The entire mammalian blood system can be restored from a single haematopoietic stem cell (HSC). These self-renewing cells reside in the bone marrow, where their proliferation and differentiation are tightly regulated to ensure a continued, lifetime supply of blood cells. The high regenerative capacity of HSCs makes them central to life-saving bone-marrow transplants, but also makes them prone to acquiring leukaemia-initiating genetic mutations. The ability to manufacture HSCs in the laboratory therefore holds enormous promise for cell therapy, drug screening and studies of leukaemia development. Two studies1,2 in this issue outline protocols to achieve just that.
In the embryo, the first HSCs are thought to develop from specialized 'haemogenic' endothelial cells that line the walls of a blood vessel called the dorsal aorta3,4. The precise molecular mechanisms behind this intriguing conversion remain to be defined, but are widely thought5 to be important for successful in vitro generation of HSCs. The current studies took this supposition as the basis for the development of their protocols.
In the first study, Sugimura et al.1 (page 432) started with human pluripotent stem cells (PSCs), which can give rise to any cell type in the body. The authors adapted a previous method6 to derive haemogenic endothelial cells from human PSCs. They then identified a subset of seven transcription factors that could induce the haemogenic endothelial cells to become immature HSCs.
In the second study, Lis et al.2 (page 439) directly converted adult mouse endothelial cells to immature HSCs through transient expression of a panel of four transcription-factor genes, including two, Runx1 and Spi1, that were also used by Sugimura and colleagues. The group's strategy opens up the exciting prospect that HSCs could eventually be directly induced from a person's own tissues.
The development of HSCs has been likened to progression from a maternity ward to a finishing school7, with the former being where HSCs are born, and the latter where they mature in response to external signals. This analogy seems fitting for the current papers (Fig. 1). In both cases, programming by transcription factors was required for HSC birth, and extracellular cues were essential for subsequent maturation.
Sugimura et al. provided 'finishing school' cues by transplanting their HSCs into the bone marrow of adult mice. By contrast, Lis and colleagues grew their cells on a layer of endothelial cells taken from the umbilical cord, which released as-yet-undefined factors that support HSC development, before transplantation into mice. HSCs generated by both protocols could engraft when transplanted into recipients, giving rise to all major blood lineages, and the cells of these animals could engraft in a second recipient — a major step forward compared with previous methods.
The dorsal aorta is believed to act as both a maternity ward and a finishing school for HSCs during normal embryonic development8. The layer of endothelial cells used by Lis et al. captures some aspects of this native setting. By contrast, in injecting their cells directly into adult-mouse bone marrow to achieve HSC maturation, Sugimura et al. exposed the cells to signals from the niche in which adult HSCs normally reside. The finishing-school site for Sugimura and colleagues' cells might be the perivascular HSC niche — a region close to blood vessels that is partly made up of endothelial cells. There is accumulating evidence9 that this niche is involved in the maintenance of adult HSCs. It is therefore plausible that there is more flexibility than previously thought in the type of environment that can support HSC maturation, and that one of the main requirements is an endothelial wall.
When asked why they should be given public money, developmental biologists have long argued that a better understanding of normal developmental processes will be crucial to the generation of authentic cell types for cell-based therapies and drug screening. The current papers provide excellent examples of this, because their protocols rely on features from embryonic mechanisms. Both studies used endothelial cells as a starting material, in analogy to HSC development in the embryo, and their transcription-factor cocktails included factors, such as RUNX1 and SPI1, that regulate10,11,12 the birth of the first HSCs during embryonic development in vivo.
Despite the excitement that the current papers will generate, there are some notable limitations. For instance, only Lis and colleagues addressed the potential of the programmed HSCs to become cancerous, finding no evidence of leukaemia for up to 20 weeks after engraftment. Longer follow-up will be required — especially for the human cells produced by Sugimura and co-workers.
Future investigations are also likely to focus on how the transcription-factor cocktail is introduced. The current studies used genetic vectors derived from retroviruses that integrate into random sites in the host-cell genome. Expression of the transcription factors is then driven by a drug called doxycycline that can be administered as needed. 'Cleaner' genome-engineering strategies (such as CRISPR–Cas9 technology) that can direct the insertion of sequences into specific sites in the genome are advancing rapidly. These techniques could eliminate concerns that retroviral integration might activate cancer-causing genes13, or that expression of the transcription factors might not be completely extinguished under inducible, doxycycline-based systems. The latter is a worry because many of the transcription factors used in the current studies have also been implicated in the initiation of leukaemia.
Another caveat is that only a small minority of cells from the engineered populations produced progeny that could be detected in the bloodstream after a few months. Although this was sufficient to repopulate the blood, further studies will be required to fully define the molecular make-up of HSCs produced in vitro. This is particularly pressing because both studies suggest that, although the programmed cells are similar to HSCs, they are not identical.
HSCs are defined by their ability to engraft and give rise to all blood-cell lineages over many months. This is arguably the most stringent assay of any adult stem cell. Cells fulfilling these criteria have now been produced in two ways — a fact that not only opens up exciting opportunities in the haematopoiesis field, but also represents a milestone for the wider stem-cell community. Although further studies are needed, the long journey to translate the promise of stem-cell research into direct patient benefit may just have become a little shorter.