Capturing chemical reactions inside biomolecular condensates with reactive Martini simulations

Biomolecular condensates, formed through liquid-liquid phase separation (LLPS), are crucial for cellular organization and are hypothesized to have played a significant role in the origin of life. These condensates can act as 'reaction crucibles,' influencing reaction rates and specificities. Experimental investigation of condensate interiors is challenging, making computational methods like molecular dynamics (MD) valuable. Atomistic MD is computationally expensive for large, slow-moving condensates; therefore, coarse-grained (CG) models like Martini are used. The Martini force field maps multiple atoms onto a single bead, allowing efficient simulation of complex systems. A recent extension, reactive Martini, enables the simulation of chemical reactions. This study utilizes reactive Martini to investigate the formation of benzene-1,3-dithiol macrocycles within a peptide-based condensate, aiming to understand how the condensate environment affects reaction kinetics and product distribution. The formation of such macrocycles, and the investigation into their formation in the presence of condensates, is particularly interesting as such molecules may be relevant to the creation of a synthetic form of life and the early emergence of life from inanimate matter.

Significant research highlights the role of biomolecular condensates in organizing biological systems through LLPS. Their involvement in protocell evolution is suggested by their ability to recruit molecules like RNA. Condensates are known to alter enzymatic activity and control cellular redox reactions. Furthermore, chemical reactions, such as phosphorylation, can modify and control condensate formation and dissolution. While experimental studies are limited in probing the dynamics within condensates, computational methods like MD provide valuable molecular-level insights. Coarse-grained models like Martini have been successfully applied to study condensate formation and dynamics. Reactive Martini, an extension of Martini, facilitates the simulation of chemical reactions in complex environments.

The study uses the Martini 3 force field to model the LLPS behavior of a short model peptide, LFssFL. Phase separation at high pH was confirmed computationally and the properties of the condensate, including water content, were characterized. The potential of mean force (PMF) was calculated to quantify the partitioning of benzene-1,3-dithiol into the condensate phase. This computational partitioning was validated experimentally using an assay measuring the partition coefficient of the molecules between the condensate and supernatant phases. Reactive Martini simulations were performed to model the formation of macrocycles from benzene-1,3-dithiol within the LFssFL condensate. Simulations were conducted with and without condensates to compare macrocycle size distributions and reaction rates. The effects of reactant concentration and the possibility of reversible disulfide bond formation were investigated through additional simulations. Simulations of systems where the condensate and macrocycle formation occurred concurrently were performed and compared against simulations where the reactants were pre-placed within the condensate. Experiments were performed to measure the partitioning of benzene-1,3-dithiol monomers and their resulting macrocycles into the LFssFL condensates. These experimental partition coefficients were then compared to the values obtained computationally from the PMF calculations. Simulations were performed using Gromacs 2018.8. Specific parameters for the LFssFL peptide, benzene-1,3-dithiol, and the reactive interactions were defined based on previous work and adjustments to the reactive potential were made to study the effects of reversible reactions. The AWH method was used to calculate PMFs. Experimental partitioning was measured using UPLC to determine concentrations in supernatant and condensate phases after centrifugation.

The Martini force field accurately reproduced the pH-dependent LLPS of the LFssFL peptide. Benzene-1,3-dithiol showed a strong preference for partitioning into the condensate, confirmed by both simulation (PMF calculations) and experiment (partition coefficient measurements). Macrocycle formation within the condensate resulted in a shift toward larger ring sizes compared to aqueous environments. This shift was primarily attributed to the increased local concentration of reactants within the condensate, although a small effect of the peptides themselves was observed in simulations. Simulations where phase separation and reaction occurred concurrently indicated that the rate of monomer consumption (reaction rate) was coupled to the rate of peptide phase separation. These findings suggest that the condensate-forming peptides act as chaperones, aiding the recruitment of reactants into the condensate, thereby increasing the effective concentration and enhancing reaction rates. The study also demonstrated that the reaction products were sensitive to condensate characteristics such as water content, and that reversible reactions could be simulated by adjusting the reactive potential parameters. Additional experiments showed that larger macrocycles also preferentially partitioned into the condensate phase, and a similar trend of increased macrocycle size was observed in a second peptide-based condensate system (WGR-1).

This study demonstrates that reactive Martini simulations can effectively probe chemical reactions within biomolecular condensates. The observed shift in macrocycle size distribution and increased reaction rates highlight the significant influence of the condensate environment on reaction outcomes. The finding that the rate of reaction is coupled to condensate formation suggests a chaperone-like role for the peptides, enhancing reactant concentration and accelerating reactions within the condensate. The sensitivity of reaction products to condensate properties such as water content emphasizes the importance of considering the specific microenvironment in studying reactions within these complex systems. The use of a coarse-grained model enables the simulation of larger systems and longer timescales compared to atomistic models, leading to valuable insights into the role of biomolecular condensates in cellular processes and early life.

This research successfully utilized reactive Martini simulations to investigate chemical reactions within biomolecular condensates, revealing the importance of condensate microenvironment on reaction kinetics and product distribution. The findings support the notion of condensates as reaction hubs, influencing both reaction rates and product diversity. Future work could explore a wider range of reactions, condensate systems, and reaction conditions. Improvements to the reactive Martini framework, including more efficient implementations and the incorporation of reversible reactions and pH dependence, are also key areas for future development.

The current study focuses on a specific type of reaction and condensate system, limiting the generalizability of the findings. The timescale accessible with the current reactive Martini implementation in Gromacs might restrict the ability to fully capture the interplay between slow phase separation and faster chemical reactions. Further optimization of the reactive potential might also be needed to accurately reflect the formation of all benzenedithiol-based macrocycles. The model’s simplification of pH effects could be improved with further development of the titratable Martini model.