Amyloids: the Adam (and Eve) of Protein Evolution?

Greenwald, J., Riek, R. (2012) On the Possible Amyloid Origin of Protein Folds. Journal of Molecular Biology 421 (4-5): 417-426

There are conflicting views on the possible origins of life on this planet and, in the absence of a fossil or genetic record reaching back several billion years, it is only possible to speculate on the probability of each.  

Three main theories are proposed.  The first is that out of the random peptide sequences emerging in a prebiotic world, that is a world before the appearance of organic life, a single protein fold originated which was capable of providing the key functions of life, with all other proteins evolving from this common ancestor.  The second theory is similar but involves the serendipitous creation of several ancestral protein folds which interacted to create a sustainable life system.  In the last proposal, life originated without peptides, as in the “RNA world” hypothesis which was highlighted by Clare in her Feb 2012 post to this blog, and then common ancestor folds evolved within this existing system, either from a single fold or as a set of interacting folds.

 

This paper examines the first proposal, that of a single common ancestral fold, which may be considered to be the most likely by virtue of being the simplest.  In examining the hypothesis, a list of properties requisite for any common ancestor fold has been developed by (Greenwald, J., Riek, R. (2012)) against which the various folds in contention have been tested.  The common ancestor fold must:

 

(i)          have a short and simple sequence, since early replication would have been relatively inaccurate,

(ii)        withstand sequence modifications, for the same reason,

(iii)      provide a function which promotes life, giving it a selective advantage,

(iv)       be composed of amino acids which have been shown to generate in abiotic conditions, since by definition biological synthesis had not developed,

(v)        and be amenable to evolution which could extend the functionality of the fold.

These stipulations have been applied across known protein folds to establish a list of candidates to be investigated for their theoretical fit.

The PDB provided several potential common ancestor folds but most could be discarded on the basis that they were either too long for successful replication, they required residues for stabilization that no-one has yet been able to generate in prebiotic lab conditions or they could not provide an obvious selective advantage as isolated peptides.

The most persuasive argument for a common ancestor fold is brought by the peptide amyloids.  An amyloid is a β strand, potentially as short as four residues, which oligomerizes into parallel or anti parallel β sheets which stack on top of each other with indefinite numbers of repeats to form amyloid fibrils.

Image adapted from (Greenwald, J., Riek, R. (2012)).  A view of a four residue amyloid peptide microcrystal from yeast prion SUP35 (PDB 2OLX)

 

These amyloid fibrils are famously associated with several neurodegenerative conditions, such as Alzheimer’s disease, but there are also so called functional amyloids with productive biological functions.  The fibrils can comprise thousands of repeats of the single peptide and this creates the potential for an impressive degree of complexity to evolve with large concentrations of functional residues allowing high specificity, reactivity and/or binding affinity.  Furthermore, a broad range of functional amyloids are known, exhibiting a diverse functionality which demonstrates that the fold possessed  both an initial selection advantage and the ability to develop further life promoting activities.

This argument demonstrates that amyloids meet all of the initial conditions required of a common ancestor, but Greenwald and Riek go on to explore further characteristics of amyloids which make them persuasive candidates for the originating fold.  

Several amyloid crystal structures have been solved recently, which have illustrated that amyloids can exhibit either dry interfaces, which are highly interlocked, or hydrated interfaces which have twice the distance between β sheets.  The fact that some amyloid fibrils have been demonstrated to employ both interface types as well as the solution of more complex amyloids, adopting structures such as a β solenoid, indicate that from the very simple starting point of a short amyloid peptide, it is possible to evolve structures capable of catalytic action.  

An example of an amyloid in β solenoid structure: HET-s(218-289) prion.  Image adapted from PDB 2RNM.

 

Repetition within amyloids allows breadth of form and therefore function, but it also allows the peptide to act as a template for the seeding of new amyloids in an established conformation.  This replicative function is illustrated by prions, misfolded proteins based on amyloid structures which can infect healthy proteins and convert them into the diseased form.  Deadly though it is in prions, this ability of amyloids to store and replicate conformational information is a further argument in favour of its survival as a common ancestor.

The inherent danger of this ability to replicate has led to evidence that selection is biased against peptide sequences which are prone to β aggregation and the observation that more complex life forms have fewer proteins with this propensity.  This need not argue against amyloids as a common ancestor, however, since the fold could have initiated the evolution of proteins but become more of a liability as replacement folds evolved which were more complex, specific and less prone to aggregation.

In summary, amyloids not only fulfill all of the immediate requirements of a common ancestor but they also have several other characteristics that recommend them to the role. They can be short, composed of prebiotic residues, provide a range of functions, be amenable to modifications and extension, show a variety of binding surfaces and can self replicate.  In addition they have been shown to withstand all of the predicted conditions of the environment 3.5 billion years ago, namely extremes in temperature, UV radiation and pH.   

Whilst there is no evidence to indicate directly that a single ancestral fold lies behind the universal proteome, it is certainly a fascinating idea and one that has been shown by this paper to be at least possible.

Twenty Years of Structural Biology in Drug Discovery

Approximately once a term, scientists working in structural biology and related areas in and near London, meet under the auspices of the London Structural Biology Club to discuss their recent research. These Club meetings generally involve four or five lectures followed by an informal discussion over beer and pizza, generously provided by a sponsoring company.

Most speakers at LSBC meetings are academic researchers and a typical presentation will highlight a novel structure or two, emphasising, for a specialist audience, some of the trickier aspects of how they were solved. The most recent meeting, however – held on 3 July 2012 in Birkbeck’s Clore Management Centre, and sponsored by specialist light scattering company Avid Nano – included one rather unusual talk. Dave Brown recently left the pharmaceutical giant Pfizer after sixteen years in its structural biology group. He now combines an academic post at the University of Kent at Canterbury with work in a new biotech company, Cangenix Molecular Solutions. During his time at Pfizer, Brown contributed his expertise in structural biology to drug discovery programmes for a wide range of cardiovascular, inflammatory and infectious diseases. The topic of his LSBC presentation was the evolution of structural biology in the pharmaceutical industry as he had experienced it during the last two decades.

The history of structural biology in drug discovery goes back a little longer than that, to the use of NMR by companies such as Abbott and Agouron to determine structure-activity relationships for compound series. The first drugs to be largely designed based on structural principles were the HIV protease inhibitors, which are covered in some depth in both section 5 and section 10 of PPS. The first protease inhibitor to be licensed for treating AIDS was Saquinavir (Invirase™), which entered the clinic in 1995, only ten years after the protease gene was first detected in the newly sequenced HIV genome.

By the mid-1990s, most major pharma companies had specialist structural biology groups. The most important developments since then have been driven by technical improvements in molecular biology, gene cloning and protein purification, and in diffraction technology. Synchrotrons such as the UK’s Diamond Light Source make their facilities available for commercial use at a price that large companies, at least, have no trouble affording. X-ray data collection is now extremely fast and can largely be automated so it has become accessible to non-expert users. Although dedicated structural biology groups in industry are no larger than they were in the 1990s, there are many more researchers there – medicinal chemists and molecular biologists – who spend some of their working life doing structural biology. Arguably, a reasonable knowledge of protein structure is essential for all scientists working in drug design.

But what, exactly, does structural biology contribute to the complex process of drug discovery? Once a likely protein target for a new discovery programme has been identified – an enzyme to be targeted by an inhibitor, for example – knowing its three-dimensional structure can help both in the selection of likely starting molecules or “hits” and in their development into “lead” compounds that are potent enough inhibitors to go forward to the later stages of drug discovery. In the related technique of fragment-based drug discovery, a huge number of very small compounds or fragments are screened against a target and those that bind to different parts of the protein’s active site are selected and linked together to form a larger and theoretically more potent compound. Typically, individual fragments bind to their target so weakly that they can only be identified experimentally rather than through computer modelling. Structural experiments can also be used to probe unexpected ligand-binding interactions, perhaps even identifying previously unknown functional binding sites, and to build selectivity for one member of a large protein family into a drug molecule. Selectivity is essential for the development of safe, non-toxic drugs that bind proteins such as kinases and G-protein coupled receptors (see section 11) that have hundreds of homologs in the human proteome.

Brown then went on to describe some of the work that he had been involved with during his career at Pfizer. Phosphodiesterases are enzymes that catalyse the breaking of phosphodiester bonds; this function is responsible for regulating the concentration of the essential cyclic nucleotide monophosphates in cells, and enzymes in this class are drug targets for a range of diseases. Cyclic GMP-specific phosphodiesterase type 5 or PDE5 is the target of Pfizer’s most famous drug, sildefanil, which is known worldwide as Viagra™ (link is to the Wikipedia entry). Researchers at Pfizer have used structure-based techniques to modify the basic structure of this compound into drugs with the potential to treat very different diseases, including Reynaud’s syndrome (in which the blood supply to the extremities is reduced in cold temperatures) and stroke recovery. The recent structure-led discovery of the mechanism by which another protein in this family, PDE4, regulates cyclic AMP may even lead to the design of drugs for a range of devastating neurological conditions including Alzheimer’s disease and schizophrenia.

Brown summed up his lecture with a series of recommendations for structural biologists working in industry (and by extension, for industrial researchers doing structural biology there). He advised them to get involved in drug discovery programmes at the earliest stage, to use structure to understand mechanisms of action rather than simply to design inhibitors and to collaborate between disciplines and sometimes even across companies. And, overall (and more prosaically), to work as quickly and cheaply as reasonable. These are not bad recommendations for academic researchers, either.