On Dec. 7th, 2020, Labs of Chao Tang (Peking University), Zhongying Zhao (Hong Kong Baptist University) and Hong Yan (City University of Hong Kong) published a collaborative research article on Nature Communications, entitled Establishment of a morphological atlas of the Caenorhabditis elegans embryo using deep-learning-based 4D segmentation (Link: <a href=""_blank" title="MetInfo CMS">https://www.nature.com/articles/s41467-020-19863-x), which quantitatively constructed a standard morphological atlas of C. elegans embryo (Figure 1).
Figure 1. A conceptual picture.
C. elegans has accurate developmental programs at cellular level, i.e., each cell has reproducible division timing, division orientation, migration trajectory, and identity and fate among individual embryos. Thus, it has been one of the most popular model organisms in developmental biology. In 1983, Sulston J. E. et al. firstly mapped the embryonic cell lineage of C. elegans through a big workload of manual identification. In 2006, Waterston R. H. et al. developed a nucleus-based experimental and computational system for automatic tracing within the whole embryo. However, the nucleus is merely a point inside the cell body, and cannot provide details in the complex 3D morphology. On the other hand, since the cell membrane is distributed across the whole space unlike the discrete distribution of cell nuclei, it’s hard to be identified especially when the cell number is large. Therefore, the morphological atlas is yet to be established.
In this work, we improved a strain that can express nucleus-labeled and membrane-labeled markers precisely and persistently, and imaged 17 embryos from 4- to 350-cell stages. Then, we segmented the fluorescent images of membrane signal using a deep-learning network, and traced the cell lineage with nucleus signal (Figure 2). The segmented regions were subjected to spatial and temporal normalization, generating a set of standardized embryo structures placed in a cuboid framework. Last but not least, we performed full analysis on the cell volume, cell surface area, cell-cell contact relationship, cell shape irregularity, signaling transduction network, etc (Figure 3). Our work on C. elegans standard morphological atlas is expected to enhance researches in developmental biology, cell biology and biomechanics.
Figure 2. Fluorecent imaging and segmentation pipeline.
Figure 3. Standard morphological atlas and multi-dimensional developmental properties.
In this paper, Prof. Chao Tang (PKU), Prof. Zhongying Zhao (HKBU) and Prof. Hong Yan (CityU) are the corresponding authors; Ph.D students Jianfeng Cao (CityU) and Guoye Guan (PKU) and Dr. Vincy Wing Sze Ho (HKBU) are the co-first authors; Ming-Kin Wong and Lu-Yan Chan provided experimental assistance.
