At first glance, our bodies seem impossibly complex, with dozens of organs built to precise specifications in exactly the right places. It seems almost miraculous that all this could develop automatically from a single fertilized egg.
But look a little closer and you’ll see that evolution, the master architect, has been economical with that complexity, relying on the same components again and again in different contexts. Take tubes, for example. “We’re basically a bag of tubes,” says Celeste Nelson, a developmental bioengineer at Princeton University. “We have a tube that goes from our mouth to our rear end. Our heart is a tube. Our kidneys are tubes.” So, too, are lungs, pancreas, blood vessels and more — most of them intricate systems of tubes with many branches.
Branching tubes appear so often because they are the best solution to a key problem that organisms face as they get bigger: As an animal grows, its volume goes up faster than its surface area. That simple physical relationship means that the logistical challenges of supplying oxygen and nutrients, and removing waste products — all of which ultimately depend on diffusion through the surfaces of cells — get more daunting with size.
But a dense forest of branching tubes increases the available surface area enormously. “They allow us to be big,” says Jamie Davies, a developmental biologist at the University of Edinburgh.
In recent years, Davies, Nelson and a few other developmental biologists have made great progress in understanding how the body makes tubes and branches in a variety of organs. Though the details usually vary from one organ to the next, some basic principles are beginning to emerge, as outlined in an article coauthored by Nelson in the Annual Review of Biomedical Engineering. So far, it looks like there are only a few ways to make a tube, only a few ways to control how it branches, and only a few ways to regulate when branching should stop.
Finding success in simplicity
At the most general level, it’s not surprising that development is based on a few simple processes. Every tissue is made of cells, and those cells have only limited options to choose among, such as moving (individually or en masse), changing shape, dividing or undergoing self-destruction. “I normally tell my students that about 90 percent of what we make we can account for with only about a dozen actions,” Davies says.
And once evolution forged a few ways to create tubes and branches (the two go together, more often than not), it makes parsimonious sense that bodies would fall back on that same handful of methods again and again.
Start with a dimple, then extend: Many tubes start from a flat sheet of tissue that develops dimples, or pits. It’s likely that these pits originate when a ring of contractile protein molecules scrunches up on one face of the sheet, causing that face to cup as the opposite face bulges outward.
In organs like lungs, mammary glands and kidneys, this initial pit can then get deeper, like dough as a finger pushes into it, until the pit deepens so much it becomes more like an extending tube. In one well-studied example, the ducts of the mammary glands, each growing duct has an unruly mob of cells at its tip. The cells in this mob respond to the hormones of puberty by dividing rapidly. As they pioneer the advance into new territory, some cells insert themselves into the lining of the tube, pushing the mob forward as the tube lengthens. Continued cell division keeps generating new cells that will in turn go on to line the tube.
“The cool thing about this mechanism is that puberty says ‘Go,’ and as long as hormones are still available, you’re going to keep making cells, and they’re going to keep inserting,” says Andrew Ewald, a developmental cell biologist at Johns Hopkins University School of Medicine, who led the work. “In a mouse, this might be an inch of elongation. In a blue whale, you’re talking about yards. You just leave the motor running longer.”
Hollow out a rod: Cells in the interior of a solid rod die or release their contacts with one another to allow a space to form between them. The mammalian vagina forms by this sort of hollowing, as do the ducts of the pancreas, and probably the salivary glands.
Roll up, roll up: Still other tubes — especially the tiniest capillaries of the circulatory system — form when a single elongated cell rolls up to enclose a space. And the tube that will go on to form the nervous system arises from a much larger roll-up, in which two ridges of tissue atop the early embryo bend toward each other, like two breaking waves, until they meet in the middle and fuse, leaving a tube — the barrel of the waves, in essence — enclosed beneath a cover of cells.
From tubes to branches
Almost all the body’s tubes form in one of these ways. And there’s another level where developing organs rely repeatedly on a small set of tricks and techniques: the construction of elaborate networks of branches from all those tubes.
Branches generally form either when a single growing tip encounters two different zones of attraction and sends a tip in each direction, or when something physically restricts the tip’s progression. In the lung, for example, branching occurs when a band of smooth muscle fibers forms across the tip of the growing tube, creating a barrier and forcing growth to both sides.
The developing embryo must also manage the spatial growth of branching tubes so that, for example, the lung fills with just the right amount of tiny, branched airways or the circulatory system delivers capillaries to every part of the body, all without overcrowding or gaps. Researchers are only beginning to understand this control process, although a few key points are emerging.
One simple management strategy is for tubes to branch if space is available and stop when they get crowded. That straightforward system seems to apply for the mammary glands, which are little more than masses of branching milk ducts embedded in a fatty matrix.
To better understand the process, Ben Simons, a developmental biologist at the University of Cambridge, and his colleagues examined preserved mouse mammaries in meticulous detail and mapped out where, and in what context, each individual branching event must have taken place to give rise to the final structure they saw.
They found that each tube continued to grow and branch only if it was not surrounded by other tubes. Actively growing tips formed a front at the edges of the mammary, advancing into new territory, but any new tips that turned inward, to territory already colonized, would shut down. These rules, played out over time, led the ducts to fill in the available space.
The molecular signals that govern this behavior have not been fully worked out, though presumably some sort of inhibition is involved. Simons suspects that the same signaling system may go awry in breast cancer, since the early stages of that disease are characterized by extra branching. “It’s interesting to ask how tumors reactivate that branching program, and how come it doesn’t terminate,” he says — and he’s actively working to understand this.
This system of branching to fill space has the virtue of simplicity, Simons adds. “Everything is local. The cells only have to sense what’s happening in their neighborhood, and it doesn’t require any memory. Cells don’t have to remember what decision they made way back when.”
But the downside is that the gland doesn’t always fill the space perfectly. Occasionally, it leaves gaps in the interior — ones that can no longer be filled because the growing tips are now all out at the periphery.
Taking a different path
The pancreas also uses local rules to build a branched structure, but by a totally different route. The organ starts its life as a mass of cells that buds from the tube that forms the gut. Gradually, holes begin to appear in the mass, and these holes eventually fuse to create an interconnected meshwork of passages. “It doesn’t look like branches initially — it looks like a net, or a road network in a city,” says Anne Grapin-Botton, a developmental biologist at the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Germany. But the process isn’t done yet.
The passages initially are tiny. As the cells that line them begin to secrete their fluids into the open spaces, though, Grapin-Botton hypothesizes that the channels with higher flow rates begin to widen, and those with low flow rates narrow. When she simulates this process using the same mathematical equations that describe how rivers shift from braided rivulets to channels with branching tributaries, she gets a pattern matching that of a real pancreas. But she has not yet observed this flow-related adjustment in a living pancreas.
Such a process has an element of randomness, and that is exactly what anatomists see: Every pancreas has its own branching pattern — even to the extent that some people have two ducts draining the pancreas, while others have just one. “There is no primary design in the pancreas, as far as we know,” says Grapin-Botton. “What guarantees the reproducibility is the feedback from the flow.” The salivary glands and perhaps the tear glands of the eyes may develop their branching networks in the same way, she says.
But not every organ can tolerate the little imperfections that come with this sort of random space-filling. It’s easy to imagine, for example, that an animal might need every bit of its potential lung capacity when fleeing a predator, so unfilled spaces could prove lethal. Not surprisingly, then, evolution has shaped a more precise developmental program for the lungs.
Like the pancreas, the mammalian lung begins as a tube-shaped outpocketing — again, off the embryonic gut. It then branches into two tubes, and each of those goes on to branch again and again, many times, until the lung is filled with millions of tiny airways. Detailed analyses of mouse lungs suggest that the first 15 of these cycles of branching occur in the same location in every lung, so these branchings must be following a preset plan, Nelson says. After that, the lung switches to a space-filling strategy, so that the final lung conforms to the space available in the chest cavity even if other organs take more or less space than usual.
Even more strikingly, researchers can put embryonic lungs into artificial chambers, and the lungs grow to conform to the space of those, too. “You can make cubic lungs or cylindrical lungs,” Nelson says.
This two-stage branching strategy might deliver the best of both worlds, Nelson adds. The early hardwired branches ensure a basic structure that fills the whole chest cavity, and the later space-filling branches finish the detailing. “From a design perspective, it makes a lot of sense,” she says. “As an engineer, I love that. But we don’t really know how that happens.”
Some hints are starting to emerge, however. Those early, pre-programmed branches depend on a molecule called FGF10, a growth factor that helps orchestrate development by carrying signals from one cell to another. In mouse embryos genetically engineered to lack FGF10, lung passages lengthen but don’t branch. No one knows exactly how FGF10 determines the location of branch points, but many researchers lean toward an explanation first proposed by the mathematician Alan Turing more than half a century ago. Turing showed that under certain conditions, signaling molecules that diffuse freely among cells can spontaneously form regular spatial patterns, even in the absence of any external cue.
Mathematical simulations by Dagmar Iber, a computational biologist at ETH Zurich, and her colleagues have shown that such Turing patterns could indeed cause the regular branching patterns seen in the lung. Iber’s team has also shown that the signaling pathways used by real lung cells meet the conditions necessary for Turing patterns to form, though they have not yet demonstrated that this mechanism does indeed direct the branching of a living embryonic lung.
A similar Turing-like organizing principle seems to be at work in another branched organ, the urine-collecting ducts of the kidney. There, too, researchers have found a highly predictable pattern governing early branches, also directed by a key signaling molecule — but the kidneys use a different one, called GDNF.
Much remains unknown about branching in these organs. In the case of the lungs, for example, researchers have known since the 1990s that the size of each successive branch within the organ fits a fractal pattern in which the volume of each length of tube is equal to the volume of the two daughter tubes that it gives rise to. This allows air pressure to remain constant as air is drawn in and out of the airways — but air pressure itself cannot help shape this pattern, which arises long before the lung is actually used. “When the baby takes its first breath, it needs to have a perfect lung,” Iber says. “How does nature manage to arrive at that architecture?”
Filling out the framework
Yet another variation on the programmed-then-local-control theme plays out in the blood vessels of the body. Here, too, the developing embryo needs to ensure that a basic framework of major blood vessels is reliably in place. “The first and most important thing is to get a vessel to every important part of the body. There, you have to hardwire it,” says Markus Affolter, a developmental biologist at the Biozentrum of the University of Basel, Switzerland.
But once that basic scaffold is in place, the embryo switches to a supply-and-demand system. Tissues that find themselves short of oxygen send out a signal, a molecule known as VEGF, which prompts existing blood vessels to sprout new branches that grow toward the oxygen-starved area. Once the new blood vessels begin delivering oxygen, VEGF secretion drops off, and no further vessels sprout. Eventually, the newly created vessels with high blood flow stabilize, while those with minimal blood flow are pruned away, and the network of blood vessels settles into a stable, efficient configuration.
Developmental biologists are encouraged by these common themes in the genesis of tubes and branches. “We wouldn’t want to claim that all branching processes are the same — that would be much too strong,” Simons says. “But we think there is a conservation of principles. That doesn’t mean that the molecular underpinnings are the same, but the rules are.”
Where possible, it seems, evolution has usually chosen relatively local controls to determine where and when branches are made — as when mammary ducts keep branching until they bump into a boundary, or when blood vessels grow toward cells starved of oxygen. “It’s neater in terms of evolvability to have these simple programs that you run over and over,” Davies says.
But when the system needs to meet more stringent specifications — as when the early embryo needs to guarantee that blood vessels serve every organ — it looks as though evolution opted to pay the higher costs for a more precise, preordained script.
And even when different organs implement the same strategy, the particular molecular tools they use in each case can differ. “The devil is in the details, and the details are different from organ to organ,” says Nelson. “I think that’s beautiful.”