key: cord-0969044-hh4vqo57
authors: Martineau, Corine; Najyb, Ouafa; Signor, Céline; Rassart, Éric; Moreau, Robert
title: Apolipoprotein D deficiency is associated to high bone turnover, low bone mass and impaired osteoblastic function in aged female mice
date: 2016-05-19
journal: Metabolism
DOI: 10.1016/j.metabol.2016.05.007
sha: 980fab09f09736ecdcae23f598486d65025efdba
doc_id: 969044
cord_uid: hh4vqo57

BACKGROUND: Apolipoprotein D (ApoD) is a member of the lipocalin family known to transport small hydrophobic ligands. A major site of ApoD expression in mice is the central nervous system where evidence suggests that it plays a protective role. Gene expression of ApoD was reported in bone-forming osteoblasts but its impact on bone metabolism remains undocumented. METHODS: We compared basic bone parameters of ApoD(−/−) (null) and transgenic (tg) mice to wild-type (wt) littermates through microCT and histochemistry, as well as ApoD expression and secretion in osteoblasts under various culture conditions through real-time PCR and immunoblotting. RESULTS: ApoD-null females displayed progressive bone loss with aging, resulting in a 50% reduction in trabecular bone volume and a 23% reduction in cortical bone volume by 9 months of age. Only cortical bone volume was significantly reduced in ApoD-null males by an average of 24%. Histochemistry indicated significantly higher osteoblast surface and number of osteoclasts in femora from ApoD-null females. ApoD gene expression was confirmed in primary cultures of bone marrow mesenchymal cells (MSC), with higher expression levels in MSC from females compared to males. ApoD-null MSC exhibited impaired proliferation and differentiation potentials. Moreover, exogenous ApoD partially rescued the osteogenic potential of null MSC, which were shown to readily uptake the protein from media. ApoD expression was upregulated under low proliferation conditions, by contact inhibition and osteoblastic differentiation in MC3T3-E1 osteoblast-like cells. CONCLUSION: Our results indicate that ApoD influences bone metabolism in mice in a gender-specific manner, potentially through an auto-/paracrine pathway.

Bone is a dynamic tissue that undergoes continuous remodeling through coordinated processes of bone resorption and formation. The balance between both processes preserves bone tissue integrity and mineral homeostasis of the organism [1] . Osteoclasts and osteoblasts are specialized cells which respectively break down old bone tissue and promote the formation of new bone matrix. Osteoblasts, under the control of specific transcription factors, sequentially differentiate from bone marrow mesenchymal stem cells (MSC) into proliferating preosteoblasts, bone matrix-producing osteoblasts, and eventually into osteocytes embedded within the bone matrix [2] . Differentiated osteoblasts synthesize and secrete type I collagen, the main bone matrix protein, and regulate bone mineralization through expression of alkaline phosphatase and osteocalcin. In addition, these cells orchestrate bone remodeling by regulating the activation and differentiation of bone-resorbing osteoclasts from the monocyte/macrophage lineage [1] . Any imbalance between resorptive and formative functions of these cell populations may lead to excessive bone resorption or inadequate formation, resulting in loss of bone density, lower bone mass and increased risk of bone fractures [3] as observed in osteoporosis.

Apolipoprotein D (ApoD) is a 29-kDa glycoprotein first isolated from high density lipoprotein (HDL) plasma fractions [4] . Unlike other apolipoproteins, which are essentially produced in liver and intestine, ApoD is mainly expressed in the central nervous system (CNS) [5] . Molecular cloning and structural analyses revealed that this protein is part of the lipocalin superfamily [6] . Lipocalins share a conserved 8 beta barrel strand structure, forming a pocket with high affinity for small hydrophobic ligands. Several molecules were identified as potential ligands for ApoD: arachidonic acid, progesterone, pregnenolone, bilirubin, cholesterol and E-3-methyl-2-hexenoic acid (reviewed in Ref. [7] ). Binding studies revealed that ApoD is able to discriminate between closely related compounds [8] . Due to this apparent heterogeneity, a role as a 'multi-ligand:multifunction' protein has been proposed [7] . ApoD expression is induced by growth arrest [9, 10] , inflammation and oxidative stress [11, 12] . The CNS was shown to be the major site of ApoD expression in mice and rats [5] and therefore, its physiological functions have mainly been studied in the nervous tissue. In rats, ApoD levels increase considerably in peripheral nervous tissue after experimental nerve injury [13] [14] [15] . In humans, ApoD expression is induced by several neurodegenerative and psychiatric disorders (for a review [16, 17] ). Thus, there is accumulating evidence suggesting that ApoD plays a protective role in the CNS. Of interest, higher ApoD expression levels are measured in women; moreover, these levels increase with age in women but not in men [18] . The protein is also overexpressed in several types of cancer [19, 20] , in lipid metabolism disorders such as Tangier disease [21] , abetalipoproteinemia [22, 23] , LCAT deficiency [24] , hypertriglyceridemia at birth [25] , and in pathological conditions leading to lipid accumulation such as non-insulin dependent diabetes mellitus (NIDDM) [26, 27] .

Microarray analyses suggest that the basal ApoD gene expression in human MSC is increased during osteogenic differentiation [28] . Similar observations were reported in the murine MSC-like cell line C3H10 [29] and in mouse primary osteoblasts [30] . Although limited in number, these studies suggest a potentially significant involvement of ApoD in bone metabolism. The present work explored the regulation of ApoD expression in osteoblasts under various culture conditions, and tried to determine the extent of its contribution to bone metabolism using ApoD-deficient (null) and human ApoD transgenic (tg) mouse models.

Materials and methods

ApoD-null mice were generated on a C57Bl/6J background by inserting a targeting vector with the neomycin phosphotransferase (Neo) gene interrupting exon 6 of the ApoD gene (for the complete protocol, see Ref. [12] ). Heterozygous females were crossbred with null males: the resulting progeny were analyzed. The human ApoD transgenic (hApoD-tg) mice were generated as described in Ref. [12] . Briefly, these mice carry a construct (~4.5 kb) containing the promoter, the first exon, the first intron and the 5′ noncoding region of the second exon of the human Thy-1 gene (generous gift from J. Silver, New York University Medical Center). This fragment was fused to the hApoD coding sequence followed by the bovine growth hormone (BGH) polyadenylation signal. Mice were backcrossed with C57BL6 mice for at least 10 generations. This study was conducted according to protocols approved by the Institutional Animal Care and Use Committee (IACUC, #706) of Université du Québec à Montréal. No significant difference was noted in body weight for either mouse strain at all ages used in this study (Table 1) . Following isoflurane general anesthesia, whole blood was harvested from 9-month-old mice through cardiac punctures and collected in 3 mL heparinated tubes (68 USP, BD Bioscience, Mississauga, ON, Canada) and mice were sacrificed by cervical dislocation. Plasma was Table 1 -Weight of 3 to 9-month-old wt, ApoD-null and hApoD-tg male and female mice.

Male 

Following euthanasia, femora from 3-, 6-and 9-month-old mice were harvested and fixed in 4% paraformaldehyde-phosphate buffered saline (PF) at 4°C for 16-20 h. MicroCT analyses were performed using a Skyscan 1172c X-ray computed microtomograph (Skyscan, Kontich, Belgium) equipped with an X-ray tube working at 70 kV/100 μA. Femora of wt, ApoD-null and hApoD-tg mice were scanned at a 5 μm pixel size, a 180°r otation with a 0.5°rotation increment using a 0.5 mm aluminum filter. Stacks of 2D X-ray shadow projections were reconstructed with a 0 to 0.09 dynamic range to obtain crosssectional images using NRecon (Skyscan), and subjected to morphometric analyses using CTAn (Skyscan). Trabecular parameters were measured at the metaphysis (400 slices -2 mmwere selected 100 slices below the growth plate); the cortical region consisted of 100 slices -0.5 mmselected 600 slices below the growth plate reference point. The parameters measured included trabecular bone volume (bone volume/tissue volume (BV/TV)), trabecular thickness (Tb.Th), number of trabeculae (Tb.N), trabecular separation (Tb.Sp), as well as cortical bone volume (Ct.BV), cortical thickness (Ct.Th) and cortical diameter (Ct.Dm). Three-dimensional renderings were generated from these volumes of interest using CTvol (Skyscan). Paraformaldehyde (PF)-fixed undecalcified bones were embedded in low-temperature polymerizing polymethylmethacrylate (PMMA) as previously described [31] , sectioned at 6 μm with a Thermofisher rotary HM 360 microtome and used for Von Kossa, alkaline phosphatase (ALP) (Millipore, Billerica, MA, USA) and tartrate resistant acid phosphatase (TRAP) (K Assay, Dako, Burlington, ON, Canada) histochemical stainings. Von Kossa (bone area per tissue area -B.Ar/T.Ar), ALP (osteoblast surface -Ob.S) or TRAP-stained (number of osteoclasts -Oc.N/mm) bone sections were visualized with an inverted phase contrast microscope (TE-300, Nikon, Mississauga, ON, Canada) and analyzed with ImageJ.

Long bones from 2 to 4 month-old female mice hind limbs were harvested and sterilized in phosphate-buffered saline (PBS) containing 200 U/mL penicillin-200 μg/mL streptomycin and 1% FungiZone (all from Invitrogen). Epiphyses were cut off under sterile hood, marrow was flushed out, suspended in αMEM supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin, 2 mmol/L L-glutamine, 10% FBS, and 25 μg/mL L-ascorbic acid (Sigma) and plated in 100 mm culture dish (Sarstedt, Montréal, QC, Canada). Bones from 1 mouse were used per dish, unless specified otherwise. Cells were left to adhere for 7 days, and then thoroughly washed with PBS to eliminate non adherent cells. Adherent cells were left to reach confluence prior to harvest and experimentation. For primary cultures of osteoblasts, bone fragments from femora and tibiae were subjected to three consecutive digestions with collagenase IA (Sigma); fragments were plated with αMEM medium containing 1% FungiZone (Invitrogen) and 10% FBS in 100 mm dishes (Sarstedt) until cell outgrowth to confluent monolayers. MC3T3-E1 osteoblast-like cells derived from mouse calvaria and constitute a well-known in vitro model for studying osteoblast function [32] . MC3T3 cells (American Type Culture Collection, ATCC; Rockville, MD, USA) were maintained in alphamodified minimum essential medium (αMEM; Sigma, Oakville, ON, Canada) containing 10% fetal bovine serum (FBS; NorthBio, Toronto, ON, Canada), 2 mmol/L L-glutamine (Invitrogen, Burlington, ON, Canada), 100 U/mL penicillin (Invitrogen), and 100 μg/mL streptomycin (Invitrogen), and subcultured every 7 days with 0.05% trypsin-0.02% EDTA solution (Invitrogen).

For proliferation assays, primary cells were plated in 96-well plates at 20,000 cells per cm 2 and cell metabolic activity was evaluated 24 h later (day 0) as well as on days 1, 3, 6, 10 and 14, through tetrazolium microtiter (MTT) assay (Sigma). For analyses of DNA synthesis, cell cycle and gene expression, MC3T3 cells were cultured in medium containing 10% FBS (control) or serumstarved for 48 h (low proliferation condition), or cultured in medium containing 10% FBS until monolayers reached subconfluence (70%), 100% confluence, and 3 days after reaching confluence (post-confluence). For evaluation of DNA synthesis, cells were seeded in 96-well plates (Sarstedt) at 2500 cells/cm 2 and cultured under the proliferation or confluence conditions described above. DNA synthesis was determined by BrdU incorporation and detected using an ELISA kit (Roche Diagnostics, Laval, QC, Canada) following manufacturer recommendations. Briefly, cells were incubated with BrdU for the last 2 h of treatment, then fixed and treated with a nuclease solution. Incorporated BrdU was detected with anti-BrdU-POD for 30 min and revealed with a peroxidase substrate. For cell cycle analysis, cells cultured in parallel under the same conditions were stained with propidium iodide (PI) and analyzed by flow cytometry (FACScan, Becton Dickinson) using FlowJo software (version 8.7.3).

To verify the short term effect on Opg, Rankl and Sod1 expression, wt and null MSC were treated with 250 ng/mL of hApoD for 24 h. The impact of hApoD on osteogenic differentiation was compared to that of basal MEM medium and to regular osteogenic medium. Briefly, confluent cells were treated for 21 days (MC3T3-E1 cells) or 14 days (wt and null MSC) with or without 250 ng/mL hApoD or 50 μg/mL ascorbic acid (Sigma) and 5 mmol/L glycerol-2-phosphate (Sigma); media were changed 3 times a week. The effectiveness of treatments was monitored through alkaline phosphatase (ALP) activity versus cultures in basal medium, with cytochemical staining (Millipore, Billerica, MA, USA) and pnitrophenylphosphate conversion to p-nitrophenolate (as described in Ref. [33] ). In parallel, marrow-derived primary cells were cultured onto glass slides and exposed to 250 ng/mL FITC-labeled hApoD for 24 h. Cells were then rinsed with PBS and fixed with 10% formalin, then permeabilized with 0.1% triton X-100 (w/v) and stained with PI. hApoD and PI fluorescence were visualized with an inverted phase contrast microscope (TE-300; Nikon, Mississauga, Ontario, Canada) at a 40× magnification.

Total RNA was extracted using TriZol (Invitrogen) according to the manufacturer's instructions. Reverse transcription (RT) reactions were carried out with Omniscript RT kit (Qiagen, Mississauga, ON, Canada) using random hexamers (Invitrogen). PCR amplifications for semi-quantitative analysis were conducted with Taq PCR core kit (Qiagen) using specific primer sets for murine ApoD (F: 5′-CCACCGGCACCCTACTGGATC-3′, R: 5′-CGGGCAGTTCGCTTGATCTGT-3′) and human ApoD (F: 5′-TTA ACCTCACAGAGCCTGCC-3′, R: 5′-GAGTCCACTGTTTCTGGAG GG-3′); Gapdh (F: 5′-AGAACATCATCCCTGCATCC-3′, R: 5′-AGTT GCTGTTGACGTCGC-3′) was used as a reference gene for normalization. Amplifications were carried out for 40 cycles of 1 min at 94°C, 30 s at 58°C and 1 min at 72°C. PCR products were resolved in 2% agarose gel and visualized under UV with Redsafe (FroggaBio, North York, ON, Canada) staining. Real-time PCR analysis for mouse ApoD, Opg, Rankl and Sod1 was performed using the iCycler IQ detection system (Bio-Rad, Hercules, CA, USA) and SYBR Green I (Bio-Rad) as a doublestrand DNA-specific binding dye, using β-microglobulin (B2m) as a reference gene (F: 5′-TACTCAGCCACCCACCGGCCG-3′, R: 5′-GCTCGGCCATACTGGCATGCT-3′) and the following sets of specific primers: Opg (F: 5′-GTGTGGAATAGATGTCACCCTGT-3′, R: 5′-CTTGTGAGCTGTGTCTCCGT-3′), Rankl (F: 5′-CCCATCG GGTTCCCATAAAGT-3′, R: 5′-AGCAAATGTTGGCGTACAGG-3′) and Sod1 (F: 5′-CACTTCGAGCAGAAGGCAAG-3′, R: 5′-CCCCATA CTGATGGACGTGG-3′). Each sample was run in triplicate, and fluorescence data were collected at the end of each cycle. Each run ended with a melting curve to ensure amplification specificity (from 60 to 95°C at a 0.5°C/10 s rate). Expression levels were calculated using the comparative CT method [i.e., 1/(ΔΔCT), where ΔCT is the difference between CT target and CT reference] after normalization to B2m expression level, and relative fluorescence units (RFU) were analyzed using optical system software Version 3.1 (Bio-Rad).

MC3T3-E1 cells were grown to confluence in 12-well plates (Sarstedt), cultured in control or osteogenic medium for 21 days and analyzed for ApoD expression and secretion on days 0, 7, 14 and 21. Twenty four h prior to protein or RNA extraction, medium was changed for FBS-free MEM which was collected at time of extraction to recuperate secreted ApoD. Cellular proteins were extracted in lysis buffer for immunoblotting, as RNA was retrieved by TriZol (Invitrogen) for PCR analysis. For protein secretion, 40 μL of media was loaded per well; for protein expression, 10 μg of cell extract was loaded per well. All samples were mixed with 5× denaturing loading buffer and resolved on a 12% SDS-PAGE (Bio-Rad Laboratories, Mississauga, Canada). Gels were transferred onto PVDF membranes and immunoblotted with a mouse-specific goat anti-ApoD (1:1000, sc34760, Santa-Cruz sc34760 biotechnology, Dallas, TX, USA), followed by an anti-goat secondary antibody (1:10,000, Santa Cruz) coupled to HRP. For ApoD secretion, membranes were dyed with Amido Black as a loading reference; for ApoD expression, membranes were incubation with a mouse monoclonal anti-β-actin (1:5000, Sigma-Aldrich). Proteins were revealed with chemiluminescence (ECL, GE Healthcare, Quebec, Canada) using the Fusion FX7 (Vilber Lourmat, France) apparatus and software.

Data are expressed as mean ± SEM, and analyzed using Prism 5.0 Software (GraphPad, La Jolla, CA, USA). Significant differences between groups were evaluated either by Student t test (2 groups) or two-way ANOVA followed by a Bonferroni post-test (3 or more groups), as detailed in the legends. Differences were considered significant at p < 0.05.

Microcomputed tomography (microCT) analysis of femora from the three mouse strains revealed reduced bone volume in trabecular and cortical regions of ApoD-null females when compared to wt females (Tables 2 and 3 ). This parameter was exacerbated by aging, ranging from 18% in 3-month-olds to 50% 9-month-olds. Bone loss was associated to lower trabecular number and reduced trabecular thickness (Fig. 1A , top row and Table 2 ). Trabecular structure of ApoD-null males and hApoD-tg mice of either gender remained unchanged. As shown in Table 3 and Fig. 1A (bottom row), cortical bone volumes were progressively reduced with age in both female and male ApoD-null mice, ranging from 13% at 3 months (non-significant) to 23% at 9 months. This parameter in ApoD-null mice was also 24% lower in all age groups when compared to gender-matched wt mice. Loss of cortical bone volume was associated to reduce cortical thickness and diameter for both male and female null mice. Calcium and phosphorus plasma levels were measured in 9month-old mice; no difference was noticed for either parameter between wt, tg and null mice (Table 4 ). Total plasma ALP levels were 120% higher in females relatively to males, independently of genotype; this parameter was augmented by 40% in ApoD-null females versus gender-matched wt mice. Histological analysis was performed on 9-month-old mice of all three genotypes; Von Kossa (VK) staining of mineralized bone area was reduced in femora sections from ApoD-null female mice compared to wt mice (Fig. 1B, top row; Table 4 , not significant). Over 50% greater osteoblast surface was observed in bone tissue of females versus that of males, independently of genotype. This parameter was also 20% higher in null versus wt females (Fig. 1B , middle row; Table 4 ). In null females, this was accompanied by a 20% rise in osteoclast number (TRAP positive cells) (Fig. 1B , bottom row; Table 4 ); the null male bones also showed a 15% increase. No significant differences were noted in tg mouse femora through either microCT or histological analysis.

Since the low bone mass phenotype was more pronounced in female femora, ApoD expression was compared between MSC harvested from male and female wt mice. ApoD gene expression was found to be 1.5-fold higher in MSC isolated from females ( Fig. 2A) . In parallel, the Rankl to Opg mRNA ratioan index of osteoclastogenesis [34] was twice as high in ApoD null MSC derived from females (Fig. 2B) ; though this index was also elevated in male null MSC, it was not found to be significant. All further experiments were therefore carried out on MSC harvested from female marrow. ApoD-null MSC showed significantly reduced in vitro survival at day 14 (Fig. 2C) , with a proliferation rate of 0.5-fold that of wt cells. These cells are also capable of internalizing exogenous hApoD (Fig. 2D) ; we therefore tested its effect on wt and null osteoblasts compared to basal and osteogenic media.

We first verified whether exogenous hApoD could influence the Rankl:Opg ratio; this treatment corrected the ratio back to wt cell control values (Fig. 3A) . Because some studies associate higher serum Rankl:Opg ratio to oxidative stress [35] , the expression of superoxide dismutase 1 (Sod1) was observed under the same conditions: Sod1 expression paralleled the Rankl:Opg mRNA ratio (Fig. 3B) , suggesting the involvement of Table 2 -Trabecular bone morphometric parameters in 3-to 9-month-old wt, ApoD-null and hApoD-tg mice.

Age ( oxidative stress in the anomalies observed in ApoD-null cells. An osteogenic treatment in mouse MSC was then performed for 14 days, since null MSC become difficult to maintain passed this time point. No difference was seen in basal ALP activity between wt and null MSC; however, null MSC did not respond to osteogenic treatment (Fig. 3C, left panel) . Addition of hApoD showed an osteogenic effect on wt cells, and partially rescued the phenotype of null cells. Note that obtaining uniform monolayers of null cells was challenging in the untreated and osteogenic conditions; the presence of hApoD in the medium seemed to correct that situation (Fig. 3C, right panel) . To explore the relationship between ApoD and osteoblastic differentiation, we investigated its expression and secretion in differentiating MC3T3 cells, using a classical ascorbic acid and glycerol-2-phosphaste medium (Fig. 4A) . As shown in Fig. 4B , ApoD gene expression was upregulated over 2-fold at 7 days of osteogenic treatment and onwards. In accordance, ApoD cellular protein levels were increased under these conditions (Fig. 4C, top panel) , showing concomitant increase of ApoD secretion in culture medium (Fig. 4C, bottom panel) . As was performed on MSC, ALP activity was monitored following 21 days in basal or osteogenic medium, or following addition of hApoD (Fig. 4D) . hApoD alone also significantly enhance ALP activity in MC3T3 cells, although not being as potent as regular osteogenic medium.

Because ApoD deficiency seemed to impair MSC survival and confluence, we evaluated ApoD gene expression regulation in MC3T3 cells under various growth conditions. Given that growth arrest is associated with ApoD upregulation in fibroblasts and several other cell types [9, 10] , we observed its expression in serum-starved MC3T3 cells. As shown in Fig. 5A (left panel) , serum starvation reduced by 15% the number of cells in S and G2/M phases, with a concurrent increase of cells in G1 phase. In accordance, BrdU incorporation was significantly reduced by serum deprivation (Fig. 5A , middle panel). Under these conditions, ApoD gene expression in MC3T3 cells was increased 1.5-fold (Fig. 5A, right panel) . Given that MC3T3 cells are sensitive to contact inhibition, we verified ApoD gene expression in sub-confluent (70% of totalarea), confluent (100% of total area) and post-confluent (3 days after confluence) cell monolayers. As shown in Fig. 5B (left panel), cell percentage in the S and G2/M phases was significantly reduced by 16% to 26% in confluent and postconfluent monolayers, shifting cells to G1 phase with a concurrent reduction in DNA synthesis (Fig. 5B, middle  panel) . Under these culture conditions, ApoD gene expression was increased 1.42-fold (Fig. 5B, right panel) . independent experiments. Student t test (*p < 0.05) vs. gender-matched wt. (C) MSC from wt and ApoD-null female mice were seeded in 96-well plates and cultured for 14 days. MTT activity was measured to monitor the cell growth. Values are the mean ± SEM of 3 cell preparations. Bonferroni (*p < 0.05, **p < 0.01) vs. MTT activity for MSC from wt mice. (D) MSC from female wt mice were exposed to FITC-labeled hApoD and uptake was confirmed by confocal microscopy. Images are representative of 3 independent cell preparations.

Osteoporosis is a health issue with significant morbidity and social cost [36] . This pathology is defined as lower bone mass and higher bone fragility [37] . Several risk factors have been identified that alter osteoblast and osteoclast activities; besides age, ethnicity and gender, numerous metabolic and genetic disorders translate to inadequate bone turnover [37, 38] . The aim of the present study was to confirm the role of ApoD in osteoblastic cells and consequently, in proper bone turnover.

Our results indicate that ApoD is upregulated in osteoblasts under conditions that reduce proliferation rate. Contact inhibition and osteogenic differentiation both upregulated ApoD gene expression; moreover, its addition to culture media enhances osteogenic differentiation to some extent, suggesting a paracrine contribution to this process. In accordance with its suggested role in osteoblast function, ApoD-null female mice display lower trabecular and cortical bone volumes with concurrent higher osteoblast surface and number of osteoclasts. Only cortical bone volume was reduced in ApoD-null male mice, arguing for gender-specific interactions.

The contribution of ApoD to bone metabolism in vivo was investigated by documenting bone status in ApoD-null and hApoD-tg mice. Our results show that bone mass in null female mice was significantly reduced at both trabecular and cortical levels. Low bone volume in ApoD-null females was associated to reduced trabecular thickness and number of trabeculae particularly accentuated with age, hinting that ApoD may play a significant role in age-related bone loss. Histochemical analysis of 9-month-old mouse femora revealed greater relative osteoblast staining and higher number of TRAP-positive cells, suggesting accelerated bone turnover in null females. Mineral plasma homeostasis of 9-month-old null and tg mice was similar to that of wt mice; however, a significant increase in plasma ALP activity was noted in the null females only. Low bone mass accompanied by high serum ALP has been reported in cases of accelerated bone turnover in post-menopausal women [39] . This phenotype bears similarities with aged human bone, which commonly shows reduced wall thickness in histological sections, an index of reduced bone matrix deposition. This condition may be associated to an increased activation frequency of bone remodeling units [40] .

In contrast, only the femoral cortex was altered in ApoD-null male mice. Higher ApoD gene expression was measured in MSC from females compared to those from male mice, suggesting that ApoD contributes to a higher extent in female bone metabolism, at least as age progresses. In accordance, higher ApoD expression levels in the CNS are also measured in women, and increase with aging in women but not in men [37] . The link between the gender-specific bone effects of ApoD and sex steroids remains unclear. The ApoD promoter shows estrogen response elements and thus, its expression and secretion could be influenced by sex steroids in breast cyst fluid [9, 41] and breast cancer cells [20, 42] . Other gender-specific effects have been reported; for instance, the neuroprotective effect of ApoD against human coronavirus OC43-induced encephalitis is greater in male mice, as they are more sensitive to this infection [43] . Perhaps the ability of ApoD to bind progesterone and pregnenolone might explain some of the gender-related differences. Since ApoD deletion predominantly impacts bone in aged female mice, ovariectomy-induced bone loss in this model could discriminate between ApoD-dependant and hormone-dependant effects.

ApoD-null MSC were found to have significantly impaired proliferation and differentiation. These data corroborate the idea of accelerated bone turnover, as reduced osteoblastic life span due to microenvironmental stress has been associated to accelerated remodeling [44] . We observed enhanced ApoD gene expression in MC3T3 cells under osteogenic culture conditions. This result agrees with the reported ApoD upregulation with osteogenic differentiation of human MSC [28] and mouse primary osteoblasts [30] , as well as high-throughput gene expression profile (biogps.org, ID11815) and microarray analyses of C3H10 cells [29] . Moreover, we report that osteogenic differentiation is associated with increased cellular and secreted ApoD protein levels in MC3T3 cells, arguing for an auto/paracrine function of ApoD in bone metabolism. Mouse MSC readily uptake the protein from media; accordingly, treatment with exogenous hApoD increased ALP activity in undifferentiated MC3T3 cells, and partially rescued the osteogenic potential of ApoD-null MSC.

The purpose for enhanced ApoD expression during osteogenic differentiation is unknown, but evidence suggests its requirement for osteoblast maturation and function. Robust energetic demands are placed on mature osteoblasts for matrix secretion and mineralization. In accordance, cellular respiration and mitochondrial activities are increased, likely maintaining higher ATP levels observed in mature osteoblast [45] . Accordingly, antioxidant enzymes manganese-dependent superoxide dismutase (SOD) and catalase were shown to be upregulated in human MSC upon osteogenic differentiation, likely to prevent the accumulation of intracellular ROS. Of note, there is increasing evidence for a prominent neuroprotective role of ApoD because of its antioxidant activity (reviewed in Ref. [46] ). Loss of ApoD function increases, whereas ApoD overexpression reduces sensitivity to oxidative stress, as well as lipid peroxidation in brain tissue after oxidant treatment in mice [12] . Specifically, hApoD has been shown to reduce arachidonic acid hydroperoxides to lipid hydroxides, thus acting as a lipid antioxidant and preventing lipid peroxidation [47] . Therefore, ApoD upregulation during osteogenic differentiation may act as a protection mechanism against oxidative stress. Interestingly, the higher Rankl:Opg ratio and Sod1 expression observed in female-derived null MSC were both rescued by exogenous hApoD, reinforcing the hypothesis of an antioxidant role of ApoD in bone tissue.

Oxidative stress was shown to decrease osteoblast life span as evidenced in osteoblasts from Sod1-null mice while preventing proper osteoclastogenesis, leading to low-turnover bone loss [48] ; ApoD-null mice on the contrary display signs of high-turnover bone loss. Perhaps the impact of oxidative stress on bone remodeling is biphasic, with low or intermediate levels accelerating bone turnover while high levels impair it altogether. In addition, ApoD may also influence RANKL expression and activation through inflammatory pathways in bone microenvironment. By binding to arachidonic acid, its main ligand, ApoD may restrain its availability. Arachidonic acid is the major precursor of leukotrienes and prostaglandins, two groups of potent inflammatory modulators known to impact bone metabolism [52] . Oxidative stress is known to enhance osteoclastogenesis and inhibit osteoblastogenesis through NF-κBmediated RANKL regulation [35, 49] . It would be of interest to compare the levels of ROS and antioxidant components, as well as Rankl:Opg ratio between ApoD-null and Sod1-null MSC. Evidence from pharmacological and genetic studies in mice has provided support for a deleterious effect of oxidative stress in bone and has strengthened the idea that increased ROS levels with age represent a pathophysiological mechanism underlying age-related bone loss (reviewed in Ref. [50] ).

Because increased ApoD gene expression occurs in several cell models following growth arrest and cell differentiation [9, 42, 51] , we investigated ApoD gene expression in osteoblasts with regards to cell density. Low proliferation rate induced by serum starvation, as well as contact inhibition caused by confluence, both enhanced ApoD gene expression in MC3T3 cells. These conditions were associated with reduced BrdU incorporation and lower proportion of cells in the S and G2/M phases, confirming proliferation arrest. Similar ApoD upregulation by growth arrest was observed in human and murine fibroblasts, as well as in various cancer cell lines [9, 10, 11, 51] . Alternating purine pyrimidine (APP) stretches and serum response elements (SRE) in the promoter of hApoD gene were found to significantly contribute to growth arrest-induced ApoD gene expression [9] . Further mechanistic studies revealed the importance of Parp-1, APEX-1 and ERK1 catalytic activities in the growth arrest-induced ApoD gene expression [51] .

In light of the present data, we suggest that ApoD deficiency causes high bone turnover by impairing osteoblast life span and differentiation. Considering ApoD's property to limit oxidative stress [12, 45, 46] , it may interact with key processes in osteoblasts. ApoD might not only bind and transport lipophilic molecules, but may also actively interfere with their metabolism and signaling in an antioxidant and anti-inflammatory manner. Overall, our results indicate that ApoD contributes to the regulation of bone metabolism in mice and further studies are required to pinpoint its mechanisms. The evidence of ApoD secretion and uptake by osteoblasts suggest a paracrine function, potentially representing a novel marker of age-related bone loss.

RM and ER designed the study; CM, ON and CS conducted the study, collected and analyzed the data; CM, ON, CS, RM and ER interpreted the data; RM, CM, ON and CS drafted the manuscript; CM, ON, CS and ER revised and approved the final version.

This work was supported by the Canadian Institutes of Health Research grants MOP-37994 (ER) and MOP-89958 (RM) and a catalyst grant to both ER and RM.

Cellular mechanisms of bone remodeling

Networks and hubs for the transcriptional control of osteoblastogenesis

Pathogenesis of osteoporosis: concepts, conflicts, and prospects

Isolation and partial characterization of apolipoprotein D: a new protein moiety of the human plasma lipoprotein system

Molecular characterization and differential mRNA tissue distribution of mouse apolipoprotein D

Cloning and expression of human apolipoprotein D cDNA

Bacterially produced apolipoprotein D binds progesterone and arachidonic acid, but not bilirubin or E-3M2H

Modulation of apolipoprotein D and apolipoprotein E mRNA expression by growth arrest and identification of key elements in the promoter

Apolipoprotein D transcription occurs specifically in nonproliferating quiescent and senescent fibroblast cultures

Modulation of apolipoprotein D expression and translocation under specific stress conditions

Apolipoprotein D is involved in the mechanisms regulating protection from oxidative stress

Fate of myelin lipids during degeneration and regeneration of peripheral nerve: an autoradiographic study

Regeneration-associated high level expression of apolipoprotein D mRNA in endoneurial fibroblasts of peripheral nerve

Modulation of apolipoprotein D and apolipoprotein E expression in rat hippocampus after entorhinal cortex lesion

Increased CNS levels of apolipoprotein D in schizophrenic and bipolar subjects: implications for the pathophysiology of psychiatric disorders

The plasma lipocalins alpha1-acid glycoprotein, apolipoprotein D, apolipoprotein M and complement protein C8gamma

Gender differences in apolipoprotein D expression during aging and in Alzheimer disease

Apolipoprotein D expression in retinoblastoma

Additive stimulatory action of glucocorticoids and androgens on basal and estrogen-repressed apolipoprotein-D messenger ribonucleic acid levels and secretion in human breast cancer cells

Plasma apolipoprotein concentrations in familial apolipoprotein A-I and A-II deficiency (Tangier disease)

Lipoprotein abnormalities associated with a familial deficiency of hepatic lipase

Factors affecting the lipid and apolipoprotein levels of cord sera

Apolipoproteins and lipoprotein families in familial lecithin: cholesterol acyltransferase deficiency

The serum apolipoprotein profile of patients with glucose-6-phosphatase deficiency

Apolipoprotein D gene polymorphism: a new genetic marker for type 2 diabetic subjects in Nauru and South India

Apolipoprotein-D polymorphism: a genetic marker for obesity and hyperinsulinemia

Molecular markers distinguish bone marrow mesenchymal stem cells from fibroblasts

Kruppel-like zinc finger protein Glis3 promotes osteoblast differentiation by regulating FGF18 expression

Increased bone formation in mice lacking apolipoprotein E

Embedding of bone samples in methylmethacrylate: an improved method suitable for bone histomorphometry, histochemistry, and immunohistochemistry

Serial passage of MC3T3-E1 cells down-regulates proliferation during osteogenesis in vitro

Pharmacological and biochemical evidence for the regulation of osteocalcin secretion by potassium channels in human osteoblast-like MG-63 cells

Changing RANKL/OPG mRNA expression in differentiating murine primary osteoblasts

Association of oxidative stress with postmenopausal osteoporosis and the effects of hydrogen peroxide on osteoclast formation in human bone marrow cell cultures

The invisible disease: making sense of an osteoporosis diagnosis in older age

The pathogenesis, diagnosis, investigation and management of osteoporosis

Looking back: developments in our understanding of the occurrence, aetiology and prognosis of osteoporosis over the last 50 years

Elevation of serum alkaline phosphatase (ALP) level in postmenopausal women is caused by high bone turnover

Increased bone turnover in late postmenopausal women is a major determinant of osteoporosis

Structure of the human apolipoprotein D gene promoter region

Apolipoprotein D gene induction by retinoic acid is concomitant with growth arrest and cell differentiation in human breast cancer cells

Neuroprotective effect of apolipoprotein D against human coronavirus OC43-induced encephalitis in mice

Bone marrow stress decreases osteogenic progenitors

Bioenergetics and mitochondrial transmembrane potential during differentiation of cultured osteoblasts

Apolipoprotein D takes center stage in the stress response of the aging and degenerative brain

Selective reduction of hydroperoxyeicosatetraenoic acids to their hydroxy derivatives by apolipoprotein D: implications for lipid antioxidant activity and Alzheimer's disease

Cytoplasmic superoxide causes bone fragility owing to low-turnover osteoporosis and impaired collagen cross-linking

Oxidative stress inhibits osteoblastic differentiation of bone cells by ERK and NF-kappaB

Basic biology of skeletal aging: role of stress response pathways

Characterization of nuclear factors modulating the apolipoprotein D promoter during growth arrest: implication of PARP-1, APEX-1 and ERK1/2 catalytic activities

Effects of leukotrienes on osteoblastic cell proliferation

We thank Denis Flipo for his help with the confocal microscopy analysis. Special thanks to Robert Moreau for investing time and energy in spite of his illness.

No conflicts of interest, financial or otherwise, are declared by the authors.