Quality control of antibodies is strongly important to know the specificity and the sensitivity of an antibody. If these kinds of controls are missing the value of results is highly questionable. We aimed to compare the anti-MIA antibody which was used in a recent study showing MIA expression in Purkinje cells [1
] with an antibody we generated and to question the expression of MIA in the cerebellum. The antibody used for the immunohistochemical analyses of cerebellum in the study by Tokunaga et al. was raised against a C-terminal fragment of MIA. In our hands this antibody is able to detect 100 ng of recombinant human MIA in western blot analyses but not lower amounts (), hinting to a relatively low sensitivity. This was confirmed by Western blot analysis of melanoma cell lysate as a positive control where it was unable to detect endogenous MIA (). In addition, in human cerebellum and melanoma cell lysates this antibody detected several bands, thus suggesting that it is not highly specific. In contrast, an antibody (#7638 III pure) produced by us via immunization of a rabbit with KLH-conjugated full-length MIA and subsequent affinity purification was able to detect as little as 1 ng of recombinant MIA (). To test for specificity, Western blot analyses of lysates of human cerebellum and melanoma were performed. Using the highly sensitive antibody we generated, no MIA was detected in human cerebellum lysate, whereas MIA was specifically detected in the melanoma lysate ().
Figure 1 (A) Western blot analysis of recombinant human MIA (100ngand 10ng) using the anti-MIA antibody used in the discussed study  (Santa Cruz sc-17048) allowed the detection of 100ng MIA. (B) Western blot analysis of human cerebellum and melanoma lysates (more ...)
Subsequently, we performed immunohistochemical stainings of human cerebellum sections using our anti-MIA antibody. Here, no staining of Purkinje cells was observed, although they could be detected using the established staining for Fussel-15, recently shown to be expressed in Purkinje cells [11
] (). As further positive controls, we stained human melanoma sections, which were strongly positive for MIA (). As negative controls, we also stained human colon, mammary gland, uterus and skin for MIA, all of which were negative for MIA (). We therefore conclude that the stained Purkinje cells reported by Tokunaga et al. are due to unspecific binding of the antibody used in their study and not due to expression of MIA by these cells. It should be noted that other members of the MIA family such as TANGO might also be detected with an unspecific antibody due to the high homology of the MIA family members. In addition, with an antibody generated against human MIA careful tests are mandatory before using this antibody in other species. The reported study claims a 100% amino acid conservation of MIA in several species which is not the case. In fact, the sequence identity of the secreted protein between homo sapiens, mus musculus, bos taurus and rattus norvegicus is only 89%.
To conclude, we have demonstrated that careful establishment of every antibody used for every single technique and every species analyzed is highly mandatory. The best controls include the staining of protein-negative and -positive tissues as well as performing Western blot analyses to confirm that the size of the detected protein is correct and the antibody is specific. Further, competition with peptides against which the antibody was generated can help to determine the specificity. Finally, a careful comparison of the amino acid sequences of the antigen of interest and any homologues as well as the conservation between different species should be done.
In this study, by using a highly specific anti-MIA antibody, we can rule out the expression of MIA in the cerebellum through Western blot analyses as well as immunohistochemical stainings. Our results therefore underline that it is crucial to evaluate the specificity of an antibody to be used in immunohistochemistry in addition to the need to examine positive as well as negative control tissues. By avoiding the described potential pitfalls, the risk of misinterpreting results in immunohistochemistry can be drastically reduced.