Osteoblasts support B-lymphocyte commitment and differentiation from hematopoietic stem cells

It has long been hypothesized that early B lymphopoiesis in mammals is induced within the bone marrow (BM) microenvironment, but which cell or cells specifically constitute this B-cell niche is not known. Osteoblasts are the differentiated stromal cell type unique to the bone marrow environment, and therefore are a candidate for a cell type to play a specific role in perhaps multiple stages in hematopoiesis, including perhaps B lymphopoiesis. Previously, we found that purified human osteoblasts play a pivotal role in myelopoiesis, both by secreting numerous hematopoietic cytokines that support terminal neutrophil maturation and support the in vitro expansion of primitive long-term culture-initiating cells (LTCICs).^1,2  More recently, analyses of bone morphogenetic protein receptor 1A–deficient (BMPR1A^−/−) mice suggest that increased osteoblast number and/or activity increases the pool size of repopulating hematopoietic stem cells (HSCs) in vivo,³  suggesting that osteoblasts' stimulatory effect may extend as far back in hematopoietic differentiation to the level of the uncommitted HSCs.^4,5  Whether osteoblasts play a specific role in supporting and/or directing the differentiation of HSCs to lymphoid commitment and differentiation is, however, less clear.

We sought to test directly whether osteoblasts play a pivotal role in B lymphopoiesis over and above any role in HSC maintenance. To ask whether osteoblasts are sufficient to support and induce HSCs to B lymphopoiesis, we have measured the production of early B-cell precursors from enriched HSCs cultured on purified osteoblasts in vitro. Conversely, to ask whether osteoblasts are necessary to support and induce HSCs to B lymphopoiesis and whether osteoblasts are necessary for early B lymphopoiesis, we have examined early alterations in hematopoiesis immediately following osteoblast depletion in Col2.3Δ-TK transgenic mice treated with GCV.⁶  The results of these studies show that osteoblasts are both necessary and sufficient for the support of all stages of B lymphopoiesis, from the first transition from HSC to common lymphoid progenitor to the generation of mature IgM⁺ B cells. We therefore conclude that the osteoblast comprises a key component of the bone marrow B-cell niche.

Rat PTH was purchased from Bachem Biochemicals (Torrance, CA). Dexamethasone, insulin, ascorbic acid, and β-glycophosphate were purchased from Sigma (St Louis, MO). Murine c-Kit ligand, IL-3, IL-6, IL-7, and BMP-2 and antibodies, including lineage-specific antibodies for HSC isolation as well as blocking antibodies against IL-7, SDF-1, TSLP, M-CSF, and VCAM-1, were obtained from R&D Systems (Minneapolis, MN). Blocking antibodies against mouse integrin α4 and ICAM-1 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). S17 stromal cells (provided by Dr John Monroe, University of Pennsylvania) were cultured first in αMEM + 20% FBS and then in DMEM + 10% FBS. MS1 endothelial cells (provided by Dr Mark Kahn, University of Pennsylvania) were cultured in DMEM + 5% FBS. All stromal cultures were maintained in 25-cm² flasks and passaged prior to confluence.

Female and male B6.SJL-Ptprc^aPep3^b/BoyJ (SJL, CD45.1⁺) and C57Bl/6J (B6, CD45.2⁺) mice at 8 to 16 weeks of age were purchased from The Jackson Laboratory (Bar Harbor, ME). RAG2 GFP NG BAC mice (B6) were kindly provided by Dr David Allman (Department of Pathology, University of Pennsylvania, Philadelphia, PA). Transgenic mice bearing a fusion gene composed of the 2.3-kb fragment of the rat type I collagen α1 (Col1α1) promoter (whose activity is largely restricted to differentiated osteoblasts) and HSV-TK (Col2.3Δ-TK)⁷  were graciously provided by Dr David W. Rowe (Department of Genetics and Developmental Biology, University of Connecticut Health Center, Farmington, CT).

Six-week-old mice bearing the Col2.3ΔTK transgene and aged-matched control CD-1 mice were injected intraperitoneally daily for up to 28 days with 5 mg/kg gancyclovir (GCV; Roche Pharmaceutical, Nutley, NJ) in saline. Mice were killed at weekly intervals and hematopoietic cells harvested from all long bones and spleens. Loss of osteoblasts was confirmed by measurement of osteocalcin mRNA in situ.

Osteoblasts were isolated from calvaria of newborn mice by collagenase digestion. Briefly, calvaria of 6 to 8 mice from whom soft tissue was removed were pretreated in 50 mL of 10 mg/mL collagenase P for 20 minutes at 37°C (Roche Applied Science, Indianapolis, IN) to remove residual connective tissue debris. After washing, the treated calvaria were minced into (approximately 1 mm²) fragments and digested 3 times with 50 mL of 10 mg/mL collagenase P at 37°C for 10 to 15 minutes. The supernatants collected from all 3 digestions were pooled together and centrifuged. The pelleted cells were resuspended in 10% FCS/α-medium supplemented with ascorbic acid and β-glycerophosphate and kept at culture conditions of 37°C and 5% CO₂. Under these conditions, the cells quickly adhered to the plastic and grew to confluence in 3 to 4 days. Aliquots of the adherent cells were released by digestion with 1 mg/mL collagenase P and 0.25% trypsin for immunostaining and in situ mRNA analyses. The maturation and purity of osteoblasts were examined by the in situ detection of osteopontin, as well as by flow cytometric detection of multiple antigens including CD45, PECAM-1, Pan-endo, and VCAM-1.

To compare osteoblastic versus adipocytic differentiation of primary calvarial cells, osteoblast differentiation was induced by the addition of 10 mM β-glycerol phosphate, 50 μM ascorbate-2-phosphate, 10⁻⁷ M dexamethasone, and 100 ng/mL human BMP-2 into 10% FCS/α-medium for 5 to 6 days until the cells reached confluence, followed by digestion, and replating in the presence of ascorbic acid and β-glycerophosphate as well as 1 × 10⁻⁷ M PTH for 4 to 5 days. For adipocytic induction, the primary calvarial cells were cultured in the presence of 10⁻⁷ M dexamethasone and 10 μg/mL human insulin for 10 to 14 days.^8,9 

Osteoblasts were preseeded onto sterile glass cover slides in 6-well plate for 5 days. Then the glass slides were picked up and fixed and stained by standard immunochemistry procedure. Antibodies against osteopontin, osteocalcin, and control IgG were purchased from Santa Cruz Biotechnology. Stained cells were visualized with a Nikon Alpha phot-2 Y52 microscope (Nikon, Tokyo, Japan) under a 160/0.17 (100×) objective lens. The measurement of IL-7 or SDF-1 concentration by enzyme-linked immunosorbent assay (ELISA) in the supernatant of monolayer culture was performed as per the manufacturer's protocols (R&D Systems). Intracellular staining of IL-7 was performed following the protocol provided by Cytofix/Cytoperm Plus Kit (Pharmingen, San Diego, CA).

Freshly isolated BM cells from the hind legs of 8- to 10-week-old mice were washed once with 2% FCS in PBS, and their cell concentration was adjusted to 5 × 10⁷/mL. To this suspension was added APC-coupled c-Kit antibody, FITC-coupled Sca-1 antibody, and PE-coupled cocktail of lineage antibodies including those against Mac-1, Gr-1, B220, Ter119, and CD3 (Pharmingen). The Lin⁻Sca-1⁺ or Lin⁻Sca-1⁺c-Kit⁺⁺ HSCs were isolated on a MoFlo cell sorter (Dako, Carpinteria, CA).

To isolate RAG2⁻ HSCs, Sca-1⁺ cells were first isolated by magnetic sorting (VarioMACS; Miltenyi Biotec, Auburn, CA) of bone marrow mononuclear cells from RAG2 GFP NG BAC mice. The positively selected Sca-1⁺ cells were then stained with Lin-PE and c-Kit-APC. GFP⁻ LSK (0.4% of total Sca-1⁺ BM cells) and GFP⁺ LSK (0.1% of total Sca-1⁺ BM cells) cells were purified by fluorescence-activated cell sorting (FACS).

Osteoblasts and osteoblastic or endothelial cell lines were plated into 6-well plates in 10% FCS/α-medium supplemented with ascorbic acid, β-glycerophosphate, and 1 × 10⁻⁷ M PTH for 3 to 4 days, allowing the cells to grow to confluence and forming a gellike collagen pad. Then FACS-sorted Lin⁻ (B220, Mac-1, Gr-1, CD3, Ter119) Sca-1⁺ (CD45.1⁺) BM cells from SJL mice were seeded into the precultured osteoblasts with or without cytokines in 10% FCS/α-medium supplemented with ascorbic acid and β-glycophosphate for 5 to 6 days. The adherent hematopoietic cells were released by digestion with 1 mg/mL collagenase P and 0.25% trypsin for 10 minutes at 37°C. For noncontact coculture, a 0.4-μm Transwell membrane insert (Corning Company, Acton, MA) was placed into the well of precultured osteoblasts, and the HSCs were added to the insert chamber.

Hematopoietic cells adherent to osteoblast monolayers were released into suspension with 1 mg/mL collagenase P and 0.25% trypsin for 10 minutes at 37°C. The remaining gel pad and debris were removed by passaging the digested solution through a 70-mm screen. The recovered cells containing both hematopoietic and osteoblastic cells were stained with APC-labeled CD45.1 antibody, FITC-labeled B220 antibody, and PE-labeled combined antibodies against Mac-1, Gr-1, Ter119, and NK1.1. The FACS-sorted CD45.1⁺B220⁺Mac-1/Gr-1/Ter119/NK1.1⁻ cells were verified to be CD19⁻. Then, 2 × 10³ sorted CD45.1⁺ cells were injected into the tail veins of sublethally irradiated B6 mouse. Three weeks after transplantation, the hematopoietic tissues were collected, and CD45.1⁺ cells were detected and analyzed for lineage-specific markers.

To ask whether osteoblasts are able to play a role in supporting lymphopoiesis, we examined the ability of murine osteoblasts to support the differentiation of lineage-depleted Sca⁺ (Lin⁻Sca-1⁺) BM cells to B-cell precursors. We first purified murine osteoblasts from newborn C57Bl/6 mice (CD45.2) in the presence of 10⁻⁷ M PTH.^10,11  Cytospins of these cultured cells demonstrated a morphologically homogenous population of cells that expressed membrane-associated osteoblast-related protein osteopontin (Figure 1A). These cells were not detectably contaminated by either hematopoietic cells (CD45.2⁺) or endothelial cells (Figure 1B). Of interest, the osteoblasts expressed VCAM-1, which has been hypothesized to mediate early B-precursor adhesion to a previously uncharacterized population of BM stromal cells,^12,13  as well as Sca-1, CD61, and ICAM-1 (Figure 1C).

When Lin⁻Sca-1⁺ (LS) BM cells from adult congenic SJL mice (CD45.1) were added to these osteoblast monolayers, the numbers of hematopoietic cells increased 20-fold over 6 days in culture. This hematopoietic proliferation was strongly dependent on contact with osteoblasts, as interposition of a porous (Transwell) membrane prevented more than 95% of the inductive effect (Figure 2A). Most of the cells produced in the Lin⁻Sca⁺/OB contact cultures were myeloid precursors, but approximately 5% of the hematopoietic cells recovered from the adherent monolayer following collagenase/trypsin digestion were B220⁺. In addition to these adherent hematopoietic cells, a similar number of B220⁺ cells were found released into the culture supernatant (Figure 2A right-hand panel). No B220⁺ cells were found in Transwell noncontact cultures. The adherent B220⁺ cells recovered from the adherent compartment of the Lin⁻Sca⁺/OB contact cultures expressed low to medium levels of B220⁺ but no CD19 or IgM. Of these, 14% were CD43⁺CD24^−/lo pre-pro-B cells and 45% were CD43⁺CD24⁺ pro-B cells.¹⁴  In contrast to the adherent cells, B-lymphoid cells found in suspension were more mature, expressing higher levels of B220 and including subpopulations expressing CD19 (7%) and IgM (9%) (Figure 2B).

The production of IgM⁺ B lymphocytes from Lin⁻Sca-1⁺ cell/osteoblast cocultures suggests that osteoblasts induce all developmental transitions required for full B-cell development. However, since many other cell types are generated during these cocultures, it is theoretically possible that the later stages of B-lymphocyte maturation are stimulated indirectly, by other hematopoietic or stromal cells derived from the multipotential Lin⁻Sca-1⁺ cells. To examine whether osteoblasts alone directly stimulate these stages of B-cell development, we isolated IgM⁻Mac1⁻NK1.1⁻B220⁺CD19⁻ pre-pro-B and IgM⁻Mac1⁻NK1.1⁻B220⁺CD19⁺ pro-B/pre-B cells from normal BM and directly seeded them onto congenic osteoblast monolayers. As shown in Figure 2C, osteoblasts stimulated the survival and expansion of B220⁺CD19⁻ pre-pro-B and B220⁺CD19⁺ pro-B/pre-B cells over 6 days, and pretreatment of the osteoblast monolayer with PTH enhanced this ability. In addition, a substantial fraction of both purified B220⁺CD19⁺ pro-B and B220⁺⁺CD19⁺ pre-B cells differentiated to become IgM⁺ B lymphocytes (Figure 2D). Purified B220⁺CD19⁻ pre-pro-B generated very few IgM⁺ B lymphocytes over these same 6 days, but one fifth became CD19⁺. These data demonstrate that osteoblasts play a direct role in stimulating all stages of BM B lymphopoiesis, from the commitment of HSCs to lymphopoiesis to the terminal differentiation of IgM⁺ B lymphocytes.

Next we asked whether osteoblasts support the earliest lineage decisions of primitive HSCs to B-lymphoid commitment per se. To focus on this early step, we initiated cultures with more highly purified early hematopoietic stem cells that excluded common lymphoid or committed lymphoid progenitors. For this purpose, we isolated GFP⁻ Lin⁻Sca⁺c-Kit⁺ (LSK) BM cells from RAG2 GFP NG BAC mice (B6),¹⁵  a recombinant B6 transgenic strain in which all cells that express recombinase activate a GFP transgene. Three-thousand sorted GFP⁻ LSK cells were cultured on either primary osteoblasts, the endothelial cell line MS-1, or the permissive stromal line S17. Over 14 days in coculture with osteoblasts, 15% of the cells generated from these GFP⁻ LSK cells became GFP⁺, nearly identical to the numbers of GFP⁺ cells produced by GFP⁻ LSK cells cultured with S17 cells (1.64 × 10⁴ versus 1.67 × 10⁴). In contrast, no GFP⁻ LSK cells cultured on endothelial monolayers became GFP⁺ (Figure 3A). Of note, these GFP⁻ LSK cells cultured on OB-generated B220⁺CD19⁺GFP⁺ B-lymphocyte precursors, at efficiency very similar to that of cultured GFP⁺ LSK cells, which already expressed recombinase (Figure 3B-C). Thus, osteoblasts stimulate early transitions from HSCs to committed lymphoid progenitors. Taken together, these data demonstrate that osteoblasts play a direct role in stimulating all stages of BM B lymphopoiesis, from the commitment of HSCs to lymphopoiesis to the differentiation to IgM⁺ B lymphocytes.

To document that the adherent population of B220⁺CD19⁻ cells induced and maintained by osteoblasts included true functional precursors that develop into mature B cells, we transplanted 2 × 10³ FACS-purified B220⁺CD19⁻ (CD45.1⁺) cells from LS/OB cocultures, released and purified from 10-day osteoblast cocultures, into sublethally irradiated congenic (CD45.2) mice. After 3 weeks, a substantial contribution of CD45.1⁺ cells was found in peripheral blood, spleen, and bone marrow, but not thymus (Figure 4A), and flow cytometric analysis confirmed the presence of CD45.1⁺ mature B220⁺⁺CD19⁺IgM⁺ B lymphocytes in both BM and spleen (Figure 4B). Of interest, we also detected donor-derived B220⁻CD19⁻ non-B cells in recipient BM, most of which were Mac-1⁺CD11c⁺CD3⁻NK1.1^−/lo (Figure 4C). This suggests that the adherent B220⁺CD19⁻ population also includes precursors for macrophages and dendritic cells, whether originating from a distinct or inherently bilineage precursor, as indicated by previous studies of Pax5^−/− lymphoid progenitors.^16–18 

Next, we investigated which cytokines and adhesion molecules derived from osteoblasts are required for B-cell commitment and maturation.^19,20  Previously, we had demonstrated that SDF-1 is secreted by human osteoblasts, as well as in other BM stromal and endothelial cells.²¹  In addition to high concentrations of SDF-1, we found that the cultivated osteoblasts synthesized and secreted measurable quantities of IL-7 when PTH was added (Figure 5A), confirming previous reports that IL-7 is required for early B lymphopoiesis.²⁰  To examine the relationship between osteoblast development and their ability to synthesize IL-7 and SDF-1, we either induced (by the addition of BMP-2 plus PTH) or inhibited (by the addition of adipocyte-inducing agents insulin plus dexamethasone) osteoblastic differentiation from freshly isolated calvarial cells. Whereas PTH-stimulated, fully differentiated osteoblasts secreted both SDF-1 and IL-7, and induced B lymphopoiesis, inhibition of osteoblast differentiation and support of adipocyte differentiation with insulin and dexamethasone blocked IL-7 production and differentiation of B220⁺ precursors, despite continued elevated level of SDF-1 (Figure 5B). Conversely, inclusion of neutralizing antibodies to control LS/OB cocultures confirmed that IL-7 and SDF-1 as well as TSLP (thymic stromal cell–derived lymphopoietin), but not MCSF, were required for the stimulation of B-cell development by osteoblasts (Figure 5C). In addition, we found that inclusion of antibody against either VCAM-1 or integrin α4 significantly reduced the generation of B220⁺ B lymphocytes, emphasizing the importance of the VCAM-1/integrin α4 ligation-mediated signaling in the contact-mediated induction of B lymphopoiesis by OBs (Figure 5D).

Two recent reports suggest that HSCs may localize preferentially to BM endothelial cells.^22,23  These results, combined with the previous observations of the role of osteoblasts in maintaining HSC pool size, raise the possibility that the HSC niches may be complex, requiring the interaction of osteoblasts, endothelial cells, and perhaps other mesenchymal components. To begin to test this concept functionally, we measured the effects of adding a cocktail of cytokines produced by nonosteoblast BM stromal cells, including c-Kit ligand, IL-3, and IL-6 into osteoblast/Lin⁻Sca⁺ (LS) BM cell cocultures. In these cultures, cytokine addition significantly augmented the ability of osteoblastic monolayer to support the survival and expansion of LS cells, but inhibited the production of B220⁺ B lymphocytes. Cultures initiated on OB monolayers with 10⁴ LS cells produced 1.0 × 10³ Sca⁺Kit⁺ cells and 1.4 × 10³ B220⁺ cells (of 1.2 × 10⁵ total adherent plus nonadherent hematopoietic cells), whereas cultures initiated on OB monolayers supplemented with c-KitL/IL-3/IL-6 produced 1.8 × 10⁴ Sca⁺Kit⁺ cells and 0.8 × 10³ B220⁺ cells (of 2.6 × 10⁵ total hematopoietic cells) (Figure S1, available on the Blood website; see the Supplemental Figure link at the top of the online article). These results suggests the possibility of specific microniches on the osteoblastic surface that support varying degrees of HSC self-renewal, myelopoiesis, or B lymphopoiesis depending on the relative proximity of other BM stromal cells such as endothelial cells.

If the ability of osteoblasts to support B lymphopoiesis in vitro reflects a key role in vivo, we hypothesized that elimination of osteoblasts in vivo would result in loss of B-cell precursors, beyond any effect on HSC numbers per se. Recently, Visnjic et al⁶  performed elegant studies demonstrating that conditional ablation of endosteal osteoblasts in Col2.3Δ-TK transgenic mice by GCV administration in vivo for 21 to 28 days resulted in the depletion of a wide range of hematopoietic cells from the bone marrow, including HSCs as well as more mature myeloid and B-lymphoid precursors. Since BM HSCs were themselves depleted, the simplest explanation for the loss of myeloid and lymphoid precursors was the loss of the HSCs from which they derive. However, an alternative possibility was that OBs might play a more direct role in B lymphopoiesis in vivo, beyond the level of HSC support. Given the present findings that cultured OBs directly stimulate all stages of B lymphopoiesis, we sought to directly test this hypothesis by examining early alterations in hematopoiesis immediately following osteoblast depletion, prior to HSC loss, by GCV in Col2.3Δ-TK transgenic mice. BM biopsies from untreated mice documented robust hematopoietic activity, and showed abundant immunostaining signals for the osteoblast-specific matrix protein osteocalcin. (Figure 6). Osteocalcin-specific osteoblasts were depleted as early as 8 days following GCV treatment. In parallel, total BM hematopoietic cellularity, but not splenic cellularity, declined by 50% during this 8-day GCV treatment (Figure 6A). Flow cytometric analysis of small mononuclear cells (Mac-1⁻NK1.1⁻) from BM of control mice identified 4 distinct stages of B-lineage development: B220⁺CD19⁻ pre-pro-B cells, B220⁺CD19⁺ pro-B cells, B220⁺⁺CD19⁺ pre-B cells, and B220⁺⁺CD19⁺⁺ immature B cells. Osteoblast ablation resulted in the depletion of each of these B-lymphocyte subsets by day 8, with the greatest reduction in pre-pro-B (64%) and pro-B (81%) cells (Figure 6B). Consistent with previous findings showing that osteoblasts support in vitro myeloid progenitor survival/expansion,²  the number of Gr-1^loc-Kit⁺ myeloid progenitors in the BM was also reduced (60%), as were the number of mature monocytes and granulocytes by 28% and 42%, respectively (Figure 6C). Erythroid (gly A⁺) and megakaryocytic (CD41⁺) cells remained at 90% to 100% of controls. Despite these effects on B lymphopoiesis, and to a lesser extent on myelopoiesis, the number of LSK cells, which includes both long-term and short-term HSCs, was not yet reduced by day 8 after osteoblast depletion with GCV (Figure 6D). In contrast, LSK cells were not detectably depleted until day 21 (Figure 7A). ⁶  Of note, the number of mature B lymphocytes in the spleen did not decline at day 8 (Figure 7B), underscoring the organ-specific depletion of early stages of B cells from BM tissue by GCV treatment. These results confirm that osteoblasts play a direct role in BM B-cell precursor production, beyond simply supporting HSC survival. Continued evaluation over 28 days following osteoblast depletion confirmed that BM pre-pro B and pro-B subsets continued to decline further and more rapidly than LSK, with BM pre-pro and pro-B cells falling to less than 1% of control by 28 days (Figure 7A, C-D). These data indicate that acute depletion of osteoblasts causes an abrupt decline in B-cell precursors in the bone marrow, confirming their ability to directly support B lymphopoiesis.

These studies provide direct evidence that osteoblasts are necessary and sufficient for supporting all stages of BM B lymphopoiesis, via inductive signals that include VCAM-1–mediated adhesion and locally secreted IL-7 and SDF-1. Thus, whereas in birds specialized cells in the bursa of Fabricius may form the B-cell niche, in mice this role has been subsumed by BM osteoblasts.

The present direct in vitro culture studies show that osteoblasts support HSC differentiation to B lymphopoiesis. Moreover, by isolating specific developmental stages and assaying cellular differentiation on osteoblasts, we have found that osteoblasts support each stage of differentiation from HSC to IgM⁺ B cell. In particular, osteoblasts support the transition from Rag2⁻ HSC to Rag2⁺ committed lymphoid progenitor, thereby dictating B-lymphoid commitment per se as well as subsequent steps in B-lymphoid differentiation. Of note, this inductive effect is not shared by adipocytes or endothelial cells, but only by osteoblasts, and the inductive effect of osteoblasts is comparable with that seen with the S17 line, a stromal cell line of uncertain lineage with known B-cell inductive properties. Of note, our data also show that the early B-cell precursors produced in HSC-osteoblast cocultures are completely developmentally competent, as they are able to proliferate and expand to produce mature B cells after transplantation in vivo. Thus, osteoblasts appear to be able to support all developmental transitions in B lymphopoiesis. What molecular events are required for each step is at present uncertain, but the present system should allow for its careful molecular dissection in the future.

Several previous studies had suggested that osteoblasts contribute to both HSC proliferation and myeloid differentiation.^1–3,10  The present results also confirm and extend the studies of Visnjic et al, who showed that osteoblasts are required for HSCs in vivo, as their postnatal depletion leads to widespread hematopoietic failure, including the loss of HSCs, myeloid progenitors and precursors, and B lymphoid precursors over one month.⁶  Although the most parsimonious explanation for these findings was that HSC loss led to multilineage hematopoietic failure, the present studies demonstrate that the loss of B-cell precursors following osteoblast depletion actually precedes any detectable loss of HSCs, and is detectable as soon as maximal osteoblast depletion can be documented, at day 8. In addition, throughout the lifetime of the mice after osteoblast depletion, the degree of B-precursor depletion exceeds the reduction in HSC numbers. These results strongly suggest that osteoblasts are essential for some aspects of B lymphopoiesis, over and above any role they may play in HSC maintenance and/or proliferation. Of course, our data also demonstrate that OBs support the production of myeloid precursors in vitro, and acute depletion of OBs in vivo results in rapid declines in the numbers of BM granulocytic and monocytic cells. Thus, OBs clearly have multiple roles in BM hematopoiesis, of which supporting the differentiation of LSKs to B-lineage commitment and subsequent support of B-precursor differentiation is only one.

Of course, the present results do not exclude the contribution of other BM stromal cell types to B lymphopoiesis, either adjacent to or separated from the endosteal region,²³  in a sense analogous to Kiel et al's recent report that BM and spleen sinusoidal endothelium might serve as additional HSC niches when HSCs are mobilized.²⁴  In fact, we believe that multiple stromal cell types are likely to be critical in the HSC niche per se, since several cytokines implicated in HSC proliferation (eg, c-Kit ligand) are secreted by whole BM stromal cultures, but not in pure human osteoblast cultures.^2,10  Thus HSC maintenance, which is clearly required for the continued production of B-cell precursors in the bone marrow, may indeed require a complex multicellular niche. However, for the developmental steps leading from the HSC to mature B cells, osteoblasts appear to play the major necessary and sufficient roles.

One key implication of these data is that bone formation is directly linked to the integrity of the humoral immune system. This finding therefore raises the possibility that deficiencies in osteoblast number or function might underlie deficiencies in B-cell production. Such a decline could, for example, contribute to the decline in humoral immunity seen in aging. At a more speculative level, since osteoblast depletion prevents the earliest detectable transitions from the HSC compartment, it is likewise formally possible that declines in osteoblast function might lead to declines in the number of available T-cell progenitors available for emigration to the thymus. Overall, our data suggest that the further characterization of the diverse effects of osteoblasts on HSC fate decision, survival, and differentiation will greatly improve our understanding of local topological mechanisms regulating stem cell lineage commitment and differentiation.

Contribution: J.Z., R.G., Y.C., E.H., R.T., and S.G.E. designed experiments; J.Z., Y.J., N.K., J.W., and G.J.J. performed experiments; J.Z., R.G., Y.Z., E.H., R.T., and S.G.E. interpreted results; J.Z., R.G., R.T., and S.G.E. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Stephen G. Emerson, Division of Hematology/Oncology, University of Pennsylvania School of Medicine, Maloney 510, 3600 Spruce St, Philadelphia, PA 19104; e-mail: emersons@mail.med.upenn.edu.