RON is a tyrosine kinase receptor that triggers scattering of normal cells and invasive growth of cancer cells on ligand binding. We identified a short RON mRNA, which is expressed in human lung, ovary, tissues of the gastrointestinal tract, and also in several human cancers, including ovarian carcinomas and cell lines from pancreatic carcinomas and leukemias. This transcript encodes a truncated protein (short-form RON; sf-RON), lacking most of the RON receptor extracellular domain but retaining the whole transmembrane and intracellular domains. Sf-RON shows strong intrinsic tyrosine kinase activity and is constitutively phosphorylated. Epithelial cells transduced with sf-RON display an aggressive phenotype; they shift to a nonepithelial morphology, are unable to form aggregates, grow faster in monolayer cultures, show anchorage-independent growth, and become motile. We show that in these cells, E-cadherin expression is lost through a dominant transcriptional repression pathway likely mediated by the transcriptional factor SLUG. Altogether, these data show that expression of a naturally occurring, constitutively active truncated RON kinase results in loss of epithelial phenotype and aggressive behavior and, thus, it might contribute to tumor progression.
The RON receptor belongs to the MET family of tyrosine kinase receptors (1) . This family includes the human and mouse MET and the human RON and its homologues in mouse (stk; Ref. 2 ), cat (f-stk; Ref. 3 ), and chicken (sea; Ref. 4 ). Members of the family share biochemical features and biological properties. MET and RON receptors both trigger complex morphogenetic programs of epithelial cells that lead to “invasive growth” (5) . In physiological conditions, cell invasive growth occurs during organ development and regeneration. In cancer cells, inappropriate activation of invasive growth might lead cells to invade and metastasize.
The RON receptor is expressed in epithelia and hematopoietic cells. First findings (6) indicated that RON receptor activation regulates tissue macrophage motility. Additional findings demonstrated that the RON receptor induces epithelial cell scattering, which involves not only cell motility but also dissociation and matrix invasion, suggesting that RON might be involved in the pathogenesis of certain epithelial cancers (7) .
Activation of oncogenes of the MET family occurs in cancer and follows rules that are common to all of the tyrosine kinase oncogenes. MET and RON are overexpressed and amplified in human cancers of specific histotypes (7, 8, 9, 10) . Both MET and RON receptors can be activated by missense mutations in the tyrosine kinase domain. MET mutations have been found in families suffering from hereditary papillary renal cell carcinoma (11 , 12) . When the same mutations were introduced in the RON cDNA, mutated receptors converted transfected cells into tumorigenic transformants (13) . Rearrangement of both MET and RON genes with the dimerizing TPR sequences led to kinase activation, cell transformation, and tumorigenesis (14 , 15) .
Both the human RON and the homologous mouse stk genes give rise to two transcripts: in humans a 5.0 kb and a 2.0 kb mRNAs (16) ; and in the mouse a 4.5 kb and a 1.9 kb mRNAs (2) . In both species, the longer transcript encodes for the full-size receptor. The shorter mouse transcript (sf-stk) was found to be expressed in animals susceptible to Friend leukemia virus-induced erythroleukemia and absent in resistant mice (17) . In the latter, sf-stk-enforced expression restored susceptibility. In Friend-infected erythroblasts, the sf-stk protein directly interacts with viral gp55 but not with the erythropoietin-R (18) and plays a key role in the signal transduction that regulates erythropoietin-independent erythroblast expansion. Conversely, the sf-stk role in cell physiology is still elusive, although the sf-stk transcript is highly expressed in mouse erythroleukemic cells (2) and is detectable in mouse stomach and intestine (19) .
The shorter (2.0 kb) human RON mRNA was detected, at a very low level, in colon, skin, and lung tissues but not in bone marrow, bone-marrow-derived cells, human muscle, nor in human brain (16 , 20) . It has been also found in some human tumor cells lines derived from gastric (16) and lung cancers (21) . Although it is known that the shorter transcript includes sequences encoding the RON kinase domain, its role in normal and cancer tissues is still unknown.
In this article, we show that a 5′-truncated RON mRNA, analogous to sf-stk transcript, is expressed in human normal and cancer cells including ovarian, pancreatic, gastrointestinal, and leukemia cells. In well-differentiated breast carcinoma cells, its enforced expression induced loss of epithelial features and progression toward a more aggressive phenotype.
MATERIALS AND METHODS
Cell Lines and Human Tissues.
The GTL16 cell line is a clonal cell line derived from a poorly differentiated gastric carcinoma line (22) . All of the other cell lines were purchased from American Type Culture Collection (Manassas, VA). Normal and cancer human tissue samples were obtained and processed as described previously (23) .
The purified polyclonal antibodies C-20 and N-20, antihuman RON, anti-SLUG, and the monoclonal antibodies against N-cadherin were from Santa Cruz Biotechnology (Santa Cruz, CA); monoclonal antibodies anti-E-cadherin were from Transduction Laboratories (BD Biosciences, Canada); monoclonal pan anticadherins antibodies were from Sigma (Sigma-Aldrich, Germany); and monoclonal antiphosphotyrosine antibodies were from Upstate (Upstate, MA). Monoclonal anti-myc antibodies (clone 9E10) were obtained from Dr. Luca Tamagnone (I.R.C.C., Candiolo, Torino, Italy).
End-Point Reverse Transcription-PCR (RT-PCR).
Cytoplasmic mRNA was prepared using the Concert Reagent from Invitrogen (Life Technology, Carlsbad, CA). Total RNA from tissue samples was prepared and analyzed as described previously (22) . Retrotranscription was carried out with Moloney murine leukemia virus reverse-transcriptase RNase H minus (Promega, Madison, WI) and oligodeoxythymidine nucleotide. To examine RON transcript expression, PCR was performed with the following primers: sense, 5′ TAT CCT GCA GGT GGA GCT G 3′; and antisense, 5′ ATG AAA TGC CAT GCC CTT AG 3′. PCR conditions were 10 min at 95°C for 1 cycle; 1 min at 95°C, 1 min at 55°C, and 1 min at 72°C for 30 cycles; and 7 min at 72°C for 1 cycle, with DNA polymerase TaqGold (Perkin-Elmer). To detect sf-RON expression, seminested PCR was performed as follows. In both rounds of amplification, the sense primer was (P1, that anneals on an intron 10 sequence) 5′ CCT CAT GAC CCT CTC TGC AGT 3′. In the first PCR round, the antisense primer was either P2, that anneals on a exon 14 sequence: 5′CAG CAG TGG CAC ACA GGA T3′or P3, that anneals on an exon 16 sequence: 5′ TAC CAA TGA GAG CCA GCA CA 3′; in the second amplification round, the antisense primer was either P4, that anneals on a exon 12 sequence: 5′ GCC ACC AGT AGC TGA AGA CC 3′ or P5, that anneals on a exon 14 sequence: 5′ CAG CAG TGG CAC ACA GGA T 3′. Both PCR amplifications were performed with DyNAzyme DNA (Finnzymes) polymerase at 3 min at 95°C for 1 cycle, 1 min at 95°C, 1 min at 56°C, 1min at 72°C for 20 cycles, and 7 min at 72°C for 1 cycle. To rule out the possibility that the identified RT-PCR products could derive from unprocessed nuclear RNA, cytoplasmic RNA was purified; control amplifications were performed with antisense primers corresponding to other intron sequences (data not shown) and showed that unprocessed RNA was not present.
To detect E-cadherin gene expression, RT-PCR was carried out with the following primers: sense, 5′ TGG GTT ATT CCT CCC ATC AG 3′; and antisense, 5′ TTT GTC AGG GAG CTC AGG AT 3′. PCR conditions were 3 min at 95°C for 1 cycle, 1 min at 95°C, 1 min at 59°C, 1 min at 72°C for 30 cycles, and 7 min at 72°C for 1 cycle, with DyNAzime (Finnzymes). For N-cadherin expression, the following primers were used: sense, 5′ CGG GTA ATC CTC CCA AAT CA 3′; and antisense, 5′CTT TAT CCC GGC GTT TCA TC 3′. PCR conditions were 3 min at 95°C for 1 cycle, 1 min at 95°C, 1 min at 57°C, 1 min at 72°C for 30 cycles, and 7 min at 72°C for 1 cycle, with DyNAzime (Finnzymes).
Immunoprecipitation, Western Blotting, and in Vitro Kinase Assay.
Immunoprecipitation and Western blotting were carried out as described previously (23) . To precipitate sf-RON protein, antibodies against RON COOH-terminus were cross-linked to Protein A-Sepharose with dimethylpimelimidate (Pierce) according to manufacturer’s instructions.
Immunocomplex kinase assay was performed on proteins immunoprecipitated with either myc or RON antibody in 25 mm HEPES (pH 7.4), 5 mm MnCl2, and 100 μm DTT supplemented with 20 μCi [γ-33P]ATP (specific activity > 2500 Ci/mmol; Amersham, United Kingdom) for 10 min at 25°C. When reported, cells were pretreated with 0.1 mm sodium orthovanadate for 1 h at 37°C. Proteins were separated on SDS-PAGE. To evaluate phosphotyrosine, gels were alkali treated.
Cell Transduction with Lentiviral Vectors (LV).
Cells were transduced using third-generation LVs (24) . As transfer vector, we used the pRRL.sin.PPT.hCMV.GFP.pre, where enhanced green fluorescent protein cDNA (XbaI-SalI fragment of 754 bp) was substituted with the full-length RON cDNA (XbaI-SalI fragment 4590 bp). Sf-RON cDNA was obtained by PCR amplification with the following primers: sense, 5′ GGT ATC AAC GTG ACC GGT GGT GGT 3′; and antisense, 5′ AGC TAA GCA GGT CGA CCC CAA GAA CTA AG 3′. The obtained AgeI-SalI fragment of 1562 bp was subcloned in the transfer vector above. The pRRL.sin.PPT.hCMV.GFP.pre vector was used as a control. Sf-RON-myc-tagged cDNA sequence was obtained by PCR amplification with the following primers: sense, 5′ GCT CTA GAA GAG CTG CCA GCA CGA GTT C 3′; and antisense, 5′ CGC GGA TCC TCA ATT CAG ATC CTC TTC TGA GAT GAG TTT TTG TTC AGT GGG CCG AGG AGG CTC 3′. The PCR product was digested XbaI-BamHI and subcloned in the transfer-multicloning site vector pRRL.sin.PPT.hCMV.MCS.pre digested with the same restriction enzymes. The luciferase reporter gene constructs pRRL.sin.PPT.Ecadpr.Luciferase.pre containing wild-type E-cadherin promoter sequence was obtained by substituting in the transfer vector pRRLsin.PPT.TetOminMMTV. luciferase.pre the sequence TET07minMMTV (XhoI-SalI fragment of 511 bp) with the sequence of E-cadherin wild-type promoter from −176 to +93. The transfer vector pRRL.sin.PPT.Tet07.Luciferase (8306 bp) was used as a control in the luciferase assay. Vectors stocks were produced by transient transfection of 293T cells. Serial dilutions of freshly harvested media were used to infect 105 cells in a six-well plate in the presence of Polybrene (8 μg/ml). The viral p24 antigen concentration was determined by HIV-1 p24 core profile ELISA Innotest HIV antigen monoclonal antibody (Innogenetics N.V.).
To assay luciferase activity driven by E-cadherin or TET promoter, cells (0.15 × 106 in six-well plates) were transduced with LVs and, after 3 days, solubilized according to the Luciferase Assay System kit (Promega). Luciferase activity was measured with luminometer (Lumat LB 9507; EG&G Berthold). Three LV concentrations were assayed. Each concentration was assayed in two separate transduction experiments. Luciferase activity was expressed as relative light units normalized for total protein in each sample.
Quantitative Real-Time RT-PCR.
Expression of E-cadherin repressors was measured with quantitative real-time RT-PCR with TaqMan assay using SYBR green (Applied Biosystems, Foster City, CA) on cDNAs reverse transcribed from mRNA of sf-RON, over-RON, mock-transduced, and wild-type T47D cells. Forward and reverse primers for each gene were as follows: (a) to detect SLUG expression, forward primer 5′ GTT TTC CAG ACC CTG GTT GCT 3′ and reverse primer 5′ TTC TCC CCC GTG TGA GTT CTA 3′; (b) to detect SNAIL expression, forward primer 5′ CTC TGG TCT GAC CGA TGT GTC TC 3′ and reverse primer 5′ ACC TGT CGG GCC CCC 3′; (c) to detect E12/E47 expression, forward primer 5′ TTT TGC CAT GAG GGT AAC CAG 3′ and reverse primer 5′ ACC CAA GAC AGT CCC AGC C 3′; and (d) to detect SIP1 expression, forward primer 5′ GCT TGG TTA GCA GGT ATT TTG ACC 3′ and reverse primer 5′ CAA GAT GGC TCA TCA GCT AAA TCA 3′. Preliminary experiments were carried out to select primer pairs and concentrations. PCR reactions were carried out using an ABI Prism 7700 Sequence Detection System (Perkin-Elmer Applied Biosystems, Oak Brook, IL). Glyceraldehyde-3-phosphate dehydrogenase expression was measured using the Perkin-Elmer glyceraldehyde-3-phosphate dehydrogenase kit.
Cell Growth Assays.
To measure proliferation rate, cells were plated (2 × 105 cells/well) in duplicate in six-well plates. Cell viability was evaluated by trypan blue exclusion assay. For soft agar growth assay, cells were suspended in 0.5% SeaPlaque agar (BMA, Rockland, ME) with complete RPMI 1640 and plated at the density of 250 or 750/ml. Colonies were stained with tetrazolium salt.
Cell Adhesion, Aggregation, and Motility Assays.
For cell adhesion assays, 96-well microtiter plates (Nunc, Naperville, IL) were coated with 0.1 μg/μl either laminin or fibronectin (Sigma) in Tris-buffered saline (pH 7.4) for 16 h at 4°C; wells were then rinsed and treated with 0.2% BSA in PBS (pH 7.4). Cells were harvested with 5 mm EDTA in PBS and seeded in serum-free medium (1.25–5 × 104 cells/well). After 15-, 30-, and 60-min incubation at 37°C, adherent cells were fixed in 11% glutaraldehyde, stained with 0.1% crystal violet, and counted. As control, cells were left to adhere to wells coated with BSA. For hanging drop assay, cells were harvested with 5 mm EDTA in PBS and resuspended in RPMI without serum and with or without 5 mm EGTA. Then, 50 μl of medium containing 10000 cells were seeded on the cover of a plastic culture plate and inverted; cell drops were incubated 5 h at 37°C and then photographed. Motility assay was carried out as described previously (23) .
Expression of the sf-RON in Human Cancers.
The homology between the sequence of human RON and mouse stk genes (Ref. 25 ; data not shown) suggested that the short RON transcript could be generated in the same way as the mouse sf-stk transcript, which starts in intron 10 and is translated from an ATG codon located in exon 11 (2 , 17) .
We first amplified human RON short transcript (sf-RON) sequences from normal human tissues (Fig. 1) ⇓ . As the RON/stk tyrosine kinase domain showed homologies with other kinase receptors (2 , 20) , we performed a seminested RT-PCR. We used intron 10 sequence as a sense primer in both rounds of amplification and two sequences as antisense primers selected to avoid amplification of the highly homologue human MET sequences. Two RT-PCR products were amplified consistently from tissues of the gastrointestinal tract, lung, and ovary but not from the liver nor both cardiac and skeletal muscle samples (Fig. 1) ⇓ . The two RT-PCR products were directly sequenced. One was found containing intron 10 and subsequent RON exons and was analogous to the sf-stk transcript. The second PCR product, more evident in lung tissues, contains also intron 11 and corresponded to an expressed sequence tag cloned previously from human lung tissues (PMO-ET0274-180301-001-c06). The two products are consistent with two spliced forms of the sf-RON transcript.
Then, we evaluated sf-RON expression in 41 cell lines, representing different human tumor histotypes (Table 1) ⇓ . Approximately 50% of these cells showed expression of the sf-RON. These were mostly pancreatic carcinoma and hematopoietic cell lines (Table 1 ⇓ ; Fig. 1 ⇓ ). The transcript coding the full-size RON receptor was present in several cell lines, where the sf-RON transcript was not detectable (Table 1) ⇓ .
We also examined breast and ovarian cancer samples and found sf-RON expression in 9 of 9 ovarian samples (9 serous carcinomas) examined and in ∼30% of breast ductal carcinoma samples (Table 1) ⇓ .
We used immunohistochemistry to localize RON proteins in ovarian cancer samples and in the normal ovarian superficial epithelium, from which >95% of cancers derive. Antibodies were against the COOH-terminal tail of the RON kinase and, therefore, able to label both the full-size RON receptor and the sf-RON. The RON kinase domain was undetectable in the human ovarian superficial epithelium (Ref. 23 ; data not shown), demonstrating that neither the full-size nor the short-form RON protein is detectable in ovarian superficial epithelium. The nine ovarian carcinoma samples were all positive (data not shown).
Kinase Activity of the sf-RON Protein.
To study sf-RON biochemical and biological activities, expression in human cells was obtained using lentivirus-mediated gene transfer. The LVs allow efficient gene transfer (transduction) without cell selection, by driving the random integration of one or more copies of transgenes in each cell at different sites. As nondividing cells also are transduced, transgenes are integrated into the large majority of infected cells. Random integration ensures expression in almost all of the cells, because expression is not influenced by integration site. Because of a deletion in the long terminal repeat promoter region of LVs, integrated transgenes are transcribed only from the internal promoter that is present in the expression cassette.
Transduction of human cells with myc-tagged sf-RON showed that the identified transcript encodes a Mr 55,000 protein (Fig. 2A) ⇓ , which shows constitutive intrinsic tyrosine kinase activity (Fig. 2B) ⇓ .
To evaluate the biological outcome of sf-RON protein expression, we transferred sf-RON cDNA into T47D breast carcinoma cell lines, again using LVs. We selected T47D cells as a model of well-differentiated transformed cells that maintain in vitro the ability to form cell cell-cell junctions (26) . In addition, in T47D cells, the RON receptor-encoding transcript is expressed highly, whereas RON short transcript is barely detectable (Fig. 3 ⇓ ; Ref. 16 ). As controls, the same cells were transduced with the cDNA encoding the full-size RON receptor to obtain overexpression, and with a nonrelated transgene, encoding the enhanced green fluorescent protein, as a control. As the RON gene behaves as an oncogene, overexpression or expression of its possibly activated counterparts results in activation of cellular functions (27) .
After transduction of T47D cells, we measured the number of integrated transgene copies and the corresponding mRNA by Southern blot analysis, PCR, and RT-PCR. As expected in both sf-RON-transduced T47D cells (sf-RON T47D) and in cells transduced with the full-size RON (over-RON T47D) transgenes, we found multiple and random transgene integration sites and the transgene mRNAs encoding the RON proteins (data not shown).
Fig. 3A ⇓ shows that in T47D cells transduced with the sf-RON transgene, a newly expressed protein with the Mr 55,000 is recognized by an anti-RON antibody raised against the COOH-terminal sequence of the full-length RON receptor. This sf-RON protein appeared phosphorylated on tyrosine in transduced cells (Fig. 3B) ⇓ . In the in vitro kinase assay (Fig. 3C) ⇓ , the p55 sf-RON protein showed intrinsic kinase activity (Fig. 3C) ⇓ , which was stronger than that of both the endogenous and the overexpressed RON full-size receptors. In fact, sf-RON phosphorylated to a higher intensity the coprecipitated full-size RON receptor.
Loss of Cell-Cell Adhesion of sf-RON Expressing Cells.
Both expression of either sf-RON or overexpression of the RON receptor caused morphological changes of T47D cells (Fig. 4A) ⇓ . Cells expressing the sf-RON appeared spherical, whereas those overexpressing the full-length RON receptor were more elongated than mock-transfected T47D. As these morphological changes can be attributed to adhesive molecule changes, we tested cell ability to adhere to extracellular matrix glycoproteins and to form cell aggregates independently from anchorage to substratum. Sf-RON T47D cells did not show any differential adhesion to laminin and to fibronectin (data not shown) but demonstrated a total inability to form cell aggregates in an in vitro assay (Fig. 4B ⇓ , top panels). To assess calcium-dependence of cell ability to form aggregates in this assay, we also performed the assay in the presence of the chelating agent EGTA. As shown in the bottom panel of Fig. 4B ⇓ , cell aggregation was completely inhibited by EGTA.
Change of morphology and loss of calcium-dependent aggregation in sf-RON expressing cells suggested that cadherin function was impaired in these cells. Therefore, we tested whether nontransduced or transduced T47D cells expressed E-cadherin, other epithelial cadherins, or nonepithelial cadherins (Fig. 4C) ⇓ . Nontransduced T47D cells expressed E-cadherin and showed a modest reactivity with pan-specific antibodies that recognize a series of other cadherins (data not shown in Fig. 4C ⇓ ). Sf-RON T47D cells have lost E-cadherin and did not express detectable levels of the cadherins recognized by pan-specific antibodies (data not shown). Over-RON T47D cells have lost E-cadherin expression but showed neoexpression of other cadherins that might account for their ability to form aggregates. In particular, these cells showed neoexpression of the neuronal specific N-cadherin. This suggested that sf-RON activation directly affected cadherin(s) expression.
To assess whether E-cadherin loss was because of transcriptional or posttranscriptional regulation, mRNAs encoding either E- or Ncadherin were evaluated using end-point RT-PCR. As shown in Fig. 4D ⇓ , sf-RON T47D cells did not contain either E- or N-cadherin-specific transcript. On the contrary, E- but not N-cadherin transcript was detected in nontransduced T47D cells, whereas N-cadherin alone but not E-cadherin mRNA was found in over-RON T47D cells (Fig. 4D) ⇓ . This suggested that E-cadherin expression was turned off in both sf-RON and over-RON T47D cells at transcriptional level.
Sf-RON Expressing Cells Showed Faster Growth and Motility.
There are several truncated tyrosine kinases (28) that are able to confer a transformed phenotype to human cells. Furthermore, it is known that E-cadherin loss is directly associated to proliferation of transformed cells (29) .
Therefore, we measured proliferation of sf-RON expressing and RON overexpressing T47D cells in monolayer cultures and in soft agar. Sf-RON T47D cells grew faster than both the nontransduced ones, the mock-transduced, and over-RON T47D cells in monolayer culture (Fig. 5A) ⇓ . In addition, sf-RON expression conferred to T47D cells the ability of very fast growth in the absence of anchorage (Fig. 5B) ⇓ .
A role of E-cadherin in tumor progression is attributed to the acquisition of a motile and invasive phenotype by transformed cells. We measured cell motility in an in vitro assay and found that sf-RON expression increased significantly the ability of T47D cells to move through filter micropores (Fig. 5C) ⇓ .
Transcriptional Repression of E-Cadherin in sf-RON Expressing Cells.
E-cadherin gene expression in human cancers is silenced by a variety of mechanisms. Among them, in breast carcinoma cells, dominant transcriptional defects prevail because of a repression pathway mediated by molecules acting on E-cadherin proximal promoter (30) . To test this hypothesis in sf-RON and over-RON cells, we subcloned reporter gene construct made of luciferase gene in LVs under the control of the E-cadherin proximal promoter critical region extending from −176 to +93 of the E-cadherin gene sequence. The latter region has been identified as the minimal portion of E-cadherin promoter, demonstrating strong activity in breast cancer cell lines with intact E-cadherin transcription and greatly reduced activity in breast cancer cell lines defective for E-cadherin transcription (30 , 31) . Three days after transduction, on obtaining transgene integration, luciferase gene expression was evaluated. The reporter gene expression was repressed in both sf-RON and over-RON T47D cells, in the former to <0.1 and in the latter to <0.25 of the expression found in both nontransduced or mock-transduced cells (Fig. 6A) ⇓ .
Several repressors of the E-cadherin promoter activity have been identified. We measured the expression of SNAIL (32 , 33) , SLUG (31) , E12/E47 (34) , and SIP1 (35) . As shown in Fig. 6B ⇓ , the expression of SLUG was increased strongly in both sf-RON and over-RON T47D, although it was barely detectable in nontransduced cells, whereas the expression of the other repressors was very modestly affected. Altogether, these data show that the RON kinase activation induced E-cadherin transcription repression that could be because of increased expression of the SLUG repressor.
In this article, we provide evidence that a truncated RON kinase is expressed in human cancers. The presence of a RON transcript, shorter than that encoding the full-size RON receptor, had been already demonstrated by Northern blot analysis in some normal tissues, cancer samples, and cell lines (16 , 20 , 21) . However, its role in physiology and pathology has never been explored. The murine homologue short-form stk is expressed highly in some mouse tissues, but its physiological role has not been established as yet, except in the Friend leukemia virus infection.
We found expression of sf-RON, detected as a transcript containing RON gene intron 10 sequences, homologous to the sf-stk transcript, in human tissues of the gastrointestinal tract, lung, and ovary. Expression in the gastrointestinal tract and lung was expected, because the shorter RON transcript has been detected already in the intestine (16) and was found to be the major one in the lung (20) . Sf-RON expression in the ovaries is a novel finding and is particularly intriguing. We found previously that the RON longer transcript was not expressed in a number of samples of human adult ovaries and in a percentage of ovarian carcinomas (23) . In mice, germ-line ablation of the RON TK domain leads to reduced ovarian size, due mainly to reduced number and density of corpora lutea and defects in ovulation (36) . The latter data imply that the RON kinase plays a role in the development of mouse ovary and, in particular, in follicles and their derivatives. The discovery of sf-RON expression in the human adult ovaries suggests that the RON kinase might be important in human ovaries, too.
Expression of sf-RON was found in ∼50% of the tumor cell lines analyzed and in a number of cancer samples. All of the ovarian cancer samples and cell lines derived from gastrointestinal carcinomas, including all of the pancreatic carcinoma cell lines, expressed the short RON transcripts. It is worth noting that normal pancreas tissues did not show short RON transcript expression. In the normal ovary, using immunohistochemistry, we did not detect RON protein in the surface epithelium, from which >95% of human ovarian cancers derive (Ref. 23 ; data not shown), whereas we detected positive staining in ∼50% of ovarian carcinomas examined (23) . This information suggests that sf-RON expression increase during ovarian cancer tumorigenesis. It is worth mentioning that we fully confirmed RON longer and shorter transcript expression in cell lines and tissues, where either transcript was identified previously with Northern blot analysis (16) .
Expression of the sf-RON transcript is low in most cancer cell lines (16 , 21) . However, we show that aberrant sf-RON expression provides striking biological outcomes, because it causes well-differentiated breast carcinoma cells to change morphology, move, and proliferate faster, even if in the absence of anchorage. Overexpression of the full-size RON receptor has milder biological outcomes. Difference is likely because of the higher and constitutive kinase activity of the sf-RON protein. Truncated forms of other tyrosine kinase receptors show ligand-independent activation leading to cell transformation (28) . For instance, the truncated epidermal growth factor receptor is expressed in a consistent percentage of human glioblastoma and transforms cells (37) .
Other naturally occurring RON variants show oncogenic potential (7) . All of these variants have in-frame deletions that lead to the RON kinase constitutive activation and cause activation of either cell growth in transformation assays or cell motility in invasion assays. These variants have been found only in a few cell lines and human tumor samples derived from gastrointestinal tract tissues. Before the discovery of these variants, the RON oncogene was converted into an oncogene by molecular approaches, including translocation with dimerizing sequences and experimental mutagenesis. Altogether, experimental approaches confirm that deregulation of RON kinase activity is essential to release RON oncogenic potential.
We report here that morphological changes and the transformed phenotype of sf-RON expressing cells are associated to loss of E-cadherin. Several observations led also to establish a direct correlation between E-cadherin loss and both cancerogenesis and tumor progression. We found that sf-RON expressing cells show not only a morphological change but also an increased proliferation rate. This reflects what occurs in cancer cells, where E-cadherin loss is associated to an increased proliferation rate, which, in some instances, directly depends on the protein loss, because it is counteracted by E-cadherin addition (29) .
E-cadherin is a tumor and invasion suppressor in human carcinomas. Its inactivation by mutations has been only occasionally found (38) . In fact, in most cases, E-cadherin down-modulation during tumor progression occurs by either promoter hypermethylation (39) or dominant transcriptional repression (40) . In particular, the identification of specific transcriptional repressors like SLUG, SNAIL, E12/E47, and SIP1 has greatly increased our understanding of E-cadherin regulation (31 , 33) . In addition, in most cancer cells where tyrosine kinases are activated, E-cadherin is lost as a consequence of its tyrosine phosphorylation leading to endocytosis and degradation or relocalization of the protein. For instance, MET tyrosine kinase activation results in E-cadherin loss by endocytosis because of E-cadherin ubiquitination (41) . We show that in sf-RON expressing cells, E-cadherin transcriptional repression is associated to increased SLUG expression, whereas expression of SNAIL, E12/E47, and SIP1 was substantially unaffected. We cannot rule out that other repressors might be involved, but SLUG was described already as E-cadherin-specific repressor in breast cancer cells (31) .
In conclusion, we show a novel mechanism by which an activated tyrosine kinase of the MET family leads to E-cadherin loss, which accompanies the acquisition of a more transformed phenotype by cancer cells. Receptors of the MET family have been already implicated in tumor progression toward an invasive and metastatic phenotype (5) . This phenotype acquisition has been shown to depend on qualitative and quantitative activation of specific signal transduction cascades but can be limited or rapidly reversed in many cases. Here, we report that a constitutively active kinase of the same family, which is expressed in human tumor cells, determines a morphological subversion and the acquisition of a more aggressive phenotype by suppressing E-cadherin expression at mRNA level. This suggests that besides transient and acute alteration of cell behavior, kinases of the MET family can orchestrate the transformed and aggressive phenotype of cancer cells by stably inducing transcriptional activation and repression.
We thank Enzo De Sio, Lucia Sergi Sergi, and Raffaella Albano for technical help, Elaine Wright for reading the English, and Prof. Luigi Naldini, Dr. Elisa Vigna, and Dr. Livio Trusolino for precious suggestions and helpful discussion.
Grant support: Italian Ministry of Research and Education (Ministero per l’Istruzione, l’Universita’ e la Ricerca Scientifica; M. Di Renzo, P. Comoglio, M. De Bortoli, and G. N. Ranzani), the Italian National Research Council (Consiglio Nazionale delle Ricerche-Ministero per l’Istruzione, l’Universita’ e la Ricerca Scientifica Progetto Oncologia), and the Italian Association for Cancer Research (Associazione Italiana Ricerca sul Cancro; M. Di Renzo and P. Comoglio).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Maria Flavia Di Renzo, Laboratory of Cancer Genetics, Istituto per la Ricerca e la Cura del Cancro, SP 142, Km 3.95, 10060 Candiolo (TO), Italy. Phone: 39-11-9933343; Fax: 39-11-9933524; E-mail:
- Received February 19, 2004.
- Revision received April 29, 2004.
- Accepted May 19, 2004.
- ©2004 American Association for Cancer Research.