Targeting autophagy and plasminogen activator inhibitor-1 increases survival and remodels the tumor microenvironment in glioblastoma

Abstract

Background: Glioblastoma (GBM), the most common and aggressive type of primary brain tumor, engages multiple survival mechanisms, including autophagy. GBM exploits both degradative and secretory autophagy pathways to support tumor growth and limit the efficacy of standard-of-care treatments. We have previously shown that lucanthone, a blood-brain barrier permeable autophagy inhibitor, reduces tumor burden. However, although lucanthone-treated tumors are significantly smaller in size, they are not completely obliterated, suggesting compensatory survival mechanisms. A critical factor for GBM survival is communication with the tumor microenvironment (TME), which can be programmed by glioma cells to support growth and immunosuppression. Plasminogen activator inhibitor-1 (PAI-1), a secreted serine protease inhibitor, has been implicated in the progression of several cancers, including GBM, and has been shown to be modulated by autophagy in other cancers. The role of PAI-1 in GBM, namely its relationship with intracellular autophagy dysregulation and extracellular TME as a mechanism of tumor survival, remains incompletely understood.

Methods: Murine glioma models were established using intracranial injection of GL261 cells in C57BL/6 mice, followed by autophagy inhibition with intraperitoneal lucanthone and/or PAI-1 inhibition with MDI-2268 chow, and tumors were assessed by immunohistochemistry. In culture, glioma cell lines were challenged with MDI-2268, lucanthone, mitoxantrone, or siRNA-LNPs targeting PAI-1, and assessed by MTT assay, q-RT-PCR, ELISA, invasion assay, immunoblot, and immunocytochemistry. Lysosomal markers and transient transfection with fluorescent vesicular proteins were utilized to evaluate PAI-1 intracellular localization via confocal microscopy. Synergy was analyzed using the HSA model in Combenefit, and statistical analyses included t-tests, ANOVA, and log-rank tests for survival.

Results: Lucanthone treatment increased intracellular PAI-1 and autophagy markers while reducing active extracellular PAI-1. PAI-1 colocalized with lysosomal markers, suggesting impaired secretory autophagy. PAI-1 inhibition reduced glioma cell viability and invasion. Combination therapy with lucanthone and MDI-2268 drastically decreased tumor volume, prolonged survival, and promoted a pro-inflammatory state in the tumor microenvironment.

Conclusions: Our findings suggest that PAI-1 may be a compensatory survival mechanism in GBM after autophagy inhibition, and that dual targeting of autophagy and PAI-1 disrupts tumor progression and enhances anti-tumor immunity, providing promising evidence for targeting this axis.

Fig. 1

PAI-1 expression correlates with poor GBM prognosis and is increased in tumors treated with lucanthone. (A) Kaplan-Meier survival curve of GBM patients with tumors of all subtypes that express low or high levels of PAI-1 (SERPINE1) (Gliovis/TCGA_GBM dataset). (B) Kaplan-Meier survival curve of GBM patients with tumors of the proneural subtype that express low or high levels of PAI-1 (SERPINE1) (Gliovis/TCGA_GBM dataset). (C) mRNA expression of PAI-1 (SERPINE1) in GBM compared to non-tumor tissue, unpaired t-test, p-value = 0.00011 (Gliovis/TCGA_GBM dataset using the HG-U133A platform). (D) Correlation plots of PAI-1 expression vs. expression of autophagy markers p62 (SQSTM1) and Cathepsin D (CTSD) in GBM tumors, labeled with respective Pearson correlation coefficients (Gliovis). (E) Cartoon of in vivo mouse GBM model detailing implantation of GL261 cells and treatment course. (F) Representative confocal images of tumors treated with saline control or lucanthone for two weeks and stained with PAI-1, Cathepsin D (CATD), and DAPI nuclear stain. Colocalization of PAI-1 and CATD are shown in the reslice and orthogonal views. Scale bars measure 10 μm. (G) Quantification of the signal integrated density divided by the number of DAPI + nuclei, i.e. per cell, shown in F. Each dot represents one animal, average of 3–5 images per animal, n = 11 animals per group, unpaired t-tests

Fig. 2

Lucanthone increases autophagy markers and intracellular PAI-1 both within and outside lysosomal machinery. (A) Representative western blot of PAI-1 and LC3 expression in GL261 cells treated with DMSO control or 10 µM lucanthone for 24 h. GAPDH was used as loading control. Numbers next to blots are kDa based on protein ladder. (B) Quantifications of PAI-1 and LC3-II signal intensity, normalized to loading control (LC). Data presented as fold change of ctrl (DMSO). Each dot represents one independent experiment, unpaired t-tests. (C) Representative images of GL261 cells treated with DMSO control or 10 µM lucanthone for 24 h in regular media, stained with PAI-1, Cathepsin D (CATD), and DAPI. Colocalization of PAI-1 and CATD are shown in orthogonal views. Scale bars measure 10 μm. (D) Quantification of C. Each dot represents one independent experiment, average of 3–5 images per experiment, unpaired t-tests

Fig. 3

PAI-1 and autophagy inhibition decrease glioma cell proliferation. (A, B). MTT assay measuring percentage of live GL261 cells treated with DMSO control, 1.25 and 2.5 µM MDI-2268, 5 and 10 µM lucanthone, or the combination, respectively, for 72 h in regular media (10% serum) relative to the average of control. n = 3 independent experiments, one-way ANOVA. (C) HSA synergy analysis of the two agents using Combenefit software. (D, E) Cell viability dose-response curves for MDI-2268 and lucanthone, respectively, in GL261 cells, and their EC50 values (% change = % change in relative live cells) plotted using Combenefit. Each data point is the mean (marked with an X) and error bar for 3 independent experiment values. (F) Representative images of GL261 cells treated with DMSO control, 2.5 µM MDI-2268, 10 µM lucanthone, or the combination for 48 h in regular media (10% serum), and stained with Ki67, cleaved Caspase 3, and DAPI. Scale bars measure 10 μm. (G) Quantification of F, each dot represents one independent experiment, average of 3–5 images per experiment, one-way ANOVA

Fig. 4

Lucanthone and combination treatments increase autophagy markers and intracellular PAI-1 both within and outside lysosomal machinery. (A) Representative western blot of PAI-1, p62, CATD, and LC3 expression in GL261 cells treated with DMSO control, 2.5 µM MDI-2268, 10 µM lucanthone, or the combination, for 24 h. Vinculin was used as loading control. Numbers next to blots are kDa based on protein ladder. (B) Quantification of A, normalized to loading control (LC). Data presented as fold change of DMSO control. Each dot represents one independent experiment, one-way ANOVA. (C) Representative images and orthogonal views of GL261 cells treated with the above conditions, stained with PAI-1, CATD, and DAPI. Scale bars measure 10 μm. (D) Quantification of C. Data presented as fold change of DMSO control. Each dot represents one independent experiment, average of 3–5 images per experiment, one-way ANOVA

Fig. 5

Lucanthone abrogates extracellular active PAI-1. (A-C) Active PAI-1 ELISA on conditioned media of GL261 cells treated with DMSO control, either 1.25 or 2.5 µM MDI-2268, 5 or 10 µM lucanthone, or the combination, for 12, 24, or 72 h. Gray boxed plots are zoomed-in views of the y-axis of the depicted conditions. Data presented as fold change of the average of control (DMSO), which was measured in ng/mL of active PAI-1. Each dot represents one independent experiment, one-way ANOVAs, or unpaired t-test for boxed plot in A. (D) Total PAI-1 ELISA on culture media of GL261 cells treated with the labeled conditions. Data presented as fold change of the average of control (DMSO). Each dot represents one independent experiment, one-way ANOVA. (E-H) Active or total PAI-1 ELISA on conditioned media of GL261 cells treated with DMSO control, 3 µM mitoxantrone, or 3 µM mitoxantrone + 10 µM lucanthone, for 1–24 h. Data presented as fold change of the average of control (DMSO). Each dot represents one independent experiment, one-way ANOVA. (I, J) Representative images of GL261 cells treated with control (DMSO), 3 µM mitoxantrone, 10 µM lucanthone, or the combination, at 1 and 24 h, stained with PAI-1, CATD, and DAPI. Quantifications of PAI-1 signal in select conditions shown. Each dot represents one independent experiment, one-way ANOVA in I, one-way ANOVA or unpaired t-tests in J. Scale bars measure 10 μm

Fig. 6

Combination therapy improved animal survival and reduced tumor volume. (A) Schematic diagram of in vivo animal study design. (B) Kaplan-Meier survival curves for animals receiving control (CTRL chow + saline injections), MDI-2268 (MDI chow + saline injections), Lucanthone (CTRL chow + lucanthone injections), and combination (MDI + lucanthone injections) treatments. 7 animals per group, log-rank (Mantel-Cox) tests. (C) Representative images of H&E-stained brain sections for animals in each of the conditions, euthanized at day 21. Comparison of more anterior, bregma 2.25-2.0, (left panel) vs. more posterior, bregma 0.5 − 0.25, (right panel) in the brain. Scale bars measure 1 mm. (D) Tumor volume quantification, n = 5 animals per condition, one-way ANOVA

Fig. 7

PAI-1 inhibition enhanced cytotoxic T-cell infiltration, while lucanthone reduced vascularization. (A) Representative images of tumors at day 21 in the 4 conditions stained with CD8a, CD31, and DAPI. Scale bars measure 20 μm. (B-E) Quantifications of A. Each dot represents one animal, average of 3–5 images per animal, one-way ANOVA

Fig. 8

MDI-2268 and lucanthone treatments decreased Arginase 1 and CD206 expression in vivo. (A) Representative images of tumors at day 21 in the 4 conditions stained with Arginase 1, CD206, and DAPI. Scale bars measure 20 μm. (B-G) Quantifications of A. Each dot represents one animal, average of 3–5 images per animal, one-way ANOVA

Fig. 9

Inhibition of both autophagy and PAI-1 activated immune-stimulatory myeloid cells. (A) Representative images of tumors at day 21 in the 4 conditions stained with Iba1, iNOS, and DAPI. Scale bars measure 50 μm. (B-G) Quantifications of A. Each dot represents one animal, average of 3–5 images per animal, one-way ANOVA

Fig. 10

Schematic diagram of the proposed role of PAI-1 in lucanthone treatment response and how the combination of PAI-1 and autophagy inhibition may act on the intra- and extra-cellular level in GBM. The bottom half of the image depicts a zoomed-in view of the boxed area of a glioma cell in the top portion, challenged with MDI-2268 and lucanthone, and the top depicts a complex TME (not to scale), with labeled tumor cells in light red, blood vessels in dark red, cytotoxic T-cells in blue, microglia in green, and peripheral macrophages in orange, responding to this treatment. Lucanthone increases PAI-1 intracellularly, results in a buildup of undegraded proteins, including PAI-1, inside deacidified lysosomes, and blocks PAI-1 in the extracellular space. Thus, lucanthone is proposed to inhibit both degradative and secretory autophagy. MDI-2268 inhibits PAI-1’s activity and vitronectin binding in the extracellular matrix (ECM), and its combination with lucanthone results in a dramatic reduction in GBM growth, prolonged animal survival, and an immunostimulatory myeloid cell response. Question marks depict mechanisms that remain to be elucidated, i.e. precisely how lucanthone leads to increased PAI-1 outside autophagic machinery, and the method of activation of microglia and macrophages. This figure was created in Biorender