Copper Availability Contributes to Iron Perturbations in Human Nonalcoholic Fatty Liver Disease

GASTROENTEROLOGY 2008;135:680 – 688

Copper Availability Contributes to Iron Perturbations in Human Nonalcoholic Fatty Liver Disease ELMAR AIGNER,* IGOR THEURL,‡ HEIKE HAUFE,§ MARKUS SEIFERT,‡ FLORIAN HOHLA,* LUDWIG SCHARINGER,* FELIX STICKEL,储 FREDERIC MOURLANE,储 GÜNTER WEISS,‡ and CHRISTIAN DATZ* *General Hospital Oberndorf, Department of Internal Medicine, Oberndorf, Austria; ‡Department of General Internal Medicine, Clinical Immunology, and Infectious Diseases, Medical University of Innsbruck, Innsbruck, Austria; §Department of Pathology, Private Medical University Salzburg, Salzburg, Austria; and 储Department of Clinical Pharmacology, University of Berne, Berne, Switzerland

See Acton RT et al on page 934 in CGH.

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

Background & Aims: Iron perturbations are frequently observed in nonalcoholic fatty liver disease (NAFLD). We aimed to investigate a potential association of copper status with disturbances of iron homeostasis in NAFLD. Methods: We retrospectively studied 140 NAFLD patients and 25 control subjects. Biochemical and hepatic iron and copper parameters were analyzed. Hepatic expression of iron regulatory molecules was investigated in liver biopsy specimens by reverse-transcription polymerase chain reaction and Western blot analysis. Results: NAFLD patients had lower hepatic copper concentrations than control subjects (21.9 ⴞ 9.8 vs 29.6 ⴞ 5.1 ␮g/g; P ⴝ .002). NAFLD patients with low serum and liver copper concentrations presented with higher serum ferritin levels (606.7 ⴞ 265.8 vs 224.2 ⴞ 176.0 mg/L; P < .001), increased prevalence of siderosis in liver biopsy specimens (36/46 vs 10/47 patients; P < .001), and with elevated hepatic iron concentrations (1184.4 ⴞ 842.7 vs 319.9 ⴞ 451.3 ␮g/g; P ⴝ .020). Lower serum concentrations of the copper-dependent ferroxidase ceruloplasmin (21.7 ⴞ 4.1 vs 30.4 ⴞ 6.4 mg/dL; P < .001) and decreased liver ferroportin (FP-1; P ⴝ .009) messenger RNA expression were found in these patients compared with NAFLD patients with high liver or serum copper concentrations. Accordingly, in rats, a reduced dietary copper intake was paralleled by a decreased hepatic FP-1 protein expression. Conclusions: A significant proportion of NAFLD patients should be considered copper deficient. Our results indicate that copper status is linked to iron homeostasis in NAFLD, suggesting that low copper bioavailability causes increased hepatic iron stores via decreased FP-1 expression and ceruloplasmin ferroxidase activity thus blocking liver iron export in copper-deficient subjects.

N

onalcoholic fatty liver disease (NAFLD) is the hepatic manifestation of the metabolic syndrome.1 The term insulin resistance-associated hepatic iron overload

(IR-HIO) syndrome describes the frequent association between hepatic steatosis and iron accumulation, as reflected by increased serum ferritin along with normal or only slightly elevated transferrin saturation.2 Iron accumulation in NAFLD is usually mild and typically deposited in hepatocytes and sinusoidal Kupffer cells and, thus, distinct from the pattern commonly encountered in hereditary hemochromatosis.3 The understanding of iron metabolism has grown recently by the identification of key iron regulatory molecules.4 Whereas cells acquire iron via different pathways including the uptake of transferrin bound iron via transferrin receptors and of ferrous iron via the transmembrane protein divalent metal transporter-1,4 respectively, so far only 1 iron exporter has been characterized, comprising a transmembrane protein termed ferroportin (FP-1) or IREG-1.5,6 Hepcidin is a master iron regulatory peptide,7 secreted mainly by hepatocytes in response to iron perturbations, inflammation, anemia, and hypoxia.8 Hepcidin exerts its regulatory functions on iron homeostasis via binding to FP-1, causing phosphorylation, internalization, and degradation of FP-1 and thus leads to the sequestration of iron by blocking its cellular export.9 It is well acknowledged that iron and copper metabolism are closely linked10: whereas the enterocyte brush border enzyme cytochrome B reductase is involved in both ferric and cupric reduction,11 the multicopper ferroxidase hepaestin, located at the basolateral membrane of duodenal enterocytes, is essential for cellular iron export via FP-1.12,13 Mice deficient in hephaestin (sla-mice) present with normal apical iron uptake but defective iron transfer to serum transferrin thus resulting in anemia.14 Ceruloplasmin ferroxidase activity is crucial for mobilization of iron from storage sites for incorporation into transferrin and metabolic utilization.15,16 Hence, aceruloplasminemia results in increased tissue iron with high ferritin levels and mild anemia,17 and ceruloplasmin knockout mice accumulate iron in the Abbreviations used in this paper: FP-1, ferroportin-1; IR-HIO, insulin resistance-associated hepatic iron overload; MS, metabolic syndrome; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis. © 2008 by the AGA Institute 0016-5085/08/$34.00 doi:10.1053/j.gastro.2008.04.007

liver and the reticuloendothelial system.18 Likewise, dietary copper deficiency induces characteristic changes in expression of iron metabolism genes and impairs iron transport across the basolateral enterocyte membrane.19 In this respect, copper has been shown to affect the expression of FP-1 in J774 macrophages20 and Caco-2 cells.21 Moreover, membrane-bound ceruloplasmin is required for the stability of FP-1, possibly explaining cerebral iron accumulation in patients with aceruloplasminemia.22 Finally, ceruloplasmin availability modifies iron loading in HFE knockout mice.23 The present study aimed to investigate a possible contribution of variations of copper status and expression of the multicopper ferroxidase ceruloplasmin to iron perturbations in human NAFLD.

Patients and Methods Patients One hundred forty consecutive patients with a diagnosis of NAFLD identified from our liver biopsy database were studied retrospectively. These patients had diagnostic liver biopsies performed at 2 different academic institutions between January of 2000 and December of 2006. Subjects with a history of relevant alcohol intake (⬎20 g/day), viral hepatitis, autoimmune hepatitis, primary biliary cirrhosis, Wilson’s disease, or ␣-1-antitrypsin deficiency or patients on known steatogenic medication were excluded. None of the study patients had signs of cardiac or renal insufficiency or suffered from cancer, autoimmune diseases, or systemic infections. Patient charts were reviewed with regard to cigarette smoking and use of oral contraceptive medication because these factors are known to affect ceruloplasmin levels and functional properties.16,24 To evaluate the contribution of copper to changes in NAFLD iron homeostasis, NAFLD patients were divided into tertiles according to their copper status. In 76 of 140 NAFLD patients, intrahepatic copper concentration was available. Patients were categorized into 3 groups according to low hepatic copper (⬍18 ␮g/g, 26 patients), intermediate hepatic copper (18 ␮g/ g–28 ␮g/g, 25 patients), and high hepatic copper status (⬎28 ␮g/g, 25 patients), respectively. Because there was a highly significant correlation between hepatic and serum copper concentrations in these 76 patients (R ⫽ 0.451; P ⫽ .005), the remaining 64 of 140 NAFLD patients without available hepatic copper concentrations were also stratified into tertiles (see above) based on serum copper levels (normal range 70-130 ␮g/L; low serum copper levels [⬍95 ␮g/L, 21 patients], intermediate serum copper levels [96 –121 ␮g/L, 22 patients], and high serum copper levels [⬎122 ␮g/L, 21 patients]). In total, 47 patients were studied as NAFLD with low serum and liver copper levels, 47 as NAFLD with intermediate liver and serum copper levels, and 46 patients as NAFLD with high liver and serum copper levels. Because liver biopsy

COPPER AND IRON IN NONALCOHOLIC FATTY LIVER

681

specimens from truly healthy subjects could not be obtained, 25 subjects (16 females, 9 males) who underwent liver biopsy for unexplained elevation of liver enzymes were studied as “control” subjects. These patients had no evidence of liver disease and normal liver histology and were considered an acceptable control cohort for our analysis. All control subjects had normal biochemical iron and copper parameters. In none of these individuals were iron deposits detected histologically. Hepatic iron and copper concentrations were available in 18 of 25 control subjects. To screen for hereditary hemochromatosis associated, genetic testing for C282Y and H63D mutations of the HFE gene was performed in all patients as described.25 Among NAFLD patients and control subjects included in the analysis, neither homozygosity for the C282Y and H63D mutation nor compound heterozygosity for these mutations was detected. A diagnosis of diabetes was made when fasting glucose levels were above 110 mg/dL, glycosylated hemoglobin (HbA1c) was 6.0% or higher, or patients were on oral antidiabetic medication. Because insulin is known to influence ceruloplasmin expression, diabetic individuals on insulin therapy were excluded from the analysis.26 To assess the prevalence of the metabolic syndrome (MS) in the study groups, we aimed to identify subjects fulfilling World Health Organization (WHO) clinical criteria for MS.27 Because the waist/hip ratio and urinary albumin excretion were not available in our study population, patients were diagnosed as having MS when they suffered from insulin resistance defined as type 2 diabetes, impaired fasting glucose (100 –126 mg/dL), or impaired glucose tolerance (pathologic glucose tolerance test) plus any 2 of the following: (1) arterial hypertension (use of antihypertensive medication and/or high blood pressure ⬎140 mm Hg systolic or ⬎90 mm Hg diastolic), (2) serum triglycerides ⬎150 mg/dL, (3) low high-density lipoprotein (HDL) cholesterol ⬍35 mg/dL in men or ⬍39 mg/dL in women, or (4) body mass index (BMI) ⬎30 (kg/m2) were present. Written informed consent was obtained from all study participants to use clinical data and obtain biopsy specimen for scientific purposes, and the study was performed in accordance with the ethical standards set forth by the Helsinki Declaration of 1975 and revised in 1983.

Histologic Examination of Liver Biopsy Samples Liver biopsy specimens were fixed in buffered formalin and embedded in paraffin. Sections were stained with H&E and Mallory trichrome for morphologic evaluation and Perl’s stain to determine liver iron deposition. All liver biopsy specimens were first assessed independently by 2 pathologists unaware of clinical data and the study objective. In cases of discrepant results of the histologic examination, samples were jointly reevaluated

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

August 2008

682

AIGNER ET AL

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

by the 2 pathologists, and the result agreed on was used for analysis. Histologic examination of liver biopsy specimens was performed according to criteria proposed by Brunt et al.28 Biopsy specimen were evaluated for the degree of macrovesicular steatosis 0 –3 (0, ⬍5%; 1, up to 33%; 2, 34%– 66%; 3, ⬎66% of hepatocytes affected) and hepatocellular ballooning (0, no ballooning; 1, mild; and 2, marked ballooning). Inflammation was graded 0 –3 (mild, moderate, and severe) according to criteria proposed, and portal inflammation was graded 0 –3 as described. Fibrosis was scored as fibrosis stage 1, zone 3 pericellular fibrosis; stage 2, pericellular and portal fibrosis; stage 3, bridging fibrosis; stage 4, cirrhosis. The diagnosis of NAFLD was established in liver biopsy specimens showing steatosis with or without evidence of steatohepatitis (inflammation and hepatocyte ballooning, with or without Mallory’s hyaline or fibrosis) associated with an increased BMI and alcohol consumption of less than 20 g per day. Furthermore, biopsy specimens with steatosis and mild lobular inflammation but without ballooning or perisinusoidal fibrosis were grouped with steatosis and were not classified as nonalcoholic steatohepatitis (NASH). Biopsy specimens with steatosis and any grade of fibrosis other than above were classified as NASH. In addition, biopsy specimens with steatosis and lobular inflammation (1–3) and hepatocellular ballooning (1–2) were considered as NASH in the absence of fibrosis. Siderosis was determined semiquantitatively upon histopathologic examination of Perl’s-stained liver biopsy specimens: score 0, granules absent or barely discernible at a magnification of 400-fold (400⫻); 1, barely discernible at a magnification of 200⫻ but easily confirmed at 400⫻; 2, discrete granules resolved at 100⫻ magnification; 3, discrete granules resolved at a magnification of 25⫻; 4, massive granules visible even upon 10⫻ magnification.29

Laboratory Evaluation Venous blood was drawn following an overnight fast for determination of liver function tests; a full blood count; serum iron status including ferritin, transferrin, transferrin saturation and serum iron; copper; ceruloplasmin; C-reactive protein; fasting glucose; and lipids and erythrocyte sedimentation rate by standardized automated laboratory methods. Hepatic iron concentration and hepatic copper concentration were determined by automated mass spectroscopy analysis in patients and control subjects and calculated as micrograms/grams of dry weight. Insulin was measured by standard laboratory techniques, and insulin resistance was calculated using homeostasis model assessment (HOMA-IR; fasting insulin (␮mol/L)*fasting glucose (mmol/dL)/22.5). Serum for determination of fasting insulin was available from 12 control subjects and 75 NAFLD patients (29 patients with low, 26 with intermediate, and 20 with high serum

GASTROENTEROLOGY Vol. 135, No. 2

or liver copper concentrations). Details of RNA extraction and reverse-transcription polymerase chain reaction (RT-PCR) as well as protein extraction and Western blot analysis are provided as online supporting material (see Supplementary Material online at www.gastrojournal. org).

Experimental Copper Deficiency in Vivo Eighteen Sprague Dawley rats were kept on Kliba Petfood Purified Diet (CH-4303; Kliba Petfood, Kaiseraugst, Switzerland), 15% casein, for 8 weeks. Thereafter, these rats were grouped randomly to receive either copper-depleted or normal diet. Copper-depleted diet contained 2 ppm copper and resulted in a daily copper intake of 0.05 mg. Normal diet contained 100 ppm of copper resulting in a daily copper intake of 2.47 mg. After 4 weeks of feeding specific diets, rats were killed. Liver protein was extracted and FP-1 Western blot analysis performed as described.30

Statistical Analysis Statistical analyses were carried out using SPSS statistics package (SPSS, Inc, Chicago, IL). Calculations for statistical differences in clinical and laboratory characteristics between the various groups were carried out by ANOVA or nonparametric Kruskal–Wallis test. Proportions were compared using Fisher exact test and ␹2 method. Student t test or Mann–Whitney U test in case of non-Gaussian distribution of parameters was used to calculate differences in hepatic mRNA expression of iron metabolism genes as determined by RT-PCR technique. Associations among the various parameters in the different groups were calculated using Spearman rank correlation technique and a Bonferroni correction for multiple testings.

Results General Characteristics and Metabolic Features of the Study Population Patient groups were not different regarding age and rates of heterozygosity for carriers of HFE C282Y and H63D mutations. HFE mutation carrier rates were comparable with the prevalence of the HFE mutations of the white population studied.25 Among NAFLD patients, we detected a higher degree of hepatic steatosis, a higher BMI, an increased prevalence of diabetes, and higher triglyceride levels in patients with low as opposed to those with high serum and liver copper concentrations. No differences were found with regard to the prevalence of fibrosis on liver biopsy, histologic diagnosis of NASH, liver transaminase levels, fasting glucose, total cholesterol, and low-density lipoprotein (LDL) and HDL cholesterol levels as well as the number of patients smoking (for details, see Table 1). Only 4 women took oral contraceptive medication, which excludes a relevant effect of

August 2008

COPPER AND IRON IN NONALCOHOLIC FATTY LIVER

683

Table 1. Clinical, Histologic, and Biochemical Characteristics of Control Subjects and NAFLD Patients According to Copper Concentrations

Number of Pts Female (%) Age, y HFE C282Y Heteroc. (%) HFE H63D Heteroc. (%) Pts with NASH (%) Hepatic steatosis (%) Pts with fibrosis (%) Degree of siderosis, 0⫺4 Hepatic copper, ␮g/gb Serum copper, 70⫺130 mg/L BMI, kg/m2 BMI ⬎30 Fasting glucose, mg/dL Diabetes HOMA-IRb Triglycerides, mg/dL Triglycerides ⬎150 Low HDL Hypertension Metabolic syndrome Serum iron, 59⫺158 mg/L Tf Sat (%) AST, 10⫺32 U/mL ALT, 10⫺32 U/mL AP, 40–129 U/mL GGT, 10⫺71 U/mL Cholesterol, mg/dL LDL cholesterol, mg/dL HDL cholesterol, mg/dL Hemoglobin, g/dL Smokers (%)

25 16 (64.0) 43.6 (⫾9.3) 3 (12) 4 (16) 0 (0) 0 (⫾0) 0 (0) 0 29.6 (⫾5.1) 119.7 (⫾25.7) 25.5 (⫾2.2) 1 (4%) 89.8 (⫾6.6) 0 (0%) 1.69 (⫾0.47) 107.0 (⫾54.4) 4 (16%) 1 (4%) 2 (8%) 0 (0%) 103.6 (⫾33.8) 26.9 (⫾10.3) 32.5 (⫾12.9) 46.6 (⫾22.1) 96.7 (⫾46.3) 109.8 (⫾52.6) 224.7 (⫾48.9) 140.3 (⫾48.4) 70.0 (⫾18.9) 14.6 (⫾1.5) 4 (16)

Low copper

Intermediate

46 16 (35.8) 53.7 (⫾8.4)a 4 (8.7) 7 (15.2) 10 (21.3)a 36.5 (⫾19.4) 8 (17.4) 0.91 (⫾0.62)a 11.9 (⫾3.9) 93.4 (⫾11.8) 29.0 (⫾3.4)a 17 (37.0%) 108.8 (⫾14.9)a 18 (39.1%) 5.09 (⫾3.71)a 211.1 (⫾116.1)a 26 (56.5%)a 9 (19.6%) 18 (39.1%)a 18 (39.1%)a 135.8 (⫾37.4)a 36.4 (⫾8.1)a 40.5 (⫾21.5) 66.4 (⫾41.9)a 82.0 (⫾23.1)a 61.2 (⫾36.9) 223.1 (⫾48.7) 141.0 (⫾45.6) 51.8 (⫾17.3) 15.6 (⫾1.2)a 5 (10.8)

47 21 (44.7%) 51.8 (⫾12.2)a 4 (8.5) 6 (16%) 7 (14.9%)a 32.6 (⫾16.5) 6 (12.8) 0.68 (⫾0.63)a 22.5 (⫾4.0) 106.3 (⫾13.7) 28.3 (⫾2.3)a 13 (27.7%) 106.1 (⫾18.0)a 14 (29.8%) 4.24 (⫾3.44) 156.6 (⫾84.9)a 22 (46.8%) 6 (12.3%) 13 (27.7%) 9 (19.1%) 115.4 (⫾39.6) 33.3 (⫾9.7) 40.8 (⫾17.8) 70.9 (⫾39.5)a 82.5 (⫾26.9)a 98.1 (⫾77.1) 215.5 (⫾44.1) 141.1 (⫾37.9) 54.2 (⫾14.4) 15.5 (⫾1.2)a 6 (12.7)

High copper 47 19 (40.4) 49.4 (⫾9.6)a 3 (6.4) 7 (14.9) 6 (12.7%)a 25.8 (⫾16.7) 7 (14.9) 0.21 (⫾0.41)a 34.4 (⫾8.0) 128.6 (⫾13.1) 27.8 (⫾3.0)a 11 (23.4%) 103.3 (⫾17.2)a 10 (21.3%) 2.89 (⫾1.74) 153.0 (⫾68.1)a 22 (46.8%) 5 (10.6%) 9 (19.1%) 6 (12.7%) 103.2 (⫾39.6) 28.4 (⫾6.8) 34.6 (⫾17.3) 59.6 (⫾35.9) 98.9 (⫾42.6) 90.2 (⫾55.7) 233.1 (⫾46.1) 149.5 (⫾42.0) 55.9 (⫾17.9) 15.1 (⫾1.5) 5 (10.6)

P value (ANOVA) — — — — — — .007 — ⬍.001 ⬍.001 ⬍.001 .016 — .321 .073 .009 .023 — — .041 .005 ⬍.001 .013 .181 .451 .029 .229 .324 .371 .294 .101 —

NOTE. Data are shown as means (⫾SD). P value (ANOVA) denotes significance level comparing NAFLD patients grouped as low serum and liver copper compared with patients grouped as high serum and liver copper as calculated by ANOVA. Proportions were compared using Fisher exact test. Pts, patients; NASH, nonalcoholic steatohepatitis; HIC, hepatic iron concentration; AST, aspartate aminotransferase; ALT, alanine aminotransferase; AP, alkaline phosphatase; GGT, ␥-glutamyl-transpeptidase; Tf Sat, transferrin saturation; (%) refers to percentage of patients in each study group. aDenotes significance at P ⬍ .05 as compared with control subjects. bData available from 76 (hepatic copper concentration) or 75 patients (HOMA-IR).

these drugs to ceruloplasmin concentrations in our cohort because these patients did not differ from the other patients of the respective group. In general, NAFLD subjects had lower hepatic copper concentrations (21.9 ⫾ 9.8 ␮g/g) compared with control subjects (29.6 ⫾ 5.1, P ⫽ .002). Patients with low serum or liver copper concentrations revealed a higher prevalence of isolated criteria of MS and fully established MS according to the modified WHO criteria applied and increased insulin resistance as calculated by HOMA-IR (Table 1). Moreover, HOMA-IR showed an inverse correlation with serum and liver variables of copper metabolism (Table 2). The comparison of 23 NASH patients with 117 NAFLD patients did not reveal a significant difference in serum or liver iron and copper parameters between these groups. However, we found that patients who presented with positive criteria

of MS had significantly lower hepatic copper concentrations as compared with patients who did not meet criteria for MS (P ⫽ .029). We found the prevalence of NASH to be higher in patients with low serum or liver copper concentrations, although this did not reach statistical significance (Table 1).

Serum Iron Parameters and Hepatic Iron Accumulation NAFLD patients with low liver and serum copper concentrations had significantly higher serum ferritin concentrations than NAFLD patients with high serum and liver copper concentrations or control subjects (Figure 1A). Generally, NAFLD patients with intermediate serum and liver copper presented with higher biochemical indices of iron metabolism compared with NAFLD patients with high serum and liver copper.

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

Control

684

AIGNER ET AL

GASTROENTEROLOGY Vol. 135, No. 2

Table 2. Selected Clinical and Biochemical Correlations Observed in NAFLD Patients Parameter Serum Cu R P Caeruloplasmin R P HCC R P Siderosis R P Steatosis R P Ferritin R P BMI R P

Caeruloplasmin

HCCa

Siderosis

Steatosis

Ferritin

BMI

HOMA-IRa

0.704 ⬍.001

0.451 .005

⫺0.487 .001

⫺0.226 .018

⫺0.521 ⬍.001

⫺0.217 .038

⫺0.555 ⬍.001

0.279 .074

⫺0.247 .009

⫺0.106 .266

⫺0.247 .009

⫺0.193 .068

⫺0.535 ⬍.001

⫺0.336 .006

⫺0.398 .001

⫺0.411 ⬍.001

⫺0.221 .084

⫺0.406 .007

⫺0.105 .241

0.772 .001

0.181 .048

0.272 .009

0.106 .215

0.188 .039

0.265 .010

0.255 .005

0.406 .001 0.320 .003

NOTE. Values of Spearman rank correlation analysis are presented as R represents correlation coefficient and P represents significance level. Serum Cu, serum copper; HCC, hepatic copper concentration; BMI, body mass index; HOMA-IR, insulin resistance as calculated by homeostatic model assessment. aData available from 76 (hepatic copper concentration) or 75 patients (HOMA-IR).

Hepatic siderosis was most prevalent among patients with low serum and liver copper concentrations with 36 of 46 patients (78.3%) showing various degrees of hepatic siderosis and only 10 of 47 (21.3%) patients with high copper levels (Figure 1B). In addition, the degree of sid-

erosis was found to be higher in patients with low serum and liver copper concentrations than in NAFLD patients with high serum and liver copper concentrations (Table 1). In accord, hepatic iron concentration was found to be higher in NAFLD patients with low liver and serum

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

Figure 1. Distribution of serum ferritin (A), transferrin saturation (B), hepatic iron concentration (HIC; C) and prevalence of patients with histologic siderosis (D) between the groups studied. Values are presented as means (horizontal lines), 25th and 75th percentiles (boxes), and minimum/maximum ranges. Calculations for statistically significant differences between groups were performed by ANOVA (A, C, D) and by nonparametric KruskalWallis test (D). Abbreviations: control, control group; low, NAFLD with low serum or liver copper; intermediate, NAFLD grouped as interm. serum or liver copper; high, NAFLD with high serum or liver copper; HIC, hepatic iron concentration.

August 2008

COPPER AND IRON IN NONALCOHOLIC FATTY LIVER

685

copper concentration as compared with NAFLD patients with high serum and liver copper levels and control subjects (Figure 1C). Inverse correlations as calculated by Spearman rank correlation analysis between low liver or serum copper concentrations and parameters of iron accumulation are summarized in Table 2.

Serum Ceruloplasmin Levels and Expression of Iron Metabolism Genes in Liver Biopsy Specimens

sion was significantly lower in NAFLD patients with low hepatic copper levels as compared with NAFLD with high liver copper concentrations. In contrast, mRNA expression of the key iron regulatory molecule hepcidin was significantly increased in NAFLD patients with low copper as compared with NAFLD patients with high liver copper and control subjects. However, no differences were observed regarding transferrin receptor 1 (data not shown) and hemojuvelin mRNA expression (Figure 2D) between the various groups. In addition, hepatic FP-1 protein expression was determined by Western blot analysis, which showed significantly lower FP-1 protein expression in NAFLD patients than in control subjects. However, among NAFLD patients, no differences in FP1 protein expression were observed (Figure 3).

Patients with low liver copper concentrations also presented with significantly lower serum copper (Table 1) and ceruloplasmin levels (Figure 2A). To investigate potential mechanisms underlying the inverse relationship between copper and iron concentrations in NAFLD patients, we determined messenger RNA (mRNA) expression of key iron regulatory molecules in liver biopsy specimens. The mRNA expression of the iron exporter FP-1 was significantly decreased in all NAFLD patients as compared with controls. Moreover, FP-1 mRNA expres-

To investigate a putative causative effect of copper depletion on the expression of hepatic FP-1 in vivo,

Figure 3. Western blot analysis of FP-1 protein expression in liver biopsy specimens. Representative liver biopsy specimens with high and low liver copper concentrations were used for protein extraction and detection of FP-1 protein by Western blot analysis. Abbreviations: control, control subjects; low copper, NAFLD with low liver copper concentration; high copper, NAFLD with high liver copper concentration.

Figure 4. FP-1 protein expression in copper deficient rats. Rats were fed a copper-deficient diet for 4 weeks. In these rats, we detected significantly lower expression of the iron export molecule FP-1 as compared with rats on a normal diet.

Copper Deficiency Down-Regulates FP-1 Protein Expression in Rats

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

Figure 2. Hepatic mRNA expression of critical iron molecules and serum ceruloplasmin levels in the study population. Values are expressed as relative abundance normalized to ␤-actin cDNA levels and are presented as means (horizontal lines), 25th and 75th percentiles (boxes), and minimum/maximum ranges. Abbreviations: control, control subjects, n ⫽ 19; low, NAFLD with low liver copper concentration, n ⫽ 20; intermed., NAFLD with intermediate copper concentrations, n ⫽ 20; high, NAFLD with high copper concentrations, n ⫽ 23.

686

AIGNER ET AL

Sprague Dawley rats were subjected to identical diets with varying amounts of copper as described in the Materials and Methods section. We found that rats kept on a copper-deficient diet showed decreased FP-1 protein expression in the liver as compared with rats on a normal copper diet (Figure 4), thus indicating a direct linkage between dietary copper intake and hepatic iron export via modulation of FP-1 expression.

Discussion

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

In the present study, we aimed to investigate the potential contribution of copper status to iron perturbations in human NAFLD. We found that NAFLD patients with the lowest liver and serum copper concentrations were significantly more likely to have iron accumulation as reflected by more pronounced histologic iron deposition, increased hepatic iron concentrations, and higher biochemical surrogates of iron overload. Thus, our results indicate that low copper bioavailability is associated with characteristic perturbations of NAFLD iron homeostasis. Because the copper-dependent ferroxidase ceruloplasmin is required for the mobilization of iron from storage sites such as the liver, low levels of ceruloplasmin in NAFLD patients with low copper levels may be a causative molecular mechanism underlying iron accumulation in NAFLD. Caeruloplasmin knockout mice18 and patients with aceruloplasminemia17 present with iron deposition in hepatic parenchymal and sinusoidal cells, thus resembling the iron phenotype of human NAFLD.3 Recently, dietary copper and ceruloplasmin deficiency were found to induce increased hepatic hemosiderin deposition in a rat model.31 Moreover, hepatic iron accumulation can be abrogated by ceruloplasmin administration in pigs with dietary copper and ceruloplasmin deficiency.32 In line with these observations, our data suggest that decreased concentrations of the serum ferroxidase ceruloplasmin in such patients may be linked to impaired iron mobilization inducing iron deposition in the liver. Because NAFLD is characterized by increased levels of proinflammatory cytokines,33 which also induce ceruloplasmin gene expression,34 a lack of such enhanced ceruloplasmin expression in NAFLD patients was particularly surprising. Besides low levels of the serum ferroxidase ceruloplasmin, we detected significant differences in mRNA expression of key molecules of hepatic iron homeostasis. We found FP-1 mRNA expression to be lowest in NAFLD patients with low hepatic copper concentrations, which is in line with cell culture studies identifying copper as an inducer of FP-1 expression in J774 human macrophages20 and Caco-2 enterocytes.21 Our results support a role of copper availability for FP-1 protein expression because we detected significantly lower FP-1 protein levels in rats kept on a copper-deficient diet. Similarly, dietary copper deficiency in pigs induced iron accumulation resembling iron perturbations typically encountered in human

GASTROENTEROLOGY Vol. 135, No. 2

NAFLD.32 These intriguing similarities make inappropriate dietary copper availability the most likely cause of low copper stores in our patients, which is compatible with our finding of low FP-1 expression in copper-deficient rats. Other recent data indicate that membrane-bound GPIlinked ceruloplasmin ferroxidase activitiy is required for cell surface FP-1 stability in astrocytes and glioma cells, thus explaining iron accumulation in affected regions of the central nervous system in patients with aceruloplasminemia.22 These findings suggest that low FP-1 expression, low hepatic or serum copper concentrations, and decreased ceruloplasmin ferroxidase activity might be functionally related. Correspondingly, by Western blot analysis, we found FP-1 protein expression in NAFLD patients to be low despite increased hepatic iron stores. Increased hepatic iron stores are expected to posttranscriptionally up-regulate FP-1 protein expression via iron-responsive element/iron regulatory protein (IRE/ IRP)-interaction,35 which may account for the difference observed between FP-1 mRNA and protein expression. We therefore speculate that low serum ceruloplasmin activity may have a similar effect on hepatic FP-1 protein expression and stability. Thus, increased liver iron concentrations associated with inappropriately low hepatic FP-1 expression correspond to an impaired capability to facilitate iron export from liver cells. In contrast to low FP-1 expression, we found hepcidin mRNA levels to be increased in NAFLD patients with low liver copper. Because the master iron regulatory peptide hepcidin is secreted in response to iron accumulation as well as proinflammatory cytokines such as interleukin-1, interleukin ⫺6, or LPS, increased hepcidin levels in the NAFLD group with low copper most likely reflect an adequate metabolic response to hepatic iron accumulation. Because important molecular links between iron and copper metabolism are located both in absorptive enterocytes and in the liver, further studies are required to elucidate potential molecular interactions between these 2 metals with regard to iron perturbations in human NAFLD. Besides the inverse relationship between iron and copper status, we found a higher BMI, increased grades of steatosis, and serum triglyceride levels in NAFLD patients with low serum and liver copper concentrations. Our observation of significantly lower hepatic copper concentrations in NAFLD patients compared with control subjects suggests that a relevant proportion of NAFLD patients, in fact, can be considered copper deficient. This has been already suggested from the results of a previous investigation in which hepatic copper concentrations of NAFLD patients were lower than those of patients with various other liver diseases and of control subjects.36 Copper deficiency has previously been linked to cardiovascular disease37,38 and several features of the MS including elevated blood pressure, atherogenic dys-

lipidemia, and especially high serum triglyceride levels.39,40 Herein, we describe for the first time that low copper bioavailability might be a typical feature of pathophysiologic relevance to the hepatic manifestation of the MS NAFLD. In addition, our findings suggest that relatively deficient copper stores associated with signs of iron overload deserve further investigation, with regard to clinically important histologic or cardiovascular end points. We furthermore detected higher insulin resistance (HOMA-IR), a higher prevalence of diabetes, and the MS in the group of patients with low serum or liver copper concentrations. Patients who received a diagnosis of MS had significantly lower levels of hepatic copper compared with patients who did not fulfil MS criteria. Thus, low body copper stores appear to be associated with more pronounced clinical and biochemical features of the MS and, particularly, the severity of insulin resistance. So far, adverse effects of copper deficiency on lipid metabolism have been studied in rodents.41– 43 These studies demonstrate conclusively that copper deficiency is capable of inducing hypertriglyceridemia, hypercholesterolemia, and modification of the low-density lipoprotein and very-low-density lipoprotein composition. A unique role of high dietary fructose in aggravating these metabolic effects of copper deficiency, especially hypertriglyceridemia, has also been reported.39 Importantly, an inverse relationship between liver copper concentrations and hepatic cholesterol and triglyceride biosynthesis has been reported.42 Additionally, high dietary iron intake has been shown to adversely influence lipidemia associated with copper deficiency.44 Moreover, mitochondrial dysfunction and structural distortion have been implicated in the pathogenesis of NAFLD.45 Notably, systemic copper deficiency leads to similar morphologic changes because of its key role in mitochondrial respiratory chain physiology, especially in cytochrome C oxidase activity.46 We therefore speculate that low copper levels may additionally aggravate impaired mitochondrial respiratory chain function in NAFLD, thereby leading to higher rates of steatosis or serum triglycerides compared with NAFLD with adequate copper stores. Our findings also suggest that the currently applied “normal range” of serum copper levels may include a significant proportion of patients with copper stores that, in fact, should be considered depleted because all our patients were found to have serum and liver copper concentrations well within the limits of normal according to current laboratory standards. Hence, the pathophysiology as well as the prevalence of inadequately low body copper stores in patients with the insulin resistance syndrome requires clarification. Because of the retrospective nature of our analysis, we are not able to determine the causative link of the associations between low copper, high iron, and adverse metabolic conditions in humans. However, the in vivo experiments with Sprague Dawley rats receiving a low-copper

COPPER AND IRON IN NONALCOHOLIC FATTY LIVER

687

diet, revealed a significant effect of dietary copper intake on FP-1 protein expression in the liver. Because the prevalence of NASH in our cohort is low, the pathophysiologic contribution of low copper stores to disease severity could not be assessed. Therefore, the potential pathophysiologic impact of copper status on the histologic progression of NAFLD should be evaluated in a prospective human study. In conclusion, we report iron accumulation in NAFLD patients with low serum and liver copper concentrations along with low serum ceruloplasmin levels, hepatic FP-1 mRNA, and protein expression. Therefore, it appears that low copper bioavailability results in decreased ceruloplasmin ferroxidase activity and therefore impaired FP-1mediated cellular iron export, thus contributing to iron deposition in human NAFLD. To our knowledge, this is the first report of low liver copper stores associated with increased insulin resistance and other features of the MS. The contribution of low hepatic copper stores to the pathophysiology of the insulin resistance syndrome and NAFLD in particular will have to be investigated in detail.

Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2008.04.007. References 1. Neuschwander-Tetri BA. Fatty liver and the metabolic syndrome. Curr Opin Gastroenterol 2007;23:193–198. 2. Mendler MH, Turlin B, Moirand R, et al. Insulin resistance-associated hepatic iron overload. Gastroenterology 1999;117:1155– 1163. 3. Turlin B, Mendler MH, Moirand R, et al. Histologic features of the liver in insulin resistance-associated iron overload. A study of 139 patients. Am J Clin Pathol 2001;116:263–270. 4. Hentze MW, Muckenthaler MU, Andrews NC. Balancing acts: molecular control of mammalian iron metabolism. Cell 2004; 117:285–197. 5. Donovan A, Brownlie A, Zhou Y, et al. Positional cloning of zebrafish ferroportin1 identifies a conserved vertebrate iron exporter. Nature 2000;403:776 –781. 6. Abboud S, Haile DJ. A novel mammalian iron-regulated protein involved in intracellular iron metabolism. J Biol Chem 2000;275: 19906 –19912. 7. Ganz T. Hepcidin, a key regulator of iron metabolism and mediator of anemia of inflammation. Blood 2003;102:783–788. 8. Nicolas G, Chauvet C, Viatte L, et al. The gene encoding the iron regulatory peptide hepcidin is regulated by anemia, hypoxia, and inflammation. J Clin Invest 2002;110:1037–1044. 9. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science 2004;306:2090 –2093. 10. Sharp P. The molecular basis of copper and iron interactions. Proc Nutr Soc 2004;63:563–569. 11. Knopfel M, Solioz M. Characterization of a cytochrome b(558) ferric/cupric reductase from rabbit duodenal brush border membranes. Biochem Biophys Res Commun 2002;291:220 –225.

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

August 2008

688

AIGNER ET AL

BASIC–LIVER, PANCREAS, AND BILIARY TRACT

12. Frazer DM, Vulpe CD, McKie AT, et al. Cloning and gastrointestinal expression of rat hephaestin: relationship to other iron transport proteins. Am J Physiol Gastrointest Liver Physiol 2001;281: G931–G939. 13. McKie AT, Barrow D, Latunde-Dada GO, et al. An iron-regulated ferric reductase associated with the absorption of dietary iron. Science 2001;291:1755–1759. 14. Anderson GJ, Murphy TL, Cowley L, et al. Mapping the gene for sex-linked anemia: an inherited defect of intestinal iron absorption in the mouse. Genomics 1998;48:34 –39. 15. Osaki S, Johnson DA. Mobilization of liver iron by ferroxidase (ceruloplasmin). J Biol Chem 1969;244:5757–5758. 16. Hellman NE, Gitlin JD. Ceruloplasmin metabolism and function. Annu Rev Nutr 2002;22:439 – 458. 17. Harris ZL, Klomp LW, Gitlin JD. Aceruloplasminemia: an inherited neurodegenerative disease with impairment of iron homeostasis. Am J Clin Nutr 1998;67:S972–S977. 18. Harris ZL, Durley AP, Man TK, et al. Targeted gene disruption reveals an essential role for ceruloplasmin in cellular iron efflux. Proc Natl Acad Sci U S A 1999;96:10812–10817. 19. Auclair S, Feillet-Coudray C, Coudray C, et al. Mild copper deficiency alters gene expression of proteins involved in iron metabolism. Blood Cells Mol Dis 2006;36:15–20. 20. Chung J, Haile DJ, Wessling-Resnick M. Copper-induced ferroportin-1 expression in J774 macrophages is associated with increased iron efflux. Proc Natl Acad Sci U S A 2004;101:2700 – 2705. 21. Tennant J, Stansfield M, Yamaji S, et al. Effects of copper on the expression of metal transporters in human intestinal Caco-2 cells. FEBS Lett 2002;527:239 –244. 22. De Domenico I, Ward DM, di Patti MC, et al. Ferroxidase activity is required for the stability of cell surface ferroportin in cells expressing GPI-ceruloplasmin. EMBO J 2007;26:2823–2831. 23. Gouya L, Muzeau F, Robreau AM, et al. Genetic study of variation in normal mouse iron homeostasis reveals ceruloplasmin as an HFE-hemochromatosis modifier gene. Gastroenterology 2007; 132:679 – 686. 24. Galdston M, Levytska V, Schwartz MS, et al. Ceruloplasmin. Increased serum concentration and impaired antioxidant activity in cigarette smokers, and ability to prevent suppression of elastase inhibitory capacity of ␣ 1-proteinase inhibitor. Am Rev Respir Dis 1984;129:258 –263. 25. Datz C, Lalloz MR, Vogel W, et al. Predominance of the HLA-H Cys282Tyr mutation in Austrian patients with genetic haemochromatosis. J Hepatol 1997;27:773–779. 26. Seshadri V, Fox PL, Mukhopadhyay CK. Dual role of insulin in transcriptional regulation of the acute phase reactant ceruloplasmin. J Biol Chem 2002;277:27903–27911. 27. World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications. Technical Report 99.2. Geneva, Switzerland: WHO, 1999. 28. Brunt EM, Janney CG, Di Bisceglie AM, et al. Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94:2467–2474. 29. Searle JLB, Crawford DHG, Powell LW. Iron storage disease. In: MacSween RNM, Burt ADPB, eds. Pathology of the liver. 4th ed. Churchill Livingstone, 2002. 30. Theurl I, Ludwiczek S, Eller P, et al. Pathways for the regulation of body iron homeostasis in response to experimental iron overload. J Hepatol 2005;43:711–719.

GASTROENTEROLOGY Vol. 135, No. 2

31. Welch KD, Hall JO, Davis TZ, et al. The effect of copper deficiency on the formation of hemosiderin in Sprague-Dawley rats. Biometals 2007;20:829 – 839. 32. Lee GR, Nacht S, Lukens JN, et al. Iron metabolism in copperdeficient swine. J Clin Invest 1968;47:2058 –2069. 33. Lalor PF, Faint J, Aarbodem Y, et al. The role of cytokines and chemokines in the development of steatohepatitis. Semin Liver Dis 2007;27:173–193. 34. Gitlin JD. Transcriptional regulation of ceruloplasmin gene expression during inflammation. J Biol Chem 1988;263:6281– 6287. 35. McKie AT, Marciani P, Rolfs A, et al. A novel duodenal ironregulated transporter, IREG1, implicated in the basolateral transfer of iron to the circulation. Mol Cell 2000;5:299 –309. 36. Ferenci P, Steindl-Munda P, Vogel W, et al. Diagnostic value of quantitative hepatic copper determination in patients with Wilson’s disease. Clin Gastroenterol Hepatol 2005;3:811– 818. 37. Klevay LM. Cardiovascular disease from copper deficiency—a history. J Nutr 2000;130:S489 –S492. 38. Klevay LM. Dietary copper and risk of coronary heart disease. Am J Clin Nutr 2000;71:1213–1214. 39. Fields M, Lewis CG. Dietary fructose but not starch is responsible for hyperlipidemia associated with copper deficiency in rats: effect of high-fat diet. J Am Coll Nutr 1999;18:83– 87. 40. Relling DP, Esberg LB, Johnson WT, et al. Dietary interaction of high fat and marginal copper deficiency on cardiac contractile function. Obesity (Silver Spring) 2007;15:1242–1257. 41. al-Othman AA, Rosenstein F, Lei KY. Copper deficiency alters plasma pool size, percent composition and concentration of lipoprotein components in rats. J Nutr 1992;122:1199 –1204. 42. al-Othman AA, Rosenstein F, Lei KY. Copper deficiency increases in vivo hepatic synthesis of fatty acids, triacylglycerols, and phospholipids in rats. Proc Soc Exp Biol Med 1993;204:97–103. 43. Hing SA, Lei KY. Copper deficiency and hyperlipoproteinemia induced by a tetramine cupruretic agent in rabbits. Biol Trace Elem Res 1991;28:195–211. 44. Johnson MA, Murphy CL. Adverse effects of high dietary iron and ascorbic acid on copper status in copper-deficient and copperadequate rats. Am J Clin Nutr 1988;47:96 –101. 45. Pessayre D, Mansouri A, Fromenty B. Nonalcoholic steatosis and steatohepatitis. V. Mitochondrial dysfunction in steatohepatitis. Am J Physiol Gastrointest Liver Physiol 2002;282:G193–G199. 46. Nose Y, Kim BE, Thiele DJ. Ctr1 drives intestinal copper absorption and is essential for growth, iron metabolism, and neonatal cardiac function. Cell Metab 2006;4:235–244. Received September 25, 2007. Accepted April 10, 2008. Address requests for reprints to: Christian Datz, Department of Internal Medicine, General Hospital Oberndorf, Paracelsusstrasse 37, A-5110 Oberndorf, Austria. e-mail: [email protected] or Guenter Weiss, Department of General Internal Medicine, Clinical Immunology and Infectious Diseases, Medical University of Innsbruck, Anichstr. 35, A-6020 Innsbruck, Austria. e-mail: guenter.weiss@i-med. ac.at. Financial Disclosure: F.M. was supported by grant 3100A0-109703 from the Swiss National Foundation and a grant from the International Copper Association. Financial Support by the Austrian Research FundsFWF (P-19664 to G.W.) and SPAR, Austria (to C.D.) is gratefully acknowledged. Conflicts of interest: None of the authors have potential conflicts of interest to disclose with regard to this study. G.W. and C.D. contributed equally to this work.

August 2008

Supplementary Materials RNA Extraction From Liver Biopsy Specimens and Quantitative ReverseTranscription Polymerase Chain Reaction When available, a small tissue portion of the liver biopsy specimen was separated and preserved for RNA and protein extraction in RNAlater (Ambion’s RNAlater solution, Applera, Brunn am Gebirge, Austria). Total RNA was extracted from liver tissue by a guanidine misothiocynate-phenol-chloroform-based procedure as described.1 Reverse transcription was performed with 1 ␮g of total RNA, random hexamer primers (5 ␮mol/L), and dNTPs (62.5 ␮mol/L; Roche, Mannheim, Germany) and 200 U Molony murine leukemia virus reverse transcriptase (GIBCO, Gaithersburg, MD). TaqMan real-time polymerase chain reaction (PCR) primers and probes were designed intron spanning using the Primer Express Software from Applied Biosystems (Vienna, Austria) and synthesized by Microsynth (Balgach, Switzerland). TaqMan probes were labelled with the reporter dye 6-carboxyflurescein (FAM) at the 5=-end and with 6-carboxytetramethyl-rhodamine (TAMRA) at the 3=-end. For quantification of messenger RNA (mRNA) expression of genes of interest, the PCR reaction was carried on the MX4000 Multiplex Quantitative PCR System (Stratagene, Amsterdam, The Netherlands) as described.2 Specific complementary DNA levels were normalized to the amount of ␤-actin in liver biopsies. Commercially available probes for ␤-actin were obtained from Applied Biosystems and used according to the manufacturer’s instructions. The following sense and antisense primers and TaqMan probes were used (primer sense; primer antisense; probe): Transferrin receptor 1: 5=-TCCCAGCAGTTTCTTTCTGTTTT-3=,

COPPER AND IRON IN NONALCOHOLIC FATTY LIVER

688.e1

5=-CTCAATCAGTTCCTTATAGGTGTCCA-3=, 5=-CGAGGACACAGATTATCCTTATTTGGGTACCACC-3=; Ferroportin: 5=-TGACCAGGGCGGGAGA-3=, 5=-GAGGTCAGGTAGTCGGCCAA-3=, 5=-CACAACCGCCAGAGAGGATGCTGTG-3=; Hepcidin: 5=-TTTCCCACAACAGACGGGAC-3=, 5=-AGCTGGCCCTGGCTCC-3=, 5=-CAGAGCTGCAACCCCAGGACAGAGC-3=; Hemojuvelin: 5=-CCCCCATGGCGTTGG-3=, 5=-GCATGTTCTTAAATATGATGGTGAGC-3=, 5=-CAACGCTACCGCCACCCGGA-3=.

Protein Extraction From Liver Biopsy Specimens and Western blot Analysis Western blots were carried out as described previously.3 Briefly, protein extracts from duodenal and liver tissue were prepared using a radioimmunoprecipitation assay buffer (150 mmol/L NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mmol/L Tris HCl, pH 8.0, 0.2 mmol/L phenylmethylsulfonyl fluoride, 1 ␮g/mL pepstatin, 0.5 ␮g/mL leupeptin), and 10 ␮g total protein were used for immunoblotting. Blots were incubated with 0.5 ␮g/mL mouse anti-human ferroportin-1 (kindly provided by Andy McKie, King’s College, London) or ␤-actin antibody for 1 hour at room temperature and then stained with a horseradish peroxidase-conjugated antimouse IgG antibody (Dako, Copenhagen, Denmark). References 1. Theurl I, Ludwiczek S, Eller P, Seifert M, et al. Pathways for the regulation of body iron homeostasis in response to experimental iron overload. J Hepatol 2005;43:711–719. 2. Welch KD, Hall JO, Davis TZ, et al. The effect of copper deficiency on the formation of hemosiderin in Sprague-Dawley rats. Biometals 2007;20:829 – 839. 3. Theurl I, Mattle V, Seifert M, et al. Dysregulated monocyte iron homeostasis and erythropoietin formation in patients with anemia of chronic disease. Blood 2006;107:4142– 4148.