Scientists have built the most complete and detailed single-cell map of embryo development in any animal to date, using the fruit fly as a model organism.
Published in Science, and co-led by Elaine Furlong of EMBL and Jay Shendor at the University of Washington, the study harnesses data from more than one million embryonic cells covering all stages of embryo development and represents a major advance on multiple levels. This basic research also aids scientists’ ability to pursue questions such as how mutations lead to various developmental defects. In addition, it provides a pathway to understanding the broad non-coding portion of our genome that contains the most disease-associated mutations.
said Eileen Furlong, head of the genomic biology unit at EMBL. “But what I’m really excited about is using deep learning to get an ongoing view of the molecular changes that drive embryonic development — right up to the moment.”
Embryonic development begins with the fertilization of an egg, followed by a series of cell divisions and decisions that result in a highly complex multicellular embryo that can move, eat, sense and interact with its environment. Researchers have been studying this process of embryonic development for more than a hundred years, but it is only in the past decade that new techniques have enabled scientists to identify the molecular changes that accompany cellular transformations at the single-cell level.
These single-cell studies generated tremendous excitement because they demonstrated the complexity of cell types in tissues, even identified new types of cells, and revealed their growth pathways as well as fundamental molecular changes. However, attempts to profile the entire embryo development with a single-cell resolution have been elusive due to numerous technical challenges in sampling, costs and techniques.
In this regard, the fruit fly (Drosophila melanogaster), a prominent model organism in developmental biology, gene regulation and chromatin biology, has some major advantages when it comes to developing new approaches to address this. The embryonic development of Drosophila occurs very quickly; Within just 20 hours after fertilization, all tissues, including the brain, intestine, and heart, have formed so that the organism can crawl and eat. This, combined with the many discoveries made in fruit flies that have prompted an understanding of how genes and their products function, encouraged Furlong’s lab and collaborators to take on this challenge.
“Our goal was to have a continuous view of all stages of embryonic development, to capture all the dynamics and changes as the embryo develops, not only at the RNA level, but also at the controls that regulate this process,” said the co-author. Stefano Secchia, PhD student at Furlong Group.
Initial work with “reinforcers”
In 2018, Furlong and Shendure’s groups demonstrated the feasibility of characterizing “open” chromatin at single-cell resolution in embryos and how these DNA regions often represent active developmental promoters. Enhancers are parts of DNA that act as control switches to turn genes on and off. The data showed which types of cells in the fetus were using which enhancers at a given point in time and how that use changed over time. Such a map is essential to understanding what drives certain aspects of embryonic development.
“I was so excited when I saw those results,” Furlong said. “Going beyond RNA to look upstream at these regulatory switches in single cells was something I didn’t think would be possible for a long time.”
The 2018 study was state of the art at the time, identifying about 20,000 cells in three different windows of embryo development (at the beginning, middle and end). However, this work still only gives snapshots of cellular diversity and organization during specific discrete time points. So the team explored the possibility of using samples from overlapping time windows and, as a proof of principle, applied the concept to a specific strain – the muscle.
This then paved the way for a significant upgrade using the new technology developed in the Shendure Lab. The team’s current work identified open chromatin of nearly one million cells and RNA from half a million cells from overlapping time points that span the entire development of the Drosophila embryo.
Using a type of machine learning, the researchers took advantage of the overlapping of time points to predict time with more precise accuracy. Co-author Diego Calderon, a postdoctoral researcher in Shendure’s lab, trained a neural network to predict the exact development time of each cell.
“Although the samples collected contained embryos of slightly different ages over a time period of 2 to 4 hours, this method allows you to zoom in on any part of this embryogenesis timeline on a scale of minutes,” Calderon said.
Shandour added, “I was amazed at how well this worked. We can capture the molecular changes that happen so quickly in time, within minutes, that previous researchers had detected by handpicking the embryos every three minutes.”
In the future, such an approach would not only save time but could serve as a reference for the normal development of the fetus to see how things might change in different mutant embryos. This can determine exactly when, and in what cell type, the mutation’s phenotype originates, the muscle researchers have shown. In other words, this work not only helps understand how evolution occurs naturally, but also opens the door to understanding how different mutations can spoil it.
The new predictive capabilities that this research portends, based on samples from much larger time windows, can be used as a framework for other model systems. For example, developing mammalian embryo, in vitro cell differentiation, or even after drug treatment in diseased cells, where gaps in sampling times can be designed to facilitate accurate prediction of the optimal time for a finder.
Going forward, the team plans to explore the predictive capabilities of the atlas.
“By bringing together all the new tools available to us in single-cell genomics, computation, and genetic engineering, I’d like to see if we can predict what happens to the fate of individual cells in vivo after a genetic mutation,” Furlong said. “…But we are not there yet. However, prior to this project, I also thought that the current work would not be possible any time soon.”