INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

During the course of hematogenous metastasis, cancer cells escape from the primary tumor, enter the blood stream, arrest in the vasculature of a secondary organ, and extravasate to form new tumor colonies (reviewed in Ref. 1). The fate of tumor cells in the blood circulation has been traced by injecting labeled cells via intravenous and intracardiac routes. These studies conclude that cancer cells initially arrest in the microvasculature of the first organ they enter. Most tumor cells die in this location (2), and only a few succeed to form metastases or recirculate to colonize other organs in a tumor type-specific pattern (3). Clinical assessment supports these data indicating that some cancers favor select secondary locations for metastasis (4). For example, prostatic carcinomas and small cell carcinomas of the lungs preferentially colonize bones and the brain, respectively, while breast carcinomas most frequently metastasize to the lungs, but also to liver, bones, brain, and adrenals. There is mounting evidence that the initial selection of an organ for metastasis occurs at the time of attachment of blood-borne cancer cells to microvascular endothelia of that site. Vascular arrest appears to be mediated by "organotypic" molecules that are expressed on the endothelial cell surface of select vascular branches (i.e. postcapillary venules) (reviewed in Refs. 5-7). A specific example of such a molecule includes recent work in this laboratory detailing the isolation and characterization of the 90-kDa lung endothelial cell adhesion molecule-1 (Lu-ECAM-1)¹ (8-11). Lu-ECAM-1 selectively binds lung-metastatic melanoma cells, and its expression on endothelia of pulmonary venules correlates closely with the formation of melanoma metastases in these locations (11). Antiadhesive, anti-Lu-ECAM-1 monoclonal antibodies (mAbs) inhibit colonization of the lungs by lung-metastatic murine B16 melanoma cells but have no effect on lung colonization by other types of lung-metastatic cancer cells tested thus far (9).

In related work more recently, outside-out luminal membrane vesicles isolated from rat lung microvascular endothelia by in situ perfusion with a low strength paraformaldehyde solution were shown to bind in significantly larger numbers to lung-metastatic than to nonmetastatic rat breast carcinoma cells (12, 13). In contrast, vesicles prepared from the vasculature of a nonmetastasized organ showed no binding preference for either lung-metastatic or nonmetastatic mammary carcinoma cells. The mAb 6A3 generated against lung-derived endothelial cell membrane vesicles was shown to inhibit specific adhesion of lung endothelial vesicles to lung-metastatic breast cancer cells. The antibody identified a 110-kDa membrane glycoprotein of rat lung capillary endothelia, and N-terminal sequencing established identity with dipeptidyl peptidase IV (DPP IV; also known as CD26 or gp110) (13). Two basic properties of DPP IV may account for the putative ability to serve as an adhesion molecule for cancer cells. First, consistent with its enzymatic function (reviewed in Ref. 14), DPP IV may use its substrate binding domain to form transient, adhesive bonds with substrates associated with the tumor cell surface. Such binding might be mediated by x-proline dipeptide sequences (e.g. RP, KP, and GP) of the putative DPP IV substrate. However, there is no direct evidence that such a mechanism may lead to cancer cell binding, although this mode of action has been proposed for human immunodeficiency virus adhesion to and entry into T lymphocytes (15). Alternatively, the ability of DPP IV to bind to fibronectin (FN) via a domain distinct from its substrate recognition site (16) and the previously recognized association between cell surface expression of FN and lung metastasis of rhabdomyosarcoma cells (17) prompted us to investigate whether tumor cell surface-associated FN served as the ligand for DPP IV.

Here, we confirm that DPP IV is an endothelial cell adhesion molecule for rat breast cancer cells and mediates lung metastasis by these tumor cells. The DPP IV ligand is identified as tumor cell surface-associated FN, and concomitantly, a correlation between the level of FN expression and the tumor cells' ability to bind to DPP IV and metastasize to the lungs is established. DPP IV/FN-mediated adhesion and metastasis are blocked when tumor cells are incubated with soluble DPP IV prior to conducting adhesion and lung colony assays. Adhesion is also blocked by anti-DPP IV mAb 6A3 and by anti-FN antiserum but is unaffected by soluble plasma FN (pFN) and thus may readily happen during hematogenous spread of cancer cells in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Rat Mammary Carcinoma Cells and Their Metastatic Potential-- The rat breast carcinoma cell lines R3230AC-MET and R3230AC-LR were obtained from Dr. J. A. Kellen (Sunnybrook Medical Center, University of Toronto, Toronto, Canada) (18). The R3230AC-MET cell line was selected in vivo for high lung colonization. The R3230AC-LR cell line was concanavalin A- and wheat germ agglutinin-resistant and nonmetastatic. The lung-metastatic MTF7 clone of the rat mammary adenocarcinoma cell line 13762NF was received from Dr. D. R. Welch (Pennsylvania State College of Medicine, Hershey, PA) (19). RPC-2 cells were isolated from lung metastases of the in vivo transplantable Dunning R3327 rat carcinoma MatLyLu donated by Dr. J. T. Isaacs (Johns Hopkins Oncology Center, Baltimore, MD) (20). Detailed tissue typing of Dunning R3327 tumors suggest that these cancers are not, as originally thought, of prostatic origin but are likely derived from mammary epithelium (21). The metastatic potential of the four breast cancer cell lines was expressed as the median (range in parentheses) number of tumor colonies observed in the lungs 3 weeks after intravenous inoculation of 2 × 10⁵ tumor cells into female Fischer 344 rats. R3230AC-MET produced 204 (176-237) lung colonies, R3320AC-LR 0 (0), MTF7 385 (312-397), and RPC-2 285 (78-327). Tumor cells were used for subsequent experiments within 10 passages following evaluation of their metastatic potential. They were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Life Technologies, Inc.).

Antibodies-- Anti-DPP IV mAb 6A3 was produced in Balb/c mice and purified from hybridoma supernatant by Protein G chromatography (RepliGen, Cambridge, MA) as described (13). mAb 6A3 (IgG₁) recognized rat DPP IV expressed by endothelia of lung capillaries, splenic sinusoidal venules, and renal vasa recta as well as by epithelia of bile canaliculi, kidney proximal tubuli, and small intestinal villi (13). Polyclonal anti-DPP IV antibodies CU31 were prepared in rabbits, using purified rat endothelial DPP IV. Control mAbs were directed against the endothelial cell adhesion molecule Lu-ECAM-1 (6D3; IgG_2a) (8, 11). Polyclonal antisera raised against rat pFN were obtained from Dr. J. L. Guan (22), and antisera against human pFN were from Life Technologies, Inc. A polyclonal, anti-peptide antiserum generated against human adenosine deaminase was a gift from Dr. J. W. Belmont (Institute for Molecular Genetics and Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX) (23).

Immunoaffinity Purification of Endothelial DPP IV-- Thirty rat lungs, homogenized and washed in ice-cold 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, were extracted for 30 min at 4 °C with 150 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM benzamidine chloride, 1 mM phenylmethylsulfonyl fluoride, 2 µg/ml leupeptin, 0.27 trypsin inhibitory units/ml aprotinin, 30 µg/ml DNase, and 1% Nonidet P-40 (13). The lysates were first precleared on a 1-ml column containing nonimmune mouse IgG immobilized on Protein G-agarose (Life Technologies, Inc.), and the flow-through was directly applied onto a second 1-ml column of anti-DPP IV mAb 6A3 coupled to Protein G-agarose. Columns were washed and eluted as described in detail elsewhere (8, 10). The purity of the isolated DPP IV was monitored by SDS-PAGE (8% polyacrylamide) and visualized by silver or Coomassie Blue staining.

Affinity Purification of the Metabolically Labeled DPP IV Ligand-- Breast cancer cells in logarithmic growth phase were washed for 20 min at 37 °C in methionine-free RPMI 1640 medium and then metabolically labeled overnight at 37 °C with 0.4 mCi of [³⁵S]methionine in methionine-free RPMI 1640 medium containing 20 µM methionine and 10% dialyzed, FN-free FBS. To differentiate between labeled surface-associated and labeled cytoplasmic proteins, tumor cells were extracted in lysis buffer (30 min; 4 °C) either immediately or after treatment with -chymotrypsin (10 µg/ml; 30 min; 37 °C) as suggested by Hynes (25). Extracts were cleared by centrifugation, and the DPP IV-tumor cell ligand was precipitated with DPP IV immobilized on Affi-Gel 10 beads (Bio-Rad Laboratories). DPP IV-ligand complexes were resolved by SDS-PAGE (5% polyacrylamide) under nonreducing and reducing (2% -mercaptoethanol (BME)) conditions and visualized by autoradiography.

Incorporation of ¹²⁵I-pFN into the Tumor Cell Glycocalyx-- The ability of breast cancer cells to incorporate pFN into their surface coat was tested under conditions that mimicked hematogenous spread. In brief, tumor cells (5 × 10⁶ cells) enzymatically released from their growth surface (0.25% trypsin in PBS; 5 min; 37 °C) were washed once in RPMI 1640 medium containing 10% FN-free FBS to stop the enzyme action and then incubated for various periods of time in rotating suspension cultures in RPMI 1640 medium supplemented with 10% FN-free FBS, 10 µg/ml unlabeled rat pFN (Life Technologies, Inc.), and 1 µg/ml ¹²⁵I-pFN labeled by the IODO-BEAD^TM method as described by the manufacturer (Pierce). Cells were extracted for 30 min at 4 °C in either lysis buffer or 2% deoxycholate (DOC) in lysis buffer without Nonidet P-40 (26). Extracts were precipitated with immobilized DPP IV as described above, and the precipitates were subjected to SDS-PAGE (5% polyacrylamide) under both nonreducing and reducing conditions. The DOC-insoluble fraction was directly applied to SDS-PAGE. FN was visualized by autoradiography.

Plasmid Construction and Transfection-- All transfection studies were performed with rat kidney DPP IV cDNA obtained from Dr. D. Doyle (State University of New York, Buffalo, NY) (27). The nucleotide sequence of kidney DPP IV cDNA was 100% identical to that of rat lung endothelial DPP IV cloned in our laboratory.² HEK293 cells were transiently transfected with DPP IV cDNA cloned into pRcCMV (Invitrogen, San Diego, CA), using Lipofectamine according to the manufacturer's instructions (Life Technologies). Control HEK293 cells were transfected with the pRcCMV vector alone.

Rosette Assay-- A rosette assay was performed between DPP IV-transfected HEK293 cells and MTF7 breast cancer cells. MTF7 cells (1 × 10⁴ cells/well) were seeded onto Lab-Tek^® two-well chamber slides (Nunc Inc., Naperville, IL) and grown overnight at 37 °C in RPMI 1640 medium supplemented with 10% FN-free FBS. Cells were then labeled with the cytoplasmic dye Calcein-AM (Molecular Probes, Inc., Eugene, OR) as described by El-Sabban and Pauli (28). A 100-fold excess of either DPP IV- or mock-transfected HEK293 cells (1 × 10⁶ cells/well) previously labeled with SNARF-1-AM (Molecular Probes) was seeded onto the adherent MTF7 breast cancer cells in serum-free medium. MTF7 and HEK293 cells were co-cultured under a gentle rocking motion for 30 min at 37 °C. Nonadherent HEK293 cells were removed by washing, and slides were examined under a fluorescent microscope (excitation, 490 nm; emission, 520-540 nm for Calcein and 550-650 nm for SNARF-1). Rosetting was indicated by the binding of six or more DPP IV-transfected HEK293 cells to MTF7 breast cancer cells. A total of 100 cells were counted in each of three experiments.

Flow Cytometry-- Fluorescence-activated cell sorting (FACS) was performed to quantify FN expression on breast cancer cell surfaces. Tumor cells released from their growth surface and recovered as described above were suspended in 10% donkey serum in PBS for 15 min at 4 °C and then incubated with rabbit anti-rat pFN antiserum (diluted 1:100 in PBS) for 1 h at 4 °C. Cells were stained with fluorescein isothiocyanate-conjugated donkey anti-rabbit antiserum in PBS containing 10% donkey serum for 1 h at 4 °C and fixed in 2% paraformaldehyde in PBS. FACS analysis was performed on a Coulter Epics Profile (Coulter Electronics, Hialeah, FL). Nonspecific fluorescence was accounted for by incubating tumor cells with nonimmune serum instead of primary antibody.

Enzyme-linked Immunosorbent Assay-- Immulon^® 4 Microtitration flat bottom plates (Dynatech Laboratories Inc., Chantilly, VA) were coated with pFN (5 µg/ml in PBS) overnight at 4 °C. Immunopurified DPP IV (0, 0.1, 0.3, 0.5, 1.0, and 2.0 µg/ml) was added to the FN-coated wells in the presence or absence of 10 µg/ml pFN and incubated for 1 h at room temperature. Unbound DPP IV was removed by washing, and bound DPP IV was detected by enzyme-linked immunosorbent assay with anti-DPP IV antiserum CU31 (1:500 in PBS) (8).

Tumor Cell Adhesion Assay-- Tumor cell adhesion assays were performed as described (9). The amount of DPP IV adsorbed per unit well surface area of Immulon^® 4 Microtitration plates was determined from peptidase activity measurements relative to standard DPP IV enzyme activity curves (29). Assays were conducted in the presence or absence of the following components in PBS: (a) mAbs 6A3 and 6D3 (both tested at 50 µg/ml); (b) DPP IV substrate GPA and control peptide GGA (Sigma) (20 mM); (c) serine proteinase inhibitors 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF) (Calbiochem) or phenylmethylsulfonyl fluoride (Sigma) (1-5 mM); (d) soluble DPP IV (wild-type DPP IV solubilized in 0.05% OG in PBS; truncated DPP IV in PBS alone) or glycophorin (50-1000 ng/ml); (e) soluble rat pFN (1-10 µg/ml); and (f) anti-FN antiserum (diluted 1:50 or 1:100). With the exception of soluble DPP IV and anti-FN antiserum, which were incubated with the tumor cells before addition to DPP IV-coated wells, these compounds were preincubated in DPP IV-coated wells for 1-2 h at room temperature and kept in the assay media throughout the tumor cell binding period unless indicated otherwise.

Lung Colony Assay-- Breast cancer cells (2 × 10⁵ cells/0.3 ml of PBS/rat) were inoculated via the lateral tail vein of 6-week-old, female Fischer 344 rats (Charles River Laboratories) to determine their metastatic potential. Rats were sacrificed 3 weeks after tumor cell injection, and lung colonies were counted using a dissecting microscope. Median and range of the number of lung colonies were determined for each cell line. Metastasis inhibition experiments were conducted only with MTF7 cells. For this purpose, MTF7 cells were incubated for 1 h at 37 °C in the presence or absence of purified, truncated DPP IV (80 µg/ml in PBS) prepared as described by Yamaguchi et al. (24) and then inoculated into rats as indicated above. Statistical comparisons between treatment groups were performed with Student's t test for unpaired data.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Lung Endothelial DPP IV Mediates Adhesion and Metastasis of Lung-metastatic Rat Breast Cancer Cells-- Lung-metastatic rat mammary carcinoma cells (MTF7; R3230AC-MET; RPC-2) adhered to DPP IV isolated and purified from rat lungs (capillary endothelia) in a dose-dependent manner (Fig. 1). Adhesion of the three metastatic tumor cell lines plateaued at a coating concentration of 500 ng of DPP IV/50 µl of PBS/well (equal to 2.25 pmol of DPP IV bound per 1 mm² of well bottom surface), yielding adhesion values of 60-70% for MTF7 carcinoma cells and 40-50% each for R3230AC-MET and RPC-2 carcinoma cells. By comparison, adhesion of nonmetastatic R3230AC-LR tumor cells reached only 8-12% at the same DPP IV coating concentration. The specific adhesion of these cancer cells to DPP IV was inhibited in a statistically significant manner (approximately 95%) upon incubation of the DPP IV-coated wells with monospecific anti-DPP IV mAb 6A3 (50 µg/ml) (Fig. 2). Control mAbs of the same immunoglobulin class had negligible effects on specific tumor cell binding. Participation of the peptidase substrate domain in the DPP IV binding to breast cancer cells was ruled out when neither the peptide substrate GPA nor the serine proteinase inhibitor AEBSF had any inhibitory effect on the adhesion of lung-metastatic breast cancer cells to DPP IV-coated dishes (Fig. 2). The observed DPP IV binding characteristics were not the result of a possible coprecipitation of adenosine deaminase, since rat lung DPP IV preparations were free of detectable adenosine deaminase (30), as determined by both enzyme assay and Western blotting, and since purified, commercially supplied adenosine deaminase (Sigma) did not support adhesion to lung-metastatic breast cancer cells (data not shown).

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Lung-metastatic breast cancer cells adhere to DPP IV-coated wells in a dose-dependent manner. Wells of Microtitration plates (Immulon® 4, Dynatech) were coated overnight at 4 °C with 50, 100, 250, 500, and 1000 ng of purified endothelial DPP IV, 50 µl of PBS, 0.25% OG/well, respectively. The pmol amounts of DPP IV adsorbed to the bottom surface of the wells were calculated as described under "Materials and Methods" and identified in brackets beneath the ng amounts of DPP IV contained in the coating solution. DPP IV-coated wells were seeded with 3 × 104 tumor cells (R3230AC-MET (); R3230AC-LR (); MTF7 (); RPC-2 ()) per well, and an adhesion assay was performed as described in Ref. 9. Means ± S.D. are shown from six different experiments.

View larger version (28K): [in this window] [in a new window]   Fig. 2.   Adhesion of breast cancer cells to DPP IV is inhibited by anti-DPP IV mAb 6A3 but not by the DPP IV enzyme substrate GPA and the serine proteinase inhibitor AEBSF. Microtitration wells were coated with 500 ng of DPP IV, 50 µl of PBS, 0.25% OG/well (overnight at 4 °C); incubated with 50 µg/ml mAb (6A3 and 6D3), 20 mM GPA or GGA, or 5 mM AEBSF in PBS (30 min at 37 °C); and then seeded with 3 × 104 tumor cells and incubated for 20 min at 37 °C. Antibodies, GPA, and GGA were kept in the assay medium throughout the adhesion assay, while AEBSF was removed before tumor cells were added to the wells. Control adhesion assays were performed in PBS alone. Coating and adhesion assay was conducted as described in Ref. 9. Means and S.D. are shown from three experiments. , p < 0.01.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Soluble DPP IV competitively inhibits adhesion of lung-metastatic breast cancer cells (R3230AC-MET; MTF7) to DPP IV-coated dishes. Microtitration wells coated as described in the legend to Fig. 2 were seeded with 3 × 104 tumor cells that had been preincubated for 30 min at 37 °C in various concentrations of purified DPP IV (0, 10, 50, 100, or 200 µg/ml in PBS containing 0.05% OG) and then washed three times with PBS. Control experiments were performed with tumor cells preincubated with glycophorin also dissolved in 0.05% OG in PBS. The adhesion assay was performed as described in Ref. 9. Means and S.D. values are shown from three experiments. Adhesion values between DPP IV- and glycophorin-treated cancer cells (10-200 µg/ml) were statistically significantly different (p < 0.01). , MTF7 and DPP IV; , MTF7 and glycophorin; , R3230AC-MET and DPP IV; , R3230AC-MET and glycophorin.

DPP IV-transfected HEK293 Cells Form Rosettes with Breast Cancer Cells-- Adhesion of lung-metastatic MTF7 breast cancer cells to recombinant DPP IV was tested in a rosette assay, using MTF7 breast cancer cells and HEK293 cells transiently transfected with DPP IV cDNA (DPP IV-HEK293). MTF7 cells were allowed to spread for 12 h on a tissue culture plastic surface and labeled with the green fluorescent dye Calcein, then seeded with DPP IV- or mock-HEK293 cells tagged with the red fluorescent dye SNARF-1. DPP IV-HEK293 cells adhered in large numbers to MTF7 cells, forming multicellular aggregates (rosettes) of six or more cells around 32 ± 4% (S.D.) of the MTF7 cells. Most DPP IV-HEK293 cells adhered to the MTF7 cell body (Fig. 4A), but individual or rows of DPP IV-HEK293 cells were also bound along slender cytoplasmic processes of MTF7 cells. In contrast, mock-transfected HEK293 cells formed rosettes with only 2 ± 1% (S.D.) of the MTF7 breast cancer cells and, thus, were mostly removed from the dishes during the washing procedure (Fig. 4B). Adhesion between MTF7 and DPP IV-HEK293 cells correlated well with the amount of surface expression of recombinant DPP IV on HEK293 cells as assessed by FACS (Fig. 4C).

View larger version (55K): [in this window] [in a new window]   Fig. 4.   DPP IV-transfected HEK293 cells form rosettes with MTF7 breast cancer cells. MTF7 cells were prepared for a rosette assay with DPP IV- or mock-transfected HEK293 cells as described under "Materials and Methods." A, DPP IV-transfected HEK293 cells form typical, multicellular rosettes of six or more cells around MTF7 breast cancer cells attached and spread on a plastic tissue culture surface. B, mock-transfected HEK293 cells are unable to adhere to MTF7 cells. Magnification is × 300. C, FACS analysis of mock- (open area) and DPP IV-transfected (shaded area) HEK293 cells.

Tumor Cell Surface-associated FN Is Identified as the Ligand for DPP IV-- MTF7 cancer cells, which produced the highest DPP IV adhesion and lung colonization values of the three lung-metastatic breast cancer cell lines, were used in the isolation and purification of the tumor cell ligand of endothelial DPP IV. Hence, extracts from metabolically labeled MTF7 cells, grown to approximately 70% confluence in RPMI 1640 medium containing 10% FN-free FBS, were precipitated with Affi-Gel 10-immobilized DPP IV. Upon SDS-PAGE, the DPP IV precipitate resolved as two high molecular weight (HMW) protein bands. The first and major band resided on top of the stacking gel and represented 95% of the DPP IV-precipitable radioactivity (Fig. 5A, lane 1, arrowhead). The second, minor protein band representing the remainder of the DPP IV-precipitable radioactivity was on top of the running gel (Fig. 5A, lane 1, double arrow). Both of these protein bands were reduced with BME to a single protein band of approximately 230 kDa that contained the sum of the radioactive counts present in the two HMW bands of the nonreducing gel (Fig. 5A, lane 2). The DPP IV-precipitated, labeled proteins were associated with the cell surface, since almost no DPP IV precipitate was obtained from tumor cells that had been treated with -chymotrypsin prior to extraction (Fig. 5A, lanes 3 and 4). The exclusive composition of the DPP IV-precipitated HMW complexes by a 230-kDa protein was also demonstrated when DPP IV precipitates from surface-biotinylated MTF7 cell extracts were analyzed (data not shown).

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Breast cancer cell surface-associated FN is composed of intrinsic and extrinsic FN that is precipitable with DPP IV. A, MTF7 cells were metabolically labeled with [35S]methionine in RPMI 1640 medium in 10% FN-free FBS. Prior to extraction in lysis buffer, tumor cells were treated without (lanes 1 and 2) or with (lanes 3 and 4) 10 µg/ml -chymotrypsin (30 min; 37 °C; cell viability >95%). Extracts were precipitated with Affi-Gel 10-immobilized DPP IV, and the precipitates were electrophoresed (5% polyacrylamide) under nonreducing (lanes 1 and 3) or reducing (lanes 2 and 4) conditions and visualized by autoradiography. B, immunoblots were prepared by cutting DPP IV-precipitated protein bands from the 5% polyacrylamide gel run under nonreducing conditions (lane NR, bands 1 and 2). Proteins were extracted from each of the two bands as described under "Materials and Methods," reelectrophoresed under reducing conditions (lane number corresponds to band number), and then subjected to autoradiography (I) and Western analysis with anti-FN antiserum (1:200 in PBS) (II). Both autoradiograph and Western blot reveal a single band in the FN monomer position. Although no signal was recorded in the nonreduced gel (lane NR), cuts were also made at the calculated dimeric (*3) and monomeric (*4) positions of FN, and the excised gel sections were processed for autoradiography and Western analysis in a manner identical to the HMW bands. C, a single-cell suspension of MTF7 cells was incubated for 4 h (lanes 1-4) or 12 h (lanes 5 and 6) at 37 °C in FN-rich medium (10 µg/ml pFN, 1 µg/ml 125I-pFN, 10% FN-free FBS in RPMI 1640 medium) to allow incorporation of 125I-pFN into the tumor cell surface coat and then extracted with 2% DOC (30 min; 4 °C) (lanes 1-4) or Nonidet P-40 lysis buffer (lanes 5 and 6). SDS-PAGE (5% polyacrylamide) was performed with the DOC-soluble fraction precipitated with immobilized DPP IV (lanes 1 and 2), the DOC-insoluble fraction (lanes 3 and 4), and the Nonidet P-40 extract precipitated with immobilized DPP IV (lanes 5 and 6). Odd numbered lanes, nonreducing; even numbered lanes, reducing; M, FN monomer; D, FN dimer; closed arrowhead, HMW FN on top of stacking gel; *, band positions calculated from standards.

Conditions Mimicking Hematogenous Spread Promote pFN Incorporation into the Surface Coat of Metastatically Competent Breast Cancer Cells and Enhance Adhesion to DPP IV-- The possibility that blood-borne breast cancer cells use pFN to augment their FN surface coat and, by this action, increase their capability of adhering to DPP IV was tested under conditions that mimicked hematogenous dissemination of tumor cells. A single-cell suspension of MTF7 cells was prepared and incubated in pFN-rich medium (10 µg/ml pFN, 1 µg/ml ¹²⁵I-pFN, and 10% FN-free FBS in RPMI 1640) for a time period that was equivalent to the presumed time that cancer cells normally spend in circulation before entering and colonizing a secondary organ (up to 4 h). Over the 4-h incubation period in the FN-rich medium, MTF7 cells accumulated significant amounts of ¹²⁵I-pFN on their surfaces that could be harvested as DOC (2%)-soluble and DOC-insoluble fractions. The DPP IV-precipitate from the DOC-soluble fraction consisted primarily of dimeric FN and a small amount of HMW FN residing on top of the polyacrylamide stacking gel. Both of these FN forms reduced to monomeric ¹²⁵I-pFN in the presence of 2% BME (Fig. 5C, lanes 1 and 2). In contrast, the DOC-insoluble fraction consisted of prominent HMW (top of stacking gel) and dimeric ¹²⁵I-pFN bands. Although much of these materials were again reducible to monomeric FN, a significant portion of the HMW FN was resistant to reduction with 2% BME (Fig. 5C, lanes 3 and 4). With continued exposure of breast cancer cells to soluble pFN beyond the 4-h period, the HMW FN fraction precipitated with immobilized DPP IV from Nonidet P-40 extracts of MTF7 cells became more and more prominent, while the dimeric FN fraction decreased (Fig. 5C, lanes 5 and 6). This HMW FN fraction also became increasingly more resistant to reducing agents and was largely nonreducible after 12 h of incubation with ¹²⁵I-pFN (Fig. 5C, lanes 5 and 6). Identical results were obtained for the other lung-metastatic breast cancer cells used in this study (R3230AC-MET; RPC-2), while ¹²⁵I-pFN incorporation into the surface coat of nonmetastatic breast cancer cells (R3230AC-LR) was minimal (data not shown).

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Immunocytochemistry with anti-FN antiserum reveals pFN incorporation into surface-associated FN globules of MTF7 cells in suspension culture. A, cells freshly trypsinized and recovered in 10% FN-free FBS in medium (1 min); B, cells incubated for 4 h in medium plus 10% FN-free FBS; C, cells incubated for 4 h in medium, 10% FN-free FBS, 10 µg/ml pFN. Magnification is × 300.

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Quantification of cell surface-associated FN by FACS using anti-FN antiserum. Lung-metastatic (MTF7; R3230AC-MET) and nonmetastatic (R3230AC-LR) breast cancer cells grown for 4 h in medium, 10% FN-free FBS, 10 µg/ml pFN were stained with anti-FN antiserum and processed for FACS analysis as described under "Materials and Methods." Histograms are from breast cancer cells stained with preimmune serum (open area) and anti-FN antibodies (shaded area), respectively. A representative experiment is shown (n = 2).

View larger version (50K): [in this window] [in a new window]   Fig. 8.   Anti-FN antibodies inhibit the binding of rat breast carcinoma cells to DPP IV. A tumor cell adhesion assay was performed in DPP IV-coated plates (see Fig. 2) in the presence of anti-FN antiserum (1:50 and 1:100) or normal rabbit serum (1:50), as described under "Materials and Methods." Tumor cells used were MTF7 and R3230AC-MET. Means and S.D. were from three experiments. , p < 0.01. Dotted bar, tumor cells alone; shaded bar, with nonimmune rabbit serum (1:50); dashed bar, with rabbit anti-FN antiserum (1:100); open bar, with rabbit anti-FN antiserum (1:50).

The DPP IV Binding Specificity for Immobilized FN Explains Tumor Cell Adhesion to DPP IV in the Presence of Excess Soluble pFN-- If this newly discovered binding interaction between endothelial DPP IV and cell surface-associated FN is effective in causing vascular arrest of blood-borne breast cancer cells in the lungs, it must happen in an environment that is rich with soluble pFN (normal blood plasma pFN concentration: 300 µg/ml (31)). Hence, we tested the effect of increasing concentrations of soluble pFN on the adhesion of breast cancer cells to DPP IV in solid-state adhesion assays. None of the pFN test concentrations had any inhibitory effect on the tumor cell binding (Fig. 9). On the contrary, a slight increase in the tumor cell adhesion of both MTF7 and R3230AC-MET cancer cells to DPP IV was observed at higher pFN concentrations, suggesting ongoing incorporation of pFN molecules into the tumor cell surface-associated FN coat during the adhesion assay. These data reflect an inability of DPP IV to recognize and bind to "conformationally inadequate" pFN in solution. This select binding behavior of DPP IV was further exemplified when soluble pFN failed to inhibit the binding interaction between purified DPP IV and immobilized FN in an in vitro enzyme-linked immunosorbent assay (Fig. 9, inset).

View larger version (30K): [in this window] [in a new window]   Fig. 9.   Soluble pFN does not inhibit breast cancer cell adhesion to DPP IV and is not recognized by DPP IV. Lung-metastatic MTF7 () and R3230AC-MET () breast cancer cells were tested for their ability to bind to DPP IV-coated plates (see Fig. 2) in the presence of various concentrations of soluble rat pFN (0, 1, 5, and 10 µg/ml PBS). There was no inhibitory effect recorded. Inset, soluble pFN (10 µg/ml) fails to inhibit the in vitro binding interaction between purified endothelial DPP IV (0, 0.1, 0.3, 0.5, 1.0, and 2.0 µg/ml) and immobilized rat pFN (5 µg/ml). Shown is an enzyme-linked immunosorbent assay with rabbit anti-rat DPP IV antiserum CU31 (1:500).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Dipeptidyl peptidase IV (DPP IV; CD26) is a serine exopeptidase that was originally isolated and cloned from rat kidney (27) but is now recognized in a variety of tissues including capillary endothelia of the lungs (13, 32, 33). It is a transmembrane sialoglycoprotein that anchors to the plasma membrane by a hydrophobic domain near its N terminus such that the bulk of its molecular mass is exposed to the outside of the cell (27, 34). While this glycoprotein has been extensively investigated with respect to its enzyme and T-cell activation activities (reviewed in Ref. 14), little has been published on its adhesion properties although it is recognized as a collagen- and FN-binding protein (16, 35, 36). Here, we show that the FN binding property of DPP IV is responsible for the adhesion of lung-metastatic breast cancer cells, mediating lung vascular arrest and lung metastasis by these cancer cells. The binding interaction between tumor cell-surface associated FN and DPP IV occurs independently of the exopeptidase substrate domain of DPP IV but appears to be critically dependent upon the conformation of the FN substrate. Extensive DPP IV precipitation studies performed on extracts of breast cancer cells with metastatic and nonmetastatic phenotypes consistently show that the preferred FN form precipitated by immobilized DPP IV is cell surface-associated multimeric and dimeric FN. Evidence for this DPP IV/FN binding preference includes (a) the direct correlation between the number of lung-metastatic breast cancer cells that were able to bind to DPP IV-coated dishes and the number of tumor cells that expressed prominent FN globules on their surface; (b) the precipitation of various cell surface-associated FN forms from extracts of lung-metastatic breast cancer cells with immobilized DPP IV; (c) the direct interaction between soluble DPP IV and immobilized FN in vitro; (d) the formation of rosettes between lung-metastatic breast cancer cells expressing numerous FN globules on their surface and DPP IV-transfected HEK293 cells; (e) the specific inhibition of adhesion of lung-metastatic breast cancer cells to dishes coated with DPP IV by monospecific anti-DPP IV mAb 6A3, polyclonal anti-FN antisera, and immunopurified DPP IV; and (f) the inhibition of the DPP IV-mediated breast cancer cell adhesion and lung colonization after masking the DPP IV-binding sites on tumor cell surface-associated FN with soluble DPP IV.

The binding of DPP IV to immobilized FN and the inability of DPP IV to recognize soluble pFN implies that the DPP IV binding site is inaccessible in soluble pFN but becomes available when pFN binds to the cancer cell surface, uncoils, and subsequently associates with other pFN molecules in forming linearized, supermolecular aggregates (31, 37, 38). Uncoiling and linearization of the FN structure and the associated gain in DPP IV binding avidity is reminiscent of the increasing binding avidities of collagen and heparin for their respective FN binding sites with progressive deletion of the FN peptide strand from the C to the N terminus (39) and, as reported more recently, of the binding of the III-1 FN peptide to the truncated (III-10A) but not the complete III-10 FN peptide (40). The process of uncoiling and linearization of pFN may occur under a variety of in vivo and in vitro conditions including the binding of pFN to cell surfaces (31, 41), gelatin-conjugated agarose beads (39), and even plastic surfaces (42)³ and seems to proceed by the opening of the pFN arms from a V-shape to a more extended form that requires little energy and therefore can happen relatively easily (37, 43). In the absence of uncoiling, the FN V-shape may obscure the DPP IV-binding site, thereby making it impossible for DPP IV to interact with pFN in solution, even under the most sensitive assay conditions. For cancer metastasis, these binding properties of DPP IV imply that cancer cells expressing abundant uncoiled FN on their surface can dock to DPP IV-expressing lung endothelia although cancer cells are bathed during their hematogenous dissemination in high concentrations of blood pFN (300 µg/ml (31)).

The molecular basis of the initial immobilization of FN on the surface of breast cancer cells is poorly understood. It appears that rat breast cancer cells growing as a solid tumor mass in syngeneic animals support the synthesis of a modest FN-containing matrix.⁴ As tumor cells escape the confinement of the primary mass, they enter neighboring blood vessels, thereby becoming blood-borne and traveling passively with the blood stream to other organ sites (1, 7). In the pFN-rich environment of the plasma, tumor cells decorated with a sparse coat of cellular FN use this FN scaffold to acquire pFN molecules for the buildup of a prominent FN coat that is visualized as multiple, densely distributed FN globules by immunocytochemistry. The presence of these randomly dispersed FN globules suggests that the initial FN binding to the cell surface occurs around focal adhesion points from which polymerization is then started by FN self-association (40, 44-47). Such adhesion points are proposed to be sites of integrin clusterings, most likely the classic FN receptor 51 (38, 48, 49). Indeed, this receptor is strongly expressed by all lung-metastatic rat breast cancer cells tested⁵ and could promote the initial immobilization of cellular FN molecules on the cancer cell surface. Alternatively, a cellular FN-51 complex could already form intracytoplasmically (e.g. in Golgi vesicles) that upon transport to and incorporation into the plasma membrane could serve as the necessary scaffold upon which self-assembly to supermolecular FN aggregates might occur.

The FN buildup on cancer cell surfaces is comparable with that reported first for normal fibroblasts incubated in vitro with ¹²⁵I-pFN (50); i.e. after a 4-h incubation period of lung-metastatic breast cancer cells with ¹²⁵I-pFN, the total cell surface-associated FN can be harvested as DOC-soluble and DOC-insoluble fractions. Although the 2% DOC extraction used in the present study did not allow a clear partition of the disulfide-bonded FN into the DOC-insoluble fraction as was achieved after extraction of ¹²⁵I-pFN-incubated fibroblasts with 1% DOC (50), the FN multimers observed on MTF7 breast cancer cells clearly increased with time of incubation with ¹²⁵I-pFN as reported for normal fibroblasts and hepatocytes (50, 51). However, the multimeric FN on breast cancer cells and normal hepatocytes, both grown in suspension, gradually converted to nonreducible, seemingly covalently bonded FN complexes (51) that were not observed on anchorage-dependent normal fibroblasts (50). Similar to hepatocytes, this conversion is perhaps mediated by a cancer cell-associated transglutaminase activity (51). The rapid accumulation of HMW FN on breast cancer cell surfaces, which plateaued after only 4 h of incubation with ¹²⁵I-pFN, might be essential for allowing blood-borne cancer cells to become arrested in the lung vasculature, since the large FN aggregates (globules) facilitate binding of multiple endothelial DPP IV molecules, thereby providing an adhesion strength between cancer cell and endothelial cell that can withstand the rigors of hemodynamic shear stresses.

In conclusion, cell surface-associated FN is shown here to mediate lung vascular arrest by binding to endothelial DPP IV. The in vivo validity of this adhesion principle is underscored by a more than 80% competitive inhibition of lung metastasis when cancer cell surface-associated FN is masked by preincubation in a DPP IV solution. Given the ubiquity of cancer cell surface receptors (5, 7) that are able to bind FN on their surfaces, upon which FN self-association to HMW structures could occur during their dissemination in the blood, the DPP IV/FN binding mechanism may be a more frequent event in lung metastasis than currently realized (7). This notion is supported by a previously observed strong association between FN expression on the surface of rat rhabdomyosarcoma cell clones and the ability of these tumor cells to bind to lung endothelium and to colonize the lungs (17) and by a similar association between cell surface expression of globular FN complexes and DPP IV adhesion of Chinese hamster ovary cell variants.³ Finally, the DPP IV/FN binding mechanism appears also to be relevant to lung vascular arrest of blood-borne human breast cancer cells, since various breast cancer cell lines currently investigated in our laboratory have been found to be decorated with FN and to adhere to endothelial DPP IV.