Investigating Amino acid Enrichments and Patterns in Phase-Separating Proteins: Understanding Biases in Liquid-Liquid Phase Separation

Abstract

Liquid-Liquid Phase Separation (LLPS) forms membraneless organelles, enhancing biochemical processes. The stickers-and-spacers model explains LLPS but is mainly validated in Prion-like RNA Binding Proteins. We explore peptide motifs in LLPS in broader protein contexts. We developed a computational approach for motif discovery, implemented in 178 Phase-Separating Proteins (PhSePs), complemented by the FuzDrop and CIDER servers, which identified droplet-promoting regions (DPRs) and examined disorder-related characteristics. Our database of PhSePs was analyzed against proteins with low propensity for LLPS. This comparative analysis revealed 129 enriched peptide motifs with folds higher than 0.2, consisting of 3 to 6 residues, with tetrapeptides being the most prevalent. Key features of the enriched motifs included Gly-rich sequences punctuated with aromatic, charged, and polar residues, as well as homopeptide repeats (e.g., GGDR, SRGG, YGGG, QQQQ, PPPP). Analysis of motif presence, frequency, and co-occurrence revealed widely distributed motifs across different DPRs, identified motifs with significant repetitive patterns, and highlighted motif trios that are more likely to co-occur within a sequence. By harvesting this analysis, we developed a data-driven approach for minimalistic peptide design with LLPS propensity, further using the CIDER server for peptide characterization and peptide design refinement. We designed 8 peptides with various motif combinations and amino acid distributions, which were experimentally validated to undergo LLPS, exhibiting liquid-like behavior with diverse molecular mobility patterns and droplet dynamics. Our approach bridges a non-biased computational approach with experimental validation, offering insights into sequence determinants of phase separation, with the potential for designing minimalistic synthetic condensates with tailored properties.