Anti-glioblastoma effects of phenolic variants of benzoylphenoxyacetamide (BPA) with high potential for blood brain barrier penetration

Glioblastomas are the most aggressive brain tumors for which therapeutic options are limited. Current therapies against glioblastoma include surgical resection, followed by radiotherapy plus concomitant treatment and maintenance with temozolomide (TMZ), however, these standard therapies are often ineffective, and average survival time for glioblastoma patients is between 12 and 18 months. We have previously reported a strong anti-glioblastoma activity of several metabolic compounds, which were synthetized based compounds, which were synthetized based on the chemical structure of a common lipid-lowering drug, fenofibrate, and share a general molecular skeleton of benzoylphenoxyacetamide (BPA). Extensive computational analyses of phenol and naphthol moieties added to the BPA skeleton were performed in this study with the objective of selecting new BPA variants for subsequent compound preparation and anti-glioblastoma testing. Initially, 81 structural variations were considered and their physical properties such as solubility (logS), blood–brain partitioning (logBB), and probability of entering the CNS calculated by the Central Nervous System—Multiparameter Optimization (MPO-CNS) algorithm were evaluated. From this initial list, 18 compounds were further evaluated for anti-glioblastoma activity in vitro. Nine compounds demonstrated desirable glioblastoma cell toxicity in cell culture, and two of them, HR51, and HR59 demonstrated significantly improved capability of crossing the model blood–brain-barrier (BBB) composed of endothelial cells, astrocytes and pericytes.

glioma initiating cells (GICs) have been identified as a primary mechanism involved in the development of drug resistance [11][12][13][14][15] . GIC's display an embryonic stem cell phenotype 11,[16][17][18] and express stem-cell markers including CD133, CD44, SOX2, L1CAM, CD15, integrin α6, and BMI1 [18][19][20][21][22][23][24] . In addition, GICs express ATP binding cassette (ABC) transporter proteins, which mediate extensive drug efflux contributing to radio-and chemo-resistance 25 , neovascularization 26 , and invasiveness 27 ; Fourth, early diagnosis of glioblastoma is rare, therefore, large highly infiltrating and vascularized tumors are often present at the time of diagnosis 28 ; Fifth, the optimization of clinical protocols for glioblastoma treatment requires the use of a reliable preclinical model/s. Unfortunately, commonly used rodent syngeneic and xenograft models have one major problem-the experimental tumors are typically ~ 10 3 -10 4 times smaller than human tumors, and therefore, drug delivery, drug retention, and effective tissue penetration by the drug, cannot be tested in a reliable manner in small animal models 29 ; and finally Six, the blood brain barrier (BBB) prevents the majority of anticancer drugs from reaching the tumor, and current methods that enhance the BBB penetration are still not effective for glioblastoma patients 30 .
One of the drugs that readily crosses the BBB is temozolomide (TMZ). Upon oral administration, TMZ maximum plasma concentration can be reached in about one hour, and the elimination half-life is approximately 2.1 h. Importantly, penetration efficiency of TMZ into the CNS is experimentally estimated to be about 20% of the plasma levels 31 . Applying this estimate to calculate the logBB (Brain-Blood Distribution) for TMZ, this equation produces a value of − 0.7, which indicates sufficient capability of the compound to cross the BBB 32 . In spite of these positive features, TMZ-treated glioblastoma patients develop TMZ-resistance and recurrent tumors are practically incurable 33 . There are also several studies of the use of TMZ in combination with other drugs, which show beneficial therapeutic effects 34,35 . One interesting example is a combination of TMZ with lipid lowering drugs, including statins 36 . In addition, another class of lipid-lowering compounds, fibrates, have also attracted attention as a possible anticancer drugs [37][38][39][40] . We have previously reported that 50 µM fenofibrate (FF) has a strong anti-glioblastoma activity in cell culture, and in glioblastoma mouse models following intratumoral injection 41 (Fig. 1). However, FF does not cross the BBB, and is quickly processed by the blood and tissue esterases to form fenofibric acid (FFA), which is no longer effective in triggering tumor cell death 41,42 .
We have previously made several chemical modifications to the FF molecular skeleton, to address the FF low stability in human blood, low water solubility, and inability of penetrating the BBB. Indeed, one of the initial compounds, PP1, demonstrated improved water solubility and stability in human blood. In addition, PP1 was capable of triggering extensive glioblastoma cell death in vitro at concentrations over fourfold lower than FF 43 (Fig. 1). To further improve anti-glioblastoma efficacy, we created other FF derivatives, which share the benzoylphenoxy-acetamide (BPA) molecular skeleton, and decided to test the addition of phenol and naphthol residues to the BPA structure due to the potential anti-cancer effects of these moieties [44][45][46][47] . As a result, 18 new compounds were generated and were analyzed during this study.

Results and discussion
Overall chemical design. In our previous studies we have explored the importance of a basic BPA skeleton 48 , and concluded that BPA could serve as a "pharmacophore", necessary to retain anti-glioblastoma activity 49,50 . The amide part of the BPA skeleton can be specifically modelled to obtain a more desirable antitumor activity. This includes, among other properties, chemical and physical parameters (described below) that contribute to the increased BBB penetration, and possibly drug retention within the tumor tissue. In this regard, we have selected phenol and naphthol residues due to mounting evidence supporting the role of different derivatives of these compounds in health benefits 51 , including anti-cancer activities 45,46 . In this paper, three variants of BPA are discussed: a substituted phenol (Phenolic-BPA), and two naphtholic BPAs (1-Naphtholic-BPA and 2-Naphtholic-BPA) (Fig. 2) that serve as prototype molecules for further modifications.
The starting point for the preparation of all phenolic BPAs is fenofibric acid (FFA), and the corresponding aminophenol or aminonaphthol residues (Fig. 3) are added through amide (peptide) coupling reactions 52,53 . As previously reported 48 , due to the steric hindrance of the carboxylic group of FFA, which includes two methyl groups in the alpha position of carboxylic acid, combined with the lower amine nucleophilicity of anilines in DCC-or EDC-coupling, these reactions do not produce acceptable isolated yields. This occurs even with more reactive aminophenols and EDC or DCC, which are stronger nucleophiles compared to nonactivated anilines, which is expected to produce corresponding BPA compounds in acceptable yields 54 Figure 1. Comparison between FF, FFA, and PP1 structural and anti-cancer properties. The information regarding the compounds water solubility, stability in human blood, and in vitro cytotoxicity were previously reported 48  www.nature.com/scientificreports/ more reactive fenofibrate chloride (FFC), followed by coupling with aminophenols or aminonaphthols (Fig. 3).
We have explored several variations of this procedure and finally selected one that is very simple and can be applied to multigram and even multikilogram production scales. In particular, the FFC was prepared fresh and immediately used, for the next step of aminophenol addition (Fig. 3). The most common method of preparation of an acid chloride is by using thionyl chloride. This requires heating of thionyl chloride with the corresponding acid (in this case FFA) with appropriate traps for hydrochloric acid and sulfur dioxide, which are undesirable byproducts of the reaction 55 .
Biological, chemical and computational testing of the compounds. Previously, we have reported that some amide derivatives of FF, including PP1, are more potent than others in eliminating glioblastoma cells 43 and belong to the large family of BPA 56 . Further modifications of the BPA structure were designed to produce compounds with the improved ability to penetrate BBB, and superior glioblastoma cytotoxicity. This is highly relevant for glioblastoma patients, since one fundamental challenge for new drug development is the need for effective BBB penetration. So far, known BBB-permeable compounds form a very small subset of current oral drugs, and experimental models for testing BBB penetration by new drugs are complex and expensive. Therefore, an independent indicator of the BBB penetration is necessary for faster and more cost effective analyses of new anti-glioblastoma drug candidates. This is critical for the initial screening of a large number of compounds, including BPA variants, and for selection of best candidates for subsequent measurements of intracranial tumor drug penetration and anti-tumoral efficacy. Therefore, we performed extensive molecular modeling prior to preparation of new BPA variants in order to evaluate their specific physicochemical properties considered relevant for the BBB penetration. We have applied first a weighted scoring approach named "Central Nervous System-Multiparameter Optimization" (CNS-MPO) 57,58 . The CNS-MPO algorithm uses 6 key physicochemical properties (clogP, clogD, MW, TPSA, HBD, and pKa) with the scores ranging between 0 and 6.0. Importantly, scores ≥ 4.0 are widely accepted for selecting compounds with high potential for CNS penetration 58 . The validation of this approach was based on a library of 616 compounds for which experimental distribution of the drug in CNS was determined and the corresponding  www.nature.com/scientificreports/ parameters of the compound incorporated into CNS-MPO scores 57,58 . It was found that CNS-MPO scores of 1-2 (0%), 2-3 (11.6%), 3-4 (40.8%), 4-5 (53.8%) and 5-6 (81.1%) correlated with the increased probability of drugs to be found in the brain 59 .
In addition to CNS-MPO, other chemical and physical parameters of the Quantitative Structure Activity Relationship (QSAR) studies were also used in our calculations to further increase the probability of BBB penetration by the compound. These include: molecular polarizability (MP), minimal molecular projection area (MPA), water solubility (LogS), and lipophilicity (cLogP), to name a few. These parameters are not incorporated in the CNS-MPO score, however, they also play a role in the estimation of compound ability to penetrate the BBB 60 . Therefore, we have calculated and incorporated all these additional parameters when evaluating and selecting new HR compounds as potential anti-glioblastoma drugs (Figs. 4,5,6,7,8,9); and Supplementary materials). In particular, MP is a response of electron distribution to an externally-applied static electrical field. Comparing MP values helps in understanding how different substrates may change polarization-and dispersion-type during interactions with the active sites of their interacting proteins 61 . It was postulated that MP values between 30 and 40 (see Supplementary data) are optimal for a molecule to bind to a biotarget 62 . Minimal projected area (MPA) represents another parameter that is important for drug transport and ultimately for drug activity. For instance, a distinct phenotypical pattern of drug recognition and transport for the G616N variant was reported, indicating that drug substrates with MPA over 70 Å 2 are less likely to be transported compared to drugs with smaller MPA 63 . In addition, LogS of − 4.5 and greater are indicators of acceptable water solubility 64 , and the rate of passive diffusion is inversely proportional to the square root of molecular size (Graham's law 65 ), which are also included in our compound analysis.
Finally, the ability of the compound to penetrate the BBB can also be expressed as a decimal logarithm of brain to plasma concentration ratio (logBB), which is derived from the modified Clark's equation: logBB = 0.152 ClogP-0.0148PSA + 0.139 30. It has been shown that compounds with logBB > 0.3 readily cross the BBB, while the compounds with logBB < − 1.0 have a problem with the CNS penetration 32 . Therefore, LogBB values for all HR compounds were also calculated and included (Figs. 4,5,6,7,8).
Once all above calculations were performed, initial cell viability (CV) tests were performed for all HR compounds using LN229 human glioblastoma cell line. The cells were treated with proposed HR compounds at 25 μM, and cell viability was evaluated using MTT assay, following a continuous cell exposure to a single dose of the drugs for 72 h. (Figs. 4,5,6,7,8,9).
As a result of this initial screening, we have identified two lead drug candidates, HR51 and HR59, with phenolic moieties that contain BPA structural skeleton similar to our prototype anti-glioblastoma compound PP1 43 . This is in addition to our recently reported BPA-based compounds (HR28, HR32, HR37, and HR46), which also demonstrated high potential as anti-glioblastoma drugs 48 . Anti-glioblastoma effects of HR51 and HR59 were subsequently confirmed using four different human glioblastoma cell lines, LN229, U-87 MG, U-118 MG, T98G, and the cytotoxicity data were compared to normal human astrocytes (NHA). Results in Fig. 9A demonstrate that all tested glioblastoma cells were partially responsive to 10 µM HR51 and HR59, but were almost completely eliminated following 72 h exposure to 25 µM HR51 or HR59. In contrast, these two compounds were significantly less cytotoxic to normal human astrocytes (NHA), indicating that these two new compounds may have low CNS toxicity. In addition, results in Fig. 9B show that IC 50 concentrations for HR51 and HR59 are below 10 μM, which is an acceptable therapeutic concentration for clinically relevant anticancer drugs.
Another important feature shared by HR51 and HR59 is the compounds ability to interfere with the sphere formation and proliferation of GBM12-patient-derived glioblastoma cells in low density 3-D cultures designed for propagation of glioma initiating cells (GICs) 66 . In this sphere formation assay, low density single cell suspension cultures of GBM12 were exposed to serum-free, growth factor supplemented medium (see "Methods"). As shown in Fig. 9C, only DMSO-treated GBM12 cells (vehicle control) were able to proliferate and formed large multicellular gliosphere structures. In the presence of 25 μM HR51 or 25 μM HR59 the cells did not proliferate (not shown), and their ability to form 3-D gliospheres was severely attenuated, indicating that not only bulk tumor cells but also tumor stem-like cells are the targets for HR51 and HR59.
We have also tested if the mechanism of action of HR51 and HR59 is similar to our prototype drugs, PP1 and fenofibrate, which have been previously shown to inhibit mitochondrial respiration at the level of Complex 1 of the electron transport chain (ETC) 41,43 . Indeed, results in Fig. 9C confirmed that both HR51 and HR59 inhibit mitochondrial respiration by the magnitude similar to PP1.
In addition to a strong in vitro anti-glioblastoma activity (Fig. 9), HR51 and HR59 have physical properties that may contribute to the improved brain tumor penetration. Specifically, HR51 and HR59, have a minimal projection area (MPA) of 46.23 Å 2 and 43.73 Å 263 , respectively; water solubility (LogS) of − 6.61 and − 6.11 64 ; and brain to plasma concentration ratio (LogBB) of − 0.15 and − 0.49 32 , which are all considered as highly promising for compounds suspected of being capable of penetrating the brain tumor tissue.
Importantly, HR51 and HR59 can also cross the triple-coculture model of the blood brain barrier (BBB), which consists of astrocytes, pericytes and epithelial cells cultured on 24-well (3 µm pores) transwell membranes (Fig. 10A), prepared according to the protocol provided by Stone et al. 67 . All experiments in which the ability of HR compounds to cross the BBB were evaluated using the BBB inserts that demonstrated a significant increase of the electric resistance (Ω) in comparison to inserts without cells (Ω BBB − Ω Insert ) (Fig. 10B). In particular, Ω BBB − Ω Insert values of 105.3 ± 16; 110.6 ± 14; 117.8 ± 19; 132.3 ± 23; and 129.0 ± 24 were measured for inserts used to evaluate BBB penetration of FF, caffeine, HR51 and HR59, respectively (Fig. 10B). Accordingly, transendothelial electric resistance (TEER = Ω BBB − Ω Insert × area of the membrane) 67 42 ) were added to the corresponding insert and the plates were incubated in 37 °C and 5% CO 2 for 24 h. Subsequently, aliquots of media from the corresponding inserts and from the wells were collected for HPLC measurements and to calculate BBB permeability (P = V A ·C A /(t·S·C L ) 70 . Results in Fig. 10C show that HR51, HR59 and caffeine cross the in vitro BBB at levels 4.4-fold, 3.5-fold and 22.0-fold higher compared to our internal negative control, fenofibrate, which although has a similar molecular weight and structure to the tested HR compounds (Fig. 4), its ability of crossing natural BBB is very low 42 .
By exploring the computed structural variations of phenolic-BPAs, and testing their cell toxicity, we have demonstrated that the addition of phenol moieties improves anti-glioblastoma activity with acceptable LogBB   www.nature.com/scientificreports/ and LogS properties. This could be further improved by adding additional substituent(s) to the phenol moiety, including hydroxy group, halogen, alkyl, nitro on carbonyl to name a few. However, replacing the phenol residue with the larger naphthol, although it might not compromise anti-glioblastoma activity in cell culture, it may result in decreased compound bioavailability mostly due to low water solubility (Fig. 7) 71 . Therefore, these naphthol compounds are less likely to be considered as anti-glioblastoma candidates. Presented in silico analyses of phenolic BPAs in combination with the in vitro BBB penetration model, represent a quick and relatively inexpensive way of testing a large number of new glioblastoma drug candidates. The observed effective penetration of the BBB model membranes by HR51 and HR59 is therapeutically promising, and it also validates our overall strategy for designing and selecting compounds with the most favorable

Methods
All starting materials were reagent grade and purchased from Sigma-Aldrich, ArkPharm, and TCI America. 1

5-methylphenyl)-2-methylpropanamide (HR52).
A dichloromethane (10 mL) suspension of fenofibric acid (FFA) (191 mg; 0.6 mmol) and oxalyl chloride (0.2 mL; 384 mg; 2 mmol) was stirred at room temperature for five minutes. A few drops of N,N-dimethylformamide (DMF) were then added to the suspension, which induced bubbling, and resulted in a clear reaction mixture after approximately 30 min. This solution was then stirred at 60 °C to promote slow solvent evaporation. The solvent residue and oxalyl chloride were removed by drying under an Argon flow at room temperature. The resulting yellow solid substance was dissolved in dichloromethane (10 mL) and mixed with 2-amino-4-methylphenol (62 mg; 0.5 mmol) in THF (10 mL) and Na 2 CO 3 (1.06 g; 10 mmol) in water (10 mL). The subsequent mixture was stirred at room temperature for five hours. The solvent was then evaporated under reduced pressure and the residue was mixed with dichloromethane (50 mL) and water (50 mL). This final mixture was sonicated at room temperature until all solid was dissolved. From this bilayer solution, the water layer was discarded, and the dichloromethane layer was washed with 5% Na 2 CO 3 (3 × 50 mL), water (50 mL), 5% HCl (3 × 50 mL), water (50 mL) and dried over anhydrous Na 2 CO 3 . After solvent evaporation, the final product was purified by crystallization from dichloromethane (~ 3 mL) and hexane (20 mL   Cell culture and viability assays. All cell culture procedures used for HR compound testing were previously described 43,48 . Briefly, human glioblastoma cell line, LN-229 (ATCC# CRL-2611), U-87MG (ATCC# HTB-14), U-118 MG (ATCC# HTB-15) and T98G (ATCC# CRL-1690) were maintained as semi-confluent monolayer cultures in DMEM (1 g/L glucose; with sodium pyruvate and l-glutamine) supplemented with 100 U/ml penicillin, 100 μg/mL streptomycin, and 10% fetal bovine serum (FBS) at 37 °C in a 5% CO 2 atmosphere. GBM12 are patient-derived human glioblastoma cells established in Dr. Sarkaria's lab 33,73 . GBM12 cells are routinely propagated in subcutaneous tissue of nude mice (IACUC protocol # 3750,LSUHSC, New Orleans), and were isolated from the tumor tissue for short-term monolayer and suspension cultures as previously reported 41 .
For three-dimensional (3-D) gliosphere cultures, the cells aggregates were dissociated with Accutase (Innovative Cell Technologies: AT104), and low density single cell suspension cultures of GBM12 (1 × 10 4 /mL) were incu- In the sphere formation assay, low density single cell suspension of GBM12 cells was exposed to the serum-free stem-cell supporting medium (StemPro NSC SFM; Gibco: A1050901) supplemented with Recombinant Human FGF Basic and EGF (10 μg each), as previously reported 66,72 . The cells were allowed to form spheres for 5 days in the presence or absence of HR51 and HR59. DMSO-treated cultures (vehicle) were used as positive control for the sphere formation. In this condition, GBM12 cells grow in suspension and display stem-like phenotype-formation of multicellular gliospheres in low density cultures 66 . (D) Metabolic effects of HR51 and HR59 compared to the prototype drug, PP1. Metabolic responses to the drugs were evaluated in LN229 using Extracellular Flux Analyzer XF96 (Seahorse/Agilent). The oxygen consumption rate (OCR; indicative of mitochondrial respiration) was evaluated after injecting DMSO, (negative control), PP1 (positive control) and two experimental drugs, HR51 and HR59, followed by sequential injections of oilgomycin, FCCP, and rotenone (mitochondrial stress assay). Average OCR data were calculated from three independent experiments. Data represent average values ± SD. Compared to negative control (DMSO), all tested metabolic compounds (PP1, HR51 and HR59) triggered an immediate drop in OCR. In addition, the cells treated with these three compounds did not respond to FCCP injection, indicating loss of the proton gradient across the mitochondrial membrane. www.nature.com/scientificreports/ bated in the serum-free growth factor-supplemented stem-cell medium (StemPro NSC SFM: A1050901, which consists of KnockOut DMEM/F-12 Basal Medium (500 mL), plus StemPro NSC SFM Supplement (10 mL), 10 μg recombinant human basic FGF and 10 μg EGF) 66 . In the sphere-formation assay, the cells were allowed to form 3-D gliospheres, and the sphere morphology was evaluated 5 days following the plating. Normal human astrocytes (NHA) were used to evaluate effects the selected HR compounds may have on normal non-transformed cells. NHA were cultured according to the manufacturer protocol (LONZA/Clonet-ics™). Prior to the treatment with HR compounds, the cells were plated in 96-well plates (BD Falcon) at the initial density of 2 × 10 4 /well. Twenty-four hours after plating, stock solutions of the HR compounds were prepared in DMSO, diluted in cell culture medium, to the final concentration of 25 µM, and added to the cells in triplicate for every experimental condition. For the vehicle control, DMSO was used at 0.5%. After 72 h incubation, MTT assay (surrogate for cell viability) was performed according to our previous publications 48 . Briefly, following 1.5 h incubation with MTT, formazan crystals were dissolved in 5 mM HCl in isopropanol and the absorbance read at 540 nm. Data represent mean values expressed as % cell viability of control (DMSO) ± SD. Phase contrast images of treated cells were taken 72 h after treatment with HR compounds with a BZ-X800 fluorescence microscope (Keyence) equipped with a 20 × objective. The drug dose causing 50% inhibition in the MTT assay at 72 h time point (half maximal inhibitory concentration-IC50) was calculated using GraphPad Prism 8.
In vitro model of the blood brain barrier (BBB). The blood brain barrier was re-created in vitro using a modified protocol provided by Stone et al. 67 . Briefly, 24-well transwell inserts (Falcon, catalog number 353096) were coated with 10 μg/cm 2 of Collagen Type IV from human placentas (Sigma) for 24 h at 4 °C. Inserts were washed with sterile water and air-dried for 2 h. Next, the inserts were coated with 2 μg/cm 2 poly-l-lysine (Scien-Cell) for 1 h at 37 °C, then washed twice with sterile H 2 0 and air-dried for 2 h. 1.57 × 10 5 primary human astrocytes (ScienCell) and 3.125 × 10 4 primary human pericytes (ScienCell) were resuspended in 25 μL of astrocyte medium and pericyte medium (ScienCell), respectively, then combined in a 1:1 ratio for 50 μL total volume. Dried, coated inserts were turned upside down such that the basolateral surface was exposed at the top, and 50 μL of the cell mixture was added to the membrane, covered with the plate lid, and incubated for 2 h at 37 °C to allow cell adherence. Any medium remaining on top of the membrane was carefully removed before return- Figure 10. Penetration of the selected HR compounds through in vitro BBB model membrane: (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. Trans-endothelial electric resistance (TEER) was measured using a EVOM 2 meter with a STX3 electrode (World Precision Instruments). (B) Measurements of the electric resistance of inserts used for specific compounds (Ω); Difference between resistance of the insert with established BBB versus empty insert (Ω BBB − Ω insert ); and trans-endothelial electric resistance (TEER; Ω BBB − Ω insert /cm 2 ). * TEER coefficient = TEER compound /TEER caffeine . TEER coefficients were used to normalize BBB permeability (P) values to compensate for differences in TEER values between inserts selected for each compound. (C) Differences in BBB permeability (P) calculated using P = V A ·C A /(t·S·C L ) eqution 70  www.nature.com/scientificreports/ ing inserts to their upright position with the apical surface facing upward as they were placed in a 24-well plate containing 500 μL per well of astrocyte/pericyte medium (1:1). An additional 300 μL of medium was added to the apical compartment. Four days after plating, the apical compartment medium was removed, and 3.75 × 10 4 of telomerase-immortalized vain endothelial cells (TIVE; provided by Dr. Rolf Renne) in 50 μL of TIVE medium 74 were added and incubated for 5 h at 37 °C to allow cell adherence, followed by the addition of an extra 250 μL of TIVE medium. Half the volume of the corresponding media in the lower and upper compartment was replaced with fresh media every third day. Ten days after initial plating, trans-endothelial electric resistance (TEER) was measured using a EVOM 2 meter with a STX3 electrode (World Precision Instruments). The ability of selected HR compounds to pass through in vitro BBB was tested using inserts with effectively reconstructed BBB as confirmed by TEER 67,68 .
HPLC detection of selected HR compounds. Following TEER measurement, the medium from apical compartment of the in vitro BBB model (Fig. 10A) was replaced with 350 μL of fresh TIVE medium containing corresponding compounds [HR49 (PP99), HR51 (PP12), HR59 (PP171)] all used at 25 μM. In addition, 25 μM fenofibrate (FF), which does not cross the BBB 42 was used as negative control, and 50 μM caffeine was used as a positive control 69 . Plates containing the inserts were returned to the incubator (37 °C, 5% CO 2 ), and after 24 h of incubation conditioned media from basolateral (well) and apical (insert) compartments were collected and frozen for quantitative analyses by HPLC. 100 μL aliquots of the collected samples were subsequently mixed with 100 μL of 100% acetonitrile, samples were centrifuged at 16,000 rpm at 4 °C for 10 min and supernatants collected for HPLC analyses using UltiMate 3000 system (Thermo Scientific) equipped with analytical YMCbasic, 3 µm, 150 × 4.6 mm column (octyl silane C8; YMC America, Inc.). Isocratic elution of the compounds was performed using mobile phase composed of solvent A (50 mM acetic acid in dH 2 O) and solvent B (acetonitrile) mixed at ratios predetermined for each compound (Table 1). All separations were carried out with sample volume of 5 μL at flow rate of 1 mL/min, at 20 °C. Concentration of each compound was calculated using serial dilutions of the known concentration of the compound separated at the same run with experimental and control samples. After separation, integrated areas under the peak were used to prepare calibration curves and to determine concentration of the compounds.
Evaluation of metabolic parameters. Metabolic responses of human glioblastoma cells were evaluated with Extracellular Flux Analyzer XFe96 (Agilent Technologies) according to our previously established protocol 43 . During the day prior to each assay the cells were plated at 2 × 10 4 cells/well in Agilent Seahorse 96-well XF cell culture microplates with growth supporting media and incubated overnight. At the time of measurement, growth media were replaced with serum-free XF assay medium (Seahorse XF Base Medium supplemented with 1 mM sodium pyruvate, 2 mM glutamine, and 5.5 mM glucose) and cartridges equipped with oxygen-sensitive and pH-sensitive fluorescent probes were placed above the cells. The oxygen consumption rate (OCR; indicative of mitochondrial respiration) was evaluated after injecting HR compounds or PP1 (all used at 25 μM), or DMSO (0.1%; vehicle control). These initial injections were followed by sequential injections of metabolic toxins to execute mitochondrial stress assay: oligomycin (inhibitor of ATP synthase; 0.5 μM); carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP; uncoupling factor; 0.5 μM), rotenone (inhibitor of mitochondrial complex I; 0.3 μM), and antimycin A (inhibitor of mitochondrial complex III; 0.3 μM).