Stem cells are a unique cell type with the remarkable ability to serve as building blocks for all cell types in the body. The studies of Till and McCulloch were instrumental in shaping our understanding of stem cells, and their work continues to have a lasting impact on both basic research and clinical applications in the field of regenerative medicine and beyond McCulloch and Till (1). Over the past half-century, researchers have discovered combinations of physical properties, cell-surface markers, and devised assays to measure stem cell function that allow for the identification and separation of stem cell populations in adult tissues like the skin, intestinal epithelium, brain, and the hematopoietic system.

The best characterized stem cells are the hematopoietic stem cells (HSCs), which reside primarily within the bone marrow (BM) during adulthood. Murine HSCs are distinguished by the absence of cell-surface markers present on lineage (Lin)-committed hematopoietic cells. Additionally, these cells express high levels of the stem cell factor receptor (c-kit, CD117) (2) and stem-cell antigen 1 (Sca1) (3, 4); this Lin⁻c-kit⁺Sca1⁺ subset is commonly referred to as LSK cells (4, 5).

HSCs account for 0.00125–0.00425% (12–42 cells per million) of whole BM in mice (6). These HSCs are comprised of two main types: Long-term HSCs (LT-HSCs) or dormant HSCs and short-term HSCs (ST-HSCs) or activated HSCs. These populations of HSCs can be differentiated by expression of the anti-adhesive sialomucin CD34 (7); mouse LT-HSCs do not express CD34 until they become activated ST-HSCs (6, 8, 9). Other methods of HSC isolation often begin with enrichment of LSK cells, a heterogeneous population of hematopoietic precursors, with only a small subset consisting of LT-HSCs. According to flow cytometry analysis, the frequency of LT-HSCs within the LSK population is approximately 10% (10, 11). However, in vivo analyses may vary in the frequencies of long-term reconstituting cells measured among LSK cells: frequencies as low as 2% (12) or as high as 20% (13) having been reported. Frequency variation in the BM progenitor pool may be related to mouse age (14), sex (15), and strain variation (16, 17). For example, the frequency of LT-HSCs cells increases with age (10). In addition to CD34, other markers of HSCs that have been used to differentiate LSK cells are the SLAM family receptors (CD150 and CD48) (13), CD90 (5), and the capacity of HSCs to remove intravital dyes like Rhodamine-123 (3, 18).

In humans, the HSC population, similar to mice, is characterized by the absence of maturity markers and the expression of CD34 (Lin⁻CD34⁺) (19). Furthermore, HSCs are found to be more abundant among CD34⁺ cells expressing CD4 (20), CD90 (21), CD133 (22), and EPCR (23), while lacking expression of CD38 (24–26), CD45RA (27), and CD71 (28). Similar to murine HSCs, there exists a rare subset of HSCs characterized by the absence of an extensive lineage marker panel and CD34 (29). These CD34⁻ HSCs have been identified in human cord blood and are known to express GPI-80 (30), CD90 (31), CD93 (31), and CD133 (32). While transplant success using CD34 selection suggests the presence of functional HSCs within the CD34⁺ fractions of neonatal and adult cells, experimental evidence indicates that pre-natal CD34⁻ HSCs may be more primitive than their CD34⁺ counterparts (31, 33).

During embryonic development, the earliest hematopoietic progenitor cells emerge in the yolk sac, appearing around day 17 in humans and embryonic day 7.5 (E7.5) in mice (34). These progenitors, originating from the mesodermal layer, form clusters known as “blood islands” within the yolk sac, closely associated with endothelial and hematopoietic cells. Human yolk sac development progresses through three phases: 1) a formative period, 2) a functional period, and 3) a period of regression. During the functional phase, which typically occurs during the second stage, the yolk sac comprises cells of both mesodermal and endodermal origin, housing endothelial and early hematopoietic cells (35). In humans, at 3–4 post-conception week (PCWs), the first generation of hematopoietic cells proliferate in the yolk sac and extraembryonic mesenchyme, primarily consisting of “primitive” erythroblasts (megaloblasts) which are present in circulating blood from 4 PCWs onward (36). Shortly after the initiation of blood island formation and primitive hematopoietic progenitor generation, endothelial cells within the para-aortic splanchnopleura region begin producing definitive HSCs. In the mouse, this process typically takes place around E9.5 (37), while in humans, it occurs briefly around Carnegie Stages (CS) 9–12 which is equivalent to 19–27 days of embryonic development (38). Subsequently, the para-aortic splanchnopleura region gives rise to most tissues in the aorta-gonad-mesonephros (AGM) (37–39), which emerges approximately after E10.5 in mouse and after the 27th day of pregnancy in humans as an active site of hematopoiesis (38, 40–42).

In mice, HSCs are present in the placenta at the same time as they are in the AGM region (43). Similarly, hematopoietic progenitors are present in the human placenta at approximately 5–6 PCWs (CS 14–16) (44–46). Human HSCs capable of long-term multilineage engraftment in immunodeficient mice were detected from placenta as early as 6 weeks of gestation (CS 17) (45). The human chorion, an extra-embryonic tissue that shares a developmental origin with placenta but is much less vascularized, also contains hematopoietic precursors and HSCs at 5–6 weeks of gestation capable of long-term reconstitution (46, 47). Hematopoietic progenitors in the early stages of placental/chorionic development may originate from extra-embryonic sources. A recent study demonstrated that primitive erythromyeloid progenitors in the placenta are distinct from definitive hematopoietic stem cells due to their absence of HLF expression and their limited lineage commitment primarily to the erythromyeloid lineage (48).

As embryonic development progresses, the liver becomes another important hematopoietic niche. Intraembryonic HSCs are located within the fetal liver of rodents (49) and humans (50) shortly after the liver has formed. It is known that HSCs migrate from the AGM to fetal liver, although the exact migration pathways are still not fully understood (51) it is known to be guided in response to stromal CXCL12 (52).

In mice, the liver begins to form as a diverticulum from the floor of the embryonic gut around E9 (53). Subsequently, the murine liver bud experiences accelerated growth as it becomes vascularized (54). It is believed that primitive progenitors from the yolk sac migrate to the fetal liver through the vitelline vein around E11 (55). By this stage, the murine liver harbors primitive erythromyeloid progenitors derived from the yolk sac (56). Additionally, AGM-derived HSCs begin to populate the murine fetal liver starting at E11, coinciding with maximal AGM activity (57). Subsequently, fetal liver cells at E12 undergo a significant expansion, up to 38-fold, and exhibit long-term reconstitution capability when transplanted into an irradiated mouse, indicating the presence of definitive HSCs by E12, as opposed to E10 and E11 cells, which did not result in reconstitution (58). This observation aligns with the decrease in HSCs observed in the AGM region by E12 (57).

An early and noteworthy investigation illustrated various hematopoietic cell lineages within the human embryonic liver as early as 5 PCW (CS 14–15), delineating their morphological characteristics (59). However, the cells may have comprised circulating primitive erythro-myeloid progenitors and their progeny, as they do not express CD34 and have been observed as early as 23 days of development (CS 10) (50). Recent studies indicate that the liver is initially populated by molecularly defined HSCs around 6 PCWs (CS 16) (60). Although CD34⁺ progenitors are detected in the human fetal liver as early as 30 days of development (CS 13) (50), several investigations indicate that transplantable HSCs are first present in the embryonic liver shortly thereafter, at 5–6 PCWs (CS 14–16) (61–64).

From this time forward until midgestation, the human liver is the primary intra-embryonic hematopoietic organ (65). Erythropoiesis dominates the hematopoietic output of the liver in order to meet the demands of an expanding blood volume (66, 67). Indeed, flow cytometric analysis of whole fetal liver indicate that CD235a in vitro erythroid cells are the dominant cell type in the liver during midgestation (68). In addition to erythropoiesis, liver hematopoiesis includes elements of myeloid and lymphoid precursors (26, 69–71). Single-cell transcriptome profiling of human fetal liver further confirmed the presence of HSCs and MPPs with the capability to generate functional innate lymphoid cells, T-, and B-lymphocytes, natural killer (NK), and CD34⁺ myeloid progenitors (72).

Interestingly, some long-lived myeloid cells are found in adult tissues like the lungs or peritoneal cavity. These so-called “tissue resident macrophages” emerge from the yolk-sac instead of progenitors derived from the AGM Gomez (56) or fetal liver (73). Also, some lymphoid populations like the B-1 B cells originate in the yolk-sac, which represent part of the humoral immune response but not the classic adaptive immune system. Similarly to tissue resident macrophages, this subset of B cells are rarely found circulating in the blood stream (74), and undergo self-renewal in the periphery (75). It is noteworthy that the nature of these yolk-sac derived leukocytes have more in common with innate immunity than adaptive immunity as reviewed in (76). While the presence of B-1 cells in humans has been controversial (74).However, a recent comprehensive comparative single-cell sequencing analysis of human fetal and adult tissues has presented compelling evidence, according to these authors, supporting the origin of human fetal B-1 cells (77).

Finally, the definitive hematopoietic organ, the BM, becomes hematopoietic at E15.5 in mouse (78) and at 10.5 PCW in humans (79). Although the first HPSCs in the murine bone marrow are found as early as E15.5 and E16 (80) the amount of LT-HSCs found a these stages is limited and steadily increasing until birth (78). The mouse BM is seeded by HSCs originating in the fetal liver (81). In humans fetal BM is also colonized by circulating hematopoietic precursors derived from fetal liver found in relative abundance during early gestational periods compared to neonates (82, 83). Human fetal BM is first found in the long bones of the fetus as these bones are the first to become large enough to contain a marrow. A recent investigation revealed that although HSCs with defined phenotypes are present in the human BM at 10 PCW, they attain full functionality only after reaching 12 PCWs (84). The BM becomes the dominant site of hematopoiesis early in the second half of gestation as the marrow increases in size relative to the liver.

It is worth mention that the fetal spleen is an important hematopoietic tissue in mice bridging the peri-natal period when hematopoiesis diminishes in the liver and before the BM forms. Humans and mice differ in this regard as the spleen contains but an insignificant number of hematopoietic precursors at any time during gestation (85, 86). Figure 1 illustrates the temporal migration of HSCs in human and mouse as well as the tissues where hematopoiesis is normally found.

EMH can occur in response to both homeostatic and pathologic conditions, such as acute infection or inflammation. Organs like the spleen and liver can start producing blood cells due to local production of hematopoietic growth factors like GM-CSF, TNFα, and IL-3 (87, 88). In chronic pathologies, the BM may lose its ability to function effectively, leading to increased reliance on organs like the liver, spleen, or even the lungs for blood cell production. In mice, splenic EMH readily occur when the capacity of the BM to produce blood cells is exceeded, recalling the significant role the spleen plays during the perinatal period in hematopoietic development. However, large mammals possessing substantial bone marrow reserves may not demonstrate a comparable propensity to transition blood cell production from the bone marrow to the spleen.

Myelofibrosis, diffuse osseous metastatic disease, leukemia, sickle cell disease, and thalassemia are among the primary factors contributing to human extramedullary hematopoiesis. (89). Many reports have found the presence of hematopoietic progenitors in extramedullary organs due to defective hematopoiesis in the BM due to any of these diseases ( Table 1 ). Interestingly, some cases of EMH are found after inflammatory conditions such as inflamed joints (115), or lung infection (116, 117). Table 1 summarizes reports were EMH is found secondary to disease.

Pathological EMH is a secondary problem diagnosed principally with radiological studies, computed tomography (CT) or magnetic resonance imaging (MRI) ( Table 1 ). Following resection, the anomalous mass is examined by pathologists, the most common techniques used to determine a hematological signature include staining for glycophorin (GPA, CD235a) for erythroid progenitors and Giemsa (118) or H&E (Hematoxylin and Eosin) staining for morphology (89). Sometimes immunohistochemistry (IHC) (104, 106, 111) or flow cytometry (FC) (98) with antibodies against known progenitor and mature immune cells or colloidal scans (93, 105) are used.

Although limited, there are some reports showing that HSPCs are present in tissues outside the BM under homeostatic conditions both in human and animal models. Furthermore, it has been established that these cells exhibit re-populating properties similar to those found in the BM. For example, tissues like the small intestine in humans contain hematopoietic progenitors expressing CD34 and CD45 (119). Later work showed not only a distinct phenotype but also confirmed the presence of hematopoietic colony-forming cells using clonal analysis in vitro (CFU-c). Additionally, it demonstrated the potential for long-term chimerism in individuals receiving intestinal allografts (120).

Another potential host for hematopoietic progenitors in humans is the stromal-vascular fraction (SVF) of adipose tissue. In pathological conditions such as chronic myeloid leukemia, adipose tissue has the potential to transform into a hematopoietic niche. In this context, leukemic cells express a fatty acid transporter to enhance their communication with the adipose microenvironment and evade chemotherapy (121). Under homeostatic conditions, this tissue harbors a subset highly enriched in CD34⁺ and VEGFR2⁺ (KDR) cells with absent CD45 expression. Interestingly, despite the lack of CD45 expression, these cells were able to originate hematopoietic colonies in a CFU-c assay (122). Notably, murine SVF within adipose tissue contains similar progenitors defined by a LSK phenotype and are capable of long-term multi-lineage reconstitution when transplanted into a recipient (123).

Among other organs known to exhibit pathology associated EMH, the lungs are included. While there is no conclusive evidence confirming the presence of HSCs in human lungs, there is evidence supporting their existence in mice. Initial observations in mouse lungs have revealed the occurrence of a cell subset, VEGFR2⁺CD31⁺ cells, which can produce endothelial cells, smooth muscle cells and form hematopoietic colonies in vitro (124). Anatomical descriptions of lung CD34⁺Sca1⁺-kit⁺ cells showed these cells are located in close proximity to blood vessels, and possibly belong to a unknown subset of hematopoietic progenitors (125). Furthermore it has been reported that mouse lungs are active hematopoietic organs, indeed they account for 50% of platelet production and are a source of hematopoietic progenitor and HSCs with a phenotype and repopulating activity similar to its counterparts in the murine BM (116).

Since the lungs are not classically known to be hematopoietic niches, it has been hypothesized that lung vasculature may function as an adaptive niche for HSCs, principally ST-HSCs, since LT-HSCs were not found by Lefrançais et al. in the lungs (116, 126). The lungs harbor more than 50 different cell types (127), the most likely subset to be nurturing HSCs are pericytes or distinct blood vessels, further work is needed to elucidate this question.

Besides the lungs, reports have indicated that the liver and spleen also display EMH associated with pathologies. Over the last decade, numerous studies have revealed that both organs harbor functional hematopoietic progenitors along with the necessary microenvironment to sustain them. A recent report has indicated the presence of an LSK population in the mouse spleen, along with its ability to expand when the Txl1 gene is overexpressed in splenic stromal cells (128). As for hepatic hematopoiesis, there is evidence that supports that liver sinusoid endothelial cells provide an appropriate hematopoietic environment that supports the proliferation and differentiation of hematopoietic progenitors in the mouse (129). Moreover, adult murine liver was shown to support human HSC engraftment, showing long-term multilineage hematopoiesis (130). In addition, donor chimerism after liver transplantation is a common finding, this comes from having a significant amount of donor-derived hematopoietic cells in the liver graft (131, 132). Liver-derived hematopoietic engraftment has even been documented to lead to a complete blood group change in the recipient (133).

In summary, although EMH has been observed under both homeostatic and emergency conditions, it remains uncertain whether extramedullary precursors are remnants from early ontogeny or if they come from and are constantly replenished from the BM ( Figure 2 ). Consequently, we wish to highlight the importance of thoroughly researching all mechanisms involved in these phenomena.

The HSCs in the BM require a special microenvironment (niche) to maintain stemness. In the BM, two types of niches have been identified. First, the endosteal niche located in the inner cellular lining within bone cavity in contact with BM of trabecular bone, which is called endosteum. This region comprises specialized osteoblasts, CXCL12-abundant reticular cells (CARs), osteoclasts, and mesenchymal stromal cells (MSCs). Secondly, we have the vascular niche where HSCs are found next to CARs adjacent to BM sinusoids (134).

The molecules secreted from niche cells promote HSC maintenance. These include stem cell factor (SCF, KL) (135–136), CXCL12 (137, 138), angiopoietin 1 (139), and thrombopoietin (TPO) (140). Further detailed analysis of the BM niche showed that approximately 80% of dividing and non-dividing hematopoietic progenitors reside in close proximity to a sinusoid and nearly all HSCs are in contact with leptin receptor and CXCL12 high expressing niche cells (LEPR⁺CXCL12 ^hi ). The rest of the BM hematopoietic progenitors are located near arterioles (10%) and transition zones (10%) with only a few HSCs residing adjacent to the endosteum (141). In general, it is well-established that BM HSCs primarily inhabit these perivascular environments.

Other molecules are produced by a wide variety of cells that are essential for the regulation of hematopoiesis such as: SCF, notch ligands, bone morphogenic protein, transforming growth factor β (TGF-β), TPO, fibroblast growth factors, and insulin-like growth factor 2 as reviewed by (142). These molecules are necessary for maintaining and activating the resident HSCs on demand.

On the other hand, HSC and progenitor proliferation and differentiation is primarily regulated by cytokines like erythropoietin, granulocyte-monocyte colony-stimulating factor (GM-CSF) (143), α-interferon (INF-α) (144), and γ-interferon (INF-γ) (145). Hematopoiesis activated by inflammatory signals is often referred to as “emergency hematopoiesis” and it has been widely studied in pathogen associated inflammation (146).

Furthermore, factors produced during inflammatory processes in the mouse stimulate hematopoietic progenitors, some Toll-like receptor (TLR) ligands like Pam₃CSK₄ and lipopolysaccharide (LPS) promotes myelopoiesis in vitro (147). IFN-γ promotes activation and proliferation of HSCs both in vitro and in vivo (145). TLR3 ligand, poly(I:C), and its downstream effector INF-α also promotes the proliferation of dormant HSCs in vivo but chronic treatment caused the opposite effect (144). Other pro-inflammatory cytokines like G-CSF, interleukin (IL)-6, IL-11, and IL-12 promote not only HSC proliferation and differentiation but also HSC mobilization from BM to peripheral blood (148). The mobilization of HSCs is a mechanism by which sites of EMH may be seeded with hematopoietic precursors.

There are multiple cell sources of hematopoietic growth factors beyond innate immune cells that may contribute to EMH. In addition to the immune response, various cell types assist in the restoration and replenishment of hematopoietic niches. In vitro experiments involving the cultivation of HSCs with endothelial cells (ECs) or their soluble products have demonstrated that factors secreted by ECs are essential for enhancing the numbers of CFU-c (149). These factors are only partly defined. It was described that ECs are the main source of G-CSF in BM during inflammation, a molecule known to promote myelopoiesis (150). ECs also secrete IFN-γ and TNF-α, which promote MSC production of IL-13, another hematopoietic regulator in mice (151, 152). BM derived MSCs are affected by infection and inflammation, they respond to cytotoxic CD8⁺ T effector-cell signals and produce high levels of IL-6, thus promoting myelopoiesis (153). BM-MSCs primed by IFN-γ secretes hepatocyte growth factor, prostaglandin E2, that are known by its immune-suppressive effect but also stimulates growth of hematopoietic progenitors (154, 155). While BM niche molecules are known and widely described, our understanding of hematopoietic regulation outside the BM remains less understood.

Hematopoietic progenitors respond to multiple signals, both in steady state and during inflammation, some of these signals are secreted molecules derived by numerous cell types throughout life as described before (156). Furthermore, the same pro-inflammatory molecules that promotes blood progenitor growth and mobilization in BM are linked to EMH in adults, such as GM-CSF, G-CSF, IL-3, IL-6 and IFN-γ (87, 157). Unfortunately, it is unknown which molecules outside the BM niche are specifically regulating the hematopoietic progenitors found in the extramedullary sites. While there are a number of potential candidates, some of them are discussed in more detail below:

In summary, the significance of extramedullary hematopoiesis (EMH) extends from embryonic development to adulthood, showcasing the dynamic changes in hematopoietic locations. Throughout development, the regulation of hematopoiesis by crucial cell types and molecules shapes the transition from EMH to bone marrow hematopoiesis. Some of these developmental EMH sites may maintain conditions conducive to hosting primitive and definitive hematopoietic progenitors. Recent research revealed that hematopoiesis originating from embryonic hemogenic ECs continues into adulthood, as evidenced even in 12-month-old mice (187). Notably, this research demonstrated that the majority of MPPs, T cells, and B-2 cells within the first month after birth were fetal EC-derived. Whether a similar phenomenon occurs in humans remains to be elucidated. Thus, understanding the origin of HSPCs as well as the variances in niches between normal and aberrant sites is vital, as it influences our comprehension of hematopoietic cell replenishment mechanisms. While BM niche is extensively characterized, with knowledge about cellular components and soluble factors, much remains unknown about the factors influencing adult EMH sites.

Most stem cell niches are situated near blood vessels and stromal cells (188–190). HSCs are no exception (191). Considering the nature of EMH sites, it is highly likely that the cells supporting these hematopoietic precursors are primarily endothelial and stromal cells. There is evidence showing that the vascular fraction of murine extramedullary adult organs like brain, heart, lung and liver are able to maintain LSK cells in vitro (192). In this context, the reliance on endothelial and stromal cells within EMH sites underscores the pivotal role of these cellular components in sustaining hematopoietic potential beyond the bone marrow.

Nevertheless, there exist numerous inquiries that require further resolution. Are these possible niches similar to one another? What distinctions exist between the BM vascular niche and the vascular fraction of the organs displaying typical EMH? Additional research is needed to uncover the mechanisms that regulate the EMH niche-like functions. Undoubtedly, it is crucial to assess and contrast the various cell types participating in EMH with those of the perivascular and/or endosteal cells found in the BM niche.

The more recent models for investigating native hematopoiesis, exemplified by studies such as (193), hold promise for significant advancements in our understanding of hematopoietic processes. These models often incorporate advanced technologies, such as single-cell sequencing and sophisticated imaging techniques, allowing for a finer resolution of cellular interactions and dynamics within the hematopoietic microenvironment.

For instance, in a study where a high-resolution characterization of the bone marrow (BM) niche was conducted in both healthy conditions and acute myeloid leukemia (AML), researchers established a singlecell gene expression database encompassing 339,381 BM cells. The findings revealed significant alterations in cell type proportions and gene expression in AML, suggesting a comprehensive disruption of the entire niche (194). Such high-resolution analyses have the potential to enhance our comprehension of the pathological and homeostatic distinctions within extramedullary niches, spanning from development to adulthood.

GR-T: Conceptualization, Investigation, Writing – original draft, Writing – review & editing. MM: Funding acquisition, Supervision, Writing – review & editing. RV-R: Conceptualization, Supervision, Writing – review & editing.