A. OLIG2 and NeuroD1; B. Sequence alignment of transcription factors that we called “E2A set” (outlined by the dashed rectangle) including E12, E47, and HTF4 with negative residues N1-N3 in the interface region and the TFs complementary to them; P1-P4 are positive residues in the interface area.

A. General structure of OLIG2 (cyan)-E47 (magenta) heterodimer, inset depicts the topological scheme of the interface, 1-upper interface, 2-middle interface (red arrow indicates the region of pharmacophore design), and 3-interface with DNA. B. Close-up of the middle interface interactions. One can see locations of key residues involved in the specific interaction between OLIG2 and E47 and distances between them. The interaction zone includes the E47 negative residues Asp561, Glu564, and Glu568 and the OLIG2 positive residues Lys148 and Arg156. C. Scheme of the OLIG2-E47 interface created by TF features. Details in the area depicted by the rectangle are presented in the zoomed-in view. Complementary combinations to E47: set 1 (S1, OLIG2 or NeuroD1)-P2, H1, H2, P4; set 2 (S2)-P1, P3, H1, H2, P4; set 3 (S3)-P2, P3, H1, H2, P4. This organization leads to the definition of the main features of four pharmacophores including the parental and 3 daughters: Pharmacophore 0 (five features, parental): P1 or P2; H1; P3; H2; P4; Pharmacophore 1 (daughter, four features): P1 orP2; H1; H2; P4 (OLIG2 and similar); Pharmacophore 2 (daughter, four features): P1 or P2; H1; P3; H2; Pharmacophore 3 (daughter, four features): H1; P3; H2; P4. D. Pharmacophore hypothesis mimicking OLIG2 complementary interaction. Purple spheres P2 and P4 are Cationic and Donor (Cat&Don) centers based on Arg156-Glu568 and Lys148-Asp561 positive-negative interactions; purple sphere P3 (Cat&Don center) added to bind E47's residue Glu564. Green spheres are Hydrophobic (Hyd) centers based on Leu152-Phe566 and Ile149-Ile562 hydrophobic interactions.

A. Five-feature parental (i) and four-feature daughter (ii, iii, and iv) pharmacophores. Ribbon diagram and residues presented by lines belong to the superimposed OLIG2 protein. B. Venn diagram for four sets of compounds identified from a four pharmacophore-hypotheses based search in conformational database derived from the Open NCI compounds in silico library. For example, using the gr1 daughter pharmacophore the program selected 545 compounds from which 147 were also selected using gr2, gr3, daughter and the parental (gen) pharmacophore. C. Two compound examples, SKOG104 and SKOG102, are theoretically able to engage all three subpharmacophores within the dimerization region. The individual figure panels illustrate the predicted fit of SKOG102/4 to the three daughter pharmacophores.

Figure depicts structural classes and representative scaffolds A. while B. shows the associated IC₅₀ curves for Ink4a/arf EGFR-VIII cells treated with the two most potent compounds from each structural class, or cluster. In addition to the 75 compounds belonging to the 5 classes, another 20 disparate structures were also tested, and these showed little activity. Error bars represent mean ± standard deviation.

A. One of the most potent OLIG2 inhibitors identified by our modeling methodology clearly inhibits human GBM4 and GBM8 cells grown as neurospheres B. in a dose-dependent fashion. DMSO was the vehicle control. *, ** indicate p-values between control and 5 μM SKOG102 treated cells, * for GBM4 cells p = 0.0003 and for GBM8 cells p = 0.001. C. Western blot data for OLIG2 expression in patient-derived GBM lines (GBM4 and GBM8), in an immortalized, serum-grown GBM cell line (U87), and in normal human astrocytes (NHA) freshly acquired from patient material. The blot image was prepared by cropping out middle lanes, the gel and blotting conditions were identical for the lanes shown in the figure. D. The GBM4/8 GBM lines express OLIG2 and exhibit an IC₅₀ between 1 and 2 μM. U87 cells, which express much less OLIG2 than GBM4/8 cells, were also suppressed, but a roughly 7-fold higher dose was needed. NHA, which express no detectable OLIG2, had an IC₅₀ of about 20 fold greater than GBM4 cells. All samples run in duplicate. Error bars represent mean ± standard deviation.

The expression of p21 and OMG were determined by RT-PCR. A. Escalating OLIG2 inhibitor doses led to increased levels of p21. B. Increasing concentrations of OLIG2 inhibitor suppressed the expression of OMG. C. GBM4 cells were treated with vehicle or TMZ (1 or 100 μM) for 12 h followed by RT-PCR analyses to determine p21 and OMG expression. * p <0.01 between vehicle control and SKOG102 treated groups. D. SKOG102 derepresses OLIG2 mediated repression of luciferase activity in p21 reporter assay. p = 0.0089. All these results suggest that the OLIG2 inhibitor compound specifically targets OLIG2.

GBM8 patient-derived cells were treated with vehicle or 5 μM Olig2 inhibitor for 20 hr. A. Nuclear fractions were isolated and used for an electrophoretic mobility shift assay (EMSA). The full length blot is shown in . B. Nuclear fractions were probed for phosphorylation of OLIG2 and total OLIG2. Cytoplasmic extracts were also examined for re-localization of OLIG2 from the nucleus to the cytoplasm. OLIG2 does not re-localize to the cytoplasm after inhibitor treatment. Each panel was run under the same conditions and the full length blots are shown in . C. Flank tumor study. GBM4 cells were implanted subcutaneously into NSG mice, dosing was started when tumors were palpable. Dosing schedule indicated by horizontal bars on graph was first 2 weeks (from day 33 to 46) 10 mg/kg, third week (from day 47 to 54) 15 mg/kg, fourth week (from 55 to 66) first 3 days 20 mg/kg, 2 days break, then alternate days from day 60 to 66. GBM4 tumor growth was inhibited after treatment with SKOG102 when compared vehicle treated group. Error bar represent mean standard deviation, * indicates p = 0.02 between vehicle control and SKOG102 treated group. D. Brain concentration of SKOG102. Mice were injected with 5 mg/kg intraperitoneally and the graph indicates the brain concentration at 1 and 4 hours. E. Intracranial tumor study. GBM4 cells were pretreated with SKOG102 or vehicle control for 14 h, only viable cells were injected intracranially into NSG mice (n = 8). Tumor volumes were measured using MRI after 4 weeks. Bar graph indicates tumor volumes, error bars represent mean ± standard deviation, * indicates p < 0.05 between vehicle control and SKOG102. F. Renderings of mouse brain tumors imaged with contrast enhanced MRI, where cells were pretreated with DMSO vehicle (upper panels) and SKOG102 (lower panels), and viable cells were injected intracranially into NSG mice. Area of the enhanced relative intensity was color coded to produce a 3D model of the extent and distribution of the tumor cells in the brain tissue.