Glioblastoma multiforme (GBM) are the most common primary malignant brain tumor in adults, with a median survival of about one year. This poor prognosis is attributed primarily to therapeutic resistance and tumor recurrence after surgical removal, with the root cause suggested to be found in glioblastoma stem cells (GSCs). Using glial fibrillary acidic protein (GFAP) as a reporter of astrocytic differentiation, we isolated multiple clones from three independent GSC lines which express GFAP in a remarkably stable fashion. We next show that elevated expression of GFAP is associated with reduced clonogenicity in vitro and tumorigenicity in vivo. Utilizing this in vitro cell-based differentiation reporter system we screened chemical libraries and identified the non-depolarizing neuromuscular blocker (NNMB), Atracurium Besylate, as a small molecule which effectively induces astroglial but not neuronal differentiation of GSCs. Functionally, Atracurium Besylate treatment significantly inhibited the clonogenic capacity of several independent patient-derived GSC neurosphere lines, a phenomenon which was largely irreversible. A second NNMB, Vecuronium, also induced GSC astrocytic differentiation while Dimethylphenylpiperazinium (DMPP), a nicotinic acetylcholine receptor (nAChR) agonist, significantly blocked Atracurium Besylate pro-differentiation activity. To investigate the clinical importance of nAChRs in gliomas, we examined clinical outcomes and found that glioma patients with tumors overexpressing CHRNA1 or CHRNA9 (encoding for the AChR-α1 or AChR-α9) exhibit significant shorter overall survival. Finally, we found that ex-vivo pre-treatment of GSCs, expressing CHRNA1 and CHRNA9, with Atracurium Besylate significantly increased the survival of mice xenotransplanted with these cells, therefore suggesting that tumor initiating subpopulations have been reduced.

The stem cell-associated transcription co-factor ZNF521 has been implicated in the control of hematopoietic, osteo-adipogenic and neural progenitor cells. ZNF521 is highly expressed in cerebellum and in particular in the neonatal external granule layer that contains candidate medulloblastoma cells-of-origin, and in the majority of human medulloblastomas. Here we have explored its involvement in the control of human and murine medulloblastoma cells. The effect of ZNF521 on growth and tumorigenic potential of human medulloblastoma cell lines as well as primary Ptc1-/+ mouse medulloblastoma cells was investigated in a variety of in vitro and in vivo assays, by modulating its expression using lentiviral vectors carrying the ZNF521 cDNA, or shRNAs that silence its expression. Enforced overexpression of ZNF521 in DAOY medulloblastoma cells significantly increased their proliferation, growth as spheroids and ability to generate clones in single-cell cultures and semisolid media, and enhanced their migratory ability in wound-healing assays. Importantly, ZNF521-expressing cells displayed a greatly enhanced tumorigenic potential in nude mice. All these activities required the ZNF521 N-terminal motif that recruits the nucleosome remodeling and histone deacetylase complex, which might therefore represent an appealing therapeutic target. Conversely, silencing of ZNF521 in human UW228 medulloblastoma cells that display high baseline expression decreased their proliferation, clonogenicity, sphere formation and wound-healing ability. Similarly, Zfp521 silencing in mouse Ptc1-/+ medulloblastoma cells drastically reduced their growth and tumorigenic potential. Our data strongly support the notion that ZNF521, through the recruitment of the NuRD complex, contributes to the clonogenic growth, migration and tumorigenicity of medulloblastoma cells.

During normal development, heterogeneous expression of Notch ligands can result in pathway suppression in the signal-sending cell, a process known as lateral inhibition. It is unclear if an analogous phenomenon occurs in malignant cells. We observed significant induction of Notch ligands in glioblastoma neurospheres and pancreatic carcinoma cells cultured in low oxygen, suggesting that this phenomenon could occur around hypoxic regions. To model lateral inhibition in these tumors, the ligand Jagged1 was overexpressed in glioblastoma and pancreatic carcinoma cells, resulting in overall induction of pathway targets. However, when ligand high and ligand low cells from a single line were co-cultured and then separated, we noted suppression of Notch pathway targets in the former and induction in the latter, suggesting that neoplastic lateral inhibition can occur. We also found that repression of Notch pathway targets in signal-sending cells may occur through the activity of a Notch ligand intracellular domain, which translocates into the nucleus. Understanding how this neoplastic lateral inhibition process functions in cancer cells may be important in targeting ligand driven Notch signaling in solid tumors.