ASCL1 phosphorylation and ID2 upregulation are roadblocks to glioblastoma stem cell differentiation

Abstract

The growth of glioblastoma (GBM), one of the deadliest adult cancers, is fuelled by a subpopulation of stem/progenitor cells, which are thought to be the source of resistance and relapse after treatment. Re-engagement of a latent capacity of these cells to re-enter a trajectory resulting in cell differentiation is a potential new therapeutic approach for this devastating disease. ASCL1, a proneural transcription factor, plays a key role in normal brain development and is also expressed in a subset of GBM cells, but fails to engage a full differentiation programme in this context. Here, we investigated the barriers to ASCL1-driven differentiation in GBM stem cells. We see that ASCL1 is highly phosphorylated in GBM stem cells where its expression is compatible with cell proliferation. However, overexpression of a form of ASCL1 that cannot be phosphorylated on Serine-Proline sites drives GBM cells down a neuronal lineage and out of cell cycle more efficiently than its wild-type counterpart, an effect further enhanced by deletion of the inhibitor of differentiation ID2, indicating mechanisms to reverse the block to GBM cell differentiation.

Figure 1

ASCL1 phosphorylation regulates proliferation of human GSCs. (A) ASCL1 protein expression in GBM cell lines; positive control is after ASCL1 overexpression. (B) Western blot showing endogenous ASCL1 phosphorylation in GBM cells, with and without phosphatase (λ-PP) treatment. (C) Overexpression of WT and 5S-A ASCL1 in G144 cells, with and without λ-phosphatase treatment. White arrowheads in B and C indicate phosphorylated ASCL1, while black arrowheads indicate unphosphorylated ASCL1. (D) Representative images of G144 cells after growth factor withdrawal and dox-induced WT and 5S-A ASCL1 expression. Scale bars: 300 μm. (E) Quantification of cell confluence. Each data point is mean ± SEM. n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; *p < 0.05; ***p < 0.001. All data points from day 3 onwards statistically differ between control and WT ASCL1 and between control and 5S-A ASCL1 (p < 0.05; not shown). (F) Quantification of cell number at 2, 7 and 14 days of ASCL1 induction. Data: mean ± SEM n = 3 independent experiments, t-test; **p < 0.01; ****p < 0.0001 (G) Immunofluorescence showing EdU incorporation. Scale bars: 100 μm. (H) Quantification of the % of EdU-positive cells. Data: mean ± SEM n = 3 independent experiments, t-test; *p < 0.05. (I) Expression of negative cell cycle regulators in G144 cells after 7 days of ASCL1 induction. Data: mean ± SEM, normalized to TBP. n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; **p < 0.01; ***p < 0.001. Full length western blots are provided in Supplementary Fig. S1A, B and C.

Figure 2

Dephosphorylated ASCL1 promotes neuronal differentiation. (A–C) Immunostaining for the neuronal marker TUBB3 (red) and quantification of the % of TUBB3⁺ cells over the total DAPI⁺ cells cultured for 7 (A,B) or 16 days (C,D) without growth factors and in the presence of dox to induce ASCL1 expression. Scale bars: 100 μm. Data: mean ± SEM n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; **p < 0.01; ***p < 0.001. (E,F) Relative mRNA expression of different neuronal (E) and oligodendrocyte (F) markers at 7 days. Data: mean ± SEM, normalized to TBP. n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 3

Dephosphorylated ASCL1 promotes stable mitotic exit and differentiation. (A) Representative images of G144 cells after removing ASCL1 induction and re-addition of growth factors. Scale bars: 300 μm. (B) Quantification of cell confluence. Data: mean ± SEM n = 2 independent experiments for Control + GFs and n = 3 independent experiments for the other conditions; one-way ANOVA followed by the Bonferroni post-hoc test; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. All data points from day 3 onwards are statistically different between control and WT ASCL1 and between control and 5S-A ASCL1 (p < 0.05; not shown). (C) Immunostaining showing persistent expression of the neuronal marker TUBB3 (red) at 25 days of culture. Scale bars: 50 μm.

Figure 4

ID2 restrains differentiation induced by dephosphorylated ASCL1. (A) Relative mRNA expression of different members of the ID family at 7 days after WT and 5S-A ASCL1 expression. Data: mean ± SEM, normalized to TBP. n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; ***p < 0.001; **** p < 0.0001. (B) Quantification of cell confluence. Each data point is mean ± SEM n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; *p ≤ 0.05 between ID2^−/− and 5SA-ASCL1; #p ≤ 0.05 between ID2^−/− and ID2^+/+. (C,D) Representative images of unedited control cells, ID2 wild type cells and ID2 homozygous knock-out cells, without (C) or with (D) dox-induced ASCL1 expression at increasing time points, as labelled. Scale bars: 100 μm. (E) Immunostaining for neuronal marker TUBB3 (green) at 14 days of culture in differentiation conditions. Scale bars: 100 μm. (F) Quantification of the % of TUBB3⁺ cells over the total DAPI⁺ cells at day 14. Data: mean ± SEM n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; *p < 0.05. (H) Quantification of the total neurite length per cell. Data: mean ± SEM n = 3 independent experiments; one-way ANOVA followed by the Bonferroni post-hoc test; ****p < 0.0001.

Figure 5

Model showing that ASCL1 dephosphorylation and reduction of ID2 levels can enhance neuronal differentiation of glioblastoma stem cells.