This is a continuation-in-part of application serial number 07/642,971, filed January 18, 1991, which is a continuation-in-part of application serial number 07/457,566 filed December 27, 1989 and application serial number 07/582,063 filed September 14, 1990, all applications pending. The entire contents of these applications are hereby incorporated by reference.

Background of the Invention

The human met protooncogene product (Met or Met protein) a member of the family of tyrosine kinase growth factor receptors, was identified via the activated oncogene tpr-met (C.S. Cooper et al . , Nature 311: 29 (1984); M. Park et al . , Cell 45: 895 (1986)). Met is synthesized as a glycosylated 170-kD precursor and cleaved in the external (ligand binding) domain to yield a mature disulfide-linked α- (50-kD) , β- (140-kD) heterodi er (C.S. Cooper et al . , Nature 311: 29 (1984); S. Girodano et al . , Oncogene 4: 1383 (1989); D.L. Faletto et al . , Oncogene , in press (1992)). The Met receptor is expressed in a wide variety of tissue and cell types, but the highest levels are found in epithelial cells (M.F. Di Renzo et al . , Oncogene 6: 1997-2003 (1991); A. Iyer et al . , Cell Growth & Diff . 1: 87 (1990)).

Hepatocyte growth factor (HGF) was first purified from human and rabbit plasma and rat platelets on the basis of its ability to stimulate mitogenesis of rat hepatocytes (E. Gohoda et al., J. Clin . Invest . 81: 414 (1988); R. Zarnegar and G. Michalopoulos, Cancer Res . 49: 3314 (1989); T. Naka ura et al. FEBS Lett . 224: 311 (1987)) . Thus, HGF may act as a humoral factor promoting liver regeneration after partial hepatectomy or liver injury (G.K. Michalopoulos, FASEB J . 4: 176 (1990)).

The same factor was purified from human fibroblast culture medium and shown to act on elanocytes and a variety of epithelial and endothelial cells (J.S. Rubin et al., Proc . Natl . Acad . Sci . U.S .A. 88: 415 (1990)). Together with evidence of HGF expression in several organs (J.S. Rubin et al., Proc . Natl . Acad . Sci . U .S .A. 88: 415 (1990); K. Tashiro et al. Proc. Natl . Acad. Sci . U.S .A. 87: 3200 (1990); R. Zarnegar et al., Proc . Natl . Acad . Sci . U.S .A. 87: 1252 (1990); T. Kinoshita et al. Biochem . Biophys . Res . Comm . 165: 1229 (1989)), these findings indicate that HGF may also act as a paracrine mediator of proliferation for a broad spectrum of cell types. Molecular cloning of HGF revealed a remarkable structural homology to plasminogen and related serine proteases (J.S. Rubin et al., Proc . Natl . Acad. Sci . U.S.A. 415 (1990); T. Nakamura et al., Nature 342: 440 (1989); K. Miyazawa et al., Biophys . Res . Comm . 163: 967 (1989)) .

HGF is structurally related to the family of serine proteases that includes plasminogen, prothrombin, urokinase, and tissue plasminogen activator (J.S. Rubin et al., Proc . Natl . Acad . Sci . U.S .A . 88: 415 (1990)); T. Nakamura et al. , Nature 342: 440 (1989)). As defined in the present invention, HGF includes a variant of HGF previously characterized as a broad-spectrum mitogen called plasminogen like growth factor (PLGF) . Several proteases, including members of the serine protease family, stimulate DNA synthesis presumably through a proteolytic mechanism similar to tryptic activation of the insulin receptor (S.E. Shoelson et al. J. Biol . Chem . 263: 4852 (1988)). Only urokinase has been found to associate with a specific cell-surface receptor, which itself bears no homology to any known tyrosine kinase receptors (A.L. Roldan et al., EMBO J . 9: 467 (1990)). U.S. Patent Application Serial Number 07/642,971, incorporated by reference above, describes the complex comprising HGF and met protooncogene protein and identifies the met protooncogene as the receptor for HGF.

Scatter factor (SF) originally had been considered to be related to but different from HGF, SF being associated with cell motogenicity (motility) , and HGF being associated with cell mitogenicity (growth) . However, recently it has been shown that HGF is identical to SF, and this factor is now referred to as "HGF/SF" (E. Gerardi et al . , Cancer Cells 3: 227 (1991); E.M. Rosen et al . , Cell Growth & Diff . 2: 603 (1991); L. Naldini et al . , EMBO J. 10: 2876 (1991b); K.M. eidner et al . , Proc . Natl . Acad . Sci . USA 88: 7001 (1991)) and was independently shown both to promote epithelial cell motility (scattering) and to cause certain epithelial cell lines to become invasive in in vitro assays (E.M. Rosen et al .. Cell Growth & Diff . 2: 603 (1991); M. Stoker et al . , Nature 327: 239 (1987)).

The scattering response of HGF/SF establishes that this factor is responsible for cell motility and differentiation (G.F. Vandewoude Japan . J. Can . Res . 83: cover (1992)). For instance, MDCK cells grown in collagen gels in the presence of HGF/SF form bracing tubules, suggesting that the three-dimensional geometry of cell-substrate interactions directs MDCK cells to organize into tubules in response to HGF/SF (R. Montesano et al . , Cell 67: 901 (1991)). It is known that mammary gland epithelia undergo developmental changes during pregnancy to become secretory. In culture, mammary epithelial cells can regain their differentiated phenotype only when appropriate hormonal and substratum conditions are provided (M.J. Bissell et al . , The Mammary Gland, M. Neville and C. Daniel, Eds. (Plenum Press Publishing Corp., New York, pp. 97-146 (1987); CH. Streuli et al . , J . Cell Biol . 115: 1383 (1991)). Thus, a need existed to determine whether there was met expression in normal breast duct cells and to determine the role that HGF/SF and met play in the differentiation of ductal epithelium in the mammary gland.

In this connection, a need also existed to determine whether there was a general phenotype in epithelial cancers arising from organs that normally involve met expression for differentiation. The work of Bieche et al . , The Lancet , 339: 139 (1992) has shown that the loss of heterozygosity on chromosome 7q is associated with aggressive primary breast cancer. Specifically, Bieche et al . used the c-met proto-oncogene probe, which detects sequences on chromosome 7q31, to analyse tumor and blood leucocyte DNA samples from 245 patients with primary breast cancers. The pmet H polymorphic probe detected a high frequency of allele loss (40-50%) among the 121 informative (heterozygous) patients. This genetic alteration was not significantly associated with standard prognostic features including tumor size, histopathologic grade, and lymph-node or steroid receptor status. However, patients with loss of heterozygosity on chromosome 7q31 in primary tumor DNA had significantly shorter metastasis-free survival and overall survival after surgery than patients without this alteration. Based upon this observation, Bieche et al. hypothesized that the region on chromosome 7 detected by the probe might be the site of a breast cancer or metastasis suppressor gene. Bieche et al . did not identify a specific genetic locus responsible for poor prognosis in breast cancer patients.

Thus, one object of the present invention is to provide a method for predicting the progression of breast cancer based upon the presence of Met protein in normal and breast cancer tissue. Such a method would be advantageous in the management of breast cancer therapy, used either alone or in conjunction with other prognostic features such as tumor size, histo- and cytopathological grade and lymph node or steroid receptor status.

Summarv of the Invention

The present invention relates to a method for predicting breast tumor progression. This method involves contacting a section from a breast tumor with an antibody reagent specific for Met protein under antibody binding conditions, wherein the section contains normal breast duct tissue and tumor tissue; determining the binding of the reagent to Met protein in the normal tissue and the tumor tissue; and comparing the binding of the reagent to met protein in the normal tissue with the binding in the tumor tissue; wherein, greater binding of the reagent to normal than to tumor tissue indicates a high likelihood of tumor metastasis.

Brief Description of the Drawings

Figure 1 shows the expression of Met in cells bordering lumen-like structures of the breast cancer cell line T47D. T47D cells were treated with Bouin\'s fixative, embedded in paraffin, serial sectioned, and stained by indirect immunofluorescence. Cells were analyzed using confocal laser scanning microscopy (CLSM) . Figure IA shows a Nomarski image of the lumen-like structures in T47D cells in culture.

Figure IB shows immunofluorescent staining of the same region as in Figure IA with Met C28 antibody. Figures IA and IB are X2.45 magnification of the lower right hand portion of Figures IA and IB. Overall magnification is X490.

Figure 1C through IE show quantitative determination of the Met-specific immunofluorescence. The histograms show the distribution of fluorescence of the bracketed areas.

Figure 1C specifically shows the fluorescence in the apical region of the lumen for the right lumen.

Figure ID specifically shows the fluorescence in the apical region of the lumen for the upper left lumen.

Figure IE specifically shows the fluorescence in the apical region of the lumen for an area adjacent to the upper left lumen.

Figure 2 shows subcellular localization of Met in T47D cells. Immunoelectron microscopy was performed using secondary antibody labeled with 10-nm colloidal gold particles (magnification 30,OOOX) (For postembedding immunoelectron microscopy, fixed cells were embedded in L.R. gold resin (Electron Microscopy Science) at -25°C, sectioned with an LKB Nova Ultratome, and picked up with Formvar-coated 200 mesh gold grids. The grids were washed three times in PBS 10 in and incubated in 1% BSA (in PBS) for 2 hours and in rabbin anti-human C28 Met antibody (diluted 1:50 in 1% BSA) at room temperature for 1 hour. Controls were incubated either in the presence of C28 competing peptide or in the absence of the primary antibody. The grids were washed again in PBS, incubated in RPM11640 medium for 20 min in 1% BSA as above, and reacted with goat anti-rabbit IgG gold (10-nm diameter; 1:10 diluted in 1% BSA; Amersham) at room temperature for 1 hour. The grids were finally washed in PBS and distilled water and stained with uranyl acetate and lead citrate. The sections were observed and photographed using a Philips EM 410 electron microscope.). Met is localized in microvilli that protrude into lumen (L) and could account for the apical staining observed in cells bordering the lumen (A and B) .

Figure 3 shows expression of Met in breast tissue biopsies. A paraffin-embedded section of biopsy from breast adenocarcino a was subjected to indirect immunofluorescence staining with C28 antibody or monoclonal antibody 4G10. The cells and immunofluorescence were analyzed by CLSM.

Figure 3A shows immunofluorescence of normal mammary duct, performed with C28. Magnification X250.

Figure 3B shows immunofluorescence of normal mammary duct, performed with anti-P-Tyr. Magnification X250.

Figure 3C shows immunofluorescence of normal mammary duct, performed with C28. Magnification X250.

(4G10) ; (F) analysis of the overlap between the Met and the anti-P-Tyr staining, using the confocal image processing program (red indicates increased overlap) .

Figures 4A-F relate to expression of Met in the mouse embryonal digestive tract. Eleven-day-old BALB/c mouse embryos were fixed, embedded in paraffin, and subjected to immunofluorescence staining with SP260 and 4G10 antibodies. Magnification is X220.

Figure 4A shows apical staining of the lumen of the digestive tract with SP260.

Figure 4B shows staining in the absence of the primary SP260 antibody. Figure 4C shows staining in the presence of SP260 competing peptide.

Figure 4D shows apical staining of the lumen of the digestive tract with SP260.

Figure 4E shows immunofluorescence with 4G10. Figure 4F shows an analysis of the overlap between the Met and the anti-P-Tyr staining by a confocal image processing program.

Figures 5A-F relates to HGF/SF induction of lumen formation in S 480 and HT29 human colon carcinoma cell lines. The cells were treated with the specified concentration of HGF/SF when they were -70% confluent.

Magnification is X164.

Figure 5A shows S 480 cells grown for 18 hours in complete medium without HGF/SF. Figure 5B shows S 480 cells grown for 18 hours in complete medium with HGF/SF ^■ -.\' ^■ :. 1 ng/ml.

Figure 5C shows SW480 c-sils grown for 18 hours in complete medium with HGF/SF at 5 ng/ml.

Figure 5D shows S 480 cells grown for 18 hours in complete medium with HGF/SF at 10 ng/ml.

Figure 5E shows SW480 cells grown for 18 hours in complete medium with HGF/SF at 100 ng/ml.

Figure 6 shows the influence of HGF/SF concentration on lumen formation. The number of cells forming lumen structures was calculated for each HGF/SF concentration.

Detailed Description of the Invention

Applicants have discovered that Met protein and RNA products are prognostic in evaluating tumor progression towards metastasis in human breast cancer. The expression of met in certain organs is associated with epithelial cell differentiation in tumor cells. More specifically, applicants have discovered that expression of the met protooncogene product in normal tissues is associated with the formation of highly ordered, fully differentiated lumen structures and that the ligand for Met, HGF, plays a key role in inducing the formation of these structures. It also has been discovered that while met expression is very high in cells that form normal breast ducts, expression of met in adjacent, poorly differentiated, cancer tissue is markedly reduced. This information has led to the further discovery that reduced met expression correlates with poor prognosis in human breast cancer.

The method of the present invention involves comparing the amount of Met in normal breast tissue with the amount of Met in tumor tissue. This is accomplished by determining one or more of met DNA abundance, met mRNA abundance, or Met protein abundance in normal breast tissue and breast tumor tissue from the same patient. The detection of a higher abundance of met DNA, met mRNA or Met protein in normal tissue than in tumor tissue is indicative of a poor prognosis.

In one embodiment of the present invention, the method involves contacting a sample of breast tissue containing both normal and tumor cells with an antibody reagent specific for Met protein, under antibody binding conditions. Greater binding of the reagent to normal

than to tumor tissue indicates a high likelihood of tumor metastasis and a poor prognosis.

Preferably, the breast tissue sample is a "section," i.e., a histological section obtained from a surgical or biopsy procedure, according to techniques well known in the art. The term "contacting" is intended to encompass any technique which permits the antibody reagent to bind with Met protein.

By antibody reagents is meant whole antibodies and parts thereof, either alone or conjugated with other moieties. Antibodies include polyclonal antibodies, monoclonal antibodies, and single chain antibodies. Antibody fragments are those the bind the Met protein, including Fab and F(ab) ₂ fragments, inter alia . The antibody reagents may comprise antibodies made in animals or by recombinant DNA techniques. Also the antibody reagents include antibody and antibody fragments conjugated to, among other moieties, detectable labels, such as enzymatic labels and fluorescent labels. Other useful labels which may be comprised by the antibody reagents include radionuclides.

Antibody binding conditions generally are well known in the art and, for the most part will include neutral pH, moderate salt, temperatures between 2-3 C and 37° C, incubation times between several minutes and overnight or longer. Preferred conditions include those described by M. Gonzatti-Haces et al . , Proc . Natl . Acad . Sci . USA 85: 21 (1988) for a rabbit antibody prepared against the extracellular domain of the Met peptide, those of M. Park et al . , Proc . Natl . Acad . Sci . USA 84: 6379 (1987) for a rabbit anti-mouse Met and the conditions described by D.K. Morrison et al . Cell 58: 649 (1989)) for a mouse monoclonal antibody to phosphorylated tyrosine (P-Tyr) . It will be readily appreciated by those of ordinary skill in the art that the details of an antibody binding procedure may be adjusted to favor improved signal to noise ratios or sensitivity, rapidity or completeness, and the like. Moreover, conditions may be adjusted to

accommodate different histological procedures for fixation and staining. Finally, it will be appreciated that titers and appropriate dilutions will be different for different antibody reagent preparations. Binding of the antibody reagent with Met protein may be determined by any method known to the skilled artisan and may include visualization by conventional or confocal laser scanning microscopy.

The term "greater binding" is readily understood to be a relative term which derives meaning by comparing the amount of bound Met in normal with the amount of bound Met in tumor cells. An 80- to 150-fold or greater reduction in Met in tumor cells is considered predictive of a poor prognosis. hether a prognosis is good or poor depends upon the likelihood of metastasis. A "high likelihood" of tumor metastasis means that there is about an 80-90% chance of tumor metastasis within about two years.

In the preferred embodiment of the present invention, a histological tumor breast biopsy section containing both normal duct tissue and adjacent tumor tissue is embedded in paraffin and blocked for about 10 minutes using a commercially available immunostaining blocking reagent. A primary antibody is then incubated with the tissue. The preferred primary antibody is selected from the group consisting of C28 (rabbit anti-human Met) , C200 (rabbit antibody against the extracellular domain of the Met peptide) , SP260 (rabbit anti-mouse Met) , 4G10 (mouse monoclonal antibody to phosphorylated tyrosine (P-Tyr) and the 19S monoclonal (previously described in U.S. patent application serial no. 07/457,556) . The tissue is then further incubated with a secondary antibody coupled to a fluorochrome. The secondary antibody may be donkey anti-rabbit coupled to phycoerythrin or donkey anti-mouse coupled to fluorescein isothiocyanate. Antibody binding is determined by examining the fluorochrome labeled cells with a conventional light or confocal laser scanning microscope, such as the Zeiss laser scan microscope. The

fluorescent intensity is then quantitated using methods known in the art. In the preferred embodiment, the relative fluorescence intensity (the intensity of the normal compared with tumor tissue) is quantitated with an Indec Laser Scan Microscope Image Processor Option. Optionally, photographs may be taken to document visualized tissue. A poor prognosis is indicated when an 80- 150-fold reduction in Met in the tumor tissue as compare with normal tissue is observed. In yet another embodiment of the present invention, the abundance of met mRNA or met DNA in a breast tissue sample may be detected by in situ hybridization using met sequence specific probes, or by hybridization of met sequence specific probes to mRNA or DNA from normal and tumor tissue. Probe complementary to met are prepared by methods known in the art and allowed to hybridize to mRNA or DNA within a section of a tissue sample (either embedded by standard techniques, e.g. paraffin, or otherwise preserved) . Unhybridized probe are removed by nuclease. Hybridization can be detected by autoradiography or other methods. The intensity of hybridization reflects the amount of met within the cells of the tissue. The observation of tumor tissue containing reduced levels of met relative to levels in normal tissue, indicates a poor prognosis.

In yet another embodiment of the present invention, the polymerase chain reaction ("PCR") is used to detect met DNA or mRNA in a breast tissue sample. To carry out the PCR, a pair of met sequence specific primers is employed, which hybridize to opposite strands of the met gene at offset positions on the double helix. The primers provide initiation points for DNA synthesis.

In the presence of DNA polymerase, the four nucleotide triphosphates ("NTPs") , and other necessary co-factors, which are all well known to the art, new DNA strands are synthesized complementary to the templates which hybridized with the primers. Several rounds of synthesis are carried out, the double stranded products

being denatured between rounds. Preferably, a thermal stable DNA polymerase is used so that it is not necessary to add enzyme anew for each round of synthesis.

The PCR produces a double stranded DNA amplification product which has the same sequence as the original stretch of the met DNA defined by the ends of the primer pair sequences. The amount of PCR product indicates the amount of met DNA or met mRNA in the sample. The product can be detected by a variety of methods well-known in the art.

Where the products are produced in a test tube, or the like, they can be resolved by agarose or polyacrylamide electrophoresis and detected by fluorescence staining, such as ethidium bromide. Alternatively, one of the NTPs may be labelled and the PCR products may be determined by measuring incorporation of the labelled NTP. A variety of other methods for resolving, detecting and measuring the amount of PCR product are well-known to the art that are suitable for use in the present invention.

In situ techniques may employ the use of fluorescent and radiolabels which can be easily quantitated by fluorescence microscopy or autoradiography, respectively. Generally, fluorescent labels will be preferred. Another labelling technique may employ enzymatic tags which generate readily quantifiable colorimetric or chemiluminescent signals.

PCR may be rendered specific for met DNA or met mRNA in in situ and in liquid PCRs. For instance, RNAse or DNAse may be used to remove one template or the other from the sample, and the use of primers that distinguish between the gene and the message (e.g. a primer that hybridizes to a sequence in the untranscribed region of the promoter will be gene specific) . In accordance with the methods of the present invention, PCR is used to detect the presence of met DNA or met mRNA in both normal and tumor tissue. Therefore, when liquid PCR is employed, it is essential

to distinguish samples containing normal and tumor tissue. Reduced amounts of met DNA or met mRNA in tumor, relative to normal tissue, is predictive of tumor metastasis. In another embodiment, RNA ("Northern") blotting may be used in the methods of the invention. According to this method, RNA is isolated from tumor and normal breast tissue by any of a number of standard procedures. (Lehrach, H. , Biochemistry, 16: 4743 (1975)) Again, it is important that separate assays be run on the tumor and normal breast tissue so that respective hybridization results can be compared. RNA is subjected to denaturing gel electrophoresis and transferred to nitrocellulose or other support matrix. The met mRNA can be detected by hybridization of radioactively or non- radioactively labelled met . The presence of met mRNA is detected by the intensity of hybridization. The intensity observed in normal tissue is compared with that in tumor tissue; a reduced amount of intensity in tumor tissue is indicative of poor prognosis.

Other techniques suitable to the claimed methods are readily apparent to the skilled artisan and include Nuclease Protection Assays, ELISA and Western blotting, as described in Example 1. Although the present invention is exemplified with a method of predicting the progression of breast cancer, the claimed methods are likewise applicable to predicting the progression of any tumor associated with tubule forming epithelial tissues, such as those found in the digestive tract.

Although the method of the present invention involves the detection of met DNA, met mRNA or Met, the detection and comparison of HGF/SF DNA, mRNA or protein in normal and tumor tissue is likewise predictive of tumor progression, in accordance with the teachings of the present application.

It is noted that unless defined otherwise, all technical and scientific terms used herein have the same

meaning as commonly understood by one of ordinary skill in the art of the invention. Although any methods and materials similar or equivalent to those described herein can be used in the practice of the invention, the preferred methods and materials have been described. Unless mentioned otherwise, the techniques employed or contemplated herein are standard methodologies well known to one of ordinary skill in the art. The materials, methods and examples are illustrative only and not limiting.

Example 1

Expression of Met in cells and tissues of both human and mouse origin. Immunoprecipitation and Western analyses showed that met is expressed in human breast carcinoma cell line T47D (I. Keydar et al., Eur. J. Cancer 15: 659 (1979)) Applicants then examined the distribution of Met in T47D cells by immunofluorescence and confocal laser scanning microscopy (CLSM) , using antibody C28 (M. Gonzatti-Haces et al., Proc . Natl . Acad . Sci . USA 85: 21 (1988)) or C200 rabbit antibodies directed against a C- terminal or an N-terminal peptide of Met, respectively, as shown in Figure 1. Controls were prepared with the C28 antiserum in the presence of competing peptide or without the C28 antiserum. It is known that T47D cells in suspension form lumen¬ like structures resembling mammary ducts (I. Keydar et al . , Eur . J . Cancer 15: 659 (1979)). Analyses of T47D cells in paraffin sections stained with C28 antibody showed intense fluorescent staining in cells bordering lumen-like structures, as shown in Figure IA. A marked decrease in intensity was observed when competing C28 peptide was added to the primary antibody or when the primary antibody was omitted. Moreover, T47D cells stained with the C200 antibody, directed against the met extracellular domain, also showed intense fluorescent staining in cells bordering the lumen-like structures, as

shown in Figure IA. By CLSM, applicants observed an 80- fold greater fluorescence in cells forming the lumen border than in adjacent cells, as shown in Figure IB.

Applicants further investigated the subcellular localization of Met by electron microscopy using indirect immunogold labeling and the C28 Met antibody. Through this analysis, applicants discovered that Met was localized to microvilli that protrude into lumen, as shown in Figure 1C. The intense met-specific staining of cells lining the lumen borders in T47D breast carcinoma cells led the applicants to examine met expression in normal and abnormal human breast tissue. Specifically, applicants examined fifty human breast carcinoma biopsies by CLSM using the C28 Met antibody. The normal and tumor tissue from a representative breast biopsy is shown in Figure 2. Applicants observed intense Met fluorescence in cells that form a normal mammary duct, as shown in Figure 2A. The intensity of staining was 80-fold greater in the duct-forming cells than in the adjacent nonductal cells. The pattern of Met staining in the mammary duct was similar to, but higher than, that observed in the T47D lumen-like structures. Applicants further discovered that, in contrast, Met staining in adjacent breast tumor tissue was always reduced, but even in the disrupted architecture of the tumor, met fluorescence was evident in lumen- or duct-like structures, as shown in Figure 2C.

It is known that Met is rapidly phosphorylated in tyrosine residues after HGF/SF activation (D.P. Bottaro et al . , Science 251: 802 (1991); L. Naldini et al., Oncogene 6: 501 (1991)) . Applicants have further observed that anti-phosphotyrosine (anti-P-Tyr) monoclonal antibody labeling colocalizes with Met staining in cells activated with the met ligand, HGF/SF. In light of this, applicants costained breast tissue samples with the anti- P-Tyr antibody and found intense anti-P-Tyr fluorescence that colocalized with Met staining in the cells forming

normal breast ducts (Figure 2B) . Applicants conclude from these results that Met and its substrates are activated in the cells bordering the normal breast duct. As with the lower Met staining of tumor tissue (Fig. 2C) , applicants also observed reduced levels of anti-P-Tyr staining in the tumor cells (Fig. 2D) ; however, even the lower levels of anti-P-Tyr fluorescence colocalized with Met staining.

Thus, applicants have also observed that human met expression both in vitro and in vivo is concentrated in cells bordering lumen-like structures. Since extensive lumen and duct formation occurs in the early development of the digestive tract, applicants further investigated met expression in this tissue in mouse embryos. These analyses, performed using a C-terminal rabbit peptide antibody, SP260 (A. Iyer et al . , Cell Growth & Diff . 1: 87 (1990)), showed that in 11-day-old mouse embryos met is expressed in the lumen-bordering cells, as shown in Figure 3. Applicants noted that, as in human samples, intense Met staining was evident on the apical side of the cells forming the lumen of the digestive tract, as shown in Figures 3A and 3D, and was 50-fold higher in these cells than in the surrounding tissue. Serial sections stained in the absence of the primary C-terminal mouse Met antibody or in the presence of competing peptide showed much lower levels of fluorescence, as shown in Figures 3B and 3C. Moreover, as in the cells lining the human breast duct, mouse Met staining colocalized with anti-P-Tyr staining in the lining of the embryonic digestive tract, as shown in Figure 3E. Thus, applicants have discovered that the Met receptor appears to be preferentially expressed in the border cells of the breast duct and the gastrointestinal tract lumen. Similarly, Met staining in lumen-like structures has been observed in human gastrointestinal lumen and lumen of biliary ducts and esophagus (M. Pratt et al . Int . J . Cancer 49: 323 (1991)). The colocalization of Met

staining with anti-P-Tyr immunofluorescence implies that the Met receptor is activated in the lumen-forming cells of the breast duct and gastrointestinal tract, a shown in Figures 2 and 3. Having established a correlation between the expression of Met and lumen formation, applicants tested whether Met-HGF/SF could induce lumen formation in human epithelial carcinoma cells in vitro . Specifically, cells expressing Met were exposed to varying concentrations of HGF/SF (1-100 ng/ml) . More specifically, cells were grown on 16-chamber Labtek slides (Nunc) and treated with specific concentrations of HGF/SF (95% pure, Collaborative Research) for 24 hours. After two washes in PBS, cells were fixed for 10 minutes in cold methanol (-20°C) and washed extensively with PBS. Cells were also stained for 10 minutes with 0.1% methylene blue in PBS and were visualized and photographed using a Zeiss microscope. The results with two human colon carcinoma cell lines, SW480 and HT29, are shown in Figure 4. Indeed, applicants observed dramatic formation of lumen structures in these cells (as well as in breast carcinoma cell lines) when the cells were grown on glass (70% confluent) and treated with HGF/SF. Lumen formation was cell density dependent and not obvious when cells were grown in plastic tissue culture flasks or treated with epidermal growth factor instead of HGF/SF under the same conditions. Furthermore, lumen formation was dependent on the dose of HGF/SF, as shown in Figures 4A and 4B. Thus, when cells were treated with 1-10 ng of HGF/SF per milliliter of medium, applicants observed lumen-like structures with a uniform size. The regularity in the size indicates that the number of cells forming the lumen border is controlled and that cell-cell interactions must play an important role in this formation.

Example 2

Immunofluorescence Assays

For immunofluorescence assays, fixed cells or paraffin-embedded tissues, cells, and embryo sections (I. Keydar et al . , Eur. J . Cancer 15: 659 (1979)) were blocked for 10 minutes using the Biomeda immunostaining kit blocking reagents. Primary antibody (C28, rabbit anti-human Met at 1:100 dilution in PBS according to the methods of M. Gonzatti-Haces et al . , Proc . Natl . Acad . Sci . USA 85: 21 (1988); C200, rabbit antibody prepared against the extracellular domain of the Met peptide, amino acids 643-663, according to the methods of M. Park et al . , Proc . Natl . Acad. Sci . USA 84: 6379 (1987); SP260, rabbit anti-mouse Met at a 1:100 dilution in PBS; 4G10, mouse monoclonal antibody to phosphorylated tyrosine (P-Tyr) , according to D.K. Morrison et al . Cell 58: 649 (1989)) was added and incubated for 2 hours at room temperature. Secondary antibody incubation (donkey anti-rabbit coupled to phycoerythrin at 1:50 and donkey anti-mouse coupled to fluorescein isothioσyanate at 1:100 (Jackson ImmunoResearch Laboratories, Inc.) were performed for 1 hour at room temperature. After an extensive washing, cells were fixed using gel mount (Biomedia) . Fluorochrome-labeled cells were examined using a Zeiss laser scan microscope (LSM) having the following configuration: 25 nW argon and HeNe lasers, 488, 514, and 543 maximum lines; control Indec (Sungate) software for image acquisition of X-Y scan, Z-series scan three- dimensional visualization. Photographs were taken using a Sony color video printer mavigraph and Sony UPC-5010a color print paper. When comparing the fluorescence intensity, applicants used identical parameters for each image (e.g., scanning line, laser light, contrast brightness) and assessed quantitation of the relative fluorescence by using an Indec LSM Image Processor Option, "Histogram."

Example 3

Immunoelectron microscopy

For postembedding immunoelectron microscopy, fixed cells were embedded in L.R. gold resin (Electron Microscopy Science) at -25°C, sectioned with an LKB Nova Ultratome, and picked up with Formvar-coated 200 mesh gold grids. The grids were washed three times in PBS 10 min and incubated in 1% BSA (in PBS) for 2 hours and in rabbin anti-human C28 Met antibody (diluted 1:50 in 1% BSA) at room temperature for 1 hour. Controls were incubated either in the presence of C28 competing peptide or in the absence of the primary antibody. The grids were washed again in PBS, incubated in RPM11640 medium for 20 min in 1% BSA as above, and reacted with goat anti-rabbit IgG gold (10-nm diameter; 1:10 diluted in 1% BSA; Amersham) at room temperature for 1 hour. The grids were finally washed in PBS and distilled water and stained with uranyl acetate and lead citrate. The sections were observed and photographed using a Philips EM 410 electron microscope.