Protein folding is vital to living organisms because it adds functional flesh to the bare bones of genes. But errors in this process generate misfolded structures that can be lethal.
The functions of most proteins depend upon the precise three-dimensional structures of their mature, folded forms. Take, for example, the most toxic substances known — proteins such as ricin, found in the seeds of the castor oil plant. The entry of a single molecule of ricin into an animal cell is in principle enough to kill that cell, because ricin is an enzyme that inactivates ribosomes, the cell's protein-synthesizing machines, one after another. The enzymatic activities of such toxins depend on their mature three-dimensional forms, but two papers in this issue1,2 suggest that some proteins pose a danger that is independent of either their mature structure or their activity.
On page 507, Bucciantini et al.1 show that when two normally harmless proteins are each allowed to aggregate into fibrils in vitro, structures that form early in the aggregation process are highly toxic to cells, whereas the fibrils themselves are non-toxic. In similar vein, Walsh et al.2 report on page 535 that early protein aggregates secreted from cultured cells expressing a mutant gene from a patient with Alzheimer's disease impair neuronal function in rat brains. The implication is that intermediates in protein folding can be dangerous in themselves, quite apart from any toxicity shown by mature folded proteins.
Most normal proteins are compact structures whose surface features determine their functions by presenting binding sites that are specific for other molecules. Such structures are produced by the process of protein folding, during which extended polypeptide chains, synthesized by ribosomes, collapse into more globular states in a manner determined by the amino-acid sequence of each chain. This collapse occurs because amino acids bearing hydrophobic side groups tend to huddle together inside the folded structure and thus avoid the aqueous environment. Correctly folded proteins are usually either soluble or associated with cell membranes. But the process of protein folding is prone to errors that generate aggregates of varying size, some of which are large enough to be insoluble. Such aggregates do not have the functions seen in correctly folded proteins, and are associated with some of the most distressing human diseases.
Aggregates arise because some intermediate structures that are formed during the folding process briefly expose hydrophobic regions on their surfaces, and these regions may bind to similar surfaces in nearby folding molecules instead of becoming buried inside the final structure (Fig. 1). The problem is made much worse by the high concentrations of macromolecules, including proteins, that are present in cells3, but is normally kept under control by proteins called molecular chaperones, which shield the exposed hydrophobic surfaces from one another4. Despite this control, several human neurodegenerative disorders such as Alzheimer's disease, and 'prion' diseases such as Creutzfeldt–Jakob disease, are associated with the occurrence of protein aggregates called amyloid fibrils, or plaques, as are protein-deposition diseases affecting other parts of the body such as the heart and liver. How these aggregates form, and how they cause disease, are topics of intense study5.
An intriguing property of amyloid fibrils that are associated with human diseases is that they can be formed from at least 20 unrelated proteins, but nevertheless show common structural features, such as a central core of 'β-sheets' (a common type of structural element) and the ability to bind dyes such as Congo red and thioflavin T. This commonality spawned the idea that aggregation into amyloid fibrils is not specific to certain amino-acid sequences, but is a generic feature of all polypeptide chains under appropriate conditions, because β-sheet formation involves interactions between the atoms of main-chain amino acids that can occur in all proteins6. In support of this idea, proteins not known to be involved in disease, such as myoglobin7, can be induced to form amyloid fibrils in vitro when they are incubated under conditions that partly unfold their normal structure.
Bucciantini et al.1 now find that when a structural region, or domain, from the bacterial regulatory protein HypF is incubated at pH 5.5 in the presence of the unfolding agent trifluoroethanol, aggregates form within a few minutes. The aggregates exhibit β-sheet structures and bind thioflavin T, but look amorphous when examined by electron microscopy, with no sign of fibrils. After 48 hours, short 'protofibrils' with disordered ends are seen, and after 20 days these are replaced by long, unbranched structures characteristic of mature amyloid fibrils.
Toxicity tests using cultured rat and mouse cells show that preparations of the amorphous aggregates or protofibrils kill these cells, as judged by the loss of their ability to exclude the dye trypan blue, but preparations containing only mature fibrils do not1. Similar experiments with a domain of a non-toxic bovine kinase enzyme also indicate that early in vitro aggregates kill the cells, unlike mature fibrils formed from the same part of the protein.
The second new paper2 looks at the role of protein aggregates in Alzheimer's disease. Here, two types of aggregate are associated with brain damage — fibrillar tangles of the tau protein inside neurons, and amyloid fibrils of another protein outside. The amyloid fibrils contain many copies of the 40–42-amino-acid amyloid-β (Aβ) peptide. This is a fragment of a much larger protein of unknown function that is attached to the plasma membrane. The amyloid fibrils are toxic to cultured neurons, but whether they are the main cause of neuronal damage in Alzheimer's patients is unclear. The density of amyloid plaques in the brains of these patients is only weakly correlated with the severity of dementia. Moreover, previous work in several laboratories has shown that soluble forms of the Aβ peptide damage neuronal junctions (synapses)8.
Walsh et al.2 now report that soluble dimers and trimers of the Aβ peptide, secreted from cultured cells, impair synaptic function when injected into rat brains, but that Aβ monomers, protofibrils and fibrils do not. The small Aβ aggregates specifically impair the ability of synapses to respond more strongly after receiving a rapid succession of stimulatory electric shocks. This property is termed long-term potentiation and is thought to be a key element in memory and learning. Encouragingly, drugs that inhibit the enzymes that produce the Aβ peptide reduce the formation of these dimers and trimers in cultured cells, hinting at potential treatments for Alzheimer's disease9.
Together, the results of Bucciantini et al.1 and Walsh et al.2 suggest that damage to cells can be caused by misfolded intermediates generated during the production of amyloid fibrils, whether or not the fibrils — or the normal proteins from which they are derived — are also toxic. The toxicity of these early aggregates depends upon some as-yet- undefined structural features, and not upon their amino-acid sequence. It is possible that some diseases not associated with the accumulation of amyloid fibrils may be caused by the sporadic production of this type of aggregate by mistakes that occur during the folding of other proteins.
These ideas highlight the importance of understanding the cellular mechanisms that combat these potentially lethal mistakes, and identifying the precise cellular targets of aggregates that escape such surveillance. The high variability in the age of disease onset suggests that aggregate production is triggered by the random, chance escape of a folding polypeptide chain from the chaperone control machinery. Moreover, in the case of prion diseases, some misfolded chains are 'infectious' — they convert normal chains into the same misfolded structures.
One way forward may lie in the report10 that certain strains of Escherichia coli produce amyloid fibrils, using them to adhere to surfaces. This discovery suggests another approach to studying the mechanism of aggregate production — we can take advantage of how easy it is to genetically and biochemically manipulate this bacterium.
Bucciantini, M. et al. Nature 416, 507–511 (2002).
Walsh, D. M. et al. Nature 416, 535–539 (2002).
Minton, A. P. Curr. Opin. Struct. Biol. 10, 34–39 (2000).
Ellis, R. J. Trends Biochem. Sci. 26, 597–604 (2001).
Dobson, C. M., Ellis, R. J. & Fersht, A. R. (eds) Phil. Trans. R. Soc. B 356, 127–227 (2001).
Dobson, C. M. Trends Biochem. Sci. 24, 329–332 (1999).
Fandrich, M., Fletcher, M. A. & Dobson, C. M. Nature 410, 165–166 (2001).
Kim, J. H. et al. J. Neurosci. 21, 1327–1333 (2001).
Selkoe, D. J. Physiol. Rev. 81, 742–761 (2001).
Chapman, M. R. et al. Science 295, 851–855 (2002).
About this article
The cytoplasmic tail of human mannosidase Man1b1 contributes to catalysis-independent quality control of misfolded alpha1-antitrypsin
Proceedings of the National Academy of Sciences (2020)
International Journal of Biological Macromolecules (2019)
A method of predicting the in vitro fibril formation propensity of Aβ40 mutants based on their inclusion body levels in E. coli
Scientific Reports (2019)
Exploring the intermolecular interactions and contrasting binding of flufenamic acid with hemoglobin and lysozyme: A biophysical and docking insight
International Journal of Biological Macromolecules (2018)
Analytical Chemistry (2018)