Research ArticleINFECTIOUS DISEASE

IL-22 controls iron-dependent nutritional immunity against systemic bacterial infections

See allHide authors and affiliations

Science Immunology  03 Feb 2017:
Vol. 2, Issue 8, eaai8371
DOI: 10.1126/sciimmunol.aai8371

Starving the pathogen

Actively killing pathogens is an important function of the immune response; equally important are mechanisms that limit nutrient availability to the pathogen, termed nutritional immunity. The cytokine interleukin-22 (IL-22) plays an essential role in resolution of infections at epithelial barrier sites, including skin, lungs, and intestines. Using a systemic model of Citrobacter rodentium infection, Sakamoto et al. have uncovered an unexpected role for IL-22 in limiting availability of iron to the pathogen by promoting increased production of heme scavengers from the liver. Their studies extend the role of IL-22 beyond barrier sites and establish a previously unappreciated role for IL-22 in regulating nutritional immunity in the context of systemic bacterial infections.

Abstract

Host immunity limits iron availability to pathogenic bacteria, but whether immunity limits pathogenic bacteria from accessing host heme, the major source of iron in the body, remains unclear. Using Citrobacter rodentium (a mouse enteric pathogen) and Escherichia coli (a major cause of sepsis in humans) as models, we find that interleukin-22 (IL-22), a cytokine best known for its ability to promote epithelial barrier function, also suppresses the systemic growth of bacteria by limiting iron availability to the pathogen. To understand the mechanistic basis of IL-22–dependent iron retention in the host, using an unbiased proteomic approach, we have identified that IL-22 induces the production of the plasma hemoglobin scavenger haptoglobin and the heme scavenger hemopexin. Moreover, the antimicrobial effect of IL-22 depends on the induction of hemopexin expression, whereas haptoglobin was dispensable. Impaired pathogen clearance in infected Il22−/− mice was restored by hemopexin administration, and hemopexin-deficient mice had increased pathogen loads after infection. These studies reveal a previously unrecognized host defense mechanism regulated by IL-22 that relies on the induction of hemopexin to limit heme availability to bacteria, leading to suppression of bacterial growth during systemic infections.

INTRODUCTION

Iron is an essential nutrient for nearly all microorganisms, including pathogenic bacteria (1). In mammals, most of the iron is found inside erythrocytes, contained in the prosthetic heme groups of hemoglobin (2). Given their strict requirement for iron, pathogenic bacteria evolved strategies to trigger hemolysis as a means to extract iron from hemoglobin (1). Bacterial pathogens have several mechanisms to overcome the limited availability of iron in the circulation or at local niches. These include production of siderophores that bind iron at extremely high affinity, heme acquisition systems, and mechanism for the uptake of transferrin- and lactoferrin-bound iron (1). Host immunity encompasses several mechanisms that limit host iron availability to pathogenic bacteria (36), a defense strategy known as nutritional immunity (7). This host defense strategy involves a reduction of circulating or local iron by inhibition of cellular iron export and the induction of cellular iron import by host cells via hepcidin-dependent and hepcidin-independent mechanisms (36). Whether nutritional immunity limits pathogenic bacteria from accessing host heme, the major source of iron, remains unclear.

Interleukin-22 (IL-22) is an important cytokine that promotes early host defense, epithelial barrier function, and tissue repair at mucosal surfaces (8, 9). In response to bacterial infection, IL-22 is produced by several immune cells, leading to the induction of innate antimicrobial molecules that are thought to promote host defense and intestinal barrier protection against pathogens (10). The protective function of IL-22 was revealed by the observation that Il22−/− mice are highly susceptible to the attaching and effacing (A/E) mouse pathogen Citrobacter rodentium. In the absence of IL-22, there was marked intestinal damage and bacterial translocation in orally infected mice (8, 9). IL-22 signals through a heterodimeric receptor complex comprised a ubiquitously expressed IL-10R2 (IL-10 receptor 2) subunit and an epithelial-specific IL-22RA1 subunit (11, 12). Because of the highly restricted expression of the IL-22RA1 subunit, IL-22 stimulates epithelial cells on certain organs, such as the liver and kidney, as well as epithelial barriers, including the skin and the intestine (13). Using C. rodentium and Escherichia coli infection models, we report that IL-22 induces a systemic protective response that is mediated by hemopexin (HPX), a plasma heme scavenger produced in the liver, which limits the availability of heme iron to the microbes and suppresses bacterial systemic growth.

RESULTS

IL-22 promotes pathogen clearance and host survival after systemic infection with C. rodentium

Consistent with earlier studies (8, 9), Il22−/− mice succumbed early to oral infection with C. rodentium, which was associated with increased pathogen load in blood, liver, and spleen (fig. S1). Systemic C. rodentium infection induced rapid production of IL-22 in plasma (Fig. 1A). To determine whether IL-22 controls pathogen growth systemically, we infected wild-type (WT) and Il22−/− mice with C. rodentium intravenously and monitored them for survival and pathogen load in the blood. Surprisingly, >90% of the Il22−/− mice succumbed to intravenous infection compared with ~20% of the WT mice (Fig. 1B). Mortality of infected Il22−/− mice was associated with impaired pathogen clearance, resulting in increased pathogen load in blood (Fig. 1C). Pathogen-induced lethality in Il22−/− mice required live bacteria because the animals did not succumb when injected intravenously with heat-killed C. rodentium (fig. S2). Although antibody production is critical to control C. rodentium oral infection (14, 15), the production of pathogen-specific immunoglobulin M (IgM) and IgG was not impaired in Il22−/− mice, as compared with WT mice (fig. S3). To ascertain the role of IL-22 in controlling pathogen growth, we pretreated Il22−/− mice with recombinant IL-22 before intravenous C. rodentium infection and assessed the pathogen loads in the blood. Administration of IL-22 reduced pathogen load (Fig. 1D). In our mouse colony, ~90% of the bacteria present in the liver of orally infected Il22−/− mice were C. rodentium, although a few commensals, particularly Lactobacillus spp., were also identified (fig. S1G). To determine whether C. rodentium infection could induce mortality in the absence of commensals, we pretreated WT germ-free (GF) mice with an IL-22–neutralizing antibody and infected the mice with the pathogen via the oral route. Neutralization of IL-22 reduced the survival of GF mice after oral C. rodentium infection (Fig. 1E), which was associated with increased pathogen loads in blood, liver, and spleen (Fig. 1, F to H). Collectively, these results indicate that IL-22 limits the systemic expansion of C. rodentium and promotes host survival.

Fig. 1 IL-22 promotes pathogen clearance and host survival after systemic infection with C. rodentium.

(A) Production of IL-22 after a single intravenous injection of C. rodentium (5 × 107 CFU) into WT mice (n = 4). IL-22 amounts in plasma on indicated time points after injection are shown. (B) Survival of WT (n = 12) and Il22−/− (n = 17) mice after intravenous pathogen infection. Mice were injected daily with 5 × 107 CFU of C. rodentium from days 0 to 9 to mimic bacteremia resulting from oral infection of Il22−/− mice. (C) Pathogen loads in the blood on the indicated time points after systemic C. rodentium infection using bacteremia model used in (B). NS, not significant. (D) Effect of IL-22 administration on bacterial clearance. Il22−/− mice were treated with Fc–IL-22 or control protein (sham), and pathogen loads in blood were measured 18 hours after a single intravenous injection with 5 × 107 CFU of C. rodentium. (E) Mouse survival of GF mice pretreated with anti–IL-22–neutralizing antibody (Ab) (n = 11) or isotype-matched control antibody (n = 9) infected orally with C. rodentium (1 × 109 CFU). (F to H) Pathogen loads in the blood (F), liver (G), and spleen (H) of GF mice pretreated with anti–IL-22–neutralizing antibody or control antibody. Pathogen loads were measured on day 7 after oral C. rodentium infection. Data in (A) to (H) are representative of at least two independent experiments. Each circle represents one mouse (C, D, and F to H). Median values are indicated by a horizontal bar. *P < 0.05, **P < 0.01, ***P < 0.001, log-rank test (B and E) and Mann-Whitney test (C, D, and F to H). N.D., not detectable.

IL-22 regulates plasma HPX and haptoglobin after pathogen infection

To understand further the mechanism by which IL-22 promotes systemic pathogen clearance, we set to identify, by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) analysis, plasma proteins regulated by IL-22 before and after infection of WT and Il22−/− mice with C. rodentium intravenously. Consistent with previous studies (16, 17), several proteins including serum amyloid A-1, serum amyloid A-2, inter-α-trypsin inhibitor heavy chain, adiponectin, C-reactive protein, α-2-macroglobulin, and complement factors were reduced in the plasma of infected Il22−/− mice compared with infected WT mice (Fig. 2). In addition, the induction of the plasma extracellular heme scavenger HPX (18) and the hemoglobin scavenger haptoglobin (HP) (19) in response to C. rodentium infection was severely impaired in Il22−/− mice compared with WT mice (Fig. 2). This was associated with the accumulation of extracellular hemoglobin α- and β-globin chains in the plasma of infected Il22−/− mice compared with WT mice (Fig. 2). Intraperitoneal administration of IL-22 restored the induction of HP and HPX expression in the plasma of Il22−/− mice compared with Il22−/− mice treated with control protein (Fig. 2).

Fig. 2 IL-22–mediated regulation of plasma proteins after C. rodentium infection.

Heat map analysis of protein abundance in plasma samples from indicated mice. Protein analysis was performed by LC-MS/MS. (Left) Abundance of all detected proteins. (Right) Proteins increased (top) or decreased (bottom) in samples from infected Il22−/− mice compared with samples from infected WT mice or Il22−/− mice treated with Fc–IL-22 compared with Il22−/− mice treated with control protein. Plasma samples were collected from mice on day 7 after intravenous C. rodentium infection (infected WT and Il22−/− mice). Data are representative of two different experiments. HPX, HP, α-globin, and β-globin are boxed in light red.

Hemolysis driven by bacterial virulence factors promotes pathogen growth in Il22−/−mice

Accumulation of extracellular hemoglobin α- and β-globin chains in the plasma of Il22−/− mice suggested that IL-22 deficiency is associated with augmented hemolysis after systemic C. rodentium infection and/or with impaired disposal of extracellular hemoglobin from plasma. To verify that hemoglobin proteins are regulated by IL-22, we assessed the amounts of α-globin in the plasma of uninfected and infected WT and Il22−/− mice by immunoblotting. The analysis confirmed that the amounts of α-globin were increased in the plasma of Il22−/− mice in response to intravenous or oral infection with C. rodentium (Fig. 3, A and B). Because scavenging of extracellular hemoglobin dimers by HP prevents the release of the prosthetic heme groups of hemoglobin (20, 21), we asked whether induction of HP by IL-22 was associated with inhibition of heme accumulation in plasma. Il22−/− mice infected via the intravenous or oral route accumulated heme in plasma, whereas infected WT mice did not (Fig. 3C). Consistent with a role of hemoglobin release from lysed erythrocytes in promoting pathogen growth, intravenous administration of lysed erythrocytes to C. rodentium–infected WT mice increased their pathogen loads in the blood (Fig. 3D). The filament protein EspA as well as EspB and EspD are required for the assembly of the translocon of the enterocyte effacement (LEE)–encoded type III secretion system (T3SS) and promote the formation of EspB/EspD pores that mediate erythrocyte lysis in vitro (22). To assess whether C. rodentium translocators induce erythrocyte lysis, we incubated mouse erythrocytes in vitro with WT and C. rodentium mutants deficient in EspA, EspB, and Ler, the global regulator of the LEE that is critical for pathogen virulence (23). The WT bacterium, but not the EspA, EspB, and Ler mutants, induced erythrocyte lysis (Fig. 3E). To determine whether EspB is important for hemolysis associated with systemic growth and lethality of C. rodentium in vivo, we infected Il22−/− mice with WT or the EspB mutant and monitored the pathogen loads in blood as well as mouse survival. Deficiency of EspB was associated with reduced pathogen loads and the absence of lethality when compared with the WT bacterium that caused 100% mortality in Il22−/− mice (Fig. 3, F and G). Ler expression was detected in the blood of mice after intravenous infection with C. rodentium (fig. S4A). Furthermore, the reduced ability of the EspB mutant to grow systemically was not associated with reduced expression of ler in infected WT or Il22−/− mice (fig. S4B). In contrast to the EspB mutant, the pathogen load of a C. rodentium mutant deficient in EspH, a T3SS effector that regulates phagocytosis (24), was comparable with that of the WT bacterium in infected Il22−/− mice (fig. S5). The Ler mutant that regulates the expression of LEE-encoded proteins, including EspB, also exhibited reduced pathogen loads and failed to cause mortality in Il22−/− mice when compared with the WT bacterium (fig. S6). Growth of the EspB mutant was enhanced in vivo by intravenous administration of lysed erythrocytes (Fig. 3H). Furthermore, EspB and Ler promoted heme accumulation in plasma after intravenous infection of Il22−/− mice with C. rodentium (Fig. 3I and fig. S6). In contrast, the amounts of iron and the unsaturated iron-binding capacity in plasma before and after infection with C. rodentium were comparable in WT and Il22−/− mice (fig. S7). Thus, IL-22 regulates heme, but not free iron, accumulation in the plasma after systemic C. rodentium infection. These results indicate that hemolysis driven by LEE-encoded EspB translocator and ensuing release of heme in plasma promotes C. rodentium growth in vivo.

Fig. 3 Erythrocyte lysis and heme release via the EspB translocator promote pathogen growth in vivo.

(A and B) Detection of hemoglobin subunit α (HBα) in the plasma of uninfected and infected Il22−/− mice. Mice were infected intravenously (IV) using bacteremia model (A) or orally (B) with C. rodentium. Plasma samples (n = 3 mice for each group) were immunoblotted with an antibody against HBα. Ponceau S staining of the gel is also shown. (C) The amounts of heme in plasma were measured in uninfected mice and mice infected intravenously (left) or orally (right) with C. rodentium (C.r.). (D) Administration of lysed erythrocytes enhances pathogen growth in vivo. Il22−/− mice were co-injected intravenously with lysed red blood cells (RBCs) or vehicle, and 5 × 107 CFU of C. rodentium and pathogen loads in blood were measured 18 hours after infection. (E) Erythrocyte lysis induced by WT and indicated mutant C. rodentium strains. Data are means of quintuplicates ± SD. (F) Pathogen loads in blood in bacteremia model used in Fig. 1B. Il22−/− mice were infected intravenously with WT or espB mutant C. rodentium strains. Blood samples were collected at the indicated time points. “†” indicates that all mice succumbed. (G) Survival of Il22−/− mice in the bacteremia model after infection with WT (n = 6) or espB mutant (n = 6) C. rodentium strains. (H) Loads of espB mutant C. rodentium in blood of Il22−/− mice at 18 hours after infection with or without lysed RBCs. (I) The amounts of heme in plasma of indicated mice were measured after intravenous infection with WT or espB mutant C. rodentium strains. Data in (A) to (I) are representative of at least two independent experiments. Each circle (C, D, F, H, and I) represents one mouse. Median values are indicated by a horizontal bar. **P < 0.01, ***P < 0.001, Dunn’s test [C (left) and E], Mann-Whitney test [C (right), D, F, H, and I], and log-rank test (G).

IL-22dependent induction of HPX inhibits heme-mediated bacterial growth

Analysis of plasma proteins by LC-MS/MS revealed that HPX and HP are positively regulated by IL-22 in response to systemic C. rodentium infection (Fig. 2). Immunoblotting analysis of plasma proteins confirmed that IL-22 is required for induction of HPX and HP in response to intravenous and oral C. rodentium infection (Fig. 4, A to C). The IL-22R is expressed on epithelial cells including hepatocytes, the major site of HPX and HP production (25, 26). Consistently, intraperitoneal administration of IL-22 induced HPX and HP mRNA in the liver (fig. S8). To determine whether C. rodentium infection induces antibacterial activity in plasma, we monitored C. rodentium colonies in vitro after incubation with plasma from uninfected and infected mice. Plasma from infected WT mice inhibited C. rodentium colony formation compared with plasma of uninfected WT mice (Fig. 4D). Plasma from infected Il22−/− mice failed to limit colony formation compared with the plasma of infected WT mice (Fig. 4D). Consistently, intravenous administration of recombinant IL-22 to Il22−/− mice restored the plasma antibacterial activity (Fig. 4E). To identify the mechanism underlying this antibacterial activity in plasma, we fractionated plasma proteins from IL-22–treated mice by column chromatography and assessed the pathogen-inhibitory activity in the fractions by agar plating assay. Immunoblotting analysis showed that the presence of the heme scavenger HPX, but not the hemoglobin scavenger HP, correlated with the ability of the plasma fractions to suppress C. rodentium colony formation (Fig. 4F). To determine whether heme promotes C. rodentium growth, we incubated the pathogen in vitro with and without synthetic heme (i.e., hemin) in the presence of increasing concentrations of HPX. In the absence of HPX, hemin promoted robust C. rodentium colony formation in vitro (Fig. 4G), whereas addition of HPX, at a concentration range found in mouse plasma, that is, ~500 to 1000 μg/ml (27), inhibited hemin-driven pathogen growth in a dose-dependent manner (Fig. 4G). Hemoglobin also promoted robust C. rodentium colony formation in vitro, which was inhibited by HPX in a dose-dependent manner (Fig. 4H), whereas HP did not (fig. S9). Furthermore, exogenous HPX largely restored the antibacterial activity of the plasma isolated from infected Il22−/− mice in vitro (Fig. 4I). Collectively, these results suggest that once released from extracellular hemoglobin, heme promotes C. rodentium growth, a pathogenic effect countered via a mechanism mediated by the IL-22–dependent induction of the plasma heme scavenger HPX.

Fig. 4 Induction of HPX by IL-22 is essential for inhibition of C. rodentium growth in the presence of heme.

(A and B) Detection of HPX and HP in the plasma of uninfected and infected animals. WT (n = 2 to 3 per group) or Il22−/− (n = 2 to 3 per group) mice were left uninfected or infected with WT C. rodentium intravenously (A) or orally (B), and the plasma samples were collected and immunoblotted with antibodies against HPX (Hpx) and HP (Hp). Staining of gels with Ponceau S is also shown. (C) Detection of HPX and HP in the plasma of Il22−/− mice left untreated (−) or treated with recombinant Fc–IL-22. (D) Growth of C. rodentium in the presence of plasma from WT and Il22−/− mice. Bacterial growth was measured by plating. (E) Growth of C. rodentium in the presence of plasma from Il22−/− mice treated with Fc–IL-22 or control protein (sham). Experiments were performed as in (D). (F) Mice were treated with Fc–IL-22 or control protein, and plasma samples were fractionated by column chromatography. Sequential fractions were pooled into six fractions that were analyzed for their ability to promote C. rodentium colony formation and for the presence of HPX or HP by immunoblotting. Analysis of unfractionated plasma samples is shown on the right panel. (G) Hemin promotes C. rodentium colony formation, which is inhibited by HPX. Bacteria were incubated with and without hemin in the presence of indicated concentrations of HPX. Bacterial growth was measured by plating. Data are means of sextuplicate cultures ± SD. (H) Hemoglobin promotes C. rodentium colony formation, which is inhibited by HPX. Bacteria were incubated with and without hemoglobin in the presence of indicated concentrations of HPX. Bacterial growth was measured by plating. (I) Addition of HPX enhances the ability of plasma from Il22−/− mice to inhibit C. rodentium colony formation. Data in (D) and (E) represent means of six different mouse samples ± SD. Data in (G) and (H) represent means of six technical replicates. Data in (A) to (I) are representative of at least two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, Dunn’s test (D and I) and Mann-Whitney test [C (right) and E to H].

IL-22 limits heme-mediated E. coli growth via HPX

E. coli, a Gram-negative bacterium that normally inhabits the intestine, is a major cause of bacteremia and sepsis in humans (28, 29). As it was observed with C. rodentium, hemin promoted robust colony formation of an E. coli strain isolated from the mouse intestine that was inhibited by HPX in vitro (Fig. 5A). Furthermore, administration of recombinant IL-22 to Il22−/− mice restored the ability of plasma to suppress E. coli colony formation in vitro (Fig. 5B), which was strictly dependent on the expression of endogenous HPX (Fig. 5C). Furthermore, Il22−/− and Hpx−/− mice were more susceptible and had more bacterial loads than WT mice after intravenous infection with E. coli (Fig. 5, D to G). These results indicate that E. coli colony formation is promoted by heme, which is inhibited by IL-22–mediated induction of HPX.

Fig. 5 Induction of HPX by IL-22 inhibits E. coli growth.

(A) Hemin promotes E. coli growth in vitro, which is inhibited by HPX. Bacteria were incubated with and without hemin in the presence of indicated concentrations of HPX. Bacterial growth was measured by plating. Data are means of quadruplicate cultures ± SD. (B) Growth of E. coli in the presence of plasma from Il22−/− mice treated with Fc–IL-22 or control protein (sham). Data are means ± SD. (C) Role of HPX in IL-22–mediated inhibition of E. coli growth. Growth of E. coli in the presence of plasma from WT and Hpx−/− mice pretreated with Fc–IL-22 or control protein (−). Data are means of 13 technical replicates ± SD. (D and E) Bacterial loads in the blood after a single intravenous injection of E. coli. Blood samples were collected from WT (D and E), Il22−/− (D), and Hpx−/− mice (E) at 18 hours after infection. (F and G) Survival of WT (n = 9) and Il22−/− mice (n = 11) (F) and WT (n = 15) and Hpx−/− mice (n = 15) (G) after intravenous E. coli infection. Data in (A) to (E) are representative of at least two independent experiments. Data in (F) and (G) are results from two pooled experiments. Each symbol represents one mouse. Median values are indicated by a horizontal bar. *P < 0.05, Mann-Whitney test (A, B, D, and E); ***P < 0.001, Dunn's test (C); *P < 0.05, log-rank test (F and G).

IL-22 requires HPX, but not HP, to suppress pathogen growth in vitro and in vivo

We next assessed the requirement for HPX and HP in IL-22–mediated suppression of bacterial growth. The ability of plasma from IL-22–treated WT mice to suppress C. rodentium colony formation was comparable with that of plasma from Hp−/− mice treated with IL-22 (Fig. 6A). In contrast, the plasma of Hpx−/− and Hpx−/−Hp−/− mice treated with IL-22 was impaired in suppressing C. rodentium growth, which was comparable with the plasma of untreated WT mice (Fig. 6, B and C). To confirm that IL-22 suppresses pathogen growth in vivo via a mechanism involving heme neutralization by HPX, we intraperitoneally administered IL-22 to WT, Hpx−/−, Hp−/−, and Hpx−/−Hp−/− mice and then infected them with C. rodentium intravenously. Whereas administration of IL-22 induced a comparable reduction of pathogen loads in the blood of WT and Hp−/− mice (Fig. 6D), this antibacterial effect was lost in Hpx−/− and Hpx−/−Hp−/− mice (Fig. 6, E and F). Thus, IL-22 requires HPX, but not HP, to suppress pathogen growth in vitro and in vivo. To assess whether HPX is sufficient to promote pathogen clearance in the absence of IL-22 administration, we infected WT, Hpx−/−, Hp−/−, and Hpx−/−Hp−/− mice with C. rodentium intravenously and monitored pathogen loads and mouse survival in infected mice. There was impaired pathogen clearance in Hpx−/− and Hpx−/−Hp−/− mice, but not Hp−/− mice, compared with WT mice (Fig. 6G). Furthermore, administration of HPX reduced the pathogen load in Il22−/− mice compared with mock-treated mice after intravenous (Fig. 6H) and oral infection (Fig. 6, I and J). Together, these results indicate that IL-22 limits the growth of C. rodentium via HPX in vivo.

Fig. 6 HPX, but not HP, is essential for inhibition of bacterial growth in vivo.

(A to C) Role of HPX and HP in IL-22–mediated inhibition of C. rodentium growth. Growth of C. rodentium in the presence of plasma samples from WT, Hp−/−, Hpx−/−, or Hp−/−Hpx−/− mice pretreated with Fc–IL-22 or control protein (−). Data are means of at least six different mouse samples ± SD. (D to F) WT, Hp−/−, Hpx−/−, or Hp−/−Hpx−/− mice were treated with Fc–IL-22 or control protein and infected with 5 × 107 CFU of C. rodentium intravenously after Fc–IL-22 or control protein administration. Pathogen loads in blood 18 hours after infection are shown. (G) Pathogen loads in the blood of indicated mice after C. rodentium infection using the bacteremia model shown in Fig. 1B. Pathogen loads in blood at the indicated time points are shown. (H) HPX or saline (sham) was injected intravenously into Il22−/− mice, and then the mice were infected with 5 × 107 CFU of C. rodentium intravenously. Pathogen loads in blood were measured 18 hours after infection. (I and J) Il22−/− mice were infected orally with 109 CFU of C. rodentium, and purified HPX or saline (sham) was injected intravenously into the mice on days 3 to 6 after infection. Pathogen loads in blood (H and I) and spleen (J) were measured on day 7 after infection. Each symbol represents one mouse (D to J). Median values are indicated by a horizontal bar (D to J). Data in (A) to (J) are representative of at least two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, Dunn's test (A to G) and Mann-Whitney test (H to J).

DISCUSSION

IL-22 has been primarily linked to the regulation of host defense, cellular proliferation, and tissue repair at intestinal barriers (10). In the current studies, we provide evidence for a role of IL-22 in protecting the host against the enteric pathogen C. rodentium and the commensal E. coli by mediating the production of HPX in plasma. Although produced by immune cells in response to microbial stimuli, IL-22 stimulates its heterodimeric IL-22RA1/IL-10R2 on epithelial cells, including hepatocytes that are known to produce large amounts of HPX (18). Induction of HPX in response to systemic infections was so far thought to protect the infected host from heme-induced cell toxicity, inflammation, and multiorgan dysfunction (30), without interfering with the host pathogen load and, as such, conferring disease tolerance to systemic bacterial infections (30, 31). This protective effect of HPX and its removal by macrophages and parenchymal cells have also been extended to several other pathologies associated with varying degrees of hemolysis (25). Our work demonstrates that IL-22–induced HPX also confers resistance to systemic bacterial infections by limiting the heme availability to bacteria. Pathogens have several mechanisms to overcome iron-limiting defenses induced by the host, including siderophores, hemophores, and heme/hemoprotein receptors (1). Thus, pathogens capable of colonizing iron-poor niches are likely to be less susceptible to IL-22–mediated HPX induction. C. rodentium and E. coli appear to rely on strategies to acquire iron from heme when they colonize the plasma and systemic organs. C. rodentium and E. coli produce enterobactin, a siderophore, and its receptor FepA to acquire iron and may express functional heme uptake systems (32, 33). Our results suggest that C. rodentium promotes hemolysis and heme acquisition through pore formation via T3SS-mediated factors and specifically EspB/EspD. EspH, a T3SS effector that regulates phagocytosis, was dispensable for pathogen growth in Il22−/− mice. These findings suggest that the T3SS promotes bacterial growth by inducing hemolysis. However, there may be additional factors, including compromised ability to block phagocytosis, that might contribute to the impaired ability of T3SS mutants to infect the mice. Further work is needed to understand how C. rodentium and E. coli acquire iron from heme in vivo.

Our studies indicate that pathogenic bacteria like A/E pathogens take advantage of virulence strategies such as expression of Ler-dependent factors normally used to colonize the intestinal epithelium to survive in alternative niches such as the blood in the event of causing bacteremia. Administration of IL-22–binding protein (IL-22–BP), a natural antagonist of IL-22 signaling, has been reported to enhance the systemic clearance of commensal bacteria in a model of polymicrobial sepsis (34). Further work is needed to determine whether the protective function of IL-22–BP is mediated through the induction of HPX and/or another mechanism. Our results suggest that strategies to scavenge heme in plasma, such as that afforded by the administration of IL-22, HPX, or other approaches, may be beneficial in the treatment of systemic bacterial infections.

MATERIALS AND METHODS

Study design

The aim of this study was to elucidate and characterize the mechanism by which IL-22 mediates systemic protection against bacterial infection using the C. rodentium and E. coli models in mice. The experimental design involved in vivo and in vitro experiments, including protein identification by LC-MS/MS, chromatographic fractionation of plasma proteins, immunoblotting, histological analysis, reverse transcription polymerase chain reaction analysis, and bacterial colony enumeration. The animal experiments were not randomized. The investigators were not blinded to the allocation during experiments and analyses unless otherwise indicated. Experimental replication is indicated in the figure legends.

Animals

Six- to 8-week-old WT C57BL6 mice were bred in our animal facility. Il22−/−, Hp−/−, Hpx−/−, and Hp−/−Hpx−/− mice on C57BL6 background have been described (35, 36) and bred under specific pathogen–free conditions at the University of Michigan Comprehensive Cancer Center. Il22−/− mice were obtained from Genentech. GF mice were bred and maintained at the GF Animal Core Facility of the University of Michigan. The animal studies were conducted under protocols approved by the University of Michigan Committee on Use and Care of Animals.

C. rodentium infection

Kanamycin-resistant WT C. rodentium strain DBS120 (pCRP1::Tn5) was a gift of D. Schauer, Massachusetts Institute of Technology. The isogenic C. rodentium Δler mutant strain has been described (37). C. rodentium strains with nonpolar deletion mutants of espA, espB, and espH have been described (23). For inoculations, bacteria were grown overnight in Luria-Bertani (LB) broth supplemented with kanamycin (50 μg/ml) with shaking at 37°C. Mice were infected by oral gavage with 0.2 ml of phosphate-buffered saline (PBS) containing about 1 × 109 colony-forming units (CFU) of C. rodentium, or by intravenous injection with 0.2 ml of PBS containing about 5 × 107 CFU of C. rodentium. The investigators were not blinded to allocation during experiments. To mimic the continuous source of bacteremia observed in Il22−/− mice orally infected with the pathogen, we infected mice with 5 × 107 CFU of C. rodentium intravenously daily for 9 to 10 days. To assess the role of HPX in C. rodentium clearance in vivo, we orally infected Il22−/− mice with the pathogen (1 × 109 CFU) and treated them intravenously with purified HPX (Athens Research and Technology; 2 mg per mouse) or saline on days 3 to 6 after infection. For systemic infections, HPX (2 mg per mouse) or saline (sham) was injected intravenously on three consecutive days before infection or control (sham) into Il22−/− mice, and then the mice were infected with 5 × 107 CFU of C. rodentium intravenously. To determine pathogen loads in the feces, we collected fecal pellets from individual mice, homogenized them in cold PBS, and plated them at serial dilutions onto MacConkey agar plates containing kanamycin (50 μg/ml), and we determined the number of CFU after overnight incubation at 37°C. To determine bacterial number in the blood, liver, and spleen, we sacrificed mice at various time points after infection, plated blood or tissue samples onto MacConkey agar plates containing kanamycin (50 μg/ml), and determined the number of CFU after overnight incubation at 37°C.

IL-22 neutralization and administration

To deplete endogenous IL-22 in mice, we injected anti–IL-22 (150 μg per mouse per dose; 8E11; Genentech) or the equivalent amount of mouse isotype-matched control antibody (to ragweed; 10D9.1E11.1F12; Genentech) to the mice three times a week. To treat mice with IL-22, we intravenously injected them with Fc–IL-22 (50 μg per dose; PRO312045; Genentech) or the equivalent amount of mouse isotype-matched control antibody (to ragweed; 10D9.1E11.1F12; Genentech) daily for 3 days and 6 hours after from final injection.

E. coli infection

E. coli strain NI1076 was isolated from the mouse intestine mice, as described (17). For infections, bacteria were grown overnight in LB broth with shaking at 37°C. Mice were infected by intravenous injection with 0.2 ml of PBS containing about 2 × 108 CFU of E. coli.

Bacterial growth assays

C. rodentium or E. coli (1 × 105 CFU/ml) were incubated in Dulbecco’s modified Eagle’s medium in the presence of Apo-transferrin (500 μg/ml) (Athens Research and Technology) to chelate free iron. To assess bacterial growth, we added hemin (2.5 μg/ml; Sigma) or hemoglobin (25 μg/ml; Sigma) to the culture medium in the presence and absence of different concentrations of HPX (Athens Research and Technology) or HP (Athens Research and Technology).

Statistical analysis

Statistical significance was calculated as indicated in the figure legends using GraphPad Prism 6 software. Log-rank test was used to assess mouse survival. Nonparametric Mann-Whitney test was used for pair comparisons, and Dunn’s test was used for multiple comparisons. Differences were considered significant when P values were less than 0.05.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/8/eaai8371/DC1

Materials and Methods

Fig. S1. Increased pathogen loads in blood and organs of Il22−/− mice orally infected with C. rodentium.

Fig. S2. Lack of lethality by heat-killed C. rodentium.

Fig. S3. Induction of pathogen-specific IgM and IgG is not impaired in Il22−/− mice.

Fig. S4. ler expression of WT and espB mutant of C. rodentium in vivo.

Fig. S5. EspH is not required for systemic growth of C. rodentium during infection.

Fig. S6. ler-regulated factors are essential for C. rodentium–induced hemolysis and lethality in Il22−/− mice.

Fig. S7. IL-22 does not regulate free iron and unsaturated iron-binding capacity in plasma.

Fig. S8. Hpx and Hp expression are induced by IL-22 in the liver.

Fig. S9. HP does not affect hemoglobin-promoted C. rodentium growth in vitro.

Table S1. Excel file containing tabulated data for Figs. 1 to 6 and fig. S1.

References (38, 39)

REFERENCES AND NOTES

Acknowledgments: We thank L. Haynes for animal husbandry, J. Pickard for manuscript review, V. Basrur for help with MS analysis, and Genentech for providing mutant mice and IL-22 reagents. We thank the Department of Pathology Proteomics Resource Facility, University of Michigan Host Microbiome Initiative and the University of Michigan GF Core Facility for support. Funding: K.S. was supported by fellowships from the Japan Society for the Promotion of Science, the Kanae Foundation for the Promotion of Medical Science, and the Mishima Kaiun Memorial Foundation. H.H. was supported by a fellowship for Research Abroad from the Japan Society for the Promotion of Science. G.C.-F. was supported by a postdoctoral fellowship from the Consejo Nacional de Ciencia y Tecnología of Mexico (CONACYT) (454848). This work was supported by NIH grants DK091191 and DK095782 (G.N.) and Fundação Calouste Gulbenkian, Fundação para a Ciência e a Tecnologia (PTDC/SAU TOX/116627/2010 and HMSP-ICT/0022/2010), and the European Community 7th Framework (ERC-2011-AdG 294709-DAMAGECONTROL) (M.P.S.). Author contributions: K.S., N.I., and G.N. designed the research. K.S. and G.N. wrote the manuscript. K.S. conducted the experiments, analyzed data, and performed statistical analyses with the help from H.H., N.K., Y.-G.K., N.I., and G.N. G.C.-F., E.T., M.P.S., and J.L.P. generated and provided critical material. All authors discussed the results and commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The proteomics data have been deposited to the ProteomeXchange Consortium database under accession number PXD005666. IL-22–, HPX-, and HP-deficient strains of mice were obtained under material transfer agreements from Genentech Inc. and the University of Torino. Requests for IL-22–deficient mice strains should be addressed to Genentech Inc. Requests for HPX- and HP-deficient strains of mice should be addressed to G.N. and E.T.
View Abstract

Navigate This Article