Molecular basis of scrapie strain glycoform variation.

Transmissible spongiform encephalopathies (TSE) are characterized by the conversion of a protease-sensitive host glycoprotein, prion protein or PrP-sen, to a protease-resistant form (PrP-res). PrP-res molecules that accumulate in the brain and lymphoreticular system of the host consist of three differentially glycosylated forms. Analysis of the relative amounts of the PrP-res glycoforms has been used to discriminate TSE strains and has become increasingly important in the differential diagnosis of human TSEs. However, the molecular basis of PrP-res glycoform variation between different TSE agents is unknown. Here we report that PrP-res itself can dictate strain-specific PrP-res glycoforms. The final PrP-res glycoform pattern, however, can be influenced by the cell and significantly altered by subtle changes in the glycosylation state of PrP-sen. Thus, strain-specific PrP-res glycosylation profiles are likely the consequence of a complex interaction between PrP-res, PrP-sen, and the cell and may indicate the cellular compartment in which the strain-specific formation of PrP-res occurs.

strains differ biologically by differences in disease incubation times and neuropathology (11). The existence of different TSE strains is difficult to explain in terms of a proteinaceous agent that shares the same amino acid sequence with a normal host protein. If PrP-res is the infectious agent, strain-specific information must be encoded within the PrP-res molecule either structurally or via post-translational modification. Interestingly, TSE strains can be defined by differences in the conformation, glycosylation, protease resistance, and aggregation state of PrP-res (7,(12)(13)(14)(15)(16). Several lines of evidence demonstrate that PrP-res molecules associated with different TSE strains have distinct protein conformations. In two strains of hamster-adapted transmissible mink encephalopathy, treatment of PrP-res with proteinase K resulted in PrP-res molecules with different sizes. These size differences are due to strain-specific, conformational changes resulting in the exposure of different proteinase K cleavage sites (17,18). This size difference could even be transmitted to newly formed PrP-res in vitro, demonstrating that the three-dimensional structure of PrP-res molecules can be self-propagated and suggesting a molecular basis for TSE strains (19). However, it is still unclear how conformational differences in the PrP-res molecules may relate to strain-specific disease phenotypes.
PrP-res molecules associated with different TSE strains also differ in their relative amount of glycosylation. The prion protein contains two N-linked glycosylation sites and is present in the cell in three different forms (high molecular mass glycoform, low molecular mass glycoform, and unglycosylated) that vary in the number and complexity of sugars added (20,21). It has been speculated that differences in PrP-res glycosylation may encipher strain-specific characteristics in vivo (22). Recently, differences in PrP-res glycosylation have been used as specific markers for TSE strains as well as in support of the hypothesis that human CJD originates from bovine spongiform encephalopathy (22)(23)(24). However, the molecular basis of TSE strain-specific glycoform variability has remained elusive. Understanding how these strain-specific glycosylation differences in PrP-res are controlled would provide important insights into the existence of strains in a disease for which no nucleic acid genome has yet been identified. Several hypotheses have been forwarded to explain PrP-res glycosylation variability, including the following proposals: 1) different cells expressing differently glycosylated PrP-sen populations are targeted by different TSE strains (25,26); 2) the TSE agent directly alters the post-translational glycosylation of PrP-sen (21,27); 3) PrP-res itself dictates strain-specific PrP glycosylation (28).
In this study we demonstrate that PrP-res can dictate its own glycosylation profile. The final PrP-res glycosylation profile is, however, influenced by the glycosylation state of PrP-sen as well as the cell in which the formation of PrP-res takes place. Our data suggest that although PrP-res is the primary determinant of its own glycosylation profile, subtle changes in the PrP-sen pool can modify the final glycosylation pattern of PrP-res. Thus, the glycosylation pattern of PrP-res in vivo may be indicative of the cellular environment and the glycosylation state of the PrP-sen pool present in this environment. Based on the fact that the PrP-res glycosylation pattern can be drastically changed depending upon the PrP-sen glycosylation state, our results also suggest that it is unlikely that in vivo strainspecific phenotypes are enciphered in the carbohydrate moieties of PrP-res.

MATERIALS AND METHODS
Preparation of Brain Homogenates-The mouse-adapted scrapie strains ME7, 87V, and 22L were the kind gift of Dr. James Hope (Vitechnologies, Boston, MA) and were passaged once in C57BL/10 (ME7, 22L) or VmDk mice (87V). The mouse-adapted scrapie isolate RML Chandler (RML) was passaged in RML mice. Brains were stored at Ϫ80°C, and brain homogenates (10% w/v) were prepared fresh in phosphate-buffered saline (PBS) containing complete protease inhibitors (Roche Molecular Biochemicals) according to a previously published protocol (29).
Preparation of Cell Lysates-The mouse neuroblastoma cell line Mo3F4-MNB has been described previously (30). These cells express high levels of mouse PrP-sen containing the epitope for the mouse monoclonal antibody 3F4 (31). For cell lysates, Mo3F4-MNB cells were grown to confluence in 10% FBS/DMEM, rinsed twice in PBS, and scraped off the flask. Mo3F4-MNB cells treated with tunicamycin were washed twice with warm phosphate-buffered balanced salt, cultured in Opti-MEM for 1 h, and incubated with 8 g/ml tunicamycin in Opti-MEM (Invitrogen) supplemented with 10% FBS for 4 h. Cells were harvested and frozen in liquid nitrogen. Harvested cells were subsequently homogenized (10% w/v) in PBS with protease inhibitors. Following addition of Triton X-100 and SDS to 0.5% and 0.05%, respectively, cleared cell lysates were used for the in vitro formation of PrP-res.
Inoculation of Cells with Scrapie Agent and Detection of Newly Formed PrP-res-Mo3F4-MNB cells were grown in 24-well microtiter plates. Cells were overlaid with 200 l of a detergent-free brain homogenate diluted to 1% in Opti-MEM. After 4 h, 400 l of 10% FBS/DMEM were added, and the cells were incubated for 96 h before being lysed in lysis buffer (5 mM Tris-HCl, pH 7.4, 140 mM NaCl, 5 mM EDTA, 0.5% sodium deoxycholate, and 0.5% Triton X-100). Cell lysates were cleared of cell debris by centrifugation and incubated with 20 g/ml proteinase K for 60 min at 37°C. The reaction was stopped by the addition of phenylmethylsulfonyl fluoride (PMSF) to a final concentration of 3 mM, and the samples were centrifuged at 356,000 ϫ g for 1 h at 4°C. Pellets were sonicated into sample buffer (2.5% SDS, 3 mM EDTA, 2% ␤-mercaptoethanol, 5% glycerol, 0.02% bromphenol blue, and 63 mM Tris-HCl, pH 6.8), and the samples were analyzed by SDS-PAGE as described previously (32). PrP was visualized using the enhanced chemifluorescence detection system (Amersham Biosciences) according to the manufacturer's instructions. The relative amounts of the PrP-res glycoforms were quantified in terms of integrated peak volume using ImageQuant.
In Vitro Formation of PrP-res Using Cell Lysate-For the in vitro formation of PrP-res, brain homogenates of healthy or scrapie-infected mice containing 0.5% Triton X-100 and 0.05% SDS were mixed with a Mo3F4-MNB cell lysate at a ratio of 1:8. Samples were sonicated to break up PrP-res aggregates using a microsonic sonicator (Misonix) with the probe immersed in the sample. The samples were then incubated for 96 h at 37°C with agitation. Following treatment with proteinase K (20 g/ml) for 1 h at 37°C, the reaction was stopped by adding 3 mM PMSF. Sample buffer was added, and the samples were analyzed by SDS-PAGE. Centrifugation of samples at 356,000 ϫ g following proteinase K treatment did not significantly alter the PrP-res glycosylation profile (data not shown).
Cell-free Conversion Assay-Mo3F4-MNB cells were metabolically labeled, and Mo3F4 PrP-sen was immunoprecipitated from the cell lysate using the monoclonal antibody 3F4 as described previously (32). For experiments that required purification of cell surface PrP-sen released from its glycophosphatidylinositol anchor, radiolabeled Mo3F4 MNB cells were treated with 14 units/ml phosphatidylinositol-specific phospholipase C (PIPLC, ICN) as previously described (33). PrP-sen was then purified by immunoprecipitation from either the tissue culture supernatant (PIPLC-released PrP-sen) or from the cell lysate (cellassociated PrP-sen) using the monoclonal antibody 3F4.
PrP-res (25 ng) isolated from mouse brain infected with scrapie strains 87V, ME7, or Obihiro (the kind gift of Dr. Motohiro Horiuchi, Obihiro University Department of Agriculture and Veterinary Medi-cine, Hokkaido, Japan) was partially unfolded in 2.5 M guanidine hydrochloride and incubated with radiolabeled Mo3F4 PrP-sen in reaction buffer (0.75 M guanidine hydrochloride, 1.25% sarcosyl, 5 mM cetyl pyridinium chloride, and 50 mM sodium citrate buffer pH 6.0) at 37°C for 48 h. One-tenth of the reaction was methanol-precipitated and served as a control for input PrP-sen. The remaining samples were incubated with 12 g/ml proteinase K for 1 h at 37°C. Proteolysis was stopped by the addition of PMSF, and the proteins were methanolprecipitated. Radiolabeled products were analyzed by SDS-PAGE and quantified using the Storm PhosphorImager system (Amersham Biosciences).
Statistical Analysis-The relative intensities of the high molecular mass glycoforms were compared statistically using Dunn's Multiple Comparison Test.

Distinct PrP-res Glycosylation Patterns Can Be Produced in
Vitro-To study the molecular basis of PrP-res glycoform variation, we focused on four mouse-adapted scrapie strains: 87V, RML, 22L, and ME7 (11). After treatment of scrapie brain homogenates with proteinase K, PrP-res molecules associated with different strains exhibited distinct proportions of the three differently glycosylated forms of PrP-res (34) (Fig. 1A). For example, the high molecular mass glycoform (Fig. 1A, bracket) predominated in PrP-res associated with the strain 87V, whereas the low molecular mass glycoform (Fig. 1A, asterisk) was more abundant in RML-associated PrP-res. In accordance with previously published protocols (22,34), the PrPres glycoform ratio for each strain was determined by comparing the high molecular mass glycoform to the low molecular mass glycoform (Fig. 1B). Quantification of the relative intensities of the high molecular mass and low molecular mass glycoforms associated with each strain (Fig. 1B) demonstrated that the brain-derived PrP-res glycoform profiles of RML and 87V strains significantly differed (p Ͻ 0.001), whereas the profiles for 22L and ME7 strains overlapped (p Ͼ 0.5).
To analyze whether or not distinct PrP-res glycoform patterns could be produced in vitro, we developed a novel assay to monitor the de novo formation of PrP-res. In this assay, mouse neuroblastoma cells expressing high levels of 3F4 epitopetagged mouse PrP-sen (Mo3F4-MNB) were lysed, and the lysate was mixed with brain homogenates from mice either infected with 87V, ME7, 22L, and RML scrapie or from an uninfected mouse. The 3F4 epitope (31) was used to detect newly generated 3F4-positive PrP-res. Thus, unlike in previously published protocols (29,35), newly formed glycosylated PrP-res can be detected by using a specific antibody that only reacts with epitope-tagged PrP and not with PrP molecules present in the brain homogenate. After 4 days, samples were assayed by Western blot for new PrP-res formation. Newly formed 3F4-positive PrP-res could be detected for each strain, whereas no PrP-res was detected when the cell lysate was mixed with the brain homogenate of an uninfected mouse (Fig.  1C). The glycoform ratio of newly formed PrP-res differed depending upon the scrapie strain. Strikingly, and consistent with brain-derived PrP-res, the RML and 87V PrP-res glycoform profiles were significantly different (p Ͻ 0.001) (Fig. 1D). In fact, for all strains the glycosylation pattern of newly formed PrP-res followed the same distribution on the graph as the brain-derived PrP-res glycosylation patterns (compare Fig. 1, B to D). Thus, distinct glycosylation profiles can be imposed on newly formed PrP-res in the absence of any cellular processes. These data suggest that the PrP-res present in the scrapieinfected brain homogenate dictates the PrP-res glycoform pattern.
The PrP-res Glycoform Pattern Is Strongly Influenced by the Glycosylation State of PrP-sen-The glycosylation patterns generated in vitro did not exactly match the brain-derived PrP-res glycosylation patterns in that the unglycosylated PrP-res form was less abundant in PrP-res produced in vitro (compare Fig. 1, A-C). This suggested that other factors could influence the final PrP-res glycosylation pattern. Because PrPsen is the only other identified component in the formation of PrP-res, we reasoned that changes in the glycosylation state of PrP-sen could be influencing PrP-res glycosylation. To determine whether a change in the relative amounts of the different PrP-sen glycoforms could influence the strain-specific glycosylation pattern of newly formed PrP-res, Mo3F4-MNB cells were incubated with tunicamycin. Tunicamycin blocks N-linked glycosylation and leads to a population of PrP-sen with an increased amount of the unglycosylated form. Quantification of the amount of unglycosylated PrP-sen present in normal and tunicamycin-treated cells revealed that the tunicamycin treatment led to an approximate 3-fold increase in the amount of unglycosylated PrP-sen in the total PrP-sen pool, although the high molecular mass and low molecular mass PrP-sen glycoforms were still most abundant ( Fig. 2A).
Cell lysates of tunicamycin-treated cells were then mixed with homogenates of scrapie-infected brain and assayed for newly formed PrP-res. Surprisingly, the slight increase in unglycosylated PrP-sen led to a drastic increase in unglycosylated PrP-res independent of the scrapie strain (Fig. 2B, compare left  and right panels). The distribution of the strain-specific glycosylation profiles on the graph remained unaltered, again demonstrating that PrP-res can dictate strain-specific glycoform profiles. However, the increase in unglycosylated PrP-res shifted the glycosylation profile lower (Fig. 2C). Quantification of the three PrP-res forms demonstrated that the percentage of unglycosylated to total 3F4-positive PrP-res increased ϳ5-fold (87V) to 6-fold (22L, ME7, and RML) when compared with PrP-res generated with normal, untreated cell lysate (Fig. 2D). Thus, our data suggest that the PrP-res glycoform ratios observed in the brain could differ from those observed in vitro due to differences in the relative amounts of the PrP-sen glycoforms. Overall, our results clearly demonstrate that the relative amounts of the different PrP-sen glycoforms can strongly influence the glycosylation pattern of newly formed PrP-res.
PrP-res Can Dictate Its Own Glycosylation Profile-It was still possible that components in the cell lysates or brain homogenates other than PrP-sen and PrP-res were influencing strain-specific differences in PrP-res glycosylation patterns. To exclude this possibility, cell-free conversion assays were performed that utilized purified PrP-sen and PrP-res as described FIG. 1. Different PrP-res glycoform ratios can be generated in vitro. A, brain-derived PrP-res. Western blot of proteinase K-treated brain homogenates from uninfected mice (Mock) or mice infected with the scrapie strains 87V, 22L, ME7, or RML. The blot was developed using the rabbit polyclonal antiserum R30 (32). Bracket, high molecular mass glycoforms of PrP-res; asterisk, low molecular mass glycoforms of PrP-res; closed circle, unglycosylated PrP-res. Molecular mass markers are indicated in kilodaltons. B, ratio of high molecular mass to low molecular mass PrP-res glycoforms in brains of scrapie-infected mice. Values were determined from a minimum of three individual animals per strain. Each brain sample was analyzed several times (n ϭ 2-4). C, Western blot analysis of 3F4-positive PrP-res generated by mixing Mo3F4-MNB cell lysate with brain homogenate. Bracket, high molecular mass glycoform; asterisk, low molecular mass glycoform; closed circle, unglycosylated. D, quantification of high molecular mass to low molecular mass 3F4-positive PrP-res glycoforms generated in cell lysates mixed with brain homogenates. Shown are the results of two independent experiments with a total of 10 -11 individual samples per strain.
previously (32). Radiolabeled Mo3F4 PrP-sen was immunoprecipitated from Mo3F4-MNB cells and mixed with highly purified PrP-res fractions isolated from the brains of mice infected with one of three mouse scrapie strains: 87V, ME7, or Obihiro (36). RML and 22L derived PrP-res were not tested because the PrP-res generated by these strains was variably truncated, and it was difficult to identify specific PrP-res glycoforms (30). 2 Although the input PrP-sen was similar for all three strains (Fig. 3A, left panel), differences in the glycosylation pattern of newly formed PrP-res were observed (Fig. 3, A, right panel and B, compare Obihiro to 87V and ME7). These differences were clearly dependent on the PrP-res input. Thus, no additional components were required to produce strain-specific differences in the glycosylation pattern of newly formed PrP-res demonstrating that PrP-res itself is capable of dictating strain glycoform variation.
Influence of the Cell on the PrP-res Glycosylation Profile-In vivo, following cellular uptake of the scrapie agent, PrP-res formation is believed to occur either on the cell surface of the infected cell or along the endocytic pathway (5). Thus, it was possible that active cellular processes could also influence the glycosylation pattern of newly formed PrP-res. However, no currently available cell culture or transgenic mouse model is capable of detecting the rapid, de novo formation of fully gly-cosylated PrP-res. By utilizing Mo3F4-MNB cells, we have now established a cell culture system that allows the detection of acute PrP-res formation within 4 days after exposure to scrapie brain homogenate. 3 To determine the influence of the cell on the glycoform pattern of newly formed PrP-res, intact Mo3F4-MNB cells were exposed for 4 days to brain homogenates from uninfected mice or from mice infected with one of the scrapie strains 87V, 22L, ME7, and RML. Newly formed 3F4-positive PrP-res was detected for all strains tested, although some background staining was also present in cells that were exposed to brain homogenate from uninfected mice (Fig. 4A). Exposure of the cells to the scrapie strains 22L, ME7, and 87V resulted in PrP-res glycosylation profiles that followed the same distribution on the graph compared with brain-derived PrP-res (compare Fig.  1B to Fig. 4C). By contrast, the PrP-res glycosylation profile associated with RML scrapie was quite variable (Fig. 4B), although the high molecular mass glycoforms (Fig. 4A, brackets) generated in the tissue culture cells tended to be more prevalent than those in brain-derived RML PrP-res. The fact that the RML PrP-res glycosylation profile differed from the glycosylation profile obtained when a cell lysate was mixed with an RML brain homogenate (compare Fig. 1D to Fig. 4B) demonstrated that the cell is also capable of influencing the 2 S. A. Priola, unpublished results. 3 I. Vorberg, unpublished data. glycosylation pattern of newly formed PrP-res. Furthermore, these results also show that different scrapie strains can interact with one cell type and produce distinct PrP-res glycosylation patterns (Fig. 4C, compare 87V to 22L/ME7) and are in agreement with data published previously (37). Thus, although the cell can influence PrP-res glycosylation, different cell types are not required to produce strain-specific differences in PrPres glycosylation patterns.

PrP-sen from Different Cellular Compartments Influences the Glycosylation Pattern of Newly Formed PrP-res-
The above results demonstrated that subtle changes in PrP-sen glycoform ratios as well as the cell could influence the PrP-res glycosylation profile. Together, these data suggested that the cell might influence strain-specific glycosylation patterns by determining the pool of PrP-sen to which PrP-res has access. To determine whether different cellular pools of PrP-sen could influence strain-specific glycosylation patterns dictated by PrP-res, two pools of PrP-sen were isolated: 1) cell-surface PrP-sen released from its glycophosphatidylinositol anchor using PIPLC, and 2) the pool of PrP-sen that remained associated with the cell following exposure to PIPLC. When the PrP-sen in these two pools was compared, the glycoform pattern of PIPLC-released PrP-sen differed subtly but significantly from that of cell-associated PrP-sen (Fig. 5A). Radiolabeled PrP-sen from both of these fractions was mixed in a cell-free conversion assay with purified PrP-res from 87V, ME7, or Obihiro scrapie-infected mice. The glycoform pattern of the newly formed PrP-res was then analyzed. PrP-res derived from PrP-sen released from the cell surface (SN, Fig. 5) had the same glycoform pattern as PrP-res derived from the unfractionated PrP-sen pool (Total, Fig. 5). Thus, the lack of a glycophosphatidylinositol anchor in the PIPLC-released fraction did not affect PrP-res glycosylation patterns. However, for all three strains, small but significant differences were seen when the glycosylation pattern of newly formed PrP-res derived from PIPLC-released PrP-sen (SN) was compared with that of newly formed PrP-res derived from cell-associated PrP-sen (Fig. 5, B-D). These results are consistent with the hypothesis that the cell may affect overall PrP-res glycosylation patterns by influencing the pool of PrPsen with which PrP-res interacts. Thus, even small differences in the glycoform composition of PrP-sen isolated from different cellular compartments can modulate the relative amounts of the different PrP-res glycoforms. DISCUSSION We have shown that TSE strain glycoforms are primarily determined by PrP-res but can be significantly influenced by PrP-sen. An interesting finding of this study was that the PrP-res glycosylation profile associated with RML scrapie generated in vitro drastically differed from the PrP-res glycosylation profile obtained by mixing scrapie brain homogenate with cell lysate. These data demonstrate that the cell can also influence PrP-res glycoform profiles. However, it is unclear why the cell only strongly affected the glycosylation pattern of RML PrP-res. One possible explanation for this is that RML scrapie initiated PrP-res formation in mouse neuroblastoma cells in a different cellular compartment than the scrapie strains 22L, ME7, or 87V.
Thus, in terms of the cellular biology of TSE diseases, our data suggest that the strain-specific glycosylation pattern of PrP-res could be indicative of the cellular compartment where PrP-res formation occurs for that strain. PrP-sen is a glycosylphosphatidylinositol-anchored cell-surface protein that is internalized and either recycled to the cell surface or degraded intracellularly (5, 38 -41). The relative amount of glycosylation appears to be dependent on the cellular localization of PrP-sen (Fig. 5A). Thus, strains that are associated in vivo with PrP-res molecules with a relatively high proportion of the unglycosylated form may induce the formation of PrP-res most effectively in a cellular compartment where this form is more prevalent. Conversely, strains in which glycosylated PrP is over-represented may convert PrP-sen to PrP-res in a cellular compartment with an abundance of glycosylated PrP-sen but lower levels of unglycosylated PrP-sen, such as the cell surface. This model of strain-specific glycosylation is supported by our data demonstrating that both the cell and the pool of PrP-sen can significantly change the PrP-res glycoform profile.
In vitro studies have suggested that unglycosylated PrP-sen is preferentially converted to PrP-res (5). Thus, if PrP-res always prefers unglycosylated PrP-sen, this form should be equally present in PrP-res from different TSE strains. In vivo, however, this is clearly not the case as the amount of unglycosylated PrP-res can vary significantly from strain to strain (34). If PrP-res glycosylation patterns indicate where in the cell PrP-res formation occurs, this discrepancy could be explained by distinct TSE strains having differential access to unglycosylated PrP-sen. Thus, it would be the PrP-sen glycoforms present in the cellular microenvironments along the endocytic and/or secretory pathways where PrP-res formation occurs most efficiently (39,42) FIG. 5. PrP-sen from different cellular compartments influences the glycosylation pattern of newly formed PrP-res. A, relative amounts of the different PrP glycoforms associated with radiolabeled Mo3F4 PrP-sen that had either been removed from the cell surface with PIPLC or that remained cell-associated following PIPLC digestion (n ϭ 20). For A only, significance was calculated by comparing each PrP glycoform in the cell-associated (Cell) and the PIPLC-released fraction (SN) to the corresponding glycoform in the unfractionated pool (Total). High, high molecular mass glycoform; low, low molecular mass glycoform; none, unglycosylated. B-D, radiolabeled and immunoprecipitated Mo3F4 PrP-sen from A was mixed with purified PrP-res from the brains of mice infected with the mouse scrapie strains 87V (B), ME7 (C), or Obihiro (D). The data shown represent the relative amounts of each of the different PrP glycoforms in the newly formed, radiolabeled PrP-res. The legend is the same as for A. For each strain, the results shown represent duplicate experiments for a total sample size of n ϭ 4. Significance was calculated by comparing the cell-associated and PIPLC-released fractions for each glycoform using the paired Student's t test. No significant difference was observed when the PIPLC-released fraction was compared with the unfractionated pool. *, p Յ 0.03; **, p ϭ 0.01; ***, p Յ 0.0001. that determine the final PrP-res glycoform pattern.
Alternatively, PrP-res glycosylation profiles may be indicative of the PrP-sen glycoform pattern of the brain area in which PrP-res is formed. In this instance, PrP-res glycosylation patterns could be the consequence of strains specifically targeting neuronal subsets with a PrP-sen glycosylation pattern that equals the final PrP-res glycosylation pattern associated with that strain (22). Attempts to define PrP-sen glycosylation patterns in different brain areas have led to conflicting results, suggesting that any glycosylation differences between PrP-sen species in different brain regions may be subtle (27,43). These subtle differences could be sufficient to generate cell typespecific PrP-res glycosylation patterns. Indeed, our data demonstrate that even minor changes in the relative amounts of the different PrP-sen glycoforms can have a significant impact on the PrP-res glycosylation pattern (Fig. 2). Unfortunately, this model cannot explain how different scrapie strains can induce the formation of different PrP-res glycosylation patterns in the same cell line as observed in our study and by others (37). A direct influence of the TSE agent on post-translational cellular processes would explain how one cell type could support the formation of strain-specific PrP-res glycosylation patterns (21,27,44). However, our experiments demonstrate that neither active modulation of the cellular glycosylation process nor cellular degradation of specific PrP-res glycoforms is necessary for strain-specific differences in PrP-res glycosylation. Thus, although we cannot exclude the possibility that different cell types or an as yet unknown interaction of the TSE agent with cellular processes could influence PrP-res glycosylation, our data show that they are not necessary.
It has been speculated that strain-specific biochemical and pathological properties are enciphered not only by the conformation of PrP-res but also by the attached carbohydrate moieties (22). Our data demonstrating that the relative amounts of the different PrP-sen glycoforms can strongly influence the strain-specific glycosylation profile of newly formed PrP-res suggest that this may not be the case. Furthermore, the PrPres glycosylation profile of one strain can differ between brain areas and/or the brain and other organs like spleen and tonsil (45,46), yet strain-specific characteristics are retained upon transmission independently of the organ from which the agent has been isolated (47). Therefore, it seems unlikely that the carbohydrate moieties on PrP-res are encoding any specific strain characteristic.