Midkine signaling maintains the self-renewal and tumorigenic capacity of glioma initiating cells

Glioblastoma (GBM) is one of the most aggressive forms of cancer. It has been proposed that the presence within these tumors of a population of cells with stem-like features termed Glioma Initiating Cells (GICs) is responsible for the relapses that take place in the patients with this disease. Targeting this cell population is therefore an issue of great therapeutic interest in neuro-oncology. We had previously found that the neurotrophic factor MIDKINE (MDK) promotes resistance to glioma cell death. The main objective of this work is therefore investigating the role of MDK in the regulation of GICs. Methods: Assays of gene and protein expression, self-renewal capacity, autophagy and apoptosis in cultures of GICs derived from GBM samples subjected to different treatments. Analysis of the growth of GICs-derived xenografts generated in mice upon blockade of the MDK and its receptor the ALK receptor tyrosine kinase (ALK) upon exposure to different treatments. Results: Genetic or pharmacological inhibition of MDK or ALK decreases the self-renewal and tumorigenic capacity of GICs via the autophagic degradation of the transcription factor SOX9. Blockade of the MDK/ALK axis in combination with temozolomide depletes the population of GICs in vitro and has a potent anticancer activity in xenografts derived from GICs. Conclusions: The MDK/ALK axis regulates the self-renewal capacity of GICs by controlling the autophagic degradation of the transcription factor SOX9. Inhibition of the MDK/ALK axis may be a therapeutic strategy to target GICs in GBM patients.

and SOX9 protein levels (as determined by Western blot) of GH2 and 12O12-GICs cultures. A representative Western blot for each cell type is shown (n=2). (B) Flow cytometry analysis of NESTIN expression in GH2, 12O12 and HCO1-GICs. Representative dot plots of NESTIN immunostaining are shown (the insets correspond to control-unstained cells). Values in the upper right corner of each dot plot correspond to the percentage of NESTIN positive cells in each cell culture. (C) MDK protein levels (as determined by ELISA) in the medium of GICs cultures or of their corresponding serum-differentiated cells. Data are expressed as the mean fold-change ± SEM (n=3). *P <0.05; **P <0.01; ***P <0.001 from their corresponding serum-differentiated cells. (D) MDK mRNA levels (as determined by qPCR) of three different GICs-cultures (GH2, 12O12 and GH11). Data are expressed as the mean foldchange ± SEM (n=3) relative to each serum-differentiated culture. *P <0.05; **P <0.01 from their corresponding serum-differentiated cells.
(F) Effect of expressing shMDK 39 (by treating with doxycycline (+Dox.) cells stably transduced with this doxycycline inducible MDK-selective shRNAs) on MDK protein levels (as determined by ELISA) in the medium of three different GICs-cultures. Data correspond to MDK concentration in the medium and are expressed as the mean fold-change in MDK protein levels relative to the corresponding (-Dox.) shMDK 39 culture ± SEM (n=2). (G) Effect of incubation with an anti-MDK antibody (MDK Ab., 40 µg/ml, 72 h) on MDK protein levels (as determined by ELISA) in the medium of three different GICs-cultures. Data correspond to MDK protein levels in the medium and are expressed as the mean fold-change relative to vehicletreated cells. (n=3 GH2 and HCO1). ***P < 0.001 from GH2 or HCO1 vehicle-treated cells. 12O12 a representative experiment of 2 is shown. (H) Effect of MDK genetic inhibition (by treating with doxycycline (+Dox.) cells stably-transduced with a doxycycline-inducible MDK-selective shRNA, shMDK) or incubation with an anti-MDK antibody (MDK Ab., 40 µg/ml) during 2 consecutive passages on the growth of 12O12-GICs spheroid cultures. Data correspond to the total number of cells counted upon disaggregation of spheroid cultures in each passage and are expressed as the mean fold-change from the number of cells plated at P0 ± SEM (n = 3). ***P < 0.001 from vehicle or shMDK (-Dox.) cells.
(C) Effect of ALK genetic inhibition (by treating with doxycycline (+Dox.) cells stably-transduced with a doxycycline-inducible ALK-selective shRNA; shALK) or pharmacological ALK inhibition (by incubation with the ALK tyrosine-kinase inhibitors crizotinib and lorlatinib) during 2 consecutive passages on the growth of 12O12 (upper panel) and HCO1 (bottom panel)-GICs. Data correspond to the total number of cells counted upon disaggregation of spheroid cultures in each passage and are expressed as the mean fold-change from the number of cells plated at P0 ± SEM. n = 3. **P < 0.01 from vehicle or shALK (-Dox.) cells.
(F) Effect of nucleofection with a control plasmid (pWPXL GFP, control) or a plasmid encoding murine SOX9 (pWPXL SOX9 plasmid, SOX9) on murine SOX9 mRNA levels (as determined by qPCR) of GH2-shC or GH2-shMDK-GICs (n=3). (G) Effect of MDK genetic inhibition (by treating with doxycycline cells stably-transduced with a doxycycline-inducible MDK-selective shRNA, shMDK) and nucleofection with a control plasmid (CP) or a plasmid encoding murine SOX9 (SOX9) on the self-renewal ability (as determined by LDA) of GH2-GICs. n=2; a representative experiment is shown. # P < 0.05 from shMDK (+Dox) CP cells. Other symbols of significance are omitted for clarity. Full χ 2 statistical analysis is included in LDA statistics supplementary section. (H) Effect of the incubation with an anti-MDK antibody (MDK Ab.; 40 µg/ml) and TAE (0.75 µM) for 72h on cMYC and CYCLIND1 mRNA levels (as determined by qPCR) of GH2-GICs. Data are expressed as the mean fold change from vehicle-treated cells ± SEM (n=3). *P < 0.05; **P < 0.01; ***P < 0.001 from vehicle-treated cells.  (B) Effect of lorlatinib (25 mg/kg daily oral administration) on the expression (as determined by Western blot) of NESTIN, MUSASHI-1 (MSI1), SOX2 and SOX9 in samples derived from tumor xenografts generated by the subcutaneous injection of 2 x 10 6 12O12-GICs. Data correspond to the densitometric analysis of the corresponding protein/β-ACTIN ratio relative to one of the vehicle-treated samples and are expressed as the mean fold change ± SEM relative to vehicle treated tumors. n=6 (vehicle, VEH) and n=5 (lorlatinib)-treated tumors. * P < 0.05 and ** P < 0.01 from vehicle-treated tumors. (C) Effect of lorlatinib (25 mg/kg daily oral administration) on MSI1 (upper panel), NESTIN (middle panel) and SOX9 (lower panel) immunostaining of 12O12 GICs-derived subcutaneous tumor xenografts. Values in the bottom right corner at each photomicrograh correspond to MSI, NESTIN or SOX9-stained area relative to the number of nuclei in each field (estimated by the DAPI-stained area) and are expressed as mean ± SEM (10 photomicrographs/fields per tumor of 4 different tumors for each experimental condition were analyzed). * P < 0.05; ** P < 0.01 from vehicle-treated tumors. (E) Effect of crizotinib (CZT, 12.5 mg/kg daily oral administration)and TMZ (5 mg/kg twice a week IP administration) on the number of apoptotic cells (as determined by TUNEL) in samples derived from tumor xenografts generated by the subcutaneous injection of 2 x 10 6 12O12-GICs in the flank of nude mice. Representative images of tumors of each experimental condition are shown. Values in the bottom right corner at each photomicrograph correspond to the number of TUNEL-positive nuclei relative to the total number of nuclei per field determined by DAPI staining and are expressed as the mean ± SEM (6 photomicrographs/fields per tumor of 5 different tumors for each experimental condition were analyzed). ** P < 0.01 from vehicletreated tumors. ## P < 0.01 from crizotinib-treated tumors.
(F) Effect of treatment with lorlatinib (25 mg/kg daily oral administration), TMZ (5 mg/kg twice a week IP administration) on the growth of glioma xenografts generated by subcutaneous injection of 2 x 10 6 12O12 GICs in the flank of nude mice (mean ± SEM; n=5-6 mice for each condition). Representative pictures of the tumor xenografts in the last day of the treatment are shown for each experimental condition. Symbols of significance are omitted for clarity. Lorlatinib-treated tumors were significantly different from vehicle-treated tumors from day 8 to day 10 (P < 0.01), and from day 11 until the end of the treatment (P < 0.001); TMZ-treated tumors were significantly different from vehicle-treated tumors from day 8 to day 10 (P < 0.05), and from day 11 until the end of the treatment (P < 0.001); lorlatinib + TMZ-treated tumors were significantly different from vehicle-treated tumors at day 5 and day 6 (P < 0.05) and from day 7 until the end of the treatment (P < 0.001).