Article Text

Download PDFPDF

Original article
Gut microbial translocation corrupts myeloid cell function to control bacterial infection during liver cirrhosis
  1. Carl-Philipp Hackstein1,
  2. Lisa Mareike Assmus1,
  3. Meike Welz1,
  4. Sabine Klein2,
  5. Timo Schwandt1,
  6. Joachim Schultze3,
  7. Irmgard Förster4,
  8. Fabian Gondorf4,
  9. Marc Beyer3,
  10. Daniela Kroy5,
  11. Christian Kurts1,
  12. Jonel Trebicka2,
  13. Wolfgang Kastenmüller1,
  14. Percy A Knolle1,6,
  15. Zeinab Abdullah1
  1. 1Institute of Experimental Immunology, Universität Bonn, Bonn, Germany
  2. 2Department of Internal Medicine I, Universität Bonn, Bonn, Germany
  3. 3Genomics and Immunoregulation, LIMES Institute, Universität Bonn, Bonn, Germany
  4. 4Immunology and Environment, LIMES Institute, Universität Bonn, Bonn, Germany
  5. 5Department of Medicine III, University Hospital Aachen, Aachen, Germany
  6. 6Institute of Molecular Immunology and Experimental Oncology, Technische Universität München, München, Germany
  1. Correspondence to Professor Percy A Knolle, Institute of Molecular Immunology, Klinikum München rechts der Isar, Technische Universität München, Ismaningerstr. 22, München 81675, Germany; percy.knolle{at}


Objective Patients with liver cirrhosis suffer from increased susceptibility to life-threatening bacterial infections that cause substantial morbidity.

Methods Experimental liver fibrosis in mice induced by bile duct ligation or CCl4 application was used to characterise the mechanisms determining failure of innate immunity to control bacterial infections.

Results In murine liver fibrosis, translocation of gut microbiota induced tonic type I interferon (IFN) expression in the liver. Such tonic IFN expression conditioned liver myeloid cells to produce high concentrations of IFN upon intracellular infection with Listeria that activate cytosolic pattern recognition receptors. Such IFN-receptor signalling caused myeloid cell interleukin (IL)-10 production that corrupted antibacterial immunity, leading to loss of infection-control and to infection-associated mortality. In patients with liver cirrhosis, we also found a prominent liver IFN signature and myeloid cells showed increased IL-10 production after bacterial infection. Thus, myeloid cells are both source and target of IFN-induced and IL-10-mediated immune dysfunction. Antibody-mediated blockade of IFN-receptor or IL-10-receptor signalling reconstituted antibacterial immunity and prevented infection-associated mortality in mice with liver fibrosis.

Conclusions In severe liver fibrosis and cirrhosis, failure to control bacterial infection is caused by augmented IFN and IL-10 expression that incapacitates antibacterial immunity of myeloid cells. Targeted interference with the immune regulatory host factors IL-10 and IFN reconstitutes antibacterial immunity and may be used as therapeutic strategy to control bacterial infections in patients with liver cirrhosis.

View Full Text

Statistics from

Significance of this study

What is already known on this subject?

  • Patients with liver cirrhosis show increased susceptible to bacterial infection.

  • Infection-associated morbidity is high in patients with liver cirrhosis and is one of the major mortality factors.

  • Bacterial translocation from the gut is increased during liver cirrhosis.

  • Liver serves as filter to eliminate bacteria from the blood.

What are the new findings?

  • Presence of tonic type I interferon (IFN) signalling in livers of mice with liver fibrosis that is triggered by translocation of gut microbiota.

  • Bacterial infection eliciting cytosolic pattern recognition receptors elicit massive IFN expression that in turn leads to interleukin (IL)-10 production in myeloid cells from mice with liver fibrosis.

  • Key findings in mice are confirmed with human material showing an IFN signature in cirrhotic liver and IL-10 production by myeloid cells upon bacterial infection.

  • Interference with IFN or IL-10 receptor signalling rescues from infection-associated mortality after bacterial infection during liver fibrosis.

How might it impact on clinical practice in the foreseeable future?

  • Identification of patients at risk for developing uncontrolled bacterial infection by measuring IL-10 production in myeloid cells—thus providing a novel biomarker that is causally related to infection-associated mortality.

  • Targeted intervention of regulatory IFN and IL-10 receptor signalling on the host side to accompany antibiotic pharmacotherapy to overcome intractable bacterial infection.


Liver fibrosis and cirrhosis are the outcome of chronic liver damage originating from various aetiologies such as chronic viral hepatitis, alcohol abuse, steatohepatitis and cholestasis, among others. Fibrosis in the liver is characterised by excessive extracellular matrix deposition that upon continuous liver damage will result in liver cirrhosis, leading to portal hypertension that causes complications such as the formation of ascites and oesophageal varices.1 Translocation of gut microbiota and delivery of viable bacteria or bacterial degradation products are observed during liver cirrhosis. Although the liver functions as a filter for gut-derived bacteria that escaped surveillance by immune cells located in the gut, systemic spread of such translocated microbiota is facilitated by portosystemic shunts that deviate portal blood from the liver to the systemic circulation because of portal hypertension.2–5 Liver cirrhosis in patients is accompanied by immune dysfunction that leads to increased susceptibility to intractable bacterial infections, especially spontaneous bacterial peritonitis, pneumonia and urinary tract infection. More than 30% of patients requiring hospitalisation suffer from bacterial infections, and this incidence increases to >45% during complications of cirrhosis such as ascites formation or GI bleeding from oesophageal varices.6–9 Bacterial infections are the leading cause of mortality in patients with liver cirrhosis.10 Previous reports have found an impaired functionality of several immune cell populations7 ,11 ,12 and others have reported an involvement of interleukin (IL)-10 as a factor contributing to reduced antibacterial immunity during liver inflammation.13 Yet, the mechanistic basis of immune dysfunction during liver cirrhosis that is responsible for the increased susceptibility towards infections with pathogenic bacteria awaits identification.

The healthy liver is a site of active local regulation of immune responses favouring induction of T cell tolerance rather than immunity in a process determined by tolerogenic organ-resident cells and the unique hepatic microenvironment.14 The role of local immune regulation in the liver for clearance or persistence of hepatitis virus or parasites through the adaptive immune response has been intensively studied. In addition, the contribution of potent innate immune response in the liver during infection with different pathogens has been unravelled.15 ,16 However, it has remained unclear how innate immunity is regulated in the already diseased liver and how this impacts on pathogen immune defence. Here, we report that gut microbial translocation during experimental liver fibrosis following bile duct ligation (BDL) leads to induction of type I interferon (IFN) in myeloid cells. Infection with pathogenic bacteria that elicit activation of cytosolic pattern recognition receptors (PRRs) results in enhanced IFN and IL-10 expression in those myeloid cells from mice with experimental liver fibrosis but also humans with liver cirrhosis. This causes a severe impairment of antibacterial innate immunity and development of infection-associated mortality. Blockade of signalling through the IFN receptor or the IL-10 receptor rescues antibacterial innate immune functions of myeloid cells during liver fibrosis and allows for clearance of pathogenic bacteria as well as prevention of infection-associated mortality.


Failure to control Listeria monocytogenes infection in mice with liver fibrosis following BDL and CCl4 application

We characterised the consequences of chronic liver damage associated with fibrosis and cirrhosis for innate immune control of infection with pathogenic intracellular bacteria. To this end, we performed BDL in mice17 that resulted in experimental liver fibrosis (see online supplementary figure S1A) and then infected fibrotic mice with the pathogenic intracellular bacteria Listeria monocytogenes. In contrast to control-operated or untreated mice, all mice with liver fibrosis succumbed to infection with an LD50 of L monocytogenes (5×104) already by day 6 post infection (p.i.) (figure 1A), indicating deficient innate immunity that fails to control bacterial infection. Even after infection with a 10-fold lower sublethal dose of L monocytogenes (5×103), 30% of fibrotic mice succumbed to infection (figure 1B). While healthy mice controlled bacterial load from day 2 p.i., mice with liver fibrosis did not control bacterial infection in spleen and liver, resulting in a 25-fold higher bacterial load in the liver compared with healthy mice (figure 1C). We detected a similarly enhanced susceptibility to L monocytogenes infection and infection-associated mortality in a second fibrosis model, that is, CCl4-induced liver fibrosis (figure 1D, E and online supplementary figure S1B). Time kinetic analyses revealed that liver damage measured as increase in serum alanine transaminase (ALT)/aspartate transaminase (AST) levels after BDL or CCl4 application declined over time and was reduced to low levels at the time of bacterial infection (figure 1F and online supplementary figure S1C). In both models, we observed a sharp increase in liver damage, reflected by ALT and AST, levels shortly after infection (figure 1F and online supplementary figure S1C). While increased IL-6 and tumour necrosis factor serum levels found in fibrotic mice after bacterial infection could still elicit hepatocyte production of the acute-phase protein C reactive protein (CRP) (figure 1G, H and online supplementary figure S1D,E), we also observed a drop in serum albumin levels over the first days after bacterial infection, indicating loss of liver synthesis function (figure 1I and online supplementary figure S1F). These changes resemble the clinical situation when acute or chronic liver failure occurs, which is frequently triggered by bacterial infections.18 ,19

Figure 1

Impaired antibacterial immunity in fibrotic liver. (A and B) Kaplan-Meier survival curves for naive, sham or bile duct ligation (BDL)-operated mice infected with Listeria monocytogenes (L.m.) at LD50 (A) or with a sublethal dose (5×103 CFU) (B); n=10 mice per group. (C) L.m. burden in liver (left) and spleen (right) of sham or BDL-operated mice infected as in (A). (D and E) Kaplan-Meier survival curves for naive, oil or CCl4-treated mice infected as in (A). (F) Time kinetic of alanine transaminase (ALT) and aspartate transaminase (AST) in the serum of mice treated as in (A). (G) ELISA for interleukin (IL)-6 and tumour necrosis factor of mice treated as in (A) at the indicated time points after infection. (H, I) Time kinetic of C reactive protein (CRP) and albumin measured in the serum of mice treated as in (A) or (D). Symbols represent individual mice, mean±SEM. Data are representative of 2–5 independent experiments. *p<0.05; **p<0.01.

Thus, like in patients with liver cirrhosis, experimental liver fibrosis in mice is associated with impaired antibacterial immunity and increased infection-associated mortality, which allowed us to study the cellular and molecular mechanisms responsible for the failure of the immune system to control bacterial infection in experimental liver fibrosis.

Impaired innate antibacterial immunity of myeloid cells during liver fibrosis

Since phagocytic myeloid cells are key for control of infection with intracellular bacteria, we determined their number and functionality in fibrotic compared with healthy livers. In experimental liver fibrosis, we found increased numbers of monocytes and other CD11b cells (likely granulocytes) by immune-fluorescence of liver tissue (figure 2A) and determination of their numbers by flow cytometry (figure 2B). Upon infection, there was a severe drop in the numbers of granulocytes found in fibrotic liver compared with mice with healthy livers. In contrast, monocyte numbers remained high (figure 2B). Detailed phenotypic characterisation and quantitative assessment by flow cytometry revealed that after bacterial infection total numbers of CD11b+F4/80negLy6C+ inflammatory monocytes, Ly6G+ granulocytes and CD11c+ dendritic cells in the fibrotic livers were comparable to those in healthy livers. CD11b+F4/80+Ly6Cneg liver-resident macrophages were reduced in the fibrotic livers (figure 2B and online supplementary figure S2A, B). This suggests that the failure to control bacterial infection is the result of functional impairment of monocytes/macrophages that are present in high numbers in infected fibrotic livers in combination with a reduction in the numbers of granulocytes.

Figure 2

Myeloid cell innate immune functions during bacterial infection in liver fibrosis. (A) Liver immune fluorescence staining. (B) Flow cytometric analysis of liver myeloid cells of on day 3 post infection (left) and quantification (right). (C) Phagocytosis of viable Listeria monocytogenes (L.m.) and (D) killing of intracellular L.m. in vitro by Kupffer cells and monocytes isolated from sham or bile duct ligation (BDL)-operated mice. (E) Reactive oxygen species production in vitro by CD11b+ cells isolated from liver or spleen of sham or BDL-operated mice. (F) Liver IL-12p35 and interferon (IFN)γ mRNA expression and ELISA for interleukin (IL)-1β from liver tissue lysates. Symbols represent individual mice, mean±SEM. Data are representative of 2–5 independent experiments. *p<0.05; **p<0.01.

Both macrophages that are known to be highly active in phagocytosis of viable L monocytogenes and monocytes that have less pronounced phagocytosis when isolated from fibrotic liver showed reduced phagocytic uptake of bacteria (figure 2C and online supplementary figure S2C). Even after phagocytic uptake of bacteria, these cells failed to control intracellular bacterial growth (figure 2D and online supplementary figure S2D). Important for control of infection with bacteria in monocytes and macrophages is the production of reactive oxygen species (ROS).20 ,21 ROS production upon infection in vitro was reduced in myeloid cells from fibrotic livers but not the spleen (figure 2E), indicating a liver-specific mechanism regulating innate immune functions of myeloid cells. The impaired bactericidal function of myeloid cells suggested reduced expression of cytokines that operate to enhance bactericidal activity, namely IL-12, IFNγ and IL-1β.22 ,23 Consistent with this assumption, we found reduced expression of IL-12p35, IFNγ mRNA and IL-1β protein in fibrotic compared with healthy livers after L monocytogenes infection (figure 2F), but no reduction in spleen or blood (see online supplementary figure S2E–H). Thus, liver monocytes and macrophages show impaired innate immune functions and fail to control infection with pathogenic bacteria during liver fibrosis, prompting us to search for liver-specific mechanisms.

Tonic type I IFN expression in fibrotic and cirrhotic livers

Previous studies have reported type I IFN to exert detrimental effects and to exacerbate infections by different pathogenic bacteria, including the intracellular pathogens L monocytogenes24–26 and Mycobacterium tuberculosis.27 Strikingly, we found enhanced tonic expression of type I IFN in fibrotic livers, exemplary shown by detection of IFNβ and increased expression levels of IFN-induced genes such as Mx1 (figure 3A), Oas1a, Oas2, Oasl1, Oas1g and Ptgs2 in fibrotic livers (figure 3B). Immunohistochemistry showed increased Mx1 expression and confirmed functional IFN signalling at protein level. Mx1 expression was selectively induced in myeloid cells in fibrotic liver, whereas upon poly I:C application widespread Mx1 expression also in hepatocytes was observed (figure 3C and online supplementary figure S3A, B). To identify the cellular source of IFN, we sorted immune cell populations from BDL-operated mice and identified myeloid cells, in particular F4/80+ macrophages and dendritic cells, as main cell populations expressing IFNβ (figure 3D). Importantly, we also found increased IFNβ expression levels in liver tissue from patients with liver cirrhosis of non-viral origin (figure 3E) that was accompanied by upregulation of IFN-induced genes (figure 3F). This demonstrates tonic IFN production in fibrotic murine liver and cirrhotic human livers and identifies liver myeloid cells as the main source and target of IFN production.

Figure 3

Tonic type I interferon (IFN) expression during liver fibrosis/cirrhosis in mice and humans. (A) Time kinetics of IFNβ or Mx1 mRNA expression. (B) Liver mRNA expression of IFN-inducible genes Oas1a, Oas2, Oasl1, Oas1g and Ptgs2 2 weeks after sham- or bile duct ligation (BDL). (C) Liver immunofluorescence on day 10 post BDL or healthy mice±polyI:C. (D) IFNβ mRNA expression in F4/80+ macrophages, CD11c+ dendritic cells, Ly6C+ monocytes and CD146+ liver sinusoidal endothelial cells (LSEC) sorted from livers of sham or BDL-operated mice. (E) IFNβ mRNA expression in livers of healthy donors or patients with cirrhosis. (F) mRNA expression of OAS1a, OAS2, OAS3, OASL, MX2 and PTGS2 in livers of healthy donors or patients with cirrhosis. Symbols represent individual mice and mean±SEM. Data are representative of 3–5 independent experiments. *p<0.05; **p<0.01; ***p<0.001.

Translocated gut microbiota induce tonic IFN signalling in fibrotic liver and predispose to failure to control bacterial infection

Next, we searched for the molecular cues that triggered tonic IFN expression in the fibrotic liver. A known consequence of liver fibrosis is translocation of gut microbiota into the circulation entering the liver via portal venous blood.28 ,29 Immune recognition of bacteria or bacterial components by myeloid cells leads to production of IFN.30 We detected a four-log increase in bacterial burden in portal venous blood and in fibrotic livers compared with sham-operated mice (see online supplementary figure S4A), indicating that gut microbial translocation may initiate tonic IFN production in liver myeloid cells. Indeed, portal venous blood from fibrotic mice added in vitro to F4/80+ liver macrophages, isolated from different knockout mouse lines, caused IFN release that was induced through TLR4, TLR2 and TLR9 (figure 4A). Similar results were obtained when bone marrow-derived macrophages from these knockout mouse lines were used (not shown). We also detected increased levels of lipopolysaccharide in portal venous blood and to a lesser extent in peripheral blood of patients with liver cirrhosis, but not peripheral blood of healthy donors (figure 4B), corroborating previous reports on detection of bacteria/debris in portal venous blood of patients with cirrhosis31 and indicating its involvement in tonic IFN expression in liver cirrhosis. To characterise the role of gut microbiota in IFN expression, we investigated germ-free mice. We found a strong reduction in tonic IFN expression in livers of germ-free mice at 3 weeks after BDL (figure 4C) that was not caused by reduced fibrosis (see online supplementary figure S4C). IFNβ expression in fibrotic germ-free mice was not reduced to the same level as in sham-operated mice (figure 4C), suggesting that further factors contribute to tonic IFNβ expression in fibrotic livers. Further, hepatic expression of IL-12p35 and IFNγ upon L monocytogenes infection normalised in germ-free mice with liver fibrosis compared with mice with normal gut microflora (figure 4D). Consequently, myeloid cells in germ-free mice showed normal phagocytic activity, ROS production and bactericidal capacity despite liver fibrosis (figure 4E, F and online supplementary figure S4D). While these results suggested a central role of translocated gut microbiota in impaired antibacterial immunity during liver fibrosis, it remained unclear which mechanism affected myeloid cell functionality.

Figure 4

Translocation of gut microbiota triggers interferon (IFN) expression in fibrotic liver. (A) IFNβ mRNA expression in F4/80+ liver macrophages isolated from different TLR-knockout lines 6 hours after incubation with portal venous blood. (B) Serum endotoxin concentration in portal venous blood or peripheral blood of patients with cirrhosis or peripheral blood of healthy donors. (C) IFNβ mRNA expression in livers of bile duct ligation (BDL)-operated germ-free (GF) or specific pathogen-free (SPF) mice on day 2 after Listeria monocytogenes (L.m.) infection. (D) Time kinetics of liver interleukin (IL)-12p35 and IFNγ mRNA expression. (E) In vitro phagocytosis of viable L.m. by liver CD11b+ cells. (F) Reactive oxygen species production by liver CD11b+ cells after re-stimulation in vitro. Symbols represent individual mouse and mean±SEM. Data are representative of 2–5 independent experiments. *p<0.05; **p<0.01; ***p<0.001.

Massive IFN induction after bacterial infection during liver fibrosis triggers IL-10 expression and impaired myeloid cell function

Since myeloid cells showed impaired innate immune functions upon L monocytogenes infection, we next investigated the role of inhibitory cytokines. Infection with pathogenic intracellular bacteria such as L monocytogenes causes higher IFN expression compared with bacteria that remain extracellular due to triggering cytosolic PRRs such as RIG-I, cGAS/STING but also NOD.32 ,33 Infection with L monocytogenes in mice with liver fibrosis led to an even higher hepatic IFN expression compared with healthy livers (figure 5A), whereas little difference in IFN expression was observed in spleen (see online supplementary figure S5A), reiterating the organ-specific regulation of IFN expression. In contrast, infection with the extracellular bacterium Escherichia coli did not raise IFN expression levels in fibrotic livers in comparison to normal livers (figure 5B), suggesting that high-level IFN expression levels were related to infection with pathogenic bacteria triggering cytosolic pattern recognition. Consequently, the L monocytogenes mutant Δhly that fails to enter the cytosol and activate cytosolic PRRs33 did not elicit such strong IFN expression in fibrotic liver (figure 5B).

Figure 5

Enhanced interleukin (IL)-10 expression after infection with pathogenic intracellular bacteria in myeloid cells from fibrotic mice or humans with liver cirrhosis. (A) Time kinetics of liver interferon (IFN)β mRNA expression after infection. (B) Liver IFNβ mRNA expression after infection with Escherichia coli, Listeria monocytogenes (L.m.) or Δhly mutant. (C) Time kinetics of liver IL-10 mRNA and protein in blood after infection. (D) IL-10 mRNA expression by indicated cell populations flow cytometry activated cell sorting (FACS)-sorted from the liver of sham or bile duct ligation (BDL)-operated mice at day 2 post infection with L.m. (E) IL-10 protein released from CD11b+ cells FACS-sorted from the liver of sham or BDL-operated mice and infected in vitro with L.m. (multiplicity of infection (MOI) of 20). (F) IL-10 mRNA expression in peripheral blood monocytes from healthy donors or patients with cirrhosis 6 hours after in vitro infection with intracellular bacteria. Graphs show mean values±SEM. Data are representative of 2–5 independent experiments. *p<0.05; **p<0.01; ***p<0.001.

Since IFN has been reported to trigger IL-10 expression in the context of M tuberculosis infection,34 we wondered whether the massive IFN expression during L monocytogenes infection in mice with liver fibrosis caused IL-10 expression. Indeed, we found enhanced IL-10 mRNA expression after L monocytogenes infection that further increased over time in fibrotic but not healthy livers and was accompanied by augmented IL-10 protein levels in blood (figure 5C), whereas no IL-10 mRNA increase was observed in spleen (see online supplementary figure S5B). F4/80+ macrophages and Ly6C+ inflammatory monocytes isolated from fibrotic livers at day 2 p.i. with L monocytogenes expressed augmented levels of IL-10 mRNA compared with myeloid cells isolated from healthy livers (figure 5D). Furthermore, myeloid cells infected in vitro with L monocytogenes produced more IL-10 protein when derived from fibrotic livers compared with healthy livers (figure 5E). We confirmed this increased IL-10 production in humans using CD14+ monocytes isolated from the peripheral blood of patients with advanced liver cirrhosis. Upon in vitro infection with various clinically relevant pathogenic bacterial species that activate cytosolic PRRs, such as Listeria, Legionella, Mycobacterium and Salmonella, monocytes isolated from patients with cirrhosis produced significantly higher levels of IL-10 compared with monocytes from healthy individuals (figure 5F). To establish a functional link between tonic IFN expression and IL-10 expression after infection, we preincubated myeloid cells with recombinant IFNβ before in vitro infection with L monocytogenes. Such IFNβ pretreatment augmented subsequent IL-10 expression after L monocytogenes infection in myeloid cells from healthy liver to the level of IL-10 expression by myeloid cells isolated from fibrotic liver (see online supplementary figure S5C). These results suggest that tonic IFN expression and signalling enhances subsequent IL-10 production by myeloid cells selectively in response to infection with intracellular bacteria.

To provide evidence for a causal link between interferon alpha receptor (IFNAR) signalling and IL-10 expression in vivo, we induced liver fibrosis in Ifnar−/− mice. Upon L monocytogenes infection of Ifnar−/− mice with liver fibrosis, no upregulation of IL-10 expression was observed compared with healthy liver (figure 6A). Thus, IFNAR signalling is responsible for massive IL-10 expression after infection that goes far beyond the levels known to be induced during infection of healthy liver35 or during chronic liver inflammation and fibrosis.36 Importantly, Ifnar−/− mice showed enhanced IL-12p35 and IFNγ expression despite liver fibrosis (see online supplementary figure S6A, B) and efficiently controlled L monocytogenes infection (figure 6A). Consistent with the role of gut microbiota in tonic IFN expression, germ-free mice with liver fibrosis showed a pronounced reduction in IFN and IL-10 expression (figure 6B) but did not completely control infection (figure 6C), suggesting a role of microbiota-independent yet IFN-mediated attenuation of antibacterial immunity. Myeloid cells from Ifnar−/− mice showed improved phagocytic uptake of L monocytogenes and bactericidal activity (figure 6D and online supplementary figure S6C). Accordingly, survival of Ifnar−/− mice with liver fibrosis was similar to healthy Ifnar−/− mice after infection (figure 6E). There were no differences in liver fibrosis between wild-type and Ifnar−/− mice (not shown). These results point to a key role of IFN signalling through induction of IL-10 in the failure to control L monocytogenes infection during liver fibrosis.

Figure 6

Mice deficient for IFNAR signalling are protected from infection-associated mortality during liver fibrosis. (A) Hepatic interleukin (IL)-10 mRNA expression on day 2 after infection relative to non-infected liver (upper panel), liver bacterial burden at day 4 post infection (lower panel). (B) Hepatic interferon (IFN)β or IL-10 mRNA expression on day 2 after infection relative to non-infected liver in germ-free (GF) or specific pathogen-free (SPF) mice. (C) Liver bacterial burden in GF or SPF mice. (D) In vitro phagocytosis of viable bacteria by liver CD11b+ myeloid cells isolated. (E) Kaplan-Meier survival curve of sham and bile duct ligation (BDL)-operated wild-type and IFNAR−/− mice after infection with Listeria monocytogenes (L.m.) (LD50). (F) Liver bacterial burden and survival after infection (LD50) of sham and BDL-operated LysMcre-IFNARflox/flox (LysM-Cre+) mice or littermates (LysM-Cre) after infection with L.m. (LD50). Graphs show mean values±SEM. Data are representative of 3–5 independent experiments. *p<0.05; **p<0.01; ***p<0.001.

As myeloid cells were the main producers of IL-10 during L monocytogenes infection in the fibrotic livers, we next examined whether myeloid cells were also direct targets of IFN signalling. To this end, we induced liver fibrosis in transgenic mice where the IFNAR was selectively deleted in myeloid cells LysMcre-IFNARflox/flox (Lyz2-CrexIfnarflox/flox). Sublethal but also LD50 infection doses of L monocytogenes were efficiently controlled by LysMcre-IFNARflox/flox mice without any infection-associated mortality (figure 6F and online supplementary figure S6D, E). Thus, during liver fibrosis myeloid cells render themselves dysfunctional with IFN being important at two stages: first, tonic IFN signalling triggered by translocated gut microbiota facilitates massive IFN production during infection with pathogenic intracellular bacteria, and second, IFN triggered by intracellular PRR activation causes induction of IL-10 in myeloid cells.

Since we observed a drop in granulocytes after bacterial infection of mice with liver fibrosis (see figure 2), we explored whether IFN also acted on granulocytes. Time kinetic analyses revealed that granulocyte numbers in peripheral blood were lower during the initial injury phase but eventually returned to levels comparable to healthy mice. After infection, there was a severe and sustained reduction in circulating granulocyte numbers that persisted over time (see online supplementary figure S6F). We did not observe a reduction in the growth factor granulocyte-macrophage-colony stimulating factor (GM-CSF) (see online supplementary figure S6G) that is required for granulocyte generation. Moreover, we have observed a selective reduction in the expression of CXCL2, the chemokine that allows for exit of granulocytes from the bone marrow and localisation to peripheral tissue but not other chemokines such as CCL2, CCL5 (see online supplementary figure S6H). In contrast, granulocyte numbers in the bone marrow were low, indicating defective granulopoiesis (see online supplementary figure S6I). Strikingly, in the absence of IFNAR signalling we detected normalisation of circulating granulocyte numbers before and after infection in mice with liver fibrosis. Also, granulocyte numbers in bone marrow of mice with liver fibrosis normalised and reached similar levels as in healthy mice (see online supplementary figure S6F, I). These results point to a universal role of IFNAR signalling during liver fibrosis in impairment of monocyte/macrophage functions and granulocyte production.

Therapeutic targeting of the IFN–IL-10 axis rescues mice with liver fibrosis from infection-associated mortality

We next investigated whether neutralisation of IFNAR signalling would also function in a therapeutic setting using an IFNAR-blocking antibody. Intravenous application of anti-INFAR1 antibodies was started at day 6 after BDL and given every third day. Such treatment with anti-INFAR1 antibodies prevented IL-10 expression upon L monocytogenes infection in fibrotic livers (figure 7A), fully restored the capacity for bacterial elimination from fibrotic livers and protected from infection-associated mortality (figure 7B, C). Thus, interference with IFNAR signalling at a time point when liver disease is already present is still able to reconstitute myeloid cell function to control L monocytogenes infection and prevent mortality.

Figure 7

Therapeutic targeting of interferon (IFN) or interleukin (IL)-10 signalling to rescue from infection-associated mortality mice during liver fibrosis. (A) Hepatic IL-10 mRNA at day 2 post infection (p.i.) with Listeria monocytogenes (L.m.) and (B) bacterial burden in liver. (C) Kaplan-Meier survival curve of sham and bile duct ligation (BDL)-operated mice treated with anti-IFNAR1-blocking antibody or isotype prior to infection with L.m. (5×103). (D) IL-12p35, IFNγ mRNA expression and ELISA of IL-1β protein from liver tissue lysates of liver tissues of mice at day 2 p.i. with L.m. (E) Liver bacterial burden of sham and BDL-operated mice at day 4 p.i. and treatment with IL-10R-blocking antibody. (F) Kaplan-Meier survival curves after infection (LD50) and treatment with IL-10R-blocking antibody. Data are representative of 2–5 independent experiments. *p<0.05; **p<0.01; ***p<0.001.

This encouraged us to investigate whether IL-10 can be therapeutically targeted to restore protective antibacterial immunity. Blockade of IL-10 with a neutralising anti-IL-10R antibody in fibrotic mice fully restored the production of pro-inflammatory cytokines IL-12p35, IFNγ and IL-1β upon L monocytogenes infection (figure 7D). Functionally, IL-10R blockade led to clearance of L monocytogenes from fibrotic livers and prevented infection-associated mortality (figure 7E, F). This rescue effect was caused by a direct improvement of myeloid cell immune function and not amelioration of liver fibrosis because no changes in collagen or α−SMA were detected by immunohistochemistry or western blot (see online supplementary figure S7A, B). The reduction in liver damage reflected by ALT levels in the serum of anti-IL-10R or anti-IFNAR-treated mice (see online supplementary figure S7C) might therefore be related to regain of immune control over bacterial infection rather than improvement of the underlying liver disease. Thus, antibody-induced interference with IFN or IL-10 receptor signalling was sufficient to reconstitute protective innate immunity against bacterial infection in myeloid cells and to rescue mice with liver fibrosis from infection-associated mortality.


Development of liver cirrhosis regardless of the aetiology of chronic liver damage is associated with high mortality that is often related to intractable bacterial infections. Dysfunctional immune cells have been found in patients with liver cirrhosis,7 ,11 ,12 yet the mechanisms determining the failure to contain bacterial infections are unclear. Here, we provide evidence that translocation of gut microbiota or bacterial degradation products during liver fibrosis and their delivery via portal venous blood to the liver triggers tonic type I IFN expression and signalling that results in dysfunctional innate immunity in myeloid cells failing to control bacterial infection. Upon infection with bacteria that activate cytosolic PRRs, myeloid cells in fibrotic liver produce excessive amounts of IFN. Signalling through the IFN receptor in myeloid cells then triggers expression of the regulatory cytokine IL-10 that reduces their phagocytic capacity and bactericidal functions. This identifies IFN and IL-10 receptor signalling as molecular targets for therapeutic intervention to control intractable bacterial infections during liver fibrosis and cirrhosis.

Microbial translocation is a known consequence of severe liver fibrosis and liver cirrhosis that is considered to be related to infection-associated mortality.10 ,37 Such microbial translocation drives progression of liver fibrosis through TLR4-dependent recognition of gut-derived microbial products perpetuating inflammation and liver damage.28 Development of hepatocellular carcinoma in the context of chronic liver damage is also accelerated through microbial translocation and TLR4 signalling.38 During liver cirrhosis, substantial changes occur in the gut microbiome and microbial dysbiosis or bacterial overgrowth can ensure that may further exacerbate inflammation-mediated liver disease.39 ,40 Recently, liver macrophages have been described to function as firewall that clear translocated gut microbiota from the portal venous blood.29 ,37 We report here on the tonic expression of IFN and IFNAR signalling, predominantly in liver myeloid cells, within fibrotic livers in mice and also in human cirrhotic liver that resulted from activation through TLR2, TLR4 and TLR9. Using germ-free mice, we found that translocation of gut microbiota was responsible to a large part for such tonic IFN induction and signalling in fibrotic liver. The small remaining IFN expression in germ-free mice may result from sterile inflammation caused by cell death in chronic liver damage that also triggers IFN expression. We conclude that during liver fibrosis and cirrhosis translocated gut microbiota trigger tonic IFN expression and signalling in the liver.

The role of IFN in clearance or persistence of infectious diseases is controversially discussed.34 IFN is induced during infection with viruses but also bacteria through activation of membrane-associated PRRs such as TLRs but also cytosolic PRRs such as STING, MDA5 and RIG-I.34 ,41 Triggering of cytosolic receptors in myeloid cells by viable intracellular bacteria that escaped phagolysomal destruction is considered the strongest stimulus for innate immune responses in the concept of scalable innate immunity to infection that results in strong IFN induction.32 ,33 Our findings reveal that such IFN induction upon infection with pathogenic intracellular bacteria was much more pronounced in myeloid cells from fibrotic mice or patients with liver cirrhosis compared with healthy individuals. Initial exposure to tonic IFN signalling, as we observed in fibrotic and cirrhotic livers, enhanced subsequent IFN expression in liver myeloid cells upon L monocytogenes infection. Such augmented IFN expression may have been caused by increased expression of PRRs or the IFNAR itself, resulting in auto-amplification of IFN induction.42 Not mutually exclusive, the impaired clearance of intracellular bacteria may enhance PRR signalling and consecutive IFN induction, which is corroborated by our finding of augmented L monocytogenes burden in mice with fibrotic livers. Irrespective of the underlying mechanisms causing this increased IFN production during bacterial infection, the consequence of IFNAR signalling was induction of IL-10 expression since no IL-10 expression was observed if fibrotic mice lack the IFNAR. IFN has been reported to induce IL-10 expression.43 ,44 Here, we define a key role for IFN-induced IL-10 in dampening of innate immunity to bacterial infection during liver fibrosis. IL-10 downregulated IFNγ, IL-12 and IL1β that are key mediators in induction of antibacterial immunity,45 ,46 and thereby blunted phagocytic and bactericidal functions of myeloid cells. Yet, IL-10 failed to control IFN expression after bacterial infection. This selective lack of IL-10-mediated control of IFN may be the reason for the simultaneous increase in IFN and IL-10 expression that we observe in human and mouse myeloid cells. In the absence of IFNAR signalling, mice with liver fibrosis controlled bacterial infection and showed no infection-associated mortality, which clearly demonstrates the decisive pathophysiological role of IFN in dampening of antibacterial immunity. Further experiments will need to identify the mechanism in myeloid cells that are responsible for the selective failure of IL-10 receptor signalling to control IFN expression. The massive IL-10 production, although originating from the liver, likely impairs antibacterial immunity at the systemic level and may account for the simultaneous occurrence of bacterial infections in different organs and for the difficulties in treating bacterial infections with antibiotics in patients with cirrhosis. While mainly Gram-negative bacteria have been isolated from patients with cirrhosis,47 ,48 it is possible that unapparent infection with pathogenic intracellular bacteria may incapacitate innate immunity in myeloid cells and subsequently facilitates infection with ubiquitous bacteria. In particular, infections with mycobacteria species are often detected in patients with liver cirrhosis47 ,48 and may contribute to myeloid cell immune dysfunction. Our observation that myeloid cells from patients with liver cirrhosis produce excessive amounts of IL-10 upon infection with various bacterial species that can activate cytosolic PRRs confirms the findings in experimental liver fibrosis and suggests that IL-10 production by myeloid cells may serve as marker to identify patients with liver cirrhosis at risk for developing bacterial infection.

We have identified myeloid cells as the main population producing IFN and responding to IFN-receptor signalling with the production of IL-10, which impairs their capacity to clear intracellular bacterial infections. Thus, myeloid cells are involved at all levels, that is, detection of bacterial infection, bacterial clearance and production of regulatory mediators inducing innate immune dysfunction. The closely linked production of IFN and IL-10 after L monocytogenes infection suggests that therapeutic blockade of IFNAR signalling in myeloid cells may restore antibacterial immunity. Indeed, antibody-mediated blockade of IFNAR signalling allowed mice with liver fibrosis to control L monocytogenes infection and rescued them from infection-associated mortality. Similarly, application of antibodies preventing IL-10–receptor signalling reconstituted antibacterial immunity in mice with liver fibrosis. These results demonstrate the therapeutic potential of IFN and IL-10-targeted interventions in liver cirrhosis to overcome impaired antibacterial immunity.

Other immune regulatory pathways such as the transforming growth factor (TGF)β signalling pathway and prostaglandin PGE2 are also found to be expressed in cirrhotic livers.49 PGE2 is induced by cyclo-oxygenase that is a target of IFN receptor signalling, and its increased local expression found in liver cirrhosis may be related to the tonic IFNAR signalling that we report here. TGFβ is not a direct target of IFNAR signalling but TGFβRII is an IFN-sensitive gene. We detected a non-redundant organ-protective role of TGFβ since antibody-mediated blockade of TGFβRII during liver fibrosis led to the death of mice (not shown). This separates IFN-induced IL-10 expression that impairs antibacterial immunity in myeloid cells from TGFβ as an organ-protective cytokine in the context of liver fibrosis and bacterial infection. Taken together, our results indicate that targeted interference of IFN or IL-10–receptor signalling may reconstitute innate immune competence in myeloid cells in patients with liver cirrhosis. This suggests the usefulness of host-targeted immune intervention together with conventional antibiotic therapy to regain immune control of intractable bacterial infections bacteria with liver cirrhosis.

Materials and methods


C57BL/6N mice, MX1 transgenic mice, TLR2–/–, TLR4–/–, TLR7–/– TLR9–/–, IFNAR–/– and LysM-Cre-IFNARflox/flox mice were bred and maintained under specific pathogen-free (SPF) conditions in the animal facility of the University of Bonn. Germ-free mice were kept at the gnotobiotic facility of the University of Ulm. In vivo experiments were approved by the local animal care commission.

BDL and CCl4 models of liver cirrhosis

BDL and sham operations were carried out under anaesthesia (isoflurane 1.5%) as described.17 ,50 Before operation, mice were injected intraperitoneally with 5 mg/kg carprofen, followed by a midline abdominal incision and isolation and occlusion of the common bile duct. In sham-operated controls, the bile duct was mobilised but not ligated. CCl4 was administered intraperitoneally (1:4 dissolved in olive oil; 60 μL per animal) twice weekly, while control mice received 60 μL olive oil. Two weeks after BDL operation or 12 weeks of continuous CCl4 administration and 10 days of recovery, mice were used for infection experiments. Germ-free mice were kept for 3 weeks after sham or BDL operation to reach a similar fibrosis score as mice kept under SPF conditions.

Infection with L monocytogenes

Mice were infected intraperitoneally with 5×103 colony forming unit (CFU) (sublethal dose) or 5×104 CFU (LD50 for C57BL/6N mice) of wild-type L monocytogenes (EGDe strain), 1×107 CFU of Δhly mutant of L monocytogenes or E coli (ATCC no. 26). Bacteria were acquired during log phase growth. Bone marrow-derived macrophages were generated from tibia and cultured in RPMI 1640 medium supplemented with fetal calf serum, glutamate, βME and 30% L929 cells conditioned medium for 7 days. Cells were cultured and infected as previously described.51

Western blot analysis

Shock-frozen liver tissues were lysed in cold-lysed, heat-denaturated and loaded on 10% SDS-PAGE. Membranes were then blotted overnight with primary anti-α-smooth muscle antigen (SMA) and anti-GAPDH antibodies (Abcam) followed by secondary antibodies (horse radish peroxidase (HRP)-linked anti-rabbit IgG (Cell Signaling) and HRP-linked anti-mouse IgG (Santa Cruz)).

Statistical analysis

Results are expressed as mean±SEM. Comparisons were drawn using a two-tailed paired Student's t-test or the non-parametric Wilcoxon signed-rank test (cortex and medulla from one mouse) or unpaired Student's t-test, one-way analysis of variance in combination with Bonferroni multiple-comparison test, or the two-sided non-parametric Mann-Whitney U test (comparison between individual mice) (Prism 4, GraphPad Software).


The authors thank T Chakraborty for providing Listeria monocytogenes.


View Abstract


  • Carl-Philipp Hackstein and Lisa Mareike Assmus are first co-authors.

  • Correction notice This article has been corrected since it published Online First. The first co-author statement has been added.

  • Contributors C-PH and LMA contributed equally to this work. PAK and ZA serve both as senior authors. Acquisition of data: C-PH, LMA, MW, SK, TS, MB, JT, DK, FG and ZA. Study concept and design: CK, JT, JS, WK, ZA and PAK. Analysis and interpretation of data: ZA, CK, WK, IF, JT and PAK. Drafting the manuscript: ZA, PAK, WK and JT.

  • Funding ZA, PAK, JT and CK were funded by the German Center for Infection Research the following part:, partner sites München and Bonn. Research Foundation (SFB TRR57 and SFB TR179) and the German Centre for Infection Research DZIF. PAK, WK, MB, JS and CK are members of the excellence cluster ImmunoSensation. PAK, WK, MB, JS and CK were funded by the German Research Foundation (SFB 704). JT was supported by the H. J. & W. Hector Foundation. WK is supported by NRW-Rückkehrerprogramm of the German state of Northrhine-Westfalia.

  • Competing interests None declared.

  • Ethics approval Germany Northrhine Westfalia.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles