Liver iron modulates hepcidin expression during chronically elevated erythropoiesis in mice


  • Víctor Díaz,

    1. Institute of Veterinary Physiology, Vetsuisse Faculty, and Zurich Center for Integrative Human Physiology (ZIHP), and University of Zurich, Switzerland
    2. Department of Health and Human Performance, Faculty of Sports Science, INEF, Technical University of Madrid, Spain
    Search for more papers by this author
    • These authors contributed equally to the work.

  • Elena Gammella,

    1. Department of Biomedical Sciences for Health, University of Milano, Italy
    Search for more papers by this author
    • These authors contributed equally to the work.

  • Stefania Recalcati,

    1. Department of Biomedical Sciences for Health, University of Milano, Italy
    Search for more papers by this author
  • Paolo Santambrogio,

    1. Department of Biomedical Science and Technology, San Raffaele Scientific Institute, Milano, Italy
    Search for more papers by this author
  • Arianne Monge Naldi,

    1. Institute of Veterinary Physiology, Vetsuisse Faculty, and Zurich Center for Integrative Human Physiology (ZIHP), and University of Zurich, Switzerland
    2. Clinic for Otolaryngology, Head and Neck Surgery, University Hospital Zurich, Switzerland
    Search for more papers by this author
  • Johannes Vogel,

    1. Institute of Veterinary Physiology, Vetsuisse Faculty, and Zurich Center for Integrative Human Physiology (ZIHP), and University of Zurich, Switzerland
    Search for more papers by this author
  • Max Gassmann,

    1. Institute of Veterinary Physiology, Vetsuisse Faculty, and Zurich Center for Integrative Human Physiology (ZIHP), and University of Zurich, Switzerland
    2. Universidad Peruana Cayetano Heredia, Lima, Perú
    Search for more papers by this author
    • Contribution of both senior authors was equivalent.

  • Gaetano Cairo

    Corresponding author
    1. Department of Biomedical Sciences for Health, University of Milano, Italy
    • Address reprint requests to: Gaetano Cairo, Department of Biomedical Sciences for Health, University of Milano, Via Mangiagalli 31, 20133 Milano, Italy. E-mail:

    Search for more papers by this author
    • Contribution of both senior authors was equivalent.

  • Potential conflict of interest: Nothing to report.

  • Supported by grants from Ministero Istruzione, Università e Ricerca (PRIN project) and World Anti Doping Agency (to G.C.), Technical University of Madrid, Program Marie Curie, COFUND, (contract UNITE 246565), Swisslife Jubiläumsstiftung (to V.D.), Zurich Center Integrative Human Physiology (ZIHP), Zurich Center Integrative Rodent Physiology (ZIRP) and Swiss National Science Foundation (to M.G.).


The liver-derived peptide hepcidin controls the balance between iron demand and iron supply. By inhibiting the iron export activity of ferroportin, hepcidin modulates iron absorption and delivery from the body's stores. The regulation of hepcidin, however, is not completely understood and includes a variety of different signals. We studied iron metabolism and hepcidin expression in mice constitutively overexpressing erythropoietin (Epo) (Tg6 mice), which leads to excessive erythropoiesis. We observed a very strong down-regulation of hepcidin in Tg6 mice that was accompanied by a strong increase in duodenal expression of ferroportin and divalent metal tranporter-1, as well as enhanced duodenal iron absorption. Despite these compensatory mechanisms, Tg6 mice displayed marked circulating iron deficiency and low levels of iron in liver, spleen, and muscle. To elucidate the primary signal affecting hepcidin expression during chronically elevated erythropoiesis, we increased iron availability by either providing iron (thus further increasing the hematocrit) or reducing erythropoiesis-dependent iron consumption by means of splenectomy. Both treatments increased liver iron and up-regulated hepcidin expression and the BMP6/SMAD pathway despite continuously high plasma Epo levels and sustained erythropoiesis. This suggests that hepcidin expression is not controlled by erythropoietic signals directly in this setting. Rather, these results indicate that iron consumption for erythropoiesis modulates liver iron content, and ultimately BMP6 and hepcidin. Analysis of the BMP6/SMAD pathway targets showed that inhibitor of DNA binding 1 (ID1) and SMAD7, but not transmembrane serine protease 6 (TMPRSS6), were up-regulated by increased iron availability and thus may be involved in setting the upper limit of hepcidin. Conclusion: We provide evidence that under conditions of excessive and effective erythropoiesis, liver iron regulates hepcidin expression through the BMP6/SMAD pathway. (Hepatology 2013; 58:2122–2132)


bone morphogenetic protein 6


divalent metal tranporter-1










growth differentiation factor 15


inhibitor of DNA binding 1






transferrin saturation


transmembrane serine protease 6


twisted gastrulation 1.

Iron is a key element for all organisms, as it participates in essential processes, including the formation of the oxygen carrier molecule heme, but also catalyzes the formation of toxic reactive oxygen species. Therefore, complex mechanisms maintain iron balance and prevent equally dangerous iron deficiency or overload. The greatest amount of iron is present in erythrocytes, and thus iron demand is mainly dictated by the erythropoietic activity, also known as the erythroid regulator.[1] In addition, iron metabolism is also controlled by body iron deposits, referred to as store regulators,[1] which are mainly located in the liver. Accordingly, both regulators require a fine-tuned coordination that ultimately allows the maintenance of iron homeostasis.

The hepatic peptide hepcidin, which induces internalization and degradation of the iron exporter ferroportin (Fpn), and thus negatively modulates iron absorption and recycling,[2] is the main regulator of systemic iron metabolism.[3] Hepcidin expression is stimulated when the body's iron stores increase, and also by inflammatory stimuli and endoplasmic reticulum stress.[3] The BMP6/SMAD pathway, which plays a key role in iron-induced stimulation of hepcidin,[4-6] also induces negative regulators of hepcidin, such as SMAD7, inhibitor of DNA binding 1 (ID1), or transmembrane serine protease 6 (TMPRSS6),[5, 7] so that a negative feedback loop controlling hepcidin expression may be established.

Conversely, studies in animal models[3] and human subjects[8-11] have shown that hepcidin is inhibited by anemia/hypoxia and also by increased erythropoietic activity, a response that is aimed at increasing iron availability to meet the need for iron for hemoglobin synthesis and erythroid proliferation. While the molecular and cellular pathways underlying iron- and inflammation-dependent hepcidin induction are reasonably well understood, those linking erythropoiesis and hepcidin suppression remain to be clarified. On the one hand, erythropoietin (Epo) itself may down-regulate hepcidin expression, as in healthy humans elevation of Epo levels caused a reduction of hepcidin before any change in hematological or iron parameters occurred.[8-11] Moreover, it has been demonstrated that Epo inhibits hepcidin transcription in hepatoma cell lines.[12] On the other hand, blockade of erythropoietic activity prevented the Epo effect, indicating that Epo suppresses hepcidin through erythropoietic stimulation.[13, 14] However, some of the candidates proposed as the “erythroid regulator” have been excluded,[15] and the role of other mediators, such as GDF15[16] and TWSG1[17] is still unclear.

In addition, the interaction between the erythroid and iron storage regulator and their relative importance remains to be fully revealed, although studies in animals showing that hepcidin expression is dependent on the degree of erythropoiesis,[18] and the fact that β-thalassemia patients display elevated erythropoiesis and low hepcidin levels despite massive liver iron overload,[19] suggest a dominance of the erythropoietic drive over the storage regulator.[13, 20, 21]

We have taken advantage of our transgenic mouse line constitutively overexpressing human Epo (termed Tg6)[22] to evaluate the interplay between the erythroid and the store regulator on the control of hepcidin expression in the liver, and to investigate the effect of chronically elevated, effective erythropoiesis on systemic iron metabolism. We show that Tg6 mice have scarce liver iron stores, very low hepcidin expression, and high duodenal iron import. The modulation of iron availability reversed this phenotype in the face of constantly high plasma Epo levels and sustained erythropoiesis. Our data suggest that under conditions of elevated and effective erythropoiesis, liver iron levels modulate hepcidin expression.

Materials and Methods

Animals and Treatments

Six- to eight-week-old male wild-type (Wt, n = 6) and transgenic (Tg6, n = 6,) mice constitutively overexpressing human Epo were used. Parenteral iron overload in Tg6 (Tg6_DXT, n = 6) and Wt (Wt_DXT, n = 7) mice was induced with a single intraperitoneal injection of iron-dextran (Sigma, St. Louis, MO) containing 5 mg of iron. A group of Tg6 mice was splenectomized (Tg6_SPL, n = 5) to reduce erythropoiesis-dependent iron consumption, as described in the Supporting Information.

Statistical Analysis

Differences in hematocrit were analyzed using a Student t test for paired samples. The Kruskal-Wallis one-factor analysis of variance (ANOVA) was used to detect any effect of treatments. When ANOVA was significant (significance level was set to P < 0.05), differences between groups were revealed by pairwise comparisons using the Mann-Whitney U test.

Detailed Materials and Methods can be found in the Supporting Information.


Hematologic Parameters

As expected, in untreated Tg6 mice overexpressing Epo the hematocrit was greatly elevated (77.7 ± 2.5%) (Fig. 1A) compared to Wt controls. Iron supplementation led to an increase in Tg6_DXT mice, while Wt_DXT animals showed a slight but significant reduction of hematocrit compared to controls. Two weeks after surgery, splenectomized mice presented a 15.3% decrease in hematocrit. A time course showing the detailed progressive reduction of the hematocrit after splenectomy is presented in Supporting Fig. S1A. We measured Epo concentration in a separate group of animals and found that the reduction of erythropoiesis by splenectomy was accompanied by increased plasma Epo levels (Fig. S1B). Additionally, we used flow cytometry to identify reticulocytes, and CD71 and CD44-positive cells as markers of erythropoiesis. Tg6 mice had higher reticulocyte count as well as CD71-positive mature cells than Wt. Iron loading and splenectomy led to a significant decrease in reticulocyte count and CD71 compared to untreated Tg6. CD44 did not change in any of the groups (Fig. S2A,B).

Figure 1.

Analysis of hematocrit, hepcidin, and BMP6 expression. Data were obtained from wild-type controls (Wt, n = 6), Wt mice treated with 5 mg of iron dextran (Wt_DXT, n = 7), mice overexpressing Epo (Tg6, n = 6), Tg6 mice treated with 5 mg of iron dextran (Tg6_DXT, n = 6), and splenectomized Tg6 mice (Tg6_SPL, n = 5). Hematocrit (A) was measured before and 2 weeks after treatment. Data are presented as mean ± SEM. Hepcidin (HAMP) (B) and BMP6 (C) expression were measured by reverse-transcription polymerase chain reaction (RT-PCR). Samples were analyzed in triplicate and normalized to the housekeeping gene 18S RNA. Dots and black line represent single animals and the mean, respectively. For better visualization, Y-axis scale has been set up using a logarithmic scale (B,C plots). *P < 0.05 with Wt; **P < 0.01 with Wt; #P < 0.05 with Tg6; ##P < 0.01 with Tg6; $P < 0.05 with Pretreatment.

Hepcidin and BMP6 Expression

Hepcidin and BMP6 messenger RNA (mRNA) levels in the liver (Fig. 1B,C) were markedly suppressed in untreated Tg6 mice (130- and 6-fold lower, respectively, than those of their Wt controls). In Wt_DXT mice, iron treatment led to a 7-fold and 4-fold increase in hepcidin and BMP6 mRNA expression, respectively. Interestingly, hepcidin expression was up-regulated 90-fold in Tg6_DXT mice as compared to untreated Tg6, and was accompanied by a 12-fold increase in BMP6 expression. Of note, iron supply to Tg6 animals led to an elevation of hepcidin and BMP6 levels, which returned close to that of the Wt controls, indicating that despite high erythropoietic drive the organism up-regulates hepcidin and BMP6 in response to iron. By reducing iron utilization for erythropoiesis, splenectomy also increased hepcidin expression significantly, but less than iron loading (7-fold), and hepcidin mRNA levels in Tg6_SPL animals remained lower than in Wt. This trend was reflected by parallel changes in BMP6 expression.

Circulating Iron and Tissue Ferritin Content

Since BMP6 expression is a positive regulator of hepcidin in response to iron,[4-6] we measured circulating iron availability and tissue ferritin (Ft) content as an indicator of cellular iron stores (Fig. 2). Compared to the untreated Wt group, in Wt_DXT mice there was an increase of almost 2-fold and 1.5-fold in serum iron and %Tf-sat, respectively. In untreated Tg6, elevated erythropoiesis was accompanied by low levels of serum iron and %Tf-sat. Since iron administration enhanced iron utilization in Tg6_DXT mice (as demonstrated by the increased hematocrit), %Tf-sat remained lower than in Wt and similar to the untreated Tg6 group. On the other hand, splenectomy increased both parameters. In contrast, in Tg6_SPL mice return of iron availability to control levels was observed, which was in line with the reduction of iron demand after splenectomy (Fig. 2A,B).

Figure 2.

Circulating iron and tissue ferritin content. Serum iron (A), transferrin saturation (B), and ferritin content in the liver (C), duodenum (D), spleen (E), and skeletal muscle (quadriceps) (F) were assessed in the groups of mice described in the legend to Fig. 1. H (gray bars) and L (black bars) ferritin subunits were measured by enzyme-linked immunosorbent assay (ELISA) and the values normalized to the protein content. Data are presented as mean ± SEM; *P < 0.05 with Wt; **P < 0.01 with Wt; #P < 0.05 with Tg6; ##P < 0.01 with Tg6.

In the liver, Tg6 mice had decreased levels of both H- and L-Ft subunits, which are related to rapid iron sequestration and long-term iron storage, respectively,[23] this indicated a depletion of hepatic iron stores. Treatment with iron dextran increased H- and L-Ft in the liver of both Wt and Tg6 mice (Fig 2C). In Tg6_SPL mice, Ft content increased to levels comparable to Wt controls and was significantly higher than in untreated Tg6, suggesting that reduction of erythropoiesis and increased circulating iron availability was sufficient to restore hepatic iron stores (Fig. 2C). These results were confirmed by measurements of liver nonheme iron content (Fig. S3A), which also showed high correlations with L-Ft and H-Ft (Table S3).

Since enterocytes do not represent iron-storing cells, as indicated by the high H-Ft/L-Ft ratio in these cells, Tg6 mice presented normal values of L-Ft (Fig. 2D). However, iron treatment or splenectomy increased L-Ft in Wt_DXT, Tg6_DXT, and Tg6_SPL animals. In the spleen, H-Ft levels in Tg6 animals were higher than in Wt (Fig. 2E), in line with increased iron recycling due to extramedullary erythropoiesis and high macrophage activity in these mice.[24] When additional iron was provided (Tg6_DXT), H-Ft was reduced and L-Ft significantly elevated compared to untreated Wt. Spleen iron content was decreased in Tg6 and restored to control levels in Tg6_DXT (Fig. S3B). Finally, since changes in muscle Ft levels have been found in subjects exposed to hypoxia[25] or treated with Epo,[10] we analyzed Ft content in the skeletal muscle. Figure 2F shows that untreated Tg6 mice had lower Ft content in the muscle that was restored after either iron administration or splenectomy.

Using Spearman's correlation coefficient, all variables (%Tf-sat, liver L-Ft, liver H-Ft, and liver nonheme iron content) showed a positive and significant correlation with BMP6 mRNA expression (Table S3). However, linear regression analysis revealed that L-Ft was the only single predictor of BMP6 expression, reinforcing the hypothesis that liver iron content modulates BMP6.[26]

Iron Absorption, Uptake, and Mobilization

Subsequently, we investigated duodenal iron absorption in order to elucidate how Tg6 mice cope with their extreme iron demand. First, we observed an up-regulation of Fpn in the duodenum of untreated Tg6 mice compared to Wt. Iron administration or splenectomy reduced Fpn to levels comparable with Wt mice (Fig. 3B). In Tg6 mice, changes in Fpn were accompanied by increased protein and mRNA levels of divalent metal tranporter-1 (DMT1), which is up-regulated by iron deficiency to increase iron absorption[27] (Fig. S4). To assess whether changes in Fpn and DMT1 were functional, we measured 59Fe absorption in a different group of animals and we found that modifications in intestinal iron absorption mirrored the changes in Fpn and DMT1 expression; however, it should be noted that, despite normalization of Fpn levels in Tg6_DXT and Tg6_SPL mice (Fig. 3B), DMT1 and intestinal iron absorption remained elevated compared to Wt (Fig. S4).

Figure 3.

Ferroportin expression. Ferroportin levels were detected in the groups of mice described in the legend to Fig. 1 by western blotting in liver (A), duodenum (B), spleen (C), and skeletal muscle (D) extracts. Each panel shows a representative blot obtained with extracts of two animals for each group and the densitometric quantitation of the analysis of the extracts from all of the mice. The values were normalized to Wt controls and are presented as mean ± SEM. *P < 0.05 with Wt; **P < 0.01 with Wt; #P < 0.05 with Tg6; ##P < 0.01 with Tg6.

Fpn expression in the liver of Tg6 mice was not significantly altered, although both iron administration and splenectomy caused a significant reduction (Fig. 3A). Similar to the liver, Fpn was lower than in Wt in the spleen of untreated Tg6 mice (Fig. 3C). Since this latter finding was not in line with the reduced hepcidin expression, we also analyzed both IRE-containing and non-IRE-containing Fpn transcripts in the spleen. We did not find statistically significant variations in Tg6, although the expected induction upon iron challenge was present (Fig. S5). In skeletal muscle, none of the treatments affected Fpn levels (Fig. 3D).

In line with their depleted iron deposits, Tg6 mice presented increased transferrin receptor (TfR1) expression in liver, duodenum, and spleen (Fig. 4A-C). Moreover, iron administration and splenectomy (Fig. 2) were accompanied by a concomitant decrease in TfR1, except in the spleen, reflecting the high erythropoietic activity of the spleen under these conditions. In the skeletal muscle, TfR1 showed a trend to increase, albeit nonsignificantly, in Tg6 mice, and remained elevated after iron loading, but was significantly reduced after splenectomy (Fig. 4D).

Figure 4.

Transferrin receptor 1 expression. TfR1 was detected in the groups of mice described in the legend to Fig. 1 by western blotting, as described in the legend to Fig. 3 except for duodenum (B), in which RT-PCR was performed due to low TfR1 protein levels in this tissue. Duodenal TfR1 mRNA was normalized to the housekeeping gene 18S and data are presented as mean ± 95% confidence interval. *P < 0.05 with Wt; **P < 0.01 with Wt; #P < 0.05 with Tg6; ##P < 0.01 with Tg6.

BMP6/SMAD and ERK Pathways

The BMP6/P-SMAD1/5/8 pathway plays a central role in the modulation of hepcidin in response to iron,[4, 28] as shown by the iron overload due to low hepcidin levels found in BMP6-deficient mice.[6] Moreover, hepcidin can be regulated through transferrin receptor 2 and the activation of the MAPK/ERK pathway.[29, 30] Therefore, we investigated both pathways in order to elucidate the signals affecting hepcidin expression in our mouse models. Tg6 mice showed a decrease in the P-SMAD1/5/8/total SMAD1 ratio, but a normal ratio was restored after iron supplementation (Fig. 5A). Splenectomy-dependent reduction of iron consumption caused a nonsignificant trend to increase the ratio P-SMAD1/5/8/total SMAD1.

Figure 5.

Evaluation of P-SMAD1/5/8/SMAD1 and P-ERK1/2/ERK1/2 ratio. P-SMAD1/5/8 and SMAD1 (A), as well as P-ERK1/2 and ERK1/2 (B) were assessed in the groups of mice described in the legend to Fig. 1 by western blotting. Each panel shows a representative blot obtained with extracts of two animals for each group and the phosphorylated/nonphosphorylated ratios obtained by densitometric quantitation of the analysis of the extracts from all of the mice. The values were normalized to the corresponding loading control protein prior to the calculation of the ratio and are presented as mean ± SEM. **P < 0.01 with Wt; #P < 0.05 with Tg6.

On the other hand, excessive erythropoiesis did not alter the P-ERK/total ERK ratio, while iron challenge showed a nonsignificant trend to increase P-ERK (Fig. 5B). These results are in line with a previous study in which the participation of the MAPK/ERK pathway in the regulation of hepcidin in vivo was questioned.[26]

ID1, SMAD7, and TMPRSS6 Expression

The BMP6/SMAD pathway has also been shown to stimulate the expression of target transcripts, such as ID1, TMPRSS6, and SMAD7.[5, 7, 31] Thus, we measured ID1, TMPRSS6, and SMAD7 mRNA expression in livers to elucidate their role in modulating hepcidin levels during chronically elevated erythropoiesis. Figure 6 shows that ID1 expression mirrored the variations of BMP6 (Fig. 1C), as it was up-regulated 15-fold in Wt_DXT, 7-fold in Tg6_DXT, and 3-fold in Tg6_SPL; however, it was not significantly affected in Tg6 mice.

Figure 6.

ID1, SMAD7, and TMPRSS6 expression. ID1 (A), SMAD7 (B), and TMPRSS6 (C) expression were measured by RT-PCR as described in the legend to Fig. 1 in the groups of mice reported in the legend to Fig. 1. *P < 0.05 with Wt; **P < 0.01 with Wt; ∼P = 0.055 with Wt; #P < 0.05 with Tg6; ##P < 0.01 with Tg6.

SMAD7, which has been shown to effectively inhibit hepcidin in vitro,[32] showed a 5-fold increase in Wt_DXT and a trend to increase in Tg6_DXT mice. Elevated erythropoiesis in Tg6 caused a small but significant down-regulation of SMAD7, which is an observation that is in line with the low levels of liver iron. Finally, increased iron availability in Tg6_SPL elevated SMAD7 to values comparable with Wt animals but higher than those of Tg6 (Fig. 6B).

It has been suggested that TMPRSS6 is indirectly regulated by BMP6 through ID1 in response to iron, and participates in the fine tuning of hepcidin levels.[7] Accordingly, TMPRSS6 was up-regulated in the liver of Wt_DXT mice, while a nonsignificant trend to increase was observed in mice overexpressing Epo (Fig. 6C). Overall, these results suggest that erythropoiesis-induced repression of hepcidin does not require TMPRSS6.


Regulation of Systemic Iron Homeostasis During Chronically Elevated Erythropoiesis

Increased erythropoiesis is associated with impressive alterations in iron metabolism that meet the need of iron for hemoglobin synthesis and erythroid cell proliferation. To understand how elevated erythropoiesis affects systemic iron homeostasis, we investigated the regulation of hepcidin in Tg6 mice characterized by excessive erythropoiesis.[22] In line with the negative impact of erythropoiesis on hepcidin,[8-13, 21, 33] we observed a very strong down-regulation of hepatic hepcidin mRNA levels in Tg6 mice (Fig. 1B), accompanied by a remarkable increase of Fpn expression in the duodenum (Fig. 3B). Furthermore, we found in Tg6 enhanced duodenal iron absorption and increased expression of DMT1 (Fig. S4). Despite these compensatory mechanisms enhancing iron absorption, Tg6 mice displayed low serum iron, low %Tf-sat, and low ferritin content in liver and muscle, but not in the duodenum, probably due to the iron absorbing function of this organ (Fig. 2A,B). These results, indicating tissue iron deficiency in Tg6, were supported by the decreased iron content in liver and spleen (Fig. S3), elevated expression of TfR1 in liver, spleen, and duodenum (Fig. 4A-C), and by the lack of changes in Fpn expression in the liver and muscle of Tg6 mice compared to controls (Fig. 3A-D). Overall, our results suggest that chronically elevated erythropoiesis causes circulating iron deficiency and low iron levels in liver, spleen, and muscle. Therefore, we conclude that repression of hepcidin synthesis and increased duodenal absorption are the major mechanisms by which erythrocytotic mice cope with persistently high iron demand.

Fpn was differently regulated in the various tissues. In particular, in the spleen of Tg6 mice we unexpectedly observed a slight but significant reduction of Fpn, accompanied by decreased L-Ft but elevated H-Ft and a 5-fold increase in TfR1, despite the strong decrease in hepcidin expression. We thus suggest that Fpn expression in the spleen of Tg6 animals is regulated by a hepcidin-independent mechanism.[34] As we did not find any significant differences in Fpn mRNA levels (see Fig. S5), the underlying mechanism may reflect IRP-dependent translational repression of the IRE-containing Fpn mRNA triggered by low intracellular iron availability.[34] This is consistent with the low levels of the iron storage-related L-Ft subunits[23] (Fig. 2), the reduced iron content (see Fig. S3), and the extramedullary erythropoietic role of the spleen in Tg6.[24]

Our observations also suggest a different regulation of TfR1, depending on the tissue and the degree of iron accumulation. Tg6 animals presented an up-regulation of TfR1 in all of the tissues examined, which is in line with lower iron stores, as shown by the significant correlation (ρ = 0.721, P < 0.001) between liver nonheme iron content and hepatic TfR1 expression. After iron administration, TfR1 was restored to normal levels in the liver but not in the spleen, probably because of high extramedullary erythropoiesis. Conversely, TfR1 expression was reduced after splenectomy in liver, duodenum, and muscle. A possible explanation for this different behavior may reside in the fact that iron was available at different rates, with splenectomy characterized by a more progressive increase of iron.

Liver Iron Modulates Hepcidin Expression During Chronically Elevated Erythropoiesis

Hepcidin expression is inhibited not only by the stimulation of erythropoiesis (reviewed[15]), but also by iron deficiency per se.[35] The observation that both elevated erythropoiesis and iron deficiency are present in Tg6 mice led us to analyze whether iron or erythropoiesis is the main signal controlling hepcidin expression. We increased iron availability in Tg6 mice by (1) providing additional iron, and (2) reducing erythropoiesis and iron use by splenectomy. Both treatments restored liver Ft and iron content and increased serum iron levels in Tg6_DXT and Tg6_SPL mice. These treatments increased and reduced hematocrit, respectively, but reticulocyte count and CD71-positive cells were decreased in both cases (Fig. S2). Since the last stage of erythropoiesis is iron-dependent, we suggest that additional iron favored the maturation of reticulocytes and CD71 cells. This is in line with the persistence of low %Tf-sat in Tg6_DXT mice because of erythropoiesis-dependent iron consumption. On the other hand, the reduced erythropoietic activity and hematocrit in splenectomized mice is in agreement with the 2-week timepoint of our experimental design. In fact, at that time, hematocrit is still decreasing (Fig. S1) while Epo is further increased to counteract the loss of extramedullary erythropoiesis and recover Tg6 “normal” hematocrit.

Hepcidin expression in Tg6 mice was strongly up-regulated by higher iron availability, the effect of iron dextran being more pronounced (Fig. 1B,C), possibly because of the different degree of liver iron overload (higher in Tg6_DXT than in Tg6_SPL). BMP6 expression changed in parallel with hepcidin when iron availability increased; notably, in agreement with a previous study,[26] correlation (Table S2) and linear regression analysis revealed that liver L-Ft was the only single predictor of BMP6 expression.

The interplay between the storage and erythroid regulators[1] has been previously studied in models of β-thalassemia characterized by reticulocytosis, elevated Epo levels, and low hepcidin expression despite liver iron overload.[19] Accordingly, the erythropoietic drive has been suggested to represent the major regulator of hepcidin expression.[13, 18, 20, 21] The present results showing that treatments with opposite effects on erythropoiesis (enhancement by iron dextran and down-regulation by splenectomy) result in the same inducing effect on both liver iron content and hepcidin expression indicate that erythropoietic signals do not control hepcidin expression directly in this setting. Rather, our data suggest that iron consumption by erythroid cells modulates liver iron content, which, in turn, regulates hepcidin synthesis. Therefore, under conditions of chronically elevated and effective erythropoiesis, the erythroid regulator corresponds to iron consumption by erythroid cells and liver iron appears to be an important regulator of hepcidin expression. A limitation of this study is the lack of information about the role of circulating iron, which has been shown to be involved in acute hepcidin induction.[26] Despite the lack of changes in the P-ERK/total ERK ratio, which is downstream of %Tf-sat,[37, 38] we cannot rule out a participation of increased %Tf-sat in hepcidin induction in Tg6-DXT animals. However, the elevated erythropoietic rate and high intestinal absorption of Tg6 mice, combined with the up-regulation of TfR1 in the liver, would make it difficult to find experimental conditions in our model similar to those that allowed Corradini et al.[26] to separate the effects of liver iron from those of %Tf-sat. It is reasonable to expect that in Tg6 the increase in %Tf-sat obtained by gavage iron feeding would be transient. Indeed, iron-DXT enhanced erythropoiesis-dependent iron consumption and restored tissue iron levels but did not increase %Tf-sat (Fig. 2).

Previous work in vitro and in vivo[10-12] suggested that Epo may have a direct effect on hepcidin. Our results show that a direct effect of Epo on hepcidin expression is unlikely, as iron treatment and splenectomy increased hepcidin expression in the face of constantly high Epo levels.

Molecular Mechanisms Regulating Hepcidin Expression During Chronically Elevated Erythropoiesis

We also investigated molecular pathways at the basis of liver hepcidin regulation in Tg6 animals. We observed up-regulation of the SMAD pathway after iron treatment and down-regulation in Tg6 mice (Fig. 5), which is in line with the very low level of hepcidin and BMP6 expression. Given the iron-dependent induction of hepatic BMP6,[4] and considering the subsequent activation of hepcidin expression through the SMAD pathway,[5] these findings confirm that liver iron is an important signal setting hepcidin expression. Iron has been suggested to regulate hepcidin also through the MAPK/ERK pathway.[29, 30] In our results, P-ERK showed only a nonsignificant tendency to increase in response to iron in both genotypes. In line with previous data,[26] this suggests that the MAPK/ERK pathway is not involved in the regulation of iron homeostasis in vivo.

Subsequently, we analyzed the expression of ID1, TMPRSS6, and SMAD7, all of which act downstream of the SMAD pathway.[5] We found a remarkable increase of ID1, which is a negative-feedback inhibitor acting to avoid continuous up-regulation of hepcidin in response to iron,[7] and SMAD7, a potent inhibitor of hepcidin,[32] in Wt_DXT animals, whereas in Tg6_DXT and Tg6_SPL, ID1 was up-regulated significantly but to a lower extent. Since hepcidin expression was up-regulated 7-fold in WT_DXT and 90-fold in Tg6_DXT, we suggest that ID1 and SMAD7 may be responsible for setting the upper limit of hepcidin rather than inhibiting hepcidin expression in response to iron. This hypothesis is in line with results[26] reporting a plateau for iron-induced hepcidin expression, accompanied by up-regulation of SMAD7 and ID1, despite a continued increase in liver iron.

TMPRSS6 encodes the ID1-regulated[7] serine protease MT-2 that participates in iron sensing and inhibition of hepcidin.[36-38] In our case, TMPRSS6 was not significantly altered by iron and its modest changes did not mirror ID1 expression. Although TMPRSS6 may be involved in the suppression of hepcidin at a later timepoint than the 2 weeks used in our study, our results are in agreement with previous studies[33, 39] suggesting that TMPRSS6 is not required for erythropoiesis-induced repression of hepcidin. Since TMPRSS6 deletion ameliorated iron overload in β-thalassemic mice,[40] further investigation of the mechanisms regulating hepcidin in ineffective versus effective erythropoiesis appears necessary.

In summary, we showed that hepcidin-dependent increased duodenal iron absorption is the primary source of iron to sustain the iron demand triggered by excessive, chronic and effective erythropoiesis. Moreover, we have provided evidence that under these conditions liver iron regulates hepcidin through the BMP6/SMAD pathway.


The authors thank Alessandra Alberghini and Paolo Buratti for support.

Author Contributions

V.D. and E.G. designed and performed the experimental work, analyzed the data, and wrote the article. S.R. and J.V performed research, interpreted the data and revised the article. P.S. and A.M.N. performed research. M.G. and G.C. designed the study, provided vital reagents, interpreted the data and wrote the article.