ApoE is a plasma protein that plays a major role in lipoprotein metabolism. Here we describe that ApoE expression is strongly induced on mineralization of primary osteoblast cultures. ApoE-deficient mice display an increased bone formation rate compared with wildtype controls, thereby showing that ApoE has a physiologic function in bone remodeling.
Introduction: Apolipoprotein E (ApoE) is a protein component of lipoproteins and facilitates their clearance from the circulation. This is confirmed by the phenotype of ApoE-deficient mice that have high plasma cholesterol levels and spontaneously develop atherosclerotic lesions. The bone phenotype of these mice has not been analyzed to date, although an association between certain ApoE alleles and BMD has been reported.
Materials and Methods: Primary osteoblasts were isolated from newborn mouse calvariae and mineralized ex vivo. A genome-wide expression analysis was performed during the course of differentiation using the Affymetrix gene chip system. Bones from ApoE-deficient mice and wildtype controls were analyzed using radiography, μCT imaging, and undecalcified histology. Cellular activities were assessed using dynamic histomorphometry and by measuring urinary collagen degradation products. Lipoprotein uptake assays were performed with125I-labeled triglyceride-rich lipoprotein-remnants (TRL-R) using primary osteoblasts from wildtype and ApoE-deficient mice. Serum concentrations of osteocalcin were determined by radioimmunoassay after hydroxyapatite chromatography.
Results:ApoE expression is strongly induced on mineralization of primary osteoblast cultures ex vivo. Mice lacking ApoE display a high bone mass phenotype that is caused by an increased bone formation rate, whereas bone resorption is not affected. This phenotype may be explained by a decreased uptake of triglyceride-rich lipoproteins by osteoblasts, resulting in elevated levels of undercarboxylated osteocalcin in the serum of ApoE-deficient mice.
Conclusion: The specific induction of ApoE gene expression during osteoblast differentiation along with the increased bone formation rate observed in ApoE-deficient mice shows that ApoE has a physiologic role as a regulator of osteoblast function.
BONE REMODELING IS an important physiologic process that is required to maintain a constant bone mass from the end of puberty until gonadal failure. This is achieved through a balanced activity of bone-resorbing osteoclasts and bone-forming osteoblasts. If bone resorption outweighs bone formation, net bone loss occurs, which can result in osteoporosis, one of the most common degenerative diseases.(1) For now, most of the therapeutic options to prevent osteoporosis are long-term treatments that target osteoclastic bone resorption. In contrast, there is still a need to identify therapeutic strategies leading to a local or systemic increase of bone formation and thereby to a reversal of aging-associated bone loss.
Bone formation is controlled on several levels.(2) First, there are many factors acting locally to regulate proliferation, differentiation, and activity of osteoblasts. Besides well-known effectors of bone formation such as transforming growth factor (TGF)-β and bone morphogenetic proteins (BMPs),(3, 4) another class of signaling molecules has entered the field of bone research recently. In fact, the discovery that activating or inactivating mutations of the human low-density lipoprotein receptor-related protein 5 (LRP5) gene result in high or low bone mass, respectively, suggested that extracellular ligands belonging to the Wnt family could play a critical role in bone remodeling.(5–7) Besides this local regulation, osteoblasts are also under endocrine control. Several hormones have been shown to affect bone formation in vivo, including parathyroid hormone (PTH), members of the calcitonin family, and steroids.(8–11) Just recently, another level of osteoblast regulation was discovered. In fact, the adipocyte-specific hormone leptin acts as a potent inhibitor of bone formation through binding to its receptor in the ventral hypothalamus.(12, 13) This indirect effect on bone remodeling is mediated through the sympathetic nervous system, which was confirmed by the finding that β-adrenergic receptor antagonists can increase bone mass in mice and humans.(13, 14) Additionally, the combined phenotypes of leptin signaling-deficient ob/ob mice showed that there is a connection between lipid and bone metabolism, as has already been suggested by others based on several clinical observations.(12, 15–17)
One approach to studying the underlying mechanisms of bone formation is the analysis of gene expression at certain stages of osteoblast differentiation. This should lead to the identification of important signaling pathways involved in osteoblast differentiation and also to the discovery of novel regulators of bone formation that were not previously described. One possibility to achieve this goal is a genome-wide expression analysis using the recently developed array technology.(18, 19) However, having identified a gene whose expression is regulated during osteoblast differentiation does not necessarily mean that this gene is physiologically relevant for bone formation. Therefore, one good approach to address this question is the analysis of mouse deficiency models that proved to be an excellent tool to study gene function in all aspects of vertebrate biology, including bone remodeling.(20)
In this study we have identified apolipoprotein E (ApoE) as one gene whose expression is strongly induced on mineralization of primary osteoblast cultures derived from newborn mouse calvariae. To determine the physiologic role of ApoE in bone remodeling, we have analyzed the skeletal phenotype of ApoE-deficient mice. Compared with wildtype controls, these mice display a high bone mass phenotype that is caused by an increased bone formation rate. This phenotype may be explained by the finding that ApoE-deficient osteoblasts show a decreased uptake of vitamin K-containing triglyceride-rich lipoproteins. This results in an incomplete γ-carboxylation of osteocalcin, an osteoblast-specific gene product with a physiologic role as an inhibitor of bone formation.(21)
MATERIALS AND METHODS
Cell culture and RNA isolation
Primary osteoblasts were obtained by sequential collagenase digestion of calvariae from 3-day-old mice as described.(12) Osteoblast differentiation was induced at 80% confluency in αMEM containing 10% FBS, 50 μg/ml ascorbic acid, and 10 mM β-glycerophosphate. Mineralization of the cultures, as determined by von Kossa staining, was initially observed at day 12 and was completed at day 20 (data not shown). Total RNA was extracted using the Trizol reagent (Invitrogen) immediately before differentiation (d0), as well as 5 and 25 days thereafter (d5, d25). This RNA was further purified using the RNeasy Midi Kit (Qiagen) according to the manufacturer's instructions.
Microarray expression analysis
For all experiments, Affymetrix Murine U74v2 GeneChips containing 36,000 genes and expressed sequence tags (ESTs) were used. The targets for Affymetrix DNA microarray analysis were prepared as described by the manufacturer. The amount of total RNA used for the cDNA synthesis was 10 μg for each reaction. GeneChip microarrays were hybridized with the targets for 16 h at 45°C, washed, and stained using the Affymetrix Fluidics Station according to the GeneChip Expression Analysis Technical Manual. Microarrays were scanned with the Hewlett-Packard-Agilent GeneChip scanner, and the signals were processed using the GeneChip expression analysis algorithm (Affymetrix). To compare samples and experiments, the trimmed mean signal of each array was scaled to a target intensity of 100. Absolute and comparison analyses were performed with Affymetrix MAS 5.0 and DMT software using default parameters. To assist in the identification of genes that were positively regulated in the time course, we selected genes that were increased or decreased at least 2-fold compared with the baseline (day 0).(22) Annotations were further analyzed with interactive query analysis at www.affymetrix.com.
RT-PCR expression analysis
To independently confirm the hybridization data, we performed an RT-PCR expression analysis. Total RNA was isolated from a different preparation of primary osteoblasts as well as from several mouse tissues using the Trizol reagent (Invitrogen). This RNA was reverse transcribed using the cDNA cycle kit (Invitrogen). The resulting cDNA was used for a PCR reaction using gene-specific primers for Bgp (5′-TCCAAGCAGGAGGGCAATAAG-3′ and 5′-GCGTTTGTAGGCGGTCTTCAAG-3′) ApoE (5′-TTTTGGTGACCGCATCCGAG-3′ and 5′-CAGGACAGGAGAAGGATACTCATTG-3′), and Gapdh (5′-GACATCAAGAAGGTGGTGAAGCAG-3′ and 5′-CTCCTGTTATTATGGGGGTCTGG-3′). PCR products were separated on a 1.5% agarose gel, visualized by ethidium bromide staining, and subsequently verified by automated sequencing.
Radiographic and μCT analysis
ApoE-deficient mice and corresponding wildtype controls were purchased from the Jackson Laboratory. The genetic background was C57Bl/6. At least six mice per group were injected with calcein (2 mg/kg) 9 and 2 days before death at the age of 3 and 8 months. After removal of the internal organs, the skeletons were fixed in 3.7% PBS-buffered formaldehyde for 18 h at 4°C and stored in 80% ethanol. Vertebral bodies L3–L6, femora, and tibias were dissected out for further analysis.
Whole skeletons were analyzed by contact radiography using a Faxitron X-ray cabinet (Faxitron X-ray, Wheeling, IL, USA). For 3D visualization, the lumbar vertebra L6 was scanned (40 kV/114 μA) in a μCT 40 (Scanco Medical, Bassersdorf, Switzerland) at a resolution of 12 μm. For the assessment of the cortical thickness, femora were scanned at the midshaft at a resolution of 10 μm. The raw data were manually segmented and analyzed with the μCT Evaluation Program V4.4A (Scanco Medical). For visualization, the segmented data were imported and displayed in μCT Ray V3.0 (Scanco Medical). Transversal femoral thickness was measured with the Distance3D tool of the μCT Evaluation Program V4.4A.
Undecalcified histology and histomorphometric analysis
The lumbar vertebral bodies (L3–L5) and one tibia of each mouse were dehydrated in ascending alcohol concentrations and embedded in methylmethacrylate as described previously.(23) Sections (5 μm) were cut in the sagittal plane on a Microtec rotation microtome (Techno-Med, Munich, Germany). Sections were stained by toluidine blue, van Gieson/von Kossa, and Giemsa staining procedures as described.(23) Nonstained sections (12 μm) were used to determine the bone formation rate.
Parameters of static and dynamic histomorphometry were quantified on toluidine blue-stained undecalcified proximal tibia and lumbar vertebral sections (5 μm). Analysis of bone volume, trabecular number, trabecular thickness, and trabecular spacing, and the determination of osteoblast, osteocyte, and osteoclast numbers and surface indices were carried out according to standardized protocols using the OsteoMeasure histomorphometry system (Osteometrics, Atlanta, GA, USA).(24) Fluorochrome measurements for the determination of the bone formation rate were performed on two nonconsecutive 12-μm sections for each animal.
Both femora from each mouse were transferred to isotonic saline and stored at 4°C for 12 h before testing. Three-point bending assays were performed as previously described using a Z2.5/TN1S-device (Zwick, Ulm, Germany).(23) In brief, the ends of the femora were supported on two fulcra separated by 5 mm. With the posterior aspect of the femur resting on the fulcra, a load was applied from above to the anterior midshaft midway between the two fulcra at a constant speed of 10 mm/minute to failure. A chart recorder was used to generate a force-deformation curve. The ultimate force (maximum load) and the ultimate deformation (maximum displacement) were determined directly from the curve. The stiffness was assessed as the slope of the force-deformation curve through its linear region. Experiments were performed in a blinded fashion.
Biochemical assays for bone resorption parameters
To visualize functional osteoclasts on the bone surface, TRACP activity assays were performed on decalcified bone sections. Sections were preincubated for 1 h in 10 mM sodium tartrate dissolved in 40 mM acetate buffer (pH 5). The activity staining was performed in the same buffer including 0.1 mg/ml naphtol AS-MX phosphate (N-5000; Sigma) and 0.6 mg/ml Fast Red Violet LB salt (F-3881; Sigma). To quantify osteoclastic bone resorption, we measured the urinary excretion of deoxypyridinoline (DPD) cross-links with the Pyrilinks-D ELISA (Metra Biosystems, Mountain View, CA, USA). Values are expressed relative to creatinine concentrations as determined by a standardized colorimetric assay using alkaline picrate (8009; Metra Biosystems).
Triglyceride-rich lipoproteins (TRLs) were obtained by density ultracentrifugation (density ≤ 1.006 g/ml) from plasma of a nonfasted patient with an apoC-II deficiency and hydrolyzed in vitro to obtain TRL remnants (TRL-Rs). Apoproteins were labeled with125I by the iodine monochloride method as described previously.(25) For the radioactive TRL-R uptake experiments, osteoblasts from wildtype and ApoE-deficient mice were differentiated in 6-well plates for 10 or 20 days. The experiments were performed in DMEM containing 5% bovine serum albumin (BSA; fraction V; Sigma) and 0.02 M HEPES (pH 7.4).125I-TRL-Rs were added in a concentration of 1 μg ligand protein/ml incubation media. Cellular uptake was allowed for 60 minutes at 37°C, followed by a wash with PBS containing 2 mg/ml BSA. Surface-bound lipoproteins were released with heparin (770 U/ml in PBS). Finally, cells were dissolved in 0.1 M NaOH, and radioactivity and protein content of the lysates were determined. Uptake was calculated as nanograms ligand protein per milligram cell protein.
Measurement of serum osteocalcin
Serum from wildtype and ApoE-deficient mice (250 μl each) was applied onto a Macro-Prep ceramic hydroxyapatite column connected to a DuoFlow chromatography system (BioRad Laboratories). The column was equilibrated and washed with buffer A (10 mM sodium phosphate, pH 6.8). Bound proteins were eluted with a 5-ml gradient from 0–100% buffer B (500 mM sodium phosphate, pH 6.8). Fractions of 0.75 ml were analyzed for the presence of osteocalcin using a radioimmunoassay (50-1300; Immutopics).
The statistical significance of differences between control and experimental groups were determined by double-sided unpaired Student's t-test (Excel; Microsoft, Redmont, WA, USA). p < 0.05 was chosen to be statistically significant, and p < 0.01 was chosen to be highly statistically significant. Cell culture experiments were performed at least in triplicate and repeated twice. For static and dynamic histomorphometry, five animals per group and age of wildtype and ApoE-deficient mice were studied.
Induction of ApoE expression in mineralizing osteoblast cultures
In an attempt to identify genes whose expression is induced during osteoblast mineralization, we isolated total RNA from primary mouse calvarial osteoblasts at days 0, 5, and 25 of differentiation. This RNA was used to generate cRNA probes that were hybridized with the Affymetrix Murine U74v2 GeneChips. Using the Affymetrix MAS 5.0 and DMT software, we first analyzed the expression pattern of known osteoblast differentiation markers. In fact, we found that the expression of Runx2, a transcription factor required for osteoblast differentiation, was strongly induced during the first 5 days of osteoblast differentiation (Fig. 1A). The same was the case for α1(I)-collagen, the gene encoding one subunit of the most abundant protein of the bone matrix. During the later stage of osteoblast differentiation, when mineralization of the extracellular matrix occurs, we found a strong induction of osteopontin (Opn), bone sialoprotein (Bsp), and osteocalcin (Bgp) expression, thereby confirming the observations made by several other investigators.(26) We did not find any expression of chondrocyte or osteoclast markers, thus verifying the purity of the cultures and the specificity of the differentiation process.
We next screened for additional genes, whose expression is induced during osteoblast differentiation. We found that the expression of ApoE is much higher in mineralized (d25) compared with nonmineralized osteoblast cultures (d5; Fig. 1B). The importance of this unexpected finding is corroborated by two further arguments. First, the 22-fold upregulation of ApoE expression during osteoblast mineralization is almost as strong as the one observed for Bgp (28-fold). Second, ApoE is the only apolipoprotein whose expression is induced during osteoblast differentiation, raising the possibility that it has a specific function in bone formation.
Given the possible implications of this novel finding, we first confirmed our expression results using a different strategy. This was achieved by RT-PCR expression analysis using gene-specific primers for Bgp, ApoE, and Gapdh, the latter one as a control for cDNA quality. Whereas Bgp expression was found to be bone-specific, we observed that ApoE is expressed in several tissues obtained from adult mice (Fig. 1C). Most importantly however, we could confirm the strong expression of ApoE in mineralized primary osteoblast cultures, whereas nonmineralized cultures did not express detectable amounts. Taken together, these data show that ApoE, unlike other apolipoproteins, is specifically expressed in mineralized primary osteoblast cultures, thereby suggesting a potential role in the regulation of bone formation.
High bone mass in mice lacking ApoE
To analyze this possibility, we studied the skeletal phenotype of mice lacking ApoE. Contact X-rays from female wildtype and ApoE-deficient mice (n = 5) at the ages of 3 and 8 months revealed no gross abnormalities of skeletal development and growth (data not shown). However, on higher magnification, we observed an increased BMD in the vertebral bodies from ApoE−/− mice compared with wildtype littermates (Fig. 2A). This result was confirmed by 3D μCT imaging, showing an increased number of trabeculae in the vertebral bodies of ApoE−/− mice (Fig. 2B). We next analyzed von Kossa-stained undecalcified sections of vertebrae and tibias from wildtype and ApoE−/− mice. Again, we observed an increased trabecular bone volume in ApoE−/− mice compared with wildtype controls (Fig. 2C). Bone volume per tissue volume (BV/TV) was significantly higher in the vertebra of ApoE−/− mice at both ages and in the tibia of ApoE−/− mice at the age of 8 months.
To further quantify this high bone mass phenotype of the ApoE−/− mice, we performed static histomorphometry at the age of 8 months. The increase of the trabecular bone volume in ApoE−/− mice compared with wildtype controls was found to be caused by a higher number of bone trabeculae, whereas their thickness was not significantly altered (Fig. 3A). We also analyzed the cortical bone area by cross-sectional CT scans of the femora. Thereby we observed that the cortical thickness was significantly increased in ApoE−/− mice compared with wildtype controls (n = 10). However, when we analyzed the biomechanical stability of the same femora in three-point bending assays, this difference did not reach significance at the functional level (Fig. 3B). Taken together, the full analysis of the skeletal phenotype of ApoE-deficient mice by radiographic, μCT, and histomorphometric analysis showed that ApoE has a physiologic function as a regulator of bone remodeling. The underlying mechanisms explaining this novel function were subsequently analyzed by further experiments.
Increased bone formation in ApoE-deficient mice
We first asked the question of whether the high bone mass phenotype of the ApoE-deficient mice is caused by a decreased bone resorption or by an increased bone formation. Given the fact that we found a strong expression of ApoE in primary osteoblast cultures, we determined the number and functional activity of bone-forming osteoblasts using standard histomorphometric procedures.(24) Whereas the osteoblast numbers were normal in ApoE−/− mice, we observed a significant increase in the rate of bone formation, as determined by measuring calcein-labeled bone surfaces and the distance between the two calcein labeling fronts (Fig. 4A).
We next analyzed the number and functional activity of bone-resorbing osteoclasts. Standard histomorphometry revealed that the osteoclast number was the same in wildtype and ApoE−/− mice (Fig. 4B). By measuring the urinary excretion of deoxypyridinoline cross-links and by performing TRACP activity assays, we could further show that the resorptional activity of osteoclasts is not affected in ApoE−/− mice. Taken together, these data show that the high bone mass phenotype of the ApoE−/− mice is caused by an increased bone formation rate, thereby confirming our hypothesis raised by the microarray expression analysis.
Decreased lipoprotein uptake and undercarboxylation of osteocalcin in ApoE-deficient mice
One possibility to explain these unexpected findings comes from observations described by others.(27, 28) Niemeier et al.(27) show that ApoE stimulates the uptake of vitamin K-containing chylomicron remnants by osteoblasts in vitro. A similar finding has also been described in a study by Newman et al.,(28) which further showed that vitamin K uptake by osteoblasts is reduced by antibodies against ApoE. Vitamin K is required as a co-factor for the γ-carboxylation of osteocalcin, whose physiologic function as an inhibitor of bone formation was revealed through the analysis of osteocalcin-deficient mice.(21) Because the bone phenotype of these mice is very similar to the one described here, we analyzed the possibility that the high bone mass phenotype of the ApoE-deficient mice could be caused by alterations of vitamin K uptake and osteocalcin carboxylation.
Vitamin K is almost exclusively transported by TRL-Rs. Therefore, we first analyzed the uptake of125I-labeled TRL-R particles by wildtype and ApoE-deficient primary osteoblasts ex vivo. Whereas we observed a significant increase in TRL-R uptake in the ApoE-producing wildtype osteoblasts from days 10 to 20, TRL-R uptake by ApoE-deficient osteoblasts remained at base level (Fig. 5A). These findings suggest that the endogenous production of ApoE in mineralized osteoblast cultures (day 20) is required for a maximal uptake of vitamin K-containing TRL-R particles.
To analyze whether this difference observed ex vivo leads to an alteration of osteocalcin γ-carboxylation in vivo, we next determined the amount of osteocalcin in the serum of wildtype and ApoE-deficient mice. Because there is no antibody-based assay available that allows the discrimination between carboxylated and noncarboxylated osteocalcin in mice, we separated these two variants based on their binding to hydroxyapatite, according to what has been described by others.(29) To achieve full resolution of both variants, we applied serum from wildtype and ApoE−/− mice onto a hydroxyapatite chromatography column, eluted the bound proteins with a linear phosphate gradient, and analyzed the obtained fractions for the presence of osteocalcin using a radioimmunoassay. Thereby we found that the detectable osteocalcin in wildtype serum eluted from the hydroxyapatite column in two fractions, the majority (about 80%) having a higher affinity to hydroxyapatite and most likely resembling fully carboxylated osteocalcin. When we performed the same experiment with serum from ApoE-deficient mice, we observed that the pattern of elution was completely reversed (Fig. 5B). The finding that the majority of osteocalcin detectable in the serum of ApoE−/− mice has a lower affinity to hydroxyapatite indicates that osteocalcin is indeed undercarboxylated in the absence of ApoE (Fig. 5C). Thus, the high bone mass phenotype of the ApoE-deficient mice is possibly explained by a decrease of functional osteocalcin and thereby a reproduction of the phenotype of osteocalcin-deficient mice.(21)
ApoE is a major protein component of TRLs and is required for their plasma clearance by binding to lipoprotein receptors.(30) The importance of ApoE in lipoprotein metabolism is underscored by several genetic findings. First, human ApoE is polymorphic with three common isoforms, designated apoE2, apoE3, and apoE4.(31) Among these isoforms, apoE2 has a reduced affinity for the low-density lipoprotein receptor and is associated with type III hyperlipoproteinemia.(32) Second, ApoE-Leiden and other rare variants of ApoE are associated with a dominant inheritance of type III hyperlipoproteinemia.(33, 34) The complete absence of the human ApoE gene also leads to hyperlipidemia and premature cardiovascular diseases.(35) Third, two mouse models lacking ApoE have been shown to display severe hypercholesterolemia and spontaneously develop atherosclerotic lesions.(36, 37)
Besides this overwhelming genetic evidence showing an important contribution of ApoE to lipid metabolism and atherogenesis, there is also evidence for an involvement of ApoE in neurological disorders, as the apoE4 allele was found to be associated with late-onset Alzheimer disease.(38) Interestingly, the same allele has been reported to be associated with reduced BMD and increased fracture risk by several investigators.(39–41) Given the potential impact of these findings, it is surprising that there are still no published results on the analysis of a skeletal phenotype in the absence of ApoE, especially because other studies could not confirm an association of the human apoE4 allele with BMD.(42–44)
Based on our initial observation that ApoE expression is specifically induced during late osteoblast differentiation, we embarked on a full characterization of the skeletal phenotype of ApoE-deficient mice. This led us to the demonstration that the complete lack of ApoE in mice leads to a high bone mass phenotype that is caused by an increased bone formation rate. In an attempt to understand the underlying mechanisms explaining these observations, we further performed TRL-R uptake assays using primary osteoblasts from wildtype and ApoE−/− mice and found a decreased TRL-R uptake by ApoE-deficient osteoblasts ex vivo. Because vitamin K is almost exclusively transported by TRL-R particles, we reasoned that the γ-carboxylation of osteocalcin could be altered in ApoE-deficient mice, which was confirmed by radioimmunoassays after serum fractionation by hydroxyapatite chromatography. The fact that osteocalcin is undercarboxylated in ApoE-deficient mice provides one probable explanation for their high bone mass phenotype, because the lack of functional osteocalcin should reproduce the phenotype of osteocalcin-deficient mice.(21) However, we can not rule out the existence of other explanations for the high bone mass phenotype of the ApoE-deficient mice at the moment, and future experiments are required to analyze this possibility.
Taken together, our results are in full agreement with several observations made by others. First, Bachner et al.(45) have previously described an induction of ApoE expression in BMP-2-treated mesenchymal progenitors as well as in mineralized primary osteoblast cultures. Second, Newman et al.(28) have described an influence of ApoE on the uptake of lipoprotein-borne vitamin K by osteoblasts. Similar findings are provided by Niemeier et al.,(27) describing that exogenously added ApoE enhances the uptake of vitamin K-containing chylomicron remnants by human osteoblasts. The study by Niemeier et al.(27) further shows that this uptake is mediated by lipoprotein receptors, especially LRP1, that was also found to be expressed at all stages of osteoblast differentiation using our microarray analysis (data not shown). Therefore, the decreased uptake of TRL-R particles that we observed in ApoE-deficient osteoblasts provides additional evidence supporting the hypothesis that the endogenous production of ApoE expression by mineralized osteoblasts is required for the sufficient uptake of lipoprotein-borne vitamin K.(27) Such a secretion-recapture mechanism has already been described for the hepatic uptake of lipoproteins and for the uptake of lipids from degenerating axons by macrophages.(46, 47)
The best argument, however, for an impaired osteoblastic vitamin K uptake in the absence of ApoE comes from the finding that serum osteocalcin is undercarboxylated in ApoE-deficient mice, because this specific difference is found in vivo and can not be explained by overlapping effects of the culture conditions. Furthermore, this finding provides a possible explanation for the high bone mass phenotype of ApoE-deficient mice that shares striking similarities with the phenotype of osteocalcin-deficient mice.(21) In both cases, the trabecular bone volume increases with age, which is caused by an elevation of osteoblast functional activity, whereas bone resorption is not affected. The phenotype of the osteocalcin-deficient mice seems to be more severe, especially in terms of cortical bone remodeling, where the increased cortical thickness leads to an improvement of biomechanical stability, which we did not observe in the ApoE-deficient mice. This difference of severity, however, can be explained by the fact that the γ-carboxylation of osteocalcin, albeit strongly decreased, is not fully abolished in ApoE-deficient mice. Although this underlying mechanism could indeed explain why the expression of osteocalcin and ApoE are co-regulated during osteoblast differentiation, it needs to be supported by future experiments. This is especially important because there are conflicting results concerning the association of the human apoE4 allele with low BMD, but also concerning the impact of vitamin K on bone remodeling.
In the case of the human apoE4 allele, the number of studies showing an association with low BMD is well balanced with the number of studies that do not confirm these observations.(39–44) Nevertheless, because none of these reports showed an association with increased BMD, these findings seem to be in contradiction with the high bone mass phenotype of the ApoE-deficient mice described here. This can be explained by the fact that the human ApoE4 isoform has the highest affinity to lipoprotein receptors and was shown to be the most effective isoform concerning the stimulation of vitamin K uptake by osteoblasts in vitro.(28) Thus, it is reasonable to speculate that subjects with the apoE4 allele have an unphysiologically high TRL-R uptake by osteoblasts, thereby possibly leading to a phenotype that is opposite to the one observed in the ApoE-deficient mice.
Concerning the effects of vitamin K on bone metabolism, there are also conflicting results reported in the literature. Whereas some investigators did show that the use of warfarin, an antagonist of vitamin K, is associated with low BMD in humans, other studies did not confirm these observations.(48–51) Animal experiments could not clarify this issue either, especially because warfarin treatment also results in extensive arterial calcification, because it impairs the γ-carboxylation of matrix GLA protein (Mgp), an inhibitor of arterial calcification produced by vascular smooth muscle cells.(52, 53) The fact that Mgp-deficient mice also display an osteopenic phenotype, although Mgp is not expressed by bone cells, suggests that the arterial calcification induced by warfarin can lead to a reduction of bone formation in an osteocalcin-independent manner.(53, 54) Therefore, we believe that the high bone mass phenotype of the osteocalcin-deficient mice provides more valuable information concerning the specific function of osteocalcin as an inhibitor of bone formation in vivo, although the question of whether the carboxylated or the noncarboxylated variant of osteocalcin is important for this function could not be addressed by the gene deletion experiment.(21)
Although the hypotheses raised by this study need be proven by further experiments, the high bone mass phenotype observed in the ApoE-deficient mice is important for several reasons. First, it shows a novel function of ApoE as a physiologic regulator of bone remodeling. Second, together with the results described in Niemeier et al.,(27) it expands our knowledge about the uptake of TRLs and vitamin K by bone-forming osteoblasts. In fact, the expression of functional lipoprotein receptors by human osteoblasts described here raises the possibility that they play predominant roles as regulators of bone formation. Third, the undercarboxylation of osteocalcin observed in the ApoE-deficient mice may reveal the importance of osteocalcin as an inhibitor of bone formation, although further experiments are now required to examine whether the high bone mass phenotype of the ApoE-deficient mice is directly caused by alterations of osteocalcin carboxylation.
This work was supported by DFG Grant AM 103/8-3 (MA).