Nitric oxide (NO) generated by inducible NO synthase 2 (NOS2) affects cellular iron homeostasis, but the underlying molecular mechanisms and implications for NOS2-dependent pathogen control are incompletely understood. In this study, we found that NO up-regulated the expression of ferroportin-1 (Fpn1), the major cellular iron exporter, in mouse and human cells. Nos2−/− macrophages displayed increased iron content due to reduced Fpn1 expression and allowed for an enhanced iron acquisition by the intracellular bacterium Salmonella typhimurium. Nos2 gene disruption or inhibition of NOS2 activity led to an accumulation of iron in the spleen and splenic macrophages. Lack of NO formation resulted in impaired nuclear factor erythroid 2-related factor-2 (Nrf2) expression, resulting in reduced Fpn1 transcription and diminished cellular iron egress. After infection of Nos2−/− macrophages or mice with S. typhimurium, the increased iron accumulation was paralleled by a reduced cytokine (TNF, IL-12, and IFN-γ) expression and impaired pathogen control, all of which were restored upon administration of the iron chelator deferasirox or hyperexpression of Fpn1 or Nrf2. Thus, the accumulation of iron in Nos2−/− macrophages counteracts a proinflammatory host immune response, and the protective effect of NO appears to partially result from its ability to prevent iron overload in macrophages
Iron homeostasis and nitric oxide (NO) biology are reciprocally interconnected. Iron affects the expression of inducible or type 2 NO synthase 2 (iNOS or NOS2) by negatively modulating the binding affinity of the iron-sensitive transcription regulators hypoxia-inducible factor (HIF) and nuclear factor (NF)-IL6 to their respective consensus sequences within the Nos2 promoter (Weiss et al., 1994; Melillo et al., 1997; Dlaska and Weiss 1999). This is important to the host immune response because NOS2 is a major component of the antimicrobial effector machinery, and the formation of NO by macrophages has been shown to mediate protection from infections with a myriad of pathogens, including viruses, bacteria, protozoa, and parasites (MacMicking et al., 1997; Bogdan, 2001). Some of these effects have been linked to direct cellular toxicity exerted by the labile radical NO and its chemical congeners, which are generated upon interaction with reactive oxygen intermediates (Stamler and Hausladen, 1998). In addition, the high affinity of NO for iron leads to targeting of heme and nonheme iron and destabilization of iron-sulfur clusters within the active centers of enzymes that are crucial for DNA synthesis and metabolic processes (Drapier and Hibbs, 1988). Moreover, NO controls cellular iron homeostasis by activating iron regulatory protein-1 (IRP1). IRP1 binds to specific RNA stem–loop structures termed iron-responsive elements (IREs), which are located within the 3′- or 5′-untranslated regions of genes that play critical roles in iron metabolism, thereby controlling their posttranscriptional expression. Activation of IRP1 binding to IREs by NO prolongs the mRNA half-life of transferrin receptor 1 (Tfr1), leading to increased iron uptake, and at the same time inhibits the mRNA translation of the iron-storage protein ferritin (Ft), so that the concentration of free intracellular iron increases (Drapier et al., 1993; Weiss et al., 1993; Wardrop et al., 2000; Wang et al., 2005).
Cellular iron homeostasis in macrophages is regulated at multiple steps and by numerous genes (Weiss, 2009). Macrophages can acquire iron by Tfr1-mediated uptake of transferrin-bound iron, acquisition of molecular iron via the divalent metal transporter 1 (Dmt1) and, most importantly, phagocytosis of senescent erythrocytes with subsequent recycling of iron. The diversion of cellular iron is then orchestrated by the IRP/IRE interaction resulting in iron reutilization, iron storage within Ft or iron export. Importantly, there is only one well-characterized pathway by which iron can exit cells, ferroportin-1 (Fpn1), which is expressed on the cell surface of macrophages, hepatocytes, duodenal enterocytes, and erythroid progenitors and acts as the exclusive trans-membrane export protein for ferrous (Fe2+) iron (Abboud and Haile, 2000; Donovan et al., 2000; McKie et al., 2000). Thus, Fpn1 expression is a key determinant of cellular iron content. Hepcidin, a mainly liver-derived peptide induced by iron and cytokines and master regulator of body iron homeostasis, exerts its regulatory effects via binding to Fpn1, which is thought to be the hepcidin receptor. This interaction results in Fpn1 internalization, proteasomal degradation and blockage of iron export (Nemeth et al., 2004).
There are two splicing variants of Fpn1 in mammalian cells. Many cell types, including macrophages, mainly express the Fpn1 mRNA variant (Fpn1A), which harbors an IRE within its 5′-untranslated region (Hentze et al., 2004), thus making it susceptible to IRP-mediated translational regulation. In contrast, duodenal enterocytes and erythroid progenitor cells exclusively express the Fpn1B transcript, which lacks this IRE (Zhang et al., 2009).
The expression of these crucial mediators of iron homeostasis is significantly affected during infection and inflammation. Cytokines such as IL-6 or IL-1β, in concert with LPS, induce the expression of hepcidin to result in cellular Fpn1 degradation. This causes iron restriction within macrophages, blockage of duodenal iron export into the circulation, and the development of hypoferremia (Hentze et al., 2004; Kemna et al., 2005; Nemeth and Ganz, 2006).
In addition, cytokines stimulate the expression of iron uptake genes Tfr1 and Dmt1, while inhibiting Fpn1 transcription, with the net result of macrophage iron retention (Yang et al., 2002; Ludwiczek et al., 2003). Also, macrophages can produce small amounts of hepcidin in response to IL-6 or pathogen-associated molecular patterns (PAMPs), resulting in autocrine inhibition of macrophage iron efflux (Armitage et al., 2011). These inflammation-mediated effects on iron homeostasis reflect a defense strategy of the body to limit iron availability for invading extracellular pathogens, which require the metal for their growth and pathogenicity (Ganz, 2009; Nairz et al., 2010).
The iron retention in macrophages, however, bears the risk of improved iron supply to intracellular pathogens (Ganz, 2009; Nairz et al., 2010). In addition, the infected host organism can develop an anemia of chronic disease (ACD) as the restriction of iron also impairs the erythropoiesis (Weiss and Goodnough, 2005). Moreover, cellular iron retention is associated with reduced production of TNF and IL-6 by inflammatory monocytes and macrophages (Oexle et al., 2003; Nairz et al., 2009a). Thus, via increasing macrophage iron content, hepcidin might have immunomodulatory effects (Pagani et al., 2011).
Given the established regulatory circuits between NOS2 activity and iron homeostasis, and considering the important roles of both molecules for the course of infection, we hypothesized that part of the protective effect of NOS2-derived NO during infections might result from the modulation of macrophage iron content and systemic iron homeostasis.
Nos2-deficient macrophages have increased iron content and reduced Fpn1 expression
We first isolated thioglycolate-elicited primary peritoneal exudate macrophages (PE-MΦ) from Nos2+/+ and Nos2−/− mice. Nos2−/− PE-MΦ had significantly higher cellular iron content as compared with Nos2+/+ cells (Fig. 1 A). Although hepcidin mRNA levels were comparable, Nos2−/− PE-MΦ showed reduced Fpn1 mRNA expression (Fig. 1 B). To investigate the temporal regulation of iron metabolic genes in response to NO, wild-type macrophages were stimulated with NO donors of two different chemical classes, i.e., the oxide Nor5 and the oxatriazole GEA5583 (unpublished data), which provided similar results. The addition of Nor5 to wild-type macrophages resulted in a significant time-dependent increase in Fpn1 mRNA and protein levels (Fig. 1, C–E), whereas hepcidin levels were not significantly affected (Fig. 1 F).
Next, we asked whether NO alters Fpn1 mRNA half-life or affects Fpn1 transcription. Primary macrophages stimulated with Nor5 or solvent for 6 h and subsequently exposed to actinomycin D for different time periods showed comparable decay of Fpn1 mRNA with a calculated Fpn1 mRNA half-life of ∼3 h for both treatments (unpublished data).
Using a dual luciferase reporter system, we found that Nor5-treatment resulted in a significant induction of Fpn1 transcription in cells transiently transfected with the full-length 8.4-kb murine Fpn1 promoter lacking an IRE so that iron/IRP/IRE-mediated effects on mRNA stability could be excluded (Fig. 1 G, right). In contrast, no effect was observed in cells transfected with the empty pGL3 reporter plasmid (Fig. 1 G, left). To see whether these effects also occur in human cells, CD14+ monocytes were isolated from peripheral blood of healthy volunteers and stimulated with Nor5. Of note, Nor5 resulted in a substantial induction of FPN1 mRNA expression (Fig. 1 H), as did iron sulfate (FeSO4). These data suggest that NOS2-derived NO mediates iron export via transcriptional up-regulation of Fpn1.
NOS2 activity is required for Fpn1-dependent iron restriction in Salmonella-infected macrophages
Next, we evaluated the putative contribution of NOS2 to the induction of Fpn1 in response to infection with the intracellular bacterium Salmonella enterica serovar Typhimurium (S. typhimurium). Wild-type PE-MΦ were infected with S. typhimurium at a multiplicity of infection (MOI) of 10 and treated with the NOS2 inhibitor L-NIL. As published earlier (Nairz et al., 2007), the infection resulted in an up-regulation of the Fpn1 mRNA and protein expression. This effect, however, was blocked by the addition of L-NIL to S. typhimurium–infected macrophages (Fig. 2, A–C). In parallel, the cellular iron release was significantly reduced in S. typhimurium–infected macrophages concurrently treated with L-NIL (Fig. 2 D).
As the induction of Fpn1 is a mechanism for depriving intracellular microbes of iron (Ganz, 2009; Nairz et al., 2010), we evaluated the effect of NOS2 inhibition on the ability of S. typhimurium to acquire iron within infected macrophages. We observed that L-NIL resulted in a substantial increase in the access of intracellular Salmonella to radioactive ferrous iron added to infected PE-MΦ (Fig. 2 E), whereas it did not affect iron acquisition by Salmonella grown in liquid cultures (Fig. 2 F).
Salmonella-induced expression of Fpn1 is dependent on NO and Nrf2
Next, we investigated the mechanisms underlying NO-mediated induction of Fpn1 expression. Fpn1 transcription is stimulated by NF erythroid 2-related factor-2 (Nrf2), which binds to an antioxidant response element (ARE) within the Fpn1 promoter (Marro et al., 2010; Harada et al., 2011). Salmonella infection of wild-type PE-MΦ resulted in a significant activation of Nrf2 binding, which was impaired upon addition of L-NIL (Fig. 2 G).
When analyzing cells transiently transfected with Fpn1 promoter reporter constructs, we observed that Salmonella infection resulted in a significant induction of Fpn1 in comparison to solvent-treated cells (Fig. 2 H). This induction was dependent on the presence of an intact Nrf2-binding sequence within the Fpn1 promoter and on the production of NO. From these data, we conclude that Salmonella-infected macrophage increase iron export due to the up-regulation of Fpn1 expression that results from an NO-dependent binding of Nrf2 to the Fpn1 promoter and limits the bacterial access to iron.
NOS2 mediates antibacterial activity of macrophages via iron restriction
To further explore the idea that an altered iron homeostasis contributes to the well-known susceptibility of Nos2−/− macrophages to infections with intracellular bacteria, PE-MΦ from Nos2+/+ and Nos2−/− mice were infected with S. typhimurium. In line with the data obtained with uninfected macrophages (Fig. 1), Nos2+/+ PE-MΦ expressed higher amounts of Fpn1 protein (Fig. 3, A and B). Furthermore, in transiently transfected PE-MΦ, the 8.4-kb Fpn1 promoter construct was more efficiently transcribed in Nos2+/+ than in Nos2−/− macrophages after bacterial infection (Fig. 3 C). Accordingly, iron release from Salmonella-infected Nos2−/− cells was significantly lower as compared with their Nos2+/+ counterparts (Fig. 3 D).
As previously described (Shiloh et al., 1999), the bacterial load of Salmonella-infected Nos2−/− macrophages was higher than that of Nos2+/+ macrophages (Fig. 3 E). We therefore tested whether iron depletion could reverse this phenotype. Indeed, treatment of Salmonella-infected PE-MΦ with the iron chelator deferasirox (DFX) resulted in a significant reduction of the number of intracellular Salmonella and virtually eliminated the differences in Salmonella survival between Nos2−/− and Nos2+/+ macrophages (Fig. 3 E).
The addition of IFN-γ to Salmonella-infected RAW264.7 macrophage-like cells resulted in a further induction of Fpn1 mRNA levels, as did the overexpression of functional natural resistance-associated macrophage protein 1 (Nramp1R; Fig. 3 F). NOS2-derived NO was required for both effects, as L-NIL prevented the up-regulation of Fpn1 and the subsequent reduction of macrophage iron content (Fig. 3, F and G).
The increased capacity of DFX-treated Nos2−/− macrophages to control intracellular Salmonella proliferation was associated with an up-regulation of the proinflammatory cytokine TNF. Also, DFX abrogated the differences otherwise observed between Nos2+/+ and Nos2−/− macrophages (unpublished data).
As a complementary approach, Nos2+/+ and Nos2−/− PE-MΦ were transiently transfected with three different FPN1 expression plasmids. Western blots confirmed the expression of high Fpn1 protein levels in transfected control cells of either Nos2 genotype (Fig. 4, A and B). The overexpression of wild-type FPN1 or a N144H mutant FPN1, which is insensitive to hepcidin-mediated degradation (Drakesmith et al., 2005), resulted in a significant and comparable reduction of intracellular Salmonella survival and abrogated the differences in bacterial counts seen between untransfected Nos2+/+ and Nos2−/− macrophages. In contrast, overexpression of a A77D FPN1 mutant that, unlike wild-type FPN1 and the N144H FPN1 mutant, is not expressed on the cell surface but retained in the endoplasmatic reticulum, had no effect on differences in bacterial numbers between Nos2+/+ and Nos2−/− macrophages (Fig. 4 C). Moreover, transfection of cells with wild-type FPN1 and the N144H variant resulted in increased TNF and IL-12p70 secretion after infection with S. typhimurium (Fig. 4, D and E).
Supernatants from macrophages transiently transfected with empty p-cDNA3.1 plasmid or human FPN1 constructs were analyzed for the accumulation of nitrite after S. typhimurium infection. Overexpression of wild-type FPN1 resulted in a significant increase of nitrite accumulation in culture supernatants. The same phenotype was observed in cells expressing the FPN1 N144H mutant and in macrophages transfected with an empty plasmid vector but treated with DFX. Conversely, the stimulatory effects of wild-type FPN1 overexpression were abrogated upon addition of 50 µM FeSO4 or 1 µM synthetic human hepcidin (Fig. 4 F). Together, these data show that the iron overload and the increased survival of Salmonella seen in Nos2−/− macrophages can be fully reversed by iron chelation or forced iron export. In addition, iron depletion also restored the proinflammatory cytokine response of Nos2−/− macrophages.
Having seen that in uninfected macrophages NOS2-derived NO regulates the expression of Fpn1 via the nuclear protein Nrf2 (Fig. 2), we tested whether the same regulatory circuit also holds true in Salmonella-infected macrophages. As expected from the NOS2 inhibitor data (Fig. 2), Nrf2 activity was significantly higher in Nos2+/+ as compared with Nos2−/− cells (Fig. 5 A). Furthermore, overexpression of Nrf2, which is paralleled by enhanced transcriptional binding activity (unpublished data), caused a significant increase in iron release from Salmonella-infected macrophages (Fig. 5 B). In parallel, the intracellular bacterial loads were significantly reduced after Nrf2 hyperexpression in Nos2+/+ or Nos2−/− macrophages (Fig. 5 C). The reduction of Salmonella growth within macrophages overexpressing Nrf2 most likely results from an Fpn1-mediated limitation of bacterial iron access, because it was abrogated upon addition of synthetic murine hepcidin which mediates the degradation of Fpn1 (Fig. 5 D).
Improved control of Salmonella by individual macrophages hyperexpressing Fpn1
To analyze the impact of Fpn1 expression on the antimicrobial activity of macrophages on a single-cell level, we transiently transfected PE-MΦ with plasmids coding for FPN1 and Emerald green fluorescent protein (EmGFP) linked to the N or C terminus of FPN1 or to respective control plasmids lacking FPN1. After infection with S. typhimurium, macrophages were evaluated for their expression of GFP and categorized into four groups based on their infection status (i.e., 0, 1–2, 3–5, and >5 bacteria/macrophage) using multicolor immunofluorescence analysis.
In GFP+ macrophages, the expression of FPN1 resulted in lower bacterial loads on a single-cell basis compared with FPN1-negative controls (Fig. 6, A and B, compare right and left panels). Furthermore, many GFP− cells treated with L-NIL contained at least three bacteria per cell, whereas most GFP+ cells (i.e., FPN1+ cells) did not contain more than two bacteria (Fig. 6, C and D, compare right and left panels). Similarly, most GFP+ Nos2+/+ macrophages overexpressing FPN1 contained two or fewer bacteria (Fig. 7, A [right] and D), whereas in GFP− Nos2+/+ macrophages, the proportion of cells containing 3–5 bacteria was significantly higher (Fig. 7, A [right] and C). As expected, in Nos2−/− macrophages the differences between GFP+ cells overexpressing FPN1 (Fig. 7, B [right] and F) and GFP− cells lacking FPN1 (Fig. 7, B [left] and E) were more pronounced.
Thus, forced Fpn1 expression clearly impairs the survival of Salmonella in PE-MΦ in a cell-autonomous manner.
Nos2 deficiency results in iron overload in mice
After these striking in vitro results, we investigated whether NOS2 might contribute to the regulation of iron homeostasis within mononuclear phagocytes in vivo. First, we assessed tissue iron distribution by Prussian blue staining. In line with the increased iron content measured in Nos2−/− macrophages (Fig. 1 A), we observed increased iron storage in spleens of Nos2−/− mice (Fig. 8, B and D). Iron storage was most prominent in the red pulp (rp), which is rich in mononuclear cells, whereas the white pulp (wp) of Nos2−/− mice showed hardly any iron accumulation (Fig. 8, B and D). Only very few and small iron foci were seen in (the red pulp of) Nos2+/+ mice (Fig. 8, A and C). When measuring the iron content of the spleen and liver of uninfected mice, we observed that Nos2−/− mice had increased levels of tissue iron compared with Nos2+/+ control mice (Fig. 8 E and not depicted). Notably, Nrf2 activation was reduced in nuclear extracts of spleens (Fig. 8 F) and livers (not depicted) from Nos2−/− mice relative to Nos2+/+ controls.
In livers, iron storage was more pronounced in Kupffer cells than in hepatocytes and the number of cells staining positive with Prussian blue was higher in Nos2−/− as compared with Nos2+/+ livers (unpublished data). In line with the well-established NOS2 expression by liver parenchymal cells (Bogdan, 2001), the cellular iron content was also higher in isolated primary Nos2−/− as compared with Nos2+/+ hepatocytes (Fig. 9 A).
Moreover, primary Nos2−/− hepatocytes expressed more hepcidin mRNA both under control conditions and in response to stimuli such as LPS, IL-6, and bone morphogenetic protein-6 (Bmp6). Conversely, exposure of NOS2−/− hepatocytes to the NO donor Nor5 rather reduced hepcidin mRNA formation in comparison to untreated control cells (Fig. 9 B). Thus, a lack of NOS2 is associated with an increased tissue iron load in vivo, which in the liver might cause an up-regulation of hepcidin.
NOS2-dependent iron restriction promotes host resistance to Salmonella infection in vivo
To see whether the NOS2- and Nrf2-dependent induction of Fpn1 is relevant for host defense in vivo, Nos2+/+ and Nos2−/− mice were infected with 500 CFUs of S. typhimurium and analyzed after 24 and 72 h, respectively. At both time points, Nos2−/− mice contained significantly higher bacterial loads in spleens and livers than the respective wild-type controls (Fig. 10 A and not depicted).
In accordance with our results obtained with PE-MΦ, infection of Nos2+/+ mice with S. typhimurium resulted in increased splenic Fpn1 protein levels, whereas in Nos2−/− mice Fpn1 remained at the level of uninfected mice (Fig. 10, B and C). Moreover, Nrf2-binding activity was induced after infection and was significantly higher in splenic nuclear extracts prepared from Nos2+/+ mice as compared with Nos2−/− mice (Fig. 10 D).
Oral treatment of Salmonella-infected mice with DFX not only caused a significant reduction of the bacterial burden in the spleen and liver of Nos2+/+ and Nos2−/− mice, but also led to indistinguishable numbers of bacteria in the tissues of both genotypes (Fig. 10 E and not depicted). TNF, IL-12p35, and IFN-γ mRNA expression was higher in the spleens of S. typhimurium–infected Nos2+/+ mice compared with Nos2−/− mice. DFX treatment caused an up-regulation of the mRNA expression of TNF, IL-12p35, and IFN-γ in the spleen of Salmonella-infected mice and abrogated the differences initially seen between Nos2+/+ than Nos2−/− mice (Fig. 10 F and not depicted). In contrast, IL-23p19 and IL-17A mRNA levels remained unaffected by DFX treatment or Nos2 genotype (unpublished data).
Prussian blue staining confirmed the efficacy of oral DFX-treatment in reducing splenic iron overload in Nos2−/− mice (Fig. 10, H vs. I). Supporting the concept that the increased susceptibility of Nos2−/− animals to S. typhimurium may result from increased microbial iron access, splenic and hepatic bacterial loads were comparable in Nos2+/+ and Nos2−/− mice when infected with an S. typhimurium entC sit feo mutant that lacks three important iron acquisition systems (Fig. 10 E and not depicted).
Hepatic hepcidin mRNA levels were up-regulated in Nos2+/+ mice in response to Salmonella infection, although a significant induction was only observed 24 h after the initiation of infection (Fig. 10 J). In Nos2−/− mice, in contrast, there were no further alterations of the higher basal hepcidin mRNA levels. After infection, Nos2+/+ mice displayed a significant and prolonged reduction of serum iron levels, which was less evident in the absence of Nos2 (Fig. 10 K).
In this study, we evaluated the putative role of NOS2 and its product NO as a mediator of host iron homeostasis and antimicrobial defense. We observed that NO induced the expression of Fpn1 in macrophages by activating the transcription factor Nrf2, thereby regulating cellular iron export under both resting conditions and in response to Salmonella infection. NO limits the ability of phagocytosed bacteria to obtain radiolabeled iron within infected macrophages, suggesting that NO-induced Fpn1 transcription resulted in iron deprivation of intracellular bacteria. Moreover, the addition of the iron chelator DFX to Nos2−/− macrophages reduced intracellular Salmonella numbers to levels seen in Nos2-expressing wild-type cells and restored the capacity of Salmonella-infected Nos2−/− macrophages to produce large amounts of TNF and IL-12. Thus, our findings suggest that the NOS2–Nrf2–Fpn1 axis is an important determinant of macrophage iron homeostasis and antimicrobial effector functions, which may counterbalance the hepcidin-induced Fpn1 degradation during infection with intracellular pathogens.
Iron, NO, and host defense
Several studies demonstrated the critical role of iron homeostasis in host defense (Byrd and Horwitz 1989; Weinberg, 2000; Collins 2003; Oexle et al., 2003; Schaible and Kaufmann, 2004; Schrettl et al., 2004; Nairz et al., 2010; Pagani et al., 2011; Johnson and Wessling-Resnick, 2012). During infections with intracellular pathogens, Fpn1-mediated iron export was a key component of the innate immune response (Chlosta et al., 2006; Nairz et al., 2007; Paradkar et al., 2008; Johnson et al., 2010; Mair et al., 2011). Hitherto, the molecular mechanism underlying Fpn1 induction in the presence of intracellular microbes has been unknown. Our observations show that NOS2, which has long been recognized as a central antimicrobial mechanism in a variety of infectious diseases (MacMicking et al., 1997; Bogdan, 2001), is a central mediator in the adaptation of iron metabolism after infection with the intracellular bacterium S. typhimurium (Mastroeni et al., 2000; Vazquez-Torres et al., 2000). In support of this conclusion, Salmonella residing within Nos2−/− macrophages was able to acquire significantly more iron in comparison to bacteria within wild-type macrophages. Thus, NO-mediated activation of Nrf2-dependent Fpn1 transcription appears to be a key component of innate immunity.
Because iron is an essential cofactor for enzymes involved in energy generation, we believe that reactive nitrogen species (RNS) and the NOS2–Fpn1–dependent iron restriction may act in concert to disrupt bacterial metabolism (Mastroeni et al., 2000; Vazquez-Torres et al., 2000). A recent report has shown that NO has multiple protein targets within bacterial metabolic pathways of S. typhimurium (Richardson et al., 2011). For instance, NO inactivated lipoamide-dependent oxidative decarboxylation reactions in Salmonella and thereby blocked the synthesis of methionine and lysine during nitrosative stress in vivo. Inhibition of the bacterial tricarboxylic acid cycle by NO also occurred by the targeting of iron–sulfur clusters within central enzymes and the transcriptional regulator FNR.
As a protective measure against RNS-induced damage, S. typhimurium expresses the flavohemoprotein Hmp to detoxify NO via a NsrR-regulated pathway (Crawford and Goldberg, 1998; Bang et al., 2006). Moreover, Salmonella Pathogenicity Island 2 reportedly inhibits trafficking of NOS2 to Salmonella-containing vacuoles to reduce nitrosative stress (Chakravortty et al., 2002).
Apart from its direct antimicrobial actions, NOS2 exhibits diverse immunomodulatory effects on several types of immune cells (Bogdan, 2000, 2001). For instance, NOS2 is a co-stimulatory factor for the expression of IFN-γ mRNA during the innate phase of experimental cutaneous leishmaniasis (Diefenbach et al., 1998). NO may also affect IL-12 production by macrophages, although reports disagree on whether there is a positive, a negative, or no role of NOS2 in IL-12 secretion (Rothe et al., 1996; Huang et al., 1998; Lasarte et al., 1999; Mullins et al., 1999). Our present data demonstrate that macrophage functions in the setting of Salmonella infection are also critically dependent on NO, as Fpn1-mediated iron export was required for TNF, IL-12, and IFN-γ mRNA expression.
NO-induced Fpn1 transcription reduced cellular iron content. This may serve two purposes. First, it has long been recognized that iron inhibits Nos2 transcription by reducing the binding affinity of NF-IL6 (Weiss et al., 1994; Dlaska and Weiss, 1999). Moreover, the reduced intracellular iron content of macrophages expressing the functional (i.e., resistant) variant of Nramp1-enhanced NOS2 transcription via IFN-regulatory factor-1 (IRF-1; Fritsche et al., 2003). The up-regulation of Fpn1-mediated iron export in response to NO may thus act as a feed-forward mechanism to allow sustained NO production despite NO-mediated activation of the IRP/IRE system, which would tend to increase iron uptake and intracellular iron storage (Drapier et al., 1993; Weiss et al., 1993; Wardrop et al., 2000; Wang et al., 2005). Our results with macrophages overexpressing functional human FPN1 variants show that increased RNS production in response to Salmonella infection is a result of macrophage iron depletion, as it could be inhibited by the addition of either iron sulfate or hepcidin. Our finding that FPN1 overexpression also promoted TNF and IL-12 production is in line with the idea that the cellular iron status is a central determinant of macrophage immunostimulatory and effector functions (De Domenico et al., 2010; Nairz et al., 2010; Pagani et al., 2011). Iron is known to impair Th1-type immune responses (Weiss et al., 1992; Mencacci et al., 1997; Oexle et al., 2003). Accordingly, Nos2−/− mice, which have increased splenic iron content, expressed lower levels of TNF, IL-12p35, and IFN-γ mRNA, but unaltered amounts of IL-17A mRNA in the spleen. This phenotype was completely reversed by DFX treatment. Thus, the Th1 pathway rather than Th17-mediated mechanisms were negatively modulated by a lack of NOS2-derived NO caused by an increased splenic iron content.
Additionally, NO has the potential to inhibit vital metabolic processes by directly targeting iron-sulfur cluster– and heme-containing centers of crucial enzymes (Drapier and Hibbs, 1988; Fang 2004; Henard and Vazquez-Torres, 2011). The induction of Fpn1 may thus aim at reducing cellular iron content to limit collateral damage by iron-catalyzed reactive oxygen species (ROS) and toxic congeners of NO and ROS (Stamler and Hausladen, 1998).
Mechanism of regulation of Fpn1 by NO
The role of NO for the regulation of the IRP/IRE system is well established (Hentze et al., 2004). Although NO activates IRP1 by disrupting its iron-sulfur cluster (Weiss et al., 1993), thus converting its activity from the enzymatic, cytoplasmic aconitase function to its IRE-binding function, IRP2 is activated when NO specifically inhibits its degradation through the proteasome (Wang et al., 2005). As cells lacking both IRP1 and IRP2 have increased Fpn1 protein levels but unchanged Fpn1 mRNA levels (Galy et al., 2008), it appears unlikely that the posttranslational control of Fpn1 expression via the IRP/IRE system substantially contributes to the NO-dependent regulation observed in our studies. Rather, NO-mediated activation of Nrf2 and subsequent transcriptional induction of Fpn1 expression appear to be responsible. In line with this mechanism, addition of the RNA polymerase II inhibitor actinomycin D abrogated the effects of NO on Fpn1 mRNA levels (unpublished data). Moreover, in our reporter studies, we used a Fpn1 promoter construct in which the IRE was specifically deleted (Marro et al., 2010). Despite the well-known interactions between NO and the IRP/IRE system, we did not observe differential IRP1 or IRP 2 protein content in the spleen of Nos2−/− and Nos2 wild-type mice by means of Western blotting (unpublished data).
Nrf2 is an iron–sulfur cluster–containing basic leucine zipper transcription factor, which acts as a cellular stress sensor and initiates appropriate responses to oxidative stress. Nrf2 mediates both the constitutive and inducible expression of a variety of cytoprotective genes, including antioxidative enzymes and export proteins (Ishii et al., 2000). Recent studies have implicated Nrf2 as modifier of iron homeostasis. In macrophages, pharmacological activation of Nrf2 with sulforaphen was found to induce Fpn1 transcription (Harada et al., 2011). Nrf2−/− mice exhibit reduced iron content in the dental enamel but increased hepatic iron deposition. In a murine model of steatohepatitis, Nrf2 deficiency was associated with reduced hepatic Fpn1 expression and, consequently, reduced hepatocellular iron export (Okada et al., 2012). Furthermore, renal Nrf2 is activated after injection with iron nitrilotriacetate, and Nrf2−/− mice display dramatically increased susceptibility to iron nitrilotriacetate-induced oxidative kidney damage caused by their inability to mount an appropriate nephroprotective response (Kanki et al., 2008; Tanaka et al., 2008). Thus, the Nrf2–Fpn1 pathway may be seen as a countermeasure to limit ongoing production of toxic radicals via iron-mediated Fenton chemistry.
Although the splenic iron distribution observed in Nos2−/− mice is similar to that seen in mice with myeloid-specific deletion of Fpn1, the immune phenotype is quite different between Nos2−/− mice with reduced Fpn1 levels and Fpn1-deficient Fpn1LysM/LysM mice (Zhang et al., 2011). We observed diminished production of TNF and IL-12p70 in Salmonella-infected Nos2−/− PE-MΦ that could be specifically reversed by overexpression of FPN1. In contrast, Fpn1LysM/LysM bone marrow–derived macrophages exhibit increased TNF and IL-6 secretion in response to LPS challenge that could be reversed by treatment with the iron chelator desferroxamine (DFO; Zhang et al., 2011).
In contrast to hepcidin, which causes ligand-induced lysosomal degradation of Fpn1, we identified NO as transcriptional Fpn1 inducer. It would thus be interesting to test the effects of NO donors in pathophysiological conditions associated with reduced Fpn1 activity such as ACD and cancer (Weiss and Goodnough, 2005; Nemeth and Ganz, 2006). We speculate that, by virtue of their capacity to up-regulate Fpn1, NO donors may counteract the functional iron deficiency underlying the pathogenesis of ACD.
We did not observe any significant effect of NO on hepcidin expression. Primary hepatocytes isolated from Nos2−/− mice responded appropriately to well-known hepcidin-inducing stimuli. In fact, Nos2−/− hepatocytes expressed elevated hepcidin levels under control conditions and showed up-regulation in response to IL-6, Bmp6, or LPS. In contrast, Nor5 did not affect basal hepcidin transcription in Nos2+/+ hepatocytes, but NO treatment caused a significant decrease in hepcidin mRNA expression in Nos2−/− hepatocytes. We speculate that this regulation results from the induction of Fpn1 and the subsequent reduction of hepatocellular iron levels. Our data are compatible with the idea that the elevation of hepcidin observed in the absence of NOS2 in vitro and in vivo is secondary to iron overload rather than to a direct negative regulation of hepcidin gene expression by NO. In line with this assumption, we failed to observe a down-regulatory effect of NO on hepcidin mRNA expression (Fig. 1 F). Furthermore, the analysis of 6-wk-old mice revealed reduced Fpn1 levels in the spleen and liver of Nos2−/− mice as compared with Nos2+/+ controls, whereas at this age, hepcidin mRNA levels in the liver of both mouse strains were indistinguishable (not depicted).
We observed that the hypoferremia in Salmonella-infected Nos2+/+ mice was more pronounced and prolonged than in Nos2−/− mice, which presented with higher hepcidin expression than the infected wild-type animals. This suggests that additional, hepcidin-independent mechanisms may exist to mediate an appropriate reduction of serum iron levels in the setting of Salmonella infection. TNF has a well-established role in the hypoferremia of the acute-phase response and the control of cellular iron homeostasis (Ludwiczek et al., 2003; Laftah et al., 2006). Given its lower levels in Nos2−/− mice, TNF might also contribute to the observed differences between Salmonella-infected Nos2+/+ and Nos2−/− mice.
We propose that in parallel to the hepcidin-mediated posttranslational down-regulation of Fpn1, NOS2/NO-dependent activation of Nrf2 and the subsequent transcriptional induction of Fpn1 controls cellular iron release and macrophage function. The NOS2–Nrf2–Fpn1 pathway is essential for macrophage responses to S. typhimurium, but presumably also for other intracellular pathogens. In line with our results, J774 macrophages overexpressing Fpn1 suppressed the intracellular growth of Mycobacterium tuberculosis. At the same time, however, the LPS-induced translation of NOS2 in these cells was impeded, an effect that remained mechanistically unexplained but reversible by IFN-γ (Johnson et al., 2010).
Iron chelators may be useful for the treatment of infections with intracellular pathogens. Given the key role of iron for microbial pathogenicity, the iron chelator DFO has also been tested in models of Salmonella infection. Although its in vivo administration is associated with reduced survival of mice (Collins et al., 2002), this might be attributable to the ability of extracellular Salmonella to use ferrioxamines as an iron source (Kingsley et al., 1999). In contrast, DFO restricts intracellular Salmonella replication in vitro due to its ability to withhold iron from the microbe (Nairz et al., 2007). It is important to note that in our study, the water-soluble chelator DFX had beneficial effects on Salmonella infection both in vitro and in vivo. The difference between DFO and DFX may be related to diverse cellular penetration and a differential ability of these compounds to be used by Salmonella in a metal-bound state. In this context, it is worth noting that both DFX and deferiprone (DFP) have been used to limit the proliferation of Chlamydia psittaci, Chlamydia pneumoniae, and Legionella pneumophila in infected macrophages (Paradkar et al., 2008; Bellmann-Weiler et al., 2012). However, data on the potential use of oral iron chelators in the treatment of systemic infections remains scarce. DFX has been demonstrated to be effective in mucormycosis in animal models (Ibrahim et al., 2007). When mice subjected to chemotherapy-induced neutropenia or diabetic ketoacidosis were experimentally infected with Rhizopus oryzae, DFX-treatment reduced mortality, which was paralleled by the induction of Th1 cytokines TNF and IFN-γ in the spleens of DFX-treated mice. Our studies disclose the novel finding that oral application of DFX may beneficially affect the course of systemic infections with the intracellular bacterium S. typhimurium. However, whether this holds also true for other intracellular microbes during in vivo infection and whether these findings will translate into clinical benefits for infected humans, remains to be investigated. Nevertheless, DFX has proven clinical efficacy and safety in the treatment of secondary iron-overload in the setting of hemoglobinopathies (Cappellini et al., 2011).
In summary, the data presented here provide novel mechanistic insights into the antimicrobial role of NO in infection and into the regulation of macrophage iron homeostasis. NOS2-derived NO activates Nrf2, resulting in increased Fpn1 transcription, cellular iron export and limitation of iron availability for pathogens residing in macrophages. In addition, NOS2-mediated reduction of the macrophage iron content promotes a proinflammatory cytokine response. Our results imply that the NOS2-initiated iron export and its subsequent immunological and metabolic effects are important for the efficient control of infections with intracellular pathogens such as S. typhimurium.
MATERIALS AND METHODS
Cell isolation and culture.
Thioglycolate-elicited PE-MΦ were harvested from C57BL/6J Nos2+/+ and Nos2−/− mice and cultured as detailed elsewhere (Nairz et al., 2009b). Macrophages were incubated in DMEM and treated with the NO donors Nor5 ((±)-2-((E)-4-Ethyl-3[(Z)-hydroxyimino]6-methyl-5-nitroheptenyl)-3-pyridinecarboxamide) or Gea5583 (1,2,3,4-Oxatriazolium, 3-(3-chloro-2-methylphenyl)-5(((cyanomethylamino)carbonyl)amino), hydroxide inner salt; 50 µM), both from Enzo Life Science, and dissolved in DMSO (final concentration 0.1%; Sigma-Aldrich) or the respective solvent control for the indicated time periods. For other experiments, the RNA polymerase II inhibitor actinomycin D (Sigma-Aldrich) was dissolved in DMSO and added at a concentration of 5 µg/ml. To determine mRNA half-life, cells were stimulated for 6 h, and then washed with fresh medium and subsequently incubated with actinomycin D or DMSO for different time periods.
RAW264.7 murine macrophage-like cells, originally isolated from Nramp1S BALB/c mice, were stably transfected with the pHbβA-1-neo expression plasmid containing the full-length Nramp1 cDNA (Nramp1 resistant) or an antisense-Nramp1 construct (Nramp1 susceptible) and used as described previously (Nairz et al., 2009a).
Primary hepatocytes were isolated and cultured as described in detail elsewhere (Theurl et al., 2008b). Hepatocytes were incubated in Williams E medium and stimulated with phosphate buffered saline (PBS; obtained from PAA), 100 ng/ml LPS (from Salmonella Abortus equi; Sigma-Aldrich), 10 ng/ml recombinant IL-6 (R&D Systems), 25 ng/ml recombinant bone morphogenic protein (Bmp)-6 (Abcam), or a combination thereof for 6 or 24 h, respectively.
Intracellular iron concentrations in primary hepatocytes and macrophages were determined by atomic absorption as previously described (Theurl et al., 2008a). Blood samples were obtained from seven healthy individuals during routine vein puncture after written informed consent had been obtained in accordance with the Declaration of Helsinki. The study was approved by the MUI ethics committee at the Medical University of Innsbruck (approval no. AN3468-272/4.7). After isolation of white blood cells by Ficoll-Paque separation (Pharmacia), CD14+ monocytes were freshly separated by magnetic sorting using specific MicroBeads (Miltenyi Biotec) after Ficoll-Paque separation (Pharmacia). The purity of the resulting cells suspension was tested by fluorescent-activated cell sorting analysis and was beyond 95% (Theurl et al., 2008b).
Salmonella infection in vitro.
Before in vitro infection, macrophages were extensively washed with PBS and incubated in complete DMEM without antibiotics. Wild-type S. typhimurium strain ATCC 14028 was used for all experiments and grown under sterile conditions in LB broth (Sigma-Aldrich) to late-logarithmic phase. Macrophages were infected with S. typhimurium at a multiplicity of infection (MOI) of 10 for 24 h. Where appropriate, cells were subsequently treated with PBS, 600 µM L-NIL (L-N6-[1-iminoethyl]-lysine.2HCl; purchased from Enzo Life Sciences), 50 µM deferasirox (DFX; from Novartis), 50 µM FeSO4 (Sigma-Aldrich) or 1 µM synthetic murine hepcidin (PeptaNova). For experiments in which human FPN1 variants were overexpressed, and 1 µM synthetic human hepcidin (PeptaNova) was used. To determine intracellular bacterial loads, infected cells were harvested in 0.1% sodium deoxycholic acid (Sigma-Aldrich) as previously described (Nairz et al., 2009b).
Salmonella infection in vivo.
C57BL/6J type mice (Nos2+/+) and Nos2−/− mice (Laubach et al., 1995) that were backcrossed onto a C57BL/6J background for at least 11 generations (JAX stock number 2609) were obtained at the age of 4–6 wk from The Jackson Laboratories via the Charles River Breeding Laboratories. Congenic mice were housed in neighboring cages under specific pathogen–free conditions at the central animal facility of the Medical University of Innsbruck or at the Franz Penzoldt Center for Animal Research of the Universitätsklinikum Erlangen. All animal experiments were performed according to the guidelines of the Medical University of Innsbruck based on the Austrian Animal Testing Act of 1988. All animal experiment protocols were approved by the MUI Animal ethics committee and the Austrian Ministry for Science and Education (approval no. BMWF-66.011/0151-II/3b/2011). Unless otherwise indicated, male mice were used at 18–20 wk of age and infected i.p. with 500 CFU of S. typhimurium diluted in 200 µl PBS. Unless otherwise specified, S. typhimurium wild-type strain ATCC 14028s was used for experiments. In some experiments, its isogenic triple mutant derivative entC::aph sit::bla feo::Tn10 (Kanr Apr Tetr) was used. This mutant strain was constructed and grown as previously described (Crouch et al., 2008) and used as detailed above. Where indicated, mice received DFX ad libitum via the oral route by dissolving 500 mg DFX in 1,000 ml drinking water. The bacterial load of organs was determined by plating serial dilutions of organ homogenates on LB agar (Sigma-Aldrich) under sterile conditions and the number of bacteria was calculated per gram of tissue as previously described (Nairz et al., 2011).
RNA extraction and quantitative real-time PCR.
Total RNA was prepared and mRNA expression quantified by quantitative RT-PCR after reverse transcription exactly as described (Nairz et al., 2009b).
Western blot analysis.
Protein extraction and Western blotting were performed exactly as described (Theurl et al., 2006) using a rabbit Fpn1 antibody (1:400; Eurogentec), a rabbit actin antibody (1:1,000; Sigma-Aldrich), and appropriate HRP-conjugated secondary antibodies (1:1,000; Dako).
PE-MΦ were seeded on glass slides, transfected as described below, and subsequently infected for 16 h. After fixation with ice-cold methanol for 15 min on ice, cells were rinsed twice with PBS containing 0.1% Tween-20 and sequentially incubated with antibodies as follows: a chicken GFP antibody (1:1000; Invitrogen), a goat CSA1 (for Salmonella common structural antigen) antibody (1:1,000; KPL), and appropriate fluorochrome-conjugated secondary antibodies, i.e., rabbit anti–chicken Alexa Fluor 488 (green; 1:1,000; Dianova) and donkey anti–goat Alexa Fluor 594 (red; 1:1,000; Invitrogen). For microscopic analysis, the cells were covered into Fluorescent Mounting Medium (Dako) containing DAPI and analyzed under the Axioskop 2 microscope (Carl Zeiss). Acquisition was done using the AxioCam MRc 5 camera, a 100×/1.25 objective, and the AxioVision imaging software version 4.0 (Carl Zeiss).
Transcription factor assay.
Nuclear protein extracts were prepared with the Nuclear and Cytoplasmic Extraction Reagent (Thermo Fisher Scientific). Nrf2-binding activity of nuclear extracts was assessed with a commercially available colorimetric transcription factor assay kit exactly according to the manufacturer’s instructions (Active Motif).
Plasmids and transient transfection.
To generate the 8-kbΔIRE-Fpn1 construct, a fragment of the mouse Fpn1 promoter containing the 5′-UTR and 8,071 bp upstream of the transcription start site was cloned into the pGL3 luciferase reporter gene vector (Promega). A corresponding variant lacking the Nrf2-binding site at position −7007/−7016 was constructed by site-specific mutagenesis as previously described (Marro et al., 2010). Plasmids were kindly provided by S. Marro, E. Messana, and S. Altamura (University of Heidelberg, Heidelberg, Germany; and Molecular Biotechnology Center, University of Torino, Torino, Italy).
The 1.6-kb full-length murine Nos2 promoter was cloned into the pGL3 luciferase reporter gene vector as previously described (Dlaska and Weiss, 1999). Promoter activity was determined by the Dual Luciferase system (Promega) according to the manufacturer’s instructions. Firefly luciferase activity was corrected by co-transfection of cells with the constitutively expressed Renilla luciferase vector pRL-SV40.
Mouse Nrf2 was cloned into the p-cDNA3.1 expression plasmid as previously specified (Song et al., 2009), and the construct was generously provided by M.-Y. Song, H.-S. So, and B.-H. Park (Medical School, Chonbuk National University, Jeonju, Jeonbuk, South Korea). Human FPN1 and the N144H and A77D FPN1 mutants, respectively, were cloned into the p-cDNA3.1/c-Myc expression vector as described previously (Drakesmith et al., 2005; L. Schimanski and H. Drakesmith, Weatherall Institute of Molecular Medicine, Oxford, England, UK). Transient transfections of primary PE-MΦ were performed by means of lipofection using Lipofectamine (Promega).
For immunocytochemistry, we used EmGFP N- or C-terminally tagged human FPN1 expression clones, provided by H. Zoller (Department of Internal Medicine II, Medical University of Innsbruck, Austria), as previously described (Mayr et al., 2011). The FPN1 open reading sequence was transferred into Vivid Colors pcDNA 6.2/EmGFP-DEST vectors to create EmGFP FPN1 expression clones. As control vectors, we used Vivid Colors pcDNA 6.2 EmGFP/CAT (chloramphenicol acetyltransferase) plasmids. Efficacy of transfections was at least 70–85%.
Detection of cytokines and reactive species.
Determination of cytokines in culture supernatants was performed with ELISA kits for TNF and IL-12p70 (BD). Determination of nitrite, the stable oxidation product of NO, was performed with the Griess-Ilosvay’s nitrite reagent (Merck) as previously described (Nairz et al., 2008).
Quantification of macrophage iron export and bacterial iron acquisition.
For macrophage iron export assays, cells were treated and infected with S. typhimurium as detailed above and iron export was determined with 5 µM 59Fe-citrate exactly as previously described (Nairz et al., 2009a). In parallel to each iron release study, a Trypan blue exclusion assay was performed to ensure that neither treatment interfered with the macrophage integrity and viability. For the determination of iron acquisition by intramacrophage S. typhimurium we used a protocol as described (Nairz et al., 2007).
Statistical analysis was performed using a SPSS statistical package. We determined significance by unpaired two-tailed Student’s t test to assess data, when only two groups were compared. ANOVA combined with Bonferroni correction was used for all other experiments. Unless otherwise specified, data are depicted as lower quartile, median, and upper quartile (boxes), and minimum/maximum ranges. For the comparison of organ bacterial loads, data were log-transformed before Student’s t test or analysis of variance. Individual values and means of log-transformed values are depicted. Generally, p-values less than 0.05 were considered significant in any test.
The authors are grateful to Nadja Baumgartner, Sylvia Berger, Ines Brosch, Sabine Engl, and Markus Seifert for excellent technical support.
The authors are indebted to Mrs. Lisa Schimanski and Drs. Sandro Altamura, Howard C. Barton, Hal Drakesmith, Samuele Marro, Erika Messana, Byung-Hyun Park, Hong-Soeb So, Mi-Young Song, and Heinz Zoller for providing plasmids and cell lines.
This work was supported by grants from the Austrian Research Fund (FWF; project TRP-188 to G. Weiss), the Medical University of Innsbruck (START program; to M. Nairz), and the German Research Foundation (DFG; SFB643 to U. Schleicher and C. Bogdan), by the Interdisciplinary Center for Clinical Research Erlangen (project grant A49 to U. Schleicher and C. Bogdan), and by the Verein zur Förderung von Forschung und Weiterbildung in Infektiologie und Immunologie an der Medizinischen Universität Innsbruck. F. Fang is supported by the National Institutes of Health (AI39557).
The authors declare no competing financial interests.
anemia of chronic disease
antioxidant response element
bone morphogenic protein
divalent metal transporter 1
Emerald green-fluorescent protein
iron regulatory protein
multiplicity of infection
natural resistance-associated macrophage protein 1
NF erythroid 2-related factor 2
pathogen-associated molecular pattern
peritoneal exudate macrophage
- S. typhimurium
Salmonella enterica serovar Typhimurium
transferrin receptor 1