Myeloid differentiation factor 88-dependent signalling controls bacterial growth during colonization and systemic pneumococcal disease in mice

Authors

  • Barbara Albiger,

    Corresponding author
    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
    • Contributed equally.

  • Andreas Sandgren,

    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
    • Contributed equally.

  • Hiroaki Katsuragi,

    1. Nippon Dental University, Niigata, Japan.
    Search for more papers by this author
  • Ulf Meyer-Hoffert,

    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
  • Katharina Beiter,

    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
  • Florian Wartha,

    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
  • Mathias Hornef,

    1. Institut Für Medzinische Mikrobiologie und Hygiene, Universität Freiburg, Freiburg, Germany.
    Search for more papers by this author
  • Staffan Normark,

    1. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author
  • Birgitta Henriques Normark

    1. SMITTSKYDDSINSTITUTET, Swedish Institute for Infectious Disease Control, Karolinska Institutet, Nobelsväg 18, Solna, 171 77 Stockholm, Sweden.
    2. MTC, Microbiology and Tumorbiology Center, Karolinska Institutet, Stockholm, Sweden.
    Search for more papers by this author

E-mail Barbara.Albiger@mtc.ki.se; Tel. (+46) 8457 23 91; Fax (+46) 830 25 66.

Summary

The Toll-like receptors (TLRs) and the myeloid differentiation factor 88 (MyD88) are key players in the activation of the innate immune defence during microbial infections. Using different murine infection models,  we  show  that  MyD88-dependent  signalling  is crucial for the activation of the innate immune defence against Streptococcus pneumoniae. Our data demonstrate that both local and systemic inflammatory response to S. pneumoniae depends on the presence of MyD88 to clear bacterial colonization of the upper  respiratory  tract  and  to  prevent  pulmonary  and systemic infection in mice. Finally, we described a strong correlation between enhanced bacterial growth in the bloodstream of MyD88-deficient mice and the inability to lower the serum iron concentration in response to infection.

Introduction

The Gram-positive bacterium Streptococcus pneumoniae represents one of the most frequent cause of community-acquired upper and lower respiratory tract (URT and LRT) infections ranging from mild otitis media to pneumonia with the potential to cause life-threatening systemic diseases (also referred to as invasive diseases) such as septicaemia and meningitis (Bogaert et al., 2004). S. pneumoniae also commonly colonizes the URT and is found in up to 50% of healthy pre-school children where it can be present for several months and represents a reservoir for infection (Henriqus Normark et al., 2003; Bogaert et al., 2004). It can therefore be considered part of the commensal microflora. However, a period of nasopharyngeal colonization is also thought to occur during the development of pneumococcal invasive disease (Bogaert et al., 2004; Tuomanen, 2004). Despite its medical importance, the microbial and host factors that influence the establishment of asymptomatic carriage and facilitate the transition to invasive infection are not yet fully understood.

The mammalian family of Toll-like receptors (TLRs) plays a crucial role in the host innate immune defence against microbial infections. They recognize the presence of microbial pathogens via the detection of conserved microbial structures or so-called pathogen-associated molecular patterns (PAMPs) (Takeda and Akira, 2004). Myeloid differentiation factor 88 (MyD88) is the central signalling adaptor that mediates cellular stimulation from most of TLRs as well as from the IL-1 receptor (IL-1R) and IL-18R (Yamamoto et al., 2004). This ultimately leads to the activation of the nuclear factor-κB (NF-κB) pathway and to the production of proinflammatory cytokines such as IL-1, IL-6 and TNF, chemokines such as IL-8 and upregulation of costimulatory molecules which initiate activation of the adaptive immune system (Takeda and Akira, 2004; Yamamoto et al., 2004). Additionally some TLRs, such as TLR3 and TLR4, signal via a MyD88-independent pathway leading to the activation of IRF3-dependent genes (Theofilopoulos et al., 2005).

The importance of the Toll/Interleukin-1 receptor signalling in human diseases was highlighted in recent clinical studies examining patients with IL-1 receptor-associated kinase 4 (IRAK-4) deficiency. IRAK-4 is an essential intracellular mediator downstream of MyD88 in the Toll/Interleukin-1 receptor-signalling pathway. IRAK-4-deficient patients exhibited an impaired NF-κB response to IL-1, IL-18 and to most of the TLRs ligands (Picard et al., 2003a) and were more prone to develop severe and invasive pneumococcal diseases (Picard et al., 2003a,b). Using gene-targeted mice lacking TLR2 or TLR4 expression, several recent studies have assessed the role of those receptors in experimental murine pneumococcal meningitis or pneumonia models. Although TLR2–/– mice were more susceptible to pneumococcal meningitis after intrathecal challenge, the inflammatory response was mainly TLR2-independent. Furthermore, TLR2–/– mice infected intranasally (i.n.) exhibited a modest reduced inflammatory response but an overall normal host defence when compared with wild-type (wt) mice. A distinct role for TLR4 has been suggested in pneumococcal infections, mediated by an interaction between pneumolysin, a pneumococcal cytolytic toxin, and TLR4. In the nasopharynx this interaction may be critically involved in the innate response to S. pneumoniae. However, it plays a limited role in the LRT defence (Echchannaoui et al., 2002; Koedel et al., 2003; Malley et al., 2003; Branger et al., 2004; Knapp et al., 2004). The specific role(s) played by individual TLRs in any given model requires further investigations and continues to be defined.

To investigate the role of MyD88 in the host control of carriage as well as in the development of invasive systemic disease, we infected MyD88-deficient mice with two clinical isolates of S. pneumoniae representing two clones primarily associated either with carriage or invasive disease in humans respectively. Our data show that MyD88-dependent signalling has essential functions in the bacterial clearance in the URT and the control of acute invasive pulmonary infection.

Results

Myeloid differentiation factor 88 significantly contributes to host defence mechanisms that control colonization of the respiratory tract and prevent invasive infection and proliferation of S. pneumoniae in the blood.

To assess the role of MyD88 in the respiratory tract, wt and MyD88–/– mice were inoculated i.n. with two pneumococcal strains, TIGR4 and ST16219F. Whereas pneumococcal strains such as the TIGR4 strain have been associated with invasive disease in humans, some other strains such as ST16219F strain are found in the nasopharyngeal microflora of asymptomatic healthy human carriers (Sandgren et al., 2004). Also in mice, TIGR4 is associated with invasive disease and ST16219F exhibits low virulence (Sandgren et al., in press).

Initially, wt and MyD88–/– mice were inoculated with both 1 × 107 colony-forming units (cfu) (high dose) and 1 × 105 cfu (low dose) of the TIGR4 strain. While none of the MyD88–/– mice survived the high dose infection, i.n. challenge  of  wt  mice  resulted  in  33%  survival  after  7  days (Fig. 1A). All surviving wt mice remained colonized in the URT at a stable viable bacterial count (5 × 104−5 × 105 cfu ml−1 nasopharyngeal lavages) for up to 10 days (data not shown). No further data were gathered at later time-points. Following low dose infection, almost all (90%) of the wt mice survived 15 days post challenge, whereas the survival rate of MyD88–/– mice was only 33% (Fig. 1B). At day 15 post infection (p.i.), nasopharyngeal lavages of the surviving mice were performed. While the surviving wt mice exhibited only low bacterial numbers in the URT, the few surviving MyD88–/– mice showed 2–3 log higher bacterial numbers (P < 0.005) (Fig. 1C). It is worth noticing that one wt mouse cleared pneumococci from the URT at 15 days p.i. (Fig. 1C) and up to 24% cleared the bacteria at 20 days p.i. (data not shown).

Figure 1.

MyD88–/– mice are more susceptible to pneumococcal intranasal challenge. Survival of wt and MyD88–/– mice after inoculation with either 1 × 105 or 1 × 107 cfu of S. pneumoniae. The survival was analysed by Kaplan-Meier log-rank test. MyD88–/– mice have higher bacterial titres in the URT. The results of individual mice are shown. Horizontal lines, medians. The counts were analysed by non-parametric Mann–Whitney test.
A. High dose challenge with TIGR4 (wt; n = 25 and MyD88–/–; n = 13). P = 0.005.
B. Low dose challenge with TIGR4 (wt; n = 10 and MyD88–/–; n = 10). P < 0.05.
C. cfu in nasopharyngeal-tracheal washes in survivors after low dose challenge with TIGR4 (wt; n = 9 and MyD88–/–; n = 4). P < 0.005.
D. High dose challenge with ST16219F (wt; n = 12 and MyD88–/–; n = 12). P < 0.0001.
E. cfu in nasopharyngeal-tracheal washes in survivors after high dose challenge with ST16219F (wt; n = 12 and MyD88–/–; n = 1).

When challenged with a high dose (1 × 107) of ST16219F, all wt animals survived (Fig. 1D). Half of the wt mice remained colonized in the upper airways at 15 days p.i., as evidenced by positive cultures from nasopharyngeal lavages, while the other half cleared pneumococci from the URT (Fig. 1E). In contrast, infected MyD88–/– mice exhibited significant morbidity and mortality. Although a delayed disease progress was observed following infection with ST16219F as compared with the above-described TIGR4 strain, all but one MyD88–/– mouse had succumbed to the infection after 7 days (Fig. 1D). Taken together, these results reveal that MyD88 plays an essential role in host resistance in the respiratory tract and show that MyD88-signalling has a strong influence on the control of pneumococcal growth in the respiratory tract, preventing spread from the respiratory tract to the bloodstream.

Subsequently, bioluminescent TIGR4 and ST16219F strains expressing luciferase were engineered, as described in Experimental procedures, to follow the kinetics of the systemic bacterial spread over time. No difference in infection kinetics was observed between the luciferase expressing and the isogenic parental bacterial strains (Fig. 1A and D and data not shown). Wt and MyD88–/– mice were infected with a high challenge dose (1 × 107) of TIGR4 and ST16219F and light emission images were taken on a daily basis p.i. (Fig. 2A and B). Light emission images illustrated the marked difference between wt and MyD88–/– mice following pneumococcal infection. Challenge with the TIGR4 strain resulted in significant bacterial proliferation within the LRT in wt and MyD88–/– mice but demonstrated a much faster dissemination of bacterial infection in MyD88–/– mice. In contrast, the less virulent ST16219F strain led to confine nasopharyngeal colonization in all wt mice in the absence of mortality. Interestingly, significant bacterial proliferation and subsequent systemic dissemination within the respiratory tract was noted in MyD88–/– mice.

Figure 2.

Real-time in vivo imaging using bioluminescent strains of S. pneumoniae. The wt and MyD88–/– mice were challenged with a high infectious dose (1 × 107 cfu). Pictures from individual mice of one representative experiment are shown. The same mouse is found at the same position (1–10) at the various time-points during the experiment. The missing mice have succumbed to the infection. Quantification of emitting light was performed on the pictures using the LivingImage software. The data were analysed by unpaired t-test.
A. Xenogen pictures using TIGR4 tagged with luciferase in wt and MyD88–/– mice (wt; n = 10 and MyD88–/–; n = 10). At day 3, the MyD88–/– mouse at the position 7 has been misplaced and put at the position 8. At day 4, the wt mouse at the position 2 has been misplaced and put at the position 1.
B. Xenogen pictures using ST16219F tagged with luciferase in wt and MyD88–/– mice (wt; n = 10 and MyD88–/–; n = 10).

To analyse whether the higher susceptibility of MyD88–/– mice to develop lethal pneumococcal infection resulted from an uncontrolled bacterial growth in the respiratory tract, the bacterial load in the respiratory tract was determined by quantification of the light emission (Fig. 2A and B). At day 1 p.i., MyD88–/– mice infected with TIGR4 showed 1.2-fold higher emission of light as compared with wt mice. However, at day 2 p.i., most of the MyD88–/– mice were already highly bacteraemic or had succumbed to infection and therefore no statistical comparison could be performed. In addition, the bacterial viable counts in nasopharyngeal and in broncho-alveolar lavages (BAL) of TIGR4 infected mice sacrificed at day 1 after high dose challenge were determined by plating (Fig. 3A and B). In accordance with the light emission measurements, MyD88–/– mice had 2.5-fold more bacteria in the URT as compared with wt mice (P < 0.05) (Fig. 3A). In the BAL, MyD88–/– mice had 132-fold more bacteria (Fig. 3B). Thus, the higher susceptibility of MyD88–/– mice to TIGR4 was primarily a consequence of uncontrolled bacterial growth in the LRT.

Figure 3.

MyD88–/– mice have higher bacterial titres in the trachea and in the BAL at day 1 after intranasal inoculation with 1 × 107 cfu of TIGR4. The results of individual mice are shown. Horizontal lines, medians. The counts were analysed by non-parametric Mann– Whitney test.
A. cfu in nasopharyngeal-tracheal washes (wt; n = 10 and MyD88–/–; n = 10).
B. cfu in BAL (wt; n = 10 and MyD88–/–; n = 10). P < 0.05.

Similarly, MyD88–/– mice infected with ST16219F presented higher light emission at day 1 post challenge in the respiratory tract as compared with wt mice (Fig. 2B). Determination of light emission showed that MyD88–/– mice emitted 1.3-, 5.8- (P < 0.05) and 32.2- (P < 0.05) fold higher light as compared with wt mice at day 1, 2 and day 3 p.i. Thus, the higher susceptibility of MyD88–/– mice to ST16219F was a consequence of uncontrolled bacterial growth, both in the URT and in the LRT.

The higher susceptibility of MyD88–/– mice to pneumococcal challenge correlated with the development of severe bacteraemia. As early as day 1 after high dose i.n. infection with TIGR4, a significant difference in the rate of bacteraemic individuals between wt and MyD88–/– mice was observed (52% vs. 69% respectively; Fig. 4A). At day 2 p.i., all MyD88–/– mice had developed severe bacteraemia with significantly higher bacterial numbers in blood (P < 0.005) while 36% of wt mice were still non-bacteraemic. In contrast, none of the MyD88–/– or wt mice infected with ST16219F were bacteraemic at day 1 p.i. However, while the wt mice remained non-bacteraemic, as many as 42% of the MyD88–/– mice showed positive blood culture at day 2 p.i. Bacteraemia in the MyD88–/– mice infected with ST16219F was retarded as compared with infection with the invasive TIGR4 strain (Figs 2A, B and 4A, B), in agreement with the delayed mortality (Fig. 1D) but preceded rapid death in all cases. These data suggest that MyD88 may also be involved in the activation of the host defence in the LRT controlling systemic spread of S. pneumoniae from the lungs.

Figure 4.

MyD88–/– mice have a more severe bacteraemia after intranasal inoculation with 1 × 107 cfu of S. pneumoniae. Results from individual mice of one representative experiment are shown. Horizontal lines, medians. The counts were analysed by non-parametric Mann–Whitney test.
A. cfu in blood after i.n. challenge with TIGR4 (wt; n = 25 and MyD88–/–; n = 13).
B. cfu in blood after i.n. challenge with ST16219F (wt; n = 12 and MyD88–/–; n = 12).

Myeloid differentiation factor 88-signalling is required for local cytokine production, leukocyte infiltration and tissue destruction

The entry of polymorphonuclear leukocytes (PMNs) to the site of infection is essential for early bacterial control and clearance. In the absence of local leukocyte infiltration, bacteria rapidly cause a disseminated infection (Rijneveld et al., 2002). Therefore, the proportion of PMNs among the cellular constituents in the BAL of mice sacrificed at day 1 p.i., with a high dose challenge of TIGR4, was determined by FACS analysis. MyD88–/– mice showed significantly lower numbers of neutrophils in the airway lumen as compared with wt mice (P < 0.01) (Fig. 5A and C).

Figure 5.

MyD88–/– mice present less inflammation in the lungs. Data are presented as means and SEM. The counts were analysed by unpaired t-test.
A. The polymorphonuclear cell (PMN) in the total cell population from the BAL day 1 p.i. The fraction of Ly-6G positive cell was determined by FACS analysis. P < 0.001.
B. PMN counts in the lung tissue day 1 p.i. were analysed by immunohistochemistry staining. P = 0.05
C. Immunohistochemical staining for Ly-6G positive cells in lung tissue sections day 1 p.i. Counterstaining was performed with haematoxylin (H). Magnification ×100. Pictures from representative lungs are shown.
D. IL-6 responses in the lungs after intranasal challenge with 1 × 107 cfu of TIGR4. P < 0.001.
E. KC responses in the lungs after intranasal challenge with 1 × 107 cfu of TIGR4. P < 0.001.
F. Histology of lung tissue of wt and MyD88–/– mice after intranasal inoculation with 1 × 107 cfu of TIGR4. Counterstaining was performed with H. Magnification ×400. Pictures from representative lungs are shown.

In addition, the number of PMNs infiltrating the lung tissue was determined by immunostaining. At day 2 and 3 p.i., MyD88–/– mice showed a significantly lower number of infiltrating PMNs as compared with wt mice (Fig. 5B and C).

The production of proinflammatory cytokines (TNF, IL-1, IL-6) and chemokines (MIP-2, KC) is crucial to induce inflammatory infiltration and protective immunity against bacterial pathogens. IL-6 is a multifunctional proinflammatory cytokine that plays an important role both in local as well as systemic immune stimulation. Elevated IL-6 levels have been observed during bacterial infections. KC is a potent neutrophil attractant and activator and its expression is associated with neutrophil influx in inflamed tissue. At day 1 after i.n. challenge with a high dose (1 × 107 cfu) of TIGR4, MyD88–/– mice showed significantly lower levels of IL-6 and KC in the lung homogenate as compared with wt mice (Fig. 5D and E). Similarly, the levels of IL-6 and KC in BAL of infected MyD88–/– mice were significantly reduced (data not shown).

The histopathological examination was consistent with the entry of professional immune cells into the lung tissue and the local production of cytokines (Fig. 5F). In wt animals, a marked inflammatory reaction with extensive recruitment of leukocytes, local bleeding, and alveolar exudates was noted in the lung tissue. Large organ areas were severely congested and collapsed with accumulation of amorphic material from desquamation of the respiratory epithelium in the bronchi leading to airway obstruction. In contrast, the lungs of infected MyD88–/– mice revealed a reduced inflammatory reaction with lower leukocyte infiltration at all time-points examined. The alveoli had a uniform and airy appearance without the congestion and the amorphic intraluminal material seen in wt mice. No oedema or bleeding was observed accompanied with reduced parenchymal inflammation.

Together, our data demonstrate a reduced cytokine production, impaired neutrophil infiltration, and enhanced bacterial growth in the absence of MyD88. The impaired local inflammatory response in the lung tissue of MyD88–/– mice might allow uncontrolled bacterial growth and eventually lead to precocious systemic spread. This in turn could explain the higher susceptibility of MyD88–/– mice to pneumococcal infection.

Myeloid differentiation factor 88 is required for a systemic immune response and for control of bacterial proliferation in the bloodstream

To study the systemic phase of the infection and to assess the ability of the pathogen to withstand the antibacterial clearing capacity in the bloodstream and internal organs like the spleen, wt and MyD88–/– mice were inoculated intraperitoneally (i.p.) with either a low (5 × 104) or a high (5 × 106) dose of TIGR4 (Fig. 6A and B). At 10 h following high dose inoculum, all wt and MyD88–/– mice had to be sacrificed for ethical reasons due to the clinical status. At this time-point bacterial numbers in the blood stream were more than one log higher in MyD88–/– mice as compared with wt mice (Fig. 6A). Conversely, MyD88–/– mice had significantly lower clinical scores until 9 h p.i., as compared with wt animals (data not shown).

Figure 6.

Role of MyD88 in bacterial clearance and in systemic host defence after intraperitoneal challenge with S. pneumoniae TIGR4. Wt and MyD88–/– mice were infected i.p. with either a low dose (5 × 104 cfu S. pneumoniae) or a high dose (5 × 106) (wt; n = 5 and MyD88–/–; n = 5 for each time-points).
A. Bacterial outgrowth in blood in wt and MyD88–/– mice after high dose i.p. challenge. Values are resented as medians and were assessed by non-parametric Mann–Whitney test.
B. Bacterial outgrowth in blood in wt and MyD88–/– mice after low dose i.p. challenge. Values are resented as medians and were assessed by non-parametric Mann–Whitney test. P < 0.001 and P < 0.05.
C. Serum TNF response to high dose systemic infection. Data are presented as means and SEM. The counts were analysed by unpaired t-test. P < 0.001 and P < 0.05.
D. Serum IL-6 at 8 h p.i. Data are presented as means and SEM. The counts were analysed by unpaired t-test. P < 0.005

At low dose i.p. challenge, bacterial numbers in the blood increased at similar rates during the first 6 h in both wt and knock-out mice. Interestingly, bacterial growth in blood ceased in wt animals at this time-point after infection. In contrast, bacterial numbers continued to increase in the blood of MyD88–/– mice. As a result, mutant mice exhibited about 10-fold higher bacterial numbers as compared with wt mice at 14 h p.i. (Fig. 6B). Similar to the high dose bacterial challenge, MyD88–/– mice had significantly lower clinical scores as compared with wt animals until 14 h p.i. (data not shown). However, the median survival time for MyD88–/– and wt mice was 16 h and 21 h respectively (P < 0.01).

Tumour necrosis factor (TNF) and IL-6 are proinflammatory cytokines and represent prognostic markers for sepsis. TNF and IL-6 serum levels were therefore determined in systemically infected mice with high i.p. challenge of TIGR4. While wt mice mounted a marked time-dependent response following infection, MyD88–/– mice showed no detectable amounts of serum TNF (Fig. 6C). Similarly, the determination of serum IL-6 levels demonstrated significantly lower concentrations in MyD88–/– as compared with wt mice at 8 h p.i. (Fig. 6D). Those results are in agreement with previous studies (Koedel et al., 2004; Khan et al., 2005).

These data show that MyD88–/– mice are severely impaired both in the production of systemic cytokines and in the control of bacterial growth in the bloodstream. In addition, the absence of significant proinflammatory cytokines production in MyD88–/– mice might illustrate the association of the activation of the innate immune system with the exhibition of clinical signs and symptoms of infection.

Enhanced bacterial proliferation in MyD88–/– mice correlates to higher serum iron levels

Iron is an essential cofactor for many basic cellular functions and metabolic pathways of microorganisms and its access in the host is a critical factor that pathogens have to overcome in order to grow and cause infection (Brown and Holden, 2002; Schaible and Kaufmann, 2004). Anaemia due to iron deficiency is frequently seen in patients with chronic infectious diseases or inflammation. It represents a way for the host to limit the availability of iron to the pathogen and thereby restrict microbial growth. We hypothesized that the enhanced bacterial growth within the bloodstream of MyD88–/– mice could be due to an impaired control of free serum iron as a direct or indirect consequence of lower IL-6 levels (Figs 6D and 7A). It has been shown that IL-6 induces hepcidin, a recently discovered peptide produced in the liver. Synthesis of hepcidin leads to hypoferraemia during infection and inflammation (Nemeth et al., 2003; Schaible and Kaufmann, 2004). Serum iron levels were therefore determined in wt and MyD88–/– mice following a low dose (1 × 105) i.p. challenge with TIGR4 (Fig. 7B). At 8 h p.i., both wt and MyD88–/– mice showed an increase in serum iron as compared with non-infected healthy controls, probably due to liberation of iron from damaged cells following infection. However, at 12 h p.i., the serum iron level in wt mice decreased to reach similar levels as in non-infected control mice (Fig. 7B). It is worth noticing that the reduction of serum iron levels in the wt between 8 h and 12 h p.i. corresponded temporally to the above mentioned growth arrest observed in infected wt mice (Fig. 6B). Interestingly, this decrease of free iron was not detected in MyD88–/– mice. In contrast, a continuous increase of free iron was seen at 12 h p.i. and at this time-point the median iron concentration in serum was twofold higher than in non-infected control mice (P < 0.05). Finally, serum hepcidin levels were determined in wt and MyD88–/– mice following infection (Fig. 7C). In accordance with the observed elevation of IL-6 secretion (Fig. 7A), infected wt mice exhibited a significant elevation of the hepcidin concentration already 8 h post challenge. In contrast, no enhancement of hepcidin levels was seen in MyD88–/– mice at 8 h and 12 h post challenge (Fig. 7C). Thus, our data demonstrate a correlation between elevated IL-6 and hepcidin levels, reduced free iron availability, and control of bacterial proliferation in the blood of wt animals. In contrast, MyD88–/– mice lacked enhanced IL-6 and hepcidin secretion and showed elevated iron levels and enhanced mortality to bacterial infection. These data suggest that the enhanced bacterial growth within the bloodstream of MyD88–/– mice could at least in part be due to an impaired bacterial growth control represented by a high serum iron availability.

Figure 7.

Role of MyD88 in induction of hepcidin and in serum iron. Wt and MyD88–/– mice were infected i.p. with a low dose (5 × 104 cfu).
A. Serum IL-6 at 8 h p.i. (wt; n = 5 and MyD88–/–; n = 6). Data are presented as means and SEM. The counts were analysed by unpaired t-test. P < 0.05.
B. Free serum iron at 8 h (wt; n = 5 and MyD88–/–; n = 7) and 12 h p.i. (wt; n = 5 and MyD88–/–; n = 8). P < 0.05. Results of individual mice are shown. Horizontal lines, means. The counts were analysed by unpaired t-test. The mean serum iron in control mice is in the same range than previously published by (Rivera et al., 2005).
C. Serum hepcidin level at 8 h (wt; n = 5 and MyD88–/–; n = 6) and 12 h (wt; n = 5 and MyD88–/–; n = 8). P < 0.001. Results of individual mice are shown. Horizontal lines, means. The counts were analysed by unpaired t-test.

Discussion

Respiratory tract infections caused by S. pneumoniae are typically associated with an acute inflammation illustrated by massive infiltration of PMNs and secretion of high levels of stimulating mediators (Bogaert et al., 2004; Tuomanen, 2004). These clinical observations suggest a prominent role of the innate immune system in the pathogenesis of pneumococcal disease. However, the importance of specific pattern recognition receptors (PRRs) in the immune defence activation against pneumococci in any given model requires further investigations (Echchannaoui et al., 2002; Koedel et al., 2003; Malley et al., 2003; Branger et al., 2004; Knapp et al., 2004). The ability of TLR1, TLR2 and TLR6 to recognize bacterial components of Gram-positive microorganisms have made them obvious candidates for playing a role during pneumococcal infection (Takeda and Akira, 2004). In vitro studies have clearly demonstrated that TLR2 recognizes pneumococcal lipoteichoic acid (LTA) (Yoshimura et al., 1999), although it seems less potent than staphylococcal LTA (Han et al., 2003). However, challenge of TLR-deficient mice with S. pneumoniae has shown only a modest role of individual receptor molecules in the immune defence (Echchannaoui et al., 2002; Koedel et al., 2003; Malley et al., 2003; Branger et al., 2004; Knapp et al., 2004). The limited effect observed in studies using single TLR-deficient mice might illustrate the redundancy in the recognition of Gram-positive PAMPs by different innate immune receptors simultaneously in vivo.

In the present study, we have focused on the central TLR-signalling adaptor MyD88 to examine the importance of the innate immune system during different steps of pneumococcal pathogenesis, namely URT colonization, pulmonary infection and systemic spread. We have selected two murine infection models: (i) a model of nasopharyngeal challenge, mimicking the natural route of infection reflecting the transition from the respiratory tract to the blood stream causing an invasive disease and (ii) an i.p. challenge, as a model for systemic infection. Certain S. pneumoniae strains such as ST16219F have rarely been associated with invasive diseases but seem to asymptomatically colonize the URT (Sandgren et al., 2004). In contrast, other pneumococcal strains such as TIGR4 have mainly been isolated from patients with invasive infection (Sandgren et al., 2004). A similar disease correlation of these two bacterial strains has also been observed in mice (Sandgren et al., in press). Both, the ST16219F and the TIGR4 strain have been used in the present study to mimic the human situation as reflected by asymptomatic carriage or invasive infection.

Here, we show that MyD88–/– mice have a significantly increased susceptibility to pneumococcal infection associated with enhanced morbidity and mortality. The increased susceptibility was illustrated by the higher degree of bacterial colonization within the URT, enhanced bacterial proliferation in infected lung tissue, precocious bacterial spread into the bloodstream, and increased mortality. The enhanced susceptibility was observed both at low and high infectious inoculums. The protective role of MyD88 indicates that signalling from either TLRs and/or IL-1R/IL-18R are critically involved in the host immune activation during pneumococcal infection. This is in agreement with earlier studies showing that IL-1R knock-out mice treated with blocking TNF antibody were highly susceptible to i.n. pneumococcal infection (Rijneveld et al., 2001).

A colonization phase is likely to precede the development of pneumococcal pneumonia and systemic disease in humans. Similar to the situation in the human population, the carrier strain ST16219F asymptomatically colonized wt mice. All wt mice survived and either remained colonized or cleared pneumococci without developing systemic disease. This clinical picture was also seen in wt mice that survived a low dose TIGR4 challenge. In contrast, the high dose infection with the TIGR4 strain resulted in significant mortality even in wt animals. Interestingly, wt mice that survived this high dose TIGR4 challenge also remained colonized in the URT, however, no secondary infection was noted for up to 10 days post challenge. Thus, the applied murine pneumococcal infection model using two human clinical isolates associated with either asymptomatic carriage or invasive disease reflects the situation of pneumococcal infections in humans. Thereby it allows studying the involvement of host factors in the initial host defence. Furthermore, these results underline the important contribution of strain specific virulence factors to the pathogenesis of pneumococcal disease.

In sharp contrast to the wt situation, the MyD88–/– mice were severely affected by carrier strain ST16219F resulting in systemic infection with significant mortality. Persistent colonization of the respiratory tract was almost never seen in the absence of systemic disease. Induction of chemokine secretion contributed to the recruitment of PMNs to the site of infection, the lung tissue. Therefore, the lack of proinflammatory cytokines and chemokine secretions seen in MyD88–/– mice might explain the enhanced systemic spread. Thus, the innate immune system seems to play a critical role to confine the infection with this carrier strain to the respiratory tract. Lack of local innate immune control renders this low pathogenic strain into a highly virulent bacterium. Interestingly, unlike in pneumococcal infection, MyD88-dependent signalling is not required to control a Staphylococcus aureus pneumonia after i.n. challenge (Gomez et al., 2004) even though local cytokine responses were also blunted. It has recently been shown that staphylococcal protein A can directly bind to the TNFR1 receptor on pulmonary epithelial cells eliciting a signalling response resulting in neutrophil entry and bacterial clearance (Gomez et al., 2004). It is possible that a direct activation of the TNFR1 receptor through protein A can substitute for the absence of TNF in MyD88–/– mice and explain the notable difference in MyD88-mediated signalling in the innate defence against pneumococcal and staphylococcal pneumonia.

We also demonstrate the importance of MyD88-mediated host defence activation during systemic infection. Similarly to two recent studies demonstrating enhanced susceptibility of MyD88–/– mice to intrathecal and i.p. injection of S. pneumoniae (Koedel et al., 2004; Khan et al., 2005), we attributed the increased susceptibility to a compromised inflammatory response associated with bacteraemia. In addition we could identify a potentially important effector mechanism of systemic antibacterial control. During bacterial infection and inflammation, humans restrict iron uptake and sequester iron within macrophages, a mechanism known to lead to anaemia of inflammation (Jurado, 1997). Iron represents an essential bacterial metabolite and iron limitation restricts pneumococcal proliferation (data not shown) and virulence (Tai et al., 1993; Brown and Holden, 2002; Schaible and Kaufmann, 2004). IL-6 has recently been shown to be required for the induction of hepcidin, a highly conserved antimicrobial peptide secreted by the liver that lowers free serum iron (Nemeth et al., 2003). Hepcidin is an iron-regulatory hormone and represents an important link between host defence, inflammation and iron metabolism (Ganz, 2005). In mice, fish and humans, hepcidin synthesis is induced by infection and inflammation (Shike et al., 2002; Nemeth et al., 2003; Lauth et al., 2005). Careful analysis of the time kinetic of serum iron and bacteraemia revealed that the induction of a hypoferraemic response in wt animals precisely coincided with elevated level of hepcidin and reduced bacterial growth. In contrast, MyD88–/– mice showed unrestricted serum iron and enhanced bacterial proliferation between 8 and 12 h p.i., which correlated to lower hepcidin levels. We hypothesize that MyD88-mediated signalling via IL-6 and hepcidin reduces the free serum iron levels in order to restrict proliferation of invading pathogens.

In conclusion, we demonstrate the central role of MyD88-mediated cell signalling for the induction of local and systemic cytokine secretion and leukocyte recruitment in pneumococcal infections. Also, we identified a strong correlation between the induction of hepcidin and serum iron restriction suggesting iron limitation as important effector mechanism of the systemic innate host defence to control pneumococcal proliferation. Finally, we show that one major role of MyD88 in the local innate defence of healthy individuals might be to control bacterial growth in the respiratory tract to maintain an asymptomatic carriership which in turn might ultimately lead to bacterial clearance. In patients with immunodeficiency of the innate host recognition, failure of pneumococcal control might thereby lead to recurrent and invasive bacterial infections (Medvedev et al., 2003).

Experimental procedures

Bacterial strains

We used two pneumococcal strains with different capacity to cause invasive disease in humans. S. pneumoniae TIGR4 is a clinical encapsulated isolate of serotype 4, which recently has been sequenced by The Institute for Genomic Research TIGR (ATCC BAA-334; http://www.tigr.org). TIGR4 belongs to a major clone, SWE4-1, of sequence type, ST205 that accounted for 8.4% of the invasive isolates from Sweden in 1997. In a previous study, we showed that the isolates of the ST205 clone have an odds ratio (OR) > 1 and were never found among healthy carriers, suggesting a high potential to cause invasive disease (Sandgren et al., 2004); OR corresponds to the odds ratio for invasiveness of the clone [OR = (ad )/(bc)], where a was the number of invasive X clones, b was the number of carriage X clones, c was the number of invasive non-X clones, and d was the number of carriage non-X clones). ST16219F is a clinical encapsulated isolate of serotype 19F that belongs to a clone with sequence type (ST) 162. ST16219F has previously been identified as a hyper-colonizing clone in mice (Sandgren et al., in press). In our previous study, isolates of type 19F were common among carriers and rarely caused invasive pneumococcal disease, and therefore showed a low disease potential in humans (Sandgren et al., 2004).

For in vivo challenges, pneumococci were grown overnight from frozen stocks on blood agar plates at 37°C and 5% CO2. Colonies were inoculated into the semi-synthetic medium C+Y and grown to midlogarithmic phase (OD620 = 0.5). Appropriate dilutions were made to obtain the desired concentration. The concentration was retrospectively confirmed by viable counting on blood agar plates.

Construction of pneumococcal strains expressing luciferase

Stable bioluminescent derivatives of the clinical isolates were created using chromosomal DNA from the highly bioluminescent S. pneumoniae D39 Xen 7 strain (Francis et al., 2001). D39 Xen 7 carries the Tn4001 luxABCDE Kmr cassette, which encodes for luciferase and its substrate. Transformation was performed using a mixture of competence stimulatory peptides CSP1 and CSP2 (100 ng ml−1) and was performed as previously described (Bricker and Camilli, 1999; Francis et al., 2001). Transformants were selected by plating on blood agar plates containing 400 µg ml−1 of kanamycin and screened for bioluminescence using the IVIS (CCD camera, Xenogen).

Mice

Six to eight-week-old C57BL/6 (wt) and MyD88-deficient mice (mice had been backcrossed for six generations on the C57BL/6 background) were kept under specific pathogen-free conditions. The animals were housed five per cage in a standardized 12 h light/dark cycle and received commercial food and water ad libitum. All the experiments were conducted in conformity with the European Communities Council Directive 86/609/EEC and the Swedish animal protection legislation. All animal groups were aged and sex-matched for each experiments.

Mouse challenge

Mice were anaesthetized by light inhalation of isoflurane (Forene, Abbott) and subsequently inoculated either i.p. or i.n. with control medium or live S. pneumoniae (5 × 104−1 × 105 cfu or 5 × 106−1 × 107 cfu) in a 200 µl and 20 µl volume respectively.

For i.n. challenge, the mice were held after inoculation in a vertical position for 1 min. The mice health status and survival were assessed during 7 days by two independent observers and clinical scores were given. The clinical score comprises the following criteria: (i) healthy and exhibited normal motor activity; (ii) showed a decreased spontaneous activity and hunched posture; (iii) presence of tremor and piloerection and loss of vigilance; (iv) turned upright in > 5 s; (v) positional passivity and did not turn upright and (vi) did not move or moribund. Mice were sacrificed at 24 h and 48 h after challenge or when they reached a clinical score of > 3–4. After i.p. challenge, mice were sacrificed at 4 h, 6 h, 8 h, 10 h, 12 h and 14 h p.i.

To evaluate bacteraemia, blood samples (5 µl) were obtained from the tail vein at various time-points (cfu ml−1 blood).

The lungs were removed, weighted and homogenized in 1 ml of PBS containing complete cocktail protease inhibitor (Roche Diagnostics Scandinavia) and used for serial plating to quantify viable bacteria (cfu mg−1 of lung). The homogenates were centrifuged at 4°C for 30 min at 5000 r.p.m. The supernatants were stored at −80°C for later cytokine analysis.

Broncho-alveolar lavage was sampled by lavaging the lung with 2 × 1 ml of PBS containing complete protease inhibitor cocktail (Roche diagnostics). Previous to BAL sampling, the lungs were perfused through the heart with 2.5 ml of PBS + 5 mM EDTA. BAL was centrifuged at 4°C for 30 min at 500 r.p.m and the supernatants were stored at −80°C for later cytokine analysis. Total cell numbers were counted from each sample and differential cell counts were done by FACS analysis.

For colonization assays, nasopharyngeal-tracheal lavages were performed in animals post-mortem with a 20-gauge catheter inserted into the proximal trachea, flushing the nasopharynx through the trachea and the nares with 200 µl PBS and used for serial plating to quantify viable bacteria (cfu ml−1 of tracheal wash).

For in vivo imaging with the IVIS CCD camera, the mice were anaesthetized by a continuous light inhalation of isoflurane and imaged for 5 min at 24 h, 48 h, 72 h, 96 h and 168 h after challenge. Total photon emission from the regions of interest within the images of each mouse was quantified using the LivingImage software package. Note that the data are strictly numerical, in the form of photons emitted per unit area (Xenogen Corporation).

FACS analysis

The fraction of neutrophils in the total cell population from the BAL was determined by FACS analysis using APC-conjugated rat anti-mouse Ly-6G (BD Biosciences) and FITC-labelled rat anti-mouse F4/80 antibodies (serotec).

Histology and immunochemistry

Upper parts of the left lung were fixed in 4% formalin and were embedded in paraffin. Slices of 5 µm were deparaffinized through a series of xylene and graded ethanol baths followed by a rinse with bidest, and stained with haematoxylin and eosin (H&E). A pathologist analysed the sections in blinded manner.

Polymorphonuclear cell counts in the lung tissue were analysed by immunohistochemistry staining with an anti-Ly-6G monoclonal antibody (BD Biosciences). Counterstaining was performed with H&E. Magnification ×400. Three sections of lung tissue per mice were photographed at a magnitude of ×200 and then analysed with NIHimage software using a density slice and counting particles.

Determination of cytokine levels and serum iron

Tumour necrosis factor, IL-6 and KC in serum, in BAL and in lung were measured by using commercial ELISA kits (R&D systems, Diaclone and BD Biosciences). Serum hepcidin levels were assessed with commercial ELISA kits (DRG Diagnostics, Hamburg, Germany). Serum iron was analysed using the Roche Analytics for the Serum Work Area at the clinical laboratory of Huddingue, Sweden. Using ascorbic acid, the iron was reduced to Fe2+, which then forms a complex with ferrozin. The amount of serum Fe was determined by absorption spectrophotometry.

Statistical analysis

Data of cfu were analysed using the non-parametric Mann–Whitney test. The TNF, IL-6, KC, hepcidin responses, serum iron levels and cell counts were analysed using the student t-test. Two-tailed tests were used. For quantification of photon emission (in units of radiances), Living Image software from Xenogen was used, and comparisons were made with student t-test. For survival studies, differences were analysed by the Kaplan-Meier analysis log-rank test. A P < 0.05 was considered significant.

Acknowledgements

This work was supported by grants from the Swedish Medical Research Council, Cancerfonden, Augusta Hedlund Foundation, Svenska Läkaresällskapet, Magnus Bergvall Foundation, Deutsche Forschungsgemeinschaft Me 2037/2-1 and EU 6th framework program (PREVIS project). We thank Prof. Tuomanen, St Jude Children's Research Hospital, Memphis, USA for kindly providing the strain TIGR4, and Prof. Akira, Dr Takeda and Dr Takeuchi, Department of Host Defense, Research Institute For Microbial Diseases, Osaka, Japan for the MyD88-deficient mice. Ingrid Andersson, Christina Johansson and Gunnel Mölleberg are also greatly acknowledged for skilful technical assistance, Anna Törner for statistical analysis assistance and Prof. Peter Biberfeld for pathology analysis.

Ancillary