⚛️ biophysics

DIELECTRIC CONTINUUM MODELING FOR THE DE NOVO DESIGNED CALCIUM CHANNEL CALC6.3: STRUCTURE PREPARATION, MESHGENERATION, AND CALCIUM SELECTIVITY

This paper presents an integrated software package that automates the workflow from structure preparation to finite element simulation for the de novo designed calcium channel CalC6.3, successfully reproducing its experimentally observed calcium selectivity and demonstrating that smoother molecular surface meshes enhance solver performance.

Original authors: Eisenberg, R. S., Xie, D.

Published 2026-10-23✓ Author reviewed ⓘ

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Eisenberg, R. S., Xie, D.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Life depends on a constant, controlled flow of electricity, but not the kind that powers a lightbulb. Inside every living cell, tiny charged particles called ions zip across membranes to trigger muscle contractions, release hormones, and send signals between neurons. Among these, calcium ions are particularly crucial, acting as a master switch for many biological processes. However, cells are crowded places filled with various ions, including sodium, which is far more abundant. The challenge for a cell is to let the right amount of calcium in while keeping the flood of sodium out. Nature solves this with specialized protein tunnels called ion channels, which act as highly selective gates. These gates must be precise enough to distinguish between calcium and sodium, even though the two are chemically similar and present in the same solution.

For decades, scientists have tried to understand exactly how these gates work by looking at their physical shapes. While modern imaging tools have revealed the structures of many closed channels, capturing them in the open state—when they are actually letting ions through—has remained elusive. The open state is fleeting and unstable, making it nearly impossible to photograph directly. Without a clear picture of the open gate, researchers have struggled to simulate how ions move through the channel and why calcium is chosen over sodium. This gap in knowledge has left a hole in our understanding of how life manages its electrical signals.

Recently, a team of researchers turned to a different approach: instead of trying to capture a natural channel in the act of opening, they looked at a channel that was designed from scratch. This artificial channel, known as CalC6 3, was built by scientists to be a perfect, open hexagonal tunnel lined with specific chemical groups that attract calcium. Because it was designed to stay open and has been confirmed by high-resolution imaging, it offers a rare, stable model to study the physics of ion selection. The researchers set out to build a complete digital simulation of this channel, not just to see if it worked, but to understand the invisible forces that guide the ions.

To do this, the team created a specialized software package that acts as a bridge between the raw data of the channel's structure and the complex mathematics needed to simulate ion movement. The process began with cleaning up the digital file of the channel's structure. The original data contained thousands of hydrogen atoms that were added by a computer during the design phase, along with some missing heavy atoms. The researchers wrote scripts to strip away the artificial atoms and fill in the gaps, creating a chemically accurate model of the protein. They then placed this model into a virtual box that represented the cell membrane and the surrounding fluid, aligning the channel so its opening faced the correct direction.

The most difficult part of the simulation was creating a digital skin for the protein. The channel has a very narrow, intricate tunnel in the center where ions must pass, surrounded by a complex, bumpy outer surface. If the digital skin is too rough, the computer simulation crashes; if it is smoothed out too much, the narrow tunnel gets distorted, and the simulation loses the very geometry that makes the channel selective. The researchers used a new method to generate a surface mesh that kept the narrow tunnel perfectly intact while smoothing out the rest of the protein's exterior. This allowed them to build a high-quality 3D grid, or mesh, that filled the entire simulation box, dividing the space into tiny tetrahedral shapes that the computer could analyze.

With the digital model ready, the team ran a simulation to see how calcium, sodium, and chloride ions would behave in a solution mimicking the inside of a cell. They used a sophisticated mathematical model that accounts for the size of the ions and the electric charges of the protein, calculating how the ions would distribute themselves around the channel. The results were striking. The simulation showed that calcium ions piled up inside the narrow filter of the channel, reaching concentrations much higher than in the surrounding fluid. In contrast, sodium ions were largely excluded, even though they were present in the same amount and are actually smaller than calcium when surrounded by water.

This finding confirmed that the channel's ability to select calcium is not just about size. The simulation revealed that the specific arrangement of negative charges inside the tunnel, combined with the crowded environment, creates a powerful attraction for the doubly charged calcium ions that sodium cannot match. The researchers also tested two different versions of the digital skin for the protein: one that was rougher and one that was smoother. They found that the smoother version allowed the computer to solve the equations much faster and with greater stability, suggesting that refining the digital representation of the protein can significantly improve the efficiency of these complex simulations.

The entire process, from cleaning the raw data to solving the final equations for a system with over 140,000 variables, took less than a minute on a single computer processor. This speed and accuracy demonstrate that the new software package is a powerful tool for exploring how ion channels work. By successfully simulating the behavior of the de novo designed CalC6 3 channel, the researchers have provided strong evidence that the physical structure and electric charge of the channel are sufficient to explain its ability to pick calcium over sodium. This work offers a clear, computationally verified view of a mechanism that has been difficult to observe directly, bringing scientists one step closer to understanding the fundamental rules that govern the flow of electricity in living things.

Technical Summary: Dielectric Continuum Modeling for the De Novo Designed Calcium Channel CalC6 3

Problem Statement
The study addresses the challenge of computationally modeling calcium selectivity in ion channels using dielectric continuum models. While high-resolution structures of native voltage-gated calcium (CaV) channels exist, experimentally determined structures in the open state are currently unavailable, limiting direct simulation of ion permeation. Although template-based modeling exists, it lacks experimental validation for the open state. Conversely, de novo designed channels like CalC6 3 offer an experimentally resolved, open-pore structure (PDB ID: 9DZW) with demonstrated calcium selectivity. However, applying dielectric continuum models (specifically the nonuniform size-modified Poisson–Boltzmann, or nuSMPBic, model) to CalC6 3 presents significant technical hurdles:

Structural Preparation: The original PDB file contains computationally generated hydrogen atoms and lacks 175 heavy atoms, requiring cleaning and repair.
Mesh Generation: Generating a high-quality triangular molecular surface mesh is difficult because the channel possesses a narrow, complex selectivity filter. Standard surface generation methods often distort the pore geometry to achieve exterior smoothness, which compromises the physical accuracy of the selectivity filter.
Workflow Integration: There is a need for an integrated software package that automates the workflow from raw PDB data to the numerical solution of the electrostatic and ionic concentration equations.

Methodology
The authors developed an integrated software package that streamlines the entire simulation workflow for CalC6 3. The methodology proceeds in four distinct stages:

Structure Preparation: Python scripts were written to process the original 9DZW.pdb file. This involved removing computationally generated hydrogen atoms and the CRYST1 record, translating the structure so the pore axis aligns with the z-axis, and centering the channel at the origin. The repaired structure was then converted to a PQR file using PDB2PQR. This step reconstructed the 175 missing heavy atoms, assigned protonation states at pH 7.0, optimized hydrogen bonding, and assigned atomic charges and radii based on the AMBER force field.
Surface Mesh Generation: The Gaussian Ion Channel Molecular Surface Generation (GICMSG) package was employed to create triangular molecular surface meshes. Unlike previous methods (e.g., TMSmesh) that use a single decay parameter, GICMSG utilizes two spatially varying parameter functions ( $d_p$ and $d_m$ ) to independently control the smoothness of the exterior surface and the geometry of the channel pore. Two cases were tested: Case 1 with default parameters and Case 2 with parameters tuned to produce a smoother exterior surface while preserving the pore geometry.
Volume Mesh Construction: The generated surface meshes were used to construct interface-fitted, unstructured tetrahedral meshes for a simulation box ( $\Omega$ ) containing protein, membrane, and solvent regions. The box dimensions were set to enclose the van der Waals surface of the channel.
Numerical Solution: The integrated package coupled the mesh generation with the nuSMPBic finite element solver. Numerical experiments were conducted for an ionic solution containing 0.1 M NaCl and 0.1 M CaCl2 (bulk concentrations: 0.1 M Ca²⁺, 0.1 M Na⁺, 0.3 M Cl⁻). Hydrated radii were assigned based on the BioNumbers database. The solver computed the electrostatic potential and ionic concentration distributions by solving a nonlinear finite element system.

Key Contributions

Integrated Software Package: The paper presents a unified tool that automates the workflow from raw PDB data (9DZW) to the visualization of electrostatic potentials and ionic concentrations. This reduces manual intervention and standardizes the preparation of de novo designed channels for continuum modeling.
Pore-Preserving Mesh Generation: The application of GICMSG successfully generated high-quality triangular meshes that preserve the intricate geometry of the narrow selectivity filter while maintaining sufficient smoothness on the exterior surface for finite element calculations.
Validation of Continuum Models: The study demonstrates that the nuSMPBic model, when applied to the experimentally resolved CalC6 3 structure, can reproduce the experimentally observed calcium selectivity without requiring ad hoc simplifications of the pore geometry or charge distribution.

Results

Calcium Selectivity: Numerical experiments confirmed preferential accumulation of Ca²⁺ within the narrow selectivity filter (formed by six Glu residues) compared to Na⁺, despite Na⁺ having a smaller hydrated radius. The Ca²⁺ concentration reached approximately 0.8 mol/L near the intracellular entrance of the filter, significantly exceeding its bulk concentration of 0.1 mol/L. Na⁺ was largely excluded from the pore.
Mesh Smoothness and Solver Performance: Comparing Case 1 (rougher exterior surface) and Case 2 (smoother exterior surface) revealed that while both preserved the pore geometry and yielded similar ionic concentration profiles, Case 2 significantly improved solver performance. Case 2 required only 72 iterations to converge compared to 151 for Case 1, and the total computation time decreased from ~49 seconds to ~44 seconds on a single CPU core (Mac Studio, M4 Max). This suggests that smoothing the exterior molecular surface enhances the convergence of the finite element solver without distorting the critical pore physics.
Ion Distribution: The simulations showed distinct distribution patterns: Ca²⁺ accumulated in the pore and near the intracellular entrance; Na⁺ was excluded; and Cl⁻ accumulated primarily near the extracellular entrance and in the bulk solvent, reflecting the combined effects of ionic charge, size, and the channel's electrostatic environment.

Significance
The paper claims that CalC6 3 serves as a valuable model system for investigating calcium selectivity because it provides an experimentally determined, open-pore structure that is simpler than native CaV channels (lacking voltage-sensing domains) yet retains the essential selectivity filter. The work demonstrates that dielectric continuum models, specifically the nuSMPBic approach, are capable of capturing the equilibrium features of calcium selectivity in a realistic structural and dielectric environment. The integrated package provides a computationally efficient tool (solving systems with ~142,000 unknowns in ~44 seconds) for studying ion channels and may serve as a foundation for coarse-grained molecular dynamics treatments where physiological ion concentrations are low. The authors note that while the package is currently applied to CalC6 3, it is designed to be adaptable for simulating other ion channel proteins.

More like this