During productive infection, host cells are modified and damaged by intracellular chlamydiae. To enable infection of neighboring cells, elementary bodies must be released in the extracellular space. Both these facts could contribute to the strong activation of the CS that has been demonstrated here for the first time in vivo. In mice infected with C. psittaci, C3a levels increased before clinical symptoms were detectable, thus indicating that complement activation could be an early trigger of the inflammatory cascade in chlamydial infection.
Using C3 knockout mice, this study has found that the CS plays an essential and complex role in chlamydial lung disease. Notably, C5aR and MAC have been ruled out as being responsible for the findings obtained with C3−/− mice, in particular for the increased susceptibility and the severe and lethal disease at the late stage of infection. C3a/C3aR and C3b/CRs are the remaining, so far not experimentally excluded known functional members of the CS downstream of C3 and upstream of C5 and, therefore currently the best candidates for future studies.
Moreover, only mice lacking C3 were partially protected early in infection and this early phenotype was also absent in C5def
mice. Thus, this deleterious early effect of the CS could be due to C3b and its binding to the highly condensed infectious elementary bodies of C. psittaci
. Mycobacterium (M.) tuberculosis
activates the CS and binds C3b to its surface 
. CR3 expressed on macrophages acts as one major phagocytic receptor for the improved uptake of these ‘opsonized’, facultative intracellular bacteria 
. In contrast to our findings on chlamydiae, the later course of M. avium
infection is not altered in C3−/−
mice during a 5-week observation period 
, which indicates that additional interactions must take place between chlamydiae and the CS. Chlamydia
) has been demonstrated to use alveolar macrophages and peripheral blood monocytes as ‘Trojan horses’ for covert dissemination 
. Indeed, as we and others 
have shown for spleen tissue, C. psittaci
DC 15 disseminates after lung infection in the host. For this reason, opsonization could be assisting the intracellular bacterium early in infection. The lack of opsonin in C3−/−
mice could temporarily counteract the overall negative consequences of a non-functional CS for the host. However, the equal bacterial loads in lung homogenate of C3−/−
and WT mice do not support this explanation. The analysis of earlier time points in the mouse model and uptake experiments on purified cells of the two mouse strains in cell culture are needed to address this issue in more detail.
CR3 in cooperation with C3, as well as complement component C1q also augment the uptake and killing of the facultative intracellular bacterium Listeria (L.) monocytogenes
into pre-activated peritoneal mouse macrophages contributing to improved antigen presentation, and, ultimately, to an enhanced adaptive T-cell response 
. C3 but not C5a/C5aR promotes the expansion of antigen-specific CD8+
T-cells in L. monocytogenes
. Similarly, based on decreased antigen presentation and subsequently hampered cellular immunity, C3b/CR3 could also cause the late protective effects in chlamydial infection.
However, it will not be easy to further dissect the role of C3b and CRs using the current model, because there are no knockout mice for each CR available and because some gene deletions do not only affect certain CRs but also other protein complexes with additional functions. As an example, CD11b is not only a subunit of CR3, but also part of an integrin.
The elevated levels of MPO indicate an increase of granulocytes in chlamydia-infected lungs with a peak around day 4. The higher levels, compared to control mice, of MPO (i.e. granulocytes) and certain cytokines in the lungs of C3−/− mice are not that surprising. The observed increase of other chemokines such as IL-6, MCP-1 or KC (data not shown) could be compensating for the lack of C5a/C5aR signaling downstream of C3 in the knockout mice. The anaphylatoxins C5a (and C3a) are not only chemotaxins for granulocytes, but also strong activators of these cells. Thus, even higher levels of MPO/granulocytes in lung tissue of C3−/− mice are not necessarily resulting in higher tissue damage because the activation of these cells might be hampered in the absence of anaphylatoxins. In addition, the decrease of MPO levels in lung homogenate from day 4 to 9, the rather small differences between MPO levels and the similar numbers of granulocytes in BALF of C3−/− and WT mice on day 9, and the almost identical histological scores of infected mice of both strains on that day also argue against a deleterious effect of granulocytes as an explanation of the high lethality in C3−/− mice occurring between day 9 and 17.
In general, the high cytokine levels in the C3−/− mice on day 4 might be partially explained by the slightly higher bacterial load in the lung. However, this cannot be the reason for higher IFN-γ levels on day 9 p.i. when the bacterial load is even significantly lower in the C3−/− mice.
In the time course of the experiments, all mice of the C3−/− strain either died or had to be sacrificed to minimize suffering when the clinical score became higher than 11. Obviously, from the particular day of their death until the originally intended end point of the experiment (on day 4, 9, or 21 p.i.), the animals with the most severe clinical course and thus, most likely, the highest ‘disease activity’ were no longer part of the observed group for daily determination of the clinical score and the body weight. They were also absent in the final assessment of lung tissue for histology and other in vitro analyses. As a result, there is a seemingly smaller increase of the clinical score, the loss of body weight and of histological changes and a smaller increase in the concentration of inflammatory mediators in this group. Thus, the effect of the C3 gene knockout at later time-points was most likely even higher than determined by the analysis of the less affected survivors.
The protective effect of the CS in C3−/− mice did not become apparent before the second or third week p.i. This time course indicates that complement activation by chlamydiae is not essential for the immediately acting innate immune defense. Instead, the kinetics suggests a functional link of the CS (and most likely C3a or C3b) to the adaptive and protective immune response in the case of chlamydial infections.
In the low-dose model, C3−/− mice showed signs of severe pneumonia even later then in high dose infection alongside high bacterial load in the lung. Notably, the majority of them recovered and finally survived, in contrast to the 100% lethality in the high dose infection model. Therefore, we hypothesize that complement activation is not an absolute requirement for a specific protective anti-chlamydial response, but that complement might enhance or accelerate it, thereby improving its efficiency.
In the current study, the slightly delayed humoral response (IgG) was ruled out as a putative link between complement and specific immune response. The serum transfer experiment demonstrated that i) pre-existing antibodies can be protective early in infection, i.e. most likely also in C. psittaci
infection, but ii) transferred antibodies alone contribute only minimally to protection later in the course of the ongoing infection with these bacteria. Considering the observed cytokine pattern in the lung homogenate and the unchanged IgG2a
) ratio, there was no indication that a modified Th1
polarization took place in C3−/−
mice. Nevertheless, future studies including adoptive cell transfer should address in more detail a possible relationship between the CS in chlamydial infection and T-cell (and NK-cell) functions, because the latter are critical factors for the elimination of chlamydiae. This view is supported by data from influenza virus infection, where C3 promoted T-cell priming 
As described above, the CS influences the course of infections with facultative intracellular bacteria, such as Listeria and Mycobacteria. However, most if not all of these interactions, including uptake or invasion by C3b/CR3, are caused by the extracellular, metabolically active form of these bacteria. Therefore, it is straightforward to suggest interaction of these bacteria with the CS, whose components are found in body fluids or are expressed on the surface of host cells. For the ‘late’ phenotype of C3−/− mice observed in infection with obligate intracellular C. psittaci (at a time point when adaptive immunity becomes essential in the defense), the situation is different and bound to be more complex.
Preliminary evidence suggests that the observed effect of the CS is not limited to C. psittaci infection. Initial experiments in the authors’ laboratory have revealed that C3−/− mice developed an identical phenotype after infection with C. pneumoniae strain CWL029 (data not shown). Moreover, a similar functional link might also exist for facultative intracellular bacteria, although this will be more difficult to prove.
In summary, our data suggest that functions of the CS downstream of C3 and upstream of C5 are to some degree harmful in early C. psittaci infection. However, more importantly, the results clearly show that in the second and third week p.i., when specific immunity becomes essential for the elimination of C. psittaci, complement activation is crucial for a successful defense against these intracellular bacteria. Additional experiments are necessary to elucidate the role of the remaining biologically active cleavage products of C3 (i.e. C3a and C3b) and identify a potential link to the specific cellular immune response against these pathogens. It is possible that the protective effect observed in C. psittaci lung infection represents a general mechanism and a specific function of the CS in infections with intracellular bacteria.