass soluble AR (8).
Analysis of AR Secretory Mutants Suggests Soluble 43-kDa AR Contains N-terminal Pro-region-Our results indicate that posttranslational modifications, such as glycosylation and GAG attachment, did not fully account for the increased molecular mass of 43-kDa soluble AR. Using an antibody to the AR pro-region, previous studies in transfected human mammary epithelial cells indicated that a 42-kDa cellular AR form con- FIG. 3. The AR precursor is modified by N-glycosylation. MDCK-AR cells grown on Transwell filters were metabolically labeled with Tran 35 S-label for 2 h and chased for 2 h. Total cell lysates and conditioned medium were prepared and immunoprecipitated with AR mAb. Immunoprecipitates were washed in appropriate buffers and digested with either endo H (A) or various combinations of N-glycosidase F, neuraminidase, and O-glycosidase (B) as described under "Experimental Procedures." All digests were analyzed by SDS-PAGE and fluorography. Arrowheads indicate the various AR forms; dashes indicate deglycosylated AR forms, and molecular masses are indicated in kilodaltons. * represents a minor AR precursor form with minimal glycosylation. tains the full-length N-terminal pro-region (6). To address the possibility that 43-kDa soluble AR may retain the N-terminal pro-region, we stably transfected two different secreted AR constructs containing the N-terminal pro-region (Fig. 5A) into MDCK cells. One construct (ARsec184) contains the first 184 amino acids of pro-AR terminating at the C-terminal end of the predicted mature AR (3) and therefore does not contain the six C-terminal residues of the EGF-like domain. The other construct (ARsec190) contains the first 184 amino acids of pro-AR plus the 6 amino acid C-terminal extension, and therefore contains the full-length EGF-like domain. Wild type and mutant soluble AR were immunoprecipitated from conditioned medium, and the immunoprecipitates were analyzed on SDS-PAGE. As shown previously, wild type 43-kDa, 19-and 21-kDa doublet, and 9-kDa soluble AR forms were detected. ARsec190 and ARsec184 are represented with one form each of 41 and 39 kDa, respectively (Fig. 5B). N-Glycosidase F digestion of wild type and mutant AR resulted in the same pattern of electrophoretic mobilities, which confirmed that N-glycosylation did not completely account for the increased molecular mass of 43-kDa soluble AR (Fig. 5C). The observation that the electrophoretic mobility of high molecular mass 43-kDa wild type AR was similar to both secreted AR mutants suggests that it retains the N-terminal pro-region. In addition, previous studies using PMA-treated MCF-7 cells had demonstrated that the distal cleavage site for the 78 and 84 amino acid AR forms is within the EGF-like domain (3). In the present study, this cleavage is represented by the ARsec184 mutant. Our results show that wild type 43-kDa soluble AR migrates significantly slower than either secreted AR mutant, and particularly AR-sec190, raising the possibility of a cleavage site more distal to the EGF-like domain.
Cell Surface AR Forms Are Membrane-anchored-These results suggested that the various soluble AR forms were derived by sequential processing of different cell surface forms; however, the nature of AR cell surface forms remained unclear. Because the only antibodies available are those raised against portions of the AR ectodomain, it has not been possible to distinguish whether cell surface AR forms are membrane-anchored or cell-associated. To address these two possibilities, we stably expressed wild type pro-AR containing a C-terminal c-MYC epitope tag in MDCK cells (MDCK-AR/c-MYC) (Fig.  6A). Cell surface AR forms which contain the MYC epitope tag would be predicted to possess both transmembrane and cytoplasmic domains, whereas those that did not would be predicted to have resulted from ectodomain cleavage. Cell surface biotinylation combined with AR or MYC mAb immunoprecipitation and SA-HRP Western blotting was used to specifically examine cell surface AR forms. Total cell lysates from biotinylated MDCK-AR or MDCK-AR/c-MYC cells were immunoprecipitated with AR mAb, separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. As shown in the presence of MYC-tagged pro-AR at the cell surface, cell lysates from cell surface biotinylated MDCK-AR/c-MYC cells were immunoprecipitated with either AR or MYC mAb, separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. With the exception of the 16-kDa cell surface AR form, similar AR forms were detected in MDCK-AR/c-MYC cells with both antibodies (Fig. 6C). Although the reason for lack of detection of the 16-kDa cell surface AR is unknown, the consistency of its detection after metabolic labeling (Figs. 1-4) suggests that there may be some variability in the sensitivity of detection of 16-kDa AR by cold cell surface biotinylation (see also Fig. 7). We predict that the 16-kDa cell surface form has lost much of the heparin-binding domain, and therefore the sensitivity of detection of this form by cold cell surface biotinylation may be decreased due to loss of many of the lysines that serve as potential sites for biotinylation. These results demonstrated that all cell surface AR forms contain the transmembrane and cytoplasmic domains of pro-AR.
Soluble Forms Are Derived from Ectodomain Cleavage of Cell Surface Pro-AR-We next used MYC-tagged pro-AR in a similar strategy to determine whether soluble AR forms were derived directly by ectodomain cleavage of membrane-bound cell surface forms. MDCK-AR/c-MYC cells were cell surface biotinylated and chased for 2 h to release biotinylated AR forms into the conditioned medium. Conditioned medium was immunoprecipitated with either AR or MYC mAb, separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. Unlike cell surface AR forms which can be immunoprecipitated with both AR and MYC mAbs (Fig. 6C), soluble AR forms were only immunoprecipitated by AR mAb (Fig. 6D), suggesting that soluble AR forms do not contain the transmembrane and cytoplasmic domains. It is also of interest to note that the 19-and 21-kDa soluble AR doublet detected in conditioned medium of MDCK-AR cells was not detected in conditioned medium of MDCK cells expressing secreted AR mutants (Fig. 5B). Absence of the 19-and 21-kDa AR doublet in conditioned medium of AR mutants further supports pulse-chase data (Fig. 2) suggesting these forms derive from cell surface forms of AR and not from processing of 43-kDa soluble AR. Taken together, these data support the hypothesis that soluble AR forms are derived directly from ectodomain cleavage of membrane-bound cell surface forms.
AR Ectodomain Cleavage Is Stimulated by PMA and Inhibited by Batimastat-To characterize further the release of AR from the cell surface, we performed cold biotinylation chase experiments. MDCK-AR cells grown on Transwell filters were biotinylated for 30 min and chased up to 60 min. Total cell lysates were prepared, separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. A serum-free chase was performed to determine the basal rate of cell surface pro-AR cleavage. At 0 min chase, the 50-kDa, 26-and 28-kDa doublet, and less apparent 16-kDa forms of pro-AR were detected at the cell surface (Fig. 7A). By 30 min chase, the intensity of the 50-kDa and 26-and 28-kDa doublet forms of pro-AR had begun to diminish, and at 60 min chase, the 50-kDa form was significantly reduced. The intensity of the 16-kDa pro-AR peaked at 30 min and was diminished by 60 min chase. Concomitantly, the 43-kDa, 19-and 21-kDa doublet, and 9-kDa soluble AR forms are detected in the conditioned medium at 30 min chase (Fig. 7B) and continued to accumulate up to 60 min FIG. 6. Cell surface AR forms contain transmembrane and cytoplasmic domains. A, a schematic illustrating wild type human pro-AR and wild type human pro-AR with a C-terminal c-MYC epitope tag (AR/c-MYC). MDCK-AR or MDCK-AR/c-MYC cells grown on Transwell filters were cell surface-biotinylated and chased for 1 h. B, total cell lysates from MDCK-AR and MDCK-AR/c-MYC cells were immunoprecipitated with AR mAb. Total cell lysates (C) and conditioned medium (D) from MDCK-AR/c-MYC cells were immunoprecipitated with either AR or MYC mAb. All immunoprecipitates were separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. Immunoreactivity was detected by using the ECL Western blotting detection kit in combination with fluorography. Arrowheads indicate the various AR forms, and molecular masses are indicated in kilodaltons. chase confirming that soluble AR forms represent forms released from the cell surface. The difference in rates of cleavage of the 50-kDa pro-AR form and the 26-and 28-kDa pro-AR doublet combined with differences in rates of appearance of the soluble 43-kDa and soluble 19-and 21-kDa AR doublet suggest that cleavage of the doublet is inefficient.
Since phorbol esters have been shown to activate cleavage of both pro-TGF␣ and pro-HB-EGF (32-37), we decided to evaluate what effects, if any, they had on AR release. As shown by the 30-and 60-min time points, PMA dramatically increased the intensity of the bands detected for soluble AR forms (Fig.  7B). For cell surface pro-AR forms, PMA treatment stimulated cleavage of all forms (Fig. 7A). Unlike the pro-AR processing under basal conditions, PMA efficiently activated cleavage of the 26-and 28-kDa pro-AR doublet relative to the 50-kDa form. These results indicate that PMA stimulates release of AR into the conditioned medium.
If pro-AR was being cleaved from the cell surface, then inhibition of this proteolytic process should result in a decrease in AR release and a corresponding accumulation of pro-AR at the cell surface. Since metalloproteases have been implicated in the proteolytic processing of pro-EGF (38), pro-TGF␣ (39), and pro-HB-EGF (37), we attempted to inhibit basal and PMAstimulated AR release with the metalloprotease inhibitor, batimastat. MDCK-AR cells were grown on Transwell filters, cell surface-biotinylated, and chased for up to 1 h with serum-free medium containing either Me 2 SO, batimastat, PMA, or a combination of PMA and batimastat. When cells were chased with batimastat only, the cell surface forms were maintained, and no soluble AR forms were detected at 30 or 60 min of chase, indicating that batimastat completely blocked release of AR from the cell surface. Similar results were obtained for cells chased with PMA and batimastat. A similar profile was observed for HCA-7 cells (data not shown). Taken together, these results demonstrate that AR release is a regulated proteolytic process, which can be activated by PMA and inhibited by batimastat. In addition, these data strongly suggest that metalloproteases play a role in the proteolytic processing of AR. DISCUSSION Amphiregulin was originally isolated and characterized from serum-free conditioned medium of phorbol ester-treated MCF-7 cells (1). Amino acid sequencing of mature AR indicated it exists in two forms, a 78 amino acid truncated form and an 84 amino acid form which has a 6 amino acid N-terminal extension (1,2). It is predicted that these forms arise from differential processing at two distinct N-terminal, or proximal, cleavage sites within the pro-AR ectodomain, which is different from the single proximal site observed with pro-TGF␣ processing (33,40). Another difference between processing of pro-AR and other EGF-like ligand precursors is that the predicted C-terminal, or distal, cleavage site for mature AR (3) does not encompass the entire EGF-like domain. Both mature AR forms are truncated at the C terminus by six amino acids and therefore lack conserved amino acids required for high affinity binding to EGFR/HER-1 (41)(42)(43). More recent studies have demonstrated that C-terminal extension of mature AR increases its biological activity (25,27) suggesting that different forms of AR may perform distinct biological functions.
Since the initial characterization of AR, several groups have identified other cellular and soluble AR forms from a variety of cell types (1,3,(5)(6)(7)(8). In general, these studies have analyzed AR expression under steady state conditions, and most have focused on AR isolated from MCF-7 cells after prolonged PMA treatment. Only limited biochemical characterization of these novel AR forms has been performed. In the current study, we sought to define the biosynthesis and processing of pro-AR by epithelial cells under basal conditions. High Molecular Mass Cellular and Soluble AR Forms Expressed in Epithelial Cells-In non-transformed MDCK cells transfected with pro-AR as well as in two transformed colon cell lines, HCA-7 and Caco-2, which express endogenous pro-AR (data not shown), we show that newly synthesized AR is modified during transit through the secretory pathway and is rapidly delivered to the plasma membrane as an N-glycosylated 50-kDa pro-AR. Once at the cell surface, pro-AR is rapidly cleaved (t1 ⁄2 ϭ 20 min) to release soluble AR forms. Interestingly, the only other report on the biosynthesis and processing of pro-AR in a transformed mammary epithelial cell line (184A1N4-TH) indicated that cellular pro-AR persisted for at least 4 h and was only partially processed into a soluble form (6). The reasons for the differences in pro-AR processing observed between the present study and this previous report are not known, although similar findings have been described for pro-TGF␣ processing in different cell types (44).
In this study, we show that preferential processing of 50-kDa pro-AR at the distal cleavage site releases a major 43-kDa soluble AR form into conditioned medium. All other cellular and soluble AR forms are derived by sequential and/or alternate processing of cell surface 50-kDa pro-AR. In addition to the 43-kDa soluble form, we have identified a 19-and 21-kDa soluble AR doublet which probably represents the original 78 and 84 amino acid AR forms (1). This doublet is probably derived from cleavage of a 26-and 28-kDa cell surface pro-AR doublet. A minor 9-kDa soluble AR form, which is released by cleavage of a 16-kDa cell surface form, was also detected in these studies and may be similar to a previously isolated 9.5-kDa soluble AR form (5). The 9-kDa AR band is not diffuse, indicating minimal posttranslational modification and suggest-

FIG. 7. Cleavage of the AR ectodomain is stimulated by PMA and inhibited by the metalloprotease inhibitor, batimastat. MDCK-AR cells grown on
Transwell filters were cell surface-biotinylated and chased for various lengths of time in the presence of Me 2 SO (control), PMA, batimastat, or PMA and batimastat. Total cell lysates (A) and conditioned medium (B) were immunoprecipitated with AR mAb. Immunoprecipitates were separated by SDS-PAGE, transferred to nitrocellulose, and probed with SA-HRP. Immunoreactivity was detected by using the ECL Western blotting kit in combination with fluorography. Arrowheads indicate the various AR forms, and molecular masses are indicated in kilodaltons.
ing that proteolytic cleavage of 16-kDa cell surface pro-AR occurs in the heparin-binding domain and removes the Nlinked glycosylation sites. A proposed model for the sequential cleavage of cell surface pro-AR is shown in Fig. 8.
Since the two original AR forms isolated from phorbol estertreated MCF-7 cells (1), which migrated between 20 and 25 kDa on SDS-PAGE, appear to be considerably smaller than the high molecular mass AR forms identified in this study, we addressed the issue of how to account for this large discrepancy in molecular masses. Using pulse-chase analyses combined with deglycosylation and tunicamycin (data not shown) studies, we demonstrate that unmodified pro-AR has an electrophoretic mobility of ϳ39-kDa which is much slower than the predicted electrophoretic mobility of ϳ26-kDa based on the sequence for full-length pro-AR (3). A similar electrophoretic mobility for pro-AR has been proposed based on in vitro transcription and translation of the full-length pro-AR cDNA (8). In addition, posttranslational modifications of pro-AR during its transit through the secretory pathway also contribute to the masses of these pro-AR forms. We show that N-glycosylation, but not O-glycosylation or GAG attachment, contributes ϳ8-kDa to the molecular mass of pro-AR. Surprisingly, after deglycosylation, pro-AR still migrates more slowly than predicted (3) and remains diffuse (Fig. 3B) suggesting the possibility of other posttranslational modifications within pro-AR. Interestingly, pro-AR contains three potential consensus sites for tyrosine sulfation within the N-terminal pro-region of its ectodomain (3,45). One possible explanation for differences in electrophoretic mobilities of deglycosylated pro-AR and pro-AR after tunicamycin treatment is that AR is tyrosine-sulfated. Variable tyrosine sulfation may alter the electrophoretic mobility and diffuseness of pro-AR in SDS-PAGE, and studies to address this issue are in progress. Furthermore, analyses of AR-secreted mutants provide evidence that wild type 43-kDa soluble AR and 50-kDa pro-AR from which it is derived contain the full N-terminal pro-region. This is in agreement with a previous report which showed that the N-terminal pro-region was present in both 35-kDa soluble and 42-kDa cellular AR forms expressed in transformed mammary cells when transfected with either a secreted AR or a full-length pro-AR construct, respectively (6). Taken together, these data indicate that the molecular masses of 50-kDa pro-AR and 43-kDa soluble AR forms can be accounted for by their slower mobilities in SDS-PAGE, N-glycosylation, and retention of the entire Nterminal pro-region.
Nuclear Localization of AR-Another interesting point raised by rapid delivery and processing of newly synthesized pro-AR at the cell surface concerns the mechanism for nuclear localization of AR. Several studies have immunolocalized AR to the nucleus in a variety of cell types both in vitro (23,46) and in vivo (9). Two stretches of basic residues within the pro-AR heparin-binding domain are predicted not only to confer heparin binding capacity but also to act as a bipartite nuclear localization signal (2,3). Earlier studies with Schwannomaderived growth factor (47), the rat homolog of AR, suggested that these nuclear targeting signals are required for transport of Schwannoma-derived growth factor into the nucleus and for induction of its mitogenic activity in mouse NIH 3T3 cells (48). Although we have not directly addressed nuclear localization of AR in these studies, our pulse-chase data suggests that most, if not all, newly synthesized 50-kDa pro-AR is delivered directly to the cell surface where it is rapidly cleaved to release a major 43-kDa soluble AR form into the medium. This result indicates that a nuclear localization signal does not function in the secretory pathway. Interestingly, Martinez-Lacaci et al. (8) have suggested that lower molecular mass AR forms are differentially expressed in the nucleus of MCF-7 cells after PMA and 17␤-estradiol treatment. This observation raises the possibility that nuclear transport of AR occurs only after initial cell surface cleavage of 50-kDa pro-AR and involves lower molecular mass cell surface and soluble AR forms derived from secondary and tertiary cleavage events. The mechanism(s) by which dif- Alternatively, in the absence of primary proteolytic cleavage at the distal site, secondary cleavage at proximal sites removes the N-terminal domain resulting in 26-and 28-kDa cell surface pro-AR forms, which appear as a doublet in our experiments (#2). B, the 26-and 28-kDa pro-AR doublet is subject to cleavage at a distal site to release a 19-and 21-kDa soluble AR doublet from the cell surface (#3). This soluble doublet probably represents the originally described mature AR forms (1). The basal rate of cleavage of the pro-AR doublet appears to be slower than for the 50-kDa AR form, thereby increasing the cell surface half-life of the doublet. In the absence of cleavage at the distal site, further proteolytic processing at a proximal site may result in a 16-kDa pro-AR cell surface form, which has lost much of the heparin-binding domain and is not glycosylated (#4). C, the 16-kDa cell surface pro-AR may be cleaved at a distal site to release a 9-kDa soluble AR, which has lost much of the heparin-binding domain, into the conditioned medium (#5). ferent low molecular mass cellular and soluble AR species are internalized from the plasma membrane and subsequently gain access to the nucleus are not known.
All Cell Surface AR Forms Are Membrane-anchored-A distinguishing feature of several EGF-like ligands, including high molecular mass EGF, HB-EGF, AR, and betacellulin, is that they can bind heparin or heparan sulfate proteoglycans (13, 49 -51). In the case of HB-EGF, there is a heparin-binding region in the extracellular domain of the precursor (52). HB-EGF may be presented at the cell surface as either a membrane-bound precursor or as a soluble form which is cell-associated through interactions with cell surface heparan sulfate proteoglycans (37). Similarly, pro-AR contains a heparin-binding region within its ectodomain (3) raising the possibility that cell surface AR forms also may be membrane-anchored or cellassociated. We sought to distinguish between these two possibilities using a pro-AR construct that contains a C-terminal epitope tag. Cell surface biotinylation studies with MYCtagged pro-AR indicate that all cell surface AR forms are membrane-anchored (Fig. 6C). This is a surprising finding considering numerous reports have demonstrated that soluble AR can bind to heparin (6,8,13,22,53), but it is in agreement with our pulse-chase data which show that cell surface pro-AR is rapidly released and that soluble AR accumulates in the conditioned medium (Fig. 2). Interestingly, the inability of 43-kDa soluble AR to bind to cell surface heparan sulfate proteoglycans may correlate with its significantly lower affinity for heparin (8).
Although we cannot rule out the possibility that pro-AR processing can occur within the plasma membrane compartment, cell surface biotinylation experiments using MYC-tagged pro-AR provide direct evidence that pro-AR ectodomain cleavage does occur at the cell surface. This is confirmed by the inability to detect the MYC-tag in any soluble AR species (Fig.  6D). In addition, analysis of soluble AR forms released from MDCK cells expressing AR-secreted mutants provides supporting evidence that all soluble AR forms originate directly from cell surface pro-AR. In the secreted AR mutants, the abundance of the 43-kDa soluble AR and the absence of the 19-and 21-kDa AR doublet in conditioned medium (Fig. 5B) indicate that the 43-kDa soluble AR is not directly processed to produce other AR soluble forms.
Sequential Cell Surface Processing of AR-Cell surface processing of pro-TGF␣ has been studied extensively in CHO and NIH 3T3 cells and occurs by a sequential two-step process (54,55). The first cleavage step, which removes the glycosylated N-terminal domain, occurs rapidly (t1 ⁄2 ϭ 15 min) leaving a membrane-anchored form of pro-TGF␣ that contains the sequence of mature TGF␣ (32,33,56). The second cleavage step, which releases soluble mature TGF␣, occurs slowly (t1 ⁄2 ϭ 120 min) but can be stimulated by a variety of factors, including PMA, serum, and Ca 2ϩ (32,34). The same sequential cleavage of pro-TGF␣ occurs in polarized MDCK cells (31). In retrovirally transformed embryo fibroblasts and hepatocellular carcinoma cells, however, high molecular mass soluble TGF␣ forms are found in the medium, suggesting that sequential cleavage of pro-TGF␣ may be regulated differently in some transformed cell lines (44). Our analysis of the cell surface processing of pro-AR in transformed colon cell lines and in non-transformed MDCK cells indicates that the initial cleavage event for pro-AR is quite different to that for pro-TGF␣. The distal cleavage site of pro-AR is the preferred site, which causes release of 43-kDa soluble AR into the medium. In contrast to pro-TGF␣, several tyrosine sulfation sites are predicted to be within the pro-AR N-terminal domain (3) and would be juxtaposed to the proximal cleavage sites. Interestingly, tyrosine sulfation has been shown to regulate proteolytic processing of several proteins, including caerulein and gastrin (45). Although the protease(s) involved in cleavage at the proximal sites of pro-TGF␣ and pro-AR have not been characterized, one possible explanation for lack of proximal pro-AR cleavage would be that tyrosine sulfation interferes with efficient cleavage at this site.
A comparison of the electrophoretic mobilities of secreted AR mutants with wild type 43-kDa soluble AR also suggests the possibility of an alternate distal cleavage site for cell surface pro-AR. Wild type 43-kDa soluble AR migrates significantly more slowly than the ARsec190 mutant, which contains both the N-terminal domain and a C-terminal extension of six amino acids, suggesting that the C-terminal cleavage site for pro-AR is more distal to what has been predicted previously (3). Similar results were obtained with the deglycosylated AR proteins (Fig. 4). Extension of the C terminus to contain the complete EGF-like domain increases the biological activity of mature AR (25)(26)(27). It is worthwhile to note that AR contains a methionine in place of a conserved leucine found in other EGF-like ligands (2). This conserved leucine is essential for high affinity binding of TGF␣ and EGF to EGFR/HER-1 (41)(42)(43), and replacement of methionine with leucine in the C-terminally extended AR leads to a greater increase in biological activity (25,26). In addition, a recent study has demonstrated that recombinant AR84 which contains the original 84 amino acid sequence has much lower bioactivity, whereas C-terminally extended recombinant AR forms have comparable biological activity to naturally derived AR isolated from PMA-treated MCF-7 cells (27). Taken together, these studies suggest wild type 43-kDa soluble AR may have different and possibly higher affinity for EGFR/HER-1.
The 50-kDa cell surface pro-AR has a high basal rate of processing at the distal cleavage site (t1 ⁄2 ϭ 20 min), compared with the slow basal rate of cleavage of pro-TGF␣ at the distal site (t1 ⁄2 ϭ 120 min). Because of the efficient cleavage of cell surface pro-AR, it would be predicted that AR is promptly lost from the cell surface; however, immunofluorescence and cell surface biotinylation data suggest that high levels of pro-AR are expressed on the cell surface under steady state conditions. Even more surprising are the apparent high levels of the 26and 28-kDa pro-AR doublet at the cell surface under steady state conditions. Although there is preferential cleavage of pro-AR at the distal site, other less efficient proteolytic cleavage events do occur. As shown in Fig. 8, in the absence of distal cleavage, proximal cleavage of pro-AR results in the appearance of the 26-and 28-kDa cell surface AR doublet, which contains the sequence of mature AR. The high cellular levels of the pro-AR doublet detected in cell surface biotinylation experiments are in apparent contradiction with pulse-chase experiments that suggest these forms are made very inefficiently. In pulse-chase experiments, under-representation of the 26-and 28-kDa pro-AR doublet cannot be attributed to the use of Tran 35 S-label which may increase the specific activity of AR species containing the N-terminal domain because similar results were obtained using [ 35 S]cysteine which should be incorporated equally in all AR forms (data not shown). Interestingly, in cell surface biotinylation-chase experiments, the 26-and 28-kDa pro-AR doublet is less efficiently cleaved from the cell surface than 50-kDa pro-AR under basal conditions (Fig. 7). A more likely explanation for the increased levels of the 26-and 28-kDa pro-AR doublet at the cell surface under steady state conditions is that inefficient cleavage of the pro-AR doublet increases its half-life at the plasma membrane, and therefore it may accumulate at the cell surface relative to the 50-kDa pro-AR.
AR Ectodomain Cleavage Is Stimulated by PMA and Blocked by a Metalloprotease Inhibitor-Cell shedding is a process in which membrane proteins undergo limited proteolysis to release ectodomains into the extracellular medium (57,58). For membrane-anchored pro-TGF␣, pro-HB-EGF, and pro-EGF, ectodomain cleavage causes release of soluble ligands into the conditioned medium (49,59,60). In addition, cell shedding is a regulated process for many membrane proteins (55,57). For pro-TGF␣ (32)(33)(34) and pro-HB-EGF (35-37), but not pro-EGF (38), ectodomain cleavage can be stimulated by phorbol esters. In the present study, we demonstrate that ectodomain cleavage of membrane-anchored pro-AR can also be stimulated by the phorbol ester, PMA. Surprisingly, distal cleavage of the 26-and 28-kDa pro-AR doublet, which is inefficiently processed under basal conditions, is stimulated upon PMA activation to a level equivalent to that of the 50-kDa pro-AR form. At present, the mechanism(s) by which differential cleavage of the various membrane-anchored pro-AR forms occurs under basal conditions and after PMA activation are not known. One interesting possibility, however, is that under basal conditions the presence of the N-terminal domain in 50-kDa pro-AR allows a more favorable conformation of the stalk region (61) of the AR ectodomain for proteolytic activity and that this difference is circumvented upon PMA stimulation.
TNF-␣-converting enzyme, a member of the adamalysin family of metalloprotease disintegrins (62), has been shown to cleave pro-TNF␣ from the cell surface (63,64). TAPI, a metalloprotease inhibitor which blocks cleavage of TNF-␣ (63), can also inhibit spontaneous and PMA-stimulated cleavage of pro-TGF␣ (39) and pro-HB-EGF (37). In addition, basal ectodomain cleavage of pro-EGF is sensitive to another broad spectrum metalloprotease inhibitor, batimastat (38). In the present studies, batimastat efficiently inhibited both basal and PMA-stimulated cleavage of the AR ectodomain. Ectodomain cleavage of the 50-kDa pro-AR form and the 26-and 28-kDa pro-AR doublet were both blocked. Interestingly, an increased accumulation of the 26-and 28-kDa pro-AR doublet at the cell surface was observed in experiments where cell surface-biotinylated AR forms were chased in the presence of batimastat (Fig. 7). This could be explained by an insensitivity of the proximal cleavage site within 50-kDa pro-AR form to batimastat; therefore, by proteolysis at the proximal site, 50-kDa AR would slowly be converted to the 26-and 28-kDa AR doublet. This result implies that the proteases involved in the proximal and distal cleavages of pro-AR may be distinct, having different protease inhibitor specificities. A similar observation has been reported for pro-TGF␣ in which the distal, but not the proximal cleavage site, was sensitive to the metalloprotease inhibitor, TAPI (39).
In summary, multiple cellular and soluble AR forms are generated by sequential proteolytic cleavage of a predominant N-glycosylated 50-kDa pro-AR. Analyses of a predominant 43-kDa soluble AR form and secreted AR mutants suggest that high molecular mass AR forms may contain the full-length N-terminal pro-region. It has been shown previously that the N-terminal pro-region of pro-AR is required for proper folding and release of soluble AR forms (6). In addition, the presence of the N-terminal pro-region may alter interactions with CD9/ DRAP27 (65), as well as affect heparin binding and/or binding to EGFR/HER-1, thereby influencing the biological activities of high molecular mass AR forms. Studies of the N-terminal proregion, which is to be confirmed by purification and sequencing, should provide insights into structure-function relationships which may be unique to individual AR forms. In addition, cleavage of the AR ectodomain at the distal site is a regulated process, which may be stimulated by PMA and blocked by the metalloprotease inhibitor batimastat. Further definition of unique AR forms and elucidation of the mechanism(s) by which pro-AR ectodomain cleavage occurs may provide information regarding the roles of AR in normal and aberrant cellular processes, as well as potential targets for therapeutic intervention in aberrant cellular processes.