Adhesion or plasmin regulates tyrosine phosphorylation of a novel membrane glycoprotein p80/gp140/CUB domain-containing protein 1 in epithelia.

Suspension of cultured human foreskin keratinocytes (HKs) with trypsin phosphorylates tyrosine residues on an 80-kDa membrane glycoprotein, p80 (Xia, Y., Gil, S. G., and Carter, W. G. (1996) J. Cell Biol. 132, 727-740). Readhesion dephosphorylates p80. Sequencing of a p80 cDNA established identity to CUB domain-containing protein 1 (CDCP1), a gene elevated in carcinomas. CDCP1/p80 cDNA encodes three extracellular CUB domains, a transmembrane domain, and two putative cytoplasmic Tyr phosphorylation sites. Treatment of adherent HKs with suramin, a heparin analogue, or inhibitors of phosphotyrosine phosphatases (PTPs; vanadate or calpeptin) increases phosphorylation of p80 and a novel 140-kDa membrane glycoprotein, gp140. Phosphorylated gp140 was identified as a trypsin-sensitive precursor to p80. Identity was confirmed by digestion and phosphorylation studies with recombinant gp140-GFP. Plasmin, a serum protease, also converts gp140 to p80, providing biological significance to the cleavage in wounds. Phosphorylation of gp140 and p80 are mediated by Src family kinases at multiple Tyr residues including Tyr(734). Dephosphorylation is mediated by PTP(s). Conversion of gp140 to p80 prolongs phosphorylation of p80 in response to suramin and changes in adhesion. This distinguishes gp140 and p80 and explains the relative abundance of phosphorylated p80 in trypsinized HKs. We conclude that phosphorylation of gp140 is dynamic and balanced by Src family kinase and PTPs yielding low equilibrium phosphorylation. We suggest that the balance is altered by conversion of gp140 to p80 and by adhesion, providing a novel transmembrane phosphorylation signal in epithelial wounds.

Based on an in vitro model for wound activation, we reported that trypsin detachment of cultured human foreskin keratinocytes (HKs) promotes phosphorylation of tyrosine residues on an 80-kDa membrane glycoprotein (p80) (4). Phosphorylated p80 (P-p80) in suspended HKs is dephosphorylated upon readhesion to laminin 5 via integrins ␣ 6 ␤ 4 and ␣ 3 ␤ 1 . In work here, we purified and characterized p80. We wished to understand whether phosphorylated p80 identified in an in vitro deadhesion/readhesion screen (4) might be involved in physiologically significant wound activation.
Quiescent epidermis adheres to laminin 5 in the BM via integrin ␣ 6 ␤ 4 in hemidesmosome (HD) cell junctions. Wounding epidermis generates leading and following subpopulations of keratinocytes at the wound margin (5). Leading keratinocytes migrate over exposed dermal collagen via integrin ␣ 2 ␤ 1 and fibronectin via integrin ␣ 5 ␤ 1 . However, leading cells also deposit laminin 5 as a provisional BM and interact with these deposits via integrin ␣ 3 ␤ 1 (5)(6)(7). The interaction of leading cells with deposited laminin 5 generates distinct transmembrane signals when compared with interaction with dermal ligands. For example, adhesion of HKs to laminin 5 via ␣ 3 ␤ 1 promotes gap junction intercellular communication when compared with adhesion to collagen or fibronectin (5). Interactions of leading cells with collagen are inhibited by toxin B, an inhibitor of Rho GTPases, which does not inhibit the phosphatidylinositol 3-kinase-dependent adhesion or spreading on laminin 5. Deposition of laminin 5 over exposed dermal collagen switches adhesion and spreading from toxin B-sensitive to toxin Bresistant (6). Cell interaction with laminin 5 (6) or laminin 10 (8) via ␣ 3 ␤ 1 assembles minimal focal adhesions with low phosphorylation of focal adhesion kinase (FAK) when compared with adhesion on dermal ligands. Consistently, we also found that adhesion to dermal ligands was less effective than laminin 5 in regulating phosphorylation of p80 (4). Dephosphorylation of P-p80 upon adhesion to laminin 5, but not collagen, is resistant to cytochalasin D, an inhibitor of the actin cytoskeleton. Vanadate, an inhibitor of PTPs, prevents dephosphorylation of P-p80 upon adhesion, indicating that a PTP is required for the dephosphorylation. Phosphorylated p80 is detected in primary cultures of many epithelial cells (epidermal, esophageal, cervical, and gastric) but not primary cultures of fibroblasts or some immortalized epithelial cell populations (4). We concluded that adhesion-dependent phosphorylation and dephosphorylation of p80 on laminin 5 may report unique cell signals and functions in many epithelia.
Here, we describe the purification of P-p80 from trypsinsuspended HKs and its characterization. The results integrate five novel observations. (i) A novel 140-kDa transmembrane glycoprotein, gp140, is identified as the trypsin-sensitive precursor to p80 and contains extracellular CUB protein-protein interaction domains. (ii) gp140/p80 is identified as the product of the CUB domain-containing protein 1 (CDCP1) gene that is overexpressed in human colorectal and lung cancers (9), is a marker for hematopoietic stem cells, (10) and is identical to SIMA135, a membrane glycoprotein elevated in metastatic human tumor cells (11). (iii) Tyrosine phosphorylation of p80 is mediated by Src family kinase(s) on several tyrosines including tyrosine 734 upon detachment of HKs with trypsin or treatment with suramin, a membrane-impermeable polysulfonated napthylurea (12). (iv) Significantly, trypsin conversion of gp140 to p80 is duplicated by serum plasmin. (v) This conversion decreases the rate of dephosphorylation of p80 relative to gp140 in response to detachment or suramin. The studies here suggest roles for plasmin and adhesion in regulating phosphorylation of gp140/p80 in epidermal wounds.
Cell Culture, Drug Treatments, and Cell Extraction-HKs were isolated and grown in monolayer culture as previously described (13), using defined medium (KGM, Cambrex, San Diego, CA). For phosphorylation studies, HKs were (i) treated with or without inhibitors for 30 min at the designated concentrations prior to (ii) treatment with or without suramin (35 M) for 20 min and (iii) extraction with 2% Triton X-100/phosphate-buffered saline/1-2 mM each phenylmethylsulfonyl fluoride, N-ethylmaleimide (NEM), sodium fluoride (NaF), and sodium orthovanadate (Na 3 VO 4 ; collectively buffer A) for 30 min at 4°C. The Triton-soluble extract was collected by centrifugation, and the Tritonresistant pellet was extracted with 2 M urea, 1 M sodium chloride plus 1-2 mM each phenylmethylsulfonyl fluoride, NEM, NaF, and Na 3 VO 4 to collect the Triton-insoluble cellular material.
Phosphatase Assay-Total phosphatase activity in lysates of HKs was assayed as described (14) using p-nitrophenyl phosphate as substrate. Tyrosine phosphatase activity (vanadate-inhibitable) was calculated as the difference between total phosphatase activities of a given sample measured with and without pervanadate.
Purification of Tyrosine-phosphorylated p80 and gp140 -p80 and gp140 were purified from HKs using similar protocols as outlined below. However, p80 was purified from HKs suspended with trypsin-EDTA to induce tyrosine phosphorylation of p80 as previously described (4). In contrast, gp140 was purified from adherent HKs that were treated with suramin to induce tyrosine phosphorylation of gp140 and p80. Adherent HKs were removed from culture dishes with a cell scraper. Trypsin-EDTA or scrape suspended HKs were homogenized with a Dounce homogenizer in phosphate-buffered saline plus 1-2 mM each phenylmethylsulfonyl fluoride, NEM, NaF and Na 3 VO 4 . Unless specifically stated, all subsequent steps used buffers that included these protease and phosphatase inhibitors. Total cellular membrane and cytosolic fractions were collected from the homogenate after centrifugation for 1 h at 100,000 ϫ g. The membrane fraction was extracted with buffer A for 30 min at 4°C and centrifuged (4 min, 2000 ϫ g). The soluble material was incubated with WGA-agarose for 1 h at 4°C; unbound or loosely attached protein was washed away with 1% Empi-gen BB, 50 mM NaCl, 50 mM Tris, pH 7.4; and the bound material was eluted with 400 mM N-acetylglucosamine, 0.1% Triton X-100, 50 mM Tris, pH 7.4. Phosphotyrosine-containing proteins were isolated from the WGA-eluted material by immunoprecipitation with anti-phosphotyrosine mAb (4G10-agarose; Upstate Biotechnology, Inc., Lake Placid, NY) and eluted with 8 M urea with 0.1% Triton X-100 detergent, 20 mM Tris, pH 7.4. The eluted phosphoproteins were further purified by preparative SDS-PAGE and analyzed as follows.
Partial Amino Acid Sequencing of Purified p80 -Partial amino acid sequencing of purified p80 was performed by the Harvard microsequencing facility under the direction of William Lane. The purified p80 protein in gel was transferred to polyvinylidene difluoride membranes (24 nmol of purified p80 based on AA analysis). Trypsin-digestion released peptides were fractionated by reverse phase HPLC on a Zorbax C18 column, and peaks of candidate peptides were analyzed by matrixassisted laser desorption time-of-flight mass spectrometry (MALDI-TOF MS) to obtain mass number and homogeneity followed by sequencing by phenylthiohydantoin derivatization. This provided amino acid sequence from five different p80 peptides (peptides 65, 98, 38, 118, and 31; see "Results" and Fig. 2).
Identification of GP140 by Mass Spectrometry-Identification of gp140 was performed in the FHCRC mass spectrometry facility by Philip Gafkin and Angela Norbeck. The SDS-PAGE-purified gp140 was subject to in-gel trypsin digestion, and the soluble samples were desalted using Millipore C18 ZipTips and dried. The sample was then resuspended in 5 l of water and analyzed by LC MS/MS with a ThermoFinnigan LCQ DECA XP mass spectrometer (15). Data were collected in the data-dependent mode in which an MS scan was followed by MS/MS scans of the three most abundant ions from the preceding MS scan. The MS/MS data were searched against the NCBI nonredundant protein data base and a human subset of this data base using SEQUEST TM software. The resulting peptide matches were scored by SEQUEST TM , and protein identifications were considered valid if the identified protein contained at least two peptides with Xcorr scores above 2.0 and the identification did not appear in a control sample from a blank portion of the gel.
Preparation of Nondegenerated Oligonucleotide Probe from p80 Peptide Sequence-An RT-PCR protocol was used to obtain a nondegenerate oligonucleotide probe from amino acid sequence of p80 peptide 65. Degenerate oligonucleotide primers were synthesized based on the amino acid sequence of p80 peptide 65 and were kinase-labeled with [␥-32 P]ATP. The labeled primers were used in RT-PCR reactions with total RNA from HKs, isolated with Trizol (Invitrogen). For synthesis of the first strand cDNA, labeled degenerate antisense primer B (RAANA-CYTCNGGRTTGGT) was used followed by PCR amplification with a second labeled degenerate sense primer A (GTNGARTAYTAYATHCC). Both labeled primers were run in two PCRs run in parallel. The PCR products were then identified by autoradiography following electrophoresis on a 12% sequencing gel. One labeled band with approximate size of 45 bp was detected in each reaction. The DNA fragment was eluted from the gel with buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 200 mM NaCl). The purified 45-bp fragment was subcloned into PCRII vector and sequenced. Double-stranded DNA sequencing was performed on purified DNA using dye terminators (Applied Biosystems, Inc.).
Screening of cDNA Library and Sequencing-selected cDNA Clones-The p80A/B oligonucleotide probe was labeled by random priming (109 cpm/g DNA; Stratagene) and used to screen a human keratinocyte gtll cDNA library. For screening, phage plaques were immobilized on nitrocellulose (Schleicher & Schü ll) membranes and prehybridized at 37°C for 4 h in buffer containing 50% formamide, 5ϫ SSPE, 1ϫ Denhardt's solution, 0.1% SDS, and 100 g/ml denatured salmon sperm DNA. Blocked filters were incubated with labeled p80A/B probe at 37°C for 14 h and washed (2ϫ SSC and 0.1% SDS, room temperature), and positive plaques were purified. Phage were grown in Y1090 bacteria, and DNA was isolated (Lambda Preps DNA Purification, Promega, CA) and digested with EcoRI to release insert, which was gel-purified and transferred to BlueScript vector for sequencing. This screening approach identified and sequenced a single 2.0-kb p80 cDNA.
Construction and Expression of GFP-gp140 -A full-length gp140/ CDCP1 cDNA was constructed by sequential insertion of three partial cDNAs into the multiple cloning site of an pEGFP-N1 plasmid (Clontech). The correct sequence of the full-length cDNA was confirmed. The resulting full-length gp140/CDCP1 cDNA without a stop codon was connected in frame at its 3Ј-end to the 5Ј-end of GFP. In this orientation, expression of gp140 would orient the GFP at the cytoplasmic, carboxyl-terminal tail of gp140. Therefore, the GFP would remain after proteolytic conversion of gp140-GFP to p80-GFP. The gp140-pEGFP-N1 plasmid was transfected into 293T cells or mouse keratinocytes with LipofectAMINE Plus reagent (Invitrogen). Mouse keratinocytes were grown in KGM (Cambrex). 293T cells were grown in Dulbecco's modified Eagle's medium without serum after transfection as indicated. Transient transfection of the gp140-pEGFP-N1 plasmid in cells produced GFP-gp140 on the surface of cells. However, stable transfection by selection in G418 was not successful. Apparently, toxicity of the GFP-gp140 construct limited stable transfection.
Immunostaining-Confluent cultures of HKs were incubated with and without PP2 kinase inhibitor (1 M for 30 min) followed by or not followed by suramin (35 M for 10 min). Cells were fixed in 2% formaldehyde in 0.1 M sodium cacodylate, 0.1 M sucrose, pH 7.2, containing both vanadate and NEM to inhibit phosphatases and then permeabilized with 0.1% Triton X-100 detergent. Cells were stained with Ab FAK Tyr(P) 861 with and without synthetic gp140 phosphopeptide P-Y734 (10 g/ml), and bound Ab was detected with rhodamine-conjugated secondary Ab. Images were collected with a Zeiss fluorescence microscope with a Photometric SenSys cooled CCD digital camera (Roper Industries, Trenton, NJ) using MetaMorph software (Universal Imaging Corp., Downingtown, PA).

RESULTS
Adhesion Regulates Phosphorylation and Dephosphorylation of p80 -As previously shown (4), deadhesion of HKs with trypsin induces tyrosine phosphorylation of a prominent 80-kDa membrane glycoprotein, p80, and decreases phosphorylation of FAK ( Fig. 1, lanes 1 and 2). Subsequent readhesion of the suspended HKs onto laminin 5 via integrins ␣ 6 ␤ 4 and ␣ 3 ␤ 1 for 10 min increases phosphorylation of FAK and dephosphorylates p80 (Fig. 1, lane 3). This indicates that phosphorylation of p80 occurs upon trypsin suspension and dephosphorylation occurs upon readhesion. Vanadate addition prevents the dephosphorylation of p80 upon readhesion, indicating that dephosphorylation by a PTP is the primary reason for the disappearance of P-p80 (4). Surprisingly, in the subsequent 1-6 h, p80 is transiently rephosphorylated and then dephosphorylated ( Fig. 1, lanes 4 -6). This indicates that phosphorylationdephosphorylation of p80 also occurs in adherent HKs possibly as a result of transient changes in spreading or migration. These changes in phosphorylation of p80 raised questions regarding its identity and the outside-in signals that regulate tyrosine phosphorylation of p80. Below, we purify and characterize p80, evaluate tyrosine kinase and PTP that balance phosphorylation of p80, and evaluate proteolysis in regulating phosphorylation.
Purification of P-p80 from Trypsin-detached HKs and Partial Amino Acid Sequencing-Candidate studies failed to identify p80, necessitating purification of P-p80 for characterization. HKs were suspended with trypsin/EDTA to promote phosphorylation of p80. The P-p80 was isolated by (i) preparation of a membrane-rich fraction, (ii) extraction of the membranes with Triton X-100 detergent to obtain a soluble membrane extract, (iii) sequential purification of P-p80 from the extract on immobilized WGA to collect membrane glycoproteins followed by elution and affinity purification on immobilized anti-phosphotyrosine antibodies (PY20 or 4G10), and (iv) preparative SDS-PAGE. The purified P-p80 was detected with silver stain and digested with trypsin, peptides were fractionated by HPLC, and peptides were assigned mass by MALDI-TOF MS and amino acid-sequenced. This provided amino acid sequences for five p80 tryptic peptides: (peptide 65, NH 2 -EERVEYYIPGSTT-NPEVFK-COOH, mass ϭ 2261. Sequencing of a p80 cDNA Encoding a Transmembrane CUB Domain Protein-A nondegenerate p80 oligonucleotide probe was prepared based on the sequence of p80 peptide 65 and used to screen a HK cDNA library. The peptide 65 oligonucleotide probe identified a single 2-kb cDNA. This cDNA was cloned and sequenced (GenBank TM accession number AY375452). The nucleotide sequence of the 2-kb p80 cDNA was translated in all three reading frames, one of which encoded 716 amino acids, including four of the five p80 tryptic peptides (peptides 65, 98, 118, and 38; Fig. 2) but not peptide 31 (Fig. 2). This established that the p80 cDNA encodes most of the 80-kDa protein that was purified and sequenced. In this reading frame, an in-frame stop codon was present in the p80 cDNA sequence at positions 2155-2157. PCR experiments that followed indicated that the stop codon was an error in the p80 cDNA possibly originating in the cDNA library used for screening and identifying the original p80 cDNA. The 716 amino acids included 26 residues upstream of the initiating methionine identified by a Kozak consensus sequence. Numbering from the initiating methionine, the coding sequence includes 691 amino acids and the following protein motifs (Fig. 2): an amino-terminal signal peptide with a predicted cleavage site that would yield a phenylalanine amino terminus, three extracellular CUB protein-protein interaction domains (16), 11 potential N-glycosylation sites consistent with the known binding of p80 to WGA, and a transmembrane domain with a short carboxyl-terminal cytoplasmic sequence. However, there were no detectable cytoplasmic tyrosine phosphorylation sites. This is inconsistent with the observed tyrosine phosphorylation of p80. Further, the absence of p80 peptide 31 in the identified 716-amino acid sequence suggested that we did not have a full-length cDNA or the complete amino acid sequence of the carboxyl terminus of p80.
p80 Is a Fragment of CDCP1 Containing Cytoplasmic Tyrosine Phosphorylation Sites-Nucleotide sequence of the 2.0-kb p80 partial cDNA was compared with the human genome data base (available on the World Wide Web www.ncbi.nlm.nih.gov) ). Cells were extracted with Triton X-100 detergent containing vanadate, NEM, and NaF as phosphatase inhibitors (Buffer A). Soluble cell extracts (75 g of protein) were fractionated by SDS-PAGE, transferred to nitrocellulose, and Western blotted with mAbs against phosphotyrosine residues (4G10). Migration of P-p80 is indicated (P80). Migration of tyrosine-phosphorylated FAK and epidermal growth factor receptor (EGFR) were established by immunoblotting or immunoprecipitation with specific antibodies. *, an uncharacterized phosphorylated protein migrating above FAK. and identified nearly identical sequences that localized to chromosome 3p21. Significantly, published gene sequence designated CUB domain containing protein 1 (CDCP1; accession number NM022842) and derived amino acid sequence (accession number AA033397) (9) was 99% identical to the amino acid sequence of p80. Only Arg 525 of p80 was distinct from Gln 525 in CDCP1. Both MS-based amino acid sequencing of p80 (peptide 98) and LC MS/MS-based sequencing of peptides in gp140 confirmed that amino acid 525 was Arg (see sequence of peptides P6, 7 in Fig. 2 and peptides 6 and 7 under gp140 peptides in Fig. 4B). Consistently, a second published sequence for SIMA135 (11) (accession number AAK02058) was 100% identical to derived p80 amino acid sequence. Two additional unpublished but submitted sequences (accession numbers BAB15511 and AA033397) were compared and found to be nearly identical to p80 when using the MULTALIN alignment program (available on the World Wide Web at npsapbil.ibcp.fr/cgi-bin/align_multalin.pl).
The CDCP1 gene was originally identified based on its elevated expression in human gastric and lung carcinomas (9). Although no protein was identified, the CDCP1 gene encodes a putative transmembrane protein of 836 amino acids that extends 145 amino acids beyond the carboxyl terminal end of the 691 amino acids encoded by the p80 cDNA (Fig. 2). The additional 145 amino acids include the missing p80 nucleotide sequence for peptide 31 in the p80 cDNA. This established that p80 is CDCP1. The additional 145 residues contain 5 tyrosine residues. Two of these tyrosine residues define canonical phosphorylation sites for tyrosine kinases (residues 734 and 806; Fig. 2). In addition, two type II Src homology 3 ligand-binding sites (residues 716 -721 and 772-777) were identified. The 836 amino acids and glycosylation sites of the CDCP1 gene should generate a glycoprotein considerably larger than p80. This raised the possibility that a larger or precursor form of p80 exists that is encoded by the CDCP1 gene.
Suramin Promotes Tyrosine Phosphorylation of p80 and gp140 -We wished to determine whether a possible trypsinsensitive precursor to p80 existed and whether phosphorylation of p80 might be regulated by known signaling pathways. Soluble factors (growth factors or cytokines: transforming growth factor ␤1, epidermal growth factor, keratinocyte growth factor, hepatocyte growth factor, interferon-␥) or drugs that regulate protein kinases (retinoic acid, TPA, suramin) were evaluated. The compounds were added to both adherent and trypsin/EDTA-suspended HKs. Effects of the compounds on phosphorylation of p80 were detected by Western blotting with anti-phosphotyrosine mAbs (PY20 or 4G10). Most of these compounds had no detectable or inconsistent effects on phosphorylation of p80. In contrast, suramin, a polysulfonated napthylurea with multiple reported activities, stimulated a rapid and selective phosphorylation of an 80-kDa protein (Fig. 3A, lanes 2  and 4). The phosphorylated 80-kDa protein induced by suramin (35 M) co-migrated with P-p80 induced by trypsin detachment (Fig. 3A, lane 5). Incubation times as short as 1 min and as long as 30 min had similar activating effects on phosphorylation of p80. The identity of the detachment-dependent p80 and the suramin-dependent 80-kDa protein is established below, and for continuity both shall be referred to here as p80.
Significantly, suramin treatment of adherent HKs also induced phosphorylation of a major 140-kDa protein, gp140 (Fig.  3A, lane 2) that was not detected in the trypsin-rounded or trypsin-suspended HKs (Fig. 3A, lanes 4 and 5). In our gels, phosphorylated gp140 (P-gp140) migrates as a diffuse band above phosphorylated FAK. This suggested that trypsin diges- tion of adherent HKs may convert GP140 into p80 by removal of an extracellular domain. To test this idea, HKs were suspended with trypsin, which removes gp140 (compare lanes 1 and 2 in Fig. 3B), followed by readhesion for increasing time periods as indicated (Fig. 3B, lanes 3-9). In the last 10 min of each time period, HKs were treated with suramin to induce phosphorylation of gp140. In response to suramin, the HKs re-expressed P-gp140 5-6 h after readhesion, returning to predigestion levels overnight (Fig. 3B, lanes 3-9). In the absence of suramin, P-gp140 was not detectable (Fig. 3B, lane 10). The low levels of P-p80 seen at 6 h were increased significantly by suramin addition (Fig. 3B, compare lanes 8 and 10). This indicated that the decrease in P-p80 resulted from dephosphorylation, not protein catabolism.
In a series of controls (results not shown), cycloheximide, an inhibitor of protein translation, blocked re-expression of gp140 after trypsin digestion. Trypsin-suspended HKs re-expressed gp140 even when left in suspension, indicating that re-expression of gp140 did not require readhesion. Treatment of adherent HKs with Ca 2ϩ (1 mM) for 24 h prior to trypsin digestion or inclusion of Ca 2ϩ (1 mM) or EDTA (5 mM) during the trypsin digestion had no effect on levels of phosphorylated p80 or gp140 induced by suramin. Suramin treatment of cultured mouse keratinocytes also promoted tyrosine phosphorylation of p80 and gp140 and had similar effects on cultures of esophageal and cervical epithelial cells but not fibroblasts. This indicates that P-gp140/P-p80 is widely expressed in epithelium. Suramin homologues were also found to promote phosphorylation of p80 and gp140: PPADS and NF023, antagonists of ATP binding to purinergic receptors, duplicated the effects of suramin and promoted phosphorylation of p80 and gp140. Surprisingly, the membrane-impermeable dye trypan blue, used routinely to assay cell viability, is also an analogue of suramin and also induced phosphorylation of gp140/p80. XAMR 0721, a nonsymmetric suramin analogue containing only one of the two polysulfonated arms (17), failed to duplicate the effects of suramin. This established that suramin or trypan blue promotes phosphorylation of gp140 and p80 by interacting with an extracellular binding site(s) or receptor(s). The results also suggested that trypsin may convert gp140 to p80.
Purification of P-gp140 and Identification as a Trypsin-sensitive Precursor to p80 -gp140 was purified and identified by LC MS/MS to determine whether it is a trypsin-sensitive precursor to p80 and the native form of CDCP1. Adherent HKs were treated with suramin to promote phosphorylation of gp140. P-gp140 was purified utilizing a variation of the protocol used for purification of P-p80. A membrane-rich fraction of suramin-treated HKs containing P-gp140 was solubilized with Triton X-100 detergent and sequentially purified with immobilized WGA and then with immobilized anti-phosphotyrosine mAb (4G10) and finally with preparative SDS-PAGE (Fig. 4A,  lanes 2, 4, and 6). In-gel purified gp140 was compared with blank lanes of the same gel by LC MS/MS. CDCP1 was the only protein identified in the gp140 samples above common background proteins (keratins, serum albumin, titin). Identical results were obtained in two separate purifications/identifications of P-gp140 (Fig. 4B, Sample A and Sample B). The most abundant peptide ions in the two samples of P-gp140 (Fig. 4B) were searched against the NCBI nonredundant data base. This search identified gp140 as CDCP1. The localization of the identified peptides within gp140/CDCP1 is summarized in Fig. 2. We conclude that cell surface gp140 is the native protein encoded by the CDCP1 gene and a trypsin-sensitive precursor to p80. Trypsin removes ϳ35 kDa of peptide and 25 kDa of carbohydrate from the amino terminus of gp140 to generate transmembrane p80 (see Fig. 2). p80 tryptic peptide 65 was the most amino-terminal peptide identified in p80 after trypsin digestion of gp140 in HKs. Therefore, tryptic cleavage must occur at an unidentified site of gp140 on the amino-terminal side of peptide 65 in CUB domain 1.
The results in Fig. 5, A and B, suggest that phosphorylation of gp140 and p80 in response to suramin required SFKs. However, equilibrium levels of phosphorylated gp140 were also  861 Ab. This Ab cross-reacts with P-gp140 and P-p80 and is characterized in Fig. 6. P-gp140 reappeared 2-4 h after the initial trypsin cleavage. dependent on PTPs that dephosphorylate P-gp140 in culture. Vanadate or calpeptin, as inhibitors of PTPs, increased P-gp140 (Fig. 5C) and P-p80 (results not shown). Consistently, the increase in phosphorylation of gp140 in response to vanadate was inhibited by prior treatment with SU6656 (Fig. 5C,  lane 3). Thus, a PTP is active in adherent cultures of HKs, resulting in only low equilibrium levels of P-gp140. Suramin and trypan blue have a significant advantage in promoting phosphorylation of gp140 or p80 due to their membrane impermeability, selectivity, and lack of toxicity when compared with vanadate. However, it is unclear how extracellular suramin increases phosphorylation of the cytoplasmic tail of gp140/p80. Suramin has been reported to inhibit PTPs (20) in cell-free extracts. Conceivably, suramin may mimic vanadate to directly inhibit cytoplasmic PTPs to increase P-gp140. This is unlikely, since suramin is not membrane-permeable and does not increase phosphorylation of FAK, whereas vanadate does. More likely, extracellular suramin may indirectly inhibit an intracellular PTP to increase P-gp140. These two possibilities were examined by assaying PTP activity in HK homogenates in the presence or absence of suramin and vanadate. Vanadate inhibited 60% of total phosphatase activity in the homogenates of HKs using p-nitrophenyl phosphate substrate. In contrast, suramin (35 M) inhibited 13% of total phosphatase activity. Assuming that vanadate inhibits 100% of the total PTP activity in the homogenates (14) and that suramin inhibits only PTPs, then suramin inhibits 22% of total PTP activity (values are representative of five independent experiments). We conclude that suramin is a relatively poor general inhibitor of PTPs in homogenates of HKs when compared with vanadate. However, when added extracellularly, suramin strongly and selectively promotes phosphorylation of gp140/p80. We suggest that suramin may interact with a cell surface intermediate indirectly inhibiting a PTP and/or activating SFKs to increase phosphorylation of gp140 and p80.
Suramin increased phosphorylation of gp140 and p80 via SFKs but did not increase phosphorylation of other SFK substrates including FAK (Fig. 3A, compare P-FAK in lanes 1 and 2). This indicated that suramin was not generally activating SFKs. We confirmed this finding by blotting FAK with a rabbit polyclonal antibody against an Src phosphorylation site, phosphotyrosine residue 861 in FAK (FAK Tyr(P) 861 Ab; Bio-Source). Although the anti-FAK Tyr(P) 861 Ab reacts with P-FAK, it also cross-reacts with P-p80 and P-gp140 after treatment with suramin (Fig. 6, lanes 1 and 7). In contrast, Abs against phosphotyrosine residue 397 in FAK (Ab FAK Tyr(P) 397 ; BioSource) did not cross-react with P-p80 or P-gp140 (Fig. 6, lane 13).
The reaction of Ab FAK Tyr(P) 861 with P-gp140 and P-p80 was inhibited with a mimetic phosphopeptide for FAK tyrosine 861 (BIOSOURCE catalog no. 04-626P; Fig. 6, lanes 2 and 8) but not the identical nonphosphorylated peptide (BIOSOURCE catalog no. 04-626N; Fig. 6, lanes 3 and 9). The cytoplasmic tail of gp140 contains 5 tyrosine residues; Tyr 734 and Tyr 806 are candidates for phosphorylation by SFKs (Fig. 2). Consistently, a mimetic phosphopeptide for gp140 tyrosine 734 (NH 2 -GRKDNDSHVpY 734 AVIEDT-COOH, where pY represents phosphotyrosine) was synthesized and inhibited binding of Ab FAK Tyr(P) 861 to P-p80 and P-gp140 (Fig. 7, lanes 4 and 10). In contrast, synthetic phosphopeptides from gp140 tyrosines 743 (NH 2 -DTMVpY 743 GHLLQDSSG-COOH) and 806 (NH 2 -SESEPpY 806 TFSHPNNGD-COOH) had negligible inhibitory effects on binding of Ab FAK Tyr(P) 861 to P-gp140 or P-p80 (Fig. 6, lanes 5 and 6 and lanes 11 and 12). Surprisingly, phosphopeptide for gp140 tyrosine 734 was more inhibitory for binding of Ab FAK Tyr(P) 861 to P-gp140 than to P-FAK. In controls, phosphopeptide for gp140 Tyr 734 had no inhibitory effect on binding of Ab FAK Tyr(P) 397 to FAK in adherent HKs (Fig. 6, lane 14). This indicates that gp140 Tyr 734 is phosphorylated by SFKs and that Ab FAK Tyr(P) 861 cross-reacts with this phosphorylated epitope. These results do not indicate that gp140 Tyr 734 is the only phosphorylation site. Apparently, Ab FAK Tyr(P) 861 is as specific for gp140 Tyr(P) 734 as it is for FIG. 4. A, purification of P-gp140 from suramin-treated adherent HKs. P-gp140 was purified utilizing the basic protocol described for purification of P-p80 (see "Experimental Procedures") except that suramin-treated adherent HKs were used. Lanes 1-8 are immunoblots with anti-phosphotyrosine mAb (4G10) of the cell extracts prepared as follows. HKs were Dounce-homogenized in buffer A (without Triton X-100 detergent), centrifuged (100,000 ϫ g) to collect a membrane-rich pellet. The soluble cyotosolic fraction (C, lane 1) contains ϳ50% of the total cell protein but no gp140. The membrane pellet was resuspended in buffer A with Triton X-100 and then recentrifuged to collect the soluble membrane fraction containing gp140 (M, lane 2). The soluble membrane extract was incubated with immobilized WGA, and an unbound (U, lane 3) and a bound fraction (E, lane 4; eluted with 400 mM N-acetylglucosamine) were collected. The eluted gp140 was incubated with immobilized anti-phosphotyrosine mAb (4G10), and unbound (U, lane 5) and bound fractions (lanes 6 -8; eluted with 8 M urea containing 0.1% Triton X-100) were collected. A silver stain of pooled fractions is shown (lane 9). The eluted gp140 was gel-purified on preparative SDS-PAGE gel and analyzed by LC-MS/MS. B, identification of gp140 as CDCP1 by LC MS/MS. In-gel purified gp140 and a negative control gel sample were trypsin-digested and analyzed by LC MS/MS. The MS scan was followed by MS/MS scans of the three most abundant ions from the MS scan that were not present in the negative control samples. The identified peptides unique to the gp140 sample were searched against the NCBI nonredundant protein data base. The peptide matches were considered valid if the identified protein contained at least two peptides with Xcorr scores above 2.0. We obtained results on two separate purifications of gp140 (Samples A and B), both indicating that gp140 is CDCP1.

Adhesion or Plasmin Regulates Phosphorylation of gp140/CDCP1
P-FAK. Ab FAK Tyr(P) 861 provides a useful antibody reagent for preferential detection of P-p80 and P-gp140.
converted cell surface gp140-GFP to p80-GFP (now 95 kDa due to the GFP tag) (Fig. 7B, lane 9). This confirmed that gp140 is a trypsin-sensitive precursor to p80. The trypsin digestion converted only part of the total gp140-GFP to p80-GFP. Presumably, the trypsin-resistant gp140-GFP is cytoplasmic or cryptic and not accessible to extracellular trypsin. (iii) Suramin treatment of gp140-GFP 293T cells increased phosphorylation of gp140-GFP detected by immunoblotting with anti-FAK Tyr(P) 861 Ab that reacts with gp140 Tyr(P) 734 (Fig. 7C, lane 6). The anti-FAK Tyr(P) 861 Ab reacted only with the transfected cells and only with the gp140-GFP or p80-GFP bands. This confirms the specificity of the anti-FAK Tyr(P) 861 Ab for gp140/ p80. Suramin had no detectable effect on the levels of gp140-GFP protein expressed, confirming that suramin affected phosphorylation, not protein expression. Surprisingly, even without suramin, the gp140-GFP in 293T cells was phosphorylated at Tyr 734 (Fig. 7C, lane 3) at a basal level that was higher than the basal level observed for gp140 in HKs. This confirms that gp140 is phosphorylated in the absence of suramin or protease. In controls (results not shown), we also expressed nontagged gp140 in 293T cells. Nontagged gp140 also had a high basal phosphorylation detected with anti-FAK Tyr(P) 861 Ab. Conceivably, 293T cells may have higher levels of SFKs or lower PTPs than HKs. The cause of the elevated basal phosphorylation in the 293T cells is not resolved but does not alter the following conclusions. These findings confirm that the gp140/p80 cDNA encodes the same plasma membrane protein(s) that we originally identified based on tyrosine phosphorylation (4). Further, phosphorylation of gp140 and p80 in their tagged or native form detected by either anti-GFP or anti-FAK Tyr(P) 861 Ab responded similarly to changes in trypsin digestion and suramin treatment. These finding ensure that the trypsin and phosphorylation studies on gp140/p80 were specific for gp140/ p80 and that both gp140 and p80 are phosphorylated in the absence of suramin.
P-gp140/P-p80 and gp140-GFP Localize to the Plasma Membrane-gp140-GFP localized to the plasma membrane when expressed in mouse keratinocytes (Fig. 8A) or 293T cells (results not shown). Therefore, we determined whether immunostaining of HKs with Ab FAK Tyr(P) 861 could detect a subcellular localization for P-gp140/P-p80 and whether suramin and PP2 could regulate the subcellular localization. Adherent HKs were treated with or without suramin, fixed, and permeabilized and then stained with Ab FAK Tyr(P) 861 that reacts with gp140 phosphotyrosine 734 (Fig. 8B). Without suramin, there was minimal organized staining with Ab FAK Tyr(P) 861 in the apical-lateral focal planes of the adherent HKs (Fig. 8B, a). In controls (results not shown), Ab FAK Tyr(P) 861 stained the expected focal adhesions in the basal focal plane co-localized with anti-FAK Ab staining. However, in the presence of suramin, Ab FAK Tyr(P) 861 stained plasma membrane on lateral and apical plasma membranes of the adherent HKs above the basal focal plane and distinct from anti-FAK staining in focal adhesions (Fig. 8B, b). Phosphopeptide gp140 Tyr(P) 734 blocked the apical and lateral membrane staining (Fig. 8B, c). Suramin-induced staining at lateral membranes was inhibited by pretreatment of cells with PP2, consistent with phosphorylation of Tyr 734 being mediated by SFKs (Fig. 8B, d).
We conclude that phosphorylation of gp140/CDCP1 at tyrosine 734 in the plasma membrane of confluent HKs is in a state of dynamic equilibrium mediated by SFK and PTP activities generating a low net phosphorylation of gp140 unless suramin is added.
Proteolytic Conversion of gp140 to p80 Alters Tyr 734 Phosphorylation of p80 Distinct from gp140 -Below, we evaluated the consequences of proteolytic conversion of gp140 to p80 and the effect of the conversion on the phosphorylation of Tyr 734 of gp140 versus p80 in response to suramin or changes in adhesion. Further, we evaluated trypsin-like proteases that may mediate the conversion of gp140 to p80 in epidermal wounds. Phosphorylated p80 is detected in both trypsin-suspended and spreading cultures of HKs either with or without suramin treatment (Figs. 1 and 3). In contrast, P-gp140 is detected in HKs only after treatment with suramin or inhibitors of PTPs (Fig. 3 or 5C). In possible explanation of the distinction, it was determined whether P-p80 and P-gp140 have different rates of dephosphorylation by PTPs. Tryspin rounding of HKs converts gp140 to p80 (Figs. 3 and 6). Adherent cells expressing gp140 were treated with and without trypsin, yielding p80 and gp140, respectively. These HKs were treated with a suramin pulse (20 min) to generate P-gp140 (Fig. 9A, Adherent) and P-p80 (Fig.  9A, Trypsin-rounded), washed to remove suramin, and then chased for the indicated times (Fig. 9A). Within 1 h after the removal of suramin, P-gp140 was not detectable. In results not shown, dephosphorylation was complete within 15 min of the removal of suramin. Retreatment of the adherent HKs with suramin restored the P-gp140 (Fig. 9A, ϩSuramin) and demonstrated that gp140 was not catabolized. As seen in Fig. 5C,   FIG. 7. Expression of P-gp140-GFP in 293T cells. 293T cells were transiently transfected with nothing (lane 1), GFP cDNA (lane 2), or gp140-GFP cDNA (lane 3) and then cultured in the absence of serum. Triton X-100 extracts (buffer A) were prepared from the cells, fractionated by SDS-PAGE, and then detected by protein stain (A) or immunoblotted with anti-GFP Ab JL-8 (B) and anti-FAK Tyr(P) 861 Ab (C) that cross-reacts with gp140 Tyr(P) 734 . Where indicated, the 293T cells were treated before extraction with suramin (35 M for 10 min), trypsin for detachment, or mechanical agitation for detachment. gp140-GFP is detected as a single 165-kDa band by both anti-GFP and anti-FAK Tyr(P) 861 Abs and only in cells transfected with gp140-GFP. Tryspin converts gp140-GFP to p80-GFP, detaches cells, and dephosphorylates noncleaved gp140-GFP but does not dephosphorylate p80-GFP. Suramin increases phosphorylation of gp140-GFP above base-line phosphorylation. Mechanical detachment by shaking resulted in the dephosphorylation of gp140-GFP.
inhibitors of PTPs were able to mimic suramin in elevating phosphorylation of gp140 and consistently were able to inhibit dephosphorylation after the suramin washout (results not shown). These results further suggested that the dephosphorylation of P-gp140 was due to a PTP and not catabolism of gp140 protein. In comparison with gp140, phosphorylation of p80 was prolonged, and P-p80 was still detectable 4 h after the removal of suramin (Fig. 9A, Trypsin-rounded). This difference between gp140 and p80 may explain the relative abundance of P-p80 upon deadhesion or in readherent cells (Fig. 1).
Based on these findings, we determined whether phosphorylation of gp140 and p80 was distinguishable based on response to suramin. Suramin concentration curves were used to evaluate phosphorylation of gp140 and p80. Phosphorylation of gp140 was half-maximal at 1.5 M suramin compared with 40 M for p80 (Fig. 9B). Thus, phosphorylated p80 has a longer half-life in culture than gp140, and phosphorylation of p80 is less sensitive to suramin than gp140, clearly distinguishing the two proteins. Further, it suggests that proteolytic conversion of gp140 to p80 in the extracellular domain alters phosphorylation of the cytoplasmic tail.
Adhesion-dependent Phosphorylation of gp140 Is Distinct from p80 -We determined whether adhesion also regulates phosphorylation of gp140 differently than phosphorylation of p80. Evaluating this difference is complicated by the fact that trypsin used to suspend cells also converts gp140 to p80. To overcome this problem, we used the following approaches. First, we found that suspension of HKs with EDTA failed to induce phosphorylation of gp140, whereas suramin treatment of the EDTA-suspended cells still induced phosphorylation of gp140 (data not shown). In contrast, trypsin-detachment of HKs that convert gp140 to p80 increased phosphorylation of p80 (Fig. 1,  lane 2, and Fig. 3, lane 5). Apparently, detachment affected phosphorylation of p80 differently than gp140. Second, we determined whether deadhesion alters phosphorylation of gp140-GFP distinct from p80-GFP. Trypsin detachment of 293T cells converts gp140-GFP to p80-GFP, and the p80-GFP remained phosphorylated in the detached cells (Fig. 7, B and C, lane 9). Surprisingly, the gp140-GFP that was resistant to trypsin was not phosphorylated. This raised the possibility that the trypsin detachment dephosphorylated gp140 but not p80. This possibility was evaluated as follows. The 293T are easily detached from the substratum by agita- FIG. 8. Localization of gp140 in keratinocytes. A, mouse keratinocytes were transfected with a gp140-GFP cDNA, and then gp140-GFP protein was detected in the plasma membrane in live cells. B, HKs grown on glass coverslips were incubated with or without suramin (35 M, 20 min), fixed (2% formaldehyde in cacodylate buffer containing vanadate and NEM), permeabilized (0.1% Triton X-100 for 1 min), and then stained with Ab FAK Tyr(P) 861 . All photos were taken in a focal plane vertically midway through the cells. a, in the absence of suramin, staining was localized to focal adhesions (not visible) determined by double staining with anti-FAK antibody. b, in the presence of suramin, staining was elevated in the apical and lateral areas of the plasma membrane and not co-distributed with FAK in the basal focal plane. c, treatment of HKs with phosphopeptide gp140 Tyr(P) 734 before (Ͼ) reaction with Ab FAK Tyr(P) 861 blocked staining of apical/lateral areas. d, treatment of HKs with PP2 to inhibit SFKs before (Ͼ) the addition of suramin inhibited the apical/lateral plasma membrane staining.

FIG. 9. Phosphorylation of Tyr 734 is distinct in gp140 and p80.
A, dephosphorylation of gp140 and p80 by PTP after removal of suramin. HKs were grown for 3 days to generate predominantly gp140 or trypsin-rounded to generate p80. The adherent or rounded cells were treated with a pulse of suramin (20 min, 35 M), washed to remove suramin, and cultured for the indicated times. In controls, identical cultures of the adherent HKs were treated similarly, except that suramin was also added during the final 15 min prior to extraction (ϩSuramin) to show the presence of gp140 protein. Extracts were prepared with buffer A, fractionated by SDS-PAGE, and Western blotted with Ab FAK Tyr(P) 861 . P-gp140 was dephosphorylated in less than 1 h, whereas P-p80 was dephosphorylated between 4 and 6 h. P-FAK detected with Ab FAK Tyr(P) 861 was unaffected. Thus, P-gp140 and P-p80 are dephosphorylated by a PTP at different rates, explaining the relative ease of detection of P-p80 in cultured HKs. B, concentration of suramin to promote phosphorylation of gp140 is distinct from p80. As indicated, adherent HKs were treated with various concentrations of suramin for 20 min prior to extraction, SDS-PAGE, transfer to nitrocellulose, and analysis of P-gp140 and P-p80 levels by immunoblotting with the FAK Tyr(P) 861 Ab. Band intensity was quantified using LiCor Odyssey software. p80 and gp140 exhibited distinct sensitivities to suramin concentration. The half-maximal stimulation of phosphorylation (SC50) occurred at 1.5 M (gp140) and 40 M (p80). tion without trypsin digestion or EDTA. Detachment of the 293T cells with agitation dephosphorylated gp140-GFP (Fig. 7B, lane 12, compared with Fig. 7C, lane 12). We conclude that phosphorylation of gp140 and p80 are both altered by adhesion, but they are regulated differently; detachment dephosphorylates gp140 and phosphorylates p80. Proteolytic conversion of gp140 to p80 distinguishes phosphorylation of p80 from phosphorylation of gp140. After conversion, both suramin-and adhesion-dependent phosphorylation of p80 is distinct from gp140.
Conversion of gp140 to p80 by Plasmin-Here, we evaluated possible trypsin-like proteases that might convert gp140 to p80 as a physiological event in wounds. Wounding of epidermis exposes leading keratinocytes to trypsin-like proteases in tissue and serum that regulate coagulation, fibrinolysis, and tissue remodeling and might convert gp140 to p80. Therefore, we determined whether exogenous plasmin, a trypsin-like protease in serum that contacts leading keratinocytes in wounds, could convert gp140 to p80. Digestion of gp140 in HKs (Fig.  10A) or gp140-GFP in 293T cells (Fig. 10B) with purified plasmin was sufficient to convert gp140 to p80. A plasmin titration curve indicated that 1 g plasmin/ml on adherent HKs for 1-3 h (plasmin/substrate ratio ϭ 1:1000) was sufficient to decrease gp140 and increase p80. This level of plasmin is well below the concentration of plasminogen in plasma (120 g/ml) (21). Thrombin, a trypsin-like protease involved in clot formation, failed to convert gp140 to p80 under conditions where it converts precursor laminin 5 to mature laminin 5 (results not shown). Digestion with plasmin had no detectable effect on cell-cell or cell-substrate adhesion in contrast to trypsin that rounds and detaches cells. Consistently, plasmin conversion, like trypsin (Fig. 9A), prolonged phosphorylation of p80 relative to gp140 (data not shown).
Finally, it was determined whether endogenous plasmin or other trypsin-like protease in epidermis is sufficient to convert gp140 to p80. Epidermis was collected from euthanized, newborn mice and then treated with suramin to promote phosphorylation of gp140 and p80. For these studies, neonatal mice with null defects in laminin 5 were used, allowing the removal of the epidermis without exogenous protease. Extracts from the epidermis were immunoblotted with anti-FAK Tyr(P) 861 Ab. Both P-gp140 and P-p80 were detected in the epidermal extract (Fig. 10C). Thus, exogenous plasmin converts gp140 to p80. Similarly, an endogenous plasmin-like protease in epidermis converts gp140 to p80, suggesting that gp140 may be decreased in epithelial wounds in vivo.

DISCUSSION
In cultures of epidermal, prostate, cervical, and esophageal epithelial cells, phosphorylation of p80 is increased by cell detachment, whereas reattachment to the endogenous substrate, laminin 5, dephosphorylates P-p80 (Fig. 1). 2 Here, we purified P-p80 from trypsin-suspended HKs and P-gp140 from suramin-treated adherent HKs and expressed a recombinant gp140-GFP protein. Conclusions from the findings under "Results" are integrated in a diagram in Fig. 11 and are as follows. (i) gp140 is a transmembrane glycoprotein and the trypsinsensitive precursor to p80 (Figs. 1, 3, and 7). gp140 and p80 contain three extracellular CUB protein-protein interaction domains of unknown function (Fig. 2). (ii) gp140 is the product of the CDCP1 gene that is overexpressed in human colorectal and lung cancers (9). (iii) Plasmin in serum that contacts leading cells in wounds converts gp140 to p80 (Fig. 10). (iv) Proteolytic conversion increases phosphorylation of p80 relative to gp140 in response to extracellular signals from adhesion or suramin (Figs. 7 and 9). (v) Tyrosine phosphorylation of gp140 on Tyr 734 (Fig. 6) and possibly other tyrosine residues is mediated by SFK(s) (Fig. 5). Phosphorylation of gp140 is balanced by dephosphorylation by an unidentified PTP(s), yielding low equilibrium phosphorylation of gp140 (Fig. 5). (vi) Extracellular suramin, plasmin, or adhesion regulates phosphorylation of gp140 and p80 by perturbing the balance of SFK and PTPs (Figs. 9 and 11). We conclude that plasmin conversion of gp140 to p80 alters the phosphorylation balance mediated by SFKs and PTPs. Conversion prolongs phosphorylation of p80 relative to gp140 in response to adhesion or suramin. We suggest that plasmin cleavage and adhesion provide transmembrane sig-  10. Plasmin converts gp140 to p80. A, exogenous plasmin converts gp140 to p80 in HKs. HKs, grown in serum-free, defined medium, were incubated with (ϩ) or without (Ϫ) exogenous purified plasmin (1 g of plasmin/ml/mg of cell protein) and/or EDTA (5 mM) for 3 h. The cells were treated with suramin (35 M, 20 min) and then extracted with 1% Triton X-100 (Buffer A). These soluble Triton extracts (75 g of protein of each) were immunoblotted with anti-FAK Tyr(P) 861 Ab. Plasmin converts gp140 to p80, whereas EDTA added to round the cells had no detectable effect on the cleavage. B, exogenous plasmin converts gp140-GFP to p80-GFP in 293T cells. 293T cells expressing gp140-GFP cDNA as described under "Experimental Procedures" and in the Fig. 7 legend were digested with purified plasmin (1 g of plasmin/ml/mg of cell protein) for 1 or 3 h as indicated and then extracted with 1% Triton X-100 detergent (buffer A). The soluble Triton extracts (75 g of protein each) were immunoblotted with anti-GFP Ab. Plasmin converted gp140-GFP to p80-GFP. C, an endogenous protease converts gp140 to p80 in epidermis. Epidermis from a neonatal mouse was treated with suramin (35 M, 20 min) and then homogenized in 1% Triton X-100 in buffer A. The soluble Triton extract (75 g of protein) was immunoblotted with anti-FAK Tyr(P) 861 Ab. Both P-gp140 and P-p80 bands were detected, indicating the presence of endogenous plasmin or trypsin-like protease that cleaves gp140 to p80. The epidermis was obtained from a neonatal mouse with homozygous null mutations in the LAMA3 gene encoding the ␣ 3 chain of laminin 5. This laminin 5 defect generates neonatal lethal blistering, allowing for easy separation of the epidermis from dermis without use of exogenous protease (26). nals that regulate phosphorylation of gp140 distinct from p80 in epithelial wounds.
gp140/p80 is encoded by the CDCP1 gene, which is elevated in human lung and colon tumors. During the preparation of this manuscript, a novel cell surface glycoprotein, SIMA135, was reported as the product of the CDCP1 gene and a marker for metastasis (11). P-p80 was originally identified in a screen for adhesion-dependent phosphorylation in an epidermal wound model (4). Chronic wound activation associated with inherited blistering diseases is a cancer promoter (25). It will be interesting to determine whether gp140/p80 phosphorylation or expression is altered in mice with adhesion defects generating chronic wounds (26).
Physiological Plasmin Converts gp140 to p80 -Extracts of epidermis prepared without exogenous protease express both P-gp140 and P-p80 (Fig. 10C). This indicates that proteolytic conversion of gp140 to p80 occurs in tissue via an endogenous protease. As a source of epidermis, we utilized mice with adhesion defects in laminin 5, allowing for separation of the epidermis without exogenous protease. This raises the interesting possibility that conversion of gp140 to p80 is elevated in laminin 5 defects relative to wild type epidermis. Plasmin, a trypsin-like protease in serum, converts gp140 to p80 (Fig. 10, A and B). Together, these results suggest that serum plasmin contacts leading keratinocytes in wounds and may convert gp140 to p80.
Studies are under way to identify the location of the trypsin and plasmin cleavage site(s) in gp140. Conversion of gp140 to p80 leaves most of the first CUB domain intact along with the complete second and third CUB domains, the transmembrane, and the cytoplasmic tyrosine phosphorylation sites. Sequencing of tryptic peptides from purified p80 identified peptide 65 (residues 278 -296 of gp140) as the most amino-terminal of the five identified p80 peptides (Fig. 2). When trypsin cleaves gp140, the resulting p80 contains peptide 65. Reasonably, trypsin cleaves gp140 at Lys/Arg residues 261, 276, or 277 on the amino-terminal side of peptide 65. The extended optimal substrate specificities for plasmin have been identified as follows: P4-Lys/norleucine/ Val/Ile/Phe, P3-Xaa, P2-Tyr/Phe/Trp, P1-Lys/Arg (27). Candidate sequences in gp140 that are optimal for plasmin cleavage include EVFK296 in peptide 65. If correct, trypsin and plasmin cleave gp140 at similar but not identical sites.
Tyrosine Phosphorylation of gp140 Is Balanced by SFKs and PTPs-Membrane-impermeable suramin or trypan blue increases phosphorylation of p80 and gp140 (Fig. 3A) possibly by outside-in signaling through an unidentified cell surface component(s) (Fig. 11). Cell detachment via trypsin phosphorylates p80, whereas suramin phosphorylates both gp140 and p80 at Tyr 734 by increasing SFK activity and/or inhibiting PTP activity. Consistently, inhibition of PTPs with vanadate also elevates phosphorylation of gp140 and p80. However, prior inhibition of SFKs with SU6656 prevents the effects of vanadate, suramin, or detachment, indicating that the active kinase is also required for the elevation in phosphorylation. This indicates that there is a dynamic balance of active SFK, and active PTP activity yields low equilibrium phosphorylation of gp140 in adherent HKs and mouse keratinocytes. The dynamic phosphorylation is unbalanced by conversion of gp140 to p80 or cell detachment or suramin that increases phosphorylation of gp140 and p80.
Suramin is a membrane-impermeable polysulfonated napthylurea (12) and heparin analogue. Suramin has previously been reported to induce tyrosine phosphorylation of similar 80 and 140-kDa proteins, but these proteins were not identified (28). Suramin has diverse effects including inhibiting melanoma invasion (29) and increasing colony formation of keratinocytes (30), morphogenic activity in tumor invasion (31), angiogenesis (32), and cell migration on collagen and laminin (33). The effects of suramin on cell motility are frequently attributed to inhibitory effects on growth factor signaling (30). However, suramin also inhibits PTPs (34), heparanase (29), and urokinase plasminogen activator (uPA) binding to its receptor (35) and activates protein kinase C (36). Extracellular suramin promotes maximal phosphorylation of gp140 and p80 at concentrations that are insufficient to inhibit PTPs in cell homogenates (see "Results"). As has been suggested (28), events other than growth factor signaling, including phosphorylation of gp140/CDCP1, may explain some effects of suramin.
Outside-in Signals Differentially Regulate Phosphorylation of gp140 and p80 -Trypsin rounds and then detaches cells in culture and removes the amino terminus of gp140, converting it to p80. Concomitantly, trypsin increases phosphorylation at Tyr 734 and other cytoplasmic tyrosine residues of p80, relative to gp140 (Fig. 9). The trypsin conversion of gp140 to p80 and/or subsequent cell detachment appear to contribute to the relative increase in phosphorylation of p80. First, trypsin conversion associated with cell rounding prolongs phosphorylation of p80 relative to P-gp140 (Fig. 9A). Plasmin conversion of gp140 to p80 does not detectably round the cells but still prolongs phosphorylation of p80. This suggests that conversion, not rounding, prolongs phosphorylation. The prolonged phosphorylation FIG. 11. Plasmin conversion of gp140 to p80 prolongs phosphorylation of p80 in response to suramin or changes in adhesion.
Results are summarized in a model that integrates outside-in signals from plasmin cleavage, substrate adhesion, and suramin that differentially regulate phosphorylation of gp140/CDCP1 and p80 in keratinocytes. gp140/CDCP1 is a transmembrane glycoprotein containing three extracellular CUB domains. Phosphorylation at cytoplasmic Tyr 734 of gp140 is dynamically balanced by SFK and PTP activities, yielding low equilibrium phosphorylation. Plasmin or trypsin converts gp140 to p80 by removal of an extracellular fragment(s). After conversion, signals from extracellular suramin or substrate extracellular matrix through uncharacterized surface receptors (?) unbalance SFKs and PTPs (arrows) to prolong phosphorylation (Pi) of p80 relative to gp140.