Adipocyte-derived soluble factor(s) inhibits early stages of B lymphopoiesis

Abstract

B lymphopoiesis declines with age, and in rabbit this occurs by 8 weeks of age. We found that colony-forming units fibroblasts (CFU-F) in the bone marrow (BM) decrease 10-fold by a few weeks of age, and that the CFU-Fs preferentially differentiate into adipocytes instead of osteoblasts. BM becomes filled with fat spaces during this time, making rabbit a unique model to study the effects of accelerated fat accumulation on B lymphopoiesis. We show that adipocytes of both rabbit and human secrete a soluble factor(s) that inhibits B lymphopoiesis and we tested if this inhibition was due to effects on the BM stroma or hematopoietic progenitors. Pre-treatment of BM mononuclear cells (MNCs) with adipocyte conditioned medium (CM) dramatically inhibited their differentiation into proB cells in co-cultures with OP9 stromal cells. In contrast, pre-treatment of OP9 stromal cells with adipocyte CM had no effect on B lymphopoiesis. Using human HSCs we show that inhibition by the adipocyte-derived factor occurred at the common lymphoid progenitor (CLP) to pre-proB cell stage. We propose that the age-related decline in B lymphopoiesis is due to a decrease in CFU-F, an increase in adipocytes, and an adipocyte-derived factor that blocks B lymphopoiesis at the CLP to pre-proB cell stage.

INTRODUCTION

BM is the site of B lymphopoiesis in which hematopoietic stem cells (HSCs), in the presence of BM stroma, give rise to B cell progenitors. B lymphopoiesis wanes in humans (1–3) and mice in mid (4) and late stages of life (5, 6). In rabbits, however, B lymphopoiesis declines by 6–8 weeks of age, a time at which only a few proB or preB cells are found in BM (7, 8). The decline in lymphopoiesis appears limited to the cells of B lineage as shown by Kalis et al (9) who found T cell progenitors in thymus at 15 and 34 weeks of age, indicating that T lymphopoiesis was unaffected at an early age. The generation of immature B cells in the BM is highly dependent on the interaction of B cell progenitors with BM stromal cells (10). Hence, the decline in B lymphopoiesis could be due to changes in the B cell progenitors and / or to changes in the BM microenvironment. We previously tested these possibilities in both in vitro and in vivo experiments, by co-culturing total BM of rabbits over 5 months of age on OP9 stromal cells and also by injecting them into young rabbits (9). The lymphoid progenitors in BM from older rabbits were capable of generating B cells if provided with an appropriate microenvironment, suggesting that the decline in B lymphopoiesis was not due to defects in early B lineage progenitors, but instead, to changes in the BM stroma. In contrast to rabbits, decline in B lymphopoiesis in mice has been attributed to defects in both the BM stroma and HSCs (4–6), rendering rabbit a useful model to study the effects of changes in BM stroma on B cell development.

BM stroma is comprised of a network of heterogeneous stromal cell types, including adipocytes, osteoblasts, endothelial cells and fibroblasts. Adipocytes and osteoblasts, both of which arise from CFU-F, play an important role in the regulation of B lymphopoiesis (11, 12). Osteoblasts have been shown to support B lymphopoiesis while there is evidence that adipocytes inhibit hematopoiesis, including B cell development (12,13). A decrease in the number of CFU-Fs and/or an increased tendency of these cells to form adipocytes could contribute to the decline of B lymphopoiesis. Changes in the BM microenvironment with respect to the number of CFU-Fs and their altered capacity to differentiate into adipocytes and osteoblasts have been reported in mice and humans (14–17). However, prior to this study a defined effect of age-related increase in BM fat on B lymphopoiesis has not been investigated. We have used both rabbit and human adipocytes and hematopoietic progenitor cells to understand the mechanism(s) by which adipocytes negatively affect B lymphopoiesis making our findings applicable to immune responses in aged humans. We tested the effect of adipocytes on B-lymphopoiesis in vitro using adipocyte conditioned medium (CM), and we show that adipocytes secrete a soluble factor that inhibits B-lymphopoiesis from the CLP to pre-proB cell stage.

MATERIALS AND METHODS

Animals

Rabbits used were maintained in a colony at Loyola University Chicago and used as per protocols approved by Institutional Animal Care and Use Committee of Loyola University Chicago.

Microscopy

All microscopy was performed on Leica DM IRB microscope (Leica Microsystems, IL) and images were taken using magnafire 2.1C digital camera system.

Human blood and bone samples

Human cord blood (CB) and bone samples were obtained from Loyola University Medical Center and approved by the institutional review board. Human spongy bone samples were taken during knee and hip replacement surgery. Samples were collected in accordance with the principles of Helsinki Declaration. Single cell suspensions of BM were obtained and RBCs were lysed with 0.85% NH₄Cl.

Flow cytometry

Antibodies used were mouse monoclonal (MAb) with specificity for human target antigens (Table 1). Flow cytometric analysis was performed on BD Canto II flow cytometer (BD Bioscience) and FlowJo software (Ashland, OR). All cells analyzed were in the lymphocyte gate; quadrants for positive and negative population were based on cells stained with isotype control Ab.

Table 1.

MAb used in this study.

Antibody	Clone	Fluorochrome	Source
Anti – CD141	TUK4	Pacific Blue	Serotec
Anti - CD79a1	HM47	PE	BD Pharmagin
Anti- CD34	563	PE	BD Pharmagin
Anti- CD38	HIT2	APC	BD Pharmagin
Anti - CD3	OKT3	FITC	eBioscience
Anti - CD14	61D3	FITC	eBioscience
Anti - CD15	H198	FITC	eBioscience
Anti - CD19	H1B19	FITC	eBioscience
Anti - CD19	SJ25-C1	APC-Cy7	Invitrogen
Anti - CD56	MEM188	FITC	eBioscience
Anti - CD7	EBio124-D1	PE-Cy5	eBioscience
Anti-CD10	HI 10a	PE-Cy7	BioLegend
Anti - IgM	G20-127	APC	BD Pharmagin

¹Cross-reactive with rabbit

CFU-F assay

Heads of rabbit bones were removed and BM was flushed from the shaft of femurs. Single cell suspensions were passed through 100 μm cell strainer, and RBCs were lysed with 0.85% NH₄Cl. BM MNCs (50 to 40,000 per well from 1 to 6 wk-old rabbits; 1000 to 500,000 per well from 7 wk to 11 month-old rabbits) were plated in 96-well plates for 48–72 hours. After removal of unattached cells, alpha MEM medium (Invitrogen, Grand Island, NY) with 20% fetal calf serum (FCS) (Atlanta Biologicals, Lawrenceville, GA) was added and the medium was changed every 2–3 days. The number of wells with or without fibroblast colonies was counted at day 21. CFU-F was defined as a colony of 50 or more cells with fibroblast-like morphology (18, 19). The frequency of CFU-Fs was determined (20) on the basis of Poisson distribution as a reciprocal of total BM MNCs that resulted in 37% of wells with no fibroblast colonies.

Differentiation of CFU-Fs into adipocytes and osteoblasts

Rabbit CFU-Fs obtained as described above, were cultured in osteoblast differentiation-inducing (ODI) medium containing 10nM dexamethasone, 20mM β-glycerol phosphate and 50μM ascorbic acid, or in adipocyte differentiation-inducing (ADI) medium containing 0.5μM isobutylmethylxanthine, 50μM indomethacin and 0.5 μM dexamethasone (21). All chemicals were obtained from Sigma Aldrich (St.Louis, MO). Colonies grown in ODI medium were stained for alkaline phosphatase and with alizarin red S while colonies grown in ADI medium were stained with oil red O.

Co-culture of OP9 stromal cells and total BM MNCs with human and rabbit adipocyte conditioned medium (CM)

Human and rabbit BM MNCs (10⁵/cm²) were plated in 12-well plates and adipocytes were generated as described above. The adipocytes (as indicated by their morphology upon phase contrast microscopy) were cultured in alpha MEM medium without FCS, and CM was collected after 72 hours. Adipocytes showed healthy morphology (as evidenced by phase contrast microscopy) in culture without FCS for 72 hours (data not shown). Human or rabbit adipocyte CM medium (or alpha MEM without FCS as a negative control ) was added (1:2) to co-cultures of OP9 stromal cells (5000 cells/well) and rabbit BM MNCs (40,000 cells/well) with alpha MEM FCS (20%) containing a final concentration of 10ng/ml each of recombinant human IL-7, stem cell factor (SCF), and Flt-3 ligand (Flt-3L) (Peprotech, Rocky Hill, NJ) in 48-well plates. OP9 cells were plated 24 hrs prior to the addition of total BM MNCs. Co-cultures were fed with fresh medium containing cytokines and adipocyte CM (or alpha MEM without FCS for the negative control) at day 4. Cells were harvested on day 7 or 8, and live cell counts from three wells were obtained using trypan blue exclusion dye. Cells were stained with anti-CD14, permeabilized with cytoperm / cytofix (BD Biosciences), stained with anti-CD79a (Igα), and analyzed by flow cytometry. The dose dependency of CM was tested by adding decreasing amounts of CM to the cultures (1:2 to 1:32). Alpha MEM without FCS was diluted 1:2 in alpha MEM with 20% FCS and used as a negative control for these experiments . For some experiments, rabbit and human adipocyte CM were centrifuged (2000 rpm) through a filter device with10kDa cut off membrane (Centricon, Millipore, Billerica, MA) at 4°C. The less than 10kDa fraction of human CM was tested for the presence of adiponectin by ELISA (Kit #ab99968; Abcam, Cambridge, MA) and no detectable adiponectin was found (< 25pg/ml).

Collection of conditioned medium (CM) from human and rabbit undifferentiated BM CFU-Fs

Rabbit and human BM stroma were generated as described in Materials and Methods. After generating confluent layers of undifferentiated fibroblast like cells, (as evidenced by phase contrast microscopy) alpha MEM without FCS was added to the cultures and CM was collected at 72 hr. Undifferentiated BM CFU-Fs showed healthy morphology upon microscopy in culture without FCS for 72 hr (data not shown).

Collection of conditioned medium (CM) from human and rabbit osteoblasts

Rabbit and human osteoblasts were generated using osteoblast differentiation-inducing (ODI) medium as described in Materials and Methods. Alpha MEM without FCS was added to the cultures and CM was collected at 72 hr. By microscopy, osteoblasts appeared healthy in cultures without FCS for 72 hr (data not shown). The presence of osteoblasts was confirmed by staining cells for alkaline phosphatase (data not shown).

Transwell Assay

Adipocytes were generated in 12-well plates (as described above) in the bottom of transwells. Transwells of 1.0 μm pore size (Millipore, Billerica, MA) were plated with 5 × 10³ OP9 stromal cells 24 hr prior to addition of 20 × 10³ total BM MNCs /well. Co-cultures containing alpha MEM (20% FCS) and IL-7, SCF, and Flt-3L (10 ng/ml each) were fed with fresh medium at day 4; cells were harvested at day 10. Live cell counts and analysis by flow cytometry were performed as described above.

Pre-treatment of OP9 stromal cells and rabbit BM MNCs with adipocyte CM

OP9 stromal cells (10⁵ ) or rabbit BM MNCs (4×10⁶) were treated 24 hr with alpha MEM (20%FCS) and adipocyte CM (1:2) (or with alpha MEM with no FCS as negative control) for 5 days. OP9 stromal cells were then detached using 0.25% trypsin (Sigma, St.Louis, MO). Treated OP9 stromal cells (detached with 0.25% trypsin) and BM MNCs were washed with alpha MEM (20% FCS) and used in co-culture experiments.

Co-culture of human HSCs with rabbit adipocyte CM

CD34⁺ cells were enriched from human CB using EasySep CD34⁺ positive selection cocktail (Stem cell technologies, Vancouver, Canada). CD34⁺ CD38⁻ Lineage⁻ (Lineage markers: CD3, CD14, CD15, CD19, CD56) HSCs (5000 cells/well) were then sorted directly onto OP9 stromal cells (5000 cells/well) using the FACSARIA flow cytometer (BD Biosciences) in 48-well plates containing alpha MEM 20% FCS and adipocyte CM (1:2) (or alpha MEM with no FCS as negative control) with IL-7, SCF and Flt-3L (10 ng/ml each). Co-cultures were fed with fresh medium and adipocyte CM (1:2) (or alpha MEM no FCS as a negative control) at days 4 and 10; cells were harvested at days 10 and 14. Live cell counts from three wells were obtained using trypan blue exclusion dye. Analysis for the presence of CLP and pre-proB was performed by surface anti-CD7, anti-CD10, anti-CD34 and anti-CD19, and cytoplasmic anti-CD79a staining. ProB cells were identified by surface anti-CD7 and anti-CD19, and cytoplasmic anti-CD79a and anti-μ staining. All samples were analyzed by flow cytometry. In our co-culture system both CLPs and pre-proB cells were CD34⁺ but CD10⁻. In agreement with Ichii et al (22), committed B-lineage cells expressed CD10 in addition to CD19 and CD79a.

Statistical Analysis

Mann-Whitney test and student’s t test were performed using Prism Statistical Software. P value of < 0.05 was considered statistically significant.

RESULTS

Changes in the frequency and differentiation potential of CFU-Fs

We hypothesized that the decline in B lymphopoiesis in rabbits was due to changes in the BM environment. We tested if there is a decrease in the number of CFU-Fs, which differentiate into osteoblasts and adipocytes, and/or to a change in the capacity of CFU-Fs to differentiate into osteoblasts with age. We quantified CFU-Fs (Fig. S1A) (23) by limiting dilution analysis of BM MNCs from rabbits of various ages. We used the CFU-F assay because the resulting colonies derived from BM were shown previously to be multipotent and clonal in origin (19). By phase contrast microscopy, the CFU-Fs had fibroblast-like morphology, and no macrophages were seen (Fig. S1A), as reported previously (19). The frequency of CFU-Fs in 1-to 4-week-old rabbits was generally greater than 100 in 10⁶ BM MNCs and declined dramatically by 9 weeks of age to less than 10 in 10⁶ cells (Fig. 1A–B). The low number of CFU-Fs was maintained for at least a year, as evidenced by the frequency of only 9 CFU-Fs per 10⁶ total BM MNCs in an 11 month-old rabbit, the latest age we tested (Fig. 1B). We conclude that the number of CFU-Fs decreased nearly 10-fold by 9 weeks of age.

FIGURE 1. Frequency and differentiation potential of CFU-Fs in rabbits of various ages.

(A) Representative example of limiting dilution analysis of CFU-F from 3 wk-old and 6 wk-old rabbits. R² = square of multiple correlation coefficient. (B) Frequency of CFU-Fs in 10⁶ BM MNCs from 16 rabbits 1 to 44 wks-of-age. (C) Ratio of osteoblastic to adipocytic colonies in 14 day cultures from 2–8 wk old (●) and 9-wks to 11 month-old (■) rabbits (data from Table 2). The average number of alkaline phosphatase positive osteoblastic and oil red O positive adipocytic colonies from three to six wells was used to calculate the ratio of osteoblastic and adipocytic colonies. Data were analyzed by Mann-Whitney test. Horizontal line represents the average of the ratio of osteoblastic and adipocytic colonies in both age groups.

CFU-Fs give rise to mature BM stromal cells, osteoblasts, adipocytes, and chondrocytes (24). Osteoblasts reportedly support all stages of B lymphopoiesis in BM (12) while adipocytes negatively regulate hematopoiesis (13). We hypothesized that the decline in B lymphopoiesis may reflect not only a decrease in the number of CFU-Fs, but also increased capacity of the CFU-Fs to differentiate into adipocytes instead of osteoblasts. We tested this possibility by culturing CFU-Fs from rabbits 2 weeks- to 11 months-of-age in osteoblast- or adipocyte-differentiation-inducing medium referred to as ODI and ADI medium, respectively. Colonies grown in ODI medium were stained for the presence of osteoblastic colonies using alkaline phosphatase, a marker of osteogenic differentiation of CFU-Fs (Fig. S1B – S1C); and for the presence of minerals deposited by osteoblasts using alizarin red S stain (Fig. S1D – S1E). Colonies grown in ADI medium were stained for the presence of adipocytic colonies by oil red O which stains fat globules (Fig. S1F – S1G). Since these CFU-Fs can differentiate into osteoblasts and adipocytes we assume that they are mesenchymal stem cells. We calculated the ratio of osteoblastic to adipocytic colonies and found that CFU-Fs from 2- to 8- wk-old rabbits readily differentiated into osteoblasts as evidenced by an average ratio of osteoblastic to adipocytic colonies of 1.3 (Table 2; Fig. 1C). In contrast, for rabbits 9 weeks of age and older, the average ratio was 0.4, indicating that the CFU-Fs from older rabbits preferentially differentiated into adipocytes (Table 2; Fig. 1C). We conclude that in vitro, CFU-Fs from rabbits over 2 months of age preferentially differentiate into adipocytes rather than into osteoblasts. Although we cannot rule out the possibility that the apparent increase in differentiation to adipocytes is due to preferential survival and/or expansion of other cell populations, the finding correlates with the increase in fat spaces at 2 months of age, as observed in H&E stained BM sections (Fig. S2) and as described previously (25).

Table 2.

Number and ratio of osteoblastic and adipocytic colonies/10⁶ BM MNCs generated from CFU-Fs in 14 rabbits of various ages.

AGE	OSTEOBLASTIC COLONIES ± SD	ADIPOCYTIC COLONIES ± SD	RATIO
2 week	38.3 ± 4.1	33.3 ± 4.1	1.15
3 week	32.3 ± 5.3	20.7 ± 5	1.56
4 week	22.3 ± 7	10 ± 2	2.23
5 week	31.3 ± 2.5	38.7 ± 5.7	0.81
6 week	5 ± 2.3	1.7 ± 1.7	3.01
6 week	3 ± 1	16 ± 1.8	0.18
7 week	19.2 ± 6.3	24 ± 7	0.79
7 week	22 ± 17	11.6 ± 7.2	1.89
8 week	8.2 ± 3.8	10.3 ± 2.1	0.82
9 week	3.6 ± 2.6	20.7 ± 7	0.17
4 month	4.7 ± 4.6	3 ± 2.6	1.55
5 month	6.8 ± 4.2	8.5 ± 7.5	0.8
6 month	3.7 ± 3.2	30.8 ± 7.1	0.11
6 month	0.3±0.5	2.7±3	0.11
6 month	0.3±0.5	3.7±1.5	0.08
7 month	0.5±0.8	3±2.2	0.16
11 month	0.3 ± 0.5	3 ± 3	0.11

Adipocyte inhibition of B lymphopoiesis

The increase in fat spaces in rabbits over 8 weeks of age, as seen in the H&E stain of BM (Fig S2), led us to ask if adipocytes inhibit the generation of proB cells, thereby contributing to the decline in B lymphopoiesis. We tested this possibility by adding primary rabbit adipocyte CM to co-cultures of OP9 stromal cells and BM MNCs from rabbits over 8 weeks of age (in which BM is devoid of CD79a⁺ [Igα] progenitor B-lineage cells (8) and measuring the number of CD79a⁺ B-lineage cells generated. We found that in the presence of CM, almost no CD79a⁺ cells were generated (Fig. 2A–C); inhibiton of the generation of CD79a⁺ B-lineage cells by CM was dose dependent (Fig 2E–F). These data indicate that adipocytes secrete a factor(s) that inhibits B lymphopoiesis. This finding was confirmed using a transwell system in which rabbit BM MNCs were co-cultured with OP9 stromal cells above the transwell with primary rabbit adipocytes below the transwell. We found a striking decrease in the capacity of OP9 stromal cells to promote the generation of CD79a⁺ B-lineage cells in the presence of adipocytes (Fig. 2G–H). In contrast, the adipocyte CM did not decrease generation of CD14⁺ myeloid-lineage cells, and in fact, appeared to increase myelopoiesis (Fig. 2B, D and G). We also generated adipocytes from human CFU-Fs (21) and found that like rabbit, human adipocyte CM impaired generation of CD79a⁺ B-lineage cells from rabbit BM MNCs (Fig. 3A and B). We conclude that adipocytes secrete soluble factor(s) that selectively inhibits B lymphopoiesis. By filtering the CM through a 10kDa membrane, we found that inhibitory activity resided in the <10 kDa fraction (Fig. 4A and B), indicating that the inhibitor is small in size. Similar results were obtained with the <10kDa fraction of human adipocyte CM (data not shown).

FIGURE 2. Effect of adiopocytes on the generation of B-lineage cells in co-cultures of OP9 stromal cells and rabbit BM MNCs.

Representative example of lymphocyte gate (A); Flow cytometric analysis of lymphoid cells for the presence of cytoplasmic CD79a⁺ B- and CD14⁺ myeloid-lineage cells after 7 day co-cultures of total BM MNCs with OP9 stromal cells with or without (w/o) adipocyte CM (B); Number of CD79a⁺ B-lineage (C) and CD14⁺ myeloid-lineage cells (D) with (+) and without (−) adipocyte CM; Dose dependence of inhibitory activity of adipocyte CM as measured by addition of adipocyte CM (ACM) diluted 1:2 – 1:32 (E); Number of CD79a⁺ B-lineage cells generated upon addition of serial-diluted adipocyte CM (F); Flow cytometric analysis for CD79a⁺ B- and CD14⁺ myeloid-lineage cells after 10 day co-culture of BM MNCs with OP9 stromal cells above the transwell with or without (w/o) adipocytes below the transwell (G); Number of CD79a⁺ B-lineage cells generated with (+) or without (−) adipocytes below the transwell (H); Data in B, C and D are representative of more than 3 independent experiments. Data in E and F are representative of 3 independent samples of adipocyte CM; data in F and G are representative of 3 independent experiments. Error bars show average ± SD of triplicate wells and data in D were analyzed using student’s t test.

FIGURE 3. Effect of human adipocyte CM on generation of B-lineage cells.

(A) Rabbit BM MNCs were co-cultured with human adipocyte CM on OP9 stromal cells for 7 days and then tested for CD79a⁺ B-lineage cells by flow cytometry. (B) Number of CD79a⁺ B-lineage cells generated in the absence (−) and presence (+) of human adipocyte CM. Error bars show average ± SD of triplicate wells. Data are representative of three independent experiments.

FIGURE 4. Effect of <10kDa fraction of adipocyte CM on generation of B-lineage cells.

(A) OP9 stromal cells and BM MNCs were co-cultured for 7 days in the presence of <10kDa fraction of adipocyte CM. Cells were analyzed by flow cytometry for the presence of CD79a⁺ B-lineage and CD14⁺ myeloid-lineage cells in the lymphocyte gate (as shown in Fig 2A). Co-cultures without (w/o) (negative control) or with adipocyte CM (positive control) are included. (B) Number of CD79a⁺ B-lineage cells. Error bars show average ± SD of triplicate wells. Data are representative of three independent experiments.

As negative controls, we tested CM from rabbit BM undifferentiated CFU-F and osteoblasts, as well as CM from human BM undifferentiated CFU-Fs and osteoblasts. Although we found some inhibitory activity in the generation of CD79a⁺ B-lineage cells (Fig. S3A–C), the adipocyte-enriched cultures had greater inhibitory activity.

The effect of adipocytes on stromal cells and hematopoietic progenitors

With age, mouse BM stroma loses the capacity to support development of early B cell precursors (26) and we asked if the adipocyte-derived soluble factors could inhibit B lymphopoiesis by altering the lymphopoietic capacity of BM stroma, or if it could directly alter the capacity of hematopoietic stem cells (HSCs) and/or early lymphoid progenitors to differentiate into B lineage cells. We first tested these possibilities by treating the OP9 stromal cells, which are known to support B lymphopoiesis (27), with rabbit adipocyte CM and then co-culturing these pre-treated cells with rabbit BM MNCs in the absence of adipocyte CM. Such treatment did not alter the capacity of OP9 stromal cells to support the generation of CD79a⁺ B-lineage cells (Fig. 5A and B) (Fig. 5A). These data suggest that short-term exposure of the stromal cells to the adipocyte-derived factor does not alter their B lymphopoietic capacity, although we cannot rule out the possibility that continuous presence of inhibitory factor(s) with the stromal cells may have an effect on them.

FIGURE 5. Effect of pre-treatment of OP9 stromal cells with adipocyte CM on the capacity to support B lymphopoiesis.

(A) OP9 stromal cells treated with adipocyte CM for 5 days, washed and co-cultured with BM MNCs in the absence of adipocyte CM for 8 days. Cells were analyzed by flow cytometry for CD79a⁺ B-lineage and CD14⁺ myeloid-lineage cells. Untreated OP9 stromal cells were co-cultured with BM MNCs without (negative control) or with adipocyte CM (positive control). (B) Number of CD79a⁺ B-lineage cells. Error bars show average ± SD of triplicate wells and data were analyzed using student’s t test. Data are representative of 2 independent experiments.

In contrast, treatment of BM MNCs from rabbits over two months of age, which have HSCs and early B cell progenitors such as common lymphoid progenitors (CLPs) in their BM, but no CD79a⁺ proB or pre-B cells (8), with adipocyte CM for 24 hr followed by culturing them with OP9 stromal cells in the absence of adipocyte CM, we found greatly impaired capacity to differentiate into CD79a⁺ B-lineage cells (Fig. 6A and B) even after the removal of adipocyte CM (Fig. 6A). We conclude that the adipocyte-derived inhibitory molecule does not appear to change the capacity of BM stromal cells to support B lymphopoiesis, but instead to directly alter the capacity of HSCs and/or early lymphoid progenitors such as CLPs to differentiate into B lineage cells. Although we cannot rule out the possibility that the inhibitory molecule acts by interfering with access to growth factors, especially IL-7, we do not think this is likely because IL-7, SCF and Flt3L were added to the cultures after removal of CM.

FIGURE 6. Effect of pre-treatment of BM MNCs with adipocyte CM on differentiation into B-lineage cells.

(A) BM MNCs treated with adipocyte CM for 24 hrs were washed and co-cultured with OP9 stromal cells and then analyzed at day 7 by flow cytometry for CD79a⁺ B-lymphoid and CD14⁺ myeloid-lineage cells. Untreated BM MNCs were co-cultured with OP9 stromal cells in the absence (negative control) or presence of adipocyte CM (positive control). (B) Number of CD79a⁺ B-lineage cells. Error bars show average ± SD of triplicate wells and data were analyzed using student’s t test. Data are representative of 3 independent experiments.

Adipokine inhibition of differentiation of CLP to pre-proB cells

To investigate the stage of differentiation at which the adipocyte CM inhibits B-lineage development, we used HSCs from humans because reagents are readily available to identify HSCs and CLPs. We first showed that the rabbit adipocyte CM inhibited differentiation of human CD34⁺ HSC-enriched cells into CD79⁺ B-lineage cells in co-cultures with OP9 cells (data not shown). We then co-cultured human HSCs and OP9 cells with rabbit adipocyte CM and analyzed the cells for CLPs, pre-proB and proB cells (28–30). At day 10, we found that the percentages and absolute numbers of CLPs (CD7⁺ CD34⁺) were similar in cultures with or without adipocyte CM (Fig. 7A and B). However, we found a significant decrease in the percentage and absolute numbers of pre-proB cells (CD79a⁺ CD34⁺) in the co-cultures that contained the inhibitory factor (Fig. 7A and C). In addition, when we analyzed CD7⁺ cells, of which ~70% are CD34⁺, we also found the percentage and absolute numbers of CD7⁺CD79a⁻early lymphoid progenitors (CLPs) were similar in cultures with or without CM (23.9% vs. 22.7%; 4,834 vs. 5,417 cells) (Fig S4). Similarly, when we analyzed CD79a⁺ cells (~60% of which are CD34+), we found a significant decrease in the percentage and absolute numbers of early B-lineage cells (pre-proB) in cultures that contained CM (13.6% vs. 5.6%; 3,194 vs. 1,143 cells) (Fig S4). By day 14, almost no proB cells were found in the presence of the inhibitory factor (Fig. 7D and E). We determined the total number of live cells at days 7, 10 and 14 using trypan blue exclusion dye and did not find a significant difference at any time point in the number of live cells in the presence or absence of adipocyte CM. In addition we searched for pre-proB cells at day 7, and similar to days 10 and 14, we found almost no pre-proB cells in the presence of adipocyte CM (data not shown). These data demonstrate that rabbit adipocytes do not affect generation of CLPs from HSCs but do inhibit differentiation of these cells to pre-proB cells.

FIGURE 7. Effect of rabbit adipocyte CM on the early stages of B cell differentiation from human HSCs.

(A) Human HSCs were co-cultured with or without (w/o) rabbit adipocyte CM on OP9 stromal cells for 10 days and analyzed by flow cytometry for the presence of CLPs (CD7⁺ CD34⁺ CD10⁻ CD79a⁻ ) and pre-proB cells (CD7⁻ CD34⁺ CD10⁻ CD79a⁺ ) in CD19⁻ cells; B) Number of early lymphoid progenitors /CLPs per well; (C) Number of pre-proB cells per well; (D) Analysis for proB cells (CD7⁻CD79a⁺CD19⁺ cyto-μ⁻) at day 14 by flow cytometry; (E) Number of proB cells. Error bars show average ± SD of triplicate wells. Data in B and C were analyzed using student’s t test. Data in A-C were obtained using 3 independent samples of adipocyte-derived CM (I-III); similar data for CLPs and pre-proB cells were obtained in three additional experiments in which cells were not stained with anti-CD10 and anti-CD34 (FigS4). Data in D and E are representative of 3 independent experiments.

DISCUSSION

B lymphopoiesis declines with age in various species including humans, rabbits and mice. In mice, this decline is attributed to defects in both HSCs and the BM microenvironment (4–6), while in rabbits the defect appears to be with BM stroma (9). The BM stroma is composed of various cell types including osteoblasts and adipocytes, which are derived from CFU-Fs. Osteoblasts support early stages of B cell development while adipocytes have been shown to negatively regulate hematopoiesis (12, 13). Whereas the frequency of CFU-Fs does not change appreciably in aged mice or humans (14,15), we found a 10-fold decline in CFU-Fs in rabbit BM within a few weeks of age. In addition, these CFU-Fs, like those in aged humans and mice, appear more likely to differentiate into adipocytes than into osteoblasts (15–17). This observation is based on using CFU-Fs that were enriched by their adherence property, and we assume that this was similar in all age groups of rabbits. The rapid decline in the frequency of CFU-Fs in rabbit BM, along with the decreased potential to differentiate into osteoblasts coincides with the time at which B lymphopoiesis in rabbit BM wanes (Fig. 8A). We suggest that these changes result in fewer osteoblasts and that the increase in fat leads to secretion of large amounts of inhibitor and a concomitant decline in B lymphopoiesis (Fig. 8B).

FIGURE 8. Model for the mechanism(s) by which B lymphopoiesis declines in rabbit BM.

(A) Decline in the number of pro- and pre-B cells in rabbit has a remarkable correlation with a decrease in the number of CFU-Fs and an increase in fat in the first eight weeks of life. (B) Decline in the numbers of CFU-Fs and their capacity to differentiate into osteoblasts rather than adipocytes, leads to an increase in fat spaces in adult rabbit BM. The adipocytes secrete a soluble factor that inhibits B lymphopoiesis at the CLP to pre-proB cell stage. Consequently, no proB or preB cells are found in adult rabbit BM.

Age-related increases in fat are also found in the thymus, and are associated with reduced numbers of thymic progenitors and thymic epithelial cells(31,32). In 1974, Tavassoli et al (33) found that by removing adipocytes from rabbit BM, hematopoiesis was increased, and more recently, Naveiras et al (13) removed adipocytes from mouse BM and showed that hematopoiesis, including B lymphopoiesis, was increased. Here we showed that rabbit BM-derived adipocytes secrete a soluble factor(s) that inhibits the generation of B cells at the CLP to pre-proB cell stage. Since human BM shows an increase in fat with age (34), we suggest that adipocytes negatively affect early stages of B lymphopoiesis in BM of the elderly. Adipocytes secrete numerous molecules, including leptin, resistin, and adiponectin, all of which play a role in immune responses (35). Of these, adiponectin was shown to inhibit B lymphopoiesis(11, 36); however, using a <10kDa adiponectin-free fraction from CM, we showed that the inhibitory activity in our study is independent of adiponectin.

Adipocytes could inhibit B lymphopoiesis by altering the BM stroma and/or by direct action on hematopoietic progenitors. Treatment of the stroma with the adipocyte inhibitor did not alter the capacity of stroma to support differentiation of HSC to B-lineage cells. However, we could only selectively treat the stroma with CM prior to the addition of hematopoietic progenitors, and we cannot rule out the possibility that CM affects the stroma, but that the changes are transient and not observed in our cultures. In contrast, 24 hr pre-treatment of the inhibitor with hematopoietic progenitors profoundly affected differentiation of the progenitors into B-lineage cells, indicating that the inhibitor induced a stable change in the progenitors, making them unable to differentiate into B-lineage cells even in the presence of BM stroma and B lymphopoietic cytokines.

While adipocytes were shown to alter the proliferation and differentiation capacity of HSCs (37), no effect on other hematopoietic progenitor cells including MPPs or CLPs has been reported. Here, we show that adipocytes block B-lymphopoiesis at the CLP to pre-proB cell stage. We think that the inhibitor affects the differentiation of CLPs into pre-proB cells rather than compromising the survival of pre-proB cells because in the presence of the inhibitory molecule we did not find a significant decrease in the number of live cells, nor did we find a distinct population of pre-proB cells at earlier time points in the cultures. We suggest that the block in differentiation of CLPs into pre-proB cells by adipocytes explains the lack of proB cells and the arrest of B lymphopoiesis in rabbit BM. Although a decrease in the differentiation of CLP to pre-proB cells in aged mice has been reported; however, the possibility that this is due to changes in BM stroma was not addressed (38). The transition of CLPs to pre-proB cells requires transcription factors early B cell factor-1 (EBF-1), E2A and Pax-5, and we suggest that the block in differentiation of CLPs into pre-proB cells is due to changes in expression of one or more of these transcription factors.

CLPs can differentiate into myeloid cells as well as B-lineage cells, and they can express transcription factors required for myeloid cell differentiation (39). In some experiments, we found an increase in myeloid cells in cultures containing adipocyte CM, suggesting that adipocytes may promote myelopoiesis. If confirmed, then adipocyte inhibitor(s) may up-regulate myeloid-specific transcription, or promote survival of early myeloid progenitors such as common myeloid progenitors and granulocyte-macrophage progenitors. Alternatively, adipocytes may alter the differentiation potential of HSC. Recent data suggest that subsets of human HSCs are myeloid- or lymphoid-biased (40), and we suggest that adipocytes contribute to regulating the balance between the numbers of myeloid- and lymphoid-biased HSCs with age.

In summary, we conclude that the decline in numbers of CFU-Fs and their decreased capacity to form osteoblasts, combined with an apparent increase in fat which secretes an as-yet-unidentified inhibitory molecule, likely contribute to the decline in B lymphopoiesis in rabbits (Fig 8B). We show that adipocytes inhibit the differentiation of human CLPs into pre-proB cells by direct action on hematopoietic progenitors. Our studies include both rabbit and human, showing that adipocyte CM inhibits B-lymphopoiesis using either human or rabbit HSC. We suggest that rabbit is a useful model to elucidate the mechanism by which adipocytes inhibit B lymphopoiesis. Further, we propose that B lymphopoiesis and immune responses can be enhanced by modulating the activity of fat cells.

Abbreviations

BM

Bone Marrow

CM

conditioned medium

ACM

Adipocyte conditioned medium

MNC

mononuclear cells

CLP

common lymphoid progenitors

HSC

hematopoietic stem cell

ADI

adipocyte differentiation-inducing

ODI

osteoblast differentiation-inducing

CB

cord blood

FCS

fetal calf serum

CFU-F

colony-forming unit-fibroblast

MPP

multipotent progentiros

EBF-1

early B cell factor-1

CFU

Colony Forming Unit