|Home | About | Journals | Submit | Contact Us | Français|
Transmissible spongiform encephalopathies are associated with an autocatalytic conversion of normal prion protein, PrPC, to a protease-resistant form, PrPres. This autocatalytic reaction can be reproduced in vitro using a procedure called protein misfolding cyclic amplification (PMCA). Here we show that, unlike brain-derived PrPC, bacterially-expressed recombinant prion protein (rPrP) is a poor substrate for PrPres amplification in a standard PMCA reaction. The differences between PrPC and rPrP appear to be due to the lack of the glycophosphatidylinositol anchor in the recombinant protein. These findings shed a new light on prion protein conversion process and have important implications for the efforts to generate synthetic prions for structural and biophysical studies.
Prion diseases, or transmissible spongiform encephalopathies (TSEs), are infectious neurodegenerative disorders that affect many mammalian species and include scrapie in sheep, bovine spongiform encephalopathy in cattle, chronic wasting disease in elk and deer, and Creutzfeldt-Jakob disease in humans [1–3]. These diseases are associated with conformational conversion of the cellular prion protein, PrPC, to a misfolded form, PrPSc. In contrast to PrPC, which is α-helical and sensitive to proteolytic digestion, PrPSc is rich in β-sheet structure and shows resistance to proteolytic enzymes, with the proteinase K (PK)-resistant core corresponding the C-terminal ~140 residues [1–3]. An increasing body of evidence indicates that the infectious TSE agent is devoid of replicating nucleic acids, consisting mainly–if not solely–of PrPSc [1–3]. This highly unusual pathogen is believed to self-propagate by an autocatalytic mechanism involving binding to PrPC and templating its conformational conversion to the PrPSc state.
Many efforts have been made to recapitulate prion protein conversion and prion propagation in cell-free systems. While early studies have shown that PrPC can be converted to PK-resistant form, PrPres, simply by incubation with PrPSc from TSE-affected animals , the yields of this reaction were very low and no infectivity could be attributed to the newly converted material . An important recent advance was the development of a procedure called protein misfolding cyclic amplification (PMCA) which, using successive rounds of incubation and sonication, is able to replicate and indefinitely amplify the PrPres conformer employing PrPC present in brain homogenate as a substrate . Remarkably, the newly generated PrPres was shown to cause TSE disease in experimental animals , suggesting that it faithfully replicates the structure of brain PrPSc. Furthermore, PMCA-based reactions could reproduce the phenomena of species barriers and prion strains [8,9]. Infectious prions could also be produced by PMCA employing purified PrPC as a substrate (though only in the presence of co-purified lipids and synthetic poly(A) RNA molecules) , providing strong support to the protein-only hypothesis of prion diseases.
Despite growing importance of PMCA technology in prion research, the molecular basis of these conversion reactions remains unclear. Here we report data pointing to an important role of the glycophosphatidylinositol (GPI) anchor as a modulator of PrPres amplification in vitro by PMCA.
Phosphatidylinositol-specific phospholipase C (PI-PLC) and proteinase K (PK) were purchased from Sigma Aldrich, and peptide:N-glycosidase F (PNGase F) was from New England BioLabs. Mouse monoclonal anti-PrP antibodies 3F4 (recognizing epitope 109–112) and 3F10 (recognizing epitope 135–150) were kindly provided by Drs. Richard Kascsak and Yong-Sun Kim, respectively. Recombinant full-length Syrian hamster prion protein (rShaPrP) was expressed in E. coli and purified as described previously .
Ten % (w/v) homogenates of normal Syrian golden hamster and PrP-knockout (FVB/Prnp0/0) mice brains were prepared in PMCA buffer (PBS containing 1% Triton X-100, 0.15 M NaCl, 4 mM EDTA, and the Complete protease inhibitor cocktail [Roche]) as described previously . Samples containing delipidated PrPC were prepared by incubating normal hamster brain homogenate (NBH) with PI-PLC (1 unit/ml) for 2 hrs at 37°C with shaking. In control experiment, NBH was subjected to the same treatment using PI-PLC inactivated by boiling for 1 h. For enzymatic deglycosylation, samples were treated with PNGase F according to the manufacturer’s instruction. Hamster brains infected with 263K scrapie strain were kindly provided by Dr. Richard Carp.
An aliquot (1 μl) of scrapie brain homogenate (SBH) was mixed with 100 μl of NBH or NBH pretreated with active PI-PLC or heat-inactivated PI-PLC, and the samples were subjected to 48 cycles of PMCA as described previously . For serial PMCA, an aliquot of the product of previous PMCA reaction was diluted ten times into fresh NBH, and the mixture was again subjected to 48 cycles of PMCA. For PMCA reactions with the recombinant prion protein (rPrP), rShaPrP substrate at various concentrations (1–25 μg/ml) was added to PMCA buffer containing SBH (1/100 dilution) and the samples were subjected to 24, 48, or 80 cycles of PMCA. When indicated, 10% (w/v) PrP-knockout brain homogenate was included in the reaction mixture. For competition experiments, rShaPrP (2.5–25 μg/ml) was added to PMCA mixtures containing SBH seed and NBH (1:100 ratio), and the samples were subjected to 24 cycles of PMCA.
Western blot analysis of PrP using anti-PrP antibodies (3F4 [1:10,000] or 3F10 [1:5,000]) was performed as described previously . For the analysis of PrPres amplification by PMCA, samples were treated with PK (50 μg/ml, 37°C, 1 hr) prior to SDS-PAGE. For slot blotting analysis, samples in SDS-PAGE sample buffer were diluted 25-fold with Tris-buffered saline (TBS, pH 7.5) and applied to nitrocellulose membrane using Bio-Dot SF microfiltration apparatus (Bio-Rad). The membrane was then treated with 3 M guanidine HCl (10 min, room temperature), rinsed extensively with TBS, and probed with anti-PrP 3F10 antibody (which shows greater sensitivity in probing slot blots than 3F4 antibody). Although it has been suggested that negatively charged nylon membrane may be more useful for slot blotting of delipidated PrPC , using our protocol we found nitrocellulose membrane to be equally suitable.
The standard PMCA protocol  involves suspension of normal brain homogenate (containing PrPC) in PBS containing 1% Triton X-100, 0.15 M NaCl and 4 mM EDTA. A minute quantity of PrPSc from brain of TSE-infected animals is then added, and the mixture is subjected to multiple rounds of cyclic amplification by incubation and sonication. Consistent with the previous report , we found that this procedure results in very efficient conversion of Syrian hamster PrPC to PK-resistant form (PrPres), with the electrophoretic profile of newly generated PrPres faithfully replicating that of input PrPSc (Fig. 1A, lane 2). Here we have made a systematic effort to adopt this PMCA protocol to bacterially expressed rShaPrP. To this end, we dissolved rShaPrP in the standard PMCA buffer, added small quantity of 263K scrapie brain homogenate and subjected the samples to cycles of incubation and sonication. However, these experiments consistently failed to produce measurable quantities of rShaPrP-derived PrPres (Fig. 1A, lanes 3–6). Since additional cellular cofactors may be required for prion protein conversion [10,14], we repeated these experiments using the PMCA buffer supplemented with 10% brain homogenate from PrP-knockout mice. However, also in this case no measurable conversion of rShaPrP could be detected (Fig. 1A, lanes 7–10), even upon increasing the number of PMCA cycles to eighty and using different concentrations of rShaPrP (1–25 μg/ml).
Next we tested whether rShaPrP can interfere with PMCA conversion of brain-derived PrPC. Consistent with previous data , we found that the recombinant protein has a strong inhibitory effect, essentially completely blocking formation of PrPres at concentrations above ~10 μg/ml (i.e., corresponding to ~4-fold excess to PrPC) (Fig. 1B). Thus, rPrP appears to be recognized by PrPSc, competing for the same binding sites with brain-derived PrPC. However, unlike PrPC, it appears to lack the ability to propagate the structure of the PrPSc template.
The apparent inability of rShaPrP to convert to PrPres under standard PMCA conditions is both unexpected and intriguing, especially since tertiary and secondary structures of rPrP appear to be identical to those of brain-derived PrPC . An important difference between these two proteins, however, is the lack of N-glycosylation and the GPI anchor in bacterially-expressed rPrP. The first of these factors appears to have little effect on prion protein conversion in vitro, as indicated by recent data showing that unglycosylated PrPC (obtained by treatment of brain PrPC with PNGase F) can be used as a substrate in PMCA reactions to produce infectious, PrPres-containing prions . On the other hand, the role of the GPI anchor in these conversion reactions has not yet been explored.
To assess the importance of GPI anchor in PrPres formation in vitro by PMCA we have prepared NBH pretreated with PI-PLC. As shown in Fig. 2A, treatment with this enzyme results in a somewhat slower migration of PrPC during electrophoresis, indicating the removal of GPI anchor. This effect is best evident when electrophoretic profiles are compared using proteins that have been deglycosylated by PNGase F treatment (Fig. 2A, lanes 4–6). The abnormal electrophoretic behavior of delipidated PrPC (i.e., slower rather than faster migration when compared with the GPI anchor-containing protein) is consistent with previous reports [13,18]. No such shift in electrophoretic profile was observed in control samples treated with PI-PLC that had been inactivated by heating at 100 °C (Fig. 2A, lanes 2 and 5).
Intriguingly, the electrophoretic band corresponding to PI-PLC-treated sample was much less intense, corresponding to ~25–30% of that in non-treated samples. This effect was consistently observed regardless whether Western blots were probed with 3F4 or 3F10 antibody. Such an apparent ‘loss’ of band intensity in Western blots upon PrPC treatment with PI-PLC was noted previously and attributed to differences in the efficiency of transfer and/or binding to nitrocellulose membrane between GPI anchor-containing and delipidated PrPC . As this effect could interfere with our assessment of the efficiency of PrPres formation by PMCA (see below), the samples used for Western blotting were reanalyzed by slot blotting, a technique that does not include the transfer step. As shown in Fig. 2B, in this case there was no difference in the intensity of bands corresponding to PI-PLC-treated and untreated PrPC. Thus, using our protocol slot blotting is better suited for quantitative comparison of samples containing native and delipidated PrPC.
In contrast to robust PrPSc-templated PMCA conversion of native PrPC to PrPres, no amplification of PrPres could be detected by Western blotting when PrPC without the GPI anchor (i.e. pretreated with active PI-PLC) was used as a substrate (Fig. 3A). To verify that this was not due to poor detection of delipidated PrPC in Western blotting analysis as discussed above, the presence of PK-resistant form of PrP in PMCA-derived samples was further tested by slot blotting. As shown in Fig. 3B, also using this technique we were unable to detect any PrPres material in the product of PMCA reaction performed on PI-PLC treated samples, again in contrast to the presence of large quantities of such material in samples containing PrPC with the native GPI anchor. Next we performed three more rounds of serial PMCA, mixing in each round fresh PI-PLC-treated NBH with a small aliquot (1:10 ratio) of the product of the previous round reaction. Also in this case, we failed to detect any amplification of PrPres in samples containing delipidated PrPC, in contrast to robust serial amplification observed using native (i.e. PI-PLC untreated) brain homogenate as a substrate (Fig. 3C). Altogether, these data clearly demonstrate that the GPI anchor plays an important role in prion protein conversion in vitro, and that the removal of this anchor by PI-PLC treatment greatly diminishes–if not completely eliminates–the ability of PrPC to act as a substrate for amplification of PrPSc conformation by a standard PMCA procedure.
PMCA, a procedure that allows essentially indefinite amplification of PrPres and TSE infectivity in vitro using PrPC present in normal brain homogenate as a substrate, has emerged as a powerful tool in prion research [7–9]. The molecular mechanism of this amplification reaction remains, however, largely unexplored. Furthermore, there is major interest in expanding this approach to the recombinant prion protein, both from the perspective of prion diagnostics as well as fundamental studies of the PrPC→PrPSc conversion mechanism .
Here we report that, under the conditions of the standard PMCA protocol, bacterially expressed rPrP is a very poor substrate for amplification of PrPres from scrapie-infected hamsters. Furthermore, we found that, when added to NBH, rPrP acts as an efficient inhibitor of the PMCA conversion of brain PrPC. Building on these observations, we tested the role of PrPC GPI anchor, finding that removal of this anchor by PI-PLC treatment renders PrPC incompetent to act as a substrate for the amplification of PrPres by the standard PMCA protocol.
The finding that native GPI anchor is important for the amplification of PrPres conformation in vitro by the PMCA protocol is unexpected, especially since previous studies have shown that both GPI-free PrPC derived from mammalian cell culture [4,20] as well as bacterially expressed rPrP  can be converted to PrPres in a “discontinuous” cell-free conversion assay developed by Caughey and coworkers [4,22]. However, the findings of those previous cell-free conversion studies and our present PMCA experiments are not necessarily contradictory as–despite an apparent similarity–the assays used in these two types of studies probe distinct phenomena. Thus, in PMCA experiments a minute quantity of PrPSc is used as a seed, and essentially an infinite number of substrate PrPC molecules is converted to PrPres conformation that appears to faithfully replicate that of the PrPSc template, with the newly generated material being infectious . This implies that, upon binding to the surface of PrPSc seed, each molecule of the substrate protein acquires catalytic properties of the seed, acting as a new template for the conversion of additional substrate molecules. By contrast, the reactions in discontinuous cell-free conversion assays are characterized by very low yields and appear to lack any autocatalytic properties, with the number of PrPC molecules converted to PrPres being substantially lower compared to input PrPSc [4,22]. Given these substoichiometric yields, it is likely that in these reactions only single substrate molecule can be recruited and converted to PrPres state by each catalytic site of PrPSc oligomers, effectively ‘capping’ these sites and preventing the conversion of additional substrates. While PrP molecules converted in this reaction acquire seed-like PK-resistance, their three-dimensional structure might be not identical to that of PrPSc. This would explain both the lack of autocatalytic properties of this reaction as well as an apparent lack of infectivity of the newly converted material . The capping mechanism described above may also be operational in our PMCA experiments using rPrP as a substrate, accounting for the observed inhibitory effect of the latter protein on the formation of PrPres from brain-derived PrPC.
It should be noted that a modified PMCA protocol has been recently developed in which the original PMCA buffer (containing Triton X-100 as a sole detergent) was supplemented with an acidic detergent, SDS . Using this modified PMCA method, it is possible to convert, in an autocatalytic fashion, bacterially expressed rPrP to fibrils apparently displaying PrPSc-like PK-digestion pattern (with a 16–17 kDa PK-resistant fragment similar to that observed for nonglycosylated PrPSc). However, this scrapie-like PK-resistance persists only if PK-digestion is performed in the presence of specific detergents used in the PMCA buffer (especially SDS), whereas upon removal of the detergents from already converted material the size of the longest PK-resistant fragment is reduced to ~12 kDa . Thus, while this modified PMCA protocol allows efficient, PrPSc-seeded conversion of rPrP, the major structure propagating in this reaction does not fully match that of the PrPSc seed.
The present findings in the cell-free environment seem to echo previous observations that the absence of the GPI moiety in PrPC diminishes propagation of PrPSc in cell culture models of prion infectivity [23,24]. However, the GPI anchor is not obligatory for prion replication in vivo. This is clearly indicated by the recent study showing that transgenic mice expressing anchorless PrP were susceptible to infection with the scrapie agent, though these mice did not develop clinical symptoms and replication of infectivity was reduced as compared to wild-type mice . In view of these data in vivo, the apparent inability of GPI-deficient PrP to support amplification of PrPres in PMCA reactions is highly intriguing. Our present data suggest that the conversion reaction of the anchorless PrP may require specific microenvironment and/or cofactors that are present in vivo, but are lost in homogenized samples used in PMCA reactions. One critical factor in this regard may be the interactions of the protein (via the anchor or directly) with biological membranes. In the case of GPI-containing PrP, these surface interactions (occurring largely through the anchor) are likely to be partially preserved even upon tissue homogenization in the presence of Triton X-100 micelles, allowing for highly efficient PrPres amplification by PMCA. In the absence of the GPI anchor, however, nonspecific surface interactions are likely to be compromised much more readily. It appears that in the latter case, these interactions and a conversion-conducive environment can be mimicked to a certain degree in vitro only in the presence of acidic detergents such as SDS, consistent with a highly basic character of the prion protein.
The present findings are also relevant to understanding the origin of prion infectivity in body fluids such as urine and amniotic fluid . Recent data indicate that PrP present in these fluids is both N-terminally truncated as well as deficient in the GPI anchor (Notari, S., Gambetti, P and Chen, S., unpublished data). Our finding that, at least under certain conditions, PrP lacking the native GPI anchor is a poor substrate for PrPres amplification suggests that PrPSc associated with prion infectivity detected in urine  may be derived from other sources than the delipidated PrP normally present in this body fluid.
The authors thank Dr. Quingzhong Kong for providing PrP-knockout mice. This work was supported in part by National Institutes of Health grants NS44158 and AG14359.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.