Model

Introduction

---

In the realm of synthetic biology, innovations are ceaselessly emerging, driving forward the potential of biological systems. Among the plethora of bioengineering strategies, the concept of 'Tag' and 'Catcher' has gained substantial attention. Drawing inspiration from the SpyTag and SpyCatcher systems, our team sought to develop a unique protein-protein interaction module. Our design bifurcates into two key entities:

• Entity One: A fusion protein comprising our Catcher, adjoined to laccase, an enzyme specialized in decolorization.
• Entity Two: Another fusion protein, connecting our Tag with GFP (Green Fluorescent Protein) and a hydrophobin protein, tailored to bind PET surfaces.

The crux of our design hinges on the strategic interaction between these entities: Entity Two latches onto PET surfaces, serving as an anchoring point, while Entity One embarks on its decolorization mission. Given the size constraints for bacterial expression, splitting these functionalities was crucial. This strategic bifurcation ensured efficient bacterial expression, all the while preserving the functionality through the integral Tag-Catcher interaction.

Objective

---

The primary goal of our modeling is to visualize and predict the 3D structural interaction between our designed Tag and Catcher. With the vast complexities of protein interactions, a computational simulation can offer preliminary insights into how these proteins might interact in vivo, setting the stage for further experimental validation.

Methodology

---

3D Structure Simulation with AlphaFold

To comprehend the intricate details of our proteins, we turned to AlphaFold - a state-of-the-art protein structure prediction tool. Its deep learning-based approach provides a robust platform for predicting protein folding patterns.

• Procedure: After retrieving the sequences for our proteins, we input them into AlphaFold. The system then predicted the 3D conformation based on the sequence information.

Protein-Protein Docking with ClusPro

Once individual structures were established, the next challenge was to simulate their interaction. ClusPro, a recognized protein docking tool, was our choice.

• Procedure: Using the 3D structures from AlphaFold, we input our Tag and Catcher proteins into ClusPro to predict their potential binding sites and interaction energetics.

Results and Analysis

---

Using AlphaFold for 3D structural predictions

To kick off our exploration, we first delved into understanding the predicted 3D conformations of our proteins. The visualization from AlphaFold brought to light the intricate features of our proteins, distinguishing between high and low confidence regions.

Figure 1: 3D Structures of the proteins predicted by AlphaFold. Color coding: High confidence (green), Medium confidence (yellow), Low confidence (red).

This figure is comprised of three subfigures, representing distinct aspects of our protein model:

• 1A (leftmost): This displays the complete 3D structure of the synthesized protein with color-coding to indicate confidence levels across all residues: High confidence (green), Medium confidence (yellow), Low confidence (red).
• 1B (center): This image isolates the laccase segment of the structure. The laccase is vividly colored representing its respective confidence levels, while the Catcher portion is de-emphasized using a grayscale.
• 1C (rightmost): Contrasting 1B, this visualization emphasizes the Catcher segment, with colors indicating its confidence levels, while the laccase portion is presented in grayscale.

Particularly, the laccase segment's high-confidence regions aligned well with known structures, such as the 2WSD from the PDB, supported by a TM-score of 0.98461 and an RMSD of 1.14 Å from TM-align, indicating a strong structural similarity and close atomic alignment, respectively. These metrics not only validate our predictive model but also suggest that our laccase structure accurately reflects real, biologically relevant conformations, establishing a solid foundation for further applications and studies.

The figure displays the superposition of our predicted laccase (blue) and the 2WSD from the PDB (red) using TM-align, showcasing their structural similarity. Numerical validation is provided by a high TM-score and a low RMSD, indicating a close topological and spatial alignment between the structures.

Detailed Analysis on pLDDT Score Variation

In our detailed assessment of the pLDDT scores across the laccase and Catcher sections, we discerned distinctive patterns in the confidence levels of the structural predictions.

Figure 3: The pLDDT scores along the residue sequence of the combined protein structure, revealing confidence fluctuations and regions of interest.

Within the laccase segment, we noted several regions where the pLDDT scores dipped, often correlating with sections where the structural model indicates a transition from the core to the exterior and then looping back towards the core.

In contrast, the Catcher segment, particularly past the residue 510 (number assumed and needs confirmation), exemplified a drastic decline in pLDDT scores, aligning with previous research assertions about the unreliable nature of these sequence stretches.

Entity Two: Hydrophobin Simulation Challenges

In our extensive modeling endeavors, we recognized significant variability in the simulated results for Entity Two, especially the hydrophobin protein component. Several factors could be attributed to this variability:

• Environmental Dependencies: Hydrophobin's structure and functionality are potentially influenced by pH and temperature. Accurately capturing these factors in simulations can be a daunting task.
• Interactions: The interactions of hydrophobin with other molecules, such as lipids, water, and other proteins, can further complicate its structural prediction.

This figure illustrates four distinct aspects of Entity Two's predicted 3D structure and its pLDDT confidence scores:

• XA (top left): This panel shows the complete 3D structure of Entity Two with color-coding to represent the pLDDT scores across all residues: High confidence (blue), Medium confidence (orange), and Low confidence (red).
• XB (top right): Focusing on the Tag, this image shows its structure in vivid colors representing pLDDT confidence scores, while other structures are de-emphasized using grayscale.
• XC (bottom left): This portion emphasizes the GFP, displaying it in vibrant colors according to its pLDDT scores, while rendering other structures in grayscale.
• XD (bottom right): Contrasting XC, this visualization highlights the hydrophobin segment, color-coding it based on pLDDT scores, while other structural components are displayed in grayscale.

These visualizations help underscore the distinct confidence levels across various regions of the protein, revealing patterns and areas that might necessitate further investigation or experimental validation due to their lower predictive confidence.

Given this pronounced variability and the inherent challenges associated with hydrophobin simulation, it was deemed necessary to omit the detailed results of Entity Two from our main findings. However, it's crucial to note that understanding such complexities and uncertainties is integral to the iterative process of scientific inquiry and protein engineering.

Protein-Protein Interaction and Docking: Initial Interaction Between Tag and Catcher

Delving deeper into the intricate functional dynamics, our exploration into the protein-protein interaction between our Tag and Catcher unveiled several challenges.

Initially, the docking results between the full-length Catcher and the Tag were not as promising as we'd hoped. In fact, there was no substantial binding observed between them, leading us to pivot our strategy. Rather than persevering with an incompatible interaction, we chose to streamline the Catcher by removing extraneous sequences. This adjustment painted a different picture altogether: the docking profile underwent a significant improvement, exhibiting a more congruent and cooperative interaction.

Comparative Analysis with Known Structures

Literature dives and structural databases like the PDB serve as repositories of knowledge. Drawing from this, our Catcher's sequence bore a striking resemblance to the structure of the 4MLI PDB file, especially post-refinement.

Figure 6: Side-by-side structural comparison of the 4MLI PDB file (top) and our refined Catcher/Tag model (bottom), with the Catcher in orange and the Tag in blue.

The initial hindrances in docking, as seen in the protein-protein interaction stage, can be attributed to possible steric hindrances or sequence extensions - a testament to the real-world challenges in protein engineering.

Discussion

---

Comparing our findings with the literature, our Catcher's sequence closely resonates with the structure of the 4MLI PDB file, especially after refinement. The initial inhibitions in protein docking were attributed to possible steric hindrances, echoing real-world challenges in protein engineering.

The challenges we faced with hydrophobin's structural predictions underscore the importance of rigorous experimental validations and iterative design in synthetic biology. Such challenges aren't anomalies but rather real-world intricacies that synthetic biologists often grapple with. Addressing these uncertainties not only aids in refining our designs but also contributes to the broader scientific understanding of complex protein behaviors.

Conclusion and Future Work

---

Our modeling journey reveals the importance of precise protein design and the invaluable insights computational tools can offer. While our streamlined Catcher presents promising results, further experimental validations are paramount.

Future endeavors could focus on in vivo validation, refining the Tag and Catcher system for enhanced interaction, and expanding the applications of this system in synthetic biology.