Liquid-liquid phase separation (LLPS) of putative assembly scaffolds has been proposed to drive the biogenesis of membraneless compartments. LLPS scaffolds are usually identified through in vitro LLPS assays with single macromolecules (homotypic), but the predictive value of these assays remains poorly characterized. Here, we apply a strategy to evaluate the robustness of homotypic LLPS assays. When applied to the chromosomal passenger complex (CPC), which undergoes LLPS in vitro and localizes to centromeres to promote chromosome biorientation, LLPS propensity in vitro emerged as an unreliable predictor of subcellular localization. In vitro CPC LLPS in aqueous buffers was enhanced by commonly used crowding agents. Conversely, diluted cytomimetic media dissolved condensates of the CPC and of several other proteins. We also show that centromeres do not seem to nucleate LLPS, nor do they promote local, spatially restrained LLPS of the CPC. Our strategy can be adapted to purported LLPS scaffolds of other membraneless compartments.

已提出假定的组装支架的液-液相分离（LLPS）来驱动无膜室的生物发生。 LLPS 支架通常通过单大分子（同型）的体外LLPS 测定来鉴定，但这些测定的预测价值仍然缺乏表征。在这里，我们应用一种策略来评估同型 LLPS 测定的稳健性。当应用于在体外经历 LLPS并定位到着丝粒以促进染色体生物定向的染色体乘客复合物 (CPC) 时，体外LLPS 倾向成为亚细胞定位的不可靠预测因子。常用的拥挤剂可增强水性缓冲液中的体外CPC LLPS。相反，稀释的细胞模拟介质溶解了 CPC 和其他几种蛋白质的冷凝物。我们还表明，着丝粒似乎不会使 LLPS 成核，也不会促进 CPC 的局部、空间限制的 LLPS。我们的策略可以适应其他无膜室的 LLPS 支架。