Wenxiang 3.0: Evolutionary Visualization of α,π,and 3/10 Helices

Abstract

Wenxiang diagrams illustrate protein helices as spirals on a plane and thus have the advantage over helical wheels of being planar graphs. Wenxiang 3.0 extends the original version by adding 3 major features: (1) individual amino acid residues can be colored according to their evolutionary conservation in comparative multiple sequence alignments using CONSURF encoding; (2) α, π, and 3/10 helices can be illustrated by overlaying arcs representative of the pitches of these helices; and, (3) the physico-chemical properties of amino acids residues in the protein sequence can be re-presented by colored geometric shapes.

Introduction

Wenxiang diagrams ¹ illustrate protein helices as spirals on a plane (Figure 1) and thus have the advantage over helical wheels ² of being planar graphs. Wenxiang 3.0 extends the original version by adding 3 major features: (1) individual amino acid residues can be colored according to their evolu-tionary conservation in comparative multiple sequence alignments using CONSURF ³ encoding (Figure 1a and b); (2) α, π, and 3/10 helices can be illustrated by overlaying arcs representative of the pitches of these helices ((Figure 1b and c); and, (3) the physicochemical properties of amino acids residues in the protein sequence can be represented by colored geometric shapes (Figure 1c).

Figure 1.

(a) Wenxiang 3.0 interface wherein the sequence of a 3/10 helix from 1GPB (glycogen phosphorylase B from rabbit (Oryctolagus cuniculus)) and its notation of evolutionary conservation using the key from CONSURF applied to a correlated multiple sequence alignment. (b) Arcs representing hydrogen bonding interactions for an alpha helix in 1GPB (glycogen phosphorylase B) with sequential residue numbers on each amino acid residue in the peptide and again with evolutionary conservation coloring. (c) Arcs representing hydrogen bonding interactions for a Pi helix in 1GPB (glycogen phosphorylase B) with coloring of physicochemical properties (acidic, basic, hydrophobic, polar) and geometry of size (triangular = small, rhombic = medium, pentagonal = large, and hexagonal = very large).

Besides helical wheels, 3 additional planar visualization tools of alpha helices in protein secondary structure include helixvis, ⁴ HELIQUEST, ⁵ : and helical net-diagrams. ⁶ We examined all 4 approaches to develop 5 different design criteria that we felt would be particularly helpful in featuring different aspects of the structure, function, and evolution of helices in proteins. In order to find relevant α, π, and 3/10 helical regions in protein sequences, we used STRIDE. ⁷

Software Features

Our 5 different reasons for developing Wenxiang 3.0 are: First, π and 3/10 helices are not currently represented in any helix planar visualization tool of protein secondary structure that we are could find. Kumar and Bansal ⁸ review the importance of π helices and Cooley et al ⁹ and Khrustalev et al ¹⁰ review the importance of 3/10 helices. Ren et al ¹¹ stress that large winding and unwinding of helices occur in membrane proteins such as rhodopsin and these result in transformations among these 3 configurations. Thus, a tool for analyzing potential helical interactions of a single amino acid sequence can provide helpful insights. As educators, we have extensively used origami models of α, π, and 3/10 helices ¹² and 3D printed models of helices (Herman et al ¹³ and Davenport et al ¹⁴ ). Yet our students still struggled with understanding the differences between these 3 types of helices and the importance of using evolutionary conservation to comprehend which residues or residue types were important to sustaining structure and function of proteins.

Second, in order to understand the different pitches of α, π, and 3/10 helices we wanted to overlay arcs representative of whether hydrogen bonding was occurring between either every third, fourth, or fifth residue in the pitches of these helices. These are in close correspondence to the various pitches which are usually reported to be pitches of 3 residues per turn for 3/10 helices ²⁰ , 3.6 for α helices, and 4.4 for π helices. While there is not a one-to-one correspondence between our integer values with measured values fractional values around these 3 integers, we believe our visualization of arcs representing different helical pitches has heuristic value in inferring vertical interactions between different amino acid residues involved in helical structures that contribute to their stability. We are particularly interested in identifying weak non-covalent potential interactions including hydrophobic interactions, salt bridges, van der Waals forces, etc. that have been evolutionarily conserved.

Third, residue size (small, medium, large, very large) was based upon the Voronoi polyhedral volumes (in cubic Angstroms) of amino acid residues which were divided into these bins based upon breaks in properties.

Fourth, chemical properties residues (acidic, basic, polar, hydrophobic) could be represented by different geometries and colors.

Fifth, and finally, we adopted a numerical identification of successive residues in the helical region of the amino acid sequence.

Use of the software

For generating visualizations in Wenxiang 3.0 like Figure 1a and b, we recommend that users identify proteins with published 3D structures determined by X-ray crystallography and that they have generated multiple sequence alignments of sequences similar to that of the published 3D structure. If a user is only interested in visualizing the physicochemical interactions and the arcs of α, π, and 3/10 helices of an individual protein such as in Figure 1c, the user can simply enter an amino acid sequence (our current limit is 18 residues long). Specific choices of whether to implement: curves for which of the 3 helices, coloring of degrees of evolutionary conservation, and chemical properties of different amino acid residues are simple check boxes on the user interface (Figure 1a).

Future directions

The decision of which features were most important to include was difficult. The number of physicochemical categories of amino acids varies enormously from 2 (H: hydrophobic/hydrophilic) to the 4 shown herein to Dayhoff et al ’s ¹⁵ 7 and Knight et al ¹⁶ 8. Since there are so many different multiple sequence alignment software packages, we will be implementing some data wrangling routines for importing from ones identified by our users. In terms of 3D representations of evolutionary conservation, 3DPatch ¹⁷ has the advantage that it “does not require the user to have prior knowledge of any 3D structures matching the query sequence.” However, because it does not have as many features as CONSURF and depends upon entropy information measures rather than section, we decided to use CONSURF in our current version of wenxiang 3.0. While we have used STRIDE to find examples of α, π, and 3/10 helices, it would be helpful to have a general-purpose tool to find various helical forms more generally. Le and Nguyen ¹⁸ and Tng et al ¹⁹ have used deep learning techniques to explore proteins that span membranes in mitochondria such as members of the electron transport chain. Their use of measures of specificity, sensitivity, and accuracy as well as visualization with sequence logos may provide useful extensions for identifying α, π, and 3/10 helices.