[Paper Review] Designability of alpha-helical Proteins

Eldon Emberly, Ned S. Wingreen+1 more|arXiv (Cornell University)|Jun 20, 2002

Glycosylation and Glycoproteins Research4 citations

TL;DR

This paper presents a computational method to generate compact, stable packings of four alpha-helices connected by turns, reproducing known natural four-helix bundle folds with high accuracy (within 3.6 Å RMSD per residue). It identifies a small set of highly designable structures—most matching known fold families, but also revealing novel topologies—using a hydrophobicity-based energy model to assess sequence compatibility, enabling targeted design of new protein folds.

ABSTRACT

A typical protein structure is a compact packing of connected alpha-helices and/or beta-strands. We have developed a method for generating the ensemble of compact structures a given set of helices and strands can form. The method is tested on structures composed of four alpha-helices connected by short turns. All such natural four-helix bundles that are connected by short turns seen in nature are reproduced to closer than 3.6 Angstroms per residue within the ensemble. Since structures with no natural counterpart may be targets for ab initio structure design, the designability of each structure in the ensemble -- defined as the number of sequences with that structure as their lowest energy state -- is evaluated using a hydrophobic energy. For the case of four alpha-helices, a small set of highly designable structures emerges, most of which have an analog among the known four-helix fold families, however several novel packings and topologies are identified.

Motivation & Objective

• To develop a systematic method for generating compact, stable packings of secondary structural elements (specifically four alpha-helices) connected by turns.
• To reproduce known natural four-helix bundle folds from the SCOP database with high structural accuracy.
• To identify novel, non-natural packings that could serve as targets for de novo protein design.
• To quantify the designability of each structure as the number of amino acid sequences that stabilize it as the lowest energy state.
• To evaluate the potential of these structures for ab initio design by assessing their sequence compatibility using a hydrophobic energy model.

Proposed method

• The method generates ensembles of four-helix packings by random sampling of possible spatial arrangements, with helices treated as rigid bodies connected by short turns.
• Structures are compared using the root-mean-square deviation (RMSD) of alpha-carbon positions (crms), with similarity defined as crms < 1.5 Å.
• Sampling stops when 95% of newly generated stacks are similar to existing ones in the ensemble, ensuring completeness.
• A clustering procedure reduces redundancy by eliminating stacks within 1.5 Å crms of a more compact representative, compressing the ensemble by a factor of 3–5.
• Designability is computed by assigning binary hydrophobic (H) and polar (P) residues to sequences, with energy E_h = Σ h_i s_i, where s_i is side-chain exposure and h_i = ±5k_B T for H and P residues.
• The compactification energy h_0 = 2k_B T is calibrated to match surface area distributions of natural four-helix bundles.

Experimental results

Research questions

• RQ1 Which compact four-helix packings can be generated using a systematic sampling method of rigid helices connected by turns?
• RQ2 To what extent can known natural four-helix bundle folds be reproduced by this method?
• RQ3 What fraction of the generated packings are novel, with no known natural counterparts?
• RQ4 Which of the generated packings are highly designable, i.e., stabilized by a large number of sequences?
• RQ5 How does the hydrophobic energy model predict the relative designability of different helix packings?

Key findings

• The method successfully reproduces all known natural four-helix bundle folds from the SCOP database with a root-mean-square deviation of less than 3.6 Å per residue.
• A small subset of packings emerges as highly designable, with most corresponding to known four-helix fold families such as the helical bundle and the up-and-down motif.
• Several novel packings and topologies are identified that have no natural counterparts, suggesting potential for de novo protein design.
• The designability of a structure correlates strongly with its compactness and low surface exposure, as expected from energy minimization principles.
• The hydrophobic energy model with h_0 = 2k_B T and δh = 3k_B T per residue (H: 5k_B T, P: -1k_B T) provides a good fit to the surface area distribution of natural four-helix bundles.
• The clustering procedure reduces the initial ensemble size by a factor of 3–5 while preserving the most compact and designable structures.