Mesenchymal Progenitors and the Osteoblast Lineage in Bone Marrow Hematopoietic Niches

Abstract

The bone marrow cavity is essential for the proper development of the hematopoietic system. In the last few decades it has become clear that mesenchymal stem/progenitor cells as well as cells of the osteoblast lineage, besides maintaining bone homeostasis, are also fundamental regulators of bone marrow hematopoiesis. Several studies have demonstrated the direct involvement of mesenchymal and osteoblast lineage cells in the maintenance and regulation of supportive microenvironments necessary for quiescence, self-renewal and differentiation of hematopoietic stem cells. In addition, specific niches have also been identified within the bone marrow for maturing hematopoietic cells. Here we will review recent findings that have highlighted the roles of mesenchymal progenitors and cells of the osteoblast lineage in regulating distinct stages of hematopoiesis.

Hematopoietic and mesenchymal progenitors differentiate in the bone marrow

In the classical model of hematopoietic differentiation, pluripotent hematopoietic stem cells (HSCs) give rise to common lymphoid progenitors (CLP) and common myeloid progenitors (CMP). CLP differentiate into the entire lymphoid lineage (B lymphocytes, T lymphocytes, natural killer (NK), and NK-T cells), while CMP give rise to granulocyte/macrophage progenitors (GMP), which differentiate into the myeloid lineages (including neutrophils, basophils, eosinophils and monocytes) and megakaryocyte/erythroid progenitors (MEP) that in turn become megakaryocytes and erythrocytes (reviewed in [1]). The differentiation of hematopoietic progenitors is directed by an intrinsic network of transcription factors (reviewed in [2]) and also by inputs from the cellular niches in which they reside.

Just as HSCs give rise to hematopoietic lineages, mesenchymal stem/progenitor cells differentiate into various lineages including adipocytes, chondrocytes and osteoblasts. Commitment to the osteoblast lineage is marked by expression of Runx2, a transcription factor essential for osteoblastogenesis, followed by osterix (Osx) [3-6]. Osteoblast progenitors (osteoprogenitors) differentiate into osteoblasts and express progressively mature markers such as alkaline phosphatase (ALP) and type I collagen, ultimately becoming terminally differentiated osteoblasts located on the bone surface and expressing markers such as osteocalcin [7]. Mature osteoblasts may undergo apoptosis, become quiescent lining cells, or become embedded in the bone matrix as osteocytes expressing Dmp1 [8]. Osteocytes account for almost 95% of all bone cells, and are morphologically very distinct from their parent cuboidal osteoblasts.

Given the physiological proximity of the bone and hematopoietic systems during development, it is not surprising that both tissues interact and influence each other. Interactions and interdependence between osteoblasts, of mesenchymal lineage, and osteoclasts, of hematopoietic origin, are well documented [9]. Distinct niches are required for different hematopoietic cell lineages and stages, in recent years it has become clear that these distinct niches can be provided by unique subsets of stromal cells. In this review we will focus on the role of mesenchymal stem cells and the osteoblast lineage in providing stage-specific support for hematopoiesis.

Osteoblasts support hematopoiesis

Osteoblasts have been demonstrated to produce many of the cytokines and growth factors that play important roles in hematopoietic and myeloid development, including granulocyte colony-stimulating factor (G-CSF) [10], granulocyte-macrophage colony-stimulating factor (GM-CSF) [11], interleukin-6 (IL6) [12], leukemia inhibitory factor (LIF) [13], transforming growth factor β (TGFβ), tumor necrosis factor α (TNFα), and c-kit ligand [14].

That the osteoblast lineage could support hematopoiesis in the bone marrow was first demonstrated in vivo by two articles that reported an increase in bone marrow hematopoietic stem and progenitor cells (HSPCs) in mice with increased trabecular bone and osteoblast numbers [15, 16]. More functional evidence for a role for osteoblasts in supporting hematopoiesis came from mice in which expression of herpesvirus thymidine kinase is targeted to maturing osteoblasts by the 2.3kb rat type I collagen promoter (Col2.3ΔTK mice). Treatment of Col2.3ΔTK mice with ganciclovir (GCV) resulted in osteoblast depletion within 7 days and progressive bone loss, accompanied by a decline of >85% in bone marrow cellularity and extramedullary hematopoiesis [17]. Subsequently, B cell precursors and erythroblast precursors were found to be drastically reduced in the bone marrow followed by a loss of HSC subsets [18], while cells of the myeloid lineage gradually disappeared over longer time periods. In contrast, an analogous transgenic mouse model expressing herpesvirus thymidine kinase targeted to terminally differentiated osteoblasts by the osteocalcin promoter (OCΔTK mice) did not show an appreciable loss of bone marrow cellularity, although a complete characterization of the hematopoietic compartment was not performed [19]. Taken together, these results demonstrated that committed early stage osteoblasts support hematopoiesis at multiple stages.

Mesenchymal progenitors are perivascular

To understand the mechanisms by which the skeleton can support hematopoietic development in the bone marrow, it is important to consider that cells of the osteoblast lineage themselves are at varying stages of differentiation. While mature osteoblasts and osteocytes are easily identifiable by their anatomic locations on and within mineralized bone, respectively, the exact nature and localization of in vivo mesenchymal stem/progenitor cells (MSPCs) and osteoprogenitors within the bone marrow stroma remain poorly understood. However, several recent studies have indicated a perivascular location for MSPCs (Table 1). Shi and Gronthos demonstrated that BMSCs expressing STRO-1 and CD146 are perivascular in human bone marrow frozen sections, and can form ectopic bone with marrow on transplantation [20]. Sachetti et al showed that CD146+ cells in human bone marrow stroma contain all colony forming unit-fibroblasts (CFU-F), and that CD146 labels subendothelial reticular cells [21]. Upon transplantation CD146+ cells can self-renew and differentiate into osteoblasts. Tormin and co-workers demonstrated that in human bone marrow, CFU-Fs are enriched in CD146+ CD271+ cells that are found in a sinusoidal subendothelial location. In contrast, bone-lining CFU-Fs express CD271 alone [22].

Table 1.

Identification of mesenchymal stem/progenitor cells in vivo.

Markers/ reporter systems	Organism	Tissue	Findings	Reference
In Vivo Localization
STRO-1, CD146	Human	Bone marrow	Perivascular location Can form bone with marrow	[20]
CD146	Human	Bone marrow	Reticular adventitial cells around sinusoids Can form bone with marrow Can self-renew in vivo	[21]
CD146 & CD271	Human	Bone marrow	Perivascular reticular cells Multilineage mesenchymal potential in vitro Can form bone with marrow	[22]
Sca-1 & PDGFR-α	Mouse	Bone marrow	Multilineage mesenchymal potential in vitro Arterial perivascular location adjacent to vascular smooth muscle cells Can give rise to perivascular stromal cells, osteoblasts and adipocytes in vivo	[23]
CD146, NG2, PDGFR-β	Human	Muscle, pancreas, adipose, placenta	Perivascular location Multilineage mesenchymal potential in vitro Can give rise to muscle and ectopic bone in vivo	[25]
In Vivo Lineage Tracing
Nes-GFP+ Nes-GFP; Col2.3− Cre; R26-LacZ Nes-CreERT2; RCE:loxP	Mouse	Adult bone and bone marrow	Nes-GFP Perivascular location Nes-GFP; Col2.3-Cre; R26-LacZ Multilineage mesenchymal potential in vitro Self-renew and contribute to ectopic bone formation Nes-CreERT2 Contribute to osteoblasts, osteocytes, and chondrocytes in vivo	[29]
Prx1-Cre; Ai9; CXCL12-GFP		Adult bone and bone marrow	Prx1-Cre Contribute to osteoblasts, osteocytes and CXCL12+ reticular cells in vivo	[36]
LepR-Cre; loxP- EYFP		Adult bone and bone marrow	LepR-Cre Perivascular stromal cells but not osteoblasts Express CXCL12, PDGFRα	[31]
Mx1-Cre; Rosa- YFP Osx-CreERt;Rosa- YFP Ocn-CreER; Rosa- YFP	Mouse	Adult bone and bone marrow	Mx1-Cre Contribute to osteoblasts in vivo Multilineage mesenchymal potential in vitro Osx-CreERt Transient source of osteoblasts Ocn-CreER Rapid turnover of mature osteoblasts on the endosteal surface Contribute to osteocytes	[40]
αSMA-GFP αSMA-Cherry αSMA-CreERT2; Ai9 (TdTomato)	Mouse	Adult bone and bone marrow	αSMA-GFP or αSMA-Cherry Perivascular location Osteogenic and adipogenic potential in vitro Contribute to osteoblasts in vivo αSMA-CreERT2 Contribute to osteoblasts and osteocytes in vivo Multilineage mesenchymal potential in vitro	[44, 45]
Osx-CreERt; R26R-LacZ ColI(3.2 kb)- CreERt; R26R- LacZ	Mouse	Embryonic bone	Osx-CreERt Labeled while in perichondrium Closely associated with invading vasculature Give rise to trabecular osteoblasts, osteocytes and stromal cells ColI-CreERt Not associated with vasculature Remain in cortical bone	[41]
Osx-Cre; mTmG; β-catfl/fl Osx-Cre; Ai9 Osx-CreERt; Ai9		Adult bone and bone marrow	Osx-Cre; β-catfl/fl Adipogenic potential in vitro Contribute to osteoblasts, osteocytes and adipocytes in vivo in adult mice Osx-Cre Contribute to perivascular stromal cells, osteoblasts and osteocytes Multilineage mesenchymal potential in vitro Osx-CreERt Embryonic cells contribute to perivascular stromal cells, osteoblasts, osteocytes and adipocytes in vivo	[42, 43]
PPARγ-GFP	Mouse	Adipose tissue	PPARγ-GFP Adipogenic potential in vitro Perivascular location Express aSMA, NG2 Contribute to adipocytes in vivo	[47]
Zfp423-GFP	Mouse	Adipose tissue	Zfp423-GFP Adipogenic potential in vitro Expressed by perivascular and endothelial cells	[48]
VE-Cad-Cre; R26R-LacZ VE-Cad-Cre; R26R-eGFP	Mouse	Adipose tissue	VE-cad-Cre Expressed by perivascular and endothelial cells, preadipocytes and adipocytes	[49]

In mice, a similar approach was used to demonstrate that BMSCs expressing both PDGFRα and Sca-1 (PαS cells) are enriched in CFU-Fs and located in the arterial perivascular space near the inner surface of cortical bone [23]. Single cell clonal analysis demonstrated that PαS cells exhibit uniform expression of conventional mesenchymal markers such as CD29, CD49e, CD44 and Sca-1 while CD105 and CD90 are more heterogeneous in expression [23]. When transplanted, PαS cells can reconstitute the hematopoietic niche, and are enriched in the expression of Angiopoietin 1 (Ang1), a niche-supporting factor [24].

In other tissues besides bone marrow, MSPCs have also been demonstrated to localize to the perivascular space. Crisan et al., identified perivascular pericytes that express NG2, CD146, and PDGFR-β in multiple tissues throughout human fetal and adult organs. These cells exhibit clonal chondrocyte, osteoblast, and adipocyte potential in vitro, and are able to differentiate into myocytes and ectopic bone in vivo [25].

Perivascular MSPCs support the HSC niche

Recent studies have highlighted the critical role of perivascular bone marrow MSPCs in the regulation of the hematopoietic system. In the following paragraphs we briefly describe the recent research advances focusing on the interaction between MSPCs and hematopoietic cells (Table 2).

Table 2.

Skeletal and hematopoietic phenotypes from experimental manipulation of niche cellular populations in vivo.

Cell Population	Experimental Manipulation	Bone Phenotype	Effects on Hematopoiesis	References
CD146+	Subcutaneous transplantation	Ectopic bone formation	Can recruit hematopoietic marrow	[20-22]
CD105+ Thy1.1− 6C3−	Renal capsule transplantation	Ectopic bone formation	Multilineage mesenchymal potential in vitro Can give rise to stromal cells, chondrocytes and osteoblasts in vivo Can recruit hematopoietic marrow	[32, 33]
CD105+ Thy1.1+ 6C3−	Renal capsule transplantation	Ectopic bone formation	Enriched in expression of osteocalcin Cannot recruit hematopoietic marrow	[32, 33]
CD105+ Thy1.1− 6C3+	Renal capsule transplantation	Ectopic bone formation	Contribute to HSC niche	[33]
CD105− Thy1.1+ 6C3−	Co-culture and transplantation of HSCs		Contribute to B lymphocyte niche	[33]
CXCL12+	Ablation by diphtheria toxin	Impaired osteogenic and adipogenic potential of bone marrow cells	Loss of bone marrow CLPs, GMPs, MEPs, pro-B cells and proerythroblasts Reduced frequency of bone marrow HSCs Increased HSC quiescence Increased expression of myeloid genes in HSCs	[37]
Nestin+ mesenchymal progenitors	Subcutaneous transplantation	Ectopic bone formation	Can recruit hematopoietic marrow	[29]
Nestin+ mesenchymal progenitors	Ablation by diphtheria toxin		Loss of bone marrow HSPCs	[29]
Nestin+ mesenchymal progenitors	Deletion of SCF		No loss of bone marrow HSPCs	[31]
Nestin+ mesenchymal progenitors	Deletion of CXCL12		No loss of bone marrow HSPCs	[30]
LepR+ mesenchymal progenitors	Deletion of SCF		Loss of bone marrow HSPCs	[31]
LepR+ mesenchymal progenitors	Deletion of CXCL12		Peripheral mobilization of HSPCs	[30]
Prx1+ mesenchymal progenitors	Deletion of CXCL12		Loss of bone marrow HSPCs, CLPs, prepro-B cells Loss of HSC quiescence Peripheral mobilization of HSPCs	[30, 36]
Osx+ osteoprogenitors	Deletion of Gsα	Reduced trabecular and cortical bone	Arrest in B cell development at pro-B stage	[52]
Osx+ osteoprogenitors	Deletion of CXCL12		Loss of bone marrow CLPs and prepro-B cells Peripheral mobilization of HSPCs	[36]
Osx+ osteoprogenitors	Deletion of VHL	Increased trabecular bone	increase in HSCs and erythroid progenitors Decrease in peripheral blood lymphocytes	[75]
2.3 kb ColI+ osteoblast	Ablation by GCV administration	Reduced trabecular bone	Decrease in B cell precursors, erythroid progenitors Later decline in HSCs	[18]
2.3 kb ColI+ osteoblast	Expression of constitutively active PPR	Increased trabecular bone and peritrabecular stromal cells	Increase in HSCs	[15]
2.3 kb ColI+ osteoblast	Deletion of CXCL12		Loss of CLPs	[30]
2.3 kb ColI+ osteoblast	Deletion of SCF		No hematopoietic phenotype	[31]
2.3 kb ColI+ osteoblast	Expression of constitutively active Gs- coupled 5HT4 serotonin receptor	Increased trabecular bone	Decrease in HSCs Delayed recovery of megakaryocytes and erythroid cells after injury	[55]
Osteocalcin+ osteoblast	Ablation by GCV administration	Increased bone mass and cortical thickness	No hematopoietic phenotype	[19]
Osteocalcin+ osteoblast	Deletion of CXCL12		No hematopoietic phenotype	[36]
Dmp1+ osteocytes	Deletion of Gsα	Osteopenia	Increased neutrophils and platelets Decreased erythroid progenitors	[54]
Dmp1+ osteocytes	Deletion of PPR		No hematopoietic phenotype	[54]
Dmp1+ osteocytes	Expression of constitutively active PPR	Increased trabecular bone but no increase in peritrabecular stromal cells	No change in frequency of HSCs	[50]
Tie2+ endothelial cells	Deletion of SCF		Loss of bone marrow HSPCs	[31]
Tie2+ endothelial cells	Deletion of CXCL12		No hematopoietic phenotype (Greenbaum) Slight reduction in bone marrow HSC frequency (Ding)	[30, 36]

CD146+ cells

As described earlier the melanoma-associated cell adhesion molecule MACM/CD146, a cell adhesion molecule that belongs to the immunoglobulin superfamily, is expressed in mesenchymal stromal cells and to a lesser extent, the endothelium. Sachetti et al. demonstrated that human bone marrow perivascular reticular CD45-CD146+ cells are enriched for CFU-Fs [21]. In addition, undifferentiated CD146+ BMSCs expressed numerous markers for HSC niche related transcripts (CXCL12, Jagged-1 and SCF) and mural cells/pericytes (α-SMA, NG2 and PDGFRβ) [21]. When transplanted into immunocompromised mice, purified CD45-CD146+ cells formed in in vivo the bone tissue, and recruited the recipient endothelial cells and HSPCs to form heterotopic bone marrow in which the CD146+ cells are positioned in the vicinity of the endothelium. Furthermore, the CD146+ cells exhibit many features of vascular mural cells, or pericytes. Co-culture of endothelial cells with CD146+ cells promoted endothelial tube formation and stability. Such findings coincide with other reports that HSPCs reside in the perivascular area and receive trophic support from endothelial cells, perivascular cells and even the osteoblasts. HSPC-supportive CD146+ cells are not only found in the adult bone marrow but also in human fetal bone marrow and adipose tissue [26]. In co-culture studies CD146 participates in the maintenance of HSC quiescence [21, 27]. Recent reports further demonstrated that human bone marrow CD45-CD31-CD71-CD146+CD105+nestin+ cells can be cultured as multipotential mesenchymal spheres that support long term repopulating HSC (LT-HSC) activity in culture [28].

Nestin+ cells

Nestin is an intermediate filament protein that is commonly expressed in the neural or glial progenitors. Mendez-Ferrer et al. discovered that Nestin+ cells represent a small population of the adult mice bone marrow that have MSPC characteristics [29]. Bone marrow-derived Nestin+ stromal cells have self-renewal capacity and contribute to the normal osteoblast lineage cell turnover in the adult bone. When transplanted into recipient mice, they formed ectopic bone complete with hematopoietic bone marrow. Similar to CD146+ cells, bone marrow Nestin+ cells are mostly found in the perivascular area and express a myriad of HSPC trophic factors, including Ang1, SCF and Cxcl12. Diphtheria toxin-mediated Nestin+ cell ablation led to the dysregulation of the HSPC homeostasis [29]. However, ablation of CXCL12 or SCF using Nestin-Cre did not significantly affect the HSPC functions and numbers in the bone marrow [30, 31], suggesting that the population of cells expressing Nestin-Cre are not the primary source of either of those factors in vivo.

CD105+Thy1.1− cells

Another niche-forming BMSC is found in the mice fetal bones [32]. Chan et al. isolated the CD105+ Thy1.1− cells from the fetal bones and transplanted them under the kidney capsule of recipient mice. The cells are capable of forming ectopic bone via endochondral ossification with recruitment of host vasculature and can support host-derived LT-HSCs. In contrast, CD105+ Thy1.1+ cells, enriched for markers of mature osteoblasts and therefore likely representing a more differentiated osteoblast population, can form ectopic bone but fail to generate a hematopoietic marrow [32]. Further analyses revealed that CD105+ Thy1.1− cells that are also negative for 6C3 (ENPEP/Aminopeptidase A) are clonogenic stem cells that can generate the skeletal, chondrogenic and adipogenic cells. In vitro culture of the sorted CD105+ Thy1.1− 6C3− cells generates progeny cells that are CD105+ Thy1.1− 6C3+ and can effectively support the in vitro HSPC maintenance and their transplantability. CD105+ Thy1.1+ 6C3− cells, on the other hand, are osteogenic but lack HSPC supportive functions [33]. Interestingly, more differentiated CD105− Thy1.1+ 6C3− cells, though lacking the HSPC supportive functions, specifically support B-lymphopoiesis. The results indicate that multiple types mesenchymal or osteoblast progenitors are involved in the HSPC maintenance in the bone marrow; furthermore, a lineage-specific niche may exist for other hematopoietic lineages.

Leptin receptor-expressing cells

SCF is required for maintenance of HSPCs in vivo [34]. SCF-GFP knock-in mice demonstrated that SCF is expressed by perivascular cells [31], which also express the leptin receptor (LepR). Ablation of SCF in LepR-expressing cells leads to reduced frequency of bone marrow HSCs [31]. Similarly, ablation of CXCL12 in LepR-expressing cells results in HSC mobilization into the periphery [30]. Thus LepR marks a population of perivascular cells with roles in HSPC retention and maintenance.

CXCL12+ cells

CXCL12 is one of the critical HSPC maintenance factors in the bone marrow. Genetic ablation of CXCL12 leads to the HSPC pool alteration and loss of long-term repopulating HSCs [35]. However, the expression of CXCL12 in the bone marrow is heterogenous. CXCL12-GFP or RFP knock-in mice have revealed that cells that express CXCL12 in the bone marrow are mostly perivascular in location [30, 36-38]. Isolated CXCL12-GFP cells from the bone marrow exhibited both osteogenic and adipogenic potential in vitro. Cell ablation using the diphtheria toxin receptor model revealed that loss of CXCL12-expressing cells immediately resulted in an imbalance of the HSPCs. Not surprisingly given the importance of MSPCs to the HSC niche, ablation of CXCL12 from mesenchymal progenitors with Prx1-Cre, expressed in the limb bud mesoderm, leads to loss of quiescent HSCs in the bone marrow and mobilization of hematopoietic progenitor cells into the periphery [30, 36]. Sorted Prx-1-Cre-expressing cells contain CFU-F activity and are capable of differentiating into both osteoblasts and adipocytes in vitro [36].

In vivo Lineage Tracing of MSPCs and osteoprogenitors

More recently, mouse genetics has been used to identify and track the fate of various mesenchymal progenitor populations in murine bone marrow in vivo during bone development, remodeling and fracture healing. In mice carrying green fluorescent protein (GFP) under control of the Nestin promoter, non-hematopoietic GFP+ cells are perivascular but distinct from CD31+ endothelial cells. In culture, Nestin-GFP+ cells contain all CFU-Fs, and can differentiate into adipocytes, chondrocytes and osteoblasts. In transplantation experiments, Nestin-GFP+ cells can self-renew and contribute to chondrocyte and osteoblast lineages in vivo [29].

Although widely used to delete genes in hematopoietic cells, the interferon-inducible Mx1 promoter is also expressed in mesenchymal stromal cells [39]. In particular Mx1+ cells are enriched in CFU-Fs and while they can give rise to adipocytes, chondrocytes and osteoblasts in vitro, lineage tracing in vivo revealed that Mx1+ progenitors are largely restricted to the osteoblast lineage [40].

Osterix (Osx) is a transcription factor expressed early in osteoblast development [5]. Osx+ osteoprogenitors are closely associated with invading vasculature during endochondral bone development, and can give rise to stromal cells, trabecular osteoblasts, and osteocytes [41]. In contrast, more mature osteoblasts expressing type I collagen are not found perivascularly, and cannot give rise to trabecular osteoblasts [41]. More recently, Song et al reported that Osx-expressing cells lacking β-catenin, a mediator of canonical Wnt signaling, can be fate shifted into the adipocyte lineage [42]. This was confirmed by Liu et al, who further demonstrated that in late embryogenesis Osx expression labels a population of stromal cells that persist into adulthood and can give rise to both osteoblasts and adipocytes in vivo [43]. However, Osx+ cells in adult mice are not self-renewing, and therefore represent only a transient population of osteoprogenitors [40], highlighting the potential differences between embryonic and adult progenitor populations.

The expression of alpha-smooth muscle actin (αSMA) has also been reported to mark an osteoprogenitor population in mice [44, 45]. During recovery from osteoblast ablation, αSMA expression is detected in early osteoblast precursors, and sorted αSMA-GFP+ cells have robust osteogenic and adipogenic potential [45]. In the bone marrow, αSMA-expressing cells are found in close association with the microvasculature, and in lineage tracing studies αSMA marks a population of osteoprogenitors that mature into osteoblasts and osteocytes in vivo. Cultured αSMA-expressing progenitors have chondrocyte, adipocyte, and osteoblast potential in vitro [44].

Osx expression marks a perivascular population of osteoprogenitors that also exhibit adipogenic potential [42, 43]. αSMA+ cells harvested from bone marrow have adipogenic potential in vitro [44, 45]. In adipose tissue, αSMA-GFP+ cells are also found in close association with vasculature [45]. Similarly, adipocyte precursors within the stromovascular fraction of adipose tissue [46] and expressing PPARγ [47], Zfp423 [48] and VE-cadherin [49] have all be reported to occupy a perivascular location.

Together, these studies demonstrate that mesenchymal stem/progenitor cells are located in the perivascular niche, where osteoblast and adipocyte progenitors have also been identified. Where these cells stand in the lineage hierarchy of mesenchymal precursors and how these populations overlap and differ in bone marrow and other tissues remains to be clarified.

Cells of the osteoblast lineage at specific stages of differentiation serve unique functions in supporting hematopoiesis

Beyond MSPCs, cells of the osteoblast lineage play stage-specific roles in supporting distinct hematopoietic niches. The effects of osteoblast lineage cells on specific hematopoietic populations depend on the maturational status, cytokine production and specific signaling pathways in the osteoblast lineage cells. The importance of the osteoblast differentiation stage in serving unique functions in the hematopoietic niche is best demonstrated by the differential hematopoietic phenotypes that result when the same gene is ablated by different osteoblast lineage stage-specific Cre drivers.

Regulation of HSPCs

Targeted expression of constitutively active parathyroid hormone receptor (Col2.3-caPPR) to maturing osteoblasts using the 2.3 kb type I collagen promoter leads to a significant increase in HSCs associated with increased osteoblast numbers [15]. However, targeting of the constitutively active PPR to DMP1-expressing osteocytes fails to increase HSCs despite an increase in osteoblasts [50]. Of note, Col2.3-caPPR mice exhibit a dramatic expansion of peritrabecular stromal cells that express markers of the osteoblast lineage [15, 51]. In contrast, Dmp1-caPPR mice lack such an expansion of stromal cells [50]. Given recent evidence that mesenchymal progenitors are crucial to HSC support, these studies raise the possibility that it is the expansion of the stromal cells, likely a population of osteoprogenitors, rather than an increase in mature osteoblasts, that is promoting an increase in HSCs ca Col2.3-caPPR but not Dmp1-caPPR mice.

Regulation of lymphopoiesis and myelopoiesis

The PPR is a G-protein coupled receptors (GPCR), and Gsα is a central major mediator of signaling downstream of GPCRs including PPR. Gsα deletion in Osx+ osteoprogenitors leads to an impaired B lymphopoiesis with a specific block in the transition from prepro-B to pro-B precursors and an increase in granulocytes but no major changes in the other hematopoietic lineages. This stage is critically dependent upon stromal cell-derived IL-7, and indeed IL-7 expression is significantly reduced in osteoprogenitors of Gsα^OsxKO mice [52]. Although IL-7 cytokine has not been specifically ablated yet from osteoblastic cells in vivo, the targeted overexpression of human IL-7 to maturing osteoblasts rescued the osteopenia and the B cell development of IL-7 KO mice without interfering with T lymphopoiesis [53].

In contrast, ablation of Gsα in osteocytes does not result in any alterations in B- and T-cells in bone marrow, spleen, and peripheral blood, demonstrating that osteocytes are not involved in lymphoid development. Instead mice carrying Gsα-deficient osteocytes (Gsα^DmpKO mice) developed a myeloproliferative-like syndrome characterized by neutrophilia, leukocytosis, thrombocytosis, and splenomegaly [54]. Erythroid cells are marginally but significantly decreased in the bone marrow indicating that Gsα-deficient osteocytes may also regulate erythropoiesis. Gsα^DmpKO mice transplanted with wild-type bone marrow rapidly developed a myeloproliferative phenotype whereas transplantation of myeloproliferative bone marrow from mutant mice into wild-type mice completely rescued the bone marrow phenotype, demonstrating that the myeloproliferative disorder in the Gsα^DmpKO mice is caused by an altered bone marrow microenvironment. Using a novel ex vivo system to co-culture osteocyte enriched bone explants with bone marrow cells, the authors found that Gsα-deficient osteocytes produce large amounts of G-CSF, an essential factor to promote granulopoiesis. Fulzele et al. also examined the hematopoietic phenotype of mice with osteocyte specific disruption of PPR (PPR^DmpKOmice). However, PPR^DmpKO mice did not show any hematological abnormalities [54]. Therefore, this study identifies that the osteocytes, differentiated cells of osteoblast lineage, interact with hematopoietic cells for myeloid but not lymphoid lineage development.

While ablation of Gsα in Osx+ osteoprogenitors results in osteopenia and impaired B lymphopoiesis, overexpression of constitutively active Gs-coupled 5HT4 serotonin receptor in osteoblasts driven by Col2.3 promoter (Rs1 mice) leads to a massive increase in trabecular bone formation, but no changes in B lymphocyte development [55]. Despite increased osteoblastic-lineage cell numbers, Rs1 mice showed decreased bone marrow cellularity and progressive loss of HSC numbers, perhaps due to decreased expression of key HSC-supportive factors that may explain the decrease in HSCs.

The model of distinct hematopoietic niche formations for MSCs, osteoprogenitors and osteoblasts has recently been strengthened by two studies that dissected the relative contributions of varying mesenchymal or stromal populations in the production of CXCL12, a chemokine with important roles in HSPC niche maintenance and B lymphopoiesis [30, 36]. As described earlier, deletion of CXCL12 in Prx1 and LepR-expressing mesenchymal stem-like cells leads to a loss of HSCs [30, 36]. The chemokine CXCL12 has been shown to regulate both HSCs and lymphoid progenitors and is expressed by a variety of stromal cells [21, 29, 56-58]. Deletion of CXCL12 from MSCs and Osx+ osteoprogenitors results in HSPC mobilization, while deletion from Osx+ osteoprogenitors and Col2.3+ maturing osteoblasts, but not terminally differentiated OC+ osteoblasts, led to a loss of B-lymphoid progenitors [30, 36]. These studies suggest that while maturing osteoblasts create an endosteal niche for certain early lymphoid progenitors, B cell progenitors and HSPCs depend on a distinct perivascular niche which includes MSCs and osteoprogenitors.

B lymphopoiesis is disrupted in models of altered skeletal homeostasis where stage-specific effects on the osteoblast lineage have not been fully characterized. Bone homeostasis results from a fine balance between bone-forming osteoblasts and bone-resorbing osteoclasts. Regulation of hematopoiesis by the osteoblastic cell lineage is also influenced by the intimate relationship of osteoblastic cells with osteoclasts. The absence of resorption leads to an osteopetrotic phenotype with increased bone mass and severe alteration in the bone microenvironment accompanied by reduced B-cell development [59-62]. In the osteopetrotic mice B lymphopoiesis is dramatically altered in the transition between pro- and pre-B cells [59, 63, 64] with a significant reduction in the expression of CXCL12, IL-7 and osteoblastic genes, and reduction in BM osteoprogenitors as assayed by CFU-ALP [65].

Although osteocytes appear not to play a direct role in regulating B lymphopoiesis, they do regulate osteoblast differentiation by production of sclerostin (SOST). Sclerostin is a secreted glycoprotein which inhibits the canonical Wnt signaling and, thus, has anti-anabolic effects on bone formation [66]. Despite a high bone mass, SOST KO mice displayed a specific reduction in B lymphocyte number only in bone marrow due to increased apoptosis rather than impaired differentiation [67]. Whether osteoprogenitors are specifically involved was not discussed but the CXCL12 mRNA levels are reduced [67]. No changes were noted in bone marrow HSPC frequency or numbers, but effects on HSPC mobilization were not assessed. In addition, as Wnt signaling regulates CXCL12 gene expression [68], it is also possible that an overactive Wnt signaling, in the absence of SOST, in stromal cells and osteoprogenitors results in a short supply of CXCL12 that is not sufficient for B lymphocyte survival.

Regulation of megakaryopoiesis

The osteoblast lineage has also been implicated in the regulation of platelet production, or megakaryopoiesis. Early megakaryocyte progenitors can be expanded in vitro when cultured in the presence of human osteoblasts [69] or MSCs [70], an interaction mediated by Kit ligand and SCF [71]. In the bone marrow, megakaryocyte progenitors interact with perivascular sinusoidal cells identified as endothelial cells that express VE-cadherin, VCAM1 and CXCL12 [72]. However, perivascular osteoprogenitors also express these factors, and osteoblasts are a source of thrombopoietin (TPO), a major megakaryocyte growth factor [24]. Of note, several murine models of increased megakaryocytes also have increased bone mass [73], and migration of megakaryocytes to the endosteal niche is required for HSC engraftment after irradiation [74]. These studies hint at complex interactions between megakaryocytes, HSCs and the osteoblast lineage within the bone marrow.

Regulation of erythropoiesis

Erythrocytes play a vital role in oxygen delivery to tissues. Erythropoiesis occurs in the bone marrow, where differentiation of erythroblast precursors is regulated by the glycoprotein hormone erythropoietin (EPO). EPO is produced primarily by peritubular cells of the kidney in response to hypoxia, a process mediated by the hypoxia-inducible factor (HIF) signaling pathway. However, recently Rankin et al [75] found that osteoblasts can unexpectedly be a major producer of EPO in bone marrow. Disruption of VHL, an inhibitor of the HIF signaling pathway, in osteoprogenitors using Osterix promoter (Osx-VHLKO mice) resulted in excessive trabecular bone accumulation due to increased osteoblast numbers. Increased and dilated vascularization was also observed along with increased VEGF expression by osteoblasts as was previously reported [76]. Osx-VHLKO mice also had a three-fold increase in HSCs in a HIF-dependent manner indicating that HIF signaling in osteoblasts controls HSC numbers. In addition, Osx-VHLKO mice had a 60% increase red blood cells (leading to polycythemia) and a 75% decrease in the numbers of lymphocytes. Erythroid progenitors were significantly increased both in bone marrow and spleen of Osx-VHLKO mice due to markedly increased overexpression of EPO by osteoblasts. Surprisingly, bones from Osx-VHLKO mice showed a 6-fold increase in EPO mRNA expression over the expression in kidney from control mice, indicating that VHL-deficient osteoblasts can be a significant source of EPO in the bone marrow compartment [75]. In contrast, ablation of HIF signaling in osteoblasts has no impact on erythropoiesis, suggesting that HIF signaling can augment erythropoiesis but is not required to maintain normal erythrocyte levels [75].

Conclusions

In summary, the bone marrow microenvironment is a complex interaction of mesenchymal, hematopoietic and other cellular lineages. It is becoming increasingly apparent that cells at various stages of differentiation play distinct roles in supporting specific subsets of hematopoietic cells and lineages. Future studies will need to clarify how these mesenchymal and osteoblast lineage populations overlap, as well as the underlying cellular and molecular mechanisms of hematopoietic support.