Bypassing Blood-Brain Barrier and Glucose Dependency of Anti-Glioblastoma Drug Candidates Targeting Mitochondrial Respiration

Abstract

We attempt to address two key therapeutic obstacles affecting glioblastoma patients: low ability of anticancer drugs to penetrate the blood-brain barrier (BBB), and temozolomide (TMZ) resistance, by targeting mitochondrial respiration of glioblastoma cells. We designed and tested over 100 new compounds based on the chemical structure of fenofibrate (FF), which in its prodrug form is cytotoxic to cancer cells by causing severe impairment of mitochondrial respiration. The compounds were designed using two key predictive tools: central nervous system-multiparameter optimization (CNS-MPO) and BBB_SCORE. These algorithms assess how effectively compounds can penetrate the BBB. We initially selected PP1 as a lead compound by testing its BBB penetration, metabolic performance, and antitumoral efficacy. PP1 accumulated in brain tumors and triggered glioblastoma cell death. However, PP1-induced inhibition of mitochondrial respiration was followed by an immediate glycolytic response, which attenuated PP1 toxicity in a glucose-dependent manner. To bypass this limitation, we tested two strategies: (1) the use of PP1 in combination with glycolysis inhibitors; and (2) introduction of a new compound, PP211, which inhibited mitochondrial respiration in the absence of a concomitant increase of glycolysis. Although the combination of PP1 with glycolysis inhibitors was very effective in vitro, this drug combination demonstrated elevated toxicity in mice. PP211, instead, attenuated TMZ-resistant tumor growth and prolonged mouse survival with only minimal general animal toxicity. In summary, we developed and tested a novel mitochondria-targeting drug candidate, PP211, which effectively crosses the BBB, overcomes TMZ resistance, and induces tumor cell death independently of glucose levels-while exhibiting minimal systemic toxicity in preclinical models. These findings support further development of PP211 for glioblastoma therapy.

Keywords: blood‐brain barrier; drug development; glioblastoma; mitochondrial respiration.

FIGURE 1 |

Validation of PP211 BBB penetration. (A) BBB-relevant physiochemical properties of PP211. Information regarding compounds structure, IC₅₀, water solubility and theoretical ability to penetrate the blood-brain barrier (CNS-MPO and BBB-score) are included. All computed molecular descriptors were generated by ChemAxon MarvinSketch version 23.2 (https://chemaxon.com/products/marvin). Numbers in green indicate values desirable for improved BBB penetration and low toxicity. Abbreviations: BBB-Score = blood-brain barrier score (probability of a compound to cross BBB); CNS-MPO = central nervous system multifactorial optimization (probability of a compound to accumulate in CNS); hERG = estimated pIC₅₀ value for hERG (the human ether-a-go-go [hERG] potassium channel – surrogate of cardiac toxicity); IC₅₀ = half maximal inhibitory concentration determined for LN229 cells at low (1 g/L) and normal (4.5 g/L) glucose concentration; WM = molecular weight. *hERG was updated according to recently recalculated risk assessment (Sanches et al. 2024). More details regarding synthesis and purity of PP211 is provided in supporting materials. PP211 is covered by the LSU patents US2023/029978. (B) Comparison between PP211 and PP1 BBB permeability (P). The p values were calculated using P = V_A · C_A/(t · S · C_L) equation (Wang et al. 2019) and normalized by corresponding TEER values (see below). FF was used as negative control and caffeine as positive control. Data represent average values from 2 independent experiments in duplicate with standard deviation, n = 4. *Indicates values significantly different from FF, **indicates values significantly different from PP1 (p ≤ 0.05). Inset: Schematic representation of a triple-coculture model of the BBB. Trans-endothelial electric resistance (TEER) was measured to confirm proper formation of the BBB. (C) Comparison of PP1 and PP211 levels in plasma and in brain tumor tissue following ip drug delivery. Brain tumor [GBM12 (TMZ-resistant)]-bearing immunodeficient mice (Foxn1) were ip injected either with PP1 or PP211 both at 100 mg/kg/day. Concentration of the compounds in plasma and brain tumors were evaluated by HPLC following previously developed protocols (Grabacka et al. 2015; Wilk et al. 2015). *Indicates values significantly different (p ≤ 00.5) from IC₅₀ values measured for PP1, and PP211, respectively (IC₅₀ values for PP1 and PP211 are based on monolayer culture of LN229 glioblastoma cells growing in DMEM containing 1 g/L of glucose + 10% FBS). (D) PP211 IC₅₀ values determined for human patient-derived glioblastoma cells, GBM12 (TMZ-sensitive), GBM12 (TMZ-resistant); GBM43; human glioblastoma cell lines (from ATCC), LN229 and U118MG; mouse glioblastoma (GL-261-luc); breast cancer metastases to brain (MDA MB 231-brain); human glioma cancer stem cells; and normal human astrocytes (NHAs). All cells were cultured in normal glucose environment (4.5 g/L). Details related to PP211 synthesis, structure, compound purity and physiochemical properties are in Supporting Information S1 (Datasets S1–5).

FIGURE 2 |

Validation of PP1 BBB penetration. (A) Schematic representation of a triple-coculture model of the BBB, which consists of astrocytes, pericytes and epithelial cells cultured on 24-well transwell membranes with 3 μm pores (Ingraham et al. 2023). Trans-endothelial electric resistance (TEER) was measured to confirm proper formation of the BBB. (B) BBB permeability (P) was calculated using P = V_A · C_A/(t · S · C_L) equation (Wang et al. 2019) and normalized by corresponding TEER values. FF was used as negative control (Grabacka et al. 2015), and caffeine, which effectively crosses the BBB, was used as positive control (Z. Zhang et al. 2011). Data represent average values from two independent experiments in duplicate with standard deviation, n = 2. *Indicates values significantly different from FF, **indicates values significantly different from caffeine (p ≤ 0.05). (C) Levels of PP1 in plasma and in brain tumor tissue were evaluated at 2 h following ip drug delivery. Brain tumor [GBM12 (TMZ-resistant)] bearing immunodeficient mice (Foxn1) were ip injected with PP1 at 100 mg/kg/day. Concentrations of the compound in plasma and brain tumors were evaluated by HPLC following previously developed protocols (Grabacka et al. 2015; Wilk et al. 2015). *Indicates values significantly different (p ≤ 00.5) from PP1 glioblastoma IC₅₀ value depicted in the inset (based on monolayer culture of LN229 glioblastoma cells growing in DMEM containing 1 g/L of glucose + 10% FBS).

FIGURE 3 |

Metabolic effects of PP1. Metabolic responses to PP1 were evaluated in LN229 human glioblastoma cells using Extracellular Flux Analyzer XF96 (Seahorse/Agilent). (A) The oxygen consumption rate (OCR, indicator of mitochondrial respiration) and (B) extracellular acidification rate (ECAR, indicator of glycolysis) were evaluated after injecting DMSO (control), PP1 (25 μM) or FF (50 μM), followed by the Mitochondrial Stress Assay: sequential injection of oilgomycin (OLIG), FCCP, and rotenone (ROT). Data represent average values ± SD (n = 5). *Indicates values significantly different from DMSO control (p ≤ 0.05). (C) Effects of PP1 on intracellular ATP levels. LN229 glioblastoma cells were harvested, counted and analyzed for intracellular ATP (CellTiter-Glo 2.0, Promega) at indicated time points following PP1 treatment (25 μM). Values below the histogram show % LN229 cell death evaluated by trypan blue exclusion test. Data represent average values with SD, n = 3. (D) Western blot showing a severe metabolic disturbance triggered by PP1. Total proteins were extracted at indicated time points for the detection of: pAMPK (Thr172) (low-energy sensor); p70S6K (Tyr389) (marker of active translation); SQSTM1/p62 (marker of active autophagy); and loading control, GAPDH.

FIGURE 4 |

PP1-induced glioblastoma cell death is glucose dependent. LN229 glioblastoma cells were cultured in low or normal glucose medium. (A) 7ADD-based cell death assay using easyCyte flow cytometry. Cell death was evaluated at 72 h following exposure to DMSO (control), PP1 (25 μM), PP1 + glucose, PP1 + glutamine, or PP1 + pyruvate in low glucose. For all panels, the treatment with PP1 on its own or in combination with other compounds was applied once, at 24 h following the cell plating. (B) Effects of glycolysis inhibitors (^#GH and ^##LND) on PP1-induced cytotoxicity. The cells were treated with PP1 (25 μM), GH (10 μM), LND (200μM), PP1 + GH, PP1 + LND and DMSO (control). Panels (A) and (B) data represent average values with standard deviation (n = 3). *Indicate values significantly different from PP1 (p ≤ 0.05). (C) PP1 IC₅₀ comparison between LN229 human glioblastoma cells cultured in low (1 g/L) and normal (4.5 g/L) glucose concentration. Data represent average MTT values, (n = 3). (D) Effects of PP1 on intracellular ATP and Glucose content in the medium. ATP level was measured in low glucose (1 g/L) using CellTiter-Glo 2.0 (Promega) and glucose content by Gluco-Glo (Promega) at indicated time points following DMSO (control) PP1, PP1 + LND, and PP1 + GH treatments. Data represent average values with standard deviation (n = 3). *Indicate values significantly different from 1 h time point (p ≤ 0.05). #GH—gnetin H (a natural product found in Vitis vinifera and potent inhibitor of glycolysis); ^##LND—Lonidamine (BBB -penetrable inhibitor of glycolysis). Values below the histogram show % LN229 cell death evaluated by trypan blue exclusion test. The cells were cultured in DMEM low glucose medium (1 g/L) + 10% FBS. (E) Western blot showing a severe metabolic disturbance triggered by PP1. Total proteins were extracted at indicated time points for the detection of: pAMPK (Thr172) (low-energy sensor); p70S6K (Tyr389) (marker of active translation); SQSTM1/p62 (marker of autophagy); Caspase 3 antibody that recognizes both pro-Caspase 3 and cleaved-Caspase 3 (marker of apoptosis), and loading control, GAPDH.

FIGURE 5 |

Characterization of PP211 as a new compound candidate that preferentially targets glioblastoma mitochondria. (A) Metabolic responses to PP1 and PP211 evaluated in LN229 cells using XF96. OCR and ECAR values were measured after a single injection of PP1, PP211 (25 μM each) or DMSO (control), followed by mitochondria-stress assay. Data represent average values ± SD (n = 4) *indicates values different from DMSO; **indicates values different from DMSO and PP1. (B) Time course analyses of intracellular ATP and cell survival following PP211 treatment. GBM cells treated with 25 μM PP211 were collected at indicated time points for ATP measurement (CellTiter-Glo 2.0, Promega) and trypan blue exclusion test. Data represent average values with SD (n = 3). (C) Cytotoxic effects of PP211 on GBM12 (TMZ-resistant) 3-D gliospheres. The gliospheres were treated with 25 μM PP211 in high glucose medium supporting growth of glioma initiating cells (GICs). Data represent average values with SD (n = 3). Representative phase contrast image of gliospheres in the presence or absence of PP211. In Panels (C) and (D) *indicates values different from DMSO controls, p ≤ 0.05. (D) Capillary Western blot analysis showing a severe metabolic disturbance triggered by PP1. Total proteins were extracted at indicated time points for the detection of: pAMPK (Thr172) (low-energy sensor); p70S6K (Tyr389) (marker of active translation); SQSTM1/p62 (marker of active autophagy); Caspase 3 antibody that recognizes both pro-Caspase 3 and cleaved-Caspase 3 (marker of apoptosis), and loading control, GAPDH. (E) Mitochondrial Membrane Potential (MMP) was measured in LN229 and in normal human astrocytes (NAHs) using live-cell imaging system (Incucyte) set for Incucyte MMP Orange Reagent [Orange (Ex/Em ~552/578 nm]. The cells were plated in 96-well plates at 1.9 × 10⁴/ well. Following attachment cells were loaded with 20 nM Incucyte MMP Orange Reagent for 1 h at 37°C. The cells were subsequently treated with PP211 (25 μM), DMSO (0.25%—vehicle), FCCP (positive control for loss of MMP), and Oligomycin (OM—negative control for loss of MMP) according to manufacturer recommendations. Fluorescent data (total area of fluorescence) were normalized by corresponding fluorescence values at time 0. Data represent average values from three separate experiments in triplicates with standard deviation.

FIGURE 6 |

Anti-glioblastoma effects of PP211. (A) Following daily ip injection of PP221 (100 mg/kg/day) or solvent (DMSO in 20% cyclodextrin) tumors were evaluated by IVIS CT. Before treatment, animals were randomly selected to control and experimental group. Data represent average values with standard deviation (n = 6 per group). *Indicates values statistically different from DMSO. Inset: IVIS luminescence showing difference in the tumor size between control (DMSO) and experimental (PP211) mice at day 25. (B) Kaplan-Meier survival curve. Additional 11 and 14 tumor bearing mice were treated either with DMSO (control) or PP211, respectively. *Indicates statistically significant difference in mouse survival between PP211 and DMSO -treated groups. (C) Detection of PP211 in normal brain tissue (brain homogenate was prepared from brain area corresponding to the tumor cell injection (see Section 4) at indicated time points following ip injection of PP211 (100 mg/kg). Red line indicates average PP211 IC₅₀ concentration determined in GBM12/TMZ-resistant gliospheres at 4.5 g/L of glucose. Data represent average values (HPLC-detected PP211 concentrations in μM) with standard deviation, n = 3. (D) Body weight determined at 1, 14, and 21 days following DMSO or PP211 treatment. Please note PP211-treated mice maintain their body weight, in contrast to control mice (DMSO) which show a significant drop in body weight at day 21, most likely due to the tumor growth. For Panels (A) and (D), data represent average values with SD (n = 6). *Indicates values significantly different from DMSO (in panel A), or values different from day 1 (in panel D) (p ≤ 0.05). (E) Intracranial glioblastoma at low (×20) and high (×200) magnification. Mice were euthanized on day 25 of the treatment and brain sections containing tumors (rectangle) and unaffected brain areas were stained with H&E. In lower panel, H&E staining of the heart, liver kidney and spleen are presented showing lack of pathological changes between control (DMOS-treated) and experimental (PP211-treated) mice on day 21 of the treatment.