Two HHMI teams rebuild the family trees of millions of cells in developing mice

A mouse goes from one fertilized egg to hundreds of millions of cells in a few weeks ^[3]. The new trees span millions of cells, and the phys.org account calls them the most complete lineage maps yet made for a mammal ^[1]. It does not say what share of each embryo was sequenced, or how often a reconstructed branch matches a division that actually happened.

The roundworm could be mapped because it can be watched ^[5]. Its embryo is transparent, so researchers saw every cell divide under a microscope. A mouse embryo grows inside its mother and cannot be watched in real time over long periods ^[5]. Both labs had the cells keep the record instead. Each division adds a small, permanent mark to the genome. The marks are inherited, so the record can be read long after the divisions occurred ^[6]. "The cell divides and each of the sisters gets a mark, and those are inherited by their daughters, and they get additional marks, and so on and so forth," Weissman said. "And so, by looking at the end at the marks in this DNA, we're able to reconstruct what this relationship is." ^[7]

Reading the marks at the end is what separates this from the worm work ^[5]. "Most biological measurements are based on either live imaging, which is limited by the fact that most animal tissues are not transparent, or genomics, which is destructive and only measures a single time point," Shendure said ^[13]. "Recording techniques like the ones in these studies enable measurements over time, including in settings that we can't directly visualize." ^[14] The worm's lineage was observed one division at a time. The mouse lineages are inferred from marks found in sequenced cells ^[8].

The two labs also started recording at different points. Weissman's group at the Whitehead Institute engineered stem cells to carry the marks and injected them into an embryo, and the marks ended up in nearly every cell the embryo produced ^[12]. Shendure's group at the University of Washington injected the parts of DNA Typewriter straight into a fertilized egg, and the marks built up as its descendants divided ^[10]. In my view that difference is an asset. Where the two methods cover the same tissues, one tree can be checked against the other. I'd want to see that comparison before calling either map complete.

Weissman describes the result in measured terms. "It's really proof of principle that we can do what was done with the roundworm in 1983, but for mammals like you and me," he said ^[4]. The animals here were mice ^[1]. I think proof of principle is the claim the evidence earns. The recorders work in a living mammal at the scale of millions of cells, and the phys.org account says the teams have worked out how to build a mammalian fate map, a goal that had been out of reach ^[2].