Copper concentrations are lower and CP expression is higher in fatty liver and diabetes models than in controls
To investigate the roles of copper and CP in lipid metabolism, we then measured the expression of CP and copper content in human liver biopsies and mouse models of metabolic diseases. Notably, copper levels in human liver biopsies were significantly lower in NAFLD patients than healthy controls (p < 0.05) (Fig. 1a). The hepatic copper content in ob/ob and HFD-induced obese (DIO) mice was also lower than that in control mice (Fig. 1b, Extended Data Fig. 1a). These results suggest that a reduction in hepatic copper content might be associated with NAFLD in humans and mice.
Next, we analyzed the expressions of genes involved in copper metabolism and transportation in human liver biopsies, including CP, CTR1 and SOD1, and found that the expression of CP, SOD1, SOD2, and COX17 was higher in liver biopsies of NAFLD patients than controls (Fig. 1c). To verify which gene regulates hepatic copper abundance, we knocked down these four genes respectively in primary hepatocytes. Only knocking-down of Cp substantially increased the intracellular copper concentration (Fig. 1d), which is consistent with the essential role of CP in hepatic copper homeostasis2. In addition, we found that expression levels of hepatic CP were significantly increased in all three NAFLD mouse models (Fig. 1e,f and Extended Data Fig. 1b-f).
In view of the close link between NAFLD and type 2 diabetes, CP expression was measured in diabetes models, including STZ-injected mice (Extended Data Fig. 1g-i) and mice with genetic inactivation of hepatic insulin receptor substrate Irs1 and Irs2 double knockout (LDKO) (Extended Data Fig. 1j-l). Similar to the obese NAFLD mouse models, CP was highly expressed in diabetic mice. These results suggest that excess hepatic CP might lead to reductions in hepatic copper level, and subsequently results in the development of NAFLD.
Acute CP-knockdown attenuates HFD-induced hepatic steatosis
To determine the function of CP on hepatic glucose and lipid metabolism and NAFLD development, we knocked down hepatic CP in DIO mice (Extended Data Fig. 2a). Relative to controls, knockdown of hepatic CP significantly reduced liver weight (Fig. 1g,h), ectopic lipid accumulation in the liver (Fig. 1i,j), and plasma free fatty acids (FFAs) (Extended Data Fig. 2b). Lipidomic analysis showed that the levels of diacylglycerides (DAGs), FFAs and triglycerides (TGs) were substantially reduced in Ad-shCP-treated mouse livers (Fig. 1k). Moreover, glucose clearance and insulin sensitivity were substantially improved in CP knockdown mice (Extended Data Fig. 2c-f), but no differences in body weight or food intake were observed (Extended Data Fig. 2g,h).
Then, we used AAV (which is less immunogenic and widely used in clinical trials) to express Cre in mouse liver by transduction of CPflox/flox mice with CreAAV−TBG viruses (CPflox/flox-CreAAV−TBG). Consistent with Ad-shCP injected mice, CPflox/flox-CreAAV−TBG mice displayed lower ectopic hepatic lipid accumulation (Extended Data Fig. 2i) and substantially less insulin resistance (Extended Data Fig. 2j). However, there were no differences in glucose tolerance between CPflox/flox-CreAAV−TBG and CPflox/flox-GFPAAV−TBG mice (Extended Data Fig. 2k).
LKO mice are protected from HFD-induced hepatic steatosis
To confirm the results observed in mice with virus-mediated CP-knockdown, we generated liver-specific CP knockout (LKO) mice (Extended Data Fig. 2l). LKO mice exhibited less lipid accumulation in the liver, less liver weight and hepatic triglyceride (Fig. 1l-n). Glucose tolerance and insulin sensitivity were improved as well in LKO mice (Extended Data Fig. 2m,n). There were no significant differences in body weight and food intake between the two groups of mice (Extended Data Fig. 2o,p).
We also investigated the effect of CP knockout (CP-KO) on lipid accumulation in primary hepatocytes, and the result showed that CP-KO resulted in less palmitic acid (PA)-induced lipid droplet and TG accumulation in primary hepatocytes than controls (Fig. 1o-q).
Taken together, these data indicate that deletion of CP ameliorates HFD-induced hepatic steatosis, and improves insulin sensitivity and glucose homeostasis.
CP-knockdown increases mitochondrial biogenesis and FAO
Given that mitochondria play a key role in lipid metabolism, we examined whether CP ablation affected mitochondrial function. Electron microscopy showed that CP deletion in mouse liver ameliorated HFD-induced mitochondria swelling and deteriorations of mitochondrial morphology (Fig. 2a). In line with this, CP ablation substantially increased mitochondrial number (Fig. 2b) and the copy number of mitochondrial DNA in primary hepatocytes (Fig. 2c). Moreover, expressions of genes encoding mitochondrial respiratory chain complex was upregulated, including Nd1, Nd2, Co1 and Atp6 (Fig. 2d). Consistent with this, immunoblotting confirmed that CP deletion efficiently upregulated protein expression of the subunits of mitochondria respiratory chain complexes, such as ATP5A and UQCRC2 (Fig. 2e). These results suggest that CP deletion rescues mitochondrial morphology under HFD feeding and increases the number of healthy mitochondria.
Next, we explored the effects of CP ablation on mitochondrial function. Notably, CP-KO cells exhibited a significantly higher capacity for basal respiration, maximal respiration, spare respiration, and ATP production (Fig. 2f,g), indicating that CP-KO resulted in higher mitochondrial activity. In addition, mitochondrial FAO rate was also increased (Fig. 2h). Consistent with this, CP-KO hepatocytes showed significantly higher expressions of genes involved in mitochondrial FAO, including Cpt1α, Cpt2 and Hmgcs2 (p < 0.01) (Fig. 2i), but without significant changes in the levels of lipogenic genes, including Srebp1c, Acc1, Fasn, Chrebpα, Chrebpβ, and Acly (Extended Data Fig. 2q). Furthermore, we found that overexpressing CP in primary hepatocytes substantially suppressed the expressions of genes related to FAO (Fig. 2j,k). These results indicate that CP depletion substantially enhances mitochondrial FAO activity.
AMPK is required for amelioration of ectopic lipid accumulation caused by hepatic CP-KO
To elucidate the molecular mechanisms by which CP deletion reduces hepatic steatosis, we performed RNA-seq analysis of Ad-shCP or Ad-shCON treated mouse livers. Ingenuity Pathway Analysis showed that the AMPK pathway was one of the most enriched biological process terms (Fig. 3a). Incubating CP-KO primary hepatocytes with inhibitors against various pathways, reveals that only inhibitors of AMPK and SIRT1 (a direct downstream target of AMPK) blocked CP deletion-mediated upregulation of Hmgcs2 and Creb3l3 (Fig. 3b, Extended Data Fig. 3a,b). Moreover, AMPK was activated in CP-deleted hepatocytes (Fig. 3c) and LKO mice liver (Fig. 3d). In contrast, inhibition AMPK with Compound C (CC) impaired CP-deletion-induced up-regulation of mitochondrial respiration (Fig. 3e, Extended Data Fig. 3c). Consistent with this data, genetic ablation of AMPK attenuated CP-knockdown-mediated reduction of lipid accumulation in hepatocytes (Fig. 3f, Extended Data Fig. 3d,e), also suppressed CP-knockdown-caused increases in mitochondria mass (Fig. 3g) and the expressions of FAO genes (Fig. 3h) and other mitochondrial genes (Fig. 3i).
Collectively, these results indicate that AMPK is a critical downstream target of CP that regulates mitochondrial function and lipid metabolism in the liver.
CP deletion-induced upregulation of FAO is dependent upon the AMPK-PGC1α-PPARα axis
Previous reports show that AMPK increases mitochondrial FAO by enhancing transcriptional activity of PGC1α and PPARα19. Notably, we found that CP ablation decreased the abundance of acetylated PGC1α, an inactivated form of PGC1α (Fig. 3j); by contrast, overexpressing CP elevated acetylated PGC1α levels (Fig. 3k). Activated PGC1α promotes its interaction with PPARα, thereby increasing its co-activator ability20. Here, we found that CP deletion efficiently enhanced the interaction between PGC1α and PPARα (Fig. 3l). By contrast, overexpression of CP substantially decreased the interaction of PGC1α and PPARα (Fig. 3m). We further examined PGC1α-PPARα-mediated transcription of FAO genes, and revealed that knocking down CP increased the activity of PPARE-Luc, CPT1α-Luc and HMGCS2-Luc reporters (Extended Data Fig. 4a), indicating that CP deletion enhances PPARα-transactivation. By contrast, overexpressing CP suppressed the GW9578 (PPARα agonist)-induced increase in promoter activities of these luciferase reporters (Extended Data Fig. 4b). In line with this, chromatin immunoprecipitation assays confirmed that CP deletion enhanced recruitment of PGC1α (Fig. 3n) and PPARα (Fig. 3o) to the promoters of PPARα target genes; whereas knocking down PGC1α (Extended Data Fig. 4c) or PPARα (Extended Data Fig. 4d) completely blocked the elevated transcription of FAO genes caused by CP deletion. In line with this, inactivation of PPARα via GW6471, an antagonist of PPARα, abolished the effects of CP-knockdown on the reduction of lipid accumulation in hepatocytes (Extended Data Fig. 4e,f).
Restoration of hepatic copper content reduces excess lipid accumulation in liver through AMPK activation
Next, we explored how CP deletion activated AMPK. Unexpectedly, we didn`t detect significant alteration in AMP/ATP or ADP/ATP in the livers of LKO mice (Extended Data Fig. 4g,h), suggesting that CP deletion activates AMPK via an AMP/ADP-independent mechanism. Notably, CP deletion efficiently prevents the decrease in hepatic copper levels caused by HFD feeding (Fig. 4a,b and Extended Data Fig. 5a). Interestingly, treating primary hepatocytes with a copper solution increased AMPK phosphorylation in a dose-dependent manner (Fig. 4c). Furthermore, depleting intracellular copper with copper cation-specific chelator, bathocuproinedisulfonic acid disodium salt (BCS), attenuated the effect of CP deletion on upregulation of Hmgcs2 (Fig. 4d, Extended Data Fig. 5b) and Cpt1α (Fig. 4e, Extended Data Fig. 5b). By contrast, an iron cation-specific chelator, deferasirox (DFX), had little effect on Hmgcs2 (Fig. 4d) or Cpt1α (Fig. 4e) mRNA levels. In addition, we performed similar experiments with additional copper chelator, tetrathiomolybdate (TM) and iron chelators, pyridoxal isonicotinoyl hydrazine (PIH) and deferoxamine (DFO) in primary hepatocytes, and obtained consistent conclusion that chelating copper, but not iron, blocked CP-deletion-upregulated FAO gene expressions (Fig. 4f, Extended Data Fig. 5c-e). Moreover, BCS and TM attenuated increases in phosphorylated AMPK by CP deletion in primary hepatocytes (Fig. 4g). Chelating intracellular copper with BCS also prevented CP ablation-induced reductions in lipid accumulation; whereas the iron chelator DFX failed to do so (Fig. 4h,i and Extended Data Fig. 5f,g).
The global CP KO (WKO) mouse is an animal model of aceruloplasminemia that exhibits excess iron accumulation in the liver 21. We also observed an increased level of hepatic iron in the WKO mouse (Extended Data Fig. 5h). By contrast, Prussian blue staining of the LKO mouse liver did not reveal obvious iron accumulation (Extended Data Fig. 5h). Although hepatic iron levels of LKO mice (measured by ICP-MS) were modestly increased compared to controls (Extended Data Fig. 5i), they are comparable to hepatic iron levels in normal mice as previously reported22–24.
Taken together, these results indicate that the effect of CP deletion on the activation of AMPK and suppression of fatty liver is dependent on upregulation of hepatic copper content.
Copper upregulates hepatic FAO via SCO1
Given that copper activates AMPK and that AMPK is not reported as a copper-binding protein25, we hypothesized that a copper-binding protein mediates AMPK activation. To test this hypothesis, we used a collection of siRNAs targeting key copper-binding proteins in primary hepatocytes, to identify candidate protein(s) mediating the effect of CP deletion on upregulation in FAO genes, including Hmgcs2 and Creb3l3 (Fig. 4j, Extended Data Fig. 6a-c).
We found that only knockdown of SCO1 consistently blunted the upregulation of FAO gene expression caused by CP deletion (Fig. 4j, Extended Data 6c). Moreover, we found that knocking down SCO1 entirely abrogated the effect of CP ablation on increasing phosphorylated AMPK levels in hepatocytes (Fig. 4k). Finally, SCO1 deficiency blocked CP knockout-dependent suppression of lipid accumulation in hepatocytes (Fig. 4l,m).
Taken together, these results suggest that SCO1 plays a critical role in copper-dependent AMPK activation and lipid catabolism.
SCO1 promotes AMPK activation by directly interacting with AMPK and LKB1
Next, we explored the mechanism by which SCO1 regulates AMPK activity. SCO1 was reported as a mitochondrial protein26. however, we detected that some SCO1 signals were well overlapped with that of β-actin (Fig. 5a), suggesting that a number of SCO1 proteins localized in cytoplasm. Moreover, co-immunoprecipitated assay showed that exogenous SCO1 interacted with exogenous AMPKα1 and endogenous LKB1 (Fig. 5b). Reverse co-immunoprecipitation further confirmed the interaction among AMPK, SCO1 and LKB1 (Fig. 5c). Besides, the endogenous interaction between SCO1 and AMPK was also detected in L02 cells (Fig. 5d). Furthermore, pull-down assays showed a direct interaction between SCO1 (Fig. 5e,f) and AMPK as well as between SCO1 and LKB1 (Fig. 5g,h). In addition, confocal images revealed the colocalization of AMPK and SCO1 in hepatocytes (Fig. 5i). Next, we performed similar co-immunoprecipitated assay in subcellular fractionation. The cytoplasm and mitochondria were respectively purified from primary hepatocytes (Extended Data Fig. 6d), and we only detected the interaction between SCO1 and AMPK in cytosol but not in the lysate of mitochondria (Fig. 5j,k). The co-immunoprecipitated assay in AMPK-DKO primary hepatocytes showed that SCO1 interacted with LKB1 even in the absence of AMPK, suggesting that SCO1 interacts with LKB1 independent of AMPK (Fig. 5l). Taken together, these results suggest that SCO1 acts as a scaffold protein to tether the activating kinase LKB1 to AMPK, resulting in AMPK phosphorylation. Next, we determined whether copper or CP regulates the assembly of the SCO1-LKB1-AMPK complex. A co-immunoprecipitation assay showed that overexpression of CP reduced the interaction between SCO1 and AMPK, but had little effect on SCO1-LKB1 interaction (Fig. 5m). Interestingly, we failed to detect CP protein in SCO1-LKB1-AMPK precipitation (Fig. 5m), which further confirmed that CP modulation of AMPK activity is not through its direct physical interaction with AMPK. Given that overexpression n of CP might lead to a reduction in intracellular copper via copper excretion, we hypothesized that copper affects the assembly of the SCO1-LKB1-AMPK complex. To confirm this hypothesis, we depleted intracellular copper with BCS, and the co-immunoprecipitation result showed that BCS treatment significantly reduced the interaction between SCO1 and AMPK (Fig. 5n), indicating the important role of copper in the SCO1-AMPK interaction. In another experiment, we constructed an SCO1 copper-binding deficient mutant (C152/156A) by substituting CyS152 and CyS156 with alanine, as reported27,28. An obvious shift of the SCO1 (C152/156A) band (Fig. 5o, line 7), similar with BCS treatment (Fig. 5o, line 6), revealed that its copper-binding capacity was abolished. Notably, a co-immunoprecipitation assay showed that the C152/156A mutation in SCO1 impaired its interaction with AMPK, but had no effects on its interaction with LKB1 (Fig. 5p). Moreover, a pulldown assay demonstrated that the a truncated SCO1 (only containing SCO domain where the copper-binding motif resides, His-Truncated) displayed a strong interaction with AMPK, compared to full length SCO1 (Fig. 5q). We next performed a similar pull-down assay with copper-loaded His-Truncated (Cu-Truncated) or copper-deficient His-Truncated protein (apo-Truncated). An obvious up-shift of apo-Truncated band in non-reducing gel electrophoresis confirmed its copper-deficient status, and the pull-down result demonstrated that Cu-Truncated had a stronger binding affinity to AMPK than apo-Truncated (Fig. 5r).
Taken together, these results reveal that intracellular copper increase activates AMPK through promoting the assembly of the SCO1-LKB1-AMPK complex.
Adequate hepatic copper levels are necessary for CP deletion-induced amelioration of hepatic steatosis
Given that SCO1 deficiency blocked CP knockout-dependent suppression of lipid accumulation in hepatocytes (Fig. 4l,m), and depleting intracellular copper by BCS produced copper-deficient SCO1 (Fig. 5o, line 6), we thus deprived the copper of SCO1 by treating mice with BCS to explore its effect on lipid metabolism in vivo (Fig. 6a, Extended Data 7a). ICP-MS analyses showed that BCS treatment substantially blunted CP LKO-caused elevation of hepatic copper content (Fig. 6b). Consistently, BCS administration reduced AMPK phosphorylation levels upregulated by CP LKO (Fig. 6c, Extended Data Fig. 7b-d), presumably due to depletion copper from SCO1. In line with this, BCS treatment also abolished the effects of CP LKO on the amelioration of hepatic lipid accumulations (Fig. 6d-f). Moreover, BCS-treatment impaired the effects of CP-LKO on the upregulation of FAO (Fig. 6g) and other mitochondrial genes (Fig. 6h), as well as maintaining healthy mitochondria morphology in the liver (Fig. 6i). There were no significant differences in body weight and food intake between the two groups of mice (Extended Data Fig. 7e,f).
Taken together, these findings suggest that copper is required to increase AMPK activation and hepatic lipid catabolism in mouse liver with CP deletion.