Biomolecular condensates formed via liquid–liquid phase separation (LLPS) are dynamic assemblies composed primarily of proteins and nucleic acids, where nucleic acids act as natural polyelectrolytes driving key cellular functions. In recent years, advances in protein engineering have shifted the paradigm from simply characterizing these natural systems to actively controlling and replicating their behaviors using synthetic mimetics. In our study, we integrated coarse-grained molecular dynamics simulations with a computational de novo protein design approach to investigate condensate formation between engineered globular proteins and polyelectrolytes. Central to our methodology is an in-house developed probabilistic algorithm that leverages fundamental physical chemistry principles to precisely dictate the placement and distribution of charged residues within the protein sequence without disrupting folding structures. Unlike recent machine learning based methods such as AlphaFold3, our algorithm assigns a probability for each amino acid at every position, enabling detailed exploration of both natural and unnatural protein designs while requiring significantly fewer computational resources. This targeted, probability-based design greatly reduces interference from irrelevant residues, allowing us to isolate and assess the impact of specific noncovalent interactions that drive protein–polyelectrolyte associations. This approach not only enhances the efficiency and precision of protein engineering for condensate formation but also establishes a robust framework for tailoring protein properties to meet specific functional demands. By systematically varying the protein charge states, we were able to investigate their direct effects on coacervation and dynamic behaviors. Our simulations revealed that fine-tuning the charge state not only alters the phase separation threshold but also modulates the material properties and dynamics of the resulting condensates. These observations allowed us to identify the primary molecular interactions and driving forces behind phase separation in these complex systems. The insights gained offer robust design rules for engineering proteins with enhanced coacervation probabilities, paving the way for precise computational modulation of biomolecular condensates. The ability to design condensates with predetermined dynamic and structural features has profound implications for applications in drug delivery, synthetic organelles, and responsive biomaterials.
