Cell iron status influences macrophage polarization

Macrophages play crucial roles in innate immune response and in the priming of adaptive immunity, and are characterized by their phenotypic heterogeneity and plasticity. Reprogramming intracellular metabolism in response to microenvironmental signals is required for M1/M2 macrophage polarization and function. Here we assessed the influence of iron on the polarization of the immune response in vivo and in vitro. Iron-enriched diet increased M2 marker Arg1 and Ym1 expression in liver and peritoneal macrophages, while iron deficiency decreased Arg1 expression. Under LPS-induced inflammatory conditions, low iron diet exacerbated the proinflammatory response, while the IL-12/IL-10 balance decreased with iron-rich diet, thus polarizing toward type 2 response. Indeed, in vitro macrophage iron loading reduced the basal percentage of cells expressing M1 co-stimulatory CD86 and MHC-II molecules. Further, iron loading of macrophages prevented the pro-inflammatory response induced by LPS through reduction of NF-κB p65 nuclear translocation with decreased iNOS, IL-1β, IL-6, IL-12 and TNFα expression. The increase of intracellular iron also reduced LPS-induced hepcidin gene expression and abolished ferroportin down-regulation in macrophages, in line with macrophage polarization. Thus, iron modulates the inflammatory response outcome, as elevated iron levels increased M2 phenotype and negatively regulated M1 proinflammatory LPS-induced response.


Introduction
Macrophages play a central role in both immunity and tissue homeostasis, protecting the organism from infection as well as completing essential tissue-specific functions. Indeed, macrophage immune functions include pathogen recognition, antigen processing, inflammation, phagocytic clearance, immune regulation as well as resolution of immune responses and tissue repair [1]. Macrophages also play an essential role in iron homeostasis by recycling iron, a crucial transition metal for living organisms. Indeed, in mammals, most of the iron used for hemoglobin production is recycled after phagocytosis of senescent red blood cells by macrophages, with a limited intake of dietary iron absorbed through enterocytes. Clearly, interactions between iron homeostasis and immune functions of macrophages have been reported. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Systemic iron balance is maintained via hepcidin hormone and ferroportin cellular iron exporter [2][3][4][5][6] and macrophage polarization may affect their expression [7][8][9]. Indeed, hepcidin 1 upregulation, controlled at the transcriptional level, is exacerbated by M1 polarization while ferroportin transcription is influenced by M2 polarisation [10,11]. Conversely, several publications indicated a change in inflammatory response according to the iron status, but conflicting results have been reported [12][13][14][15]. Here, we questioned the influence of iron on macrophage polarization.
The couple hepcidin and ferroportin regulate systemic iron balance [2][3][4][5][6]. Hepcidin is mainly synthesized from hepatocytes and upregulated during iron overload through Bmp6-Smad1/5/8/4 pathway activation. Hepcidin is released in blood circulation and binds to the cellular iron exporter ferroportin, leading to its internalization and degradation [2,6,36,37]. Thus, the systemic signal given by hepcidin decreases the transfer of iron from the enterocytes to the rest of the body by ferroportin and iron recycling from macrophages. Splenic macrophages and liver Kupffer cells contribute to erythrophagocytosis and influence the daily turn-over of iron by increasing ferroportin mRNA transcription [38,39]. In isolated macrophages, while hepcidin 1 gene is not upregulated by increased iron levels, the expression of ferroportin mRNA is upregulated upon iron cell loading and can be downregulated after addition of exogenous hepcidin [40,41]. The expression of hepcidin 1 and ferroportin genes is also altered upon inflammation. Liver and macrophage hepcidin 1 expression is upregulated through Smad1/5/8/4 and Stat3 pathways in response to inflammation, while ferroportin transcription is downregulated at the transcriptional level, independently of hepcidin, both leading to hypoferremia [42][43][44][45][46][47].
Here we addressed the influence of iron on innate immune phenotype and macrophage polarization in vivo and in vitro, and we investigate the cross-talk between inflammation and iron balance regulation. We show that, while iron depletion reduces M2 markers, an iron-rich status promotes M2-like phenotype and reduces type 1 immune response markers, impairing LPS-induced pro-inflammatory cytokine expression, and this is accompanied by a differential regulation of hepcidin 1 and ferroportin expression under inflammatory conditions.

Ethics statement
All animal experimental protocols complied with the French ethical and animal experiments regulations (see Charte Nationale, Code Rural R 214-122, 214-124 and European Union Directive 86/609/EEC) and were approved by the "Ethics Committee for Animal Experimentation of CNRS Campus Orleans" (CCO), registered (N˚3) by the French National Committee of Ethical Reflexion for Animal Experimentation, under N˚CLE CCO 2015-1085.

In vivo experiments
C57BL/6 wild-type mice (Janvier Labs, Le Genest-Saint-Isle, France) housed in the Transgenose Institute animal facility (TAAM, CNRS UPS44, Orleans) were fed with either iron-rich diet (25000 mg of iron carbonyl/kg) for 3 days, iron-deficient diet (less than 20 mg iron carbonyl/kg) for 14 days, or iron-replete diet as control diet (280 mg iron carbonyl/kg, R03 from Scientific Animal Food & Engineering; Augy, France). For some experiments mice received 0.2 g/kg of iron-dextran by peritoneal injection 48 hours before necropsy. Mice were observed daily and monitored for signs of distress including ruffled fur, immobility or hunching, and body weight loss. Acute inflammation was induced by a single intraperitoneal injection of lipopolysaccharide (LPS E. coli 055:B5, Sigma-Aldrich, St Louis, MO, USA; 50 μg/kg), and equivalent volume of sterile saline solution (0.9% NaCl) was used as vehicle control. After 4 hours, mice were bled under isoflurane anesthesia, then killed with carbon dioxide inhalation, liver and spleen removed, and peritoneal exudate cells harvested as a source of peritoneal macrophages.

Measurement of hepatic iron content
Hepatic iron content (HIC) was measured from dried liver samples dissolved in nitric acid:sulfuric acid (1:1) at 90˚C, then after adding 1 mL of H 2 O 2 , an equivalent volume of sample and 10 mM ferrozine / 32.6 mM ascorbic acid / 50 mM Tris HCl pH4 was incubated 30 min before reading at 560 nm. The values were expressed as ng of iron per mg of dry tissue.

Hematology and blood markers
Blood was obtained from retro-orbital plexus under isoflurane anesthesia. Serum iron and unbound iron binding capacity was measured by using ferrozine as described by Goodwin et al. [48]. Transferrin saturation was calculated as follows: serum iron/ (serum iron + unbound iron binding capacity) × 100%. Red blood cell count (RBC), mean corpuscular volume (MCV), hematocrit (Ht), hemoglobin (Hb), were measured from EDTA-plasma with MS9-5V Melet schloesing apparatus.

Nitric oxide production
The production of nitric oxide (NO) was measured in culture supernatants by colorimetric Griess assay for nitrites/nitrates [50]. An equal volume of culture supernatant sample and Griess reagent 0.5% sulfanilamide / 0.05% N-1-napthylethylenediamine dihydrochloride / 2.5% phosphoric acid) (all chemicals were from Sigma-Aldrich) was mixed and incubated for 30 min. A standard curve was established with NaNO 2 and absorbance was measured at 562 nm.

Flow cytometry analysis
Macrophages were plated in Petri dish at 5×10 6 cells per dish. Cells were then stimulated with FAC, IFN-γ or IL-4 as indicated. Cells were washed and stained with 7-aminoactinomycin D

Quantitative PCR analysis
Total RNA from liver and spleen homogenates or cells was isolated using TRI Reagent (Sigma-Aldrich) and reverse transcription was performed with Promega synthesis system (Promega, Madison, WI, USA) according to the manufacturer's instructions. Quantitative PCR was performed with QuantiTect SYBR Green PCR system (Qiagen, Hilden, Germany) on Stratagene Mx3005P (Agilent Technologies, Santa Clara, CA, USA) by using hepcidin 1 forward and reverse primers: 5'-CCTATCTCCATCAACAGATG-3' and 5'-TGCAACAGATACCACAC TG-3', ferroportin forward and reverse primers: 5'-CTCTGTCAGCCTGCTGTTTG-3' and 5'-TCAGGATTTGGGGCCAAGATG -3', other primers were Quantitect primers (Qiagen, Hilden, Germany). The mRNA levels were normalized to the housekeeping gene Gapdh in the same RT sample and relative transcript expression of genes was given as ΔCt = C t target −C t Gapdh . Fold change of genes expression compared with untreated mice was determined as 2 −ΔΔCt values (ΔΔCt = ΔC t treated −ΔC t control ).

Statistical analysis
Student's t-test and one-way Anova test were used to determine the significance of the data by using GraphPrism software (version 5.04 for Windows, GraphPad Software, La Jolla, CA, USA). Significant levels were indicated as Ã , p < 0.05, ÃÃ , p < 0.01, ÃÃÃ , p < 0.001.

Host iron status controls M2 marker expression in liver and peritoneal macrophages
Adult wild type C57BL/6 mice received specific iron-containing diet to induce either dietary iron loading or dietary iron deficiency. These diets had little effect on the overall body weight and relative liver weights (Fig 1A). Liver gene expression of hepcidin anti-microbial peptide (Hamp1) and bone morphogenic protein (Bmp6), the key iron-responsive inducer of Hamp1 transcription, were then assessed to confirm the change of iron metabolism parameters induced by the specific diets. As expected, Hamp1 and Bmp6 mRNA expression in liver homogenates was increased with iron-rich diet and decreased with iron-deficient diet (Fig 1B), confirming their value as biomarkers of iron status. We next determined whether the modification of dietary iron and liver iron genes level was correlated with markers of immune  homogenates (B, C), and peritoneal macrophages (D) gene expression was then analyzed using quantitative PCR analysis. Data are expressed as mRNA fold change relative to control mice fed with IR diet. Data are from two independent experiments and presented as mean ± SD (n = 6 mice per group). ns, non significant; Ã p 0.05; ÃÃ p 0.01; ÃÃÃ p 0.001. response. Interestingly, in liver the expression of type 2 immune response markers Arginase 1 (Arg1) and Chitinase-like 3 (Ym1) was increased after high iron diet correlating with a trend of increase expression of IL-10 anti-inflammatory cytokine ( Fig 1C). Conversely, after low iron diet liver Arg1 was reduced. Since hepatocytes express Arg1 [51] we further assessed the impact of iron status on macrophages in the peritoneum compartment. Peritoneal macrophages were isolated from mice receiving iron-rich or iron-deficient diet. Indeed, the gene expression of Arg1, Ym1, IL-10 and Stat6, a series of genes expressed in M2 polarized macrophages, was increased after high iron diet condition, and, conversely, the expression of Arg1 and IL-10 was significantly decreased after dietary iron deficiency (Fig 1D). Thus, in vivo dietary-induced modifications of iron levels influence the expression of type 2 immune response markers.
We documented the iron loading induced by feeding with a high iron-containing diet (25 g/kg) for 3 days and compared it with a systemic injection of iron-dextran (0.2 g/kg). Although overall body weight were barely affected liver relative weight, and furthermore spleen relative weight were increased after iron-dextran treatment (Fig 2A). Iron-dextran intraperitoneal injection resulted at 48h in high iron deposition in liver, including hepatocytes and macrophages, in spleen red pulp, and in cells from peritoneal exudates (Fig 2B), associated with 120-fold increased hepatic iron concentration (HIC) and elevated transferrin saturation ( Table 1). The digestive route of iron loading resulted mainly in iron deposition in periportal hepatocytes (Fig 2B), and was associated with a more moderate increase in hepatic iron content and transferrin saturation (Table 1), as compared to control mice fed with iron-replete diet (IR). Spleen iron staining was equivalent for both conditions, with no iron staining of peritoneal cells (Fig 2B). The level of hepcidin 1 gene upregulation was similarly increased in the liver of mice fed with iron-rich diet or injected with iron-dextran. Hepcidin 1 gene expression was not altered in spleen and peritoneal cells in response to either iron treatments (Fig 2C). The expression of ferritin L gene, encoding the predominant subunit of ferritin in liver and spleen [52], was increased in liver under both iron overload conditions, while it was strongly enhanced in spleen and peritoneal cells after iron-dextran systemic injection, but was not affected after iron rich diet regimen (Fig 2C). The transcription of ferritin L gene is induced in response to oxidative challenge [53,54] independently of iron level. Here, ferritin L gene overexpression was indicative of an increased ROS formation in correlation with tissue iron level. In blood, transferrin saturation percentage was increased under both iron overload conditions, whereas serum iron concentration was highly increased after iron-dextran injection and less so after iron-rich diet, as compared to mice fed with iron-replete diet (Table 1). Erythrocyte markers, including red blood cell counts, mean corpuscular volume, and hemoglobin were not affected by either iron overload conditions, but hematocrit tended to increase in iron loaded mice (Table 1). Thus, both iron-rich diet and iron-dextran injection led to liver iron overload and hepcidin 1 upregulation, while iron-dextran also lead to systemic elevated iron deposits in spleen and peritoneal cells.
We next further characterized how iron overloading influenced markers of immune response. Hence, the gene expression of a series of markers expressed in M2 (Arginase-1, Chitinase-like 3, mannose receptor) or M1 (iNos) polarized macrophages was assessed in liver, spleen and peritoneal macrophages. Interestingly, in mice fed with high iron diet the expression of type 2 immune response markers Arginase-1 (Arg1) and Chitinase-like 3 (Ym1) was increased in liver (Fig 3A), spleen ( Fig 3B) and peritoneal macrophages (Fig 3C), yet Mrc was not affected and iNos expression did not change significantly after iron rich diet in liver and spleen, but increased in peritoneal macrophages (Fig 3A). In mice injected with iron dextran, Arg1, Ym1, mannose receptor (CD206) as well iNos expression was enhanced in liver homogenate, spleen and peritoneal macrophages (Fig 3A, 3B and 3C). Thus, in vivo high iron levels influence the expression of both type 1 and type 2 immune response markers.  1 (Hamp1) and ferritin L (Ftl) gene expression in liver tissue homogenates, spleen and peritoneal cells was analyzed using quantitative PCR. Data are expressed as mRNA fold change relative to control mice fed with IR diet. Data are from two independent experiments and presented as mean ± SD (n = 6 mice per group). ns, non significant; Ã p 0.05; ÃÃ p 0.01; ÃÃÃ p 0.001.

Iron status influences the IL-12/IL-10 axis balance during endotoxininduced inflammation in peritoneal macrophages
Our results suggested that in vivo diet-induced iron status impacted immune polarization. We next hypothesized that changes of iron status could alter the inflammation process in macrophages. We thus assessed the effect of iron diet on lipopolysaccharide (LPS)-induced proinflammatory response in peritoneal macrophages. Adult C57BL/6 mice fed with high or low iron diet were treated with a sublethal concentration of LPS (50 μg/kg) or vehicle for 4 hours and cytokine gene expression in peritoneal cells was determined. In naïve cells from mice fed with either high or low iron diet Il-12 and Il-1β gene expression were not affected as compared to iron replete, while Il-10 was influenced by the iron diet (Fig 4A). In response to LPS treatment, the expression of Il-12, Il-10 and Il-1β was clearly modulated by the iron status (Fig 4A). Iron deficiency promoted LPS-induced Il-12 gene expression and decreased Il-10 gene expression in response to LPS, while Il-1β was significantly reduced after iron supplementation ( Fig  4A). In vivo extreme iron loading after iron-dextran injection yielded a decrease in Il-1β and Il-10 ( Fig 4B). Thus, in vivo iron status modulates the immune responses, dietary iron deficiency exacerbated pro-inflammatory effects of endotoxin whereas dietary iron loading decreased some inflammatory responses such as Il-1β expression.

Macrophages iron loading impairs IFN-γ-induced M1 polarization
To further assess the influence of iron loading on macrophage polarization, we next determined the profile of macrophages in response to cellular iron loading by using bone marrow derived macrophages incubated with ferric ammonium citrate (FAC). We first verified cell viability in response to iron loading, and showed that the FAC concentrations of 50-100 μM used in this study did not influence cell viability (Fig 5A). We then confirmed that increasing the concentration of ferric ammonium citrate in culture medium correlated with increased intracellular iron using calcein fluorescence quenching method (Fig 5B). Further, the increase of intracellular reactive oxygen species was determined using the general oxidative stress indicator 2',7'-dichlorodihydrofluorescein diacetate (H2DCFDA). The increase in intracellular iron was correlated with an increase of reactive oxygen species, which was significantly induced Table 1

. For each group of mice hepatic iron content and iron blood parameters, including serum iron and percentage of transferrin saturation, as well as red blood cell counts (RBC), mean corpuscular volume (MCV), hematocrit (Ht) and hemoglobin (Hb) are indicated.
Data are from two independent experiments and presented as mean ± SD (n = 6 mice per group).  after 16 hours of incubation with ferric ammonium citrate (Fig 5B). Thereafter, we determined macrophage polarization based on cell surface expression of the CD86 co-stimulatory molecules and I-A/I-E MHC class II molecules, versus CD206 mannose receptor, as markers highly expressed on M1 and M2 polarized macrophages, respectively. BMDM were gated on 7-AAD -, F4/80 + cells and analysed by flow cytometry for the membrane expression of CD86, I-A/I-E, and CD206. Interestingly, 24 hours of incubation with ferric ammonium citrate alone decreased CD86 + and I-A/I-E + cells within the F4/80 + population (Fig 5C), suggesting that cell iron loading decreased the basal constitutive expression of these molecules. IFN-γ-induced Iron modulates macrophage polarization CD86 and I-A/I-E over-expression was impaired by ferric ammonium citrate pretreatment and this was dose-dependent. The decrease of CD86 and I-A/I-E expression induced by iron loading was accentuated by IL-4 ( Fig 5C). In contrast, the expression of CD206, increased after incubation with IL-4, as expected, was not further increased by iron loading. Thus, cell iron loading affects the expression of membrane costimulatory molecules associated with macrophage polarization, higher iron levels notably reducing M1 marker expression.

Alteration of endotoxin-induced pro-inflammatory response by cell iron overload
To further address the anti-inflammatory role of cell iron loading, we next assessed the functional consequences of iron supplementation on LPS-induced inflammation response in macrophages. BMDM were treated with ferric ammonium citrate overnight followed by TLR4 stimulation with LPS during 4 h for gene expression and 24 h for cytokine protein release and mediator analysis. The mRNA overexpression of Il-12b, Il-1β, Il-6 and Tnfα genes induced by LPS (Fig 6A) was drastically reduced in BMDM pretreated with ferric ammonium citrate and this correlated with a reduction in the respective pro-inflammatory cytokines Il-12p40, Il-1β, Il-6 and Tnfα release (Fig 6B). In addition, the expression of iNos, a marker of M1 macrophages and its chemical product nitric oxide (NO) in response to LPS were also strongly impaired by iron loading (Fig 6A and 6B). Early activation of TLR4 pathway leads to M1 polarization of macrophages characterized by the activation of NF-κB p65/p50 heterodimers, leading to the expression of pro-inflammatory cytokines and nitric oxide production. To assess the role of NF-κB p65/p50 in the modulation of the immune response by iron loading we evaluated the activation NF-κB p65. Macrophages were treated with ferric ammonium citrate overnight followed by TLR4 stimulation with LPS during 30 min to analyse NF-κB p65 nuclear translocation. As shown in Fig 6C, neither medium nor ferric ammonium citrate incubation allowed NF-κB p65 nuclear translocation, whereas LPS treatment on macrophages led to translocation of NF-κB p65. Interestingly, pretreatment of macrophages with 100 μM of ferric ammonium citrate impaired the LPS-induced nuclear translocation of NF-κB p65 (%21%). Thus, in vitro cellular iron loading reduced NF-κB p65 nuclear translocation, iNos and pro-inflammatory cytokines expression, indicating that iron overload directly prevents macrophage proinflammatory response to endotoxin.

Intracellular iron load differentially modulates macrophage hepcidin and ferroportin expression in response to LPS
Iron-loading of macrophages impairs pro-inflammatory gene expression associated with a decrease of the expression of CD86 and I-A/I-E cell surface molecules associated with M1 polarization. Further, hepcidin 1 upregulation and/or ferroportin downregulation have been associated with different inflammatory stimuli such as TLR agonists or cytokines [45,47,10]. Thus, we next analyzed how iron cellular loading might influence the expression of hepcidin 1 and ferroportin in macrophages. As shown in Fig 7, intracellular iron loading decreased LPSinduced hepcidin 1 mRNA expression dose-dependently while iron treatment alone had no effect. The ferroportin expression was increased in response to iron, and strongly downregulated after LPS exposure in macrophages. Interestingly, the LPS-induced downregulation of ferroportin mRNA in macrophages was abrogated by ferric ammonium pretreatment (Fig 7).

Fig 6. Iron loading of primary macrophages impairs LPS-induced pro-inflammatory responses.
Bone marrow derived-macrophages were preincubated with ferric ammonium citrate (FAC, 0, 50 or 100 μM) overnight followed by LPS stimulation (100 ng/mL) for 4 h or 24 h to assess Il-12b, Il-1β, Il-6, Tnfα and iNos mRNA gene expression by quantitative PCR analysis (A) or cytokine Il-12p40, Il-1β, Il-6, Tnfα and nitric oxide measurement in cell medium by ELISA and Griess assay (B), respectively. Bone marrow derived-macrophages were preincubated with ferric ammonium citrate (FAC 100 μM) or medium overnight followed by LPS stimulation (100 ng/mL) for 30 min to assess NF-κB nuclear translocation by immunofluorescence, with NF-κB p65 in green and DAPI in red (C). Data of mRNA gene expression are given as fold change gene expression relative to the expression in untreated cells. Data are representative of at least two independent experiments and presented as mean ± SD (n = 4). Ã p 0.05; ÃÃ p 0.01; ÃÃÃ p 0.001.
Conversely, LPS stimulation prevented the upregulation of ferroportin in response to iron. Thus, the impaired expression of proinflammatory genes, cytokines and costimulatory receptors in response to LPS after macrophage iron overloading was associated with a reduced expression of hepcidin 1 gene and restoration of ferroportin gene expression.

Discussion
Our data show that increased cell iron loading triggers the expression of monocyte polarization markers of M2-like phenotype in resting macrophages and dampens pro-inflammatory immune responses, while iron deficiency has the opposite effect. Indeed, in vivo increased iron status leading to increase transferrin saturation and tissue iron deposition, stimulated expression of M2-associated markers and cytokines, such as arginase-1, Ym1 chitinase-like and mannose receptor. Modifications in iron balance also altered cellular inflammatory responses, as increased iron levels repressed some type 1 immune responses to TLR4 stimulation, such as iNos and Il-12 gene expression. Indeed, after TLR4 stimulation iron-loading of primary bone marrow derived macrophages repressed the expression of M1-associated markers, decreased NF-κB p65 nuclear translocation, pro-inflammatory cytokines release and NO formation, while promoting M2-associated markers. Conversely, iron deficiency promoted type 1 proinflammatory cytokine expression in response to TLR4 stimulation. Therefore, changes in cell iron concentration can modulate macrophage phenotype and function with clear implications for the immune responses.
The effect of iron status on macrophage is consistent with previous reports of decreased CD86 and I-A/I-E expression in bone marrow derived macrophages in response to the iron chelator deferoxamine in the absence of an inflammatory context [11], or impaired iNos expression and NO release, M1 associated markers, in IFNγ treated RAW264.7 monocytic cell Iron modulates macrophage polarization line after iron loading [55]. Interestingly, intracellular iron dysbalance disrupted the L-arginine metabolism as a precursor for iNos and arginase-1, M1 and M2 activity, respectively [16,21,56]. We showed that cellular iron status influences macrophage metabolic programs, which are a key component of macrophage plasticity and polarization, instrumental to their function in homeostasis, immunity, and inflammation [16,17,22,23,26,30]. We also document that changes in cell iron status interfere with the inflammation process by impairing LPS-induced pro-inflammatory response. The IL-12/IL-10 balance is one of the most striking features of macrophage polarization. Activated macrophages M1 and M2 produce IL-12 hi IL-10 lo and IL-12 lo IL-10 hi , respectively, which are cytokines involved in their pro-inflammatory and anti-inflammatory functions [1,16,17]. Indeed, IL-12/IL-10 ratio in response to LPS was significantly increased under iron deficiency, suggesting a pro-inflammatory role of iron deficiency in macrophages. Conversely, iron overload tended to decrease the IL-12/IL-10 balance as well as other pro-inflammatory cytokine response. Furthermore, we showed that NF-κB p65 translocation into the nucleus was reduced in our model of iron loaded macrophages, which can account for the reduced pro-inflammatory cytokine expression. NF-kB is a redoxsensitive transcription factor, and ROS are clearly implicated in its activation [57][58][59]. Here, the decrease in NF-κB p65 nuclear translocation after LPS stimulation was accompanied by an increase in ROS production. High ROS production by phagocytes mediates bacteria killing, but it has been previously shown that non-pathogenic levels of ROS, such as induced by certain commensal bacteria via TLRs or by H 2 O 2 treatment, had suppressive activity on host inflammatory pathway and dampened the innate immune response [60,61]. Indeed, ROS trigger inactivation of redox sensitive ubiquitin ligase involved in NF-κB inhibitor IκB-α degradation, reducing nuclear translocation of the p65 subunit of NF-κB [60,61]. The decrease in NF-κB p65 nuclear translocation evidenced in iron loaded macrophages may be a consequence of low levels of ROS formation, and result in the suppression of NF-κB pathway activity and the alteration of pro-inflammatory genes expression.
Changes in inflammatory responses according to iron status were reported, although with different results [12][13][14][15]. We showed a reduced pro-inflammatory response to LPS in mice with iron overload induced by either iron-rich diet or intraperitoneal iron-dextran. These results corroborated a previous study showing that in vivo iron deprivation exacerbated the production of pro-inflammatory cytokines TNFα and IL-6 in liver, spleen macrophages and serum in LPS treated mice and this was reversed after hepcidin injection [12]. Other reports revealed an exacerbation of pro-inflammatory response associated with high iron status. Indeed, a model of high dose iron-dextran intraperitoneal injection (1 g/kg) exacerbated the pro-inflammatory response to a high LPS endotoxin dose (5 mg/kg) [13]. This was associated with elevated serum iron, but no data on tissue iron load was reported. In our hands, very high iron overload after iron-dextran injection (0.5 g/kg on day 8 and 4 before necropsy) and high LPS (10 mg/kg) increased LPS-induced iNos and pro-inflammatory cytokines expression S1 Fig). Indeed, high iron content due to erythrophagocytosis / extravasion of erythrocytes in a model chronic venous leg ulcers, or after repeated massive iron-dextran overload (0.25 g/kg every 3 days for 21 days), induced a proinflammatory M1 macrophage population [14]. Further, local iron oxide nanoparticles treatment or systemic ferumoxytol supplementation suppressed tumor growth by inducing pro-inflammatory macrophage responses with strong oxidative bursts [15]. Thus, under these conditions, the high iron tissue deposition associated with high macrophage iron levels was pro-inflammatory. In contrast, our studies show that under dietary iron supplementation, moderate iron overload is rather dampening inflammation. Thus, depending on the level of iron overload, through systemic or enteric route, the effect on inflammation might be pro-inflammatory at extremely high doses but rather antiinflammatory at more moderate iron overload doses. In conclusion, we show here that under physiological iron loading conditions, iron-enriched diet modifies the expression of markers of macrophage polarization and influences the inflammatory responses.
Supporting information S1 Fig. Expression of pro-inflammatory markers in liver upon high iron dextran overload. C57BL/6 wild type mice were fed with iron replete diet (C) or received an injection of irondextran (Fe-dextran) 0.5 g/kg on day 8 and 4 before necropsy or dextran as vehicle (C). Mice were then treated with LPS (i.p 10 mg/kg) or NaCl 0.9% as vehicle. After 4 h liver homogenates gene expression of pro-inflammatory cytokines (Il-6, Tnfα, Il-1β, Ifnγ) and iNos was analyzed using quantitative PCR analysis. Data are expressed as mRNA fold change relative to control mice fed with IR diet. Data are from two independent experiments and presented as mean ± SD (n = 4 mice per group). ns, non significant; Ã p 0.05; ÃÃ p 0.01; ÃÃÃ p 0.001. (TIF)