The Effect of Class A Scavenger Receptor Deficiency in Bone*

Class A scavenger receptor (SR-A) is predominantly expressed by macrophages, and because osteoclasts are of monocyte/macrophage lineage, SR-A is of potential interest in osteoclast biology. In addition to modified low density lipoprotein uptake, SR-A is also important in cell attachment and signaling. In this study we evaluated the effect of SR-A deletion on bone. Knock-out animals have 40% greater body weight than wild type. Body composition analyses demonstrated that total lean and fat body mass were greater in knock-out animals, but there was no significant difference in percent fat and lean body mass. Bone mineral density and content were significantly greater in knock-out compared with wild type animals. Micro-computed tomography analyses confirmed that total volume, bone volume as well as trabecular number, thickness, and connectivity were significantly greater in knock-out mice. As expected, trabecular separation was greater in wild type mice. The phenotype appears to be explained by 60% fewer osteoclasts in females and 35% fewer in males compared to wild type mice with a paradoxical increase in nuclei/osteoclast in knock-out animals. Furthermore, there were no differences in adipocyte number and osteoblast number or activity. The addition of the soluble extracellular domain of SR-A to RAW264.7 cells stimulated a concentration-dependent increase in osteoclast differentiation that was receptor activator of nuclear factor-κB ligand (RANKL)-dependent. Soluble SR-A had no effect on cell proliferation in the presence of RANKL but stimulated a 40% increase in numbers in the absence of RANKL. We conclude that SR-A plays a role in normal osteoclast differentiation, suggesting a novel role for this receptor in bone biology.

The class A scavenger receptor (SR-A) 2 or modified low density lipoprotein receptor is a multifunctional receptor associ-ated with several important pathological conditions including atherosclerosis (1) and prostate cancer (2,3). SR-A is most abundantly expressed in macrophages and has three isoforms, SR-AI/II/III (1). SR-A III is trapped in the endoplasmic reticulum and acts as a dominant negative protein with respect to SR-A activity (4). The SR-A are trimeric, integral membrane glycoproteins that bind an unusually broad array of macromolecular ligands including modified proteins and lipoproteins, nucleic acids, and a variety of plant and microbial cell wall constituents (1). Recent advances in understanding scavenger receptors indicate that these proteins have multiple diverse roles (5) and that SR-A is also important in cell attachment (6) and signaling (7). These roles include macrophage growth, adhesion to the substratum, cell-cell interactions, phagocytosis, and host defense (5,8). In addition, SR-A appears to be involved in intracellular signaling (6,9,10).
There are several subpopulations of resident macrophages including perivascular, peritoneal, synovial, intestinal, alveolar, and bone marrow macrophages (11)(12)(13). In bone marrow, hematopoietic stem cells give rise to osteoclasts that are derived from monocyte/macrophage precursors (14 -16). Osteoclasts form by a process of differentiation and cell fusion (17) that requires macrophage-colony stimulating factor (M-CSF) and receptor activator of nuclear factor-B ligand (RANKL) resulting in multinucleated, terminally differentiated cells (18,19). M-CSF production by monocytes is stimulated by oxidized lipids that are ligands for scavenger receptors (20). Differentiation of osteoclast precursors requires cell-cell contact with osteoblasts, whereas osteoclast attachment to bone is primarily integrin-mediated (15,21). In view of the role of SR-A in macrophage cell attachment and signaling, we investigated the role of SR-A in bone using SR-A knock-out (KO) mice. In this study we report that deletion of SR-A in vivo results in animals that are ϳ40% larger than the wild type controls while maintaining the same percentage lean and fat body mass. Histomorphometric, micro-computed tomography (CT), and body composition analyses demonstrate increased bone mineral density and content and decreased numbers of osteoclasts as well as increased trabecular thickness and connectivity. The bone phenotype results from an approximate 50% reduction in osteoclast numbers and resorption area with no effect on osteoblast number or bone formation rate. Furthermore the extracellular domain of SR-A synergistically stimulates differentiation of RAW264.7 cells into osteoclasts in vitro in a RANKL-dependent manner.
We conclude that SR-A has an important role in normal osteoclast development.

EXPERIMENTAL PROCEDURES
Animals-SR-A KO animals deficient in both SR-AI, AII, and AIII were generated through homologous recombination by disrupting exon 4 of the SR-A gene on a 129/ICR background (22). KO animals (generous gifts of T. Kodama Kyoto, University, Kyoto Japan) were subsequently backcrossed onto a 129/ SvJ background a minimum of 10 generations. SR-A Ϫ/Ϫ 129/ SvJ male and female mice were compared (10 animals/group) to WT age-matched 129/SvJ (The Jackson Laboratory, Bar Harbor, ME). All experimental procedures were reviewed and approved by the University of Kentucky Institutional Animal Care and Use Committee and the Lexington Veterans Affairs Medical Center Institutional Animal Care and Use Committee.
Mineralized Bone Histology and Bone Histomorphometry-Femurs were harvested from 4-week-old WT and KO mice that had undergone dual intraperitoneal calcein injections (5 and 2 days before euthanasia). Calcein is deposited at sites of bone mineralization and allows assessment of new mineralization and bone turnover. Bones were defleshed, fixed in 100% ethanol, dehydrated, and embedded in methylmethacrylate (23). Serial sections of 4 m were cut with a Microm microtome, model HM360 (Microm, C. Zeiss, Thornwood, NY). Alternate sections were stained with modified Masson-Goldner trichrome stain (24). Unstained sections were prepared for fluorescent light microscopy. Structural, cellular, and dynamic parameters of bone were measured at standardized sites under the growth plate in femoral metaphyses in mice using the semiautomatic method (Osteoplan II, Kontron, Munich, Germany) (25). The sites evaluated were confined to the secondary spongiosa to ensure that only areas of remodeling were analyzed (25). All parameters comply with the guidelines of the nomenclature committee of the American Society of Bone and Mineral Research (26). The Osteoplan II software is programmed to automatically transfer data to the statistical software SPSS for Windows (Version 13; SPSS, Chicago, IL).
Body Composition-Body composition analyses, fat mass, soft lean body mass, and bone mineral density (BMD) and content (BMC) were performed with a GE-Lunar PIXImus densitometer (software Version 1.44; Lunar, Madison, WI) as previously described (27). Animals were thawed to room temperature and scanned once. During analysis, the head of the animal was excluded using the exclusion tool. Thus, all body composition data are only from the body. After completion of the single dual-energy x-ray absorptiometry scan, necropsies were performed, and weights of the following fat pads were obtained to corroborate the densitometry data: paired gonadal, retroperitoneal, inguinal, and mesenteric fat pads. Peripheral dual-energy x-ray absorptiometry has been demonstrated to be a useful tool for the measurement of body composition in mice (28).
Osteocalcin Secretion-Serum was prepared from blood collected at euthanasia and stored at Ϫ80°C until assay. Serum osteocalcin was quantified with the murine osteocalcin enzyme-linked immunosorbent assay (Biomedical Technologies Inc., Stoughton MA) by comparison to known standards as suggested by the manufacturer.
Proliferation Assay-RAW264.7 cells (ATCC, Manassas, VA) were plated at 1.25 ϫ 10 4 cells/well in 24-well plates and cultured in ␣-minimal essential medium with 10% fetal bovine serum (Atlanta Biological, Atlanta, GA) in the absence or presence of 25 ng/ml murine RANKL and the indicated concentrations of soluble SR-A (construct was a generous gift of Alan Daugherty and Debra Rateri, University Kentucky). The sequence of the soluble extracellular SR-A is the same as originally reported (29). The His 6 tag-SR-A fusion protein was expressed in bacteria and purified on ProBond-Sepharose (Invitrogen) dialyzed against phosphate-buffered saline, sterile filtered, and stored at Ϫ80°C. RAW264.7 cell proliferation was assessed spectrophotometrically after 48 h of soluble SR-A treatment by the MTT assay at 570 nm according to the manufacturer's protocol (Sigma). Data are presented as a percent of the control cell density. Data were analyzed with GraphPad Prism software using nonlinear regression and F test to compare the fitted lines.
Osteoclast Differentiation-RAW264.7 cells were plated at 10 ϫ 10 4 cells/well in 6-well plates and cultured in ␣-minimal essential medium in the absence or presence of 25 ng/ml soluble murine RANKL and the indicated concentrations of soluble SR-A. Media were changed after 48 h. Cells were incubated an additional 16 -24 h, and differentiation was assessed by TRACP staining as described in the manufacturer's protocol (Sigma).
Statistics-Unless otherwise noted, data are presented as mean Ϯ S.E. because the data were obtained from many different litters of mice. Statistical comparisons were performed by analysis of variance or Student's paired t test. Normality of distribution was assessed by the Lilliefors test (56), and homogeneity of variance was tested with the Levene test (30). Histomorphometric results were analyzed using the Bonferroni post hoc comparison. Significant differences were concluded only if the null hypothesis is rejected at or below the 5% confidence level.
Sample Size Determination-There were no bone-related data in 129 SvJ mice available at the time of this study. However, we know the variance of bone formation rate/bone surface (BFR/BS) in FVB-N wild type mice and have data in FVB-N mice with targeted overexpression of insulin-like growth factor-1 (31). The sample size calculations used in this study were based on the mean difference in BFR/BS comparing insulin-like growth factor-1 transgenic and wild type mice (31). We anticipated that the mean difference in BFR/BS between SR-A KO and WT mice would be 6.3 mm 3 /cm 2 /yr with a common within-group S.D. of 4.5. With an ␣ level of 0.05 and the proposed sample size of 10 mice/group, the study has a power of 84.1% to yield a statistically significant result. Moreover, interim analyses are routinely conducted when data from 6 mice/group are obtained to adjust sample size if needed.

RESULTS
Four-week-old female WT and KO mice are illustrated in Fig. 1. Because the KO animals were significantly larger than the WT (Fig. 1A), we quantified the differences by comparing weights of WT and KO animals from at least five different litters (Fig. 1B). Female WT mice weighed 13.7 Ϯ 0.6 g (n ϭ 24), whereas the KOs weighed 19.7 Ϯ 0.8 g (n ϭ 18). Male WT mice weighed 14.6 Ϯ 0.8 g (n ϭ 17) compared with 20.2 Ϯ 0.9 g (n ϭ 22) for the KOs. On average, 4-week-old KO animals weighed ϳ40% more than age-and sex-matched WT animals. There was no significant difference in weight between male and female WT or KO animals (Fig. 1B).
Given the apparent difference in size, we harvested femurs from five female KO and WT animals to determine the average femur weight (Fig. 2) as an index of growth. The weight of femurs from KO mice were ϳ35% greater than from WT mice, consistent with the percent differences in body mass (see Fig. 1).
To determine whether growth was proportional between the KO and age-matched WT animals, we tested for differences in body composition between the phenotypes (Table 1). Body composition analyses were performed measuring total and per-cent fat mass, soft lean body mass, BMD, and BMC (27). There were no significant differences in the percent lean and fat mass (Table 1), although there were significant differences in total lean and fat body mass between WT and KO animals as would be expected in view of the differences in mass. Because femur weight, BMC, and body mass are ϳ40% greater in KO compared with WT animals, we conclude that growth is proportional between phenotypes. In addition, femur length (another index of growth) and BMD in KO animals were significantly greater than WT animals. Femur lengths in 4-week-old KO animals were 12.5 Ϯ 0.11 mm compared with 10.9 Ϯ 0.14 mm in WT animals (p Ͻ 0.001).
The mineral composition data were confirmed by CT analysis. Transverse CT slices were obtained from the femurs of three WT and KO mice to compute total cross-sectional area. The cross-sectional areas and the relative trabecular density at the mid-shaft are illustrated (Fig. 3). The increased total area and trabecular number are evident by comparing the transverse slices from the KO and WT gray scale images. The data from the CT analyses are summarized in Table 2. Changes in cortical bone thickness appear to be evident in Fig. 3, but these apparent changes did not reach statistical significance (p Ͻ 0.11, females), whereas there was a similar trend toward increased cortical thickness in males (p Ͻ 0.07). In addition, total volume, bone volume, and bone surface area in KO animals were all significantly greater than WT animals ( Table 2).     FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4655
The three-dimensional reconstructions of femurs from KO and WT animals (Fig. 4) demonstrates the increase in trabecular bone and connectivity (Fig. 4A) and the total bone content of KO animals compared with WT controls (Fig. 4B). In addition, increased trabecular density is readily apparent in the distal femur. The distal femur of the KO animals have taken on the characteristic club-like appearance reported in osteopetrotic knock-out models (18).
Because both BMD and BMC were significantly greater in KO mice, we compared dynamic histomorphometric parameters in the distal femurs (Table 3) to determine the nature of the differences at the cellular level. The only histomorphometric parameters that were significantly different between sexmatched mice were osteoclast-related. There was a significant reduction in the number of osteoclasts/100-mm bone surface in KO compared with WT animals. The percent reduction in osteoclasts was greatest in females (60%), whereas males had 35% fewer osteoclasts than sex-matched WT animals. There was a similar reduction in both the osteoclast surface/bone surface and erosion surface/bone surface in KO compared with WT animals (Table 3). Furthermore, there were no significant differences in any osteoblast-related histomorphometric parameters (see Table 3). In addition, there were no differences in the number of adipocytes counted in bone between KO and WT mice (not illustrated).
Serum osteocalcin levels were also compared between genders and phenotypes as an indicator of endogenous osteoblast activity. There were no significant differences in osteocalcin concentrations between males and females as well as WT and KO animals. Osteocalcin concentrations averaged 198 Ϯ 2.4 ng/ml for all groups and ranged from 194 to 205 ng/ml.
The reported role of SR-A in cell attachment led us to hypothesize that deletion of SR-A may reduce the ability of osteoclast precursors to fuse, thereby resulting in smaller, less active osteoclasts. Therefore, we quantified the number of nuclei/osteoclast between WT and KO animals during the histomorphometric analysis (500ϫ final magnification) of at least 10 mice/group. Counter to our hypothesis, KO animals had significantly more nuclei/osteoclast than WT animals (females, 2.3 Ϯ 0.2 compared with 1.2 Ϯ 0.3; p Ͻ 0.009; males, 2.1 Ϯ 0.8 compared with 1.3 Ϯ 0.5; p Ͻ 0.03). The data were more variable in males than females, but the increase was statistically significant for both genders.
Bone marrow macrophages from WT and KO animals were isolated for osteoclast differentiation experiments. There was no significant difference in osteoclast precursor cell recovery from the bone marrow cell isolations as determined by fluorescence-activated cell sorter analysis (not illustrated). Standard conditions (32,33) for murine macrophage differentiation into osteoclasts were problematic in KO-derived macrophages due to poor cell attachment. Although this makes functional studies on the role of SR-A in osteoclast differentiation much more difficult, it clearly suggests that a SR-A-mediated step is an important component of the process of osteoclast attachment/differentiation.
Because of the differences in cell attachment between KO and WT bone marrow macrophages, we reasoned that we could not adequately control this variable in a manner that would allow solid conclusions regarding the role of SR-A in osteoclasts because of the complexity of cell attachment-dependent signaling. An important factor regulating the number of osteoclasts present at a given time is the balance between proliferation of precursor cells and apoptosis of osteoclasts and/or the precursor cells. Therefore, we investigated the role of the soluble extracellular portion of SR-A (sSR-A) on proliferation and differentiation of murine RAW264.7 cells into osteoclasts in the absence and presence of soluble RANKL. Soluble SR-A stimulated a modest but significant (p Ͻ 0.05) concentrationdependent increase in RAW264.7 cell density in the absence of soluble RANKL, as determined by the MTT assay (Fig. 5). Halfmaximal stimulation of the proliferative response occurred at

Role of SR-A in Bone
0.2 nM, whereas the effect was maximal at 2 nM sSR-A. Overall, there was a 40% increase in cell density in response to sSR-A compared with control cells. There were no significant differences in cell numbers in similar assays performed in the presence of RANKL. RAW264.7 cells were cultured in the absence or presence of 25 ng/ml soluble murine RANKL and the indicated concentrations of sSR-A, and differentiation was assessed by TRACPstaining cells (Fig. 6). Soluble SR-A synergistically stimulated a concentration-dependent increase in osteoclast differentiation that was RANKL-dependent. Fig. 6A (20ϫ magnification) Fig. 6B, (100ϫ) illustrates the TRACP-positive multinucleated osteoclasts.

DISCUSSION
A potential role for SR-A in bone was hypothesized because SR-A is a receptor protein predominantly expressed by macrophages. Given the established role of macrophages in foam cell formation and numerous reports on the role of foam cells in    FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

Role of SR-A in Bone
formation of atherosclerotic lesions and vascular calcifications, we reasoned that since osteoclasts are of monocyte/macrophage origin, SR-A may have a potential role in bone remodeling (1,9,34). Furthermore, because SR-A is important in cell attachment and cell-cell interactions (9), deletion of SR-A may alter the ability of developing osteoclasts to undergo fusion and or cell spreading. In this study we focused on the role of SR-A in bone; the role of SR-A in atherosclerotic plaque formation has been reported in the atherosclerosis prone C57Bl/6 mice (35,36). SR-A is predominantly expressed by macrophages and to a lesser extent by smooth muscle cells and endothelial cells (37,38). Therefore, we predicted that deletion of a macrophage protein would result in an osteoclast phenotype, and clearly there is a significant reduction in osteoclast numbers in male and female KO mice. The percent reduction was greatest in female mice, but the percent reduction in both male and female mice was statistically significant. There is no readily apparent reason for the gender-based differences in osteoclast numbers based on the data obtained at this time.
This is the first report demonstrating a role of SR-A in skeletal growth and osteoclast differentiation. Low bone turnover due to decreased osteoclast numbers results in increased bone mineral content. In other animal models imbalances in bone remodeling resulting in alterations in bone homeostasis leading to osteoporosis or osteosclerosis and decreased bone turnover have been linked to vascular calcification (39). Deletion of SR-A has a protective effect against development of atherosclerotic lesions (40), whereas SR-A expression promotes foam cell formation (7). Taken together, this suggests that SR-A is likely to be important in both models.
Classically, SR-A is viewed as an important element in lipid related pathology (1). It is also well established that osteoporosis is associated with increased bone adiposity (41). Increased adiposity generally occurs at the expense of osteoblast differentiation because adipocytes and osteoblasts arise from the same progenitor cells. In fact, Tintut et al. (42) have shown that hyperlipidemia promotes osteoclast formation. They demonstrated a significant correlation between levels of atherogenic lipids and number of osteoclasts in low density lipoprotein receptor knock-out (LDLR Ϫ/Ϫ ) mice compared with wild type animals on a high fat diet. In a separate study the same group demonstrated that increases in oxidized lipids inhibited osteoblast differentiation (43). Deletion of SR-A protects against atherosclerosis since there is a significant reduction in foam cell formation (44). In this study loss of SR-A resulted in reduced osteoclast numbers but had no effect on static or dynamic osteoblast-related histomorphometric parameters. Furthermore, there were no changes in adipocyte number upon histomorphometric analysis. The current study focused on a detailed analysis of the SR-A deletion phenotype in bone and did not utilize primary culture data because the KO cells do not attach well to tissue culture plates. Differences in attachment efficiencies exclude in vitro comparisons of KO and WT cell differentiation because we cannot conclusively control alterations in attachment-related signaling. The SR-A knock-out model used in these studies was bred on a SvJ129 background, whereas the low density lipoprotein Ϫ/Ϫ studies were in C57BL/6J mice (42,43). The C57BL/6J stain is the standard model for atherogenic studies.
In view of the reported role of SR-A in cell attachment and cell-cell interactions, we hypothesized that osteoclasts from KO animals may have an impaired ability to undergo fusion and, therefore, have fewer nuclei/osteoclast than WT animals. The number of nuclei/osteoclast has previously been used as a positive index of osteoclast activity (45) (i.e. the more nuclei, the greater the activity). Fewer less-active osteoclasts should result in increased bone mass consistent with the observed phenotype. Paradoxically, osteoclasts from SR-A KO animals have more nuclei/cell than WT animals independent of gender. In fact, the histomorphometric data do not support a positive relationship between the number of nuclei/osteoclast and resorption area in this model since the percent reduction in erosion surface/bone volume is the same as the percent reduction in osteoclast number in KO compared with WT animals (see Table 3). The fact that deletion of SR-A results in fewer osteoclasts with more nuclei and substantially reduced bone resorption suggests SR-A is important in osteoclast development but that the integral membrane protein may somehow limit the extent of normal osteoclast precursor cell fusion.
The phenotype observed with deletion of SR-A is not as severe an osteopetrotic phenotype as has been reported in other KO models (18,46,47) which are characterized by very low osteoclast abundance, whereas models with enhanced osteoclastogenesis have osteoporosis (48,49). Typically, KO animals in other osteopetrotic phenotypes are smaller compared with WT animals (18,47). It seems unlikely, therefore, that a reduction in osteoclast numbers alone would result in the overall increase in body mass and increased bone density. In fact, the increased trabecular density evident by CT analysis in proximity to the growth plate in the KO animals suggests that SR-A, in addition to being important in normal osteoclast development, may have a regulatory role on bone growth. The nature of this putative regulatory role remains to be determined but could be either direct or indirect. A direct negative interaction of SR-A expressing macrophages on osteoblastic precursors seems less likely in view of the overall increase in growth, although Rowe and coworkers (50) recently reported that SR-A is transiently up-regulated in differentiating osteoblastic precursor cells. This observation raises the possibility that SR-A may be important developmentally and that low-level expression in smooth muscle and osteoblastic precursor cells may be important in growth and development as well as osteoclast formation.
M-CSF and RANKL are the two factors essential for osteoclast differentiation (51). Deletion of M-CSF results in the osteopetrotic op/op mouse (52). M-CSF has many effects on osteoclastic precursors including effects on attachment and proliferation and, interestingly, also induces increased expression of SR-A by macrophages (53). The relationship between M-CSF and SR-A is not known, but the fact that loss of each protein individually results in increased bone mass suggests that SR-A has an important role in bone growth, whereas the role of M-CSF has been well established (51,54).
Deletion of SR-A alters normal osteoclast development in vivo in this mouse model. It is not known whether the bone phenotype observed is due solely to interactions within the bone microenvironment. The decrease in osteoclast numbers resulting from loss of SR-A may be an indirect effect due to altered secretion of other factors from either macrophages (inside or outside of the bone microenvironment), osteoclasts, or osteoclast precursor cells that have subsequent effects on bone growth. Alternatively, in view of the role of SR-A in cellcell interactions, loss of SR-A may interfere with the ability of osteoclast precursors to associate with osteoblasts, an essential step in osteoclast differentiation that is dependent on RANK (on the osteoclast precursor membrane) binding to RANKL on the osteoblast plasma membrane. The absolute requirement of RANKL for the synergistic effect of sSR-A on osteoclast differentiation is interesting since RAW264.7 cells normally express SR-A (55). These data suggest that sSR-A somehow has a potentiating effect on the RANK/RANKL-signaling complex resulting in the synergistic effect on osteoclast differentiation.
Deletion of SR-A in vivo altered osteoclast development as originally hypothesized and, surprisingly, also substantially increased overall growth in a proportional manner. The data demonstrate for the first time that SR-A is necessary but not sufficient for normal osteoclast development. The data also suggest that SR-A is important in bone growth/development and that SR-A has roles in vivo that are likely to be independent of modified low density lipoprotein uptake.