Autoantibodies in systemic autoimmunity are directed against only ~5% of the proteome. The purpose of this study was to assess whether the properties of assemblages (also known as Membraneless Organelles and Biological Condensates) and their protein constituents partly explain the immunological selectivity of autoimmunity. Assemblages arise from phase separation of their protein components, akin to partitioning of oil droplets in water. We obtained from a prediction algorithm (Vernon et al., elife7, 2018) the propensity scores (PScores), i.e., likelihood, for phase separation of autoantigens and non-autoantigens. We then compared autoantigens with the highest PScores to identify shared structural properties. The mean PScores for autoantigens (n = 1050) and the entire human proteome of non-autoantigens (n = 17,532) were 1.46 and 1.09 (p = 1.2E-08). To varying extents, the 25 autoantigens with the highest phase separation propensities shared additional features such as compositional bias, repeated domains, coiled coil regions, nucleic acid binding, and disorder. Most of these properties were present with greater frequencies than their frequencies in the non-autoantigens. We conclude that, on average, autoantigens have a higher predisposition to undergo phase separation, thus, they are more likely to exist in assemblages compared with the average non-autoantigen. We suggest that assemblage formation and the greater than average presence of certain structural features are key factors in selection of a portion of the autoimmune repertoire. Other properties of assemblage proteins, such as high concentration and tendency to form novel complexes with other proteins, may partially explain why assemblages are potent sources of autoantigens.

中文翻译：

---

通过相分离形成的组合中的蛋白质具有促进其转化为自身抗原的特性：对自身免疫的影响。

全身性自身免疫中的自身抗体仅针对约 5% 的蛋白质组。本研究的目的是评估组合物（也称为无膜细胞器和生物凝聚物）及其蛋白质成分的特性是否部分解释了自身免疫的免疫选择性。组合产生于其蛋白质成分的相分离，类似于油滴在水中的分配。我们从预测算法 (Vernon et al., elife7, 2018) 中获得了自身抗原和非自身抗原相分离的倾向得分 (PScores)，即可能性。然后，我们将自身抗原与最高 PScore 进行比较，以确定共享的结构特性。自身抗原 (n = 1050) 和非自身抗原的整个人类蛋白质组 (n = 17,532) 的平均 PScores 为 1.46 和 1.09 (p = 1.2E-08)。在不同程度上，具有最高相分离倾向的 25 种自身抗原具有其他特征，例如组成偏差、重复结构域、卷曲螺旋区域、核酸结合和紊乱。大多数这些特性以比它们在非自身抗原中的频率更高的频率存在。我们得出的结论是，平均而言，自身抗原具有更高的发生相分离的倾向，因此，与平均非自身抗原相比，它们更有可能以组合形式存在。我们认为组合形成和某些结构特征的大于平均存在是选择部分自身免疫组库的关键因素。组合蛋白的其他特性，例如高浓度和与其他蛋白质形成新复合物的倾向，