|Home | About | Journals | Submit | Contact Us | Français|
Prion diseases, in which the conformational transition of the native prion protein (PrP) to a misfolded form causes aggregation and subsequent neurodegeneration, have fascinated the scientific community as this transmissible disease appears to be purely protein-based. Disease can arise due to genetic factors only. At least 30 single point mutations have been indicated to cause disease in humans. Somehow, these mutations must influence the stability, processing and/or cellular interactions of PrP, such that aggregation can occur and disease develops. In this review, the current evidence for such effects of single point mutations is discussed, indicating that PrP can be affected in many different ways, although questions remain about the mechanism by which mutations cause disease.
The prion protein (PrP) is a small protein mostly found attached to the outer membrane of neuronal cells. The biological function is not yet fully determined, but there are strong indications that it has a role in signal transduction (Linden et al., 2008), e.g. for olfactory signals (Le Pichon and Firestein, 2008), and the regulation of metal metabolism (Millhauser, 2007; Viles et al., 2008; Singh et al., 2009). PrP is best known for its potential to cause disease: similar to amyloid diseases, the protein can adopt a misfolded conformation (PrPSc), which in turn can lead to aggregation, neurodegeneration and the manifestation of transmissible spongiform encephalopathy. Prion diseases can be of an infectious, genetic or sporadic nature. In genetic or familial disease, single nucleotide mutations in the gene encoding PrP (PRPN) can increase the likelihood of aggregation and neurodegeneration (Kovacs et al., 2005; Mead, 2006). For humans, many mutations have been reported that are linked to prion diseases: familial Creutzfeldt–Jakob's disease (fCJD), Gerstmann–Sträussler–Scheinker disease (GSS) and fatal familial insomnia (FFI) (Fig. 1).
After the cleavage of the N- and C-terminal signal peptides, human PrP is a 208 residue protein (23–230) that is exported to the cell surface with up to two carbohydrate moieties (linked by N-glycosylation) and a glycosylphosphatidylinositol (GPI)-anchor. In many reported studies, however, recombinant PrP (recPrP, usually residues 90–231) is used, which lacks these post-translational modifications. The natively folded human protein (PrPC) has been solved (as a monomer) by NMR at acidic pH (Zahn et al., 2000) and neutral pH (Calzolai and Zahn, 2003) and by X-ray crystallography as an unusual domain-swapped dimmer (Knaus et al., 2001) or in complex with an antibody (Antonyuk et al., 2009). These studies have shown that PrPC consists of a ~100 residue flexible N-terminal domain, a globular domain of residues 125–228 and a short flexible C-terminal domain (229–230/231). The globular domain consists of three helices and a short antiparallel β-sheet (Fig. 2). A core of hydrophobic residues can be defined that will stabilize the tertiary contacts between these elements (Riek et al., 1998), together with several salt bridge and hydrogen bonding interactions.
The mechanism of formation and structure of PrPSc is yet unknown, but a significant shift toward more β-structure and less helical structure is evident (Caughey et al., 1991; Pan et al., 1993; Jackson et al., 1999). Numerous experiments have linked a low pH environment to PrP misfolding (reviewed by DeMarco and Daggett, 2005), and molecular dynamics (MD) simulations of PrP at low pH indicate that significant conformational changes take place that are consistent with experiment (Alonso et al., 2001, 2002; DeMarco and Daggett, 2004, 2007). The main changes in conformation are (i) the extension of the native sheet through addition of one or more strands and extension of the existing strands and (ii) changes in the position and conformation of helix HA and the loop between S1 and HA, with the latter becoming more extended (Fig. 3B). Conformers resulting from low pH simulations that show these changes have been used to build spiral protofibril models, intermediates en route to the fibrillar form (DeMarco and Daggett, 2004; Scouras and Daggett, 2008) (Fig. 3C), in line with a range of experimental data (DeMarco et al., 2006). Recent experimental evidence indicates that in the final PrPSc fibrils, refolding of helices HB and HC into a (parallel) β-sheet core may occur (Bocharova et al., 2006; Cobb et al., 2007, 2008; Lu et al., 2007), but this is at odds with numerous antibody-binding epitopes that are retained upon conversion (Horiuchi et al., 1999; Perrier et al., 2004; Matucci et al., 2005; Jones et al., 2009). It is further important to note that it is the intermediate protofibrillar structures, or soluble oligomers, that harbor the toxicity and infectivity (Wille et al., 1996; Hartley et al., 1999; Bucciantini et al., 2002; Caughey and Lansbury, 2003; Kayed et al., 2003), not the mature fibrils.
For familial prion disease, the question is how the identified pathogenic mutations in PrP affect the conversion of PrP to PrPSc and aggregation, resulting in neurodegeneration and eventually death. Pathogenic mutations are found throughout the PRPN sequence (Fig. 1) and display a variety of effects on the mature PrPC protein. There are several mutations that significantly destabilize the PrPC conformer in favor of a folding intermediate (Apetri et al., 2004) that is likely to be a PrPSc precursor (Apetri and Surewicz, 2002; Apetri et al., 2006). Some of these mutations also affect the overall thermodynamic stability of PrPC (Swietnicki et al., 1998; Liemann and Glockshuber, 1999). More subtle effects are also possible, such as modification of cellular trafficking (Harris, 2003). Furthermore, PrPSc fibrils with significantly different characteristics than those from wild type (WT) are formed in combination with certain mutations (Piccardo et al., 1998, 2001; Apetri et al., 2005; Tunnell et al., 2008).
Here, we review the current evidence (or lack thereof) of the effects of pathogenic single residue mutations on structure, dynamics and cellular interactions of PrP in humans. On the basis of their occurrence in the sequence and the type of amino acid replacements involved, the reported pathogenic mutations can be classified in different groups, which will be discussed separately below. Insertions or deletions in the octarepeat region of the PRPN gene have also been linked to prion disease in humans (Goldfarb et al., 1991; Beck et al., 2001), but they are not discussed here. For a number of mutations (G114V, S132I, A133V, T193I, E196K), no information is available apart from their occurrence in cases of inherited spongiform encephalopathies. Note, however, that virtually all the studies deal with the unglycosylated forms of the protein, and the effects on stability, structure and dynamics may be different in vivo with the glycosylated, membrane-bound form (see DeMarco and Daggett, 2009, for further discussion and references).
There are four residues linked to GSS causing mutations in the flexible N-terminus of human PrP: P102L, P105L/S/T, G114V and A117V. It is likely that the flexible N-terminus is involved in the conversion from PrPC to PrPSc (Alonso and Daggett, 2001; DeMarco and Daggett, 2007). Furthermore, residues 112–119 are required for the formation and propagation of PrPSc (Norstrom and Mastrianni, 2005). The importance of the flexible N-terminus in causing disease is supported by the fact that a short neurotoxic peptide (106–126) exists (Forloni et al., 1993; Brown et al., 1994), and ‘abridged’ mouse PrP (with residues 23–88 and 141–176/221 deleted) can form infectious, protease resistant and neurotoxic PrPSc aggregates (Supattapone et al., 1999, 2001).
Most studies on mutations in the flexible N-terminus have focused on small fragments of the PrP protein. Forloni et al. (1999) reported that a peptide representing residues 89–106 became neurotoxic when containing the P102L mutation, but not with P105L. Neurotoxicity of the 106–126 peptide was not significantly affected by the A117V mutation, although others did find this to be the case (Brown, 2000). Before the latter finding, MD simulations of a 109–122 peptide in a helical starting conformation showed that the WT peptide remained largely helical, whereas the A117V peptide rapidly lost helicity (Kazmirski et al., 1995). This was later confirmed by MD simulations of the 106–126 peptide (Levy et al., 2001).
Jones et al. (2006) compared the fibril formation of the 23–144 fragment (which is also relevant due to the Y145stop mutant, see below) with the mutations P102L, P105L and A117V to the WT. No significant effects on the amyloidogenicity were found, confirming the results obtained with small peptides (Forloni et al., 1999). The P102L mutation also has no or little effect on the thermodynamic stability of recPrP (Swietnicki et al., 1998) or the population of the proposed recPrP folding intermediate (Apetri et al., 2004). MD simulations of 4 and 8 ns at 320 K of the P102L mutant in combination with V129 and M129 of Syrian hamster PrP 90–231 revealed no significant differences with the WT structure (either the globular domain or the flexible N-terminus) (Santini and Derreumaux, 2004). Circular dichroism (CD) measurements on full-length mouse PrP with the equivalent mutation (P101L), however, show a decrease in α-helical content compared with WT (Cappai et al., 1999).
Overall, there is little evidence for a direct molecular effect of the mutations in the flexible N-terminus on the globular structure of PrPC or conversion to PrPSc. It is possible that β-sheet formation in the N-terminal region is enhanced, but it could also be that the disease-causing ability of these mutants arises due to more complex interactions at the cellular level. For the mutations P105L and A117V, an increased amount of PrP was found in an alternative membrane topology (Hegde et al., 1998; Kim and Hegde, 2002). Recently, Schiff et al. (2008) showed that the co-existence of WT PrP with several different mutant PrPs, including A117V, can alter their cellular location. Mutations in the N-terminus may also influence the structure and propagation of PrPSc, as found for P102L (Piccardo et al., 1998; Wadsworth et al., 2006) and P105S (Tunnell et al., 2008).
Two stop mutations have been observed in patients that result in a truncated version of PrP: Y145stop and Q160stop. Correspondingly, Watzlawik et al. (2006) studied aggregation in vitro of PrP fragments 23–144 and 23–159. This study revealed that the construct that includes helix HA (144–156) aggregates much faster than the shorter construct. Further research has mostly been performed on the Y145stop mutant. The largely missing C-terminal globular domain in this mutant is normally required to import the complete PrP into the ER (Heske et al., 2004). Without this happening, a large portion of the truncated PrP is not post-translationally processed (keeping its N-terminal signal peptide) and rapid degradation by the proteasome occurs (Zanusso et al., 1999; Drisaldi et al., 2003). It was found that a recombinant fragment containing residues 23–144 can convert spontaneously to a fibrillar form, with residues 138–141 being essential for its formation (Kundu et al., 2003). Fibrils of this 23–144 fragment were recently investigated using solid-state NMR (Helmus et al., 2008). The lack of signals detected at temperatures above 253 K for the majority of this fragment indicates that the fibrils are largely unordered and display significant conformational dynamics. Signals were only detected for residues 112–141 and likely β-strand segments were identified, including residues 130–139, which corresponds to a strand first identified by the MD simulation (Alonso et al., 2001) (Fig. 3B).
Salt bridges between residues in different parts of a protein structure are often involved in stabilization of its tertiary structure. Six known disease-causing mutations can directly affect (potential) salt-bridge interactions in the globular PrP domain: R148H, D178N, E196K, E200K, R208H and E211Q (Fig. 4). R148 can form salt bridges with D144 and E146, thereby forming the core of the rare pattern of salt bridges (also including D147–R151) along helix HA. This network of interactions was shown to confer stability of HA, both experimentally (Speare et al., 2003; Megy et al., 2004) and theoretically (Dima and Thirumalai, 2004; Santini and Derreumaux, 2004). But apart from the similarity of R148H PrPSc aggregates to those observed in a sporadic form of CJD (Pastore et al., 2005), no information on the effect of this mutation is available. Salt bridges R164–D178 and R156–E196 can be inferred from the NMR structures of human PrP (Zahn et al., 2000; Calzolai and Zahn, 2003), and they are highly populated in several independent MD studies (see e.g. Zuegg and Gready, 1999; Langella et al., 2006; Bamdad and Naderi-Manesh, 2007; DeMarco and Daggett, 2007). Both these interactions link separate structural elements (S2 to HB and HA to HB/HC, see Fig. 4). E200 and K204 are in close proximity in NMR structures, but due to their location (solvent exposed and directed away from other structural elements), a salt bridge between them is unlikely to be of great structural importance. This was confirmed by the NMR structure obtained for the E200K mutant, which was found to be nearly identical to WT PrP (Zhang et al., 2000). For R208, the situation is less clear; the NMR structures suggest the proximity of R208 to E211, whereas MD studies report the formation of a salt bridge with E146 (Zuegg and Gready, 1999; DeMarco and Daggett, 2007).
Zuegg and Gready (1999) argued that a salt bridge between E146 and R208 could stabilize the tertiary structure of PrPC and its loss may facilitate conversion to PrPSc. MD simulations of WT PrP (residues 125–228), R208H PrP and WT PrP with neutralized D144 and E146 (to break the putative salt-bridge interaction), however, showed no significant rearrangements upon breaking the interaction, whereas the introduction of the R208H mutation did (Bamdad and Naderi-Manesh, 2007). The authors attribute the latter to global changes in the backbone dynamics. These changes do not affect the misfolded form: PrPSc extracted from an fCJD patient with the R208H mutation was shown to have the same characteristics as WT PrPSc (Capellari et al., 2005). Sequestering into aggresomes in the cytosol was reported for E211Q PrP, in conjunction with inhibition of proteasomes (Mishra et al., 2003).
E200K is a major causative mutation related to fCJD (Mead, 2006) [and its phenotype is influenced by the M/V129 polymorphism (Puoti et al., 2000)]. It was argued that the mutation may affect the stability of helix HC (Gallo et al., 2005), but it does not seem to significantly affect the overall thermodynamic stability of PrPC (Swietnicki et al., 1998). It does, however, change the surface electrostatic potential of the protein (Zhang et al., 2000), which may cause abnormalities in interactions of PrP with other proteins in the cell or the cell membrane itself (Schiff et al., 2008). Glycosylation (particularly at N197) was also found to be affected (Capellari et al., 2000), and an abnormal ratio of mono-, di- and unglycosylated protein in PrPSc aggregates of E200K PrP was observed (Hill et al., 2006). A theoretical investigation of the energy landscape of E200K PrP (using high-temperature, implicit solvent MD simulations) indicated that the HA helix may be significantly less stable in the mutant compared with WT (Levy and Becker, 2002), but experiment indicates that the population of a folding intermediate is not affected (Apetri et al., 2004).
The D178N mutant is probably the most extensively studied PrP mutation and is involved in a large number of disease cases (Mead, 2006). Intriguingly, the mutation causes FFI in combination with M129 and fCJD in combination with V129 (Goldfarb et al., 1992), although this distinction may not be so clear cut (McLean et al., 1997; Brown, 2000). Several reports indicate that the D178N mutation greatly increases the aggregation propensity. When trying to produce recombinant huPrP (90–231) in E. coli, the D178N–M129 mutant aggregated into inclusion bodies (Swietnicki et al., 1998), as was the case for D178N–M129 and D178N–V129 recombinant mouse PrP (94% sequence identity) (Liemann and Glockshuber, 1999). It was further found that the recPrPSc structure of the D178N mutant is significantly different from WT PrPSc obtained under the same conditions (Chen et al., 2000; Apetri et al., 2005).
Apart from the salt bridge with R164 (located on the second native strand S2), D178 can also be involved in hydrogen bond interactions with the Y128 and Y169 side chains (located at the beginning of strand S1 and the loop between S2 and HB, respectively, Fig. 2). Protonation of D178 in MD studies leads to the loss of these interactions, which may facilitate conversion (Alonso et al., 2001). The D178N mutation would also seem to be able to affect the region around the native β-sheet S1/S2. This was confirmed in a recent study using spin-directed spin labeling combined with ESR, in which it was found that the D178N mutation increases the instability of S2 (Watanabe et al., 2008).
Many computational studies of D178N mutant PrP have been conducted because experiment indicates that this mutation significantly affects PrPC stability and may therefore influence the structure and dynamics of the protein. Initial 1.5 ns (explicit solvent) MD simulations of WT and D178N mouse PrP showed no differences in flexibility (Gsponer et al., 2001). Interestingly, the most stable WT simulation in their study only showed the R164–D178 salt bridge for 15% of the time. The authors concluded that the R164–D178 salt bridge may not be important for WT PrP stability, but these simulations seem too short to substantiate such a statement. Using high temperature, implicit solvent MD simulations, Levy and Becker (2002) found that HA was very stable in WT but not in D178N, which they attributed to changes in the charge distribution that affect internal salt bridges in HA. In more extensive explicit solvent MD simulations of human PrP 125–228 (in combination with M129 or V129), however, no significant differences were observed (Shamsir and Dalby, 2007). MD simulations of WT and D178N mouse PrP (124–226) were also performed, both in water and CCl4 (in order to reveal structural weaknesses in an unfavorable environment) (Barducci et al., 2005). Although the overall fold stayed largely the same in the simulations, it was found that with D178N (and in particular in CCl4), the S1/S2 β-sheet was unstable (due to weakened interactions with R164 and Y128) in line with the recent experimental findings by Watanabe et al. (2008). This was later argued to be a first step for conversion to PrPSc (Barducci et al., 2006).
Overall, it is clear that the D178N mutant reduces the barrier for PrPSc aggregate formation. Conflicting ideas exist on the mechanism by which this occurs, but it may well be related to the loss of an important salt-bridge interaction. For E200K, it is more likely that subtle interactions at the cellular level are involved, as PrPC conformation and stability are largely unperturbed. The importance of the disruption of the other salt-bridge interactions (i.e. in the R148H, E196K, R208H and E211Q mutations) for stability of PrPC and conversion to PrPSc is yet to be determined.
There are several known pathogenic mutations of polar residues in the PrP globular domain that do not involve the disruption of salt-bridge interactions, but they may be involved in other stabilizing interactions (Fig. 5). Three of the polar residue mutations are found in helix HB. Mutation of T183 to A can cause abnormal glycosylation (Grasbon-Frödl et al., 2004), and it affects both in vivo folding and GPI-anchor attachment (Kiachopoulos et al., 2005). Mouse T183A PrP 121–231 expressed in E. coli aggregates and shows a significantly decreased stability compared with WT (Liemann and Glockshuber, 1999). This instability may be related to a disrupted hydrogen bond interaction between the T183 side chain and the backbone nitrogen of Y162, which can help to anchor the native sheet to the core of HB and HC (Fig. 5A). The H187R mutation, which has been related to both fCJD (Gu and Singh, 2004) and GSS disease with abnormal PrPSc deposits (Butefisch et al., 2000; Colucci et al., 2006), may disrupt a putative interaction between helices HB and HA (Langella et al., 2006) (Fig. 5B). H187R PrP accumulates in lysosomes instead of moving to the cell surface, but doxycycline and protein folding agents can change this (Gu and Singh, 2004). Directly adjacent to H187 is T188, which can mutate to either R, K or A, causing fCJD. A study employing a mouse cell model system indicated that T188R and T188K mutant PrPs are glycosylated and inserted in the cell membrane similar to WT PrP (Lorenz et al., 2002). These mutant PrPs did, however, show an increased proteinase K resistance and detergent insolubility that can be indicative of a (higher) degree of aggregation.
Three other pathogenic mutations of polar residues are located in HC. The first is D202, which forms a (conserved) stabilizing ‘capping box’ with T199 at the top of the helix (Riek et al., 1998) (Fig. 5C). A combined CD and NMR study of peptides corresponding to helix HC showed that the D202N mutation completely destabilizes its structure (Gallo et al., 2005). In a human cell model, D202N PrP accumulates and aggregates in the ER without reaching a mature conformation (Gu et al., 2007). The mutation of Q212P is remarkable as it inserts a proline in the middle of helix HC, which would be expected to affect its structure. The only effect of this mutation currently known is that Q212P PrP accumulates and eventually forms aggresomes in the cytosol upon proteasomal inhibition (Mishra et al., 2003). The final mutation in HC is Q217R. Q217R PrP lacks a GPI-anchor, which probably causes the observed impaired transport to the cell surface, and it exhibits increased aggregation and proteinase resistance (Singh et al., 1997). As these effects are temperature-dependent, they could be due to the misfolding of PrP. Q217 is in close contact with S1 and S2, and MD simulations of sheep and human PrP have shown that it is involved in a ‘tight hydration site’: a water molecule can bind between the backbone carbonyl of S132 (in S1), the backbone amide of V161 (in S2) and the side chain oxygen of Q217 (De Simone et al., 2005) (Fig. 5D). This site also may be affected by the recently discovered S132I mutation (Hilton et al., 2009). When Q217R PrP was simulated, the large side chain of the arginine replaced the tightly bound water and forced other rearrangements, ultimately causing the S1/S2 β-sheet to extend and stabilize (De Simone et al., 2005). The authors argue that this may facilitate fibril formation.
The remaining pathogenic mutations in the globular domain of PrP affect hydrophobic residues. Santini et al. (2003) performed MD simulations of G131V recPrP (90–231) that indicated that this mutation may cause the extension of the S1/S2 β-sheet but does not affect the overall stability. Four other hydrophobic mutations are part of the hydrophobic core (Fig. 6). Similar to T183A (described above), three of the mutations have been indicated to cause atypical glycosylation [V180I (Nixon et al., 2000; Chasseigneaux et al., 2006), F198S (Zaidi et al., 2005)], interference with GPI-anchor attachment [F198S (Kiachopoulos et al., 2005)] and/or changes in PrP folding (or refolding after denaturation) [V180I and V210I (Apetri et al., 2004), F198S (Apetri et al., 2004; Kiachopoulos et al., 2005; Zaidi et al., 2005)]. Thompson et al. (2001) reported that in a peptide corresponding to a large part of helix HC, the V210I mutation increased both helical and aggregation propensity compared with the WT sequence. Apart from a decrease in the (cell-death preventing) anti-Bax function of PrP (also found for 12 other mutants studied) (Jodoin et al., 2007) and formation of cytosolic aggresomes when proteasomes are inhibited (Mishra et al., 2003), V203I has not yet been investigated.
The F198S mutation leaves a gap in the hydrophobic core (Riek et al., 1998) and as expected, the thermodynamic stability of PrPC is significantly affected, as determined by urea-induced transitions of mouse PrP 121–231 (Liemann and Glockshuber, 1999) and MD simulation (Zhang et al., 2006a, 2006b). A direct influence on the conversion from PrPC to PrPSc is also evident for this mutation: conversion from a predominantly α-helical conformation into an oligomeric β-sheet-like structure in the presence of guanidine HCl was found to be ~50 times faster than that of WT PrP (Vanik and Surewicz, 2002). Furthermore, this conversion also occurred in the absence of denaturants.
There are three mutations in the GPI-signal peptide associated with fCJD: M232R, M232T and P238S. The mechanism of cytotoxicity of these mutants remains unknown and is puzzling, as the signal peptide is cleaved off within 5 min of PrP synthesis and translocation into the ER (Linden et al., 2008). Recently, Gu et al. (2008) showed that M232R and M232T do not interfere with GPI-anchor addition, but do cause PrP being bound to the membrane in an alternative (C-transmembrane) orientation.
The many single point mutations linked to inherited prion disease can cause widely varying effects, from severe destabilization of the PrP, changes to its cellular trafficking and processing, to subtle changes in local structure or electrostatics. When charged or polar residues are involved, tertiary or secondary stabilizing hydrogen bonding interactions are often affected. Mutations in the hydrophobic core of PrP usually result in defective post-processing and/or folding. The effect of mutations in the flexible N-terminus of PrP is not yet clear, whereas mutations resulting in a truncated protein cause rapid aggregation. Overall, the current literature indicates that mutations in PrP display a multitude of effects, both on the molecular and the cellular level. Many questions, however, remain as to how the mutations cause such effects and how these effects in turn lead to neurodegeneration and disease.
We are grateful for support by the National Institutes of Health (GM 81407).
Molecular images were generated using PyMOL (DeLano, 2002).
Edited by Alan Fersht