MET inhibition overcomes radiation resistance of glioblastoma stem‐like cells

Abstract Glioblastoma (GBM) contains stem‐like cells (GSCs) known to be resistant to ionizing radiation and thus responsible for therapeutic failure and rapidly lethal tumor recurrence. It is known that GSC radioresistance relies on efficient activation of the DNA damage response, but the mechanisms linking this response with the stem status are still unclear. Here, we show that the MET receptor kinase, a functional marker of GSCs, is specifically expressed in a subset of radioresistant GSCs and overexpressed in human GBM recurring after radiotherapy. We elucidate that MET promotes GSC radioresistance through a novel mechanism, relying on AKT activity and leading to (i) sustained activation of Aurora kinase A, ATM kinase, and the downstream effectors of DNA repair, and (ii) phosphorylation and cytoplasmic retention of p21, which is associated with anti‐apoptotic functions. We show that MET pharmacological inhibition causes DNA damage accumulation in irradiated GSCs and their depletion in vitro and in GBMs generated by GSC xenotransplantation. Preclinical evidence is thus provided that MET inhibitors can radiosensitize tumors and convert GSC‐positive selection, induced by radiotherapy, into GSC eradication.


Appendix Fig S6. MET inhibition impairs AKT-dependent ATM activity.
A, Western blot of BT308 NS showing phosphorylation of ATM (pATM) and Chk2 (pChk2), and accumulation of RAD51 at the indicated time points after IR in the absence (5 Gy

Neurosphere derivation from human patients and culture
Surgical samples were obtained from consecutive primary GBM provided by the Fondazione IRCCS Istituto Neurologico C. Besta, according to a protocol approved by the institutional Ethical Committee. All Patients signed an informed consent. NS were derived as described (De Bacco et al., 2012), and plated at clonal density in standard NS medium containing human recombinant bFGF (20 ng/ml) and EGF (20 ng/ml). For in vitro experiments, NS were kept in a modified medium with a lower concentration of EGF and bFGF (10 ng/ml), and supplemented with human recombinant HGF (10 ng/ml). Both media sustain the same NS proliferation rate (Appendix Fig   S9A). For immortalized cell lines, a passage below 15 was used.

NS irradiation and treatment with inhibitors
NS were irradiated with a 200 kV X-ray blood irradiator (Gilardoni), and a 1 Gy/min dose-rate. In (an average of 95% and 99% was obtained).

Xenografts models
All animal procedures were approved by the internal Ethical Committee for Animal Experimentation (FPRC-CESA) and by the Italian Ministry of Health. Mus musculus, female NOD.CB17-Prkdc scid /NcrCr mice (Charles River Laboratories) were kept and manipulated under pathogen-free conditions. In details, mice were housed in filtered bottomed, individually sterilized and ventilated cages. Every cage contained a maximum of 6 mice and suitable amounts of sterilized food, water and bedding. Cages were changed weekly to prevent the introduction of minimalinoculating doses of opportunistic or commensal organisms in the cage environment, and mice were manipulated under a laminar-flow hood.
To generate xenograft models, neurospheres were dissociated into single-cell suspensions, and injected subcutaneously or orthotopically into 6-8 weeks old female NOD-SCID mice. Accordingly, at least 6 mice/group were used.

Nucleic acid extraction
From NS, nucleic acids were extracted as follows: genomic DNA (gDNA), using the Wizard® Genomic DNA Purification kit (Promega); total RNA, using the mRNeasy Mini kit (Qiagen), according to manufacturer's instructions. Extracted purified nucleic acids were quantified with Nanodrop ND1000 (Thermo Scientific). polysomy, EGFR copy number was normalized against the copy number of a gene mapped on chr7

Gene sequencing and evaluation of copy number alteration
and usually not amplified (HGF), and defined amplified when EGFR/HGF copy number was > 5.
Reported values are the mean ± SEM of two independent experiments in triplicate.

Microarray analysis
Gene expression profiling and NS classification according to Verhaak's subtypes (Verhaak et al., 2010) was performed as previously described (De Bacco et al., 2012). nucleoid with long tail (long comet): severe DNA DSBs (representative images of nucleoid with no tail, and short and long comet were shown in Appendix Fig S9B). Experiments were repeated at least twice in duplicate, and a representative experiment was shown.

Western blotting
Protein expression and phosphorylation were analyzed in whole-cell lysates solubilized in boiling

Immunophenotypical analysis and Fluorescence-Activated Cell Sorting
NS were dissociated and resuspended in PBS 3% BSA at a concentration of 2 × 10 5 cells/100 µl.
The following antibodies were used: phospho-H2AX-FITC (Ser139; #17-344, Merck Millipore); Olig2-Alexa-Fluor 488 (#MABN50A4, Merck Millipore); HGF R/c-MET (clone 95106; MAB3582, R&D System), conjugated by the manufacturer with PE or APC, or in house with PE-Cy7, according to manufacturer's instructions (R&D System). The three antibodies display comparable specificity (Appendix Fig S9C). The % of positive cells was calculated by comparing unstained vs. stained cells in each sample, as to avoid bias associated with morphological, and ensuing autofluorescent, changes due to irradiation (Maecker and Trotter, 2006). Before analysis, DAPI (Roche) was added to exclude dead cells. Analysis was performed on a CyAn ADP

PKH-26 staining
Dissociated NS cells were stained with PKH-26 dye (1:2000, Sigma), according to manufacturer's instructions, and plated at clonal density. Derived NS were passaged for 2 weeks to allow adequate PKH-26 dilution before treatments. Percentage of PKH-26 positive cells was evaluated by flow cytometry as above. Experiments were repeated at least twice, and a representative experiment was shown.

Annexin V evaluation
Annexin V Apoptosis Detection Kit (BD Pharmingen TM ) was used according to manufacturer's instructions for apoptosis evaluation by flow cytometry. To discriminate apoptosis in MET high and MET neg subpopulation, cells were incubated with anti-MET-APC as above and washed twice with Binding Buffer prior to Annexin V staining. Reported values are the mean ± SEM of two independent experiments.

Cell cycle analysis
Cell cycle was analyzed according to standard procedures. Briefly, 10 6 cells were fixed with 70% Ethanol overnight at -20°C, stained with propidium iodide (50 µg/ml) in RNaseA solution (DNAcon3 kit ConsulTS) for 3 h at 4°C in the dark, and analyzed by flow-cytometry as above. Data are the mean ± SEM of at least two independent experiments.

Cell viability and caspase activation assays
To assess NS viability and apoptosis, ATP cell content and Caspase 3/7 activity were measured using Cell Titer Glo® and Caspase-Glo® 3/7 Assay (Promega) respectively, according to manufacturer's instructions. Cells were plated at clonal density (10 cells/µl) in 96-well plates, and treated as described 24 h after seeding (day 0). Cell viability and caspase 3/7 activity were measured using a GloMax 96 Microplate Luminometer (Promega). Data are reported with respect to non-treated cells as mean ± SEM of at least two independent experiments in quadruplicate.

Radiobiological clonogenic assay
Met high and Met neg cell subpopulations were sorted as described above and seeded as single cell/well. 24 h after sorting, cells were irradiated in the presence or in the absence of MET inhibitors, and then cultured for 14 days. The number of growing NS was counted, and the surviving fraction was calculated using the formula, modified from (Franken et al., 2006):

Limiting dilution sphere forming assay in vitro
Limiting dilution sphere forming assay was performed to assess the frequency of GSC in NS established from original tumors, or in cells freshly dissociated from xenografts obtained by NS transplantation. Viable cells (trypan blue exclusion test) were plated into 96-well plates at decreasing concentrations (ranging from 150 cells/well to 1 cell/well). The number of growing NS was counted 14-21 days after seeding. Data were evaluated through the ELDA software (http://bioinf.wehi.edu.au/software/elda/) (Hu et al., 2009), and reported as percentage of stem-like cells ± CI.

Serial transplantation assay
NS were irradiated in vitro (p0) as described above, and 24 h after irradiation 10 3 cells were resuspended in 100 µl v/v PBS/Matrigel, and subcutaneously injected in 4 mice (p1). Tumor formation was monitored once a week by caliper as described above. From p1 tumors, cells were dissociated as described above for human tumor specimens, and, when NS formed, 10 3 cells in 100 µl v/v PBS/Matrigel were subcutaneously injected in 6 mice (p2). From p2 tumors, cell isolation and NS formation was repeated as above, and cells were assessed by in vivo limiting dilution assay, by injecting 10, 10 2 , 10 3 , or 10 4 cells in 6 mice per condition (p3). Frequency of stem-like cells was calculated using ELDA software as above and reported as stem-like cells per 10,000 cells.

Limiting dilution assay in vivo
In vivo serial dilution tumor-propagating assay was performed to evaluate the relative frequency of GSCs in subcutaneous xenografts after radiosensitization treatment. Subcutaneous xenografts were obtained as described above. Mice were randomized into four groups: (i) vehicle treated; (ii) treated with IR (2 Gy × 3 consecutive days); (iii) treated with JNJ38877605 (50 mg/kg, by daily oral gavage); (iv) treated with association of JNJ38877605 and IR as above (combo). The inhibitor JNJ-38877605 was administered starting from the day before irradiation (day -1) and prolonged up to day 10 in which 50% of tumor volume regression was reached in the combo group, as compared to day 0. At this point, tumors from each group were explanted and dissociated to re-derive tumor cells immediately cultured in stem conditions as described above. After cell recovery (about 10 days),

single-cell suspensions of viable tumor-derived cells for each group (trypan blue exclusion test)
were injected in the right flank of NOD-SCID mice at the following dilution dose 10, 10 2 , 10 3 or 10 4 in 100 µl v/v PBS/Matrigel (6 mice/condition). The frequency of stem-like tumor propagating cells within heterogeneous cell populations was evaluate on the efficiency of secondary xenograft formation and data were evaluated through the ELDA software as above and reported as stem-like cells per 10,000 cells.

Irradiation and radiosensitization of xenograft models
To establish treatment groups and to minimize the effect of subjective bias, mice with either subcutaneous tumors of approximately 400 mm 3 , or orthotopic tumors with a BLI average radiance signal of 7 × 10 5 p/s/sr/cm 2 , were randomized and group allocated by LAS software (Baralis et al., 2012 In the orthotopic model, mice were euthanized at day 62 (at the onset of neurological symptoms in the control group). Explanted brains were immediately analyzed for GFP signal detection (e x 465 nm, e m 520 nm; IVIS® SpectrumCT). Data were collected by A.D.A. with no blinding.
Anti-mouse Alexa-Fluor 555 (1:1000, Molecular Probes by Life Technologies) was used as secondary antibody. Nuclei were counter-stained with DAPI, according to standard protocols.
Images were acquired with Leica DMI 3000 B or Leica TCS SP2 AOBS confocal laser-scanning microscope (Leica). Images have been analyzed with NIH ImageJ software (National Institutes of Health, Bethesda, MD); the Relative Fluorescence Intensity was estimated with Leica LAS AF lite software. For analysis of cytoplasmic/nuclear localization of p21 in irradiated MET-pos-NS, at least 10 HPF (63 × magnification) were evaluated; for MET quantification, 6 HPF (63 × magnification) in two sections from each tumor, for a total of 3 mice/group, were counted.

Association of MET expression with disease free and overall survival in public dataset
MET mRNA expression data were obtained from the public TCGA GBM dataset available on cBioPortal (http://www.cbioportal.org), selecting the provisional dataset (June 2015, Agilent microarray data, n = 401), and applying an mRNA Z-score threshold of ± 2. Disease-free and overall survival curves were obtained with the cBioPortal software using the Kaplan-Meier method (Cerami et al., 2012;Gao et al., 2013).

Collection of matched primary and recurrent human GBMs.
Surgical samples were derived from primary and matched recurrent GBMs surgically removed at the Spedali Civili di Brescia, according to a protocol approved by the institutional Ethical Committee. All Patients signed an informed consent. Patients were treated with standard fractionated radiotherapy (60 Gy) and concomitant chemotherapy (75 mg/m 2 of Temozolomide) on daily basis for 6 weeks, and then with Temozolomide alone (150-200 mg/m 2 ) for 5 days in a 28-day for 6 cycles, and up to 12 cycles, if no treatment related adverse events were noted or there was no evidence of tumor progression based on both clinical evaluation and MRI findings. At tumor progression, second surgery was performed, and a second line treatment was offered to selected Patients (Fotemustine).

Immunohistochemical staining of MET on matched primary/recurrent GBMs.
MET immunohistochemical staining were performed on representative paraffin embedded sections (2 µm thick), incubated with rabbit polyclonal anti-Met C12 (1:50 in TBS/1% BSA, Santa Cruz Biotechnology), after blocking of endogenous peroxidase activity with 0.3% H 2 O 2 in methanol and antigen retrieval in 1mM Citrate buffer (pH 6.0) in a thermostatic bath. Signal was revealed using the NovoLinkTM Polymer Detection System (NovocastraTM), followed by Diaminobenzydine (DAB) as chromogen and Hematoxylin as counterstain. Images were acquired with an Olympus Bx60 microscope equipped with a DP70 camera and CellF imaging software (Soft Imaging System GmbH). Expression of MET was evaluated by P.L.P. and M.C., and was scored semi-quantitatively as percentage of positive cells and staining intensity. For the percentage of positive immunoreactive cells, the following scores were used: 0, 0-5%; 1, 6-29%; 2, 30-69%; and 3, 70-100%. The staining