0270~6474/82/0201-0113$02.00/O The Journal of Neuroscience Copyright 0 Society for Neuroscience Vol. 2, No. 1, pp. 113-119 Printed in U.S.A. January 1982 IMMUNOSTAINING OF NEUROFIBRILLARY TANGLES IN ALZHEIMER’S SENILE DEMENTIA WITH A NEUROFILAMENT ANTISERUM1 DORIS DAHL,*,’ DENNIS J. SELKOE,$ ROBERT T. PERO,* AND AMICO BIGNAMI* *Spinal Cord Injury Research, West Roxbury Veterans Administration Medical Center, and fMailman Research Center, McLean Hospital, Department of Neuropathology, Harvard Medical School, Boston, Massachusetts 02115 Received June 2, 1981; Revised September 24, 1981; Accepted September 24, 1981 Abstract Using anti-chicken brain neurofilament antisera, Alzheimer’s neurofibrillary tangles from two patients with senile dementia were stained by immunofluorescence and by the peroxidase-antiper- oxidase procedure in cryostat sections of hippocampus and frontal cortex. In sections of cerebellum obtained from the same patients, the distribution of immunostaining was the same as that observed in experimental animals: Purkinje cell baskets and nerve fibers in the inner half of the molecular layer were demonstrated selectively. The immunostaining of the tangles was abolished when the antisera were absorbed by their own antigen, by bovine brain filament preparations, or by the fraction of bovine brain filament preparations nonabsorbed on anti-glial fibrillary acidic (GFA) protein immunoaffinity columns. Absorption with a bovine microtubule preparation isolated by two cycles of the assembly-disassembly procedure did not abolish the staining. Immunostaining experi- ments conducted on bovine brain filament preparations resolved on sodium dodecyl sulfate-poly- acrylamide gel electrophoresis showed that the antisera staining the tangles reacted with the 200,000-, 150,000-, and 70,000-dalton neurofilament polypeptides. Antisera raised to the 150,000- dalton bovine neurofilament polypeptide isolated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis stained the tangles much less intensely, although Purkinje cell baskets in the cerebellum appeared well stained. No staining of neurofibrillary tangles was observed with antisera to other classes of lo-nm filament proteins (GFA protein, vimentin, and desmin). The neurofibrillary tangle is a structural alteration hippocampus and amygdala in a majority of neurologi- affecting neurons during normal and pathological aging tally normal individuals after age 70. The number of of the human brain. The lesions consist of masses of tangle-bearing hippocampal neurons is 6 to 40 times abnormal fibrils found within the cytoplasm of certain greater in patients with dementia of the Alzheimer type neuronal perikarya and identified at the light microscopic (Ball, 1976), which is the most common cause of progres- level with silver stains. As far as is known, neurotibrillary sive intellectual failure in late life (Katzman, 1976). In tangles are unique to the human brain, since they have Alzheimer’s disease, neurofibrillary tangles also occur not been identified in aged non-human primates and with high frequency in many areas of cerebral cortex lower mammals (Wisniewski et al., 1973). Small numbers besides hippocampus. Their density appears to correlate of neurofibrillary tangles are found within neurons of strongly with the duration and severity of the dementia (Tomlinson et al., 1970; Farmer et al., 1976). ’ This work was supported by National Science Foundation Grant The distinctive electron microscopic appearance of the BNS-7912962 (D. D.), National Institutes of Health Grant AG-01307 neurofibrillary tangle was first described by Terry (1963) (D. J. S.), and by the Veterans Administration (A. B. and D. D.). Dr. and Kidd (1963). It is formed by bundles of filaments Selkoe is the recipient of a Teacher-Investigator Award from the approximately 20 nm in diameter and characterized by National Institute of Neurological and Communicative Disorders and constrictions about 10 nm in diameter at 80-nm intervals. Stroke. Mr. Pero was supported by a Veterans Administration Summer These have been interpreted as paired filaments coiled Student Stipend. We thank Dr. Klaus Weber for the gift of vimentin antisera and Dr. Friederich H. Klutzow for brain samples of Alzheimer’s as double helixes making a full turn every 160 nm (Kidd, dementia. 1963; Wisniewski et al., 1976) or, alternatively, as con- ‘To whom correspondence should be addressed at Spinal Cord stricted microtubules, for which the term twisted tubules Injury Research, Veterans Administration Medical Center, 1400 VFW was proposed (Terry, 1963). Parkway, West Roxbury, MA 02132. The chemical composition of the paired helical fila- 114 Dahl et al. Vol. 2, No. 1, Jan. 1982 ments in aged and diseased neurons is still controversial primary antisera. Immunoperoxidase staining by the (Grundke-Iqbal et al., 1979a; Eng et al., 1980). The top- PAP procedure was conducted according to Van Raams- ographic selectivity of neurofibrillary tangles in the cere- donk et al. (1977). Longitudinal cryostat sections 100 pm bral cortex, with sparing of many neuronal and non- thick were fixed for 10 min in 1% formalin at 4°C. The neuronal element+s,makes their quantitative purification sections were washed in phosphate-buffered saline, pH problematic. For this reason, an antiserum which can 7.5, for 10 min and incubated overnight at 4°C in 1:50 react reliably with Alzheimer’s neurofibrillary tangles dilutions of the primary antisera. A 30-min wash in would provide an important tool for further morpholog- phosphate buffer at room temperature and a distilled ical and biochemical investigations. water rinse were followed by incubation with goat anti- rabbit IgG at 1:lO dilutions for 60 min at room tempera- Materials and Methods ture. The phosphate buffer wash was repeated and the Antisera to neurofilament proteins isolated by hydrox- sections were incubated with PAP at 1:25 dilutions for 60 yapatite chromatography from 8 M urea extracts of min at room temperature. The sections then were rinsed chicken brain were raised in three New Zealand albino with phosphate buffer for 30 min at room temperature rabbits as described (Dahl and Bignami, 1977; Dahl, and dipped for 1 min in a 0.05 M Tris solution, pH 7.6. 1980a). The antisera were shown to react with the While the sections were still moist, the substrate diami- -150,000- and 72,000-dalton polypeptides of the mam- nobenzidine activated with HzOz was added. After a few malian neurofilament “triplet” (Hoffman and Lasek, minutes, the brown stain appeared. Finally, the gel sec- 1975; Schlaepfer and Freeman, 1978; Thorpe et al., 1979; tions were rinsed in distilled water and mounted in Lazarides, 1980) by immunoaffinity chromatography glycerol. Preimmune sera were used as a control in each (Dahl, 1981). They also were shown to stain neurofibrils experiment. selectively by indirect immunofluorescence and by the peroxidase-antiperoxidase (PAP) procedure in cryostat Results sections (Dahl and Bignami, 1977; Dahl et al., 1981a) and Due to the large amount of autofluorescent lipofuscin to decorate neurofilaments at the electron microscopic pigment in aged human brain, most observations were level (Dahl et al., 1981b). Absorption experiments were made with rhodamine-conjugated antisera. We found conducted as previously reported (Dahl and Bignami, that, in freshly cut sections of human cortex, lipofuscin 1979). The antisera were absorbed by their own antigen, pigment was more brightly autofluorescent with fluores- 100 pg of protein/ml, bovine brain filament preparations, cein optics than with rhodamine optics. By changing 300 pg of protein/ml, the fraction of a bovine brain filters in the fluorescence microscope, the distinction filament preparation not absorbed on anti-glial fibrillary between immunofluorescence and autofluorescence of acidic (GFA) immuno-Sepharose, 250 pg of protein/ml intraneuronal material thus was easily made. (Dahl and Bignami, 1979), and by a microtubule prepa- The specificity of the neurofilament antisera in autop- ration isolated from bovine brain by two cycles of the sied human brain was assessed in the cerebellum. As assembly-disassembly procedure, 1 mg of protein/ml. originally reported in the rat (Dahl and Bignami, 1977), Antisera to the -150,000-dalton neurofilament polypep- the following criteria allowed the characterization of our tide isolated by sodium dodecyl sulfate-polyacrylamide neurofilament antisera: (i) intense staining of Purkinje gel electrophoresis (SDS-PAGE) from bovine brain Ma- cell baskets and of a delicate mesh of nerve fibers in the ment preparations were prepared as reported (Dahl, lower half of the molecular layer and (ii) absence of 1980b). Antisera to GFA protein, vimentin, and desmin staining of Purkinje cells (perikarya and dendrites) and were prepared as previously described (Dahl and Big- of Bergmann glia in the molecular layer. As shown in nami, 1976, 1981; Franke et al., 1978). Indirect immuno- Figures lA, 2, and 5, these criteria were met when applied fluorescence with rhodamine- and fluorescein-conjugated to postmortem human brain. antisera and immunostaining with the PAP procedure Due to their intense staining, Alzheimer’s neurofibril- were conducted on cryostat sections. Immunoreagents lary tangles were readily identified in the pyramidal layer were from Cappel Laboratories, Cochranville, PA. The of the hippocampus with the anti-chicken brain neuro- dilutions of the primary antisera varied from 1:5 to 1:40 filament antisera using both immunofluorescence and for immunofluorescence