In June, 100 fruit fly researchers gathered on the Greek island of Crete for their biennial meeting. Among them was Cassandra Extavour, a Canadian geneticist at Harvard University. Her lab works with fruit flies to study evolution and development – “evo devo.” Most often, such researchers choose as their “model organism” the species Drosophila melanogaster – a winged workhorse that has served as an insect collaborator on at least a few Nobel prizes in physiology and medicine.
But Dr. Extavour is also known for cultivating alternative species as model organisms. She is particularly keen on crickets, particularly Gryllus bimaculatus, the two-spotted field cricket, although it does not yet enjoy anything close to the fruit fly’s entourage. (Approximately 250 principal investigators had applied to attend the meeting in Crete.)
“It’s crazy,” she said during a video interview from her hotel room, swatting away a beetle. “If we tried to have a meeting with all the lab heads working on that species of cricket, there might be five of us, or 10.”
Crickets have already been enlisted in studies on circadian clocks, limb regeneration, learning, memory; they have served as disease models and pharmaceutical factories. Veritable polymaths, crickets! They are also increasingly popular as food, chocolate covered or not. From an evolutionary perspective, crickets provide several opportunities to learn about the last common insect ancestor; they have more features in common with other insects than fruit flies do. (Remarkably, insects make up more than 85 percent of animal species).
Dr. Extavour’s research aims at the basics: How do embryos work? And what might it reveal about how the first animal came to be? Every animal embryo follows a similar journey: One cell becomes many, then they arrange themselves in a layer at the egg’s surface, providing an early blueprint for all adult body parts. But how do embryonic cells—cells that have the same genome but don’t all do the same with that information—know where to go and what to do?
“That’s the mystery to me,” Dr. Extavour said. “It’s always where I want to go.”
Seth Donoughe, a biologist and computer scientist at the University of Chicago and an alumnus of Dr. Extavour’s lab, described embryology as the study of how a developing animal makes “the right parts in the right place at the right time.” In some new research using stunning videos of the cricket embryo – which show certain “right parts” (the cell nuclei) moving in three dimensions – Dr Extavour, Dr Donoughe and their colleagues found that good old fashioned geometry plays a major role.
Humans, frogs, and many other widely studied animals start out as a single cell that immediately divides again and again into separate cells. In crickets and most other insects, the cell nucleus first divides, forming many nuclei that travel through the divided cytoplasm and only later form their own cellular membranes.
In 2019, Stefano Di Talia, a quantitative developmental biologist at Duke University, studied the movement of fruit fly nuclei and showed that they are carried along by pulsating currents in the cytoplasm—kind of like leaves traveling in the eddies of a slow-moving current.
But a different mechanism was at work in the cricket embryo. The scientists spent hours watching and analyzing the microscopic dance of nuclei: glowing nuclei dividing and moving in a bewildering pattern, not quite orderly, not quite random, in different directions and speeds, neighboring nuclei more synchronized than those farther away. The performance refuted a choreography beyond pure physics or chemistry.
“The geometries that the nuclei come to assume are a result of their ability to sense and respond to the density of other nuclei in their vicinity,” Dr. Extavour said. Dr. Di Talia was not involved in the new study, but found it moving. “It’s a beautiful study of a beautiful system with great biological relevance,” he said.
The journey to the cores
The cricket researchers first took a classic approach: Look closely and pay attention. “We just saw it,” said Dr. Extavour.
They took videos using a laser light sheet microscope: Snapshots captured the dance of the nuclei every 90 seconds during the embryo’s first eight hours of development, when 500 or so nuclei had gathered in the cytoplasm. (Crickets hatch after about two weeks.)
Usually, biological material is translucent and difficult to see even with the most sophisticated microscope. But Taro Nakamura, then a postdoctoral fellow in Dr. Extavour’s lab, now a developmental biologist at the National Institute for Basic Biology in Okazaki, Japan, had engineered a special strain of crickets with nuclei that glowed fluorescent green. As Dr. Nakamura recounted, when he recorded the embryo’s development, the results were “amazing.”
That was the “starting point” of the exploratory process, Dr. Donoughe said. He paraphrased a remark sometimes attributed to science fiction author and biochemistry professor Isaac Asimov: “Often you don’t say ‘Eureka!’ when you discover something, you say, ‘Huh. It is strange.'”
At first, the biologists watched the videos on loop, projected onto a conference room screen – the cricket equivalent of IMAX, given that the embryos are about a third the size of a grain of (long-grain) rice. They tried to discover patterns, but the data sets were overwhelming. They needed more quantitative knowledge.
Dr. Donoughe contacted Christopher Rycroft, an applied mathematician now at the University of Wisconsin-Madison, and showed him the dancing nuclei. ‘Wow!’ said Dr. Rycroft. He had never seen anything like it, but he recognized the potential for a data-driven collaboration; he and Jordan Hoffmann, then a graduate student in Dr. Rycroft’s lab, joined the study.
Over the course of a series of screenings, the math-bio team considered many questions: How many nuclei were there? When did they start splitting up? In which direction did they go? Where did they go? Why did some slide around and others crawl?
Dr. Rycroft often works at the crossroads between the life and natural sciences. (Last year he published on the physics of paper curling.) “Math and physics have been very successful in deriving general rules that apply broadly, and this approach may also help in biology,” he said; Dr. Extavour has said the same thing.
The team spent a lot of time swirling around ideas on a white board, often drawing pictures. The problem reminded Dr. Rycroft of a Voronoi diagram, a geometric construction that divides a space into non-overlapping subregions—polygons, or Voronoi cells, each of which originates from a seed point. It’s a versatile concept that applies to things as varied as galaxy clusters, wireless networks, and the growth pattern of forest canopies. (The tree trunks are the seed points and the crowns are the Voronoi cells, which huddle tightly but do not interfere with each other, a phenomenon known as crown shyness.)
In the context of cricket, the researchers calculated the Voronoi cell surrounding each nucleus and observed that the cell’s shape helped predict the direction the nucleus would move next. Dr. Donoughe basically said, “Nuclei tended to move into nearby open spaces.”
Geometry, he noted, offers an abstract way to think about cell mechanics. “For most of the history of cell biology, we couldn’t directly measure or observe the mechanical forces,” he said, although it was clear that “motors and squishes and pushes” were at play. But scientists could observe higher-order geometric patterns produced by these cellular dynamics. “So, when we think about the spacing between the cells, the size of the cells, the shapes of the cells — we know that they come from mechanical constraints at very fine scales,” Dr. Donoughe said.
To extract this kind of geometric information from the cricket videos, Dr Donoughe and Dr Hoffmann tracked the nuclei step by step, measuring their location, speed and direction.
“This is not a trivial process and it ends up involving many forms of computer vision and machine learning,” said Dr Hoffmann, an applied mathematician now at DeepMind in London.
They also verified the software’s results manually, clicking through 100,000 positions, and connecting the cores’ lines through space and time. Dr. Hoffmann found it boring; Dr. Donoughe thought of it as playing a video game, “zooming at high speed through the tiny universe inside a single embryo, stitching together the threads of each nucleus’ journey.”
They then developed a computational model that tested and compared hypotheses that can explain the nuclei’s movements and positioning. All in all, they ruled out the cytoplasmic currents that Dr. Di Talia saw in the fruit fly. They disproved random motion and the notion that nuclei physically pushed each other apart.
Instead, they arrived at a plausible explanation by building on another known mechanism in fruit fly and roundworm embryos: miniature molecular motors in the cytoplasm that extend clusters of microtubules from each nucleus, not unlike a forest canopy.
The team proposed that a similar type of molecular force pulled the cricket nuclei into unoccupied space. “The molecules may well be microtubules, but we don’t know for sure,” Dr. Extavour said in an email. “We need to do more experiments in the future to find out.”
The geometry of the manifold
This cricket odyssey would not be complete without a mention of Dr. Donoughe’s custom-made “embryoconstriction device,” which he built to test various hypotheses. It recreated an old-fashioned technique, but was motivated by earlier work with Dr. Extavour and others on the evolution of egg sizes and shapes.
This device enabled Dr. Donoughe to perform the delicate task of looping a human hair around the cricket egg—thus forming two regions, one containing the original nucleus, the other a partially pinched-off annex.
Next, the researchers looked again at nuclear choreography. In the primordial region, the nuclei slowed down when they reached an overcrowded density. But when a few cores snuck through the tunnel at the constriction, they picked up speed again and let loose like horses in an open pasture.
This was the strongest evidence that the nuclei’s movement was governed by geometry, Dr. Donoughe said, and “not controlled by global chemical signals, or currents, or pretty much all the other hypotheses out there for what could plausibly coordinate the behavior of an entire embryo.” “
By the end of the study, the team had amassed more than 40 terabytes of data on 10 hard drives and had refined a computational, geometric model that was added to cricket’s toolkit.
“We want to make cricket embryos more versatile to work with in the lab,” Dr. Extavour said — that is, more useful in the study of even more aspects of biology.
The model can simulate any egg size and shape, making it useful as a “testing ground for other insect embryos,” Dr. Extavour said. She noted that this will make it possible to compare different species and go deeper into evolutionary history.
But the study’s greatest reward, all the researchers agreed, was the spirit of collaboration.
“There is a place and time for specialized knowledge,” Dr. Extavour said. “Just as often in scientific discovery, we have to expose ourselves to people who are not as invested as we are in any particular outcome.”
The questions asked by the mathematicians were “free of any kind of bias,” Dr. Extavour said. “Those are the most exciting questions.”