The discovery of biomolecular condensates—dynamic, membraneless organelles formed via liquid-liquid phase separation (LLPS)—has reshaped our understanding of cellular organization. Many critical drug targets, including those in oncology (ESR1, EGFR) and neurodegeneration, are now known to reside within these unique compartments [2]. This presents both a tremendous opportunity and a formidable challenge. While we have become adept at designing drugs with high binding affinity for their targets, a crucial question has remained largely unanswered: how does a drug efficiently navigate the crowded, complex interior of a cell to find its target within a condensate? This "last-mile problem" represents a significant bottleneck, as traditional affinity-based design often fails to account for the distinct physicochemical microenvironment of the condensate itself, leading to unpredictable drug efficacy.
The journey to address this challenge began with foundational work establishing that condensates are not uniform mixtures but possess heterogeneous internal properties [3]. A pivotal advance came from studies demonstrating that small molecules partition into various condensates based on general physicochemical features, primarily hydrophobicity, rather than specific binding interactions [4]. This confirmed that the condensate microenvironment acts as a selective filter. However, these insights were largely descriptive. They revealed that partitioning was important but did not provide a predictive framework for how to design a drug for a specific condensate-sequestered target. The field needed a unifying principle to connect a drug's molecular properties, the target's condensate environment, and the ultimate therapeutic potency.
A landmark 2025 study in Nature Chemical Biology by researchers at Peking University and Westlake University provides this missing link, proposing a powerful new rule for drug design [1]. By systematically analyzing the properties of thousands of FDA-approved drugs and endogenous metabolites, the team made a striking initial observation: small molecules that interact with phase-separating (PS) proteins are significantly more hydrophobic than those targeting non-PS proteins.
This finding sparked a brilliant hypothesis: if drugs for PS targets are different, perhaps the condensates they target are also different. The researchers classified the intrinsically disordered regions (IDRs) that drive phase separation, identifying distinct clusters based on amino acid composition—notably, nonpolar, hydrophilic, and charged clusters. They then employed a sophisticated combination of optogenetics and fluorescence lifetime imaging microscopy (FLIM) to directly measure the micropolarity inside condensates formed by these different IDRs. The results were unequivocal: condensates driven by nonpolar-residue-enriched IDRs, such as that of the estrogen receptor 1 (ESR1), created a markedly more hydrophobic microenvironment (i.e., had a lower dielectric constant) [1].
This set the stage for the study's central conclusion. Analyzing inhibitors of ESR1, a key breast cancer target known to form hydrophobic condensates, the team found that the inhibitors' potency correlated strongly with their hydrophobicity, but not with their binding affinity. Using in vitro partitioning assays, they demonstrated that more hydrophobic molecules preferentially accumulated within the ESR1 condensate. This elegantly explained the potency data: a drug's hydrophobicity governs its ability to enrich itself within the target's hydrophobic condensate, dramatically increasing its local concentration and thereby boosting its effective potency. The research distills this complex interplay into a simple, powerful principle: "like dissolves like." To effectively target a protein in a hydrophobic condensate, a drug must itself be sufficiently hydrophobic.
The implications of this work are profound, shifting a fundamental paradigm in drug discovery. The condensate is no longer a passive background but an active design parameter. The new formula for success is Potency = Affinity + Partitioning. This principle suggests that for many "difficult-to-drug" targets residing in condensates, the path forward may not be a relentless pursuit of higher affinity but a more nuanced optimization of properties, like hydrophobicity, to match the target's microenvironment.
This opens several exciting future directions. A critical first step in future drug campaigns against PS targets will be to characterize the micropolarity of the relevant condensate. However, challenges remain. The "double-edged sword" of hydrophobicity must be managed carefully, as excessive hydrophobicity can lead to poor solubility and off-target toxicity. Furthermore, native condensates are far more complex than the model systems studied, containing a multitude of proteins and RNA that could influence the microenvironment.
Addressing this complexity and scaling the new design principle will require a new generation of tools. The study's use of machine learning to predict inhibitor efficacy from molecular properties highlights a critical need for large-scale, structured datasets. Future work could leverage AI-native platforms that link high-throughput screening of combinatorial libraries to predictive model training, accelerating the discovery of compounds with optimal partitioning properties. By systematically mapping the design space between molecular features and biological function, we can transition from serendipity to a more rational, scalable, and ultimately more effective era of drug discovery.
Ailurus Bio is a pioneering company building bioprograms, which are genetic codes that act as living software to instruct biology. We develop foundational DNAs and libraries to turn lab-grown cells into living instruments that streamline complex procedures in biological research and production. We offer these bioprograms to scientists and developers worldwide, empowering a diverse spectrum of scientific discovery and applications. Our mission is to make biology a general-purpose technology, as easy to use and accessible as modern computers, by constructing a biocomputer architecture for all.
