Modeling a Circular AFP

After weighing the pros and cons, we decided to work with Type III AFP, from the ocean pout. Relatively active in antifreeeze activity, it serves as an ideal AFP for use in industrial purposes. Furthermore, its termini are only 19.8 Angstroms apart, making it easier to circularize using a smaller linker (Figure 1).

The modeling process for the circularization of an antifreeze protein can be explained by a three-step approach to protein design (seen below).

Linker Design & Spatial Fitting

Approach 1: The idea for this project was meant to be a continuation of the work of team Heidelberg 2014 and validation of their parts. Using their intein BioBrick we wanted to circularize our own protein, a Type III AFP. Using the CRAUT software , we ran our protein through the program. After fixing syntax errors, the following linkers were suggested:

Where sequences in red represent the extein sequences (scars that will be left after the intein reaction), and sequences in purple represent a secondary structure (alpha-helix). Preliminary spatial evaluation using PyMOL suggested these were not ideal for linking the termini of our protein of interest. For the purpose of progression we adapted the second sequence and opted for running simulations on our own linker designs.

Approach 2:Using the extein sequence tested by Heidelberg 2014, linkers were built within PyMOL and Coot to effectively joined the N- and C- termini. Flexible linkers were chosen to enable the termini to arrange such that the protein remains in its functional conformation.

Modeling Methods & Stability Analysis

Selected linkers, from Approach 2, were then run through molecular dynamic (MD) simulations using the GROMACS package . Following energy minimization, the stability of the protein under physiological conditions was determined. Protein files were edited to circularize the backbone and solvated in water at 298 K to mimic those functional conditions (Figure 2).

Statistical analysis was used to compare the stability of each circularized protein with our 3 linkers found in Approach 2. Root mean square (RMS) deviation fluctuation plots are shown in Figures 3 and 4 below.

AFP-Scaffold Design

Scaffold

A self-assembling protein scaffold introduces the ability to create a platform from which to create attachment between different subunits. Baker and Yeates have successfully created a number of such scaffolds with verified crystal structures⁴.

After examination of the different constructs available, we have chosen to use the T3-10 model (Figure 5). This single subunit 12-mer has a less complicated assembly procedure as compared to a 2-unit scaffold such as T33-21. The position of the C-terminus points away from the assembled structure enabling the attachment of an engineered part. Expression and assembly requirements are easily met at room temperature from a single vector. With 12 identical subunits, the scaffold could potentially house 12 antifreeze proteins.

Antifreeze Protein

The Type III antifreeze protein from ocean pout, the same used in the circularization project was selected for the scaffold system as well (See Figure 1). This is a well characterized protein, whose sequence was easily accessible by QGEM this year.

The E/K-coil system

We sought a strong, highly efficient means to connecting AFPs to the self-assembling scaffold. After considering our options, we opted for the E/K coil method used by Calgary iGEM 2013 . These non-covalent interactions are highly specific and should enable selective binding between AFPs and the scaffold subunit. Read more about coiled-coils on our Background page.

The coil sequences to be used:

Upon examination of their work and the sequences used, we noted a discrepancy in their data. According to the PDB file and NMR structure elicited in 2004 ⁵, the coils interact in a parallel fashion, incorrectly identified as anti-parallel in the registry: K coil and E coil. This has been reviewed by us to indicate the correct interaction according to the PDB file 1U0I.

To generate a model of our complex, the E-coil was fused to the Type 3 AFP and the K-coil was fused to the scaffold subunits. The expected interaction is a parallel alignment as depicted in Figure 6. The introduction of a flexible region between the C-terminus of the scaffold subunit and the coil is introduced to allow the coil movement to enable to interaction to occur without steric hindrance from the protein units.

Coiled-Coil Stability Simulations

Before using these coils we sought out to test their stability and method of interaction in its native form. The coils selected for our project is represented by the PDB file 1U0I, an NMR-solved crystal structure. These coils are described to interact in a parallel fashion (Figure 7a). To simulate this interaction, the individual coils were solvated and run thorugh MD simulations in GROMACS.

To simulate an antiparallel coiled-coil interaction, the K-coil's amino acid sequences were inverted such that it sat antiparallel to the E-coil. The file was then run though PyRosetta docking programs to undergo energy minimization, generating the most stable conformation. This file, with separated coil subunits, was then also run through GROMACS simulations (Figure 7b).

PyRosetta Docking & Linker Testing Program

AFP & Scaffold Docking

In order to test the self-assembly of the AFPs and scaffold proteins with E/K coils, docking simulations were run. These were used to assess the energetic stability of the coiled coil interaction and determine the orientation of the ice-binding surface of the AFP. This was done using PyRosetta by following the standard procedure for initial low resolution docking prior to high resolution docking on favourable protein structures. Sorting of low energy dockings was used with consideration given to the proximity of the E/K coils to choose a final selection of proteins for refinement and scoring. The general procedure used is outlined in Figure 8 and the final docked structure reached shown in Figure 9.

Circularization Program

In order to run molecular dynamic simulations of the circularized AFP, the linkers were individually fit with the scar sequence to optimize positioning to allow the terminal nitrogen and carbon groups in bond length proximity (approximately 1.3 angstroms). To save time and energy in spatially testing multiple linkers, a script was written to utilize PyRosetta’s capabilities to circularize the protein for easy visual assessment of atom positioning. This involved using the small mover method in PyRosetta to allow each movement to be checked in ‘fitting’ the linker to ensure the dihedral angles were within acceptable ranges. The script for this program can be located here.

References

1. The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC.

2. Chao et al. (1994). "Structure-function relationship in the globular type III antifreeze protein: Identification of a cluster of surface residues required for binding to ice". Protein Science. 3(10):1760-1769.

3. Sonnichsen et al. (1996) "Refined solution structure of type III antifreeze protein: hydrophobic groups may be involved in the energetics of the protein-ice interaction". Structure. 4(11):1325-1337.

4. King et al. (2012). "Computational Design of Self-Assembling Protein Nanomaterials with Atomic Level Accuracy." Science. 336:1171-1174

5. Lindhout et al. (2004)."NMR solution structure of a highly stable de novo heterodimeric coild-coil." Biopolymers. 75:367-375.

6. Atkins and de Paula (2006). "Atkins' Physical Chemistry". 8th edn, W.H. Freeman. pg.637.