Megan Scudellari
In a field rife with dissension, four recent papers may resolve one of the haematopoietic community's greatest disputes
Shortly after fertilization, a group of rare but nondescript cells begin to form in the mammalian embryo. They are the haematopoietic stem cells, able both to self-renew and differentiate into the entire blood system of an animal. Since their discovery almost 50 years ago, haematopoietic stem cells have been studied by legions of researchers, quickly becoming the best characterized of all known stem cells. Today, these regenerative cells are a mainstay therapy for cancers, genetic disorders and bone marrow diseases.
But despite the wealth of information on haematopoietic stem cells (HSCs) and their widespread use in clinics and labs, their origins remain a mystery. And researchers aren't happy about it. Throughout the haematopoietic community, a contentious debate persists over the birth of these cells.
Blood cells emerge from various locations in the embryo and, once developed, they circulate throughout the body, making it no simple task to pinpoint their source. Piles of conflicting evidence have built up, especially over the past five years. "There is a passion by which people think about these processes. People are very polarized about their work," says Leonard Zon of Children's Hospital Boston, a veteran in the field. "Every time a paper is published, a counter-argument will appear to interpret what the experiment 'really did'."
The dispute
Researchers are divided on two major issues: the anatomical and cellular origins of the adult blood system — the 'definitive haematopoietic system'. In the 1970s researchers believed the yolk sac to be the major site of blood formation, but later experiments led to the prevailing dogma that the aorta-gonad-mesonephros (AGM) region of the embryo is the principal site of definitive haematopoiesis1. A pool of red blood cells does form in the mouse yolk sac seven days after fertilization to assist the embryo through a rapid growth spurt, but these cells are not believed to contribute to the adult blood system. A second, definitive wave of haematopoiesis occurs three days later when clusters of blood cells form in the AGM. Cells harvested from these clusters can reconstitute the blood system of an irradiated adult mouse — the trademark property of HSCs2. Subsequently, there is significant haematopoietic activity in the placenta, yolk sac and liver.
By the 1990s, a general consensus had emerged — adult HSCs are born in the AGM region of the embryo. Then, in 2007, a paper from the lab of Shin-Ichi Nishikawa at the RIKEN Centre for Developmental Biology in Kobe, Japan, shocked the field with results suggesting that definitive HSCs may be derived from yolk sac precursors3. The paper sparked immediate controversy, says Zon. Since then, several technical issues about the paper have arisen, he adds, and many scientists still believe all HSCs originate in the AGM. But "there's still a lot of controversy about yolk sac versus intra-embryonic sites," says Marella de Bruijn, a molecular haematologist at the Weatherall Institute of Molecular Medicine in Oxford.
Close on the heels of where HSCs form comes the argument over which cell type is responsible. "Which cell type gives rise to the first blood cells?" asks Timm Schroeder, deputy director of the Institute for Stem Cell Research at the Helmholtz Centre in Munich, Germany. "It's a deceivingly simple question."
There are experimental data in support of three alternatives, says Georges Lacaud, a stem cell biologist at the University of Manchester, UK. Definitive HSCs could be generated — like their primitive precursors in the yolk sac — from a proposed bipotential haemangioblast that directly gives rise to both blood cells and endothelial cells. Alternatively, as some research proposes, they could arise directly from a layer of endothelial cells on the floor of the aorta, dubbed the 'haemogenic endothelium'. Or, as a surprising 2005 paper suggests, the stem cells might originate from patches of subendothelial mesenchymal cells in the aorta that migrate to the aortic floor, giving the false impression of haemogenic endothelial activity4.
The evidence
A few years ago, researchers weren't sure endothelium capable of producing HSCs existed. Now, thanks largely to recent papers from four labs in Germany, the United States and Britain, that hypothesis is gaining strength, and the debate over the cellular origin of HSCs is on the brink of resolution.
The first paper to hit the shelves was from the lab of Luisa Iruela-Arispe at the University of California, Los Angeles and was published in the December issue of Cell Stem Cell5. To determine which cells in the mouse AGM contribute to haematopoiesis, her team selectively labelled cells in both the AGM endothelium and the underlying mesenchyme, then tracked their progeny in vivo for up to a year. The endothelial progeny migrated to both the fetal liver and the adult bone marrow and could reconstitute an irradiated mouse's blood system on transplantation. The progeny of the underlying mesenchyme seemed incapable of haematopoiesis.
"Our studies confirm that endothelial cells are able to give rise to haematopoietic cells," says Arispe. But, she cautions, that's not to say they are the only cells that can do it — she cites the evidence for haemangioblasts, which are mesenchymal cells.
Two months after Arispe's results were published, three papers in the 12 February issue of Nature, sandwiched between articles celebrating Darwin's 200th birthday, contributed more evidence for the existence of a haemogenic endothelium.
The first study examined the effect of Runx1, a transcription factor required for production of HSCs, in the mouse embryo. Nancy Speck, previously at Dartmouth College and now at the University of Pennsylvania in Philadelphia, and colleagues investigated which cells require Runx1 to produce HSCs. They first tracked whether mouse blood cells were derived from embryonic cells that had at one point expressed VEC, a cell-adhesion protein specific to endothelial cells, and found that 96% of adult bone marrow cells are derived from VEC-expressing cells. They also engineered transgenic mice to systematically delete Runx1 in VEC-expressing cells and found that the loss of Runx1 prevented the generation of HSCs. In addition, once a haematopoietic cell marker (Vav1) was expressed in the cells, marking the end of the endothelial-to-haematopoietic cell transition, Runx1 was no longer required for HSC emergence and function. Referring to Arispe's results along with her own, Speck says: "The evidence is that almost all, if not all, adult blood cells are born from this endothelium in this very brief period of time in development."
The final two Nature papers complement the in vivo analyses of Arispe and Speck with in vitro imaging studies of blood-cell genesis. Schroeder and colleagues in Munich designed their experiment to overcome two experimental pitfalls. To track cell type, they relied on morphology as well as multiple molecular markers (many studies use only one marker). And to track cell changes, they maintained continuous observation instead of taking individual snapshots. "From a single picture, you can always find alternative explanations to the data you're looking at," says Schroeder.
The researchers took high-resolution images every 2–3 minutes of developing mouse mesodermal cells in vitro and then manually tracked individual cells over a period of up to seven days7. The team observed colonies that were clearly endothelial in morphology — tightly adhering cells forming flattened sheets — and in protein expression. From those colonies, they observed individual endothelial cells, about 1 in 1,000, pulling away from the sheets and proliferating as free-floating cells, the trademark morphology of blood cells. These cells also began expressing multiple blood-specific proteins: CD41, CD45 and CD11b.
"It was a really beautiful paper," says Speck. "There's something really gratifying about watching in real time." It took months of work tracking thousands of cells by hand, but the effort was worth it, says Schroeder. "We're working on questions that have been unanswered for decades. If to answer them you have to sit down for half a year and look at these movies, then I think it's worth it," says Schroeder. Besides, he laughs, it beats working in a cold room.
A final piece of evidence came from Lacaud and colleagues at Manchester. They investigated the mechanism by which blast colony–forming cells — the proposed haemangioblasts responsible for the primitive wave of haematopoiesis — generate blood cells8. Using time-lapse photography, the researchers watched blast colonies first give rise to a tight adherent structure and then saw non-adherent cells emerge from it. Using haematopoietic and endothelial markers, they showed that the tight adherent structure contained endothelial cells, from which the haematopoietic cells are generated. They found that about 1% of endothelial cells displayed the capacity to generate haematopoietic cells. Lacaud suspects the frequency is actually higher: "By physically isolating the endothelial cells from the blast colony, we probably impact the rate of haematopoiesis negatively," he says. Lacaud says his results demonstrate a link between the two precursor populations: haemangioblasts do not give rise to blood cells directly, but rather through a haemogenic endothelium intermediate.
The outcome
Overall, declares Lacaud, echoing many other researchers, "I'm convinced." Together, the four recent papers strongly support the endothelial origin of haematopoietic cells, he says. It's especially encouraging to see four groups with four different points of view reach the same conclusion, adds Arispe. Zon agrees: "The findings are consistent with a haemogenic endothelial cell model. I think most people will believe that."
But not all do. "I'm not totally convinced," says Ana Cumano of the Pasteur Institute in Paris and senior author of the 2005 paper proposing that patches of mesenchymal cells below the aorta, and not haemogenic endothelium, are the source of HSCs. "What we have seen, and we still see, is that there are cells with haematopoietic markers underneath the endothelium." The recent imaging studies did not visualize the cells in situ, she notes, nor did they prove the presence of stem cells, just that of typical blood cells. "There has to be something that's still missing," says Cumano.
It seems a divided field will not unite overnight. And more questions linger: Do true haemangioblasts, able to directly produce blood cells and endothelial cells, exist or not? What percentage of adult blood is derived from a haemogenic endothelium? Do any adult HSCs originate in the yolk sac? Each new paper moves the field forward, says Zon, but that momentum continues to be plagued by caveats. "It's still strange that after all these years, we don't know what the truth is," he says with a sigh.