A PDGFRα-driven mouse model of glioblastoma reveals a stathmin1-mediated mechanism of sensitivity to vinblastine

Abstract

Glioblastoma multiforme (GBM) is an aggressive primary brain cancer that includes focal amplification of PDGFRα and for which there are no effective therapies. Herein, we report the development of a genetically engineered mouse model of GBM based on autocrine, chronic stimulation of overexpressed PDGFRα, and the analysis of GBM signaling pathways using proteomics. We discover the tubulin-binding protein Stathmin1 (STMN1) as a PDGFRα phospho-regulated target, and that this mis-regulation confers sensitivity to vinblastine (VB) cytotoxicity. Treatment of PDGFRα-positive mouse and a patient-derived xenograft (PDX) GBMs with VB in mice prolongs survival and is dependent on STMN1. Our work reveals a previously unconsidered link between PDGFRα activity and STMN1, and highlight an STMN1-dependent cytotoxic effect of VB in GBM.

Fig. 1

Spatiotemporal activation of PDGFRα in the CNS produces proneural GBM in adult mice. a Schematic of the conditional human PDGFRα cDNA transgene driven by the CAG promoter whose activity is prevented by a floxed stop cassette (LSL) until removed by Cre recombinase. The transgene was knocked into the 3′-UTR of the Col1α1 gene. b Representative photomicrograph of an H&E-stained FFPE section of a P3 brain tumor (scale bar, 1 mm). c Anti-hPDGFRα IHC (scale bar, 250 μm). d Tumor-free survival (Kaplan–Meier) analysis of three separate cohorts of mice of indicated genotypes fed the DOX diets. e qPCR analysis of PDGF-A mRNA from tumors of mice fed a 25 and 625 mg kg⁻¹ DOX diet. *p < 0.03, (Student’s t-test). f Quantitative western blottings for phospho-PDGFRαTyr762 from lysates of P3 end-stage brain tumors (error bars denote SD; n = 3; NS, not significant by the Student’s t-test). g Percent of cells from end-stage tumors scored for proliferation using Ki67 marker by IHC on FFPE sections from P3 tumors from mice fed with 25 mg kg⁻¹ and 625 mg kg⁻¹ DOX diets. Average values of positive cells per field of view (FOV) (× 40). (error bars denote SD; n = 12 (4 FOVs per mouse); NS, not significant by the Student’s t-test). h P3 tumors have features of glioblastoma. Representative photomicrographs of FFPE tumor sections stained with H&E, IHC for GFAP and Tuj1 (NeuN) (scale bars, top and bottom row 50 μm, middle left 200 μm, middle right 500 μm). T, tumor; N, normal brain. i Photomicrographs of serial T2 flair MR imaging of a representative P3 GBM tumor. j Gene expression profiles of P3 GBM tumors from both low (25 mg kg⁻¹ DOX) and high (625 mg kg⁻¹ DOX) PDGF-A were used to perform GSEA against gene-set lists representing all four GBM subtypes. For scatter plots, the center line represents the mean and upper and lower lines SD

Fig. 2

Differential chronic signaling from weakly and strongly activated hPDGFRα. a–c Log₂ fold change of representative phospho-sites from quantitative phospho-proteomics dataset. d Global quantitative MS/MS phospho-proteomic analysis of low and high chronic hPDGFRα activation. Plotted are the Log₂ fold change (FC) of low hPDGFRα activation over unstimulated hPDGFRα for 5849 normalized phosphopeptides (x axis) against the differential between Log₂FC high vs. control and Log₂FC low vs. control hPDGFRα activation (y axis). Top right inset, 15 top scoring GO categories of outlier (Log₂FC < − 0.5, > + 0.5) phospho-events. Green data points are hPDGFRα autophosphorylation sites. Results are from single measurements of pooled biological triplicates

Fig. 3

hPDGFRα-STMN1 signaling axis confers sensitivity to vinblastine. a Phospho-STMN1 Ser16, Ser25, and Ser38 levels decrease in a dose-dependent manner. Representative images and quantitation of phospho-STMN1 Ser16, Ser25, and Ser38 immunoblots (error bars denote SD; n = 3, *p < 0.0001, by Student’s t-test when compared with 0 DOX controls). b PDGFRα-positive GBM cell sensitivity to microtubule drugs. Cells were treated with the indicated drugs and viability assessed via trypan blue assays. MDA-231 breast carcinoma cell line serves as a positive control for drug sensitivities. Cells were treated with docetaxel and paclitaxel (500 nM, 96 h, MDA-231 cells were treated with 100 nM), vinblastine (100 nM, 24 h), and vincristine (200 pM, 24 h). Cells were grown in the absence or presence of activated hPDGFRα (error bars denote SD; n = 3, *p < 0.0001 by Student’s t-test when compared with 0 DOX control). c Synthetic decreases in cell viability between hPDGFRα activity and vinblastine treatment. Dose-response curves of vinblastine and vincristine in the presence and absence of PDGFRα activity. d Inhibition of hPDGFRα activity using ponatinib (1 μM) blocks the synthetic sensitivity between hPDGFRα and vinblastine (error bars denote SD; n = 4, *p < 0.0001 by Student’s t-test when compared with 0 DOX control). e Synthetic induction of apoptosis in vinblastine and hPDGFRα active cells. Quantitative western blotting of cleaved and total caspase 3 of GBM cells treated for 24 h with vinblastine (100 nM), in the absence or presence of PDGF-A with and without the PDGFR inhibitor ponatinib (1 μM) (error bars denote SD; n = 3). f The hPDGFRα-vinblastine synthetic lethal effect is mediated through STMN1. STMN1 CRISPR/Cas9 knockout abrogates the heightened lethal outcome of hPDGFRα activity and vinblastine. Representative clonal cultures from four independent sgRNAs against STMN1 are shown (error bars denote SD; n = 4, *p < 0.0001 by Student’s t-test when compared withto 0 DOX control)

Fig. 4

hPDGFRα-STMN1 signaling exacerbates vinblastine toxicity during mitosis. a Representative photomicrographs of phase-contrast time-lapse microscopy of cells untreated or treated with vinblastine. In the absence of vinblastine cells entering mitosis proceed to divide normally within an average of ~120 min, whereas in the presence of vinblastine cells entering mitosis undergo arrest followed by either apoptosis or mitotic slippage. b Quantification of time spent in mitosis from a. c hPDGFRα activity promotes cell death in vinblastine during mitosis in an STMN1-dependent manner. Percentage of cells entering mitosis that undergo slippage or apoptosis for parental and STMN1 CRISPR KO cells (error bars SD; n = 24, *p < 0.0001 by Student’s t-test, when compared with no DOX control). For scatter plots, the center line represents the mean and upper and lower lines SD

Fig. 5

Prolonged survival with vinblastine in GBM is PDGFRα activity and STMN1 dependent. a Tumor-free survival (Kaplan–Meier) analysis of mice intracranially injected with hPDGFRα GBM cells wild-type or null for STMN1 fed a DOX diet (625 mg kg⁻¹). Upon tumor detection (as determined by an 8–10% weight loss), mice were enrolled randomly on control or vinblastine (0.5 mg kg⁻¹ I.P. every 3 days) until moribund. Cohorts of mice were also withdrawn from DOX at tumor detection. *p = 0.0067 (log-rank Mantel–Cox test). b PDGFRα-activity-dependent decrease in levels of phospho-STMN1 Ser16 in the human GBM6 PDX line. Shown are representative western blotting of the indicated antibodies and graphical representation of quantitative measurements from biological triplicates. *p < 0.0001 by Student t-test. c Human GBM PDX sensitized to vinblastine treatment in vivo by chronic activation of PDGFRα. Tumor volumes of the GBM6 PDX line genetically modified to express human PDGF-A ligand in a DOX-inducible manner grown in flanks of immunocompromised Ncr^Nu/Nu mice. Tumor response is expressed as the percentage change from the baseline tumor volumes at the time of treatment initiation (~100 mm³, Day 0). Baseline tumor volumes are listed in Methods. n = 5, *p < 0.0001 by Student t-test. d Tumor-free survival (Kaplan–Meier) analysis of mice intracranially injected with human GBM6 PDX cells from c. Upon tumor detection (as determined by an 8–10% weight loss), mice were enrolled randomly on control or vinblastine (0.5 mg kg⁻¹ IP every 3 days) until moribund. *p = 0.0027, **p = 0.018 (log-rank Mantel–Cox test)

Fig. 6

PDGFRα and STMN1 cooperate to exacerbate the cytotoxic effects of vinblastine. Cartoon depicting the mechanism by which PDGFRα, through STMN1 dephosphorylation, leads to enhanced microtubule depolymerization. Vinblastine-poisoned microtubules undergo reduced polymerization at the + end effectively resulting in depolymerization, which during mitosis, triggers SAC and results in either apoptosis or mitotic slippage