About two thirds of polypeptides in existing native folded proteins occur as α-helix, as in Fig 1a, or β-sheet, shown in Fig 1b. Native proteins also have a few percent of short pieces of polyproline type II helix (PPII; polypeptides with this conformation, in spite of the name, are not necessarily composed of prolines), as in Fig 1c. However, when it comes to unfolded native proteins, or to short polypeptides, evidence 31 suggests that PPII is the most common conformation. At first sight, it might seem likely that the earliest proteins would also tend to consist of PPII. However, the earliest peptides are expected to be made of mixtures of L and D amino acids 7. In the same peptide, they would be incompatible with the chiral PPII helical structure. The same applies for chiral α-helix. This leads us to consider enantiomeric peptides, which tolerate, and can be favored by, such amino acid compositions.

In regularly-repeating polypeptides the main chain conformations of successive amino acids are the same, or at least roughly so. Polypeptides can also fold to form a different type of structure where the main chain conformations of successive amino acids are enantiomers (mirror images). The φ, ψ values of one amino acid (the φ, ψ torsion angles, measuring rotation about the N⋯Cα and Cα⋯C bonds, are major determinants of polypeptide conformation) are equal to the φ, ψ values of its neighbour multiplied by -1. Such polypeptides 3233 are achiral and without pitch, forming rings or chains, but not helices, as in Figs 1d–g and 1j–l; they are expected to be favored by achiral or alternating racemic amino acids. This may be contrasted with the various chiral helices (strands of β-sheet are geometrically types of helix) observed in modern proteins with mainly chiral amino acids. Enantiomeric polypeptides occur as two kinds, nests and α-sheet, described below.

Nests are motifs made from 3–6 amino acid residues where successive main chain NH groups bind anionic atoms or groups 343536373839. One with 3 residues is shown in Fig 1d. Currently, 5–8% of residues in proteins belong to nests 3436. Their residues have main chain φ, ψ conformations alternating between about -90°, 0° and +90°, 0°. In most nests an anionic atom or molecule is firmly hydrogen bonded to the NH groups of residues i and i+2 and less strongly or hardly at all to the i+1 NH group, which often points slightly away. In typical present day nests glycine is preferred at alternating residues resulting in a conformation where the anion is partially encircled. This is because glycines are achiral, and can adopt the φ = 90°, ψ = 0° conformation favoured by D-amino acids equally with the φ = -90°, ψ = 0° one favoured by L-amino acids, i.e., they can behave like the D enantiomer. As glycine was probably the main amino acid generated within the hydrothermal reactor 40, it is likely nests were more abundant in early peptides than in present day proteins. But equally the synthesis of alternating L-amino acid/glycine, D-amino acid/glycine or L-amino acid/D-amino acid peptides 41 in the mound would give rise to nest conformations with anion binding capacity.

The clue as to how inorganic phosphate was first ligated is given by the P-loop, the commonest and best known of the protein motifs that bind ATP/ADP or GTP/GDP 12, occurring in a wide variety of intracellular enzymes. It has been referred to as a giant anion hole 13 and has a characteristic GxxxxGKS/T consensus sequence 34. Fig 1e shows that the P-loop has a nest incorporated within it made from five successive main chain NH groups which wrap around the β-phosphate of the nucleotide phosphate ligand, forming hydrogen bonds with the phosphate oxygen atoms. It seems probable that, in the earliest proteins, nests formed binding sites for the anionic groups of polyphosphate and acetyl phosphate (for immediate energy provision) and nucleoside polyphosphate (later, for coenzyme and RNA synthesis).

In about a half of all current protein 3D structures with iron-sulfur centres, nests with between 3 to 6 adjacent main chain NH groups are employed to help bind the Fe₂S₂ or Fe₄S₄ centres 34. One is illustrated in Fig 1f. Like the P-loop, these features could also be evolutionary relics. Centers bound to two separate nests are not unusual. Fig 1c shows a cubane iron-sulfur center in the protein ferredoxin bound by a 6 amino acid nest. Most iron-sulfur centres are bound by cysteine thiol groups. In the earliest stages of evolution it seems probable the places of the cysteine thiol groups were occupied by organic sulphides from the hydrothermal fluid 1142, and that cysteine binding was a later development. Nanocrystals of FeS are assembled from [2Fe-2S] rhombs 14. The first step in growth to a potential active center is the formation of [Fe₂S₂] and [Fe₄S₄] clusters. At this stage hydrothermal thiols (RS-) tend to inhibit further growth as they sequester the rhombs as [Fe₂S₂(SR)₄]^1-/2- and the cubanes as [Fe₄S₄(SR)₄]^2-/3- in what may be termed "protoferredoxins" 1143. The sulfur atoms confer an overall negative charge to the complexes, complementing the δ+ on the NH hydrogen atoms of the nests. It seems likely nickel-bearing FeS clusters are sequestered similarly.

Amyloid is the name given to characteristic insoluble proteinaceous fibres that are resistant to detergents and proteases. Amyloid is thought to be the primary cause of many diseases, called amyloidoses, including Alzheimer's, Parkinsons', Type II diabetes and CJD. The amyloid of prions, as in CJD, is special in being highly infective because their fibres propagate as well as being self-templating 44. In each of these diseases, a particular protein slowly becomes converted into amyloid, giving rise to the typical late onset condition. It appears, however, that amyloid is more than an aberrant form of certain proteins produced in diseases but rather a major alternative form of most, if not all, proteins. A few native proteins do form amyloid naturally for functional reasons, such as some spider silk and bacterial surface fibrils, but these are the exception. The majority of native proteins fold to form complex, but well defined and soluble, three-dimensional structures often with α-helices and relatively small clumps of β-sheet (1–4 sheets, each clump having only 10–300 amino acids). However, most of these proteins left in aquaeous solution away from organisms eventually turn into amyloid. It seems that evolutionary pressure has ensured the majority of native proteins produced in organisms avoid turning into amyloid in vivo.

The insoluble fibres of mature amyloid are thought to consist of stacks of say hundreds of β-sheets each with hundreds of long β-strands. Its size differentiates it from the β-sheet of soluble proteins. A small portion of a β-sheet is in Fig 1i. Current work 4546 favours parallel β-sheet, with the in-register strands perpendicular to the fibre axis. A feature making amyloid specially suitable in early evolution was shown 47 by randomly shuffling yeast prion domains and finding they still form amyloid. It seems the amino acid composition rather than the sequence is all that is needed to form amyloid. Amyloid three-dimensional structure is simpler than the complicated arrangements of native proteins, yet, because it is sticky and does not crystallise easily, precise knowledge about its structure has been difficult to gather and still presents problems for scientists. However, in early evolution this stickiness of amyloid is likely to have been useful in forming protective insulating layers, so is a plausible candidate for the material of the first proper organic membranes, as has been suggested 484950. This idea is illustrated in Fig 2b.

Certain short, non-native, polypeptides form a substance called α-sheet 51525354. Conformationally, a single polypeptide chain of it is another example of an enantiomeric polypeptide and, as seen in Fig 1g, looks like a flattened version of the nest. Typical φ, ψ angles are -60°, -60°; 60°, 60°. Like nests, polypeptides consisting of regularly-repeating D and L amino acids favour this conformation. Multiple-stranded α-sheet resembles β-sheet, except that it is polar. One edge has a row of δ- charges, and the opposite edge has a row of δ+ charges, as in Fig 1l.

The toxic feature of amyloid in diseases is not the mature amyloid but a soluble precursor. Computer simulations by Armen et al. 55 give rise to the suggestion this intermediate is composed of α-sheet. Peptide plane flipping provides an easy way for α-sheet to be converted into the β-sheet of mature amyloid. Although α-sheet is rare in native proteins, peptide plane flipping of this sort occurs commonly 56, providing support for this amyloid formation pathway. The special charge distribution mentioned may underlie its toxicity on the one hand and its ability to polymerize on the other.

The rarity of α-sheet in native proteins compared to its relative stability and ease of formation in small polypeptides is striking. In any case, from the expected φ, ψ angles it would be expected to be more common than it is in native proteins. This apparent anomaly could be explained 565745 by it being so toxic as an amyloid precursor that natural selection has eliminated most proteins that form α-sheet. Overall, whether or not α-sheet does prove to be the main precursor to amyloid, it seems likely it was a common conformation of the earliest peptides.

One problem when considering the nature of early membranes is that the constituent materials are either too permeable or too impermeable to ions and water molecules. What is needed is a semi-permeable membrane, the traditional name for the plasma membranes of existing living cells.

Although α-sheet with hydrogen bonds between strands is rare in native proteins in their natural state, polypeptides with the same conformation but not arranged as α-sheet do occur in some integral membrane proteins. A group of four strands four amino acids long line the selectivity filter of the potassium channel such that the main chain carbonyl oxygen atoms form a cation channel 5859, as shown in Fig 1j. A feature of this sort between blocks of α-sheet as in Fig 3a might have been used for cation transport early in evolution. In the water-transporting protein aquaporin a single strand with the same conformation lines one side of its water channel 60. The row of main chain carbonyl oxygen atoms in this strand binds the δ+ protons of a corresponding row of water molecules in the channel, as in Fig 1k. Such channels can transport either protons (via a so-called proton wire) or water molecules, depending on the organisation. Aquaporin has two apparently similar but opposing channels arranged head to head such that water, rather than proton, transport is favored.

These observations, deriving from the polarity of α-sheet, lead to the idea that the edges of randomly arranged blocks of otherwise impermeable α-sheet would be likely to spontaneously form either proton, water or cation channels. In a membrane α-sheet has to be in one polarity or the other. Blocks of sheet that grow independently may happen to meet another block so that the electrostatic interaction between them, δ+ to δ-, is favorable. In this case they combine to form a larger sheet as in Fig 3b. However, if they meet in the δ- to δ- orientation, a potential cation channel is formed as in Fig 3a, whereas, if they meet side-on as in Fig 3c, a proton or water channel is liable to be formed. These effects are also indicated in Fig 2b.

Copper, nickel, cobalt and iron cations readily complex with peptides in alkaline solutions by binding to successive main chain amide nitrogen atoms 61. Amide NH protons are displaced by the metal during the formation of these tight complexes in which the main chain atoms adopt a flat conformation (φ = 180°, ψ = 0°), as in Fig 1h. Such complexes, with up to four amide protons substituted by metals, also occur naturally within a number of native proteins (prion protein of CJD) and enzymes (acetyl CoA synthase), sometimes at active sites. The Ni-tetraglycine complex of Fig 1h62 exhibits a remarkable similarity to present day macrocyclic tetrapyrrole cofactors. Also the analogous Co-tetraglycine resembles a corrinoid group. In these types of complex the metal has octahedral coordination to four planar nitrogens with two positions free to assist in catalytic reactions. The peptide ones could thus have performed much the same catalytic functions as hemes, corrinoids and F₄₃₀, before being largely, but not entirely, superceded by them. Coenzyme F₄₃₀ is a nickel tetrapyrrole with the ability to bind and reduce methyl groups. As such it is an essential cofactor in two enzymes in the methanogenic branch of the acetyl-coenzyme A pathway, considered to be one of the earliest pathways at the origin of life 6364. The simpler, acetogenic branch of the acetyl coenzyme-A pathway involves an iron-sulfur protein with a corrinoid cofactor – a cobalt-tetrapyrrole – that transfers a methyl group to acetyl coenzyme-A synthetase 646566.

This is an interesting paper pointing out the possible types of interactions that could have taken place between short heterochiral peptides and inorganic catalysts during chemical evolution. It makes several novel and worthwhile points.

This article continues the series of studies by Russell and coworkers on the origin of life in networks of inorganic compartments existing at hydrothermal vents. This paper focuses on primordial amino acid and peptide chemistry and attempts to infer the distribution of conformations of primordial, primarily, abiogenic peptides and small proteins. Predictably, these abiogenic peptides are thought to be heterochiral and as a result, the population of the conformational space would be very different from that in modern proteins. I am particularly interested in the authors' conjecture that abiogenic peptides, such as the anion-binding nests, could facilitate the formation of the first transmembrane potential and so could be central to the emergence of bioenergetics. Other unusual structures would contribute to the formation of the primordial membranes themselves and membrane channels, and yet others would become catalysts. By and large, I find the scenario of the participation of biogenic peptides in the emergence of biochemistry and bioenergetics outlined in this paper to be interesting and plausible. I also agree that P-loops and iron-binding loops must have been among the earliest peptide structures.

The biggest remaining question is how did the transition from the abiogenic, enantyomeric was brought about. Recently, we outlined a scenario under which the utility of amino acid and peptides in the emerging biological systems drove the evolution of translation, through the intermediate stage of peptide ligases (Wolf, Koonin, On the origin of the translation system and the genetic code in the RNA world by means of natural selection, exaptation, and subfunctionalization. Biol Direct 2007 May 31;2:14). I believe that, at least, conceptually, this line of thinking about the origin of life is remarkably well compatible with the ideas on the role of abiogenic peptides in primordial biochemistry developed by Milner-White and Russell. Perhaps, this connection could be briefly discussed in the paper.

In a somewhat unusual review, the authors discuss potential conformations of ancient pre-biotic peptides and present hypotheses about their possible interactions and functions. While nothing in the manuscript contradicts the main principles of physics and evolution, I am left with a mixed feeling. Why would the conformation of early peptides (obviously prebiotic, since they are DL mixes) be of interest when we are not sure whether the hypothesis about the importance of pre-biotic peptides has any merit? We know that chemical, pre-biotic peptide formation is not an easy process (e.g. experimental papers the authors cite), so there is no basis to believe that polypeptides of reasonable length were any abundant, and that they have played a significant role early on in prebiotic evolution. It is equally not clear why regular DLDLDL mixtures should form in pre-biotic conditions. Thus, it seems to me that prior to addressing the question about conformations, it would be nice to discuss these questions. The hypothesis that polypeptides were rare and not significance in prebiotic evolution makes better sense to me. The fact that peptides are so common and important today, does not imply that they were significant early on. In any case, this is such an imprecise area of research, that it would be good to describe any alternative scenarios and potential pitfalls, other hypotheses and rationalizations. Presentation of these alternatives would make for a more interesting reading.