a Immunofluorescence assays showing that USP21 colocalizes with FOXD1 in the nuclei of MES 21 and 505 GSCs. Scale bar: 30 μm (b, c). Primary neurosphere formation showing that the silencing of USP21 considerably attenuated cell growth, and the effect of USP21-knockdown could be largely rescued by FOXD1. Right panel (c) showing the quantification of neurosphere formation efficiency (spheres/cells plated). Data are shown as mean ± S.D., n = 3, *P < 0.05, **P < 0.01, Student’s t-test. d In vitro limiting dilution sphere-forming frequency showing that the knockdown of USP21 remarkably reduced the tumorsphere formation frequency of MES21 and 505 GSCs, and the effect of USP21-knockdown could be rescued by FOXD1. Stem cell frequencies were estimated as the ratio 1/x with the upper and lower 95% confidence intervals, where 1 = stem cell and x = all cells. *P < 0.05. e Western blotting showing that the USP21/FOXD1 axis has great importance in the maintenance of core MES-specific markers, including ALDH1A3, CD44, C/EBPβ, TAZ, phosphorylated STAT3 (p-STAT3), and c-MET. f Representative bioluminescent images showing that USP21-knockdown MES 21 and 505 GSCs had lower tumorigenicity abilities than control GSCs while the tumorigenicity abilities could be rescued by FOXD1. n = 10. g Kaplan–Meier survival curves showing that mice bearing xenograft tumors formed by USP21-knockdown GSCs had a longer lifespan than control GSCs, and the survival time in mice intracranially injected with USP21-depleted GSCs significantly decreased after FOXD1 overexpression. n = 10, ***P < 0.001, Log-rank test. h Representative H&E and IHC images showing that the xenografts carrying USP21 shRNA GSCs displayed restricted tumor growth, whereas this effect could be reversed by overexpression of FOXD1.