Hepatic ENTPD5 Is Critical for Maintaining Metabolic Homeostasis and Promoting Brown Adipose Tissue Thermogenesis

Abstract

Although hepatocyte-released adenosine triphosphate (ATP) plays important roles in maintaining metabolic homeostasis, how its hydrolyzation outside hepatocytes impacts on glucolipid metabolism remains unclear. The authors aim to identify the enzyme(s) that hydrolyzes hepatocyte-secreted ATP to regulate metabolic homeostasis. All known ATP-hydrolyzing enzymes are expressed with the highest expression of ectonucleoside triphosphate diphosphohydrolase 5 (ENTPD5) in hepatocytes. ENTPD5 expression is reduced in steatotic mouse and human livers. Hepatic ENTPD5 overexpression ameliorates the deregulated glucolipid metabolism and obesity with increased brown adipose tissue (BAT) thermogenesis, while hepatic ENTPD5 silencing exerted the opposite effects in obese mice. Mechanistically, ENTPD5 hydrolyzes extracellular ATP to ADP, which activates purinergic receptor, P2Y₁₂, to inhibit gluconeogenesis and lipid deposition, and repress adrenomedullin (ADM) expression. Hepatic ENTPD5 repression promotes ADM expression and release to inhibit uncoupling protein 1 (UCP1) expression and thermogenesis in BAT. Hepatic ADM expression is increased in NAFLD patients. Serum ADM level is correlated positively with Body mass index in overweighted or obese humans. Hepatic ADM silencing promotes UCP1 expression and thermogenesis in BAT of obese mice. Overall, ENTPD5-mediated hydrolysis of extracellular ATP to ADP of hepatocytes is critical for maintaining hepatic glucose/lipid metabolism and promoting BAT thermogenesis by inhibiting ADM expression and secretion.

Keywords: adenosine triphosphate metabolism; adrenomedullin; brown adipose tissue thermogenesis; ectonucleoside triphosphate diphosphohydrolase 5.

Figure 1

ENTPD5 expression is reduced in steatotic mouse and human livers, and in cultured hepatocytes treated with free fatty acids. A,B) Gel electrophoresis of PCR products determination the expression profile of ENTPDs in mouse livers (A) and cultured hepatocytes (B). C) Relative mRNA levels of ENTPDs in cultured mouse hepatocytes (N = 4). D–F) Changes in ENTPD mRNAs in primary hepatocytes treated with FFAs (D, N = 4), and HFD‐fed and db/db mouse livers (E,F, N = 5–8). Hepatocytes were treated with FFAs (0.1 mm oleic acid and 0.2 mm palmitic acid) for 24 h before assays. G) Confocal imaging analysis revealed that ENTPD5 protein was reduced in the livers of patients with NAFLD (N = 3). Scale bar, 75 and 25 µm. The relative expression data was marked in the images. H–J) Confocal imaging analysis indicated that ENTPD5 protein was reduced in the livers of obese mice (H,I), and primary hepatocytes treated with FFAs (J). Scale bar: 75 and 25 µm for (H,I); 50 and 25 µm for (J). The relative expression data was marked in the images (N = 3). K) Immunoblotting assays confirmed the reduction of ENTPD5 protein in obese mouse livers and FFAs‐treated hepatocytes (N = 3). L,M) The lipid staining (L) and TG quantification (M) in mouse primary hepatocytes infected with AAV8‐GFP, AAV8‐ENTPD5, or AAV8‐shENTPD5, respectively, in the presence of FFAs for 48 h (N = 4). Scale bar, 50 and 10 µm. N) Glucose production assays in mouse primary hepatocytes infected with AAV8‐GFP, AAV8‐ENTPD5, or AAV8‐shENTPD5, respectively, for 48 h (N = 3). Hepatocytes were infected with AAV8 at the dose of 100 vg per cell. FFAs (0.1 mm oleic acid (OA) + 0.2 mm palmitic acid (PA)); ND, normal diet; HFD, high fat diet. db/m, db/m mice, db/db, db/db mice. TG, triglycerides. *P < 0.05, ** P < 0.01 versus control mice.

Figure 2

Hepatic ENTPD5 overexpression improved the dysregulation of glucose and lipid metabolism with activated BAT thermogenesis in db/db mice. A) Bodyweight curve of db/db mice post AAV8‐ENTPD5 injection (N = 10). B–D) AAV8‐ENTPD5 injection attenuated glucose intolerance (B), hepatic glucose production (C), and global insulin resistance (D) as evaluated by OGTT, PTT, and ITT in the indicated time point of mice (N = 10). E) The fasting blood glucose curve in db/db mice transduced with AAV8‐ENTPD5 (N = 10). F) AAV8‐ENTPD5 injection increased energy expenditure (EE) of mice as assayed in metabolic cages (N = 4–5). G) AAV8‐ENTPD5 injection activated BAT thermogenesis and increased core body temperature after cold exposure (N = 6). H) MRI analysis revealed that AAV8‐ENTPD5 injection reduced global and hepatic fat content of mice. Representative images were shown in left panel, and quantitative data shown in right panel (N = 5). I,J) The impact of AAV8‐ENTPD5 injection on serum TG/CHO (I), and AST/ ALT (J) levels (N = 8). K) AAV8‐ENTPD5 injection specifically increased ENTPD5 mRNA expression in the liver (N = 6). MRI: magnetic resonance imaging. Data of control mice were marked in red, and those of mice infected with AAV8‐ENTPD5 marked in blue. BAT, brown adipose tissues; TG, triglycerides; CHO, cholesterol; AST, alanine aminotransferase; ALT, aspartate aminotransferase. * P < 0.05, ** P < 0.01 versus control mice.

Figure 3

Hepatic ENTPD5 inhibition aggravated HFD‐induced glucose and lipid metabolic disorders, and obesity with reduced BAT thermogenesis in mice. A) Bodyweight curve of mice infected with AAV8‐shENTPD5 on ND or HFD feeding condition, AAV8‐GFP as control (N = 8–10). Mice were treated with AAV8‐GFP or AAV8‐ENTPD5, and then subjected to ND or HFD feeding, respectively. B–D) Impact of AAV8‐shENTPD5 injection on glucose tolerance (B), hepatic glucose production (C), and overall insulin sensitivity (D) in mice subjected to a normal diet (ND) or high‐fat diet (HFD) (N = 8–10). E) The fasting blood glucose curve in mice post AAV8‐shENTPD5 injection (N = 8–10). F) The EE curve and AUC data obtained in metabolic cage analyses of HFD mice at 15 weeks post AAV8‐shENTPD5 injection (N = 5). G) AAV8‐shENTPD5 injection reduced BAT thermogenesis and decreased core body temperature after cold exposure at 16 weeks post AAV8‐shENTPD5 injection (N = 6). H) MRI analysis revealed that AAV8‐shENTPD5 injection increased global and hepatic fat content. Representative images were shown in the left panel, and quantitative data shown in the right panel (N = 5). I,J) Impact of AAV8‐shENTPD5 injection on serum TG/CHO (I), and serum AST/ALT (J) levels in mice (N = 8). K) AAV8‐shENTPD5 injection specifically reduced ENTPD5 mRNA levels in the livers of mice (N = 6). Data of HFD‐AAV8‐GFP mice were marked in purple, and those of HFD‐AAV8‐shENTPD5 marked in red. * P < 0.05, ** P < 0.01 versus control mice.

Figure 4

Modulation of ENTPD5 on lipid deposition and the expressions of metabolic genes in the livers of mice. A,B) Hepatic ENTPD5 overexpression attenuated lipid deposition in the livers of db/db mice as evidenced by morphological observation, H.E staining, Oil Red O staining (A), and lipid quantification (B) (N = 7). Scale bar: 50 µm. C,D) Hepatic ENTPD5 overexpression reduced the weight of liver and WAT (C), and the ratios to bodyweight (D) (N = 9). E,F) Hepatic ENTPD5 overexpression on the mRNA (E) and protein (F) levels of lipid metabolic and gluconeogenic genes in the livers of db/db mice (N = 6). G,H) Hepatic ENTPD5 silencing exaggerated HFD‐induced lipid deposition in the livers of mice as evidenced by morphological observation, H.E staining, Oil Red O staining (G), and lipid quantification (H) (N = 7). Scale bar: 50 µm. I,J) Hepatic ENTPD5 silencing increased the weight of liver and WAT (I), and the ratios to bodyweight (J) of HFD mice (N = 10). K,L) Hepatic ENTPD5 silencing on the mRNA (K) and protein (L) levels of lipid metabolic and gluconeogenic genes in the livers of HFD mice (N = 6–7). * P < 0.05, ** P < 0.01 versus control mice.

Figure 5

Hepatic ENTPD5 influenced BAT thermogenesis with altered ADM expression and secretion. A,B) Hepatic ENTPD5 overexpression increased the expressions of key thermogenic genes with decreased brown adipocyte size in BAT of db/db as indicated by H.E staining and immunohistochemical staining (A), as well as the mRNA and protein expressions of key thermogenic genes (B). Bar graphs represent the statistical analysis of brown adipocyte diameters and UCP1 grayscale values from immunohistochemical staining (A, N = 4), and the statistical analysis of densitometric values from WB (B, N = 6). Scale bar: 100 µm. C,D) Hepatic ENTPD5 silencing increased brown adipocyte size, bar graphs represent the statistical analysis of brown adipocyte diameters and UCP1 grayscale values from immunohistochemical staining (C, N = 4), while decreased the mRNA and protein expression levels of key thermogenic genes (D, N = 6) in BAT of HFD mice, bar graphs represent the statistical analysis of densitometric values from WB. Scale bar: 100 µm. E) The volcano plot of RNA‐sequencing analyses of different expression genes in mouse primary hepatocytes infected with AAV8‐GFP or AAV8‐ENTPD5 for 36 h (N = 4). F) Heat map of hepatocyte‐secreted protein genes affected by ENTPD5 overexpression in mouse primary hepatocytes. The color of the map represented the relative expression fold changes (N = 4). G) Change in the mRNA and protein levels of ADM in mouse hepatocytes with ENTPD5 overexpression or silencing, respectively (N = 3). H) Confocal imaging analysis of ADM protein in mouse primary hepatocytes after ENTPD5 overexpression or silencing. Scale bar: 50 µm, 10 µm. I) Confocal imaging staining of ADM protein in liver tissue specimens from patients with NAFLD. Scale bar, 75 µm, 25 µm. J) The mRNA and protein expressions of ADM were increased in the livers of obese mice (N = 6–7). K) Serum ADM level was increased in obese mice (N = 6). L) Hepatic overexpression or silencing of ENTPD5 on the mRNA and protein levels of ADM in obese mouse livers (N = 6). M) Confocal imaging analysis of ADM protein in db/db mouse livers after ENTPD5 overexpression. Scale bar: 75 and 25 µm. N) Impact of ENTPD5 overexpression or silencing on ADM level in the medium of cultured primary hepatocytes (N = 4). O) Serum ADM level in obese mice with hepatic ENTPD5 overexpression or silencing (N = 6). P) The correlation analysis between serum ADM level and BMI in overweighted or obese adolescent (N = 38). Q) The correlation analysis between serum ADM level and BMI in overweighted or obese adults (N = 30). * P < 0.05, ** P < 0.01 versus control mouse or control cell groups or control subjects.

Figure 6

Hepatic ADM knockdown ameliorated glucose/lipid metabolic disorders and obesity with enhanced BAT thermogenesis in db/db mice. A) Bodyweight curve of db/db mice infected with AAV8‐shADM or AAV8‐GFP (N = 7). B–D) Injection of AAV8‐shADM improved glucose intolerance, gluconeogenesis, and insulin resistance of mice as assayed by OGTT (B), PTT (C), and ITT (D) (N = 7). E) Fasting blood glucose curve of db/db mice with AAV8‐shADM injection (N = 7). F) The EE curve and AUC data in metabolic cage analyses of db/db mice after AAV8‐shADM injection (N = 4). G) AAV8‐shADM injection activated BAT thermogenesis and increased key body temperature after cold exposure (N = 5). H) MRI assay indicated that AAV8‐shADM injection reduced global and hepatic content (N = 5). I,J) Change in serum AST/ALT (I) and TG/CHO (J) levels of db/db mice after AAV8‐shADM injection (N = 7). K) Serum ADM level of db/db mice after AAV8‐shADM injection (N = 7). L) Tissue expression profile of ADM mRNA in db/db mice after AAV8‐shADM injection (N = 7). *P < 0.05, ** P < 0.01 versus control mouse.

Figure 7

Thermogenic gene expression in BAT was activated by hepatic ADM knockdown but repressed by treatment with rADM in mice. A–C) Hepatic ADM inhibition reduced lipid deposition in the livers of db/db mice as evidenced by morphological observation (A), H.E staining (B, top), Oil Red O staining (B, bottom), and lipid quantification (C) (N = 6). Scale bar: 100 µm. D,E) Hepatic ADM inhibition reduced the weight of liver and WAT, and their ratios to bodyweight of db/db mice (N = 7). F,G) Hepatic ADM inhibition decreased the mRNA and protein levels of lipid metabolic and gluconeogenic genes in db/db mouse livers (N = 6–7). H,I) Hepatic ADM inhibition upregulated the expressions of key thermogenic genes with decreased brown adipocyte size in BAT of db/db as indicated by morphology, bar graphs represent the statistical analysis of adipocyte diameters from H.E staining (H‐left, N = 3), grayscale values statistics of UCP1 from immunohistochemical images (H‐right, N = 4) and key thermogenic protein levels (I, N = 6), bar graphs illustrate densitometric values from WB. Scale bar: 100 µm. J) Dynamic serum ADM in C57BL/6 mice after tail vein injection of rADM (5 nmol kg⁻¹, N = 5). K) Treatment with rADM for 5 days reduced BAT thermogenesis and core body temperature after cold exposure in C57BL/6 mice (N = 6). L) Treatment with rADM for 1 week decreased EE of mice as assayed in metabolic cages (N = 4). M) Serum ADM in C57BL/6 mice after scarification (N = 6). N–P) Treatment with rADM inhibited the expressions of key thermogenic genes in BAT of wild type C57BL/6 mice in morphology, bar graphs represent the statistical analysis of adipocyte diameters and grayscale values from H.E staining (N‐left, N = 3) and immunohistochemical staining (N‐right, N = 4), as well as the mRNA (O, N = 6) and protein levels of key thermogenic genes (P, N = 6). rADM: recombinant ADM protein. * P < 0.05, ** P < 0.01 versus control mouse.

Figure 8

ENTPD5 activated ADP‐P2Y₁₂ pathway to repress ADM expression, leading to the activation of BAT thermogenesis via NONO. A) Determination of extracellular ATP/ADP/AMP (eATP/eADP/eAMP) concentrations in mouse primary hepatocytes after infection with AAV8‐GFP, AAV8‐ENTPD5, or AAV8‐shENTPD5 for 36 h, respectively (N = 3). B) Relative P2 receptor subtype expression identified by RNA‐sequencing analyses in mouse primary hepatocytes (N = 4). C) ENTPD5‐induced suppression of gluconeogenesis and TG deposition was reversed by the treatment with suramin (50 µm), PPADS (50 µm), and P2Y₁₂ inhibitor PSB‐0739 (20 µm) in hepatocytes (N = 3). D) Treatment with P2Y₁₂ inhibitor reversed the regulatory effects of ENTPD5 on ADM and metabolic proteins in hepatocytes (N = 3). E) ENTPD5‐induced reduction in ADM secretion was reversed by P2Y₁₂ inhibitor in hepatocytes (N = 4). F,G) Treatment with rADM reduced the mRNA(F) and protein levels(G)of UCP1 and PGC1α in HIB1B cells (N = 3). Cells were treated with 2 and 20 nm rADM for 36 h. H,I) In Hepatocte‐HIB1B cell coculture system, ENTPD5 overexpression in hepatocytes activated UCP1 and PGC1α mRNA (H) and protein (I) expressions in HIB1B cells (N = 3). J,K) In Hepatocte‐HIB1B cell coculture system, ENTPD5 silencing in hepatocytes reduced UCP1 and PGC1α expressions in HIB1B cells and were reversed by the incubation with anti‐ADM antibodies (J) or ADM receptor antagonist ADM (AM) 22–52 (K) (N = 3). L) Representative silver‐stained gel image of DNA pull‐down products in mouse primary hepatocytes. The protein bands pulldown using various mouse ADM gene promoter fragments in the red circles were cut and subjected to mass spectrometry analysis. 36 potential nuclear proteins were identified, while RNA sequencing revealed a total of 1909 transcribed transcription factor genes in hepatocytes. M) ChIP assay revealed that MECP2 bound to −532 to 0 region of mouse ADM gene promoter. N) Dual‐luciferase reporter assays indicated that MECP2 activated the promoter activity of mouse ADM gene in HEK293T cells (N = 3). O) ENTPD5 overexpression reduced the mRNA level of MECP2 in mouse primary hepatocytes, reversed by P2Y₁₂ inhibitor (N = 3). P) The impact of CaM overexpression on the protein levels of ENTPD5, MECP2, and ADM in mouse primary hepatocytes (N = 3). Hepatocytes were infected with Ad‐CaM or Ad‐GFP (50 MOI) for 24 h, respectively. Q,R) The effect of ENTPD5 knockdown followed by MECP2 inhibition on the expression of JNK downstream proteins and ADM in primary hepatocytes (Q), bar graphs depict the statistical analysis of densitometric values derived from WB assays (R) (N = 3). S) AP1 overexpression activated mouse MECP2 gene promoter activity in HEK293T cells (N = 3). T) NONO overexpression activated the mouse UCP1 gene promoter activity in HEK293T cells (N = 3). U) ChIP assay revealed that NONO bound to −2000 to 1471 region of mouse UCP1 gene promoter, but the binding was inhibited by rADM treatment. V) Treatment with NONO inhibitor (R)‐SKBG‐1 (5 µm) for 24 h reduced the mRNA and protein expressions of UCP1 in HIB1B cells, bar graphs depict the statistical analysis of densitometric values derived from WB assays (N = 3). * P < 0.05, ** P < 0.01 versus control cells or between two indicated groups. W) Proposed mode for ENTPD5's regulatory mechanism(s) in hepatic glucose/lipid metabolism and BAT thermogenesis. Hepatocyte‐released ATP (eATP) is hydrolyzed by ENTPD5 on hepatocyte membrane to eADP, which acts on P2Y₁₂ to activate CaM. Activation of CaM on one hand represses gluconeogenesis and lipogenesis. On the other hand, activated CaM reduces the expression and secretion of ADM in hepatocytes via the inhibition of JNK‐AP1‐MECP2 pathway. A decrease in circulating ADM activates NONO to induce key thermogenic gene expression and promote thermogenesis in BAT. Under overnutrition condition, inhibition of hepatic ENTPD5 expression will promote gluconeogenesis, lipogenesis and inhibit BAT thermogenesis to cause obesity. ADM: adrenomedullin; ADP: adenosine diphosphate; AP1: activated protein 1; ATP: adenosine triphosphate; CaM: calcium‐calmodulin; ENPTD5: ectonucleoside triphosphate diphosphohydrolase 5; JNK: c‐Jun N‐terminal kinase; MCEP2: methyl CpG binding protein 2; NONO: non‐POU domain‐containing octamer‐binding protein; P2Y₁₂ receptors: purinergic receptor P2Y₁₂; UCP1: uncoupling protein 1.