Negative pressure wound therapy improves bone regeneration by promoting osteogenic differentiation via the AMPK-ULK1-autophagy axis

ABSTRACT

Deficient bone regeneration causes bone defects or nonunion in a substantial proportion of trauma patients that urges for novel therapies. To develop a reliable therapy, we investigated the effect of negative pressure wound therapy (NPWT) on bone regeneration in vivo in a rat calvarial defect model. Negative pressure (NP) treatment in vitro was mimicked to test its effect on osteoblast differentiation in rat mesenchymal stem cells (MSCs) and MC3T3-E1 cells. Transcriptomic analyses, pharmaceutical interventions, and shRNA knockdowns were conducted to explore the underlying mechanism and their clinical relevance was investigated in samples from patients with nonunion. The potential application of a combined therapy of MSCs in hydrogels with negative pressure was tested in the rat critical-size calvarial defect model. We found that NPWT promoted bone regeneration in vivo and NP treatment induced osteoblast differentiation in vitro. NP induced osteogenesis via activating macroautophagy/autophagy by AMPK-ULK1 signaling that was impaired in clinical samples from patients with nonunion. More importantly, the combined therapy involving MSCs in hydrogels with negative pressure significantly improved bone regeneration in rat critical-size calvarial defect model. Thus, our study identifies a novel AMPK-ULK1-autophagy axis by which negative pressure promotes osteoblast differentiation of MSCs and bone regeneration. NPWT treatment can potentially be adopted for therapy of bone defects.

Abbreviations: ADP, adenosine diphosphate; AICAR/Aic, acadesine; ALP, alkaline phosphatase; ALPL, alkaline phosphatase, biomineralization associated; AMP, adenosine monophosphate; AMPK, AMP-activated protein kinase; ARS, alizarin red S staining; ATG7, autophagy related 7; ATP, adenosine triphosphate; BA1, bafilomycin A₁; BGLAP/OCN, bone gamma-carboxyglutamate protein; BL, BL-918; BS, bone surface; BS/TV, bone surface per tissue volume; BV/TV, bone volume per tissue volume; C.C, compound C; CCN1, cellular communication network factor 1; COL1A1, collagen type I alpha 1 chain; COL4A3, collagen type IV alpha 3 chain; COL4A4, collagen type IV alpha 4 chain; COL18A1, collagen type XVIII alpha 1 chain; CQ, chloroquine; GelMA, gelatin methacryloyl hydrogel; GO, Gene Ontology; GSEA, gene set enrichment analysis; HIF1A, hypoxia inducible factor 1 subunit alpha; HPLC, high-performance liquid chromatography; ITGAM/CD11B, integrin subunit alpha M; ITGAX/CD11C, integrin subunit alpha X; ITGB1/CdD9, integrin subunit beta 1; KEGG, Kyoto Encyclopedia of Genes and Genomes; MAP1LC3B/LC3B, microtubule associated protein 1 light chain 3 beta; micro-CT, microcomputed tomography; MSCs, mesenchymal stem cells; MTOR, mechanistic target of rapamycin kinase; NP, negative pressure; NPWT, negative pressure wound therapy; PRKAA1/AMPKα1, protein kinase AMP-activated catalytic subunit alpha 1; PRKAA2, protein kinase AMP-activated catalytic subunit alpha 2; PTPRC/CD45, protein tyrosine phosphatase receptor type C; ROS, reactive oxygen species; RUNX2, RUNX family transcription factor 2; SBI, SBI-0206965; SPP1/OPN, secreted phosphoprotein 1; THY1/CD90, Thy-1 cell surface antigen; SQSTM1, sequestosome 1; TGFB3, transforming growth factor beta 3; ULK1/Atg1, unc-51 like autophagy activating kinase 1.

KEYWORDS:

• AMPK
• bone nonunion
• bone regeneration
• macroautophagy/autophagy
• mesenchymal stem cells
• negative pressure wound therapy
• osteoblast differentiation
• ULK1

Introduction

Bone defects due to various causes such as trauma, infections, and bone tumors, are common diseases with substantial clinical and economic impacts [1–3]. Although major advances in surgical technology and medical equipment have been made, 5–10% of fracture patients whose normal healing process is inhibited still develop delayed union or nonunion [4,5]. The treatment of bone defects and nonunion remains controversial [1]. Development of a reliable and feasible therapy to augment the impaired or “insufficient” bone regeneration process in bone defects is urgently needed.

Osteogenesis is a key process in bone defect repair. Mesenchymal stem cells (MSCs) in bone marrow, periosteum, vessel walls, muscle, circulation, and other tissues can migrate to the site of bone injury and differentiate to the osteoblast-lineage cells to participate in bone regeneration [6,7]. To promote osteogenesis for bone defect repair, researchers have applied costly growth factors (such as BMP2 [bone morphogenetic protein 2], TGFB [transforming growth factor beta], etc.) in the clinic or in preclinical trials [8,9]. In particular, MSCs are proposed for cell therapy due to their ability to differentiate into multiple cells of mesenchymal tissues, including bone, cartilage, fat, tendon, muscle, and marrow stroma [10]. Interestingly, the osteogenic potential of MSCs can be regulated by mechanical stimulation [11]. Extrinsic mechanical cues and suitable stiffness can promote osteoblast differentiation in vitro and influence endochondral ossification of MSCs [12,13]. However, the potential application of mechanical stimulation for bone regeneration has not been explored.

Negative pressure wound therapy (NPWT) has been developed as a surgical technique widely used in the clinic that promotes wound healing by continuous drainage of secretions [14,15]. Our previous clinical research also found that open bone grafting combined with NPWT can be a feasible alternative for the treatment of infection in tibia [16]. Negative pressure (NP) treatment has been known to affect various types of cells. For example, NP accelerated monolayer podia formation and influenced the proliferation of fibroblasts [17,18] and stimulated endothelial migration and proliferation [19] that potentially promoted angiogenesis. Interestingly, NP treatment promoted osteoblast differentiation in human bone marrow MSCs [20] in vitro. However, the in vivo effect and potential therapeutic efficacy have not been tested. Moreover, the mechanism underlying osteoblast differentiation by NP has not been addressed.

In this research, we applied NPWT to calvarial defects in rats. We found that NPWT accelerated bone regeneration and significantly improved osteogenic activity in defects. We further identified a novel macroautophagy/autophagy axis involving AMP-activated protein kinase (AMPK) and ULK1 (unc-51 like autophagy activating kinase 1) by which NP promotes osteoblast differentiation. Based on our findings, a combined therapy of MSCs in hydrogels with NPWT on rat critical-size calvarial defects significantly improved the repair of bone injury. Our study potentially provides a novel surgical treatment for bone defects.

Results

NPWT accelerates bone regeneration in rat calvarial defects

To determine whether NPWT can promote bone regeneration, we used a rat calvarial defect model. A 4-mm diameter defect was created and NPWT was carried out above the defect for 7 days (Fig. S1A-C). Four and eight weeks after surgery, microcomputed tomography (micro-CT) scanning was performed and a 3D view of the craniums was constructed. Bone defects of the NPWT group were much smaller than those of the control group 8 weeks after surgery (Figure 1A). Quantitative analysis revealed that the NPWT group had a significantly higher bone volume per tissue volume (BV/TV) and bone surface per tissue volume (BS/TV) than the control group (Figure 1B). Histological analyses, including H&E, Masson’s trichrome and Von Kossa staining, showed more osteoid and newly formed bone tissues at the edge and center of defects in the NPWT group than that in the control group (Figure 1C). Alkaline phosphatase (ALP) staining exhibited an increased number of osteoblasts with high osteogenic potential in the NPWT group compared with the control group (Figure 1D). Moreover, immunohistochemistry revealed that tissues in the defects had significantly increased expression levels of osteogenic biomarkers such as COL1A1 (collagen type I alpha 1 chain) and SPP1 (secreted phosphoprotein 1) under NPWT (Figure 1E and F). We observed similar results in rats 4 weeks after surgery (Fig. S1D-H). These observations suggest that NPWT significantly improves bone regeneration and promotes osteogenesis in vivo.

Figure 1. NPWT accelerates bone regeneration in rat calvarial defects. (A) Micro-CT 3D reconstruction of craniums in the control and NPWT groups 8 weeks after surgery; scale bar: 2 mm. Dotted circle indicates the initial defect region. (B) Quantitative analysis of micro-CT imaging of defect regions (n = 5 in the control group and n = 7 in the NPWT group). (C) Histological staining of craniums from the control and NPWT groups, including H&E, Masson’s trichrome and von Kossa staining; scale bar: 700 µm. (D) ALP staining of craniums from the control and NPWT groups. Dotted squares indicate magnification zones; scale bar: 700 µm (for magnification, scale bar: 50 µm). (E) Immunohistochemistry of COL1A1 and SPP1 in craniums from the control and NPWT groups; scale bar: 50 µm. (F) Quantitative analysis of the immunohistochemistry in E (n = 4 in each). Data are represented as the mean ± SEM. *P < 0.05 versus control; **P < 0.01 versus control; ***P < 0.001 versus the control, by unpaired Student’s t test.

NPWT promotes osteoblast differentiation of rat MSCs

Osteogenic activity is closely related to osteoblast differentiation of MSCs [6,21]. To further confirm the effect of NPWT on osteogenesis, we used customized cell chambers, in which the atmospheric pressure could be decreased to mimic NPWT in vitro (Figure 2A). We isolated primary rat MSCs from bone marrow and the identity was confirmed by flow cytometry with positive staining for MSCs markers, ITGB1/CD29 (integrin subunit beta 1) and THY1/CD90 (Thy-1 cell surface antigen), but negative staining for hematopoietic markers PTPRC/CD45 (protein tyrosine phosphatase receptor type C), ITGAM/CD11B (integrin subunit alpha M) and ITGAX/CD11 C (integrin subunit alpha X) (Fig. S2A-D). Rat MSCs were induced to differentiate into osteoblasts under different NPs. Although no significant differences in ALP activity were observed (Figure 2B), the NP-treated cells exhibited increased calcium deposits measured by increased alizarin red S (ARS) staining compared with control cells (Figure 2C and D). Cells treated with −200 mmHg pressure showed the highest level of ARS staining (Figure 2D). Quantitative RT-PCR verified the expression of multiple osteogenesis-related genes including Col1a1, Spp1, Runx2 (RUNX family transcription factor 2), and Alpl (alkaline phosphatase, biomineralization associated) (Figure 2E), and the upregulation of RUNX2 and SPP1 expression was further confirmed by immunofluorescence (Figure 2F and G). These findings verified the effect of NP on osteoblast differentiation of rat MSCs in vitro.

Figure 2. NPWT promotes osteoblast differentiation of rat MSCs. (A) Illustration of NP treatment in vitro. (B) ALP staining of MSCs after osteoblast differentiation under different pressures for 7 days; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (C) Alizarin Red staining of MSCs after osteoblast differentiation under different pressures for 18 days; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (D) Quantitative analysis of the Alizarin Red staining in C (n = 5 in each). (E) Relative expression levels of osteogenic genes (Runx2, Col1a1, Spp1, Alpl) in MSCs under different pressures for 48 h measured by quantitative RT-PCR (relative to Gapdh). (F-G) Immunofluorescence staining for RUNX2 (F) and SPP1 (G) in MSCs under different pressures for 48 h; scale bar: 50 µm. Data were presented as mean ± SEM. *P < 0.05 versus control; **P < 0.01 versus control; ***P < 0.001 versus control tested by one-way ANOVA followed by Bonferroni’s post hoc test.

Transcriptomic analysis of rat MSCs undergoing NPWT-induced osteoblast differentiation

To reveal the underlying mechanism by which NP promotes osteoblast differentiation in rat MSCs, we performed RNA-seq. Volcano analysis identified 1316 differentially expressed genes after NP treatment, of which 678 genes were upregulated and 638 genes were downregulated (Figure 3A and B). Gene Ontology (GO) analysis revealed a significant enrichment of genes related to extracellular matrix organization and ossification (Figure 3C and D). The enriched items related to the osteogenic effects of NPWT were further subjected to GOChord analysis. The enriched genes were assigned to the following categories: “extracellular structure organization”, “response to mechanical stimulus”, “bone mineralization”, “angiogenesis”, “ossification”, and “mesenchyme development” (Figure 3E). Consistently, the upregulated expression of osteogenesis-promoting genes such as Bglap (bone gamma-carboxyglutamate protein), Col4a3 (collagen type IV alpha 3 chain), Col4a4 (collagen type IV alpha 4 chain) and Col18a1 (collagen type XVIII alpha 1 chain) and the downregulated expression of Ccn1 (cellular communication network factor 1) and Tgfb3 (transforming growth factor beta 3) were verified by quantitative RT-PCR (Figure 3F). These observations suggest that NP treatment systemically induces osteoblast differentiation in rat MSCs.

Figure 3. Transcriptomic analysis of rat MSCs undergoing NP-induced osteoblast differentiation. (A) Volcano map of the RNA-seq data of MSCs in the control versus NP (−200 mmHg treatment for 48 h) groups calculated by DESeq2 (n = 3 in each). (B) Heatmap of differentially expressed genes (DEGs) for MSCs in the Control versus NP groups (n = 3 in each). (C) GOBubble plot of GO enrichment analysis of the DEGs. (D) GOCircle plot of GO enrichment analysis of the DEGs. (E) GOChord plot of GO enrichment analysis, displaying of the relationship between genes and the following terms: “extracellular structure organization”, “response to mechanical stimulus”, “bone mineralization”, “angiogenesis”, “ossification”, and “mesenchyme development”. (F) Verification of DEGs (Bglap, Col4a3, Col4a4, Col18a1. Ccn1 and Tgfb3) in MSCs from the control and NP groups measured by quantitative RT-PCR (n = 3 in each, relative to Gapdh). Data were presented as mean ± SEM. *P < 0.05 versus the control; **P < 0.01 versus the control; ***P < 0.001 versus control tested by unpaired Student’s t test.

AMPK mediates the osteogenic effect of NP

To explore the initial driver of the osteogenic effect of NP treatment, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. We noticed the “AMPK signaling pathway” as AMPK is a key sensor of oxidation-reduction and energy status in cells (Figure 4A) [22,23]. Indeed, immunofluorescence showed increased phosphorylation of PRKAA/AMPKα (protein kinase AMP-activated catalytic subunit alpha) at Thr172 in osteoblast-lineage cells at rat calvarial defects from the NPWT group compared with the control group (Figure 4B). In consistent, the increase of PRKAA phosphorylation was further confirmed by immuinoblotting in NP-treated cells compared with the control cells (Figure 4C). To test whether PRKAA activation is required and sufficient for the effect of NP on osteoblast differentiation, we added PRKAA inhibitor compound C (C.C) or agonist acadesine (AICAR/Aic) to rat MSCs. Apparently, NP effectively promoted PRKAA phosphorylation whereas addition of compound C abrogated the effect of NP on the increase of PRKAA phosphorylation (Figure 4D). In contrast, AICAR effectively stimulated PRKAA phosphorylation and normalized the difference in two groups (Figure 4D). More importantly, compound C effectively abolished the effect of NP on osteogenesis. In contrast, AICAR alone efficiently promoted osteogenesis and further enhanced NP-induced osteogenesis (Figure 4E and F). To verify the specific requirement of PRKAA for NP treatment on osteoblast differentiation, we chose to knock down Prkaa1 (protein kinase AMP-activated catalytic subunit alpha 1) by shRNA since Prkaa1 expression was much higher than Prkaa2 (protein kinase AMP-activated catalytic subunit alpha 2) in our MSCs transcriptome dataset (Fig. S2E). Prkaa1 knockdown (Fig. S2F) abrogated the enhanced osteoblast differentiation by NP treatment evidenced by fewer calcium deposits (Figure 4G and H) and lower expression of osteogenesis-related genes (Fig. S2G).

Figure 4. AMPK mediates the osteogenic effect of NP on osteogenesis. (A) KEGG enrichment analysis of DEGs (KEGG disease pathway filtered). (B) Immunofluorescence staining (frozen section) for p-Thr172-PRKAA, THY1 and DAPI in craniums in control and NPWT-treated rats; scale bar: 25 µm. (C) Western blot analysis of PRKAA activation in MSCs under control and different pressures for 48 h. (D) Western blot analysis of PRKAA activation in MSCs treated with PRKAA inhibitor compound C (C.C, 0.2 µM) or agonist AICAR (Aic, 15 µM) under control and NP treatment (−200 mmHg treated for 48 h). (E) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days with C.C or Aic under control and NP treatment; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (F) Quantitative analysis of the E-panel Alizarin Red staining (n = 4 in each). (G) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days infected with scramble or shPrkaa1 shRNA lentivirus under control and NP treatment; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (H) Quantitative analysis of the G-panel Alizarin Red staining (n = 3 in each). (I) HPLC analysis of ATP, ADP and AMP in MC3T3-E1 cells under control and NP treatment. (J) Quantitative analysis of the HPLC results (n = 3 in each). Data were presented as mean ± SEM. *P< 0.05 versus control; **P< 0.01 versus control; ***P< 0.001; ### P< 0.001 versus NP group tested by unpaired Student’s t test and one-way ANOVA followed by Bonferroni’s post hoc test.

Multiple factors are known to activate PRKAA/AMPKα [24]. We did not detect consistent changes of HIF1A (hypoxia inducible factor 1 subunit alpha) expression in response to NP treatment (Fig. S3A). In addition, neither hypoxia nor hyperoxia consistently increased Hif1a, PRKAA phosphorylation, or osteoblast differentiation in rat MSCs (Fig. S3B and C). Neither did we detect significant differences in reactive oxygen species (ROS) production or calcium influx in the NP-treated cells versus the control cells (Fig. S3D and E). These observations suggest that NP treatment may not be involved in hypoxia or oxidative stress-induced PRKAA activation. Notably, we measured ATP, ADP and AMP products by high performance liquid chromatography (HPLC) in MC3T3-E1 cells, which were widely used as a cell model for osteoblast differentiation [25] and nicely mimicked the effect of NP treatment on osteoblast differentiation in our study (data not shown). We found significant increases in AMP:ATP ratio and (AMP+ADP):ATP ratio in the NP-treated MC3T3-E1 cells compared with the control cells (Figure 4I and J). These results demonstrate that PRKAA activation mediates the effect of NP treatment on osteogenesis and the increase of AMP:ATP ratio and/or (AMP+ADP):ATP ratio may be responsible for PRKAA activation.

Taken together, our findings suggest that NP treatment promote osteoblast differentiation by activating PRKAA signaling.

AMPK activation causes autophagy during NP-induced osteoblast differentiation

AMPK activation alters several biological processes including autophagy [26,27]. Interestingly, gene set enrichment analysis (GSEA) showed that genes upregulated in NP-treated cells are significantly enriched in cellular process of lysosome (Figure 5A) and NP-treated cells did exhibit more autophagosomes than control cells detected by transmission electron microscopy (Figure 5B). Activation of autophagic flux in NP-treated cells was further evidenced by the decreased ratio of MAP1LC3B/LC3B (microtubule associated protein 1 light chain 3 beta) yellow puncta to total LC3B red puncta compared with that of the control cells (Figure 5C and D), as LC3B is involved in expansion of the phagophore and its acidic- and/or proteolytic-sensitive GFP-tagged signal decreases during formation of the autolysosome [28]. Increased autophagy was also confirmed in osteoblast-lineage cells at rat calvarial defects from the NPWT group compared with those from the control group, as measured by immunofluorescence for LC3B and THY1 (Figure 5E). Consistent with this finding, Western blot analysis revealed increases in ATG7 (autophagy related 7), LC3B-I and LC3B-II levels and decreases in SQSTM1 (sequestosome 1) levels in the NP-treated rat MSCs (Figure 5F), which means formation and degradation of autophagosomes [28]. To test whether AMPK activation drives autophagy, we added AMPK inhibitor compound C to the NP-treated cells. Compound C efficiently decreased ATG7, LC3B-I, and LC3B-II and increased SQSTM1 in NP-treated and control to comparable levels (Fig. S3F). These findings demonstrate that PRKAA activation-mediated autophagy is involved in NP-induced osteoblast differentiation.

Figure 5. PRKAA activation causes autophagy during NP-induced osteoblast differentiation. (A) GSEA analysis demonstrated significant enrichment of lysosome-regulated genes in the NP (−200 mmHg treated for 48 h) versus control group. (B) Transmission electron microscopy of MSCs under control and NP treatment. Arrows indicate autophagosomes; scale bar: 1 µm. (C) Autophagic flux detection for MSCs transfected with stubRFP-senseGFP-LC3B under control and NP treatment; scale bar: 25 µm. (D) Quantitative analysis of the autophagic flux in C (n = 3 in each). (E) Immunofluorescence staining (frozen section) for LC3B, THY1 and DAPI in craniums from control and NPWT rats; scale bar: 25 µm. (F) Western blot analysis of autophagic activation in MSCs under control and different pressures for 48 h. (G) Western blot analysis of autophagic activation in the MSCs treated with autophagy inhibitors chloroquine (CQ, 20 µg/ml) or bafilomycin A₁ (BA1, 1.5 µM) under control and NP conditions. (H) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days with CQ or BA1 under control and NP conditions; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (I) Quantitative analysis of the H-panel Alizarin Red staining (n = 4 in each). (J) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days infected with scramble or shAtg7 lentivirus under control and NP treatment; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (K) Quantitative analysis of the J-panel Alizarin Red staining (n = 3 in each). Data were presented as mean ± SEM. *P < 0.05 versus control; **P < 0.01 versus control; ***P < 0.001; ### P < 0.001 versus NP group tested by unpaired Student’s t test and one-way ANOVA followed by Bonferroni’s post hoc test.

To further examine whether autophagy is required for NP-induced osteoblast differentiation, we applied chloroquine (CQ) and bafilomycin A₁ (BA1), two autophagic inhibitors, to treat rat MSCs under NP treatment. These inhibitors obstructed autophagic flux with accumulation of SQSTM1 and LC3B-II (Figure 5G). Notably, these inhibitors completely abolished the promoting effect of NP on osteoblast differentiation (Figure 5H and I). To demonstrate the specific requirement of autophagy in NP-induced osteoblast differentiation, we knocked down Atg7 (Fig. S3G) as a critical factor in autophagy [29] in MSCs. Consistently, Atg7 downregulation abrogated the effect of NP on osteogenesis with fewer calcium deposits (Figure 5J and K) and lower expression of osteogenesis-related genes (Fig. S3H). These findings suggest that PRKAA activation-mediated autophagy is required for the promoting effect of NP on osteoblast differentiation.

AMPK phosphorylation of ULK1 is required for NP-induced autophagy and osteoblast differentiation

To further investigate how AMPK activation may induce autophagy, we measured the phosphorylation of ULK1 and MTOR (mechanistic target of rapamycin kinase), two essential factors downstream of AMPK signaling promoting autophagy [26,27]. The phosphorylation of ULK1 at Ser555 was significantly increased in osteoblast-lineage cells at rat calvarial defects from the NPWT group compared with that from control rats (Figure 6A). The increased level of p-Ser555-ULK1 was confirmed in NP-treated rat MSCs (Figure 6B). No significant increase in the phosphorylation of MTOR or its downstream factors was observed (Fig. S4A and B). Neither was significant change of p-Ser757-ULK1 observed (Fig. S4C and D). Furthermore, coimmunoprecipitation showed the interaction of PRKAA and ULK1 (Figure 6C). To test whether ULK1 mediates the effect of NP, we applied an ULK1 inhibitor SBI-0206965 (SBI) or agonist BL-918 (BL) with NP treatment. Expectedly, the activation of PRKAA by NP was not affected by adding SBI-0206965 or BL-918. Noticeably, SBI-0206965 potently suppressed ULK1 phosphorylation and formation of LC3B-II level induced by NP-treatment (Figure 6D). In contrast, BL-918 alone maintained high levels of ULK1 phosphorylation and LC3B-II production and further enhanced the effect of NP treatment on autophagy (Figure 6D). Consistently, SBI-0206965 blocked the osteogenic effect of NP whereas BL-918 further enhanced osteoblast differentiation by NP treatment (Figure 6E and F). Specifically, Ulk1 knockdown by shRNA (Fig. S4E) abrogated the enhanced osteogenesis by NP treatment evidenced by fewer calcium deposits compared with scramble cells (Figure 6G and H) and lower expression of osteogenesis-related genes (Fig. S4F). More importantly, SBI-0206965 efficiently antagonized ULK1 activation and LC3B-II formation by PRKAA agonist AICAR (Figure 6I) and effectively abrogated the effect of NP and AICAR on osteogenesis (Figure 6J and K). These observations suggest that ULK1 may mediate the effect of NPWT downstream of PRKAA on osteoblast differentiation.

Figure 6. ULK1 phosphorylation by AMPK is required for NP-induced autophagy and osteoblast differentiation. (A) Immunofluorescence staining (frozen section) for p-Ser555-ULK1, THY1 and DAPI in craniums from the control and NPWT rats; scale bar: 25 µm. (B) Western blot analysis of ULK1 activation in MSCs under control and different pressures for 48 h. (C) MSCs were treated under control and NP conditions (−200 mmHg treated for 48 h). Immunoprecipitation (IP) was performed with PRKAA antibody or pre-immune IgG. Immunoblotting (IB) was carried out to detect PRKAA and ULK1. (D) Western blot analysis of AMPK-ULK1/autophagic activation in MSCs treated with ULK1 inhibitor SBI-0206965 (SBI, 1 µM) or agonist BL-918 (BL, 15 µM) under control and NP treatment. (E) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days with SBI or BL under control and NP treatment; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (F) Quantitative analysis of the E-panel Alizarin Red staining (n = 4 in each). (G) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days infected with scramble or shUlk1 lentivirus under control and NP treatment; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (H) Quantitative analysis of the G-panel Alizarin Red staining (n = 3 in each). (I) Western blot analysis of AMPK-ULK1-autophagy activation in MSCs with an PRKAA agonist AICAR (Aic, 15 µM) or/and ULK1 inhibitor SBI-0206965 (SBI, 1 µM) under control and NP treatment. (J) Alizarin Red staining of MSCs after osteoblast differentiation for 18 days with Aic or/and SBI under control and NP conditions; scale bar: 2 mm (squares: magnification zones, scale bar: 250 µm). (K) Quantitative analysis of the J-panel Alizarin Red staining (n = 4 in each). Data were presented as mean ± SEM. *P < 0.05 versus control; **P < 0.01 versus control; ***P < 0.001; ### P < 0.001 versus NP group tested by one-way ANOVA followed by Bonferroni’s post hoc test.

Suppression of the AMPK-ULK1-autophagy axis in patients with nonunion

To address whether our findings were of clinical relevance, we collected 12 samples from patients with nonunion. Eight samples of new calli attached to internal fixations were also collected from patients with successful fracture healing and served as controls (Table S1 for patient information). Our samples from both control and nonunion patients showed regenerated mesenchymal tissues (Figure 7A). Moreover, ALP and Masson’s trichrome staining revealed a decrease in osteogenic potential and osseous collagen formation in nonunion patients (Figure 7B and C). The decreased osteogenic potential in nonunion patients was also evidenced by decreased expression of osteogenic markers including COL1A1, SPP1 and BGLAP detected by immunohistochemistry (Figure 7D). Consistently, decreases in p-Thr172-PRKAA, p-Ser555-ULK1, and LC3B were observed in the nonunion samples compared with control samples and no significant changes in p-Ser2448-MTOR were observed in these two groups (Figure 7E and F). These observations suggest that suppression of AMPK-ULK1-autophagy axis is tightly correlated to nonunion conditions.

Figure 7. Suppression of the AMPK-ULK1-autophagy axis is relevant to patients with nonunion. (A) Representative H&E staining from control and nonunion groups; scale bar: 300 µm (squares: magnification zones, scale bar: 100 µm). (B) Representative ALP staining from control and nonunion groups; scale bar: 300 µm (squares: magnification zones, scale bar: 100 µm). (C) Representative Masson’s trichrome staining from control and nonunion groups; scale bar: 300 µm (squares: magnification zones, scale bar: 100 µm). (D) Representative immunohistochemistry of BGLAP, COL1A1, and SPP1 in mesenchymal tissues from the control and nonunion groups; scale bar: 50 µm. (E) Representative immunohistochemistry of p-Thr172-PRKAA, p-Ser2448-MTOR, p-Ser555-ULK1, and LC3B from the control and nonunion groups; scale bar: 50 µm. (F) Quantitative analysis of the immunohistochemistry in E (n = 8 in the control group and n = 12 in the nonunion group). Data were presented as mean ± SEM. *P < 0.05 versus control; **P < 0.01 versus control; ***P < 0.001 versus control tested by unpaired Student’s t test.

Combinatory therapy with NPWT and MSCs transplantation promotes calvarial defect regeneration in vivo

To explore the translational application of our findings, we carried out a combinatory therapy of NPWT and MSC transplantation on rat bilateral critical-size calvarial defects (5 mm diameter). Gelatin methacryloyl hydrogel (GelMA) carrying rat MSCs (GelMA+MSCs) were crosslinked by ultraviolet rays during surgery followed by NPWT for 7 days after surgery. After 4 and 8 weeks, micro-CT scanning analysis revealed that MSCs (GelMA+MSCs) alone or NPWT alone (GelMA NPWT group) improved bone regeneration in defects compared with that in the GelMA control group. Moreover, combinatory therapy (GelMA+MSCs NPWT group) exhibited the best efficacy (Figure 8A and B). However, the promoting effect of the combinatory therapy was completely abrogated by PRKAA inhibitor compound C (GelMA+MSCs+C.C NPWT group) (Figure 8A and B). Quantitative analysis revealed that the combinatory therapy group had the highest BV/TV and BS/TV values (Figure 8B). Since rats from GelMA groups failed to regenerate newborn tissues, these groups were excluded in further analysis. Histological analyses including H&E and Masson’s trichrome staining showed that the most osteoid and newly formed bone tissues were observed in the defects of the combinatory therapy group (GelMA+MSCs NPWT group) (Figure 8C). ALP staining revealed the highest osteogenic potential in the combinatory therapy group (Figure 8C). We observed similar phenotypes in rats 4 weeks after surgery (Fig. S5A-C). Consistently, the combinatory therapy group also exhibited the highest levels of osteogenic biomarkers such as COL1A1 and SPP1, and highest levels of autophagic biomarkers including p-Thr172-PRKAA, p-Ser555-ULK1, and LC3B in rats 4 and 8 weeks after surgery (Figure 8D, Fig. S5D, and Fig. S6A and B). Again, the enhanced osteogenic capacity and autophagy induced by NPWT were abrogated by compound C (Figure 8D, Fig. S5D, and Fig. S6A and B). These observations demonstrate that the combinatory therapy of NPWT and MSCs transplantation has the potential to treat critical-size bone defects. NPWT improves bone regeneration by promoting osteoblast differentiation via the AMPK-ULK1-autophagy axis in vivo.

Figure 8. Combinatory therapy with NPWT and MSC transplantation promotes calvarial defect regeneration in vivo. (A) Micro-CT 3D reconstruction of craniums treated by GelMA, GelMA+MSCs, and GelMA+MSCs+C.C (compound C) with or without NPWT 8 weeks after surgery; scale bar: 2 mm. Doted circles indicate the initial defect regions. (B) Quantitative analysis of micro-CT imaging of defect regions in A (n = 6 in each group except n = 5 in the GelMA+MSCs+C.C under control and NPWT groups). (C) Histological analysis of craniums from each group, including H&E, Masson’s trichrome staining and ALP staining; scale bar: 500 µm. (D) Quantitative analysis of the immunohistochemistry of COL1A1, SPP1, p-Thr172-PRKAA, p-Ser555-ULK1, and LC3B in each group (n = 5 in each). Data were presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 tested by one-way ANOVA followed by Bonferroni’s post hoc test.

Discussion

This study was designed to test whether NPWT promoted bone defect healing and investigate its underlying mechanisms. Our experiments showed that NPWT could accelerate bone regeneration in rat calvarial defects and significantly improve osteogenic activity. To mimic NPWT, we used NP treatment on rat MSCs in vitro and dissected the underlying mechanism. The impaired activation of AMPK-ULK-autophagy axis was verified in human nonunion samples. Based on these findings, we believe that NP treatment in vitro faithfully mimics many features of in vivo microenvironment in NPWT treatment for rat critical-size calvarial model. Finally, we carried out a combinatory therapy of NPWT and MSCs transplantation on rat critical-size calvarial defects and showed significant improvement in bone regeneration.

Although the biological effect of NPWT in multiple types of cells has been reported in several studies and NP has been reported to promote osteoblast differentiation in bone marrow MSCs [30], its underlying mechanism remains unknown and the in vivo function was not addressed before. While the AMPK-ULK1-autophagy axis was already studied in the process of autophagy and reported in other fields, its role in promoting osteogenesis and its clinical relevance to bone nonunion were not previously reported. AMPK is a heterotrimeric serine-threonine protein kinase complex composed of α, β, and γ subunits [31] and is known to have multiple functions in many pathophysiological conditions [22,23]. PRKAA was activated by several factors, such as glucose, AMP:ATP ratio, and ROS, has been well defined [31–33]. In our study, NP did not significantly alter the levels of intracellular ROS, calcium, or HIF1A (Fig. S3A, D and E). Neither hypoxia nor hyperoxia did consistently promote osteoblast differentiation (Fig. S3C). Instead, we found that the ratio of AMP:ATP ratio and (AMP+ADP):ATP ratio significantly increased after NP treatment (Figure 4I and J), suggesting that the increased AMP:ATP ratio may be responsible for PRKAA activation in NPWT. NWPT potentially affects energy metabolism involving glycolysis and oxidative phosphorylation in MSCs. How NPWT is linked to energy metabolism remains unclear and is worthy of further investigation.

We also revealed that PRKAA activation in osteoblast differentiation caused autophagy through ULK1 rather than MTOR, both of which may mediate the effect of AMPK [26]. Autophagy, a major catabolic process in eukaryotic cells for the degradation and recycling of damaged organelles and macromolecules, is enhanced during osteoblast differentiation and mineralization and plays an essential role in bone homeostasis [34]. PRKAA stimulates autophagy through MTOR and ULK1 [26,27]. Moreover, TSC2 and RPTOR play a role in the interaction between AMPK and MTOR [35,36]. In this research, the phosphorylation levels of TSC2 at Thr1462, RPTOR at Ser792, and MTOR at Ser2448 were not significantly changed under NP treatment (Fig. S4A and B) suggesting that MTOR may not contribute to NP-induced autophagy. Consistently, the phosphorylation level of ULK1 at Ser757, which can be modified by MTOR, resulting in a decrease in ULK1 activity, was not significantly changed (Fig. S4C and D). Instead, phosphorylation at Ser555, which is modified by PRKAA and increases in ULK1 activity [26,27,37], was enhanced after NP treatment. Moreover, the coimmunoprecipitation results revealed that AMPK and ULK1 interacted under NP. By using ULK1 inhibitor, shRNA and agonist, we demonstrate that AMPK-ULK1 is required for triggering NP-induced autophagy and osteoblast differentiation.

Most importantly, our findings are of clinic relevance that the underlying mechanisms potentially provide novel therapeutic strategy for clinical treatment. Patients with nonunion showed a decrease in osteogenic activity [38], and multiple risk factors, including medical comorbidities, age, and the characteristics of the injury [39], have been identified. In this research, we detected suppression of autophagy in nonunion tissues with decreasing levels of p-Thr172-PRKAA, p-Ser555-ULK1, and LC3B, which is consistent to our findings. These observations suggest that PRKAA-induced autophagy is related to osteogenic activity in the clinic. In present, a second surgery with autologous bone grafts is the gold standard for the treatment of nonunion, which potentially cause secondary injury. Moreover, various approaches including stem cell therapy have been applied [40]. To investigate whether NPWT could be applied to treat nonunion, we chose a rat critical-size calvarial defect model with MSCs embedded in the hydrogel to simulate the clinical scenario. Our results demonstrate that NPWT combined with MSC transplantation has the potential to treat critical-size bone defects and that autophagy induced by PRKAA activation plays a critical role in this process. Since secondary damage of autologous bone graft transplantation is inevitable, stem cell therapy with biomaterials can be a good replacement [40].

There are several limitations in this study. Firstly, genetic models were not available to confirm the axis in NPWT due to poor development of Cre-loxP system in rats. Using inhibitors may not exclude the off-target effect. To overcome this, we used shRNA technique to specifically knock down each gene in vitro and reproduce phenotypes achieved by inhibitors. Secondly, the therapeutic effect of the combinatory therapy of NPWT and MSCs transplantation remains to be tested in bone defects in large animals such as dog before it can be clinically applied.

In conclusion, we reveal that NPWT can promote bone regeneration in vivo and in vitro via the AMPK-ULK1-autophagy axis, and combinatory therapy of NPWT and MSC transplantation with hydrogels can promote critical-size bone defect healing. Our study potentially provides a new strategy to improve the treatment of nonunion fractures.

Materials and methods

Rat models

Male Sprague-Dawley rats (8 weeks old) were purchased from Vital River Laboratories. The calvarial defects were assessed using previously validated protocol [41]. The size of the bone defect was adjusted in this protocol to the design of the study. Briefly, after deep anesthesia with isoflurane gas (2%–3% inhaled) was confirmed, rats were shaved and disinfected from the bridge of the snout between the eyes to the caudal end of the calvarium. Then, a 1.5 cm incision down to the periosteum over the scalp from the nasal bone to just caudal to the middle sagittal crest was made. Round calvarial defects at the center of parietal bones were carefully conducted with a trephine under dropwise irrigation with sterile normal saline. For evaluating the effect of NPWT on calvarial defects without padding, a single 4-mm diameter defect was generated on the left side. For experiments evaluating NPWT combined with MSC transplantation, bilateral critical-sized 5 mm diameter defects were generated, and each defect was filled with GelMA (Yongqinquan Intelligent Equipment Co., Ltd, EFL-GM-90) embedded with MSCs (5 × 10⁴ cells per gel) and inhibitors. Next, a customized 8 × 6 × 4 mm foam (VSD Medical Technology Co., Ltd, customized) was placed and sutured with cranial fascia. Then, the skin was closed above the foam and covered by a vacuum-assisted closure device with constant negative pressure values at continuous −200 mm Hg (Fig. S1B and C). Rats in the control group were connected to the machine without NP. NPWT treatment lasted for 7 days, and the foams and devices were removed by secondary surgery. Rats were sacrificed at 4 and 8 weeks after surgery for the subsequent experiments. All animal procedures were carried out in accordance with protocols approved by the Wuhan University Institutional Animal Care and Use Committee.

Human samples

Bone nonunion samples were obtained from twelve patients at their secondary surgery for nonunion. Eight patients with successful fracture healing served as controls. New calluses attached to the internal fixations were collected at their secondary surgery to remove their internal fixations. The human study of this research was conducted in accordance with the principles expressed in the Declaration of Helsinki and was approved by the ethical committee of the Medical Ethical Committee of Zhongnan Hospital of Wuhan University (approval number: 2,021,013). Written informed consent was obtained from each enrolled patient. The patient information of clinical samples was provided in Table S1.

Micro-CT imaging and analysis

After rats were euthanized with CO₂, craniums were dissected free of skin and evaluated using a high-resolution micro-CT imaging system (Bruker, SkyScan 1176, USA). Each cranium was scanned separately at 55 kV and 385 μA with a 1.0-mm aluminum filter to obtain a 9-μm voxel size. NRecon (Bruker) was used to reconstruct images and quantitative analysis was performed using CTAn (Bruker) in accordance with the recommendations of the American Society for Bone and Mineral Research [42]. The entire initial round defect was used for analysis

Isolation and culture of rat bone MSCs

Primary MSCs were harvested from Sprague-Dawley rats (4 weeks old) as previously described [43]. Briefly, mechanically dissociated diaphyses of femurs and tibiae were digested with collagenase II (1 mg/ml; Gibco, 17,101,015) and dispase II (2 mg/ml; Millipore Sigma, D4693). Cells were collected and cultured in α-MEM medium (HyClone, SH30265.01B) supplemented with 10% fetal bovine serum (Gibco, 10,099,141), 1% penicillin-streptomycin (Procell, PB180122), and 2 mM L-glutamine (Procell, PB180420) at 37°C. Adherent cells were harvested at passage 3 for flow cytometry (Beckman, Cytoflex, USA) by staining cells with fluorescence-labeled antibodies to ITGB1 (Invitrogen, 11–0291-80), THY1 (Invitrogen, 11–0900-81), ITGAM and ITGAX (Invitrogen, 12–0110-82) and PTPRC (Invitrogen, 11–0461-80). Corresponding isotype controls (Invitrogen, 11–4724-81 and 12–4724-82) were applied in the flow cytometry. MSCs from passages 3 to 5 were used for experiments in this study.

Cell line

Mouse preosteoblast cell line MC3T3-E1 subclone 14 was purchased from ATCC (CRL-2594™) and maintained in α-MEM medium without ascorbic acid (Gibco, A1049001) supplemented with 10% fetal bovine serum (Gibco, 10,099,141) and 1% penicillin-streptomycin (Procell, PB180122). Cells were cultured at 37°C with 5% CO₂. This cell line has been identified with high osteogenic differentiating potential, and it can mimic multiple aspects of osteoblast differentiation such as synthesis of a extracellular matrix, formation mineral deposition within the extracellular matrix, and expression of differentiation markers under osteoblast differentiation induction [25].

Negative pressure (NP) treatment in vitro

MSCs were cultured in customized airtight chambers with air inlet and outlet channels placed in a CO₂ cell incubator. In NP treatment experiments, 1.5 × 10⁶ cells in a 10-cm dish, 2 × 10⁵ cells in 6-well plates, or 4 × 10⁴ cells in 24-well plates were seeded. The air outlet channel was connected to a vacuum pump (VSD Medical Technology Co., Ltd, customized, China) to generate continuous NP. The air inlet channel was installed in a valve to adjust the air exchange. The NP was controlled and adjusted by the vacuum pump with a pressure sensor. The pressure and time of NP treatment were described in Figure legends for each experiment.

In vitro osteoblast differentiation

For osteoblast differentiation, cells were switched to α-MEM supplemented with 10% fetal bovine serum (Gibco, 10,099,141), 10 nM dexamethasone (Sigma-Aldrich, D2915), 10 mM β-glycerolphosphate (Sigma-Aldrich, G6251), and 50 µM L-ascorbic acid (Sigma-Aldrich, A4403). After staining with alizarin red S (1% solution in water, pH 4.2; Sigma-Aldrich, A5533) and ALP (Beyotime, C3206), whole-well images were captured with a camera (Nikon, D100, Japan) and microscopy images were obtained using a 4× phase-contrast objective on optical microscopy (Olympus, IX73, Japan). For the quantitative analysis of Alizarin Red, the stained cultures were destained with acetylpyridinium chloride (100 mM; Sigma-Aldrich, C9002) for 1 h at room temperature. The absorbance of the released stain was measured at 562 nm with microplate reader (PerkinElmer, EnSpire, USA).

Histological analysis

After assessed with micro-CT scanning, craniums were fixed in 4% paraformaldehyde (Servicebio, G1101) for 48 h at 4°C (except for frozen sections). Craniums were then cut into sections (4 μm thickness) and calcium deposits in the bone tissue were visualized by von Kossa staining using 4% silver nitrate (Servicebio, G1043) followed by hematoxylin-eosin (H&E; Servicebio, G1005) counterstaining. For paraffin sectioning, samples were decalcified with EDTA solutions (0.5 M, pH 7.4; Sigma-Aldrich, E5134) at 4°C as previously described [44]. Then, sections (8 μm thickness) were prepared and stained with H&E and ALP (Beyotime, C3206). For frozen sectioning, the time of fixation was shortened to 6 h, and decalcification was performed [44]. Sections (20-μm thickness) were prepared by using a cryostat (Leica, CM3050S, Germany).

Immunohistochemistry and immunofluorescence

Immunohistochemical staining was performed according to a previously described protocol with modifications [45]. Antigen retrieval was performed with citrate buffer solution (pH 6.0) at 95°C for 5 min with three repeats. After incubation with primary antibodies, the corresponding biotinylated secondary antibodies were incubated using a polink-2 plus polymer horseradish peroxidase detection system (Zhongshan Biotechnologies, PV6001), and 3,3-Diaminobenzidine tetrahydrochloride (Zhongshan Biotechnologies, ZLI-9017) was used as a substrate. The nuclei were counterstained using hematoxylin. Immunochemistry was performed using antibodies against p-Thr172-PRKAA (Cell Signaling Technology, 2535), p-Ser555-ULK1 (Cell Signaling Technology, 5869), LC3B (Cell Signaling Technology, 3868), p-Ser2448-MTOR (Cell Signaling Technology, 5536), COL1A1 (Affinity Biosciences, AF7001), and SPP1 (Proteintech group, 22,952-1-AP). Immunohistochemical staining were captured and analyzed by semi-automatic histology system (Leica, Aperio VERSA 8, Germany).

Immunofluorescence staining was performed following a previous protocol with modifications [44]. After incubation with primary antibodies and fluorescence-labeled secondary antibodies (Abcam, ab150113 and ab150115) and mounting with DAPI (Abcam, ab104139), sections were captured with confocal microscopy (Leica, SP8, Germany). For cells in 24-wells, samples were fixed in 4% paraformaldehyde (Servicebio, G1101) for 30 min at room temperature, a protocol similar to that for the frozen sections was followed, and images were captured with a microscopy (Olympus, IX73, Japan). Immunofluorescence was performed using the same primary antibodies in immunohistochemistry expect antibodies to THY1 (Santa Cruz Biotechnology, sc-53,456) and RUNX2 (Cell Signaling Technology, 12,556).

Transcriptomic analysis

Primary MSCs were cultured in complete growth medium to confluence and treated with negative pressure or control for 48 h. Cells were harvested and total RNA was prepared using GenCatch TM Total RNA Extraction Kit (Epoch Life Sciences, 1,660,050) for RNA-seq analysis following the manufacturer’s instructions (Illumina, Hiseq 2500, USA). Sequencing reads were mapped to the rat genome rn6 using HISAT2 [46] and tag counts were summarized at the gene level using SAMtools [47], which allowed only one read per position per length. Differentially expressed genes (DEGs) were determined by DESeq2 [48]. DEGs were used for heatmap analysis using R package pheatmap (https://github.com/raivokolde/pheatmap). Gene set enrichment analysis, including GO and KEGG was performed using DEGs with clusterProfiler [49]. GO analysis was visualized by GOplot [50]. RNA-seq data in this research are available in SRA with BioProject accession number PRJNA732669 (https://www.ncbi.nlm.nih.gov/sra/PRJNA732669).

HPLC for ATP, ADP, and AMP

Control and NP-treated MC3T3-E1 cells were collected and disrupted by sonication to make homogenates (1 × 10⁷ cells per sample). HPLC analysis with Ultimate Plus C18 Column (Welch, 00208–31,043, China) was carried out on Agilent 1260 Infinity II (Agilent Technologies Inc, USA). The chromatogram collections and the integration of compounds were processed by Chemstation (Agilent Technologies Inc, USA).

Transmission electron microscopy

Control or NP-treated MSCs were fixed with 2.5% glutaraldehyde in 0.1 M sodium dihydrogen phosphatase (Servicebio, G1102) for 4 h at 4°C and were then fixed with 1% osmium tetroxide for 1 h at room temperature followed by dehydration using graded ethanol solutions and gradual infiltration with EMbed 812 epoxy resin (Ted Pella, 14,120). Ultra-thin sections (60–80 nm) were made by ultramicrotome (Leica, EM UC7, Germany), and were contrasted with uranyl acetate and lead citrate. The sections were observed under a transmission electron microscopy (Hitachi, HT-7800, Japan). Autophagosomes were identified as previously described [28].

MSCs transfection

Lentivirus expressing shRNA targeting Prkaa1, Atg7 and Ulk1 was purchased from Shanghai Genechem. Lentivirus were transfected to rat MSCs with the aid of polybrene (Solarbio, H8761). MSCs were harvested or performed treatments after 48 h of transfection.

Autophagy flux detection

Lentivirus expressing stubRFP-senseGFP-LC3B was purchased from Shanghai Genechem. After infections, MSCs were treated with negative pressure and control conditions for 48 h. Cells were fixed with 4% paraformaldehyde (Servicebio, G1101) for 30 min at room temperature and nuclei were counterstained with DAPI (MilliporeSigma, D9542). Immunofluorescence were observed and captured with a Leica SP8 Confocal microscopy as previously described [28].

Total RNA preparation and quantitative RT-PCR analysis

Total RNA was extracted from cultured MSCs using TRIzol (Thermo Fisher, 15,596,026) following the instructions of the manufacture. An aliquot of 400 ng total RNA was reverse-transcribed into cDNA with the reverse transcriptase kit (Vazyme, R223). Quantitative PCR was performed using a SYBR green mixture (Vazyme, Q311) and a Monad Real-Time PCR instrument (Monad, q225, China). Primers used for specific transcripts were listed in Table S2.

Co-Immunoprecipitation and Western blot

MSCs (4 × 10⁶ cells per sample) were collected and lysed in ice-cold cell lysis buffer for western and IP (Beyotime, P0013) containing both a protease inhibitor cocktail (MedChemExpress, HY-K0010) and a phosphatase inhibitor cocktail (MedChemExpress, HY-K0023). Cell lysates (1%) were preserved as inputs. PRKAA antibody (Cell Signaling Technology, 5831; diluted at 1:50) and Protein A-G Magnetic Beads (MedChemExpress, HY-K0202) was used to perform immunoprecipitation. Western blot was performed as previously described [51] with primary antibodies and horseradish peroxidase-linked secondary antibody (Cell Signaling Technology, 7074). Images were acquired with enhanced chemiluminescent imaging system (Tanon, 4600, China) without gamma-adjustment in default parameters. Uncropped Western blots involved in this research were summarized in Data S1. Antibodies to PRKAA (5831), p-Thr172-PRKAA (2535), ULK1 (8054), p-Ser555-ULK1 (5869), p-Ser757-ULK1 (14,202), MTOR (2983), p-Ser2448-MTOR (5536), RPTOR (2280), p-Ser792- RPTOR (2083), TSC2 (4308), p-Thr1462-TSC2 (3617), p-Thr389-RPS6KB1 (9234), p-Thr37/46-EIF4EBP1 (2855), and LC3B (3868) were purchased from Cell Signaling Technology. Antibodies to SQSTM1 (18,420-1-AP), ATG7 (10,088-2-AP), and GAPDH (10,494-1-AP) were purchased from Proteintech Group. Antibody to HIF1A (AF1009) was purchased from Affinity Biosciences.

Statistical analysis

Quantitative data were presented as the mean ± SEM, with P values of less than 0.05 considered significant. Individual data points were shown, and the number of samples or images analyzed was indicated in the figures and/or legends. Parametric data were analyzed using the appropriate Student’s t test when 2 groups were compared or a one-way ANOVA when more than 2 groups were compared followed by Bonferroni multiple comparisons post hoc test as indicated in the figure legends. All statistical tests were performed using Prism 8.0 software (USA).

Acknowledgments

We thank Z. Hou and W. Sun for providing advice on the rat model; We also thank Y. Huang for helping with RNA-seq analysis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [81870162]; National Natural Science Foundation of China [81870427]; National Natural Science Foundation of China [81702150].