Aging accelerates APAP-ALI and liver inflammation

We first examined the effect of aging on APAP-ALI. The analysis revealed that aging significantly aggravates liver injury induced by APAP, as manifested by increased serum ALT and AST levels (Fig. 1A) and increases in necrotic tissue areas, as indicated by H&E staining (Fig. 1B). Moreover, the number of the terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labelling (TUNEL) positive cells was significantly increased in liver tissue sections of aged mice (Fig. 1C). In addition, increased MΦ infiltration was found in the livers of aged mice after APAP-ALI had been induced (Fig. 1D). Consistent with these observations, aging promoted the expression of proinflammatory cytokines, such as SASP components (Il1b, Il6, and Tnfa), and decreased the expression of the anti-inflammatory cytokine Il10 in the liver (Fig. 1E). The severity of APAP-induced liver injury was dose dependent. In a survival study, the mortality was significantly increased in the aged mice that received a higher dose of APAP (500 mg/kg) (Fig. 1F). These results indicate that aging greatly exacerbates APAP-ALI and this increase in ALI severity is accompanied by increased infiltration of MΦs and higher proinflammatory cytokine expression.

Fig. 1

Aging accelerates APAP-ALI and inflammation. Young and aged mice were administered APAP (300 mg/kg) or PBS and analyzed 24 h later. A Serum ALT and AST levels of the young and aged mice treated with PBS or APAP. B Representative images showing H&E staining liver slices from young and aged mice treated with PBS or APAP. C Representative images showing TUNEL staining in the liver slices from young and aged mice treated with PBS or APAP. Quantitative data are shown in the right panel. D Representative images showing IHC staining for F4/80 in liver slices from young and aged mice treated with PBS or APAP. Quantitative data are shown in the right panel. E mRNA expression (Il1b, Il6, Tnfa, and Il10) in the livers from young and aged mice treated with PBS or APAP as measured by RT‒qPCR. The average target gene/Gapdh ratios of different experimental groups relative to the control group are reported. F Survival curves of young and aged mice treated with APAP when the dose of APAP was increased to 500 mg/kg. *p < 0.05, **p < 0.01, and ***p < 0.001

BMP9 promotes hepatocyte injury and senescence

RT‒qPCR- and immunoblot-based measurements revealed that the expression of both Bmp9 and Cebpa was increased in the liver tissue of aged mice (Fig. 2A, B). Since BMP9 is a secreted protein and enriched in the aged liver, we used mouse BMP9 recombinant protein (Rm-BMP9) to investigate the effect of BMP9 on APAP-treated hepatocytes. With prolonged APAP treatment in vitro, the ratio of dead-to-live hepatocytes after Rm-BMP9 treatment was significantly higher than that after APAP treatment alone, indicating that BMP9 might play a direct role in promoting hepatocyte injury induced by APAP (Fig. 2C). We further explored the mechanism by which BMP9 aggravates hepatocyte injury in vitro. Increased phosphorylation of SMAD1/5/9, representing the activation of BMP signaling, suggested that BMP9 may have activated downstream signaling cascades in hepatocytes after APAP-induced injury (Fig. 2D). Additionally, the activation of SMAD1/5/9 was increased in the livers of the aged mice, and the phosphorylation rates of SMAD1/5/9 were decreased in the livers of aged Bmp9-knockout (Bmp9−/−) mice (Additional file 1: Figure S1A).

Fig. 2

BMP9 promotes hepatocyte injury and senescence. A, B mRNA (A) and protein (B) expression in the livers from young and aged mice. (C–E) Primary hepatocytes with or without Rm-BMP9 addition were treated with APAP, and 1 h, 3 h, 6 h, 12 h and 24 h later, we measured the proportion of APAP-induced hepatocyte death to normal growing primary hepatocytes (C), the protein expression levels of SMAD1/5/9 and P-SMAD1/5/9 (D), and the protein expression levels of ATG3, ATG7, p62 and LC3 I/II (E). F, G Rm-BMP9 was injected into the tail vein, and the APAP-ALI model (300 mg/kg) was established 1 h later. Serum and liver tissue were collected from each group of mice 24 h after model establishment. We measured the serum ALT and AST levels (F), and representative images showing H&E staining (G) in each group are shown. H, I APAP-ALI model (300 mg/kg) was established with young and aged WT mice, as well as with young and aged Bmp9−/− mice. Serum and liver tissue were collected from each group of mice 24 h later. We measured the serum ALT and AST levels (H), and representative images showing H&E staining (I) in each group are shown. The average target gene/Gapdh ratios of different experimental groups to the control group are reported. *p < 0.05, **p < 0.01, and ***p < 0.001

Autophagy-related 3 (ATG3, with E2-like enzyme activity) and ATG7 (with E1-like enzyme activity) are key molecules in autophagy, mediating the transformation of microtubule-associated protein 1 light chain 3 II (MAP1LC3 II/LC3 II) and autophagosome maturation [17, 18]. Notably, Rm-BMP9 downregulated ATG3 and ATG7 expression and significantly decreased the transformation of LC3 I to LC3 II in APAP-injured hepatocytes, results that were consistent with the accumulation of SQSTM1/p62 (Fig. 2E).

These results illustrate the aggravating effect of BMP9 on aging-associated APAP-ALI in vitro, and therefore, we further investigated the effect of BMP9 on APAP-ALI in vivo by injecting Rm-BMP9 into mice. The serum ALT/AST levels (Fig. 2F) and H&E staining results (Fig. 2G) indicated significantly increased APAP-ALI severity, and the number of TUNEL- stained cells was increased (Additional file 1: Figure S1B). Then, we found that Rm-BMP9 promoted the expression of Ilb, Il6 and Tnfa in the mouse livers (Additional file 1: Figure S1C). Consistent with the results obtained from exogenous supplementation of BMP9 in mice, Bmp9 deletion reduced the severity of the liver injury in both young and aged mice (Fig. 2H, I). Additionally, the number of TUNEL-stained cells in the livers of young and aged mice treated with APAP was significantly decreased in the Bmp9−/− groups (Additional file 1: Figure S1D). RT‒qPCR assays revealed that Bmp9 deletion reduced the expression of proinflammatory factors (Ilb, Il6 and Tnfa) in liver tissue (Additional file 1: Figure S1E).

These results suggest that BMP9 inhibits hepatocyte autophagy through ATG3 and ATG7, thereby regulating the severity of APAP-ALI. The high level of BMP9 protein associated with aging may be an important factor in the aggravation of APAP-ALI during aging.

The expression of BMP9 is regulated by C/EBPα in vitro

Because the cell type that expresses BMP9 has not been clarified in previous studies, we sought to visualize the expression and distribution of BMP9 in aged livers. Therefore, we performed immunofluorescence (IF) staining for F4/80 and BMP9 in liver tissues of young and aged mice. In uninjured livers, BMP9 was mainly localized to F4/80-positive cells, and BMP9 expression was higher in aged mouse livers (Fig. 3A). In mice with APAP-ALI, BMP9 accumulated in injured area, BMP9 expression localized to F4/80-positive cells was increased, and BMP9 expression, in general, was identified in hepatocytes (Fig. 3A). Then, we isolated hepatocytes and hepatic MΦs from young and aged mice. We found that BMP9 expression was markedly upregulated in both the hepatocytes and hepatic MΦs from the aged mice (Fig. 3B, C). Notably, we also detected increased C/EBPα and decreased ATG3 and ATG7 levels in the isolated hepatocytes and hepatic MΦs from the aged mice (Fig. 3B, C).

Fig. 3

BMP9 expression is regulated by C/EBPα in vitro. A IF staining for BMP9 and F4/80 in liver slices from young and aged mice treated with PBS or APAP (300 mg/kg) for 24 h. B, C Primary hepatocytes and MΦs were isolated from young and aged mice by in situ perfusion of mouse livers. Protein expression levels of BMP9, C/EBPα, ATG3 and ATG7 in mouse primary hepatocytes (B) and hepatic MΦs (C). D–G To study the regulation of BMP9 by C/EBPα in MΦs, we altered Cebpa expression in iBMDMs. D Protein expression levels of C/EBPα and BMP9 after overexpressing Cebpa. E Protein expression levels of C/EBPα and BMP9 after inhibiting Cebpa with or without treatment with the APAP-treated AML-12 cell supernatant (Sup). F Protein expression levels of CD206 and iNOS after overexpressing Cebpa with or without Sup treatment. G mRNA expression of related cytokines (Il1b, Il6 and Tnfa) after overexpressing Cebpa with or without Sup treatment. (H, I) To study the regulation of BMP9 by C/EBPα in hepatocytes, we altered Cebpa expression in AML-12 cells. H Protein expression levels of C/EBPα and BMP9 after overexpressing Cebpa. I Protein expression levels of C/EBPα and BMP9 after inhibiting Cebpa with or without Sup treatment. The average target gene/Gapdh ratios in different experimental groups relative to the control group. *p < 0.05, **p < 0.01, and ***p < 0.001

C/EBPα is an important transcription factor that is a determinant in multiple physiological functions. High expression of C/EBPα and BMP9 was detected in both senescent hepatocytes and MΦs; however, whether BMP9 was regulated by C/EBPα remained unclear. Concomitant with Cebpa overexpression (OE), BMP9 expression was increased in the immortalized murine bone marrow-derived MΦ (iBMDM) cell line (Fig. 3D). Similarly, Cebpa expression in iBMDMs that was downregulated by siRNA reduced BMP9 expression (Fig. 3E). BMDMs derived from young and aged mice were then stimulated with APAP-treated AML-12 cell supernatant (Sup) to mimic DAMP signaling induced by hepatocyte death. Moreover, enhanced Cebpa expression promoted iNOS expression and decreased CD206 expression in the iBMDMs stimulated by Sup (Fig. 3F). Following Cebpa overexpression, the mRNA expression of Il1b, Il6, and Tnfa induced by Sup was increased in the iBMDMs (Fig. 3G). In addition, Cebpa overexpression increased the BMP9 level in AML-12 cells (Fig. 3H), and downregulated Cebpa expression correspondingly led to reduced BMP9 expression (Fig. 3I).

These results suggest that the expression of BMP9 is driven by C/EBPα signaling in MΦs and hepatocytes during aging, which might be a key intrinsic driver of aggravated APAP-ALI associated with aging.

The C/EBPα-BMP9 axis in hepatocytes promotes APAP-ALI progression

Given the link between C/EBPα and BMP9, the effect of C/EBPα on BMP9 expression in hepatocytes and APAP-ALI in vivo was investigated by overexpressing Cebpa via AAV2/8. Both the mRNA and protein expression levels of Bmp9 were upregulated after Cebpa was overexpressed in the liver 2 weeks after Cebpa AAV2/8 injection into the tail vein (Additional file 1: Figure S2A, B). Notably, IF and IHC assays confirmed successful Cebpa overexpression (Additional file 1: Figure S2C). After Cebpa was overexpressed in hepatocytes, the APAP-ALI model was established. Increased serum ALT and AST levels suggested that Cebpa overexpression aggravated APAP-ALI (Fig. 4A), paralleling the images showing H&E staining (Fig. 4B) and the increase in the number of TUNEL-stained cells (Fig. 4C). Notably, Cebpa overexpression in hepatocytes induced the mRNA expression of Il1b, Il6 and Tnfa in the liver (Fig. 4D). Moreover, we also detected increased Cxcl1, Cxcl13 and Mcp1 and decreased Arg1 and Il10 expression levels (Additional file 1: Figure S3A, B). For a survival study, we increased the dose of APAP administered (500 mg/kg) to the mice and found that Cebpa overexpression in the liver significantly increased the mortality of the mice treated with high-dose APAP (Additional file 1: Figure S3C).

Fig. 4

The C/EBPα-BMP9 axis in the liver promotes APAP-ALI progression. APAP-ALI (300 mg/kg) was established 2 weeks after confirmation that Cebpa was successfully overexpressed by Cebpa-overexpressing AAV2/8 injection into the tail vein of the mice, and serum and liver tissue were collected from each group of mice after 24 h. A Serum ALT and AST levels. B Representative images showing H&E staining. C Representative images showing TUNEL staining images. D mRNA expression of Il1b, Il6 and Tnfa. E Representative images showing IHC staining for BMP9. F Representative images showing IHC staining for P-SMAD1/5/9. G Protein expression levels of C/EBPα, BMP9, SMAD1 and P-SMAD1/5/9. H Protein expression levels of ATG3, ATG7, LC3 I/II and p62. I Representative images showing IF staining for BMP9 and F4/80 in the liver slices of each group. The average target gene/Gapdh ratios different experimental groups relative to the control group. *p < 0.05, **p < 0.01, and ***p < 0.001

IHC assay of liver sections showed that the expression of BMP9 was upregulated after Cebpa was overexpressed (Fig. 4E). Moreover, the rate of SMAD1/5/9 phosphorylation was increased (Fig. 4F). These results indicated that Cebpa overexpression promoted BMP9 production and activated BMP signaling. Next, we sought to identify the proteins that potentially control APAP-ALI progression in the liver after Cebpa is overexpressed. By performing immunoblotting, we confirmed that Cebpa was successfully overexpressed and that the expression level of BMP9 and SMAD1/5/9 phosphorylation rates were increased (Fig. 4G). Furthermore, we detected the downregulation of ATG3 and ATG7 expression in the liver, which was accompanied by blocked autophagy (Fig. 4H). In addition, we also detected increased Bmp9 expression after successful Cebpa overexpression, followed by decreased Atg3 and Atg7 mRNA levels (Additional file 1: Figure S3D).

IF staining for F4/80 and BMP9 in liver sections revealed that Cebpa overexpression increased the expression of BMP9 localized to F4/80-positive cells in the liver (Fig. 4I). Similar to our observations of aged livers, increased expression of BMP9 in F4/80-negative cells was observed in livers after APAP treatment, suggesting the diversity and complexity of BMP9 sources in the liver (Fig. 4I).

BMP9 inhibits ATG3 and ATG7 expression to block autophagy in MΦs

Our results thus far confirmed that BMP9 is abundantly expressed in hepatic MΦs and directly aggravates APAP-induced injury by inhibiting hepatocyte autophagy. Rapamycin (RAPA) has been extensively reported to activate autophagy and reduce senescence [19, 20]. Therefore, we investigated the roles of aging and BMP9 on MΦs treated with RAPA. The LC3 I-to-LC3 II conversion rate was reduced, and BMP9 was highly expressed in primary hepatic MΦs from aged mice treated with RAPA and/or bafilomycin (Ba) (Additional file 1: Figure S4). Next, after infecting Ad-mCherry-GFP-LC3 adenovirus and expressing mCherry-CFP-tagged LC3 in BMDMs, we confirmed that autophagy in aged mouse-derived MΦs was hindered compared with that in the young group (Fig. 5A). Dual IF staining for LC3 and BMP9 (two colors) was then performed with BMDMs derived from aged mice, and we found that the BMP9 level increased while LC3 level decreased in cells from the aged mice, as indicated through the assessment of high-resolution images (Fig. 5B). Furthermore, the number of mCherry puncta was increased in cells treated with RAPA after Bmp9 was deleted (Fig. 5C).

Fig. 5

BMP9 inhibits ATG3 and ATG7 and thus blocks autophagy in MΦs. A BMDMs derived from the bone marrow of young and aged mice were infected with Ad-mCherry-GFP-LC3 (MOI = 20) on the fifth day of culture when the cells were not fully mature and then treated with rapamycin on the seventh day of culture. Representative confocal laser microscopy images of each group. Quantitative data are shown in the right panel. B BMDMs from young and aged mice were treated with rapamycin on day 7. Representative images showing IF staining for LC3-I/II (red) and BMP9 incorporation (green) with DAPI counterstaining (blue) in each group. Quantitative data are shown in the right panel. C BMDMs derived from the bone marrow of WT and Bmp9−/− mice were infected with Ad-mCherry-GFP-LC3 (MOI = 20) on the fifth day of culture when the cells were not fully mature and then treated with rapamycin on the seventh day of culture. Representative confocal laser microscopy images of each group. Quantitative data are shown in the right panel. D Protein expression levels of ATG3, ATG7, LC3 I/II and p62 in BMDMs from WT and Bmp9.−/− mice with or without APAP-treated AML-12 cell supernatant (Sup). E Protein expression levels of ATG3, ATG7, LC3 I/II and p62 were in BMDMs with or without Rm-BMP9 and/or Sup treatment. *p < 0.05, **p < 0.01, and ***p < 0.001

Autophagy was inhibited in the short term after exposure to Sup (described above), which may be an important driver of MΦs proinflammatory action during the acute phase of liver injury (Fig. 5D, E). After BMP9 was depleted, ATG3 and ATG7 expression in MΦs was upregulated, the LC3 I-to-LC3 II conversion rate was increased, and the p62 level was decreased, indicating that autophagy had been activated (Fig. 5D). Additionally, Rm-BMP9 addition resulted in downregulated ATG3 and ATG7 expression, a decreased LC3 I-to-LC3 II conversion rate, and increased p62 accumulation in the MΦs (Fig. 5E).

In previous studies, a decline in autophagy promoted M1-type MΦ polarization during ALI, inducing proinflammatory and injury-promoting effects [21]. Importantly, BMP9 can directly promote APAP-induced hepatocyte death; therefore, the aging-associated excess accumulation of BMP9 may be an essential driver of aggravated liver injury.

BMP9 regulates the senescence and immunophenotype acquisition of MΦs

After etoposide (1 μg/ml) treatment for 2 days, the mRNA levels of p21 and p53 in iBMDMs were increased, and the expression of the SASP components Il1b, Il6 and Tnfa was upregulated (Additional file 1: Figure S5A). Similar to the BMP9 promotion of etoposide-induced hepatocyte senescence, BMP9 promoted etoposide-induced senescence phenotype acquisition by iBMDMs (Additional file 1: Figure S5B). In hepatic MΦs isolated from Bmp9−/− mice, Il1b, Il6, Tnfa and iNos expression levels were reduced after etoposide treatment alone and/or with APAP-treated AML-12 cell supernatant treatment (Additional file 1: Figure S5C).

Considering these results, we isolated primary hepatic MΦs from young WT and Bmp9−/− mice and aged WT and Bmp9−/− mice. SA-β-gal staining revealed that Bmp9 deletion reduced the number of MΦs that acquired the senescence phenotype (Fig. 6A). Next, after these young and aged cells were treated with Sup, the expression levels of Il1b, Il6 and Tnfa were decreased in the Bmp9−/− group (Fig. 6B). Moreover, cytokine (IL-1β, IL-6, TNF-α and IL-10) levels in the supernatants of the cells subjected to different treatments were measured by ELISAs. The results showed that Bmp9 deletion reduced the secretion of IL-1β, IL-6 and TNF-α (proinflammatory cytokines) and increased the production of IL-10 (a cytokine that suppresses inflammation and promotes injury repair), while increases in the exogenous Rm-BMP9 level significantly increased the levels of secreted IL-1β, IL-6 and TNF-α and reduced the concentration of IL-10 (Fig. 6C). The expression and intracellular distribution of iNOS and CD206 were then determined by IF staining, and the results showed that aging promoted iNOS production and decreased CD206 levels, and excess BMP9 in cells further enhanced iNOS production and reduced CD206 levels, while inhibiting Bmp9 expression led to the opposite outcomes (Additional file 1: Figure S6A).

Fig. 6

BMP9 regulates the senescence and immunophenotype acquisition of MΦs. A SA-β-gal staining of BMDMs derived from young and aged WT mice and young and aged Bmp9.−/− mice. B mRNA expression levels of Il1b, Il6 and Tnfa in each group. C Cytokine levels (IL-1β, IL-6, TNF-α and IL-10) in the supernatants of cells in each group as measured by ELISAs. To further study the effect of macrophage-derived BMP9 on liver injury, an F4/80-specific AAV2/8-overexpressing plasmid vector (AAV2/8-F4/80-Bmp9) was constructed. APAP-ALI (300 mg/kg) was established for 2 weeks after AAV2/8-F4/80-Bmp9 and AAV2/8-F4/80-CTRL injection. Serum and liver tissue were collected from each group of mice 24 h later. D, E Serum ALT and AST levels (D) and representative images showing H&E staining (E) of liver slices of each group. F Representative images showing TUNEL staining. G mRNA expression levels of Il1b, Il6 and Tnfa. The average target gene/Gapdh ratios of different experimental groups relative to the control group. *p < 0.05, **p < 0.01, and ***p < 0.001

To further study the effect of MΦ-derived BMP9 on APAP-ALI, an F4/80-specific Bmp9-overexpressing AAV2/8 virus was constructed (AAV2/8-F4/80-Bmp9) [22]. APAP-ALI was established 2 weeks via tail vein injection of AAV2/8-F4/80-Bmp9. IF staining of isolated hepatic MΦs showed that BMP9 expression in the AAV2/8-F4/80-Bmp9 group was significantly upregulated, confirming the successful overexpression of MΦ-specific BMP9 (Additional file 1: Figure S6B). As shown by serum ALT/AST levels and H&E staining, APAP-ALI severity was significantly increased in the F4/80-specific Bmp9-overexpressing group (Fig. 6D, E). Additionally, the number of TUNEL-stained cells in the livers of the F4/80-specific Bmp9-overexpressing mice treated with APAP were significantly increased (Fig. 6F). Moreover, the mRNA expression of Il1b, Il6 and Tnfa was upregulated in the livers of the F4/80-specific Bmp9-overexpressing mice (Fig. 6G).

These data suggest that BMP9 plays a key role in the senescence of MΦs and that MΦ-derived BMP9 clearly drives the progression of APAP-ALI.