Elevated copper impairs hepatic nuclear receptor function in Wilson’s disease

Dietary copper is absorbed in the duodenum and delivered to the liver via the portal circulation, where it enters hepatocytes via the membrane copper transporter CTR1 (1). Cytosolic copper is transported into the trans-Golgi network via the transmembrane copper-transporting P-type ATPase ATP7 and is incorporated into ceruloplasmin, which is secreted into blood (1). In steady-state or low hepatic copper conditions, ATP7B localizes to the trans-Golgi compartment. Elevation of hepatic copper results in sorting of ATP7B into vesicles that traffic to the canalicular domain to promote copper excretion into bile (2, 3). Loss-of-function mutations in the ATP7B gene result in Wilson’s disease (WD), characterized by excessive hepatic copper accumulation and a variety of symptoms including steatosis, cholestasis, cirrhosis, and liver failure, as well as neurological dysfunction (1). The Atp7b^–/– mouse (4) develops hepatic copper overload by 6 weeks of age, which precedes the onset of WD symptoms (5). Transcriptional profiling at 6 weeks showed a limited number of gene expression changes that did not include alterations in redox pathways but did show a decrease in expression of genes associated with lipid metabolism (5).

Nuclear receptors consist of a variable N-terminal region, a highly conserved DNA-binding domain containing 2 zinc-binding modules that are critical for DNA-binding activity, a hinge region, and a ligand-binding domain (6). Metal replacement studies using an apo-polypeptide of the estrogen receptor (ER) α DNA-binding domain (ER-DBD) demonstrated that copper binds to the ER-DBD with greater affinity than does zinc, resulting in a disordered structure that does not bind an estrogen response element (7, 8). Since hepatic lipid metabolism is regulated by nuclear receptors and disruption may result in liver toxicity (9–12), we hypothesized that some of the metabolic symptoms in WD and in Atp7b^–/– mice could be due to disrupted nuclear receptor function.

Activation of the FXR/SHP pathway inhibits production of excessive concentrations of bile acids in the liver. Bile acids activate the nuclear receptor FXR, which increases expression of short heterodimer partner (SHP), a nuclear receptor that lacks a DNA-binding domain and functions as a transcriptional repressor (13). SHP binds to another nuclear receptor, liver receptor homolog 1 (LRH-1), and recruits corepressor complexes to the promoters of genes involved in bile acid synthesis (Cyp7a1, Cyp8b1) and basolateral bile acid uptake into hepatocytes (Ntcp), thus decreasing both bile acid synthesis and uptake (10, 14, 15). FXR also induces the expression of canalicular transporters that excrete bile acids (via BSEP), glutathione, and glucuronidated and sulfate-conjugated compounds (via MRP2) into bile (16–18).

Since biliary secretion is a major route for copper excretion, cholestatic disorders may develop hepatic copper levels similar to those found in WD. Patients with progressive familial intrahepatic cholestasis (PFIC) 3 (MDR3 mutation, resulting in loss of biliary phospholipid excretion) (19), primary sclerosing cholangitis (PSC) (20), or primary biliary cholestasis (PBC) (20) have been found to have significantly elevated hepatic copper levels and could share pathological features with WD.

We hypothesized that elevated copper levels could be associated with decreased nuclear receptor function. In accord with this prediction, we found that copper treatment strongly decreased nuclear receptor function in vitro and in cell-based studies. Nuclear receptor function was also significantly impaired in Atp7b^–/– mice, WD patients, and PFIC2 and PFIC3 patients who had elevated hepatic copper levels.

In vitro copper–mediated disruption of nuclear receptor function. The zinc finger containing the DBD of the nuclear receptors is highly conserved. Given the dramatic negative effect of copper on ERα binding and structure (7, 8), we performed in vitro assays to determine the direct effect of copper on metabolic nuclear receptor function. To this end, we added various metals to either FXR or RXR produced by in vitro translation. Addition of 4 μM CuSO₄, but not 1–10 μM nickel or cobalt, to either the FXR or RXR synthetic reaction resulted in loss of FXR:RXR binding to the BSEP promoter FXRE (Figure 1A). Loss of FXR:RXR binding was not due to copper-mediated changes in protein expression (Supplemental Figure 1, A and B; supplemental material available online with this article; doi:10.1172/JCI78991DS1). In a simple competition model, excess zinc would counter the effect of copper on FXR:RXR DNA binding. As expected, coincubating in vitro–translated FXR with RXR synthesized in reactions containing CuSO₄ plus 40 μM ZnSO₄ reversed the negative effect of copper on FXR:RXR binding (Figure 1B).

Figure 1

Copper disruption of DNA binding is reversed by zinc. (A) EMSA of in vitro–translated FXR and RXR with (+) or without 4 μM copper sulfate (Cu⁺²), or 1–10 μM nickel (Ni⁺²) or cobalt (Co⁺²) chloride in the RXR reaction. One hundred–fold excess cold competition with WT (CC) but not mutant cold oligonucleotide (mCC). (B) Nuclear extracts from HepG2 cells treated with 1–4 μM Cu⁺² or 4 μM Cu⁺² plus 5–40 μM ZnSO₄ (Zn⁺²) were incubated with a radiolabeled probe containing the FXRE on the BSEP promoter. (C) HepG2 cells were treated overnight with DMSO, 75 μM CDCA, 10 μM Cu⁺², and 10–40 μM Zn⁺², as indicated. Error bars represent the mean ± SEM. n = 3; *P < 0.05 and ^#P < 0.01 by 1-way ANOVA, followed by Bonferroni’s post-hoc test. (D) EMSA analysis of binding to the human FXRE on the BSEP promoter with nuclear extracts harvested from HepG2 cells treated with 10 μM Cu⁺², 10 μM Cu⁺² plus 40 μM Zn⁺², and 40 μM Zn⁺². Binding of nuclear extracts from HepG2 cells treated with 10 μM Cu⁺² or 10 μM Cu⁺² plus 40 μM Zn⁺² to radiolabeled oligonucleotides containing an (E) HNF4α, (F) TR, (G) SP1 (complex I), (H) GATA, or (I) CREB response element. NE, nuclear extract.

In HepG2 cells, CDCA-mediated induction of BSEP mRNA levels decreased by approximately 90% with CuSO₄ treatment (Figure 1C), and this was correlated with decreased FXR:RXR DNA binding in nuclear extracts (Figure 1D). However, both BSEP mRNA induction and nuclear extract binding were restored with ZnSO₄ treatment in a dose-dependent manner (Figure 1, C and D). These changes occurred in the absence of detectable changes in RXR protein levels (Supplemental Figure 1, C and D). Likewise, binding of proteins from nuclear extracts to HNF4α and thyroid hormone receptor (TR) response elements were abolished by 10 μM CuSO₄ treatment, which was prevented by the addition of 40 μM ZnSO₄ to HepG2 cell cultures (Figure 1, E and F).

To determine whether copper could affect DNA binding of other zinc-containing transcription factors, we examined the Cys₂His₂ transcription factor SP1 and the Cys₄ transcription factor GATA4. Like the nuclear receptors, SP1 DNA binding was lost in the copper-treated cells and restored by cotreatment of HepG2 cells with 40 μM ZnSO₄ (Figure 1G); however, GATA binding was unchanged (Figure 1H). Binding of nuclear extracts to a radiolabeled probe containing the leucine zipper transcription factor CREB consensus element was not decreased with metal treatment (Figure 1I). Thus, the detrimental effects of copper occur in a protein-specific manner.

Atp7b^–/– mice have disrupted hepatic nuclear receptor function and expression of metabolic target gene expression. At 6 weeks of age, Atp7b^–/– mice have increased serum alanine aminotransferase (ALT) and hepatic copper levels, which precedes overt liver pathology (5), and we confirmed these results (Supplemental Figure 2). We found 1.8-fold and 5-fold increases in serum bile acid and bilirubin levels in 3-month-old mice (Figure 2, A and B) but no changes in the total hepatic bile acid pool size (data not shown), which suggested disrupted hepatic bile acid metabolism. Therefore, we measured the mRNA levels of hepatic FXR target genes. In 2- to 3-month-old mice, Bsep and SHP mRNA expression was decreased to 40% of control levels, and SHP was decreased further to 20% of WT control levels at 5 months of age; however, Mrp2 mRNA expression was unchanged (Figure 2C). Despite changes in FXR:RXR basal activity, Atp7b^–/– mice showed a residual response to the potent FXR ligand GW4064 (4.9-fold increase, P = 0.02 for SHP mRNA expression); however, GW4064-induced expression levels of Bsep and SHP mRNA in Atp7b^–/– mice were much lower than those detected in WT mice (Figure 2D).

Figure 2

Nuclear receptor activity in Atp7b^–/– mice. (A) Serum bile acid (BA) and (B) bilirubin (BR) levels were measured in 3-month-old WT and Atp7b^–/– mice. (C and G) Bsep, Mrp2, SHP, Cyp7a1, and Cyp8b1 mRNA expression levels were measured by real-time PCR for WT and Atp7b^–/– mice at 2, 3, and 4–5 months of age. The Atp7b^–/– target gene expression level is a percentage of that of the age-matched WT control. (D) Mice were given DMSO or 50 mg/kg BW GW4064 four hours prior to harvesting livers for analysis of Bsep and SHP mRNA levels. (E, F, and H–J) ChIP was performed using formaldehyde–cross-linked and purified nuclei and FXR, RXR, HNF4α, LRH-1, and RNA Pol II antibodies as indicated. The response elements on each promoter were amplified by real-time PCR and normalized to input. Primers for a gene desert region were used as a negative control (Neg). Error bars represent the mean ± SEM. *P < 0.05 by Student’s t test (A and B, n = 6–12 samples; E and F and H and I, n = 4 samples); *P < 0.05 by 1-way ANOVA, followed by Bonferroni’s post-hoc test (C and G, n = 4–6 samples); P < 0.05 by 2-way ANOVA, followed by Bonferroni’s post-hoc test (*WT versus KO GW4064 treatment, ^#WT versus KO vehicle, and **KO vehicle versus GW4064) (D, n = 4–6 samples).

ChIP analysis revealed that FXR and RXR binding to the Bsep and SHP promoters was reduced by 70% (FXR) and 80% (RXR) in Atp7b^–/– livers relative to WT control livers at 2 to 5 months of age (Figure 2, E and F); binding was not further decreased between 2 and 5 months of age (data not shown). Rxr mRNA levels were decreased, but changes in Fxr mRNA levels were detected only at 5 months of age (Supplemental Figure 3A), in contrast to the in vitro data, which did not show changes in RXR protein expression in samples treated with copper. Chronic copper accumulation in the Atp7b^–/– mouse liver may result in activation of hepatic stellate and Kupffer cells and pathways, such as the inflammatory pathway, that decrease RXR expression (21–25).

LRH-1 and HNF4α are orphan nuclear receptors that are essential for the regulation of bile acid synthetic genes, such as Cyp7a1 and Cyp8b1. Cyp7a1 mRNA expression levels trended lower in Atp7b^–/– mice (45% decrease, P = 0.12 at 3 months of age and 38% decrease, P = 0.14 at 5 months of age) (Figure 2G). LRH-1 binding to the Cyp7a1 promoter was unchanged (Figure 2H), whereas HNF4α binding was significantly decreased in these mice at 5 months of age (Figure 2I). Cyp8b1 mRNA expression levels decreased by 65% in the Atp7b^–/– mice at 2 to 3 months of age and by 44% at 5 months of age relative to WT levels (Figure 2G). As with Cyp7a1, recruitment of HNF4α to the Cyp8b1 promoter decreased by approximately 60% at 5 months of age in the Atp7b^–/– mice (Figure 2I). There was a 60% (P = 0.057) downward trend in LRH-1 binding to the Cyp8b1 promoter in these mice at 2 months of age (data not shown) and a 55% decrease by 5 months of age (Figure 2H). Binding of RNA Pol II to the Gapdh promoter was the same in both WT and Atp7b^–/– livers (Figure 2J). No changes in Lrh1 or Hnf4a mRNA expression were detected in Atp7b^–/– mice at 2 to 5 months of age (Supplemental Figure 3A). Other LRH-1 and HNF4α target genes involved in hepatic metabolism (Ntcp, basolateral transport) and biliary cholesterol transport (AbcG5 and AbcG8), as well as the TR and PPARα target genes Spot14 and G6Pase, had decreased expression in the Atp7b^–/– mice at 2 to 5 months of age (Supplemental Figure 4).

A zinc-enriched diet prevents decreased nuclear receptor function. WD patients are treated with zinc following successful chelation therapy (26–30). Zinc therapy is thought to decrease copper load by competing with copper for absorption in the small intestine (26, 27, 31). Given the protective effect of zinc on nuclear receptor activity in vitro, we tested whether zinc had a protective effect on hepatic nuclear receptors in Atp7b^–/– mice. To maximize zinc exposure prior to significant increases in hepatic copper levels, mice in mating cages were given a chow diet enriched with zinc acetate (1,000 ppm), and the diet was maintained after weaning until the mice were 2 months of age. We then measured target gene expression and nuclear receptor binding. Control littermate mice given the zinc-enriched diet did not have significant changes in target gene expression or nuclear receptor binding relative to control mice given a regular chow diet (data not shown). The zinc-supplemented Atp7b^–/– mice had significant increases in Bsep, SHP, and Rxr mRNA expression levels, whereas Fxr mRNA expression levels were unchanged in the KOs fed regular or zinc-supplemented chow relative to levels in WT controls (Figure 3, A–D). In accord with the mRNA expression levels, ChIP showed that zinc supplementation increased RXR binding to the Bsep and SHP promoters and FXR binding to the Bsep promoter (Figure 3, E–G), and there was an upward trend in binding of FXR to the SHP promoter (P = 0.052) (Figure 3H). Modest, yet significant, improvements were also observed in Ntcp and Cyp8b1 mRNA expression levels (Supplemental Figure 5). Zinc treatment also reduced serum bile acid and bilirubin levels and restored ALT to WT levels (Figure 3, I–K). Taken together, our findings show that zinc treatment prevented significant disruption of nuclear receptor function and improved serum markers of liver function.

Figure 3

In vivo effects of zinc therapy on nuclear receptor activity in Atp7b^–/– mice. (A) Bsep, (B) SHP, (C) Rxr, and (D) Fxr mRNA expression was measured by real-time PCR for WT and Atp7b^–/– (KO mice on a chow diet or on a diet supplemented with 1,000 ppm zinc acetate). ChIP was performed using formaldehyde–cross-linked and purified nuclei and (E and F) RXR or (G and H) FXR antibodies. The response elements on each promoter were amplified by real-time PCR and normalized to input. Primers for a gene desert region of chromosome 6 were used as a negative control. Serum (I) bile acids, (J) bilirubin, and (K) ALT levels were measured for WT, KO, and zinc-supplemented KO (KO + Zn)animals. Data represent the mean ± SEM. *P < 0.05 by 1-way ANOVA, followed by Sidek’s post-hoc test (A–C, n = 3 samples; I and J, n = 3–6 samples). *P < 0.05 by Student’s t test (E–G, n = 3 samples); ^#P = 0.052 by Student’s t test (H, n = 3 samples).

WD patients with elevated hepatic copper have disrupted metabolic nuclear receptor signaling. Aside from findings of decreased biliary copper excretion (32, 33) without pathological changes in total bile acids (34), there are a limited number of studies regarding bile acid metabolism in WD patients. To address whether WD patients exhibit changes similar to those seen in the WD mouse model, we measured nuclear receptor target gene mRNA expression and activity in adult hepatic autopsy samples obtained from the National Disease Research Interchange (NDRI). We did not have access to data on disease pathology or severity, so we measured copper concentrations in the samples by inductively coupled plasma optical emission spectrometry (ICP-OES). Surprisingly, only 1 WD sample had hepatic copper levels in the WD diagnostic criteria range (451 μg/g liver dry weight vs. 25 μg/g liver dry weight) (Figure 4A). Despite this discrepancy, all WD samples had changes in hepatic markers of liver injury and hepatic stellate cell activation, such as increased mRNA of the fibrogenic marker tissue inhibitors of matrix metalloproteinase 1 (TIMP1) (Figure 4B). Collagen 1A1 (COL1A1) mRNA expression was slightly higher in all WD samples (Figure 4B); however, α smooth muscle actin (α-SMA, ACTA2) mRNA, another marker of hepatic stellate cell activation and fibrogenesis, was not changed in the WD samples (Figure 4B). Overall, this may reflect a more advanced stage of cirrhotic liver disease, since α-SMA has been shown to increase with initial stellate cell activation and decrease with prolonged activity (35).

Figure 4

Characterization of adult WD patients. (A) Liver dry weight (μg/g) was determined by ICP-OES. Real-time PCR for mRNA expression of (B) COLA1, ACTA2, TIMP1, (C) BSEP, and (D) SHP from NDRI adult liver tissue samples. (B–D) Samples are grouped according to the degree of hepatic copper levels: control (non–liver disease samples), normal Cu WD (WD samples without elevated hepatic copper levels), and high Cu WD (WD samples with hepatic copper levels that meet WD criteria). For each experiment, samples were measured in duplicate and the average plotted for the individual sample. Each experiment was performed 3 times, and all graphs are representative of 3 separate experiments.

For further analysis, the WD samples were grouped according to copper levels: normal copper (Cu) WD and high Cu WD. The high Cu WD sample had the lowest BSEP and SHP mRNA expression levels (Figure 4, C and D) and binding of FXR and RXR to the BSEP promoter (Figure 5, A and B). Binding of LRH-1 to the BSEP promoter was also lowest in the high Cu WD sample group (Figure 5C). Similarly, binding of FXR, RXR, and LRH-1 to the SHP promoter was lowest in the high Cu WD sample; however, binding was in the range of the control samples (Figure 5, D–F). RNA Pol II binding to a positive control primer for GAPDH was similar among the groups (Figure 5G). CYP7a1 and CYP8b1 mRNA expression levels were not decreased in the WD samples, and LRH-1 and HNF4α binding was in the range of the control and normal Cu WD samples (Supplemental Figure 6). CYP7a1 has a well-defined circadian rhythm (36, 37), and CYP7A1 expression is negatively regulated by miR-122a and miR-422a (38). This finding may reflect differences in CYP7a1 mRNA stability and/or posttranslational mechanisms in the Atp7b^–/– mouse versus the those in fibrotic WD human samples. Similar to Atp7b^–/– mice, patients with WD do not have significant changes in mRNA expression of LRH1, HNF4A, or FXR; however, RXR mRNA was increased in both groups of patients with WD (Supplemental Figure 3B). Despite these differences in RXR expression, both KO mice and WD patients with elevated hepatic copper levels had decreased recruitment of RXR in binding assays.

Figure 5

ChIP analysis of adult control and WD samples. ChIP was performed with (A and D) FXR, (B and E) RXR, (C and F) LRH-1, and (G) RNA Pol II. The response elements on each promoter were amplified by real-time PCR and normalized to input. Primers designed for a gene desert region of chromosome 12 were used as a negative control. Samples are grouped according to the degree of hepatic copper levels: control (non–liver disease samples), normal Cu WD (WD tissue samples from patients without elevated hepatic copper levels), and high Cu WD (WD tissue samples from patients with hepatic copper levels meeting WD criteria). For each experiment, samples were measured in duplicate and the average plotted for the individual sample. Each experiment was performed 3 times, and graphs are representative of 3 separate experiments.

Pediatric WD and PFIC patients have elevated hepatic copper levels and disrupted nuclear receptor function. We also obtained several pediatric biopsy samples from controls (noncholestatic patients with normal hepatic copper levels); from a WD patient treated with the metal chelation therapy trientine (WD/trientine) and given a low-copper diet; and from a newly diagnosed WD patient who was not treated with trientine chelation therapy (WD/no trientine). To determine overall fibrotic changes in the liver, we measured COL1A1, ACTA2, and TIMP1 mRNA and found that ACTA2 and TIMP1 levels were approximately 3-fold and 2-fold higher in the WD/no trientine group than the average levels in the control samples, which the WD/trientine sample overlapped (Figure 6, A and B); however, COL1A1 mRNA expression levels were similar in both WD groups (Figure 6C). The liver sample from the WD/no trientine patient had lower BSEP, SHP, and NTCP mRNA expression levels than did the liver sample from the WD/trientine patient (Figure 6, D–F). The WD/no trientine patient had decreased LRH1, HNF4A, RXR, and FXR mRNA expression levels (Figure 6, G–J), and FXR mRNA expression levels were lower than those in the control samples from the WD/trientine patient. RXR mRNA expression patterns in the pediatric WD/no trientine patient were similar to those in the Atp7b^–/– mice, which may point to a central importance of RXR in the dysregulation of nuclear receptor signaling in WD.

Figure 6

Pediatric cholestasis and copper-mediated effects on nuclear receptors. (A–J) Real-time PCR was performed to measure COL1A1, ACTA2, TIMP1, BSEP, SHP, NTCP, LRH1, HNF4A, RXR, and FXR mRNA in pediatric liver biopsy samples. mRNA levels in tissue samples from controls (C, denoting pediatric patients with normal hepatic copper levels), WD/trientine patients (WD+), WD/no trientine patients (WD–), PFIC2 patients A and B (2A, 2B), and PFIC3 patients A and B (3A, 3B). For each experiment, samples were measured in duplicate and the average plotted for the individual sample. Each experiment was performed 3 times, and graphs are representative of 3 separate experiments. In panels A–J, target gene mRNA expression was normalized to 18S mRNA expression. (K) EMSA analysis of binding to a radiolabeled probe containing an FXRE on the BSEP promoter with nuclear extracts harvested from HepG2 cells or from patients with PFIC2, PFIC3, WD, or cirrhosis or cholestasis. Fifty-fold excess of cold competition with WT (CC), but not mCC, demonstrated specific and nonspecific binding.

We expanded upon the studies of others (19), who demonstrated elevated hepatic copper levels in cholestatic liver disease, and obtained liver biopsy samples from 2 patients with PFIC3 (MDR3 mutation) and from 2 patients with PFIC2 (BSEP mutation). Similar to the samples from adult WD and pediatric WD/no trientine patients, both PFIC2 and PFIC3 samples had elevated mRNA expression of TIMP1; however, COL1A1 and ACTA2 mRNA expression overlapped that of controls (Figure 6, A–C). PFIC3 patient A had hepatic copper levels that were well within the WD criteria (1136 μg/g liver dry weight) and decreased expression levels of LRH1, HNF4A, RXR, BSEP, SHP, and NTCP mRNA that were similar to those detected in the WD/no trientine patient (Figure 6, D–I). For PFIC3 patient B, rhodanine staining demonstrated significant copper staining within zone 1 hepatocytes, as was seen in a WD patient (Supplemental Figure 7), but nuclear receptor and target gene mRNA expression overlapped that of the control pediatric samples (Figure 6, D–F). Hepatic copper levels in PFIC2 patient A were elevated (177 μg/g liver dry weight) but slightly below typical WD levels. Data regarding copper concentrations were not available for PFIC2 patient B, due to the small sample size. Both PFIC2 patients had a downward trend in mRNA expression of BSEP, SHP, and NTCP (Figure 6, D–F). As in the pediatric WD/no trientine patient, LRH1 and HNF4A mRNA levels were decreased in PFIC2 patient B and PFIC3 patient A (Figure 6, G and H). All PFIC patients had RXR and FXR mRNA levels overlapping those of WD/no trientine patients (Figure 6, I and J), and hence lower than control mRNA expression levels.

We confirmed decreased FXR:RXR binding in samples from PFIC2 patient A, PFIC3 patient A, an additional WD patient (865 μg copper/g liver dry weight), and a patient with cholestasis and cirrhosis (727 μg copper/g liver dry weight) using EMSA analysis. Because of the limited sample amounts for each patient, we were unable to perform Western blot or ChIP assays. Nuclear extracts from HepG2 cells were used as a positive control for binding of FXR:RXR to the FXR response element (FXRE) of the human BSEP promoter. Binding in the PFIC2 sample was reduced, yet specific, to the FXRE (Figure 6K). No specific binding was found in the nuclear extracts from the other pediatric nuclear extract samples analyzed (Figure 6K). To the best of our knowledge, these studies are the first to describe changes in nuclear receptor function in WD and to correlate changes in nuclear receptor function with hepatic copper levels in patients with cholestasis.

Copper is an essential trace element that serves as a cofactor for many metabolic enzymes, such as cytochrome c oxidase and ceruloplasmin, but excessive copper levels can result in cell stress and cell death. While much is known about the role of excessive copper in the generation of redox-oxidative stress, copper-mediated disruption of nuclear receptor signaling in vivo has not been described. Herein, we describe changes in hepatic metabolic nuclear receptor activity in the livers of the Atp7b^–/– WD mouse model and of patients with WD and PFIC.

Studies using mass spectrometry electrospray ionization (MS-ESI) confirmed a bicoordinate interaction of the ER with copper (39) and strong copper-induced alterations in protein structure (8). In agreement with the ERα studies, in vitro translation of either FXR or RXR in the presence of copper resulted in loss of FXR:RXR binding to an FXRE. This was not due to changes in the efficacy of the in vitro translation reaction (data not shown) and occurred when a previously generated FXR:RXR complex was incubated overnight at 4°C with copper (data not shown). Strong impairment of nuclear receptor function was also observed in HepG2 cells treated with copper, as shown by inhibition in the CDCA-mediated induction of BSEP mRNA expression and loss of specific DNA binding by FXR:RXR, HNF4α, or TRα:RXR. These copper-mediated decreases in transcription factor function extended to the Cys2His2 transcription factor SP-1, but not the Cys4 transcription factor GATA4 or the leucine zipper transcription factor CREB. Differences in zinc finger coordination among different zinc-containing transcription factors may result in a spectrum of susceptibility to copper interaction with the zinc finger proteins.

In agreement with previous studies, we found that Atp7b^–/– mice at 3 months of age had elevated hepatic copper levels that were within the WD diagnostic criteria, elevated serum ALT, as well as elevated bilirubin and bile acid levels. We measured the activity of nuclear receptors involved in bile acid metabolism and transport, as well as other metabolic targets, and found that mRNA expression of FXR:RXR targets was decreased in these mice at 2 months of age, which precedes full-blown liver pathology (5, 40). Bsep and SHP mRNA were decreased in Atp7b^–/– mice, and the recruitment of FXR and RXR to the Bsep and SHP promoters was strongly decreased. RXR expression was decreased in these mice beginning at 2 months of age, whereas FXR expression was not changed until 5 months of age. While the mechanism for RXR reduction is unclear, decreased RXR could contribute to reduced binding of FXR:RXR to the response elements on the Bsep and SHP promoters.

WD does not result in elevated intestinal copper levels, therefore, copper-mediated effects on nuclear receptor activity should not be manifest in the intestine of Atp7b^–/– animals. GW4064 treatment induced duodenal Fgf15 mRNA expression by approximately 8-fold in both WT and KO animals (data not shown). Studies in intestine- or liver-specific Fxr-KO mice demonstrated that intestinal activation of FGF15 can activate SHP mRNA expression in the absence of hepatic FXR expression (15). Since FGF15 signaling is present in the Atp7b^–/– animals, activation of intestinal FGF15 expression may contribute to the GW4064-mediated activation of SHP mRNA expression.

Despite a decrease in Bsep mRNA expression, Atp7b^–/– mice did not have decreased bile flow (data not shown). Bsep-KO animals on a mixed genetic background do not have significant cholestasis, due to decreased hydrophobicity of the bile acid pool and increased alternative basolateral bile acid transport (41, 42). Likewise, Fxr^–/– mice have increased Mrp4, Cyp3a11, and Cyp2b10 mRNA expression levels as well as increased hydroxylated bile acids (43, 44), and Fxr^–/– animals have increased Mrp4 mRNA expression levels in a bile duct–ligation model of obstructive cholestasis (45). Atp7b^–/– mice also have increased Mrp4 (P < 0.05) and Cyp2b10 mRNA expression levels (P = 0.0557; Supplemental Figure 8), which should increase alternative bile acid detoxification pathways.

It is interesting that the residual FXR remained responsive to the synthetic GW4064 treatment, with ligand treatment increasing Bsep and SHP mRNA expression levels to those of the WT control animals. This suggests a potential therapeutic benefit for ligand-mediated activation of FXR in patients with WD.

We found that the activity of additional hepatic nuclear receptors was also changed in the Atp7b^–/– mouse. At 2 months of age, Cyp8b1 mRNA was significantly decreased, and by 5 months of age, recruitment of LRH-1 and HNF4α to the Cyp8b1 promoter was decreased. Cyp7a1 mRNA was not significantly decreased at this earlier time point, but HNF4α binding to the Cyp7a1 promoter was decreased in these mice at 5 months of age. Other HNF4α target genes not involved in bile acid synthesis or uptake were also decreased (histidine-rich glycoprotein [HRG] and ornithine transcarbamaylase [OTC]); however, Apoc3 mRNA expression was unchanged (Supplemental Figure 9). This supports a role for copper disruption of HNF4α activity, as shown in HepG2 cells (46). In contrast with FXR:RXR, binding of LRH-1 and HNF4α was not significantly changed in these mice until 5 months of age. The transcriptional regulation of Cyp7a1 and Cyp8b1 by nuclear receptors has complex compensatory mechanisms (12, 47). The FXR/SHP signaling axis is a well-known repressor of CYP7a1 via SHP recruitment of corepressor complexes at LRH-1–binding elements (47), and an early decrease in SHP expression may maintain expression of CYP7a1 and other LRH-1 and HNF4α target genes.

A straightforward interpretation of the in vitro and Atp7b^–/– mouse results is that copper directly decreases nuclear receptor function by competing with zinc for occupancy of the DNA-binding domain. However, such direct binding to nuclear receptors in vivo remains to be established, and other mechanisms may also contribute. In cholestatic mouse models such as bile duct ligation, decreased nuclear receptor activity has been attributed to inflammation that resulted in decreased nuclear receptor expression (43, 44, 48, 49). Activation of hepatic stellate cells and inflammatory pathways have been specifically associated with decreased RXR expression (21–25). In accord with this, we observed decreased expression of Rxr mRNA in the Atp7b^–/– livers and in the pediatric WD/no trientine and PFIC samples, but not in the adult patient samples. It is likely that decreased RXR function contributes to the decreased activity of FXR and its other heterodimer partners, but not to the observed effects on HNF4α homodimers and LRH-1 monomers.

Oral zinc administration is a therapy used to treat WD (26, 27, 29). The mechanism of zinc action is thought to compete with copper for intestinal absorption and induction of the metal chelator metallothionein (26, 27, 31). In agreement with previous in vitro studies with ERα (7), we found that increased zinc levels could counteract the impact of copper on specific DNA binding with in vitro binding assays and HepG2 cells. Administration of a zinc-enriched diet to Atp7b^–/– mice reversed the decreased nuclear receptor activity found in KO animals fed a standard chow diet. FXR:RXR target gene mRNA expression recovered to WT levels due to increased Rxr mRNA expression and FXR:RXR binding to FXRE response elements. In contrast with the in vitro studies, increased RXR expression was observed in KO animals supplemented with zinc. In a mouse model of alcohol-induced steatosis, zinc supplementation restored HNF4α and PPARα binding to response elements in EMSA assays, and this coincided with decreased expression of oxidative stress markers (50). No changes in superoxide dismutase (Sod) mRNA expression (data not shown) were observed in the Atp7b^–/– or zinc-supplemented Atp7b^–/– groups relative to WT mice, but zinc-mediated downregulation of copper-mediated redox stress could potentially protect cysteine residues in the nuclear receptor zinc finger from oxidation, in addition to competing with copper for incorporation into the DNA-binding domain.

Alterations in hepatic nuclear receptor target genes in humans were similar to those in the Atp7b^–/– mice. In the adult WD group, only 1 of the 3 patients had significantly elevated hepatic copper concentrations, which could reflect effective chelation therapy in those patients, even though the individual with elevated copper was the only one listed as being treated with trientine. Because of this unexpected finding, we measured the expression of fibrogenic markers and found elevated COL1A1 and TIMP1 mRNA expression levels in all of the WD patients relative to levels in the nondiseased controls. As expected from the Atp7b^–/– results, the high Cu individual had the lowest BSEP and SHP mRNA expression levels and decreased binding of FXR, RXR, and LRH-1 to the BSEP and SHP promoters. CYP7a1 and CYP8b1 mRNA expression in the high Cu samples overlapped that of the other groups, and recruitment of LRH-1 and HNF4α was unchanged, which may reflect loss of SHP-mediated repression of CYP7a1 and CYP8b1. The contrast in LRH-1 recruitment to the BSEP and SHP promoters versus the CYP7a1 and CYP8b1 promoters may be due to cooperative overlapping between HNF4α and LRH-1 binding in the CYP7a1 and CYP8b1 promoters (12, 14, 51, 52). OTC and APOC3 mRNA expression levels were decreased in WD patients with high Cu (Supplemental Figure 9), indicating effects on other HNF4α targets.

We also obtained 2 pediatric WD liver biopsy samples, along with records regarding disease state and drug therapy. One newly diagnosed patient had not received chelation therapy (WD/no trientine), and the other patient had received chelation therapy (WD/trientine). Both the adult patient with elevated hepatic copper and the WD/no trientine patient had decreased mRNA expression of BSEP, SHP, and NTCP, whereas the patient with controlled WD (WD/trientine patient) had mRNA expression of nuclear receptor target genes that more closely overlapped that of the control group.

A group of PFIC3 patients was recently described as having hepatic copper levels similar to those of WD patients (19), and we extended those studies to a set of PFIC3 and PFIC2 samples. Two PFIC3 patients were identified as having elevated hepatic copper levels (patient A: 1,136 μg/g liver dry weight; patient B had significant rhodanine-copper histological staining), and PFIC3 patient A had decreased nuclear receptor and target gene mRNA expression, whereas PFIC3 patient B had mRNA expression patterns similar to those of the control pediatric samples. We also obtained liver biopsy samples from 2 PFIC2 patients: patient A (hepatic copper 177 μg/g liver dry weight) and patient B (hepatic copper content not available). Both PFIC2 patients had mRNA expression levels of BSEP, SHP, and NTCP that were in the range of those detected in the WD/no trientine pediatric patient.

Overall, although only a small number of patient samples were available, the trends in nuclear receptor target gene expression and nuclear receptor DNA binding were similar to those observed in the Apt7b^–/– mice. Thus, we consistently found impaired nuclear receptor binding in both the mouse and adult and pediatric human samples when copper levels were elevated. We conclude that disruption of nuclear receptor activity presents a new, quite unexpected mechanism that could contribute to the pathology of WD and potentially of other disorders that result in excessive and chronic hepatic copper accumulation. Although we have focused on a subset of nuclear receptors, all nuclear receptors, and potentially other zinc-binding transcription factors, may be affected to a greater or lesser degree. Our results also suggest the restoration of nuclear receptor or other transcription factor function as an additional mechanism for the beneficial effects of zinc supplementation.

Animal care. Male and female Atp7b^–/– (C57BLx129S6/SvEv) mice were previously described (5). Animals were given a standard chow diet from Harlan Laboratories. They were provided food and water ad libitum and maintained on a 12-hour light/12-dark cycle and were killed at Zeitgeber time 2–3. For GW4064 (Sigma-Aldrich) treatment, vehicle or GW4064 (50 mg/kg BW) were given by i.p. injection 4 hours before sacrifice (53). For zinc-feeding studies, male and female Atp7b^+/– mice in mating cages were given either a chow-matched diet or a zinc-enriched chow diet (1,000 ppm; Teklad Diets; Harlan Laboratories). Mice were maintained on the diets after weaning until 2 months of age.

Human specimens. ICP-OES was used to determine the μg/g liver dry weight of adult liver tissue. The pediatric control group consisted of male and female patients (1 with a benign tumor, 1 with developmental tumors, and 1 with Budd-Chiari syndrome) who did not have excessive hepatic copper levels as measured by rhodanine staining. Before beginning chelation therapy and dietary copper restriction, the WD/trientine patient had hepatic copper concentrations of 1,108 μg/g liver dry weight, and the newly diagnosed patient who had not begun treatment (WD/no trientine) had a +4 score of rhodanine staining (+4 indicating the most significant degree of staining). Liver biopsy samples from an additional WD patient (865 μg/g liver dry weight) and from a patient with cirrhosis and cholestasis (727 μg/g liver dry weight) were used for EMSA analysis. PFIC3 patient A had hepatic copper concentrations of 1,136 μg/g liver dry weight, and PFIC3 patient B had a +4 rhodanine score. PFIC2 patient A had hepatic copper levels of 177 μg/g liver dry weight; however, copper data were not available for the PFIC2 patient B.

Cell culture, EMSA, and ChIP analyses. The tissue culture core at Baylor College of Medicine provided HepG2 cells. EMSA analysis was performed with 2 μl FXR and RXR synthesized with the Rabbit Reticulocyte Translation System (Promega) with or without metals (copper sulfate, zinc sulfate, nickel chloride, or cobalt chloride), as indicated in the figure legends, and γ-P³²–labeled double-stranded oligonucleotide containing the FXRE on the BSEP promoter (18). HepG2 cells treated overnight with DMSO, 75 μM CDCA, 10 μM copper sulfate, and 10–40 μM zinc sulfate and nuclear extracts were harvested (Pierce, Life Technologies), and 5 μg was used for EMSA analysis as described above. Nuclear extracts (5 μg) from HepG2 cells not treated or treated with 10 μM copper sulfate or 10 μM copper sulfate plus 40 μM zinc sulfate were used in the EMSA with γ-P³²–labeled double-stranded oligonucleotides containing HNF4α, TR, SP1, GATA, and CREB consensus response elements (Santa Cruz Biotechnology Inc.).

ChIP assays were performed as previously described with the following modifications (54): freshly harvested livers (0.5–1.0 g) were homogenized in PBS containing 1% formaldehyde and incubated at room temperature for 10 minutes; incubation was stopped by the addition of 2.2 M sucrose in 150 mM glycine, 10 mM HEPES (pH 7.6), 15 mM KCl, 2 mM EDTA, 0.15 mM spermine, 0.5 mM spermidine, 0.5 mM DTT, and 1X protease inhibitor cocktail (Roche). The liver homogenate was layered onto a 2.05 M sucrose buffer containing 10% glycerol and 150 mM glycine, 10 mM HEPES (pH 7.6), 15 mM KCl, 2 mM EDTA, 0.15 mM spermine, 0.5 mM spermidine, 0.5 mM DTT, and 1X protease inhibitor cocktail (Roche) and centrifuged at 100,000 g for 30 minutes. Purified nuclei (10–50 μg) were sonicated to between 200 and 300 bp and incubated with 10 μg FXR (H-130X), RXR (D-20X), HNF4α (H-171X) antibody (Santa Cruz Biotechnology Inc.), LRH-1 (catalog PP-H2325-10; Perseus Proteomics Inc.), and RNA Pol II (catalog 39097; Cell Signaling Technology) using the Active Motif EZ-ChIP kit. Primers for a gene desert region on chromosome 6 (mouse promoter, catalog 71011; Active Motif) or chromosome 12 (catalog 71001; Active Motif) were used as a negative control. Primers for the mouse and human CYP7A1 (55), CYP8B1 (56), BSEP (57), and SHP (57) were used for real-time PCR analysis of the immunoprecipitated regions.

Human tissue staining. Pediatric liver biopsy samples were stained with rhodanine staining and counterstained with hematoxylin using standard protocols. The degree of copper accumulation was graded according to previously described criteria (58).

Real-time PCR and Western blot analysis. RNA was harvested from approximately 100-mg liver samples in TRIzol reagent (Invitrogen, Life Technologies), and synthesized cDNA (Invitrogen, Life Technologies) was used as a template for real-time PCR with SYBR Green reagent (Roche) on a 384-well LightCycler (Roche). Primers were designed using Roche Universal Primer web-based software. Reactions using mouse cDNA were normalized to 36B4 mRNA expression, and reactions using human cDNA were normalized to 18S mRNA expression.

Western blot analysis was performed with 5 μg nuclear extract or 2 μl FXR plus RXR synthesized with the Rabbit Reticulocyte Translation System using 1:1,000 α-FXR (H-130; catalog sc-13063; Santa Cruz Biotechnology Inc.); RXR (D-20; catalog sc-553; Santa Cruz Biotechnology Inc.); HNF4α (H-171; catalog sc-8987; Santa Cruz Biotechnology Inc.); LRH-1 (catalog PP-H2325-10; Perseus Proteomics Inc.); CYP7A1 (N-17; catalog 14423; Santa Cruz Biotechnology Inc.); BSEP (F-6; catalog sc-74500; Santa Cruz Biotechnology Inc.); COL1A1 (C-18; catalog sc-8784; Santa Cruz Biotechnology Inc.); TIMP1 (catalog H-150; catalog sc-5538; Santa Cruz Biotechnology Inc.); α-SMA (catalog ab15734; Abcam); β-actin (catalog 13E5; Cell Signaling Technology); and histone H3 (catalog 9715; Cell Signaling Technology); or a 1:50,000 secondary antibody.

Serum analysis. Serum was collected ALT and bilirubin levels measured in the Comparative Pathology Laboratory at Baylor College of Medicine. Total serum bile acids were measured with an enzymatic kit (BIOQUANT Image Analysis).

Inductively coupled plasma–optical emission spectroscopy. Liver samples were dried at 60°C for 3 to 4 days to achieve a stable dry weight. Weighed samples were digested using concentrated nitric acid and hydrogen peroxide. Digestates were resuspended in 2% ultra-pure nitric acid (w/w) and analyzed for minerals using ICP-OES (CIROS ICP Model FCE12; Spectro) as previously described (59).

Statistics. Error bars represent the mean ± the SEM. A 2-tailed Student’s t test was used to compare differences between 2 groups, and a P value of less than 0.05 was considered statistically significant. For multiple group comparisons, 1-way or 2-way ANOVA, followed by either Bonferroni’s or Sidak’s post-hoc test, were performed, and a P value of less than 0.05 was considered statistically significant. All statistical analyses were performed with GraphPad Prism software, versions 5.0 and 6.0 (GraphPad Software).

Study approval. All animal experiments were performed according to NIH guidelines (Guide for the Care and Use of Laboratory Animals, 8th ed. The National Academies Press. 2011.), and animal experiments were approved by the IACUC of Baylor College of Medicine.

The IRB of Baylor College of Medicine approved the use of deidentified human tissue from the NDRI (Philadelphia, Pennsylvania, USA) and the Department of Pathology of Texas Children’s Hospital (Houston, Texas, USA). Adult male and female human liver biopsy samples were obtained from the NDRI and pediatric samples from the Department of Pathology of Texas Children’s Hospital. The protocols for sample use were approved by the IRB of Baylor College of Medicine.