Constitutive bone marrow adipocytes suppress local bone formation

Abstract

BM adipocytes (BMAd) are a unique cell population derived from BM mesenchymal progenitors and marrow adipogenic lineage precursors. Although they have long been considered to be a space filler within bone cavities, recent studies have revealed important physiological roles in hematopoiesis and bone metabolism. To date, the approaches used to study BMAd function have been confounded by contributions by nonmarrow adipocytes or by BM stromal cells. To address this gap in the field, we have developed a BMAd-specific Cre mouse model to deplete BMAds by expression of diphtheria toxin A (DTA) or by deletion of peroxisome proliferator-activated receptor gamma (Pparg). We found that DTA-induced loss of BMAds results in decreased hematopoietic stem and progenitor cell numbers and increased bone mass in BMAd-enriched locations, including the distal tibiae and caudal vertebrae. Elevated bone mass appears to be secondary to enhanced endosteal bone formation, suggesting a local effect caused by depletion of BMAd. Augmented bone formation with BMAd depletion protects mice from bone loss induced by caloric restriction or ovariectomy, and it facilitates the bone-healing process after fracture. Finally, ablation of Pparg also reduces BMAd numbers and largely recapitulates high-bone mass phenotypes observed with DTA-induced BMAd depletion.

Keywords: Adipose tissue; Bone Biology; Bone disease; Bone marrow.

Figure 1. DTA expression depletes BMAds.

Control (–) and BMAd-DTA (+) male mice at 20–24 weeks of age were sacrificed to validate the depletion of BMAds by DTA expression (n = 7 per group). Experiments were repeated more than 3 times. (A) Tibiae were decalcified with 14% EDTA for 3 weeks and stained with 1% osmium tetroxide for 48 hours. μCT scanning was performed, and 3D reconstituted images of tibiae are shown. (B) Decalcified bones were processed for paraffin embedding and sectioning. Paraffin slides were stained with H&E. Images were taken under 100× magnification. Scale bars: 200 μm. (C) Lipid staining within the BM was quantified by μCT scanning following osmium tetroxide staining. Proximal metaphysis is the region indicated by 1 to 2, and distal tibia is the region between 3 and 4, shown in A. Data are presented as mean ± SD. *P < 0.05 with a 2-tailed t test. (D) Fresh distal tibiae were collected and hammered into powder for bulk RNA purification. RNA-Seq and GSEA were performed. (E and F) BM mesenchymal stem cells were isolated from control and BMAd-DTA male mice at 16 weeks of age and were differentiated into adipocytes. Scale bar: 200 μm. After 3 weeks of differentiation, cells were collected for RNA purification, followed by qPCR. Fold changes of gene expression were normalized to the geomean of Hprt and Rpl32a. Data are presented as mean ± SD. *P < 0.05 with 2-way ANOVA with Šídák’s multiple-comparison test.

Figure 2. Depletion of BMAds increases cortical bone formation.

Control (–) and BMAd-DTA (+) male mice at 20–24 weeks of age received calcein injections 9 and 2 days before dissection. Tibiae were collected for μCT and dynamic histomorphometry (n = 8–10 per group). Experiments were repeated twice. (A) Representative images of distal tibiae from μCT scanning are shown. Scale bar: 500 μm. (B) Distal tibial cortical bone area fraction (BA/TA) and thickness (Th) were quantified for control and BMAd-DTA mice. (C and D) Calcein-labeled tibiae were cross-sectioned and imaged by fluorescence microscopy. Cortical bone interlabel widths (Ir.L.Wi) in endosteum (Ec.) and periosteum (Ps.) were quantified by BioQuant software. Data are presented as mean ± SD. *P < 0.05 with a 2-tailed t test (B and D). (E and F) Fresh distal tibial cortical bone was cross-sectioned and analyzed with Raman microscopy (n = 3–4). Representative spectra from distal tibiae are shown, and major peaks are labeled (E). Lipid/mineral ratio, mineral/matrix ratio, collagen crosslink (Xlinks), and bone crystallinity were calculated for endosteal (Endo-), mid-cortical (Mid-), and periosteal (Peri-) regions (F). Data are presented as mean ± SD. *P < 0.05 with 2-way ANOVA analyses followed by Šídák’s multiple-comparison test. (G and H) Total RNA was purified from distal tibiae and used for RNA-Seq analysis (n = 4 for control; n = 7 for BMAd-DTA). The upregulated gene set was analyzed by MetaScape to identify enriched terms and pathways (G). Z scores of genes related to ossification pathway are shown as heatmap (H).

Figure 3. Trabecular bone formation is enhanced in caudal vertebrae of BMAd-DTA mice.

Control (–) and BMAd-DTA (+) male mice at 20–24 weeks of age were sacrificed, and the fourth through sixth caudal vertebrae were collected (2 cohorts were included: one cohort with n = 11–14 per group was used for μCT analyses; the other cohort with n = 13 per group was split into histology and qPCR). (A and B) Caudal vertebrae were decalcified and used for paraffin sectioning. H&E-stained slides were scanned for an overview of the fifth caudal vertebra (A). Higher magnification images were taken at 100× (B). Scale bar: 200 μm. (C) RNA from the fourth through sixth caudal vertebrae was purified and used for qPCR to measure expression of adipogenic transcriptional factors. Relative gene expression is presented after normalization to the geomean of Hprt and Rpl32a. (D and E) Caudal vertebral trabecular bone parameters were assessed by CT. Scale bar: 200 μm. Tb., Trabecular bone; BV/TV, bone volume fraction; BMD, bone mineral density; Conn. Dens, connective density; N, number; Th, thickness; Sp, separation. (F and G) H&E-stained slides were used to count osteoblast number (N. Ob) and normalized to bone surface (BS). (H and I) Paraffin-sectioned slides were used for TRAP staining and osteoclast (Oc) number (N) and surface (S) quantification. (J) Osteogenic gene expression in caudal vertebrae were evaluated by qPCR and normalized to the geomean of Hprt and Rpl32a. Data are presented as mean ± SD. *P < 0.05 with a 2-tailed t test. Multiple unpaired t tests were performed in E and J, and P values were adjusted for multiple comparisons using 2-stage step-up (Benjamini, Krieger, and Yekutieli) with FDR method.

Figure 4. BMAd-DTA expression largely blocked caloric restriction–induced BMAT expansion.

Control and BMAd-DTA male mice at 24 weeks of age underwent 30% caloric restriction (CR) for 12 weeks (n = 6–10 per group). Tibiae were collected for BMAT quantification. – indicates ad libitum, + indicates CR. (A and B) Decalcified tibiae were stained with osmium tetroxide. 3D images (A) and quantitative data of BMAT (B) were collected from μCT analyses. (C and D) Calcified (C) or decalcified tibiae (D) were plastic- or paraffin-processed and sectioned for Goldner’s trichrome (proximal tibiae; C) or H&E staining (distal tibiae; D). Images were taken under 100× magnification. Scale bar: 200 μm. Goldner’s trichrome staining, red-hematopoietic cells; green-bone, circular void-BMAds. Data are presented as mean ± SD. *P < 0.05 with 2-way ANOVA analyses followed by Šídák’s multiple-comparison test.

Figure 5. Loss of BMAds protects caloric restricted mice from loss of cortical bone.

Control and BMAd-DTA male mice at 24 weeks of age underwent 30% caloric restriction (CR) for 12 weeks (n = 6–10 per group). Tibiae and caudal vertebrae were collected. – indicates ad libitum, + indicates CR. (A and B) Distal tibial cortical bone area fraction (Ct. BA/TA) and mineral density (Ct. BMD) were determined by μCT scanning. Scale bar: 500 μm. (C and D) Calcified distal tibiae were cross-sectioned for dynamic histomorphometry. Endosteal bone formation was measured. Ec, endocortical; Ir. L. Wi, interlabel width; dL. Pm, double-labeled perimeter; M. Pm, mineralizing perimeter. (E and F) Caudal vertebral trabecular bone variables were analyzed following μCT scanning. Scale bar: 200 μm. Tb., Trabecular bone; BV/TV, bone volume fraction; BMD, bone mineral density; Conn. Dens, connective density; N, number; Th, thickness; Sp, separation. Data are presented as mean ± SD. *P < 0.05 with 2-way ANOVA analyses followed by Šídák’s multiple-comparison test.

Figure 6. BMAd depletion protects mice from ovariectomy-induced cortical bone loss.

Control and BMAd-DTA female mice at 20 weeks of age underwent ovariectomy (OVX). Mice were euthanized 6 weeks after surgery (n = 4–7 per group). Tibiae and caudal vertebrae were collected. – indicates sham, + indicates OVX. (A) Decalcified tibiae were paraffin embedded, sectioned, and H&E stained. Representative images from proximal and distal tibiae were collected under 100× magnification. Scale bar: 200 μm. (B and C) Tibiae were used for μCT scanning. Cortical bone area (Ct. BA/TA) and mineral density (Ct. BMD) at midtibia shaft were quantified. Scale bar: 200 μm. (D and E) Caudal vertebrae were scanned by CT. Trabecular bone volume (Tb. BV/TV) and mineral density (Tb. BMD), as well as microstructure parameters, were measured. Scale bar: 200 μm. Tb. N, trabecular number; Tb. Sp, trabecular separation. Data are presented as mean ± SD. *P < 0.05 with 2-way ANOVA analyses followed by Šídák’s multiple-comparison test.

Figure 7. BMAd depletion promotes bone repair.

(A–H) Control (–) and BMAd-DTA (+) female mice at 24 weeks of age underwent surgery to fracture distal tibiae. Tibiae and serum were collected 10 or 20 days after surgery (n = 12–15 per group; mice were split into day 10 or 20 euthanization after surgery). (A) Distal tibial fracture was visualized by x-ray scanning during surgery. (B) Representative 3D reconstruction images of tibial callus after 10 or 20 days of healing. (C) Longitudinal and cross-sectional images of callus formation from μCT analyses are shown. (D) Circulating bone formation marker (P1NP) and resorption marker (CTX-1) were analyzed by ELISA. Data are expressed as mean ± SD. *P < 0.05 with 2-way ANOVA analyses followed by Šídák’s multiple-comparison test. (E–G) Safranin O fast green (SOFG) staining and μCT analysis using Dragonfly software were performed to analyze callus formation in tibiae collected from day 10 of fracture healing. Green, bone tissue; orange, cartilage; dark blue, nuclei. Callus sizes and cartilage area percentage were quantified by ImageJ (NIH) (F). (H–J) Tibiae collected at day 20 following fracture were used for callus analysis. (K–M) Control (–) and BMAd-DTA (+) male mice at 24 weeks old underwent distal tibial fracture and were euthanized at day 20 after procedure (n = 8–9 per group). Callus formation was analyzed by SOFG staining and μCT scanning. TV, total volume; BV, bone volume; BV/TV, bone volume fraction; BMD, bone mineral density; TMD, tissue mineral density. Data are presented as mean ± SD. *P < 0.05 with a 2-tailed t test. Multiple unpaired t tests were performed, and P values were adjusted for multiple comparisons using 2-stage step-up (Benjamini, Krieger, and Yekutieli) with FDR method. Scale bar: 1 mm.

Figure 8. BMAd-Pparg deficiency recapitulates bone phenotypes observed in BMAd-DTA mice.

Female mice with excision of exons 1 and 2 of Pparg in BMAds (BMAd-Pparg^–/–) and their littermate controls (BMAd-Pparg^+/+) were housed until 20 weeks of age (n = 5 per group). Tibiae and caudal vertebrae were collected. +/+ indicates control, –/– indicates BMAd-Pparg KO. (A and B) Tibiae were decalcified and stained with osmium tetroxide for 48 hours. μCT scanning was performed. Representative 3D reconstitution images of tibiae are shown (A). BMAT in proximal (rBMAT) and distal tibiae (cBMAT) were quantified and normalized by total volume (TV) (B). (C) Decalcified bones were processed for paraffin sectioning and H&E staining. Images were taken under 100× magnification. Scale bar: 200 μm. (D and E) Distal tibial cortical bone area (Ct. BA/TA) and thickness (Ct. Th) were quantified following μCT scanning. Scale bar: 500 μm. (F) Decalcified caudal vertebrae were paraffin embedded, sectioned, and H&E stained. Representative pictures were taken under 100× magnification. Scale bar: 200 μm. (G and H) Trabecular bone (Tb.) parameters in caudal vertebrae were measured by CT. Scale bar: 200 μm. BV/TV, bone volume fraction; BMD, bone mineral density. (I–L) Distal tibiae from male mice at 24 weeks of age were collected for mechanistic analyses (n = 5 per group). (I) Distal tibial total RNA was purified and used for qPCR to measure osteogenic markers. The expression of Sp7 and Bglap genes were normalized to the geomean of Hprt and Rpl32a. (J) Distal tibiae were cross-sectioned and scanned for calcein-labeled mineralizing bone surfaces. (K and L) Dynamic histomorphometry was performed in endosteal (K) and periosteal (L) surfaces. Endocortical double-labeled perimeter (Ec. dL. Pm) and mineralizing perimeter (Ec. M. Pm) were quantified by BioQuant software. These parameters were also quantified at periosteal (Ps.) mineralizing surfaces. Data are presented as mean ± SD. *P < 0.05 with a 2-tailed t test.