Endo180 Binds to the C-terminal Region of Type I Collagen*

Type I collagen is a fibril-forming heterotrimer composed of two α1 and one α2 chains and plays a crucial role in cell-matrix adhesion and cell differentiation. Through a comprehensive differential display screening of oncogenic ras target genes, we have shown that the α1 chain of type I collagen (col1a1) is markedly down-regulated by the ras oncogene through the mitogen-activated protein kinase pathway. Although ras-transformed cells are no longer able to produce and secrete endogenous collagen, they can still adhere to exogenous collagen, suggesting that the cells express a collagen binding factor(s) on the cell surface. When the region of col1a1 encompassing the C-terminal glycine repeat and C-prodomain (amino acids 1000–1453) was affinity-labeled with human placental alkaline phosphatase, the secreted trimeric fusion protein could bind to the surface of Ras-transformed cells. Using biochemical purification followed by matrix-assisted laser desorption/ionization mass spectrometry analysis, we identified this collagen binding factor as Endo180 (uPARAP, CD280), a member of the mannose receptor family. Ectopic expression of Endo180 in CosE5 cells followed by in situ staining and quantitative binding assays confirmed that Endo180 indeed recognizes and binds to placental alkaline phosphatase. The interaction between Endo180 and the C-terminal region of type I collagen appears to play an important role in cell-matrix adhesion.

The ras proto-oncogenes are important mediators of cell growth and differentiation and are frequently mutated in human cancer. Approximately 90% of pancreatic and 50% of colorectal cancers have been shown to carry an oncogenic mutation in one of the ras family members (1). These proto-oncogenes are guanine nucleotide-binding proteins that cycle between an inactive GDP-bound and an active GTP-bound state. When activated, Ras proteins function by transducing a number of extracellular signals into the nucleus of a cell via downstream effectors and signaling cascades, subsequently activating transcription factors, such as AP-1 and Ets (2,3), that control the expression of various genes (4). Oncogenic mutations in the Ras proteins lock them in the active GTP-bound state (5), resulting in constitutive Ras activation and thus permanent stimulation of the numerous downstream effectors and signaling pathways that regulate various transcription factors. These transcription factors, in turn, control the expression of a number of genes that are responsible for such cellular changes as increased cell proliferation, migration, and invasion (6). Oncogenic Ras also regulates the expression of genes that promote various phenotypic changes associated with morphological transformation in culture, although the mechanisms involved in this process remain obscure (7).
Many of the genes whose expression is controlled by the ras oncogenes are yet to be elucidated, although some target genes, such as transin/stromelysin-1 (8), MMP-1 (9,10), PAI-2 (11), mob-1/IP10 (12,13), and IL-24 (14) have been identified. However, these genes fail to account for the complex phenotype of cell transformation. Ras-transformed fibroblast cells, compared with parental fibroblast cells, appear very disorganized as a result of both altered cell-matrix interactions and lost contact inhibition, which is one of the hallmarks of cell transformation (15,16). In addition, oncogenic ras-mediated transformation results in increased cellular proliferation and enables transformed cells to form colonies in soft agar and tumors in mice (17,18). Identifying and characterizing the biological functions of additional ras target genes should help to shed light on our understanding of how the ras oncogenes regulate cell transformation and tumorigenesis.
In a search for ras target genes that mediate the complex phenotype of transformation, we completed a comprehensive differential display (DD) 1 screen (19) to compare the gene expression profiles between nontransformed rat embryonic fibroblast cells and a derivative of the same cells that were constitutively transformed by oncogenic h-ras (14). Here we show that the expression of col1a1 was essentially abrogated by oncogenic ras through the MAP kinase pathway. However, Ras-transformed cells still retained the ability to bind and respond to exogenous type I collagen. Using a defined sequence of soluble col1a1 that spanned the C-terminal region of the protein, we have biochemically identified this collagen-specific binding factor (CBF) as Endo180 (uPARAP, CD280) on the surface of parental and Ras-transformed rodent embryonic fi-broblast cells. We showed that the interaction of Endo180 with the C-terminal region of type I collagen plays a prominent role in cell-matrix adhesion.

MATERIALS AND METHODS
Cell Lines and Culture-Rat-1 embryonic fibroblast cells and the Rat-1(Ras) h-ras-transformed derivative have been previously described (12). Rat-1:iRas cells, which are a derivative of the Rat-1 cells, were grown in the presence of 200 g/ml Hygromycin B (Roche Applied Science). When these cells reached 80% confluence, they were treated with 5 mM isopropyl-␤-D-thioglucoside (IPTG; Invitrogen) to induce expression of oncogenic h-ras, essentially as described (20). RIE rat intestinal epithelial cells are spontaneously immortalized, diploid, nontransformed cells and were obtained from Dr. R. Coffey. RIE(h-Ras) and RIE(k-Ras) derivatives were established by stably transfecting RIE cells with mutant h-ras and k-ras (12V), respectively (21). CosE5 cells, developed from clonally purified COS1 cells, and 293T cells were obtained from GenHunter Corp (Nashville, TN). All of the above cell lines were routinely maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% bovine calf serum (HyClone, Logan, UT) and 1% penicillin/streptomycin (Invitrogen) at 37°C with 10% CO 2 . Phoenix (amphoteric) packaging cells were provided by Dr. A. Reynolds and were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Atlanta Biologicals, Atlanta, GA) and 1% penicillin/streptomycin.
For adhesion assays, tissue culture plates (Falcon, Franklin Lakes, NJ) were coated with rat tail type I collagen (Becton Dickinson Labware, Bedford, MA) using the thin coating procedure recommended by the manufacturer. Briefly, type I collagen was diluted to 50 g/ml in 0.02 M acetic acid (EMD Chemicals) and applied to the tissue culture plates for 1 h at room temperature or overnight at 4°C. The plates then were washed with phosphate-buffered saline (PBS), and cells were grown on the plates until they reached confluence.
Differential Display-DD was carried out essentially as previously described (19). Total RNA was purified from Rat-1 and Rat-1(Ras) cells that had been grown in the absence or presence of 50 M PD98059 (Calbiochem) for 24 h, using the RNApure Reagent (GenHunter), following the manufacturer's instructions. DNase I treatment of RNA prior to DD was carried out using the MessageClean kit (GenHunter Corp.). DD-PCR were completed using 160 combinations of one-base anchored oligo(dT) primers and rationally designed arbitrary 13-mers from the RNAimage kit (GenHunter Corp.), representing about 60% gene coverage (22). cDNAs that were generated from the 3Ј termini of the mRNAs were separated on a 6% denaturing polyacrylamide gel. Bands representing cDNAs of interest were excised from the gel, cloned into the PCR-TRAP cloning vector (GenHunter Corp.), and sequenced using the Sequenase kit from U.S. Biochemical Corp.
Northern Blotting-Northern blot analysis was completed as previously described (23). In summary, 10 g of total RNA was denatured, separated on a 1% agarose gel with 3% formaldehyde, transferred to a nylon membrane (Amersham Biosciences), and hybridized using either a 208-bp cDNA probe of col1a1 obtained by differential display or a 5.5-kb cDNA probe of Endo180 obtained from IMAGE clone 7099211. Both probes were radioactively labeled with [␣-32 P]dATP by random priming with the HotPrime DNA labeling kit (GenHunter) according to the manufacturer's instructions. After hybridization, the membranes were washed, and bands were visualized by autoradiography at Ϫ80°C.
Development of a Secreted Human Placental Alkaline Phosphatase (AP) Collagen Fusion Construct-A 1.5-kb cDNA fragment encoding the C-terminal glycine repeat region and prodomain (amino acids 1000 -1453) of rat col1a1 (accession number Z78279) was amplified by PCR using cloned ␣1(I) collagen cDNA as a template (PCR primers 5Ј-AGATCTGGTCGTGAGGGATCCCC-3Ј and 5Ј-TCTAGAGTTGGGG-GAAAGTGGGC-3Ј). The fragment was inserted into the BglII and XbaI sites of the APtag4 vector (GenHunter) to allow for in-frame fusion to the C terminus of secreted human placental AP (24).
Cell Transfection-The APtag4 vector and AP-Coll recombinant plasmid expressing secreted AP alone and AP-Coll, respectively, were stably co-transfected with the pBabe-puro plasmid (GenHunter) into 293T cells using FuGENE 6 (Roche Applied Science), according to the manufacturer's instructions. Cells were selected in 10 g/ml puromycin (Sigma) to obtain stable clones that were clonally purified. Conditioned medium containing AP and AP-Coll was collected for binding assays.
CosE5 cells were seeded at a density of 1 ϫ 10 5 cells in 6-well plates. Twenty-four hours later, the cells were transiently transfected with either pcDNA3 vector alone or the pcDNA3-Endo180 expression vector, which has been previously described (25). Forty-eight hours after trans-fection, quantitative or in situ receptor binding assays (described below) were carried out in duplicate.
Quantitative and in Situ AP and AP-Coll Binding Assays-Quantitative and in situ receptor binding assays were completed as previously described (14), using the secreted AP and AP-Coll conditioned medium produced from stably transfected 293T cells. Briefly, cells were seeded at a density of 1 ϫ 10 5 cells/well on a 6-well tissue culture plate and grown until ϳ80% confluent. Quantitative or in situ receptor binding assays then were performed in duplicate by incubating the cells with 1 ml/well of either secreted AP or AP-Coll conditioned medium containing equivalent AP activity units for 90 min at 37°C. AP assay reagent A and AP assay reagent S (GenHunter) were used to measure the AP binding activity for the quantitative and in situ receptor binding assays, respectively.
Slot Blot and Ligand Affinity Binding-Slot blot binding assays were performed on Rat-1 and Rat-1(Ras) whole cell extract (WCE) that had been prepared in cell lysis buffer A (10 mM Tris-Cl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride). Briefly, the samples were either heated for 5 min at 100°C or were not heated and then were applied to a polyvinylidene difluoride (PVDF; Millipore Corp., Billerica, MA) membrane via a slot blot apparatus. The membrane was probed with secreted AP or AP-Coll and stained with AP substrate S.
Binding assays also were performed by transferring protein extract to a PVDF membrane after subjecting the extract to nonreducing SDS-PAGE (National Diagnostics, Atlanta, GA). The membrane then was probed with secreted AP or AP-Coll, followed by either staining with AP substrate S to visualize the AP binding activity or additionally probing the membrane first with anti-AP monoclonal antibody (Seradyn, Ramsey, MN) and then with anti-mouse IgG conjugated to horseradish peroxidase (Amersham Biosciences) and visualizing the proteins with an ECL kit (Amersham Biosciences), following the manufacturer's protocol.
FACS Analysis-To perform FACS analysis, AP and AP-Coll conditioned medium was concentrated to 15 g/ml using an Amicon Ultra Centrifugal Filter device with a molecular mass cut-off of 30 kDa (Millipore). Rat-1 and Rat-1(Ras) cells were washed with PBS and detached from tissue culture plates by exposure to 0.25% trypsin/EDTA (Invitrogen) for 1 min at room temperature, and 3 ϫ 10 5 cells were incubated for 90 min at room temperature with 15 g/ml secreted AP or AP-Coll. Cells then were washed with PBS and probed with 20 g/ml mouse anti-AP monoclonal antibody for 1 h, followed by fluorescein isothiocyanate-conjugated anti-mouse Ig antibody for another 1 h. Flow cytometry analysis then was performed using a FACScan flow cytometer (BD Biosciences). This assay was performed twice in triplicate.
Competition Adhesion Assays-Rat-1 and Rat-1(Ras) cells were detached from tissue culture plates with 0.25% trypsin/EDTA and washed in PBS, and 1 ϫ 10 5 cells/sample were incubated with 293T-conditioned medium alone, 5 g/ml secreted AP, or 5 g/ml secreted AP-Coll for 90 min at room temperature with gentle agitation. The cells then were inoculated onto 96-well tissue culture plates that had been precoated with 50 g/ml of rat tail type I collagen, 5 g/ml fibronectin (Sigma), or 100 g/ml matrigel (BD Biosciences) in PBS overnight at 4°C. After 90 min, unattached cells were removed from the wells, and adherent cells were washed three times with PBS, fixed with 4% paraformaldehyde in PBS for 20 min, and then stained with crystal violet (EM Science, Gibbstown, NJ). Attached cells were lysed with 10 mM Tris-Cl, pH 8.0, 1% Triton X-100, and the absorbance was read at A 595 .
Biochemical Purification of the AP-Coll-specific Binding Protein-All steps of the following procedure were performed on ice. Rat-1 and Rat-1(Ras) membrane protein extract (MPE) was prepared by washing the cells with PBS and lysing the cells with lysis buffer B (15 mM HEPES, 5 mM EDTA, 5 mM EGTA, and 2 mM phenylmethylsulfonyl fluoride). Cells were scraped from the tissue culture plates, and the solution was pulled up and down through a 21-gauge needle syringe five times to fragment the cells. A quick spin at low speed was completed to remove large cellular debris, and the remaining supernatant was layered onto a 60% sucrose solution and subjected to ultracentrifugation at 35,000 ϫ g for 1 h at 4°C. Solubilized MPE was resuspended in cell lysis buffer B containing 0.1% Triton X-100 using a 26-gauge needle syringe, incubated with agitation at 4°C for 30 min, and centrifuged at high speed for 2 min. The supernatant then was used in fast performance liquid chromatography to isolate the CBF.
Rat-1 MPE was applied to a ResourceQ anion exchange column (Amersham Biosciences) that had been pre-equilibrated with 5 column volumes of lysis buffer B plus 0.1% Triton X-100. Unbound proteins were eluted with the same buffer, whereas bound proteins were eluted as 0.5-ml fractions by a 20-ml linear salt gradient from 0.0 to 0.5 M of NaCl with a flow rate of 1 ml/min. Eluate fractions were applied to a PVDF membrane via a slot blot apparatus, as previously described here, and the membrane was probed with secreted AP or AP-Coll conditioned medium. The fraction containing the highest level of AP-Coll-specific binding activity as well as WCE and MPE was applied to a nonreducing SDS-polyacrylamide gel and transferred to a PVDF membrane. The membrane was probed with secreted AP or AP-Coll and stained with AP substrate S. An AP-Coll-specific binding protein was identified at ϳ180 kDa in all three lanes. Alternatively, the nonreducing SDS-polyacrylamide gel was stained with Colloidal Blue (Invitrogen) for 4 h at room temperature and destained overnight. A band with the same electrophoretic motility as the AP-Coll-specific binding protein was found in the fraction of interest.
In-gel Protein Digestion and Mass Spectrometry of Peptides-Proteins from the fraction of interest were separated by SDS-PAGE under nonreducing conditions and visualized using the Colloidal Blue staining kit. An ϳ180-kDa band was excised from the gel, cut into cubes, equilibrated in 50 mM NH 4 HCO 3 , reduced with dithiothreitol (3 mM in 50 mM NH 4 HCO 3 , 37°C for 15 min), and alkylated with iodoacetamide (6 mM in 50 mM NH 4 HCO 3 for 15 min). The gel cubes were then dehydrated with acetonitrile and rehydrated with 15 l of 20 mM NH 4 HCO 3 containing 0.01 g/l modified trypsin (Promega). Trypsin digestion was carried out for ϳ2 h at 37°C. Peptides were extracted with 60% acetonitrile and 0.1% trifluoroacetic acid, dried by vacuum centrifugation, and reconstituted in 5 l of 60% acetonitrile with 0.1% trifluoroacetic acid. A small aliquot (0.4 l) was applied to a target plate and overlaid with 0.4 l of a ␣-cyano-4-hydroxycinnamic acid matrix. Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry and TOF/TOF tandem mass spectrometry was carried out using a Voyager 4700 mass spectrometer (Applied Biosystems, Foster City, CA) operated in reflectron mode.
For intact peptide mass analysis, the mass spectra were calibrated to within 50 ppm using trypsin autolytic peptides present in the sample (m/z ϭ 842.51 and 2211.09 daltons). Peptide ion masses (M ϩ H) and MS/MS fragmentation data on specific ions were collectively used for data base interrogation against the NCBInr data base using GPS Explorer software (Applied Biosystems) running the MASCOT algorithm. This search generated an unambiguous match to gi͉5174485 human mannose receptor, C type 2, also known as gi͉4835878 human endocytic receptor Endo180 (Mowse score ϭ 97 with partial oxidation of methionine and complete carbamidomethylation of cysteine ) validated the predicted amino acid sequences. A rat homolog was not present in the NCBInr data base, which most likely explains ion signals present in the MALDI-TOF MS (and corresponding MS/MS fragmentation spectra) that do not yield significant matches (data not shown).
Western Blot Analysis-AP and AP-Coll conditioned medium and WCE, cytoplasmic protein extract, and MPE from various cells were used in immunoblotting assays. Protein concentrations were determined using the Bio-Rad protein assay dye reagent (Bio-Rad). Aliquots (20 -50 g) were separated by electrophoresis on SDS-polyacrylamide gels under reducing (ϩ␤-mercaptoethanol) or nonreducing (Ϫ␤-mercaptoethanol) conditions. For Western blot analysis, proteins were blotted onto a PVDF membrane, blocked with 5% nonfat milk in PBS containing 0.05% Tween 20 (J. T. Baker Inc.), and probed with primary antibody for 2 h at room temperature or overnight at 4°C. The following primary antibodies were used in this study: anti-Ras antibody (Oncogene Science, Long Island, NY), AP-monoclonal antibody (Seradyn) and AP-polyclonal antibody (GenHunter), Endo180 monoclonal antibody A5/158 (25) and polyclonal anti-Endo180 antiserum (26) that have been described previously, and actin antibody A2066 (Sigma). Blots then were incubated for at least 1 h at room temperature with the following secondary antibodies: anti-rabbit IgG or anti-mouse IgG conjugated to horseradish peroxidase Amersham Biosciences). Reactive proteins were visualized with an ECL kit following the manufacturer's protocol.
RNA Interference-The small interfering RNA (siRNA) vector pSu-perRetro was a gift from Dr. R. Agami. The pSuperRetro-Endo180 plasmid was designed according to Brummelkamp et al. (28,29). Briefly, to establish pSuperRetro-Endo180, the following sequences from rodent Endo180 cDNA were subcloned into pSuperRetro that had been digested with BlgII and HindIII: oligonucleotides 5Ј-CCC-GATGTCTTCCTCATCT-3Ј and 5Ј-CGTGCAAACAATGCATCGA-3Ј. The constructs were confirmed by sequencing (GenHunter). Retroviral infection was performed by transfecting the recombinant plasmids into Phoenix packaging cells using FuGENE 6, following the manufacturer's recommended procedure. Forty-eight hours after transfection, the medium containing the retrovirus was collected, filtered, treated with polybrene (4 g/ml), and transferred to Rat-1 and Rat-1(Ras) target cells. Infected cells were selected with puromycin (10 g/ml), and clones were isolated using a cloning ring. Quantitative and in situ cell surface staining assays then were performed with the clones, as described previously.
cDNA Cloning and Sequencing-The full-length nucleotide sequence of murine Endo180 was used to BLAST search for rat sequences in the NCBI rat expressed sequence tag data base. One matching expressed sequence tag clone was purchased from ATCC (Manassas, VA) (IMAGE number 7099211), and the entire coding region was sequenced by primer walking (GenHunter). The protein sequences of human, mouse, and rat Endo180 were aligned with MultiAlign.

Down-regulation of col1a1
Collagen by the ras Oncogenes through the MAP Kinase Pathway-To identify genes controlled by the ras oncogenes, we performed DD to compare the gene expression profiles of Rat-1 embryonic fibroblast cells and the Rat-1(Ras) h-ras-transformed derivative (12). The cells were grown in the absence or presence of PD98059, an inhibitor of MAP kinase kinase (MEK), to rule out cell line-specific gene expression (30). Two types of genes were sought, those that were expressed in both the nontransformed Rat-1 cells and in the Rat-1(Ras) cells treated with PD98059 and those expressed only in the transformed cells in the absence of PD98059. The ␣1 chain of type I collagen (col1a1) was identified as a gene expressed in the Rat-1 cells but greatly down-regulated in the Ras-transformed cells (Fig. 1A). Expression of col1a1 was upregulated when the Rat-1(Ras) cells were treated with PD98059, suggesting that col1a1 is a gene whose expression is controlled by oncogenic ras through the MAP kinase pathway. Northern blot analysis using a 208-bp cDNA probe of col1a1 confirmed the regulation of this gene by the ras oncogene through the MAP kinase pathway. Expression of col1a1 was apparent 4 h after the Rat-1(Ras) cells were treated with PD98059, whereas the MEK inhibitor had little effect on ␣1(I) collagen expression in the parental Rat-1 cells (Fig. 1B).
To determine whether regulation of col1a1 expression by the ras oncogene was specific to fibroblast cells, rat intestinal epithelial (RIE) cells and the oncogenic h-Ras-transformed RIE(hras) derivative were analyzed by Northern blot. Both Rat-1 and RIE cells expressed high levels of col1a1, whereas expression was almost completely abolished in both cell types when the cells were transformed with the h-ras oncogene (Fig. 1C). Treatment of Rat-1:iRas cells with IPTG has been shown to induce expression of an oncogenic h-Ras protein 4 h after IPTG induction (13,20). Interestingly, a decrease in col1a1 mRNA expression was observed 8 h after IPTG induction, with almost complete abrogation of col1a1 mRNA expression by 24 h (Fig.  1D), again suggesting that expression of ␣1(I) collagen was down-regulated by the h-ras oncogene.
Disruption of the "Parallel Array" Phenotype by the ras Oncogene and Its Restoration by Exogenous Type I Collagen-To determine the functional effect of the loss of type I collagen on cell transformation, Rat-1 and Rat-1(Ras) cells were grown on tissue culture plates uncoated or coated with purified rat tail type I collagen. Collagen coating had little effect on the cellular morphology of the Rat-1 cells, which produce their own type I collagen. These cells exhibited contact inhibition and formed a monolayer of well oriented cells at confluence, a characteristic phenotype that we call a "parallel array" (Fig. 2). In contrast, collagen coating had a striking effect on the cellular morphology of the Rat-1(Ras) cells. The transformed cells, which appeared very disorganized in culture as a result of the loss of contact inhibition, were able to readopt the "parallel array" phenotype when grown on collagen-coated plates (Fig. 2). The phenotypic reversion of the Rat-1(Ras) cells led us to hypothesize that, although the Ras-transformed cells express very low levels of col1a1, they still retain a CBF(s) on the cell surface to which exogenous type I collagen can bind. Thus, identifying the CBF could be important to understanding cell transformation mediated by the ras oncogene.
Identification of a Putative ␣1(I) CBF on the Surface of Parental and ras-transformed Rat-1 and RIE Cells-To determine whether a specific ␣1(I) collagen receptor is expressed on the surface of Rat-1(Ras) cells, we employed APtag technology (24) to create AP-tagged ␣1(I) collagen fusion proteins with defined amino acid sequences. Previously, APtag was used to identify many cell surface receptors, including those for leptin (31), semaphorin III (32), and IL-24 (33). Logistically, we were most interested in identifying CBFs that would bind to the triplehelical region of ␣1(I) collagen. However, the C-prodomain of collagen is known to initiate trimerization of procollagen chains, which allows the triple helical glycine repeat region to form properly from the C terminus to the N terminus in a zipper-like manner. Therefore, three AP-collagen fusion proteins were constructed: one designated AP-Gly that was developed from the C-terminal glycine repeat region (amino acids 1000 -1189), one named AP-Pro that was developed from the C-prodomain alone (amino acids 1206 -1453), and one encompassing both of these regions, designated AP-Coll (amino acids 1000 -1453; Fig. 3A). We anticipated that the glycine repeats from AP-Gly, in contrast to those from AP-Coll, would not be able to fold correctly into a triple helix. In addition to having to include the C-prodomain for triple helical formation, we chose to focus on the C-terminal glycine repeat region to identify a novel CBF that is distinct from other known collagen receptors such as the integrins, which have been shown to bind to more upstream sequences of collagen. 293T cells were stably transfected with the recombinant expression plasmids, and the conditioned medium was analyzed via Western blot. AP-Coll was secreted as a 130-kDa fusion protein into the tissue culture medium, as visualized by Western blot analysis with an anti-AP polyclonal antibody (Fig. 3B). Interestingly, when SDS-PAGE was performed under nonreducing conditions, the AP-Coll-specific band was shifted to a higher molecular weight, suggesting that AP-Coll, similar to type I collagen, can form a trimeric structure. The 67-kDa human placental secreted AP alone was used as a control for antibody specificity.
To determine whether the AP-tagged fusion proteins could bind to the surface of parental and Ras-transformed Rat-1 cells, quantitative binding assays were completed. For these assays, cells were incubated with 293T conditioned medium containing equivalent AP activity units of either AP, AP-Pro, AP-Gly, or AP-Coll. Although AP, AP-Pro, and AP-Gly showed little binding to the surface of these cells (Fig. 3C, open bars; data not shown), AP-Coll exhibited a significant binding activity (6 -8fold over AP control) to the surface of the nontransformed Rat-1 cells (Fig. 3C, solid bars). Binding of AP-Coll was reduced, but still significant, on the Rat-1(Ras) cells, suggesting that these cells still retained the ability to bind to AP-Coll via a CBF (Fig.  3C). AP in situ cell surface staining assays also were completed on the Rat-1 cells. AP-Coll, but not AP alone, was able to bind specifically to the surface of these cells (Fig. 3D), confirming the results of the AP-Coll quantitative binding assay. Similar results were seen with the Rat-1(Ras) cells (data not shown). Finally, FACS analysis was completed to confirm the above results and to specifically evaluate the ability of AP and AP-Coll to bind to whole cells in suspension. Flow cytometric analysis demonstrated that AP-Coll could bind to both Rat-1 and Rat-1(Ras) cells, suggesting that these cells expressed detectable levels of the CBF on the membrane (Fig. 3E). These results have led us to conclude that a CBF does exist on the surface of parental and Ras-transformed fibroblast cells that binds to the region of ␣1(I) collagen encompassing the C-terminal glycine repeat and prodomain.
Characterization of the Nature of the AP-Coll-binding Protein-To further understand the nature of the AP-Coll-specific binding protein, Rat-1 and Rat-1(Ras) WCEs with or without heat denaturation at 100°C for 5 min were applied to a PVDF membrane via a slot blot apparatus. The membrane then was probed with equal amounts of either secreted AP or AP-Coll conditioned medium, followed by staining with AP substrate S. The results showed that the CBF is heat-labile (Fig. 4A), suggesting that it is a protein in nature.
To characterize the AP-Coll/CBF interaction, Rat-1 cytoplasmic protein extract and MPE were loaded onto a nonreducing SDS-polyacrylamide gel and transferred to a PVDF membrane, and the membrane was probed with either AP or AP-Coll conditioned medium and stained with AP assay reagent S. AP-Coll was shown to bind specifically to an ϳ180-kDa protein on the cell membrane (Fig. 4B), whereas AP alone failed to detect any binding (data not shown). In contrast, when the same experiment was carried out under reducing conditions, AP-Coll binding activity was not detected (data not shown). These results suggest that the CBF is a 180-kDa membrane protein or protein complex and that it must form disulfide bonds in order for AP-Coll to bind.
AP-Coll Competes with Type I Collagen for Binding to the Surface of Rat-1 and Rat-1(Ras) Cells-Type I collagen is an

FIG. 3. Construction of an AP-tagged collagen fusion protein (AP-Coll) that binds to the surface of parental and Ras-transformed Rat-1 cells.
A, schematic representation of the AP-tagged fragments used in the present study. B, Western blot analysis confirming the size of the secreted AP-Coll fusion protein described in A. 293T cells were stably transfected with the AP-Coll expression construct, and the AP-Coll fusion protein was secreted into the cell culture medium. Twenty-five l of conditioned medium were analyzed by Western blot using polyclonal antibody specific to AP, resulting in the visualization of a 130-kDa band. Secreted 67-kDa AP alone was used as a control. When loaded under nonreducing conditions, the AP-Coll band was shifted, suggesting that this protein forms a trimeric complex similar to fulllength type I collagen. C, quantitative receptor binding assays for AP-Coll. Rat-1 and Rat-1(Ras) cells were analyzed for AP-mediated (open bars) and AP-Coll-mediated (solid bars) binding activity. AP-Coll appeared to bind to the Rat-1 cells 6 -8-fold higher than AP alone. This binding activity was reduced but still significant in the Ras-transformed cells, suggesting that these cells retain the ability to bind to AP-Coll. Bars and error bars represent the average and S.D. from five independent experiments performed in duplicate. *, significant difference between AP-and AP-Coll-treated cells (p Ͻ 0.001). D, in situ staining of bound AP and AP-Coll. Rat-1 cells were assayed for bound AP activity with an AP substrate and viewed under a light microscope (ϫ 10 magnification) without phase contrast. Secreted AP-Coll bound to the surface of Rat-1 cells, whereas AP alone did not appear to bind, consistent with the results obtained from the quantitative binding assay. This assay was completed three times with duplicate samples. E, FACS analysis of AP and AP-Coll binding to the surface of Rat-1 and Rat-1(Ras) cells. Rat-1 and Rat-1(Ras) cells were incubated with 15 g/ml AP and AP-Coll and analyzed by flow cytometry. Note that whereas AP-Coll bound to the surface of both Rat-1 and Rat-1(Ras) cells, no binding was observed with AP alone. This assay was completed twice in triplicate.

FIG. 4. Characterization of the nature of the CBF.
A, the AP-Coll-specific binding factor appears to be a protein in nature. Rat-1 and Rat-1(Ras) WCEs were either heat-denatured or not and then were applied to a PVDF membrane via a slot blot apparatus. The membrane was probed with AP or AP-Coll conditioned medium and stained with AP substrate. Both cell types expressed detectable levels of a heat-labile protein to which AP-Coll appeared to bind. B, Rat-1 cytoplasmic protein extract (CPE) and MPE were subjected to SDS-PAGE under nonreducing conditions. The proteins were transferred to a PVDF membrane, probed with AP-Coll conditioned medium, and stained with AP substrate to locate the approximate molecular weight of the CBF. A band of ϳ180 kDa was identified. extracellular matrix protein that, through interactions with various cell surface proteins, mediates proper cell-matrix adhesion. Interruption of these binding contacts could damage the ability of cells to adhere as needed, leading to a more migratory phenotype. The importance of the AP-Coll/CBF interaction was investigated by performing a competition binding assay with various extracellular matrix proteins. Detached Rat-1 and Rat-1(Ras) cells were incubated with 293T conditioned medium alone or conditioned medium containing a 5 g/ml concentration of either secreted AP or AP-Coll and subsequently added to tissue culture plates precoated with type I collagen, fibronectin, or matrigel. Interestingly, incubation with secreted AP-Coll significantly inhibited the ability of the Rat-1 and Rat-1(Ras) cells to bind to type I collagen-coated plates (Fig. 5, A and B). In contrast, AP-Coll did not inhibit cell binding to either fibronectin or matrigel, suggesting that AP-Coll binds to a CBF on the cell surface that specifically interacts with collagen.
Biochemical Purification to Identify and Characterize a Novel Binding Protein for AP-Coll-Type I collagen binds to a variety of cell surface receptors, including integrins, discoidin domain receptor tyrosine kinases (DDRs), and syndecans (34 -38). In our quest to discover a novel collagen-binding protein on the surface of Rat-1 and Rat-1(Ras) cells, it was important to first determine whether AP-Coll could bind to any of the well known type I collagen receptors. Therefore, numerous binding assays were performed with AP-Coll in the presence of various anti-integrin antibodies (i.e. anti-␣1, -␣2, and -␤1 integrin, the major collagen binding receptors among the integrin family) or with cells null for ␣1 integrin (39), ␤1 integrin (40), or DDR1 (41) from knockout mice. Additionally, AP-Coll binding assays were performed on ARH-77 (human B lymphoid) cells that had been stably transfected with syndecan 1, 2, or 4 (38). In these experiments, AP-Coll did not appear to interact with any of these known collagen receptors (data not shown), again suggesting that the CBF is a novel receptor.
In order to identify the CBF, we developed a biochemical purification scheme, using the slot blot and nonreducing SDS-PAGE as assays to monitor AP-Coll binding activity. Solubilized Rat-1 MPE was loaded onto a Resource Q anion exchange column and, after washing, bound proteins were eluted in fractions with an increasing salt gradient. The fraction containing the highest AP-Coll-specific binding activity was subjected to nonreducing SDS-PAGE and then probed with AP-Coll conditioned medium. AP-Coll could bind to an ϳ180-kDa protein in the anion exchange chromatography fraction (Fig. 6A, top,  arrow). When a nonreducing SDS-polyacrylamide gel containing WCE, MPE, and the fraction with AP-Coll binding activity was stained with colloidal blue, a band with an electrophoretic mobility similar to that seen in Fig. 6A (top) was observed in the lane containing the eluted fraction with AP-Coll binding activity (Fig. 6A, bottom, arrow). This band was excised from the gel and subjected to tryptic digestion and MALDI mass spectrometry analysis, and although a rat homologue was not found in the NCBInr data base, the CBF was unambiguously identified as human Endo180.
AP-Coll Binds to Endo180 on the Cell Surface-Results from MALDI analysis suggested that the identity of the CBF was Endo180. To confirm these results, CosE5 and MCF-7 cells were transiently transfected with vector alone (ϩpcDNA3) or vector expressing human Endo180 cDNA (ϩpcDNA3-Endo180). Western blot analysis was performed using the En-do180 monoclonal antibody A5/158. CosE5 ϩ pcDNA3 cells appeared to express endogenous Endo180, and this expression was significantly up-regulated in cells transfected with End-o180 cDNA (Fig. 7A). Quantitative binding assays also were completed with the CosE5 cells transiently transfected with empty vector or with vector plus Endo180. Compared with AP, AP-Coll showed significant binding activity to the CosE5 ϩ pcDNA3 cells, although this binding activity was significantly increased with the CosE5 ϩ pcDNA3-Endo180 cells (Fig. 7B). AP in situ cell surface staining assays unambiguously confirmed the quantitative binding data and showed strong AP-Coll-positive cell staining in the wells that had been transiently transfected with Endo180 cDNA (Fig. 7C). Similar results also were observed with MCF-7 cells transfected with Endo180 cDNA when compared with cells transfected with vector alone (data not shown).
Endo180 Is Expressed as a Glycosylated Protein on the Surface of Rat-1 and Rat-1(Ras) Cells-In order to examine Endo180 expression in the rodent system where the CBF was originally discovered, Northern and Western blots were carried out with the Rat-1 and Rat-1(Ras) cells to determine Endo180 mRNA and protein expression, respectively. Both the parental Rat-1 cells and the Rat-1(Ras) cells appeared to express similar levels of Endo180 mRNA in contrast to col1a1 mRNA, which is greatly down-regulated in the oncogenic Ras-transformed cells (Fig. 8A). At the protein level, Endo180 was found to be expressed on the membrane of both Rat-1 and Rat-1(Ras) cells.

FIG. 5. The AP-Coll/CBF interaction appears to be important for proper collagen-mediated cell adhesion. Rat-1 (A) and
Rat-1(Ras) (B) cells were incubated with conditioned medium alone (open bars), 5 g/ml secreted AP (gray bars), or 5 g/ml secreted AP-Coll (solid bars) for 90 min at room temperature with gentle agitation. The cells then were plated onto 96-well tissue culture plates coated with either type I collagen (50 g/ml), fibronectin (5 g/ml), or Matrigel (100 g/ml). After 90 min, unattached cells were removed, and attached cells were washed, fixed, and stained with crystal violet. Adhesion to type I collagen-coated plates was significantly inhibited by AP-Coll, whereas the ability of these cells to attach to fibronectin or matrigel-coated plates was unaltered. The bars and error bars represent the average and S.D. from three independent experiments performed in duplicate. *, significant difference between conditioned medium alone, conditioned medium containing 5 g/ml secreted AP, and conditioned medium containing 5 g/ml secreted AP-Coll (p Ͻ 0.001).
When the membrane was probed with secreted AP-Coll instead of polyclonal Endo180 antibody, an AP-Coll-specific band with an electrophoretic mobility identical to Endo180 was found in the MPE of Rat-1 and Rat-1(Ras) cells (Fig. 8B). These results consistently support our hypothesis that Endo180 is the CBF.
siRNA-mediated Knockdown of Endo180 Decreases the Ability of AP-Coll to Bind to the Surface of Rat-1 and Rat-1(Ras) Cells-The above data demonstrate that Endo180 is localized FIG. 6. Isolation and characterization of the AP-Coll binding protein. The AP-Coll-specific binding protein was isolated by loading Rat-1 membrane extract onto a ResourceQ ion exchange column and using a salt gradient to elute the bound proteins as fractions. A, the fraction containing the highest binding activity was subjected to SDS-PAGE under nonreducing conditions, along with WCE and MPE. The proteins on the gel were transferred to a PVDF membrane and probed first with AP-Coll conditioned medium, then with an AP monoclonal antibody, and finally with horseradish peroxidase-conjugated anti-mouse IgG antibody to identify the location of the AP-Coll-specific binding protein (top, arrow). Alternatively, the gel was stained with colloidal Coomassie and destained, and the band from the fraction of interest was excised for MALDI mass spectrometry analysis (bottom, arrow). B, peptide mapping of the isolated AP-Coll binding protein by MALDI mass spectrometry. After excision from the colloidal Coomassie-stained gel, the AP-Coll binding protein was treated with trypsin using the in-gel digestion technique. The resulting peptides were analyzed directly by MALDI mass spectrometry.

FIG. 7. Confirmation that AP-Coll binds to Endo180 on the cell surface.
A, Western blot analysis of whole cell extracts from CosE5 cells transiently transfected with either the pcDNA3 vector or pcDNA3-Endo180, using a monoclonal antibody against Endo180. B, quantitative receptor binding assays for AP-Coll. CosE5 cells were transiently transfected with either pcDNA3 empty vector or pcDNA3-Endo180 and were assessed for their ability to bind to AP-Coll (solid bars) versus AP alone (open bars). CosE5 cells transfected with pcDNA3 exhibited appreciable AP-Coll-specific binding over AP alone. The ability of AP-Coll to bind to the cell surface was dramatically increased when cells were transfected with pcDNA3-Endo180. Bars and error bars represent the average and S.D. of duplicate samples from three independent experiments. Significant difference was calculated between AP-and AP-Coll-treated cells (*, p Ͻ 0.05; **, p Ͻ 0.001) and between CosE5 ϩ pcDNA3 and CosE5 ϩ pcDNA3-Endo180 cells treated with AP-Coll (***, p Ͻ 0.001). C, in situ cell surface staining of bound AP and AP-Coll to the transfected CosE5 cells. CosE5 cells transfected with either pcDNA3 or pcDNA3-Endo180 were stained for cell surface-bound AP-Coll activity and viewed under a light microscope (ϫ 20 magnification) without phase contrast. The results of this assay were consistent with the data obtained from the quantitative binding study, suggesting that AP-Coll binds to Endo180 on the cell surface. This assay was completed two times in duplicate.
to the surface of Rat-1 and Rat-1(Ras) cells and mediates AP-Coll binding activity. To further confirm our finding that the binding is due to a specific interaction between AP-Coll and Endo180, we utilized the siRNA pSuperRetro vector system to reduce the expression of Endo180 in Rat-1 and Rat-1(Ras) cells. As a negative control, cells also were infected with pSuperRetro vector alone. Two different Endo180 siRNA constructs targeting the cDNA sequence at positions 117 and 405 from the start codon were developed. Infection of Rat-1 and Rat-1(Ras) cells with the two pSuperRetro-Endo180 constructs resulted in isolation of several cell clones that expressed lower levels of Endo180. In agreement with previous binding data, the ability of AP-Coll to bind to Rat-1 (Fig. 9A) and Rat-1(Ras) (Fig. 9B) cells was significantly reduced when Endo180 expression was down-regulated. These data further confirmed that Endo180 is indeed the major AP-Coll-binding protein on the surface of Rat-1 and Rat-1(Ras) cells.
cDNA Cloning and Sequencing of Rat Endo180 -In order to determine the cDNA sequence of rat Endo180, which is not listed in the NCBInr data base, the rat expressed sequence tag data base was "probed" with the full-length nucleotide sequence of mouse Endo180. A clone was obtained from ATCC (IMAGE clone number 7099211) that showed high homology to mouse Endo180 and was sequenced with primer walking. Rat Endo180 cDNA is roughly 4.5 kb in length and encodes a protein of 1479 amino acid residues. The rat Endo180 protein shares 90 and 85% homology to that of mice and humans, respectively (data not shown).

DISCUSSION
In this study, we have demonstrated that transformation of Rat-1 and RIE cells by the h-ras oncogene resulted in marked down-regulation of col1a1 collagen expression. Expression of col1a1 collagen was quickly restored when the Rat-1(Ras) cells were treated with PD98059, suggesting that repression of its expression by the h-ras oncogene occurred through the MAP kinase pathway. Additionally, col1a1 was down-regulated in a timely manner when nontransformed Rat-1 cells containing an IPTG-inducible h-ras oncogene were treated with IPTG. Our results are in agreement with previous findings that have suggested that col1a1 expression is down-regulated by the ras oncogene at the transcriptional and post-transcriptional levels (42,43). The ␣1(I) chain is a subunit of type I collagen, a 210-kDa protein normally composed of two ␣1 and one ␣2 chains that forms an ␣1(I) 2 ␣2(I) 1 complex (44), although a homotrimeric ␣1(I) 3 complex also can form (45). Type I collagen is synthesized initially as a procollagen precursor with a central triple-helical glycine repeat region and massive amino-and carboxyl-terminal propeptide domains that help maintain the solubility of procollagen inside of the cell (46). Additionally, the C-propeptide is essential in forming the correct triple-helical structure of the glycine repeat region, since disulfide bond formation in the C-prodomain aligns the chains so that the collagen heterotrimer can fold in a zipper-like manner toward the amino terminus (47,48). Importantly, type I collagen acts as a substrate for cell attachment and binds to various cell surface proteins, including several integrin heterodimers, DDRs, and syndecans (34 -38).
The biological consequence of loss of collagen production in cell transformation has not been fully characterized. To determine the biological function of type I collagen in cell transformation, Ras-transformed fibroblast cells were plated on type I collagen-coated plates. In contrast to cells plated on mockcoated plates, these cells, when grown to confluence, displayed restored cell-matrix contacts, as visualized by the appearance of the well organized "parallel array" phenotype that is characteristic of the nontransformed parental Rat-1 cells. This discovery led us to hypothesize that the Ras-transformed cells still retained a CBF to which type I collagen could bind. Through biochemical purification and MALDI mass spectrometry analysis, we were able to identify Endo180 as a CBF that binds to the C-terminal glycine repeat region and C-prodomain of ␣1(I) collagen. This discovery was further verified first in the human system via ectopic expression of Endo180 followed by quantitative binding and in situ cell surface staining assays and then in the rodent system using Western blot analysis and RNAi.
Endo180 was independently discovered by three different groups. It was originally identified as an antigen that could bind to four different monoclonal antibodies that had been raised solely to identify novel human fibroblast cell surface receptors and was shown to be a constitutively recycling cell surface protein (26). Endo180 also was found to form a trimolecular cell surface complex with urokinase plasminogen activator and its receptor through chemical cross-linking studies and thus also is named urokinase plasminogen activator receptor-associated protein (49). A third group discovered that Endo180 encoded a novel macrophage mannose receptor type C lectin that was present in an expressed sequence tag data base (50). This cell surface protein has been identified as the fourth member of the mannose receptor family, which is also composed of the macrophage mannose receptor, the M-type phos- pholipase A 2 receptor, and DEC-205 (51). The four receptors are type I transmembrane proteins with a highly conserved structure composed of an N-terminal cysteine-rich or ricin-type domain, a fibronectin type II domain, 8 or 10 C-type lectin-like domains (CTLDs), a single transmembrane domain, and a short cytoplasmic domain (51). Endo180 previously has been shown to act as a collagen receptor by binding to gelatin, type I, type II, type IV, and type V collagens and also mediates endocytosis and turnover of these extracellular matrix proteins (52)(53)(54)(55)(56). The fibronectin type II domain of Endo180 (which shares a highly conserved sequence with other collagen-binding fibronectin type II domains, such as fibronectin, MMP-2, and MMP-9 (49, 51, 52)) appears to be required for binding to these collagens. However, the region of collagen to which Endo180 binds has not been previously defined, due to the use of native collagens from tissue sources.
Here we show that Endo180 binds to a defined sequence of ␣1(I) collagen encompassing both the C-terminal glycine repeat and prodomain (AP-Coll; amino acids 1000 -1453). To further define whether either the C-terminal glycine repeat region or C-prodomain alone was capable of initiating this binding, APTag technology was utilized to develop AP-Gly (amino acids 1000 -1189) and AP-Pro (amino acids 1206 -1453) fusion constructs, respectively. However, these fusion proteins exhibited little binding to the surface of Rat-1 and Rat-1(Ras) cells (data not shown), compared with AP-Coll. AP-Gly was unable to form a trimeric structure (data not shown), suggesting that the fusion construct could not fold in its proper conformation without the C-prodomain. Although AP-Pro could form a trimeric structure, little binding was detected (data not shown), suggesting that binding of Endo180 to type I collagen is mediated mostly through the C-terminal glycine repeat region, which must be in a triple helical conformation. This is supported by our previous findings that Endo180 could bind to commercially purified collagens without the C-prodomain (52,53). Thus, it appears that the C-prodomain of collagen is disposable for Endo180 binding once the collagen triple helix is already formed and processed. However, we cannot rule out the possibility that the C-prodomain also contributes to Endo180 binding.
The interaction between various collagen types and Endo180 appears to play a role in cell-matrix adhesion, since Endo180-deficient cells have a 50% reduction in adhesion to type V collagen and also a reduced ability to bind to purified types I and IV collagen (54). To investigate whether the C-terminal glycine repeat region and C-prodomain of the ␣1 chain collagen mediate this adhesion to Endo180, we performed competition binding experiments. Our results demonstrated that AP-Coll was able to significantly inhibit the adhesion of Rat-1 and Rat-1(Ras) cells specifically to type I collagen, suggesting that this region of collagen facilitates adhesion of cells to the extracellular matrix. Interestingly, the majority of the known bindings sites for the ␣1␤1 and ␣2␤1 integrins, which are largely responsible for mediating cellular adhesion to collagen, have been mapped outside of the C-terminal region (57). Thus, our results suggest that the C-terminal region of ␣1(I) collagen, similar to the integrin-binding domains, plays an important role in cell adhesion.
In order to determine whether Endo180 acts as a signaling receptor or serves only as an endocytotic and adhesive protein, we looked at the phosphorylation of Endo180 in serum-starved cells upon treatment with AP-Coll conditioned medium. Preliminary results suggest that the AP-Coll/Endo180 interaction is not involved in cell signaling, since Endo180 did not appear to be phosphorylated on serine, threonine, or tyrosine upon binding to AP-Coll (data not shown). In addition, the MAP kinases (extracellular signal-regulated kinase, c-Jun N-terminal kinase, and p38) did not seem to be phosphorylated following AP-Coll binding to Endo180 (data not shown). Therefore, the main biological functions of Endo180 appear to be mediating the uptake of collagen and its subsequent delivery to the lysosomes for degradation (52,55) and also cellular adhesion.
Interestingly, although similar at the mRNA level, expression of Endo180 at the protein level in the Ras-transformed fibroblasts appears to be lower than in the parental cells, consistent with AP-Coll ligand affinity staining of the Rat-1 and Rat-1(Ras) MPE and also with the results of the quantitative ligand binding assays. One possible explanation for the lower Endo180 protein level in the Ras-transformed cells could be the decreased ␣1(I) collagen production. Conceivably, in contrast to Rat-1 cells, most of the Endo180 on the surface of the Rat-1(Ras) cells would not be bound to ligand, due to almost complete abrogation of ␣1(I) collagen expression by the ras oncogene. Thus, ligand binding to Endo180 may help to main- FIG. 9. Down-regulation of endogenous Endo180 in Rat-1 and Rat-1(Ras) cells reduces the ability of AP-Coll to bind to these cells. Rat-1 (A) and Rat-1(Ras) (B) cells were infected with pSuperRetro empty vector or the pSuperRetro-Endo180 constructs, and stable cell clones were selected with 10 g/ml puromycin. Western blot analysis was completed to show down-regulated expression of Endo180 in two of the positive clones for both cell types. Quantitative binding assays were completed with the same clones, and a significant decrease in the ability of AP-Coll to bind to the surface of the pSuperRetro-Endo180 stables was visualized, compared with control cells or cells stably infected with pSuperRetro alone. Bars and error bars represent the average and S.D. of three independent experiments performed in duplicate. Significant difference was calculated between control cells, cells plus pSuperRetro, cells plus pSuperRetro RNAi-1, and cells plus pSuperRetro RNAi-2 (*, p Ͻ 0.001).
tain the stability of type I collagen. This hypothesis is supported by our finding that Rat-1(Ras) cells grown on type I collagen-coated plates appeared to express higher levels of Endo180 on the membrane (data not shown).
Loss of type I collagen due to Ras transformation appeared to be largely responsible for the disappearance of the "parallel array" phenotype, which is one of the major morphological changes associated with cell transformation. However, it is not yet clear whether the interaction between type I collagen and Endo180 is responsible for this phenotype. Rat-1 and Rat-1(Ras) cells that had undergone siRNA-mediated knockdown of Endo180 retained the ability to form the "parallel array" phenotype on type I collagen-coated plates, despite greatly reduced expression of Endo180 (data not shown). However, residual Endo180 expression in the siRNA knockdown Rat-1 and Rat-1(Ras) cells might be sufficient to mediate the "parallel array" phenotype. Future experiments with mouse embryonic fibroblasts generated from Endo180 knockout mice (53) may help to better define the role of the type I collagen/Endo180 interaction in the formation of the "parallel array." Conceivably, there could be a second CBF on the surface of these cells to which AP-Coll can bind, given the residual AP-Coll binding to the Endo180 RNAi knockdown cells. Preliminary experiments suggest that this additional CBF is unlikely to be any of the known collagen-binding integrins, DDRs, or syndecans, since we have ruled out their involvement using cells either deficient in or overexpressing these genes (data not shown). It is possible that AP-Coll could also bind to other members of the mannose receptor family, due to their highly conserved protein structure, in particular the fibronectin type II domain, which has been found to mediate collagen binding (52). Future studies should help to clarify these unsettled issues.