The remarkable degree of variation in the humoral immune response displayed by the animals of the present study indicates a high complexity of host-pathogen interactions during MmmSC infection, which can lead to acute, sub-acute to chronic or symptomless courses of disease . On the one hand, the immune status of the individual animal seems to play a role in the time course and level of specific antibody production. Naïve animals are assumed to react in a different fashion than re-infected cattle exhibiting an anamnestic response, and symptomless carriers can exhibit low antibody levels in the absence of intense host pathogen interactions. Inter-animal differences in the cellular immune response [18, 19], which have not been addressed in the present study, may also add to the overall diversity observed.
On the other hand, the pathogen has been shown to possess a genetically determined machinery for surface antigen variation [24, 25], which enables it to evade the host immune response by selecting modified phenotypes that cannot be challenged by cognate antibodies. Depending on the efficiency of the individual host defence, the progress of MmmSC infection can be expected to vary from animal to animal.
The authors wish to emphasise that the present comparative analysis of diagnostic tests is referring to the individual animal level, which is a limitation because the data cannot be simply extrapolated to herd level. While currently available serological tests are generally suitable for herd diagnosis, the present findings highlight serious limitations of these tests at the individual animal level, which have to be taken into account when field studies are conducted.
CFT titres do not represent the whole spectrum of specific antibodies present in the infected animal, nor are they long lasting. The half-life of CFT antibody titres was estimated to be approximately 30 days [20, 26]. Notably, our own data indeed show a steep decline in these titres, beginning between weeks 3 and 6 p.i. (Figure 3, see Additional File 3: Examination of Panel 3 sera using in-house CFT and Additional File 4: Examination of Panel 3 sera using CIRAD CFT). Such a drop in CFT titres was also observed in the contact challenge study of Niang and co-workers , albeit significantly delayed (from week 16 to 36) as the precise time point of each individual infection remained elusive in that infection model. Taken together, these observations imply that field studies based solely on CFT are prone to miss individual animals at the later stages of infection.
While the differences between the two CFTs used were marginal, the relative diagnostic sen-sitivity of both tests compared to culture was 50.0% (in-house CFT) and 57.1% (CIRAD CFT) with Panel 1 samples, which is in line with data of other authors . The consistently observed divergence between the results of CFT and cELISA (Table 2, Figure 3) is probably a consequence of the different immunoglobulin classes covered by each method. Thus, IgG class antibodies have a greater affinity in the cELISA, while IgG2 subclass antibodies are unable to fixate complement used in the CFT. Moreover, IgM class antibodies, which are characteristic for early infection, are easier to detect by CFT [28, 29]. This can explain the earlier detection of antibodies by CFT as observed in the present study and elsewhere [20, 30]. Furthermore, the present finding that IgG antibody levels from cELISA remained at a high level for a prolonged period is in agreement with the study by Niang and co-workers , where the kinetics of different antibody isotypes was investigated.
It is important to note that the present cELISA was given a relatively high cut-off in order to maximise specificity, which in turn diminishes the test's sensitivity . In fact, the present evidence suggests that there is some room for lowering the cut-off without loss of specificity. However, this has to be confirmed by further studies involving more field sera from cattle herds having CBPP and/or other mycoplasma infections. We hypothesise that another way to improve the test's performance includes the use of specific peptides  or recombinant MmmSC proteins [16, 33] instead of whole-cell antigen.
The IBT has been described as being more sensitive and specific than CFT , which has been confirmed by the present data (Table 1). Thus, the IBT showed specific reaction patterns for sera tested negative in CFT and cELISA (animals 515, 520, 539, 544, 546). The test's high specificity is based on the reaction to five different antigens, i.e. 110, 98, 95, 60/62, and 48 kDa proteins, which must be recognised by their specific antibodies in order to identify a positive serum . However, we observed some problems with the test's reproducibility and potential for standardisation, as immunoblot reaction patterns are complex and individual bands may be difficult to identify when non-specific bands from cross-reactions with other bacteria are interfering. Cross-reactions with closely related Mycoplasma (M.) species of the "mycoides cluster" seem to be less important here, because they are rarely encountered in cattle, but other mycoplasmas, such as M. bovis and M. bovigenitalium, may play a role [13, 31].
Culture of MmmSC from affected lung tissue was included to underpin our serological findings. While seropositive animals always showed clinical symptoms of CBPP that were confirmed by pathology, the presence of lung lesions was no guarantee for successful re-isolation of the challenge strain (Panel 2, data not shown). The findings of the present study also suggest that re-isolation was seriously hampered at the late and chronic stages of infection, i.e. isolation of MmmSC did not succeed from tissue samples of Panel 3 animals after 35 weeks p.i. The absence of a strict correlation was particularly evident with animal 506 (Panel 1, Table 1), where all serological tests were negative (and pathological signs were missing) despite successful re-isolation of the pathogen.