Runx3 regulates iron metabolism via modulation of BMP signalling

Abstract Objectives Runx3, a member of the Runx family of transcription factors, has been studied as a tumour suppressor and key player of organ development. In a previous study, we reported differentiation failure and excessive angiogenesis in the liver of Runx3 knock‐out (KO) mice. Here, we examined a function of the Runx3 in liver, especially in iron metabolism. Methods We performed histological and immunohistological analyses of the Runx3 KO mouse liver. RNA‐sequencing analyses were performed on primary hepatocytes isolated from Runx3 conditional KO (cKO) mice. The effect of Runx3 knock‐down (KD) was also investigated using siRNA‐mediated KD in functional human hepatocytes and human hepatocellular carcinoma cells. Result We observed an iron‐overloaded liver with decreased expression of hepcidin in Runx3 KO mice. Expression of BMP6, a regulator of hepcidin transcription, and activity of the BMP pathway were decreased in the liver tissue of Runx3 KO mice. Transcriptome analysis on primary hepatocytes isolated from Runx3 cKO mice also revealed that iron‐induced increase in BMP6 was mediated by Runx3. Similar results were observed in Runx3 knock‐down experiments using HepaRG cells and HepG2 cells. Finally, we showed that Runx3 enhanced the activity of the BMP6 promoter by responding to iron stimuli in the hepatocytes. Conclusion In conclusion, we suggest that Runx3 plays important roles in iron metabolism of the liver through regulation of BMP signalling.

in non-transferrin-bound iron in the blood, which is a highly reactive form that can cause cellular and visceral damage. 2 Therefore, tight regulation of plasma iron is required to avoid iron-related toxicity in the body.
Hepcidin, a key regulator of iron transport, suppresses the release of iron from macrophages or intestinal cells into the plasma via binding to ferroportin, which induces internalization and degradation of the cellular iron exporter. 3 Genetic deficiency of hepcidin causes excessive iron in blood, which is followed by the deposition of iron and consequent functional failure in the liver and other tissues. [4][5][6][7] The bone morphogenetic protein (BMP) signalling pathway is a major regulatory pathway of hepcidin expression in the liver. 8,9 In hepatocytes, the pathway is initiated by the binding of BMP6 with the BMP receptor (BMPR) complex and a membrane-anchor coreceptor hemojuvelin (HJV) at the cell surface. 8 The binding elevates kinase activity of the BMPR complex and results in phosphorylation of Smad1, Smad5 and Smad8, the cytoplasmic effectors of the BMP pathway. Phosphorylated Smad1, 5 and 8 form heteromeric complexes with the common mediator Smad4, and they then translocate into the nucleus to induce the transcription of target genes. 10 A deficiency of the BMP pathway-related genes causes low hepcidin expression, excessive iron in the blood and iron-overloaded organs in mice. [11][12][13][14][15][16] The activity of the BMP signalling pathway in the liver should be associated with the plasma iron concentration to maintain iron homeostasis. In the liver of mice fed high-iron diet, transcriptional activation of BMP6 has been observed. 13 Recent studies suggested liver sinusoidal endothelial cells (LSECs) as main sources of hepatic BMP6 responding to the iron stimuli. 17,18 Hepatocytes, once considered to serve dual roles as iron-sensor and autocrine sources of BMP6, revealed as passive producers of hepcidin regulated by paracrine BMP6 from non-parenchymal cells. 19 However, high expression of transferrin receptor 2 (TfR2) and its unveiled function in hepatocytes 18 implies possible mechanism of direct sensing of iron by the hepcidin producer.
Here, we demonstrated that Runx3 is an upstream regulator of BMP6 in the liver. Prussian blue staining revealed an iron-overloaded liver at postnatal day 1 (PN1) in Runx3 knockout (KO) mice. Hepcidin was decreased in the liver of Runx3 KO mice. Interestingly, a similar iron-overloaded liver was reported in Bmp6 KO mice. 11 To reveal the possible engagement of BMP signalling with Runx3 deficiency-induced iron overload in the liver, we detected BMP6 expression in the liver tissue. The results showed a decrease in BMP6 and BMP signalling in the liver of Runx3 KO mice. A systematic approach using RNA sequencing of primary hepatocytes isolated from Runx3 conditional KO (cKO) mice revealed that BMP6 was specifically induced by iron stimuli, and Runx3 KO using Cre recombinase-expressing adenovirus aborted the ironinduced BMP6 expression in the hepatocytes. Down-regulation of these genes was also observed in Runx3 knock-down (KD) in both HepaRG cells, which are functional human hepatocytes, 20 and HepG2 cells, which are human hepatocellular carcinoma cells. 21 The Runx3 KD abolished iron-induced BMP6 transcription and the resultant activation of BMP signalling in both cells. Furthermore, we found that Runx3 activated the promoter of BMP6 to trigger the BMP signalling-mediated hepcidin regulation by iron stimulation.
Taken together, Runx3 plays important roles in the iron metabolism of the liver through regulation of BMP signalling.

| Runx3 KO and cKO mice
Runx3 knock-out (Runx3 −/− FVB) and Runx3 cKO (Runx3 flox / flox C57BL/6) mice were generated and maintained as described previously. 22,23 The animals were maintained in pathogen-free conditions and monitored daily. All experiments were performed according to the guidelines of the Yonsei University College of Dentistry, Intramural Animal Use and Care Committee.

| Histology and immunohistochemistry
Samples were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) and then embedded in paraffin using standard procedures. Serial paraffin sections (4μm thickness) were prepared, and individual slides were stained with haematoxylin and eosin. Antigen retrieval was achieved by citrate buffer, pH 6.0. After antigen retrieval, immunohistochemical analyses were performed using fol-

| Western blotting analyses
Liver tissue and hepaRG cells underwent lysis by sonication (Next Advance Inc., Averill Park, NY) in radio-immunoprecipitation assay

| Primary hepatocyte isolation and RNA preparation
Primary hepatocytes of Runx3 cKO mice were isolated as described previously. 24 The isolated hepatocytes were infected with Cre

| RNA-sequencing and data analysis
Reverse transcription was performed, and cDNA was synthesized using 5′ adaptor forward and 3′ adaptor reverse primers. Libraries for Illumina sequencing were constructed from cDNA as described. 25 High-throughput RNA sequencing was performed by Theragen Bio Differentially expressed genes were considered in a given library when the p-value was less than 0.05 and a greater-than-or-equal to twofold change in expression across libraries was observed and used to identify the genes differentially expressed between two samples.
Clustered heat maps and volcano plots were drawn using a statistical computing software, R (https://www.R-proje ct.org/).

| Iron overload in Runx3 KO mouse liver hepatocytes at PN1
Runx3 KO mice showed lethality soon after birth, as reported previously. 27 At postnatal day 1, skin pigmentation was observed in the KO mice ( Figure 1A), which revealed a possible abnormality in iron metabolism. 28 Depletion of Runx3 in the liver tissue of Runx3 KO mice was confirmed using immunohistochemistry and immunoblotting ( Figure 1B-1E). Using Perls' Prussian blue staining, substantial iron accumulation was visualized in liver parenchymal cells (hepatocytes) of Runx3 KO mice at PN1 ( Figure 1F). The accumulation of iron was significantly higher in the central lobule region than the peripheral vein region of the liver ( Figure 1F). In addition, an increase in ferritin protein, indicating accumulation of iron in the cytosol, was observed in the hepatocytes of Runx3 KO mice compared to those of WT mice (Figures S1A, S1B).
The iron accumulation in the centrolobular region of liver tissue is a typical pattern of iron overload caused by hepcidin deficiency. 29 Therefore, we examined hepcidin expression in the liver tissue of

| Systemic regulation of BMP signalling-and iron metabolism-related genes by Runx3 KO in primary hepatocytes
The lethality of Runx3 KO mice set limitations on the systemic analysis of the Runx3 effect on iron metabolism. We introduced cKO mice of Runx3 to avoid the limitations. 22 Primary hepatocytes of Runx3 cKO mice were isolated, and Runx3 was knocked out by exogenous The analysis result showed that Runx3 was knocked out in Cre overexpressed hepatocytes ( Figure 2B). We examined expressions of the direct target genes of Runx3, which are known to be positively hTF treatment or both respectively (Figure 2A and Tables S1 and   Table S2). The expression patterns of 294 genes, which were significantly regulated in at least one treated group, were visualized as a heat map ( Figure 2C). The clustered heat map showed a cluster of genes that were highly responsive to hTF treatment ( Figure 2C, dashed-line box). Cre-mediated Runx3 KO suppressed the hTFinduced expression of the 19 genes in this cluster ( Figure 2C, 2D).
To investigate the specificity of BMP6 regulation by hTF treatment and Cre expression, we displayed the DEG analysis results

| Regulation of BMP signalling-and iron metabolism-related genes by Runx3 KD in human hepatocytes and hepatocellular carcinoma cells
Here, we aimed to confirm the role of Runx3 on the expression of BMP signalling-and iron metabolism-related genes using established hepatocyte cell lines. Firstly, we used HepaRG cells, which are functional human hepatocytes. Transfection of Runx3 siRNA successfully decreased the mRNA level of Runx3 in the hepatocytes ( Figure 3A).
The knock-down effect also confirmed in protein level using immunoblotting ( Figure 3B). Similar to the observation in the KO mice, the mRNA level of hepcidin decreased by the knock-down (KD) of Runx3 ( Figure 3C). The Runx3 KD effect also confirmed using HepG2 cells, which are human hepatocellular carcinoma cells ( Figure 3D).
Immunohistochemistry results revealed a decrease in hepcidin expression in the Runx3 KD hepatocellular carcinoma cells ( Figure 3D).
Hepcidin regulates ferroportin expression at the protein level, rather than at the transcription level. 3 We found that Runx3 KD cannot alter the mRNA level of ferroportin ( Figure 3E). The mRNA level of transferrin, a liver-originated iron carrier protein, increased by Runx3 KD in the HepaRG cells ( Figure 3F). To monitor the activity of the BMP pathway in Runx3 KD hepatocytes, we detected the mRNA levels of the BMP ligand and BMP pathway target genes. The quantitative real-time PCR results showed that mRNA levels of BMP6 decreased by Runx3 KD (Figure 3G). The mRNA levels of well-known target genes of the pathway Id1, Smad7 and Atoh8 also decreased in the Runx3 KD hepatocytes (Figure 3H-3J).

| Direct transcriptional regulation of BMP6 by Runx3
BMP6 induction and activation of the BMP pathway by iron stimuli in hepatocytes were reported previously. 40 To investigate whether

Holo-transferrin is an iron-bound form of transferrin, which
is known to induce BMP6 expression. 41 Similar to the experiment using serum, holo-transferrin increases mRNA levels of Id1 and Smad7; however, the effect was cancelled by Runx3 KD (Figure 4D-4F). These results indicate that Runx3 functions as a mediator between iron stimuli and BMP pathway activation.
The Runt domain of the Runx3 has a specificity on a conserved sequence, 5′-YGYGGT-3′. 34 Most of the target genes of Runx3 contain this sequence in their promoter region. We analysed an ~1 Kb upstream sequence from the BMP6 open reading frame to determine if the Runx3 binding sequence (RBS) existed, which was identified as one of the enriched motif in Runx3-bound promoter of natural killer cells. 42 The sequence analysis showed two putative RBSs in the BMP6 promoter region that were evolutionally well-conserved ( Figure 4G). We used a luciferase reporter plasmid containing the BMP6 promoter to investigate the role of Runx3 on the transcriptional activity of the promoter ( Figure 4H). Serum treatment showed a dose-dependent increase of promoter activity in the reporter plasmid-transfected HepG2 cells ( Figure 4I). However, Runx3 KD suppressed activation of the BMP6 promoter induced by serum treatment (Figure 4I).

| DISCUSS ION
Hereditary haemochromatosis (HH) is a term used to describe a group of genetic disorders characterized by increased iron absorption. 43 This absorption may lead to a progressive accumulation of iron in tissues and organs, resulting in impairment of organ structure | 7 of 11 KIM et al. and function, especially of the liver, pancreas, heart, pituitary gland and, likely, joints. The prevailing mechanism in most types of HH is deficiency of hepcidin, originally identified as an antimicrobial peptide 44 and then shown to play a major role in iron homeostasis. 45,46 Hepcidin is synthesized mainly in hepatocytes and controls the plasma iron concentration by binding to ferroportin (also termed SLC40A1), the only known cellular iron exporter. After binding, ferroportin is degraded, reducing both intestinal absorption of iron from enterocytes and iron released from hepatocytes and macrophages.
Increased plasma iron or cellular iron stores, as well as inflammation, generate a negative feedback loop that leads to a restriction of iron release into the plasma and blockade of dietary iron absorption through increased hepcidin production. In this study, Runx3 KO mice showed a haemochromatosis-like phenotype. Skin pigmentation and an iron-overloaded liver were observed in the KO mice. Molecular biological analyses showed that BMP6 expression and activity of the BMP pathway were suppressed in the liver of Runx3 KO mice.
Studies have shown that, at least in rodent models, increasing body iron stimulates the production of BMP6, which binds to a complex of type I and II BMP receptors on the plasma membrane where it binds to BMP responsive elements in the hepcidin promoter, stimulating transcription. The glycosylphosphatidylinositollinked membrane protein HJV binds BMP6 and acts as a co-receptor for the BMP receptor complex. 47 HJV is essential for BMP6 signalling because the disruption of the protein, as occurs in the juvenile form of the iron loading disorder, haemochromatosis, leads to the complete loss of hepcidin production. 48 However, the regulation mechanism of BMP6 by iron stimuli has not yet been revealed. Here, we showed that Runx3 KO or KD in mouse primary hepatocytes, human hepatocytes and human hepatocellular carcinoma decreased BMP6 expression and inhibited the BMP-Smad pathway in human hepatocytes. Therefore, Runx3 regulates iron metabolism of the liver via modulation of BMP signalling ( Figure 5).
The upstream regulatory mechanism of Runx3 is remained to elucidated. Extra-or intra-iron sensing proteins are possible candidates of a Runx3 modulator. The TfR2, a transferrin receptor, is functionally unknown although highly expressed in hepatocytes. 18 Increase or decrease in intracellular iron induces redox change, which is recognized by redox proteins. The iron regulatory proteins, IRP1 and IRP2, possibly modulate Runx3 in a similar way with ferritin. 49 Defining upstream modulator of Runx3 would be a first goal of future study.
Global transcriptome analysis of the Runx3 KO primary hepatocytes with or without hTF treatment showed that the BMP6 induction by iron stimuli was a specific regulation and that Runx3 KO abolished the regulation. However, regulation of the target genes of the BMP-Smad pathway was not observed in the transcriptome analysis. Of note, BMP7, another member of BMP ligands, showed the opposite pattern of expression to BMP6. BMP7 is closely related in structure to BMP6 and shares the receptor complex to activate the BMP-Smad signalling pathway. 50 Furthermore, a previous study showed that BMP7 was upregulated in the liver tissue of BMP6 null mice treated with iron-dextran, and exogenous BMP7 injected into the null mice induced hepcidin expression and reduced an abnormally high concentration of plasma iron. 39 Therefore, a compensatory effect by BMP7 is a possible explanation for the insensitivity of the BMP-Smad pathway in the experiment using primary hepatocytes. F I G U R E 4 Runx3 mediates iron stimuli to direct regulation of BMP6 expression. (A)-(F), Foetal bovine serum (FBS, A-C) or hTF (D)-(F) of indicated doses was used to treat 24-h starved HepG2 cells transfected with control siRNA or Runx3 siRNA. The relative expressions of Id1 and Smad7 were measured with quantitative real-time PCR (B-C and E-F). G, An alignment result of BMP6 promoters of various species. Similar or identical amino acid sequences are indicated by blue or yellow blocks respectively. Putative Runx3 binding sites (RBSs) are indicated by a red dash-lined box. H-I, FBS of indicated doses were used to treat 24-h starved HepG2 cells transfected with control siRNA or Runx3 siRNA with a plasmid for the reporter assay of BMP6 promoter. The activities of the BMP6 promoter were measured by luciferase activities (I). Control or Runx3 siRNA-treated groups are indicated by blue and red bars respectively (B-C, E-F and I) F I G U R E 5 A schematic diagram of iron-induced BMP6 expression via Runx3 and the consequent regulation of hepcidin. The increasing body iron stimulates the production of BMP6, which binds to a complex of type I and II BMP receptors on the plasma membrane of hepatocytes. 11,13 This leads to the phosphorylation of SMAD1, 5 and 8 in the cytoplasm, which allows the binding of SMAD4. The entire complex is then translocated into the nucleus where it binds to BMP responsive elements in the hepcidin promoter, stimulating transcription. However, the regulation mechanism of BMP6 by iron stimuli has not yet been revealed. Here, we showed that Runx3 KO or KD in mouse primary hepatocytes, human hepatocytes and human hepatocellular carcinoma decreased BMP6 expression and inhibited the BMP-Smad pathway in human hepatocytes. Therefore, Runx3 regulates iron metabolism of the liver via modulation of BMP signalling

| CON CLUS ION
In conclusion, this work depicts Runx3 as a transcription factor of regulating hepcidin expression. Our findings highlight possible role of Runx3 in human iron metabolism disorders, such as haemochromatosis, hemosiderosis and atransferrinemia.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw data supporting the conclusions of this article will be made available by the authors.