A spontaneously arising rabbit anti-centrosome serum with strong human specificity, used to identify specific antigens in isolated centrosomes, was shown to react with several noncentrosomal proteins including a 36-kDa protein that appeared to be the major cellular antigen. To explore the immunological relationship between noncentrosomal and centrosomal antigens, immunoglobblins were affimity purified using the individual noncentrosomal antigens (from lymphoblastoma KE37 cells) and were tested for their capacity to bind to human centrosomes in situ and to proteins from isolated centrosomes. In this way, the 36-kDa antigen, an abundant cytosolic protein, was shown to share at least one antigenic determinant with high molecular weight centrosomal proteins. This antigen was further identified by mild proteolysis as the glycolytic enzyme lactate dehydrogenase. In all the analyzed human cell lines, the centrosomal staining in situ was correlated with a strong labeling of purified lactate dehydrogenase in immunoblots. Conversely, the absence of centrosomal staining in rodent cells was always correlated with the absence of lactate dehydrogenase labeling. These data suggest an evolutionary relationship between centrosomal proteins and this "housekeeping" enzyme.Several cytoskeletal components have been studied at the cellular and molecular level (1-3), or even discovered (4), with nonimmune sera. In some instances, these sera are still the exclusive tools for tracing cellular structures. This is the case for microtubule-organizing centers such as kinetochores (4-6) and centrosomes (7)(8)(9)(10) Serum. In the course of a study on the inflammatory process, eight rabbits were injected with Streptococcus type 24 (17). When tested on cultured cells from various origins by the indirect immunoperoxidase technique, the preimmune serum of one rabbit (serum 0013) was shown to strongly label centrosomes and weakly label nucleoli in primate cells (12). The experimental immunization induced a strong and transient increase of this preimmune specificity. The fact that cellular labeling did not disappear after extensive preabsorption of the serum with Streptococcus type 24 showed that these specificities were antigenically unrelated to the Streptococcus (17). Moreover, these specificities were not detected in the sera of the seven rabbits immunized in the same way.Immunofluorescence Microscopy. Cells were fixed in PBS (150 mM NaCl/10 mM sodium phosphate, pH 7.4) containing 3% (vol/vol)