Receptor Activator of Nuclear Factor κB Ligand (RANKL) Protein Expression by B Lymphocytes Contributes to Ovariectomy-induced Bone Loss*

Background: The contribution of B lymphocytes to the bone loss caused by estrogen deficiency is unclear. Results: Deletion of the cytokine receptor activator of NFκB ligand from B lymphocytes, but not T lymphocytes, blunted bone loss in ovariectomized mice. Conclusion: Cytokine production by B lymphocytes contributes to ovariectomy-induced bone loss. Significance: This mechanism may be relevant to the mechanisms responsible for postmenopausal osteoporosis. Production of the cytokine receptor activator of NFκB ligand (RANKL) by lymphocytes has been proposed as a mechanism by which sex steroid deficiency causes bone loss. However, there have been no studies that functionally link RANKL expression in lymphocytes with bone loss in this condition. Herein, we examined whether RANKL expression in either B or T lymphocytes contributes to ovariectomy-induced bone loss in mice. Mice harboring a conditional RANKL allele were crossed with CD19-Cre or Lck-Cre mice to delete RANKL in B or T lymphocytes, respectively. Deletion of RANKL from either cell type had no impact on bone mass in estrogen-replete mice up to 7 months of age. However, mice lacking RANKL in B lymphocytes were partially protected from the bone loss caused by ovariectomy. This protection occurred in cancellous, but not cortical, bone and was associated with a failure to increase osteoclast numbers in the conditional knock-out mice. Deletion of RANKL from T lymphocytes had no impact on ovariectomy-induced bone loss. These results demonstrate that lymphocyte RANKL is not involved in basal bone remodeling, but B cell RANKL does contribute to the increase in osteoclasts and cancellous bone loss that occurs after loss of estrogen.

Loss of sex steroids, such as occurs after menopause or after gonadectomy, causes bone loss in both sexes (1). The bone loss is due in part to an increase in osteoclasts, which are highly specialized hematopoietic cells that resorb bone. Loss of sex steroids also increases production of bone-forming osteoblasts, but the increase in formation is insufficient to balance the increase in resorption, resulting in net bone loss (2).
Numerous mechanisms appear to contribute to the increase in osteoclast number caused by loss of sex steroids. For example, estrogens suppress production of cytokines, such as IL-6, IL-1, and TNF␣, that can stimulate proliferation of osteoclast progenitors and potentiate osteoclast differentiation and survival. Blocking the action of any one of these cytokines blunts or prevents the increase in osteoclast number caused by ovariectomy in mice (3)(4)(5)(6)(7). Estrogens can also act directly on osteoclasts to stimulate their apoptosis in vitro, and deletion of the estrogen receptor ␣ from osteoclasts or their progenitors is sufficient to increase osteoclast number in vivo and recapitulate the effects of estrogen deficiency on cancellous bone (8,9).
Another potential mechanism may involve altered expression of the TNF-family cytokine receptor activator of NFB ligand (RANKL). 2 RANKL is essential for osteoclast differentiation and also stimulates the function and survival of mature osteoclasts (10,11). Importantly, changes in the abundance of RANKL are sufficient to alter the magnitude of bone resorption. Membrane-bound and soluble forms of RANKL are produced, and both can stimulate osteoclast differentiation (10). Osteoprotegerin is a soluble decoy receptor for RANKL that inhibits its activity, and changes in osteoprotegerin expression can also alter the extent of bone resorption (12,13).
Lymphocytes have been proposed to contribute to the bone loss caused by sex steroid deficiency via a variety of mechanisms (14,15). One such mechanism involves RANKL expression by lymphocytes. Specifically, ovariectomy in mice increases the number of RANKL-expressing B lymphocytes in the bone marrow (16). Consistent with this, orchidectomy of rats increases soluble RANKL (sRANKL) levels in the bone marrow (17,18). Moreover, membrane-bound RANKL is increased on both B and T lymphocytes in the bone marrow of postmenopausal women not on estrogen replacement compared with either postmenopausal women on estrogen replacement or premenopausal women (19). An increase in the number of bone marrow * This work was supported, in whole or in part, by the Central Arkansas Veterans Healthcare System, Merit Review 5I01BX000294-04 (to C. A. O.). This work was also supported by National Institutes of Health Grants AR049794 (to C. A. O.) and AG13918 (to S. C. M.), by the University of Arkansas for Medical Sciences (UAMS) Translational Research Institute Grant UL1 RR029884, and UAMS tobacco settlement funds. 1  cells that express RANKL, within the population that includes lymphocytes, has also been observed in estrogen-deficient women (20). However, whether RANKL produced by lymphocytes contributes to the increase in bone resorption caused by sex steroid deficiency has not been determined. Herein, we examined the contribution of RANKL produced by B or T cells to the bone loss caused by estrogen deficiency in mice. We found that mice lacking RANKL in B cells were partially protected from the ovariectomy-induced loss of cancellous bone, but mice lacking RANKL in T cells lost bone similar to control animals.

EXPERIMENTAL PROCEDURES
Animal Studies-The generation of mice harboring a conditional RANKL allele (RANKL-flox) has been described previously (21). CD19-Cre knock-in mice and Lck-Cre transgenic mice, both in the C57BL/6 genetic background, as well as wildtype C57BL/6 mice, were purchased from the Jackson Laboratories (Bar Harbor, ME). Experimental animals were obtained by a three-step breeding strategy. Either heterozygous Cre knock-in mice or hemizygous Cre transgenic mice were crossed with heterozygous RANKL-flox mice to generate heterozygous RANKL-flox offspring with and without a Cre allele. These offspring were then intercrossed to generate mice homozygous for the RANKL-flox allele, with and without a Cre allele. Last, mice homozygous for the RANKL-flox allele, with or without a Cre allele, were intercrossed. Offspring were genotyped by PCR using the following primer sequences: Cre-for, 5Ј-GCG-GTCTGGCAGTAAA AACTATC-3Ј; Cre-rev, 5Ј-GTGAAA-CAGCATT GCTGTCACTT-3Ј, product size 102 bp; RANKLflox-for, 5Ј-CTGGGAGCGCAGGTTAAATA-3Ј; RANKLflox-rev, 5Ј-GCCAATAATTAAAATA CTGCAGGAAA-3Ј, product size 108 bp (wild type) and 251 bp (floxed allele). The sham or ovariectomy operations were performed at 6 -7 months of age for all of the ovariectomy studies. The animals were euthanized 2 or 6 weeks after the operations. All studies involving mice were approved by the Institutional Animal Care and Use Committees of the University of Arkansas for Medical Sciences and the Central Arkansas Veterans Healthcare System.
Flow Cytometry Analysis-Femoral bone marrow cells were stained with 2 g/ml anti-CD19-APC-Cy7 (BD Biosciences) to identify B cells, 2.5 g/ml anti-CD45R/B220-PE-Cy7 (BD Biosciences), and anti-CD43-FITC (BD Biosciences) to identify the B220 ϩ CD43 Ϫ pre-B cells, and 5 g/ml anti-CD3-FITC (BD Biosciences) to identify T cells. To detect cell surface RANKL, bone marrow cells were incubated for 30 min with 5 g/ml recombinant mouse osteoprotegerin-Fc (R&D Systems, Minneapolis, MN) or 50 g/ml human Fc-IgG (R&D Systems) as a negative control. Cells were then washed and stained for 30 min with 5 g/ml goat F(abЈ) anti-human IgG-FITC (Southern Biotech, Birmingham, AL). All samples were analyzed or sorted by flow cytometry using a FACSAria (BD Biosciences), and the data were analyzed with FACSDiva software.
Gene Expression Analysis-RNA was purified from sorted CD19 ϩ B cells and CD3 ϩ T cells using the RNeasy Mini Kit (Qiagen, Valencia, CA). RNA was purified from bone and soft tissues using Ultraspec reagent (Biotecx Laboratories, Houston, TX), according to the manufacturer's directions. cDNA was made using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) according to the manufacturer's directions. TaqMan quantitative reverse transcription-PCR (RT-PCR) was performed as described previously (22) using the following primer probe sets from Applied Biosystems: Tnfsf11-VIC (Mm00441908m1) and ribosomal protein S2 (for, 5Ј-CCCAGGATGGCGACGAT-3Ј; rev, 5Ј-CCGAATGCTGTAATGGCGTAT-3Ј; and probe, 5Ј-FAM TCCAGAGCAGGATCC-NFQ-3Ј). Relative mRNA levels were calculated using the ⌬Ct method (23).
RANKL Protein Levels in Bone Marrow Supernatants-Bone marrow supernatants were collected by removing both ends of the femur with a scalpel and then centrifuging the diaphyseal bone in a microcentrifuge tube at 850 ϫ g for 30 s. The bone marrow cell pellet and bone marrow plasma were then resuspended in 80 l of phosphate-buffered saline and then centrifuged at 2,300 ϫ g for 1 min. The supernatant was then transferred to a fresh tube and stored at Ϫ80°C until analyzed. RANKL protein levels in the bone marrow supernatants were determined using an ELISA (R&D Systems) and normalized to total protein levels determined using the BCA protein assay (Pierce Biotechnology).
Bone Mineral Density (BMD) Determinations-BMD was measured in live mice by dual-energy x-ray absorptiometry with a PIXImus Mouse Densitometer (GE Lunar Corp., Madison, WI) using the manufacturer's software as described previously (21). Using a proprietary skeletal phantom, measurements of the total body BMD performed over the past 4 years have a mean coefficient of variation of 3.1% (n ϭ 285). Growth curves were obtained by sequential BMD analysis of the same animals. The percentage change in BMD caused by ovariectomy was determined by comparing BMD measurements on the day of the operations with measurements 2-6 weeks later. CT Analysis-Soft tissue was removed from femurs or L4 vertebra, which were then fixed in 10% Millonig's Formalin for 24 h and transferred gradually from 70% to 100% ethanol. Bones were loaded into a 12.3-mm-diameter scanning tube and imaged using a CT (model CT40, Scanco Biomedical, Bruttisellen, Switzerland). Scans were integrated into three-dimensional voxel images (1,024 ϫ 1,024 pixels), and a Gaussian filter ( ϭ 0.8, support ϭ 1) was used to reduce signal noise. A threshold of 200 was applied to all scans, at medium resolution (E ϭ 55 kVp, I ϭ 145 A, integration time ϭ 200 ms). The entire vertebral body was scanned with a transverse orientation. In the distal femur, 151 transverse slices were taken from the epicondyles and extending toward the proximal end of the femur. The cortical bone and the primary spongiosa were manually excluded from the analyses. All trabecular measurements were made by drawing contours every 10 -20 slices and using voxel counting for bone volume per tissue volume and spherefilling distance transformation indices, without preassumptions about the bone shape as a rod or plate for trabecular microarchitecture. Cortical thickness was measured at the femoral mid-diaphysis. Calibration and quality control were performed weekly using five density standards, and spatial resolution was verified monthly using a tungsten wire rod. Beamhardening correction was based on the calibration records.
Corrections were made for 200 mg of hydroxyapatite for all energies. Over the past 3 years, the coefficient of variation for the fifth density standard (mean 5) was 1.28 (781 Ϯ 10 S.D. mg of HA/cm 3 ) and for rod volume was 3.16 (0.0633 Ϯ 0.002 cm 3 ).
Histology-L1-L3 vertebrae were fixed for 24 h in 10% Millonig's Formalin, decalcified in 14% EDTA for 1 week, dehydrated, and embedded in paraffin, and 5-m longitudinal sections were obtained. After removal of paraffin and rehydration, sections were stained for tartrate-resistant acid phosphatase (TRAP) activity and counterstained with fast green FCF. To determine osteoclast surface and number, quantitative histomorphometry was performed on the TRAP-stained vertebral sections using a computer and digitizer tablet (OsteoMetrics, Decatur, GA) interfaced to an Axio Scope (Carl Zeiss, Thornwood, NY) attached with a drawing tube. The number and the surface of TRAP-positive cells on the cancellous perimeter (osteoclast number and surface) were measured directly. Terminology recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research was used in this study (24).
Statistics-Data were analyzed using SigmaStat (SPSS Science, Chicago, IL) and SAS 9.3 (SAS Institute Inc., Cary, NC). All data, with the exception of those presented in Fig. 2A, passed Levene's test for constant variance and the Shapiro-Wilk test for normality, either as the data were, or after log or reciprocal transformation. The data presented in Fig. 2A were bimodal so two-sample tests for ovariectomy versus sham operation within each genotype were performed. For the RANKL fl/fl group, data were not normal so the nonparametric Wilcoxon Rank Sum test was used. For the conditional knock-out mice, the data were normal but did not have equal variances so the unequal variance t test was used. In all other cases, the differences between group means were evaluated using a two-way ANOVA or Student's t test. For comparisons using ANOVA, post hoc analysis was performed using the Holm-Sidak method. All values are reported as the mean Ϯ S.D.

Deletion of RANKL from B Cells Does Not Alter Bone Mass-
To determine whether RANKL produced by B lymphocytes contributes to osteoclastogenesis, we generated mice lacking RANKL specifically in B cells. To do this, mice harboring a RANKL conditional allele, hereafter referred to as RANKL fl/fl mice, were crossed with mice expressing the Cre recombinase under the control of CD19 regulatory elements, which express Cre beginning at an early stage of B cell differentiation (25). Mice lacking RANKL in CD19-Cre-expressing cells will hereafter be referred to as RANKL ⌬B mice. RANKL mRNA was significantly lower in CD19 ϩ B cells isolated from 7-month-old RANKL ⌬B mice compared with RANKL fl/fl littermates (Fig.  1A). However, RANKL mRNA levels in either T cells or whole spleens were not different in RANKL ⌬B compared with RANKL fl/fl mice, demonstrating the specificity of CD19-Cremediated deletion. Notably, RANKL mRNA levels were lower in intact L5 vertebra from RANKL ⌬B mice compared with RANKL fl/fl littermates, demonstrating that B cell RANKL constitutes a significant portion of the total amount of RANKL mRNA present in whole bones (Fig. 1A).
Serial analysis of BMD from 2 to 7 months of age revealed that the bone mass of RANKL ⌬B mice was similar to that of three types of control littermates: mice heterozygous for the conditional allele, mice heterozygous for the conditional allele that also harbored the CD19-Cre allele, and RANKL fl/fl mice (Fig. 1B). This was the case whether bone mass was measured in the femur, spine, or in the whole body. BMD is significantly elevated in mice with germ-line deletion of RANKL or in mice lacking RANKL in osteocytes because of a decrease in osteoclastogenesis (11,21). Therefore, these results suggest that RANKL produced by B lymphocytes does not contribute to osteoclastogenesis during development or growth.
Previous studies have shown that germ-line deletion of RANKL also leads to an ϳ50% reduction in the percentage of CD45R/B220 ϩ IgM ϩ B cells in the spleen (11). Reconstitution studies suggested that this reduction was because of loss of RANKL in hematopoietic cells and possibly because of cell autonomous actions of RANKL in B cells. Consistent with this idea, the percentages of CD19 ϩ B cells and B220 ϩ CD43 Ϫ pre-B cells were slightly reduced in the bone marrow (Fig. 1C) and spleen (data not shown) of RANKL ⌬B mice. However, the percentage of CD3 ϩ T cells in the bone marrow was not altered (Fig. 1C). These results demonstrate that RANKL expression in B lymphocytes contributes to B cell development.
B Cell RANKL Contributes to Ovariectomy-induced Bone Loss-To determine whether RANKL produced by B cells contributes to the bone loss caused by ovariectomy, 6-month-old RANKL ⌬B mice and their RANKL fl/fl littermates were either ovariectomized or sham-operated, and the changes in bone mass were measured 6 weeks after the operations. Loss of estrogen was confirmed by the decrease in uterine weight and the increase in body weight, which occurred to a similar extent in both genotypes (Fig. 2, A and B).
Ovariectomy caused a decrease in BMD in both genotypes in the femur and the spine (Fig. 2, C and D). However, the ovariectomy-induced vertebral bone loss was significantly less in RANKL ⌬B mice compared with RANKL fl/fl mice (Fig. 2D). Consistent with this, CT analysis of the fourth lumbar vertebra (L4) showed that ovariectomy decreased cancellous bone volume in RANKL fl/fl but not in RANKL ⌬B mice (Fig. 2E). In addition, the decrease in trabecular thickness caused by ovariectomy was slightly lower in RANKL ⌬B mice compared with littermate controls (Fig. 2F). Ovariectomy did not cause significant changes in trabecular number or spacing in either genotype (data not shown). Deletion of RANKL from B cells also prevented loss of cancellous bone volume in the femur caused by ovariectomy (Fig. 2, G and J). Ovariectomy did not cause a significant change in trabecular thickness at this site in either genotype (Fig. 2H). In contrast to the prevention of cancellous bone loss, deletion of RANKL from B cells had no effect on the reduction in cortical thickness caused by ovariectomy (Fig. 2I).
Histomorphometric analysis of lumbar vertebrae confirmed blunting of the ovariectomy-induced cancellous bone loss in RANKL ⌬B mice (Fig. 3, A and B). Consistent with this, ovariectomy caused an increase in osteoclast surface and osteoclast number in control mice but not in RANKL ⌬B mice (Fig. 3, C-E). These results demonstrate that RANKL produced by B lymphocytes is required for the increase in osteoclast formation on cancellous bone that occurs after loss of estrogen, at least when measured 6 weeks after ovariectomy.
We next determined whether loss of estrogen altered the levels of RANKL produced by B lymphocytes. As expected, RANKL mRNA levels were significantly lower in CD19 ϩ B cells isolated from RANKL ⌬B mice compared with control littermates (Fig. 3F). However, ovariectomy did not alter RANKL mRNA levels in B cells from either genotype (Fig. 3F). Similarly, ovariectomy did not change RANKL mRNA levels in CD3 ϩ T cells isolated from either RANKL ⌬B or RANKL fl/fl mice (Fig.  3G).
Previous studies have shown that ovariectomy induces an increase in the percentage of B cells within the bone marrow  (26,27). Because RANKL contributes to B lymphopoiesis and because we observed that deletion of RANKL from B cells decreased the percentage of B cells in the bone marrow, we assessed whether deletion of RANKL from B cells affected the ovariectomy-induced changes in B cell populations. Consistent with previous studies, ovariectomy caused an increase in the percentage of CD19 ϩ B cells in the bone marrow of both genotypes (Fig. 3H). However, this increase was significantly lower in RANKL ⌬B mice (Fig. 3H). The percentage of CD3 ϩ T cells was not altered by ovariectomy in either genotype (Fig. 3I).
Although the RANKL mRNA levels in B cells were not elevated 6 weeks after ovariectomy, it remained possible that RANKL expression is elevated at an earlier time point after loss of estrogen. Consistent with this idea, previous studies have shown that sRANKL in the bone marrow of orchidectomized rats is elevated 1-2 weeks after gonadectomy. Therefore, wildtype C57BL/6 mice were sham-operated or ovariectomized, and the expression of RANKL in B cells was measured 2 weeks after the operations. Even in this shorter experiment, significant bone loss was observed in both the femur and the spine (Fig. 4A). However, consistent with the results of the 6-week experiment, there was no difference in RANKL mRNA abundance in B cells from sham-operated and ovariectomized mice (Fig. 4B). Nonetheless, both the total number of B cells and the percentage of B cells expressing RANKL protein on the cell surface were slightly increased in the bone marrow of the ovari-ectomized group (Fig. 4, C and D), although the average intensity of RANKL staining per cell was slightly lower in the ovariectomized group (Fig. 4E). Last, similar to the situation in orchidectomized rats, sRANKL in bone marrow supernatants was significantly higher in the ovariectomized mice (Fig. 4F).
Deletion of RANKL from T Cells Does Not Alter Bone Mass-To determine whether RANKL produced by T cells contributes to osteoclast formation, we crossed RANKL fl/fl mice with Lck-Cre mice, which express the Cre recombinase beginning early in the differentiation of the T cell lineage (28). Mice lacking RANKL in Lck-Cre-expressing cells will hereafter be referred to as RANKL ⌬T mice. Compared with the RANKL fl/fl littermate controls, RANKL ⌬T mice had lower RANKL mRNA expression in sorted T cells and whole spleens but not in other tissues (Fig. 5A). In addition to demonstrating the specificity of RANKL deletion, these results indicate that RANKL produced by T cells constitutes a significant portion of the total RANKL produced in the spleen.
Deletion of RANKL from T cells did not alter bone mass, as measured by dual-energy x-ray absorptiometry, up to 7 months of age (Fig. 5B). As with the B cell deletion findings, these results suggest that RANKL produced by T cells does not contribute to osteoclast formation during development and growth. These results are also consistent with a recent report showing no change in bone mass at 8 weeks of age in mice in which a RANKL conditional allele was deleted using the same Cre  AUGUST 24, 2012 • VOLUME 287 • NUMBER 35 driver strain used here (29). Deletion of RANKL from T cells had no effect on the percentage of T and B cells in bone marrow (Fig. 5C), suggesting that RANKL produced by T cells is not required for lymphocyte development.

Lymphocyte RANKL and Bone Loss
T Cell RANKL Does Not Contribute to Ovariectomy-induced Bone Loss-To determine whether T cell RANKL is involved in the osteoclast formation caused by ovariectomy, 6-month-old RANKL ⌬T mice and control littermates were sham-operated or ovariectomized and analyzed 6 weeks after the operation. Mice of both genotypes experienced similar decreases in uterine weight in response to ovariectomy (Fig. 6A). In addition, ovariectomy increased the percentage of B cells in the bone marrow to a similar extent in both genotypes (Fig. 6B) but had no effect on the percentage of T cells (Fig. 6C). More importantly, ovariectomy decreased BMD to a similar extent in both genotypes, whether measured in the femur, spine, or whole body (Fig. 6, D-F). Neither vertebral (Fig. 6G) nor femoral (data not shown) cancellous bone volume was significantly affected by ovariectomy in either genotype in this experiment likely because of high variation in this group of mice. However, ovariectomy reduced trabecular thickness in L4 vertebra (Fig. 6H) and cortical thickness in the femur (Fig. 6I) to similar extents in both genotypes. Overall, these results suggest that T cell RANKL does not play a role in ovariectomy-induced bone loss.

DISCUSSION
Multiple studies have demonstrated that the number of B cells increases after ovariectomy in mice (26, 30 -34), raising the possibility that this cell type plays an important role in the bone loss caused by estrogen deficiency (30). The genetic evidence presented here demonstrates that RANKL produced by B lymphocytes does indeed contribute to the increase in osteoclasts and decrease in cancellous bone that occur after ovariectomy in mice. Our finding that the amount of RANKL mRNA or protein expressed by B cells was not increased, either 2 or 6 weeks after ovariectomy, supports the idea that it is the increase in the total number of B cells expressing RANKL, rather than the expression level per cell, that stimulates osteoclastogenesis after loss of estrogen. This view is also consistent with previous studies demonstrating that elevation of B cell numbers via administration of IL-7 is sufficient to increase osteoclast number and cause bone loss in mice (30).
Consistent with our results in mice, an elevated number of bone marrow cells expressing RANKL was observed in estrogen-deficient women compared with women treated with estrogen for 3 weeks or premenopausal women (20). The identity of the cells expressing RANKL was not determined, but these cells were within a fraction of bone marrow cells that contains lymphocytes. In contrast, a second study in women suggests that estrogen suppresses the cell surface expression of RANKL in B lymphocytes as well as in T lymphocytes and bone marrow stromal cells, with no change in total cell number (19). Although this latter study did not examine the functional significance of changes in cell surface RANKL, it nonetheless suggests that there may be important differences in the response to estrogen loss in mice and humans. Alternatively, such differences may reflect differences in the timing of B cell enumeration after estrogen loss. Ovariectomy in mice induces a transient increase in B cell number, with a peak ϳ2 weeks after the operation (26). Whether there is a transient increase in the number of RANKL-expressing B cells in humans after acute estrogen loss remains unclear.
Ovariectomy was also associated with an increase in sRANKL in the bone marrow of wild-type mice 2 weeks after the operation. This coincided with a small decrease in the expression of RANKL protein on the surface of B cells. This finding is consistent with the possibility that ovariectomy stimulates cleavage of the membrane-bound form from the surface of B cells. Unfortunately, bone marrow supernatants were not collected from the ovariectomy studies involving the mice lacking RANKL in B cells. Therefore, identification of the cellular source of the sRANKL and its functional significance will need to be addressed in future work. . Ovariectomy increases B cell number but not RANKL gene expression. 7-month-old wild-type C57BL/6 mice were sham-operated or ovariectomized and killed after 2 weeks. A, changes in femoral and spinal BMD (n ϭ 13-14 per group). B, RT-PCR analysis of RANKL mRNA in CD19 ϩ B cells isolated from the bone marrow (n ϭ 13-14 mice per group). C and D, percentages of CD19 ϩ B cells and CD19 ϩ B cells that also express RANKL on their surface in the bone marrow (n ϭ 6 mice per group). E, mean intensity of cell surface RANKL protein detected on CD19 ϩ B cells from the bone marrow (n ϭ 6 mice per group). F, amount of sRANKL protein present in bone marrow supernatants normalized to total protein level (n ϭ 9 -12 mice per group). *, p Ͻ 0.05 using Student's t test.
In vitro studies have demonstrated that B cells at several stages of differentiation, from early precursors to plasma cells, can support osteoclast formation (16,(35)(36)(37)(38)(39). Nonetheless, Weitzmann and colleagues have shown that ovariectomy led to similar amounts of bone loss in mice lacking mature B cells compared with control littermates, suggesting that mature B  cells are not required for the bone loss caused by estrogen deficiency (40). These results, together with this study, suggest that it is the RANKL expressed by immature B cells that contributes to osteoclastogenesis after ovariectomy. However, because mice lacking mature B cells display altered bone remodeling prior to ovariectomy (41), it will be important to address this possibility directly by deleting RANKL using Cre deleter strains that become active only in mature B cells.
It is important to note that the increase in B cell number caused by ovariectomy may contribute to osteoclast formation by supplying cytokines other than, or in addition to, RANKL that promote osteoclast differentiation or function. It is also possible that the increased number of B lymphocytes may act as an additional source of osteoclast progenitors. Multiple studies have shown that isolated B lymphocytes can act as osteoclast progenitors, at least in vitro (16,(42)(43)(44)(45), although this has been attributed by some to contamination of the lymphocyte preparation with macrophage-lineage cells (46). In vivo lineage-tracing studies will be required to address this issue directly. Nonetheless, it remains possible that B cell lineage cells can act as osteoclast progenitors in vivo and thereby contribute to the increase in osteoclast formation after ovariectomy.
The protection from ovariectomy-induced bone loss conferred by deletion of any one of several inflammatory cytokines (47), together with the finding that deletion of estrogen receptor ␣ specifically in osteoclasts protects mice from cancellous but not cortical bone loss (9), suggests that there are multiple, nonredundant mechanisms involved in the bone loss caused by loss of sex steroids. Moreover, these studies demonstrate that different mechanisms are involved in cancellous versus cortical bone loss after loss of estrogen. Our finding that deletion of RANKL in B cells prevented cancellous but not cortical bone loss in ovariectomized mice adds additional support to these ideas.
Mice with germ-line deletion of the RANKL gene have defects in early B cell differentiation leading to reduced percentages of mature B cells in the spleen (11). This defect in B cell maturation was still observed after transplantation of RANKLdeficient hematopoietic cells into immuno-deficient mice, suggesting that RANKL produced by hematopoietic cells, but not stromal cells in the bone marrow or spleen, contributes to B cell maturation. Consistent with this, we found that deletion of the RANKL gene specifically from B cells blunted the increase in B cell number caused by ovariectomy. In light of the earlier transplantation studies, one possible interpretation of this result is that RANKL functions in an autocrine manner to promote B cell maturation. However, a recent study has shown that deletion of the RANKL receptor from B lymphocytes had no impact on B cell number or function (48). The Cre driver strain used to delete the RANKL receptor becomes active at a stage of B cell differentiation very similar to the CD19-Cre model used in our study (49). Therefore, taken together, these results suggest that B cell RANKL binds to the RANKL receptor expressed on hematopoietic cells that are not B cells to promote B cell differentiation.
T lymphocytes are thought to be involved in the bone loss caused by ovariectomy in mice (15), although some studies have been unable to demonstrate a requirement for this cell type (50). The major mechanism is thought to be increased production of T cells that produce TNF␣ (51), with a possible additional contribution of RANKL produced by T cells. Consistent with the latter possibility, increased production of RANKL by T cells has been observed in postmenopausal women compared with premenopausal controls (19,52). Nonetheless, we found in this study that deletion of RANKL from T cells had no impact on the bone loss caused by ovariectomy in mice and did not alter bone mass in intact mice. The latter finding is consistent with a recent report in which RANKL was deleted from T cells using the same Cre driver strain used in this study (29). In contrast to the results in B cells, RANKL deletion from T cells did not change T cell abundance in the bone marrow. These results demonstrate that RANKL produced by T cells is not required for either skeletal or T cell homeostasis under the conditions examined here. Additional work will be required to determine whether T cell RANKL plays a role in the bone loss caused by other conditions or in immune function.
Many different cell types can produce RANKL, and each of them could potentially contribute to the increase in osteoclast formation caused by loss of sex steroids. Although the studies reported herein argue for a functional role of RANKL produced by B lymphocytes in this process, it remains possible that other cellular sources also play a role. For example, in situ hybridization studies in growing rats indicate that ovariectomy increases RANKL mRNA production by hypertrophic chondrocytes (53). Although this cell type clearly cannot support osteoclast formation in postmenopausal women, it may nonetheless contribute to osteoclastogenesis in rodent models because the growth plate remains active in these species at the age at which most gonadectomy studies are performed. In addition, we and others have demonstrated recently that osteocytes produce the majority of the RANKL that controls osteoclastogenesis in cancellous bone (21,29). Thus, it will be important to determine whether the expression of RANKL in osteocytes and other cell types is altered by loss of sex steroids, and whether RANKL produced by such cells is required for bone loss after gonadectomy.