|Home | About | Journals | Submit | Contact Us | Français|
Branched polyamines, including polyamidoamine and polypropyleneimine (PPI) dendrimers, are able to purge PrPSc, the disease-causing isoform of the prion protein, from scrapie-infected neuroblastoma (ScN2a) cells in culture (S. Supattapone, H.-O. B. Nguyen, F. E. Cohen, S. B. Prusiner, and M. R. Scott, Proc. Natl. Acad. Sci. USA 96:14529–14534, 1999). We now demonstrate that exposure of ScN2a cells to 3 μg of PPI generation 4.0/ml for 4 weeks not only reduced PrPSc to a level undetectable by Western blot but also eradicated prion infectivity as determined by a bioassay in mice. Exposure of purified RML prions to branched polyamines in vitro disaggregated the prion rods, reduced the β-sheet content of PrP 27-30, and rendered PrP 27-30 susceptible to proteolysis. The susceptibility of PrPSc to proteolytic digestion induced by branched polyamines in vitro was strain dependent. Notably, PrPSc from bovine spongiform encephalopathy-infected brain was susceptible to PPI-mediated denaturation in vitro, whereas PrPSc from natural sheep scrapie-infected brain was resistant. Fluorescein-labeled PPI accumulated specifically in lysosomes, suggesting that branched polyamines act within this acidic compartment to mediate PrPSc clearance. Branched polyamines are the first class of compounds shown to cure prion infection in living cells and may prove useful as therapeutic, disinfecting, and strain-typing reagents for prion diseases.
Prion diseases are caused by an infectious protein (20, 25). These invariably fatal illnesses cannot be cured using routine antimicrobial agents, and materials contaminated with prions cannot be disinfected by conventional methods. Therefore, it is important to identify compounds that can be used either as therapeutic or disinfecting reagents for prion diseases. Ongoing epidemics of new variant Creutzfeldt-Jakob disease and bovine spongiform encephalopathy (BSE) in the United Kingdom highlight the urgency of this task.
We recently reported that branched polyamines could purge scrapie-infected neuroblastoma (ScN2a) cells of PrPSc, the disease-causing isoform of the prion protein (33). The ability of these compounds to eliminate PrPSc from ScN2a cells depended upon a highly branched structure and a high surface density of primary amino groups. The most potent compounds identified were generation 4.0 polyamidoamine (PAMAM) and polypropyleneimine (PPI) dendrimers. Dendrimers are branched polyamines manufactured by a repetitive divergent growth technique, allowing the synthesis of successive, well-defined “generations” of homodisperse structures. In the current study, we demonstrate that branched polyamines cure prion-infected cells and identify the site and mechanism of polyamine-mediated prion clearance. We also demonstrate that these compounds can be employed in a rapid and simple assay to discriminate between different prion strains in vitro.
High-molecular-weight polyethyleneimine (PEI) was purchased from Fluka. SuperFect transfection reagent was purchased from Qiagen. All other polyamines were purchased from Sigma-Aldrich. Fluorescein-labeled PPI was synthesized by mixing 30 mg of fluorescein isothiocyanate (FITC) with 1 mg of PPI generation 4.0 in 2 ml of ethanol overnight at 4°C. Labeled PPI was separated from residual, unreacted FITC using a Sephadex P-2 column.
Cultures of ScN2a cells were maintained as described previously (33). Cytotoxicity after treatment with polyamines was assessed in ScN2a cells by the following four methods: (i) examination of morphology under phase contrast microscopy, (ii) observation of growth curves and cell counts for 3 weeks after treatment, (iii) vital staining of living cells with 0.4% trypan blue (Sigma-Aldrich), and (iv) assay of dehydrogenase enzymes with 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich). For ScN2a cells treated with either PAMAM or PPI generation 4.0 continuously for 1 week, the 50% toxic dose was ~50 μg/ml.
To prepare samples for infectivity assays, 100-mm-diameter plates (Falcon) of confluent cells were washed three times with 5 ml of phosphate-buffered saline, scraped into 2 ml of phosphate-buffered saline, and homogenized by repeated extrusion through a 26-gauge needle. Prion infectivity was determined by intracerebral inoculation of 30 μl of cell homogenate into Tg(MoPrP)4053 mice. Mice were observed for clinical signs of scrapie, and a subset of diagnoses were confirmed by neuropathological examination. Samples were prepared for sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis as described previously (33).
Brain homogenates were prepared as described previously (33). Except for Tg(BoPrP) samples, 50 μl of 1-mg/ml brain homogenate was mixed with 450 μl of 1% NP-40–50 mM sodium acetate (pH 3.0) (final measured pH = 3.6) plus or minus 60 μg of PPI generation 4.0/ml and shaken constantly for various periods at 37°C.
Purified prions were prepared as described previously (21), utilizing both proteinase K digestion and sucrose gradient sedimentation, and resuspended in 1% NP-40–1-mg/ml bovine serum albumin (BSA). For pH studies, 475 μl of 0.5-μg/ml purified RML PrP 27-30 in 1% NP-40–1-mg/ml BSA was mixed with 25 μl of 1 M buffers from pH 3 to 8 (sodium acetate for pHs 3 to 6 and Tris acetate for pHs 7 and 8) plus or minus 60 μg of PPI generation 4.0/ml for 2 h at 37°C with constant shaking. The final pH value of each sample was measured directly with a calibrated pH electrode (Radiometer Copenhagen). For compound screening, 475 μl of 0.5-μg/ml purified RML PrP 27-30 in 1% NP-40–1-mg/ml BSA was mixed with 25 μl of 1 M sodium acetate (pH 3.0) plus 60 μg of polyamine/ml for 2 h at 37°C with constant shaking.
Following incubations, each sample was neutralized with an equal volume of 0.2 M HEPES (pH 7.5) containing 0.3 M NaCl and 4% Sarkosyl. Samples not treated with proteinase K were mixed with an equal volume of 2× SDS sample buffer. For proteinase K digestion, samples were incubated with 20 μg of proteinase K (Boehringer Mannheim)/ml for 1 h at 37°C. Proteolytic digestion was terminated by the addition of 8 μl of 0.5 M phenylmethylsulfonyl fluoride. Digested samples were then mixed with equal volumes of 2× SDS sample buffer. All samples were boiled for 5 min prior to SDS-polyacrylamide gel electrophoresis. Western blotting was performed as previously described (27), using human-mouse chimeric Fab D13.
The mixture protocol was modified for Tg(BoPrP) samples. Incubations were carried out at room temperature, and deoxycholate was substituted for Sarkosyl in the neutralization buffer. Protease-treated samples were concentrated 10-fold by centrifugation for 1 h at 100,000 × g. Immunoblotting was performed with Fab clone P as the primary antibody to recognize bovine PrP.
Samples were lyophilized and resuspended in D2O. Prior to the spectroscopic measurements, the samples were centrifuged briefly (14,000 × g for 2 min), and 1.5-μl samples from the bottom of the tube were enclosed between 2 AgCl windows (International Crystal Laboratories, Garfield, N.J.), creating a path length of 50 μm. Spectra were recorded with a Perkin-Elmer (Norwalk, Conn.) System 2000 Fourier transform infrared resonance (FTIR) spectrophotometer. Blank controls identical in buffer conditions and PPI content were used to subtract any nonprotein contributions from the spectra. Spectral analysis and self-deconvolution were carried out as previously described (6) and modified (15).
Confocal images were obtained using a Bio-Rad (Hercules, Calif.) laser scanning confocal microscope (MRC-1024) outfitted with a Nikon Diaphot 200 microscope and a helium-neon laser. Laser power was set at 10% and scanned with a slow speed across the sample. Individual laser lines confirmed the lack of “bleed-through” between detection channels. The images were averaged with a Kalman filter (n = 4).
ScN2a cells were incubated with 3 μg of PPI/ml in supplemented Dulbecco's modified Eagle's medium for 4 weeks and then cultured for an additional 2 weeks in polyamine-free medium. This transient exposure to PPI was not cytotoxic (see Materials and Methods) and completely purged the cells of protease-resistant PrPSc (Fig. (Fig.1A,1A, lanes 2 and 4). In contrast, protease-sensitive PrPC bands migrating between 32 and 38 kDa appear to be present at similar levels in cells treated with PPI (lane 2) and in uninfected N2a cells (data not shown). Elimination of PrPSc was measured by the disappearance of the protease-resistant core of PrPSc, denoted PrP 27-30. The elimination of PrP 27-30 appeared to be relatively specific, since the steady-state levels of proteins in PPI-treated cells were similar to those in control ScN2a cells (Fig. (Fig.1B).1B). To assess the effect of PPI treatment on prion infectivity, homogenates prepared from polyamine-treated and control ScN2a cells were inoculated into Tg(MoPrP)4053 mice. The average scrapie incubation time was 61 ± 2 days for mice inoculated with control ScN2a cells and >200 days for mice inoculated with ScN2a cells treated with PPI (n/n0 = 0/13) (Fig. (Fig.1C).1C). These incubation times indicate that the titer of infectious prions in ScN2a cells was reduced from ~106 50% infective dose (ID50) units/100-mm plate to <102 ID50 units/plate by PPI treatment (Fig. (Fig.1D).1D). Thus, exposure to PPI eliminates measurable prion infectivity from ScN2a cells.
Having established that branched polyamines reduce prion infectivity, we sought to identify the mechanism by which these compounds eliminate PrPSc. Our first objective was to determine the molecular target of branched polyamines. Previously, we developed an in vitro assay which was used to show that these compounds could render PrPSc protease susceptible when mixed directly with crude brain homogenates (33). We performed a similar assay with PrP 27-30 purified from mouse brains infected with RML prions to determine whether or not the molecular target of branched polyamines was present in this highly purified preparation. PrP 27-30 in purified preparations of RML prions was rendered protease sensitive by branched polyamines with a similar acidic pH optimum (Fig. (Fig.2A)2A) and structure-activity profile (Fig. (Fig.2B)2B) as previously obtained in crude brain homogenates (33). Treatment of purified prions with branched polyamines in vitro also diminished infectivity. We incubated 15 μg of RML prion rods per ml in 50 mM sodium acetate (pH 3.0)–1% NP-40–1-mg/ml BSA for 2 h at 37°C, with or without 60 μg of PPI generation 4.0/ml, and measured prion infectivity using a scrapie prion incubation time assay in Tg(MoPrP)4053 mice. PPI treatment reduced prion infectivity from 107 ID50 units/ml to 105 ID50 units/ml (data not shown).
Although the PrP sequence is well conserved among mammals, a small number of amino acid substitutions retard prion transmission across species (27). Furthermore, prions can exist as different phenotypic strains that yield distinct incubation times, neuropathology, and distribution of PrPSc upon infection of susceptible hosts. In certain cases, these phenotypic differences can be correlated with differences in the conformation of PrPSc (3, 26, 29, 37). We sought to determine whether different species and strains of rodent prions, which presumably contain different conformations of PrPSc, vary in their susceptibility to branched polyamines. Homogenates were prepared from the brains of rodents infected with one of several Syrian hamster (SHa designations), mouse (Mo designations), or artificial prion strains. Individual samples were mixed with 60 μg of PPI generation 4.0/ml in vitro for 2 h at 37°C, neutralized, and subjected to limited proteolysis. The results indicate that susceptibility to the PPI dendrimer is dependent on both the prion strain and PrP sequence (Fig. (Fig.3A).3A).
The varying susceptibility of different strains is most clearly illustrated by the six mouse strains analyzed (paired lanes 7 to 12). Mo(RML), Mo(22a), and Mo(139A) were susceptible to PPI-induced conformational change (paired lanes 7, 9, and 11, respectively). In contrast, Mo(Me7) and Mo(87V) were resistant (paired lanes 8 and 10, respectively) and Mo(C506) was marginally susceptible to PPI-induced conformational change (paired lanes 12).
The effect of PrP sequence can be seen by comparing the relative susceptibilities of SHa(RML), MH2M(RML), and Mo(RML). Whereas Mo(RML) was susceptible to PPI-induced conformational change (paired lanes 7), SHa(RML) was resistant (paired lanes 4). MH2M(RML) displayed an intermediate level of susceptibility to PPI (paired lanes 5); MH2M is a chimeric PrP molecule in which amino acids 94 to 188 of the mouse sequence have been replaced by the corresponding Syrian hamster residues (28). Thus, SHaPrPSc appears to be more resistant to PPI-induced conformational change than MoPrPSc.
We investigated whether the varying susceptibilities to PPI displayed by different strains and species of prions might be caused by kinetic differences. To test this possibility, we incubated samples of each prion isolate with PPI generation 4.0 for various periods of time. Even after incubation with PPI for 3 days, PrPSc in samples of resistant isolates did not become more susceptible to protease digestion (Fig. (Fig.3B).3B). Thus, the differences in susceptibilities of different prion strains and sequences are not caused simply by differences in the kinetics of PrPSc unfolding.
Recently, it was demonstrated that Tg(BoPrP)Prnp0/0 mice were susceptible to both BSE and natural sheep scrapie (30). Furthermore, these two prion strains remain distinct during passage through Tg(BoPrP)Prnp0/0 mice (30). These transgenic mice therefore provided an opportunity to compare the susceptibility of BSE and scrapie prions to branched polyamines. We incubated brain homogenates from Tg(BoPrP)Prnp0/0 (BSE) and Tg(BoPrP)Prnp0/0 (sheep scrapie) mice with PPI generation 4.0 in vitro. The PrPSc in scrapie-infected Tg mice was not susceptible to PPI-induced conformational change (Fig. (Fig.3C,3C, lanes 1 and 2). In contrast, >90% of the PrPSc in BSE-infected Tg mice was rendered protease sensitive by treatment with PPI (Fig. (Fig.3C,3C, lanes 3 and 4).
The existence of prion strains resistant to branched polyamines suggests that PrPSc molecules in these strains might exist in conformations which are more resistant to denaturation than PrPSc molecules in polyamine-susceptible strains. To test this hypothesis, we examined the effect of adding urea to SHa(Sc237) brain homogenate treated with and without PPI generation 4.0. In the presence of urea, PrPSc was more susceptible to protease digestion in samples treated with PPI, whereas no difference in protease resistance could be detected in the absence of urea (Fig. (Fig.4A).4A). Thus, additional denaturation enables PrPSc molecules in a resistant strain to become susceptible to branched polyamines. This result suggests that the general mechanism of action of branched polyamines might be to assist PrPSc denaturation. Consistent with this concept, branched polyamines render PrPSc protease sensitive more efficiently at lower pH values (Fig. (Fig.2A).2A). Furthermore, polyamine-treated PrPSc did not regain protease resistance after prolonged neutralization (Fig. (Fig.4B)4B) or dialysis (data not shown). Finally, we excluded the possibility that acidification might be required only to activate the dendrimer by demonstrating that preacidified PPI generation 4.0 could not render PrPSc protease sensitive at a neutral pH (Fig. (Fig.4C).4C).
To visualize the effect of branched polyamines on prions, we examined the ultrastructure of purified prion rods treated in vitro with PPI generation 4.0. By electron microscopy, RML prion rods were disaggregated after incubation for 2 h at 37°C with PPI (Fig. (Fig.5B).5B). In contrast, SHa(Sc237) PrP 27-30 rods remained intact after treatment with PPI (Fig. (Fig.5D).5D). Disaggregation of purified RML prion rods by treatment with PPI was accompanied by a loss of β-sheet secondary structure, as judged by FTIR spectroscopy. Whereas control rods were 57% β-sheet, 25% α-helix, 7% β-turn, and 11% random coil, PPI-dissociated PrPSc was 47% β-sheet, 25% α-helix, 15% β-turn, and 13% random coil (Fig. (Fig.5E).5E).
To investigate further the mechanism of polyamine-induced disaggregation of PrP 27-30, we performed a kinetic study in vitro using purified mouse RML prion rods and various concentrations of PPI. The results indicate that polyamine-induced PrPSc disaggregation is not a catalytic process and requires a stoichiometry of approximately one PPI molecule per five PrP 27-30 molecules in purified RML prion preparations (data not shown).
Branched polyamines apparently require acidic conditions to render PrPSc protease sensitive when mixed with brain homogenates or purified prions in vitro (Fig. (Fig.2A).2A). However, these compounds successfully cure living cells of prion infection when added to culture media buffered at pH 7.4 (Fig. (Fig.1).1). One possible explanation for this discrepancy is that branched polyamines might localize with prions within an acidic intracellular compartment. PrPSc has previously been shown to accumulate in lysosomes (35). Therefore, we sought to determine whether branched polyamines localize to this same compartment. We incubated N2a cells with fluorescein-labeled PPI and LysoTracker Red and performed dual channel confocal microscopy to compare the localization of the two compounds. Our results indicate that fluorescein-labeled PPI accumulates in the lysosomes of living cells (Fig. (Fig.6).6).
A major finding of this study is that branched polyamines eliminate prion infectivity from living cells that were chronically infected. To our knowledge, this is the first class of compounds shown to cure an established prion infection. Polyene antibiotics, anionic dyes, sulfated dextrans, anthracylines, porphyrins, phthalocyanines, dapsone, and a synthetic β-breaker peptide all prolong scrapie incubation times in vivo but only if administered prophylactically (1, 11, 12, 16–19, 31, 34).
The unique ability of branched polyamines to cure an established prion infection in cells suggests that these compounds might also reverse disease progession in animals. However, two factors could potentially limit the use of these compounds as therapeutic reagents against prion diseases. One potential limitation is that branched polyamines might not act on all strains of prions in vivo. This possibility is shown by in vitro studies in which some strains and species of prions were more resistant than others to branched polyamine-induced disaggregation (Fig. (Fig.3).3). It remains to be determined whether prion strains resistant to branched polyamine-induced disaggregation in vitro would also be resistant to treatment by these compounds in vivo. Treatment of more resistant strains might require therapy with branched polyamines in combination with another class of prion-directed compounds. Significantly, PPI demonstrates substantial in vitro activity against BSE (Fig. (Fig.33C).
The second potential limitation of branched polyamines is that these highly charged compounds might not cross the blood-brain barrier. If this proves to be the case, it may be possible to deliver branched polyamines directly to the cerebrospinal fluid through an intraventricular reservoir or perhaps to synthesize them as prodrugs capable of crossing the blood-brain barrier. Preliminary studies indicate that continuous intraventricular infusion of PPI generation 4.0 is tolerated by FVB mice up to a total dose of approximately 0.5 mg/animal (data not shown). Further studies are required to characterize the biodistribution of dendrimers and to optimize their delivery to prion-infected neurons in vivo.
The ability of branched polyamines to render PrPSc sensitive to proteolytic digestion in purified prion preparations (Fig. (Fig.2)2) suggests that the molecular target of these compounds must be either (i) PrPSc itself, (ii) an acid-induced unfolding intermediate of PrPSc, or (iii) a very tightly bound, cryptic molecule which copurifies with PrPSc. Cross-linking experiments indicate that photoaffinity-labeled PPI generation 4.0 binds PrP 27-30 avidly (data not shown), but unfortunately these results cannot prove conclusively that PrP is the molecular target of branched polyamines. If the molecular target is PrP, at least one of the polyamine binding sites must be contained within the amino acid sequence of the PrP106 deletion mutant (32), since PPI renders PrPSc106 protease sensitive (Fig. (Fig.3A,3A, lane 6). The 106 amino acids present in PrP106 are residues 89 to 140 and 177 to 231. PPI also renders a spontaneously protease-resistant, 61-amino-acid-long PrP deletion mutant, PrP(Δ23–88,Δ141–221), susceptible to protease digestion (33a), further delimiting the boundaries of at least one putative binding site to residues 89 to 140 and 222 to 231.
Several lines of evidence suggest that branched polyamines render PrPSc molecules protease sensitive by dissociating PrPSc aggregates. (i) RML prion rods treated in vitro with PPI become disaggregated, as judged by electron microscopy (Fig. (Fig.5).5). (ii) Prion strains resistant to branched polyamines in vitro appear to be more amyloidogenic than polyamine-susceptible strains, as judged by neuropathology (4, 7, 13, 14). (iii) The ability of branched polyamines to render PrPSc protease sensitive in vitro is enhanced by conditions which favor PrPSc disaggregation. These conditions include acidic pH (Fig. (Fig.2A)2A) and the presence of urea (Fig. (Fig.44A).
Theoretically, it is possible that the mechanism by which branched polyamines remove PrPSc and prion infectivity from ScN2a cells does not relate to the ability of these compounds to disaggregate prions in vitro. However, this is unlikely because the relative potency of 14 different polyamines in eliminating PrPSc from ScN2a cells correlates with the relative ability of these same compounds to render PrPSc sensitive to proteolysis in crude brain homogenates and purified preparations of RML PrP 27-30 in vitro (33) (Table (Table1).1). The structure-activity profile obtained from these studies indicates that polyamines become more potent at eliminating PrPSc as they become more branched and possess more surface primary amines. With PPI dendrimers, this effect reaches a plateau at the fourth generation; PPI generation 5.0 is no more potent than PPI generation 4.0 at either removing PrPSc from cells or rendering PrPSc protease sensitive in vitro. Homodisperse, uniform PPI and PAMAM dendrimers were more potent than the heterogeneous preparations of PEI or SuperFect, a heat-fractured dendrimer.
We determined that the process by which PPI renders PrPSc protease sensitive in vitro was not catalytic. Instead, this process appeared to require a fixed stoichiometric ratio of PPI to PrPSc of approximately 1:5. How could PPI denature prion rods stoichiometrically? One possible explanation is that individual amino groups on the surface of PPI might bind to PrPSc monomers or oligomers that exist in equilibrium with a large aggregate under acidic conditions. The dendrimer might then pry bound PrPSc molecules apart from the aggregate and/or prevent such molecules from reaggregating.
Another possible mechanism of polyamine-induced prion clearance from ScN2a cells is that branched polyamines facilitate PrPSc transport from the plasma membrane through the endocytic pathway into secondary lysosomes. Several lines of evidence indicate that the cellular site of action of branched polyamines is secondary lysosomes. (i) Fluorescein-tagged PPI and PrPSc both localize to lysosomes (8, 35) (Fig. (Fig.6).6). (ii) The pH optimum of PrPSc denaturation in vitro is <5.0. When cultured cells were studied with fluorescent acidotropic pH measurement dyes, secondary lysosomes were the most acidic cellular compartment detected, with pH values of ~4.4 to 4.5 (2, 10). (iii) The lysosomotropic agent chloroquine attenuates the ability of branched polyamines to eliminate PrPSc from ScN2a cells (33). Our studies raise the possibility that lysosomal proteases normally degrade PrPSc in prion-infected cells at a slow rate and that polyamines accelerate this process by denaturing PrPSc.
Beyond their potential use as therapeutic agents and research tools, branched polyamines might also be useful as prion strain-typing reagents and/or prion decontaminants. Presently, typing of prion strains is time-consuming and requires the inoculation of samples into several strains of inbred animals to obtain incubation time and neuropathology profiles (4, 9). Recently, antibody-based PrPSc conformational stability assays able to distinguish prion strains have been developed (26) (D. Peretz, unpublished data). In this study, we observed that different species and strains of prions displayed varying susceptibilities to branched polyamine-induced denaturation in vitro (Fig. (Fig.3).3). These results suggest that a polyamine-based in vitro protease digestion assay could, in principle, be used as a simple and rapid diagnostic method for prion strain typing. One practical application which arises from our results is that a polyamine-based assay could be used to distinguish between BSE and natural scrapie in flocks of domestic sheep.
Currently, it is very difficult to remove prions from skin, clothes, surgical instruments, foodstuffs, and surfaces (5). Standard prion decontamination requires either prolonged autoclaving or exposure to harsh protein denaturants such as 1 N NaOH or 6 M guanidine thiocyanate (24). Branched dendrimers are nontoxic and relatively inexpensive. These compounds may therefore be suitable for use as disinfecting reagents to limit the commercial and iatrogenic spread of prion diseases.
This work was supported by grants from the National Institutes of Health (NS14069, AG02132, and AG10770) and by a gift from the Leila and Harold Mathers Foundation. S.S. was supported by the Burroughs Wellcome Fund Career Development Award and by an NIH Clinical Investigator Development Award (K08 NS02048-02).