Interleukin-1 (IL-1 ) transcriptionally activates hepcidin by inducing CCAAT enhancer-binding protein (C/EBP ) expression in hepatocytes

Yohei Kanamori, Masaru Murakami, Makoto Sugiyama, Osamu Hashimoto , Tohru Matsui, and Masayuki Funaba From the Division of Applied Biosciences, Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, the Laboratory of Molecular Biology, Azabu University School of Veterinary Medicine, Sagamihara 252-5201, and the Laboratory of Veterinary Anatomy and Laboratory of Experimental Animal Science, Kitasato University School of Veterinary Medicine, Towada 034-8628, Japan

Inflammation is a symptom of an adaptive response that is triggered by microbial infection or tissue injury (1,2). Initially, infiltrated neutrophils kill invading microbes by releasing granules containing toxic contents such as reactive oxygen species and proteases; this process is followed by macrophage-mediated resolution and repair (3,4). The inflammatory process persists when pathogens are insufficiently eliminated during the acute response (4,5). Chronic inflammation is well known to associate with a wide variety of diseases, including progressive and irreversible damage to the central nervous system, tumorigenesis, and metabolic syndrome (5)(6)(7).
One of the various disorders resulting from chronic inflammation is anemia of inflammation (8,9). The cause of anemia of inflammation is multifactorial, and precise mechanisms underlying its pathogenesis are not fully elucidated (10); however, inflammation-induced hepcidin production has been suggested to be responsible for the onset of anemia of inflammation (8,9). Hepcidin is a liver-derived peptide hormone that negatively regulates plasma iron levels (11)(12)(13). Hepcidin stimulates the internalization and degradation of ferroportin, an iron exporter expressed in macrophages and intestinal epithelial cells. Therefore, overproduction of hepcidin resulting from inflammation inhibits iron release from macrophages for erythropoiesis and the intestinal absorption of iron (11)(12)(13), leading to the onset of anemia of inflammation.
Hepcidin expression has been shown to increase with elevated hepatic iron levels; hepatic iron induces bone morphogenetic protein (BMP) 2 6 expression, which stimulates hepcidin transcription via BMP-responsive elements (BMP-RE) 1 and 2 on the hepcidin gene (12). However, proinflammatory cytokines such as interleukin (IL)-6 and oncostatin M also up-regulate hepcidin expression; these cytokines transactivate the hepcidin gene via the signal transducer and activator of transcription (STAT) 3-binding site (STAT-BS) on the hepcidin gene (11)(12)(13). Furthermore, we recently found that activin B is induced in sinusoidal endothelial cells and Kupffer cells in response to intraperitoneal lipopolysaccharide (LPS) injection, which activates hepcidin transcription via BMP-RE1 and BMP-RE2 (14). In view of the production of diverse cytokines during inflammation, other cytokines may be involved in regulating hepcidin transcription via regions other than known elements. Here, we show that IL-1␤, a proinflammatory cytokine, stimulates hepcidin transcription mainly via a CCAAT enhancerbinding protein (C/EBP)-binding site (C/EBP-BS) located in the hepcidin promoter.

IL-1␤ stimulates hepcidin transcription through region other than BMP-REs and STAT-BS
We first examined the effects of IL-1␤ on hepcidin expression in primary hepatocytes; consistent with a previous study (15), treatment with IL-1␤ for 24 h stimulated hepcidin expression in primary mouse hepatocytes (Fig. 1A). In contrast, IL-1␤induced up-regulation of hepcidin expression was not detected in primary rat hepatocytes (Fig. 1A). Considering that in primary hepatocytes from mice and rats, IL-1␤ induces iNOS, an IL-1␤-responsive gene (Fig. 1B) (16), both mouse hepatocytes and rat hepatocytes are defined as IL-1␤-responsive cells. IL-1␤ increased hepcidin expression within 4 h after the treatment in primary mouse hepatocytes, and the increased expression continued after at least 12 h of IL-1␤ treatment (supplemental Fig.  S1A). In contrast, hepcidin expression was slightly higher in primary rat hepatocytes treated with IL-1␤ within 12 h than in control hepatocytes, but this difference was due to a reduction of hepcidin expression in the control cells (supplemental Fig.  S1B). We also examined whether IL-1␤ stimulates hepcidin expression in HepG2 cells, a human liver-derived cell line. Similar to the primary mouse hepatocytes, HepG2 cells responded to IL-1␤ by increasing hepcidin expression (Fig. 1C).
IL-1␤ activates the transcription factor nuclear factor-B (NF-B) (17). To evaluate the role of the NF-B pathway in IL-1␤-induced hepcidin expression, HepG2 cells were treated with BAY 11-7085, an inhibitor of NF-B pathway by blocking phosphorylation of inhibitor B␣ (IB␣) (18). IL-1␤-induced hepcidin expression was inhibited by pretreatment with BAY 11-7085 in HepG2 cells, suggesting the up-regulation of hepcidin expression through activation of the NF-B pathway (Fig.  1D).
To examine the necessity of novel protein synthesis for IL-1␤induced hepcidin expression, cycloheximide, an inhibitor of protein synthesis (19), was added to cells, and the data showed that IL-1␤-induced hepcidin expression was cycloheximidesensitive in both HepG2 cells (Fig. 1E) and mouse hepatocytes (supplemental Fig. S2), suggesting that de novo protein synthesis is required for IL-1␤-induced hepcidin expression.
A previous study revealed that IL-1␤ stimulates hepcidin expression by inducing BMP2 expression in Huh7 cells, a human liver-derived cell line (20). In addition, IL-1␤ up-regulated IL-6 expression in various cell lines, including Huh7 cells (21). Thus, it is possible that BMP2 and/or IL-6 are produced in response to IL-1␤ treatment in hepatocytes and that these molecules induce hepcidin expression in an autocrine manner.
IL-1␤ transiently increased BMP2 expression in HepG2 cells within 2 h of treatment initiation ( Fig. 2A). In contrast, IL-1␤ did not increase Bmp2 expression in mouse hepatocytes (supplemental Fig. S3A), whereas IL-1␤-induced Bmp2 expression peaked at 8 h in primary rat hepatocytes (supplemental Fig.  S3B). After BMP forms a complex with its respective receptors, Hepcidin expression was examined by RT-qPCR analysis. The expression levels in control cells treated without either BAY 11-7085 or cycloheximide were defined as 1. The data are presented as the mean Ϯ S.E. (n ϭ 3). *, p Ͻ 0.05 and **, p Ͻ 0.01 versus cells treated with the respective inhibitor (vehicle, BAY 11-7085, or cycloheximide) in the absence of IL-1␤. †, p Ͻ 0.05 and † †, p Ͻ 0.01 versus cells with corresponding IL-1␤ treatments in the absence of inhibitor (i.e. BAY 11-7085 (D) or cycloheximide (E)).
We next examined whether the induction of BMP2 by IL-1␤ has a role in hepcidin expression in HepG2 cells; siRNA targeting BMP2 was transfected to down-regulate BMP2 expression (supplemental Fig. S4). Although knockdown of the BMP2 gene decreased basal expression of hepcidin, IL-1␤ still increased expression of hepcidin ( Fig. 2C) without an increase in phosphorylation of Smad1/5/8 (Fig. 2D). The role of the induced BMP2 in hepcidin transcription was further evaluated by luciferase-based reporter assays in HepG2 cells. BMP stimulates hepcidin transcription via BMP-REs 1 and 2 in the hepcidin promoter: BMP-RE1 spans nt Ϫ155 to Ϫ150, and BMP-RE2 spans nt Ϫ1678 to Ϫ1673 (22)(23)(24). IL-1␤ increased the luciferase activity of wild-type reporter as expected, suggesting that IL-1␤-induced hepcidin expression is transcriptionally regulated (Fig. 2E). Mutations in either BMP-RE (mBMP-RE) decreased the basal transcription of hepcidin; however, their responsiveness to IL-1␤ (i.e. fold-induction of luciferase expression after IL-1␤ treatment) was not decreased but rather increased (wild-type reporter, 18-fold; reporter with mBMP-RE1, 41-fold; reporter with mBMP-RE2, 58-fold) (Fig. 2E). In addition, mutants of both BMP-RE1 and BMP-RE2 did not decrease the fold-induction in response to IL-1␤ (Fig. 2E). In contrast, mutations of both BMP-REs in the hepcidin promoter blunted the transcriptional response to either BMP2 or ALK3(QD) expression, the latter of which is a constitutively active ALK3 (25) (supplemental Fig. S5). Furthermore, treatment with LDN-193189, an inhibitor of BMP type I receptor (26), also decreased the basal transcription of hepcidin but did not prevent the responsiveness to IL-1␤ (Fig. 2F). We concluded that the induction of BMP2 and subsequent transcription of hepcidin via BMP-REs does not contribute to IL-1␤-induced hepcidin expression based on the following results: 1) induction of Bmp2 expression was not detected in mouse hepatocytes irrespective of IL-1␤-induced hepcidin expression; 2) Bmp2 expression was increased by IL-1␤ in rat hepatocytes; however, Smad1/5/8 phosphorylation was not detected, and the hepcidin induction was minimal; 3) knockdown of the BMP2 gene blocked IL-1␤-induced phosphorylation of Smad1/5/8, but IL-1␤ still increased expression of hepcidin; and 4) neither BMP-RE mutations within the hepcidin promoter nor LDN-193189 treatment inhibited the transcriptional responsiveness of HepG2 cells to IL-1␤.
IL-6 expression was increased by IL-1␤ treatment in both HepG2 cells and mouse hepatocytes ( Fig. 3A and supplemental Fig. S6). In HepG2 cells, up-regulation of IL-6 expression was detected within 2 h after IL-1␤ treatment and maintained for at least 12 h (Fig. 3A). The rapid induction of IL-6 expression by IL-1␤ was transcriptionally regulated; IL-1␤ treatment increased the luciferase expression of the reporter containing  the IL-6 promoter (supplemental Fig. S7A). There is a putative NF-B site within the IL-6 gene located from nt Ϫ123 to Ϫ111 (supplemental Fig. S7B). Mutations of this potential NF-B site blunted IL-1␤-induced IL-6 transcription (supplemental Fig.  7A), suggesting that IL-1␤ transcriptionally stimulates IL-6 expression by activating the NF-B pathway.
Consistent with the induction of IL-6 by IL-1␤, STAT3, a molecule that is phosphorylated in response to IL-6, showed increased phosphorylated levels in HepG2 cells (Fig. 2B); furthermore, IL-1␤ also increased phosphorylated STAT3 levels in rat hepatocytes but not mouse hepatocytes (supplemental Fig. S3, C and D). Previous studies have shown that IL-6 stimulated hepcidin transcription by activating STAT3 to promote its binding to the STAT-BS spanning nt Ϫ143 to Ϫ134 of the hepcidin promoter (11)(12)(13). Mutations of the STAT-BS slightly decreased the transcriptional responsiveness to IL-1␤ (Fig. 3B); fold-induction of luciferase expression after IL-1␤ treatment was 16-fold for the wild-type reporter and 11-fold for the reporter with mSTAT. IL-6-induced hepcidin transcription was expectedly inhibited by mutations of the STAT-BS (supplemental Fig. S8). These results suggest that IL-1␤ induces IL-6 expression but that this induction does not majorly contribute to hepcidin transcription.

C/EBP␦ binding to the C/EBP-BS in the hepcidin promoter is responsible for IL-1␤-induced hepcidin transcription
Provided that IL-1␤ transcriptionally regulates hepcidin expression, the region responsible for gene induction was next explored by using a series of deleted reporters (Fig. 4A). Deletion of a region spanning nt Ϫ2018 to Ϫ1419 decreased luciferase expression in the absence of IL-1␤; this could be explained by the deletion of BMP-RE2 (24). However, the responsiveness to IL-1␤ was not decreased but rather increased (fold-induction of luciferase expression by IL-1␤: hepcidin(Ϫ2018)-luc, 13-fold; hepcidin(Ϫ1418)-luc, 59-fold). Deletion of a region with the hepcidin promoter from nt Ϫ1418 to Ϫ356 did not affect IL-1␤-induced hepcidin transcription. In contrast, deletion of the region spanning nt Ϫ355 to Ϫ271 significantly decreased the responsiveness to IL-1␤ (Fig. 4A). The nucleotide sequence of this region indicates an element closely related to a C/EBP-BS (27) that spans nt Ϫ329 to Ϫ320 (Fig.  4B). The mutations of this putative C/EBP-BS in the hepcidin promoter blunted the responsiveness to IL-1␤ (Fig. 4C).
We further evaluated the relative importance of C/EBP-BS and interactive relationships among the regulatory elements for hepcidin transcription, C/EBP-BS, BMP-REs, and STAT-BS (Fig. 4D). The mutations of C/EBP-BS greatly decreased responsiveness to IL-1␤ (fold-induction: wild-type, 20-fold; mC/EBP-BS, 4-fold), whereas that of STAT-BS slightly decreased (13-fold) and that of BMP-RE1,2 did not decrease but rather increased IL-1␤ responsiveness (38-fold); these results are consistent with those shown in Figs. 2E, 3B, and 4C. Combinational mutations of C/EBP-BS and STAT-BS further decreased responsiveness to IL-1␤ (2-fold), which was comparable with the results on the reporter with all mutations of C/EBP-BS, STAT-BS, and BMP-RE1,2 (2-fold). These results suggest that C/EBP-BS is the principle region responsible for IL-1␤-induced hepcidin transcription and that STAT-BS is also involved in the responsiveness. The present results also suggest the independent role of C/EBP-BS, STAT-BS, and BMP-RE1,2 in IL-1␤-induced hepcidin transcription. These results suggest the independent regulation of hepcidin transcription via C/EBP-BS, BMP-REs, and STAT-BS.
There are several C/EBP isoforms: C/EBP␣, -␤, -␥, -␦, -⑀, and -(28); the mRNA levels of C/EBP␦ were increased within 2 h after IL-1␤ stimulation and maintained for at least 12 h (Fig.  5A). In contrast, C/EBP␣ expression was transiently decreased by IL-1␤. In addition, IL-1␤ minimally affected the expression of C/EBP␤ and C/EBP (Ͼ2-fold), and significant expression of neither C/EBP␥ nor C/EBP⑀ was detected (data not shown). IL-1␤-induced C/EBP␦ expression was also detected at the protein level (Fig. 5B). Similar to the response in HepG2 cells, clear up-regulation of C/EBP␦ expression by IL-1␤ was detected in mouse and rat hepatocytes (supplemental Fig. S9). Although substantial up-regulation of C/EBP␦ expression was detected even at 12 h after IL-1␤ treatment in HepG2 cells and mouse hepatocytes, the marked increase in C/EBP␦ expression by IL-1␤ was relatively transient in rat primary hepatocytes; the reason of the differential response to IL-1␤ on C/EBP␦ induction is not clear. Moreover, BAY 11-7085 blocked IL-1␤-induced C/EBP␦ expression at the mRNA level (Fig. 5C) as well as at the protein level (Fig. 5D), suggesting that activation of NFB by IL-1␤ is involved in the C/EBP␦ gene induction.
To evaluate the involvement of C/EBP␦ in IL-1␤-induced hepcidin expression, we examined the effect of C/EBP␦ gene knockdown. siRNA transfection targeting C/EBP␦ decreased the C/EBP␦ mRNA levels by ϳ80% in HepG2 cells (Fig. 6A). IL-1␤ increased the expression level of C/EBP␦ even in cells transfected with C/EBP␦-siRNA; this could be explained by the imperfect suppression of gene expression by siRNA. Downregulation of C/EBP␦ expression decreased the IL-1␤-induced mRNA expression of hepcidin (Fig. 6B). Unlike IL-1␤, BMP2 and IL-6 did not increase expression of C/EBP␦ in HepG2 cells, irrespective of transfection with siRNA for C/EBP␦ (Fig. 6C). In addition, the gene knockdown of C/EBP␦ did not modulate responsiveness to BMP2 and IL-6 on hepcidin expression (Fig.  6D), suggesting that C/EBP␦ is not involved in BMP2-or IL-6mediated hepcidin expression. An oligonucleotide pulldown assay indicated that binding of C/EBP␦ to the C/EBP-BS on the hepcidin promoter was IL-1␤-dependent (Fig. 6E). All these results suggest that IL-1␤ stimulates hepcidin transcription by activating NF-B, which in turn induces C/EBP␦ production and its subsequent binding to the C/EBP-BS spanning nt Ϫ329 to Ϫ320 on the hepcidin promoter.

The nucleotide sequence of the rat C/EBP-BS on the hepcidin promoter contributes to less efficient transcription of hepcidin by IL-1␤
As shown above, hepcidin expression was slightly higher in rat hepatocytes treated with IL-1␤ than in control rat hepatocytes, resulting from reduction of hepcidin expression with time in the control cells; unlike mouse hepatocytes, hepcidin expression was not increased with time after IL-1␤ treatment in rat hepatocytes ( Fig. 1A and supplemental Fig. S1, A and B). In view of the induction of C/EBP␦ by IL-1␤ in primary rat hepatocytes (supplemental Fig. 9B), we hypothesized that the C/EBP-BS in the rat hepcidin gene cannot mediate efficient transcription in response to IL-1␤. A comparison of the nucleotide sequence among human, mouse, and rat hepcidin pro-

Up-regulation of hepcidin expression by IL-1␤
moters indicates that one nucleotide difference was detected between mouse C/EBP-BS and rat C/EBP-BS: the guanylic acid at nt Ϫ323 in the mouse hepcidin promoter is a thymidylic acid at nt Ϫ319 of the rat hepcidin promoter (Fig. 7A). Mutating the reporter construct containing the mouse hepcidin promoter to mimic that of the rat C/EBP-BS (i.e. mutation of the guanylic acid at nt Ϫ323 to a thymidylic acid) decreased IL-1␤-induced hepcidin transcription (Fig. 7B). In contrast, mutating the rat hepcidin promoter in the luciferase reporter to mimic the sequence of the mouse-type C/EBP-BS increased IL-1␤-induced luciferase expression (Fig. 7C). These changes in transcriptional activity of the swapped reporters are not nonspecific events; stimulating the BMP pathway by ALK3(QD) expression did not affect the transcription of these reporters (supplemental Fig. S10). Furthermore, mutations of the mouse and rat C/EBP-BSs to TTAtGGGcAA and TTAtGGTcAA, respectively (small characters indicate the mutated nucleic acids), decreased the responsiveness of both the mouse and rat hepcidin promoters to IL-1␤ (Fig. 7, B and C). All these results indicate that the reduced activity of IL-1␤ in up-regulating hepcidin expression in rat hepatocytes at least partly results from the nucleotide sequence of the rat C/EBP-BS.

Induction of IL-1␤ in hepatocytes and Kupffer cells in response to LPS stimulates hepcidin transcription via C/EBP␦ production
Previous studies have shown that IL-1␤ expression is up-regulated in response to inflammation in the liver (14, 29 -31). To determine the source of IL-1␤ during inflammation, immunohistochemical analyses were performed in the livers of mice injected with either phosphate-buffered saline (PBS) or LPS (Fig. 8A, Table 1). Immunoreactive IL-1␤ was strongly detected in the cytoplasm of Kupffer cells from LPS-treated livers. Additionally, a large number of hepatocytes was positively stained by an anti-IL-1␤ antibody. IL-1␤-positive hepatocytes were also slightly detected in control livers (Fig. 8A).
C/EBP␦ was localized in hepatocytes that resided in a limited area around the central vein in control mice (Fig. 8B, Table 1). LPS increased the number of C/EBP␦-positive hepatocytes, and immunoreactive C/EBP␦ was also detected in some but not all sinusoidal endothelial cells in LPS-treated mice. Consistent with the results of immunohistochemical analyses, the expression level of C/EBP␦ was higher in LPS-treated livers than in control livers (Fig. 8C). Concurrently, hepcidin expression in the liver was significantly increased by LPS (Fig. 8D).
We also isolated hepatocytes and non-parenchymal cells from livers treated with or without LPS; cells of the hepatocyte fraction exclusively expressed albumin, a gene predominantly expressed in hepatocytes, but not stabilin-1 (endothelial cell marker) and  Nramp-1 (Kupffer cell marker), and those of non-parenchymal fraction expressed vice versa (supplemental Fig. S11). LPS greatly increased the expression level of IL-1␤ in non-parenchymal cells; LPS-induced up-regulation of IL-1␤ was also detected in hepato-cytes (Fig. 8E). In addition, expression of C/ebp␦ was increased by LPS in hepatocytes as well as non-parenchymal cells (Fig. 8F).
In RAW264.7 cells, a mouse macrophage-like cell line, LPS increased the expression of IL-1␤ within 2 h after treatment,

Up-regulation of hepcidin expression by IL-1␤
but the expression levels began to gradually decrease after 4 h (Fig. 9A). In fact, IL-1␤ protein was detected in culture supernatant from LPS-treated RAW264.7 cells but not from control RAW264.7 cells (Fig. 9B). Expression of IL-6 but not inhibin ␤B, a molecule consisting of activin B, was also increased by LPS in RAW264.7 cells (supplemental Fig. S12). In contrast, significant IL-1␤ induction was not detected in HepG2 cells in response to LPS treatment (data not shown). However, treatment with conditioned medium from LPS-treated RAW264.7 cells increased expression of IL-1␤ in primary mouse hepatocytes (Fig. 9C).
Conditioned medium from LPS-treated RAW264.7 cells potentiated the induction of C/EBP␦ (Fig. 10A) and hepcidin (Fig. 10B) gene expression; this activity was inhibited by BAY   11-7085 in HepG2 cells. Primary mouse hepatocytes also showed an increase in C/EBP␦ and hepcidin expression after treatment with conditioned medium from LPS-treated RAW264.7 cells (supplemental Fig. S13). Furthermore, either mutations of the C/EBP-BS on the hepcidin promoter or downregulation of C/EBP␦ expression by siRNA targeting C/EBP␦ decreased the ability of conditioned medium from LPS-treated RAW264.7 cells to induce efficient hepcidin transcription and expression (Fig. 10, C and D). Based on these data and the results of IL-1␤ induction during hepatic inflammation, the induced IL-1␤ expression in Kupffer cells and hepatocytes stimulated hepcidin transcription via C/EBP␦ production in an autocrine/paracrine manner.

IL-1␤ enhances hepcidin expression induced by activin B and IL-6
Various molecules are produced during inflammation; among them, activin B and IL-6 stimulate hepcidin transcription via BMP-REs and STAT-BS, respectively (14,32). We explored whether IL-1␤ enhances activin B-or IL-6-induced hepcidin transcription and expression (Fig. 11). IL-1␤ increased activin B-or IL-6-induced hepcidin expression and further enhanced hepcidin expression induced by co-treatment with activin B and IL-6 ( Fig. 11A). Similar results were also obtained by hepcidin transcription assays (Fig. 11B). These results indicate that molecules produced during hepatic inflammation could independently activate hepcidin transcription, leading to excessive hepcidin production.

Discussion
Hepcidin is a liver-derived hormone that regulates plasma iron levels, and aberrant hepcidin expression leads to a severe disturbance of the intestinal absorption of iron and iron release from macrophages, these events indicate the central role of hepcidin in homeostatic regulation of iron metabolism (11)(12)(13). Previous studies have extensively revealed that hepcidin expression is transcriptionally regulated via BMP-REs and STAT-BS on the hepcidin promoter (11)(12)(13). The present study reveals that: 1) IL-1␤ up-regulates hepcidin expression by stimulating transcription; 2) BMP-REs on the hepcidin promoter are involved in IL-1␤-induced hepcidin transcription, whereas the STAT-BS slightly participate in the transcriptional regulation resulting from stimulation of IL-6 production; 3) a C/EBP-BS spanning from nt Ϫ329 to Ϫ320 is essential for transcriptional activation; 4) IL-1␤ induces expression of C/EBP␦, which binds to the C/EBP-BS on the hepcidin promoter to activate transcription; 5) IL-1␤ is localized in Kupffer cells in basal murine livers, and LPS stimulation increased IL-1␤ expression in Kupffer cells as well as in hepatocytes; and 6) molecules produced during hepatic inflammation such as IL-1␤, activin B, and IL-6 cooperatively stimulate hepcidin expression through distinct transcriptional mechanisms. Our results shown here indicate that Kupffer cells sense a proinflammatory stimulus to accelerate IL-1␤ production, leading to hepcidin production through up-regulation of C/EBP␦ expression in hepatocytes. In addition, IL-1␤-induced IL-6 production slightly contributes to hepcidin transcription via STAT-BS (Fig. 12). The relay of the proinflammatory signal from Kupffer cells to hepatocytes via IL-1␤ leads to the excessive production of hepcidin.
Previously, it was shown that interferon (IFN) ␥ and Mycobacterium tuberculosis could increase hepcidin transcription in RAW264.7 cells via the putative NF-B-binding site spanning nt Ϫ556 to Ϫ547 on the hepcidin promoter (33). However, this region is not involved in IL-1␤-induced hepcidin transcription in hepatocytes, as deletion of this region did not affect hepcidin transcription induced by IL-1␤ (Fig. 4A). Although IL-6-stimulated hepcidin transcription has been well established (supplemental Figs. 8 and 11-13), hepcidin expression was not increased by IL-6 in macrophages (34,35). These results reveal the distinct regulatory mechanisms of hepcidin transcription between hepatocytes and macrophages, implying cell type-dependent regulation of hepcidin transcription; the relative importance of which region is most responsible for hepcidin transcription may be different between cell types. In fact, in alveolar macrophages, IL-1 did not induce hepcidin expression (34). Considering that hepcidin is predominantly expressed in hepatocytes (12,36), the results of this study clarify the primary regulatory system of hepcidin expression during inflammation.
A previous study has shown the involvement of C/EBP␣ in hepcidin expression: overexpression of C/EBP␣-stimulated hepcidin transcription in U-2 OS osteosarcoma cells (37). Bind- Figure 11. Enhancement of hepcidin expression and transcription by IL-1␤, activin B, and IL-6. A, HepG2 cells were treated with the indicated combination of IL-1␤ (10 ng/ml), activin B (ActB, 50 ng/ml), and IL-6 (10 ng/ml) for 12 h. Expression of hepcidin was examined by RT-qPCR analysis with the level in the control cells treated without ligand defined as 1. The data are presented as the mean Ϯ S.E. (n ϭ 3). **, p Ͻ 0.01 versus cells treated without IL-1␤, activin B, or IL-6. B, HepG2 cells were transfected with tk-Renilla-luc and the indicated reporters. At 4 h post-transfection, cells were treated with the indicated combination of IL-1␤ (10 ng/ml), activin B (ActB, 50 ng/ml), and IL-6 (10 ng/ml) for 12 h. The firefly luciferase activity normalized to Renilla luciferase activity was calculated, and the relative luciferase activity in cells transfected with hepcidin(Ϫ2018)-luc in the absence of activin B, IL-6, and IL-1␤ was defined as 1. The data are presented as the mean Ϯ S.E. (n ϭ 3).

Up-regulation of hepcidin expression by IL-1␤
ing of C/EBP␣ to the C/EBP-BS was verified in rat liver nuclear extracts, but the functional role of this region in hepcidin transcription was not determined. Furthermore, how this C/EBP␣ activity is regulated was unclear (37). Considering that C/EBP␣ expression was decreased in response to IL-1␤ (Fig. 5A), IL-1␤induced hepcidin transcription is unlikely to be mediated by C/EBP␣ in hepatocytes during inflammation.
Our results revealed that C/EBP␦ expression is up-regulated in response to LPS-induced IL-1␤ expression, which led to efficient binding of C/EBP␦ to the C/EBP-BS on the hepcidin promoter. C/EBP␦ has an intrinsic ability to bind to the C/EBP-BS (38). However, the activity of C/EBP␦ as a transcription factor is enhanced through post-translational modifications (39,40). As compared with the increase in C/EBP␦ expression in response to IL-1␤ treatment, more C/EBP␦ bound to the C/EBP-BS (Fig.  6E). Thus, IL-1␤ may exert activities not only to increase C/EBP␦ expression but also to promote C/EBP␦ activity.
Etiological studies have shown that increased concentrations of serum IL-1␤ were detected in patients with coronary artery disease, schizophrenia, insulin-dependent diabetes, and Alzheimer disease (41)(42)(43)(44). Patients with Alzheimer disease suffer from anemia with decreased plasma iron levels (45)(46). In addition, hepcidin has been hypothesized to be involved in dysfunctional iron metabolism in patients with Alzheimer disease (47). In various pathological conditions with increased IL-1␤ levels, IL-1␤-mediated hepcidin expression may partially contribute to aberrant iron metabolism.
IL-1␤ induction was detected not only in Kupffer cells but also in hepatocytes from LPS-treated mice. In fact, IL-1␤ expression was up-regulated in RAW264.7 cells treated with LPS (Fig. 9A). However, LPS did not induce IL-1␤ expression in HepG2 cells (data not shown). These results suggest that IL-1␤ induction in hepatocytes but not Kupffer cells during hepatic inflammation is indirect. Considering that LPS stimulates Kupffer cells as well as sinusoidal endothelial cells (14, 48 -50), various molecules secreted from the non-parenchymal cells are possibly responsible for IL-1␤ induction in hepatocytes. Transcription of IL-1␤ is stimulated by activation of NF-B (51); in fact, IL-1␤ transcription was stimulated by IL-1 in U937 myeloid cells in an autoregulatory manner (52). However, we could not detect significant induction of IL-1␤ in response to IL-1␤ in HepG2 cells (data not shown). Future studies should clarify the regulation of IL-1␤ expression in hepatocytes at the molecular level.
Previous studies revealed that activin B and IL-6 production are increased in the liver during inflammation and that these cytokines stimulate hepcidin transcription via BMP-REs and STAT-BS, respectively, in hepatocytes (14,24,32). Activin B and IL-6 independently increased hepcidin expression, and cotreatment with activin B and IL-6 further enhanced hepcidin expression (53). The present study expands the available information on regulating hepcidin expression in hepatocytes during inflammation, as inflammation-induced IL-1␤ production leads to the stimulation of hepcidin transcription in hepatocytes via the C/EBP-BS on the hepcidin promoter. The concurrent stimulation of the three cis-elements cooperatively enhanced hepcidin transcription and expression in hepatocytes compared with the stimulation of each individual element (Fig.  11); the molecules induced during inflammation possibly resulted in increased hepcidin expression in hepatocytes. Considering that hepcidin was originally identified as an antimicrobial peptide (54 -56), enhanced hepcidin production via these three elements may be helpful to exclude pathogens but could potentially promote anemia of inflammation through overproduction of hepcidin.

Materials and methods
The following reagents were purchased: recombinant human IL-1␤ was from RayBiotech, Inc. (Norcross, GA); recombinant mouse IL-1␤ and rat IL-1␤ were from Bioworld Technology (Louis Park, MN); recombinant human IL-6, recombinant activin B, and goat polyclonal antibody against IL-1␤ (AF-401-NA) was from R&D Systems (Minneapolis, MN); recombinant human BMP2 was from PeproTech (Rocky Hill, NJ); cycloheximide and control mouse IgG were from Sigma; LDN-193189 was from Stemgent (San Diego, CA); BAY 11-7085 was from Cayman Chemical (Ann Arbor, MI); rabbit polyclonal antibody against phospho-Smad1 (Ser 463 /Ser 465 )/Smad5 (Ser 463 / Ser 465 )/Smad8 (Ser 426 /Ser 428 ), and mouse monoclonal antibody against phospho-STAT3 (Tyr 705 ) (3E2) were from Cell Signaling Technology (Danvers, MA); rabbit polyclonal antibody against human C/EBP␦ that cross-reacts with mouse C/EBP␦ and was used in immunohistochemical analysis, mouse monoclonal antibody against ␤-actin (AC-15), and rat monoclonal antibody against F4/80 (CI:A3-1) were from Abcam (Cambridge, MA); rabbit polyclonal antibody against C/EBP␦ (M-17) that was used in Western blot analysis was from Santa Cruz Biotechnology (Santa Cruz, CA); Alexa 488 donkey antigoat IgG antibody and Alexa 594 donkey anti-rat IgG antibody were from Thermo Fisher Scientific (Waltham, MA). IL-1␤ is expressed in Kupffer cells and hepatocytes in response to hepatic inflammation. This induced IL-1␤ stimulates the expression of C/EBP␦ and IL-6; the induced C/EBP␦ enhances hepcidin transcription via the C/EBP-BS on the hepcidin promoter spanning nt Ϫ329 to Ϫ320, and the induced IL-6 stimulates STAT3 phosphorylation and hepcidin transcription via the STAT-BS spanning nt Ϫ143 to Ϫ134 slightly. Note the species difference on IL-1␤-induced hepcidin expression between mouse hepatocytes and rat hepatocytes, resulting from difference of the nucleotide sequence of C/EBP-BS.

Cell isolation and cell culture
All procedures for animal use were approved by the Kyoto University Animal Experiment Committee. Primary hepatocytes from the livers of 4-week-old male Sprague-Dawley rats were collected as previously described (57). Primary hepatocytes were also recovered from 5-8-week-old male ICR mice by a similar procedure to isolate primary rat hepatocytes. Isolated hepatocytes were plated in 12-well collagen-coated plates at 1.5 ϫ 10 5 cells per well and cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS), insulin, dexamethasone, and antibiotics. Adherent cells were immediately used. Non-parenchymal cells were isolated as supernatant fraction to recover hepatocyte fraction as cell pellet of liver digested with collagenase after 50 ϫ g for 3 min. Subsequently, non-parenchymal cells were washed with HBSS and pelleted at 800 ϫ g for 10 min at 4°C. Furthermore, non-parenchymal cells resuspended in HBSS were layered onto a 2-step Percoll gradient (25% Percoll layer and 50% Percoll layer, respectively), followed by centrifugation at 800 ϫ g for 30 min to purify further. Non-parenchymal cells reside in the 25% Percoll layer as well as the 50% Percoll layer; interface of HBSS and 25% Percoll contains cell debris, red blood cells are at the bottom of the 50% Percoll layer. After recovery of non-parenchymal cells, the cells were pelleted at 800 ϫ g for 10 min at 4°C. HepG2 human hepatoma cells and RAW264.7 mouse macrophage-like cells were cultured in DMEM supplemented with 10% heat-inactivated FBS and antibiotics.

Preparation of conditioned medium from RAW264.7 cells
RAW264.7 cells were treated with or without LPS (100 ng/ml) for 30 h in serum-free DMEM. The conditioned medium of LPS-treated cells (CM-LPS) and control cells (CM-C) were concentrated by Centriprep-10 (Merck, Darmstadt, Germany), and the solvent was replaced with HEPES buffer (21 mM HEPES, pH 7.5, 0.7 mM Na 2 HPO 4 , 137 mM NaCl, 5 mM KCl, 6 mM dextrose). HepG2 cells were treated with the conditioned medium; concentrations of CM-C and CM-LPS were equivalent to the conditioned medium of RAW264.7 cells.

siRNA transfection
HepG2 cells (3 ϫ 10 4 cells per well) were seeded onto 24-well plates. Cells were reverse-transfected with 2 l of Lipofectamine RNAi Max (Invitrogen) and 50 pmol of siRNA. The nucleotide sequence of the double-stranded siRNA is shown in supplemental Table S1. At 48 h after seeding, cells were serumstarved with medium containing 0.2% FBS for 4 h followed by treatment with IL-1␤ (25 ng/ml) for 12 h.

RNA isolation and RT quantitative PCR
Total RNA isolation, cDNA synthesis, and real-time quantitative PCR (qPCR) were performed as previously described (57). The sequences of the oligonucleotide primers are shown in supplemental Table S2. The ⌬⌬C t method was used to normalize the levels of the target transcripts to the TBP levels (58).

Western blot analyses
Western blot analyses were performed as previously described (59). The immunoreactive proteins were visualized using the ECL Select Western blotting detection system (GE Healthcare, Buckinghamshire, UK) according to the manufacturer's protocol.

Plasmids and luciferase reporter assay
Constitutively active ALK3 (ALK3(QD)) (25) was kindly provided by Dr. K. Miyazono. A mouse hepcidin promoter fragment (nt Ϫ2018 to Ϫ35) or rat hepcidin promoter fragment (nt Ϫ1861 to Ϫ35) was inserted into the luciferase reporter vector pGL4 (mhepcidin-luc or rhepcidin-luc, respectively). In addition, a mouse IL-6 promoter fragment (nt Ϫ300 to Ϫ79) was inserted into pGL4. The translation initiation site is numbered as ϩ1. Mutations were prepared by PCR-based methods. The nucleotide sequences of the reporter constructs were verified by DNA sequencing. HepG2 cells (6 ϫ 10 4 cells per well) were seeded onto 24-well plates. The next day, 0.5 g of a pGL4based hepcidin reporter and either 0.5 g of Renilla luciferase expression vector under the control of a thymidine kinase promoter (tk-Renilla-luc) or 0.1 g of a ␤-galactosidase expression plasmid under control of a cytomegalovirus-derived promoter (CMV-␤Gal) were transfected into cells in 0.2% FBS medium using polyethylenimine Max reagent (Polysciences, Warrington, PA). After 4 h, cells were stimulated with ligands or the culture supernatant from RAW264.7 cells. Firefly luciferase activity was normalized to either Renilla luciferase activity or ␤-galactosidase activity as appropriate.

Oligo DNA pulldown assay
HepG2 cells were scraped from the plates and centrifuged at 1500 rpm. The cell pellets were resuspended in lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% (w/v) Triton X-100, 1 mM PMSF, 1% (v/v) aprotinin, 1 mM Na 3 VO 4 ), vortexed, incubated on ice for 15 min, and centrifuged to remove cell debris. The supernatants were treated with 50 pmol of 5Ј-biotinylated probe with or without 500 pmol of unlabeled probe for 12 h at 4°C followed by an incubation with 50 l of 25% (v/v) streptavidin-agarose beads for 1 h at 4°C. Subsequently, the beads were washed with lysis buffer three times, and the proteins were eluted into 6ϫ SDS-PAGE sample buffer. C/EBP␦ binding was analyzed by Western blotting. The probe was prepared from the following oligonucleotides: 5Ј-catcgtgatggggaaagggctcccc-3Ј (forward, 5Ј-biotinylated) and 5Ј-atctggggagccctttccccatcac-3Ј (reverse). The probe included the C/EBP-BS from the human hepcidin promoter.

Statistical analysis
The data are expressed as the mean Ϯ S.E. The data regarding gene expression were log-transformed to provide an approximation of a normal distribution before analysis. Differences in the gene expression among the cells were examined using unpaired t-tests. Differences of p Ͻ 0.05 were considered significant.