Going With the Flow: Mapping Vascular Systems in Fruit Flies

Unlike us, insects don’t have lungs. Instead, a complex ductwork of microscopic tubes called the trachea delivers oxygen throughout their bodies.

At the Flatiron Institute, Tatyana Gavrilchenko models the trachea of fruit flies. But her results aren’t only applicable to insect physiology. This fundamental work is helping scientists understand how biological distribution networks deliver nutrients, which has applications in studies of the structure of veins and arteries and in the development of artificial organs.

Gavrilchenko joined the Flatiron Institute’s Center for Computational Biology (CCB) in 2019 as a Flatiron Research Fellow. Prior to that, she earned a doctorate in physics from the University of Pennsylvania.

Gavrilchenko recently spoke with the Simons Foundation about her work and about the future of artificial organs. The conversation has been edited for clarity.

What projects are you currently working on?

Broadly speaking I model biological distribution networks, the vasculature system being one of the most well-known examples, existing at many scales of multicellular life. In animals, the vasculature comprises blood vessels, while plants have xylem and phloem cells. These systems are the essential support structures that shuttle material within an organism, providing nutrition and removing waste.

Despite their ubiquity, there’s little existing theory about how vascular networks grow and function. One particularly fascinating system is the vasculature of the brain — an interwoven net that permeates the entire organ, delivering oxygen to brain cells. It’s an intricate and dense system since every cell needs to be within a few microns of an oxygen source. This complexity makes it difficult to study. Instead, we’re starting by looking at a simpler system that follows similar basic principles. And we’ve focused our efforts on Drosophila melanogaster, the common fruit fly.

Insects breathe very differently from humans. Instead of lungs that transfer oxygen to red blood cells, which then flow through blood vessels, insects have a network of air-filled trachea: tubes that carry oxygen throughout the animal’s body. At the ends of this tracheal network are terminal tracheal cells. This is where oxygen leaves the tubes and gets absorbed into the animal’s muscles and organs.

These terminal cells, which have spindly shapes resembling neurons, are essentially hollow pipes surrounded by cellular walls of varying thickness that prevent too much oxygen from leaking out. The cells’ branched architecture allows for uniform oxygen delivery throughout the tissue. At the CCB, we’re working to model these cells’ functional capacity and the way they grow into their intricate structure.

How do you model these systems?

Our work starts with examining many terminal cells, looking for patterns in the cell branches and finding measurable features to quantify these patterns. My collaborators at Princeton University take microscopic photographs of tracheal terminal cells in fruit fly larvae. We simplify each cell into an abstract network, sweeping away the cellular details and distilling it into a network of connected tubes, keeping only the geometric structure. We have two main research questions: One is how the networks grow, and the other is how they deliver oxygen. It turns out that these questions are linked: A muscle that is not getting enough oxygen releases chemical signals, and in response, the terminal cell grows its tendrils to fill the undersupplied area.

Our current theoretical model for oxygen delivery is based on solving equations for how oxygen moves inside of the cell, passes through the cell wall, and gets absorbed by surrounding tissue. With the help of collaborators at the Flatiron Institute’s Center for Computational Mathematics, we have reformulated these equations into a different problem where we solve for oxygen concentration at discrete points as it decays along the path of the network. Adding up the contribution from all the points, we can make a map of how oxygen is distributed in the tissue space.

Why is it important to understand these systems?

Our work in fruit flies is a starting point for understanding function in complex mammalian vascular systems. In fruit flies, a single terminal cell — combined with the muscles that it supplies — forms a compartmentalized system; we know exactly where oxygen enters and where it is absorbed. The modeling problem for a relatively simple network with 40 or 50 individual tubes is much more tractable than for a whole brain vasculature, comprised of hundreds of millions of small tubes.

Furthermore, if you want to build a working vascular system, you must understand the details of the architectural features that allow it to work efficiently. The most prominent application of this work is the design and manufacture of artificial kidneys, livers and other organs. Scientists have been able to isolate organ cells and grow them in cultured media, but the problem of oxygen supply imposes a size limit for these systems. As the cells layer on top of each other during growth, the inner cells become starved of oxygen. If you want to build a larger, functional artificial organ, you need to build the right vasculature to support it as well.

How far off are we from making this sort of artificial vasculature?

We don’t yet have a mathematical language to speak about the design principles of distribution network growth and efficiency. However, researchers are already working on the mechanical side of engineering these systems. It is relatively simple to make microscopic networks out of materials like glass, but these materials don’t allow oxygen to leak through the vessel walls.  Only in the past decade are scientists learning how to build delivery networks from biologically compatible materials. I think that in another decade or two we will start to see the mathematical and mechanical pieces come together. Maybe by then we’ll have an artificial organ vasculature that could be grown to follow an algorithm and we’ll be studying how well it matches an actual liver or kidney vascular network.