|Home | About | Journals | Submit | Contact Us | Français|
Multicellular organisms are continuously exposed to many different pathogens. Because different classes of pathogens require different types of immune responses, understanding how an ongoing immune response to one type of infection affects the host’s ability to respond to another pathogen is essential for a complete understanding of host-pathogen interactions. Here we used a mouse model of co-infection to gain insight into the effect of respiratory influenza virus infection on a subsequent systemic bacterial infection. We found that influenza infection triggered a generalized stress response leading to a sustained increase in serum glucocorticoid levels, resulting in a systemic suppression of immune responses. However, virus-induced glucocorticoid production was necessary to control the inflammatory response and prevent lethal immunopathology during co-infection. This study demonstrates that activation of the hypothalamic-pituitary-adrenal axis controls the balance between immune defence and immunopathology, and is an important component of the host response to co-infection.
Influenza virus commonly leads to complications with pulmonary secondary bacterial infections that are able to spread systemically.
Infection with influenza virus led to an increase in bacterial burden of the systemic model bacterial pathogen Listeria monocytogenes.
Influenza virus suppressed the immune response to systemic bacterial infection in part through the induction of glucocorticoids.
While removal of glucocortioids rescued the virus-induced immunosuppression, they were necessary to control the inflammatory response to co-infection.
Most studies of host-pathogen interactions have focused on the interaction of a single pathogen with a host cell or organism. However, host organisms in a natural environment are commonly exposed to multiple pathogens simultaneously. The immune response to infection can cause vast physiological alterations in the metazoan host, including changes in metabolism, serum proteins, cellular distribution, and hematopoiesis. In addition, the class of pathogen and the site of infection are instrumental in determining the extent and type of immune response induced. Therefore, an important problem in host-pathogen interactions is how the immune response to one pathogen alters the immune response to another infectious agent (Bakaletz, 2004). Polymicrobial infection, or co-infection, can lead to complications in the treatment of infectious diseases, and there are many well-recognized instances of co-infection that occur naturally in the human population (Beadling and Slifka, 2004; Elston and Thaker, 2008; Renia and Potter, 2006). Several studies examining infection with bacteria during an ongoing response to a viral infection have found that, depending on the types of pathogens, the temporal proximity of the infections, and the route of infection, a variety of outcomes are possible (Barton et al., 2007; Gardner, 1981; Gumenscheimer et al., 2007; Humphreys et al., 2008; Navarini et al., 2006). Respiratory infection with influenza virus is particularly interesting in this regard, as it is most commonly complicated by secondary bacterial infections (Hament et al., 1999).
Influenza virus is a negative stranded RNA virus that primarily infects the lung epithelium and leads to a strong local inflammatory response (La Gruta et al., 2007). Infection with influenza virus in humans usually leads to a brief but severe illness, and fatalities are often linked to co-infection with bacterial pathogens (Beadling and Slifka, 2004; Berendt and Jaax, 1985; Hament et al., 1999; McCullers, 2006; Peltola and McCullers, 2004; Seki et al., 2004; Sun and Metzger, 2008). Several studies have demonstrated that infection with influenza virus can result in immunosuppression and subsequent secondary bacterial infection (Borrow et al., 2005; Didierlaurent et al., 2008; Marcus et al., 2005; Noone et al., 2005; Shahangian et al., 2009; Sun and Metzger, 2008; Tumpey et al., 2000; van der Sluijs et al., 2004). However, these studies focused primarily on the local effects of influenza at the site of infection. The effect of influenza virus infection on the systemic immune response is less well understood.
Influenza virus infection facilitates the ability of bacterial pathogens to gain entry and replicate in the lung, and many of these pathogens are able to spread systemically with lethal complications (Beadling and Slifka, 2004). In order to specifically examine the effect of influenza virus lung infection on the systemic immune response to bacteria, we utilized the model bacterial pathogen L. monocytogenes. Although L. monocytogenes is not commonly associated with secondary bacterial infections during influenza infection, it was used here because it provides an excellent model of a systemic bacterial infection. The innate immune system is triggered rapidly after infection with L. monocytogenes, and it is crucial for early control of bacterial replication (Kaufmann, 1993). Two cytokines are important components of the innate immune response to L. monocytogenes, IL-6 and IFN-γ (Andersson et al., 1998; Dai et al., 1997; Dalrymple et al., 1995; Kopf et al., 1994). The primary cells infected by L. monocytogenes are macrophages and hepatocytes (Selmi et al., 2007). Following infection, bacteria spread from cell-to-cell by co-opting the host cell’s actin-based motility system (Portnoy et al., 2002). As with many systemic bacterial infections, the hepatic immune response is essential for the clearance of L. monocytogenes (Cousens and Wing, 2000; Selmi et al., 2007). The early immune response in the liver consists primarily of interactions between the resident hepatocytes and Kupffer cells and infiltrating monocytes and neutrophils (Gregory and Wing, 2002). While the innate immune response is able to control many aspects of L. monocytogenes infection, the adaptive immune system, and particularly the cytotoxic T cell response, is necessary for sterilizing immunity to the pathogen (Pamer, 2004; Unanue, 1997b).
We found that influenza infection suppressed the systemic antibacterial innate immune response. The major mechanism for this suppression was influenza virus infection-induced glucocorticoid (GC) production. GC are produced by adrenal glands during times of psychological and physiological stress, and their production is regulated by the hypothalamic-pituitary-adrenal (HPA) axis (Di Comite et al., 2007; Rogatsky and Ivashkiv, 2006; Sternberg, 2006). They are known to have pleiotropic immunosuppressive effects, and are important in regulating inflammation (Sternberg, 2006). Previous studies that demonstrated production of GC during viral infection and other inflammatory immune responses proposed mechanisms of GC production caused by inflammatory cytokines such as IL-1 and IL-6 (Besedovsky and del Rey, 1989; Bethin et al., 2000; Dunn et al., 1989; Dunn and Vickers, 1994; Hermann et al., 1994; Ruzek et al., 1997). We show here that lung infection with influenza virus can induce GC production independently of elevated systemic inflammatory cytokines. Furthermore, we show that while GC production resulted in systemic immunosuppression, it was required to protect co-infected hosts from lethal immunopathology.
To examine the effect of an ongoing viral infection on the immune response to a systemic bacterial infection, C57Bl/6 mice were infected intranasally with influenza virus and subsequently infected intravenously with L. monocytogenes. Bacterial and viral loads were measured in the liver and spleen four days after L. monocytogenes infection. Mice with a concurrent viral infection had a five-log greater bacterial load in the liver, compared to mice infected with L. monocytogenes alone (Figure 1A,B). This increase in bacterial burden was dependent on the dose of influenza virus (Figure 1A). The minimal effective influenza virus dose caused an increase in bacterial CFUs even when given as much as five days before L. monocytogenes (Figure 1B). A lower dose of virus (200 PFU) was only effective in causing an increased bacterial load when given 3-5 days before bacterial infection (Figure 1C and data not shown). Mice infected with a sublethal viral dose of 200 PFU five days before infection with L. monocytogenes also displayed delayed bacterial clearance when examined 8 days after bacterial infection (Figure 1C). Mice given this sublethal dose of influenza had elevated levels of bacteria in the liver 4, 6, 8, and 12 days after bacterial infection and in the spleen at the later timepoints (Figure S1). However, by 30 days after bacterial infection the bacteria were cleared in both organs regardless of a previous infection with influenza virus (data not shown).
Type I interferons IFNs have been shown to be detrimental to the host response during L. monocytogenes infection (Auerbuch et al., 2004; Carrero et al., 2004; Decker et al., 2005; O’Connell et al., 2004). Given that type I IFNs are produced during the early stages of an active influenza virus infection (Julkunen et al., 2001; Wang et al., 2000), it seemed possible that this was responsible for the increased CFUs. However, type I IFNs were not responsible for the increase in bacterial CFUs observed in co-infected mice, as there was still a significant increase in CFUs in viral/bacterial co-infected IFNαR1 deficient mice (Figure 2A).
The increase in bacterial load caused by previous infection with influenza virus was observed as early as four days after infection with L. monocytogenes, suggesting a defect in the innate immune response. We examined further whether this increased bacterial burden was due to altered functions of the innate or acquired immune system by infecting Rag2−/− mice, which lack an acquired immune system. Rag2−/− co-infected mice exhibited increases in bacterial load that were similar to those observed in co-infected wildtype mice (Figure 2B), indicating that the acquired immune response was not responsible for the initial increase in CFUs.
Since both type I IFNs and cells of the acquired immune system are necessary components of the anti-viral immune response, both IFNαR1−/− and Rag2−/− mice had increased viral loads in the lungs when infected with the same dose of influenza as wildtype mice (data not shown). To compare the effect of influenza on subsequent bacterial infection in wildtype, IFNαR1−/−, and Rag2−/− mice independent of this increased viral load, viral doses were used that resulted in similar viral loads in the lungs of all strains of mice (Figure 2C). As shown in previous studies (Garcia-Sastre et al., 1998), influenza virus had altered tissue tropism in IFNαR1−/− mice and spread systemically to the kidneys (data not shown). While influenza virus infection had a profound effect on the bacterial burden, infection with L. monocytogenes did not alter the tissue tropism or viral titre of the influenza virus in comparison with singly infected controls (Figure 2C and data not shown).
Since co-infection with influenza virus and L. monocytogenes led to increased bacterial burden, we next examined the immune response to L. monocytogenes after infection with influenza virus. One clear change was that the spleen size and total splenocyte number of both influenza virus infected and co-infected mice was decreased compared to spleens of uninfected mice or mice infected with L. monocytogenes alone (Figure 3A). Systemic levels of inflammatory cytokines were also measured over the course of L. monocytogenes infection. When compared to mice infected with L. monocytogenes alone, the levels of IFN-γ and IL-6 in the serum were significantly decreased in co-infected mice (Figure 3B).
Since increased bacterial loads were observed mainly in the liver, we next directly examined the hepatic immune response. Recruitment of leukocytes to the liver by chemokines, cytokines, and adhesion molecules is essential for development of an appropriate immune response (Salmi et al., 1998). We therefore examined liver sections from mice four days post-infection with L. monocytogenes for recruitment of immune cells. There was a significant decrease in cellular infiltrate in the livers of co-infected mice compared to mice infected with Listeria monocytogenes alone (Figure 3C and Figure S2A). At day 1 after bacterial infection there was a significant decrease in neutrophil infiltrate (Figure S2B, C).
To examine a cause for this reduction in leukocyte infiltrates we looked at the expression of chemokines in the liver. We observed significant decreases in expression levels of the chemokines CCL7 (MCP-3), CCL2 (MCP-1), CXCL10 (IP-10), CXCL2 (MIP-2α) and CXCL9 (MIG) in co-infected mice relative to mice infected with bacteria alone (Figure 3D). CXCL10 and CXCL9, two chemokines with proposed bactericidal activity, are both induced by IFN-γ (Cole et al., 2001); it is therefore possible that reduction in systemic IFN-γ may account for their decreased expression in the liver. CCL2 and CCL7 have both been shown to be critical chemokines in the defence against L. monocytogenes (Jia et al., 2008). In addition, CCL7 can be induced by IL-6, therefore the reduction in systemic IL-6 may lead to the decrease in hepatic CCL7 expression (Romano et al., 1997). While CCL2 recruits mainly monocytes to the site of infection, CCL7 attracts many different cell types of the immune system, by binding to at least four different chemokine receptors (Menten et al., 2001). In addition to the decrease in chemokine expression, there was a decrease in the hepatic expression of the adhesion molecule Icam-1 (Figure 3D). The reduction of these critical chemokines and adhesion molecules in the liver is consistent with the observed reduction of immune cell infiltrate.
Given the delay in bacterial clearance it seemed likely that aspects of the acquired immune response were also suppressed. Indeed, there were fewer antigen specific CD8+ T cells at day 8 after bacterial infection in mice infected 5 days previously with 200 PFU of influenza virus (Figure S3). Thus, previous infection with influenza virus suppresses many aspects of the innate and acquired immune responses to systemic bacterial infection, including production of cytokines, tissue specific chemokines, immune cell infiltration into the infected organs and antigen specific T cell accumulation.
We next examined the mechanism responsible for the suppressed immune response in co-infected mice. Levels of the immunosuppressive cytokine IL-10 were not increased in the serum of influenza-infected or co-infectd mice (Figure S4A and data not shown). Although this does not formally exclude a potential contributing role of IL-10, we examined other potential mechanisms of immunosuppression. Mice infected with influenza virus showed sustained elevated levels of GC in the serum (Figure 4A). GC are produced by the adrenal glands in response to stress through activation of the hypothalamic-pituitary-adrenal (HPA) axis, and are known to have multiple immunosuppressive effects (Sternberg, 2006). Inflammatory cytokines are known to induce GC, and indeed high levels of the cytokines IL-6, IFN-γ, and IL-12 were detectable in the serum of the infected mice 24 hours after systemic infection with L. monocytogenes (Figure S4B). However, increased serum GC levels were only detectable in L. monocytogenes infected mice at the peak of the innate immune response to infection (~24hrs), after which they quickly returned to normal levels (Figure 4A). In contrast, mice infected with influenza virus showed sustained levels of serum GC (Figure 4A), despite the absence of high levels of systemic inflammatory cytokines (Figure S4A). Furthermore, influenza virus-induced GC levels were not decreased in MyD88−/−, IL-6−/−, IL-1R−/−, and TNFα−/− and IFNαR1−/− mice (Figure S5). Finally, systemic administration of LPS caused only modest and transient induction of GC while poly I:C caused no induction at all (data not shown). Collectively, these results suggest that the sustained level of serum GC induced by influenza virus infection was independent of a systemic inflammatory response.
To examine the role of influenza infection-triggered GC in the suppression of immune responses to a systemic bacterial infection, we took advantage of mice that had their adrenal glands surgically removed. Consistent with the requirement for adrenal glands in the production of GC, adrenalectomized (ADX) mice did not show an increase in serum GC in response to co-infection or influenza infection alone (Figure 4A). In addition, many of the ADX mice had levels of GC that were below the limit of detection. Importantly, adrenalectomy reversed the effect of influenza infection on the hepatic bacterial load in co-infected mice as the co-infected ADX mice had bacterial loads similar to mice infected with L. monocytogenes alone (Figure 4B). Viral titres in the lung were not statistically different between co-infected, co-infected sham-operated, and co-infected ADX (Figure 4C).
The improved clearance of bacteria in co-infected ADX mice indicated that the influenza virus infection-induced production of GC is an important mechanism involved in suppression of the immune system during co-infection. We next analyzed the specific aspects of the innate immune response that could be the targets of GC signaling. In contrast to co-infected mice or co-infected sham-operated controls, co-infected ADX mice had higher levels of IL-6 in the serum 24 hours after L. monocytogenes infection (Figure 5A). The levels were similar to those seen in mice infected with L. monocytogenes alone, consistent with in vitro observations showing that IL-6 is regulated by GC (Ray and Sehgal, 1992). IL-6 has also been shown to induce expression of CCL7 (Romano et al., 1997). In correlation with the observed rescue of IL-6 production there was an increase in CCL7 expression in the liver (Figure 5B). In addition, in ADX co-infected mice, Icam-1 expression in the liver was also restored to nearly the same level as in mice infected with L. monocytogenes alone (Figure 5C). In accordance with the rescue of CCL7 and Icam-1 expression in the liver in ADX mice, co-infected ADX mice had similar levels of immune cell infiltrate as mice singly infected with L. monocytogenes. (Figure 5D and Figure S2C). Another pronounced effect of influenza virus co-infection was the decrease in spleen size. This effect was also due to GC as it was eliminated in co-infected ADX mice (Figure 5E). Interestingly, the effect of adrenalectomy did not extend to all pro-inflammatory signalling pathways, as serum levels of IFN-γ were not affected by adrenalectomy, in contrast to IL-6 (Figure 5A). Accordingly, IFN-γ induced chemokines also were not rescued in co-infected ADX mice (data not shown). However, given the rescue of the bacterial burden in ADX mice, the levels of these chemokines and cytokines must be sufficient.
Further corroborating the hypothesis that virus-induced GC were responsible for the increased bacterial load, mice treated daily with the synthetic corticosteroid dexamethasone (DEX) starting one day prior to infection with L. monocytogenes also showed increased bacterial loads in the liver (Figure 4B). Also, the immunosuppression by influenza-induced GC was mimicked by DEX treatment. CCL7 and Icam-1 expression in the liver were decreased in mice treated with DEX, and there was a corresponding decrease in cellular infiltrates (Figure 5B, C, and D). In addition, the spleens of L. monocytogenes infected/ DEX treated mice showed a decrease in size (Figure 5E). Importantly, the DEX levels used in these experiments were determined by a bioassay to be physiologically similar to the levels of GC measured during influenza infection (Figure S6).
These results show that GC production triggered by an infection with influenza virus led to a suppression of key aspects of the innate immune response to L. monocytogenes, including the production of IL-6 and the hepatic expression of CCL7 and Icam-1. Suppression of these key factors of the innate immune response resulted in an increased bacterial burden. Blockade of GC production prevented the suppression of the innate immune response and reduced the early bacterial load in the liver. In addition, treatment with DEX mimicked many of the aspects of immunosuppression of the innate immune response to L. monocytogenes caused by infection with influenza virus.
While many aspects of the influenza-induced immunosuppression, including the defect in bacterial clearance, were rescued in ADX mice, prevention of GC production is not universally beneficial to the host. Indeed, GC are known to play an important role in controlling excessive inflammation. Accordingly, co-infected ADX mice had large increases in the systemic inflammatory cytokines, IL-6, IL-5 and TNFα four days after infection with L. monocytogenes, and five days after infection with influenza virus (Figure 5F). To examine the effect of the increased inflammatory response further, mice were infected with a sublethal dose of influenza virus followed 5 days later by a sublethal dose of L. monocytogenes. While all non-ADX mice survived co-infection, the majority of the co-infected ADX mice did not survive past five days after the bacterial infection, and many started to succumb as early as one day after bacterial infection (Figure 6A). In contrast, both ADX and non-ADX mice infected with a sublethal dose of either L. monocytogenes (5000 CFU) or influenza virus (200 PFU) alone were able to survive for up to 30 days after infection (Figure 6A and data not shown).
When we further examined this sudden decrease in survival of the co-infected ADX mice, we found that there was a systemic increase in two key inflammatory cytokines. There was a large increase in serum levels of TNFα and IL-12p40 one day after infection with L. monocytogenes in co-infected ADX mice, as compared to the singly infected controls (Figure 6B). Both TNFα and IL-12p40 are known to be suppressed by GC (Rogatsky and Ivashkiv, 2006).
Since our co-infected mice were infected with influenza virus for 5 days before infection with L. monocytogenes it seemed likely that there would also be significant alterations specifically in the pulmonary immune response. There were large increases in a number of inflammatory cytokines in the bronchoalveolar lavage fluid (BALF) (Figure 7A). IL-6, IL-12p40, TNFα, and IFN-γ were all increased in co-infected ADX mice to a greater extent than any other group (Figure 7A). IL-1β and IL-5 were increased in influenza infected ADX mice and co-infected ADX mice (Figure 7A). These cytokines are all known to be subject to control by GC (Besedovsky et al., 1986; Ray and Sehgal, 1992). In addition, there was a dramatic increase in immunopathology noted in the lungs of influenza infected and co-infected ADX mice (Figure 7B and Figure S7). There was also increased level of albumin in the BALF of co-infected ADX mice, which suggests increased epithelial tissue damage (Figure 7C). In contrast, there were no detectable changes in the liver determined by histological analysis and expression of inflammatory genes (data not shown).
Thus, while the production of GC triggered by influenza virus infection transiently suppressed the immune response to a secondary bacterial infection, it was necessary to control the inflammatory response during co-infection in order to prevent lethal immunopathology.
The immune system undergoes a series of dramatic changes during a response to infection. Exposure to a secondary infection while the host undergoes these alterations may affect host defence capabilities in multiple ways, depending on the types of infections. Influenza virus infection is known to increase susceptibility to secondary bacterial infections, particularly opportunistic bacterial pathogens of the respiratory tract (Beadling and Slifka, 2004). Many studies have focussed on mechanisms of influenza virus-mediated immunosuppression in the lung (Didierlaurent et al., 2008; Shahangian et al., 2009; Sun and Metzger, 2008; van der Sluijs et al., 2004). However, the effect of influenza virus infection on systemic immunity is less well understood. Many bacterial pathogens that initially infect the lung will subsequently become systemic and, therefore, it is important to understand the impact of influenza virus on the immune response to a systemic bacterial pathogen. Here we investigated how respiratory influenza infection affects the systemic innate immune response. We chose the model bacterial pathogen L. monocytogenes, to investigate the effects of influenza infection on subsequent systemic bacterial infection because the innate and adaptive immune responses to this pathogen are well characterized (Pamer, 2004; Unanue, 1997a). Using this model system we found that influenza infection caused a profound systemic immunosuppression as indicated by a blunted response to systemic infection with Listeria. Specifically, we found decreased production of inflammatory cytokines and chemokines, reduced recruitment of innate immune cells to the infection site (liver) and, as a result, a dramatic increase in bacterial burden. Surprisingly, this systemic immunosuppressive effect was mediated by a sustained production of GC triggered by the influenza infection. Surgical removal of the adrenal glands, which are the main source of GC, eliminated most of the immunosuppressive effects of influenza infection. Accordingly, adrenalectomized (ADX) mice were able to better suppress secondary bacterial infection. Interestingly, ADX mice had a relatively normal immune response to single viral or bacterial infections. However, the lack of the GC response in co-infected mice, while allowing for an improved anti-bacterial immunity, eventually resulted in an exacerbated inflammatory response leading to lethality.
GC are crucial for the homeostatic regulation of many physiological systems (Tasker, 2006; Viegas et al., 2008). In the immune response they function to regulate inflammation and maintain immune homeostasis (Di Comite et al., 2007; Sternberg, 2006). GC are produced by adrenal glands through activation of the HPA axis by different types of psychological and physiological stress (Di Comite et al., 2007; Rogatsky and Ivashkiv, 2006; Ruzek et al., 1999). One type of physiological stress that activates the HPA axis is excessive production of inflammatory cytokines during systemic inflammatory response syndrome (SIRS) (Beishuizen and Thijs, 2003; Webster and Sternberg, 2004). Many bacterial and viral infections have been shown to trigger the HPA axis because of an increase in systemic inflammatory cytokines, and often the resulting GC are necessary to prevent an excessive inflammatory response (Besedovsky et al., 1986; Besedovsky and del Rey, 1989; Dunn and Vickers, 1994; Hermann et al., 1994; Ruzek et al., 1997; Ruzek et al., 1999). In this study we demonstrate that infection with influenza virus induces a sustained production of GC in the absence of systemic inflammation. In addition, we show that in mice deficient in the classic inflammatory components that are known to induce GC (Silverman et al., 2005), influenza virus infection still induces glucocrticoid production, while a potent inflammatory response induced by systemic administration of LPS or poly I:C did not result in the same magnitude or duration of GC production. Collectively, these data suggest that lung infection with influenza virus induces GC through a novel pathway that is at least in part independent of systemic inflammatory cytokines. Influenza virus causes lung damage both directly, through viral infection of epithelial cells, and indirectly through induction of the inflammatory response (La Gruta et al., 2007). Damage to a crucial organ, such as the lung, causes a generalized stress response (Mendez and Hubmayr, 2005), which can trigger GC production. We therefore suggest that lung tissue damage caused by influenza virus infection triggers the HPA response and GC production. Tissue damage in this case may be perceived through the group C nerve fibers, which play an essential role in monitoring tissue homeostasis (Craig, 2002). However the exact mechanism linking infection-induced tissue damage to the HPA axis activation is currently unknown.
The prolonged elevated GC levels triggered by influenza virus infection affected several key aspects of the innate immune response. Here we show that the induction of the essential cytokine IL-6 in response to L. monocytogenes infection was suppressed by influenza virus-induced GC. Several key aspects of the hepatic immune response to bacterial infection (Jia et al., 2008; Lopez et al., 1999) were also suppressed during co-infection, including the chemokine CCL7 and adhesion molecule Icam-1. Correlating with the decrease in hepatic chemokine and adhesion molecule expression, there was a significant decrease in infiltrating immune cells into the liver in co-infected mice.
Corroborating the data with virus-induced GC, daily treatment with the synthetic corticosteroid DEX also affected the innate immune response in the liver. Mice treated daily with DEX had increased bacterial CFUs in the liver comparable to mice infected with influenza virus. The specific aspects of the anti-bacterial innate immune response that were suppressed in influenza-infected mice and rescued in ADX mice were also suppressed in mice treated with DEX. Mice pre-treated with DEX and infected with L. monocytogenes had decreased expression of the chemokine CCL7 and decreased expression of the adhesion molecule Icam-1, and subsequent decreased cellular infiltration into the liver.
We observed that the liver was the major target organ of early immunosuppression and the site of increased bacterial burden in response to influenza induced endogenous GC or exogenous DEX. There are several factors that may contribute to this organ specific suppression. The liver is a very important organ in the immune response to many bacterial infections of the blood and is known specifically to be important in the innate immune response to L. monocytogenes (Cousens and Wing, 2000; Ebe et al., 1999; Lopez et al., 1999). Regulation of the immune response by IL-6 is very important for the immune response in the liver and mice deficient in IL-6 have a greater defect in the immune response of the liver than the spleen (Dalrymple et al., 1995). The liver has a high expression of glucocortoid receptors, and this expression can be up-regulated during times of stress (Dong et al., 1988; Turner et al., 2006). Also, in the liver hepatocytes as well as macrophages support L. monocytogenes growth, while in the spleen growth occurs mainly in macrophages. The disappearance of these niche cells in the spleen due to GC induced apoptosis is likely to offset the immunosuppressive effect. A combination of these factors could account for the influenza virus-induced increase in bacterial load specifically in the liver.
In this study, we also found that GC are necessary to control excessive inflammation during co-infection and prevent a lethal immunopathology. ADX mice infected with sublethal levels of influenza virus followed 5 days later by a sublethal dose of L. monocytogenes succumb within days of the bacterial infection. These mice have some signs of slightly increased systemic inflammation such as elevated levels of the cytokines TNFα and IL-12p40 in the serum, however, the most dramatic effects of co-infection were observed mainly in the lung. Many inflammatory cytokines in the BALF are dramatically increased in co-infected ADX mice. Consequently the lungs of co-infected ADX mice had increased signs of immunopathology including excess immune cell infiltrates and epithelial cell necrosis. ADX mice infected with influenza virus alone also have increased inflammatory responses in the lung, but these responses are further increased in the co-infected ADX mice.
Bacterial complications from viral infections cause more morbidity and mortality than viral infections alone (Beadling and Slifka, 2004). We have delineated a mechanism by which infection with influenza virus, through the induction of GC, leads to suppression of the systemic immune response to a secondary bacterial infection and, therefore, an increased bacterial burden. However, we also found that the induction of GC is critical for survival of co-infection. When infected with influenza virus and L. monocytogenes, mice lacking the ability to produce GC had a greatly exacerbated inflammatory response, particularly in the lung, and did not survive co-infection. Collectively, our findings reveal a novel, inflammation-independent activation of the HPA axis by influenza virus infection, and demonstrate its physiological and pathological consequences for a systemic bacterial co-infection.
C57Bl/6 (B6) mice were obtained from the National Cancer Institute and the Jackson Laboratory and were between 6 and 12 weeks of age. ADX mice were obtained from Taconic laboratories. Rag2−/− and IFNαR1−/− mice were bred in the Yale University School of medicine animal facility in specific pathogen free conditions. GC levels in the serum were measured by the animal healthcare service Antech Inc. DEX was administered intraperitoneally at a daily dose of 1 mg per mouse starting 1 day before infection. All experiments were performed in accordance with the institutional animal care and use guidelines.
The A/WSN/33 (H1N1) strain of influenza virus was obtained from the laboratory of Dr. Akiko Iwasaki and was propagated using MDCK cells as described in (Okuda et al., 2001). The 10403S L. monocytogenes strain used was propagated in BHI agar as described in (Auerbuch et al., 2004). Influenza virus was administered intranasally at the indicated doses in a volume of 30 μl. Mice were infected with L. monocytogenes intravenously at the indicated doses in a volume of 100 μl.
Viral titres in the lungs were determined by titration of organ homogenates on MDCK cells as described in (Okuda et al., 2001). L. monocytogenes CFUs were determined by plating titrating doses of organ homogenate on Agar plates as previously described (Auerbuch et al., 2004).
Briefly, tissues were disrupted in RNA-bee (Teletest), and RNA was isolated according to manufacturer’s instructions. After treatment with DNase superscript III was used to produce cDNA. Quantitative PCR was done on eppendorf mastercycler with Qiagen syber green mix according to the manufacturer’s instructions. Serum cytokine levels were measured using the Luminex™ system according to manufacturer’s instructions.
Organs used for histology were fixed overnight in Formaldefresh ™ and embedded in paraffin for sectioning. Sections were stained with H&E according to the manufacturer’s instructions, and blind randomized scored. For each liver section 40 pictures were taken and the areas of infiltrate were counted.
Statistical analysis was performed using the Prism graphing and analysis software. Comparison of multiple groups was done using ANOVA and comparison of two groups was done with the student’s T test.
The authors would like to thank Sophie Holley for technical assistance; Gergana Galabova-Kovacs for advice on histological analysis; Takeshi Ichinohe, Melissa Lineham, and Akiko Iwasaki for viral strains and advice; Noah Palm and Igor Brodsky for critical reading of the manuscript. This work was supported by the Ellison Foundation, the Howard Hughes Medical Institute, and the NIH (AI055502, AI046688, and DK071754). A.M.J. is a Berger Foundation Fellow of the Damon Runyon Cancer Research Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.