New tetrahydroisoquinoline derivatives overcome Pgp activity in brain-blood barrier and 1 glioblastoma multiforme 2

5 1 Department of Oncology, University of Torino, via Santena 5/bis, 10126, Torino Italy 6 2 Department of Drug Science and Technology, University of Torino, via Pietro Giuria 9, 10125, 7 Torino, Italy 8 *Equal contribution 9 10 Corresponding authors: Dr. Stefano Guglielmo, Department of Drug Science and Technology, 11 University of Torino, via Pietro Giuria 9, 10125, Torino, Italy; Phone: +390116707178; email: 12 stefano.guglielmo@unito.it; Dr. Chiara Riganti, Department of Oncology, University of Torino, via 13 Santena 5/bis, 10126, Torino Italy. Phone: +390116705857; email: chiara.riganti@unito.it 14


Introduction
Glioblastoma multiforme (GB) is considered the most common, aggressive and lethal brain tumor in adult population, because of the high infiltration into surrounding brain tissue.GB usually occurs within the white matter as a heterogeneous lesion, but it spreads rapidly into the surrounding brain tissue [1].GB standard therapy involves surgical resection, followed by radiotherapy and chemotherapy based on temozolomide, followed by second-line therapy based on topoisomerase I and II inhibitors, or anti-angiogenic drugs.The success of chemotherapy is limited by the tumor polyclonality, the intrinsic resistance to most chemotherapeutic drugs and the presence of bloodbrain barrier (BBB) [2][3][4].
Chemotherapy is not efficient to completely eradicate tumor stem cells (SCs) that contribute to initiation, progression and recurrence of GB.Indeed, these cells show a multidrug resistance (MDR) phenotype [5][6] that prevents the intracellular accumulation and efficacy of several antineoplastic drugs.The MDR phenotype of GB SCs is sustained by the high expression of ATP binding cassette (ABC) transporters, such as P-glycoprotein (Pgp/ABCB1), MDR related protein 1 (MRP1/ABCC1), breast cancer resistance protein (BCRP/ABCG2) [6].
Chemotherapy fails against GB also because of the low drug delivery across the BBB, the microvascular endothelium that surrounds brain parenchyma.BBB is characterized by the absence of fenestrations and the presence of tight junctions (TJs) and ABC transporters [7][8].BBB is often disrupted within GB bulk, but it is competent in the "brain-adjacent to tumor" (BAT) area, where isolated GB cells can grow, inducing local tumor recurrence or spreading in other areas if not eradicated by chemotherapy [7].Pgp is abundant on GB SCs and on the luminal side of BBB, and mediates the backward efflux of doxorubicin, taxanes, Vinca alkaloids, teniposide/etoposide, topotecan, methotrexate, imatinib, dasatinib, lapatinib, gefitinib, sorafenib, erlotinib [8].
The presence of Pgp either in GB and BBB represents a double obstacle for the success of chemotherapy.Notwithstanding different approaches to circumvent the Pgp efflux activity in BBB [9][10][11][12][13][14] and GB, in particular in GB SCs [15][16][17], no satisfactory tools have been found.
Our research group has recently developed a library of Pgp ligands, based on the tetrahydroisoquinoline scaffold, a substructure often found in Pgp ligands [18][19].The compounds were designed by functionalizing the phenolic group of an already known Pgp inhibitor [MC70,20] with two types of substituents: 1,2,5-oxadiazole (furazan) moiety linked through alkyl spacers [18], and flexible alkyl chains of various length [19].From this library, we selected 6 compounds with an EC50 for Pgp ranging from 0.60 nM to 54 nM (Table 1), i.e. superimposable with the lastgeneration of Pgp inhibitors [21].-H 690 20  In the present work we investigated whether these compounds overcame the Pgp activity at BBB and GB levels.We used human brain microvascular endothelial cells and GB cells obtained from patients, isolated and propagated as differentiated (adherent cell, AC) or stem cell-enriched (neurospheres, NS) cultures.In isolated BBB and GB, as well as in co-culture systems, we studied the effects of Pgp ligands on the transport and accumulation on doxorubicin, chosen as a prototypical drug that does not cross BBB [12,22] and is ineffective against GB NS [16], being a substrate of Pgp.

Chemistry
Target compounds 1-5 [18] and 6 [19] were prepared as reported, starting from MC70.The latter one was more conveniently prepared according to a straightforward metal-free synthetic route depicted in Scheme 1.Briefly, the commercially available 4'-hydroxybiphenyl-4-carboxylic acid 7 was converted into the corresponding methyl ester by refluxing in methanol with a catalytic amount of concentrated sulfuric acid.The product was reduced to the benzyl alcohol 9 in presence of LiAlH4 at room temperature.Treatment of 9 with 37% HCl at 90 °C afforded the benzyl chloride 10 which was readily reacted with 6,7-dimethoxy-1,2,3,4-tetrahydroisoquinoline hydrochloride to give MC70.

Pgp ligands increase doxorubicin uptake and cytotoxicity in Pgp-positive neurosphere from glioblastoma
We next validated the efficacy of our compounds against primary GB cells of 3 patients.From each tumor, AC and NS were generated (Figure 2A).As previously shown [16], NS had typical stemness properties, such as self-renewal, in vitro clonogenicity and in vivo tumorigenicity.In parallel, NS had high expression of general and neural stemness markers (nestin, CD133, Musashi, SOX2, EGFR, p53) and low expression of differentiation markers (glial fibrillary acidic protein, GFAP; galactocerebroside-C, Gal-C) compared to AC (Supplementary Table 1), suggesting that they represent cultures enriched in GB-derived SCs.As shown in Figure 2B, AC had undetectable levels of Pgp and low levels of MRP1 and BCRP; by contrast, all these ABC transporters were well-detected in the corresponding NS.In keeping with this trend, fluorescence microscope analysis revealed a typical intranuclear localization of doxorubicin in AC, whereas NS had no appreciable red fluorescence, indicating a very low drug uptake (Figure 2C).This difference was confirmed by the quantitative fluorimetric measurement of doxorubicin uptake in AC and NS (Figure 2D).NS, which had a lower intracellular retention of the drug compared to AC, significantly increased doxorubicin accumulation if treated with compounds 1-3.Compounds 4-6 had no effects.
Moreover, none of the compounds increased the drug uptake in AC compared to untreated cells (Figure 2D).
Pgp ligands were not toxic on GB cells: indeed, they did not increase the release of lactate dehydrogenase (LDH; Figure 3A), indicative of cell damage and necrosis [16], they did not activate caspase 3 (Figure 3B), an index of apoptosis, and they did not reduce AC and NS viability (Figure 3C).According to the higher retention of doxorubicin in AC than in NS, the drug increased LDH release and caspase 3 activity, and decreased cell viability in AC but not NS.Compounds 1-3 partially restored doxorubicin's cytotoxic effects in NS.Again the compounds did not enhance the We finally validated the efficacy of our Pgp ligands in co-culture systems: doxorubicin-resistant NS were seeded in the lower chamber of Transwell devices, containing confluent hCMEC/D3 monolayer in the upper chamber.Doxorubicin, alone or co-incubated with compounds, was added in the upper chamber, facing the luminal side of BBB cells.In these conditions, the amount of doxorubicin delivered into NS (Figure 4A) was unable to elicit cell necrosis (Figure 4B) and apoptosis (Figure 4C), nor to decrease NS viability (Figure 4D).The co-incubation with compounds 1-3 significantly increased the amount of doxorubicin delivered to NS (Figure 4A), the release of LDH (Figure 4B) and the activity of caspase 3 (Figure 4C).At longer time-point NS viability was also reduced by the co-incubation of doxorubicin and compounds 1-3 (Figure 4D).
To discriminate whether the effects of the Pgp ligands were solely due to the inhibition of Pgp on BBB, or to their inhibitory effects on Pgp both on BBB and NS (i.e. after crossing the barrier), we added 5 μM doxorubicin, alone or with compound 1-6, in the upper chamber of Transwell devices containing BBB.After 3 h, the medium of the lower chamber was collected: part was added to on NS cultures (Figure 5A), part was used to measure the doxorubicin concentration (Figure 5B).The drug concentration in the media of the lower chamber ranged between 1 and 1.8 μM for the Transwells treated with compounds 1-3, and was significantly higher than in all the other experimental conditions (Figure 5B).The effects elicited by the media derived from these Transwells was compared with the effects produced by medium containing 1 μM doxorubicin.Of note, the intracellular doxorubicin uptake in NS (Figure 5C), the release of LDH (Figure 5D), the activity of caspase-3 (Figure 5E) were higher, the viability of NS was lower (Figure 5F) when NS were treated with media derived from Trasnwells exposed to compounds 1-3 than with medium containing 1 μM doxorubicin, suggesting a possible effects of the compounds on Pgp either in BBB cells or in NS.
Many Pgp inhibitors that achieved excellent efficacy in vitro had failed in pre-clinical and clinical models because of the great toxicity, owing to the high concentrations (i.e.millimolar-micromolar concentrations) needed to inhibit Pgp, that induced heavy side-effects and toxicities [25].Looking for more effective P-glycoprotein inhibitors, it was observed that Pgp-expressing tumor cells retain sensitivity to local anaesthetics, detergents, antimetabolites, alkylating agents, platinum compounds, metal chelators.These findings opened the opportunity to bypass MDR by treating Pgp-expressing cells with non-cross-resistant drugs, exploiting the peculiar sensitivity of resistant cells to these agents (i.e., exploiting the so-called "collateral sensitivity", CS) [26].Despite promising results in vitro, the in vivo safety of these agents, most of which exert cytotoxic effects in cell cultures, is not known.
Targeting ABC efflux transporters with new chemosensitizers is still considered the main strategy to improve drug delivery and overcome MDR [27], but it remains an unmet need.Compared to the first Pgp inhibitors, the latest generation of Pgp inhibitors, such as tariquidar, elacridar or zosuquidar, showed efficacy at lower concentrations (i.e.nanomolar concentrations) and higher specificity for Pgp over the other ABC transporters [21].
Furazan based compounds 1-5 were originally designed following some preliminary results on the activity of some furazan derivatives towards Pgp (unpublished data): in this series, stereo-electronic properties of the substituents on the heterocyclic ring were modulated.Compound 6 belongs to a series of derivatives designed mainly with the aim of verifying the effect of lipophilicity of the substituents on the phenolic group of parent compound.As can be seen from EC50 values, the selected MC70 derivatives proved much more potent and displayed a functional profile different compared to MC70: the latter behaves as a Pgp inhibitor, since it has an apparent permeability ratio < 2, determined in Caco-2 cells monolayer, and does not induce ATP depletion; the new compounds are substrates belonging to a particular class defined as class IIB3, characterized by apparent permeability > 2 and absence of ATP depletion [18,19].Kinetic parameters for these derivatives have not been determined, but, as a cautious consideration, it can be argued that, due to the large internal cavity of Pgp, binding of ligands could be approximately considered only diffusion-limited; in such cases, a typical residence time for nanomolar ligands would be around 1 s [28].
3 out of 6 compounds, namely 1-3, effectively increased the transport of doxorubicin, a virtually BBB-impermeable drug, being a substrate of Pgp [7], across BBB monolayer, when used at 1 nM concentration.Notably, at such concentration they did not reduce BBB cell viability and they did not change expression of other luminal ABC transporters and TJ proteins, nor they modify TEER values.This experimental set demonstrated that at this concentration the compounds do not compromise the integrity and the physiological properties of BBB.Of note, compounds 1-3 strongly inhibited Pgp activity, but slightly activated MRP1 and BCRP, as demonstrated by the increased transport of doxorubicin across Pgp-MDCK monolayer, the decreased transport across MRP1-MDCK and BCRP-MDCK monolayer, by the lack of increase in doxorubicin transport across hCMEC/D3 monolayer treated with MRP1 and BCRP inhibitors compared to untreated cells.These contrasting effects may reduce the efficacy of compounds 1-3 on doxorubicin transport: on the one hand the compounds increased the drug delivery by inhibiting Pgp, on the other hand they decreased the drug transport by stimulating MRP1 and BCRP.The activating effect on MRP1 and BCRP, however, was smaller compared to the inhibitory effect on Pgp, in terms of doxorubicin transport, as demonstrated by the results in MDCK cells selectively expressing Pgp, MRP1 or BCRP: therein, the net effect in hCMEC/D3 cells, where these three transporters were present, was a significant increase of the transport of doxorubicin across BBB monolayer.Compounds 4, 5 and 6 did not affect doxorubicin transport across Pgp-MDCK cells, where MRP1 and BCRP levels were undetectable, but they strongly reduced the drug transport in MRP1 and BCRP-MDCK cells, where Pgp was undetectable.This results suggest that compounds 4, 5 and 6 were likely activators of MRP1 and BCRP, but the activating effect was reversed by selective MRP1 and BCRP inhibitors According to the results obtained with compounds 1-3 (inhibitors of doxorubicin transport by Pgp, slight activators of doxorubicin transport by MRP1 and BCRP) and with compounds 4-6 (neither inhibitors nor activators of doxorubicin transport by Pgp, strong activators of doxorubicin transport by MRP1 and BCRP), we can hypothesize that in hCMEC/D3 cells the main transporter involved in doxorubicin transport is Pgp, while MRP1 and BCRP play an ancillary role.We hypothesize that this could be the reason why we did not detect any effect of compounds 4-6 on doxorubicin delivery across BBB monolayer.We are aware that all the compounds exerted an inhibition on Pgp activity, measured as calcein-acetoxy methylester (AM) transport (Table 1).Given the complex structure of Pgp, containing different drug binding sites and characterized by different affinity for different substrates, it is not surprising that compounds inhibiting the transport of one substrate, have no effect or even opposite effects on the transport of other substrates [29].Since our aim was to obtain a net increase in doxorubicin delivery across BBB to make it an effective anti-GB drug, the most promising compounds in this perspective were 1, 2 and 3.
All the compounds did not change the permeability of high molecular weight (70-kDa dextran) or low molecular weight ([ 14 C]-inulin, [ 14 C]-sucrose, lucifer yellow) compounds.Therefore, we excluded that the effects on doxorubicin permeability was due to loss of TJs integrity of change in paracellular diffusion of the drug across BBB.
Overall, our compounds showed the same properties -i.e.efficacy at low nanomolar concentrations, specificity for Pgp and lack of in vitro toxicities on not-transformed cells -of the latest generation of Pgp inhibitors or tracers, under investigation in preclinical models and clinical trials [10,14,21].
The selectivity for Pgp was demonstrated also in the experiments performed on AC and NS generated from patients GB.While AC where low Pgp-expressing cells and retained high amount of doxorubicin, NS were chosen as a prototypical model of highly Pgp-expressing GB-derived SCs (16).We recognize that NS do not reproduce the tissue organization and cell polarity occurring in vivo in the microvascular niches, where endothelial cells, pericytes, astroglial and microglial cells, differentiated and GB SCs, actively proliferating and apoptotic/necrotic GB cells were present and interconnected.However, NS have been proven to be a reliable tool to measure the chemosensitizing efficacy of Pgp-reversing agents in GB and BBB-GB co-cultures (12,13,16).We thus compared the effects of our compounds on AC and NS, alone and growing under hCMEC/D3 monolayer.The compounds did not exert any additive effects to doxorubicin in AC, where Pgp was undetectable.In AC, doxorubicin likely reached its maximal accumulation and exerted a broadest spectrum of cytotoxic effects, including cell necrosis, apoptosis and reduced viability.By contrast, the presence of Pgp in NS limited the retention and cytotoxic efficacy of doxorubicin.In NS compounds 1-3 restored the drug's accumulation to the same levels of AC, suggesting that they were able to inhibit the drug efflux via Pgp.As already observed in BBB, the compounds alone were not toxic against GB cells.Only the association of compounds and doxorubicin induced cytotoxicity, suggesting that the compounds acted as chemosensitizers agents, rescuing the efficacy of doxorubicin.
We could not directly demonstrate the transport of compounds 1-3 across BBB, since nanomolar concentrations were below the detection limit of high-pressure liquid chromatography (HPLC) device and micromolar concentrations, that were well-measured by HPLC, reduced cell viability of BBB, leading to suppose that at this concentration BBB is damaged and does not represent a physiologically competent BBB.However, in co-culture experiments, when we exposed the luminal side of BBB with the association of compounds 1-3 and doxorubicin, we observed an increased delivery and cytotoxicity of doxorubicin in NS growing under competent BBB.In BBB monolayer, Pgp is expressed on the luminal side [30], i.e. facing doxorubicin.The increased efficacy of doxorubicin may be due to a simple increase in the drug delivery across BBB, consequent to the inhibition of Pgp present on BBB.However, also NS express Pgp.According to the low intracellular retention of doxorubicin within NS, it is likely that Pgp is expressed on the outer surface of the spheres, preventing the intracellular accumulation of the drug.
Therein, doxorubicin delivered across BBB could not be efficiently accumulated within NS and exert its cytotoxic effects if the Pgp present in NS was active.The results of the experiments with medium collected from the lower chamber of Transwell devices incubated with compounds 1-3 suggest that the compounds inhibit at the same time Pgp on BBB and NS cells: this could be an indirect evidence of the delivery of compounds 1-3 across BBB.We are currently testing this hypothesis evaluating the pharmacokinetic profile of the compounds systemically administered in mice bearing orthotopically implanted GB.
Notwithstanding the excellent anti-tumor activity against GB in in vitro [31], doxorubicin is not a first-choice drug in GB treatment because of the very low delivery across BBB.However, since GB AC are sensitive to doxorubicin [31], several strategies to improve the in vivo efficacy of doxorubicin are under intensive investigation in pre-clinical and clinical settings [32][33][34].Our compounds represent a step forward this direction, since they transformed the BBB-impermeable doxorubicin into a drug with a good BBB permeability and efficacy against Pgp-positive NS.
Moreover, temozolomide, topoisomerase I and II inhibitors, that are the current first-and secondline therapy in GB, are substrates of Pgp [11,35].Therein, this work may open the way to preclinical studies combining our Pgp ligands and these drugs, in order to circumvent Pgp-mediated chemoresistance and improve the efficacy of chemotherapy against GB.

Materials and methods
Chemicals.The plasticware for cell cultures was obtained from Falcon (Becton Dickinson, Franklin Lakes, NJ).The electrophoresis reagents were obtained from Bio-Rad Laboratories (Hercules, CA).
The protein content of cell lysates was assessed with the BCA kit from Sigma Chemicals Co. (St. Louis, MO).Unless specified otherwise, all reagents were purchased from Sigma Chemicals Co.  [4'-(6,7-Dimethoxy-3,4-dihydro-1H-isoquinolin-2-ylmethyl)biphenyl-4-ol] (MC70).Compound 9 was suspended in 37% HCl and the mixture was stirred at 90 °C for 2 hours.The suspension was cooled in an ice bath, diluted with ice-cold water and filtered under reduced pressure to give 10 as a white solid, which, after being dried over KOH, was dissolved in acetonitrile.To the solution 6,7dimethoxy-1,2,3,4-tetrahydroisoquinoline hydrochloride (1.3 eq) and 4-methylmorpholine (2.3 eq) were added and the mixture was refluxed for 6 hours.The solvent was then evaporated under reduced pressure, the residue was taken up with water and extracted with ethyl acetate.The organic extracts were washed with brine, dried over Na2SO4, filtered and evaporated under reduced pressure.The crude product was purified on silica gel column, eluting with petroleum ether / acetone 70 / 30, to give the title product identical to an authentic sample [19].Yield 56% over two steps.

Synthesis and characterization of compounds.
The Pgp activity in the presence of compounds was evaluated by the Calcein-AM assay and the bioluminescent ATP assay, as described previously [18,19].Technology, Milano, Italy), were measured as previously described [12,[22][23][24] in BBB cells in the absence of GB cells.The TEER value was measured using a Voltohmetro Millicell-ERS (Millipore, Billerica, MA), according to the manufacturer instructions.The mean TEER value of the plastic insert in the absence of cells was 26.73 Ω cm 2 (n=8).This value was subtracted from each value obtained in the presence of the cells.

BBB cells, TEER
For transport assays, after 7 days of culture, the culture medium was replaced in both chambers.2 μM 70 kDa dextran-FITC, 2 μCi/ml [ 14 C]-sucrose, 2 μCi/ml [ 14 C]-inulin, 100 μM lucifer yellow were added to the upper chamber of Transwell.After 3 h the medium in the lower chamber was collected.The amount of [ 14 C]-sucrose and [ 14 C]-inulin was measured using a Tri-Carb Liquid Scintillation Analyzer (PerkinElmer).Radioactivity was converted in nmol/cm 2 , using a calibration curve previously prepared.The radioactivity of the medium alone, considered as a blank, was subtracted from each measure.
The amount of 70 kDa dextran-FITC and lucifer yellow was measured fluorimetrically, using a Synergy HT microplate spectrofluorimeter (Bio-Tek Instruments, Winooski, VT).Excitation and emission wavelengths were: 494 nm and 518 nm (70 kDa dextran-FITC); 430 nm and 540 nm (lucifer yellow).Fluorescence was converted in nmol/cm 2 , using a calibration curve previously set.
The autofluorescence of the medium, considered as a blank, was subtracted from each measure.The permeability coefficients were calculated according to [36].
MDCK, Pgp-MDCK, MRP1-MDCK and BCRP-MDCK cells were a kind gift of Dr. Maria Alessandra Contino (Department of Pharmacy, University of Bari, Italy).Culturing and seeding conditions for doxorubicin transport assay were carried out as reported in [37].

3 "Figure 1 .
Figure 1.Effects of Pgp ligands of BBB viability and integrity

Figure 2 .
Figure 2. Effects of Pgp ligands on doxorubicin retention of glioblastoma cells

Figure 3 .
Figure 3. Effects of Pgp ligands on doxorubicin cytotoxicity in glioblastoma cells