Prion propagation is controlled by discrete structural regions of PrP rather than overall stability

Abstract

Mutational analysis of the cellular prion protein (PrP^C) has revealed various regions of the protein that modulate prion propagation. However, these approaches involve deletions, insertions, or replacements in the presence of the WT PrP^C, which may mask the true phenotype. Here, site-directed alanine mutagenesis of the prion protein (PrP) was conducted to identify sites, particularly potential “surface patches” of PrP^C required for prion propagation. Mutations were targeted to the helical, sheet, and loop regions of PrP^C, or a combination thereof and the mutated proteins expressed in mouse neuroblastoma cells in which the endogenous PrP^C had been silenced. Using the scrapie cell assay, a highly sensitive cell culture-based bioassay for quantifying infectious titers of Rocky Mountain Laboratory (RML) prions, we found that all mutations within the structured 119 to 231 domain, irrespective of secondary structure, severely reduced prion propagation. We further analyzed four strongly inhibiting mutations within conformationally variable loop regions of PrP^C - residues 123 to 125 (PRP1); 134 to 135 (PRP4); 139,141 (PRP5), and connecting helix residues 146 and 153, respectively, as well as residues 188,191 to 192 (PRP13). Mutations in PRP1, 4 and 5 dominantly inhibited prion propagation even in the presence of WT PrP^C, while mutations in PRP13 had no comparable effect, suggesting that the former disrupt prion fibril structure or its replication mechanism. None of the most inhibitory mutations substantially altered PrP stability, and mutant PrPs were able to both form amyloid in vitro and seed fibril formation of WT PrP.

Mammalian prion diseases include the fatal transmissible neurodegenerative conditions affecting humans and a range of other species (1). Prions consist of a “cloud” of assemblies of misfolded prion protein (PrP) that propagate by recruitment of monomers of the cellular PrP (2, 3, 4, 5, 6, 7). Prion propagation is an autocatalytic process of seeded fibrillization and fission and involves considerable changes to the overall structure of PrP, from a more α-helical protein to one with a higher β-sheet content (2, 3, 4, 5). Recent electron microscopic analyses of exceptionally pure high-titer infectious hamster 263K and mouse Rocky Mountain Laboratory (RML), ME7, and 22L prions have identified rods comprised of single and twisted pairs of helical protofilament amyloid fibrils where a single PrP monomer is each rung of the protofilament (8, 9, 10, 11).

The disease-associated form of PrP, PrP^Sc, was originally defined in terms of its proteinase K (PK) resistance and detergent insolubility and is pathognomic of prion infection (4, 12). However, it is now clear that the majority of infectivity in hamster 263K and RML prion isolates is protease-sensitive (13, 14, 15). While these protease-sensitive infectious forms are completely digested upon limited PK treatment, they can still be competent to infect susceptible cell models and seed the formation of disease-associated PrP assemblies, which are only partially susceptible to N-terminal digestion by PK and remain detectable by anti-PrP antibodies, such as ICSM18, upon denaturation (16). Prion infection of a cell can thus be determined by detection of protease-resistant PrP^Sc deposits and indeed this forms the basis of the scrapie cell assay (SCA) (17) used to measure prion infectivity in this study.

Cellular prion protein (PrP^C) has a flexible N-terminal domain (residues 23–118) and a structured C-terminal globular domain (residues 119–231) comprised of three α-helices (H1, H2, and H3), and a short antiparallel β-sheet, consisting of two β-strands (S1 and S2) (18, 19). Helices H2 and H3 are linked by a disulphide bridge. H1 is the shortest of the helices, spanning residues 143 to 153. It is one of the most immunogenic regions of the protein as numerous anti-PrP antibodies, including ICSM18, have been mapped to this region and found to recognize PrP^C as well as misfolded disease-associated forms upon denaturation (16, 20, 21). H2 (residues 171–193) and H3 (residues 199–226) are larger in comparison and represent regions of PrP where the largest number of pathogenic mutations are clustered (22).

Previous approaches to identify regions within the structured domain of PrP^C that influence prion formation have produced conflicting results. All secondary structure components were suggested to influence prion formation via numerous mechanisms. They range from: specific protein sequence elements such as the S2-H2-loop for RML, 22L, and chronic wasting disease prions (23, 24, 25, 26); charge interactions for RML prions (27, 28, 29, 30); stability for RML and human sporadic Creutzfeldt-Jakob Disease prions (31, 32, 33, 34, 35); amyloid-seeding and fibril-forming potential of individual residues in PrPs and M1000 mouse prions (36, 37, 38, 39). Metal-binding (40) as well as the protein microenvironment (41, 42, 43) also play a role in prion formation.

Here, the effect of amino acid substitutions on prion propagation were analyzed without interference from either molecular tags or the endogenous PrP, by targeted alanine replacement, which eliminates side chains beyond the β-carbon, with minimal perturbation of the protein backbone conformation. Alanine replacement has been extensively used to probe the influence of specific amino acid side chains on bioactivity in vivo, and on protein stability, or the folding pathways in vitro (44, 45). It is applicable to a wide range of amino acids, as alanine contains an inert, nonbulky methyl group, and retains the secondary structure preferences of many other amino acids, thus minimally affecting secondary structure. In some cases, however, alanine replacement may perturb the folding or stability of the modeled protein (46), and care must be taken with regard to the nature of the replaced amino acid. Alanine replacement has previously been used to probe the stability and the folding pathway of PrP in vitro (46, 47), prion replication in N2a cells chronically infected with RML prions (48, 49), and amyloid fibril formation, and stability (50, 51, 52). To address the effect of PrP mutations on RML prion replication in vivo without possible interference from WT PrP, we performed a systematic alanine replacement of spatially proximal residues within the C-terminal domain of PrP^C (residues 119–231). The mutant proteins were expressed in cells silenced for endogenous PrP expression (PK1-KD cells) (53, 54), in contrast to previous studies which have used cells expressing endogenous PrP, or N2a cells chronically infected with RML prions (ScN2a) (48, 55). The aim was to determine if there was a critical “surface patch” of the protein or whether the hydrophobic core dominated efficient RML prion propagation.

Results

Alanine mutagenesis of the mouse PrP region 119 to 231

Mutations to alanine were introduced at various sites in mouse PrP (PDB entry 2L39) (56) to identify potential “surface patches” required for efficient prion propagation. Eighteen constructs were generated: 16 targeted surface amino acids within the globular domain, and two targeted core hydrophobic residues (Fig. 1A). Groups of three surface residues, spatially close to each other in the native 3D structure, were replaced simultaneously to create mutated “surface patches’ across the protein (Fig. 1B, shown in yellow). Within the core hydrophobic regions of the protein, replacements to alanine, as well as deletion of the disulphide bridge, were undertaken at specific sites aimed at reducing protein stability and promoting protein unfolding (57) (Fig. 1B, shown in blue).

Figure 1.

Mutagenesis of the mouse prion protein structured region (119–231). MoPrP mutations were created in the structured region of the protein, as triple alanine replacements to surface residues. A, the full list of mutations prepared is detailed in a tabular format, listed by mutation (N- to C-terminal), nomenclature used (PRP1-18), and secondary structure elements affected (loop/strand/helix). B, targeted mutations mapped on to PDB entry 2L39; images created in PyMol software (PyMol Molecular Graphics System, Version 2.0, Schrödinger LLC). Residues that remained unchanged are shown in gray; mutated surface residues are shown in yellow and mutated buried residues are shown in blue. The binding epitope of the anti-PrP ICSM18 antibody (aa 143–153, Helix 1) is indicated. PDB, Protein Data Bank; PrP, prion protein.

The mutations were stably expressed by retroviral infection of PK1-KD cells in which endogenous PrP expression has been stably silenced (54). As the level of PrP^C is reduced by about 90% in these cells, they are resistant to prion infection, but upon reconstitution with WT mouse PrP regain full susceptibility to prion infection, as determined by SCA (17, 54). We reconstituted PK1-KD cells with the 18 mutants (PRP1-18), derived bulk cultures of stably transduced cells and analyzed them for PrP expression.

Mutant PrPs were expressed and trafficked to the cell surface

In all the mutants, PrP was expressed and localized to the cell surface (Fig. 2A). The pattern of expression was the same as for WT PrP^C in PK1 cells, a derivative of N2a cells that have previously been shown biochemically to express PrP on the cell surface (57). Although the PrP expression level was variable, all mutants except PRP8, 12, 13, and 18 were expressed at levels equivalent to endogenous PrP^C and exhibited the same pattern of three bands that correspond to endogenous PrP^C that is non-, mono-, and di-glycosylated PrP in the parental PK1 cells (Fig. 2B). Mutant PRP9 expressed multiple PrP bands, which migrated slightly differently from WT PrP^C. PrP proteins expressed by mutants PRP8, 12, 13 and 18 migrated differently. PRP8 protein migrated close to mono-glycosylated PrP^C whereas PRP12, 13, and 18 proteins migrated differently from any of the three bands corresponding to WT PrP^C (Fig. 2B). Nevertheless, the mutant PrPs were all expressed on the cell surface (Fig. 2A) indicating that the altered migration probably due to impaired glycosylation did not prevent them from being trafficked to the cell surface.

Figure 2.

Analysis of PrP expression.A, immunofluorescence analysis of PK1, PK1-KD, and PK1-KD cells reconstituted with the wild-type PrP [moPrP^WT (WT)] or alanine mutants. SET1: analysis of PRP1-7 using anti-PrP antibody ICSM18 and ICSM35 for PRP7, WT and KD cells (indicated with ’); PRP7 was recognized by ICSM35 (epitope aa 93–105) but not by ICSM18 (epitope aa 143–153 overlaps mutated region). SET2: analysis of PRP8-16 using anti-PrP antibody ICSM18. SET3: analysis of PRP17 to 18 using anti-PrP antibody ICSM18. All reconstituted KD lines expressed moPrP on the cell surface. DAPI, nuclear stain; moPrP, green. Magnification at x40. B, Western blot analysis of lysates prepared from bulk cultures of PK1-KD cells stably reconstituted with the indicated mutants. PK1 correspond to lysates prepared from PK1 cells, KD from PK1-KD cells, pLNCX2 from cells reconstituted with the empty pLNCX2 retroviral vector and WT from cells reconstituted with wild-type mouse PrP. Immunoblot analysis was conducted using the anti-PrP antibody ICSM35. The data presented are representative of multiple western blots where several lysates were analyzed and show that reconstituted cells express varying levels of PrP isoforms. KD, knockdown; PK, proteinase K; PrP, prion protein.

Mouse PrP is glycosylated at asparagine 180 and 196. Even though neither of these residues was mutated in our study, the T198A mutation in PRP8 is part of the consensus sequence for N-linked glycosylation of asparagine 196 and therefore likely to affect its glycosylation. PRP8 did not express di-glycosylated PrP but the mutant PrP was still trafficked to the cell surface (Fig. 2A) in accordance with a previous study that showed that a T198A mutation in PrP eliminated glycosylation of this site but did not alter its trafficking to the cell surface (58). Mutants PRP12, 13 and 18 also exhibit aberrant migration but were trafficked to the cell surface. PRP12 contains V179A.I183A whereas PRP18 contains C178A, very close to asparagine 180; it is possible that these mutations perturb glycosylation at this residue. PRP13 contains mutations close to both glycosylation sites at asparagine 180 and 196 thereby affecting their glycosylation.

Some of the mutants (PRP3, 4, 10, 11, and 15) affected PrP trafficking leading to partial intracellular accumulation as observed previously (59, 60). It has been well documented that GPI-anchored cell surface expression of PrP^C is necessary for replicating infectivity (58), but the actual level of PrP^C expression is not important. Enari et al. found that N2a/Bos2 cells, a clone of susceptible cells, expressed PrP^C at the same low level as the original N2a cells whereas a resistant cell line expressed it at a 10 times higher level (61). PrP^C overexpression did not increase susceptibility either leading them to conclude that it is a prerequisite for prion propagation, but other factors are essential for susceptibility to prion infection (61). Marbiah et al. also showed that overexpression of PrP^C did not render revertant cells susceptible to RML prion infection and did not increase the susceptibility of prion permissive cells (62). These findings are consistent with the two-phase model of in vivo prion kinetics in mice where the exponential phase 1 of RML prion replication is not rate limited by PrP^C expression level (63).

Mutations to PrP at surface-exposed sites, as well as core hydrophobic residues, severely inhibit RML prion propagation

PrP mutations were analyzed in three sets across residues 123 to 225. Sets 1 (PRP1-7: residues 123–151; Fig. 3) and 2 (PRP8-16: residues 148–225; Fig. 4) primarily covered surface residues, whereas set 3 (PRP17–18; Fig. 5) targeted buried hydrophobic residues. Propagation of RML prions was measured by the SCA as the number of prion-infected cells detected per 25,000 PK1-KD cells reconstituted with each construct compared to WT PrP. Cells were passaged through six rounds by splitting biweekly 1:8. The number of infected cells in splits 3 to 6 defined as the “spot number” was determined by counting the number of PrP^Sc-positive (indicating prion-infected) cells identified, using anti-PrP ICSM18 antibody.

Figure 3.

RML prion infection of PK1-KD cells expressing mutations in region 123 to 151 (PRP 1–7).A, PK1-KD cells reconstituted with the WT PrP or alanine mutants of moPrP were challenged with RML prions and assayed for their relative ability to propagate prions via SCA for four consecutive passages. Box plot shows maximum, minimum, median, and 25%, 75% quartiles; n = 8. p-values of each postsplit 6 sample compared to moPrP^WT at the same split calculated in a one-way ANOVA with a Bonferroni correction for multiple comparisons of infected cell counts from the sixth split were below 10⁻⁵. B, mutations in PRP1-7 are depicted on PrP (PDB entry 2L39), with target residues for mutation highlighted. Side chain atoms are colored according to element, with nitrogen blue, carbon green, hydrogen white, and oxygen red, with mutated residues highlighted in green. C, mutations in PRP1-7 are depicted on the RML prion structure (PDB entry 7QIG), with target residues color-coded for relative inhibition of prion replication; mutants in blue severely reduce prion propagation whereas those in yellow have an even greater effect. PRP7 is uncolored, as it affects the epitope of the ICSM18 antibody and so could not be assayed for its effect on prion propagation in SCA. All images were created in PyMol. KD, knockdown; PDB, Protein Data Bank; PK, proteinase K; PrP, prion protein; RML, Rocky Mountain Laboratory; SCA, scrapie cell assay.

Figure 4.

Infection of PK1-KD cells expressing mutations in region 148 to 222 (PRP 8–16) with RML.A, PK1-KD cells reconstituted with the WT PrP or alanine mutants of moPrP were challenged with RML prions and assayed for their relative ability to propagate prions via SCA for four consecutive passages. Box plot shows maximum, minimum, median, and 25%, 75% quartiles; n = 8. p-values of each postsplit 6 sample compared to moPrP^WT at the same split calculated in a one-way ANOVA with a Bonferroni correction for multiple comparisons of infected cell counts from the sixth split were below 10⁻⁵. B, mutations in PRP 8 to 16 are depicted on PrP (PDB entry 2L39), with target residues for mutation highlighted. Side chain atoms are colored according to element, with nitrogen blue, carbon green, hydrogen white, and oxygen red, with mutated residues highlighted in green. C, mutations in PRP 8 to 16 are depicted on the RML prion structure (PDB entry 7QIG), with target residues color-coded for relative inhibition of prion replication; target residues that comprise PRP13 are highlighted. KD, knockdown; PDB, Protein Data Bank; PK, proteinase K; PrP, prion protein; RML, Rocky Mountain Laboratory; SCA, scrapie cell assay.

Figure 5.

RML prion infection of PK1-KD cells expressing mutations in region 204 to 213 (PRP 17–18).A, PK1-KD cells reconstituted with the WT PrP or alanine mutants of moPrP were challenged with RML prions and assayed for their relative ability to propagate prions via SCA for four consecutive passages. Box plot shows maximum, minimum, median, and 25%, 75% quartiles; n = 8. p-values of each postsplit 6 sample compared to moPrP^WT at the same split calculated in a one-way ANOVA with a Bonferroni correction for multiple comparisons of infected cell counts from the sixth split were below 10⁻⁵. B, mutations in PRP17 to 18 are depicted on PrP (PDB entry 2L39), with target residues for mutation highlighted. Side chain atoms are colored according to element, with nitrogen blue, carbon green, hydrogen white, and oxygen red, with mutated residues highlighted in green. C, mutations in PRP17 to 18 are depicted on the RML prion structure (PDB entry 7QIG), with target residues comprising PRP17 and 18 color-coded for relative inhibition of prion replication. KD, knockdown; PDB, Protein Data Bank; PK, proteinase K; PrP, prion protein; RML, Rocky Mountain Laboratory.

All mutations in set 1 and 2, reduced the level of prion propagation to a fraction of that for WT mouse PrP, indicating that regions 123 to 151 and 148 to 225 are highly sensitive to mutations at surface sites, as substitution of up to three residues within the native sequence to alanine resulted in a large decrease in propagation (Figs. 3A and 4A). However, for most PrP mutants some prion-infected cells were detected, albeit at low levels, allowing us to quantitatively distinguish between the mutants. Set 1 mutations (PRP1-7, region 123–151), were slightly more inhibitory than set 2 (PRP8-16, region 148–225) as fewer prion-infected cells were observed in the SCAs with these mutants (p = 0.049; comparing post sixth split data in Figs. 3A and 4A). Furthermore, mutations to conformationally variable regions, PRP1 (Fig. 3A) and PRP13 (Fig. 4A), abrogated propagation and reduced spot numbers to background levels by split 6, even lower than the control PK1-KD cells. PRP7 also produced very few spots, but since this mutant protein was not recognized by the anti-PrP ICSM18 antibody no conclusion can be drawn on prion propagation (Fig. 2A).

Set 3 mutations were replacements in the core hydrophobic region of PrP, including those thought to have destabilizing and unfolding properties: PRP17 (M204A), and PRP18 (C178A.C213A), that result in loss of the PrP disulfide bridge (Figs.1A and 5B) (47, 64). Both mutants significantly reduced prion propagation (Fig. 5A). An additional PrP mutant, in which several core residues were replaced with glutamine (M204Q.M205Q.V209Q.M212Q), similarly reduced prion propagation (unpublished data).

Mutants PRP1 and PRP13 abrogate RML prion propagation by different mechanisms

Among the mutants that exhibited diminished propagation, the largest decrease in infectivity occurred in those affecting the conformationally variable regions of PrP^C.: PRP1 (G123A.L124A.G125A), N-terminal to S1 (Fig. 1A) and PRP13 (V188A.T191A.T192A), at the C terminus of H2 (Fig. 1A). Prion propagation in PK1-KD cells reconstituted with these two mutants was akin to background PK1-KD levels if not lower by the last split, indicating complete abrogation of prion propagation (Figs. 3A and 4A). The residues mutated in PRP1 (123–125) display a high degree of conformational flexibility, as shown by NMR relaxation measurements (65), as they immediately precede the main structured domain of PrP, and are unstructured. Mutations in PRP1 are therefore unlikely to affect the structure of PrP. This was confirmed by guanidine hydrochloride (GdnHCl) denaturation experiments, in which none of the mutants analyzed (PrP 1, 5 or 13) showed decreased stability compared with WT PrP (Table 1 and Fig. S1). Similarly, the C-terminus of H2, also displays elevated conformational variability relative to other structural elements in the C-terminal domain (65, 66). It is markedly less stable than the remainder of H2 (67), and, depending on solution conditions, adopts an imperfect helical conformation as seen in a number of NMR structures (56, 68). Its structural integrity is not mandatory for 127S ovine prion replication, as insertions within this region have no limiting effect on prion replication (69), although insertions affecting the charge interaction between the disease-associated residue H186 and K193 do inhibit prion propagation (69). As in the case of PRP1, alanine substitutions in this region of the protein did not decrease stability of PrP and are thus unlikely to significantly affect the structure of this region of PrP. Instead, they may act through intermediate states or may affect the structure and stability of PrP in its pathologic conformation.

Table 1.

Mutant PrP 23-231 Gibbs free energies of unfolding

PrP mutant	ΔG, kJ/mol
PRP1	28.5 ± 0.4
PRP5	28.8 ± 0.3
PRP13	26.5 ± 0.3
WT	25.8 ± 0.4

Free energies of unfolding of PrP mutants were calculated from guanidine hydrochloride denaturation.

We selected several mutants exhibiting strong repression that represent distinct regions of PrP for further analysis (PRP1, loop1; PRP 4 and 5, loop2-H1; PRP13, H2). We first examined mutations PRP1, PRP4, PRP5, and PRP13 for their potential to inhibit prion replication driven by WT PrP at physiological levels (Fig. 6). Strikingly, mutations in the N-terminal lobe of the prion structure (Fig. 3C) abrogated prion replication, whereas PRP13 in the C-terminal lobe (Fig. 4C) had little to no effect in the presence of WT PrP (Fig. 6). These data suggest that PrP carrying mutations in the N-terminal lobe may bind to WT prions and either prevent fibril elongation or disrupt the structure of the prion fibril after being incorporated. In contrast, PRP13 either does not bind to prion fibrils or does not disrupt their structure when incorporated.

Figure 6.

RML prion infection of WT PK1 cells expressing mutations PRP1, 4, 5, and 13. PK1 cells endogenously expressing WT PrP were stably transduced with retroviruses expressing the triple-alanine PrP mutants to generate WT and mutant PrP coexpressing PK1 cell lines. These cultures were then infected with RML and passaged on a biweekly basis. Infected cell counts were collected post third, fourth, fifth, and sixth passage to determine degree of infection persisting in the cell cultures. One-way ANOVA with a Bonferroni correction was performed for multiple comparisons of infected cell counts from the sixth split of mutant PrP expressing PK1 cells to control PK1 cells with only WT PrP expression; n = 8. PK, proteinase K; PrP, prion protein; RML, Rocky Mountain Laboratory.

Mutant PrP fibrils are seeding competent in vitro

We determined whether mutant PrP PRP1, 5, and 13 are incorporated into PrP fibrils. Ideally, this would be done with infectious prions. However, none of the four mutants formed prions in cell culture, which could have served as infectious seeds (Figs. 3A and 4A). We therefore performed seeded and unseeded aggregation assays under native conditions in vitro and recorded fibril formation kinetics via Thioflavin T fluorescence (Fig. 7). All mutant PrPs formed fibrils under native conditions. Both PRP1 and PRP5 kinetics were significantly delayed when compared to WT PrP, while PRP13 aggregated faster than WT PrP under unseeded conditions (Fig. 7, A and B). However, this effect was ameliorated when homologous seeds were added with PRP5 kinetics being identical to WT PrP.

Figure 7.

Seeded and unseeded fibril formation in vitro of WT PrP and triple alanine mutations PRP 1, 5, and 13.A, monomeric recombinant PrP was incubated with shaking under near native conditions and ThT fluorescence was monitored in a fluorescence microplate reader. Representative curves for each tested mutant and corresponding fitted sigmoidal curves were plotted. B, aggregation kinetics were analyzed in GraphPad Prism and fitted against a sigmoidal curve model, generating t₅₀ values for unseeded aggregation (n = 5). One-way ANOVA with a Bonferroni correction was performed for multiple comparisons to aggregation kinetics of full-length mouse WT. C, in vitro aggregation assays were performed with the addition of seeds generated from a spontaneous aggregation assay of the same PrP species as the PrP monomer substrate. Representative aggregation kinetics and sigmoid fits for each homologous seeded amyloid aggregation experimental condition were plotted as in A and analyzed as in B. D, fitted t₅₀ values of homologous seeding assays (n = 6). E–H, mixed seeding amyloid aggregation assay was performed under same near native conditions, with the heterogeneous pairing of the PrP mutant as seed and the full-length mouse WT PrP as substrate (E–F) or vice versa (G–H, n = 6 for each condition). PrP, prion protein; ThT, Thioflavin T.

Next, we tested whether there was a seeding barrier between mutant PrP seeds and WT PrP substrate (Fig. 7, E and F) and vice versa (Fig. 7, G and H). All fibrils formed from mutant PrPs seeded WT PrP aggregation. However, PRP13 fibrils showed significantly reduced seeding capacity when compared to WT PrP or PRP1 and PRP5. All mutants aggregated when seeded by WT PrP fibrils. Their respective aggregation kinetics closely mirrored those in the homologous seeding experiment (Fig. 7, D and H), suggesting that aggregation kinetics were dominated by the PrP substrate. These data further suggest that no “species barrier” for seeding exists for fibril formation in vitro under these conditions.

Loss of RML prion propagation is more severe for mutations in H1 and H2, rather than H3

Propagation was among the lowest for PRP1 (G123A.L124A.G125A) and PRP13 (V188A.T191A.T192A) as well as PRP4 (S134A.R135 AM153A) and PRP5 (H139A.G141A.D146A; Figs. 1A and 3B). In contrast, some mutants that exclusively targeted H3 (PRP11 and PRP16, Figs.1A and 4B), although severely reduced in their ability to propagate RML prions, permitted low-level prion propagation and reported numbers of infected cells higher than those observed for mutants which are within and adjacent to H1 and H2 (Figs. 3B and 4B). We were unable to assess the effect of mutagenesis in the central H1 residues of the protein, mutant PRP7, as mutations in this region destroy the ICSM18 epitope, used for the SCA (Fig. 2A). Overall, mutations to the surface regions of the protein (PRP1- PRP16) resulted in severe loss of prion propagation, regardless of the region mutated.

Discussion

In this study, we employed site-directed alanine mutagenesis of surface residues within the structured domain (119–231) of the mouse PrP to identify whether specific regions of the protein are crucial for RML prion propagation. All mutations within the structured domain, irrespective of secondary structure, severely reduced prion propagation. The reduction was exacerbated for mutants targeting conformationally variable regions (PRP1 and PRP13) and those neighboring or within H1 (PRP4 and PRP5). These sites may represent essential surfaces on PrP^C for template-assisted conversion following prion infection or may disrupt the prion structure itself as discussed below. Our data provide evidence in support of the three previously proposed regions important for successful infection and propagation of prions: the glycine rich region (residues 123–127) (39); H1 and the neighboring residues (27); and the core residues that provide stability (34, 70). These results corroborate previous studies that have identified the C terminus of H2 as a key region in the conversion of PrP^C to its more β-sheet disease-associated form (46, 71). Pathogenic mutations associated with inherited prion disease are also found in this region: T183A, H187R, T188R/K/A, E196K, F198S, E200K, and D202N (72, 73, 74).

Complete abrogation of propagation was observed for PRP1 and PRP13 (Figs. 3 and 4), when present exclusively as substrate for prion propagation. There was a progressive drop with successive splits; split 6 values were below those observed for non-reconstituted PK1-KD cells. This suggested that the face of the globular domain in moPrP that bears these regions is an important site for prion propagation. Point mutations G122A and L124A, severely affect the ability of cells to take up and propagate M1000 mouse prions (39). Mutation G125A, also present in PRP1, was previously shown to offer partial resistance to prion infection, but not to the same extent as G122A and L124A (39). The fact that the PRP1 mutation, like PRP4 and PRP5, prevents prion replication in the presence of WT PrP suggests that it blocks the elongation of prion fibrils, disrupts fibril stability when incorporated into a WT fibril or both.

Glycine to alanine mutations at positions G123A and G125A (PRP1) are predicted to disrupt the hairpin loop in the interlobe interface of the prion fibril and thus destabilize the amyloid fold (Fig. 3C) (11). It remains to be investigated whether these sites are also required for contact between the native and disease-associated forms, native protein unfolding, association with propagation-facilitating factors, or are important residues for the stability of the disease-associated PrP assemblies. This glycine-rich region has been proposed to have lipid bilayer properties that may allow the protein to engage in transmembrane helix-helix interactions at this site (75). In support of a conformational change in this region, it has been shown that the anti-prion monoclonal antibody 1C5, whose epitope spans residues 119 to 130, can recognize abnormal PrP more effectively than the normal PrP (76).

Unlike PRP1, the PRP13 mutation did not prevent prion propagation in the presence of WT PrP. This suggests a different mechanism of inhibition, which could be either an inability to bind and be incorporated into the prion fibril or a substantial barrier to conversion from the native into the prion fold. It is not obvious how the PRP13 mutations could prevent the binding of monomers to the nascent fibril, as they lie within a conformationally variable region of the prion (Fig. 4C). Therefore, the mutation may affect the misfolding pathway of PrP rather than the stability of the prion fibril. In PRP13, the PrP region affected is the C terminus of H2 (residues 187–194), which despite its high sequence conservation displays a high degree of conformational variability. NMR data indicate that the dominant α-helical structure of this region of the protein is unstable and interchanges with other lower populated structures (19, 67, 77). Indeed, Salamat et al. have shown that peptide insertions here have no limiting effect on PrP conversion and concluded that structural integrity within this region was not mandatory for successful 127S ovine prion propagation (69). However, peptide insertion at K193, but not V188 or T190 in the loop region between H2 and H3, inhibited propagation (69). Our data showed a strong restriction on prion propagation in mutant PRP13 (V188A.T191A.T192A). This region is important in terms of prion disease, as the disease-associated mutation of the neighboring residue H187 to arginine has a dramatic effect on PrP folding and stability, resulting in a molecule that displays a markedly increased propensity to oligomerize (71). In addition, there are four consecutive threonine residues in this region (residues 190–193) that are found almost exclusively in a strand and/or loop conformation (irrespective of the identity of the flanking amino acids) (78).

Residues at the C terminus of H2, clustered around H187 and T188, display high conformational flexibility, which enables a transition from α-helical to β-sheet or random coil states. This region contains many so-called “α/β discordant” residues, proposed to be associated with amyloid formation (37). Dynamic simulation of pathogenic mutations showed specific perturbation of loop H2-H3 for all mutations tested (79). Our data from PRP13 highlighted native residues V188, T191, and T192 within the C terminus of H2 as having a key role in prion propagation (Fig. 4A). Whether viewed from peptide insertion data, dynamic simulation, or SCA, it was evident that prion propagation can be modulated by the C terminus of H2 and the H2-H3 loop. Conformationally variable regions are unlikely to have a significant effect on the structure or stability of PrP^C, and neither mutation decreased the thermodynamic stability of PrP^C. The finding that PRP1 and PRP13 mutations completely abrogate prion propagation suggests they may be acting downstream of PrP^C and perturbing the balance between propagation and cellular clearance mechanisms, possibly through an interaction partner.

The data also suggest that residues in and around H1 have an important role in prion propagation, as both PRP4 (S134A.R135AM153A) and PRP5 (H139A.G141A.D146A) exhibited a marked reduction in prion propagation (Fig. 3A) and abrogated propagation of WT prions (Fig. 6). These mutations replace polar residues on the inside of the N-terminal lobe of the prion fibril, which may disrupt the hydrogen-bonding network in this region, destabilizing H1. Indeed, the H1 region has previously been shown to play a crucial role in prion propagation in cell-free conversion studies, with aspartate residues and salt-bridges, within H1 thought to protect PrP from conversion through increased helix stability of the native state (31). Alternatively, residues neighboring and within H1 (143–153) have been proposed as interaction sites where disease-associated PrP interacts with PrP^C, consistent with the potential interaction surface identified in this study. It is notable that the epitope for the ICSM18 antibody used in this study lies within H1. This antibody inhibits prion replication as well as prion induced toxicity and delays the development of prion disease (80, 81).

None of these mutations prevented PrP amyloid fibril formation under near-native conditions in vitro when present as a substrate, and all had similar efficiencies serving as seeds for fibril formation of WT PrP (Fig. 7). This argues strongly against a mutational “species barrier” for seeding or a dominant negative effect on fibril formation. However, PrP fibrils formed under these conditions have smaller PK resistant cores than bona fide prions (82), most likely reflecting a fibril structure distinct from those of infectious prions.

Loss of prion propagation was less severe for several mutations in H3 than for those in H1, H2, and S1/S2 regions (Figs. 1, 3 and 4). This was surprising since a large number of pathogenic mutations are clustered in H3 (21). A possible explanation would be that pathogenic mutations affect primary nucleation rather than the elongation or fragmentation of fibrils during prion replication. The alanine replacements targeted all secondary structure elements within PrP (Figs. 1 and S2). Of these mutants, substitution of core hydrophobic residues also significantly affected prion propagation (Fig. 5A). Mutations to the hydrophobic core destabilize native PrP and accelerate misfolding (34, 46). Following the reasoning that lower native state stability correlates with altered folding intermediates and increased misfolding propensity (35), mutations to core residues should increase prion propagation. However, PRP2 and PRP12 which target specifically core residues, did not enhance prion replication (Fig. 4A). Similarly, oxidation of methionine residues perturbs the structural core of PrP (70). However, SCA results showed that substituting methionine with alanine at M204 or replacing M204.M205.V209.M212 with glutamine (unpublished data), resulted in considerably reduced prion propagation (Fig. 5A). The prion propagation efficiency decreased with subsequent splits and may be examples of mutant PrPs being incapable of sustaining infection (83). Similarly, destabilization of PrP through removal of the disulphide bridge (PRP18) markedly reduced prion propagation (Fig. 5A), suggesting that the stability of PrP is not a major factor in prion replication efficacy.

Conclusions

Taken together, the data presented indicate that structural mutations decrease prion propagation; however, our mutational analysis highlights key regions that have a strong influence on propagation efficiency as evidenced by mutations in residues 123 to 125, 188 to 192 or 134 to 153, either through interaction of PrP with the nascent fibril or by disrupting fibril integrity (Figs. 1, 3 and 4). When mapped onto PDB entry 2L39, sites 123 to 125 and 188 to 192 lie on the same face of the protein with region 134 to 153 on the opposite face (Fig. 6). It has been proposed that during prion conversion, there may be a conformational change in the S1-H1-S2 region away from the H2-H3 core of the PrP (84, 85). The S1-H2-S2 segment may be unfolded preferentially, whereas the H2-H3 segment remains marginally stable (86). A shift away from the H2-H3 core may bring H1 and its neighboring residues in proximity to the loop regions of PrP. Such a mechanism would complement the SCA data that highlight residues 123 to 125, the C terminus of H2 and H2-H3 loop, and H1, as the regions that most severely compromise prion propagation when mutated. These faces of the protein may represent important sites as scaffolds for, or have direct involvement in, the interaction of PrP^C with prions resulting in successful propagation. Alternatively, residues 123 to 125 lie in a core region of recently determined prion structures (8, 9, 10, 11) at the intersection between the two main lobes of the prion, where the introduction of bulkier side chain amino acids will likely destabilize the prion structure. The amino acid changes for residues 134 to 153 also lie in the N-terminal lobe of the prion and may therefore be involved in a direct interaction of PrP^C with prions or disrupt prion stability, like those in the hinge region of the fibril.

Experimental procedures

Construction of plasmid DNAs

pBluescript SK + plasmid vector containing the full-length ORF of mouse PrP was used for mutagenesis. Mutations to alanine (codon GCC) were made using the Stratagene QuikChange site-directed mutagenesis kit (Agilent Technologies). Mutants were sequence-verified and the ORFs inserted into the retroviral vector pLNCX2 (Takara Clontech).

Retroviral expression of alanine-mutants

A total of 10 μg of pLNCX2 DNA containing the mutated ORFs were used to transfect phoenix ecotropic packaging cells (ATCC, LGC Standards) with FuGENE 6 transfection reagent (Promega) according to the manufacturer’s instructions. Supernatants were collected 24 h posttransfection, passed through a 0.45 μm filter and applied to PK1-KD cells. Cells stably transduced with the pLNCX2 constructs were selected using 200 μg per ml G418 (Life Technologies, Thermo Fisher Scientific). Suppression of endogenous PrP expression was achieved through stable expression of pRetroSuper retroviral vector (87) containing a shRNA sequence targeting the 3′UTR of the endogenous PrP (53, 54); stable silencing was ensured by maintaining the cells under puromycin selection. Stably transduced PK1-KD cells were selected using 200 μg/ml G418 and 4 μg/ml puromycin. Drug resistant colonies were pooled to obtain bulk cultures.

Scrapie cell assay

SCA was carried out as previously described by Kloehn et al. (17). Briefly, cells were plated at 18,000 cells/well of a 96-well plate and RML prion infected mouse brain homogenate (a 10^-4 dilution of 10% RML prion infected mouse brain homogenate in PBS) added the following day. Cells were grown for three weeks with biweekly 1:8 or 1:6 splits. At splits 4, 5, 6, and 7, cell suspensions equivalent to 25,000 cells were filtered onto activated ELISPOT plates, lysed, digested with PK and probed with the mouse anti-PrP ICSM18 antibody (final concentration 0.5 μg/ml; D-Gen Ltd) and goat anti-mouse anti-IgG1-AP secondary antibody (1:8000 of 1 mg/ml; Southern Biotech) to identify the number of PrP^Sc-positive cells. Quantitation of cells positive for PrP^Sc was carried out using the BioSys Bioreader-7000-F imaging system (BIOSYS GmbH). Statistical significance of differences between sets PRP 1 to 7 and PRP 8 to 16 was assessed by nonpaired, one-tailed Student’s t test assuming unequal variances.

Immunofluorescence analysis

Sterile 22 mm glass coverslips were coated with 1 mg/ml Poly-L-Lysine for 30 min, washed with 1xPBS in the hood and dried, prior to seeding cells at 20,000 cells per coverslip. Cells were grown at 37 °C in 5% CO₂ and fixed using 4% w/v paraformaldehyde for 20 min at room temperature (RT). Cells were washed 3 times with 1× PBS, permeabilized with 0.05% Triton X-100 and then incubated with the mouse anti-PrP ICSM18 primary antibody (1:7000 of 1 mg/ml) for 1 h at RT, prior to 3 further washes with 1× PBS. AlexaFluor 488-conjugated goat anti-mouse antibody was subsequently added at 1:1000 for 1 h at RT. After three washes with 1× PBS, coverslips were mounted using ProLong Gold Antifade Reagent (Life Technologies, Thermo Fisher Scientific) containing DAPI for nuclear staining and left to dry. Coverslips were stored at 4 °C prior to imaging.

Image acquisition

Fluorescence images were acquired using a Zeiss LSM510 META laser scanning confocal microscope with a 40 × /1.30 oil DIC objective, equipped with ConfoCor 3 detection module. Scanning was carried out at room RT using λ_ex = 405 nm (nuclear stain: DAPI) or 488 nm (mouse anti-PrP antibodies ICSM18 or ICSM35 and goat-anti-mouse Alexa-488 conjugated secondary antibody).

Western blotting

Cell lysates were prepared from frozen cell pellets enriching for membrane associated proteins. Cells were resuspended in 10 mM phosphate buffer (P5244, Merck Sigma-Aldrich), incubated on ice for 5 min followed by centrifugation at 15,000g for 15 min. The pellet was resuspended in D-PBS, digested with benzonase (1-2 μl, 25KU equivalent to > 250 units/μl) for 15 min at RT and boiled for 10 min after addition of an equal volume of 2× sample buffer (125 mM TrisHCl pH6.8, 20% v/v glycerol, 4% w/v SDS, 4% v/v 2-mercaptoethanol, 8 mM 4-(2-aminoethyl)-benzene sulfonyl fluoride and 0.02% w/v bromophenol blue). After centrifugation at 15,000g for 1 min, the supernatant was removed and protein concentration determined by the Bradford Assay.

15-25 μg of protein was fractionated on 12% BisTris NUPAGE gel (NP0341, Thermo Fisher Scientific), transferred to PVDF and probed using anti-PrP ICSM35 antibody (0.2 μg/ml, D-Gen LTD, London, UK). Antigen-antibody complexes were identified using goat anti-mouse AP (1:10,000 dilution of A2179, Sigma-Aldrich) and CDP-Star Substrate (T2146, Thermo Fisher Scientific).

Protein expression and purification

The ORF of mouse PrP gene (Prnp) (accession number M13685, residues 23–231) with mutations at selected amino acids (PRP1: G123A,L124A, G125A; PRP4: S134A,R135A,M153A; PRP5: H139A,G141A,D146A; PRP13: V188A,T191A,T192A), was synthesized de novo by Twist Bioscience, with a thrombin-cleavable His-Tag added to the PrP N terminus. The ligated pTrcHisB/Prnp constructs were used to transform the Escherichia coli host strain BL21(DE3) (Novagen), genotype F′ompT hsdSB (rB- mB-) gal dcm (DE3), which was then plated onto Luria-Bertini agar plates containing 100 μg/ml carbenicillin. Cultures were grown for purification using a modification of previously described protocols (88). Briefly, following harvesting, cells were sonicated and the inclusion bodies containing PrP resolubilized in 6M GdnHCl, 50 mM Tris-HCl, 0.8% β-mercaptoethanol, pH 8.0. These were loaded onto a Ni-NTA column equilibrated in 6M GdnHCl, 10 mM Tris-HCl, 100 mM Na₂HPO₄, 10 mM glutathione pH 8.0, and eluted from the column using 10 mM Tris-HCl, 100 mM Na₂HPO₄, 2M imidazole pH 5.8. Residual GdnHCl was removed through dialysis against 20 mM Bis-TrisHCl pH 6.5, CaCl₂ added to a final concentration of 2.5 mM, and the N-terminal His-tag cleaved by thrombin for 16 h at RT (0.1U thrombin [Novogene]/1 mg of PrP added). The cleaved protein was loaded onto a second Ni-NTA column equilibrated with 20 mM Bis-Tris pH 6.5, 25 mM imidazole pH 6.5, and the eluted PrP peak was collected and dialyzed against 10 mM Bis-Tris pH 6.5, and aliquoted and stored at −80 °C. Protein concentrations were determined by UV absorption at 280 nm using a calculated molar extinction of 62,268 M^–1 cm^–1 for mouse PrP (https://web.expasy.org/protparam).

Native aggregation assay

As an initial step, a spontaneous aggregation assay was performed under native conditions to generate seeds. Synthesized PrP monomer substrate was filtered through a 50 kD membrane filter (Amicon Ultra-0.5) to remove large aggregates. The filtered monomer solution was then diluted in a buffer solution to achieve a desired concentration of 10 μM. Buffer solutions were freshly made on the day of the experiment by combining each of the following components to attain a final concentration of 50 mM sodium phosphate (pH: 6.8), 150 mM sodium chloride, and 5 mM of Bis-Tris (pH 6.5). Thioflavin T was also included in the working solution at 20 μM to monitor amyloid aggregation kinetics. After a quick pulse on the vortex, the reaction mixture was aliquoted in 100 μl volumes across wells of a low-binding 96-well plate (Corning 3651), each of which had been preallocated 3 silanized zirconium beads. The plate was then sealed and placed in a BMG FLUOstar OMEGA plate reader to shake at 700 rpm for 100 s followed by 20 s of rest at 42 °C. Fluorescence emission was recorded at 10 min intervals with excitation and emission wavelengths of 440 nm and 485 nm, respectively.

Products from this nonseeded aggregation assay were then diluted 1:100 in buffer solution (see previous description of spontaneous aggregation assay for composition) and sonicated in a water bath (Grant XUBA1) for 15 min to generate seeds. A seeded aggregation assay was then conducted under the same conditions as above with the addition of seeds at a final concentration of 0.1% (w/w-in relation to PrP monomer concentration). Aggregation kinetics were analyzed in GraphPad Prism by a sigmoid fit to determine times to half-maximal fluorescence (t₅₀) values as described in (82).

Circular dichroism

CD spectra were taken with a Jasco J-715 spectropolarimeter. Molecular ellipticity of natively folded PrP mutants was determined at a wavelength of 222 nm at 25 °C. Thermodynamic stability of each mutant was determined from GdnHCl denaturant-induced unfolding of the protein via a concentration series. Each data point represents the average of three denaturation curves. A 2 mm path length cuvette was used to capture spectra of 10 μM of protein in a sodium phosphate (15 mM, pH: 6.5) buffer solution.

Calculation of denaturant activity

Denaturant activity was used instead of molar concentration as a more accurate unit of measurement to extrapolate effects of GdnHCl on the folding energy of PrP. It is derived by accounting for the nonlinear relationship between denaturant and free folding energy with the following equation (89).

$$D=\left[GdnHCl\right]\left(\frac{7.5M}{7.5M+\left[GdnHCl\right]}\right)$$ (1)

Calculation of equilibrium constants and proportion folded of natively folded PrP

Using data collected from CD data of molar ellipticity during denaturant-induced unfolding of PrP, equilibrium constants were established by imposing a two-state, single transition model of folding based on the following equation where K_W is the equilibrium constant of the folded state and K is the equilibrium constant at the corresponding measure of denaturant activity. Meanwhile, m correlates to the sensitivity of the protein to unfolding based on the degree of exposure to denaturant:

$$K=K_W\ast \mathit{exp}\left(m\ast D\right)$$ (2)

Subsequently, proportion folded (α_F) was calculated at each magnitude of denaturant sampled and plotted as a visual representation based on the equation below:

$${\mathrm{\alpha }}_F=\frac{K}{\left(1+K\right)}$$ (3)

Equilibrium constants generated were used to calculate the Gibbs free energy change of unfolding for PrP^C.

Data availability

All data generated or analyzed in this study are included in the article and the supporting information. Additional information, including original blots, raw data, and reagents, are available from the corresponding authors upon reasonable request.

Supporting information

This article contains supporting information.

Conflict of interest

J. C. is a Director and shareholder of D-Gen Limited, an academic spin-out company working in the field of prion disease diagnosis, decontamination, and therapeutics. D-Gen Ltd supplies PK1 cells, PK1-KD cells, and the ICSM18 and ICSM35 antibodies used in this study. The other authors declare no conflicts of interest.

Reviewed by members of the JBC Editorial Board. Edited by Karen Fleming

Supporting information

Fig. S1.

Assessment of the stability of the natively folded PrP mutants assessed via chemical denaturation series. Molar ellipticities of natively folded PrP mutants at a wavelength of 222nm was measured in a Jasco J-715 CD spectrometer as unfolding of the protein was induced by guanidine hydrochloride (GdnHCl) addition. The folded fraction of the protein was calculated at each denaturant activity by fitting a single transition, two-state model of folding to the CD data. The proportion of the folded protein are plotted as a function of GdnHCl activity.

Fig. S2.

Visualization of mutations PRP 1-18. Mutation sites are color-coded in A, PrP^C structure (PDB: 2L39). B, PrP amino acid sequence and C, RML prion structure (PDB: 7QIG).