The Scatter Factor/Hepatocyte Growth Factor: C-Met Pathway in Human Embryonal Central Nervous System Tumor Malignancy

Research Article

The Scatter Factor/Hepatocyte Growth Factor: c-Met Pathway in Human Embryonal Central Nervous System Tumor Malignancy

Yunqing Li,1,5 Bachchu Lal,1,5 Sherwin Kwon,5 Xing Fan,2 Usha Saldanha,1 Thomas E. Reznik,5 Eric B. Kuchner,2 Charles Eberhart,2 John Laterra,1,3,4,5 and Roger Abounader1,3,5

Departments of 1Neurology, 2Pathology, 3Oncology, and 4Neurosciences, Johns Hopkins University School of Medicine; and 5Kennedy Krieger Research Institute, Baltimore, Maryland

Abstract throughout the CNS, leading to leptomeningeal metastasis (5). Understanding the molecular pathways involved in embryonal CNS Embryonal central nervous system (CNS) tumors, which comprise medulloblastoma, are the most common malignant tumor formation and growth is required for the identification of new brain tumors in children. The role of the growth factor scat- therapeutic targets. Recent research on the molecular basis of these ter factor/hepatocyte growth factor (SF/HGF) and its tyro- tumors and its implications for targeted therapeutics has focused on sine kinase receptor c-Met in these tumors has been until the Hedgehog-Patched, Wnt, and Trk signaling pathways (3, 4, 6). now completely unknown. In the present study, we show These pathways, whereas important, are recognized to represent a that human embryonal CNS tumor cell lines and surgical subset of the multiple molecular mediators of embryonal CNS tumor specimens express SF/HGF and c-Met. Furthermore, tumorigenesis and malignancy. The SF/HGF:c-Met pathway has c-Met mRNA expression levels statistically significantly cor- emerged as an important contributor to human neoplasia (7). relate with poor clinical outcome. Treatment of medullo- However, the role of SF/HGF and c-Met in human embryonal CNS tumors is, to our best knowledge, completely unknown. blastoma cells with SF/HGF activates c-Met and downstream signal transduction as evidenced by c-Met, mitogen-activated SF/HGF is a multifunctional growth factor that plays a role in the regulation of cell growth cell motility, morphogenesis, and protein kinase, and Akt phosphorylation. SF/HGF induces angiogenesis (8). SF/HGF exerts its effects through its only known tumor cell proliferation, anchorage-independent growth, receptor, the c-Met proto-oncogene product, a transmembrane and cell cycle progression beyond the G -S checkpoint. 1 receptor that possesses tyrosine kinase activity (9). SF/HGF and Using dominant-negative Cdk2 and a degradation stable c-Met are expressed in a wide variety of human carcinomas where p27 mutant, we show that cell cycle progression induced by their overexpression or misexpression often correlates with poor SF/HGF requires Cdk2 function and p27 inhibition. SF/HGF prognosis (7). Activation of c-Met in tumors occurs through also protects medulloblastoma cells against apoptosis induced paracrine and autocrine mechanisms. Ligand-independent activa- by chemotherapy. This cytoprotective effect is associated tion of c-Met has also been observed in tumor cells that express with reduction of proapoptotic cleaved poly(ADP-ribose) high levels of the receptor (10). Additionally, some types of cancers polymerase and cleaved caspase-3 proteins and requires and metastatic lesions have activating mutations of c-Met (11, 12). phosphoinositide 3-kinase activity. SF/HGF gene transfer We and others have shown that SF/HGF:c-Met affects tumorige- to medulloblastoma cells strongly enhances the in vivo nicity and malignant progression by inducing cell cycle progres- growth of s.c. and intracranial tumor xenografts. SF/HGF- sion, tumor cell migration, and tumor angiogenesis and by overexpressing medulloblastoma xenografts exhibit increased inhibiting apoptosis (13–17). The mitogenic effects of SF/HGF on invasion and morphologic changes that resemble human large tumor cells are at least partly due to its ability to mediate G -S cell cell anaplastic medulloblastoma. This first characterization 1 cycle transition (17). SF/HGF expression inhibits apoptosis of establishesSF/HGF:c-Metasanewpathwayofmalig- glioblastoma tumor cells and confers resistance to cell death nancy with multifunctional effects in human embryonal CNS induced by chemotherapy and radiation (15, 16). Many studies have tumors. (Cancer Res 2005; 65(20): 9355-62) also shown that c-Met activation induces tumor cell migration and invasion (18, 19). c-Met is also a potent tumor angiogenic factor Introduction and its inhibition leads to inhibition of tumor growth partially via Embryonal central nervous system (CNS) tumors, known as its effects on tumor angiogenesis (14, 15, 20, 21). SF/HGF:c-Met medulloblastoma in the cerebellum and primitive neuroectodermal malignant functions are mediated by a network of signal tumors (PNET) elsewhere in the CNS, are among the most frequent transduction pathways and transcriptional events. Activation of causes of cancer death in children. Despite aggressive therapy— c-Met results in the recruitment of scaffolding proteins, such as consisting of surgery, radiation, and chemotherapy—lethality and Gab1 and growth factor receptor binding protein 2, which lead to morbidity of patients with embryonal CNS tumors remains high the activation of Ras and extracellular signal-regulated kinase/ (1–3). Medulloblastoma is thought to arise from granule cell precur- mitogen-activated protein kinase (MAPK). This causes changes in sors in the external germinal layer of the developing cerebellum (4). gene expression of cell cycle regulators, such as pRb, cdk6, and p27, They are highly invasive neoplasms with a tendency to disseminate as well as extracellular matrix proteinases, such as matrix metal- loproteinases and urokinase-type plasminogen activator, leading to alterations of cytoskeletal and cell adhesive functions that Requests for reprints: Roger Abounader, Kennedy Krieger Research Institute, 707 control cell proliferation, migration, and invasion (7, 22, 23). Active North Broadway, Suite 400-K, Baltimore, MD 21205. Phone: 443-923-2671; Fax: 443-923- c-Met also directly binds to phosphoinositide 3-kinase (PI3K) 2695; E-mail: [email protected]. I2005 American Association for Cancer Research. at phosphotyrosines 1,349 (Y-1349VHV) and 1,356 (Y-1356VNV; doi:10.1158/0008-5472.CAN-05-1946 refs. 24, 25) of the receptor. This leads to the activation of Akt, www.aacrjournals.org 9355 Cancer Res 2005; 65: (20). October 15, 2005

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Cancer Research a well-characterized mediator of cell survival (26, 27). We and others Clinical samples. Frozen tissues from 32 embryonal tumors resected have previously shown that activation of PI3K/Akt mediates SF/ between 1992 and 2002 at the Johns Hopkins Hospital were used to make HGF-induced cancer cell resistance to apoptosis and cell death (16). the RNA to study the correlation between SF/HGF, c-Met, and patient In the present study, we identify for the first time a role for the survival. The cases included fourteen classic medulloblastomas, six anaplastic medulloblastomas, five nodular medulloblastomas, four supra- SF/HGF:c-Met pathway in human embryonal CNS tumor malig- tentorial PNET, two medulloepitheliomas, and one pineoblastoma. The nancy. We show that SF/HGF and c-Met expression levels in median age of patients was 8 years, and the median follow-up was 18 human tumor specimens significantly correlate with patient months. Eight additional human embryonal tumors were used for analysis survival. SF/HGF activates c-Met–dependent signal transduction of SF/HGF and c-Met protein by immunoblotting. pathways in medulloblastoma cells, leading to induction of Taq-Man real-time quantitative reverse transcription-PCR. Total multiple malignancy parameters, including anchorage-independent RNA was extracted from snap-frozen tumor tissues using TRIZOL Reagent growth, cell proliferation, cell cycle progression, and cell survival. (Invitrogen) per the instructions of the manufacturer. RNA was then treated Activation of c-Met in medulloblastoma cells also strongly induces with DNase and further purified using the RNeasy Protocol (Qiagen, in vivo medulloblastoma xenograft growth. These findings show Valencia, CA). Quantitative reverse transcription-PCR (RT-PCR) was done that the SF/HGF:c-Met pathway is an important mediator of using the iCycler (Bio-Rad Laboratories, Inc., Hercules, CA) with TaqMan medulloblastoma malignancy. The results provide a timely One-Step RT-PCR Master Mix reagents (Applied Biosystems, Foster City, CA) and normalized to actin (Applied Biosystems) as previously described (30). validation of the SF/HGF:c-Met pathway as a target for the Assays-on-Demand TaqMan probes (Applied Biosystems) were used to mea- promising small molecular inhibitors of c-Met and anti-SF/HGF sure c-Met (assay ID: Hs00179845_m1) and HGF (assay ID: Hs00300159_m1) antibodies that are expected to enter clinical trials soon. expression. All samples were analyzed in triplicate. Serial dilutions of total RNA from the medulloblastoma cell line DAOY (for c-Met) and astrocytoma Materials and Methods cell line SW1088 (for HGF) were used to generate standard curves. The expression of c-Met and HGF in all samples was calculated in relation to Cell culture and treatments. DAOY cells were grown in Improved MEM these. The GraphPad Prism 4 program (San Diego, CA) was used for log-rank Zinc Option supplemented with 10% fetal bovine serum (FBS). D425 and analysis of Kaplan-Meier survival curves. D283 cells were grown in Improved MEM Zinc Option supplemented with Immunoblotting. Protein levels of c-Met and SF/HGF, p27, Cdk2, cyclin 20% FBS. PFSK and ONS-76 cells were grown in RPMI 1640 with L-glutamine E, cyclin D1, E2F-1, caspase-3, cleaved caspase-3, cleaved poly(ADP-ribose) j supplemented with 10% FBS. All cells were grown at 37 Cin5%CO2-95% polymerase (PARP), and actin as well as phosphoprotein levels of c-Met, f O2. For treatment with SF/HGF, cells were grown to 50% confluence MAPK, and Akt in cells or tumor specimens were analyzed by immuno- before being transferred to medium containing 0.1% FBS overnight. Cells blotting. Immunoblotting was done as previously described using anti- were then incubated with 10 to 100 ng/mL recombinant SF/HGF (a kind gift bodies specific for SF/HGF, c-Met, Cdk2, cyclin E, cyclin D1, E2F-1 (Santa from Genentech, Inc., San Francisco, CA) for the appropriate time points. Cruz Biotechnologies, Santa Cruz, CA), p27 (BD Biosciences, San Jose, CA), For treatment with the MAPK inhibitor (PD-98059) and PI3K inhibitor MAPK, Akt, caspase-3, cleaved caspase-3, cleaved PARP phospho–c-Met, (LY 294002 and wortmannin), cells were handled as described above except phospho-MAPK, and phospho-Akt (Cell Signaling Technologies, Beverly, that they were preincubated with the inhibitor for 1 hour before being MA; ref. 31). Blots hybridized with SF/HGF and cleaved PARP antibodies subjected to SF/HGF treatment. For treatment with campthotecin, cells were stripped and rehybridized with antiactin antibodies. Blots hybridized were preincubated with SF/HGF for 48 hours before incubation with with phospho-MAPK and phospho-Akt antibodies were stripped and campthotecin for 48 hours. rehybridized with total MAPK and Akt antibodies. All five above-described cell lines were tested for SF/HGF and c-Met Cell proliferation and anchorage-independent colony formation in expression as well as for phosphorylation of c-Met, Akt, and MAPK in soft agar. Cell proliferation was assessed by seeding cells in medium response to SF/HGF. DAOY and D425 and ONS-76 cells were tested for cell containing 0.1% FBS (5,000/well) in 24-well plates. Cells were subse- proliferation, cell cycle progression, and cytoprotection. DAOY and D425 quently counted each day (n = 3) for 7 days using a Coulter counter and cells were additionally tested for anchorage-independent growth as growth curves were established. Anchorage-independent growth was described below. assessed by use of colony formation in soft agar as previously described Vectors and transfections. All transfections were done using Fugene (14). DAOY or D425 cells were suspended in soft agar (5,000 cells per j transfection reagent (Roche Biochemicals, Indianapolis, IN) according well) and incubated at 37 Cin5%CO2-95% O2 in medium containing to the instructions of the manufacturer. For the generation of SF/HGF- 1% FBS. Recombinant SF/HGF (10 ng/mL) was added to the experi- overexpressing medulloblastoma cells, cDNA containing the complete mental group twice per week; controls received medium only. After 2.2 kb coding sequence for SF/HGF was subcloned into the pcDNA 3.1/ 3 weeks, the cells were fixed and stained with Wright’s solution. The Zeo-mammalian expression vector (Invitrogen, Carlsbad, CA) to yield the number of colonies larger than 100 Am in diameter were determined by pcDNA/SF vector. Empty pcDNA 3.1/Zeo vector was used for control computer-assisted image analysis. transfections. Transfected cells were selected for zeocin resistance and all Propidium iodide flow cytometry. The cell cycle status was assessed resistant clones were randomly pooled for experimental (DAOY-SF) and using propidium iodide flow cytometry. Cells growing as a monolayer were control (DAOY-Control) groups. This approach insures that all cells are trypsinized (Life Technologies, Grand Island, NY), pelleted by centrifuga- stably transfected while avoiding the disadvantages associated with the tion, resuspended in 1 mL ice-cold PBS, and fixed by the addition of 4 mL use of single clones. SF/HGF expression levels were quantified by immu- ice-cold ethanol under gentle vortexing. Fixed cells were stored up to 1 week noblot analysis as described below. The dominant-negative form of Cdk2 in 80% ethanol at À20jC until ready for labeling. Stored cells were collected (pCMV-Cdk2DN) was kindly provided by S. van den Heuvel (Massachusetts by centrifugation, resuspended in 1 mL of PBS, and treated with 20 Agof General Hospital, Charlestown, MA; ref. 28). The p27 degradation stable DNase-free RNase (Boehringer-Mannheim, Mannheim, Germany) for 30 mutant p27T187A construct was provided by B. Clurman (Fred Hutchinson minutes at 37jC. Cells were labeled with propidium iodide (Sigma, St. Louis, Cancer Center, Seattle, WA; ref. 29). The GFP-expressing vector is a kind gift MO) at a final concentration of 100 Ag/mL for 10 minutes at room from Kenneth Yamada (NIH, Bethesda, MD). To assess the roles of p27 and temperature. Analyses were done on a FACScan (Becton Dickinson, Cdk2 in SF/HGF-mediated cell cycle progression, pCMV-Cdk2DN or Fullerton, CA). For experiments that did not involve transfection, 10,000 p27T187A were cotransfected with GFP-plasmid (ratio = 1:3) and GFP- events were counted per sample. For experiments involving transfection, positive DAOY cells were selected by flow cytometry for further analysis to only GFP-labeled cells were counted (5,000 events per cycle). Because all ensure that the majority of analyzed cells are transfected with the vectors either coexpressed GFP or were cotransfected with GFP-expressing corresponding vectors. vectors, this ensured that only transfected cells were counted. Raw data

Cancer Res 2005; 65: (20). October 15, 2005 9356 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. c-Met Pathway in Embryonal CNS Tumors were gated to remove doublets and cellular debris. The resultant cell cycle Results histograms were analyzed with CellQuest software (Becton Dickinson) and the area under the curve was integrated for each peak. Human medulloblastoma cells and tumors express c-Met Annexin V flow cytometry. Apoptosis and cell death was assessed using and scatter factor/hepatocyte growth factor. We analyzed SF/ Annexin V flow cytometry. To test the effect of SF/HGF on campthotecin- HGFandc-MetmRNAandproteinexpressioninhuman induced apoptosis, subconfluent proliferating cells were seeded in six-well embryonal CNS tumor cell lines and surgical tumor specimens plates and subsequently preincubated with or without HGF/SF (100 ng/mL) by quantitative RT-PCR and immunoblotting. All five cell lines for 48 hours in 10% serum before treatment with campthotecin (20 Amol/L examined (DAOY, D425, D283, PFSK, and ONS-76) expressed for D425 cells and 0.5 Amol/L for ONS-76 cells) for 48 hours. The fraction of moderate to high levels of c-Met and SF/HGF protein compared phosphatidylserine-positive cells was determined by Annexin V labeling with the levels observed in glioma cells (Fig. 1A). All eight tumor using the Annexin V-FITC Detection kit according to the instructions of the specimens that were analyzed by immunoblotting expressed both manufacturer (BD PharMingen, San Diego, CA). Briefly, cells were pelleted and then resuspended in 100 AL binding buffer and 5 AL Annexin V-FITC c-Met and SF/HGF protein (Fig. 1A). All 32 additional embryonal and 10 AL propidium iodide were added, and the samples were incubated CNS tumor specimens that were analyzed by RT-PCR were found to for 15 minutes at room temperature in the dark. The cell suspensions were express c-Met mRNA and all but one tumor expressed SF/HGF diluted with 400 AL binding buffer and were immediately analyzed with flow mRNA (not shown). We also examined expression of c-Met protein cytometry as described above. in human medulloblastoma samples immunohistochemically. Lactate dehydrogenase assays. To further determine the effects of SF/ Consistent with the immunoblot data, all but one of 20 medullo- HGF treatment on cell viability, lactate dehydrogenase (LDH) assays were blastomas examined contained cytoplasmic c-Met immunoreac- done using CytoTox96Nonradioactive Cytotoxicity Assay (Promega, Madi- tivity (not shown). These data show that SF/HGF and c-Met are son, WI). Cells were plated at 0.1 Â 105 per well in a 24-well plate. Cells were widely expressed in human embryonal CNS tumor specimens and then preincubated with SF/HGF for 48 hours and subsequently treated with campthotecin as described above for 48 hours. Cells in each well were cultured cells. assayed for LDH release as follows: 50 AL supernatant aliquots were c-Met messenger RNA expression is associated with poor transferred to 96-well assay plates with 50 AL of reconstituted substrate mix clinical outcome. We used real-time RT-PCR to measure the level and incubated at room temperature for 30 minutes in the dark. Stop of c-Met in mRNA extracted from snap-frozen tumor tissue. All solution (50 AL) was added to each well and the absorbance was read at embryonal CNS tumors examined (n = 32) contained detectable 490 nm by an automatic spectrometer. levels of c-Met mRNA. We divided these tumors into those with Tumor formation in vivo. The effects of c-Met activation on in vivo above and below median c-Met expression. Of the 16 tumors that medulloblastoma xenograft growth were tested in s.c. and intracranial exhibited higher than median expression levels of c-Met, 11 also animal models. Due to slow tumor formation in vivo, DAOY cells were had higher than median expression levels of SF/HGF. Log-rank mixed with an equal volume of Matrigel before s.c. and intracranial implantations. For s.c. xenografts, DAOY-SF medulloblastoma cells (5 Â 106 analysis of Kaplan-Meier survival curves revealed that the high cells) were implanted into the flanks of Nu/Nu immunodeficient mice. c-Met patient group had statistically significant shorter survival Tumor sizes were monitored by direct measurement with calipers and compared with low c-Met patients (P < 0.05; Fig. 1B). We also tumor volumes were calculated by using the following formula: volume = analyzed SF/HGF mRNA expression in this tumor set, and detected (length Â width2) / 2. For intracranial experiments, tumors cells (1 Â 105) expression of the ligand in all but one of the cases examined. were stereotactically implanted into the striatum of nude mice and animals Whereas not statistically significant, there was a trend toward were sacrificed at day 21. The brains were removed, sectioned, and stained shorter survival in patients whose tumors contained above median with H&E. Maximal tumor cross-sectional area was measured by computer- levels of SF/HGF (P = 0.064; Fig. 1B). assisted image analysis and tumor volumes were calculated. Scatter factor/hepatocyte growth factor activates c-Met and Statistical analysis. Two-group comparisons were analyzed by two- sided Student’s t test. Multiple-group comparisons were done with downstream signaling in medulloblastoma cells. To assess the Bonferroni/Dunn multiple comparison tests. Survival data were analyzed functionality of the c-Met receptor, we examined the effects of SF/ with log-rank analysis of Kaplan-Meier survival curves. P values were HGF on the activation of c-Met and its downstream signaling by determined for all analyses and P < 0.05 was considered significant. analyzing c-Met, MAPK, and Akt phosphorylation in response to

Figure 1. Human medulloblastoma tumors and cells express c-Met and SF/HGF protein and high c-Met expression is associated with poor clinical outcome. A, cell extracts were isolated from embryonal CNS tumor surgical specimens and cells and analyzed for SF/HGF and c-Met protein by immunoblotting. All tumors and cell lines expressed both c-Met and SF/HGF. B, log-rank analysis of Kaplan-Meier survival curves for patients with above median (high Met) or below median (low Met) c-Met mRNA levels determined by real-time RT-PCR reveals significantly worse clinical outcomes for patients with high c-Met–expressing tumors.

www.aacrjournals.org 9357 Cancer Res 2005; 65: (20). October 15, 2005

independent growth. SF/HGF significantly increased the number of formed colonies from 4.0 F 2.0 to 37.0 F 5.9 (n =6;P < 0.05) in DAOY cells and from 12.2 F 2.6 to 27.7 F 6.7 (n =6;P < 0.05) in D425 cells (Fig. 3B). These data show that SF/HGF induces embryonal CNS tumor cell proliferation and anchorage- independent growth. Scatter factor/hepatocyte growth factor induces cell cycle progression in medulloblastoma cells in a p27- and Cdk2- dependent manner. Using flow cytometry, we found that SF/HGF allows DAOY and ONS-76 medulloblastoma cells but not D425 cells to escape G0-G1 arrest induced by serum withdrawal, indicating that SF/HGF can enhance medulloblastoma growth partly by promoting tumor cell proliferation (Fig. 4A). SF/HGF reduced the F F G1-G0 fraction from 66.6 1.6% to 50.3 1.7% (P < 0.05) in DAOY Figure 2. SF/HGF activates c-Met and downstream signaling pathways in cells and from 71.7% to 63.2% (n = 2) in ONS-76 cells. Treatment of human medulloblastoma cells: DAOY, D425, D283, and ONS-76 DAOY cells with SF/HGF reduced p27 and induced Cdk2, cyclin D1, medulloblastoma cells were treated with recombinant SF/HGF and analyzed for cyclin E, and E2F-1 protein levels as measured by immunoblotting phospho-Met, phospho-MAPK, and phospho-Akt by immunoblotting. Blots were also hybridized to total MAPK and total Akt as control. (Fig. 4B). Additionally, transfection of DAOY cells with the p27 degradation stable mutant p27T187A or a Cdk2 dominant-negative (Cdk2-DN) inhibited SF/HGF-mediated cell cycle progression. F SF/HGF treatment in DAOY, D425, D283, and ONS-76 cells. SF/HGF Whereas SF/HGF reduced the G1-G0 fraction from 56.0 1.8 to strongly induced c-Met, MAPK, and Akt phosphorylation in all four 44.0 F 1.0 (n =3;P < 0.05) in control transfected cells, no reduction cell lines examined, indicating that the SF/HGF:c-Met pathway is was measured in Cdk2-DN– and p27T187A-transfected cells functional and leads to the activation of c-Met–dependent signal (Fig. 4C). This shows that the SF/HGF-induced cell cycle response transduction in human embryonal CNS cells (Fig. 2). requires p27 degradation and Cdk2 function. Scatter factor/hepatocyte growth factor induces medullo- Scatter factor/hepatocyte growth factor protects medulloblastoma cell proliferation and anchorage-independent blastoma cells against chemotherapy-induced cell death in a growth. SF/HGF induced proliferation in DAOY and ONS-76 but phosphoinositide 3-kinase–dependent manner. We found that not in D425 cells. Whereas control DAOY cells did not grow under SF/HGF protects D425 and ONS-76 but not DAOY medulloblasto- low serum (0.1%) conditions, addition of 10 ng/mL recombinant ma cells against chemotherapy-induced cell death. Pretreatment of SF/HGF dramatically and significantly increased DAOY cell D425 or ONS-76 cells with SF/HGF before treatment with numbers relative to control (Fig. 3A). Anchorage-independent campthotecin reduced the fraction of dead cells from 52.5 F growth is a well-characterized in vitro correlate of malignancy. 3.0% to 41.5 F 1.4% (P < 0.05) in D425 cells and from 63.7 F 2.0 to Using colony formation in soft agar assays, we found that SF/HGF 46.7 F 2.2 (P < 0.05) in ONS-76 cells as assessed by Annexin V flow strongly induces DAOY and D425 medulloblastoma cell anchorage- cytometry (Fig. 5A). Cytoprotection was also confirmed by LDH

Figure 3. SF/HGF induces cell proliferation and anchorage-independent growth in medulloblastoma cells: A, SF/HGF induces the proliferation of DAOY and ONS-76 medulloblastoma cells. Cells (5,000 per well; n = 3 wells) were seeded in media containing 1% serum and treated with or without 10 ng/mL SF/HGF. Cells were counted and growth curves were established. B, assay of clonogenicity in soft agar showing that SF/HGF induces anchorage-independent growth. DAOY and D425 cells were suspended in soft agar and treated with 10 ng/mL SF/HGF for 3 weeks. Colonies >100 Am diameter were measured and counted by computer-assisted image. *P < 0.05.

Cancer Res 2005; 65: (20). October 15, 2005 9358 www.aacrjournals.org

Figure 4. SF/HGF induces cell cycle progression in medulloblastoma cells in a p27- and Cdk2-dependent manner: A, flow cytometric analysis of the cell cycle in DAOY medulloblastoma cells showing a reduction of the G0-G1 fraction and increase of S and G2-M fractions in cells treated with 10 ng/mL SF/HGF. B, immunoblots showing the activation of cell cycle regulatory proteins by SF/HGF in DAOY cells. C, flow cytometric analysis of DAOY cells transfected with Cdk2-DN or p27 degradation stable mutant (p27T187A) constructs before treatment with SF/HGF (right). The results show that SF/HGF induction of cell cycle progression is dependent on p27 and Cdk2 function. *P < 0.05.

assay (not shown). Basal apoptosis/necrosis levels were slightly but toma subtype is characterized by increases in cell size, mitotic rate, not statistically significantly inhibited by SF/HGF pretreatment of apoptotic rate, and prominent nucleoli. Xenografts from wild-type the cells. Consistent with the above findings, SF/HGF decreased the or vector-transfected DAOY cells were composed of round to levels of the proapoptotic proteins cleaved PARP and cleaved spindled cells with vesicular chromatin, interspersed with occa- caspase-3 in cells treated as described above (Fig. 5B). Pretreat- sional groups of larger cells with more prominent nucleoli. Single ment of D425 cells with the PI3K inhibitors LY 294002 or apoptotic cells were present, but clusters of apoptotic cells were wortmannin, but not with the MAPK kinase inhibitor PD-98059, generally not seen. Xenografts from DAOY-SF cells contained completely inhibited the cytoprotective effect of SF/HGF, showing numerous enlarged cells, often with macronucleoli. Apoptotic that this effect is PI3K dependent (Fig. 5C). bodies were common, as were large ‘‘apoptotic lakes’’ like those Scatter factor/hepatocyte growth factor strongly enhances seen in human large cell anaplastic medulloblastoma. These the growth of s.c. and intracranial medulloblastoma tumor tumors thus recapitulate several features of human large cell xenografts. DAOY cells engineered to overexpress SF/HGF (DAOY- anaplastic medulloblastoma. Additionally, of the five anaplastic SF) or control cells expressing empty plasmids (DAOY-Control) medulloblastomas among the human tumor specimens described were implanted s.c. in immunodeficient mice. DAOY-SF cells above, four expressed higher than median levels of SF/HGF and exhibited significantly faster tumor growth rates than tumors two expressed higher than median levels of both SF/HGF and generated form DAOY-Control cells. Also, DAOY-SF xenograft c-Met. These findings are consistent with a role of SF/HGF in growth rates were proportional to the SF/HGF expression levels generating this most malignant medulloblastoma subtype. of the transfected cells. DAOY-Control cells exhibited a very slow growth rate or did not grow at all (Fig. 6A). DAOY-SF cells were stereotactically injected into the brains of Discussion immunodeficient mice. After 3 weeks, DAOY-SF xenografts grew This is, to the best of our knowledge, the first study that significantly larger (maximal tumor cross-sectional area = 17.8 F examines the role of the SF/HGF:c-Met pathway in human 2.9, n = 8) than xenografts established from DAOY-Control cells embryonal CNS tumor malignancy. The study shows a role for (maximal tumor cross sectional area = 0.64 F 0.19, n = 9; Fig. 6B). SF/HGF in embryonal CNS tumor growth and establishes the c-Met DAOY-SF xenografts also showed more invasive characteristics pathway as an interesting new potential target for experimental with fingerlike protrusions, metastatic growth, and leptomeningeal therapies. spread as opposed to DAOY-Control xenografts that exhibited a The embryonal CNS tumor cell lines used in this study were more defined and limited growth (Fig. 6B). carefully selected to represent various embryonal CNS tumor When examined histologically, s.c. as well as intracranial DAOY- subtypes and disease models. The ONS-76 cell line was derived SF xenografts exhibited dramatic histologic differences compared from a case of classic medulloblastoma and has also been used as with DAOY-Control xenografts. DAOY-SF xenografts displayed an experimental model of leptomeningeal dissemination (32, 33). features that closely resemble those of human large cell anaplastic Using expression profiling, McDonald et al. (34) compared DAOY medulloblastoma (Fig. 6C). The large cell anaplastic medulloblas- cells to medulloblastoma tumors and found them to be excellent www.aacrjournals.org 9359 Cancer Res 2005; 65: (20). October 15, 2005

Figure 5. SF/HGF protects medulloblastoma cells against chemotherapy-induced apoptosis in a PI3K-dependent manner: A, flow cytometric analysis of Annexin V–labeled D425 and ONS-76 cells that were pretreated with SF/HGF before treatment with the chemotherapeutic agent camptothecin (CPT). The results show that SF/HGF reduces the fraction of apoptotic cells. B, D425 and ONS-76 cells treated as in (A) were also analyzed by immunoblotting for mediators of apoptosis cleaved PARP and cleaved caspase-3. The results show that SF/HGF decreases the levels of these proapoptotic proteins in medulloblastoma cells. C, pretreatment of D425 cells with the PI3K inhibitors LY294002 and wortmannin before treatment with SF/HGF and campthotecin as in (A) completely inhibits the cytoprotective effects of SF/HGF in these cells. *P < 0.05.

representatives of metastatic medulloblastoma. D283 and D425, reports have shown an emerging role for the receptor tyrosine two well-characterized medulloblastoma lines, are phenotypically kinase pathways FGFR, PDGFR, Trk, and erbB in medulloblastoma different form each other as well as from DAOY (35, 36). The PFSK (34, 39–41). However, the role of c-Met receptor tyosine kinase in cell line is derived from a PNET tumor from the cerebral embryonal CNS tumors remained unexplored until now. Trks are hemisphere and exhibits tumor stem cell characteristics (37). Also, the most studied receptor tyrosine kinases in medulloblastoma. all WHO types and subtypes of embryonal CNS tumors were A few studies have shown that TrkC receptor expression is represented in the human surgical tumor specimens used in the associated with better prognosis of medulloblastoma patients and study. We found that c-Met and SF/HGF are expressed in all the that activation of TrkC induces apoptosis of tumor cells (40, 42). above-mentioned human embryonal CNS tumor cells and surgical Two reports also linked FGFR expression with antitumor activity in specimens. c-Met receptors were functional in all cell types classic medulloblastoma subtype (39, 43). Therefore, unlike c-Met examined as evidenced by the SF/HGF-induced phosphorylation activation, which enhances embryonal CNS tumor growth and of the receptor and associated downstream signaling molecules. malignancy, activation of Trk and FGF receptors is associated with This widespread expression and activation of c-Met is indicative of reduced tumor malignancy. Conversely, ERBB2 and PDGFR a critical role for the pathway in these tumors and is consistent receptors have been linked to increased malignancy of medullo- with its expression and functional activity in cerebellar granule blastoma. Both ERBB2 and PDGFR expression in medulloblastoma cells, which are believed to be the cell precursors for medulloblas- have been associated with poor prognosis, metastatic phenotype, toma (38). SF/HGF and c-Met were expressed in the same cells and enhanced medulloblastoma cell migration (34, 44, 45). indicative of autocrine activation of the receptor. Autocrine loop Although our study did not systematically investigate tumor cell formation has been associated with increased malignancy in other migration, DAOY-SF xenografts clearly displayed a more invasive human tumors. Importantly also, c-Met expression levels in tumors phenotype than controls, indicating that SF/HGF also enhances significantly correlated with poor patient survival, indicating that embryonal CNS tumor migration and invasiveness in brain. We c-Met levels could serve as a prognostic factor for medulloblasto- show a role for c-Met activation in regulating tumor cell cycle mas and providing the basis for subsequent mechanistic studies of progression, proliferation, and survival mechanisms. Considering the role of c-Met in promoting tumor growth. that SF/HGF is also a well-known and potent angiogenic factor, SF/ The molecular mechanisms that govern medulloblastoma HGF might, therefore, enhance medulloblastoma growth by acting growth and malignancy are only partially understood. A few on multiple malignancy parameters. This would explain the very

Cancer Res 2005; 65: (20). October 15, 2005 9360 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. c-Met Pathway in Embryonal CNS Tumors strong in vivo tumor-promoting effect of SF/HGF and render SF/ end points in ONS-76 cells. This heterogeneous response to SF/ HGF:c-Met a promising target for therapy. HGF might be due to the different origins of the cell lines and/or Medulloblastoma tumors exhibit a high proliferative index. also reflect the cellular heterogeneity that is observed in em- c-Met activation in medulloblastoma cells induced anchorage- bryonal CNS tumors. independent growth, cell proliferation, and cell cycle progression. SF/HGF gene transfer to DAOY medulloblastoma cells led to a SF/HGF-induced cell cycle progression required inhibition of the strong induction of in vivo xenograft growth. SF/HGF also tumor suppressor p27 as well as induction of Cdk2 and E2F-1. SF/ enhanced medulloblastoma tumorigenicity as evidenced by the HGF protected medulloblastoma cells against cell death and transformation of DAOY from low tumorigenic in experimental apoptosis induced by chemotherapy. This effect was mediated by animals to cells that form large and invasive xenografts. SF/HGF PI3K. This cytoprotective effect indicates that targeting SF/HGF:c- also changed the cell and tumor morphology. We found that SF/ Met could sensitize embryonal CNS tumors to chemotherapy and HGF-overexpressing DAOY xenografts exhibit dramatic histologic radiotherapy as already shown by our group in glioma (46). differences compared with control xenografts. DAOY-SF/HGF Therefore, the pathways that mediate SF/HGF effects on xenografts had enlarged cells with large nuclei and prominent embryonal CNS tumors are similar to the ones observed in other nucleoli (Fig. 6). These features closely resemble the ones observed human tumors, including gliomas. Of note is that SF/HGF did not in human large cell anaplastic medulloblastoma, which possesses affect all malignancy end points equally in all cell lines that were the worst prognosis among medulloblastoma subtypes. Overall, the tested. In fact, SF/HGF induced cell proliferation and cell cycle tumor growth pattern was also changed by overexpression of progression but not cytoprotection in DAOY cells. Conversely, SF/ SF/HGF. Unlike control xenografts, SF/HGF-overexpressing xeno- HGF induced cytoprotection but not cell cycle progression and grafts formed fingerlike protrusions and metastasis and invaded cell proliferation in D425 cells. SF/HGF affected all three above the subarachnoidal space. These characteristics are typical for

Figure 6. SF/HGF strongly enhances the growth of s.c. and intracranial medulloblastoma tumor xenografts and alters tumor morphology and histology: A, DAOY medulloblastoma cells were engineered to stably express full-length human SF/HGF. Stably transfected cells (mixed transfectants—not clones) were screened for SF/HGF protein levels by immunoblotting (right) and designated SF/HGF/A and SF/HGF/B. Control cultures (Control/A and Control/B) were transfected with empty vectors. All cultures were mixed with Matrigel (50% volume) and injected in the flanks of nude mice (5 Â 106 cells/animal; n = 9). Tumor growth was monitored by caliper measurements. The results show that tumor growth rates correlate with SF/HGF expression. B, SF/HGF-overexpressing and control-transfected DAOY medulloblastoma cells were stereotactically injected into the brains of immunodeficient mice. The animals (n = 10) were sacrificed after 4 weeks and maximum cross-sectional tumor areas were measure on H&E-stained brain cross sections. SF/HGF-overexpressing cultures generated significantly larger tumors than controls. SF/HGF transfectants were also more invasive and metastasized in the subarachnoidal space (arrows). C, histologic analysis of DAOY-SF xenografts showing enlarged cells with large nuclei and prominent nucleoli. These features closely resemble human large cell anaplastic medulloblastoma.

www.aacrjournals.org 9361 Cancer Res 2005; 65: (20). October 15, 2005

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Cancer Research large-cell anaplastic human medulloblastomas. Therefore, c-Met multifunctional tumor-promoting effects. Our findings provide activation in medulloblastoma might lead to the generation of this preclinical evidence that the SF/HGF:c-Met pathway constitutes a most aggressive medulloblastoma subtype. Large-cell anaplastic promising new therapeutic target in embryonal CNS tumors and human medulloblastoma is also associated with c-Myc amplifica- support testing of emerging c-Met pathway inhibitors, such as tion or overexpression. We have previously shown that c-Myc small molecule inhibitors and neutralizing anti-SF/HGF monoclo- induction by SF/HGF mediates G1-S transition in human glioma nal antibodies, in these common childhood malignancies. cells. However, a relationship between c-Met activation and c-Myc overexpression in large cell anaplastic medulloblastoma is pre- sently undefined and is subject to ongoing investigation in our Acknowledgments laboratory. Received 6/3/2005; revised 8/1/2005; accepted 8/16/2005. In conclusion, we have established for the first time a role for the Grant support: Children’s Cancer Foundation (R. Abounader), NIH grants RO1 SF/HGF:c-Met pathway in human embryonal CNS tumor malig- NS045209 (R. Abounader), and RO1 NS032148 and NS 43987 (J. Laterra). The costs of publication of this article were defrayed in part by the payment of page nancy. We showed that c-Met levels correlate with patient charges. This article must therefore be hereby marked advertisement in accordance prognosis and that activation of the pathway has widespread and with 18 U.S.C. Section 1734 solely to indicate this fact.

References 17. Walter KA, Hossain MD, Luddy C, Goel N, Reznick cyte growth factor in human glioblastoma. J Neurochem TE, Laterra J. Scatter factor/hepatocyte growth factor 2001;76:1497–508. 1. Reddy AT, Packer RJ. Medulloblastoma. Curr Opin stimulation of glioblastoma cell cycle progression 32. Yamada M, Shimizu K, Tamura K, et al. [Establish- Neurol 1999;12:681–5. through G1 c-Myc dependent, and independent of p27 ment and biological characterization of human medul- 2. Ellison D. Classifying the medulloblastoma: insights suppression cdk2 activation, or E2F1-dependent tran- loblastoma cell lines]. No To Shinkei 1989;41:695–702. from morphology and molecular genetics. Neuropathol scription. Mol Cell Biol 2002;22:2703–15. 33. Yamada M, Shimizu K, Tamura K, et al. Murine Appl Neurobiol 2002;28:257–82. 18. Lamszus K, Liang J, Laterra J, et al. Scatter factor model of leptomeningeal dissemination using human 3. Taylor MD. Mainprize TG, Rutka JT. Molecular insight promotes motility of human glioma and neuromicro- medulloblastoma cells. Neurol Med Chir (Tokyo) 1991; into medulloblastoma and central nervous system vascular endothelial cells. Int J Cancer 1998;75:19–28. 31:763–7. primitive neuroectodermal tumor biology from hereditary 19. Lamszus K, Laterra J, Westphol M, Rosen EM. Scatter 34. MacDonald TJ, Brown KM, LaFleur B, et al. syndromes: a review. Neurosurgery 2000;47:888–901. factor/hepatocyte growth factor content and function in Expression profiling of medulloblastoma: PDGFRA and 4. Wechsler-Reya R, Scott MP. The developmental human gliomas. Int J Dev Neurosci 1999;17:517–30. the RAS/MAPK pathway as therapeutic targets for biology of brain tumors. Annu Rev Neurosci 2001;24: 20. Schmidt MO, Westphal M, Hagel C, et al. Levels of metastatic disease. Nat Genet 2001;29:143–52. 385–428. vascular endothelial growth factor, hepatocyte growth 35. He XM, Skapek SX, Wikstrand CJ, et al. Phenotypic 5. Packer RJ. Brain tumors in children. Arch Neurol factor/scatter factor and basic fibroblast growth factor analysis of four human medulloblastoma cell lines and 1999;56:421–5. in human gliomas and their relationship to angiogen- transplantable xenografts. J Neuropathol Exp Neurol 6. Newton HB. Review of the molecular genetics and esis. Int J Cancer 1999;84:10–8. 1989;48:48–68. chemotherapeutic treatment of adult and paediatric 21. Rosen EM, Goldberg ID. Regulation of angiogenesis 36. Wikstrand CJ, Friedman HS, Bigner DD. Medullo- medulloblastoma. Expert Opin Investig Drugs 2001;10: by scatter factor. Basel (Switzerland): Birkha¨user Verlag; blastoma cell-substrate interaction in vitro. Invasion 2089–104. 1997. Metastasis 1991;11:310–24. 7. Birchmeier C, Birchmeier W, Gherardi E, Vande 22. Tanimura S, Chantani Y, Hoshino R, et al. 37. Fults D, Pedone CA, Morse HG, Rose JW, McKay RD. Woude GF. Met, metastasis, motility and more. Nat Activation of the 41/43 kDa mitogen-activated protein Establishment and characterization of a human prim- Rev Mol Cell Biol 2003;4:915–25. kinase signaling pathway is required for hepatocyte itive neuroectodermal tumor cell line from the cerebral 8. Matsumoto K, Nakamura T. Emerging multipotent growth factor-induced cell scattering. Oncogene 1998; hemisphere. J Neuropathol Exp Neurol 1992;51:272–80. aspects of hepatocyte growth factor. J Biochem 1996; 17:57–65. 38. Hossain MA, Bailone JC, Gomes R, Laterra J. Neuro- 119:591–600. 23. Sipeki S, Bander E, Buday L, et al. Phosphatidylino- protection by scatter factor/hepatocyte growth factor 9. Bottaro DP, Rubin JS, Faletto DL, et al. Identification of sitol 3-kinase contributes to Erk1/Erk2 MAP kinase and FGF-1 in cerebellar granule neurons is phosphati- the hepatocyte growth factor receptor as the c-met activation associated with hepatocyte growth factor- dylinositol 3-kinase/Akt-dependent and MAPK/CREB- proto-oncogene product. Science 1991;251:802–4. induced cell scattering. Cell Signal 1999;11:885–90. independent. J Neurochem 2002;81:365–78. 10. Wang R, Ferrell LD, Faouzi S, Maher JJ, Bishop JM. 24. Graziani A, Gramaglia D, Cantley LC, Comoglio PM. 39. Duplan SM, Theoret Y, Kenigsberg RL. Antitumor Activation of the Met receptor by cell attachment The tyrosine-phosphorylated hepatocyte growth factor/ activity of fibroblast growth factors (FGFs) for medul- induces and sustains hepatocellular carcinomas in scatter factor receptor associates with phosphatidylino- loblastoma may correlate with FGF receptor expression transgenic mice. J Cell Biol 2001;153:1023–34. sitol 3-kinase. J Biol Chem 1991;266:22087–90. and tumor variant. Clin Cancer Res 2002;8:246–57. 11. Jeffers M, Schmidt L, Nakaigawa N, et al. Activating 25. Ponzetto C, Bardelli A, Zhen Z, et al. A multifunc- 40. Kim JY, Sutton ME, Lu DJ, et al. Activation of mutations for the met tyrosine kinase receptor in tional docking site mediates signalling and transforma- neurotrophin-3 receptor TrkC induces apoptosis in human cancer. Proc Natl Acad Sci U S A 1997;21: tion by the HGF/SF receptor family. Cell 1994;77:261–71. medulloblastomas. Cancer Res 1999;59:711–9. 11445–50. 26. Xiao GH, Jeffers M, Bellacosa A, et al. Anti-apoptotic 41. Nakagawara A. Trk receptor tyrosine kinases: a 12. Danilkovitch-Miagkova A, Zbar B. Dysregulation of signaling by hepatocyte growth factor/Met via the bridge between cancer and neural development. Cancer Met receptor tyrosine kinase activity in invasive tumors. phosphatidylinositol 3-kinase/Akt and mitogen-activat- Lett 2001;169:107–14. J Clin Invest 2002;109:863–7. ed protein kinase pathways. Proc Natl Acad Sci U S A 42. Segal RA, Goumnerova LC, Kwon YK, Stiles CD, 13. Laterra J, Nam M, Rosen E, Rao JS, Johnston P. 2001;98:247–52. Pomeroy SL. Expression of the neurotrophin receptor Scatter factor/hepatocyte growth factor gene transfer to 27. Fan S, Ma YX, Wang J, et al. The cytokine scatter TrkC is linked to a favorable outcome in medulloblas- 9L glioma cells enhances glioma growth and angiogen- factor inhibits apoptosis and enhances DNA repair by a toma. Proc Natl Acad Sci U S A 1994;91:12867–71. esis in vivo. Lab Invest 1997;76:565–77. common mechanism involving signaling through phos- 43. Kenigsberg RL, Hong Y, Yao H, et al. Effects of basic 14. Abounader R, Ranganathan S, Lal B, et al. Reversion phatidyl inositol 3V kinase. Oncogene 2000;19:2212–23. fibroblast growth factor on the differentiation, growth, of human glioblastoma malignancy by U1 small nuclear 28. van den Heuvel S, Harlow E. Distinct roles for cyclin- and viability of a new human medulloblastoma cell line RNA/ribozyme targeting of scatter factor/hepatocyte dependent kinases in cell cycle control. Science 1993; (UM-MB1). Am J Pathol 1997;151:867–81. growth factor and c-met expression. J Natl Cancer Inst 262:2050–4. 44. Gilbertson RJ, Clifford SC, MacMeekin W, et al. 1999;91:1548–56. 29. Sheaff RJ, Groudine M, Gordon M, Roberts JM, Expression of the ErbB-neuregulin signaling network 15. Abounader R, Lal B, Luddy C, et al. In vivo targeting Clurman BE. Cyclin E-CDK2 is a regulator of p27Kip1. during human cerebellar development: implications for of SF/HGF and c-met expression via U1snRNA/ribo- Genes Dev 1997;11:1464–78. the biology of medulloblastoma. Cancer Res 1998;58: zymes inhibits glioma growth and angiogenesis and 30. Fan X, Wang Y, Kratz J, et al. hTERT gene 3932–41. promotes apoptosis. FASEB J 2002;16:108–10. amplification and increased mRNA expression in central 45. Hernan R, Fasheh R, Calabrese C, et al. ERBB2 up- 16. Bowers DC, Fan S, Walter K, et al. Scatter factor/ nervous system embryonal tumors. Am J Pathol 2003; regulates S100A4 and several other prometastatic genes hepatocyte growth factor activates AKT and protects 162:1763–9. in medulloblastoma. Cancer Res 2003;63:140–8. against cytotoxic death in human glioblastoma via PI3- 31. Abounader R, Ranganathan S, Kim BY, Nichols C, 46. Lal B, Xia C, Abounader R, Laterra J. Targeting the kinase and AKT-dependent pathways. Cancer Res 2000; Laterra J. Signaling pathways in the induction of c-met c-Met pathway potentiates glioblastoma responses to 60:4277–83. receptor expression by its ligand scatter factor/hepato- g-radiation. Clin Cancer Res 2005;11:4479–86.

Cancer Res 2005; 65: (20). October 15, 2005 9362 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. The Scatter Factor/Hepatocyte Growth Factor: c-Met Pathway in Human Embryonal Central Nervous System Tumor Malignancy

Yunqing Li, Bachchu Lal, Sherwin Kwon, et al.

Cancer Res 2005;65:9355-9362.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/20/9355

Cited articles This article cites 44 articles, 14 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/20/9355.full#ref-list-1

Citing articles This article has been cited by 20 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/20/9355.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/20/9355. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.