Main

Experimental transmission of BSE to mice13, cattle14 and two sheep PrP genotypes (A136 Q171 (ov-AQ) and V136 Q171 (ov-VQ))15 has been reported, but hamsters (J. D. Foster and J.H., unpublished data) and at least one PrP genotype of sheep (A136 R171 (ov-AR))16 seem to be resistant to clinical disease. To test for a correlation between in vitro cell-free conversion and in vivo transmission, we included types of 35S-labelled PrP-sen from each susceptible or resistant group of species in PrPBSE-driven conversion experiments. ProteinaseK-resistant 35S-labelled conversion products that were the characteristic 6K to 7K smaller than the 35S-PrP-sen substrate8,9,10,11,12,17 were generated in the reactions with both glycosylated and aglycosyl 35S-PrP-sen molecules from BSE-susceptible, but not BSE-resistant, hosts (Fig. 1). For example, the 25K aglycosyl bovine and sheep (ov-AQ and ov-VQ) 35S-PrP-sen proteins generated 18K 35S-PrP-res bands (Fig. 1, lanes 1, 6, 7, 13, 15, 16), whereas the aglycosyl hamster and ov-AR gave no proteinaseK-resistant product (Fig. 1, lanes 17, 20, 21). Thus, the efficiency of the PrPBSE-induced conversion of PrP-sen of a given host was found to be correlated with the in vivo transmissibility of BSE to that host. This correlation encouraged us to look for an in vitro indication of the transmissibility of BSE to humans, using a PrPBSE/human PrP-sen conversion assay.

Figure 1: Phosphor autoradiographic analysis of SDS–PAGE gels of cell-free conversion products formed on incubation of PrPBSE with 35S-PrP-sen from different species.
figure 1

The molecular size range of 35S-radiolabelled PrP proteins following incubation with PrPBSE is seen before (−PK) or after (+PK) digestion with proteinase K in the top two panels (+PrPBSE, −PK and +PrPBSE, +PK). The effect of omitting PrPBSE from the incubation is shown in the bottom panels (−PrPBSE, +PK). The migration of Mr markers is shown on the right. The ∼18K PK-resistant 35S-PrP product is marked by the arrowhead. Equal amounts of PrPBSE (∼0.5μg) were used in each reaction with an approximately equivalent amount of radioactive PrP-sen (∼25,000–40,000c.p.m. per reaction; apart from boPrP-sen reactions, where ∼10,000c.p.m. was used owing to inefficient labelling in the bovine cells). The reaction sample was divided for analysis ∼1:10 (−PK:+PK) between the gels shown in the top two panels (and also between the top and bottom panels). Results are shown using differing concentrations of GdnHCl in the pretreatment and conversion phases of the reaction (see Methods). Conversions using 35S-PrP-sen molecules labelled in the presence (+) or absence (−) of tunicamycin are shown. The relative intensities of bands in the +PrPBSE, +PK panel need not reflect the efficiency of conversion as the specific activities of each label are not always the same; the percentage of label converted to proteinase K-resistant fragments (summarized in Fig. 4) is a better indicator of this efficiency than absolute intensities of the PK-resistant bands in the +PK, +PrPBSE panel.

Full size image

Wild-type human (h) PrP has two common allelic forms that encode either methionine (hPrP-M) or valine (hPrP-V) at codon 129 (ref. 18). Hence we tested both types of hPrP-sen in conversion experiments. PrPBSE converted the ∼25K aglycosyl forms of 35S-hPrP-M and 35S-hPrP-V to ∼18K proteinaseK-resistant forms compatible with the proteinaseK-resistant core of PrP found in the human diseases (Fig. 1, lanes 2, 3, 4, 14; Fig. 2, lane 8). When more glycosylated 35S-hPrP-M was used, the proteinaseK-resistant conversion product was still largely restricted to an ∼18K band, suggesting preferential conversion of the 25K aglycosylated form of 35S-hPrP-M (Fig. 1, lane 9). Little or no spontaneous formation of proteinaseK-resistant PrP was observed in the absence of PrP-res (Fig. 1, bottom). The 35S-hPrP-V labels were converted by PrPBSE roughly threefold less efficiently than the 35S-hPrP-M labels (Fig. 1, lanes 2–5, 9, 10; Fig. 2, lanes 8, 9; Fig. 4; P = 0.0025). So far the new variant CJD has been found only in patients homozygous for methionine at codon 129 (ref. 2) and, although it may be premature to speculate on the basis of these conversion data, such genotypic selection may result from a more efficient conversion of hPrP-M than hPrP-V by PrPBSE.

Figure 2: Comparison of cell-free conversions induced by PrPBSE and PrPCJD.
figure 2

Bovine (bo) and human (h-M, h-V) 35S-PrP-sen proteins were labelled in the presence of tunicamycin (top panel, −PK, left) and incubated either alone (none) or with equivalent amounts (∼0.5μg) of PrPBSE or human PrPCJD-M before processing with or without PK treatment. The conversion reaction samples were split 1:10 between the −PK and +PK gel lanes. Phosphor autoradiographic images are shown of the labelled proteins from conversions using different concentrations of GdnHCl in the pretreatment and conversion phases of the reaction (see Methods): top panel, pretreatment in 2M GdnHCl followed by conversion in 1M GdnHCl; bottom panel, pretreatment in 2.5M GdnHCl, conversion in 1.5M GdnHCl.

Full size image
Figure 4: Percentages of 35S-PrP-sen proteins converted to PK-resistant forms by incubation with PrPCJD, PrPBSE, an.d sheep, mouse or hamster PrPSc.
figure 4

a, Scatter graph of percentage conversion of PrP-sen to PK-resistant PrP forms in multiple, independent experiments. Each symbol represents percentage conversions observed in independent conversion reactions using the designated PrP-res incubated with glycosylated (open circles) or unglycosylated (filled circles) 35S-PrP-sen with a pretreatment/conversion protocol of 2M to 1M GdnHCl or unglycosylated 35S-PrP-sen with the 2.5M to 1.5M GdnHCl protocol (squares). b, Bar graph of percentage conversion of PrP-sen to PK-resistant PrP forms. Mean (+s.e.m.) percentage conversion for each PrP-res/PrP-sen pair is shown calculated for the two GdnHCl pretreatment/conversion protocols and types of PrP-sen differentiated in a by the filled circles, open circles and squares (left) or summarized by combining the data from all experiments (right). The absence of a bar on the left means that no data were collected under those conditions except for the cases where the percentage conversions were negligible; such cases are shown by 0.0. As well as the results of a, this graph shows the negligible conversion produced using 6M GdnHCl-pretreated PrPBSE. Relative mean percentage conversions have been normalized against the mean of the homologous conversions induced by each type of PrP-res, with the latter set at 100%. Question marks acknowledge that it is likely, but not certain, that the human new variant CJD cases in hPrP-M129 homozygotes are due to BSE infections and that, although no new variant CJD cases have yet been observed in hPrP-V129 carriers, it is possible that all codon-129 genotypes may eventually be affected. Minus signs in brackets refer to the data indicating that 6M GdnHCl treatment inactivates hamster scrapie infectivity26, although this has not been demonstrated specifically for BSE infectivity.

Full size image

To help gauge the efficiency of these various PrPBSE-induced reactions, we compared these data with the proportion of labelled hPrP-sen (M or V) converted by a similar amount of human PrPCJD isolated from either new variant (Figs 2, 4) and sporadic CJD (not shown) patients homozygous for methionine at codon 129 (PrPCJD-M). PrPCJD-M converted the 25K 35S-hPrP-M and 35S-hPrP-V lacking N-glycans to an ∼18K proteineaseK-resistant form compatible with the proteinaseK-resistant core of PrPCJD (Fig. 2, lanes 5, 6). Using predominantly glycosylated forms of 35S-hPrP-sen, we saw that the PrPCJD again preferentially generated the same 18K, presumably aglycosyl, conversion product, although some generation of the larger glycosylated conversion products was also observed (not shown). The 35S-hPrP-M label was converted threefold more efficiently than 35S-hPrP-V by PrPCJD-M (Fig. 2, lanes 5, 6; Fig. 4; P = 0.0025), similar to the findings from conversions with PrPBSE (Figs 1 and 4).

The efficiency of cross-species conversion of hPrP-sen of either genotype by PrPBSE was significantly less than the efficiency observed in either of the homologous conversion reactions: hPrP-M or hPrP-V with PrPCJD-M (P = 0.0005 or 0.005, respectively) or bovine (bo) PrP-sen with PrPBSE (P = 0.0005) (Fig. 1, lanes 1–5, 8–10, 13, 14; Fig. 2, lanes 5–9; Fig. 4). On average, PrPBSE converted boPrP-sen 30-fold more than hPrP-V (P = 0.0005), and 10-fold greater than hPrP-M (P = 0.0005). However, this low level of conversion was significantly higher than that seen on incubation of PrPBSE with PrP-sen from hamsters (P = 0.005) or sheep homozygous for the AR genotype (P = 0.025)—hosts shown in vivo to be resistant to BSE—(Fig. 1, lanes 17, 20, 21). To put this into perspective, we evaluated the efficiency of conversion of hPrP with PrPSc from scrapie-infected sheep (ovPrPSc), a source of agent of no measurable risk to the human population.

In conversions driven by ovPrPSc(VQ), the conversion efficiency of 35S-hPrP-M (Fig. 3, lane 1) ranked with that of mouse (lane 5) and sheep ov-AR (lane 4) proteins; no conversion of hamster (ha) PrP-sen was observed (lanes 6, 7). In contrast, much higher conversion efficiencies were observed with the ov-VQ and ov-AQ PrP-sen alleles. Thus, the conversion efficiencies of the different aglycosyl ovPrP-sen alleles by ovPrPSc(VQ) correlated with their sequence homology at codon 171: ov-VQ, ov-AQ ≫ ov-AR (Figs 3, 4)11. These relative conversion efficiencies correlated with the fact that, although animals of VQ/VQ, AQ/VQ or AQ/AQ genotypes can develop natural scrapie (depending on breed) and can be experimentally induced to develop clinical scrapie (depending on agent strain) and BSE, animals with at least one copy of the AR allele seem to have enhanced survival following exposure to either natural or experimental disease16. Also, the lack of conversion of haPrP-sen by ovPrPSc(VQ) correlates with the lack of documented transmission of the sheep scrapie agent directly to hamsters (Fig. 4).

Figure 3: Cell-free conversions induced by sheep PrPSc.
figure 3

Human (h-M), mouse (mo), hamster (ha and ha-GPI−) and three allelic forms of sheep (ov-AQ, ov-VQ and ov-AR) 35S-PrP-sen proteins were labelled in the presence of tunicamycin (−PK). These aglycosyl PrP-sen molecules were incubated with equivalent amounts (∼0.5μg) of ov-PrPSc(VQ) and processed as described in Methods. The reaction samples were split 1:10 (−PK:+PK) for analysis on the gels.

Full size image

In summary, the most efficient conversion reactions were observed between homologous PrP-sen and PrP-res molecules, just as TSE transmissions are usually most efficient between hosts of homologous PrP genotype (Fig. 4). In contrast, little or no conversion was observed with non-homologous PrP combinations associated with a lack of transmission in vivo. Although overall conversion efficiencies varied and were influenced by the pretreatment and conversion GdnHCl concentrations, the optimal conversion conditions for each species fell within the range shown in Figs 1 and 2. Furthermore, variations of these conditions, the amount of PrP-res, or the glycosylation state of PrP-sen, did not seem to affect the relative efficiencies of conversion of the various PrP-sen molecules (Fig. 4). Comparison of the amino-acid sequences of human, sheep, bovine and other PrP molecules strongly implicated the loop structures of the protein19 as critical regions affecting the conversion of PrP-sen to PrP-res and for maintenance of at least part of the barrier to transmission of TSEs between species (see http;//cgi-bin/SupplData.cgi/screen/388285A0>Supplementary Information).

Our results provide further evidence for a correlation between the efficiency of in vitro conversions and the transmissibilities of TSE diseases. It extends the basis for using the cell-free conversion reaction as an in vitro indicator of the transmission barrier to new species. We must emphasize that the cell-free assay is a test only of the molecular compatibility between PrP-res and PrP-sen of different sequences. Other factors such as the dose, strain and route of infection, the stability of the infectivity in the host, and the efficiency of its delivery to the central nervous system are all likely to be important in vivo. Nonetheless, some degree of molecular compatibility is likely to be essential in the transmission of TSE diseases, especially given the remarkable correlation between PrP conversion efficiencies and transmissibilities (Fig. 4). The conversion experiments with 35S-hPrP-sen showed that the relative converting activity of PrP-res depends on its species of origin: PrPCJD-M > PrPSc, PrPBSE. The relatively low efficiency of conversion by PrPSc and PrPBSE may explain at a molecular level the observed failure to transmit BSE to transgenic mice expressing human PrP20 and the lack of an epidemiological link of CJD to sheep scrapie. It is premature to draw firm conclusions from our results about the likelihood of BSE passing to humans, although the results suggest that BSE would be no more inherently transmissible to humans than is sheep scrapie.

Methods

Purification and analysis of PrP-res. PrP-res preparations were purified21 and stored at 4°C in 1%-sulphobetaine 3-14 in phosphate-buffered saline (PBS; 20mM sodium phosphate and 120mM NaCl, pH 7.4)17. Two or more independent isolates of each type of PrP-res were tested. Yields of PrP-res per g brain tissue were assayed by immunoblotting using R505, a polyclonal antiserum raised against a peptide epitope which, N-terminal amino acid apart, is common to all these species (sheep peptide 100-111; provided by J. Langeveld, Lelystadt, The Netherlands)22: haPrPSc from Syrian golden hamsters infected with the 263K strain, ∼50μgg−1; mouse (mo)PrPSc from VM/Dk mice infected with the 87V strain, ∼10μgg−1; ovPrPSc(VQ) from natural scrapie cases in Cheviot sheep (provided by J. Foster, IAH-NPU, Edinburgh), ∼1–5μgg−1; PrPBSE from brainstem of BSE-affected cattle (provided by MAFF VI Centre, Thirsk), ∼1μgg−1; PrPCJD from cases of new variant or sporadic CJD (∼1μgg−1).

Labelling, purification and analysis of 35S-PrP-sen.35S-PrP-sen proteins were radiolabelled and purified using PrP-specific antisera as described17, except that they were eluted from the protein A-Sepharose beads using 0.1M acetic acid at 22°C for 30min and stored at 4°C. To prepare aglycosylated 35S-PrP-sen, preincubation and labelling of cells were done in the presence of tunicamycin (2.5μgml−1). Bovine adrenocortical cells (SBAC, ATCC RL-1796) were used to obtain radiolabelled boPrP-sen which was immunoprecipitated using a rabbit polyclonal PrP antiserum (R35) raised against a synthetic peptide sequence (residues 93–107) of bovine PrP (provided by R. Race, RML, Montana). Labelled hPrP-M and hPrP-V were produced from molecular clones expressed in human neuroblastoma cells and immunoprecipitated using the 3F4 monoclonal antibody23,24. Endogenous moPrP-sen from mouse neuroblastoma (N2a) cells was collected using R30 serum. Recombinant mo-3F4 PrP-sen9 expressed in N2a cells was immunoprecipitated with the 3F4 antibody. 35S-PrP-sen expressed for ovPrP clones encoding three of the sheep PrP genotypes11 were expressed in N2a cells and immunoprecipitated with R521, a sheep-specific polyclonal antiserum against sheep PrP peptide 94–105 (provided by J. Langeveld22). Hamster 35S-PrP-sen proteins were immunoprecipitated from mouse fibroblast cells expressing recombinant haGPI− PrP-sen or the full, unmodified haPrP-sen9.

Cell-free conversion reactions. To begin a conversion reaction25, the PrP-res was sonicated, mixed with guanidine-HCl (GdnHCl) to a final concentration of 2.0, 2.5 or 6M, typically in 8μl and incubated at 37°C for 1–3h. Approximately equal amounts of each 35S-PrP-sen tested were lyophilized and reconstituted in conversion buffer (final composition in the conversion mix: 50mM sodium citrate, pH 6.0, 5mM cetyl pyridinium chloride, 1.25% (w/v) N-lauryl sarcosinate). Pre-treated PrP-res (in GdnHCl) and 8M GdnHCl (to give a final concentration of 1 or 1.5M) were added into the 35S-PrP-sen solution (final volume, 16–24μl) and incubated at 37°C for 3 days. The amount of different PrP-res proteins used in any single experiment was kept constant at either ∼0.5 or 0.8μg. Using these amounts, the system was at or near saturation with PrP-res as the higher amount did not significantly increase the degree or efficiency of conversion (data not shown). After incubation, each reaction was split 1:10 and the major fraction digested with 100μgml−1 proteinase K (PK) in Tris-saline (50mM Tris-HCl, pH 8, 130mM NaCl) for 1h at 37°C. Pefabloc (2mM) and thyroglobulin (20μg) were added to each fraction (+ or −PK) and the proteins precipitated by methanol. Percentage conversions were determined by quantifying the proportion of radioactivity within a specific Mr range in each fraction (+PK and −PK) by phosphor autoradiographic imager analysis of SDS–PAGE gels. Aglycosylated 35S-PrP-sen was quantified within an Mr range of ∼24K to 26K; the glycoforms produced in non-tunicamycin-treated cells were scanned over an Mr of ∼24K to 38K. PK-resistant products of conversions using aglycosylated 35S-PrP-sen were quantified within an Mr range of ∼18K to 20K; the variably glycosylated products of conversions using 35S-PrP-sen labelled without tunicamycin were quantified over a Mr range of ∼18K to 30K. The significance of differences in percentage conversion of PrP-sen/PrP-res pairs was calculated using the Mann–Whitney U-test.