Autophagy in Osx1-Cre-targeted cells is essential for development, growth, and maintenance of bone

Abstract

Autophagy is a cellular recycling pathway in which cytoplasmic components are delivered to lysosomes for degradation. Changes in autophagy levels have been implicated in various skeletal pathologies including osteoporosis. Elimination of autophagy at different stages of the osteoblast lineage reduces bone formation and bone mass. However, whether autophagy plays a role specifically during development, growth, or maintenance of bone remains unclear. To start addressing this question, we eliminated autophagy in the entire osteoblast lineage prenatally or at weaning and subjected the mice to skeletal phenotyping at 4.5 and 10 months of age. We found that regardless of when autophagy was eliminated from the osteoblast lineage, autophagy deficiency reduced bone mineral density (BMD), cortical thickness, and cancellous bone volume in the femur and spine. Serial BMD analysis revealed that autophagy-deficient mice had consistently lower BMDs from 3 to 9 months of age, and the BMD difference between genotypes became progressively greater in the spine. The reduction in vertebral cancellous bone volume of autophagy-deficient mice was associated with reduced bone formation. To assess autophagy-induced changes at the cellular and molecular level, we performed single-cell RNA-sequencing (scRNA-seq) analysis of periosteal mesenchymal cells and compared autophagy deficiency-induced changes in periosteal and endosteal cell preparations. This analysis revealed that autophagy deficiency disrupts proteostasis, causes mitochondrial dysfunction, induces senescence, and increases stress response pathways like TNF and TGF-β. Overall, we conclude that autophagy is important for skeletal growth and maintenance of bone, and we identify potential cellular populations and cellular processes via which autophagy support bone formation.

Highlights

• •Osteoblast lineage autophagy supports prenatal and postnatal bone formation.
• •Osteoblast autophagy is essential for skeletal development, growth, and maintenance.
• •Autophagy maintains proteostasis, mitochondrial function, and response to stress.
• •Autophagy supports osteoblast function and may promote their differentiation.

1. Introduction

Macroautophagy, hereafter referred to as autophagy, is a catabolic process in which cellular components are encapsulated by autophagosomes and delivered to lysosomes for degradation and recycling (Feng et al., 2014). Autophagy is important for recycling macromolecules in response to high energy demands as well as clearing cytosolic components such as damaged organelles and misfolded proteins. As such, autophagy is critically important for highly secretory cells such as osteoblasts.

Several studies have examined the role of autophagy in osteoblast lineage cells by deleting genes essential for autophagy, namely Atg7 (autophagy-related 7), Atg5 (autophagy-related 5), and Fip200 (FAK family kinase-interacting protein of 200 kDa) (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016). These studies have shown that autophagy inhibition in the osteoblast lineage reduces osteoblast number and bone formation, leading to low bone mass. Most strikingly, deletion of Atg7 using the Osx1-Cre driver, which eliminates autophagy from the entire osteoblast lineage (osteoblast progenitors, osteoblasts, and osteocytes), causes the most prominent reduction in osteoblast number, bone formation, and bone mass, leading to a 50% fracture rate (Piemontese et al., 2016). Together, these studies establish that autophagy supports bone formation. However, the underlying cellular and molecular mechanisms remain unclear.

In recent work, we showed that autophagy stimulation in the entire osteoblast lineage starting at 3 weeks of age elevates bone formation in cancellous and cortical bone compartments, leading to increased bone mass and strength (James et al., 2025). Single cell RNA-sequencing (scRNA-seq) of periosteal cells showed that autophagy stimulation impacts potential osteoblast progenitor populations, namely Osteo-X and Fibro-2 cells, as well as osteoblasts. Together these results suggest that autophagy in the osteoblast lineage could be modulating bone formation post-developmentally by affecting osteoblast production and function. To test this idea, herein we eliminated autophagy from the entire osteoblast lineage postnatally (starting at 3-weeks of age), and examined the skeletal, cellular, and molecular changes.

2. Materials and methods

2.1. Animal experiments

To generate mice without autophagy in the entire osteoblast lineage, we performed two-step breeding. Specifically, we crossed mice hemizygous for the Osx1-Cre transgene (Rodda and McMahon, 2006) with mice heterozygous for the Atg7-flox allele (Komatsu et al., 2005). Offspring which were heterozygous for the Atg7-flox allele with and without the Osx1-Cre transgene were intercrossed to generate the following experimental mice: wild type mice, mice homozygous for the Atg7-flox allele (Atg7^fl/fl), mice hemizygous for the Osx1-Cre allele, and mice that were both hemizygous for Osx1-Cre allele and homozygous for the Atg7-flox allele (Osx1-Cre;Atg7^fl/fl). The mice were genotyped as previously described (Komatsu et al., 2005; Rodda and McMahon, 2006). Mice expressing the Osx1-Cre transgene are smaller and have low bone mass when compared to the mice lacking the transgene (Davey et al., 2012). Therefore, to account for any effects of the Osx1-Cre transgene on the skeleton, all experiments were conducted using Osx1-Cre-expressing mice as controls. Mice were maintained on either a standard chow diet or a doxycycline (dox)-containing diet (Bio-Serv, S3888) to allow temporal control of Osx1-Cre expression. Of note, mice whose Cre expression was delayed by dox feeding displayed no adverse effects of the Osx1-Cre transgene (Fig. 1). All mice were provided food and water ad libitum and were maintained on a 12-h light/dark cycle. All animal studies were carried out with approval from and following the policies of the Institutional Animal Care and Use Committee of the UAMS. The studies described here were performed and reported in accordance with ARRIVE guidelines. Each figure legend indicates the number, sex, and age of the mice used for each experiment.

Fig. 1.

Autophagy in the osteoblast lineage contributes to skeletal development pre- and post-3 weeks of age. (A) Diagram depicting the timeline of Osx1-Cre expression in Osx1-Cre;Atg7^fl/fl mice and their littermate controls. Group A indicates mice that were never fed a doxycycline diet and therefore had prenatal Cre expression; group B indicates mice that were kept on a doxycycline-containing diet through 3 weeks of age, thereby preventing Cre expression and autophagy elimination until 3 weeks of age. (B–E) Body weight (B) and global (total body) (C), vertebral (D), and femoral (E) DXA BMD measurements were collected from 4.5-month-old female Control (WT [green dots] and Atg7^fl/fl mice [gray dots]), Osx1-Cre, and Osx1-Cre;Atg7^fl/fl mice. n = 4–13 mice/group. Bars indicate mean ± standard deviation (SD). # indicates p < 0.05 for Control vs. Osx1-Cre. † indicates p < 0.05 for Control vs. Osx1-Cre;Atg7^fl/fl. * indicates p < 0.05 for Osx1-Cre vs. Osx1-Cre;Atg7^fl/fl. p values were calculated by one-way ANOVA followed by Tukey's post-hoc analysis.

2.2. Skeletal phenotyping

Bone mineral density (BMD) of the total body, spine, and femur was measured with a PIXImus Mouse Densitometer (GE Lunar Corp.) as previously described (Onal et al., 2015). Total body BMD analysis excluded the head and shoulder blades. Spinal BMD was measured in lumbar vertebrae 2–6, and femoral BMD was measured in the right femur.

For μCT analysis, lumbar vertebrae 3 (group A) and 4 (group B) and femurs were dissected, cleaned of soft tissue, fixed in 10% Millonig's formalin for 24 h, dehydrated using a series of serial dilutions of ethanol into 100% ethanol, and stored at 4 °C. Medium-resolution scans (12 μm isotropic voxel size) of vertebrae (threshold = 220 mg/cm³), 4.5-month group A femurs and 10-month femurs (threshold = 260 mg/cm³) were obtained using a model μCT40 (Scanco Biomedical) as previously described (Laster et al., 2023). Medium resolution scans (10 μm isotropic voxel size, E = 55 kVp, I = 72 μA, integration time = 1000 ms, threshold = 260 mg/cm3) of 4.5-month group B femurs were obtained using a model μCT45 (Scanco Medical) and analyzed using Scanco Eval Program Version 1.9.99.99. Fixed global thresholds were applied uniformly to all samples within each analysis. Scans were integrated into 3D voxel images (1024 × 1024 pixel matrices for each individual planar stack). A Gaussian filter (sigma = 0.8, support = 1) was applied to all analyzed scans. Cortical thickness of the femur midshaft was measured by drawing contours on the 20 slices flanking the midshaft. Cortical thickness of the femur distal metaphysis was measured by drawing contours to exclude trabecular bone on the 100 slices proximal of the distal growth plate, ending 10 slices away from the growth plate. Vertebral cortical thickness and cancellous analyses were performed by drawing contours every 10 slices between the two growth plates of the vertebrae, beginning and ending 10 slices away from the end of each growth plate. The vertebral cortical bone thickness was determined on the ventral cortical wall using contours of cross-sectional images, drawn to exclude trabecular bone. For Fig. 2, some of the vertebrae in group A were cracked during processing, so to avoid confounding the results, contours were only drawn on half of the vertebrae for all group A cancellous measurements.

Fig. 2.

Autophagy is required for postnatal maintenance of cortical and cancellous bone. μCT analysis of femurs and vertebrae from 4.5-month-old female Osx1-Cre; Atg7^fl/fl mice and their Osx1-Cre-harboring littermates with and without Cre expression through 3 weeks of age (group A and group B, respectively). (A) Femoral cortical thickness (Ct.Th.) measured at the femur midshaft. (B) Representative femur μCT image indicating regions in which Ct.Th. was measured. (C) Femoral Ct.Th. measured at the distal metaphysis. (D–E) Vertebral Ct.Th. (D) and cancellous bone volume/tissue volume (BV/TV) (E) measured in lumbar vertebrae. (F) Representative vertebral μCT images. White scale bars indicate 1 mm. Yellow outlines depict cortical region of interest (ROI). Red outlines depict cancellous ROI. n = 5–9 mice/group. Bars indicate mean ± SD. Indicated p values were calculated by unpaired t-test for equal or unequal variances.

2.3. Histology

Mice were injected with 20 mg/kg of calcein (Sigma Aldrich, C0875) and alizarin complexone (Sigma Aldrich, A3882), 7 days and 3 days prior to euthanization, respectively. L1–3 vertebrae were fixed in 10% Millonig's formalin for 24 h and then dehydrated using a series of serial dilutions of ethanol into 100% ethanol. Next, the bones were embedded in methyl methacrylate, and longitudinal sections were obtained. Dynamic histomorphometric measurements were performed as previously described (Dempster et al., 2013; Piemontese et al., 2015).

2.4. RNA isolation and gene expression analysis

Lumbar vertebrae 5 were dissected, cleaned of soft tissue, snap-frozen, and stored at −80 °C. To isolate RNA, the bones were homogenized in TRIzol Reagent (Life Technologies, 15596018), and RNA was isolated with the RNeasy Plus Mini Kit (QIAGEN, 74136). A total of 1 μg RNA was used to synthesize cDNA with a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, 4368814). Relative abundance of mRNAs was measured using multiplex qRT-PCR with TaqMan Fast Advanced Master Mix (Applied Biosystems, 4444964), FAM-labeled TaqMan gene expression assays (Thermo Fisher Scientific) Acp5 (Mm00475698_m1), Ctsk (Mm00484039_m1), and VIC-labeled mouse Actb (β-actin, Applied Biosystems, 4352341E). The relative mRNA levels were calculated using the comparative cycle threshold method (Livak and Schmittgen, 2001).

2.5. Ex vivo osteoblastic cell culture

For protein isolation, femurs and tibia were dissected, epiphyses were removed, and bone marrow was flushed using Hank's isolation media (MEM with 1× HBSS + 10% FBS + 1% PSG [penicillin/streptomycin/glutamine]). Cells were seeded into 6-well plates at a density of 12.5 × 10⁶ cells/well in osteogenic media (α-MEM + 10% FBS + 1% PSG + 50 μg/mL ascorbic acid). Once cells reached confluency, protein was isolated for downstream analysis. For ex vivo Seahorse Mito Stress Test, cells were isolated as described above and cultured in osteogenic media in 10 cm plates for 4 days.

2.6. Immunoblot analysis

Protein was extracted from cultured osteoblastic cells using RIPA buffer (Thermo Fisher Scientific, PI89901) with protease/phosphatase inhibitors (Cell Signaling Technologies [CST], 5872S). Proteins were resolved in 4–20% Mini-PROTEAN TGX gels (Bio-Rad, 4561093) and transferred onto a TransblotTurbo midi-size nitrocellulose membrane (0.2 μm pore size, Bio-Rad, 1704271). The membrane was blocked for 5 min with EveryBlot Blocking Buffer (Bio-Rad, 12010020) and incubated overnight, rocking, at 4 °C with primary antibodies. The primary antibodies were p62 (CST, 23214 s, 1:1000 dilution), LC3 (CST, 12741 T, 1:1000 dilution), and β-actin (MilliporeSigma, A5316, 1:4000 dilution). After incubation with primary antibodies, the membranes were washed with PBS 3 times for 5 min each, then incubated with secondary antibodies conjugated to IRDye 680 or IRDye 800 dyes (LI-COR, 1:2000 dilution). The membranes were washed with PBS 3 times for 5 min each and then scanned with an Odyssey IR imaging system (LI-COR) and Image Studio Software (Version 5.2).

2.7. Seahorse Mito Stress Test

After incubation in osteogenic media, osteoblastic primary bone marrow mesenchymal stromal cells (pBMSCs) were seeded into XFe96/XF Pro microplates (Agilent Seahorse Bioscience, 101085-004) at a density of 15,000 cells/well and cultured for 48 h. A Seahorse Extracellular Flux Analyzer (Agilent) was used for measuring mitochondrial respiration according to kit instructions (Agilent, 103777-100). Oligomycin (1 μM), FCCP (Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone, 1 μM), and rotenone (0.5 μM) with antimycin A (0.5 μM) were injected sequentially to quantify oxygen consumption rate (OCR). After the assay, total protein/well was quantified using a Pierce™ BCA protein assay kit (Thermo Fisher Scientific, 23225), and OCR values were normalized to protein concentration using Wave Desktop software (v 2.6, Agilent). Data were obtained from 3 to 5 biological replicates, each with seven technical replicates.

2.8. Cell isolation for 10× genomics sequencing

Periosteal cells were isolated as previously described (Nookaew et al., 2024). Briefly, femurs from 2 mice per group were cleaned of muscle, leaving the periosteum intact. The femurs were subjected to three digestions with LiberaseTM and three decalcifications with EDTA. The released cells were combined and enriched in mesenchymal cells by depletion of hematopoietic and endothelial cells using a lineage depletion kit (Miltenyi Biotec, 130-090-858) and CD45, CD117, and CD31 microbeads (Miltenyi Biotec, 130-0520301, 130-091-224, and 130-097-418) according to the manufacturer's instructions. The cells were encapsulated using a Chromium Controller (10× Genomics). Libraries were constructed using a Chromium Single Cell 3’ Reagent Kit (10× Genomics) and sequenced using Illumina NovaSeq 6000 to generate FASTQ files.

2.9. Bioinformatic analysis of single-cell RNA-sequencing

Cell Ranger software version 7.1 (10× Genomics) was used for preprocessing FASTQ files to produce feature-barcode matrixes. Alignments were performed using the mouse reference genome mm10. Feature-barcode matrixes were imported for further analysis in R suite software using Seurat package version 4.2.0 (Hao et al., 2021). Further analysis included cells containing between 1000 and 5000 transcripts, mitochondria read content less than 15%, and complexity of RNA species (log10[number of genes/number of unique molecular identifiers] > 0.85). Harmonization was performed using a reciprocal principal component analysis method based on the top 50 principal components and 6000 most variable features to minimize batch effect. Harmonized results were used for clustering using Louvain algorithm with multilevel refinement and Uniform Manifold Approximation and Projection (UMAP) for dimension reduction. The gene-specific markers of individual clusters and differentially expressed genes were identified using Model-based Analysis of Single-cell Transcriptomics (MAST) algorithm for cell type identification (Finak et al., 2015). The Gene Set Enrichment Analyses (GSEA) were performed using piano package (Varemo et al., 2013). The single cell RNA-seq data is available NCBI SRA under BioProject number PRJNA1444489.

2.10. Statistical analysis

For two-sample analyses evaluating differences between Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice, data distributions were checked for equal variance to choose the appropriate test for each. 2-tailed t-test was used when variances were equal, and Welch's t-test was used when variances were unequal. One-way ANOVA and Tukey's post hoc analysis were employed for analysis of Fig. 1. For Fig. 3A–D, repeated-measures mixed models ANOVA was used to determine the effects of genotype and time. Residuals were inspected to assure model assumptions of normality and constant variance were met. Time was treated as a categorical variable with month 3 as the reference category. To determine whether genotypes differed in their changes over time, the difference between genotypes was compared at 3 months and 9 months. Differences between the genotypes were evaluated using GraphPad Prism 10.5.5 software (GraphPad Software, Inc., La Jolla, CA, USA). All values are reported as mean ± standard deviation (SD), with the exception of Fig. 6A which was represented as mean ± standard error of the mean (SEM). A p < 0.05 was considered significant. Individual statistical tests, p values, and sample sizes are indicated in each figure legend.

Fig. 3.

Elimination of osteoblast lineage autophagy post-3 weeks of age leads to sustained low bone mass through 10 months of age. (A–D) Body weight (A) and serial global (total body) (B), femoral (C), and vertebral (D) DXA BMD measurements were collected from female Osx1-Cre; Atg7^fl/fl mice and their Cre-expressing littermates from 3 to 9 months of age. n = 13 mice/group. Error bars indicate mean ± SD. * indicates p < 0.05 as calculated by repeated-measures 2-way ANOVA. # indicates p < 0.05 comparing the difference in genotypes at 3 versus 9 months (detailed in methods section). (E–F) μCT analysis of femurs and vertebrae from 10-month-old female Osx1-Cre; Atg7^fl/fl mice and their Cre-expressing littermates. (E) Femoral cortical thickness (Ct.Th.) measured at the femur midshaft and distal metaphysis. (F) Vertebral Ct.Th. and cancellous bone volume/tissue volume (BV/TV) measured in lumbar vertebrae 4. n = 9–10 mice/group. Bars indicate mean ± SD. Indicated p values were calculated by unpaired t-test.

Fig. 6.

Loss of autophagy in osteoblast progenitors impairs mitochondrial respiration and abolishes spare respiratory capacity. Primary bone marrow stromal cells were isolated from femurs and tibias from 6-month-old male control and Osx1-Cre; Atg7^fl/fl mice and cultured in osteogenic media for 4 days. The cells were subjected to an Agilent Seahorse XF Mitochondrial Stress Test. (A) Real-time Oxygen Consumption Rate (OCR) kinetic trace normalized to total cellular protein content (pmol/min/μg protein). Sequential injection times for Oligomycin (1.0 μM), FCCP (1.0 μM), and Rotenone/Antimycin A (Rot/AA, 0.5 μM) are indicated by dashed lines. Data points represent the biological group mean ± SEM. (B) Basal respiration, maximum respiration, spare respiratory capacity as a percent of maximum respiration, and ATP production were quantified from OCR kinetic trace. n = 3–5 mice/group with 7 technical replicates. Bars indicate mean ± SD. Indicated p values were calculated by unpaired t-test.

3. Results

3.1. Autophagy in the osteoblast lineage is essential for skeletal growth during development and post-weaning

Previous studies have inhibited autophagy starting prenatally and shown that autophagy in the entire osteoblast lineage is essential for bone formation (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016). However, the importance of osteoblast lineage autophagy for skeleton during development vs. postnatal growth have not been assessed. To address this, we deleted Atg7 – an essential component of autophagy – from the entire osteoblast lineage using the Tet-off–regulated Osx1-Cre transgene. In this genetic maneuver, doxycycline (dox), provided in a diet, suppresses Cre expression and enables temporal control of Atg7 deletion, and thereby autophagy loss. To distinguish the contribution of osteoblast lineage autophagy to bone homeostasis during development vs. post-3 weeks of age (weaning), breeders and progeny were fed either standard chow or dox-containing diet until weaning. Specifically, Group A mice and their parents received chow diet, resulting in Osx1-Cre expression and loss of autophagy from prenatal period to 4.5 months of age. On the other hand, Group B mice and their parents received dox diet until 3 weeks of age, and upon weaning Group B mice were switched to chow diet, thereby restricting Osx1-Cre expression – and thus autophagy deletion – to the post-weaning period (Fig. 1A).

To confirm autophagy deficiency in osteoblasts from Group B animals (autophagy inhibition post-weaning), we measured LC3 and p62 protein levels of osteoblastic cells in culture. Upon initiation of autophagy, LC3-I is lipidated to form LC3-II which is incorporated into the autophagosome membrane. p62 is a classical autophagy cargo receptor which is degraded in the autolysosomes. Consistent with previous work (Resende-Coelho et al., 2025), osteoblasts from Osx1-Cre;Atg7^fl/fl mice have reduced conversion of LC3-I to LC3-II and increased accumulation of p62 compared to controls demonstrating successful autophagy elimination (Supplemental Fig. 1).

As previously shown (Davey et al., 2012), Cre expression starting prenatally (group A) decreased body weight of mice bearing Osx1-Cre transgene (Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice) compared to wild type (WT) or Atg7^fl/fl mice (Fig. 1B). However, inhibition of Cre expression until weaning (group B) prevented the impact of Cre expression on body weight (Fig. 1B). Nonetheless, the timing of autophagy inhibition did not make a difference between the body weights of Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice (group A and group B, Fig. 1B), indicating that the following changes in bone mass were due to the impact of autophagy deficiency in osteoblast lineage cells.

Dual-energy X-ray absorptiometry (DXA) bone mineral density (BMD) analysis revealed that regardless of when autophagy was eliminated from the osteoblast lineage, loss of autophagy in the Osx1-Cre;Atg7^fl/fl mice reduced their BMD at all sites measured (Fig. 1C–E). However, timing of the Atg7 deletion impacted the magnitude of the BMD reduction. Specifically, loss of autophagy from the Osx1-Cre targeted cells prenatally (group A) caused bone mass to be reduced more drastically than loss of autophagy post weaning (group B) (Fig. 1C–E). This highlights the importance of autophagy for skeletal development and shows that autophagy remains important as mice reach skeletal maturity (post-weaning).

3.2. Autophagy is required for postnatal growth and maintenance of cancellous and cortical bone compartments

We next conducted detailed μCT analysis of 4.5-month-old female mice to examine the impact of prenatal (group A) vs. post-weaning (group B) elimination of autophagy on different bone compartments. Consistent with previous reports (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2015), deletion of Atg7 prenatally from the Osx1-Cre-targeted cells (Group A - Osx1-Cre;Atg7^fl/fl) decreased femoral cortical thickness compared to their Cre-expressing littermate controls in both femoral midshaft and at metaphysis (Fig. 2A–C). In contrast, loss of autophagy post 3 weeks of age (Group B - Osx1-Cre;Atg7^fl/fl) reduced cortical thickness in metaphasis (Fig. 2C), but not in midshaft (Fig. 2A). Autophagy elimination at either time point (Group A or B - Osx1-Cre;Atg7^fl/fl) reduced vertebral cortical thickness measured at the ventral cortical wall (Fig. 2D). Similarly, Osx1-Cre;Atg7^fl/fl mice had decreased cancellous bone volume compared to controls independent of whether autophagy was eliminated prenatally or post-weaning (Fig. 2E–F). In both instances, the decreases in cancellous bone volume of Osx1-Cre;Atg7^fl/fl mice were due to decreased trabecular thickness (Supplemental Fig. 2A–B). Of note, we have previously observed changes in trabecular number and separation in L4 vertebrae of autophagy-deficient mice (Piemontese et al., 2016). In the present study, we utilized a smaller sample size which may be limiting our ability to detect more subtle differences in trabecular number and separation. Nonetheless, these results indicate that autophagy remains important for the growth and maintenance of cortical and cancellous bone compartments even after 3 weeks of age.

3.3. Loss of autophagy in the osteoblast lineage causes a progressive reduction in vertebral bone mass

Next, we wanted to determine whether autophagy deficient mice (Osx1-Cre;Atg7^fl/fl) continued to lose bone mass beyond rapid growth. To address this, we kept mice on a dox-containing diet until 3 weeks of age and then conducted serial BMD analysis of Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice from 3 to 9 months of age. There was no difference in body weight between Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice at any of the time points measured (Fig. 3A). Global, femoral, and vertebral BMDs were consistently lower in Osx1-Cre;Atg7^fl/fl mice compared to Cre controls across all skeletal sites (Fig. 3B–D). The magnitude of femoral or global bone mass reduction was similar at 3 and 9 months of age. However, the difference between vertebral BMDs of Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice was greater at 9 months than 3 months. This suggests that vertebral bone loss induced by autophagy-deficiency is exaggerated over time (Fig. 3D).

Next, we conducted μCT analysis in 10-month-old female Osx1-Cre and Osx1-Cre;Atg7^fl/fl mice. Autophagy deficiency starting post-weaning decreased cortical thickness at the femur metaphysis and vertebral ventral cortical wall (Fig. 3E–F). Autophagy deficiency also reduced vertebral cancellous bone of 10-month-old Osx1-Cre;Atg7^fl/fl mice (Fig. 3F). Similar to 4.5 months of age, this decrease in vertebral cancellous bone was due to decreased trabecular thickness (Supplemental Fig. 3). Together, these results indicate that autophagy remains important for the maintenance of cortical and cancellous bone beyond skeletal maturity.

3.4. Elimination of osteoblast lineage autophagy reduces bone formation

We have previously shown that the low bone mass phenotype observed in autophagy-deficient mice is due to decreased bone formation (Piemontese et al., 2016). Here, we wanted to determine whether bone formation rate of Osx1-Cre;Atg7^fl/fl mice remains low at 10 months of age. To address this, we conducted dynamic histomorphometric analysis of vertebral cancellous bone in 10-month-old female mice. Mineral apposition rate and mineralizing surface over bone surface were both lower in Osx1-Cre;Atg7^fl/fl mice compared to controls (Fig. 4A–B). These decreases led to reduced bone formation rate (Fig. 4C–D), demomstrating that autophagy remains important for maintaining bone formation beyond skeletal growth.

Fig. 4.

Elimination of osteoblast lineage autophagy post-3 weeks of age reduces bone formation at 10 months of age. Dynamic histomorphometric analysis of vertebral cancellous bone from 10-month-old female Osx1-Cre; Atg7^fl/fl mice and their Cre-expressing littermates. (A) Mineral apposition rate (MAR). (B) Mineralizing surface/bone surface (MS/BS). (C) Bone formation rate/bone surface (BFR/BS). (D) Representative histomorphometry images of unstained sections (labeled with calcein and alizarin red) imaged with 20× magnification. White scale bars indicate 100 μm. n = 7 mice/group. Bars indicate mean ± SD. Indicated p values were calculated by unpaired t-test.

Previous studies examining the role of autophagy in the osteoblast lineage agree that loss of autophagy results in decreased bone formation, though, the effect on bone resorption has been inconsistent (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016). To assess if an increase in bone resorption contributes low bone mass in autophagy deficient mice, we measured osteoclast marker genes Acp5 and Ctsk. Osx1-Cre;Atg7^fl/fl mice had lower osteoclast marker gene expression compared to controls (Supplemental Fig. 4). This result suggests that, similar to our previous findings (Onal et al., 2013; Piemontese et al., 2016), postnatal loss of autophagy also reduces bone turnover.

3.5. scRNA-seq reveals potential mechanisms via which autophagy deficiency in Osx1-Cre-targeted cells reduces osteoblast function

In previous studies, we have performed single-cell RNA-sequencing (scRNA-seq) analysis of autophagy-deficient mesenchymal cells from the femur endosteal and trabecular compartments (Onal et al., 2025). Because autophagy affects cortical thickness, we sought to assess and compare the effect of autophagy deficiency in endosteal and periosteal compartments.

To this end, we performed scRNA-seq of periosteal cells as previously described (Nookaew et al., 2024). Briefly, femurs were cleaned of muscle without damaging the periosteum. The cells were released from the intact periosteum by serial digestions with LiberaseTM and EDTA. We enriched for mesenchymal cells by depleting hematopoietic and endothelial cells, and subjected the enriched cells to single-cell RNA-sequencing (scRNA-seq) using the 10× Chromium platform.

Clustering analysis identified eight clusters representing the major mesenchymal cell populations of the periosteum, namely, osteoblasts (high Bglap), Osteo-X (high Postn), Fibro-2 (high Clec3b), Sfrp2+ fibroblasts, Prg4+ fibroblasts, chondrocytes, articular chondrocytes, and pericytes, as illustrated in uniform manifold approximation and projection (UMAP) (Fig. 5A–B). Osteo-X and Fibro-2 cells are potential osteoblasts precursors, which are targeted by the Osx1-Cre transgene (Nookaew et al., 2024). Osteo-X, Fibro-2, and osteoblasts had the highest number of differentially regulated genes in response to autophagy deficiency, as determined by model-based analysis of single-cell transcriptomics (MAST), with a random effect for the sample of origin (Fig. 5C).

Fig. 5.

Autophagy deficiency in Osx1-Cre-targeted cells reduces osteoblast function. Periosteal mesenchymal cells were isolated from femurs of 8-month-old male Osx1-Cre and Osx1-Cre; Atg7^fl/fl mice and were subjected to scRNA-seq using the 10× Genomics Chromium platform. n = 2 mice/group. (A) UMAP visualization of cell clusters obtained from the periosteum. (B) Dot plot of the representative marker genes for each cluster. Circle size indicates the percentage of cells expressing each gene, and color gradient indicates expression level. (C) Volcano plots of significantly up- or downregulated genes in each cluster. (D) Biological processes significantly downregulated (red) or upregulated (green) in response to Atg7 conditional knockout as determined by GO analysis (p_adjusted <0.05). Larger circle sizes and darker colors indicate higher significance. (E) Violin plots with box and whiskers representing the expression of select genes in osteoblasts. Wider parts indicate more data points; inner box plots indicate the median, quartiles, and range of the data.

To determine which cellular processes were altered in response to autophagy deficiency, we performed gene set enrichment analysis (GSEA) in osteoblasts, Osteo-X, and Fibro-2 cells. Previous studies (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016) and our current study (Fig. 4) show that autophagy is necessary to maintain bone formation. Consistent with this, autophagy-deficiency in osteoblasts decreased gene ontology (GO) terms related to collagen biosynthesis and organization. We have shown that autophagy stimulation increases expression of bone matrix components by the Osteo-X cells (James et al., 2025). Conversely, herein, we show that autophagy inhibition dampens the collagen biosynthetic process by the Osteo-X cells (Fig. 5D). These changes support the idea that Osteo-X cells are osteoblast progenitors and autophagy primes their differentiation into osteoblasts. A major function of autophagy is maintaining protein homeostasis. Thus, its deficiency led to decreased proteostasis as indicated by a decrease in GO terms related to the endoplasmic reticulum and Golgi apparatus, protein folding, protein trafficking, and post-translational modifications of proteins in Fibro-2, Osteo-X cells, and osteoblasts (Fig. 5D). Together these suggest that autophagy deficiency reduces osteoblast production and function.

Similar to our observations in the endosteal preparations (Onal et al., 2025), loss of autophagy caused an increase in senescence, senescence-associated secretory phenotype (SASP), and cell death in osteoblasts. These processes also increased in Osteo-X and Fibro-2 cells in periosteum (Fig. 5D). Senescence is thought to impact proliferation of osteoblast progenitor cells (Kim et al., 2021). In line with this, autophagy deficiency reduced genes involved in proliferation in both Osteo-X and Fibro-2 cells (Fig. 5D). There was a decrease in proliferation-related genes in osteoblasts as well; however, as osteoblasts are post-mitotic, this may just indicate retention of this transcriptional signature from their progenitors. Further studies are needed to validate in vivo whether apoptosis and proliferation are altered in the periosteum of autophagy-deficient mice.

GO terms related to cellular response to hypoxia and angiogenesis were also increased in periosteal cells (Fig. 5D). Similarly, we have also observed an increase in angiogenesis in the endosteum of autophagy-deficient mice (Onal et al., 2025). These increases suggest that blood vessels are being actively recruited to the cortical bone in these mice. In previous studies, we have shown that loss of autophagy in the osteoblast lineage increases cortical porosity (Onal et al., 2013). Therefore, the development of cortical porosity over time in autophagy-deficient mice may be driven by an increase in angiogenic signaling at the periosteum and endosteum.

In relation to signaling pathways, autophagy deficiency in the periosteum increased GO terms related to R-SMAD binding and response to tumor necrosis factor (TNF) (Fig. 5D). scRNA-seq analysis revealed that autophagy deficiency in osteoblasts increased expression of various genes downstream of tumor necrosis factor-α (TNF-α), including the transcription factors Nfkbia and Jun, as well as genes downstream of transforming growth factor-β (TGF-β), including the canonical pathway mediator Smad4 and TGF-β target gene Serpine1 (Fig. 5E). Upregulation of either pathway can inhibit maturation of osteoblasts and induce apoptosis (Abbas et al., 2003; Alliston et al., 2001; Gilbert et al., 2000; Jilka et al., 1998; Liu, 2025; Maeda et al., 2004), which is consistent with reduced osteoblast numbers in Osx1-Cre;Atg7^fl/fl mice (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2015). Further studies are required to assess the role of TNF-α and TGF-β signaling in autophagy-deficient osteoblasts.

Another major function of autophagy is maintaining cellular homeostasis by recycling damaged or dysfunctional organelles. Autophagy deficiency in osteoblasts leads to an accumulation of mitochondria and reactive oxygen species (Piemontese et al., 2016), which suggests that lack of autophagy results in the accumulation of defective mitochondria. Surprisingly, our scRNA-seq revealed that autophagy deficiency also reduced expression of genes related to mitochondria and cellular metabolism (Fig. 5D). To test whether autophagy-deficiency causes mitochondrial dysfunction, we subjected primary osteoblastic stromal cells to a Seahorse Mito Stress Test, and showed that autophagy deficiency reduced mitochondrial function as evidenced by a reduction in basal respiration, maximal respiration, and spare respiratory capacity (SRC) compared to control cells (Fig. 6A–B). Osx1-Cre;Atg7^fl/fl cells did not have a significant reduction in ATP production (Fig. 6B). The reduced SRC in Osx1-Cre;Atg7^fl/fl cells indicates that they are operating near their maximum respiratory capacity, working harder to maintain ATP production to meet metabolic requirements. This reduction in SRC is also thought to increases the cells' susceptibility to mitochondrial collapse under additional stress (Marchetti et al., 2020). Overall, our results suggest that autophagy may impact cell metabolism by decreasing mitochondrial function and turnover.

4. Discussion

Inhibition (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016) and stimulation (James et al., 2025) of autophagy revealed that autophagy in the entire osteoblast lineage supports bone formation. However, importance of osteoblast lineage autophagy at different stages of skeletal maturity (during development, growth, and post-skeletal maturity) has not been examined in detail. To start filling in this gap in knowledge, we inhibited autophagy in the entire osteoblast lineage starting prenatally or after weaning and performed detailed skeletal phenotyping at 4.5 and 10 months of age. We found that Osx1-Cre;Atg7^fl/fl mice had reduced bone mass compared to controls regardless of when autophagy was eliminated, but prenatal elimination of autophagy produced a greater effect. This finding demonstrates that autophagy is important for skeletal development and growth.

A study by Li et al. examined the postnatal role of autophagy in osteoblasts and osteocytes. Specifically, they inhibited autophagy between postnatal days 5 and 21, or between 8 and 22 weeks by deleting Atg7 with Col1a-Cre^ERT2 (Li et al., 2018). In a separate study, Thomas et al. showed that postnatal autophagy-inhibition (Atg5 and Fip200 deletion by Osx1-Cre transgene) also reduces bone mass in the cranial vault (Thomas et al., 2019). Here, we build on these studies by inhibiting autophagy postnatally with use of a different Cre driver which targets the entire osteoblast lineage (Osx1-Cre). These studies show that postnatal inhibition of autophagy in the osteoblast lineage decreases bone formation and bone mass. Further, herein we also show that the impact of autophagy deficiency on vertebral bone loss becomes greater as mice age, and that autophagy remains important for cancellous bone formation post-growth at 10 months of age, well after peak bone mass had been attained. Together, our results suggest that autophagy supports cancellous bone formation during growth and post-skeletal maturity.

Previous studies have utilized Osx1-Cre to inhibit autophagy in the osteoblast lineage via deletion of Atg7 and Fip200 (Liu et al., 2013; Piemontese et al., 2016). Commonly, these studies show that autophagy deficiency reduces osteoblast number, bone formation, and bone mass. However, some phenotypic differences were observed between the Fip200- and Atg7-deficient models such as differences in their impact on bone resorption and spontaneous fractures. These differences may be due to autophagy-independent functions of Atg7 or FIP200, such as cell cycle regulation and apoptosis, transcriptional regulation, and LC3-associated phagocytosis (Galluzzi and Green, 2019; Shang et al., 2024). Therefore, comparing phenotypes resulting from deletion of different essential autophagy genes may help solidify the specific contribution of autophagy to the regulation of bone formation.

Here, we show that loss of autophagy impacts cortical bone at all skeletal sites and stages of bone growth. It was unexpected, however, that elimination of autophagy after 3 weeks of age affected the femur metaphysis, but not the midshaft. This difference in midshaft and metaphysis cortical thickness could be due to differences in the rate of periosteal modeling. One study points out that up to 3 months of age, the rate of periosteal expansion of the metaphysis exceeds that of the diaphysis (Glatt et al., 2007). Therefore, the difference observed at the metaphysis could be due to its early rapid bone formation, compared to the slower bone formation at the diaphysis. Nonetheless, our study shows that autophagy is important for the cortical bone compartment.

Our scRNA-seq analysis in the periosteum showed that both inhibition (Fig. 5) and stimulation of autophagy (James et al., 2025) in the osteoblast lineage had the most impact on osteoblasts, Osteo-X, and Fibro-2 cells. Osteo-X and Fibro-2 cells are thought to be osteoblast progenitors and are targeted by Osx1-Cre (Nookaew et al., 2024). We previously hypothesized that Osteo-X cells commit to becoming osteoblasts upon autophagy stimulation (James et al., 2025). The present study supports this notion as eliminating autophagy decreased GO terms related to osteoblast function (e.g., collagen biosynthetic process, protein transport) and increased TNF and TGF-β signaling pathways, which are known to prevent osteoblast maturation (Abbas et al., 2003; Alliston et al., 2001; Gilbert et al., 2000; Maeda et al., 2004) in Osteo-X and Fibro-2 cells. Together, these results suggest a role of autophagy in osteoblastogenesis.

As mentioned above, our scRNA-seq analysis revealed that autophagy deficiency induced TNF and TGF-β signaling pathways (Fig. 5D–E). Loss of autophagy may be activating the TNF pathway by two possible mechanisms. First, autophagy deficiency in the osteoblast lineage results in accumulation of p62 (Onal et al., 2013; Resende-Coelho et al., 2025), which can induce TNF-α signaling through activation of NF-κB (Moscat and Diaz-Meco, 2009). Second, autophagy deficiency increases reactive oxygen species (ROS) (Onal et al., 2013), which has also been shown to induce NF-κB activity in osteoblasts (Almeida et al., 2010). Thus, elevated p62 and ROS represent plausible mechanisms via which autophagy deficiency may elevate TNF signaling in the osteoblast lineage. On the other hand, autophagy-deficiency may increase TGF-β signaling as a result of a crosstalk with TNF-α signaling (Liu et al., 2021), or via the elevated ROS (Liu and Desai, 2015). Nonetheless, elevation of these two pathways in response to autophagy deficiency may be underpinning the impairment in osteoblast maturation or the elevation of osteoblast apoptosis.

Our scRNA-seq analysis also revealed that autophagy-deficiency caused similar changes in both periosteal (Fig. 5) and endosteal osteoblasts (Onal et al., 2025). Autophagy deficiency in both periosteal and endosteal osteoblasts altered biological processes related to osteoblast function, proteostasis, apoptosis, cellular metabolism, and senescence. In our recent work, we confirmed that genetic inhibition of autophagy promotes senescence and SASP in osteoblast lineage cells (Onal et al., 2025). Senescence of osteoblast lineage cells contributes to age-related bone loss (Farr et al., 2016, Farr et al., 2017; Onal et al., 2025). Thus, it is reasonable to hypothesize that the age-related reduction in autophagy (Ma et al., 2018; Rubinsztein et al., 2011; Zhang et al., 2020) induces senescence to contribute to age-related bone loss. However, a context-dependent connection between autophagy and senescence is observed in many cell types. Specifically, autophagy can prevent senescence (Garcia-Prat et al., 2016; Moreno-Blas et al., 2019) or do the opposite - promote and support senescence and SASP (Capparelli et al., 2012a, Capparelli et al., 2012b; Lee et al., 2021; Liu et al., 2014; Nam et al., 2013; Peng et al., 2023; Slobodnyuk et al., 2019; Young et al., 2009). Therefore, future functional studies are needed to establish causation between autophagy and senescence in osteoblast lineage cells in the context of skeletal aging.

Previous studies from our group and others showed that inhibiting autophagy at progressively earlier stages of osteoblastogenesis results in increasingly severe skeletal phenotypes (Li et al., 2018; Liu et al., 2013; Nollet et al., 2014; Onal et al., 2013; Piemontese et al., 2016). These studies suggest that autophagy plays an important role throughout the osteoblast lineage. Although we focus mainly on osteoblasts and progenitors in this work, our model also eliminated autophagy from osteocytes. Indeed, recently we showed that postnatal autophagy-deficiency in Osx1-Cre-targeted cells decreases osteocytes connections (Resende-Coelho et al., 2025). While this decrease in the osteocyte network did not impact load-induced bone formation (Resende-Coelho et al., 2025), postnatal loss of osteocyte autophagy may nonetheless be contributing to the low bone mass phenotype we observe. Thus, future studies in which autophagy is eliminated specifically in osteocytes are necessary to clarify the functional role of osteocyte autophagy in postnatal bone formation.

One limitation of this study is that we only used female mice to evaluate changes in bone mass and architecture, and therefore, sex-dependent effects of postnatal autophagy deficiency cannot be excluded. Some studies have reported sex-specific differences in the skeletal response to autophagy deficiency. In rapidly growing mice (3–6 weeks of age) autophagy deficiency reduced BMD in male, but not female mice (Li et al., 2018; Onal et al., 2013). In contrast, in skeletally mature mice (4.5–6 months of age), both Li et al. and our group observed reduced bone mass in both autophagy-deficient males and females (Li et al., 2018; Onal et al., 2013). Nollet et al., however, reported decreased cancellous bone volume in 9-month-old female, but not male, mice (Nollet et al., 2014). Collectively, these findings suggest that the magnitude, timing, and skeletal compartments affected by autophagy deficiency may differ between sexes. Thus, future studies in male mice are necessary to determine whether the skeletal effects observed herein are sex-dependent.

In conclusion, autophagy is important for maintaining osteoblast function at multiple stages of skeletal development and maintenance, including postnatal growth and post-skeletal maturity. We believe that autophagy supports the function of osteoblasts and may promote their maturation from Osteo-X and Fibro-2 progenitor populations. Loss of autophagy may block osteoblast maturation by increasing TNF-α and TGF-β signaling. These findings illuminate potential molecular mechanisms governed by autophagy in osteoblasts, providing a foundation for future studies aimed at therapeutically targeting skeletal pathologies associated with impaired autophagy.

CRediT authorship contribution statement

Alicen James: Writing – review & editing, Writing – original draft, Visualization, Investigation, Formal analysis. Ilham Kadhim: Investigation. Jacob Laster: Investigation. Vivek Khanal: Investigation. Jeff Thostenson: Formal analysis. Neha Dole: Resources. Jinhu Xiong: Resources, Methodology. Intawat Nookaew: Visualization, Software, Formal analysis, Data curation. Melda Onal: Writing – review & editing, Visualization, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Funding

This work was supported by the National Institute of General Medical Sciences (NIGMS) grant P20GM125503 and the UAMS Bone and Joint Initiative Funds.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary figures

Data availability

Data will be made available on request.