This study was supported by a grant of the excellence program of the Landesstiftung Baden Württemberg. CH was supported by a M. von Wrangell stipend for habilitation.
7 Summarizing Discussion
Infections with the respiratory pathogen Chlamydophila pneumoniae mostly have an asymptomatic course. In many cases, C. pneumoniae are not eradicated by the immune system, but persist in host tissue for many years and establish essentially silent chronic infections (99, 246). Persistent infections with C. pneumoniae are discussed as risk factors for the pathogenesis of numerous chronic inflammatory diseases, most prominent atherosclerosis (30, 161). Since the initial recognition of bacteria by the innate immune system is crucial for the subsequent induction of inflammatory responses and elimination of the pathogen, it is challenging to identify the deviations of normal pathogen recognition which allow C. pneumoniae to escape the immune system and to establish persistent infections.
For C. pneumoniae, immune recognition has been described to depend in vitro on TLR2 but also on TLR 4 (217). In line with this, we found that the in vitro stimulated cytokine release was strongly impaired in bone marrow macrophages derived from TLR2 knock-out mice and also partially diminished in C3H/HeJ mice, which express a non functional TLR4.
To investigate the role of TLR2 and TLR4 for C. pneumoniae infection in vivo, a mouse model was established. In this model, mice were infected intranasally with low doses of C.
pneumoniae EB and the infection was monitored by serology and by determining the bacterial burden of organs via a highly sensitive, quantitative real-time PCR. The low inoculum of 106 C. pneumoniae EB led to a mild asymptomatic course of infection and appeared thereby to reflect the clinical picture in humans. C. pneumoniae infected only the lungs and were not spread to other organs. Furthermore, no persistence was observed.
Surprisingly, the course of infection, i.e. the bacterial burden of the lung, BAL and antibody development was not significantly different between TLR2-/- and C3H/HeJ mice compared to their respective wild types. In contrast, the in vivo TNF and IL-6 but not anti-inflammatory IL-10 release was impaired in TLR2-/- mice, while C3H/HeJ mice responded similar to wild-types for all measured cytokines. These in vivo findings confirm the in vitro data that at least TLR2 is necessary for C. pneumoniae-induced cytokine release, but they also show that TLR2 and TLR4 are dispensable for the eradication of C. pneumoniae in vivo. These results might be explained by receptor redundancy (263), meaning that the induction of inflammatory responses is not only mediated by activation of one receptor, but via an assembly of multi-component receptor clusters in the membrane (211, 260, 261).
TLRs of mammalians are known to signal through the cytoplasmatic adapter protein MyD88 (175). The fact that knock out of MyD88 abrogates responses towards bacterial
compounds like LPS (137) or peptidoglycan (255), and renders mice highly sensitive to infections (62, 253) supports the hypothesis that more than one receptor has to be blocked to impair anti-bacterial defence. Only recently, C. pneumoniae-induced IFNγ release has been shown to depend on MyD88 but not on TLR2 or TLR4 in human PBMCs (198). The ability of chlamydial hsp60 to induce the CXC chemokines KC and MIP2 and PMN accumulation, was completely abolished in MyD88-deficient and TLR2/TLR4 double deficient mice in vivo, while TLR single deficient mice responded almost normally (47). The failure to induce MIP2 and PMN accumulation was also observed after intranasal infection of MyD88-deficient mice with C. pneumoniae (229). Therefore, it would be interesting also to investigate C. pneumoniae infection in TLR2/TLR4 double deficient mice.
Currently the intracellularly located NOD proteins are discussed as pattern recognition receptors (PRR) for C. pneumoniae. A role for NOD has so far only been reported in one study using NOD1 and NOD2 over-expressing epithelial cells (203). In the present study, we investigated the three so far known NOD2 polymorphisms, i.e. the NOD2 3020insC frame shift mutation as well as the 2104 and 2722 missense mutations upon stimulation with C. pneumoniae. Cytokine induction by monocytes of subjects with heterozygous NOD2 3020 polymorphism was significantly increased, while the other two polymorphisms had no effect, indicating that the NOD2 3020 mutation may specifically affect the C.
pneumoniae binding region. In addition, the NOD2 3020 was shown to be a risk factor for atherosclerosis in patients which are seropositive for C. pneumoniae by R. Schumann’s group. Thus, one might speculate that increased cytokine release by monocytes of NOD2 3020 polymorphic subjects stimulates enhanced chronic inflammation of the infected vessel wall and might thereby contribute to the pathogenesis of atherosclerosis.
So far, we have shown that at least three PRR, i.e. TLR2, TLR4 and NOD, are engaged in immune recognition and subsequent induction of inflammatory responses induced by C.
pneumoniae. This is rather unusual, since most bacteria are recognized basically by one PRR. The involvement of more PRR might represent a strategy of immune modulation that enables C. pneumoniae to circumvent host defence. I was able to show that in general, C.
pneumoniae are weak inducers of inflammatory responses in human whole blood compared to E. coli, which may render them capable to escape the immune cells and to establish intracellular persistent infections. Remarkably, beside the reduced induction of pro-inflammatory cytokines, especially IFNγ, the cytokine profile induced by C.
pneumoniae is characterized by an enhanced IL-10 release in comparison to a prototypic
crucial for the elimination of intracellular pathogens. IL-10 is a cytokine with predominantly anti-inflammatory properties which down-regulates monocytic and lymphocytic activities.
C. pneumoniae-induced IL-10 release has been shown to render infected PBMC anti-apoptotic (76) and to mediate MHC class I down-regulation (38). Furthermore, participation of IFNγ in the resistance of macrophages towards C. pneumoniae and restriction of their infectious progeny was reported (1). In C. pneumoniae-infected HEp-2 or HeLa cell cultures, IFNγ induced a persistent state of C. pneumoniae via tryptophan depletion (116, 176) where an altered gene expression, i.e. down-regulation of genes for cytokinesis but not DNA replication (26, 183, 213), was observed. Whether this in vitro observed IFNγ-induced persistence also translates to the in vivo situation is speculative. An impaired induction of IFNγ release has also been shown for the TLR2 agonist LTA and is the reason for the reduced toxicity of LTA in mice (111). However, the immunological consequences of the impaired IFNγ release together with the increased induction of IL-10 might be the basis of the characteristic hallmarks of C. pneumoniae infections, i.e. the asymptomatic course of infection and the high rate of intracellular persistence.
The role of chlamydial LPS for immune activation is controversially discussed. We found that C. pneumoniae-induced TNF, IL-1β and IFNγ release was completely abrogated in the presence of an anti-CD14 antibody and the LPS-specific inhibitor LALF, while the release of anti-inflammatory IL-10 and the chemokine IL-8 was not affected. This indicates that the chlamydial LPS is most probably responsible for the induction of pro-inflammatory cytokines while a second non-LPS molecule stimulates the release of the anti-inflammatory IL-10 and IL-8. This hypothesis is furthermore supported by the finding that the enterococcal LPS-induced release of the same subset of cytokines which showed anti-CD14 and LALF-sensitivity, namely TNF, IL-1β and IFNγ, were inhibited by co-incubation with C. pneumoniae, while IL-10 and IL-8 release was not affected. In this case, the LPS of C. pneumoniae may act as a partial agonist of TLR4, what might be due to structural differences to a non-chlamydial LPS. We expected that this inhibitory interaction would result in an inhibition of the TLR4-induced signal transduction cascade. P-38 and ERK1/2 are kinases predominantly involved in activation of LPS-induced cytokine release. It was rather surprising that none of these kinases was affected by co-incubation with C.
pneumoniae. However, several other kinases are involved into the process of induction of cytokine release. Currently, a Western blot for JNK is established and the influence of C.
pneumoniae on the activation of NF-κB, which as transcription factor is located down-stream in the signalling cascade, will be followed.
Nevertheless, the observed immune modulatory effects appeared to be restricted to C.
pneumoniae, since co-incubation with C. trachomatis did not result in any alteration of enterococcal LPS-induced cytokine release. Furthermore, comparison of the cytokine profiles of the two Chlamydia species argues against a common immune stimulatory principle since stimulation with C. pneumoniae led to a lower production of pro-inflammatory cytokines and only very little amounts of IFNγ, but led to a stronger release of IL-8 and anti-inflammatory IL-10. These results in parts reflect the clinical picture, where infections with C. trachomatis, in contrast to C. pneumoniae, are characterized by pronounced inflammation, and where the property to escape the immune system and to establish chronic infections is more typical for C. pneumoniae. Differences in target cell activation of C. pneumoniae and C. trachomatis, i.e. IL-8 induction and kinase activation, have also been reported by others (144). It might be speculated that either the LPS of C.
pneumoniae and C. trachomatis are structurally different or that their cytokine inducing properties are modulated by other membrane components. Heat- or UV-inactivation of C.
pneumoniae and C. trachomatis confirmed that chlamydial outer membrane proteins or bacterial replication are not involved in the activation of immune cells. LPS from Gram-negative bacteria is in general prepared by phenol/chloroform extraction procedures. In contrast, LTA from Gram-positive bacteria, which is more hydrophobic and more labile than LPS can only be prepared in a biological active form by butanol extraction, while phenol extraction results in an inactive and partially decomposed LTA (185, 187). Since C.
trachomatis LPS was shown to exert weak endotoxic activity and its lipid anchor was more hydrophobic compared to prototypic LPS from enterobacteria (106, 121, 142), isolation of immune active components of C. pneumoniae has been carried out by butanol extraction.
Indeed, after separation by HIC, a pool of fractions was obtained which showed a comparable cytokine inducing activity and TLR2/4-dependence as whole C. pneumoniae.
NOD2-dependency had not been investigated, because the role of the NOD2 polymorphism was not known at this time point. Analysis by LAL indicated that LPS was present in the pool fraction. Our initial hypothesis was that two immune active components of C. pneumoniae exist, one, most probably the chlamydial LPS, which induces pro-inflammatory cytokines, and second a component of unknown nature, which induces potently IL-10 and IL-8. The chemical analysis of the extracted material indicated that a mixture of substances was extracted, but that at least a lipopeptide was present and LAL indicated the presence of a small amount of LPS. However, since both, the release of pro-
the presence of anti-CD14 and LALF; it seems that the second non-LPS component was not extracted. Furthermore, the fraction pool, unlike whole C. pneumoniae, did not lead to an inhibition but to an increase of enterococcal LPS-induced pro-inflammatory cytokine release. With regard to our hypothesis about immune modulation by C. pneumoniae, this indicates that rather not the chlamydial LPS acts as an agonist at the TLR4, but that the unknown component leads to interference with the TLR4 signalling cascade. To clarify these complex interactions it is essential to identify the chemical composition of the immune active structures. Since we have hints for the presence of a lipopeptide, chloroform/methanol extraction followed by HPLC, like performed for the isolation of the lipopeptide MALP-2 from Mycoplasma might be considered (194). Once obtained, the different C. pneumoniae components shall also be tested in our murine model for the elicitation of immune responses.
Beside their role as respiratory pathogens, persistent infections with C. pneumoniae are discussed as risk factors for atherosclerosis for more than two decades. Although C.
pneumoniae were shown to be present in atherosclerotic tissue and animal models provide causal evidence for a role in pathogenesis, the results of seroepidemiological as well as of treatment studies remain controversial. In this respect one factor of crucial importance is the validity of serodiagnostic assays, which are used to stratify study groups.
The current gold standard, the MIF, is a time-consuming method, difficult to perform and analyze and prone to false positive and false negative results. We have focussed on the most promising ELISAs SeroCP and SeroCP Quant of a study previously performed by our group, which showed a very good concordance of these tests with the gold standard MIF in sera from healthy individuals (109). The SeroCP and SeroCP Quant ELISA are both based on whole LPS-containing C. pneumoniae as antigens and the Quant ELISA furthermore allows quantification of antibody titers. The sELISA, which is based on synthetic antigens and which had not been available during the previous study, was also included in this study. This second assay evaluation was performed with 80 sera from pneumoniae patients. Since no convalescent sera were used, this was not a diagnostic study, but a mere assay evaluation. In comparison to MIF the concordance of all three ELISAs for the detection of IgG in sera from pneumoniae patients was ≥90% and the positive predictive values were 93% for Sero CP and 97% for SeroCP Quant and the sELISA. The negative predictive values for IgG detection in sera from pneumonia patients were 83% for SeroCP, 81% for Quant and 68% for the sELISA. It was obvious that the detection of samples with low titers in IgG and IgA like 1:64 or 1:32 show the greatest
variance between the tests, since all false-negative samples were samples with very low titer in MIF. This lack in sensitivity may not be of major significance for the diagnosis of acute infections, where a fourfold rise in IgG titer and the use of paired samples is recommended as a diagnostic criterion, but will strongly affect the patient composition of collectives analysed retrospectively or for prospective seroepidemiological studies, where no acute and paired sera will be assessed. Since SeroCP and Quant achieved the highest sensitivities, they were evaluated further by using a second batch of 50 medium to low titer patient sera and 80 negative sera from children. In conclusion, the evaluation of the SeroCP ELISA and the Quant ELISA using two different collectives of sera samples from patients with respiratory tract diseases as well as one control-collective of negative sera from children revealed a high specificity and sensitivity of these ELISAs as well as a very good agreement with the MIF results (96%) and also the specificities determined with the negative sera were high (SeroCP Quant ≥99%; SeroCP 86%). Therefore, these time-saving and more objective ELISAs represent an acceptable alternative for current C.
pneumoniae serodiagnosis of acute infection. So far, persistent C. pneumoniae infections, which are associated with various chronic diseases, cannot be distinguished from past infections on the basis of serology. Therefore, there is need to identify persistence specific antigens, for example on the basis of immunoblots.
The association of C. pneumoniae and atherosclerosis is mainly supported by the fact that viable C. pneumoniae were directly detected in atherosclerotic lesions but never in healthy tissue (84, 148, 245). There is considerable evidence that the presence of C. pneumoniae in vascular cells or in immune cells that have invaded the inflamed tissue contribute to sustained stimulation of inflammation. Since free C. pneumoniae EBs could not be identified circulating in the blood stream, it is unclear how C. pneumoniae, that primarily infect cells of the respiratory tract, are disseminated throughout the body. In vitro, all atherosclerosis relevant cells like monocytes, endothelial cells and smooth muscle cells can be infected by C. pneumoniae, leading to cell activation, i.e. expression of adhesion molecules and release of pro-inflammatory cytokines (73, 107, 218, 222, 234). In addition, C. pneumoniae were also detected in vivo in resident cells of the atherosclerotic plaque (5, 73, 190). Furthermore, key events in atherogenesis like the transformation of macrophages into fat-laden foam cells after uptake of LDL and the oxidation of lipoproteins at the site of lesion development can be promoted by C. pneumoniae infection (58, 131).
In the present thesis we demonstrate that PMN can serve as carriers for C. pneumoniae,
exposure to laminar shear stress, which can explain the focal distribution of C.
pneumoniae in the vasculature, since atherosclerotic lesions develop mainly at sites of disturbed laminar shear stress. PMN are potent phagocytes abundant in the blood which are rapidly and in large numbers recruited to the site of infection. Indeed, as demonstrated in the present study, in vitro co-culture of PMN with C. pneumoniae results in uptake of bacteria, which then can be transmitted to endothelial cells simply by co-incubation. It is surprising that phagocytosis of C. pneumoniae does not result in killing of the bacteria by PMN, but in multiplication of the pathogen and in prolonged half-life of PMN, possibly induced by autocrine IL-8 production, what makes them a suitable shuttle cell. These in vitro results are further supported by an in vivo study where intranasal infection of mice with C. pneumoniae results in recruitment of PMN into the lungs and in C. pneumoniae infection of PMN. In this model, C. pneumoniae do not only replicate in PMN, but also enhance replication in bronchial epithelial cells (229). The fact that the infection of endothelial cells can be efficiently prevented by exposure to laminar shear stress is a reasonable explanation for the focal distribution of C. pneumoniae in the vasculature. It is suggested that C. pneumoniae-infected PMN are targeted to the vessel wall, but that infection is normally prevented by the laminar blood flow. Pro-longed laminar shear stress has been shown to generate an anti-oxidant, anti-proliferative, anti-apoptotic and pro-differentiative endothelial cell type via the regulation of gene transcription (272). In “lesion prone regions”, these beneficial effects of laminar shear stress are reduced. Obviously, shear stress exposure does not prevent the transmission of C. pneumoniae to endothelial cells but the establishment of inclusions. It is known that shortly after infection C.
pneumoniae starts to influence the host cell metabolism and especially redirect their energy sources. Whether effects like this are prevented by shear stress remains to be defined. This study links for the first time two putative risk factors for the initiation of atherogenesis and explains for C. pneumoniae both a possible route of infection and the selective localization in atherosclerotic lesions.
Taken together, the results of this thesis emphasize the crucial importance of the understanding of immune recognition of bacteria. For C. pneumoniae it seems most likely that deviations of inflammatory reactions in response to recognition by innate immune cells enable them, after asymptomatic acute infections, to enter an intracellular persistent state.
Furthermore, in case of C. pneumoniae, uptake by innate immune cells does not result in killing of the pathogen but might contribute to the distribution of C. pneumoniae in the body.
8 Summary
Chlamydophila pneumoniae is an obligate intracellular pathogen which leads to mostly asymptomatic infections of the respiratory tract. After acute infection, C. pneumoniae can establish persistent infections in the host. To be able to prevent persistence, it is necessary to understand how C. pneumoniae are recognized by the innate immune cells and which evasion strategies are used to hide from the immune system. The association
Chlamydophila pneumoniae is an obligate intracellular pathogen which leads to mostly asymptomatic infections of the respiratory tract. After acute infection, C. pneumoniae can establish persistent infections in the host. To be able to prevent persistence, it is necessary to understand how C. pneumoniae are recognized by the innate immune cells and which evasion strategies are used to hide from the immune system. The association