Computational modeling reveals that a combination of chemotaxis and differential adhesion leads to robust cell sorting during tissue patterning

Robust tissue patterning is crucial to many processes during development. The "French Flag" model of patterning, whereby naïve cells in a gradient of diffusible morphogen signal adopt different fates due to exposure to different amounts of morphogen concentration, has been the most widely proposed model for tissue patterning. However, recently, using time-lapse experiments, cell sorting has been found to be an alternative model for tissue patterning in the zebrafish neural tube. But it remains unclear what the sorting mechanism is. In this article, we used computational modeling to show that two mechanisms, chemotaxis and differential adhesion, are needed for robust cell sorting. We assessed the performance of each of the two mechanisms by quantifying the fraction of correct sorting, the fraction of stable clusters formed after correct sorting, the time needed to achieve correct sorting, and the size variations of the cells having different fates. We found that chemotaxis and differential adhesion confer different advantages to the sorting process. Chemotaxis leads to high fraction of correct sorting as individual cells will either migrate towards or away from the source depending on its cell type. However after the cells have sorted correctly, there is no interaction among cells of the same type to stabilize the sorted boundaries, leading to cell clusters that are unstable. On the other hand, differential adhesion results in low fraction of correct clusters that are more stable. In the absence of morphogen gradient noise, a combination of both chemotaxis and differential adhesion yields cell sorting that is both accurate and robust. However, in the presence of gradient noise, the simple combination of chemotaxis and differential adhesion is insufficient for cell sorting; instead, chemotaxis coupled with delayed differential adhesion is required to yield optimal sorting.

稳健的组织模式形成对发育过程中的诸多环节至关重要。其中，“法国旗模型（French Flag model）”作为组织模式形成领域最被广泛认可的假说，其核心机制为：处于可扩散形态发生素（morphogen）信号梯度中的未特化细胞（naive cells），因暴露于不同浓度的形态发生素而获得不同的细胞命运。不过近期通过延时实验（time-lapse experiments）研究，人们在斑马鱼（zebrafish）神经管（neural tube）的组织模式形成中发现了细胞分选这一替代模型，但该分选过程的具体机制仍不明晰。本文中，我们通过计算建模（computational modeling）证实，稳健的细胞分选需要趋化作用（chemotaxis）与差异黏附（differential adhesion）两种机制共同参与。我们通过量化正确分选比例、正确分选后形成的稳定细胞团比例、实现正确分选所需的时间，以及不同命运细胞的大小变异程度，评估了两种机制各自的分选性能。研究发现，趋化作用与差异黏附对分选过程各有独特优势：趋化作用可实现较高的正确分选比例，因为单个细胞会根据自身细胞类型向信号源迁移或远离信号源；但细胞完成正确分选后，同类型细胞之间缺乏稳定分选边界的相互作用，因此形成的细胞团并不稳定。反观差异黏附，尽管其形成的正确细胞团稳定性更强，但正确分选比例较低。在无形态发生素梯度噪声的情况下，趋化作用与差异黏附的联合作用可实现既精准又稳健的细胞分选；但当存在梯度噪声时，单纯的趋化-差异黏附联合作用无法满足细胞分选需求，此时需要趋化作用与延迟启动的差异黏附相结合，才能实现最优的分选效果。