Design

 Conceptual model

First, we created a conceptual model to achieve our goal. As shown in the diagram, we hypothesized that BASE and TARGET could be connected by using three types of molecules. These include one that binds to BASE, one that binds to TARGET, and one that links them together. The molecules do not polymerize in solution. When the molecules bind to BASE or TARGET, their structure changes, enabling the other end to bind to additional molecules. Repeating this process creates a path formed by polymerized molecules, which eventually connects BASE and TARGET. To replicate this system, we identified five essential functions that the molecules need to possess. 

To evaluate whether the conceptual model and the molecules with the necessary functions we proposed can achieve the intended goal, we created a simple simulation. In the simulation, 

Red circle : BASE, Blue circle : TARGET, Red rod : a molecule extending from BASE, Blue rod : a molecule extending from TARGET

To replicate the process where molecules floating in solution do not bind to each other, but bind when a molecule approaches BASE or TARGET, and subsequently bind to the tip of the already-bound molecule, repeating this process to form a path of polymers connecting BASE and TARGET, the following rules were established:

• Sticks floating across the screen are not displayed.

• Sticks appear from red circles and blue circles at a certain probability.

• New sticks appear from the tips of sticks bound to red or blue circles, within a specific angle range, and at a certain probability.

• When sticks extending from red and blue circles overlap, the entire stick turn green.

Furthermore, to connect BASE and TARGET through a more direct path, an additional setting was introduced:

• Sticks bound to red or blue circles are deleted at a certain probability.

Using the above settings, we created a simulation to confirm that sticks extending from the red and blue circles overlap and turn green.

View the code and movies tested under different conditions 

This video shows the execution of the simulation we made. In the simulation, fibers extend from the red circle and the blue circle. We observed that these fibers overlap and turn green, indicating successful connection. By modifying the probabilities of rod generation and deletion, as well as the angular range for rod creation, we were able to control the growth rate of the fibers and the number of overlapping green rods. 

The results confirmed that if we design molecules with the necessary functional properties, it is possible to connect the BASE and TARGET efficiently. This validation confirmed confidence in the  functionality of our approach.

To replicate this model at the molecular level, we developed a  new protein design based on actin, a molecule known for its ability to polymerize into fibers. Our design aimed to incorporate specific features to ensure the desired behavior. 

First, we focused on preventing unwanted polymerization in solution. For this, we utilized actin and thymosin, a molecule that inhibits actin polymerization, to design a protein that prevents actin from spontaneously assembling in the solution. This step was critical to maintain control over polymerization.

Next, we worked on inducing controlled polymerization. We incorporated an actin polymerization-promoting factor into the designed protein to displace thymosin from actin, thereby initiating the polymerization process. This activation would result in the formation of actin fibers capable of effectively connecting the base and target.

Using this approach, we successfully designed molecules with the following essential functions:

1. Actin polymerizes into actin fibers: Promoting the formation of fiber-like structures.  

2. No polymerization in solution: Ensuring that polymerization occurs only under controlled conditions.  

3. Un-polymerizable actin becomes polymerizable and polymerizes into actin fibers: Activating polymerization when necessary.

This step-by-step design process demonstrates our ability to replicate the desired molecular functions, paving the way for achieving our objectives efficiently.

Overview of the design

We explored the specific functional features needed to integrate the conceptual model into the practical system.

The Bidirectional Nano Grabbers is a system where Actin polymerizes into fiber-like structures from both the TARGET and BASE, eventually connecting into a single chain. The system is conceptualized to draw two distant objects closer, focusing on the connection of these objects (base and target) through fiber-like Actin. By linking multiple fiber-like Actin filaments and shortening their length, the system reduces the distance between TARGET and BASE.

Actin required for polymerization, which is central to the reaction system, is initially inhibited from polymerizing in solution by binding to Modified Thymosin. Initially, Thymosin β4, which binds to Actin, exhibited a weak binding region. To address this, modifications were made, leading to the development of the amino acid sequences with stronger binding in its middle to posterior regions. For the Modified Thymosin, we designed a molecule by combining the strong binding region of Thymosinβ4 with an amino acid sequence obtained through simulations, creating a Modified Thymosin with overall stronger binding. Modified Thymosin ensures that Actin remains as individual molecules in solution, without spontaneous polymerization.

To initiate polymerization, a Polymerization Activator is required. This activator must specifically bind strongly to Modified Thymosin while maintaining weak interactions with Actin and other molecules. Additional details about Polymerization Activator can be found on the Future page.

Design process

We followed a systematic process to implement the design outlined above, ensuring that each component of the system was thoughtfully constructed according to the necessary functions. To organize and streamline the design process, we used a flowchart (as shown in the accompanying figure) that broke down the system into its functional elements.

• Step 1: Identifying Functional Requirements

The first step involved identifying the specific requirements that each function needed to fulfill. These requirements were distilled to their core principles, aiming for simplicity in the conceptualization of each function. This ensured a clear and focused design objective for every system function.

• Step 2: Hypothesizing Molecular Properties

Next, we hypothesized how each functional requirement could be met by specific molecular properties. For this, we focused on identifying molecules with simple structures that were widely available, preferably ones naturally found in biological systems. A thorough review of relevant literature provided a list of candidate molecules. These were evaluated against criteria such as simplicity, ubiquity, and compatibility with the system's requirements to select the most suitable candidates.

• Step 3: Simulating Molecular Functions

We tested whether the selected molecules exhibited the desired functions through simulations. The simulation software we used allowed us to:

- Modify existing molecular structures to introduce new features.  

- Predict interactions between molecular structures and sequences.  

To reduce probabilistic variations, simulations with stochastic outcomes were repeated multiple times to ensure reliable results. The simulation results were evaluated not only through 3D models and automatically generated graphs but also by examining output values to ensure they met the desired conditions. 

• Step 4: Refining Molecules and Selecting Additional Components

Once a molecule was confirmed to fulfill its intended function, we applied the same process to design or select additional molecules required for other system functions. This iterative approach ensured that all components were compatible and worked together seamlessly within the system.

By following these steps, we systematically designed and evaluated each functional element of the system, ensuring a robust and cohesive implementation process.

Further details about these processes are available on the Simulation page.

Actin with a Polymerization Inhibitor

One of the most important functions of the system is to establish a connection between the two points, TARGET and BASE. It is assumed that the spatial relationship between TARGET and BASE is not fixed and changes over time. To meet this requirement, we determined that the connection must maintain a fiber-like structure with flexibility and a certain degree of strength.

For the molecules used in this connecting function, we first selected those that can polymerize into fiber-like structures. Several candidates, including DNA and fiber-like proteins, were identified, and their characteristics were compared. The ultimate and complete system we aim to build is highly complex, requiring sufficient design freedom not only for the system's structure but also for interactions between functions. Therefore, we decided to implement the connecting function using proteins, which offer multiple functionalities and sufficient scalability for the system’s functionalities.

Actin bound to Modified Thymosin, which is the core of the system, was designed using RFdiffusion for the Modified Thymosin component. The structural predictions and simulations for its binding with Actin were performed using AlphaFold3 to evaluate structural reliability. RFdiffusion is a software tool capable of generating new proteins in arbitrary parts of an existing protein and evaluating amino acid sequences and structures. AlphaFold3 is a software tool that evaluates the structural reliability of DNA or protein structures for each amino acid sequence. Both tools enable more freedom and efficiency in protein design.

Actin (PDB:1J6Z) has the property of polymerizing into fibers, making it suitable for connecting the TARGET and BASE. While Actin polymerizes into fibers in solution, this system requires Actin to polymerize only when bound to the TARGET or BASE. To address this, Actin is bound to Modified thymosin in solution to prevent natural polymerization. For more details about the Modified thymosin, refer to the Modified Thymosin section below. Additionally, when polymerization needs to be initiated, the system is Modified so that the addition of a polymerization activator detaches the actin from the Modified Thymosin, allowing polymerization to begin. For more details about the Polymerization Activator.

Modified Thymosin

To prevent Actin from polymerizing spontaneously while existing as individual molecules in solution, we decided to use a Polymerization inhibitor to control polymerization. Thymosin β4 was chosen as the Polymerization inhibitor. Simulations revealed that there were weak binding regions in Thymosin when combined with Actin. To address this, we aimed to strengthen the binding of Thymosin β4 by modifying its amino acid sequence, designing a Modified Thymosin with stable and stronger binding properties.

To strengthen the binding of Thymosin β4, we used RFdiffusion to generate new backbones and sequences, and performed structural prediction simulations using AlphaFold3 (details are provided in the page on Modification of Thymosin β4 by RFdiffusion). Based on AlphaFold3’s structural predictions, we selected Thymosin β4 candidates with strong binding and identified the optimal amino acid sequence.

Analysis with PyMOL revealed that this amino acid sequence contained hydrophobic amino acids in the middle and posterior regions. Similarly, an investigation of Thymosin β4 using PyMOL and literature review showed that its N-terminal region also contained hydrophobic amino acids. We hypothesized that these hydrophobic amino acids play a significant role in the binding between Actin and Thymosin β4.

To test this hypothesis, we designed Modified thymosin by combining the N-terminal region of Thymosin β4 with the hydrophobic regions of the amino acid sequence. This Modified Thymosin was further optimized using AlphaFold3 to identify the best configuration.

The Modified Thymosin binds to Actin by wrapping around it, thereby inhibiting polymerization. When Actin exists freely in solution, it binds to Modified Thymosin to prevent spontaneous polymerization. When polymerization is required, the introduction of a Polymerization Activator detaches Modified Thymosin from Actin, allowing polymerization to occur.

Actin + Modified Thymosin

REFERENCE

[1] Roberto Dominguez & Kenneth C. Holmes (2011) "Actin Structure and Function" Annu. Rev. Biophys. 40:169–86

[2] Rustam Ali, Jacob A. Zahm & Michael K. Rosen (2022) "Bound nucleotide can control the dynamic architecture of monomeric actin" Nature Structural & Molecular Biology volume 29, pages320–328

[3] Pekka Lappalainen, Tommi Kotila, Antoine Jégou & Guillaume Romet-Lemonne (2022) "Biochemical and mechanical regulation of actin dynamics" Nature Reviews Molecular Cell Biology volume 23, pages836–852

[4] Oda, T. (2009) "アクチンフィラメントの構造と動態：アクチン分子の重合にともなう構造変化" Biophys. 49:238-241

[5] Oda, T. (2023) "アクチンのとりうるコンフォメーション " Biophys. 63:138-142

[6] Bruce Alberts, Alexander Johnson, Julian Lewis, David Morgan, Martin Raff, Keith Roberts & Peter Walter (2014)"Molecular biology of the cell sixth edition" Garland Publishing Inc.

[7] Tsuyoshi Mashima, Masaru Yamanaka, Atsuki Yoshida, Naoya Kobayashi, Yui Kanaoka, Takayuki Uchihashi & Shun Hirota (2024) "Construction of ligand-binding controlled hemoprotein assemblies utilizing 3D domain swapping"  Chem. Commun., 60:9440-9443

[8] Shun Hirota, Tsuyoshi Mashima & Naoya Kobayashi (2021) "Use of 3D domain swapping in constructing supramolecular metalloproteins" Chem. Commun., 57:12074-12086

[9] Dani Permana, Herlian Eriska Putra & Djaenudin Djaenudin (2022) "Designed protein multimerization and polymerization for functionalization of proteins" Biotechnology Letters 44:341–365

[10] Jessica A. Cross, William M. Dawson, Shivam R. Shukla, Johannes F. Weijman, Judith Mantell, Mark P. Dodding & Derek N. Woolfson (2024) "A de novo designed coiled coil-based switch regulates the microtubule motor kinesin-1" Nature Chemical Biology, 20:916–923

[11] Elizabeth H. C. Bromley, Richard B. Sessions, Andrew R. Thomson, Derek N. Woolfson (2009) "Designed α-Helical Tectons for Constructing Multicomponent Synthetic Biological Systems" J. Am. Chem. Soc., 131: 928–930