The kidney hepcidin/ferroportin axis controls iron reabsorption and determines the magnitude of kidney and systemic iron overload

The hepcidin/ferroportin axis controls systemic iron homeostasis by regulating iron acquisition from the duodenum and reticuloendothelial system, respective sites of iron absorption and recycling. Ferroportin is also abundant in the kidney, where it has been implicated in tubular iron reabsorption. However, it remains unknown whether endogenous hepcidin regulates ferroportin-mediated iron reabsorption under physiological conditions, and whether such regulation is important for kidney and/or systemic iron homeostasis. To address these questions, we generated a novel mouse model with an inducible kidney-tubule specific knock-in of fpnC326Y, which encodes a hepcidin-resistant ferroportin termed FPNC326Y. Under conditions of normal iron availability, female mice harboring this allele had consistently decreased kidney iron but only transiently increased systemic iron indices. Under conditions of excess iron availability, male and female mice harboring this allele had milder kidney iron overload, but greater systemic iron overload relative to controls. Additionally, despite comparable systemic iron overload, kidney iron overload occurred in wild type mice fed an iron-loaded diet but not in hemochromatosis mice harboring a ubiquitous knock-in of fpnC326Y. Thus, our study demonstrates that endogenous hepcidin controls ferroportin-mediated tubular iron reabsorption under physiological conditions. It also shows that such control is important for both kidney and systemic iron homeostasis in the context of iron overload.

F erroportin (FPN) is the only known mammalian iron export protein. It mediates iron release into the circulation from duodenal enterocytes and splenic reticuloendothelial macrophages, the respective sites of iron absorption and recycling. 1,2 FPN-mediated iron release is antagonized by the hormone hepcidin, also known as hepcidin antimicrobial peptide (HAMP). Produced primarily in the liver, hepcidin binds to and induces internalization of FPN, thereby limiting iron release into the circulation and its availability to peripheral tissues. 3,4 Thus, the HAMP/FPN axis operates at the sites of absorption and recycling to control systemic iron homeostasis.
The kidney is the site of iron reabsorption. Both nontransferrin-bound and transferrin-bound iron can cross into the glomerular filtrate. 5 The vast majority of this iron is taken up back into the tubular epithelia. Several transporters have been implicated in this reuptake, including multiligand megalin-cubilin receptor complex, transferrin receptor 1, divalent metal transporter 1, zinc transporter ZIP8, and zinc transporter ZIP14. [5][6][7][8][9][10] Once in the renal epithelia, iron is reabsorbed into the circulation. FPN is abundant in the kidney and has been implicated in iron reabsorption. [11][12][13][14] However, it remains unknown if renal FPN is also subject to regulation by endogenous HAMP under normal physiological conditions, and if so, whether such regulation is important for systemic and/or renal iron homeostasis.
To address these questions, we generated a novel mouse model with an inducible renal tubule-specific knock-in of fpnC326Y, which encodes a HAMP-resistant FPNC326Y protein. In addition, to confirm the previously reported role of FPN in iron reabsorption, we also generated a mouse model with an inducible renal tubule-specific deletion of the fpn gene. Our results demonstrate that endogenous HAMP directly regulates FPN-mediated iron absorption and that this regulation is important for both renal and systemic iron homeostasis, particularly in the setting of excess iron availability.
This study is the first to utilize mice harboring renalspecific loss of HAMP responsiveness to formally determine the importance of the renal HAMP/FPN axis. It provides new insights into the role of iron reabsorption in determining the degree of renal and extrarenal iron overload in the setting of hemochromatosis.

METHODS Mice
All animal procedures were compliant with the UK Home Office Animals (Scientific Procedures) Act 1986 and approved by the University of Oxford Medical Sciences Division Ethical Review Committee.
The conditional fpn fl and fpnC326Y fl alleles were generated as described previously. 15,16 Mice harboring the Pax8.CreERT2þ transgene were a gift from Dr. Athena Matakidou, Cancer Research UK Cambridge Institute, University of Cambridge. These mice were generated as described previously. 17 All mice were on a C57BL/6 background.

Diets
Unless otherwise stated, animals were provided with a standard rodent chow diet containing 200 parts per million (ppm) iron. In iron manipulation experiments, mice were given an iron-loaded diet (5000-ppm iron; Teklad TD.140464) or a matched control diet (200ppm iron; Teklad TD.08713) from weaning for 3 months.

Iron quantitation and iron indexes
Serum iron and ferritin levels were determined using the ABX-Pentra system (Horiba Medical). Hemoglobin values were determined by HemoCue Hb 201 Hemoglobin Microcuvettes. Serum erythroferrone levels were measured by enzyme-linked immunosorbent assay (Intrinsic Lifesciences). Determination of total elemental iron in tissues was performed by inductively coupled plasma mass spectrometry, as described previously. 15,16,18 Calibration was achieved using the process of standard additions, where spikes of 0 ng/g, 0.5 ng/g, 1 ng/g, 10 ng/g, 20 ng/g, and 100 ng/g iron were added to replicates of a selected sample. An external iron standard (High Purity Standards ICP-MS-68-A solution) was diluted and measured to confirm the validity of the calibration. Rhodium was also spiked onto each blank, standard, and sample as an internal standard at a concentration of 1 ng/g. For urinary iron quantitation, urine was collected from mice over a period of 24 hours, and subject to inductively coupled plasma mass spectrometry. Creatinine levels in the same samples were measured using a colorimetric creatinine assay kit (Abcam; ab65340), and iron values were normalized to creatinine concentration.

Western blotting
Tissues were snap frozen in liquid nitrogen, crushed, and then lysed using RIPA Lysis Buffer System (Santa Cruz; sc-24948), according to the manufacturer's instructions. Tissue lysates were cleared by centrifugation at 15 000g for 10 minutes at 4 C. Protein concentration in the lysates was measured by BCA Protein Assay (Pierce; 23225) and normalized to the same concentration for each batch. Lysates were then diluted in nonreducing Laemmli sodium dodecyl sulfate sample buffer and heated at 95 C for 5 minutes. Protein (30-50 mg) was loaded onto Mini-PROTEAN TGX Gels (Biorad; 4561096). After electrophoresis, protein was transferred onto polyvinylidene difluoride membrane using the BioRad Translotter system, and membranes were blocked for an hour in blocking buffer containing 5% bovine serum albumin. Membranes were then stained overnight at 4 C with rabbit polyclonal anti-mouse FPN antibody (NBP1-21502; Novus Biologicals) at 1:1000 or horseradish peroxidase-conjugated anti-b-actin antibody (Proteintech; HRP-60008) at 1:5000. Blots were developed using the ECL prime detection kit (RPN2232; VWR International). Signal intensities were quantified by ImageJ, and the ratio between the FPN and the b-actin signals was calculated to produce normalized intensities.

Diaminobenzidine (DAB)-enhanced Perls stain
Formalin-fixed, paraffin-embedded tissue sections were deparaffinized using xylene, and then rehydrated in ethanol. Slides were then stained for 1 hour with 1% potassium ferricyanide in 0.1 mol/L HCl buffer. Endogenous peroxidase activity was quenched, and then slides were stained with DAB chromogen substrate and counterstained with hematoxylin. They were visualized using a standard bright-field microscope.

RESULTS
The hepcidin/FPN axis in renal tubules controls iron reabsorption and is important for renal iron homeostasis in female mice We first sought to establish the exact site of FPN expression in the kidney. To that end, we costained mouse kidneys for FPN and segment-specific markers aquaporin 1 (a marker of proximal convoluted tubules and thin descending limb of Henley), aquaporin 2 (a marker of collecting ducts and connecting tubules), and calbindin (a marker of distal convoluted tubules and cortical connecting and collecting ducts). We found that FPN is strongly expressed in the cortex, colocalizing with aquaporin 1 to the proximal convoluted tubules. There was some FPN in the inner medulla, colocalizing with aquaporin 2 to the medullary collecting ducts. There was no colocalization of FPN with calbindin ( Figure 1a; larger panel shown in Supplementary Figure S2). These results confirm that FPN is most abundant in the cortical region within proximal convoluted tubules.
To determine whether renal FPN is regulated by HAMP, we used mice harboring a Pax8.CreER T2þ knock-in transgene, which drives tamoxifen-inducible expression of the Cre recombinase under control of the paired box gene 8 pax8 promoter in proximal and distal tubules and in collecting ducts. 17 We crossed Pax8.CreER T2þ mice with those harboring a conditional knock-in floxed allele FpnC326Y, which encodes a hepcidin-resistant FPN. In addition, and to confirm the previously reported role of renal FPN in iron reabsorption, we crossed Pax8.CreER T2þ mice with mice harboring the Fpn fl/fl allele. One week following tamoxifen treatment to induce the Pax8.CreER T2 transgene, renal FPN levels were increased in female and male FpnC326Y fl/ fl ,Pax8.CreER T2þ mice relative to FpnC326Y fl/fl controls (Figure 1b), demonstrating that renal FPN is subject to regulation by endogenous HAMP under normal physiological conditions, and in addition confirming the efficiency of the Pax8.CreER T2þ transgene. Conversely, renal FPN levels were reduced in female and male Fpn fl/fl ,Pax8.CreER T2þ mice relative to Fpn fl/fl controls, further confirming the efficiency of the Pax8.CreER T2þ transgene ( Figure 1c). The specificity of this Pax8.CreER T2þ transgene was also confirmed by the presence of the deletion allele (DFpn) in the kidney but not in the liver or spleen of Fpn fl/fl ,Pax8.CreER T2þ mice (Supplementary Figure S2A).
Next, we set out to determine whether the renal HAMP/ FPN axis controls iron reabsorption. To that end, we measured renal and serum iron levels at 1 week, 1 month, 3 months, and 6 months after tamoxifen treatment. We found that renal iron content was lower in FpnC326Y fl/fl ,Pax8.CreER T2þ females than in FpnC326Y fl/fl control females at all time points (Figure 1d). Renal iron content in males was not different according to genotype (Figure 1d). DAB-enhanced Perls iron stain also confirmed the reduction in iron levels within proximal tubules of FpnC326Y fl/fl ,Pax8.CreER T2þ females ( Figure 1e; larger panel shown in Supplementary Figure S2). Conversely, we found that renal iron content was higher in Fpn fl/fl ,Pax8.CreER T2þ females than in Fpn fl/fl control females from 1 month onwards (Figure 1f). Renal iron content in males was not different according to genotype (Figure 1f). DAB-enhanced Perls iron stain also confirmed iron accumulation within proximal tubules of Fpn fl/fl ,Pax8.CreER T2þ females ( Figure 1g; larger panel shown in Supplementary Figure S2).
Serum iron levels were increased transiently in both male and female FpnC326Y fl/fl ,Pax8.CreER T2þ mice relative to FpnC326Y fl/fl controls at the 1-month time point (Figure 1h). Conversely, serum iron levels were decreased in Fpn fl/ fl ,Pax8.CreER T2þ females relative to Fpn fl/fl control females at the 1-month time point, whereas they remained comparable in males of different genotypes at all time points (Figure 1i). Restoration of serum iron levels to those seen in control animals at 3 months of age could not be attributed to restoration of normal renal FPN levels. Indeed, the Pax8.CreER T2þ -induced changes in renal FPN levels were still maintained at 3 months after tamoxifen treatment (Supplementary Figure S2B and C). Decreased renal iron content and increased serum iron levels in FpnC326Y fl/ fl ,Pax8.CreER T2þ females demonstrate that FPN-dependent iron reabsorption in proximal tubules is subject to regulation by HAMP under normal physiological conditions. Increased renal iron content and decreased serum iron levels in Fpn fl/fl ,Pax8.CreER T2þ females confirm previous findings that renal FPN contributes to iron reabsorption. In addition, under normal physiological conditions, the control of iron reabsorption by the renal HAMP/FPN axis appears to be more important in females than in males, at least in the C57BL/6 strain.
The renal HAMP/FPN axis contributes to, but is not essential for, normal systemic iron homeostasis under conditions of normal iron availability Next, we set out to determine the contribution of the renal HAMP/FPN axis to systemic iron homeostasis. We found that  Figure 3B). Together, these data demonstrate that, under normal physiological conditions, the renal HAMP/FPN axis contributes to systemic iron levels but in  (Figure 3j). Hemoglobin levels were not affected by either diet or genotype, and consistent with this, serum erythroferrone levels also remained unchanged (Supplementary Figure 4A and B). These findings demonstrate that, under conditions of excess iron availability, control of iron reabsorption by the renal HAMP/FPN axis decreases systemic iron overload while increasing renal iron overload.
The renal HAMP/FPN axis determines the pattern of tissue iron overload in hemochromatosis Next, we set out to explore the role of the renal HAMP/FPN axis in the context of hereditary hemochromatosis, a genetic condition of iron overload caused by defects in hepcidin production or hepcidin responsiveness. 19 To that effect, we used mice generated in-house harboring a heterozygous ubiquitous knock-in of the fpnC326Y allele (Fpn wt/C326Y ). We had previously demonstrated that these mice develop the iron-overload phenotype characteristic of hereditary hemochromatosis. 15,18 We compared the pattern of tissue iron overload in these mice with that seen in wild-type mice fed an iron-loaded diet from weaning for 3 months. Both Fpn wt/ C326Y mice and wild-type mice fed an iron-loaded diet had increased iron content in the liver, heart, and lung compared with their respective controls (Figure 4a and b). Renal iron content in Fpn wt/C326Y mice was normal at 3 months of age ( Figure 4a) and only increased by 32% relative to controls at 6 months of age (Supplementary Figure 5A). In contrast, wildtype animals provided an iron-loaded diet had a 260% increase in renal iron content compared with those on a normal diet (Figure 4b). Relative differences in renal and liver iron overload between the 2 models were also apparent in DABenhanced Perls iron stain (Figure 4c and d; larger panels shown in Supplementary Figure S5). The magnitude of extrarenal iron overload was comparable between the 2 models, whereas the magnitude of renal overload was considerably higher in wild-type animals provided an ironloaded diet than in Fpn wt/C326Y mice (Figure 4e). FPN was increased in the cortex in both Fpn wt/C326Y mice and wildtype mice provided an iron-loaded diet (Figure 4f and g; larger panels shown in Supplementary Figure S5). Closer examination of the site of FPN expression in proximal tubules revealed that it appeared to localize intracellularly as well as to the basolateral membrane in Fpn wt/C326Y mice. In contrast, FPN appeared to localize primarily to the cytoplasm, apical membrane, and surprisingly the nucleus in mice provided an iron-loaded diet (Figure 4f and g, lower panels). The pattern of FPN localization arising from the provision of iron-loaded diet was further confirmed using Fpn fl/fl ,Pax8.CreER T2þ Fpn fl/fl ,Pax8.CreER T2þ animals as negative controls (Supplementary Figure S5B). These observations, together with the previous finding that FpnC326Y fl/fl ,Pax8.CreER T2þ animals have lower renal iron loading than controls following provision of an iron-loaded diet, demonstrate that loss of HAMP action on renal FPN protects the kidney from iron loading in the setting of hereditary hemochromatosis.

DISCUSSION
The most important finding of the present study is that endogenous HAMP controls FPN-mediated iron reabsorption. Previous studies had reported the regulation of FPN by exogenous HAMP in cultured renal cells, and an inverse relationship between HAMP and FPN levels in the kidney following unilateral ureter occlusion. 12,13 However, this is the first demonstration that such regulation operates in vivo under normal physiological conditions, and in a manner that impinges on renal and systemic iron homeostasis. A particular strength of this study is the use of novel mice, generated in-house, to harbor a renal-specific loss of HAMP responsiveness. This approach allows the study of the renal HAMP/ FPN axis without the confounding effects of altered systemic  iron homeostasis, otherwise seen in ubiquitous animal models.
The second important finding of the present study is that the renal HAMP/FPN determines the magnitude of both renal and systemic iron overload ( Figure 5). Indeed, loss of HAMP responsiveness in renal tubules increased the magnitude of liver, spleen, heart, and lung iron overload, while reducing the magnitude of renal iron overload following provision of an iron-loaded diet. Consistent with this, greater renal iron overload was seen following provision of an ironloaded diet to wild-type animals (in which the renal HAMP/FPN axis is intact) than in hemochromatosis mice (in which the renal HAMP/FPN axis is also disrupted) (graphical abstract). This finding may explain the clinical observation that the kidney is not commonly affected in patients with hereditary hemochromatosis. 19 Although both hemochromatosis and dietary iron overload increased FPN in renal tubules, they appeared to result in different patterns of localization to the apical, intracellular, or basolateral compartments. Similar observations were made in the rat duodenum, where iron gavage was found to alter the relative distribution of FPN between these compartments in duodenal enterocytes. 20 One interpretation of these results is that in the setting of dietary iron overload, increased FPN translation (mediated by iron regulatory proteins) would act to maintain high overall FPN production in the renal tubular cell, whereas increased serum HAMP would reduce the proportion of FPN that can localize to the basolateral membrane. It follows from this interpretation that the divergence between previous reports as to the localization of FPN in proximal tubules could reflect differences in renal tubular iron content and in serum hepcidin levels between animal models used in different studies. 5,[11][12][13][14]21 Alternatively, this divergence could reflect a degree of nonspecific reactivity reported previously in relation to commercial FPN antibodies, including the one used in the present study. 14 The functional significance of the apical localization of FPN in renal tubules remains unclear, and we could not detect any changes in urinary iron excretion in iron-loaded mice (Supplementary Figure 5C). Another interesting observation is that some FPN appeared localized to the nucleus in renal tubules of iron-loaded mice. Similar observations were made by others in iron-loaded macrophages and in the rat liver. The latter study found nuclear FPN to be involved in nuclear iron retention as part of the acute phase response. 22,23 Detailed studies of the subcellular localization of FPN in different iron states and in different pathologies are needed to further advance our understanding of the role of FPN in the kidney.
The regulation and function of renal HAMP are also not completely understood. We found that expression of the hamp gene in the kidney was increased in wild-type animals following provision of iron-loaded diet (Supplementary Figure 5E), but remained unaltered in hemochromatosis mice (Supplementary Figure 5D), and in those harboring renal tubule-specific loss of FPN or renal tubule-specific loss of HAMP responsiveness (Supplementary Figure S5F and G). These results do not support the notion that renal hamp gene expression is regulated locally by renal iron levels. In terms of the function of renal HAMP, a previous study in a model of unilateral ureter occlusion suggested that it is involved in controlling renal FPN. We found that renal HAMP (detected using an antibody against pro-HAMP peptide) colocalized with calbindin (a marker of distal convoluted tubules and cortical collecting ducts and connecting ducts) (Supplementary Figure S5H). On the other hand, FPN expression does not colocalize with calbindin, even in mice harboring renal-specific loss of HAMP responsiveness (Figure 1a and Supplementary Figure S5I). These results are not consistent with an autocrine role for renal HAMP in the regulation of renal FPN. Nonetheless, renal HAMP may be involved in paracrine regulation of renal FPN. In the future, it would be interesting to study the paracrine functions of renal HAMP, and to quantify the relative contributions of renal and hepatic HAMPs to the control of iron reabsorption.
Another interesting observation from the present study is that, under conditions of normal iron availability, the renal HAMP/FPN axis is not essential for maintenance of normal systemic iron homeostasis and is instead more important for regulating renal iron levels, at least in female mice of the C57BL/6 strain. Indeed, although the observed changes in renal iron levels resulting from loss of FPN or of HAMP responsiveness in the renal tubules were persistent, changes in systemic iron indexes were transient in nature. The transient nature of the systemic effects suggests the involvement compensatory mechanism(s). One possible compensatory mechanism is modulation of dietary iron absorption by hepatic HAMP. Indeed, we found that in mice harboring renalspecific loss of HAMP responsiveness, hepatic hamp gene expression was increased following the increase in serum iron levels at the 1-month time point and remained increased at later time points. Conversely, hepatic hamp gene expression was suppressed in mice harboring renal-specific loss of FPN following the decrease in serum iron levels at the 1-month time point and remained suppressed at later time points (Supplementary Figure S6A and B). In addition, increased hepatic hamp gene expression in the first setting was accompanied by decreased gut FPN levels, whereas decreased hepatic hamp gene expression in the second setting was accompanied by higher gut FPN (Supplementary Figure S6C and D). These results indicate that the action of hepatic HAMP on dietary iron absorption is an active compensatory mechanism involved in maintaining normal systemic iron homeostasis in the face of perturbed renal iron reabsorption.
Of note, under conditions of normal iron availability, the renal HAMP/FPN axis appears to be more important in female mice that in male mice, at least in the C57BL/6 strain. This observation could not be attributed to differences between males and females in the activity of the Pax8.CreER T2þ transgene. Sexual dimorphism in the levels and patterns of transporters along the nephron has been reported previously in this mouse strain. 24 Consistent with this, we found higher basal expression of FPN in the kidneys of females than in the kidneys of male littermates (Supplementary Figure S2D). A previous study using a different Cre recombinase transgene, driven by a constitutively active Nestin promoter to delete fpn in the entire nephron, also reported an increase in renal iron levels, and decrease in serum iron and liver iron stores. However, that study was conducted in a different mouse strain (129/SvEvTac) and did not analyze the phenotype according to sex or time course. 14 Together, the findings of the present study provide new understanding of the role of the renal HAMP/FPN axis in renal and systemic iron homeostasis. This new understanding has potentially important implications for the management of hemochromatosis and other iron disorders.      Figure 4c, d, f, and g, respectively. Values are shown as mean AE SEM. *P < 0.05. Figure S6. Liver hamp gene expression and gut FPN levels are altered in mice with renal tubule-specific knockout of fpn and in mice with renal tubule-specific knock-in of fpnC326Y. as mean AE SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.