| Abstract |
Hierachic organisation of protein structure makes it simple to describe and follow. Yet, the knowledge regarding any -potential- detailed relationships between sequence and structure is far from complete. This sort of knowledge finds value in the fact that, while DNA-based techniques have permitted massive sequence discovery, structure determination by experimental means remains slow, and prediction in energetic (thermodynamic) terms is extremely difficult -if not useless; not to mention that the specificity of the interactions observed at the folded molecule is not encoded at this level. However, at the tertiary structural level, each protein follows one of -relatively- very few structural motifs; this might provide a key to the case. Tertiary structural motifs do not refer to any exact details of any particular protein family, but -instead- they describe how the secondary structural parts can be arranged into a thermodynamically stable and functional whole. Key locations, which may be defined as topologically distinct positions on each scaffold, obey limitations -due to specific interactions with neighbouring positions in 3D space- which should allow compatibility to only a few amino acid types. Since the protein families, which implement each motif, share no similarity other than a common scaffold, the set of these preferences, taken as a whole, should describe the characteristics a sequence must have in order to be compatible with the motif and is irrelevant to origin or function. Do such preferences exist? Seeking an answer, a relevant analysis was performed on a simple recurrent tertiary motif, the four-α-helical bundle. The four helices comprising the motif are packed in an up-and-down or all-anti-parallel fashion; among the two possible twists, the left-handed was chosen. Stability and structural simplicity have proven this motif an attractive system for analysis and design procedures. The four helices are amphipathic and can be described as a regular repetition of seven positions (abcdefg)n. Positions a and d are buried and hence hydrophobic, b, c and f are exposed to the solvent whereas e and g lie on the boundary between the interior and the exterior. Despite the functional and evolutionary variety among the families considered, the results are clear cut: it is not just the expected distribution with hydrophobic residues inside and hydrophilic ones outside; instead, within this framework, an extensive redistribution occurs, where each position is occupied in more than 50% of the instances by no more than five amino acid types, with a and d in about 40% by only two: Leu and Ala. However, while the distribution of alanine is rather smooth over the seven positions, leucine appears more determined. The feeling of this redistribution across (but within) the bundle is reinforced by the fact that the composition of the bundle does not differ markedly from that of all-α proteins; and -sure- it differs much less than it does, when each of the seven positions is considered separately. The interconnecting segments are usually short and hence they have a limited repertoire of conformations: when they are only two or three residues long, prefer (or perhaps demand) particular conformations. These are independent from whether they turn the head of the main chain to the right or to the left relative to the hydrophobic core; however this detail settles for specific patterns of residue types along the connecting segment. Moreover, the helices comprising the bundle, prefer to begin and end at particular positions a-g. More specifically, the four positions following the N-cap prefer to be a-b-c-d or b-c-d-e, whereas the last four just before the C-cap prefer g-a-b-c or a-b-c-d. These preferences facilitate short connections, although ends connected via longer segments show similar preferences. Thanks to a multiple fit, the same amino acid types, compatible with this framework, satisfy the preferences of the end helical turns, for particular amino acid types (known from the literature). These data, taken together (distribution both around and along the helices, behaviour of the interconnecting segments) constitute part of specialised knowledge regarding the particular structural motif. Knowledge of the kind can facilitate the complete broad spectrum of protein design procedures, ranging from evaluation of data resulting from local modifications up to the de novo design of a complete protein molecule. To demonstrate this, an example is described in detail, which utilises the complete set of matrices and observations made in previous chapters. This way, the general scaffold is rapidly specified, except for details not covered by the nature of these matrices, which are best left to human intervention and/or brute force, systematic, combinatorial approaches. This work is completed after an extensive investigation, concerning the statistics of the conclusions drawn, as well as the kind of sequence-based structure prediction they can facilitate, along with manners available and conditions necessary for it. This investigation reveals the real nature of the results, the tolerance of which finds a neat explanation on molecular evolutionary grounds as well.
|
Collections
