Property-based design of a glucosylceramide synthase inhibitor that reduces glucosylceramide in the brain

      Synthesis inhibition is the basis for the treatment of type 1 Gaucher disease by the glucosylceramide synthase (GCS) inhibitor eliglustat tartrate. However, the extended use of eliglustat and related compounds for the treatment of glycosphingolipid storage diseases with CNS manifestations is limited by the lack of brain penetration of this drug. Property modeling around the D-threo-1-phenyl-2-decanoylamino-3-morpholino-propanol (PDMP) pharmacophore was employed in a search for compounds of comparable activity against the GCS but lacking P-glycoprotein (MDR1) recognition. Modifications of the carboxamide N-acyl group were made to lower total polar surface area and rotatable bond number. Compounds were screened for inhibition of GCS in crude enzyme and whole cell assays and for MDR1 substrate recognition. One analog, 2-(2,3-dihydro-1H-inden-2-yl)-N-((1R,2R)-1-(2,3-dihydrobenzo[b][1,4]dioxin-6-yl)-1-hydroxy-3-(pyrrolidin-1-yl)propan-2-yl)acetamide (CCG-203586), was identified that inhibited GCS at low nanomolar concentrations with little to no apparent recognition by MDR1. Intraperitoneal administration of this compound to mice for 3 days resulted in a significant dose dependent decrease in brain glucosylceramide content, an effect not seen in mice dosed in parallel with eliglustat tartrate.
      Two general strategies for the treatment of lysosomal storage diseases exist. The first strategy includes the replacement or restoration of the defective or absent catabolizing enzyme (e.g., the infusion of recombinant enzyme, chaperone therapy, bone marrow transplantation, or gene therapy) (
      • Barton N.W.
      • Brady R.O.
      • Dambrosia J.M.
      • Di Bisceglie A.M.
      • Doppelt S.H.
      • Hill S.C.
      • Mankin H.J.
      • Murray G.J.
      • Parker R.I.
      • Argoff C.E.
      • et al.
      Replacement therapy for inherited enzyme deficiency–macrophage-targeted glucocerebrosidase for Gaucher's disease.
      ). Enzyme replacement therapy is clinically approved for lysosomal storage diseases with peripheral manifestations but is limited by the absence of distribution of infused recombinant enzyme into the central nervous system (CNS) and by the frequent development of auto-antibodies to the protein in patients who carry null mutations. The second strategy involves synthesis inhibition therapy (
      • Radin N.S.
      Treatment of Gaucher disease with an enzyme inhibitor.
      ). Synthesis inhibition is a more recent therapeutic approach and has focused on identifying small molecule inhibitors of GCS. Two classes of these inhibitors have been described, including imino sugars and analogs of D-threo-1-phenyl-2-decanoylamino-3-morpholino-propanol (PDMP) (
      • Shayman J.A.
      • Lee L.
      • Abe A.
      • Shu L.
      Inhibitors of glucosylceramide synthase.
      ). The imino sugar N-butyldeoxynojirimycin (NBDNJ) is limited by its micromolar level inhibitory activity and limited specificity against the synthase. The latter trait is associated with a high level of untoward effects resulting from secondary sites of action unrelated to glycolipid synthesis inhibition. These effects most notably include diarrhea, weight loss, and tremor, and limit its approved use in the United States (
      • Weinreb N.J.
      • Barranger J.A.
      • Charrow J.
      • Grabowski G.A.
      • Mankin H.J.
      • Mistry P.
      Guidance on the use of miglustat for treating patients with type 1 Gaucher disease.
      ). However, one distinct advantage of NBDNJ over the PDMP-based homologs reported to date is its ability to distribute into the CNS.
      The active lead compound that is currently in clinical trials and is structurally related to PDMP is N-((1R,2R)-1-(2,3-dihydrobenzo[b][1,4]dioxin-6-yl)-1-hydroxy-3-(pyrrolidin-1-yl)propan-2-yl)octanamide (Genz-112638, eliglustat tartrate 3a) (
      • Shayman J.A.
      Eliglustat tartrate.
      ). Recent phase 2 clinical trials using this drug for type 1 Gaucher disease demonstrate a level of efficacy equal to or greater than that seen for recombinant β-glucocerebrosidase as evidenced by reversal of spleen and liver enlargement, correction of anemia, and improvements in thrombocytopenia and bone density (
      • Lukina E.
      • Watman N.
      • Arreguin E.A.
      • Banikazemi M.
      • Dragosky M.
      • Iastrebner M.
      • Rosenbaum H.
      • Phillips M.
      • Pastores G.M.
      • Rosenthal D.I.
      • et al.
      A phase 2 study of eliglustat tartrate (Genz-112638), an oral substrate reduction therapy for Gaucher disease type 1.
      ). Phase 3 trials with eliglustat tartrate are currently in progress. Experimental data also support a potential role for eliglustat tartrate in the treatment of Fabry disease, another lysosomal storage disease with peripheral manifestations (
      • Abe A.
      • Gregory S.
      • Lee L.
      • Killen P.D.
      • Brady R.O.
      • Kulkarni A.
      • Shayman J.A.
      Reduction of globotriaosylceramide in Fabry disease mice by substrate deprivation.
      ). In theory, substrate inhibition might also work for six other lysosomal storage diseases with CNS involvement that include early and late onset Tay-Sachs disease, Sandhoff disease, GM1 gangliosidosis, and types 2 and 3 Gaucher disease. Indeed, an experimental model of genetic epistasis demonstrates markedly improved survival in a mouse model of Sandhoff disease that also lack GM2 synthase (
      • Liu Y.
      • Wada R.
      • Kawai H.
      • Sango K.
      • Deng C.
      • Tai T.
      • McDonald M.P.
      • Araujo K.
      • Crawley J.N.
      • Bierfreund U.
      • et al.
      A genetic model of substrate deprivation therapy for a glycosphingolipid storage disorder.
      ). Unfortunately, drug distribution studies indicate that eliglustat tartrate is not transported across the blood-brain barrier (BBB) (
      • Shayman J.A.
      Eliglustat tartrate.
      ). A possible basis for the poor brain distribution of eliglustat tartrate may be that the drug is a substrate for the p-glycoprotein (MDR1) transporter. MDR1 is highly expressed at the BBB where it markedly limits the brain penetration of many xenobiotics (
      • Miller D.S.
      Regulation of P-glycoprotein and other ABC drug transporters at the blood-brain barrier.
      ). We confirmed this possibility and then embarked on the design of new glycolipid synthesis inhibitors using the active pharmacophore of PDMP with a goal of retaining activity against GCS but eliminating substrate specificity for the MDR1 protein. In this paper, we report on the design and synthesis of a series of PDMP analogs that are active against GCS, the evaluation of these compounds in in vitro models of MDR1 transport, and the identification of a novel analog that inhibits glucosylceramide (GlcCer) synthesis in both brain and peripheral organs.

      METHODS

       Materials

      N-((1R,2R)-1-(2,3-dihydrobenzo-[b][1,4]dioxin-6-yl)-1-hydroxy-3-(pyrrolidin-1-yl)propan-2-yl)octanamide (eliglustat tartrate, 3a) was provided by Genzyme Corporation. [3H]Vinblastine and [14C]mannitol were both purchased from American Radiolabeled Chemicals (St Louis, MO).

       Inhibitor synthesis

      The preparation of analogs of 3a is summarized in Fig. 1. The 3,4-ethylenedioxy (EDO) amino alcohol intermediate 1 was prepared as previously described (
      • Hirth B.H.
      • Siegel C.
      Synthesis of UDP-glucose: N-acylsphingosine glucosyltransferase inhibitors. Patent no. 6,855,830.
      ). Hydrogenolysis of 1 afforded common intermediate amine 2. N-acylation of 2 to afford analogs 3b-i (Table 2) was effected with various acid chlorides or carboxylic acids under standard EDC coupling conditions. Monomethyl amide 3j was prepared in three steps from 1 by Eschweiler-Clarke mono-N-methylation to afford 4, hydrogenolysis of the benzyl group to afford N-methylamine 5, and N-acylation with hydrocinammic acid/EDC. The chemical structures, experimental procedures, and spectral characterization for all final analogs are described fully in the supplementary data.
      Figure thumbnail gr1
      Fig. 1General scheme for the synthesis of new inhibitors. The structure of compound 3h (CCG-203586), 2-(2,3-dihydro-1H-inden-2-yl)-N-((1R,2R)-1-(2,3-dihydrobenzo[b][1,4]dioxin-6-yl)-1-hydroxy-3-(pyrrolidin-1-yl)propan-2-yl)acetamide, is indicated in the rectangular box.
      TABLE 2Calculated properties and biological activity of new analogs
      CmpdR1R2MWTPSA (Å2)SLogPRotBCrude Enzyme
      Inhibition of GlcCer synthesis in disrupted wild-type MDCK cell preparations (mean of n≥2 experiments).
      IC50 (nM)
      WT-MDCK
      Inhibition of GlcCer production in whole wild-type MDCKII cells.
      IC50 (nM)
      MDR-MDCK
      Inhibition of GlcCer production in MDCKII cells stably expressing human MDR1 (obtained from The Netherlands Cancer Institute).
      IC50 (nM)
      MDR/WT
      Ratio of MDR1-MDCKII IC50 divided by WT-MDCKII IC50. R1 and R2 denote substitutions shown in Fig. 1.
      3aH-C7H15404.571.03.531214913.731.72.3
      3bH-OCH2Ph412.580.33.2491447.347.16.5
      3cH-CH2CH2Ph410.571.02.80913423.586.73.7
      3dH-CH2Ph396.571.02.418>2000NDNDND
      3eH-CH2CH2CH2Ph424.571.03.19108025414405.7
      3fH- E-CH = CHPh408.571.02.88837980.73894.8
      3gH-trans-2-Ph-cyclopropyl422.571.02.9789027.378.12.9
      3hH-CH2-indan-2-yl436.571.02.9782715.324.31.6
      3iH-indan-2-yl422.571.02.587107151321021
      3jCH3-CH2CH2Ph424.562.23.14915859.4125021
      a Inhibition of GlcCer synthesis in disrupted wild-type MDCK cell preparations (mean of n≥2 experiments).
      b Inhibition of GlcCer production in whole wild-type MDCKII cells.
      c Inhibition of GlcCer production in MDCKII cells stably expressing human MDR1 (obtained from The Netherlands Cancer Institute).
      d Ratio of MDR1-MDCKII IC50 divided by WT-MDCKII IC50. R1 and R2 denote substitutions shown in Fig. 1.

       GCS activity

      Enzyme activity was measured as described previously (
      • Shayman J.A.
      • Abe A.
      Glucosylceramide synthase: assay and properties.
      ). Madin-Darby canine kidney (MDCK) cell homogenates (120 μg of protein) were incubated with uridine diphosphate-[3H]glucose (100,000 cpm) and liposomes consisting of 85 μg of octanoylsphingosine, 570 μg of dioleoylphosphatidylcholine, and 100 μg of sodium sulfatide in a 200 μl reaction mixture and kept for 1 h at 37°C. PDMP derivatives dissolved in dimethyl sulfoxide (final concentration <1%, which did not affect enzyme activity) were dispersed into the reaction mixture after adding the liposomes.

       GCS inhibition in MDCKII cells

      Parental wild type-MDCKII (WT-MDCKII) cells and MDCKII cells retrovirally transduced with human MDR-1 cDNA were obtained from the Netherlands Cancer Institute. Both cell lines were routinely maintained in medium consisting of Opti-MEM/F12 (1:1), 5% FBS, 100U/ml of penicillin, 100 µg/ml streptomycin, and 200 mM L-glutamine. MDCKII cells were newly thawed from frozen ampules every 2 months. Protein levels of MDR1 in MDR1-MDCKII cells were measured monthly, and MDR1-MDCKII cell passages were immediately terminated when a reduction of MDR1 levels was observed by Western blot using anti-human MDR1 monoclonal antibody (Abcam C219).
      Stock solutions of water-insoluble glycosphingolipid inhibitors (100 mM) were prepared by dissolving each inhibitor into 100% ethanol as previously described (
      • Shayman J.A.
      • Lee L.
      • Abe A.
      • Shu L.
      Inhibitors of glucosylceramide synthase.
      ). The inhibitor-ethanol solutions were then diluted 50× into 2 mM delipidated BSA-PBS solution to make water-soluble glycosphingolipid inhibitor-BSA complexes. The inhibitor-BSA complexes were sterile-filtered and stored at −20°C. Prior to use, portions of the inhibitor-BSA complexes were further diluted with Opti-F12 to make treatment solutions. Equal amounts of BSA and ethanol were added into the control cultures. WT and MDR1-MDCKII cells (5 × 105) were seeded into 10 cm culture dishes containing 10 ml of Opti-F12 with 5% FBS. After 24 h, the medium was replaced with fresh serum-free Opti-F12 medium, and cells were exposed to candidate GCS inhibitors at concentrations of 0, 1, 3, 10, 30, 100, and 300 nM for 24 h.

       Cell lipid analysis

      Following inhibitor treatment, whole cellular lipids of wild-type and MDR1-MDCKII cells were extracted as previously described in detail (
      • Shu L.
      • Shayman J.A.
      Src kinase mediates the regulation of phospholipase C-gamma activity by glycosphingolipids.
      ). Briefly, cells were washed with ice-cold PBS, fixed by methanol and collected with rubber scraper. Chloroform was then added to yield a theoretical ratio of chloroform-methanol-water at 1:2:0.8 (v/v/v) to form a mono-phase. Cell debris and proteins were removed by centrifugation at 2200 g for 30 min. The supernatants were portioned by adding chloroform and 0.9% NaCl. The lower organic phases containing neutral glycosphingolipids lipids were washed with methanol and 0.9% NaCl and subjected to base- and acid-hydrolysis (
      • Shu L.
      • Shayman J.A.
      Src kinase mediates the regulation of phospholipase C-gamma activity by glycosphingolipids.
      ). A portion of purified glycosphingolipids normalized to 100 nmol of total phospholipids was analyzed by high-performance TLC. The TLC separations were processed twice. The plate pretreated with 1% sodium borate was first developed in a solvent system consisting of chloroform-methanol (98:2, v/v). After air drying, the plate was then developed in a solvent system containing chloroform-methanol-water (70:30:4, v/v/v). The levels of glucosylceramide were detected by charring with 8% cupric sulfate in 8% phosphoric acid and quantified by densitometric scanning using ImageJ, NIH Image. Image data was analyzed, and the IC50 of each inhibitor was calculated using GraphPad Prism (version 5.03).

       Vinblastine transport

      MDR1- and WT-MDCKII cells were grown to confluence on Transwell filters (12-well plates) in DMEM + 10% FBS. The media was then replaced with fresh DMEM and [3H]vinblastine (0.5 µCi/ml; 10 µM final unlabeled vinblastine concentration) and [14C]mannitol (0.25 µCi/ml; an extracellular space marker) was added to the apical chamber. Uptake and, in a subset of experiments, transepithelial flux, was measured over 2 h at 37°C. At that time, the basal chamber was sampled and uptake was stopped by washing each side of the membrane three times with ice-cold PBS. Vinblastine uptake into the cells, after correction for any remaining adherent extracellular contamination and transepithelial flux, was calculated as described previously (
      • Shu C.
      • Shen H.
      • Teuscher N.S.
      • Lorenzi P.J.
      • Keep R.F.
      • Smith D.E.
      Role of PEPT2 in peptide/mimetic trafficking at the blood-cerebrospinal fluid barrier: studies in rat choroid plexus epithelial cells in primary culture.
      ). To investigate the effect of experimental drugs on MDR-mediated transport, those compounds were added to the apical chamber (1-100 µM) during vinblastine uptake and the uptake and trasepithelial flux results expressed as a percentage of that with vehicle alone.

       In vivo studies

      C57BL/6 mice were maintained on regular chow in specific-pathogen-free facilities. All animal studies were performed under the review of the University of Michigan Committee on the Use and Care of Animals and conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Injection solutions were prepared from inhibitor-ethanol stock solution (100 mM). A portion of the stock solution was evaporated under a stream of N2 gas. Dried eliglustat tartrate was directly dissolved into 1X PBS. Compound 3h was dissolved with 250 µl of water plus 13.7 µl of 0.5 N HCl by hand shaking and gentle vortexing resulting in 1 or 3 mg/ml of solution. The acidic solution was neutralized by mixing 100 µl of 10× PBS with 636.3 µl of water to bring the total volume to 1 ml. Inhibitor solutions were sterilized by passage through a 0.2 µM filter. Inhibitor recovery after filtration was confirmed by UV spectrometry and exceeded 99%. For control injections, PBS containing the same amount of HCl was used. Inhibitors were given to 6- to 8-week-old female or male C57BL/6 mice by intraperitoneal injection volume at 1% of body weight.

       Mouse tissue lipid analysis

      Lipid extractions of liver, kidney, and brain were performed as previously described (
      • Abe A.
      • Gregory S.
      • Lee L.
      • Killen P.D.
      • Brady R.O.
      • Kulkarni A.
      • Shayman J.A.
      Reduction of globotriaosylceramide in Fabry disease mice by substrate deprivation.
      ). Briefly, frozen liver (∼0.5 g), two kidneys (∼0.3 g), and whole brain (∼0.4 g) were individually homogenized in sucrose buffer (250 mM sucrose, pH 7.4, 10 mM Hepes and 1 mM EDTA), at 0.2 g tissue/1 ml of sucrose buffer, with a Tri-R homogenizer. Each 0.8 ml of homogenate was mixed with 2 ml of methanol and 1 ml of chloroform, bath sonicated for 1 min, and incubated at room temperature for 1 h. Tissue debris were removed by centrifugation at 2,400 g for 30 min. The pellets were reextracted by mixing with 1 ml of methanol, 0.5 ml of chloroform, and 0.4 ml of 0.9% NaCl (chloroform-methanol-0.9% NaCl, 1:2:0.8), incubated at room temperature for 1 h, and centrifuged at 2,400 g for another 30 min. Two extracts were combined and mixed with 4.5 ml of chloroform and 1.2 ml of 0.9% NaCl (chloroform-methanol-0.9% NaCl, 2:1:0.8). After centrifugation at 800 g for 5 min, the lower layer was washed with 3 ml of methanol and 2.4 ml of 0.9% NaCl. A second washing was carried with 3 ml of methanol, 2 ml of water, and 0.4 ml of 0.9% NaCl followed by a 5 min centrifugation at 800 g. The resultant lower phase was collected and dried under a stream of N2 gas.
      The analysis of neutral glycosphingolipids from mouse liver, kidney, and brain was processed after alkaline methanolysis. Kidney lipids were incubated with 2 ml of chloroform and 1 ml of 0.21N NaOH in methanol for 2 h (kidney) or 7.5 h (liver and brain) at room temperature. The lipid extract was normalized to 0.5 µmol of total phospholipid phosphate (liver and kidney) or 2 µmol of total phospholipid phosphate (brain) for high performance TLC analysis. After alkaline methanolysis, the brain lipids were passed through a silica gel column (
      • Abe A.
      • Gregory S.
      • Lee L.
      • Killen P.D.
      • Brady R.O.
      • Kulkarni A.
      • Shayman J.A.
      Reduction of globotriaosylceramide in Fabry disease mice by substrate deprivation.
      ). Borate-impregnated TLC plates were developed in a two solvent system. Plates were first developed in chloroform-methanol (98:2, v/v). The plates loaded with kidney and liver lipids were then developed in chloroform-methanol-water (64:24:4, v/v/v), and brain lipids were further separated in chloroform-methanol-water (60:30:6, v/v/v). GlcCer levels were quantified by comparison to known standards.

      RESULTS

      The GlcCer content of WT- and MDR1-MDCKII cells was measured as a function of inhibitor concentration to confirm that eliglustat tartrate (3a) was a substrate for the MDR1 transporter (Fig. 2A). A significant rightward shift in the concentration response curve was observed for the reduction in GlcCer levels in the MDR1 expressing cells. The IC50s for the WT and MDR1 expressing cells were 13.7 and 31.7 nM, respectively. The interaction of eliglustat with MDR1 was further examined by measuring [3H]vinblastine (an MDR substrate) transport in MDR1- MDCKII cells. Eliglustat produced a dose-dependent increase in [3H]vinblastine uptake similar to that found for verapamil, a known MDR1 inhibitor (Fig. 2B). This increase in uptake into MDR1-MDCKII cells with eliglustat was also associated with an increase in transepithelial flux (to 280 ± 23 and 558 ± 28% of vehicle control at 30 and 100 μM, respectively). In comparison, 30 μM verapamil increased transepithelial flux of [3H]vinblastine to 490 ± 26% of that in vehicle-treated cells.
      Figure thumbnail gr2
      Fig. 2A: Concentration dependent change in GlcCer content of WT-MDCKII and MDR-MDCKII cells in response to eliglustat tartrate. MDCK cells were cultured in the presence of eliglustat tartrate as described in the Methods section. After exposure to the inhibitor for 24 h, the cells were harvested and the lipids extracted and analyzed also as described in the Methods section. The data represent the mean ± SD. (n = 4). B. Effects of different concentrations of Genz 112368 (3a; eliglustat) and verapamil (a known MDR1 inhibitor) on the uptake of [3H]vinblastine into MDR1-MDCKII cells. Values are expressed as a percent of vehicle-treated cells and are given as means ± SE (n = 3). With vehicle treatment, [3H]vinblastine uptake into MDR1-MDCKII cells was 13.6 ± 1.7% of that in WT-MDCKII cells.
      The design of new PDMP analogs was based on attenuating significant differences between key physical chemical properties of 3a (eliglustat) and successful CNS drugs. Table 1 summarizes a comparison of selected computed properties of 3a and known CNS and nonCNS drugs. Data was derived from structures for 1,213 approved drugs as reported in DrugBank (
      • Wishart D.S.
      • Knox C.
      • Guo A.C.
      • Cheng D.
      • Shrivastava S.
      • Tzur D.
      • Gautam B.
      • Hassanali M.
      DrugBank: a knowledgebase for drugs, drug actions and drug targets.
      ). A number of physical chemical properties were calculated for the 198 CNS drugs and the remaining 1,015 nonCNS drugs using the Molecular Operating Environment software (MOE, version 2008.10, Chemical Computing Group Inc., Montreal, Quebec, Canada). Mean and SD of the computed properties are tabulated in the table for the two groups of drugs.
      TABLE 1Calculated physical properties of eliglustat tartrate (3a) versus CNS drugs
      Computed Property
      Values calculated using The Molecular Operating Environment (MOE), version 2008.10, Chemical Computing Group Inc., Montreal, Quebec, Canada.
      Non-CNS Drugs
      Mean values and standard deviations derived from 198 approved CNS drugs and 1015 approved nonCNS drugs (DrugBank) (14).
      CNS Drugs
      Mean values and standard deviations derived from 198 approved CNS drugs and 1015 approved nonCNS drugs (DrugBank) (14).
      3a
      MW383 ± 210288 ± 88405
      SLogP2.1 ± 2.92.8 ± 1.43.5
      TPSA (Å)98 ± 8446 ± 2671
      HBA4.8 ± 4.12.6 ± 1.35
      HBD2.8 ± 3.31.0 ± 0.92
      RotB6.2 ± 5.54.0 ± 2.612
      a Values calculated using The Molecular Operating Environment (MOE), version 2008.10, Chemical Computing Group Inc., Montreal, Quebec, Canada.
      b Mean values and standard deviations derived from 198 approved CNS drugs and 1015 approved nonCNS drugs (DrugBank) (
      • Wishart D.S.
      • Knox C.
      • Guo A.C.
      • Cheng D.
      • Shrivastava S.
      • Tzur D.
      • Gautam B.
      • Hassanali M.
      DrugBank: a knowledgebase for drugs, drug actions and drug targets.
      ).
      A comparison of the computed property values for 3a relative to the mean values for the approved CNS drugs (Table 1) highlights significant divergences (≥1 SD) in molecular weight (MW), topological polar surface area (TPSA), number of hydrogen bond acceptors (HBA), number of hydrogen bond donors (HBD) and rotatable bonds (RotB). Within the constraints of the established pharmacophore for 3a (which requires maintenance of the free OH, the tertiary amine, and the absolute relative stereochemistry of the two chiral centers) we elected to focus on reducing the number of rotatable bonds and TPSA (Table 2). Rotatable bond number was most easily reduced in the aliphatic amide side chain by incorporation of a phenyl group (3b3i), whereas TPSA was reduced by methylating the amide (3j).
      The activity of the new analogs at inhibiting GlcCer production in the disrupted cell assay is summarized in Table 2. Benzyl carbamate analog 3b retained the enzyme inhibitory activity of 3a, confirming a similar observation by Jimbo et al. (
      • Jimbo M.
      • Yamagishi K.
      • Yamaki T.
      • Nunomura K.
      • Kabayama K.
      • Igarashi Y.
      • Inokuchi J.I.
      Development of a new inhibitor of glucosylceramide synthase.
      ) in a related template. The corresponding 3-propionamide 3c, lower in TPSA than 3b, was equally potent. Adjusting the length of the tether to the phenyl group revealed that a 3-carbon tether (3e) was optimal (IC50= 80 nM). Conformational restriction of the tether to further reduce the number of rotatable bonds was explored with analogs 3f3i. Among these, 2-indanylmethyl analog 3h displayed a striking improvement in activity (IC50= 27 nM) and, to our knowledge, represents the most potent GCS inhibitor reported to date.
      The activity of the compounds at inhibiting GlcCer production in whole WT-MDCKII cells is also summarized in Table 2. Eliglustat (3a) was almost 10-fold more potent in intact cells than disrupted, possibly a result of intracellular accumulation. This phenomenon, originally reported for the D-threo-ethylenedioxyphenyl homologs of PDMP, was also observed to varying degrees with most of the other analogs (
      • Lee L.
      • Abe A.
      • Shayman J.A.
      Improved inhibitors of glucosylceramide synthase.
      ). We found that the ratio of broken cell IC50 divided by whole cell IC50 is weakly correlated negatively with MW (R2= 0.58) and positively with SlogP (R2= 0.50), consistent with well established medicinal chemistry principles that decreasing MW and increasing lipophilicity each improve passive permeability into cells (
      • Lipinski C.A.
      • Lombardo F.
      • Dominy B.W.
      • Feeney P.J.
      Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings.
      ).
      The MDR-MDCKII cell line was used to help predict BBB permeability because it expresses an efflux transporter that is physiologically relevant to the brain endothelium (
      • Garberg P.
      • Ball M.
      • Borg N.
      • Cecchelli R.
      • Fenart L.
      • Hurst R.D.
      • Lindmark T.
      • Mabondzo A.
      • Nilsson J.E.
      • Raub T.J.
      • et al.
      In vitro models for the blood-brain barrier.
      ,
      • Wang Q.
      • Rager J.D.
      • Weinstein K.
      • Kardos P.S.
      • Dobson G.L.
      • Li J.
      • Hidalgo I.J.
      Evaluation of the MDR-MDCK cell line as a permeability screen for the blood-brain barrier.
      ). We assumed that IC50s for inhibition of GlcCer production in these cells would correlate directly with intracellular drug levels, thereby providing a convenient and sensitive estimate of susceptibility to MDR-mediated efflux when compared with IC50s in WT-MDCKII cells. Results are included in Table 2. Significantly, the ratios of MDR-MDCKII IC50 divided by WT-MDCKII IC50 (MDR/WT) varied widely among the group, presumably reflecting a wide range of affinities for MDR1.
      Although several replacements for the C8 acyl group (R2) of eliglustat were associated with the retention of nanomolar range inhibition of GlcCer in the WT cell line, only the –CH2-indan-2-yl homolog 3h displayed a comparable IC50 in the MDR1-MDCKII cells when compared with WT (<2-fold increase in IC50). Remarkably, deletion of a single endocyclic or exocyclic methylene group from the indane (3e or 3i, respectively) was associated with significant loss of activity in the MDR1-MDCKII line. A comparison of the dose-dependent changes in GlcCer levels for these three homologs is shown in Fig. 3A. Representative TLC plates of the extracted lipids from the WT-MDCKII and MDR1-MDCKII cells are shown in Fig. 3B. Consistent with many reports on the association of cellular GlcCer content and MDR1 expression, the MDR1-MDCKII cells had significantly higher concentrations of glucosylceramide at baseline (
      • Gouaze-Andersson V.
      • Cabot M.C.
      Glycosphingolipids and drug resistance.
      ). However, the percent decrement in GlcCer in the presence of increasing concentrations of 3h was comparable for the two cell lines. The decrease in GlcCer occurred in concert with a fall in lactosylceramide, but in the absence of any apparent change in sphingomyelin and galactosylceramide, consistent with a specific effect of 3h on GlcCer synthase. This observation is consistent with previous reports for eliglustat as well as the other PDMP-based analogs.
      Figure thumbnail gr3a
      Fig. 3Correlation between cellular GCS inhibition and inhibition of [3H]vinblastine uptake in the presence of MDR1. A: A comparison of the dose dependent changes in GlcCer for three homologs of PDMP. B: Representative thin layer chromatography plates of WT-MDCKII and MDR1-MDCKII cells treated with 3h. SM, sphimgomyelin; GalCer, galactosylceramide. C: Effects of different concentrations of three novel GCS inhibitors on the uptake of [3H]vinblastine into MDR1-MDCKII cells. Values are expressed as a percent of vehicle-treated cells and are given as means ± SE (n = 3). 3h only differs from 3i by a single methylene spacer and from 3e by a single bridging CH2, but these changes result in marked changes in the interaction with MDR. D: Effects of different GCS inhibitors on two parameters, [3H]vinblastine uptake into MDR1-MDCKII cells expressed as a fold change between 30 μM of the inhibitor and vehicle (x axis) and the fold change in GCS IC50 between MDR1- and WT-MDCKII cells (y axis). For most compounds, there was a good correlation between the two parameters (the line represents equal changes). There was a significant correlation between both parameters although two compounds, 3i and 3j, showed noticeably greater effects on the enzyme IC50 than 3f and 3e despite similar magnitude effects on [3H]vinblastine uptake. As assessed by both assays, 3h had the least interaction with MDR.
      Figure thumbnail gr3b
      Fig. 3Correlation between cellular GCS inhibition and inhibition of [3H]vinblastine uptake in the presence of MDR1. A: A comparison of the dose dependent changes in GlcCer for three homologs of PDMP. B: Representative thin layer chromatography plates of WT-MDCKII and MDR1-MDCKII cells treated with 3h. SM, sphimgomyelin; GalCer, galactosylceramide. C: Effects of different concentrations of three novel GCS inhibitors on the uptake of [3H]vinblastine into MDR1-MDCKII cells. Values are expressed as a percent of vehicle-treated cells and are given as means ± SE (n = 3). 3h only differs from 3i by a single methylene spacer and from 3e by a single bridging CH2, but these changes result in marked changes in the interaction with MDR. D: Effects of different GCS inhibitors on two parameters, [3H]vinblastine uptake into MDR1-MDCKII cells expressed as a fold change between 30 μM of the inhibitor and vehicle (x axis) and the fold change in GCS IC50 between MDR1- and WT-MDCKII cells (y axis). For most compounds, there was a good correlation between the two parameters (the line represents equal changes). There was a significant correlation between both parameters although two compounds, 3i and 3j, showed noticeably greater effects on the enzyme IC50 than 3f and 3e despite similar magnitude effects on [3H]vinblastine uptake. As assessed by both assays, 3h had the least interaction with MDR.
      The potential interactions of the novel compounds with MDR1 was further examined by determining their effects on [3H]vinblastine transport in MDR1-MDCKII cells (Fig. 3C). Consistent with the marked rightward shifts in the concentration response curves observed for analogs 3e and 3i, a significant dose dependent increase in uptake in [3H]vinblastine was observed. In contrast, the –CH2-indan-2-yl homolog 3h resulted in no significant dose-dependent change in vinblastine uptake. These results are consistent with the absence of recognition of 3h as a substrate for MDR1.
      The full range of active carboxamide inhibitors was compared using both the GCS inhibition and vinblastine transport assays in WT- versus MDR1-MDCKII cells. There was reasonable agreement between the two assays with regard to changes in the presence of MDR1 (Fig. 3D; r= 0.50). Based on these results and the IC50 against the cell lysate synthase activity, 3h (CCG-203586) was chosen as a lead compound for in vivo studies. Six-week-old wild-type mice were initially treated with 10 mg/kg/day of 3h or vehicle for 3 days and euthanized 12 h after the last injection (Fig. 4). A 10% decrease in the GlcCer levels in the brains of 3h inhibitor-treated mice was observed without changes in the eliglustat or vehicle-treated mice. When a higher dose of inhibitor was employed (30 mg/kg q 12 h for 3 days), a more significant fall in brain GlcCer levels was observed (17%) in 3h inhibitor-treated mice without any changes in the vehicle-treated or eliglustat-treated mice. By contrast, eliglustat lowered liver and kidney GlcCer levels, consistent with its previously reported effects. These data confirmed the ability of 3h to cross the BBB and inhibit GCS. The mice tolerated treatment with both doses of 3h well, displaying no weight loss or obvious gastrointestinal, respiratory, or behavioral changes.
      Figure thumbnail gr4
      Fig. 4Effect of eliglustat and inhibitor 3h (CCG-203586) on organ GlcCer content after short term exposure. Six- to eight-week-old C57BL/6 mice were treated with either vehicle or inhibitor for 3 days. Mice received 10 or 60 mg/kg/day i.p. of the respective inhibitor and were then euthanized 12 h following the last injection. The data are expressed as μg of GlcCer normalized to 0.5 or 2 μmoles of phospholipid (PL) and represent the mean ± SD and * denotes P < 0.05 by student's t-test versus vehicle treated control, n = 4 except for 10 mg/kg/day treatment brain samples where n = 3.

      DISCUSSION

      Property modeling of compounds retaining the well-established PDMP pharmacophore and four screening assays were employed in an attempt to find GCS inhibitors that would be active in the brain. Specific findings included the following. First, a series of PDMP analogs were designed to lower or eliminate their recognition as substrates for the MDR1 transporter by comparison of selected physical properties to drugs known to cross the BBB. Several compounds retained nanomolar activity as GCS inhibitors in both crude cell enzyme and whole cell assays. The activity of these compounds confirmed the utility of the pharmacophore of PDMP as a scaffold for GCS inhibitors. Second, one analog (3h), was demonstrated to be a particularly poor substrate for the MDR1 protein relative to the other analogs using two in vitro assays based on the stable expression of MDR1. Third, the 3h analog demonstrated significant in vivo activity in lowering brain GlcCer levels after 3 days of intraperitoneal administration. The decrease in brain GlcCer was in contrast to eliglustat tartrate, which had no discernable effect on brain GlcCer when administered in parallel.
      A wealth of scholarship exists about the differences in the physical chemical properties of CNS versus nonCNS drugs (
      • Mahar Doan K.M.
      • Humphreys J.E.
      • Webster L.O.
      • Wring S.A.
      • Shampine L.J.
      • Serabjit-Singh C.J.
      • Adkison K.K.
      • Polli J.W.
      Passive permeability and P-glycoprotein-mediated efflux differentiate central nervous system (CNS) and non-CNS marketed drugs.
      ,
      • Leeson P.D.
      • Davis A.M.
      Time-related differences in the physical property profiles of oral drugs.
      ,
      • Pajouhesh H.
      • Lenz G.R.
      Medicinal chemical properties of successful central nervous system drugs.
      ) and in fact, a number of computed properties vary significantly between the two groups. Compounds possessing molecular parameters outside of the “ideal” ranges for CNS drugs tend to either have poor passive diffusion through cell membranes or to be substrates for MDR1 mediated efflux. The range of tolerated physical chemical properties observed for CNS-active agents is significantly narrower than the whole of oral therapeutics, a result of the unique physical characteristics of the BBB (
      • Cecchelli R.
      • Berezowski V.
      • Lundquist S.
      • Culot M.
      • Renftel M.
      • Dehouck M.P.
      • Fenart L.
      Modelling of the blood-brain barrier in drug discovery and development.
      ,
      • Lundquist S.
      • Renftel M.
      • Brillault J.
      • Fenart L.
      • Cecchelli R.
      • Dehouck M.P.
      Prediction of drug transport through the blood-brain barrier in vivo: a comparison between two in vitro cell models.
      ). Specifically, the endothelial cells of the cerebral capillaries have extraordinarily tight junctions, requiring compounds to pass through entirely via the transcellular route by passive diffusion. Furthermore, xenobiotic efflux transporter MDR1 in the BBB efficiently expels a wide variety of substrates.
      The retention of the pharmacophore derived from the development of D-threo-PDMP led us to focus primarily on modifications to the N-acyl group of the amide. Substitutions were made to lower the number of rotatable bonds and total polar surface area. Although a number of new PDMP homologs with nanomolar activity against GCS were identified (notably 3h), no overall correlations were apparent between the ratio of inhibition of the synthase in MDR versus WT MDCKII cells and any of the calculated parameters in Table 2. However, some interesting pair-wise comparisons between closely related structures were observed. Replacing the carbamate oxygen of 3b with CH2 (3c) reduces both TPSA and lipophilicity (SlogP), and appeared to result in less recognition by MDR1. N-methylating the amide of 3c to afford 3j resulted in a dramatic increase in MDR/WT, which is opposite to what one would expect from removal of a hydrogen bond donor and reduced TPSA. Phenylpropionamides 3e and 3h differ only in a single bridging CH2 that eliminates two rotatable bonds and conformationally restricts the propyl chain. This resulted in both a significant improvement in enzyme activity (27 nM vs. 80 nM) and an apparent dramatic reduction in MDR efflux (MDR/WT of 1.6 vs. 5.7). Finally, compound 3i, differing from 3h by the absence of a single CH2, was avidly recognized as a MDR1 substrate. Overall, these results suggest that recognition by MDR1 within this series is much more dependent on conformation than on small changes in lipophilicity or TPSA. Indeed, when energy-minimized conformations of compounds 3c,3e,3h,3i, and 3j were overlaid, significant differences were apparent (supplementary Fig. V) This is consistent with previous reports that conformational restriction can be successful at attenuating MDR1 efflux, presumably due to locking compounds into conformations that are no longer recognized by MDR1 (
      • Raub T.J.
      P-glycoprotein recognition of substrates and circumvention through rational drug design.
      ).
      The comparison in GCS inhibition between WT and MDR-MDCKII cells also served as a facile and convenient method for detecting recognition of GCS inhibitors by MDR1. However, because of the importance of MDR1 in determining brain penetration of many therapeutics (
      • Miller D.S.
      Regulation of P-glycoprotein and other ABC drug transporters at the blood-brain barrier.
      ), a second assay was used to assess MDR1 interactions by determining whether GCS inhibitors increased [3H]vinblastine uptake into MDR-MDCKII cells. Vinblastine is an MDR substrate and evidence indicates that MDR is the sole/predominant efflux mechanism present for vinblastine in MDR-MDCKII cells (
      • Ito S.
      • Woodland C.
      • Sarkadi B.
      • Hockmann G.
      • Walker S.E.
      • Koren G.
      Modeling of P-glycoprotein-involved epithelial drug transport in MDCK cells.
      ,
      • Kuteykin-Teplyakov K.
      • Luna-Tortos C.
      • Ambroziak K.
      • Loscher W.
      Differences in the expression of endogenous efflux transporters in MDR1-transfected versus wildtype cell lines affect P-glycoprotein mediated drug transport.
      ). Under control conditions, [3H]vinblastine levels were only 14 ± 2% of that found in WT-MDCKII cells [which do express some canine MDR (
      • Kuteykin-Teplyakov K.
      • Luna-Tortos C.
      • Ambroziak K.
      • Loscher W.
      Differences in the expression of endogenous efflux transporters in MDR1-transfected versus wildtype cell lines affect P-glycoprotein mediated drug transport.
      )], and inhibiting MDR (e.g., with verapamil) results in an increase in [3H]vinblastine uptake. Such an increase in uptake was found with many of the GCS inhibitors. It should be noted, however, that an increase in [3H]vinblastine uptake only indicates that the GCS inhibitor interacts with MDR; it does not necessarily imply it is a substrate.
      In general, there was good accord between the [3H]vinblastine and GCS assays. However, for two pairs of compounds (3i/3j vs. 3e/3f) that showed similar effects on [3H]vinblastine uptake, there were marked differences in the MDR effects in the GCS assay (∼20-fold vs. ∼5-fold). This may reflect that the [3H]vinblastine assay measures whole cell uptake, whereas the effective concentration for inhibition of GCS is the drug level at the outer Golgi membrane. However, these compounds are also cationic amphiphilic drugs that are protonated under the acidic conditions and trapped in the lysosome where they may exert other effects. For example, a secondary site of activity of PDMP and other cationic amphiphilic drugs is the inhibition of group XV phospholipase A2 (
      • Abe A.
      • Hiraoka M.
      • Shayman J.A.
      A role for lysosomal phospholipase A2 in drug induced phospholipidosis.
      ). This inhibition may be the basis for some forms of drug-induced phospholipidosis. Nevertheless, the use of an enzyme assay in WT- and MDR-MDCKII cells to assess the MDR affinity of enzyme inhibitors is novel and may potentially be extended to other types of enzyme inhibitors provided that MDCKII cells (or other cells overexpressing MDR) express the relevant target enzyme.
      The final assay employed was measurement of GlcCer levels in vivo. Although MDR is important for the efflux of many therapeutics at the BBB, there are also other active efflux mechanisms that impact brain penetration [e.g., breast cancer-related protein (
      • Miller D.S.
      Regulation of P-glycoprotein and other ABC drug transporters at the blood-brain barrier.
      )]. The role of such transporters can be examined using overexpressing cell lines (as used here for MDR). However, because of the complexity of the efflux systems present at the BBB, it was more efficient to employ an in vivo assay to determine whether our lead compound could penetrate the brain at sufficient concentrations to impact GCS. Importantly, with short-term dosing, inhibitor 3h was observed to significantly lower brain GlcCer levels. This effect was specific for this analog in that eliglustat failed to demonstrate any change in brain cerebroside content under identical dosing conditions, even though it significantly lowered liver and kidney glycolipids. This finding confirmed that developing a high-affinity GCS inhibitor (3h) with a lack of recognition by MDR1 is able to result in a pharmacological response in the brain.
      In summary, the present study outlines a general strategy for the design and testing of compounds that lack MDR1 affinity and identifies a new PDMP analog that appears to satisfy the properties of high inhibitory activity against GCS and limited MDR1 affinity. Because the concept of synthesis inhibition for the treatment of glycosphingolipidoses by PDMP-based GCS inhibitors is now well established on both experimental and clinical grounds, the identification of a related compound that is active within brain is encouraging. Specifically, the D-threo-ethylendioxyphenyl analogs of PDMP are characterized by inhibition of GCS at low nanomolar concentrations, high specificity, and the absence of β-glucocerebrosidase binding. Recent phase 2 clinical data for eliglustat tartrate have demonstrated a clinical response in type 1 Gaucher disease that is comparable to enzyme replacement therapy as measured by reduction in spleen and liver volume, correction of anemia, and improvement in thrombocytopenia. The untoward effects observed with NBDNJ, including weight loss, diarrhea, and tremor, were not observed in this clinical trial nor in an extension study. These observations are consistent with the high specificity of the drug and its absence of CNS penetration.
      Although the absence of eliglustat tartrate distribution into brain may be advantageous for glycosphingolipidoses without CNS manifestations, including type 1 Gaucher and Fabry diseases, our identification of a PDMP homolog that crosses the BBB is of potential therapeutic benefit for disorders such as GM2 gangliosidoses, Tay-Sachs, Sandhoff disease, and types 2 and 3 Gaucher disease. Excellent murine models for these disorders exist and, following the characterization of the pharmacokinetic profile of 3h, will provide the basis for experimental proof of principle for the use of synthesis inhibition in these disorders. Finally, recent work has defined a mechanistic association between loss of glucocerebrosidase activity, neuronal GlcCer accumulation, and α-synuclein accumulation (
      • Mazzulli J.R.
      • Xu Y.H.
      • Sun Y.
      • Knight A.L.
      • McLean P.J.
      • Caldwell G.A.
      • Sidransky E.
      • Grabowski G.A.
      • Krainc D.
      Gaucher disease glucocerebrosidase and alpha-synuclein form a bidirectional pathogenic loop in synucleinopathies.
      ). These findings provide a potential explanation for the high association of type 1 Gaucher disease with Parkinson's disease. The identification of a potent GCS inhibitor with CNS activity provides a pharmacological tool for further assessing this association and possible therapeutic intervention.

      Supplementary Material

      References

        • Barton N.W.
        • Brady R.O.
        • Dambrosia J.M.
        • Di Bisceglie A.M.
        • Doppelt S.H.
        • Hill S.C.
        • Mankin H.J.
        • Murray G.J.
        • Parker R.I.
        • Argoff C.E.
        • et al.
        Replacement therapy for inherited enzyme deficiency–macrophage-targeted glucocerebrosidase for Gaucher's disease.
        N. Engl. J. Med. 1991; 324: 1464-1470
        • Radin N.S.
        Treatment of Gaucher disease with an enzyme inhibitor.
        Glycoconj. J. 1996; 13: 153-157
        • Shayman J.A.
        • Lee L.
        • Abe A.
        • Shu L.
        Inhibitors of glucosylceramide synthase.
        Methods Enzymol. 2000; 311: 373-387
        • Weinreb N.J.
        • Barranger J.A.
        • Charrow J.
        • Grabowski G.A.
        • Mankin H.J.
        • Mistry P.
        Guidance on the use of miglustat for treating patients with type 1 Gaucher disease.
        Am. J. Hematol. 2005; 80: 223-229
        • Shayman J.A.
        Eliglustat tartrate.
        Drugs Future. 2010; 35: 613-621
        • Lukina E.
        • Watman N.
        • Arreguin E.A.
        • Banikazemi M.
        • Dragosky M.
        • Iastrebner M.
        • Rosenbaum H.
        • Phillips M.
        • Pastores G.M.
        • Rosenthal D.I.
        • et al.
        A phase 2 study of eliglustat tartrate (Genz-112638), an oral substrate reduction therapy for Gaucher disease type 1.
        Blood. 2010; 116: 893-899
        • Abe A.
        • Gregory S.
        • Lee L.
        • Killen P.D.
        • Brady R.O.
        • Kulkarni A.
        • Shayman J.A.
        Reduction of globotriaosylceramide in Fabry disease mice by substrate deprivation.
        J. Clin. Invest. 2000; 105: 1563-1571
        • Liu Y.
        • Wada R.
        • Kawai H.
        • Sango K.
        • Deng C.
        • Tai T.
        • McDonald M.P.
        • Araujo K.
        • Crawley J.N.
        • Bierfreund U.
        • et al.
        A genetic model of substrate deprivation therapy for a glycosphingolipid storage disorder.
        J. Clin. Invest. 1999; 103: 497-505
        • Miller D.S.
        Regulation of P-glycoprotein and other ABC drug transporters at the blood-brain barrier.
        Trends Pharmacol. Sci. 2010; 31: 246-254
        • Hirth B.H.
        • Siegel C.
        Synthesis of UDP-glucose: N-acylsphingosine glucosyltransferase inhibitors. Patent no. 6,855,830.
        in: Genzyme Corporation, United States2005
        • Shayman J.A.
        • Abe A.
        Glucosylceramide synthase: assay and properties.
        Methods Enzymol. 2000; 311: 42-49
        • Shu L.
        • Shayman J.A.
        Src kinase mediates the regulation of phospholipase C-gamma activity by glycosphingolipids.
        J. Biol. Chem. 2003; 278: 31419-31425
        • Shu C.
        • Shen H.
        • Teuscher N.S.
        • Lorenzi P.J.
        • Keep R.F.
        • Smith D.E.
        Role of PEPT2 in peptide/mimetic trafficking at the blood-cerebrospinal fluid barrier: studies in rat choroid plexus epithelial cells in primary culture.
        J. Pharmacol. Exp. Ther. 2002; 301: 820-829
        • Wishart D.S.
        • Knox C.
        • Guo A.C.
        • Cheng D.
        • Shrivastava S.
        • Tzur D.
        • Gautam B.
        • Hassanali M.
        DrugBank: a knowledgebase for drugs, drug actions and drug targets.
        Nucleic Acids Res. 2008; 36: D901-D906
        • Jimbo M.
        • Yamagishi K.
        • Yamaki T.
        • Nunomura K.
        • Kabayama K.
        • Igarashi Y.
        • Inokuchi J.I.
        Development of a new inhibitor of glucosylceramide synthase.
        J. Biochem. 2000; 127: 485-491
        • Lee L.
        • Abe A.
        • Shayman J.A.
        Improved inhibitors of glucosylceramide synthase.
        J. Biol. Chem. 1999; 274: 14662-14669
        • Lipinski C.A.
        • Lombardo F.
        • Dominy B.W.
        • Feeney P.J.
        Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings.
        Adv. Drug Deliv. Rev. 2001; 46: 3-26
        • Garberg P.
        • Ball M.
        • Borg N.
        • Cecchelli R.
        • Fenart L.
        • Hurst R.D.
        • Lindmark T.
        • Mabondzo A.
        • Nilsson J.E.
        • Raub T.J.
        • et al.
        In vitro models for the blood-brain barrier.
        Toxicol. In Vitro. 2005; 19: 299-334
        • Wang Q.
        • Rager J.D.
        • Weinstein K.
        • Kardos P.S.
        • Dobson G.L.
        • Li J.
        • Hidalgo I.J.
        Evaluation of the MDR-MDCK cell line as a permeability screen for the blood-brain barrier.
        Int. J. Pharm. 2005; 288: 349-359
        • Gouaze-Andersson V.
        • Cabot M.C.
        Glycosphingolipids and drug resistance.
        Biochim. Biophys. Acta. 2006; 1758: 2096-2103
        • Mahar Doan K.M.
        • Humphreys J.E.
        • Webster L.O.
        • Wring S.A.
        • Shampine L.J.
        • Serabjit-Singh C.J.
        • Adkison K.K.
        • Polli J.W.
        Passive permeability and P-glycoprotein-mediated efflux differentiate central nervous system (CNS) and non-CNS marketed drugs.
        J. Pharmacol. Exp. Ther. 2002; 303: 1029-1037
        • Leeson P.D.
        • Davis A.M.
        Time-related differences in the physical property profiles of oral drugs.
        J. Med. Chem. 2004; 47: 6338-6348
        • Pajouhesh H.
        • Lenz G.R.
        Medicinal chemical properties of successful central nervous system drugs.
        NeuroRx. 2005; 2: 541-553
        • Cecchelli R.
        • Berezowski V.
        • Lundquist S.
        • Culot M.
        • Renftel M.
        • Dehouck M.P.
        • Fenart L.
        Modelling of the blood-brain barrier in drug discovery and development.
        Nat. Rev. Drug Discov. 2007; 6: 650-661
        • Lundquist S.
        • Renftel M.
        • Brillault J.
        • Fenart L.
        • Cecchelli R.
        • Dehouck M.P.
        Prediction of drug transport through the blood-brain barrier in vivo: a comparison between two in vitro cell models.
        Pharm. Res. 2002; 19: 976-981
        • Raub T.J.
        P-glycoprotein recognition of substrates and circumvention through rational drug design.
        Mol. Pharm. 2006; 3: 3-25
        • Ito S.
        • Woodland C.
        • Sarkadi B.
        • Hockmann G.
        • Walker S.E.
        • Koren G.
        Modeling of P-glycoprotein-involved epithelial drug transport in MDCK cells.
        Am. J. Physiol. 1999; 277: F84-F96
        • Kuteykin-Teplyakov K.
        • Luna-Tortos C.
        • Ambroziak K.
        • Loscher W.
        Differences in the expression of endogenous efflux transporters in MDR1-transfected versus wildtype cell lines affect P-glycoprotein mediated drug transport.
        Br. J. Pharmacol. 2010; 160: 1453-1463
        • Abe A.
        • Hiraoka M.
        • Shayman J.A.
        A role for lysosomal phospholipase A2 in drug induced phospholipidosis.
        Drug Metab Lett. 2007; 1: 49-53
        • Mazzulli J.R.
        • Xu Y.H.
        • Sun Y.
        • Knight A.L.
        • McLean P.J.
        • Caldwell G.A.
        • Sidransky E.
        • Grabowski G.A.
        • Krainc D.
        Gaucher disease glucocerebrosidase and alpha-synuclein form a bidirectional pathogenic loop in synucleinopathies.
        Cell. 2011; 146: 37-52