PrPSc Incorporation to Cells Requires Endogenous Glycosaminoglycan Expression*

Many lines of evidence suggest an interaction between glycosaminoglycans (GAGs) and the PrP proteins as well as a possible role for GAGs in prion disease pathogenesis. In this work, we sought to determine whether the PrP-GAG interaction affects the incorporation of PrPSc (the scrapie isoform of PrP) to normal cells. This may be the first step in prion disease pathogenesis. To this effect, we incubated proteinase K-digested hamster scrapie brain homogenates with several lines of Chinese hamster ovary (CHO) cells in the presence or absence of heparin. Our results show that over a large range of PrPSc concentrations the binding of PrPSc to wild type CHO cells, which do not express detectable PrP, was equivalent to the binding of PrPSc to CHO cells overexpressing PrP. A significant part of PrPSc binding to both lines could be inhibited by heparin. Additional evidence that PrPSc binding to cells was dependent on the presence of GAGs could be concluded from the fact that the binding of PrPSc to CHO cells missing GAGs on the cell surface was significantly reduced. Interestingly, preincubation of scrapie brain homogenate with heparin before intraperitoneal inoculation into normal hamsters resulted in a significant delay in prion disease manifestation.

Many lines of evidence suggest an interaction between glycosaminoglycans (GAGs) and the PrP proteins as well as a possible role for GAGs in prion disease pathogenesis. In this work, we sought to determine whether the PrP-GAG interaction affects the incorporation of PrP Sc (the scrapie isoform of PrP) to normal cells. This may be the first step in prion disease pathogenesis. To this effect, we incubated proteinase K-digested hamster scrapie brain homogenates with several lines of Chinese hamster ovary (CHO) cells in the presence or absence of heparin. Our results show that over a large range of PrP Sc concentrations the binding of PrP Sc to wild type CHO cells, which do not express detectable PrP, was equivalent to the binding of PrP Sc to CHO cells overexpressing PrP. A significant part of PrP Sc binding to both lines could be inhibited by heparin. Additional evidence that PrP Sc binding to cells was dependent on the presence of GAGs could be concluded from the fact that the binding of PrP Sc to CHO cells missing GAGs on the cell surface was significantly reduced. Interestingly, preincubation of scrapie brain homogenate with heparin before intraperitoneal inoculation into normal hamsters resulted in a significant delay in prion disease manifestation.
During prion disease, an abnormal protease-resistant isoform of cellular PrP (PrP C ), 1 denominated PrP Sc , which is considered to be the major component of the prion, accumulates in the central nervous system (1). It has been postulated that ␤-sheet-rich PrP Sc is produced from mostly ␣-helical PrP C by a conformational-dependent conversion process (2). Although the detailed steps of such conversion as well as the mechanisms of infection and pathogenesis leading to death from prion disease are as yet unknown, the absolute requirement of endogenous PrP C in prion disease pathogenesis has been solidly demonstrated (3).
Studies in transgenic mice suggest that a host factor, designated protein X, is involved in the conversion of the normal cellular prion protein, PrP C , into the scrapie isoform, PrP Sc (4). Consequently, considerable effort has been devoted to the identification of natural receptors for the prion proteins, both to understand the prion pathogenesis process and to reveal potential targets for therapeutic intervention. Among the proteins that were shown to bind PrP C were amyloid precursorlike protein 1 (5), stress-induced proteins, and the laminin receptors (6,7). Interestingly, antibodies to laminin receptors were shown to reduce the accumulation of PrP Sc in ScN2a cells (8). Both PrP and laminin receptors have been shown to interact with glycosaminoglycans (GAGS), and especially with heparan sulfate (HS), supporting the possibility that these sugar polymers function as possible cofactors for 37-kDa laminin receptors mediating PrP binding (9).
In addition to PrP C , GAGs are by far the most important host molecules connected so far to prion pathogenesis and to the metabolism of the prion proteins (10). HS accumulates in cerebral prion amyloid plaques as it does in Alzheimer disease or other amyloidotic diseases (11,12). Also, addition of HS promotes the reconstitution of infectivity of dimethyl sulfoxidesolubilized PrP Sc (13). In addition, several lines of evidences connect GAGs, and especially HS, to the metabolism of the PrP isoforms as well as to the accumulation of PrP Sc in prioninfected cells. A variety of sulfated glycans, such as low molecular weight heparin (14), suramin (15), pentosan polysulfate, and dextran sulfate (16,17), were shown to reduce the accumulation of PrP Sc in ScN2a cells and in some cases prolong the incubation time of experimental prion diseases. In addition, long-term incubation of Scn2a cells with chlorate, which affects the sulfation of GAGs as well as of proteins and glycolipids, reduces PrP Sc accumulation (14). Finally, ScN2a cells treated with heparinase III showed a significant decrease in PrPSc accumulation (18).
In this work, we investigated whether endogenous GAGs play a role in the first step of prion disease pathogenesis, i.e. the binding of PrP Sc , the main if not the only prion component, to normal cells. We show here that, over a wide range of PrP Sc concentrations, the binding of PrP Sc to mutant cells in which expression of total GAGs or only HS was abolished was significantly reduced when compared with PrP Sc binding to wt Chinese hamster ovary (CHO) or to CHO cells overexpressing PrP. Most of this binding resulted in the internalization of PrP Sc to the cells as can be seen by the fact that only a small part of PrP Sc was bound to the cells at 4°C. PrP Sc binding to both wt CHO or to CHO cells overexpressing PrP was significantly inhibited when the cells were preincubated either with heparin or with copper. Because PrP Sc is an insoluble molecule, it is unclear whether it binds directly to the GAGs on the cell surface or through some closely associated molecule.

EXPERIMENTAL PROCEDURES
Heparin, polyaspartic acid, and heparin-agarose were purchased from Sigma. Cell culture media and serum were purchased from Beit Haemek, Biological Industries.
Preparation of Scrapie Brain Homogenates-Normal or scrapie-infected hamster brains were homogenized in 10% (wt/v) cold buffer containing 10 mM Tris-HCl, pH 7.4, 300 mM sucrose, and 5 mM EDTA in phosphate-buffered saline, pH 7.4. The homogenates were centrifuged at 800 ϫ g (2000 rpm) for 15 min at 4°C. The supernatants were frozen in aliquots for future experiments.
Binding Experiments-Binding experiments were performed as reported previously (22). Briefly, after reaching 80% of confluence in 25-cm 2 flasks, cells were preincubated with Hams F12 medium supplemented with or without 4 mg/ml heparin for 2 h at 37°C. After the preincubation, PK-digested (40 g/ml PK for 30 min at 37°) scrapie hamster brain homogenate at the different concentrations used was added for each flask to the designated flasks for an additional 2 h. Subsequently, the cells were washed four times with medium and twice with phosphate-buffered saline. Cells were lyzed in 10 mM Tris-HCL, pH 7.4,100 mM Nacl, 1% Nonidet P-40. PrP proteins were detected by immunoblotting with ␣PrP mAb 3F4, RO73, or D13 (Inpro) when required.
Binding of Brain Homogenates to Heparin-Agarose-Equal amounts of total protein (0.37 mg) from normal or scrapie hamster brain homogenate digested with PK (40 g/ml PK for 30 min at 37°), were incubated for 1 h with 25 or 50 g/ml of heparin or with 50 g/ml polyaspartic acid or with no additions. Subsequently, the different aliquots were incubated with heparin-agarose beads for 1 h at 4°before washing the beads five times with a buffer containing 10 mM Tris-HCL, pH 7.4, 100 mM Nacl, 1%Nonidet P-40. Beads were boiled in SDS to elute bound material that was then immunoblotted with ␣PrP mAb 3F4.
In Vivo Experiments-Syrian hamsters (six animals in each group) at the age of 4 weeks were inoculated intraperitoneally (IP) with 100 l (0.1%) of brain homogenate from either normal or scrapie-infected an-imals that were preincubated with or without heparin (2 or 4 mg/ml) for 2 h before injection. Animals were followed closely through the incubation time and sacrificed when showing distinct scrapie symptoms.

RESULTS
The Binding of PrP Sc to CHO Cells Is GAGs-dependent and PrP C -independent-Because PrP Sc can only be purified as an insoluble aggregate, which may not exist as such in vivo, we used PK-digested scrapie brain homogenate from Syrian hamster brains for our binding experiments. In this context, PrP Sc may be associated to components of membrane rafts (23). Therefore, we should be aware that the binding of PrP Sc to cells or heparin resins (see below) may be either direct or occur through other molecules. We have shown before (22) that this preparation can be used to study PrP Sc binding and internalization to cells. To determine the concentration of PK-resistant PrP in these preparations, we immunoblotted serial dilutions of the PK-treated brain homogenates as well as serial dilutions of purified recombinant hamster PrP and compared the appropriate bands (Fig. 1a). Recombinant Syria hamster (SHa)PrP (23-232) was produced as insoluble inclusion bodies, purified, and refolded as previously described (24). As can be seen in Fig.  1a, 1 g of purified recombinant PrP represents the PrP Sc present in ϳ45 g of total brain homogenate (after digestion with PK).
To examine the role of PrP C and cellular GAGs on the binding of PrP Sc to cells, we incubated PK-digested Syrian hamster scrapie brain homogenate (comprising ϳ7.5 g of PrP Sc ) with several types of CHO cells. These include wt CHO cells, which have been shown previously not to express detectable levels of endogenous Chinese hamster PrP (25) as well as CHO cells overexpressing MHM2 PrP (21). In addition, we used CHO (XTϪ/Ϫ) and (HSϪ/Ϫ) mutant cells, which do not express chondroitin sulfate and HS or only HS, respectively (for explanation FIG. 1. a, calibration of PrP Sc concentration in scrapie-infected brain homogenates. Scrapie-infected hamster brain homogenate, as well as purified recombinat hamster PrP, were serially diluted before immunoblotting with ␣PrP mAb 3F4. The numbers represent the amounts of recombinant PrP or of total protein in the brain homogenates. b, the binding of PrP Sc to CHO cells depends on the presence of endogenous GAGs. I, wt, XTϪ/Ϫ, HSϪ/Ϫ, and MHM2 PrP CHO cells were extracted and immunoblotted with mAb 3F4. II, CHO cells, as in panel a, after digestion of cell homogenates with 40 g/ml PK before immunoblotting with mAb 3F4. III, CHO cells, as in panel a, were incubated with PK-digested hamster scrapie brain homogenates comprising 7.5 g of PK-resistant PrP (see "Experimental Procedures"), extracted, and digested with PK before immunoblotting with mAb 3F4. IV, CHO cells, as in panel I, but cells were preincubated with 4 mg/ml heparin before the addition of the scrapie homogenate.
on mutant CHO cells, see "Experimental Procedures") (19). As can be seen in Fig. 1b, only the extract of CHO-MHM2 PrP cells reacted with the 3f4 antibody. Similar results were obtained with other ␣PrP antibodies known to react with Chinese hamster PrP (not shown).
Following incubation with the scrapie brain homogenate (see "Experimental Procedures"), cells were extracted, digested with PK, and immunoblotted with ␣PrP mAb 3F4 to assess the binding of PrP Sc to the diverse cells. Regardless of the big difference in PrP C expression, similar concentrations of PrP Sc were bound to both wt and MHM2 CHO cells (Fig. 1b, III). Contrary to the extensive binding of PrP Sc to wt CHO cells, PrP Sc binding to both mutant lines, lacking only HS or total GAGs, was very poor. These results suggest that the initial binding of PrP Sc to CHO cells is independent of PrP C concentration but dependent on the presence of GAGs, and especially HS, on the cell surface.
As opposed to PK-resistant PrP from either mouse or human, PK-resistant hamster PrP always shows a predominant band representing fully glycosylated PrP. As can be seen in the figures throughout this report, this pattern did not change after the binding of the scrapie protein to cells or heparin beads, indicating there was no preference for any glycosylated isoform. Similar results were obtained for PK-resistant mouse PrP (not shown).
Heparin Inhibits the Binding of PrP Sc to CHO Cells-To further assess the interaction of PrP Sc with GAGs on the cell surface, we examined the effect of heparin (a naturally occurring analog of HS) on the binding of PrP Sc to wt and mutant CHO cells. To this effect, cells were cultured in the presence of heparin for 2 h before the addition of PK-digested brain homo-genate, as described above. Fig. 1b, IV, shows that heparin inhibited most of the PrP Sc binding to wt and MHM2 PrP CHO cells but had no effect on the residual binding of PrP Sc to HSϪ/Ϫ or XTϪ/Ϫ cells. These results suggest that PrP Sc or a molecule closely associated with PrP Sc may bind directly to cell-associated HS and that soluble heparin inhibits this interaction in a competitive manner. Moreover, Fig. 1b, IV, shows that heparin inhibited the binding of PrP Sc to wt and PrPoverexpressing CHO cells to the same extent, consistent with the possibility that the presence of PrP C is not required for the heparin-dependent binding of PrP Sc to cells. Heparin also inhibited the binding of PrP Sc to neuroblastoma cells (not shown), which have been shown to support prion replication (26).
The Association of PrP Sc with Cells Is Concentration-and Temperature-dependent-To test further the specificity of the PrP Sc -GAG interaction on the cell surface, we performed the same binding experiments described in the previous section at several concentrations of PrP Sc (as present in brain homogenate). To this effect, different concentrations of homogenate were added to the four lines of CHO cells described above, with and without preincubation of the cells with heparin. As can be seen in Fig. 2a, in the absence of heparin, PrP Sc binding to wt CHO or to MHM2 CHO cells increased with the increasing concentration of homogenate until reaching saturation at about 45 g of PrP Sc (see also Fig. 2b). In the cells in which either HS or total GAGs were absent, the binding of PrP Sc also increased with homogenate concentration but at a much lower rate. When heparin was added to the cells in culture before the scrapie homogenate, the binding of PrP Sc was very similar in all four CHO cell lines regardless of the presence of either PrP or GAGs. These results suggest that at high PrP Sc concentra- FIG. 2. a, the binding of PrP Sc to different CHO cells is concentration-dependent. wt, XTϪ/Ϫ, HSϪ/Ϫ, and MHM2 PrP CHO cells were incubated as described at 37°C with different concentrations of PKresistant PrP in the presence or absence of 4 mg/ml heparin. The concentration of PrP Sc added to cells was calculated by comparing with the signal obtained by serial dilutions of recombinant PrP. b, PrP Sc binding to wt CHO cells is temperaturedependent. wt CHO cells were incubated with different concentrations of PK-digested scrapie hamster homogenate in the presence or absence of heparin at both 4 or 37°C. Samples were processed as described and immunoblotted. The amounts of PrP Sc added and bound to cells were calculated by comparing to serial dilutions of purified recombinant hamster PrP. Graph shows calculations for bound PrP Sc .
tions, PrP Sc can bind to cells by a GAG-independent mechanism, the nature of which remains to be determined.
To determine whether the GAG-PrP Sc interaction affects the internalization of the prion protein, we incubated increasing concentrations of scrapie homogenate with cells at both 4 and 37°C in the presence and absence of heparin (27,28). Cells were processed as described in the previous experiments. Concentrations of added or bound PrP Sc were calculated by comparing PK-resistant PrP bands on gels with serial dilutions of recombinant PrP (Fig. 1a). Because the amount of PrP Sc associated with the cells at 4°C at a range of PrP Sc concentrations was very low (Fig. 2b), we conclude most of the PrP Sc was rapidly internalized into the CHO cells. As suggested from the results in Fig. 2a, although heparin (at 37°C) inhibited very strongly the incorporation of PrP Sc to cells at low PrP Sc concentrations, it was less efficient in higher PrP Sc concentrations, when the binding of PrP Sc reached saturation. This suggests the existence of a low affinity mechanism for PrP Sc incorporation into cells that is GAG-independent. The internalization of PrP Sc into CHO cells was also shown by immunocytochemistry methods to be inhibited by heparin-like reagents (29).
The results in Fig. 2b may suggest heparin inhibited the binding of PrP Sc to cells at 4°C, indicating sulfated sugars may affect not only the internalization of PrP Sc but also the initial binding of PrP Sc to the CHO cells. However, because of the low binding of PrP Sc to all CHO cells at this temperature, the results were not significant enough to assure this conclusion.
PrP Sc Binds to Immobilized Heparin-Although the binding of PrP C to heparin has been shown previously, this is not the case for PrP Sc (14). To this effect, we tested whether the proteaseresistant core of PrP Sc present in our protease-digested scrapie brain homogenate can bind to heparin-agarose. Homogenates from normal or scrapie-infected hamster brains were incubated in the presence of heparin-agarose, washed extensively, and boiled in SDS-PAGE buffer before immunoblotting with anti-PrP mAb 3F4 (see "Experimental Procedures"). Fig. 3 depicts the results of such an experiment. Both PrP Sc and PrP C bind to heparin specifically because this binding activity could be abolished when the incubation of the homogenates with the beads was performed in the presence of soluble heparin. As opposed to soluble heparin, polyaspartic acid, another negatively charged polymer, did not inhibit PrP Sc or PrP C binding to heparin-agarose. As stated above, although these results are consistent with the possibility that PrP Sc binds directly to heparin molecules, the insoluble nature of PrP Sc prevents us from claiming a direct binding between PrP Sc and any other molecule.
Copper Also Inhibits the Binding of PrP Sc to Cells in the Absence of PrP C -In a previous report (22), we showed that the binding of PrP Sc to N2a cells was inhibited when N2a cells were first cultured in the presence of copper ions, which have been shown to bind to PrP C . Incubation of such cells with copper resulted in increased accumulation of PrP C , probably because of its internalization and delayed degradation. We speculated at the time that the inhibition by copper of PrP Sc binding to N2a cells may result from the absence of PrP C or another copper-internalized molecule from the cell surface. To test whether the presence of PrP C is required for copper to inhibit the binding of PrP Sc to cells, we cultured wt and MHM2 PrP CHO cells in the presence or absence of 300 m copper for 24 h (Fig. 4) and subsequently added PK-digested scrapie brain homogenate to the cell culture as described above. Fig. 3 shows the results of such an experiment. As was shown for N2a cells (22), whereas the concentration of PrP C in the presence of copper was largely increased in the MHM2 CHO cells (Fig. 4a), PrP Sc binding was inhibited in a similar manner in both cell lines regardless of the presence of detectable PrP C on the cell surface. These results indicate that the copper-related inhibition of PrP Sc binding to cells may not result only from the binding of copper to PrP C but rather from the binding of copper to cell surface GAGs, which have shown here to be essential for PrP Sc binding to cells. It has been suggested previously that the formation of PrP-Cu(II)-glycosaminoglycan assemblies may be crucial entities in the metabolism of PrP (24).
Incubation of Heparin with Prion Inoculum Delays Disease Onset in Hamsters-To test whether heparin can delay prion disease onset by inhibiting binding of PrP Sc to its targets in vivo, we incubated normal or PK-digested scrapie brain homogenate in phosphate-buffered saline alone or in phosphate- FIG. 3. Both PrP C and PrP Sc bind to heparin-agarose. Normal or PK-digested scrapie brain homogenate was incubated with heparin beads following preincubation of the samples with 0, 25, or 50 g/ml heparin or 50 g/ml polyaspartic acid. Following extensive washing of the beads, they were boiled in SDS buffer and the eluents immunoblotted with mAb 3F4. buffered saline containing heparin (2 or 4 mg/ml) for 2 h before intraperitoneal inoculation to normal hamsters (Fig. 5).
Administration of normal brain homogenates in the presence or absence of heparin did not produce any adverse effects in any of the animals. Because none of the control animals developed prion disease even after long incubation times, this experiment also shows that incubation of PrP C with heparin at the present conditions did not convert PrP C into PrP Sc . Preincubation of scrapie brain homogenate with heparin did delay incubation time significantly in both concentrations of heparin used. Together with our in vitro experiments, these results are consistent with the possibility that heparin delayed prion disease onset via the inhibition of PrP Sc binding to GAGs on the cell surface. However, because of the complexity of in vivo mechanisms, it is not possible to conclude at this point whether this is the only or major function of heparin as an anti-prion agent. DISCUSSION Many lines of evidence lead to the conclusion that PrP Sc is the major, if not the only, component of the prion. Recently, a new and most awaited result has been provided by the Prusiner laboratory, indicating that an amyloid form of ␤-sheet recombinant PrP is enough to produce transmissible prion disease when inoculated to mice overexpressing PrP C (31). However, the molecules with which PrP Sc approaches its cell targets and subsequently causes conversion of PrP C and neurodegeneration have yet to be elucidated.
The results presented in this report suggest that GAGs may be the main, if not the only, receptor for PrP Sc on the cell surface. The binding of PrP Sc to cells was independent from the presence of GAGs only at high concentrations of the prion protein, which were much higher than required to produce minimum incubation time. Interestingly, incubation time for scrapie disease is known to reach a saturation point and cannot be reduced further for a specific disease strain by the inoculation of larger amounts of scrapie homogenates or of purified PrP Sc . Consistent with the role of endogenous GAGs in PrP Sc binding and internalization is the fact that heparin, an analog of GAGs, can inhibit the binding of PrP Sc in cells where GAGs are present.
Interestingly, PrP C , although it seems not to be required for the primary binding of PrP Sc to the cells, has been shown to be closely associated with HS on the cell surface (9,32). GAGs may constitute the "bridge" between PrP C on the cell surface and PrP Sc in the prion aggregate and cause the internalization of both the normal and scrapie-associated prion proteins. Although HS binds directly to PrP C , it may have an affinity for the aggregated amyloid-like structure of PrP Sc as we have shown previously for Congo Red (34). It is important to note that although the results shown here for CHO cells may not be directly relevant to prion infection, because CHO cells were never shown to sustain prion replication, heparin-like compounds have been shown by us (data not shown) as well as by others to inhibit the binding of PrP Sc to neuroblastoma cells (29).
The fact that copper ions also inhibit the binding of PrP Sc to cells in the absence of detectable PrP C further indicates that the direct receptor for PrP Sc on the cell surface is another molecule. We hypothesize that copper ions bind to GAGs on the cell surface, causing the internalization of the sugar polymer with the proteins attached to it, including PrP C . Because the concentration of PrP C has been shown to determine prion disease incubation time by modulating the concentration of PrP Sc formed in the cells (35), our results suggest that the binding of PrP Sc to the cell surface is mechanistically separated from the conversion of PrP C to PrP Sc .
Reagents that inhibit binding of PrP Sc to cells may reduce the inoculated prion dose by increasing prion clearance and thereby delay disease onset. Several anti-prion agents, among them other sulfated sugar molecules, have been shown to increase the incubation time when inoculated in combination or immediately after the prion inoculum (15,30,33) but less so when applied during disease incubation time, suggesting that at least part of their mechanism of function may be inhibition of PrP Sc binding to its cell targets.