TL;DR
Researchers have developed a new coarse-grained computational model for collagen fibrils using the Martini 3 force field. This model accurately captures the mechanical properties of collagen, including the crucial role of cross-links, enabling simulations of larger and more complex systems than previously possible. This advancement opens doors for understanding age-related changes in tissue mechanics and designing biomimetic materials.
What Problem Does This Paper Address?
Collagen is a fundamental structural protein in mammals, vital for connective tissues like bone and ligaments, and subjected to significant mechanical forces. Its integrity and stiffness are heavily influenced by covalent cross-links. However, studying collagen with its cross-links computationally has been challenging. Fully atomistic (AA) simulations are computationally expensive, limiting the size and duration of studies. Existing coarse-grained (CG) models, while faster, often lack the resolution to accurately represent the triple helical structure and the force response of individual cross-links. This gap hinders a detailed understanding of how collagen's mechanical behavior changes with factors like aging, where cross-link composition and distribution vary.
Methods
The core of this work is the development and validation of a new CG model for collagen fibrils based on the Martini 3 force field. The process involved several key steps:
- Building Blocks: Individual collagen triple helices and specific cross-links (divalent hydroxylysino-5-keto-norleucine (HLKNL) and trivalent pyridinoline (PYD)) were parameterized as separate building blocks.
- Nonbonded Interactions: For cross-links, partition coefficients were determined using nonequilibrium thermodynamic calculations, comparing their distribution between water and a hydrophobic solvent to match AA simulations.
- Bonded Interactions: For both cross-links and the collagen triple helix backbone, bonded parameters (bond lengths, angles, and dihedrals) were fitted using Boltzmann inversion to match AA simulation data. This included a novel dual-potential approach for bond lengths to capture sequence-specific variations in the collagen backbone.
- Gō Model Integration: To preserve the characteristic triple helical structure, a Gō model was implemented, focusing on intrahelical interactions.
- Assembly and Validation: The parameterized building blocks were assembled into microfibrillar structures using the ColBuilder tool. The resulting CG model was extensively validated against experimental data and AA simulations under both equilibrium and nonequilibrium (force-pulling) conditions.
- Force-Extension Dynamics: A key aspect was adapting Martini 3 for out-of-equilibrium simulations. This involved stiffening bond force constants to accurately reproduce the mechanical response under high pulling forces, a departure from standard Martini 3 parameterization.
Datasets
The study primarily relies on data generated from all-atom (AA) molecular dynamics simulations of collagen molecules and fibrils. These AA simulations served as the reference data for parameterizing the CG model. The authors also compared their CG model's results against experimental data for properties like rise per residue and residues per turn. No new experimental datasets were generated for this study.
Key Insights and Ideas
- Bridging the Resolution Gap: The developed CG model successfully bridges the gap between atomistic detail and the mesoscopic scale, allowing for the study of larger collagen structures with reasonable accuracy.
- Crucial Role of Cross-links: The model highlights the critical role of cross-links in stabilizing collagen fibrils and influencing their mechanical properties, a factor often simplified or ignored in previous CG models.
- Martini 3 Adaptability: The study demonstrates the adaptability of the Martini 3 force field for complex biological systems and, importantly, for out-of-equilibrium simulations under mechanical stress, which required specific parameter tuning.
- Bond Stiffness is Key for Force: For simulating collagen under high mechanical load, optimizing the stiffness of bond potentials is more critical than fine-tuning other parameters like the Gō model's well depth.
- Computational Efficiency: The CG model offers a significant speedup (approximately 16-fold) compared to AA simulations, making large-scale investigations feasible.
Key Results
- The developed Martini 3 CG model accurately reproduces the force-extension behavior of collagen triple helices and fibrils observed in AA simulations.
- The model successfully captures the mechanical response of collagen under pulling forces up to 1000 pN.
- Simulations of a 335-nm-long microfibril showed excellent agreement with AA simulations in terms of end-to-end distance under force.
- The CG model slightly overestimates the stiffness of the overlap region relative to the gap region in fibrils, a minor deviation attributed to the loss of fine-grained details in CG models.
- The computational efficiency gain allows for simulations of systems with ~43 million atoms (AA) down to ~6.3 million beads (CG), enabling the study of much larger structures.
Limitations and Future Work
- Helical Shape Stabilization: While a Gō model was implemented, the authors acknowledge that the Martini 3 implementation alone cannot fully maintain the helical shape under very high forces without additional explicit stabilization methods. Future versions might offer new approaches for this.
- Interhelical Interactions: The current force distribution analysis has limitations in fully capturing hydrophobic interactions between collagen chains. The inherent smoothing of free energy landscapes in CG models can also reduce friction between triple helices, potentially leading to enhanced elongation compared to AA systems. Future work could explore more sophisticated ways to characterize these interactions.
- Parameter Trade-offs: The parametrization process involved trade-offs, particularly when balancing adherence to Martini 3 guidelines with reproducing AA data for cross-links. Alternative bead choices might also be suitable.
- Future Directions: The authors suggest exploring more sophisticated parameter balancing approaches. They also envision using this model for investigating age-related changes in tissue mechanics, the effects of advanced glycation end products, and for designing biomimetic materials. The compatibility with backmapping techniques also opens avenues for predicting reactivity with chemical detail.
Summary updated