Developmental biology

It takes muscle to make blood cells

Blood stem cells derive at least in part from an embryonic vessel called the dorsal aorta. It emerges that a flanking tissue called the somite contributes cells and signals to this process. See Letters p.314 & p.319

The extrinsic cues that instruct cells to become blood-cell precursors are mostly unknown. Studies in different vertebrate models have shown that these precursors, called haematopoietic stem cells, originate at least in part from the first functional intra-embryonic blood vessel, the dorsal aorta. However, the cell types and signalling molecules that promote the generation of haematopoietic stem cells in the dorsal aorta are not well understood. Two papers in this issue1,2 find that structures called somites, precursors of a range of tissues including the vertebrae and skeletal muscle, are involved at more than one stage of this developmental process.

Haematopoietic stem cells (HSCs) have the capacity to replenish all blood-cell types throughout life. Live-imaging experiments in zebrafish (Danio rerio)3,4,5 provided the first conclusive evidence that these cells derive from the endothelial cells that line the dorsal aorta (DA). However, these and other studies6 indicated that only a subset of DA endothelial cells can become HSCs.

Studies in avian embryos7 gave the first indication that the ability of endothelial cells to become HSCs was determined by their origin. Endothelial cells from a tissue called the lateral plate mesoderm populate part of the DA, and can give rise to HSCs7. Endothelial cells from the somites populate another part, and do not become HSCs8. When production of HSCs within the DA ceases, the somite-derived endothelial cells replace those from the lateral plate mesoderm8. It therefore stands to reason that somite-derived cells and signalling factors may regulate the endothelium-to-HSC transition.

Nguyen et al.1 (page 314) investigated the role of the somites in development of the DA and HSCs in zebrafish using a mutant fish strain called choker, in which signalling from somitic tissues is defective9. They expanded on the previous analysis of this defect, and found that choker fish harbour a defective copy of meox1, a gene that is normally expressed in the somites. They then examined HSC development in these mutants. Surprisingly, these animals had more HSCs than their wild-type siblings.

Next, the authors used genetic tools to trace the descendants of cells within the somites (an approach known as lineage tracing). They found that, as in birds and mice7,8, a portion of zebrafish somitic cells give rise to endothelial cells that line the DA — a progenitor population that Nguyen and colleagues named endotomal cells. Up- and downregulating meox1 function revealed that the gene inhibits the formation of endotomal cells in the somites. Moreover, the authors found that endotomal cells do not give rise to HSCs themselves, but help other endothelial cells to become HSCs (Fig. 1).

Figure 1: The birth of blood-cell precursors.

A simplified cross-section through a zebrafish embryo illustrates how haematopoietic stem cells (HSCs) arise in the dorsal aorta (DA), an embryonic structure lined by endothelial cells from two different origins. Nguyen et al.1 report that endothelial cells that migrate to the dorsal aorta from the somites (blue arrows) do not give rise to HSCs themselves, but instead help other endothelial cells in the DA to become HSCs. Kobayashi et al.2 find that endothelial precursors that migrate from the lateral plate mesoderm (red arrows) interact with the developing somite en route to the DA (dashed arrows). Some of these cells go on to become HSCs.

Accordingly, ablation of endotomal-derived endothelial cells led to a significant reduction in the number of HSCs. This outcome was, at least in part, due to reduction of the signalling protein Cxcl12b. When Nguyen and co-workers either chemically inhibited or 'knocked down' Cxcl12b function, they saw a drastic reduction in HSC numbers.

Given the small size, elongated shape and tight packing of endothelial cells and their non-endothelial neighbours, it will be important to follow up this and other lineage-tracing studies10,11 in mice, for example by using an approach that marks somite-derived cells only if they become endothelial cells. One interpretation of the studies in birds8 is that replacing the endothelial cells derived from the lateral plate mesoderm with somite-derived endothelial cells limits the DA's ability to generate HSCs. In light of Nguyen and colleagues' evidence that reducing the number of somite-derived endothelial cells actually decreases HSC formation in zebrafish, it will be essential to directly test the role of somite-derived endothelial cells in HSC emergence in mice.

Do the somites also regulate HSC precursors at earlier stages? The lateral plate mesoderm is in direct contact with somitic tissue, and there is evidence12 that somite-derived signalling proteins promote HSC formation as cells migrate from the lateral plate mesoderm to the forming DA. Wnt16 may be one such protein12, although Nguyen and co-workers question this finding, because wnt16 expression is in fact downregulated in choker mutants.

In addition to secreted factors, direct physical interactions between migrating cells and somitic tissues might regulate HSC generation. Kobayashi et al.2 (page 319) report that cell-adhesion proteins called junction adhesion molecules (Jams) are involved in interactions between HSC precursors and somites, which in turn are required for HSC formation. They found that endothelial precursors in the lateral plate mesoderm express jam1a as they migrate towards the forming DA along the somite surface, which expresses jam2a. The two Jam proteins physically interact, promoting strong cell–cell contacts.

Kobayshi and colleagues also found that the somites express the genes deltaC and deltaD, which encode Notch-binding proteins, and that the migrating endothelial precursors express Notch protein. Notch signalling is involved in cell–cell communication in many settings in the body. Knockdown of jam1a led to loss of Notch signalling and loss of HSCs, but HSC formation could be restored in these embryos with forced activation of Notch signalling in endothelial precursors. Thus, strong Jam-mediated cell–cell interactions might facilitate the activation of Notch signalling in endothelial precursors.

This work suggests that Notch signalling is required in HSC progenitors during their migration from the lateral plate mesoderm, earlier than previously reported12. This model has implications for optimizing the induction of HSCs in vitro, which remains challenging, probably because some key factors are missing from the protocols being used.

In addition to Kobayashi and co-workers' finding, another study13 has shown that somite-derived endothelial cells themselves require endogenous Notch signalling. To analyse the role of Notch signalling in more detail, tools that permit precise temporal and spatial control of gene expression will therefore be needed. These tools will include conditional genetic mutants that can be induced to lack gene function only at certain times and in certain tissues, technology that for the most part is not yet available in zebrafish.

Together, these two studies show that somites play a key part in HSC formation. The idea that cells from the somites populate embryonic blood vessels is now fairly well established, but their role in HSC formation in birds and mammals remains to be defined. Ultimately, of course, understanding where the various cell types come from should help to determine the identity and exact sequence of signalling pathways activated in HSCs, their precursors and their derivatives.


  1. 1

    Nguyen, P. D. et al. Nature 512, 314–318 (2014).

  2. 2

    Kobayashi. I. et al. Nature 512, 319–323 (2014).

  3. 3

    Bertrand, J. Y. et al. Nature 464, 108–111 (2010).

  4. 4

    Kissa, K. & Herbomel, P. Nature 464, 112–115 (2010).

  5. 5

    Lam, E. Y., Hall, C. J., Crosier, P. S., Crosier, K. E. & Flores, M. V. Blood 116, 909–914 (2010).

  6. 6

    Boisset, J. C. et al. Nature 464, 116–120 (2010).

  7. 7

    Pardanaud, L. et al. Development 122, 1363–1371 (1996).

  8. 8

    Pouget, C., Gautier, R., Teillet, M. A. & Jaffredo, T. Development 133, 1013–1022 (2006).

  9. 9

    Svetic, V. et al. Development 134, 1011–1022 (2007).

  10. 10

    Esner, M. et al. Development 133, 737–749 (2006).

  11. 11

    Mayeuf-Louchart, A. et al. Proc. Natl Acad. Sci USA 111, 8844–8849 (2014).

  12. 12

    Clements, W. K. et al. Nature 474, 220–224 (2011).

  13. 13

    Sato, Y. et al. Dev. Cell 14, 890–901 (2008).

Download references

Author information

Correspondence to Suphansa Sawamiphak or Didier Y. R. Stainier.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sawamiphak, S., Stainier, D. It takes muscle to make blood cells. Nature 512, 257–258 (2014) doi:10.1038/nature13740

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.