There’s a lot going on as a single fertilized egg cell transforms into a full-blown organism. Cells are dividing, moving and committing to specific roles; tissues and organs are forming in precise places at precise times; and molecular signals are flying everywhere.
“Fundamentally, it’s a deeply multi-scale problem,” says Ottoline Leyser, a developmental biologist with the Sainsbury Laboratory at the University of Cambridge in the UK. “I like to tell my classes that everybody used to be one cell, and maybe one of the greatest achievements of our lives is getting to the point where we’re a whole organism.”
Leyser is one of the leaders of a growing movement in developmental biology that seeks to tackle that complexity head-on by studying the developing organism as an integrated whole. “Systems developmental biology,” as it’s coming to be known, is just starting to yield results. But it offers the promise of a richer, more holistic understanding of the intricacies of development.
In some ways, this new systems biology actually represents a return to the old ways. Decades ago, long before anyone knew much about genes, biologists studied development by describing the physical changes they saw under their microscopes — but they also used mathematical modeling to try to uncover the control mechanisms that could give rise to those patterns. No less a luminary than the mathematician Alan Turing, more famous for his exploits in code-breaking and artificial intelligence, published a theory back in 1952 explaining how regular patterns such as stripes and spots could result from the interaction between two competing signals.
But then the genetic revolution came along, and suddenly everyone focused on identifying the genes that code for key developmental signals. Today, researchers know almost all the important developmental genes, at least for a few well-studied organisms like fruit flies and mice, giving them what is essentially a parts list for building an organism.
“We spent a very exciting couple of decades collecting up the parts,” says Leyser. “We’re now looping back to the original ideas.”
Beyond the intuitive
Though scientists are learning much about the roles that individual genes play, systems-level modeling of development can help them go deeper and discover processes that can’t be fathomed by studying the behavior of single genes or cells. “It’s the working together of the parts that makes these processes,” says Veronica Grieneisen, a systems biologist at Cardiff University in Wales.
As one simple example, she points to situations where a group of cells migrate en masse in a single direction. Generally, biologists had assumed that cells at the front must be leading the migration in response to an attractive signal — but modeling showed that neither leaders nor signals were necessary: The same behavior could result from a gradual synchronization of random movements of cells in a confined space. “It’s like putting many blindfolded people in a room and saying, ‘Move.’ Because they are bumping into each other, they start coordinating their movement,” Grieneisen says.
Non-intuitive results like these are common in development, largely because of feedback loops where cell types or sets of genes help regulate their own behavior. “If gene A interacts with gene B, gene B interacts with gene C and gene C interacts with gene A, that can generate some complex behaviors. Actually figuring out what the consequences would be is difficult,” says James Briscoe, a developmental biologist at the Francis Crick Institute in London. That’s where computational modeling can help.
Developmental biologists are just starting to crack this nut, and one of the best-worked examples so far comes from Grieneisen’s work on plant roots. All the tissue types of a mature root arise from precisely orchestrated divisions of stem cells near the root’s tip, including a key cell division that generates two different tissue types. “It’s really important that it makes that division, but only at that place,” Grieneisen says. It turns out that the division is regulated by multiple signals whose interaction is too complex to understand intuitively. But when the team simulated the signaling network mathematically, they found that it formed a toggle switch that would turn on the cell division at the crucial point, then switch it off again to prevent the same division from occurring elsewhere.
Biologists hope that they will eventually identify many such switches, oscillators and other components that recur in a variety of developmental contexts. This would allow them to build a library of developmental subroutines that may apply across a wide diversity of organisms even if those organisms are using different molecular signals to make the subroutines hum. “It’s a shift away from thinking about the individual parts that are doing the job, toward the job they are doing,” Leyser says.
Watching everything at once
The ultimate application of systems thinking, of course, would be to track everything that happens in a developing embryo and see how it all works together to create a fully formed creature. And the first steps toward that future are already underway. Allon Klein, a systems biologist at Harvard Medical School, is using a technique called single-cell transcriptomics to identify all the genes active in thousands of individual cells in embryos of the frog Xenopus tropicalis. These patterns of gene activation help identify when cells make the choice to turn into a particular cell type — and Klein’s findings reveal that many of those decisions happen sooner, but also more reversibly, than researchers had realized.
Klein’s technique gives a whole-embryo snapshot of what every cell is doing at a single point in time. By combining snapshots, he can see how this changes during development. “Now we can look at what a whole embryo does at one go,” he says. “Starting to think about how to put all that together — and even looking at the data — is a big challenge.”
But as researchers learn to deal with that data deluge, they can begin to ask which cells in an embryo are producing particular molecular signals, and which cells have receptors that let them receive those signals. “You can peek in on the chatter between the cells,” Klein says. His team is also adding molecular “bar codes” to individual cells that allow them to trace the cells’ lineages as they divide, from one snapshot to the next.
The biggest payoff is likely to come from a merger between this big-data experimental measurement and the theoretical insights that come from quantitative modeling, such as Grieneisen’s work in plant roots. Together, the two form an especially powerful partnership, says Scott Fraser, a systems developmental biologist at the University of Southern California who coauthored a paper on this synthesis in the 2019 Annual Review of Biomedical Data Science.
By treating the developing embryo as an integrated whole, developmental biologists will be able to ask a new sort of question, Fraser says. In the past, researchers looked at a molecular signal in isolation and asked what effect it had on development. But the better question, says Fraser, is how important that effect may be, compared to all the other things going on in the embryo at that moment. Understanding this can help biologists unravel tough problems such as why many genetic mutations cause problems in some people who carry them but leave others relatively unscathed.
“Knowing that something matters is important, but knowing how much it matters is much more important,” Fraser says. “That’s the most exciting thing right now.”