Autophagic flux is promoted by the absence of Ano5 in mCOBs. To investigate whether autophagic activity is implicated in the absence of Ano5, we designed a series of experiments to explore the difference in autophagic activity between Ano5+/+ and Ano5–/– mouse cranial osteoblasts (mCOBs). Data from experiments confirmed that the absence of Ano5 caused a significant promotion of autophagic vesicle formation. As shown in Figure 1, A and B, the expression of osteocalcin (OCN) and collagen type I α1 (COL1A1) was notably elevated in Ano5–/– mCOBs compared with Ano5+/+ mCOBs during osteogenic induction. Additionally, the autophagy marker LC3B-II exhibited a marked increase in Ano5–/– mCOBs relative to Ano5+/+ mCOBs (Figure 1B), which is consistent with the results of Gene Ontology (GO) analysis based on RNA sequencing (RNA-Seq). The GO analysis revealed that autophagic activity was substantially different in mCOBs in the Ano5–/– group compared with those in the Ano5+/+ group (Supplemental Figure 2; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.189817DS1). However, the expression of another autophagy marker, p62, did not show obvious differences between these 2 groups (Figure 1A). Then, we conducted further experiments to assess autophagic flux to distinguish between increased autophagic vesicle production and impaired autophagic vesicle degradation, both of which may contribute to the accumulation of LC3B-II. Firstly, chloroquine (CQ) was utilized to inhibit the degradation of LC3B-II. The results indicated a greater accumulation of LC3B-II in Ano5–/– mCOBs, suggesting that the production of autophagic vesicles was enhanced (Figure 1C). Secondly, the results from transmission electron microscopy (TEM) showed that after osteogenic induction, the total amount of autolysosomes in Ano5–/– mCOBs exceeded that in Ano5+/+ mCOBs, while the total amount of autophagosomes did not differ statistically between the 2 groups (Figure 1, D and E). Thirdly, the maturation of autophagic vesicles was monitored with autophagosomes labeled as yellow puncta and autolysosomes labeled as red puncta by using adenovirus expressing an mCherry-GFP-LC3B fusion protein. The results demonstrated that, prior to osteogenic induction, there was no significant change in the number of autophagosomes in Ano5–/– mCOBs compared to Ano5+/+ mCOBs, whereas the number of autolysosomes increased significantly (Figure 1, F and G). Following osteogenic induction, both autophagosomes and autolysosomes in Ano5–/– mCOBs showed a marked increase, with the change in autolysosomes being particularly pronounced (Figure 1, H and I).

Autophagic flux is promoted by the absence of Ano5 in mCOBs. (A and B) Western blot analysis of p62, OCN, LC3B-II, and COL1A1 protein levels in mCOBs from the Ano5+/+ and Ano5–/– groups during osteogenic differentiation. (C) Western blot showing the accumulation of LC3B-II in mCOBs from the Ano5+/+ and Ano5–/– groups after CQ treatment. (D) TEM images of mCOBs following 14 days of osteogenic induction, displaying autophagosomes (black arrowheads) and autolysosomes (red arrowheads). Scale bars: 2 μm (top) and 1 μm (bottom). (E) Quantitation of autophagosomes and autolysosomes observed in TEM images. (F and H) Confocal microscopy of mCOBs infected with mCherry-GFP-LC3B adenovirus, assessed on day 0 and day 14 of osteogenic induction. Representative images show autophagic activity. Scale bars: 25 μm (top rows) and 10 μm (bottom rows). The zoomed-in images in F are magnified ×800, and those in H are magnified ×1200. (G and I) Quantification of autophagosomes (mCherry+GFP+, yellow puncta) and autolysosomes (mCherry+GFP–, red puncta) in F and H, respectively. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Additionally, Lyso-Tracker was utilized to assess the number of lysosomes, which is vital for autolysosome formation and degradation. We found that the number of lysosomes was substantially higher in Ano5–/– mCOBs compared with Ano5+/+ mCOBs during osteogenic induction (Supplemental Figure 3A). Furthermore, the genes related to the integrity of lysosomes and the transportation of autophagosomes to lysosomes, including RAB7, member RAS oncogene family (Rab7a), lysosomal-associated membrane protein 1 (Lamp1), and vesicle-associated membrane protein 8 (Vamp8) were analyzed. The results showed that the expression of these genes in Ano5–/– mCOBs increased significantly compared with Ano5+/+ mCOBs (Supplemental Figure 3B). These findings further corroborate the observation that Ano5 deficiency led to notably increased autolysosomes in Ano5–/– mCOBs.

Subsequently, we examined relevant indicators in bone tissues. Notably, the thickness of the cartilage layer at the tibial epiphysis was significantly increased (Supplemental Figure 4, A and B), and OCN expression was elevated in the alveolar bone lacunae of the mandible in Ano5–/– mice (Supplemental Figure 4, C and D), indicating aberrant bone metabolic activity. Additionally, LC3B-II expression levels were markedly upregulated in various bone tissues, including the mandibles, tibiae, and femurs (Supplemental Figure 4, E and F). These findings suggest the presence of abnormal bone metabolism and autophagic activity within the bone tissues of the Ano5-deficient mice. Notably, both in mCOBs and bone tissues from Ano5–/– mice, autophagy and osteogenesis markers were significantly upregulated. This suggests a potential link between autophagy dysregulation and the abnormal osteogenic activity observed in this model.

Inhibition of autophagy mitigates the aberrant osteogenic capacity in Ano5–/– mCOBs. To further examine whether the change in autophagic activity is associated with excessive osteogenesis in Ano5–/– mCOBs, 3-methyladenine (3-MA), which inhibits the formation of autophagic vesicles, was utilized in Ano5–/– mCOBs. The results showed that LC3B-II levels in Ano5–/– mCOBs was reduced following treatment with 3-MA after osteogenic induction (Figure 2A), accompanied by decreased protein and mRNA levels of OCN and COL1A1 after osteogenic induction (Figure 2, A and B). We then assessed the degree of osteogenic differentiation in the cells. Results showed that the alkaline phosphatase (ALP) activity, which represents the early stage of osteogenic differentiation, was higher in Ano5–/– mCOBs compared with Ano5+/+ mCOBs, and it was significantly reduced following treatment with 3-MA (Figure 2, C and D). Additionally, calcium nodule formation, which signifies the late stage of osteogenic differentiation, was also substantially decreased in the 3-MA–treated group, as indicated by alizarin red staining (ARS) (Figure 2C). These results demonstrate that inhibiting the formation of autophagic vesicles significantly impairs the osteogenic potential of Ano5–/– mCOBs.

Inhibition of autophagy mitigates the aberrant osteogenic capacity in Ano5–/– mCOBs. 3-MA treatment: (A–D) mCOBs from the Ano5–/– group were treated with 3-MA (early-stage autophagy inhibitor) during osteogenic induction. (A) Western blot analysis of LC3B-II, OCN, and COL1A1 protein levels. (B) qRT-PCR analysis of Ocn and Col1a1 expression. (C) ALP staining and alizarin red staining. Scale bars: 200 μm. (D) ALP activity assay. CQ treatment: (E–H) mCOBs were treated with CQ (late-stage autophagy inhibitor). (E) ALP staining and alizarin red staining. Scale bars: 200 μm. (F) ALP activity assay. (G and H) qRT-PCR analysis of Ocn and Col1a1 expression. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Consequently, we explored how inhibiting the degradation of autophagic vesicles affects osteogenesis, given that accumulated autophagic vesicles carrying minerals can enhance osteogenic capacity by being secreted into the extracellular matrix directly (17). Therefore, CQ was utilized to disrupt the function of lysosomes, thereby inhibiting the degradation of autophagic vesicles, and the results showed that the osteogenic capacity of the Ano5–/– mCOBs was also significantly reduced following the treatment with CQ, including reduced ALP activity and calcium nodules formation (Figure 2, E and F). Furthermore, the mRNA levels of Ocn and Col1a1 were markedly downregulated in the CQ-treated group (Figure 2, G and H). These findings suggest that either inhibiting the formation or degradation of autophagic vesicles in Ano5–/– mCOBs could alleviate their abnormal osteogenic capacity, indicating a close correlation between the change in autophagy progression and the excessive osteogenesis in Ano5–/– mCOBs.

ATG9A is involved in autophagy-mediated osteogenesis regulation in Ano5–/– mCOBs. To determine which initial compartments contribute to the increased autophagy, we first assessed the mRNA levels of the autophagy initiation marker UNC-51–like kinase 1 (Ulk1) and the phosphorylation levels at ULK1 Ser555. The results indicated no significant change in Ulk1 mRNA levels (Supplemental Figure 5A), while the phosphorylation level at ULK1 Ser555 was upregulated in Ano5–/– mCOBs (Supplemental Figure 5B). Phosphorylation of ULK1 at Ser555 is known to initiate autophagosome formation (18). However, the application of MRT68921, an inhibitor targeting ULK1, did not significantly inhibit ALP activity of Ano5–/– mCOBs at various stages of osteogenic induction (Supplemental Figure 5, C and D).

To identify other autophagy-related proteins involved in the abnormal autophagy of Ano5–/– mCOBs, we analyzed the mRNA levels of several key genes related to autophagy, including Atg3, Atg4b, Atg5, Becn1, Atg7, and Atg9a. Among these, Atg9a expression exhibited the most significant upregulation in Ano5–/– mCOBs during osteogenic induction (Figure 3A). Western blot analysis further confirmed this finding, revealing that ATG9A protein expression was markedly higher in Ano5–/– mCOBs compared with Ano5+/+ mCOBs, while Beclin-1 expression showed no difference between these 2 groups (Figure 3, B and C). Immunofluorescence corroborated these results, showing that ATG9A was widely distributed in the cytoplasm, with more pronounced perinuclear localization in certain cells (Figure 3, D and E). Notably, the proportion of cells exhibiting this perinuclear aggregation was markedly increased in Ano5–/– mCOBs (data not shown).

ATG9A is upregulated in Ano5–/– mCOBs. (A) qRT-PCR analysis of Atg3, Atg4b, Atg5, Becn1, Atg7, and Atg9a expression. (B) Western blot analysis of Beclin-1 protein levels. (C) Western blot analysis of ATG9A protein levels. (D and E) Immunofluorescence analysis showing ATG9A expression and localization in mCOBs. Scale bars: 10 μm. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Subsequently, we synthesized Atg9a shRNA and administered it to Ano5–/– mCOBs to investigate the effects of silencing Atg9a. The knockdown efficiency was confirmed by a significant reduction in both the mRNA and protein levels of ATG9A in the Ano5–/– plus Atg9a shRNA group during osteogenic induction (Figure 4, A and B). This suppression of autophagy was accompanied by a significant decrease in osteogenic differentiation markers, including lower mRNA and protein levels of OCN and COL1A1 compared with the control group (Figure 4, A and C). Moreover, the early stage of osteogenic differentiation was impaired, as reflected by a significant reduction in the ALP activity (Figure 4, D and E). In addition, a substantial decrease in calcium nodule formation was observed, as demonstrated by ARS (Figure 4F), indicating a disruption in the later stage of osteogenic differentiation.

ATG9A is involved in autophagy-mediated osteogenesis regulation in Ano5–/– mCOBs. (A–F) For Atg9a knockdown, Atg9a shRNA was expressed in Ano5–/– mCOBs via lentiviral transfection. (A) Western blot analysis of ATG9A, OCN, and COL1A1 protein levels. (B and C) qRT-PCR analysis of Atg9a, Ocn, and Col1a1 expression. (D) ALP activity assay. (E) ALP staining. Scale bars: 200 μm. (F) Alizarin red staining of extracellular calcium deposits. Scale bars: 200 μm. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Collectively, these results indicate that among the key proteins involved in autophagy initiation, ATG9A plays a particularly critical role in the abnormal autophagy associated with Ano5 deficiency. The inhibition of ATG9A expression effectively attenuated the abnormal osteogenesis caused by Ano5 deficiency, underscoring a strong association between ATG9A-dependent autophagy upregulation and enhanced osteogenesis in Ano5–/– mCOBs.

Ano5 deficiency promotes osteogenesis in mCOBs via the AMPK/ATG9A pathway. The PI3K/AKT/mTOR signaling pathway is widely known as a negative regulator of autophagy. However, we found that the level of p-AKT was elevated in Ano5–/– mCOBs (Supplemental Figure 6). This prompted further exploration of other autophagy regulatory pathways. Additionally, a recent study indicated that ATG9A is also regulated by the AMPK signaling pathway during autophagy progression (Figure 5A) (19). In line with these studies, Western blot analysis in our study exhibited a marked increase in phosphorylation of AMPK (p-AMPK) in Ano5–/– mCOBs (Figure 5B), which led us to hypothesize that AMPK is more likely involved in the regulation of autophagy in this model. To test this hypothesis, the AMPK inhibitor Compound C (C.C) was utilized in Ano5–/– mCOBs. As shown in Figure 5, C–F, treatment with C.C significantly downregulated the number of both autophagosomes and autolysosomes in Ano5–/– mCOBs during osteogenic induction. Notably, the reduction in autolysosomes exceeded 90%, with their total number significantly lower than that of autophagosomes, rendering autolysosomes almost undetectable within the cells treated with C.C (Figure 5, D and F).

AMPK is involved in autophagy regulation in Ano5–/– mCOBs. (A) Schematic illustration of AMPK- and AKT-mediated regulation of autophagy. The samples were run on the same blot but the lanes are noncontiguous. (B) Western blot analysis of p-AMPK and AMPK protein levels. C.C treatment: (C–F) mCOBs from the Ano5–/– group were treated with C.C (an AMPK inhibitor) during osteogenic induction. (C and E) Confocal microscopy of mCOBs infected with mCherry-GFP-LC3B adenovirus, assessed on day 0 and day 14 of osteogenic induction. Representative images show autophagic activity. Scale bars: 10 μm. The zoomed-in images in C and E are magnified ×900. (D and F) Quantification of autophagosomes (mCherry+GFP+, yellow puncta) and autolysosomes (mCherry+GFP–, red puncta) in C and E respectively. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

In terms of osteogenic differentiation, the inhibition of AMPK by C.C led to a significant rescue of the enhanced osteogenesis observed in Ano5–/– mCOBs. Protein and mRNA levels of key osteogenic markers, including OCN and COL1A1, were markedly reduced compared with the untreated Ano5–/– group (Figure 6, A and B). Additionally, the ALP activity was significantly downregulated in the Ano5–/– plus C.C group, suggesting a restoration of osteogenic potential (Figure 6, C and D). This was further corroborated by decreased calcium nodule formation, as shown by ARS (Figure 6C).

AMPK regulates osteogenic function and is positioned upstream of ATG9A. C.C treatment: (A–F) mCOBs from the Ano5–/– group were treated with C.C during osteogenic induction. (A) Western blot analysis of p-AMPK, AMPK, OCN, and COL1A1 protein levels. (B) qRT-PCR analysis of Ocn and Col1a1 expression. (C) ALP staining and alizarin red staining. Scale bars: 200 μm. (D) ALP activity assay. (E) qRT-PCR analysis of Atg9a expression in mCOBs. (F) Western blot analysis of ATG9A protein levels. The samples were run on the same blot but the lanes are noncontiguous. (G) Upon Atg9a knockdown, Western blot analysis of LC3B-II, p-AMPK, and AMPK protein levels. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Recent research has shown that ATG9A can be regulated by AMPK directly, independently of any other initial complexes (19). To verify whether the regulation of autophagy by AMPK is mediated through ATG9A, we measured ATG9A levels in the Ano5–/– plus C.C group. Western blot analysis showed that AMPK inhibition by C.C resulted in significantly decreased ATG9A expression (Figure 6, E and F). We further assessed the relationship between ATG9A and AMPK by using Atg9a shRNA to knock down ATG9A expression in Ano5–/– mCOBs. The results demonstrated that downregulation of ATG9A could reduce autophagy levels, as evidenced by the reduction in LC3B-II expression, but did not change the expression of p-AMPK (Figure 6G). Furthermore, we applied Atg9a mimics to the Ano5–/– plus C.C group and assessed their osteogenic differentiation. The workflow of the rescue experiment is presented in Figure 7A. The results showed that the overexpression of ATG9A substantially rescued the impaired autophagy and osteogenic capacity caused by C.C during osteogenic induction in Ano5–/– mCOBs, as evidenced by partially restored protein levels of LC3B-II, OCN, and COL1A1, while p-AMPK expression was still at a low level (Figure 7B). Additionally, the ALP activity increased, and calcium nodule formation improved, as shown by ALP/ARS staining and ALP activity quantification (Figure 7, C and D).

Ano5 deficiency promotes osteogenesis in mCOBs via the AMPK/ATG9A pathway. Following Atg9a overexpression combined with C.C treatment: (A) mCOBs were first selected for Atg9a overexpression, followed by C.C treatment. (B) Western blot analysis of p-AMPK, AMPK, ATG9A, LC3B-II, OCN, and COL1A1 protein levels. (C) ALP activity assay. (D) ALP staining and alizarin red staining. Scale bars: 200 μm. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Taken together, these findings provide strong evidence that ANO5 modulates autophagy, at least in part, via the AMPK/ATG9A signaling pathway, and this modulation plays a crucial role in the regulation of osteogenesis in mCOBs.

The application of 3-MA in vivo alleviated the abnormal bone formation. 3-MA, a well-established autophagy inhibitor, is commonly used in vivo to block autophagy progression in animal models (20). Therefore, we selected 3-MA as the intervention drug for in vivo administration. The results showed that Ano5–/– mice exhibited increased cortical bone thickness (Ct.Th) in the femurs and tibiae compared with the Ano5+/+ group (Figure 8 and Supplemental Figure 7). Concurrently, the trabecular bone volume fraction (BV/TV), bone mineral density (BMD), trabecular thickness (Tb.Th), and trabecular number (Tb.N) were all significantly reduced in the Ano5–/– group relative to the Ano5+/+ controls. Following 3-MA treatment, Ct.Th in the femurs and tibiae of Ano5–/– mice decreased compared with the Ano5–/– plus NaCl group. Simultaneously, BV/TV, BMD, Tb.Th, and Tb.N all demonstrated significant increases after treatment (Figure 8, A–F).

In vivo administration of 3-MA alleviated the abnormal phenotype in Ano5–/– mice. Femoral trabecular bone mass was assessed by micro-CT. Representative 3D reconstruction (A) and relative quantification (B and C) are displayed. Tibial trabecular bone mass was assessed by micro-CT. Representative 3D reconstruction (D) and relative quantification (E and F) are displayed (n = 6 per group). Ct.Th, cortical bone thickness; BV/TV, trabecular bone volume/total volume; Tb.Th, trabecular thickness; Tb.N, trabecular number per cubic millimeter; BMD, trabecular bone mineral density. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.

Additionally, Ct.Th on the palatal side of the first molar region of the mandible was significantly greater in the Ano5–/– group; however, this thickness markedly decreased after 3-MA treatment (Figure 9, A and B). The 3-point bending experiment revealed a significant reduction in the elastic modulus of the tibiae in the Ano5–/– group. Both the load-bearing capacity and stress tolerance of the tibiae in the Ano5–/– group were markedly lower compared with the Ano5+/+ group. Following treatment with 3-MA, the tibiae demonstrated a substantial improvement in vertical load tolerance and a higher elastic modulus (Figure 9, C–I). Serological analysis showed that serum levels of ALP and procollagen type I N-terminal propeptide (PINP) were significantly higher in Ano5–/– mice than in Ano5+/+ mice. Upon 3-MA administration, ALP levels in Ano5–/– serum showed a decreasing trend without statistical significance, whereas PINP levels were significantly reduced (Figure 9, J and K). We subsequently measured the levels of ATG9A in the tibiae, and the results showed that the ATG9A levels in the Ano5–/– group were higher than those in the Ano5+/+ group. After applying 3-MA to inhibit autophagy, the ATG9A levels significantly decreased (Figure 9, L and M, and Supplemental Figure 8, A and B). These results further confirm that ATG9A, previously validated in vitro, also exerts an important role in vivo in regulating bone formation.

In vivo administration of 3-MA alleviated the abnormal phenotype in Ano5–/– mice. (A and B) Mandibles were assessed by micro-CT, with the red arrows representing the cortical bone on the palatal side of the first molar region (n = 6 per group). (C) Schematic representation of the 3-point bending test used to evaluate the biomechanical function of tibiae. (D–I) The biomechanical properties of tibiae were assessed through the 3-point bending test (n = 5 per group). (J and K) Serum levels of PINP and ALP (n = 5 per group). (L and M) The expression levels of ATG9A. Data are represented as mean ± SD. *P < 0.05; NS, P > 0.05, as assessed by 1-way ANOVA followed by Tukey’s post hoc test.