Smad7 deficiency decreases iron and haemoglobin through hepcidin up‐regulation by multilayer compensatory mechanisms

Abstract To maintain iron homoeostasis, the iron regulatory hormone hepcidin is tightly controlled by BMP‐Smad signalling pathway, but the physiological role of Smad7 in hepcidin regulation remains elusive. We generated and characterized hepatocyte‐specific Smad7 knockout mice (Smad7 Alb/Alb), which showed decreased serum iron, tissue iron, haemoglobin concentration, up‐regulated hepcidin and increased phosphor‐Smad1/5/8 levels in both isolated primary hepatocytes and liver tissues. Increased levels of hepcidin lead to reduced expression of intestinal ferroportin and mild iron deficiency anaemia. Interestingly, we found no difference in hepcidin expression or phosphor‐Smad1/5/8 levels between iron‐challenged Smad7 Alb/Alb and Smad7 flox/flox, suggesting other factors assume the role of iron‐induced hepcidin regulation in Smad7 deletion. We performed RNA‐seq to identify differentially expressed genes in the liver. Significantly up‐regulated genes were then mapped to pathways, revealing TGF‐β signalling as one of the most relevant pathways, including the up‐regulated genes Smad6, Bambi and Fst (Follistatin). We found that Smad6 and Bambi—but not Follistatin—are controlled by the iron‐BMP–Smad pathway. Overexpressing Smad6, Bambi or Follistatin in cells significantly reduced hepcidin expression. Smad7 functions as a key regulator of iron homoeostasis by negatively controlling hepcidin expression, and Smad6 and Smad7 have non‐redundant roles. Smad6, Bambi and Follistatin serve as additional inhibitors of hepcidin in the liver.

Perturbations in hepcidin expression can lead to a variety of iron-related disorders. For example, reduced hepcidin level causes iron overload in hereditary haemochromatosis (HH) and iron-loading anaemia, which is induced by ineffective erythropoiesis. 4 In HH types I, II and III, mutations either in the hepcidin-encoding gene HAMP or in genes that encode hepcidin regulators can reduce the expression of hepcidin, thereby increasing duodenal iron absorption and causing clinical iron overload. [5][6][7] In contrast, increased hepcidin expression causes iron restriction in a variety of inflammatory conditions, including autoimmune disease, critical illness, certain types of cancers and chronic kidney disease. 8 Therefore, considerable effort has been devoted to developing agents that target hepcidin and/or its regulators in order to develop novel therapeutic strategies for treating iron-related disorders. 9 In addition, hepcidin and hepcidin agonists can exert a protective effect on the liver, heart and other vital organs by redistributing iron into macrophages in the liver and spleen. Thus, given the high therapeutic potential of hepcidin, understanding how hepcidin is regulated in vivo is essential.
In hepatocytes, hepcidin expression is regulated by the BMP-Smad signalling pathway. Binding of BMP ligands (eg BMP6) to BMP receptors on the surface of hepatocytes triggers the downstream phosphorylation of Smad proteins. 10,11 Under dietary iron stimulation, hepatic BMP6 triggers the phosphorylation of Smad1/5/8, together with Smad4, to translocate to the nucleus, where they activates hepcidin expression. 12 Therefore, both Bmp6-deficient mice and mice with liver-specific Smad4 deletion have reduced hepcidin expression and develop an severe iron-overload phenotype. 12,13 Results obtained from studying patients with HH types I, II or IIItogether with their corresponding genetic mouse models-support the notion that defective BMP-Smad signalling leads to hepcidin insufficiency. 6,7,[14][15][16][17] Smad7 is a negative regulators of BMP-Smad signalling, and the function of Smad7 protein in iron metabolism is poorly understood, although a growing body of in vitro evidence supports the notion that inhibitory Smads regulate hepcidin expression. 18,19 Based on a genomewide liver transcription profiling study, the expression of Smad7 was found to be up-regulated by iron-enriched diet. 20 However, whether-and how-the Smad7 regulates dietary iron intake and hepcidin expression in the liver is currently unknown. Therefore, in this study, we generated and characterized a hepatocyte-specific Smad7-knockout mouse model to investigate the physiological role of Smad7 in regulating iron metabolism.

| Animals and treatments
Conditional Smad7-floxed 21 mice were backcrossed with wild-type C57BL/6 mice (SLRC Laboratory Animal Co., Ltd., Shanghai, China) for at least seven generations, then crossed with albumin-Cre (Alb) transgenic mice (on a C57BL/6 background) to obtain hepatocytespecific Smad7-knockout (Smad7 Alb/Alb ) mice. The Smad7-knockout mice used in this study were 8 week old of littermates. Hfe À/À mice were kindly provided by Dr. Nancy C. Andrews, 22 and Smad4 Alb/Alb mice were kindly provided by Dr. Chu-xia Deng. 12 The Hfe À/À and Smad4 Alb/Alb mice were maintained on the 129/SvEvTac background, and 8-week-old mice were used in this study. All mice were housed under specific pathogen-free conditions and fed a standard rodent diet (SLRC Laboratory Animal Co., Ltd, Shanghai, China) containing 232 mg/kg iron. 23 The iron-rich diet used for the iron-challenged experiments was composed of standard diet containing 8.3 g/kg carbonyl iron. All animal protocols were approved by the Animal Studies Serum iron (SI) and unsaturated iron-binding capacity (UIBC) were measured using a commercially available colorimetry-based detection kit (Pointe Scientific). Total iron-binding capacity (TIBC) and transferrin saturation (TS) were calculated from SI and UIBC as follows: TIBC = SI + UIBC and TS = (SI/TIBC 9 100). Tissue non-haem iron concentration was measured as previously described. 24

| Ferroportin immunohistochemistry
Intestinal ferroportin detection using immunohistochemistry and Perls' Prussian blue iron staining was performed as previously described. 23 2.4 | Isolation and culture of primary hepatocyte Primary hepatocytes were isolated as previously described, 25

| Plasmid generation and overexpression in cell lines
The open reading frames of the Smad6, Smad7, Bambi and Fst mRNAs (NCBI reference sequences NM_005585.4, NM_001042660.1, NM_ 012342.2 and NM_006350.3, respectively) were amplified from a cDNA library of the HepG2 cell line and inserted into the pCMV-3tag-3A vector (Stratagene). All constructs and their protein products were confirmed using DNA sequencing and Western blot analysis, respectively. Huh7 cells, a human hepatoma cell line, were plated in 12-well plates and cultured at 37°C in 5% CO 2 with 1 mL/well DMEM (Gibco) containing 15% (v/v) heat-inactivated foetal bovine serum (Gibco). The cells were then transfected with the respective plasmid using X-tremeGENE HP DNA transfection reagents (Roche). Where indicated, 36 hours after transfection, human recombinant BMP6 (R&D systems) was added to the wells to a final concentration of 10 ng/mL. After incubating for an additional 12 hours, the cells were collected for the following analyses.

| RNA extraction and real-time PCR analysis
RNA extraction and real-time PCR analysis of gene expression were performed as previously described. 26 Relative expression was normalized to internal control b-actin. The primer sequences are listed in Table S1.

| RNA-seq data analysis
Eight-week-old female Smad7 flox/flox and Smad7 Alb/Alb mice were fed an iron-rich diet for 3 days. Total RNA was then isolated from the livers (3 mice per genotype), and RNA sequencing libraries were generated using the TruSeq RNA Sample Preparation Kit (Illumina). The Illumina HiSeq 2000 platform was used with 100-bp paired-end reads in accordance with the manufacturer's instructions. RNA-seq reads were mapped to the mouse reference genome (mm9, NCBI build 37) using TopHat. 27 Only uniquely aligned reads were used for gene and exon quantification. The Cufflinks tool was used to quantify isoform expression. 28 Genes that were significantly up-regulated (q < 0.05) are listed in Table S2. These genes were then mapped to signalling pathways using the KEGG pathway mapping tool (http:// www.genome.jp/kegg/tool/map_pathway1.html).

| Statistical analysis
All summary data are presented as the mean AE SD. The Student's t test was used to compare two groups. For multiple group comparisons, we used an ANOVA followed by Tukey's post hoc test. If data did not meet the assumption of homogeneity of variance (Bartlett's test), log-transformed values were used in ANOVA. Differences were considered significant if P < .05. Statistical analyses were performed using R (http://www.r-project.org).

| Liver-specific deletion of Smad7 caused increased hepcidin expression and iron deficiency
Smad7 interacts with the TGF-b type I receptor via the MH2 domain, preventing phosphorylation of effector Smad proteins. 29 To generate mice with hepatocyte-specific Smad7 deletion, mice carrying the Smad7 conditional knockout allele (Smad7 flox/flox ) 21 were backcrossed with wild-type C57BL/6 mice at least seven generations and then crossed with albumin-Cre (Alb) transgenic mice, yielding Smad7 liver-specific knockout mice in which the MH2 domain in exon 4 of Smad7 is deleted. Heterozygous hepatocyte-specific knockout mice (Smad7 WT/Alb ) were used to generate Smad7 flox/flox and Smad7 Alb/Alb mice. Primary hepatocytes were isolated from Sma-d7 Alb/Alb mouse livers and had a 98% reduction in Smad7 expression ( Figure S1).
Compared with control mice, both male and female Smad7 Alb/Alb mice had reduced levels of non-haem iron in the liver and spleen ( Figure 1A-C). Smad7 Alb/Alb mice also had reduced levels of ferritin-L protein in the liver, indicating decreased iron stores ( Figure 1D).
An analysis of serum samples revealed that Smad7 Alb/Alb mice have decreased serum iron (SI) and transferrin saturation (TS) levels (Table 1). Moreover, Smad7 Alb/Alb mice have an altered haematological profile, including decreased haemoglobin concentration, mean corpuscular volume (MCV), mean corpuscular haemoglobin (MCH) and mean corpuscular haemoglobin concentration (MCHC). Taken together, these serum and haematology results indicate that Sma-d7 Alb/Alb mice have mild iron deficiency anaemia. The serum and haematology data are summarized in Table 1. 3.2 | Smad7 Alb/Alb mice fed an iron-rich diet had up-regulated expressions of Smad6, Fst and Bambi The expression of Smad7 has been linked to dietary iron. 19,20,30 We thus have been suggested that feeding Smad7 Alb/Alb mice with AN ET AL.
| 3037 an iron-rich diet might induce a more robust phenotype.
According to the report, the hepcidin level reached to its peak at the 3rd day of iron-rich diet treatment. 30 We therefore choose a 3-day iron-rich diet treatment for the following experiments.
To identify these potential factors, we performed RNA-seq analysis and examined which genes were differentially expressed in the liver between Smad7 Alb/Alb mice and Smad7 flox/flox mice under ironrich dietary condition. A total of 52 genes were significantly up-regulated in Smad7 Alb/Alb mice (q < 0.05) and were selected for further analysis; these 52 genes are listed in Table S2. The genes were mapped to signalling pathways using the KEGG pathway mapping, and the pathways with ≥5 hits are summarized in Table 2; all pathways with ≥3 hits are summarized in Table S3.
Among the top-rated pathways (  Bambi interacts with membrane BMP receptors to inhibit BMP signal transduction, 31 and Smad6 inhibits the phosphorylation of Smad proteins. Fst encoding protein Follistatin binds to activin and BMPs, thereby blocking downstream signalling. 32 Based on these functions, we investigated whether these proteins played a role in limiting iron-induced hepcidin expression in the absence of Smad7.

| Smad6, Bambi and Fst are differentially controlled by the iron-BMP-Smad pathway
Because we found no detectable change in phosphor-Smad2/3 levels ( Figure 3D) or activin expression ( Figure S4) in iron-challenged Smad7 Alb/Alb mice, we tested whether the BMP6-Smad1/5/8 pathway controls these putative negative regulators of hepcidin. Accordingly, we measured the mRNA levels of Smad6, Bambi and Fst in Hfe À/À and Smad7 Alb/Alb mice fed either a normal iron diet or an iron-rich diet; we selected these two mouse lines because Hfe À/À mice have impaired BMP6-Smad1/5/8 signalling, whereas Smad7 Alb/ Alb mice have enhanced signalling.
We found decreased hepatic expression of Smad6 in Hfe À/À mice ( Figure 4A) and increased hepatic expression of Smad6 in Smad7 Alb/ Alb mice ( Figure 4B). Moreover, Smad6 expression changed in response to an iron-rich diet ( Figure 4A and B) and BMP6 treatment ( Figure 4C) in mice and primary hepatocytes, respectively. Similarly, Bambi expression decreased slightly in Hfe À/À mice but increased in iron-challenged Smad7 Alb/Alb mice ( Figure 4A and B). Bambi expression also increased in primary hepatocytes in response to 10 ng/mL BMP6 ( Figure 4C), which indicates that Bambi is also regulated by the BMP6-Smad1/5/8 pathway. In contrast, the expression of Fst   Table S2. Full list of pathways with hits ≥ 3 genes was summarized in Table S3. Genes were mapped into pathways using KEGG pathway mapping. Tmprss6, which encodes a serine protease that represses hepcidin expression, causes decreased liver non-haem iron concentration and decreased mean corpuscular volume; however, Tmprss6 knockout mice developed a more severe phenotype than Smad7 Alb/Alb mice. 38 In addition, a recent study suggested that Tmprss6 plays a key role in erythroferrone-mediated hepcidin suppression. 39 In contrast, hepatic expression of Tmprss6 remains unchanged in our Smad7 Alb/Alb mice ( Figure S3). Taken together, these data indicate that Smad7 together with other hepcidin-negative regulators plays an essential role in maintaining iron homoeostasis.
Smad7 expression is up-regulated in mice fed an iron-rich diet and down-regulated in mice fed an iron-deficient diet. 20 Figure 4A and Figure S7) Figure 5).
In conclusion, we report that hepatic Smad7 plays an essential