Metformin and simvastatin exert additive antitumour effects in glioblastoma via senescence-state: clinical and translational evidence

Abstract

Background: Glioblastoma is one of the most devastating and incurable cancers due to its aggressive behaviour and lack of available therapies, being its overall-survival from diagnosis ∼14-months. Thus, identification of new therapeutic tools is urgently needed. Interestingly, metabolism-related drugs (e.g., metformin/statins) are emerging as efficient antitumour agents for several cancers. Herein, we evaluated the in vitro/in vivo effects of metformin and/or statins on key clinical/functional/molecular/signalling parameters in glioblastoma patients/cells.

Methods: An exploratory-observational-randomized retrospective glioblastoma patient cohort (n = 85), human glioblastoma/non-tumour brain human cells (cell lines/patient-derived cell cultures), mouse astrocytes progenitor cell cultures, and a preclinical xenograft glioblastoma mouse model were used to measure key functional parameters, signalling-pathways and/or antitumour progression in response to metformin and/or simvastatin.

Findings: Metformin and simvastatin exerted strong antitumour actions in glioblastoma cell cultures (i.e., proliferation/migration/tumoursphere/colony-formation/VEGF-secretion inhibition and apoptosis/senescence induction). Notably, their combination additively altered these functional parameters vs. individual treatments. These actions were mediated by the modulation of key oncogenic signalling-pathways (i.e., AKT/JAK-STAT/NF-κB/TGFβ-pathways). Interestingly, an enrichment analysis uncovered a TGFβ-pathway activation, together with AKT inactivation, in response to metformin + simvastatin combination, which might be linked to an induction of the senescence-state, the associated secretory-phenotype, and to the dysregulation of spliceosome components. Remarkably, the antitumour actions of metformin + simvastatin combination were also observed in vivo [i.e., association with longer overall-survival in human, and reduction in tumour-progression in a mouse model (reduced tumour-size/weight/mitosis-number, and increased apoptosis)].

Interpretation: Altogether, metformin and simvastatin reduce aggressiveness features in glioblastomas, being this effect significantly more effective (in vitro/in vivo) when both drugs are combined, offering a clinically relevant opportunity that should be tested for their use in humans.

Funding: Spanish Ministry of Science, Innovation and Universities; Junta de Andalucía; CIBERobn (CIBER is an initiative of Instituto de Salud Carlos III, Spanish Ministry of Health, Social Services and Equality).

Keywords: Glioblastoma; Metformin; Senescence; Simvastatin; Splicing; Telomere.

Fig. 1

Treatments with metformin and/or simvastatin significantly decrease proliferation rate in different glioblastoma (GBM) cell models and high grade (III) astrocytomas, but not the viability in non-tumour brain cells in vitro. (a) Metformin dose–response carried out in U-87 MG and U-118 MG cells (n = 4). (b) IC₅₀ of metformin in vitro in U-87 MG and U-118 MG cells. (c) Simvastatin dose–response carried out in U-87 MG and U-118 MG cells (n = 4). (d) IC₅₀ of simvastatin in vitro in U-87 MG and U-118 MG cells. Proliferation/viability rates of GBM cell lines [U-87 MG (e) and U-118 MG (f); n = 5], primary patient-derived cell cultures from AIII [grade III (g), n = 4] and grade IV-GBM [(h); n = 4], primary non-tumour brain cell cultures [(i); n = 4] and mouse primary-astrocytes progenitor derived cells [(j); n = 4] in response to metformin, simvastatin, and their combination compared to vehicle-treated controls. Four technical replicates (tr) were assessed in each condition. Data represent medians (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Fig. 2

Treatments with metformin, simvastatin, and specially their combination, decreased key functional parameters in glioblastoma (GBM) cells in vitro. Effect of the treatment with metformin, simvastatin, and their combination compared to vehicle-treated control cells in: (a) Migration rate of GBM U-118 MG cells ([n = 5; 3 technical replicates (tr)] representative images of the migration capacity are included); (b) Apoptosis induction in GBM U-87 MG and U-118 MG cells (n = 3); (c) Sphere number and area in GBM U-87 MG and U-118 MG cells (n = 3; tr = 4; representative images of formation of tumourspheres are shown); (d) VEGF secretion in U-87 MG, U-118 MG, and primary patient-derived GBM cells (n = 4; tr = 2); and (e) Percentage of β-galactosidase positive cells (cells positive in blue) in GBM U-87 MG and U-118 MG cells (n = 3; tr = 2) in response to metformin, simvastatin, and their combination compared to control condition and their corresponding representative images. (f) Percentage of cells in each cell cycle phase (right panel) determined by the Watson Pragmatic model (left panel) in U-87 MG and U-118 MG cells (n = 3; tr = 1) after 72 h of treatment with metformin, simvastatin, and their combination compared to control condition. (g) Colony formation in GBM U-87 MG and U-118 MG cells (n = 3; tr = 3). (h) Summary of the effects of metformin and/or simvastatin treatments (vs. control) on different functional parameters associated with the progression, development, and aggressiveness of GBM cells. Data represent medians (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Fig. 3

Treatment with metformin, simvastatin, and their combination altered the phosphorylation levels of different proteins associated with different key oncogenic pathways in GBM cells. (a)Upper panel: Membranes of the phosphoprotein array showing the spots (phosphorylation levels) quantified of 5 oncogenic signalling pathways (MAPK, AKT, JAK/STAT, NF-kB, and TGFβ pathways; 55 phosphorylated proteins) in response to the treatments with metformin, simvastatin, and their combination (24 h, n = 3; tr = 1, pooled) in GBM U-87 MG cells. Lower panel: Heatmap showing the fold-change mean corresponding to each pathway in response to metformin, simvastatin, and their combination. (b-d) Log2 (Fold Change) of individual phosphorylation protein levels in response to metformin (b), simvastatin (c), and their combination (d) compared with vehicle-treated controls cells [threshold: log2(FC) = 0.2]. The significantly altered phosphorylated proteins were analysed using the STRING database. (e) Functional association network of the significantly dysregulated proteins (upregulated in red and downregulated in blue) in response to the combination of metformin and simvastatin using the Network Analyst software.

Fig. 4

The combination of metformin and simvastatin altered the expression levels of key elements involved in the Senescence-Associated Secretory Phenotype (SASP), and in the splicing process in GBM cells. (a) Hierarchical heatmap generated using the expression levels of 32 SASP genes in all GBM experimental cell models treated with metformin, simvastatin, and the combination of both drugs (n = 3; tr = 2; mean value U-87 MG + U-118 MG + primary-derived cell cultures). Hierarchical heatmap showing the expression levels of 20 top altered genes (b) and Partial Least Squares Discriminant Analysis (PLS-DA) (c) of the expression of SASP genes in response to metformin, simvastatin, and their combination in GBM cells [cell lines (U-87 MG and U-118 MG) and primary patient-derived GBM cells (PPdC); data represent the mean value of 3 experiments (4 treatment conditions/experiment) in each GBM cell model]. d) Variable Importance in Projection (VIP) score of the expression levels of SASP key genes in GBM cells. e) Heatmap generated using the expression levels of 4 spliceosome components and (f) individual expression levels of these components in response to metformin, simvastatin, and their combination in GBM cells (n = 3; tr = 2; data represent mean value of U-87 MG + U-118 MG). Data represent medians (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Fig. 5

Conditioned media of metformin, simvastatin, and metformin + simvastatin treated cells induced a senescence state and an elongation in the telomere of GBM cells. (a) Diagram showing the workflow carried out to determine whether the treatments with metformin and/or simvastatin are able to induce a senescence state or alteration in the elongation of telomeres in GBM cells. (b) Proliferation rates in U-87 MG and (c) U-118 MG cell lines (U-87 MG/U-118 MG) in response to the conditioned media of cells treated metformin, simvastatin, and their combination compared to control condition (n = 3; tr = 4). (d) Telomere length determination in U-87 MG and (e) U-118 MG cells in response to the conditioned media of cells treated metformin, simvastatin, and their combination compared to control condition (n = 3; tr = 2). (f) Percentage of β-galactosidase positive cells (cells positive in blue) in response to the conditioned media of cells treated metformin, simvastatin, and their combination compared to control condition and their corresponding representative images (n = 3; tr = 2). Data represent medians (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Fig. 6

In vivo pharmacological administration of metformin, simvastatin,and their combination disruptsGBM progression and mimicthe molecular drug effects observed in vitro. (a) Kaplan–Meier survival curves discerning between patients with aggressive gliomas treated with metformin, statins, or their combination, and patients not treated with these drugs obtained from our pilot, observational, and retrospective study including 85 patients. Doses were administered following the prescribed daily dose: metformin (between 500 and 1000 mg/daily), and statins (between 20 and 40 mg/daily for simvastatin and 10–20 mg/daily for atorvastatin). (b) Diagram showing the generation of a preclinical-xenograft GBM model by inoculation of U-87 MG cells (n = 15; left panel), average tumour volume (right panel) and weight (c) of intra-tumour injection of metformin, simvastatin, and their combination vs. control-treated tumours. (d) Image of a representative tumour from each treatment condition at sacrifice day is shown. (e) Histopathological evaluation of mitosis number (x5 high-power field; HPF) and (f) apoptosis of all the U-87 MG xenograft tumours. (g) Representative images of H&E staining comparing tumours treated with metformin, simvastatin, and their combination vs. control-treated tumour samples. All these evaluations were determined by experienced pathologists. (h) Hierarchical heatmap generated using the expression levels of relevant SASP genes in all the U-87 MG xenograft tumours treated with control (PBS), metformin, simvastatin, and the combination of both drugs (i) and the Partial Least Squares Discriminant Analysis (PLS-DA) of the SASP genes in each condition. (j) Variable Importance in Projection (VIP) score of the expression levels of SASP key genes in the same samples. (k) Heatmap generated using the expression levels of 4 spliceosome components and (l) individual expression levels of these components in response to metformin, simvastatin, and their combination intra-tumours administration in U-87 MG xenograft tumours. Data represent medians (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Supplementary Fig. S1

(a) Left panel: Heatmaps showing the western-blot densitometric level (log2) of phosphorylated proteins in GBM cells (U-87 MG and U-118 MG) after 24 h treatment with metformin, simvastatin, and their combination compared to control condition. Right panel: Images of western-blot results showed in the heatmaps. (b) Results from principal component analysis (PCA) of all the patient samples showing homogeneity between samples using all directly observed variants available in our cohort. From left to right: 3D PCA plot, 3D loading plot showing the different variables included. PCs Biplot showing both PC scores of samples (dots) and loadings of variables (vectors). (c) Astrocytes markers expression (Sox2 and S100b as Neural Stem Cell Markers; Aqp4 as Astrocyte Marker) in mouse primary-astrocytes progenitor derived cells (n = 9) normalized by Rps11 (used as housekeeping gene). (d) Single cell data from Allen Brain map were used to corroborate the existence of astrocytes progenitor population in non-mature mice (<8 weeks-old) expressing the astrocytes markers previously measured in our in vitro model (Dataset: Whole Cortes & Hippocampus −10x genomics 2020, https://portal.brain-map.org/atlases-and-data/rnaseq). Data represent median (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Supplementary Fig. S2

Percentage of cells in each flow gate corresponding to early apoptosis, late apoptosis and total apoptosis in (a) U-87 MG and (b) U-118 MG cells together with their representative PI-Annexin V plots (c-d, respectively) after 72 h treatment with metformin, simvastatin, and their combination compared to control condition. (e) Individual expression levels of key apoptosis-related genes in response to metformin, simvastatin, and their combination in GBM cell lines (U-87 MG, U-118 MG; n = 4; tr = 1) and (f) in human primary-GBM derived cell cultures. Data represent median (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.

Supplementary Fig. S3

Scheme, membrane by membrane, of the phosphoprotein array (Fig. 3a; Upper panel) that shows the spots quantified of 5 oncogenic signalling pathways (MAPK, AKT, JAK/STAT, NF-kB, and TGFß pathways; 55 phosphorylated proteins) in response to the treatments with metformin, simvastatin and their combination (24 h, n = 3, pooled) in GBM U-87 MG cells.

Supplementary Fig. S4

Alteration in the phosphorylation levels of key elements of 5 oncogenic signalling pathways (MAPK, AKT, JAK/STAT, NF-kB, and TGFß pathways; 55 phosphorylated proteins) in response to treatments with metformin, simvastatin and their combination (24 h incubation; n = 3; tr = 1; pooled) in GBM U-87 MG cells. Data are expressed as Log2 (fold change) of individual phosphorylation protein levels in response to the different treatments compared with vehicle-treated controls showed in bar plots [threshold: log2 (FC) = 0.2].

Supplementary Fig. S5

Hierarchical heatmap (a) and Principal Component Analysis (PCA) (b) generated using the expression levels of 32 genes of the Senescence-Associated Secretory Phenotype (SASP) in response to metformin, simvastatin, and their combination in GBM cells [cell lines (U-87 MG and U-118 MG) and primary patient-derived GBM cells (PPdC); data represent the mean value of 3 experiments (4 treatment conditions/experiment; tr = 2) in each GBM cell model]. c) Representative images showing the morphology of GBM cells (U-87 MG and U-118 MG) in response to the conditioned media of cells treated metformin, simvastatin, and their combination compared to control condition (n = 3; tr = 2).

Supplementary Fig. S6

Individual expression levels of key apoptosis-related genes in response to metformin, simvastatin, and their combination after their intra-tumour administration in U-87 MG xenograft tumours (n = 4; tr = 1). Data represent median (interquartile range) or means ± SEM (error bars). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001, were set as statistically significant differences vs. control conditions.