Niches that regulate stem cells and hematopoiesis in adult bone marrow

Abstract

In mammals, hematopoietic stem cells (HSCs) engage in hematopoiesis throughout adult life within the bone marrow where they produce the mature cells necessary to maintain blood cell counts and immune function. In the bone marrow and spleen, HSCs are sustained in perivascular niches (microenvironments) associated with sinusoidal blood vessels - specialized veins found only in hematopoietic tissues. Endothelial cells and perivascular Leptin Receptor⁺ stromal cells produce the known factors required to maintain HSCs and many restricted progenitors in the bone marrow. Various other cells synthesize factors that maintain other restricted progenitors or modulate HSC or niche function. Recent studies identified new markers that resolve some of the heterogeneity among stromal cells and refine the localization of restricted progenitor niches. Other recent studies identified ways in which niches regulate HSC function and hematopoiesis beyond growth factors. We summarize the current understanding of hematopoietic niches, review recent progress, and identify important unresolved questions.

Introduction

Red blood cells, white blood cells, and platelets must be produced throughout life in the bone marrow to sustain blood cell counts and immune function (Figure 1). Hematopoiesis is sustained by HSCs, which give rise to a series of multipotent and restricted hematopoietic progenitors (Busch et al., 2015; Sawai et al., 2016; Rodriguez-Fraticelli et al., 2018; Sawen et al., 2018). As a result of the proliferation that occurs at each level of the hematopoiesis hierarchy, humans are estimated to produce more than 100 billion new hematopoietic cells each day in the bone marrow.

Figure 1. Schematic representation of adult hematopoiesis.

Rare HSCs give rise to multipotent progenitors and restricted hematopoietic progenitors that differentiate into several different types of blood cells. The approximate frequencies of each cell population in the bone marrow and thymus are shown. It is uncertain if HSCs undergo progressive lineage restriction in a stereotypical manner, as diagrammed, or whether lineage restriction occurs in a more flexible and variable way (Yamamoto et al., 2013; Paul et al., 2015; Notta et al., 2016; Velten et al., 2017; Carrelha et al., 2018; Hofer and Rodewald, 2018; Rodriguez-Fraticelli et al., 2018).

HSCs reside in specialized microenvironments, or niches, within hematopoietic tissues that promote their maintenance and regulate the production of hematopoietic cells (Morrison and Scadden, 2014). HSC niches are perisinusoidal (Kiel et al., 2005; Sugiyama et al., 2006; Nombela-Arrieta et al., 2013; Acar et al., 2015; Chen et al., 2016; Christodoulou et al., 2020; Kokkaliaris et al., 2020) (see Table 1 and Figure 2), with Leptin Receptor⁺ (LepR⁺) mesenchymal stromal cells and endothelial cells synthesizing the factors required for HSC maintenance in the bone marrow (Ding et al., 2012; Ding and Morrison, 2013; Greenbaum et al., 2013; Oguro et al., 2013).

Table 1: Studies that assessed the localization of HSCs in the bone marrow with markers that give high levels of HSC purity.

All of these markers identify HSCs that are nearly all quiescent.

Study	Markers used to identify HSCs	HSC purity (fraction of cells that give long-term reconstitution of irradiated mice)	HSC location
Kiel et al., 2005	CD150+CD48−CD41−	1 in 2.2	60% perisinusoidal in BM and spleen
Nombela-Arrieta et al., 2013	Lin−Sca1+c-kit+CD48−CD41−/low	Not determined	~90% perisinusoidal
Kunisaki et al., 2013	CD150+CD48−Lin−Sca1+c-kit+	1 in 2.1	Two thirds perisinusoidal, one third periarteriolar
Acar et al., 2015	α-catulin-GFP+c-kit+	1 in 3.1	84% perisinusoidal
Chen et al., 2016	Hoxb5-mCherry+	Not determined	94% perivascular
Sacma et al., 2019	Label-retaining Lin−CD150+CD48−CD41−cells	Not determined	83% perisinusoidal
Christodoulou et al., 2020	Mds1-GFPfl; Flt3-Cre	1 in 9	94% perisinusoidal
Kokkaliaris et al., 2020	α-catulin-GFP+c-kit+ Mds1-GFPfl; Flt3-Cre Labelling-retaining c-kit+ H2B-GFP cells	1 in 3.1 1 in 9 Not determined	~60% perisinusoidal

Figure 2. Bone marrow anatomy and HSC localization.

A) A bisected mouse femur. B) A schematic of landmarks in the bone marrow and HSC and lymphoid progenitor localization. Arterioles carry blood into the bone marrow while sinusoids carry it out. HSCs reside in perisinusoidal niches. LepR⁺ stromal cells and endothelial cells synthesize factors required for HSC maintenance (SCF, Cxcl12, and Pleiotrophin). There are many additional stromal cells and hematopoietic cells in the bone marrow that directly or indirectly regulate aspects of HSC and niche function. There are also factors critical for HSC maintenance that are not synthesized in the bone marrow, such as Thrombopoietin, which is made by the liver and transported by the blood into the bone marrow (Decker et al., 2018). There are multiple niches for early lymphoid progenitors in the bone marrow. One subset resides near the endosteum where they depend upon Cxcl12 from osteoblasts (Ding and Morrison, 2013; Greenbaum et al., 2013). Another subset resides near arterioles where they depend upon SCF from LepR⁺Osteolectin⁺ stromal cells (Pinho et al., 2018; Shen et al., 2021). Some early lymphoid progenitors depend upon IL-7 from LepR⁺ cells and it is unclear whether they reside in periarteriolar or perisinusoidal niches (Cordeiro Gomes et al., 2016). It is uncertain to what extent these subsets of lymphoid progenitors overlap with each other or if they represent distinct stages of lymphoid development.

Restricted progenitors also reside within specialized niches in hematopoietic tissues. Like HSCs, many restricted progenitors in the bone marrow derive factors for their maintenance from LepR⁺ cells (Cordeiro Gomes et al., 2016; Comazzetto et al., 2019). However, some restricted progenitors derive factors from other cell types, like osteoblasts (Ding and Morrison, 2013; Greenbaum et al., 2013). Some restricted hematopoietic progenitor niches are near sinusoids, like HSCs (Cordeiro Gomes et al., 2016; Comazzetto et al., 2019; Zhang et al., 2021), while others are near arterioles (Shen et al., 2021) or the endosteum (Ding and Morrison, 2013; Greenbaum et al., 2013) and therefore are spatially distinct from HSC niches (Figures 3 and 4). Even the restricted progenitors in perisinusoidal niches may be spatially distinct from perisinusoidal HSCs despite the fact that both depend upon LepR⁺ cells for their maintenance (Zhang et al., 2021). This raises the possibility that there are distinct domains around sinusoids that regulate the maintenance of different kinds of stem and progenitor cells. Fortunately, there is rapid progress toward better understanding the heterogeneity among stromal cells thanks to single cell RNA sequencing (Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020; Matsushita et al., 2020; Zhong et al., 2020).

Figure 3. HSCs and different kinds of myeloid restricted progenitors reside in distinct perisinusoidal niches.

Granulocyte progenitors and monocyte dendritic progenitors reside in spatially distinct niches along sinusoidal blood vessels that do not colocalize with HSCs (Zhang et al., 2021). There may be distinct domains along sinusoidal blood vessels that are specialized for the maintenance of different kinds of stem and progenitor cells.

Figure 4. Multiple waves of mesenchymal stem/progenitor cells give rise to the skeleton and to bone marrow stromal cells during development.

These mesenchymal stem/progenitor cells give rise to LepR⁺ bone marrow stromal cells postnatally (Mizoguchi et al., 2014; Ono et al., 2014b; Pineault et al., 2019). LepR⁺ cells include the fibroblast-colony forming skeletal stem cells in adult bone marrow (Zhou et al., 2014) and give rise to both osteogenic (Shen et al., 2021) and adipogenic progenitors (Zhou et al., 2017)(Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020; Matsushita et al., 2020; Zhong et al., 2020). They are the main source of new osteoblasts and adipocytes that form in adult bone marrow (Mizoguchi et al., 2014; Zhou et al., 2014). There may be multiple distinct lineages of mesenchymal stem/progenitor cells that co-exist in and around the bone marrow during the early postnatal period.

Beyond understanding the precise locations and cellular compositions of restricted progenitor niches and their relationship to the HSC niche, there are several other areas of niche biology with important recent progress. Now that we know the locations and cellular compositions of HSC and some restricted progenitor niches, we can begin to ask what else niches do to regulate stem/progenitor cell function beyond secreting growth factors. Second, niches change in response to injury and these changes are starting to be explored. Within the next 10 years we may have a detailed map of niches within the bone marrow, an understanding of the mechanisms cells use to migrate between these niches as they differentiate, and an understanding of how niches sense cellular depletion and sustain homeostasis.

There are also important aspects of hematopoietic stem and progenitor cell niche biology that fall outside of the scope of our review including niches during fetal development (Mikkola and Orkin, 2006), interactions between leukemia cells and niche cells (Batsivari et al., 2020), heterogeneity among HSCs (Hofer and Rodewald, 2018), and some aspects of the characterization of mesenchymal stem/progenitor cells (Ono et al., 2019).

Endothelial cells and perisinusoidal stromal cells promote HSC maintenance

Bone marrow endothelial cells and stromal cells were first shown to promote the maintenance of hematopoietic stem/progenitor cells in culture. Early studies demonstrated that bone marrow-derived stromal cells could support granulopoiesis and B lymphopoiesis in culture for long periods of time (Dexter et al., 1977; Whitlock and Witte, 1982; Muller-Sieburg et al., 1986). Further selection identified clones with the ability to transiently support HSC maintenance in culture (Whitlock et al., 1987). Endothelial cells also promote HSC maintenance in culture (Li et al., 2004; Butler et al., 2010). Nonetheless, the identity and location of the relevant stromal cells and endothelial cells in the bone marrow in vivo remained to be determined.

In vivo, bone marrow endothelial cells express low levels of growth factors required for HSC maintenance and regeneration, including SCF, Cxcl12, and Pleiotrophin. Conditional deletion of any of these factors from endothelial cells depletes HSCs from the bone marrow (Ding et al., 2012; Ding and Morrison, 2013; Greenbaum et al., 2013; Himburg et al., 2018; Xu et al., 2018; Chen et al., 2019). It remains uncertain whether endothelial cells synthesize these factors entirely in the bone marrow or whether endothelial cells outside of the bone marrow contribute to their production. Endothelial cells are also critical for the regeneration of HSCs and hematopoiesis after myeloablation (Hooper et al., 2009; Butler et al., 2010).

Certain perivascular stromal cells also support the maintenance of HSCs and restricted hematopoietic progenitors in the bone marrow. These perivascular stromal cells can be identified in mouse bone marrow by their expression of LepR (Ding et al., 2012), high levels of Scf(Ding et al., 2012), high levels of Cxcl12 (Sugiyama et al., 2006; Omatsu et al., 2010; Ding and Morrison, 2013; Greenbaum et al., 2013), low levels of the Nestin-GFP transgene (Mendez-Ferrer et al., 2010; Kunisaki et al., 2013), and PDGFRα (Morikawa et al., 2009; Zhou et al., 2014). Conditional deletion of Scf (Ding et al., 2012; Oguro et al., 2013), Cxcl12 (Ding and Morrison, 2013; Greenbaum et al., 2013), or Pleiotrophin (Himburg et al., 2018) from LepR⁺ cells depletes and/or mobilizes HSCs from the bone marrow. Deletion of the Ebf3 or Foxcl transcription factors from LepR⁺ cells also depletes LepR⁺ cells and HSCs from the bone marrow (Omatsu et al., 2014; Seike et al., 2018).

LepR⁺ stromal cells seem to make the vast majority of SCF and Cxcl12 in the bone marrow: 95% of Scf-GFP^high cells and 90% of Cxcl12-DsRed^high cells are LepR⁺ in normal young adult bone marrow (Ding et al., 2012; Ding and Morrison, 2013). Osteoblasts do not express Scf and express very low levels of Cxcl12 (Ponomaryov et al., 2000; Dar et al., 2005; Sugiyama et al., 2006; Ding and Morrison, 2013; Greenbaum et al., 2013). Deletion of Scf or Cxcl12 from osteoblasts, Nestin⁺ cells, megakaryocytes, or other hematopoietic cells does not affect HSC frequency (Ding et al., 2012; Ding and Morrison, 2013; Greenbaum et al., 2013). Recent single cell RNA sequencing studies also showed that LepR⁺ cells express high levels of Scf and Cxcl12 while endothelial cells express much lower levels (Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020; Matsushita et al., 2020). NG2⁺ periarteriolar cells express modest levels of Cxcl12, much lower than LepR⁺ cells (Baryawno et al., 2019; Baccin et al., 2020). These studies found little or no Scf or Cxcl12 expression by osteoblasts, Nestin⁺ cells, megakaryocytes, or other hematopoietic cells in the bone marrow.

While SCF from endothelial cells is necessary for the maintenance of normal numbers of HSCs, it does not appear to be necessary for the maintenance of most c-kit⁺ restricted hematopoietic progenitors (Comazzetto et al., 2019). These cells derive SCF from LepR⁺ cells, at least under steady state conditions. So why is the SCF made by endothelial cells necessary for HSC maintenance but not for the maintenance of restricted progenitors? One possibility is that there is a difference in the extent to which different niche cells produce membrane bound versus soluble SCF, and that hematopoietic stem/progenitor cells discriminate between these different forms of SCF. Another possibility is that there are differences in the localization of HSCs and restricted progenitors relative to these niche cell types.

While some factors are synthesized locally in the bone marrow to regulate hematopoietic stem/progenitor cell maintenance, other factors are synthesized outside of the bone marrow and act at a distance. Thrombopoietin is required to maintain HSCs and thrombopoiesis by binding the c-Mpl receptor, which is expressed by HSCs and megakaryocyte lineage cells (de Sauvage et al., 1994; Kaushansky et al., 1994; Sitnicka et al., 1996; Kimura et al., 1998; Qian et al., 2007). Thrombopoietin is mainly expressed by hepatocytes and deletion from hepatocytes depletes HSCs in the bone marrow (Decker et al., 2018). In contrast, deletion from hematopoietic cells, bone marrow stromal cells, and osteoblasts does not affect HSC frequency. HSCs are thus maintained by a combination of locally-acting and long-distance factors.

What is in the periarteriolar niche?

Some studies proposed a periarteriolar niche for the most quiescent HSCs in which Nestin^high and NG2⁺ periarteriolar stromal cells were suggested to secrete factors that promote HSC maintenance (Kunisaki et al., 2013; Asada et al., 2017). However, subsequent studies found few, if any, HSCs adjacent to arterioles (Nombela-Arrieta et al., 2013; Acar et al., 2015; Sacma et al., 2019; Christodoulou et al., 2020) and little expression of Scf or Cxcl12 in Nestin^high or NG2⁺ stromal cells (Ding et al., 2012; Ding and Morrison, 2013; Baryawno et al., 2019; Baccin et al., 2020). Conditional deletion of Scf using Nestin-Cre (Ding et al., 2012), Nestin-CreER (Ding et al., 2012), or NG2-CreER (Acar et al., 2015; Asada et al., 2017) had no effect on HSC frequency or hematopoiesis in adult bone marrow. Deletion of Cxcl12 using Nestin-Cre (Ding and Morrison, 2013) or Nestin-CreER (Ding and Morrison, 2013) also had no effect on HSC frequency or hematopoiesis.

While the data do not currently support the existence of a periarteriolar niche for quiescent HSCs in adult bone marrow, Nestin⁺ and NG2⁺ stromal cells may contribute to a niche for HSCs in early postnatal bone marrow. Deletion of Cxcl12 using NG2-CreER at 2 weeks after birth (Asada et al., 2017) and Nestin-CreER at 1 week after birth (Isern et al., 2014) each depleted HSCs. In early postnatal bone marrow, Nestin and NG2 expression are widespread and not limited to periarteriolar stromal cells. Thus, these cells appear to contribute to the formation of an HSC niche in early postnatal bone marrow but the precise location of this niche has not been determined.

Nestin can be hard to interpret as a marker as all of the Nestin reporter and Cre alleles are transgenic mice made using the Nestin second intronic enhancer (Zimmerman et al., 1994; Mignone et al., 2004; Sun et al., 2014). Each of the transgenes exhibits a somewhat different expression pattern that does not fully mirror the expression of endogenous Nestin (Zimmerman et al., 1994; Ding et al., 2012; Sun et al., 2014). LepR⁺ cells express a low level of the Nestin-GFP transgene (Kunisaki et al., 2013; Zhou et al., 2014) but do not express endogenous Nestin, Nestin-Cre, or Nestin-CreER (Ding and Morrison, 2013; Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020). The Scf and Cxcl12 expression observed among Nestin-GFP⁺ cells appears to come mainly from Nestin-GFP^low cells, not from Nestin-GFP^high periarteriolar cells.

So what is contained in the periarteriolar niche in adult bone marrow? A more refined analysis suggests that the periarteriolar niche contains lymphoid-biased HSCs (Pinho et al., 2018), which are more mitotically-active than myeloid-biased HSCs and on the verge of exiting the HSC pool. The evidence that periarteriolar niches contain lymphoid-biased HSCs also fits with the recent observation that periarteriolar niches contain a subset of early lymphoid progenitors (Shen et al., 2021). Quiescent HSCs thus appear to reside overwhelmingly in perisinusoidal niches while a subset of early lymphoid progenitors, including some that are multipotent, are periarteriolar.

An unresolved inconsistency is that SCF expression is higher in arteriolar endothelial cells than in sinusoidal endothelial cells (though much lower in both populations as compared to LepR⁺ cells) (Xu et al., 2018; Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020). Yet HSCs, which depend on SCF from endothelial cells (Ding et al., 2012; Xu et al., 2018), are perisinusoidal. SCF is thought to act locally in creating the niche, suggesting that HSCs would be expected to be near the endothelial cells from which they derive SCF. However, it remains possible that arteriolar endothelial cells produce soluble SCF that diffuses to perisinusoidal niches, or even that endothelial cells in other tissues produce soluble SCF that is transported systemically through the blood. This is difficult to test rigorously because the production of soluble versus membrane bound SCF is regulated by differential splicing and proteolytic cleavage, making it difficult to eliminate one form without affecting the other.

Bone marrow stromal cells include skeletal stem cells

There are multiple waves of mesenchymal progenitors that form skeletal tissues during development and then maintain the skeleton throughout adulthood (Takashima et al., 2007; Maes et al., 2010; Mizoguchi et al., 2014; Ono et al., 2014a; Ono et al., 2014b; Zhou et al., 2014; Worthley et al., 2015; Pineault et al., 2019) (Figure 4). Cells that are Osterix⁺, Collagen2⁺, and/or Hoxa11⁺ give rise to osteoblasts, osteocytes, chondrocytes and stromal cells in developing bones. Nestin-CreER-expressing cells transiently form osteoblasts and bone marrow stromal cells in the early postnatal period (Takashima et al., 2007; Mendez-Ferrer et al., 2010; Ono et al., 2014a). Some of the Osterix⁺ and Hoxa11⁺ mesenchymal cells give rise to LepR⁺ bone marrow stromal cells perinatally (Mizoguchi et al., 2014; Ono et al., 2014b; Pineault et al., 2019). The LepR⁺ cells are initially rare and limited to the metaphysis but become much more common by adulthood. They include the skeletal stem cells that form most of the CFU-F, osteoblasts, and adipocytes in adult bone marrow (Zhou et al., 2014; Zhou et al., 2017).

In adult bone marrow, LepR⁺ cells give rise to Adiponectin-CreER⁺ adipocyte progenitors (Zhou et al., 2017; Matsushita et al., 2020) and Osteolectin⁺ osteogenic progenitors (Shen et al., 2021), both of which also express high levels of Scf and Cxcl12. This has created some confusion as LepR⁺ cells that retain the ability to make osteoblasts (Matsushita et al., 2020) express a lower level of adiponectin as compared to the adipocyte progenitors they form (Zhong et al., 2020). Consequently, Adiponectin-Cre recombines broadly in LepR⁺ cells, including in multipotent LepR⁺ cells, but Adiponectin-CreER⁺ recombines preferentially in a much smaller population of LepR⁺ adipocyte progenitors. Thus, although LepR⁺ bone marrow stromal cells are uniformly positive for the expression of Scf, Cxcl12, and Angptl1 (Ding et al., 2012; Ding and Morrison, 2013; Zhou et al., 2015) they remain heterogeneous in terms of fate and function.

Heterogeneity among stromal cells in the bone marrow

Single cell RNA sequencing provided important new insights into the heterogeneity among bone marrow stromal cells (Baryawno et al., 2019; Tikhonova et al., 2019; Baccin et al., 2020; Matsushita et al., 2020). Most interestingly, these studies found that perisinusoidal and periarteriolar LepR⁺ cells exhibit distinct gene expression profiles. Although both express high levels of Scf, Cxcl12, and Pleiotrophin, the perisinusoidal LepR⁺ cells express genes suggesting they are poised to undergo adipogenesis while the periarteriolar LepR⁺ cells express genes suggesting they are poised to undergo osteogenesis.

This heterogeneity in LepR⁺ cells was subsequently confirmed in fate mapping studies. Cxcl12-CreER recombines in approximately 30% of perisinusoidal LepR⁺ stromal cells (Matsushita et al., 2020). Consistent with single cell RNA sequencing studies, these cells give rise to adipocytes under steady state conditions but retain the ability to form osteoblasts in response to injury. This is important because it suggests that at least a subset of perisinusoidal LepR⁺ cells are reprogrammed by bone injuries to make osteoblasts. Conversely, Osteolectin-CreER recombines in periarteriolar but not perisinusoidal LepR⁺ cells (Shen et al., 2021). These cells are rapidly-dividing and short-lived osteogenic progenitors that represent a major source of osteoblasts, but not adipocytes, under steady state conditions and after bone injury. There is, therefore, a periarteriolar niche for osteogenic progenitors in the bone marrow.

Other cell types modulate HSC and niche functions

While LepR⁺ cells and endothelial cells are functionally important sources of known factors required for HSC maintenance, many other cells directly or indirectly modulate HSC or niche function through other mechanisms. Depletion of monocytes, macrophages, or dendritic cells promotes HSC mobilization from the bone marrow by altering Cxcl12 expression by bone marrow stromal cells (Winkler et al., 2010; Chow et al., 2011) and endothelial cell permeability (Zhang et al., 2019). Neutrophil clearance by macrophages in the bone marrow also influences hematopoietic stem/progenitor cell mobilization (Casanova-Acebes et al., 2013). A subset of macrophages regulates HSC quiescence by secreting TGFβ (Hur et al., 2016). Megakaryocytes modulate HSC quiescence by secreting Cxcl4 and TGFβ1 (Bruns et al., 2014; Zhao et al., 2014) and promote the regeneration of HSCs after myeloablation by secreting FGF21 (Zhao et al., 2014). Megakaryocytes localize to sinusoids (Stegner et al., 2017) and are often near HSCs (Acar et al., 2015; Kokkaliaris et al., 2020), particularly von Willebrand factor (vWF) expressing myeloid- and platelet-biased HSCs (Sanjuan-Pla et al., 2013; Pinho et al., 2018).

Nerve fibers in the bone marrow also regulate HSC mobilization and hematopoietic regeneration. Sympathetic and nociceptive nerves innervate the bone marrow, mainly near arterioles (Katayama et al., 2006; Yamazaki et al., 2011; Gao et al., 2020). Neither sympathetic nor nociceptive nerve fibers are required for HSC maintenance or hematopoiesis under normal circumstances, though ablation of both sensory and sympathetic nerve fibers increases HSC and myeloid progenitor frequency in the bone marrow (Katayama et al., 2006; Yamazaki et al., 2011; Gao et al., 2020). Sympathetic nerves regulate circadian variation in HSC mobilization into the blood and the effects of G-CSF on mobilization, perhaps by influencing the expression of Cxcl12 by stromal cells (Katayama et al., 2006; Mendez-Ferrer et al., 2008). Nociceptive nerve fibers promote HSC mobilization by releasing Calcitonin Gene-Related Peptide, which activates receptors expressed by HSCs (Gao et al., 2020). Nerve fibers are also required for hematopoietic regeneration after myeloablation, though the underlying mechanisms are not well understood (Lucas et al., 2013).

Non-myelinating Schwann cells associated with some nerve fibers promote HSC quiescence by regulating the post-translational processing and activation of latent TGFβ (Yamazaki et al., 2011). Therefore, even though nerve fibers and non-myelinating Schwann cells are generally not in close contact with HSCs (Acar et al., 2015; Kokkaliaris et al., 2020), they appear to regulate HSCs through diffusible factors. More work is required to identify the cells on which nerve fibers synapse in the bone marrow and how signals from nerve fibers propagate through the bone marrow to influence cells at a distance.

Niches for restricted hematopoietic progenitors

Niches for restricted progenitors are distributed in different microenvironments throughout the bone marrow. For example, a subset of early lymphoid progenitors resides near the endosteum, where they depend upon Cxcl12, and perhaps other factors, synthesized by osteoblasts (Ding and Morrison, 2013; Greenbaum et al., 2013). Another subset of early lymphoid progenitors resides adjacent to arterioles, where they depend upon SCF synthesized by periarteriolar LepR⁺Osteolectin⁺ stromal cells (Shen et al., 2021). LepR⁺ cells are also an important source of IL-7 for early lymphoid progenitors, though it is not clear whether the LepR⁺ cells that supply IL-7 are periarteriolar or perisinusoidal (Cordeiro Gomes et al., 2016). It also remains unclear whether the same early lymphoid progenitors are distributed among endosteal, periarteriolar, and perisinusoidal niches or whether progenitors at distinct stages of lymphopoiesis transition among niches in different regions of the bone marrow.

Early erythroid progenitors localize adjacent to sinusoids and depend upon SCF produced by LepR⁺ cells (Comazzetto et al., 2019). This suggests that they reside in a perisinusoidal niche, though it remains unresolved whether this niche is spatially distinct from the perisinusoidal HSC niche.

Technically impressive recent work has carefully evaluated the positions of distinct myeloid progenitors in the bone marrow. Granulocyte-macrophage progenitors are scattered as single cells in the bone marrow under steady state conditions but proliferate to form clusters after myeloablation (Herault et al., 2017). These clusters tend to occur near megakaryocytes, suggesting a perisinusoidal niche (Herault et al., 2017). Consistent with this, monocyte/dendritic cell progenitors, monocyte progenitors, and granulocyte progenitors in the bone marrow are enriched near sinusoids, but not arterioles or endosteum, and monocyte/dendritic cell progenitors depend upon CSF1 from endothelial cells (Zhang et al., 2021). Although granulocyte progenitors, monocyte/dendritic cell progenitors, and HSCs all appear to occupy perisinusoidal niches, they do not appear to colocalize with each other (Zhang et al., 2021). This suggests distinct perisinusoidal domains for HSCs and myeloid progenitors (Figure 3). An important future question is whether there are different types of endothelial cells and/or Lep⁺ cells that are specialized to maintain different kinds of stem/progenitor cells.

The niche changes in response to injury

The niche changes dramatically in response to myeloablation or hematopoietic stresses such as pregnancy. Myeloablation, either by chemo or radiation therapy, destroys the sinusoidal network in the bone marrow (Hooper et al., 2009), depletes endothelial cells and LepR⁺ cells (Zhou et al., 2014), and dramatically increase adipogenesis. Regeneration of hematopoiesis depends upon the regeneration of sinusoids and stromal cells (Hooper et al., 2009; Zhou et al., 2014; Chen et al., 2019; Fang et al., 2020). Apelin⁺ endothelial cells associated with a subset of sinusoidal blood vessels appear to be particularly important for the regeneration of vascular endothelial cells after myeloablation (Chen et al. 2019).

Adipogenesis during development in mouse caudal vertebrae suppresses vasculogenesis, reducing HSC frequency and hematopoiesis in these bones (Naveiras et al., 2009). However, in most other mouse bones adipocytes are rare in developing bone marrow and accumulate during aging. During adulthood, adipocytes synthesize niche factors, like SCF, that promote HSC maintenance and hematopoietic regeneration after myeloablation (Zhou et al., 2017). Bone marrow adipocytes are different from other adipocytes, such as in abdominal fat, which do not appear to express Scf (Zhou et al., 2017). Adipocytes are likely to promote the regeneration of HSCs and hematopoiesis through multiple mechanisms. Adipocytes appear to be an emergency response system that rapidly increases the production of niche factors after the depletion of sinusoids and stromal cells.

The other major change that occurs when bone marrow hematopoiesis is insufficient is the induction of extramedullary hematopoiesis (EMH). EMH can occur nearly anywhere but most commonly occurs in the spleen or liver. EMH involves the formation of facultative niches during adulthood that persist only as long as required to augment hematopoiesis. For example, during pregnancy the blood volume increases rapidly, requiring a transient increase in erythropoiesis. This is achieved by the mobilization of HSCs to the spleen where they engage in EMH (Nakada et al., 2014; Inra et al., 2015). EMH also occurs in response to anemia, infection, and when the bone marrow is compromised by hematopoietic neoplasms or fibrosis.

The EMH niche has been characterized in the spleen, where HSCs are maintained in perivascular niches associated with sinusoidal blood vessels in the red pulp (Kiel et al., 2005; Inra et al., 2015). Splenic EMH requires SCF and Cxcl12 synthesized by endothelial cells and TCF21⁺ perisinusoidal stromal cells (Inra et al., 2015). The splenic HSC niche is thus analogous to the bone marrow HSC niche in that it is perisinusoidal and depends upon endothelial cells and perivascular stromal cells but the TCF21⁺ stromal cells in the spleen are distinct from the LepR⁺ stromal cells in the bone marrow. Other cells also modulate the spleen niche, including macrophages (Dutta et al., 2015). Other EMH niches, like in the liver, have not yet been characterized. It will be particularly interesting to characterize EMH niches outside of hematopoietic tissues, where there are no sinusoidal blood vessels, to determine whether HSCs can be maintained in niches that are not associated with sinusoids.

Many growth factors, membrane bound ligands, and other secreted gene products are necessary for the regeneration of HSCs and hematopoiesis after myeloablation but are dispensable for HSC maintenance and hematopoiesis under steady state conditions (see Table 2 in (Crane et al., 2017)). It generally remains unclear to what extent these factors promote hematopoietic regeneration by acting on stromal cells, on hematopoietic stem/progenitor cells, or both. Moreover, the cellular sources of many of these factors remain uncertain.

There is bidirectional communication between hematopoietic stem/progenitor cells and the niche. As hematopoietic stem/progenitor cells arrive in the zebrafish caudal hematopoietic tissue during development they remodel the endothelial cells around them, potentially increasing the local concentrations of niche factors (Tamplin et al., 2015; Blaser et al., 2017). Hematopoietic stem/progenitor cells also regulate the regeneration of the vasculature in adult bone marrow by secreting Angiopoietin-1, which acts on endothelial cells to regulate the regeneration of patent (non-leaky) sinusoids (Zhou et al., 2015). Mature hematopoietic cells, such as granulocytes, regulate niche regeneration by secreting proteases that remodel extracellular matrix and growth factor availability (Heissig et al., 2002) and TNFα, which accelerates vascular recovery (Bowers et al., 2018). These likely represent some of many ways in which hematopoietic cells influence the function and regeneration of their niches.

There is also immunomodulation of the niche. Regulatory T cells in the bone marrow reside near sinusoids and promote HSC quiescence by suppressing inflammation (Fujisaki et al., 2011; Hirata et al., 2018). Dietary restriction promotes the accumulation of memory T cells in the bone marrow, partly as a result of adipocyte function, preserving memory T cells and immune function (Collins et al., 2019). The suppression of inflammation within the bone marrow is an understudied element of niche biology that may play an important role in preserving HSCs during aging and after pathogen infection.

The bone marrow HSC niche changes during aging

Hematopoietic stem/progenitor cells undergo cell-intrinsic and cell-extrinsic changes during aging that increase myelopoiesis and decrease lymphopoiesis. Changes in the bone marrow microenvironment contribute to these changes (Young et al., 2016), though there remains a limited understanding of how the niche changes with age. During aging, there are changes in the vasculature, including depletion of CD31^highEndomucin^−/low arterioles (Kusumbe et al., 2016; Poulos et al., 2017; Ho et al., 2019; Sacma et al., 2019). There are also changes in bone marrow stromal cells, including a depletion of periarteriolar Osteolectin⁺ cells, which may contribute to the depletion of lymphoid progenitors (Shen et al., 2021). In contrast, sinusoids are preserved during aging, perhaps explaining why HSCs are not depleted with age in most mouse strains, in contrast to lymphoid progenitors (Sacma et al., 2019). An important outstanding question concerns the impact of increasing adipogenesis in the bone marrow during aging and the mechanisms by which this influences HSC function and hematopoiesis (Naveiras et al., 2009; Ambrosi et al., 2017; Zhou et al., 2017).

The levels of inflammation and inflammatory growth factors increase with age in the bone marrow and likely contribute to changes in hematopoiesis. For example, Interferon, IL-1β, IL-6, TNFα, and Rantes expression by stromal cells and hematopoietic cells increase with age and contribute to the increase in myelopoiesis (Ergen et al., 2012; Schurch et al., 2014; Ho et al., 2019; Yamashita and Passegue, 2019; Valletta et al., 2020). There may also be changes in nerve fibers in the bone marrow that contribute to the changes in hematopoiesis with age (Maryanovich et al., 2018; Ho et al., 2019).

What else does a niche do?

There are likely many ways in which the niche regulates HSC maintenance and hematopoiesis beyond the secretion of growth factors, though this has been explored only to a limited extent. The bone marrow is hypoxic in general but is particularly hypoxic around the sinusoids where HSCs localize (Parmar et al., 2007; Nombela-Arrieta et al., 2013; Spencer et al., 2014), raising the possibility that low oxygen tension contributes to HSC maintenance. The niche may also regulate the metabolic environment in the bone marrow. For example, LepR⁺ stromal cells may modulate the availability of aspartate for certain kinds of stem/progenitor cells (van Gastel et al., 2020). The extent to which the niche modulates the availability of nutrients is an important, open question.

Although not something mammals have to worry about, lower vertebrates, such as fish, must protect hematopoietic stem/progenitor cells from ultraviolet light-induced mutagenesis (Kapp et al., 2018). In fish, a melanocyte umbrella forms above the kidney marrow. In terrestrial animals HSCs relocated into the bone marrow, which is protected from ultraviolet light by bone. A curious phenomenon of unknown physiological significance is that melanocytes accumulate in the mammalian spleen after irradiation, when EMH is induced (Weissman, 1967). This may represent an evolutionary vestige of the melanocyte umbrella.

There is also mechanical regulation of some niches in the bone marrow. The maintenance of periarteriolar niches for osteogenic and lymphoid progenitors depends upon movement-induced mechanical forces that appear to be transmitted along arterioles from the bone surface into the bone marrow (Shen et al., 2021). LepR⁺Osteolectin⁺ periarteriolar stromal cells appear to sense these mechanical forces via the stretch activated ion channel, Piezo1. Piezo1 signaling promotes the proliferation of LepR⁺Osteolectin⁺ cells, increasing the production of SCF and the numbers of lymphoid progenitors in the periarteriolar niche. The discovery that mechanical forces are required to maintain a niche for osteogenic and lymphoid progenitors identifies a new mechanism by which exercise increases bone thickness and immune function and raises the question of whether other niches also require mechanical stimulation.

The effects of mechanical forces on niche function and the cellular organization of the bone marrow are likely mediated partly by adhesion molecules. Many studies have identified adhesion molecules that regulate hematopoiesis or the trafficking of hematopoietic cells (reviewed by Forsberg and Smith-Berdan, 2009). In some cases, these adhesion molecules promote the retention of hematopoietic stem/progenitor cells in perivascular niches (Smith-Berdan et al., 2011; Gur-Cohen et al., 2015).

Behavioral attributes, including changes in sleep, exercise, social stress, and diet all influence the bone marrow. Sleep reduces myelopoiesis and cardiovascular inflammation by increasing hypocretin expression in the hypothalamus, which reduces CSF1 expression by neutrophils in the bone marrow (McAlpine et al., 2019). Exercise also reduces myelopoiesis by reducing leptin production by adipose cells, leading to changes in growth factor expression by LepR⁺ bone marrow stromal cells (Frodermann et al., 2019). Conversely, chronic social stress promotes a more inflammatory environment by activating the sympathetic nervous system, promoting myelopoiesis and EMH (Powell et al., 2013; McKim et al., 2018). Dietary changes alter bone marrow hematopoiesis by altering the abundance of vitamin A (Cabezas-Wallscheid et al., 2017), vitamin C (Agathocleous et al., 2017), valine (Taya et al., 2016), or calorie restriction (Mana et al., 2017). However, there is not yet any evidence that these changes in hematopoiesis are mediated by changes in niche cell function.

Concluding remarks

Using the genetic and imaging approaches that have been developed over the past 10 years we have the opportunity to address many fundamental questions related to hematopoietic niches (Box 1). The ultimate goal is to be able to map the three-dimensional organization and lineage relationships within the bone marrow as HSCs give rise to differentiated cells. We already know a lot about the cell-intrinsic changes that cells undergo during lineage restriction and differentiation, from gene expression to morphology. We know much less about the three-dimensional organization of the bone marrow and remain unable to link specific stages of lineage restriction or differentiation to specific microenvironments. Where in the bone marrow do these changes occur and how does migration among distinct microenvironments trigger or enable the changes?

Box 1: Important future questions.

Important future challenges include:

• Refining our understanding of the location and cellular composition of niches for restricted progenitors in the bone marrow

• Assessing the mechanisms by which hematopoietic stem/progenitor cells migrate among distinct niches in the bone marrow as they undergo lineage restriction and differentiation

• Identifying additional EMH niches beyond the spleen and comparing them to bone marrow and spleen niches

• Identifying mechanisms that initiate and terminate EMH

• Better understanding the ways in which bone marrow stromal cells are regenerated and remodeled in response to myeloablation as well as the mechanisms that promote hematopoietic regeneration

• Identifying the signals that initiate and terminate adipogenesis after myeloablation and the ways in which adipocytes regulate hematopoietic regeneration

• Identifying bidirectional communication signals between hematopoietic stem/progenitor cells and their niches and how they influence niche regeneration

• Better understanding immunomodulation and its role in HSC maintenance

• Identifying mechanisms that act within the bone marrow to mediate quorum sensing and homeostasis by sensing the depletion of differentiated cells and increasing their production

One of the great unsolved problems in hematopoiesis is identifying the homeostatic mechanisms that sense the depletion of subsets of progenitors and/or differentiated cells and promote the proliferation of other cells to sustain normal blood cell counts. These homeostatic mechanisms appear to operate at many levels of hematopoiesis and in many different hematopoietic lineages but surprisingly few of the underlying molecular mechanisms have been identified. The increased production of erythropoietin in response to hypoxia is a notable exception. A refined understanding of the location and cellular composition of niches for restricted progenitors in the bone marrow could reveal more of these homeostatic mechanisms by allowing us to study the precise stromal cells responsible for the regulation of specific subsets of progenitors.

Hematopoietic stem cells (HSCs) and hematopoietic progenitors are essential for blood production and reside in specialized niches in the adult mammalian bone marrow. In this review, Comazzetto et al. describe our current understanding of HSC and progenitor cell niches, focusing on recent progress and important unresolved questions.