|Home | About | Journals | Submit | Contact Us | Français|
Amyloid fibrils are highly ordered crystal-like structures. It is generally assumed that individual amyloid fibrils consist of conformationally uniform cross-β-sheet structures that enable the amyloids to replicate their individual conformations via a template-dependent mechanism. Recent studies revealed that amyloids are capable of accommodating a global conformational switch from one amyloid strain to another within individual fibrils. The current review highlights the high adaptation potential of amyloid structures and discusses the implication of these findings for several emerging issues including prion strain adaptation (i.e. gradual change in strain structure). It also proposes that the catalytic activity of an amyloid structure should be separated from its templating effect, and raises the question of strain classification according to their promiscuous or species-specific nature.
Amyloid fibrils are highly-ordered, densely packed and thermodynamically stable protein structures [1;2] that are believed to be a common state of a polypeptide backbone . Despite the large diversity in fibril morphology , the key structural components in amyloid fibrils and amyloid-like structures have similar architecture in which the individual β-strands are assembled perpendicular to the fibrillar axis and separated by a distance of 4.8 Å, forming the so-called cross-β core . Considering that the fibrils are highly ordered crystal-like structures, the conformation of a cross-β core is believed to be structurally uniform within individual fibrils. Although conformationally different amyloids can be formed within the same amino acid sequence [6;7], structural uniformity of individual fibrils enables faithful copying of the conformation from the parent molecules in a fibril to newly recruited daughter molecules. The principle of template-dependent amplification is believed to underlie the self-replicating mechanism of prion propagation and protein-based memory or inheritance [8-10] (Fig. 1A). Recent studies, however, revealed that amyloid structures are capable of substantial conformational changes that can be observed even within individual fibrils . These unexpected findings highlight the high adaptation potential of seemingly rigid amyloid structures and have numerous implications for our understanding of structural and functional properties of amyloids and prions.
The self-replicating properties of amyloid fibrils are attributed to the unique assembly of the amyloid cross-β-core. Because of the perpendicular arrangement of β-strands relative to the fibrillar axis, they provide a template for recruiting and converting a monomeric precursor at the growing edge. The self-replicating process consists of two activities: (i) catalytic (which is defined as the ability of amyloid state to recruit and facilitate conversion of a monomeric precursor into an amyloid state), and (ii) templating (i.e. the ability to accurately imprint the strain-specific conformation onto a newly recruited polypeptide). A balance between template-dependent elongation and fibril fragmentation provides an effective mechanism for multiplication of seeds resulting in self-replication of the amyloid state at the expense of the monomeric state of a protein  (Fig. 1A).
The templating and catalytic activities of amyloid are believed to be intimately coupled. Faithful templating is based on self-complementation of a polypeptide chain involved in amyloid assembly . Self-complementation can be achieved trough several mechanisms including tight complementarity of amino acid side chains in the sterric zippers of the cross-β spine; the stacking of side chains in so-called “polar zippers”, where the side chain hydrogen bonds are formed between β-strands along the fibrillar axis; or domain swapping .
Although multiple amyloid structures could be produced within the same amino acid sequence [6;7], only structures with high fidelity of replication maintain their strain-specific properties through multiple rounds of replication. It is yet to be determined what classes of amyloids display high fidelity of replication and whether specific structural motifs account for the high fidelity.
The recent studies that employed single-fibril microscopy imaging revealed that the elongation of fibrils does not always support uniformity in the cross-β structure within individual fibrils [7;11;14]. In these studies, two amyloid strains referred to as S- and R-fibrils were produced in vitro from highly pure hamster (Ha) recombinant prion protein (rPrP) . In order to determine whether individual strain-specific conformations were maintained in the cross-seeding reactions, Ha S- and R-strains were used for cross-seeding the fibrillation of the closely-related mouse (Mo) rPrP [11;14]. It is important to emphasize that the original Ha S- and R-fibrils showed fundamentally different global folds, including differences in the structures of their cross-β-cores [7;15]. Ha R-fibrils were able to elongate using Mo rPrP as a substrate while replicating the R-conformation faithfully despite utilizing a heterologous rPrP variant  (Fig. 1B). Surprisingly, Ha S-fibrils also recruited Mo rPrP; however, the fibril elongation proceeded through switching to a new conformational state that resembled the R-conformation  (Fig. 1B). Despite the lack of compatibility between Mo PrP amino acid sequence and S-structure, the Ha S-template was as efficient in recruiting Mo rPrP molecules as the Ha R-template that was fully compatible with the Mo rPrP amino acid sequence (the term “compatibility” is defined here as the ability of a PrP variant to acquire certain strain-specific conformation). Unexpectedly, the cross-seeding reaction produced hybrid fibrils that were formed by two segments each of which was composed by one of two rPrP variants (Ha or Mo), and each of which adopted a unique, clearly distinctive conformation within its own segment. A separate set of experiments revealed that Mo rPrP was not able to adopt the S-structure that appeared to be Ha-specific . Therefore, the switch from S- to R-conformation within hybrid fibrils occurred due to the inability of the recruited Mo rPrP precursor to acquire the S-conformation of the fibrillar template.
The catalytic and templating activities, the two key features of the amyloid and prion replication, were believed to be intimately coupled owing to the self-replicating nature of the cross-β structure . The experiment on cross-seeding of Mo rPrP using two Ha amyloid strains revealed that amyloid structures can display catalytic activity in the absence of a templating effect .
The ability to seed fibrillation reactions without transferring seed-specific structure in not limited to the replication of amyloids formed from recombinant polypeptides. In two recently developed assays, miniscule amounts of PrPSc were found capable of seeding the fibrillation of rPrPs [16;17]. As judged from the PK-resistance profile, however, the products of PrPSc-seeded conversion did not mimic the structure of infectious PrPSc but were similar to the rPrP fibrils produced in the non-seeded reactions or in the reactions seeded with rPrP amyloids . Surprisingly, regardless of whether PrPSc or rPrP amyloids were used as seeds, all conversion products formed from rPrPs showed the major PK-resistant fragments in the range between 10 and 13 kDa. Therefore, when rPrP was supplied as a substrate, PrPSc maintained its seeding activity, but failed to transfer its PrPSc-specific infectious conformation to rPrP. To date, formation of infections prions from rPrP in PrPSc-seeded reactions including the Protein Misfolding Cyclic Amplification reaction has not been demonstrated and remains one of the most puzzling problems in the field. Why does PrPSc lack the templating effect when used with rPrP substrate or why does rPrP fail to acquire an infectious conformation in PrPSc-driven conversion?
In a search for possible answers we need to discuss the PrP energy landscape. First of all, the structure of rPrP amyloid fibrils is fundamentally different from that of PrPSc . Second, rPrP fibrils are much more thermodynamically stable than any known PrPSc strains [20-22]. Third, the structural variations between different PrPSc strains are relatively minor [23;24] when compared to the differences in structures of PrPSc and in vitro formed rPrP amyloids . Therefore, the amyloid state appears to correspond to a global energetic minimum in the PrP energy landscape, whereas the PrPSc states occupy local energy minima (Fig. 2). The reverse correlation between the conformational stability of PrPSc strains and the incubation time to the disease observed in recent studies  highlights the idea that PrPSc structure evolved to propagate efficiently within the complex environment of a cell and cause disease within an animal's life time . From a thermodynamic perspective of the prion conversion reactions, PrPSc seeds decrease the kinetic barriers for the conversion of both substrates, PrPC and rPrP (Fig. 2). In the reactions involving PrPC, however, the structure of the end-products is determined by the structure of PrPSc seeds, whereas in the reactions involving rPrP the final products are formed according to global minima in the energy landscape.
In contrast to PrPC, rPrP lacks posttranslational glycosylation and a GPI anchor, none of which however were shown to be essential for acquiring the infectious conformation. Infection of transgenic mice expressing GPI-deficient PrPC was found to produce GPI-anchorless PrPSc deposits that were fully infectious . In other studies, transgenic mice expressing only unglycosylated PrPC were shown to be susceptible to prion infection producing unglycosylated PrPSc that could be further transmitted to wild type mice . The failure of rPrP to acquire an infectious conformation could be due to its inadequate conformation of peptidyl-prolyl bonds. The peptidyl-prolyl bonds in proteins can adopt either the cis or trans conformation with the trans conformation being predominant. The cis/trans isomerization in mammalian cells was found to be under strict regulation by the peptidyl-prolyl isomerases (PPIases) . Full-length PrP has 16 proline residues, where the residues P101 and P105 seem to have the most dramatic impact on the ability of PrP to aggregate into the PrPSc-like form . It is possible that rPrP is substantially different from PrPC with respect to the conformation of the proline backbone and/or distribution of cis versus trans prolines along the PrP polypeptide chain. Such differences could be due to (i) the high level of rPrP expression in E. coli, (ii) the seclusion of rPrP in inclusion bodies or (iii) an inadequate level of PPIases in E. coli. Recent studies on β-microglobulin highlighted the importance of the proline cis/trans isomerization for in vitro conversion into amyloid fibrils . Regardless of the specific reasons why rPrP fails to acquire the infectious conformation in PrPSc-seeded reactions, the aforementioned studies revealed the ability of PrPSc to convert a rPrP precursor to an amyloid PK-resistant state without transferring PrPSc-specific conformations to the newly-converted molecules. The observations of “seeding without templating” help to develop a mechanistic explanation for a number of phenomena including prion strain conversion and adaptation as will be discussed in the next sections.
The question of why some prion strains cross the species barrier efficiently while others do not is puzzling and calls for a search for the structural determinants of strain “promiscuity” (promiscuous strains are defined as those that can propagate using PrP variants from more than one species while maintaining their individual strain-specific properties, whereas species-specific strains propagate only within a single species without change in their strain-specific features). The key difference between the promiscuous and species-specific strains appears to be in their ability to accommodate mismatches in PrP amino acid sequence without conformational changes in their individual structures. Considering that PrPSc structure might not become available in the near future, the studies that employ PrP amyloid strains generated in vitro are very timely in elucidating the molecular rules that underlie promiscuity or species-specificity of the self-replicating protein states. The R-amyloid fibrils were shown to be compatible with at least two PrP variants, Mo and Ha  and, therefore, can be used as a model for defining the molecular structure of promiscuous amyloid strains. The S-strain, on the other hand, replicated itself only within the Ha PrP amino acid sequence, and as such, it can be a model of species-specific amyloid structure.
Although the structure of in vitro-generated PrP amyloids might be substantially different from that of PrPSc , the in vitro cross-seeding experiments showed surprising parallels with the data on interspecies prion transmission of bovine prion strains, BSE and BASE (bovine spongiform encephalopathy and bovine amyloidotic spongiform encephalopathy, respectively) . Upon serial passages in wild type mice, BSE maintained its individual strain-specific characteristics showing a limited, if any, species barrier, whereas the BASE strain displayed a substantial species barrier and eventually converted into the BSE strain as judged from biochemical and neuropathological features . In both experiments on cross-seeding of amyloids in vitro or cross-species prion transmission in vivo, the replication of species-specific PrPSc or amyloid strain using a heterologous PrP variant resulted in a switching into a more promiscuous self-replicating structure. To address the phenomenon of strain switching, a “strain conversion” model was introduced [8;33]. This model, however, does not answer the questions of how does strain switching occur and whether one can predict a direction for strain switching or adaptation.
The question of how cross-β structure changes without interrupting fibril continuity is not trivial. Here, we postulate that within a hybrid fibril, two segments have different global folds while sharing common local motifs that account for the integrity of the hybrid structure (Fig. 3). To satisfy this requirement, the same polypeptide region should adopt the identical parallel β-strand conformation within two fundamentally different folding structures. Because the region that acquires the common β-strand conformation is connected by hydrogen bonds to the same region in the polypeptide molecules above and below, the parallel β-sheet propagates along the whole length of the fibril despite being part of two different global folds. Hydrogen bonds running up and down the common β-sheet provide conformational stability for the whole hybrid structure. This model proposes that the catalytic activity in recruiting and converting monomeric precursor into an alternative amyloid state is due to partial overlap in amyloid structures.
The molecular determinants for forming a stable hybrid structure have to be elucidated in future studies. It would be important to determine next whether the formation of hybrid structures requires a high level of homology within the polypeptide regions that share a common β-strand conformation. Alternatively, acquiring identical β-strand configuration by two polypeptide regions regardless of the level of their homology could be sufficient for forming hybrid fibrils. Future studies should illustrate whether the model of hybrid fibrils is valid or whether alternative models could explain the phenomenon of a conformational switch.
The observation of conformational switches within individual fibrils highlights the high adaptation potential for amyloid structures. Adaptive conformational switching allows recruitment of non-identical but homologous polypeptide molecules which otherwise are not compatible with the existing structure.
Adaptive conformational switching within individual fibrils may provide a mechanistic explanation for modification, adaptation or emergence of new prion strains that are typically observed upon cross-species prion transmission [33-35]. The studies on interspecies transmission using PMCA reactions or bioassays in animals revealed that prion strains have variable adaptation potential. Some strains are capable of maintaining their individual strain-specific properties within a range of PrP amino acid sequences, whereas other strains can faithfully replicate only within a single or closely homologous PrP sequence [32-35]. The work on cross-seeding of rPrP amyloids provided direct illustration that self-replicating amyloid structures are not equally selective with respect to the amino acid sequence of the substrate molecules that can be recruited for their replication . This difference in selectivity between amyloid strains specifies a direction in which adaptation or evolution of amyloid structures or prion strains occurs upon interspecies transmission. The conformational adaptation is expected to proceed from highly selective or species-specific structures toward promiscuous ones. A switch from a species-specific to promiscuous strain presumably occurs when a species-specific strain faces a heterologous substrate that is not compatible with the conformation of the original strain. Therefore, one can predict that inter-species transmissions could lead to formation of new promiscuous strains.
The new concept on amyloid adaptation could be useful for developing mechanistic models for a number of puzzling observations including conformational switching in amyloids formed within the same sequence or cross-talk between distant or non-related amyloidogenic proteins. Using conformation-dependent luminescent dyes, recent studies reported that changes in conformational states of Aβ fibrils occurred within single amyloid plaques formed in a mouse brain . Surprisingly, the plaque centers were characterized by loosely packed or less-ordered structures that were found to evolve into more compact or tightly packed structures toward the plaque periphery. Whether this spatial centrifugal alteration in the conformation of Aβ amyloid represents different stages of plaque growth in vivo is not yet clear. This study, however, highlights that a continuum of conformational states could be formed within individual plaques.
There are growing examples of cross-seeding between distant or unrelated amyloidogenic proteins [37-41]. While the efficiency of cross-seeding between two unrelated amyloidogenic polypeptides was found to be quite low in vitro , the bioassays on animals revealed that inoculation of β-pleated fibrils formed by one protein was sufficient to cause or accelerate the in vivo amyloidosis of an unrelated protein [39;41]. It is of particular interest that the fibrils of nonpathogenic proteins including silk fibrils, curli fibrils from E. coli, or amyloids of yeast prion protein Sup 35 were effective in accelerating the amyloidosis of amyloid protein A upon injections into wild type mice . It remains to be determined whether hybrid fibrillar structures could be formed by two unrelated proteins and what level of homology is required for maintaining the integrity of hybrid structures.
The process of amyloid cross-seeding in yeast offers an example of a functional mechanism that evolved to regulate a transition of certain yeast prion proteins between prion and non-prion states. Presence of [PIN+], which is the self-propagating or prion state of the yeast prion protein Rnq1p, was shown to be essential for inducing the conversion of another yeast prion protein Sup35 into its prion state, referred to as [PSI+]. Considering that the prion folding domains of both proteins are rich in asparagine and glutamine residues, their degenerative amino acid sequences seem to evolve for providing a sufficient level of homology for the direct cross-seeding to be effective . Remarkably, [PIN+] seeds were also found to be active in seeding fibrillation of huntingtin protein that contains an abnormally long polyglutamine region . Consistent with the model of a hybrid fibrils proposed here, the mechanism of cross-talk between the two yeast prion proteins includes direct recruitment of Sup35 by Rnq1p fibrils .
The current article introduces a new concept of hybrid fibrils and conformational switching within individual fibrils that provides a mechanistic explanation for the previously observed phenomena of prion strain adaptation, modification or convergence, as well as the cross-interaction of yeast prions or unrelated amyloidogenic proteins. This model suggests that the catalytic activity of an amyloid structure should be separated from its templating effect. It raises the question of strain classification according to their promiscuous or species-specific nature, and helps to predict a direction in adaptation of amyloid strains.
Important questions to be addressed in future studies include: does fibrillar structure switch within individual fibrils made of the same protein; what is the probability of such spontaneous conformational switching, and what role does it play in strain structural diversity; and what is the extent to which the global amyloid structure must change to encode a new strain?
The dominant view of amyloids as rigid crystalline-like structures that can be formed only through assembly of identical polypeptide chains will likely be changed in the near future. We anticipate that amyloids will be viewed as dynamic structures that are capable of accommodating significant conformational switching and of being hybrid with respect to their composition. The new view should stimulate future studies on establishing causative links between cross-seeded amyloidoses of disease-related proteins, on the functional role of amyloids, and the search for practical applications of amyloids as biomaterials or smart nanodevices.
We thank Pamela Wright for editing the manuscript. I.V.B. is supported by National Institute of Health grant NS045585.
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.