Deterioration of hematopoietic autophagy is linked to osteoporosis

Abstract

Hematopoietic disorders are known to increase the risk of complications such as osteoporosis. However, a direct link between hematopoietic cellular disorders and osteoporosis has been elusive. Here, we demonstrate that the deterioration of hematopoietic autophagy is coupled with osteoporosis in humans. With a conditional mouse model in which autophagy in the hematopoietic system is disrupted by deletion of the Atg7 gene, we show that incapacitating hematopoietic autophagy causes bone loss and perturbs osteocyte homeostasis. Induction of osteoporosis, either by ovariectomy, which blocks estrogen secretion, or by injection of ferric ammonium citrate to induce iron overload, causes dysfunction in the hematopoietic stem and progenitor cells (HSPCs) similar to that found in autophagy-defective mice. Transcriptomic analysis of HSPCs suggests promotion of iron activity and inhibition of osteocyte differentiation and calcium metabolism by hematopoietic autophagy defect, while proteomic profiling of bone tissue proteins indicates disturbance of the extracellular matrix pathway that includes collagen family members. Finally, screening for expression of selected genes and an immunohistological assay identifies severe impairments in H vessels in the bone tissue, which results in disconnection of osteocytes from hematopoietic cells in the autophagy-defective mice. We therefore propose that hematopoietic autophagy is required for the integrity of H vessels that bridge blood and bone cells and that its deterioration leads to osteoporosis.

Keywords: autophagy; hematopoietic system; osteoporosis.

Figure 1

Bone marrow hematopoietic autophagy is decreased in osteoporotic patients. (a) Statistical analysis of bone mineral density (BMD) and red blood cell (RBC) count in human osteoporotic and healthy populations. Female nonosteoporosis individuals, n = 1,766; female osteoporosis individuals, n = 127; male nonosteoporosis individuals, n = 3,051; and male osteoporosis individuals, n = 35. BMD was tested by dual‐energy X‐ray absorptiometry. RBC count was examined by routine hematologic method.*p < .05. (b) Expression of autophagy‐essential genes in the bone marrow hematopoietic and stem cells of osteoporotic patients and healthy individuals. Osteoporosis was identified by BMD using dual‐energy X‐ray absorptiometry. BMD T value ≤−2.5 is a criterion for osteoporosis. Human femoral bone marrows were collected in total hip replacement or total knee replacement surgery. Bone marrow hematopoietic stem and progenitor cells were isolated with magnetic‐activated cell sorting (MACS) against CD45 and CD34 antibodies after Ficoll‐gradient separation. mRNA levels of Atg7, Atg5, Atg12, LC3b, Lamp2a, and P62 genes were detected by real‐time quantitative PCR. Data = means ±SDs. *p < .05. n = 30/group. (c, d) Detection of autolysosome formation in healthy and osteoporotic populations. Autolysosome formation was measured by image flow cytometry for double staining of LC3 and lysosome with bone marrow CD45 or CD45CD34 cells prepared by fluorescence‐activated cell sorting (FACS). Autolysosome formation was represented by the colocalization of LC3 and lysosomal marker LAMP1. Left panel, representative flow image; right, results of flow image analysis. Data = means ± SDs. *p < .05. n = 30/group

Figure 2

Induced osteoporosis resembles the hematopoietic autophagy defect‐caused phenotype in bone loss and hematopoietic abnormality. (a, d) HSPC proportion was detected for mice in three groups by flow cytometry: wild‐type (Ctrl), ovariectomy (OVX), and iron accumulation (FAC). Generation of both mouse models started at 8 weeks of age. The OVX group and its control were sacrificed at 14 weeks of age, while FAC group and its control were sacrificed at 16 weeks of age for analysis. Bone marrow was isolated from femur and tibia for analysis. Left panel, representative images showing percentage of Lin^‐CD117⁺ Scal‐1⁺ in Lin⁻ cells; right panel, graph showing the significant difference between control and model groups. *p < .05. (b, e) Detection of apoptosis of HSPCs in the same three groups by flow cytometry. Left panel, representative images showing apoptosis in HSPCs; right panel, graph showing the significant difference between control and model groups. *p < .05. (c) The representative micro‐CT reconstructed three‐dimensional pictures of femur trabecular and cortical bones. Femora were collected from wild‐type and iron accumulation mice. (f) Functional colocalization of autophagosome marker LC3 and lysosome marker LAMP1 in the CD45‐positive hematopoietic cells sorted from the control and FAC mice, measured by image flow cytometry with fluorescent antibodies against LC3 or LAMP1. Left panel, representative images showing colocalization of LC3 and LAMP1; right panel, graph showing the significant difference between control and FAC mice. ***p < .001. n = 5. (g) P62 and LC3 protein expression in Lin⁻ cells of OVX mice. (h) Confocal microscopic detection of LC3 puncta by immunostaining of DAPI and LC3 in bone marrow hematopoietic stem cells (HSC) and mononuclear cells (MNC) of OVX mice. (i) Percentage of HSCs (CD48⁻ CD150⁺ LSK) in HPCs (LSK) was detected in OVX mice. Left panel, representative images showing percentage of CD48⁻ CD150⁺ in LSK compartment; right panel, graph showing the significant difference between control and OVX mice. **p < .005

Figure 3

Hematopoietic autophagy defect caused severe bone loss. (a) Confirmation of autophagy disruption in the Atg7^−/− mice. Bone marrow and bone tissue from 10‐week‐old mice were collected for Western blotting analysis of ATG7 and LC3. β‐Actin was used as a loading control. (b) Representative images of wild‐type and Atg7‐deleted mice at ages of 3 or 12 weeks. n = 3. (c) Micro‐CT analysis. Femora were collected from 8‐week‐old mice. Left, representative reconstructed three‐dimensional pictures of femur trabecular and cortical bones. Right, micro‐CT quantification of the trabecular index of femur samples. Measurement of distal femur spatial structure parameters include bone mineral density (BMD), trabecular space (Tb.Sp), trabecular number (TB.N), bone tissue fraction (BV/TV), trabecular thickness (Tb.Th), and cortical thickness. *p < .05. n = 3. (d) Scanning electron microscopy (SEM) analysis of trabecular microstructure of femur. Femora were collected from 8‐week‐old mice. n = 3. (e) Representative calcein double‐labeling images with quantification of mineral apposition rate (MAR). n = 3. (f) Immunofluorescence of 8‐week‐old mouse tibia metaphysis frozen section stained with phalloidin antibody (green) and DAPI (blue). Fluorescently tagged phalloidin staining was performed to measure the density of osteocyte cell projection formed during the process of embedding into bone matrix. Nuclei were visualized by DAPI staining. n = 3. (g) Three‐point bending test for bone biomechanical properties. Femora were collected from 8‐week‐old mice and were measured for load, stress, stiffness and strain. Upper panel, representative image; lower panel, bar graph. *p < .05. n = 9. Femora were collected from 8‐week‐old mice. *p < .05. n = 6. (h) HE and Masson staining of tibia paraffin section. Tibiae were collected from 8‐week‐old mice for paraffin section and immunohistochemistry. n = 6. (i) Alizarin red/Alcian blue staining of the whole‐body skeleton from neonatal mice. Mice were sacrificed at postnatal day 3. Skeleton was stained after soft tissue was removed. n = 3

Figure 4

Hematopoietic autophagy defect disturbs osteocyte homeostasis. (a) Immunofluorescence of 8‐week‐old mouse tibia frozen section stained with phalloidin antibody (green) and DAPI (blue). Fluorescently tagged phalloidin staining was performed to measure the density of osteocyte cell projections formed during the process of embedding into bone matrix. Nuclei were visualized by DAPI staining. n = 3. (b) Immunofluorescence of 8‐week‐old mouse tibia cortical bone frozen section stained with DAPI (blue), reflecting the density of osteocytes per bone area. n = 3. (c) Immunofluorescence of 8‐week‐old mouse tibia cortical bone frozen section stained with MitoTracker Deep Red (red) and DAPI (blue). Right, quantitative data for mitochondrial level measured from microscopic images. n = 3. (d) Flow cytometric detection of mitochondrial mass and ROS levels. Bone cells were prepared from femurs of 8‐week‐old mice by digesting with trypsin and collagenase I. Mitochondrial mass and ROS levels were measured with MitoTracker Deep Red and DCFH‐DA on a flow cytometer, respectively. For mitochondria staining, red represents mitochondria and blue represents DAPI. Merged cells in mitochondria staining were quantified. **p < .005, ***p < .001. n = 6. (e) Immunofluorescence of 8‐week‐old mouse tibia cortical bone frozen section stained with γ‐H2AX antibody (red) and DAPI (blue). Merged cells in γ‐H2AX staining were quantified (right). n = 3. (f) Measurement of expression levels for several genes critical in regulation of osteocyte homeostasis in bone tissue. Femora and tibiae were collected from 8‐week‐old control and Atg7‐deleted mice for total RNA extraction. The gene encoding proteins examined by real‐time quantitative PCR include SP7, RUNX2, BMP2, BMP6, CTSK, and TRAP5 in bone tissue. The primers used for amplification of these genes are given in the methods (Table S1). SP7 (a bone specific transcription factor) is essential for osteoblast differentiation and bone formation; RUNX2 (runt related transcription factor 2) is essential for osteoblast differentiation and skeletal morphogenesis; BMP2 (bone morphogenetic protein 2) and BMP6 (bone morphogenetic protein 6) both regulate bone development; BCTSK (cathepsin K) and TRAP5 (tartrate‐resistant acid phosphatase 5) are required in bone remodeling and resorption. mRNA levels are normalized to GAPDH. *p < .05, **p < .005. n = 3

Figure 5

Proteomic profiling of tibia bone of hematopoietic autophagy‐defective mice. (a) KEGG analysis of extracellular matrix (ECM) pathway. (b) STRING analysis of interaction network for downregulated proteins in ECM function. (c) STRING analysis of interaction network for downregulated collagen family members. (d) Heatmap analysis of the altered expression levels for proteins in the ECM pathway. (e) Collagen 1 staining of tibia section. The tibiae were collected from 8‐week‐old mice for immunohistochemistry. Collagen 1 was stained golden yellow

Figure 6

Examination of angiogenesis in bone tissues of the autophagy‐defective mice. (a) Expression of Vegf gene family members in Atg7^+/+ and Atg7^−/− bone tissue. The mRNA levels for analysis are normalized to the mRNA encoding GAPDH. **p < .005. (b) Identification of type H vessels. Type H vessels are a major connection between blood cells and bone formation. 8‐week‐old mouse tibia metaphysis frozen section was stained with CD31 antibody (green), EMCN antibody (red), and DAPI (blue). CD31 and EMCN merged colors represent type H vessels. Shown are the representative images by immunofluorescence from WT and Atg7‐deleted mice. n = 3. (c) A cartoon illustrating the functional connection between hematopoietic autophagy and prevention of osteoporosis. Sufficient hematopoietic autophagy maintains homeostasis in both the hematopoietic system and bone tissue. Deficient hematopoietic autophagy leads to not only abnormal hematopoiesis, but also concurrently impaired type H vessels, elevated mitochondrial mass, and oxidative stress that together disturb the ECM pathway. Ultimately, it leads to accelerated osteoporosis