Tricyclic antidepressants, quinacrine and a novel, synthetic chimera thereof clear prions by destabilizing detergent-resistant membrane compartments

Authors


Address correspondence and reprint requests to Carsten Korth, Institute for Neuropathology, Heinrich Heine University of Düsseldorf, Moorenstrasse 5, 40225 Düsseldorf, Germany.
E-mail: ckorth@uni-duesseldorf.de

Abstract

Prion diseases are invariably fatal, neurodegenerative diseases transmitted by an infectious agent, PrPSc, a pathogenic, conformational isoform of the normal prion protein (PrPC). Heterocyclic compounds such as acridine derivatives like quinacrine abolish prion infectivity in a cell culture model of prion disease. Here, we report that these compounds execute their antiprion activity by redistributing cholesterol from the plasma membrane to intracellular compartments, thereby destabilizing membrane domains. Our findings are supported by the fact that structurally unrelated compounds with known cholesterol-redistributing effects – U18666A, amiodarone, and progesterone – also possessed high antiprion potency. We show that tricyclic antidepressants (e.g. desipramine), another class of heterocyclic compounds, displayed structure-dependent antiprion effects and enhanced the antiprion effects of quinacrine, allowing lower doses of both drugs to be used in combination. Treatment of ScN2a cells with quinacrine or desipramine induced different ultrastructural and morphological changes in endosomal compartments. We synthesized a novel drug from quinacrine and desipramine, termed quinpramine, that led to a fivefold increase in antiprion activity compared to quinacrine with an EC50 of 85 nm. Furthermore, simvastatin, an inhibitor of cholesterol biosynthesis, acted synergistically with both heterocyclic compounds to clear PrPSc. Our data suggest that a cocktail of drugs targeting the lipid metabolism that controls PrP conversion may be the most efficient in treating Creutzfeldt-Jakob disease.

Abbreviations used
BBB

blood–brain barrier

BSA

bovine serum albumin

BSE

bovine spongiform encephalopathy

CJD

Creutzfeldt-Jakob disease

DRM

detergent-resistant membrane domains

GPI

glycophosphatidylinositol

PBS

phosphate-buffered saline

PrP

prion protein

ScN2a

scrapie-infected neuroblastoma cells

Prion diseases are invariably fatal, transmissible neurodegenerative diseases that present in humans as Creutzfeldt-Jakob disease (CJD), Gerstmann-Sträussler-Scheinker disease, familial fatal insomnia or Kuru. They are caused by unique pathogens consisting of a disease-specific conformational isoform of the glycophosphatidylinositol (GPI)-anchored prion protein, PrPSc that replicates by inducing conversion of the normal host prion protein, PrPC (Prusiner 1998). Replication is dependent on the presence of PrPC in detergent-resistant membrane domains (Taraboulos et al. 1995; Vey et al. 1996; Kaneko et al. 1997). The unique pathogenic mechanism that involves misfolding of a normal host protein is related to their ability to manifest as spontaneous, inherited and infectious maladies (Prusiner 1998).

Prion diseases are rare in humans but endemic in sheep (scrapie) and even cattle (bovine spongiform encephalopathy; BSE) and mule deer (chronic wasting disease). Transmission of prions to humans from cattle infected with BSE has caused variant CJD (vCJD) (Collinge et al. 1996; Will et al. 1996; Scott et al. 1999) in about 150 young adults (Giles 2004). A survey of tonsillectomies and appendectomies in Great Britain, the most severely BSE-affected country, has led to the estimate that about 50–700 per million people could be infected with vCJD (Hilton et al. 2004). Furthermore, recent reports have pointed to the possible transmission of human prions through contaminated blood (Llewelyn et al. 2004; Peden et al. 2004). So far, no pharmacotherapy exists for the prion diseases. Therefore, identifying cellular pathways involved in PrP conversion as well as elucidating the natural degradation and clearance mechanisms of PrPSc is a major goal for providing molecular targets for the pharmacotherapy of prion diseases.

The search for blood–brain barrier (BBB)-permeable compounds with antiprion effects led to the identification of acridine and phenothiazine derivatives as new antiprion lead compounds (Korth et al. 2001). In scrapie-inoculated mice, a protocol with a time-limited application of quinacrine starting at mid-incubation time led to a 25% prolongation in survival time in CD−1 and FVB mice (Ryou et al. 2004). Application of quinacrine at lower doses was not effective (Collins et al. 2002; Ryou et al. 2004) or toxic when infused intrathecally (Doh-ura et al. 2004). Application of quinacrine to single cases of CJD patients for compassionate treatment in terminal stages of the disease led to transient improvements of symptoms but did not halt the disease (Nakajima et al. 2004), though a final statement on the efficacy of quinacrine monotherapy in CJD awaits carefully conducted clinical trials. Taking quinacrine as the leading compound in antiprion therapy led to the identification of a series of bis-acridines with a 10-fold higher antiprion efficacy in cell culture (May et al. 2003). The molecular mechanism of antiprion action executed by the heterocyclic compounds has remained unclear.

In this study we demonstrate that heterocyclic antiprion compounds act by an intracellular redistribution of cholesterol, thereby causing destabilization of conversion-mandatory detergent-resistant membrane domains (DRMs). We add further weight to this conclusion by introducing the iminodibenzyl derivatives, or tricyclic antidepressants. These compounds form a novel class of heterocyclic antiprion reagents that display a strict structure–antiprion activity relationship. We also show that the structurally unrelated compounds U18666A and progesterone, which are used pharmacologically to induce cholesterol/lipid storage diseases, also exhibit antiprion effects. Based on our insight that transiently manipulating cellular lipid distribution has sustained effects on clearing prions, we developed two pharmacological antiprion strategies. First, we synthesized a novel lead compound linking the two heterocyclic moieties, quinacrine and desipramine, into one molecule that exhibited fivefold increased antiprion potency. Second, a combination of the synergistically acting antiprion compounds quinacrine, desipramine and simvastatin was demonstrated to potentiate antiprion effects, suggesting this antiprion cocktail as an immediately available pharmacological option for treating patients with Creutzfeldt-Jakob disease.

Experimental procedures

Compounds were purchased from Sigma-Aldrich (Munich, Germany), with the exception of simvastatin and doxepin (Tocris, Bristol, UK) and N-desmethyltrimipramine, N-desmethyldoxepine and dothiepine (Euriso-top, Saarbruecken, Germany). Compounds were diluted in either H2O, dimethyl sulfoxide or ethanol in a 10 mm stock solution and stored at 4°C. Cell culture media and supplements were from Invitrogen (Carlsbad, CA, USA).

PrPSc inhibition assay in ScN2a cells

Neuroblastoma cells were infected with the RML strain of mouse-adapted scrapie prions and subcloned (Bosque and Prusiner 2000). A confluent 10-cm dish was split and a drop of cells (c. 50 μL) was pipetted into 4 mL of minimal essential medium (MEM) containing 10% (v/v) fetal calf serum (FCS), penicillin-streptomycin, l-glutamine, and antiprion drugs, in a 60-mm dish. Medium was exchanged every second day, together with the drugs. Cells were lysed (lysis buffer: 10 mm Tris, pH 8.0, 150 mm NaCl, 0.5% Triton X-100, 0.5% deoxycholate) on the 7th day, having achieved about 80% confluency. All experiments were repeated at least three times.

Western blotting

Cell lysates were digested with proteinase K at 20 μg/mL for 30 min at 37°C. The reaction was stopped with 2 mmphenylmethylsulphonylfluoride (PMSF), and the lysates were centrifuged for 45 min at 100 000 × g in an ultracentrifuge (Beckman Coulter, Fullerton, CA, USA). Pellets were resuspended in sample buffer, and sodium dodecyl sulfate − polyacrylamide gel electrophoresis (SDS-PAGE)/immunoblotting was performed according to standard techniques. Immunoblots were incubated with a horseradish peroxidase-labeled chicken α-PrP-antibody and developed with enhanced chemiluminescence (Amersham Pharmacia). Effective concentrations where half the antiprion activity was achieved (EC50) were determined by densitometry of immunoreactive bands on western blot.

Generation of chicken α-mouse PrP IgY

Recombinant mouse PrP was produced in Escherichia coli as described (Korth et al. 1999) and used in 100 μg doses for intramuscular immunization of chickens (Gallus gallus) at 3-week intervals. Eggs were collected 2 weeks after the second boost with recombinant mouse PrP (recMoPrP). Egg yolks containing the IgY were diluted 1 : 5 in cold 20 mm NaAc pH 5.2 and left overnight at 4°C. After removing insoluble material (20 000 × g, 20 min), IgY was salted out by addition of 20% (w/v) (NH4)2SO4 and pelleted after 2 h at 4°C (20 000 × g, 20 min). IgY was then resolubilized in 20 mm Tris pH 8, 150 mm NaCl, 0.1% Tween-20, 1 mm EDTA.

Affinity purification

RecMoPrP was diluted to 0.5 mg/mL in a suspension of NHS-activated Sepharose (Amersham Pharmacia Biotech, Piscataway, NJ, USA; prewashed according to manufacturer's protocol) in cold 50 mm NaHCO3 pH 8.3, 1% Triton X-100, 20% dimethyl sulfoxide, at 3 mg protein per mL resin. This suspension was stirred overnight at 4°C, free NHS groups were subsequently blocked with 50 mm glycine (1 h, 20°C). The resin was then washed consecutively with 50 mm Tris pH 8, 50 mm glycine pH 3, and finally 20 mm Tris pH 8, 150 NaCl, 0.2% Triton X-100, 0.2% Tween-20, 2 mm EDTA. In this buffer, immobilized recMoPrP was mixed with crude α-MoPrP IgY isolate (approximately 200 mg IgY per mg MoPrP) and stirred overnight at 4°C. The resin was then collected and washed extensively, followed by elution with 100 mm glycine pH 3, 1 m NaCl, 1% Triton X-100 (pH adjusted to 8 immediately afterwards). Affinity-purified α-MoPrP IgY was labelled with EZ-link plus activated horseradish peroxidase (Pierce, Rockford, IL, USA) according to the manufacturer's manual. This affinity-purified antibody recognized the un- and monoglycosylated forms of PrP better than the diglycosylated one.

Immunofluorescence and filipin-staining

ScN2a cells or WAC-2 human neuroblastoma cells (Zhang et al. 2001) were grown on coverslips and treated with antiprion compounds, as described above. On day 6 of treatment, cells were washed once with cold phosphate-buffered saline (PBS), and fixed with 4% paraformaldehyde (in PBS) for 15 min at room temperature. Nonspecific antibody binding was blocked and cells were permeabilized with 5% dried milk powder/1% bovine serum albumin (BSA)/0.5% saponin in PBS for 15 min, before incubating with primary antibody [α-caveolin-1 (Pharmingen, Franklin Lakes, NJ, USA), 1 : 40 in 1% BSA/PBS/0.5% saponin) for 30 min at 20°C. Finally, incubation with rhodamine-linked secondary antibody (Pierce; 1 : 75 in 1% BSA/PBS) was performed for 30 min at 20°C, followed by filipin staining (50 μg/mL) for 30 min at 20°C. The coverslips were washed five times with PBS and mounted upside down on a drop of glycerol on glass slides. The coverslips were viewed and photographed with a Zeiss Axiovert 25 photomicroscope.

Electron microscopy of ultrathin cryosections

ScN2a cells were fixed in 0.2 m PHEM buffer (120 mm PIPES, 50 mm HEPES, 4 mm MgCl2, 20 mm EGTA, pH 6.9) containing 2% paraformaldehyde and 0.2% glutaraldehyde for 2 h at 20°C. Fixed cells were stored as pellets in 0.5% paraformaldehyde in 0.1 m PHEM at 4°C until they were processed for ultrathin cryosectioning as described (Peters et al. 2003; Peters et al. 2006). Ultrathin cryosections were chosen for electron microscopy to retain the option for later immunolabeling, if necessary. Briefly, fixed cells were washed with 0.02 m glycine in PBS, scraped gently from the dish in 1% gelatine in PBS and pelleted in 12% gelatin in PBS. The gelatin cell pellet was solidified on ice and cut into small blocks. For cryoprotection, blocks were infiltrated overnight with 2.3 m sucrose at 4°C and afterwards mounted on aluminum pins and frozen in liquid nitrogen. Ultrathin cryosections with an average thickness of 60 nm were cut with a diamond knife (Diatome, Biel, Switzerland) at − 120°C and picked up in a 1 : 1 mixture of 2.3 m sucrose and 1.8% methyl cellulose in distilled water (Liou et al. 1996). The sections were thawed to room temperature, contrasted and embedded in a mixture of 0.4% uranylacetate in 2% methyl cellulose, after which they were examined in a Tecnai 12 transmission electron microscope (FEI Company, Eindhoven, the Netherlands). Photographic negatives of light and electron microscopic images were digitized with an Epson Perfection 2450 scanner. Inspection and counting of endosomal vacuoles was done blind to treatment conditions by randomly picking 25 cells and calculating the sum of each of five different endosomal vacuoles (multivesicular bodies, multilamellar bodies, lysosomes, late endosomes, and vacuoles) as described (Griffiths et al. 2001; Peters et al. 2003).

Isolation and analysis of detergent-resistant membranes

Cells were grown in a 10 cm dish and treated with compounds as described above. On day 7, cells were washed twice with PBS, scraped on ice and collected in 500 μL homogenization buffer (HB: 250 mm sucrose in 10 mm HEPES, pH 7.4 with proteinase inhibitors; Roche Molecular Biochemicals, Indianapolis, IN, USA). For DRM isolation, standard protocols were used (Schuck et al. 2003). Briefly, after cell homogenization by passing through a 25-G needle 20 times, nuclei were removed by centrifugation (5 min, 1000 × g, 4°C). Supernatant (500 μL) was mixed with 100 μL HB containing 6% Triton-X 100, and incubated on ice for 30 min. The samples were adjusted to 42% sucrose (wt/wt) with 63% sucrose in 10 mm HEPES. A 1.2 mL aliquot of this solution was placed on the bottom of an ultracentifugation tube (Ultra clear, Beckman Coulter) and sequentially overlaid with 3 mL 38% and 400 μL 5% sucrose/HEPES. After centrifugation (270 000 × g for 18 h) six fractions (for ScN2a cells) or three fractions (for WAC cells) were collected from the top to the bottom: two (or for WAC cells one) times 500 μL, two times 1000 μL and two times 800 μL, respectively. Membranes were destroyed by three freeze thaw cycles and diluted to less than 10% sucrose with PBS. The diluted fractions were ultracentrifuged for 3 h at 100 000 × g and 4°C, and the pellet dissolved in 20 μL 2X SDS loading buffer and analyzed by western blotting.

Synthesis of (6-chloro-2-methoxy-acridin-9-yl)-(4-{4-[3-(10,11-dihydro-dibenzo[b,f]azepin-5-yl)-propyl]-piperazin-1-yl}-butyl)-amine (quinpramine)

Iminodibenzyl was converted to the respective N-piperazin-1-ylpropyl substituted derivative by deprotonation with NaNH2 in toluene followed by alkylation with 1-bromo-3-chloropropane. Subsequent amination with piperazine and conversion with bromobutyronitrile in CH3CN using Na2CO3 as a base and NaI as a catalyst resulted in the corresponding cyanopropylpiperazine, which was reduced with LiAlH4 in Et2O the respective primary amine. Finally, nucleophilic aromatic substitution with 6,9-dichloro-2-methoxyacridine in PhOH at 100°C afforded quinpramine in 19% overall yield (5 steps). The structure was confirmed by 1H-NMR, 13C-NMR and liquid chromatography-mass spectrometry using allophycocyanin ionization.

Preparation of 1,8-bis(10,11-dihydro-dibenzo[b,f]azepine-5-yl-carbonylamino)-3,6-dioxaoctane (RK-1)

1,8-Bis 10,11-dihydro-dibenzo[b,f]azepine-5-yl-carbonylamino-3,6-dioxaoctane was synthesized by Chemcon (Freiburg, Germany) according to the following procedure. Potassium carbonate (1.15 g, 7.76 mmol) was added to a mixture of 10,11-dihydro-dibenz[b,f]azepine-5-carbonyl chloride (1 g, 3.88 mmol) and 1,2-bis(2-aminoethoxy)ethane (0.288 g, 1.94 mmol) in dimethylformamide. After being stirred for 24 h at 20°C, 30 mL of water was added and the white precipitate was filtrated, washed with water and dried in high vacuum. The structure was confirmed by 1H NMR and MALDI-MS.

Results

Structure-dependent antiprion effects of iminodibenzyl derivatives

Iminodibenzyl derivatives (or tricyclic antidepressants) are a class of heterocyclic compounds very similar in their structural design to the antiprion effective acridine and phenothiazine derivatives (Korth et al. 2001), but with a different and more favorable spectrum of side effects. We therefore investigated whether this class of compounds exhibited a structure-antiprion activity relationship. We examined 12 antidepressants and found clear antiprion effects for each of them (Table 1).

Table 1.   Structure-activity relationship of tricyclic antidepressants on PrPSc inhibition
CompoundStructure EC50, μMFull Activity, μMCompoundStructure EC50, μmFull Activity, μm
  1. EC50 defined as concentration resulting in half the PrPSc immunoreactivity compared to untreated control cells, Full Activity defined as minimal concentration at which no PrPSc immunoreactivity was seen. Toxicity defined by growth to less than 80% confluency compared to untreated control: *50 μm, †40 μm, ‡20 μm, §12 μm.

Imipramine*inline image510Desipramineinline image2.56
Trimipramineinline image2.56N-Desmethyltrimipramineinline image12.5
Clomipramineinline image24Norclomipramine§inline image12.5
Amitriptylineinline image57.5Nortriptylineinline image36
Doxepininline image7.510Nordoxepininline image4.57.5
Dothiepininline image7.510    
Iminodibenzylinline image> 10> 10    

Distinct structure–antiprion activity relationships could be observed (see Table 1, and iminodibenzyl for following position numbers):

  • 1Substitutions in the middle ring moiety of the tricyclic scaffold with nitrogen at position 5 to carbon or substitution of carbon at position 10 to oxygen (doxepine vs. imipramine) or sulfur (dothiepine vs. imipramine) decreased antiprion activity.
  • 2A chlorine substituent at position 7 of the tricyclic scaffold improved the EC50 twofold (compare clomipramine and imipramine with norclomipramine and desipramine, respectively).
  • 3The monobasic sidechain was essential for antiprion activity as iminodibenzyl alone did not possess antiprion potency.
  • 4An additional methyl group in the aliphatic sidechain doubled the antiprion activity (compare imipramine and trimipramine with desipramine/N-desmethyltrimipramine, respectively).
  • 5The substitution grade of the nitrogen atom in the sidechain influenced antiprion activity: secondary amines were twice as active as tertiary amines (compare left panel with right panel in Table 1).

Antiprion effects of iminodibenzyl derivatives were permanent as in a set-off experiment, i.e. after 1 week of treatment and 3 weeks of further cultivating cells without treatment, PrPSc did not reappear (Fig. 1a). The importance of the substitution grade of the nitrogen in the aliphatic side chain could clearly be seen in this set-off experiment: a concentration of 6 μm desipramine but not 6 μm imipramine was effective in curing ScN2a cells.

Figure 1.

 Western blot of protease-digested ScN2a cell lysates depicting the presence or absence of prions (PrPSc) under different treatment conditions. (a) Treatment with different concentrations of imipramine and desipramine for 7 days, then no treatment and weekly splitting for 3 weeks. Single compound concentrations are indicated below. Sc, untreated cells as a negative control; Q, cells treated with 1 μm of quinacrine as positive control. The lack of reappearance of protease-resistant PrPSc after discontinuation of treatment demonstrates that prions have been cleared from this population of ScN2a cells. Desipramine cures the cells at a concentration of about 6 μm, whereas 10 μm of imipramine would be needed. (b) Treatment with different concentrations of reboxetine, a selective noradrenaline reuptake inhibitor, for 7 days. Drug concentrations are indicated. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm). At a concentration of 7.5 μm (where desipramine is already active) no reduction of PrPSc could be detected. (c) Treatment for 7 days with different concentrations of U18666A, a model substance inducing a lipid storage disease phenotype in cells, for 7 days. Compound concentrations are indicated. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm). A concentration of 0.1 μm leads to a clear reduction of PrPSc and a concentration of 0.5 μm completely abolishes prion infection. (d) Treatment for 7 days with different concentrations of progesterone and amiodarone, two structurally unrelated compounds known to induce lysosomal cholesterol storage, for 7 days. Compound concentrations are indicated. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm). Both compounds efficiently cleared prions.

Tricyclic antidepressants have been used for the treatment of depressive disorders since the late 1950s (Kuhn 1958; Alnaes and Kristiansen 1963). Their antidepressant activity is thought to be mediated by influencing serotonin action on neuronal synapses. Tricyclic antidepressants with a tertiary amine in the aliphatic side chain have been reported to be stronger serotonin reuptake inhibitors than tricyclic antidepressants with a secondary amine, the latter being stronger noradrenaline reuptake inhibitors (Sanchez and Hyttel 1999). However, the structurally unrelated selective noradrenaline reuptake inhibitor reboxetine did not display antiprion activity (Fig. 1b), suggesting that neurotransmitter effects of the tricyclic antidepressants do not mediate or relate to antiprion activity.

Heterocyclic compounds induce cellular cholesterol redistribution at antiprion effective concentrations

Because high dosages of cationic amphiphilic substances like quinacrine are known to induce lipidosis (Lullmann et al. 1978) and lead to a disturbance of intracellular cholesterol transport and distribution (Rodriguez-Lafrasse et al. 1990), we investigated whether the antiprion effects of other heterocyclic antiprion compounds were caused by an impairment of cellular cholesterol localization. Conversion of PrPC to PrPSc is critically dependent on the intactness of cholesterol-rich detergent-resistant membrane domains (Taraboulos et al. 1995; Vey et al. 1996; Kaneko et al. 1997). Therefore, interference with cholesterol-dependent cell functions should be a highly efficient pharmacological antiprion strategy.

To visualize cholesterol redistribution induced by treatment with antiprion compounds we performed cholesterol staining with filipin on fixed cells (Fig. 2). Cholesterol redistribution from the cell surface into endosomal/lysosomal compartments could clearly be observed in ScN2a cells treated for 1 week with antiprion active concentrations of quinacrine (Fig. 2b) and desipramine (Fig. 2c) when compared with the untreated control (Fig. 2a).

Figure 2.

 Heterocyclic antiprion compounds redistribute cholesterol from the cell surface into intracellular compartments. Filipin staining of untreated (a) and treated (b–d) ScN2a cells, and untreated (e, i, m) and treated (f–h, j–l, n–p) WAC-2 human neuroblastoma cells, the latter also stained for caveolin-1 (m–p). Cells were treated with the indicated compounds for 7 days before being processed for (immuno)fluorescence microscopy.

Potent cholesterol-redistributing agents U18666A, amiodarone and progesterone exhibit antiprion effects

In a converse experiment, we probed for antiprion activity of compounds unrelated to heterocyclic drugs that are used in the pharmacological induction of cellular cholesterol redistribution. We tested U18666A, an oxidosqualene cyclase inhibitor (Sexton et al. 1983) and cationic amphiphile that is widely used to model cellular lipid storage diseases by dramatically redistributing cholesterol in cells (Liscum and Underwood 1995; Patterson et al. 2001). U18666A had high antiprion activity with an EC50 of 0.1 μm (Fig. 1c) and a corresponding strong cholesterol-redistributing effect in ScN2a cells (Fig. 2d).

Amiodarone, a cationic amphiphilic, antiarrhythmic drug (Zipes and Troup 1978) with potent cholesterol redistributing properties (Palmeri et al. 1995), but structurally unrelated to heterocyclic compounds, also presented antiprion activity with an EC50 between 0.5 and 1 μm (Fig. 1d). Yet another compound known to induce cholesterol storage in the endosomal/lysosomal system but not a cationic amphiphile is progesterone. In ScN2a cells, progesterone cleared PrPSc in concentrations of 20 μm (Fig. 1d), comparable to concentrations used to pharmacologically induce cellular lipid storage disease (Butler et al. 1992). Amiodarone and progesterone also showed clear cholesterol-storage in the lysosomal compartment of ScN2a cells (data not shown).

Lysosome morphology of ScN2a cells treated with heterocyclic compounds

We performed electron microscopy on ScN2a cells that had been treated with quinacrine, desipramine or U18666A at antiprion effective concentrations for 1 week, and specifically examined lysosomal morphology, as lysosomes play a major role in cellular cholesterol metabolism (Mobius et al. 2003). In untreated cells, the majority of late endosomal/lysosomal structures were multivesicular endosomes (Fig. 3 and Table 2). The numbers of multilamellar endosomes, repositories for intracellular cholesterol (Lajoie et al. 2005), were increased in cells treated with quinacrine (Fig. 3b, Table 2) and U18666A (Fig. 3c), whereas in cells treated with desipramine, partially swollen lysosomal vacuoles with lower membrane content were prominent (Fig. 3d, Table 2). These results suggest that the heterocyclic compounds quinacrine and desipramine, although both affecting cholesterol redistribution (Fig. 2), have different effects on endosomal cell compartments.

Figure 3.

 Ultrastructure of endolysosomal compartments of ScN2a cells treated with different compounds (as indicated). Electron micrographs show ultrathin cryosections of an untreated control ScN2a cell (a), and of ScN2a cells treated for 7 days with 1 μm quinacrine (b), 1 μm U18666A (c) and 10 μm desipramine (d), respectively. Asterisks depict multilamellar endosomes. Bar 200 nm.

Table 2.   Quantitative effects of antiprion compounds on morphology of endosomal/lysosomal ScN2a cell compartments. In treated cells, a shift of multivesicular bodies to multilamellar bodies could be observed in late endosomal/lysosomal structures
 Non-treated*%U18666A%Quinacrine%Desipramine%
  • *

    Basal relative values of non-treated ScN2a cells in percent.

  • Relative change (in percent) compared to the non-treated cells.

  • MVB, multivesicular body, endosomal vacuole with > 5 internal vesicles.

  • §

    MLB, multilamellar body, endosomal vacuole with internal multilamellar membranes.

  • LYS, lysosome, endosomal vacuole filled with irregular electrodense material.

  • **

    LE: late endosome, electrolucent endosomal vacuole with some electrodense material.

  • ††

    VAC, vacuole, endosomal vacuole without visible content.

MVB9.8− 3.7− 8.1− 7.5
MLB§10.8+ 4.5+ 3.8− 2.4
LYS¶39.1− 22.4− 17.9− 27.7
LE**38.4+ 18+ 18.7+ 12.8
VAC††1.9+ 3.9+ 3.5+ 24.8

Heterocyclic compounds differentially redistribute caveolin-1

As cholesterol and other lipids are essential molecular components of detergent-resistant membrane domains (DRM), a subcellular membrane compartment essential for conversion of the prion protein (Taraboulos et al. 1995; Vey et al. 1996; Kaneko et al. 1997), we investigated whether this compartment was destabilized as a consequence of administering cationic amphiphilic compounds. For that purpose, we performed immunostaining experiments with the human neuroblastoma WAC-2 cell line (Zhang et al. 2001), a cell line expressing both the prion protein and caveolin-1. Caveolin-1 is a defining protein of caveolae, a particular kind of DRM where GPI-anchored PrP is localized (Peters et al. 2003). Compared to ScN2a cells, WAC-2 cells showed a slightly different baseline cholesterol distribution in the untreated condition, with most of the cholesterol present in asymmetrically distributed perinuclear compartments rather than on the plasma membrane, as for ScN2a cells (Figs 2e, i). Treatment with quinacrine, desipramine or U18666A led to changes in cholesterol distribution similar to those in ScN2a cells, with a more spread out and punctate cholesterol storage in endosomal/lysosomal compartments (Figs 2f–h, j–l). Treatment with U18666A produced a less pronounced phenotype in WAC-2 cells than in ScN2a cells; in ScN2a cells, cholesterol was redistributed into fewer endosomal/lysosomal compartments.

Parallel staining with filipin and an α-caveolin antibody (compare Fig. 2i–l with Fig. 2m–p, respectively) revealed that caveolin-1 partitioned differently upon treatment with heterocyclic compounds. In the untreated cells, there was an about equal distribution of caveolin-1 throughout the cell (Fig. 2m). Quinacrine treatment for 1 week led to the disappearance of caveolin-1 immunoreactivity in a region defined by the nucleus (Fig. 2n), whereas desipramine or U18666A treatment (Fig. 2O, and Fig. 2P, respectively) led to a relative concentration of caveolin-1 immunoreactivity in the perinuclear space. Again, these results showed that quinacrine and desipramine acted differentially on cholesterol metabolism and on components of DRMs.

Antiprion effects of heterocyclic compounds are mediated by destabilization of conversion-competent detergent-resistant microdomains

To further investigate the effects of heterocyclic antiprion compounds on detergent-resistant membrane (DRM) domains, or ‘lipid rafts’ (Simons and Ikonen 1997), we performed sucrose gradients in the presence of cold Triton-X-100 to isolate DRMs of ScN2a or uninfected WAC-2 cell lysates after treatment with U18666A, desipramine or quinacrine at antiprion effective concentrations for 1 week. Because proteins that are attached to the plasma membrane with a GPI anchor, such as PrP, are sorted into DRM domains (Vey et al. 1996), we reasoned that the cholesterol-redistributing effects of quinacrine and the iminodibenzyl derivatives might exert their antiprion effects via destabilization of DRMs. In the floating fractions containing purified DRMs, we detected less PrP immunoreactivity when ScN2a cells had been treated with desipramine, and immunoreactivity was absent in cells treated with the more potent antiprion compounds quinacrine and U18666A (Fig. 4). In the compound-treated uninfected WAC-2 cells, treatment resulted in the decrease (for desipramine) or absence (for quinacrine and U18666A) of PrP immunoreactivity in the DRM fraction.

Figure 4.

 Western blot of sucrose gradients from DRM (detergent-resistant membrane) isolations of untreated ScN2a cells and ScN2a cells treated with different compounds (as indicated). ScN2a cell homogenate was extracted with 1% Triton X-100 and floated on sucrose gradients. The distribution of PrP was analyzed by immunoblotting in ScN2a cells (left panel) and WAC-2 cells (right panel). The light fractions from the top of the gradients containing DRMs are on the left, the heavy bottom fractions on the right.

Quinpramine, a novel compound that covalently links quinacrine and desipramine has increased antiprion potency

Our results suggested that although both quinacrine and desipramine exerted their antiprion actions via their impact on cellular lipid redistribution, their effects seemed slightly different. For example, lysosomal morphology appeared more impaired in ScN2a cells treated with desipramine, and caveolin-1 redistribution was more pronounced in ScN2a cells treated with quinacrine. This prompted us to combine both antiprion drugs to potentiate the antiprion effect. When we treated ScN2a cells with a combination of quinacrine and desipramine at dosages at which each drug alone did not exhibit antiprion potency, prions were completely cleared (Fig. 5A), in that they had not reappeared 3 weeks after stopping this treatment (Fig. 5B). These additive antiprion effects led us to design a novel class of compounds covalently linking the acridine scaffold with the iminodibenzyl scaffold by a linker that corresponded to the combined aliphatic side chains of both compounds (Dollinger, Klingenstein, Löber, Korth, and Gmeiner, manuscript in preparation). The resulting novel antiprion lead compound, termed quinpramine, exhibited a fivefold increase in antiprion potency over quinacrine with an EC50 of around 85 nm (Fig. 5c). Of note, a synthesized compound linking two iminodibenzyl scaffolds (RK-01), albeit with a different linker, did not exhibit antiprion effects in the same concentration range (Fig. 5c). Quinpramine had similar effects on cholesterol redistribution and PrP dislocation from DRMs when ScN2a cells or WAC-2 cells were examined (data not shown). Quinpramine is thus a novel heterocyclic antiprion lead compound.

Figure 5.

 Western blot of protease-digested ScN2a cell lysates depicting the presence or absence of prions after a 1-week treatment with (a) quinacrine, desipramine and a combination of both after 1 week of treatment. Sc, untreated cells as negative control. Single compound concentrations are indicated below. A combination of both compounds shows antiprion activity at concentrations at which the single compounds do not abolish the prion infection. The lack of reappearance of protease-resistant PrPSc after discontinuation of treatment for 3 weeks (b) demonstrates that prions have been cleared from this population of ScN2a cells. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm). (c) Treatment with the same concentrations of quinacrine, quinpramine, a novel molecule covalently linking quinacrine and desipramine, as well as a bis-iminodibenzyl derivative, RK-01. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm).

A cocktail of approved drugs against Creutzfeldt-Jakob disease

Simultaneous administration of a combination of drugs acting on different targets is an efficient pharmacological strategy in the treatment of many infectious diseases. We investigated drug combinations including quinacrine, desipramine and other clinically approved drugs with antiprion activity. Lovastatin, a 3-hydroxy-3 methylglutaryl (HMG)-CoA reductase inhibitor, has been described to have antiprion effects through interference with proper formation of cholesterol-rich DRM domains necessary for prion replication (Taraboulos et al. 1995). Simvastatin has improved BBB permeability compared to lovastatin, which had been reported to act favourably on Aβ-levels in vitro (Fassbender et al. 2001), and patients with Alzheimer's disease (Simons et al. 2002). We reasoned that if the antiprion effects of acridine and iminodibenzyl derivatives had a different molecular target than HMG-CoA reductase, simvastatin could constitute a third drug with a non-overlapping antiprion mechanism. Simvastatin alone showed antiprion activity at concentrations of 1 μm(Fig. 6a). A combination pharmacotherapy of ScN2a cells with quinacrine, desipramine, and simvastatin at concentrations at which each compound alone or two of them had no complete antiprion activity, revealed synergistic antiprion effects (Fig. 6B).

Figure 6.

 Western blot of protease-digested ScN2a cell lysates depicting the presence or absence of prions after a 1-week treatment with (a) simvastatin alone. Dosage for antiprion effects with simvastatin was 1 μm. Single compound concentrations are indicated. The right part shows the molecular structure of simvastatin. Control cells (Sc) were untreated or treated with quinacrine (Q, 1 μm). (b) A triple combination pharmacotherapy including quinacrine (Q), desipramine (D) and simvastatin (S). The concentration of each compound is indicated. A dosage combination with 0.1 μm quinacrine, 1 μm desipramine and 0.25 μm simvastatin was effective in clearing PrPSc.

Discussion

The discovery of the heterocyclic compounds as potent antiprion substances (Doh-ura et al. 2000; Korth et al. 2001) has led to questions about their mechanism of action. The distinct structure–activity relationships of several classes of heterocyclic compounds in previous studies and in this paper (Table 1) argue for specificity of the molecular antiprion effect. Our studies suggest that the molecular mechanism of their antiprion activity can be sufficiently explained by their destabilization of DRMs, a membrane-resident subcellular compartment mandatory for prion conversion (Taraboulos et al. 1995; Vey et al. 1996; Kaneko et al. 1997). This does not formally exclude other, parallel mechanisms, for example the direct interaction of heterocyclic compounds with PrP as previously claimed (Vogtherr et al. 2003). Three arguments based on our data support our claim that heterocyclic compounds exert their antiprion effects by destabilizing DRMs:

(i) Dose-dependent changes in cholesterol redistribution parallel antiprion effects (Fig. 2).

(ii) Antiprion effective dosages of heterocyclic compounds lead to down-regulation of PrP in DRM fractions (Fig. 4).

(iii) Structurally unrelated cholesterol-redistributing compounds also possess antiprion potency (Fig. 2).

In our experiments, U18666A showed more prominent cholesterol redistribution than quinacrine or desipramine at antiprion effective dosages, which explains why heterocyclic drugs have not primarily been used to manipulate lipid metabolism. The stronger effect of U18666A on cholesterol redistribution and metabolism can be explained by the fact that U18666A influences plasma membrane cholesterol distribution in two different ways simultaneously: first, by an unknown mechanism it induces cholesterol redistribution leading to a lipid storage disease phenotype in cells (Patterson et al. 2001), and, second, it also inhibits endogenous cholesterol biosynthesis as an inhibitor of oxidosqualene cyclase (Sexton et al. 1983) blocking de novo cholesterol synthesis and resulting in a very effective disruption of cholesterol metabolism. However, from a pharmacological point of view, a weaker effect on cholesterol redistribution with sufficient antiprion activity may be desirable to avoid serious side effects. Thus, heterocyclic compounds might be seen as ‘soft’ cholesterol redistributors with a higher therapeutic potential for treating humans.

From the present data and previous work of others, it seems that DRMs are key to a proper function and conversion of PrP: caveolae-mediated PrPC uptake has been shown to be dependent on cholesterol because filipin was able to decrease PrP immunoreactivity in caveolae (Peters et al. 2003). Filipin could also be used to inhibit conversion of PrPC to PrPSc (Marella et al. 2002). Efficient antiprion activity has been shown for the statins, a class of pharmaceuticals which suppress endogenous cholesterol synthesis by HMG-CoA-reductase inhibition (Taraboulos et al. 1995). As heterocyclic compounds affect redistribution of cholesterol, albeit with different efficiencies and slightly different effects on lysosomal morphology, combining two heterocyclic antiprion compounds like quinacrine and desipramine resulted in additive, overlapping antiprion effects (Fig. 5). For example, the antiprion potency of quinacrine in vitro is 10 times higher than that of desipramine (Table 1), and similar antiprion effects could be reached by adding homologous doses when combining both drugs (data not shown). However, combining the heterocyclic antiprion compounds with a statin revealed true synergistic effects due to simultaneous interference of PrP conversion by two non-overlapping pharmacological mechanisms. The pharmacological strategy of simultaneously attacking different targets has been proven to be very successful for efficient pharmacotherapy of many infectious diseases. We destabilized DRMs simultaneously by inducing redistribution of cholesterol into lysosomes using the heterocyclic compounds quinacrine and desipramine, and by suppressing cellular cholesterol synthesis with simvastatin. This parallel approach made it possible to combine these drugs, at concentrations at which all compounds were ineffective when applied separately, into a cocktail that had full antiprion activity, when administered at the same time (Fig. 6). Inducing a transient lipid redistribution by 1 week of therapy seemed sufficient to cure ScN2a cells because misfolded PrPSc did not reappear after halting the treatment (Fig. 2A), suggesting that even a temporary disturbance of DRMs has long lasting effects on the cell's ability to clear misfolded or misprocessed proteins. The present combination therapy, or ‘antiprion cocktail’, which acts by transiently disturbing cellular cholesterol metabolism and conversion-mandatory cellular compartments, thus presents a novel pharmacological option for treating patients with Creutzfeldt-Jakob disease. The most promising drug combination for future in vivo studies consists of quinacrine, desipramine and simvastatin (Fig. 6). For the tricyclic antidepressants, as well as for quinacrine (Yung et al. 2004) and simvastatin (Saheki et al. 1994; Fassbender et al. 2001), BBB permeability has clearly been shown, a precondition of a successful in vivo pharmacotherapy for Creutzfeldt-Jakob disease.

A link between cellular cholesterol distribution and protein misfolding or misprocessing has previously been shown for Alzheimer's disease where the important extracellular amyloid precursor protein processing step to amyloidogenic Aβ by β-secretase was demonstrated to be dependent on the intactness of DRMs (Cordy et al. 2003; Ehehalt et al. 2003). Similar to their proposed antiprion effect, the antiamyloidogenic effects of the statins in treating Alzheimer's disease were thought to be a result of sorting DRM-resident β-secretase into non-functional cell compartments (Fassbender et al. 2001; Ehehalt et al. 2003).

Our strategy of combining two effective heterocyclic antiprion compounds into one molecule also proved very efficient (Fig. 5c). The compound consisting of covalently linked acridine and iminodibenzyl moieties, termed quinpramine, exhibited fivefold higher antiprion potency than quinacrine, making it a novel lead heterocyclic compound. Linking two acridine moieties to bis-acridines has been successfully used previously to increase antiprion potency (May et al. 2003). However, BBB permeability of these compounds and hence applicability in vivo has not been shown yet. Quinpramine containing the highly BBB-permeable iminodibenzyl moiety is likely to fulfil this requirement for future in vivo pharmacotherapy.

Acknowledgements

This work has been supported by a grant of the Bundesministerium für Forschung und Bildung (BMBF), Germany, to CK, a grant from the Volkswagenstiftung, Germany to CK and PG, and a grant from the European Research Commission QLK2-CT-2002–81628 to PJP.

Ancillary