1. Before the Hemogenic Endothelium: The Hemangioblast Theory
The existence of progenitor cells capable of differentiating into both endothelial and hematopoietic cells (“hemangioblasts”) was initially suggested by Florence Sabin in 1917, based on microscopical observations which noted the close physical proximity of emerging endothelial and red blood cells in the yolk sac (YS) of chicken embryos. The actual “hemangioblast” denomination was coined by Murray in 1932
[1] and referred to a mass of cells derived from the primitive streak mesoderm, containing precursors of endothelium and blood cells. The first experimental data that suggested the presence of a hemangioblast did not arrive until the 1990s when Gordon Keller and colleagues identified a clonal mesodermal precursor for blood and endothelium in embryonic stem (ES) cell differentiation cultures, the blast colony-forming cell (BL-CFC)
[2]. This work introduced a paradigm shift in which the hemangioblast no longer represented a population of cells as initially implied by Murray, but a bipotent clonal progenitor for blood and endothelium. The same laboratory went on to provide in vivo evidence of the presence of hemangioblasts in mouse embryos, first detected at the mid-streak stage of gastrulation (E6.75)
[3]. Labeling of cells fated to endothelial and hematopoietic lineages in zebrafish was also interpreted as in vivo support for the existence of the hemangioblast
[4]. However, in the same years, several experimental evidences started lending support to another theory for the origin of blood cells in the embryo: the hemogenic endothelium (discussed in the following section), which partially clashed with the hemangioblast
[5]. After few years, a study based again on ES cultures tried reconciling the two apparently conflicting hypotheses by providing a model in which the hemangioblast gives rise to blood cells through a hemogenic endothelial intermediate
[6]. More recent work using lineage tracing of individual cells in the mouse epiblast, primitive streak, and early YS provided little support for the in vivo existence of hemangioblasts as the majority of labeled clones contained either blood or endothelial cells, but very few clones harbored both lineages
[7]. These results suggest that precursors for primitive blood and YS endothelium are already specified before gastrulation. Comparison of these lineage tracing data with the BL-CFC model, which was supported primarily by ES cultures, serves as a reminder that in vitro potential does not always equal in vivo fate, which is often more restricted. The hemangioblast may therefore represent the in vitro counterpart of an early mesodermal precursor cell present prior to the commitment to the hematopoietic lineage, as opposed to an actual bipotent progenitor persisting in vivo until midgestation. Further investigation and new experimental approaches are needed to clarify this issue.
2. Heterogeneity of the Hemogenic Endothelium
Several landmark studies in the 1990s provided strong evidence that in vertebrates definitive repopulating HSCs are autonomously generated intra-embryonically within the developing Aorta-Gonad-Mesonephros region (AGM) at midgestation, prior to their appearance in other hematopoietic organs
[8][9][10][11][12][13][14]. HSCs and their precursors (pro- and pre-HSCs) are initially found within clusters of hematopoietic cells observed in the aorta and major embryonic arteries
[15][16][17][18][19]. Similar to the putative hemangioblasts, these peculiar cell clusters were first identified microscopically in the beginning of the last century, and their close association to the vascular endothelium lent support to the idea that the embryonic endothelium has a hemogenic capacity (reviewed in
[20]). Later lineage tracing studies in the chick and mouse models confirmed that this was indeed the case
[21][22]. The strongest support for an endothelial origin of hematopoietic cells came from more recent live imaging studies that visualized the transition of endothelium into blood in real-time, both in vitro
[23] and in/ex vivo
[24][25][26]. The process of endothelial cells acquiring a hematopoietic fate has been termed endothelial-to-hematopoietic transition (EHT). Cells undergoing this process experience major morphological changes independent of cell division, including breakage of tight junctions with adjacent endothelial cells, which result in their acquisition of a round shape and the expression of a hematopoietic program. Importantly, hemogenic endothelial cells are not bipotent but rather already committed to an hematopoietic fate
[27][28]. How EHT is regulated at the molecular level is complex and still incompletely understood, though the requirement for the master hematopoietic transcription factor Runx1 is well established (reviewed in
[29]). Although EHT has been mostly studied in the context of the dorsal aorta because of it being the primary site of HSC generation, it is now clear that a similar process also takes place in other locations, such as YS, placenta, and possibly other sites
[30][31]. Hemogenic endothelium (HE) is inherently transient and it is present only during a limited timeframe in embryonic development, although a potential wave with limited hematopoietic contribution has been described also in the BM of late fetus/young adults
[32]. HE represents a common source of various types of stem and progenitor cells generated in multiple anatomical locations at different times during embryogenesis, including progenitors with broad mesodermal potential
[33]. This poses an important question: are prospective hematopoietic cells intrinsically fated to become a specific type of stem or progenitor cell already at the HE stage, or do they become specified only later, in response to extrinsic cues from the microenvironment? In other words, when exactly is heterogeneity generated during developmental hematopoiesis? This is a basic research question which carries important translational relevance. Achieving a better understanding of the embryonic hematopoietic system roadmap will inform experimental strategies for production of hematopoietic stem and progenitor cells (HSPCs) ex vivo, which have recently been reported, though their efficiency is still relatively low
[34][35].
There is evidence in support of the notion that HE is intrinsically heterogeneous. Clonal assays of individual hemogenic endothelial cells demonstrated very high heterogeneity in their ex vivo output at a given developmental stage
[36]. HE could therefore consist of a mix of precursors endowed with distinct potential, and/or that differentiation process proceeds in a highly asynchronous way. These two hypotheses do not exclude each other–and, in fact, there is experimental evidence supporting both. As mentioned above, the formation of HSCs requires the transcription factor Runx1, but also its non-DNA binding partner core binding factor β (CBFβ). The same requirement is observed for the generation of Erythro-Myeloid Progenitors (EMPs), multipotent progenitors distinct and appearing earlier than HSCs. A first demonstration that EMPs and HSCs differentiate from distinct populations of hemogenic endothelial cells came from a study using a CBFβ rescue strategy, which showed that when CBFβ was reintroduced in a null background under the control of two different tissue specific expression transgenes, this caused the alternate rescue of EMPs or HSCs, but not of both
[37]. A recent study supported and extended these findings
[38]. The authors employed single cell index sorting of HE combined with a co-culture strategy to show that rare HSC-competent HE and the relatively more abundant progenitor-restricted HE can be discriminated by expression of the chemokine receptor CXCRSingle cell RNA-Seq (scRNA-Seq) of the two HE types isolated from E9–E9.5 mouse embryos showed that the HSC competent HE already expresses many of the genes known to be enriched in E11.5 AGM HSCs
[38][39]. Accordingly, CXCR4 lineage tracing has been recently used to mark a contribution of intra-but not extra-embryonic HE to innate lymphoid cells
[40] and HSC-derived monocytes from microglia and other tissue-resident macrophages (TRM)
[41]. Recently, some markers enriched in the YS HE have also been identified, such as Stab2 and Lyve1
[42][43], although the latter was shown not to be exclusively expressed in the YS
[44]. These data support the concept that HE cells are already “primed” to become specific types of stem or progenitor cells when they are still undergoing EHT. What are the signals that instruct the generation of heterogeneity in prospective hemogenic endothelial cells?
It is now clear that different signaling pathways are involved in EMP and HSC formation (
Table 1). For example, EMP formation from HE in the YS vascular plexus does not require Notch signaling, whereas HSC production is strictly dependent on Notch
[45][46][47]. The Notch pathway plays a fundamental role in the arterialization of HE, and the arterial identity of HE is a mandatory condition for the establishment of HSC generation
[48][49], but not for EMPs in the YS, as they emerge from both arterial and venous endothelium
[31]. The arterial identity of HE appears to be important not only for HSC emergence, but more in general for the appearance of progenitors with lymphoid potential
[50]. If the dependence on Notch pathway can be considered a discriminant between HSC-fated HE in the embryo proper and the HE generating EMPs in the YS, the WNT canonical pathway is instead a common regulator of both programs
[31]. Other signaling pathways appear to be differentially required in distinct types of HE. In particular, the emergence of HSCs requires mechanistic stimuli provided by the onset of circulation and blood flow, which result in the modulation of several important pathways, among which nitric oxide (NO)
[51], cAMP and BMP
[52][53], Rho-Yap
[54] and metabolic pathways
[55]. In contrast, EMP emergence is blood flow-independent
[56]. Interestingly, ex vivo modulation of hypoxia/glycolysis yielded opposite effects in embryo proper and YS cells, with a decrease of the hematopoietic output in the former and an increase in the latter
[55]. Hepatic leukemia factor (Hlf) is another discriminating factor between EMPs and HSCs
[57]. As determined by analysis of a transgenic reporter mouse model, Hlf is expressed by virtually all intra-embryonic c-Kit
+ cells, including HE and hematopoietic clusters in the dorsal aorta and vitelline artery, but not by non-hemogenic endothelium and E9.5 YS EMPs. Intriguingly, Hlf appears to be also expressed by some c-Kit clusters in the E10.5 YS. These results are another strong suggestion that the acquisition of the HSC or progenitor fate takes place at an early stage during hematopoietic differentiation.
Many studies have very recently employed scRNA-Seq and newer variants of this technique such as spatial transcriptomics to delve deeper into the study of the molecular determinants of EHT and to gain insight into the cellular and molecular heterogeneity of HE
[39][58][59][60][61][62][63][64], mostly in animal models but some also in human
[65][66]. These studies mainly focused on intra-embryonic EHT because of its obvious connection with HSC generation, and collectively highlighted how (pro-/pre-)HSC development proceeds in a fairly asynchronous way, with cells at different stages of maturation coexisting in the same embryo. scRNA-seq allowed to describe the differentiation trajectory of EHT in greater detail by revealing previously unrecognized intermediate stages, such as the “pre-HE”, which may represent a developmental bottleneck in which many important molecular pathways are active
[58]. Moreover, these studies confirmed the heterogeneity within the intra-aortic clusters population by highlighting the presence of lympho-myeloid biased progenitors, which can be distinguished molecularly
[58]; however, the precise cellular relationships between these different progenitor types and the HE are still unclear. Single cell studies not only identified unexpected differences, but also similarities. Indeed, non-HE and HE show surprisingly similar transcriptomes with only a few differentially expressed genes (DEG)
[59]; this was confirmed by another study showing that HE cells clustered mainly according to the tissue of origin and to a lesser extent according to hemogenic/non hemogenic identity
[67]. In a similar way, clusters in the ventral and dorsal aspects of the aorta also exhibited few global transcriptional differences
[59]. Transcriptomics studies have recently been instrumental in identifying microenvironmental cues as driving forces of heterogeneity in the embryonic hematopoietic niche
[62][65][68]. These and other studies
[69][70][71] showed that secreted factors (BMP, Kit ligand, Noggin as ventral signals; Shh as a dorsal signal are among the most well-characterized) contribute to generating the dorso/ventral polarization of signals that places the embryonic (pre-)HSC niche in the ventral aspect of the aorta.