For all vertebrates, the textbook picture of how blood vessels normally develop begins with the embryonic layer known as the mesoderm. That tissue gives rise to hemangioblastsstem cells that can become either hematopoietic cells (red and white blood cells) or endothelial cells (the precursors of tubular blood vessels). Although it seemed a strange notion that these two very different cell types have a common progenitor, Mark A. Krasnow, an HHMI investigator at Stanford University School of Medicine, has discovered what he believes is an evolutionary clue. Krasnow found that blood cell growth in fruit flies is regulated by vascular endothelial growth factor (VEGF), which is also important for the development of blood vessels in mammals.
Fruit flies have no true blood vessels, only an open circulatory system through which runs a network of tracheal tubes that transports oxygen and serves as the respiratory system. But Krasnow believes that the blood cells and VEGF signaling system found in fruit flies were adapted in some ancient animal to become the closed cardiovascular system of mammals. "Maybe some subset of blood cells acquired the ability during evolution to form associations with other blood cells and assemble into tubular structures," he hypothesizes. "The tubes connected up, and because the VEGF signaling system was already in place, they recruited and used this receptor and ligand to build a vascular network."
BIRTH OF A BLOOD VESSEL
The story of how a mammalian embryo develops a full-fledged vasculature goes basically like this: In the mouse, the hemangioblasts give rise to endothelial cells. Around embryonic day 8, those cells form tubules and, in a process called vasculogenesis, a loose vascular net. Next, the tubules organize, hook up to a heart that starts beating at about day 9, and remodel into a complex network by day 10. This remodeling process, called angiogenesis, can occur later in other parts of the body, such as limbs, and even during adulthood when tissues repair themselves or when tumors grow.
Although many researchers study blood vessel development in mice, some also study the process in zebrafish, which grow an intact circulatory system in just 24 hours. "You can't beat the zebrafish for a model of blood vessel development," Zon says. "I can watch vasculogenesis, angiogenesis, see the blood cells go around the circulation, see how cells get recruited to form a functional vasculatureall under a microscope and in the short lifetime of a fish."
He can also manipulate changes in gene expression and witness their effects on embryonic development. That has led Zon's group to identify some key genes in blood vessel development. For instance, cloche mutant zebrafish embryos show a severe reduction in hemangioblasts and don't develop blood vessels or blood cells. A knockdown of the transcription factor SCL leads to a failure in angiogenesis and a lack of blood cells. "We don't know if these are the first genes to set in motion vasculogenesis and hematopoiesis, but we think both cloche and scl are critical genes," says Zon. While subtle species differences exist, these genes and their pathways appear to be conserved in higher vertebrates as well, he adds. Other genes, called notch, mindbomb and gridlock, appear to be important in triggering the remodeling that occurs during angiogenesisthey help determine which blood vessels become arteries and which ones become veins.
For the past century, scientists believed that the two kinds of blood vessels were essentially the same tubular plumbing, differing only in function: Arteries carry oxygenated blood to the body's tissues, while veins return deoxygenated blood to the heart. That view was shattered six years ago in a serendipitous discovery in the laboratory of David J. Anderson, an HHMI investigator at the California Institute of Technology.
Anderson remembers the day that his graduate student, Hai Wang, came to him very excited about a pattern of blue staining he was seeing in mouse embryonic tissue. It suggested they had found a factor involved in guiding nerve development. Ten minutes later, Wang returned disappointed, saying that the staining pattern did not involve the nerves after all but rather the blood vessels.
However, Anderson noticed something strange: The carotid artery in the embryonic mouse's head was clearly stained, but none of its veins were. "Everywhere we looked, we found that the factorephrinB2was expressed in arteries but not in veins," Anderson recalls. The group then decided to look at veins more closely. They found that veins expressed ephrinB4, but arteries did not. Anderson consulted with an expert on blood vessels, who told him that no one had ever observed such a difference between arteries and veins before. Says Anderson: "I knew then that this was important. Not only was there a molecular difference between veins and arteries, but we had a ligand-receptor pair that interacts physically with each other."
Anderson's group found that arteries and veins become distinct by day 8 in mouse embryonic development, at the beginning of angiogenesis and before the heart beats. At that time, arteries start expressing ephrinB2, and veins ephrinB4. The ephrinswhich are glycoproteins anchored in cell membranesallow arteries and veins to communicate. When a primitive artery and a primitive vein come in close contact, ephrinB2 and ephrinB4 click together, triggering a change in the interiors of the two cells.
In later studies, Anderson's group found that knocking out the ephrinB2 gene in mice disrupted angiogenesis and caused embryos to die later. "What's particularly interesting about this pair of molecules is that usually one thinks of signaling as going in a single directionwith the receptor listening and the ligand talking," Anderson says. "In this case, the arteries were talking and the veins listening. But the whole network of arteries and veins was disrupted. That suggests bidirectional communication."
Several groups have recently shown that developing blood vessels will even talk to cells outside their own system, thereby influencing the growth and differentiation of organs. It's as if the developing organs must know that a blood supply will be in place before they continue to grow. "The way you guarantee the proper physical interaction is that the development goes hand in hand," explains Douglas A. Melton, an HHMI investigator at Harvard University. "You never make a pancreatic islet cell without a blood vessel. They're made together."
Melton, a biologist who studies pancreas development, came to that conclusion after pondering the special reciprocal relationship that blood vessels have with organsblood provides sustenance to the pancreas, for example, but the pancreas also monitors the blood's sugar level and secretes the appropriate amount of insulin. That raised an interesting developmental question: When you build an animal, how do you get these two separate systems to come together?
Although it's known that most organsthe lungs, liver, pancreas, gastrointestinal tractarise from the embryonic endoderm layer, only recently have researchers begun to understand how they bud off from the primordial tube and become full-fledged organs. In studies of mouse embryos, Melton has found that the endoderm begins to receive signals for pancreatic development at or before day 7.5. By day 9, the prepancreatic endoderm starts expressing a gene specific to pancreas development, and by day 10.5, the tissues begin expressing insulin. Notably, these last two events occur just after contact with endothelial cells. Melton and his colleagues Ondine Cleaver and Eckhard Lammert wondered whether endothelial cells and prepancreatic cells were signaling to each other.
In a series of experiments reported in the October 19, 2001, issue of Science, Melton's team did find evidence of two-way signaling. It was already known that many tissues, including pancreatic, attract blood vessels to grow nearby by secreting VEGF. But the group found that another factor is also required to come from the endothelial cells. "It says, Grow up and make yourself into an islet cell,' " he explains. "But we don't know what that signal is." His lab is presently pursuing its identityfor Melton, one more piece in the larger quest of finding all the steps for converting stem cells into pre-islet cells. That knowledge, he expects, could one day be used to treat patients with diabetes.
Melton's group isn't the only one uncovering this new role of blood vessels in organ development. In the same issue of Science, a research team led by Kenneth S. Zaret of the Fox Chase Cancer Center in Philadelphia reported similar findings in the liver. Using a reagent for tagging endothelial cells in mouse embryos, the team found that contact with these cells was critical for getting the liver to bud from the endoderm. Moreover, when they looked at mouse embryos mutated to have no endothelial cells, no liver buds formed at all. Says Zaret: "Endothelial cells need to be considered as important signaling components in the growth of organs and might even be important in responses to tissue damage in adults and cancer."
His group is now trying to determine whether this mystery signal coming from endothelial cells is used broadly to influence the development of other organs. So far, based on the work involving the pancreas and liver, Zaret predicts that more than one signaling system exists. Endothelial cells signal liver cells to rapidly multiply, whereas they seem to be telling pancreatic cells to specialize to become islet cells. "This underscores the diversity," he says, "of signaling systems that are out there and likely to be discovered for endothelial cells."
CHICKEN SOUP HYPOTHESIS
Learning more about how blood vessels influence organ development could yield novel medical treatments. A research group led by Jennifer LeCouter at Genentech Inc. in South San Francisco has shown that it's possible to stimulate endothelial cells in adult liver tissue in mice and thereby spur liver growth. Reporting in the February 7, 2003, issue of Science, the Genentech group also manipulated the signaling system in a way that protected mice from liver damage caused by carbon tetrachloride, which is known to damage essential cells in the liver. This is an exciting and important finding, says Zon. "It supports what's called the 'chicken soup hypothesis'that the endothelial cells are making all these goodies to make tissues and organs happy, even in adulthood."
Still, real therapies for repairing tissues are not likely to come quickly. It's not clear yet that tweaking one factor in a complex pathway would have therapeutic benefits and no serious side effects. So far, researchers have only a broad outline of how the body builds a well-functioning vasculature, and many questions remain.
For instance, how are the cardiovascular system's intricate branching patterns created? Indeed, by sight alone it's hard to tell whether blood vessel patterns are the result of precise control or random events. But molecular clues now suggest that the positioning of blood vessels is most likely a highly regulated process involving other organ systems and two very different mechanisms.
Krasnow at Stanford has studied the general question of how patterning occurs in the fruit fly tracheal system. The main branches that arise early in development show a highly stereotypic arrangement and appear to be governed by a hardwired genetic program, he says. His group has found that a secreted signaling molecule, called fibroblast growth factor (FGF), directs where each major branch sprouts and grows and where the next generation of branches sprouts.
But what about the finer tertiary branches that must reach out and contact every cell? Here he sees more variability in the tracheal pattern, as the mechanism seems to switch to one based on physiological needs. "A target cell in an oxygen crisis sends out a signal that attracts a new branch, which grows out to the cell and supplies it with more oxygen. Another cell goes through a crisis, and on and on," explains Krasnow. The result is a densely patterned and customized network designed to serve the specific tissue. Surprisingly, the same signaling molecule, FGF, is adapted for new use in this physiological phase. "The same factor and same receptor are used," says Krasnow, "but there's a switch from hardwired control of its gene expression to oxygen-controlled expression that happens in a matter of hours, at the precise time when the tracheal pattern also changes dramatically."
FACTORS IN PATTERNING
The parallels between the fruit fly tracheal system and the mammalian cardiovascular system are striking, although in mammals the key signaling molecule is VEGF. At about day 8 in mouse embryos, still in the hardwired genetic phase of development, VEGF promotes early patterning of the blood vessel network. Then at day 8.5, hypoxia (oxygen deficit) begins to play a role, says M. Celeste Simon, an HHMI investigator at the University of Pennsylvania's Abramson Family Cancer Research Institute. Hypoxia triggers more VEGF, which then stimulates vascular remodeling at the finer level.
Simon also has learned something about how the switch from hardwiring to physiological control occurs. Although it's still unclear how cells detect when oxygen levels get low, once they do so a transcription factor called hypoxia-inducible factor, or HIF, comes into play. "As soon as cells become oxygen deprived, HIF responds within minutesit's a very rapid and reversible response," explains Simon. Her group has found that HIF turns on about 100 genes, 10 of whichincluding the genes for VEGF and FGFare related to blood vessel development. From gene-knockout studies, Simon concludes, "we now know that HIF is important in just about every aspect of the cardiovascular system of the developing embryo: the heart, vasculature, placenta and blood cells."
Still, when it comes to blood vessel patterning, hypoxia is not the only guide. Anderson's group at Caltech examined how blood vessels lay down their patterns in limb skin, which develops around day 15 in mice. It turns out the arteries align themselves with developing peripheral nerves. "We found that when we misrouted the nerve pattern, the arteries were still aligned with the nerves," says Anderson. "That really told us that the nerves were leading and the arteries were following." What signal were they following? Once again it was VEGF, which is secreted by some nerve cells. Interestingly, VEGF also seems to tell the blood vessels to become arteries and not veins, Anderson says.
Applying all these new insights is clearly a challenge for the future. Drugs that manipulate VEGF, ephrins or HIF activity, for instance, would be excellent candidates for blocking blood vessel development in tumors. Many pharmaceutical companies are interested, but so far they are only cautiously optimistic about the potential for such cancer therapies. "We're learning that the way the body gets it right during development is by using a lot of factors working together," says George D. Yancopoulos, president of Regeneron Research Laboratories in Tarrytown, New York. "We need to know all the players and how they act in time, space and amount during blood vessel development. While it's getting more interesting, it's also getting more complicated."
this story in Acrobat PDF format.
Photos (from top): Paul Mead, Fred Mertz, Mark Harmel
Reprinted from the HHMI Bulletin,
June 2003, pages 18-21.
©2003 Howard Hughes Medical Institute