The presence of numerous keratin bodies in the upper dermis is a characteristic finding in skin lesions of patients with various dermatoses such as cutaneous graft-versus-host disease, lichen planus, or chronic discoid lupus erythematosus. These keratin bodies are generated by apoptotic keratinocyte death, consist largely of keratin intermediate filaments (KIF), and are constantly covered with immunoglobulins, mainly IgM. Apoptosis is also thought to occur under physiologic conditions in the skin as it does in other organs, but keratin bodies are not frequently reported as being found in nonlesional skin. In order to assess the frequency of keratin bodies in normal skin, we examined serial sections of 10 normal human skin specimens and 5 dermal sheets prepared from normal human skin for the presence of keratin bodies. They were visualized by direct immunofluorescence using a fluorescein isothiocyanate (FITC) rabbit antihuman IgM conjugate. In addition the KIF origin of keratin bodies was demonstrated by a double-staining immunofluorescence procedure using a FITC-conjugated rabbit antihuman IgM followed by a mouse monoclonal antibody against keratin and a sheep antimouse immunoglobulin conjugated with Texas Red. One specimen was also examined for keratin bodies at the ultrastructural level. In serial sections, all 10 normal human skin specimens had numerous keratin bodies as assessed by visualization of globular IgM deposits. Evaluated on dermal sheets, the number of keratin bodies ranged from 39-262 per mm2. Nearly all keratin bodies also stained with the antikeratin antibodies. Ultrastructurally the remarkable number of keratin bodies, which consist of filaments measuring approximately 10 nm in diameter or of more granular material, in normal human skin was confirmed. In order to investigate the capacity of KIF material in keratin bodies to function as autoantigen, we examined the sera of the 10 skin donors and, in addition, of 30 normal healthy individuals and 10 patients with rheumatoid arthritis for the occurrence and specificity of IgM-anti-KIF autoantibodies by an enzyme-linked immunosorbent assay and by immunoblot. IgM-anti-KIF autoantibodies were found in all 50 test sera. In the majority of the sera the specificity of these autoantibodies included the 51 kD and the 58 kD KIF protein, which are constituents of KIF in keratin bodies and basal keratinocytes. Quantitatively, the antibody activity of the IgM-anti-KIF autoantibodies varied from serum to serum, being highest in the sera of patients with rheumatoid arthritis.(ABSTRACT TRUNCATED AT 400 WORDS)