Motor neurons in Cu/Zn superoxide dismutase-deficient mice develop normally but exhibit enhanced cell death after axonal injury.

The discovery that some cases of familial amyotrophic lateral sclerosis (FALS) are associated with mutations in the gene encoding Cu/Zn superoxide dismutase (SOD1) has focused much attention on the function of SOD1 as related to motor neuron survival. Here we describe the creation and characterization of mice completely deficient for this enzyme. These animals develop normally and show no overt motor deficits by 6 months in age. Histological examination of the spinal cord reveals no signs of pathology in animals 4 months in age. However Cu/Zn SOD-deficient mice exhibit marked vulnerability to motor neuron loss after axonal injury. These results indicate that Cu/Zn SOD is not necessary for normal motor neuron development and function but is required under physiologically stressful conditions following injury.

Selective sparing of a class of striatal neurons in Huntington's disease.

A distinct subpopulation of striatal aspiny neurons, containing the enzyme nicotinamide adenine dinucleotide phosphate diaphorase, is preserved in the caudate nucleus in Huntington's disease. Biochemical assays confirmed a significant increase in the activity of this enzyme in both the caudate nucleus and putamen in postmortem brain tissue from patients with this disease. The resistance of these neurons suggests that the gene defect in Huntington's disease may be modifiable by the local biochemical environment. This finding may provide insight into the nature of the genetically programmed cell death that is a characteristic of the disease.

Gene transfer into experimental brain tumors mediated by adenovirus, herpes simplex virus, and retrovirus vectors.

Three vectors derived from retrovirus, herpes simplex virus type 1 (HSV), and adenovirus were compared in cultured rat 9L gliosarcoma cells for gene transfer efficiency and in a 9L rat brain tumor model for histologic pattern and distribution of foreign gene delivery, as well as for associated tumor necrosis and inflammation. At a multiplicity of infection of 1, in vitro transfer of a foreign gene (lacZ from Escherichia coli) into cells was more efficient with either the replication-defective retrovirus vector or the replication-conditional thymidine kinase (TK)-deficient HSV vector than with the replication-defective adenovirus vector. In vivo, stereotactic injections of each vector into rat brain tumors revealed three main histopathologic findings: (i) retrovirus and HSV vector-mediated gene transfer was relatively selective for cells within the tumor, whereas adenovirus vector-mediated gene transfer occurred into several types of endogenous neural cells, as well as into cells within the tumor; (ii) gene transfer to multiple infiltrating tumor deposits without apparent gene transfer to intervening normal brain tissue occurred uniquely in one animal inoculated with the HSV vector, and (iii) extensive necrosis and selective inflammation in the tumor were evident with the HSV vector, whereas there was minimal evidence of tumor necrosis and inflammation with either the retrovirus or adenovirus vectors.

Experimental tumor therapy in mice using the cyclophosphamide-activating cytochrome P450 2B1 gene.

Most malignant tumors of the central nervous system do not respond well to chemotherapy. The anticancer drug cyclophosphamide (CPA) is largely ineffective against these neoplasms as its conversion to DNA-alkylating, cytotoxic metabolites is restricted primarily to the liver and these metabolites do not readily cross the blood-brain barrier. Here, we show that brain tumor cells can be sensitized to the cytotoxic effects of CPA, both in culture and in vivo, by introduction of the hepatic enzyme responsible for the activation of CPA, cytochrome P450 2B1. Stable transfection of rat C6 glioma cells with the P450 2B1 gene rendered the cultured tumor cells sensitive to CPA. Further, C6 cells bearing this gene were more sensitive than parental cells to the cytotoxic action of CPA when grown subcutaneously in the flanks of athymic mice. Murine fibroblasts producing a retrovirus vector encoding P450 2B1 and expressing this enzyme were then prepared and grafted into the brains of athymic mice seeded with rat C6 gliomas. Intrathecal administration of CPA prevented the development of meningeal neoplasia and led to partial regression of the parenchymal tumor mass. By contrast, C6 glioma-bearing mice receiving fibroblasts expressing the Escherichia coli lacZ gene and CPA exhibited extensive meningeal tumors and parenchymal solid brain tumors. The in situ activation of CPA by cytochrome P450 2B1 provides a novel approach not only for brain tumor gene therapy, but also for negative, drug-conditional selection of other defined cell populations.

Thy-1 in hippocampus: normal anatomy and neuritic growth in Alzheimer's disease.

Abnormal neuritic sprouting is a prominent feature of Alzheimer's disease (AD), and the Thy-1 glycoprotein has a role in neurite growth in culture. We therefore investigated the distribution of Thy-1 immunoreactivity in the hippocampus of normal elderly patients and of AD patients. Normally, Thy-1 immunoreactivity, which was more prominent in CA1 than elsewhere in the hippocampus, was located mainly in irregular patches on the perikarya of pyramidal cells, their dendrites and axons. In AD, Thy-1-immunoreactive neurons were reduced in number in CA1, and there was diffuse staining of neurofibrillary tangle-bearing pyramidal cells, but neurofibrillary tangles themselves were not immunoreactive. There was also staining of disorganized arrays of dystrophic neurites, some with spiny processes and bizarre filopodial endings. Some Thy-1-immunoreactive dystrophic neurites entered senile plaques. The data confirm that there is extensive growth of abnormal neurites in AD and suggest that Thy-1 is involved in this process.

Senile plaques are located between apical dendritic clusters.

The location of senile plaques (SP) in human cortex has been studied in relation to the dendritic architecture of the neocortex. Senile plaques were examined in patients with normal senescence and with Alzheimer's disease. The study utilizes a monoclonal antibody to microtubule-associated protein 2 for selective immunocytochemical visualization of dendrites and thioflavin S to visualize SP. The immunocytochemical analysis demonstrated the fascicular organization of pyramidal cell apical dendrites. Senile plaques were located between dendritic clusters with pyramidal cell apical dendritic shafts excluded from plaques. The close correlation between plaque diameter and interfascicular distances suggests that the boundaries of the vertically organized bundles may be an architectural constraint on plaque size.

Hippocampal neurons predisposed to neurofibrillary tangle formation are enriched in type II calcium/calmodulin-dependent protein kinase.

The microtubule-associated phosphoprotein, tau, is an integral component of paired helical filaments in Alzheimer neurofibrillary tangles (NFT). The mechanism of NFT formation is unknown but aberrant phosphorylation of tau may be contributory. Calcium/calmodulin-dependent protein kinase type II (CaM kinase II), the most abundant kinase in the brain, phosphorylates tau in vitro. We found CaM kinase II immunoreactivity concentrated in human hippocampal pyramidal neurons of CA1 and the subiculum. In Alzheimer's disease (AD) staining intensity of CA1 and subicular neurons is strikingly increased despite NFT formation and neuronal depletion. Enhanced CaM kinase II activity, possibly a result of deafferentation, may contribute to phosphorylation of tau protein leading to NFT deposition and neuronal death in AD.

Morphologic and histochemical characteristics of a spared subset of striatal neurons in Huntington's disease.

We have previously found that a biochemically distinct subset of neurons, containing nicotinamide adenine dinucleotide phosphate diaphorase (NADPH-d), is selectively resistant to the degenerative process that affects the striatum in Huntington's disease (HD). We report the morphologic and histochemical characteristics of these striatal neurons and their distribution with respect to the histochemical compartments as defined by acetylcholinesterase (AChE) activity. Sections of striatum were stained histochemically for NADPH-d and AChE and immunocytochemically for somatostatin and neuropeptide Y-like immunoreactivity. The diaphorase end-product was contained within medium-sized neurons which corresponded morphologically to a category of aspiny interneurons. Combined techniques showed that NADPH-d, somatostatin, and neuropeptide Y coexisted within the same neurons in controls and patients with HD. The density of these neurons was greater in the ventral putamen and the nucleus accumbens than in the remainder of the striatum. The distinctive AChE pattern of high and low enzyme activity was altered in HD. The AChE-rich matrix zone was markedly reduced in size, while the total area of zones of low enzyme activity was not different from that found in control striatum. The relation between these AChE chemical compartments and the distribution of preserved diaphorase neurons remained intact; NADPH-d neurons were predominantly observed in the matrix zone.

3-Acetylpyridine produces age-dependent excitotoxic lesions in rat striatum.

The effects of 3-acetylpyridine (3-AP) were studied in rat striatum. Striatal injections of 3-AP produced dose-dependent lesions. The lesion size was significantly increased in 4- and 12-month-old rats compared to 1-month-old rats. Coinjection of the competitive N-methyl-D-aspartate (NMDA) antagonist 2-amino-5-phosphonovaleric acid (APV) or systemic administration of the noncompetitive NMDA antagonist MK-801, the competitive NMDA antagonist LY274614, or the glutamate release inhibitor lamotrigine partially but significantly attenuated striatal lesion volume. Consistent with an NMDA receptor-mediated excitotoxic effect, histologic studies showed that 3-AP lesions result in relative sparing of NADPH-diaphorase neurons. Using freeze clamp, 3-AP resulted in a marked depletion of ATP. Two-dimensional water-suppressed proton chemical shift magnetic resonance imaging showed a striatal depletion of the neuronal marker N-acetylaspartate but no focal increase in lactate during the first 3 h after intrastriatal 3-AP injections. Pretreatment with fructose-1,6-biphosphate attenuated the lesion volume significantly, which may be due to its ability to serve as a substrate for glycolytic metabolism, with resulting ATP production. The results of the present studies support the hypothesis that 3-AP produces an impairment of energy metabolism due to its substitution for niacinamide in the formation of NAD(P). Furthermore, 3-AP toxicity may involve a secondary excitotoxic mechanism mediated by NMDA receptors.

In vivo neurotoxicity of beta-amyloid [beta(1-40)] and the beta(25-35) fragment.

We examined the histological changes produced by injections of beta-amyloid [beta(1-40)], and control peptides in rat and monkey cerebral cortex. beta(25-35) injections were also studied in rat cortex. Standard immunoperoxidase procedures were used to detect the distribution of tau, MAP2, beta(1-40) and ALZ 50 immunoreactivity. All injections produced localized necrosis at the injection site surrounded by a zone of neuronal loss and gliosis. In rat cortex, lesions produced by solubilized beta(1-40) and beta(25-35) in water were generally larger than those produced by control peptides. Tau and ALZ 50 antibodies labeled neurites and diffusely positive perikarya around beta(1-40) injections, whereas MAP2 staining was reduced, paralleling the distribution of neuronal loss and gliosis. In aged primate cortex, beta(1-40) lesion size was dose dependent. Hyalinized, ALZ 50 positive neurons, and abnormal neurites were prominent around the injection site. Although beta-amyloid is acutely neurotoxic in both rat and monkey cerebral cortex, neuronal degeneration in the primate more closely resembles that found in AD.

Morphometric image analysis of neuropil threads in Alzheimer's disease.

Neuropil threads were quantitated in the neuropil (excluding senile plaques) of the superior frontal gyrus of 6 late stage patients with Alzheimer's disease (AD) and 6 age-matched control subjects using tau immunocytochemistry and computerized morphometric image analysis. The mean percent of the area of the neuropil occupied by neuropil threads was 10.6 for AD and 0.19 for controls (p < 1 x 10(-10)). The mean length of neuropil threads in AD was 21.9 mu compared with 19.7 mu for controls (p < 1 x 10(-10)). The mean area of neuropil threads was 25.3 mu 2 for AD and 21.3 mu 2 for controls (p < 1 x 10(-10)). In AD, the threads were most prominent in mid cortex (lamina 2 and 3) and least prominent in the lower cortex (lamina 5 and 6). Neuropil threads appear to lead to severe disorganization of intracortical and corticocortical connectivity and probably play a role in the cognitive failure in AD.

Subset of neurons characterized by the presence of NADPH-diaphorase in human substantia innominata.

The substantia innominata encompasses an area of the basal forebrain that is ventral to the lenticular nucleus and anterior commissure, medial to the claustrum and external capsule, and lateral to the hypothalamus. The nucleus basalis of Meynert consists primarily of large acetylcholinesterase (AchE)-positive neurons embedded within the substantia innominata. Damage to these neurons may be important in the pathogenesis of cortical dysfunction in Alzheimer's disease. In order to characterize other neuronal elements in the substantia innominata and their relationship to the nucleus basalis, we chose to study a biochemically distinct neuronal subset containing the enzyme nicotinamide adenine dinucleotide phosphate diaphorase (NADPH-d). The substantia innominata was blocked from six normal brains obtained postmortem and fixed in neutral-buffered formalin at 4 degrees C for 48 hours. Free-floating 50-micron sections from several levels were stained for NADPH-d or AchE activities. Selected sections were double stained for NADPH-d and AchE. NADPH-d activity was present in a network of pleomorphic neurons that extended through all levels of the substantia innominata and into the striatum and amygdala. NADPH-d neurons were particularly numerous at the level of the anterior commisure and were closely associated with the cholinergic neurons of the nucleus basalis. They were not seen in the ventral pallidum, or the vertical limb of the diagonal band of Broca or in the islands of Calleja. The cell bodies of NADPH-d neurons were quite varied in shape, ranging from ovoid to fusiform, and about half the cells were bipolar. Where neuronal density was high, their dendrites formed an interlacing pattern. NADPH-d-positive fibres were seen coursing through the external capsule, hypothalamus, and amygdala. This novel set of neurons in the substantia innominata may be part of a more extensive network that interacts with the magnocellular basal forebrain system at the level of the nucleus basalis. Whether other neurotransmitters are present within these neurons and whether NADPH-d neurons are involved in Alzheimer's disease remain to be elucidated.

Cytoarchitecture, fiber connections, and some histochemical aspects of the interpeduncular nucleus in the rat.

The organization of afferent and efferent connections of the interpeduncular nucleus (IP) has been examined in correlation with its subnuclear parcellation by using anterograde and retrograde tracing techniques. Based on Nissl, myelin, and acetylcholinesterase staining five paired and three unpaired IP subnuclei are distinguished. The unpaired division includes the rostral subnucleus (IP-R), the apical subnucleus (IP-A), and the central subnucleus (IP-C). The subnuclei represented bilaterally are the paramedian dorsal medial (IP-DM) and intermediate subnuclei (IP-I) and the laterally placed rostral lateral (IP-RL), dorsal lateral (IP-DL), and lateral subnuclei (IP-L). Immunohistochemical techniques showed cell bodies and fibers and terminals immunoreactive for substance P, leu-enkephalin, met-enkephalin, or serotonin to be differentially distributed over the different IP subnuclei. Substance P-positive perikarya were found in IP-R, enkephalin neurons in IP-R, IP-A, and the caudodorsal part of IP-C, and serotonin-containing cell bodies in IP-A and the caudal part of IP-L. Efferent IP projections were studied both by injecting tritiated leucine in IP and by injecting HRP or WGA-HRP in the presumed termination areas. The results indicate that the major outflow of IP is directed caudal-ward to the median and dorsal raphe nuclei and the caudal part of the central gray substance, i.e., the dorsal tegmental region. The projection appears to terminate mainly in the raphe nuclei, around the ventral and dorsal tegmental nuclei of Gudden, and in the dorsolateral tegmental nucleus. The descending projection to the dorsal tegmental region originates in virtually all IP subnuclei, but the main contribution comes from IP-R and the lateral subnuclei IP-RL, IP-DL, and IP-L. Sparser projections to the dorsal tegmental region originate in IP-C and IP-I, whereas the contribution of IP-A is only minimal. The projections from IP-R are mainly ipsilateral and those from IP-DM are mainly contralateral. IP fibers to the median and dorsal raphe nuclei originate predominantly in IP-R and IP-DM, and to a lesser extent in IP-C, IP-I, IP-RL, and IP-DL. A much smaller contingent of IP fibers ascends to diencephalic and telencephalic regions. A relatively minor projection, stemming from IP-RL and IP-DL, reaches the lateral part of the mediodorsal nucleus, the nucleus gelatinosus, and some midline thalamic nuclei. These IP fibers follow either the habenulo-interpeduncular pathway or the mammillothalamic tract.(ABSTRACT TRUNCATED AT 400 WORDS)

Mitochondrial DNA mutations in complex I and tRNA genes in Parkinson's disease.

OBJECTIVE: To identify mitochondrial DNA (mtDNA) mutations that predispose to PD. BACKGROUND: Mitochondrial complex I activity is deficient in PD. mtDNA mutations may account for the defect, but the specific mutations have not been identified. METHODS: Complete sequencing was performed of all mtDNA-encoded complex I and transfer RNA (tRNA) genes in 28 PD patients and 8 control subjects, as well as screening of up to 243 additional PD patients and up to 209 control subjects by restriction digests for selected mutations. RESULTS: In the PD patients, 15 complex I missense mutations and 9 tRNA mutations were identified. After screening additional subjects, rare PD patients were found to carry complex I mutations that altered highly conserved amino acids. However, no significant differences were found in the frequencies of any mutations in PD versus control groups. The authors were unable to confirm previously reported associations of mutations at nucleotide positions (np) 4336, 5460, and 15927/8 with PD. Complex I mutations previously linked to Leber's hereditary optic neuropathy, one of which has been linked to atypical parkinsonism, were not associated with PD. CONCLUSIONS: mtDNA mutations with a high mutational burden (present in a high percentage of mtDNA molecules in an individual) in complex I or tRNA genes do not play a major role in the risk of PD in most PD patients. Further investigations are necessary to determine if any of the rare mtDNA mutations identified in PD patients play a role in the pathogenesis of PD in those few cases.

Mice lacking cytosolic copper/zinc superoxide dismutase display a distinctive motor axonopathy.

OBJECTIVE: To characterize the motor neuron dysfunction in two models by performing physiologic and morphometric studies. BACKGROUND: Mutations in the gene encoding cytosolic superoxide dismutase 1 (SOD1) account for 25% of familial ALS (FALS). Transgenes with these mutations produce a pattern of lower motor neuron degeneration similar to that seen in patients with FALS. In contrast, mice lacking SOD1 develop subtle motor symptoms by approximately 6 months of age. METHODS: Physiologic measurements, including motor conduction and motor unit estimation, were analyzed in normal mice, mice bearing the human transgene for FALS (mFALS mice), and knockout mice deficient in SOD1 (SOD1-KO). In addition, morphometric analysis was performed on the spinal cords of SOD1-KO and normal mice. RESULTS: In mFALS mice, the motor unit number in the distal hind limb declined before behavioral abnormalities appeared, and motor unit size increased. Compound motor action potential amplitude and distal motor latency remained normal until later in the disease. In SOD1-KO mice, motor unit numbers were reduced early but declined slowly with age. In contrast with the mFALS mice, SOD1-KO mice demonstrated only a modest increase in motor unit size. Morphometric analysis of the spinal cords from normal and SOD1-KO mice showed no significant differences in the number and size of motor neurons. CONCLUSIONS: The physiologic abnormalities in mFALS mice resemble those in human ALS. SOD1-deficient mice exhibit a qualitatively different pattern of motor unit remodeling that suggests that axonal sprouting and reinnervation of denervated muscle fibers are functionally impaired in the absence of SOD1.

Morphology and distribution of nicotinamide adenine dinucleotide phosphate (reduced form) diaphorase reactive neurons in human brainstem.

Nicotinamide adenine dinucleotide phosphate diaphorase reactive neurons were found in several regions of human brainstem. Three major groups were located in the medulla: a dorsomedial group in the central gray and floor of the fourth ventricle, a ventromedial group in the vicinity of the medullary raphe, and a lateral group in the lateral reticular nucleus. In the upper pons a large cluster of reactive neurons was centered in the nucleus centralis oralis extending into the locus coeruleus and dorsal tegmental region. A second cluster in the lateral parabrachial nucleus merged with this group more rostrally and continued into the midbrain tegmentum (paracoeruleus-cuneiform group). Nicotinamide adenine dinucleotide phosphate diaphorase neurons in this region often contained acetylcholinesterase activity. A second midbrain group was seen in the nucleus paranigralis. Aside from these discrete neuronal collections, scattered reactive neurons were found in the medullary reticular formation, periaqueductal gray, inferior colliculus and superior colliculus. Nicotinamide adenine dinucleotide phosphate diaphorase neurons were classified into three groups based on somal size. Parvocellular neurons (10-20 micron) were primarily found in the ventromedial medulla and lateral parabrachial nucleus. Intermediate neurons (20-25 micron) were located in the paranigralis nucleus and dorsomedial medulla. Magnocellular neurons (25-35 micron) were characteristically found in the lateral reticular nucleus and paracoeruleus-cuneiform region. Nicotinamide adenine dinucleotide phosphate diaphorase reactive neurons are present in substantial numbers in human brainstem and their distribution is complex. They represent the caudal end of a widespread network of nicotinamide adenine dinucleotide phosphate diaphorase-enriched neurons that extend rostrally from the brainstem reticular formation into the basal forebrain, striatum, and cerebral cortex.

Substance P-like immunoreactive neurons are depleted in Alzheimer's disease cerebral cortex.

We studied the morphology and distribution of substance P-like immunoreactive elements in normal and Alzheimer's disease brain with a monoclonal anti-substance P antibody. Bands of prominent terminal-like staining were found in the dentate gyrus of normal brain. Multipolar substance P-immunoreactive neurons were seen in dentate polymorphic layer and CA4 and prominent fiber staining was present in the CA fields of the hippocampus and adjacent allocortex. Reactive perikarya, concentrated in deep cortex and infracortical white matter, were found in all isocortical regions. Greatest density was in frontal and parietal association cortex; lowest in visual cortex. Fiber density was generally greatest in layers I and II. In Alzheimer's disease, staining intensity was reduced in the dentate gyrus. Hilar neurons were unaffected but other CA field neurons were distorted with pruned dendritic trees. Isocortical perikarya and fibers were significantly depleted and distorted in all regions. Globular deposits consisting of distorted neurites or dissolving perikarya were frequently seen. Double staining methods showed that the vast majority of isocortical, but not hippocampal, substance P-like immunoreactive neurons are nicotinamide adenine dinucleotide phosphate diaphorase-positive. Despite the modest quantitative depletion of substance P in Alzheimer's disease cortex as measured by radioimmunoassay compared to somatostatin, there is a significant depletion of substance P-like immunoreactive perikarya. This disparity may be due to persistence of afferent projections which make a major contribution to substance P concentrations in cerebral cortex or to the high substance P content of dystrophic fibers in Alzheimer's disease cortex.

NADPH diaphorase histochemistry of the human hypothalamus.

The morphology and distribution of NADPH diaphorase reactive neurons was studied in the normal human hypothalamus. Reactive neurons were divided into three categories on the basis of perikaryal size. Small neurons (8-20 microns) were oval or fusiform, and pale staining. Intermediate neurons (20-30 microns) were fusiform, triangular or pyramidal with a wide range of staining intensity. Large neurons (greater than 30 microns) were triangular or pyramidal with moderate to dark staining. Reactive neurons were found in four major regions: medial preoptic, ventromedial, lateral, and perifornical. Scattered positive neurons were found in several other hypothalamic areas. Reactive fibers were present in the supraoptic decussation, medial forebrain bundle, and stria medullaris thalami. The localization of NADPH diaphorase neurons in hypothalamic nuclei affected by Alzheimer's disease and other degenerative disorders suggests that further studies of this neuronal subset are warranted.

Myocyte-specific enhancer binding factor 2C expression in human brain development.

Myocyte-specific enhancer binding factor 2C (MEF2C) activates transcription by binding to the myocyte-specific enhancer binding factor 2 (MEF2) regulatory element and has been shown previously to be expressed in muscle and in the brain. We have now studied MEF2C expression in human brain using an antiserum raised against amino acids 140-238 of MEF2C. Western blotting demonstrated that, in fetal brain, MEF2C-immunoreactive bands have the same apparent molecular weight as those in extracts of COS cells transfected with MEF2C complementary DNA. In adult brain, however, MEF2C-immunoreactive bands have a higher molecular weight. In the cerebral cortex, MEF2C immunoreactivity is present in the cortical plate, and is not found in the intermediate zone or ventricular zone. At 14 weeks of gestation, the earliest age examined, MEF2C immunoreactivity is present in cell nuclei throughout the cortical plate. Subsequently, MEF2C immunoreactivity develops a bilaminate and then a trilaminate distribution, and ultimately is expressed preferentially in layers II, IV and VI of mature neocortex. MEF2C immunoreactivity is also found in entorhinal cortex, hippocampus, claustrum, cerebellum and amygdala, and in scattered cells in the thalamus. These findings suggest a role for MEF2C in postmitotic neuronal differentiation, in particular, in the development of certain cortical layers, but also in differentiation of other neurons as well.

Aluminum-induced neurofibrillary degeneration affects a subset of neurons in rabbit cerebral cortex, basal forebrain and upper brainstem.

Neurofibrillary tangles in Alzheimer's disease show a predilection for cortical pyramidal and subcortical projection neurons. The antigenic composition, neuronal specificity and distribution of aluminum-induced neurofibrillary degeneration were examined in regions of rabbit brain analogous to those that develop neurofibrillary tangles in Alzheimer's disease. Neurofibrillary degeneration was induced by intraventricular instillation of aluminum chloride. In aluminum-treated rabbits, intensely immunoreactive filamentous aggregates were seen in affected neuronal perikarya after staining with an antiphosphorylated neurofilament antibody (SMI 31), while in controls immunoreactivity was confined to axon-like elements. Monoclonal antibodies against Microtubule-associated protein 2 and tau, which stain human neurofibrillary tangles, did not stain aluminum-induced neurofibrillary degeneration. Pyramidal neurons exhibiting neurofibrillary degeneration formed a discrete linear pattern in layers III and V of cortex. Cortical somatostatin and nicotinamide adenine dinucleotide phosphate diaphorase-reactive neurons identified in double-stained sections were unaffected. Large perikarya in the vicinity of the globus pallidus, some of which contained acetylcholinesterase, were frequently SMI 31-immunoreactive. Among the cell groups affected in the upper brainstem were the nucleus raphe dorsalis and locus coeruleus. These findings show that aluminum-induced neurofibrillary degeneration differs antigenically from neurofibrillary tangles in Alzheimer's disease. Nevertheless, many neuronal subsets that are particularly susceptible to Alzheimer's disease, including cortical pyramidal neurons, basal forebrain cholinergic neurons and upper brainstem catecholaminergic neurons, are also affected by aluminum-induced neurofibrillary degeneration.