New papers in Science magazine and the journal Cell have addressed a long-standing question of how the descendants of hematopoietic stem cells in bone marrow make the various types of blood cells that course through our blood vessels and occupy our lymph nodes and lymphatic vessels.
Hematopoietic stem cells (HSCs) are partly dormant cells that self-renew and produce so-called “multipotent progenitors” or MPPs that have reduced ability to self-renew, but can differentiate into different blood cell lineages.
The classical model of how they do this goes like this: the MPPs lose their multipotency in a step-wise fashion, producing first, common myeloid progenitors (CMPs) that can form all the red and white blood cells except lymphocytes, or common lymphoid progenitors (CLPs) that can form lymphocytes (see the figure below as a reference). Once these MPPs form CMPs, for example, the CMP then forms either an MEP that can form either platelets or red blood cells, or a GMP. which can form either granulocytes or macrophages. The possibilities of the types of cells the CMP can form in whittled down in a step-by-step manner, until there is only one choice left. With each differentiation step, the cell loses its capacity to divide, until it becomes terminally differentiated and becomes platelet-forming megakarocyte, red blood cell, neutrophil, macrophage, dendritic cells, and so on.
These papers challenge this model by arguing that the CMP does not exist. Let me say that again – the CMP, a cell that has been identified several times in mouse and human bone marrow isolates, does not exist. When CMPs were identified from mouse and human none marrow extracts, they were isolated by means of flow cytometry, which is a very powerful technique, but relies on the assumption that the cell type you want to isolate is represented by the cell surface protein you have chosen to use for its isolation. Once the presumptive CMP was isolated, it could recapitulate the myeloid lineage when implanted into the bone marrow of laboratory animals and it could also produce all the myeloid cells in cell culture. Sounds convincing doesn’t it?
In a paper in Science magazine, Faiyaz Notta and colleagues from the University of Toronto beg to differ. By using a battery of antibodies to particular cell surface molecules, Notta and others identified 11 different cell types from umbilical cord blood, bone marrow, and human fetal liver that isolates that would have traditionally been called the CMP. It turns out that the original CMP isolate was a highly heterogeneous mixture of different cell types that were all descended from the HSC, but had different developmental potencies.
Notta and others used single-cell culture assays to determine what kinds of cells these different cell types would make. Almost 3000 single-cell cultures later, it was clear that the majority of the cultured cells were unipotent (could differentiate into only one cell type) rather than multipotent. In fact, the cell that makes platelets, the megakarocyte, seems to derive directly from the MPP, which jives with the identification of megakarocyte progenitors within the HSC compartment of bone marrow that make platelets “speedy quick” in response to stress (see R. Yamamoto et al., Cell 154, 1112 (2013); S. Haas, Cell Stem Cell 17, 422 (2015)).
Another paper in the journal Cell by Paul and others from the Weizmann Institute of Science, Rehovot, Israel examined over 2700 mouse CMPs and subjected these cells to gene expression analyses (so-called single-cell transriptome analysis). If the CMP is truly multipotent, then you would expect it to express genes associated with lots of different lineages, but that is not what Paul and others found. Instead, their examination of 3461 genes revealed 19 different progenitor subpopulations, and each of these was primed toward one of the seven myeloid cell fates. Once again, the presumptive CMPs looked very unipotent at the level of gene expression.
One particular subpopulation of cells had all the trappings of becoming a red blood cell and there was no indication that these cells expressed any of the megakarocyte-specific genes you would expect to find if MEPS truly existed. Once again, it looks as though unipotency is the main rule once the MPP commits to a particular cell lineage.
Thus, it looks as though either the CMP is a very short-lived state or that it does not exist in mouse and human bone marrow. Paul and others did show that cells that could differentiate into more than one cell type can appear when regulation is perturbed, which suggests that under pathological conditions, this system has a degree of plasticity that allows the body to compensate for losses of particular cell lineages.
Fetal HSCs, however, are a bird of a different feather, since they divide quickly and reside in fetal liver. Also, these HSCs seem to produce CMPs, which is more in line with the classical model. Does the environmental difference or fetal liver and bone marrow make the difference? In adult bone marrow, some HSCs nestle next to blood vessels where they encounter cells that hang around blood vessels known as “pericytes.” These pericytes sport a host of cell surface molecules that affect the proliferative status of HSCs (e.g., nestin, NG2). What about fetal liver? That’s not so clear – until now.
In the same issue of Science magazine, Khan and others from the Albert Einstein College of Medicine in the Bronx, New York, report that fetal liver also has pericytes that express the same cell surface molecules as the ones in bone marrow, and the removal of these cells reduces the numbers of and proliferative status of fetal liver HSCs.
Now we have a conundrum, because the same cells in bone marrow do not drive HSC proliferation, but instead drive HSC quiescence. What gives? Khan and others showed that the fetal liver pericytes are part of an expanding and constantly remodeling blood system in the liver and this growing, dynamic environment fosters a proliferative behavior in the fetal HSCs.
When umbilical inlet is closed at birth, the liver pericytes stop expressing Nestin and NG2, which drives the HSCs from the fetal liver to the other place were such molecules are found in abundance – the bone marrow.
These models give us a better view of the inner workings of HSC differentiation. Since HSC transplantation is one of the mainstays of leukemia and lymphoma treatment, understanding HSC biology more perfectly will certainly yield clinical pay dirt in the future.