Ultrastructural immunogold labeling of glial filaments in osmicated and unosmicated epoxy-embedded tissue

Alexander disease: The story behind an eponym

In 1949, William Stewart Alexander (1919-2013), a young pathologist from New Zealand working in London, reported the neuropathological findings in a 15-month-old boy who had developed normally until the age of seven months, but thereafter had progressive enlargement of his head and severe developmental delay. The most striking neuropathological abnormality was the presence of numerous Rosenthal fibers in the brain. The distribution of these fibers suggested to Alexander that the primary pathological change involved astrocytes. In the next 15 years, five similar patients were reported, and in 1964 Friede recognized these cases reflected a single disease process and coined the eponym "Alexander's disease" to describe the disorder. In the 1960s, electron microscopy confirmed that Rosenthal fibers were localized to astrocytes. In 2001, it was shown that Alexander disease is caused by mutations in the gene encoding glial fibrillary acidic protein, the major intermediate filament protein in astrocytes. Although the clinical, imaging, and pathological manifestations of Alexander disease are now well known, few people are familiar with Alexander's career. Although he did not make a further contribution to the literature on Alexander disease, his observations and accurate interpretation of the neuropathology have justified the continued use of the eponym "Alexander disease."

Heat-induced Antigen Retrieval in Conventionally Processed Epon-embedded Specimens: Procedures and Mechanisms

We studied the effectiveness of heat-induced antigen retrieval (HIAR) in conventionally processed, epon-embedded specimens and the mechanisms of HIAR in the specimens. Frozen sections were first immunostained to examine the possibility of using HIAR for 18 antigens to avoid the effects of epoxy resin embedment. The antigenicity of 7 out of 18 antigens was retrieved with glutaraldehyde fixation followed by osmium tetroxide treatment whereas none were retrieved with glutaraldehyde fixation without post-osmication. Six antigens also exhibited positive immunostaining in semi-thin epon sections when the sections were deplasticized with sodium ethoxide followed by autoclaving. In the immunoelectron microscopy with the post-embedding method, positive reactions with fine ultrastructures were obtained using HIAR without deplasticization. These results suggested that osmium tetroxide binds to ethylene double bonds (which are introduced into protein crosslinks by glutaraldehyde) and forms an extremely stable resonance interaction with the Schiff bases, thus destabilizing the protein crosslinks. Heating also further degrades these crosslinks. The present study demonstrated that archival epon blocks can be useful resources for immunohistochemical studies for both light and electron microscopy.

Evidence for the existence of secretory granule (dense-core vesicle)-based inositol 1,4,5-trisphosphate-dependent Ca2+ signaling system in astrocytes

BACKGROUND

The gliotransmitters released from astrocytes are deemed to play key roles in the glial cell-neuron communication for normal function of the brain. The gliotransmitters, such as glutamate, ATP, D-serine, neuropeptide Y, are stored in vesicles of astrocytes and secreted following the inositol 1,4,5-trisphosphate (IP3)-induced intracellular Ca2+ releases. Yet studies on the identity of the IP3-dependent intracellular Ca2+ stores remain virtually unexplored.

PRINCIPAL FINDINGS

We have therefore studied the potential existence of the IP3-sensitive intracellular Ca2+ stores in the cytoplasm of astrocytes using human brain tissue samples in contrast to cultured astrocytes that had primarily been used in the past. It was thus found that secretory granule marker proteins chromogranins and secretogranin II localize in the large dense core vesicles of astrocytes, thereby confirming the large dense core vesicles as bona fide secretory granules. Moreover, consistent with the major IP3-dependent intracellular Ca2+ store role of secretory granules in secretory cells, secretory granules of astrocytes also contained all three (types 1, 2, and 3) IP3R isoforms.

SIGNIFICANCE

Given that the secretory granule marker proteins chromogranins and secretogranin II are high-capacity, low-affinity Ca2+ storage proteins and chromogranins interact with the IP3Rs to activate the IP3R/Ca2+ channels, i.e., increase both the mean open time and the open probability of the channels, these results imply that secretory granules of astrocytes function as the IP3-sensitive intracellular Ca2+ store.

State of the art in antigen retrieval for immunohistochemistry

The masking effects of antigens by chemical fixation, processing, embedding media interactions, represent a serious problem for immunohistochemical purposes. Fortunately, different approaches in antigen retrieval exist. These techniques are relatively recent and continuously expanding. This review focuses on the present state of the art in antigen retrieval methods for immunohistochemistry in light and electron microscopy. Moreover, a brief discussion on the chemical aspects of fixation, mechanism of retrieval, as well as its efficacy, is given.

Ependymomas with neuronal differentiation: a morphologic and immunohistochemical spectrum

The category of mixed glioneuronal tumors of the CNS is rapidly losing its definition as encompassing tumors composed of histologically distinct neuron variants and glia. We encountered five ependymomas with neuronal differentiation seen in two by histology, in two by immunohistochemistry alone, and in one by electron microscopy. Antibodies against GFAP, S-100 protein, neurofilament protein, chromogranin, synaptophysin, Neu-N, and EMA were applied. Ultrastructural studies were also performed. In addition, 33 randomly selected ependymomas of various histologic types were screened for these same antigens. Cases 1 and 2 were anaplastic and showed clearly defined neuropil islands or pale islands as in nodular desmoplastic medulloblastoma, respectively. The tumors affected a 16-year-old male and a 5-year-old female and involved the right frontoparietal lobe and fourth ventricle, respectively. The islands were positive for synaptophysin and Neu-N (cases 1 and 2), and chromogranin (case 1). Cases 3-5, as well as 7 of the 33 screened ependymomas, showed a suggestion of neuronal differentiation by immunohistochemistry alone, including immunoreactivity for Neu-N (n = 8), synaptophysin (n = 4), neurofilament protein (n = 4), and chromogranin (n = 2). Five tumors each were WHO grade II and III. Electron microscopy performed on the two cases with neuronal islands demonstrated microtubule bundles and dense core granules (case 1) and poorly differentiated cells with high nuclear/cytoplasmic ratios, with intermediate filament accumulation and rare cilia (case 2). Cases identified by immunohistochemistry or electron microscopy demonstrated dense core granules (n = 5) and aligned microtubules (n = 3). Neuronal differentiation occurs in ependymomas but is less frequently definitive (histologic, ultrastructural) than merely a limited immunohistochemical finding. The clinical significance of these observations is unknown but deserves further exploration.

Heat-induced antigen retrieval: mechanisms and application to histochemistry

Since the introduction of the fluorescence-labeled antibody method by Coons et al. [Immunological properties of antibody containing a fluorescent group. Proc Soc Exp Biol Med 47, 200-2002], many immunohistochemical methods have been refined to obtain high sensitivity with low background staining at both light and electron microscopic levels. Heat-induced antigen retrieval (HIAR) reported by Shi et al. in the early 1990s has greatly contributed to immunohistochemical analysis for formalin-fixed and paraffin-embedded (FFPE) materials, particularly in the field of pathology. Although antigen retrieval techniques including enzyme digestion, treatment with protein denaturants and heating have been considered tricky and mysterious techniques, the mechanisms of HIAR have been rapidly elucidated. Heating cleaves crosslinks (methylene bridges) and add methylol groups in formaldehyde-fixed proteins and nucleic acids and extends polypeptides to unmask epitopes hidden in the inner portion of antigens or covered by adjacent macromolecules. In buffers having an appropriate pH and ion concentration, epitopes are exposed without entangling the extended polypeptides during cooling process, since polypeptides may strike a balance between hydrophobic attraction force and electrostatic repulsion force. Recent studies have demonstrated that HIAR is applicable for immunohistochemistry with various kinds of specimens, i.e., FFPE materials, frozen sections, plastic-embedded specimens, and physically fixed tissues at both the light- and electron-microscopic levels, and have suggested that the mechanism of HIAR is common to aldehyde-fixed and aldehyde-unfixed materials. Furthermore, heating has been shown to be effective for flow cytometry, nucleic acid histochemistry (fluorescein in situ hybridization (FISH), in situ hybridization (ISH), and terminal deoxynucleotidyl transferase-mediated nick labeling (TUNEL)), and extraction and analysis of macromolecules in both FFPE archive materials and specimens processed by other procedures. In this article, we review mechanism of HIAR and application of heating in both immunohistochemistry and other histochemical reactions.

Unusual malignant glioneuronal tumors of the cerebrum of adults: a clinicopathologic study of three cases

Malignant glioneuronal tumors of the brain are rare and poorly characterized. Herein, we report the clinicopathologic features of three examples with unusual morphologies including both glial and neuronal differentiation. Hematoxylin and eosin-stained slides were reviewed in all cases. Immunohistochemical stains were performed on formalin-fixed, paraffin-embedded sections. Transmission electron microscopy (EM) was performed on both formalin-fixed (n=1) and paraffin embedded tissue (n=2). The immunogold technique for localization of GFAP was also performed. Two patients were male and one was female, age 66, 84, and 34 years, respectively. Radiologic studies demonstrated hyperdensity on CT (n=3), multicentricity (n=2), and a cortical based solid component with a cystic extension into underlying white matter (n=2). At surgery, all three tumors were superficial and relatively circumscribed. Histologically, they were composed of large epithelioid cells (n=3), spindle cells (n=1), and poorly differentiated smaller cells with high nuclear/cytoplasmic ratios (n=1). Brisk mitotic activity and coagulative non-palisading necrosis were present in all cases. The tumors were immunopositive for GFAP (n=3), S-100 (n=3), synaptophysin (n=3), chromogranin (n=3), Neu-N (n=2), and neurofilament protein (n=2). Stains for EMA were negative. EM demonstrated convincing neurosecretory granules in one case, some in filament-containing cells immunogold labeled for GFAP. Two patients expired 3-5 weeks after surgery. True malignant neoplasms with glial and neuronal differentiation do occur in the central nervous system of adults and may pursue a highly aggressive course. The use of minimal diagnostic criteria, e.g., immunoreactivity for a single antigen like neurofilament protein, may not be sufficient and should be discouraged.

Alexander-disease mutation of GFAP causes filament disorganization and decreased solubility of GFAP

Alexander disease is a fatal neurological illness characterized by white-matter degeneration and the formation of astrocytic cytoplasmic inclusions called Rosenthal fibers, which contain the intermediate filament glial fibrillary acidic protein (GFAP), the small heat-shock proteins HSP27 and αB-crystallin, and ubiquitin. Many Alexander-disease patients are heterozygous for one of a set of point mutations in the GFAP gene, all of which result in amino acid substitutions. The biological effects of the most common alteration, R239C, were tested by expressing the mutated protein in cultured cells by transient transfection. In primary rat astrocytes and Cos-7 cells, the mutant GFAP was incorporated into filament networks along with the endogenous GFAP and vimentin, respectively. In SW13Vim– cells, which have no endogenous cytoplasmic intermediate filaments, wild-type human GFAP frequently formed filamentous bundles, whereas the R239C GFAP formed `diffuse' and irregular patterns. Filamentous bundles of R239C GFAP were sometimes formed in SW13Vim– cells when wild-type GFAP was co-transfected. Although the presence of a suitable coassembly partner (vimentin or GFAP) reduced the potential negative effects of the R239C mutation on GFAP network formation, the mutation affected the stability of GFAP in cells in a dominant fashion. Extraction of transfected SW13Vim– cells with Triton-X-100-containing buffers showed that the mutant GFAP was more resistant to solubilization at elevated KCl concentrations. Both wild-type and R239C GFAP assembled into 10 nm filaments with similar morphology in vitro. Thus, although the R239C mutation does not appear to affect filament formation per se, the mutation alters the normal solubility and organization of GFAP networks.

Alexander disease: a review and the gene

This review presents historical and clinical information on the rare human brain disorder known as Alexander disease (ALX), and reports on the recent discovery of the gene that appears to be causative. The disease is a fatal, white matter disorder (leukodystrophy) of childhood. Adult onset cases also have been described, but it has not been clear whether they represent the same disease. Until recently the diagnosis was made by the pathological examination of brain tissue, in which abundant Rosenthal fibers were found. These abnormal structures occurred within astrocytes, but their composition was unclear. In 1985, a child underwent a diagnostic brain biopsy at this institution, which established the diagnosis of ALX. Ultrastructural immunocytochemistry revealed that the Rosenthal fibers contained abundant amounts of glial fibrillary acidic protein (GFAP), a normal component of astocytic intermediate filaments. Thus, the gene for this filament protein was considered a candidate gene for the cause of ALX, and DNA samples from children presumed or proven to have this disorder were banked for future study. Other work on the same brain biopsy showed that Rosenthal fibers also contained abundant alphaB-crystallin, a heat shock protein, but no defect was found in its gene. A decade after the biopsy, a transgenic mouse with an extra copy of the gene for GFAP was produced. These mice died early and their brains contained Rosenthal fibers. Although not an exact model for ALX, this also suggested that the gene for GFAP should be considered a candidate gene for ALX. Subsequent research has demonstrated that the great majority of childhood ALX cases contain mutations in the gene for GFAP. This work is now being extended as a diagnostic test, as well as to seek understanding of the pathogenesis of ALX and possible approaches for treatment.

Mixed ependymoma-neuroendocrine tumor of the lateral ventricle

A histologically unique glioneuronal neoplasm occupying the lateral ventricle of a child was immunohistochemically and ultrastructurally characterized. Its principal component exhibited the characteristic features of ependymoma, whereas a minor population of neuroendocrine cells, occurring singly and in small clusters, lay scattered throughout the ependymoma component. Yet another unusual finding was the presence of numerous elastic fibers within the extracellular matrix. This tumor is considered to represent a true mixed neoplasm consisting of ependymal and neuroendocrine elements.

Semiquantitative postembedding characterization of intermediate filaments in central nervous system lesions using immunoelectron microscopy

Standardized postembedding immunoelectron microscopy was performed to demonstrate glial fibrillary acidic protein (GFAP) and vimentin in individual intermediate filaments to determine the diagnostic value of demonstrating ultrastructural and immunophenotypic characteristics of intermediate filaments in routine brain biopsy specimens. Dual expression of GFAP and vimentin was observed in the astroblastoma and astrocytes of Alexander's disease. The antigen availability for vimentin, however, was too low to allow reliable assessment of the GFAP:vimentin ratio in individual intermediate filaments and/or filament bundles. In meningioma, only vimentin positive intermediate filaments were found. GFAP positive intermediate filaments were present in all other specimens except the oligodendroglial components of the mixed glioma, which were devoid of intermediate filaments. GFAP positivity in the filamentous periphery and electron-dense core of Rosenthal fibers was demonstrated. Technical and tissue processing factors had a significant effect on particle density values obtained for individual specimens. Although the number, distribution, and density of glial intermediate filaments varies in different astroglial entities, correlation of particle density values determined by immunoelectron microscopy with relative GFAP concentrations in different lesions requires utmost caution. Nevertheless, application of the postembedding approach to routinely fixed biopsy specimens indicated an association of different entities with the exclusive presence of GFAP and/or vimentin in individual intermediate filaments, thus emphasizing the diagnostic value of intermediate filament typing for pathological characterization.

Tuber and subependymal giant cell astrocytoma associated with tuberous sclerosis: an immunohistochemical, ultrastructural, and immunoelectron and microscopic study

The cellular nature of the giant eosinophilic cells of tuber and of the cells comprising subependymal giant cell astrocytoma (SEGA) in tuberous sclerosis (TS) remains unclear. To assess the characteristics of these lesions, 13 tubers and 6 SEGA were immunohistochemically studied with glial and neuron-associated antigens. In addition to conventional ultrastructure, 6 tubers and 8 SEGA were subjected to immunoelectron microscopic study for glial fibrillary acidic protein (GFAP) and somatostatin. Eosinophilic giant cells of tubers were positive for vimentin (100%), GFAP (77%) and S-100 protein (92%); such cells were also found to a various extent to be reactive for neuron-associated antigens, including neurofilament (NF) proteins (38%) or class III beta-tubulin (77%). SEGA also showed variable immunoreactivity for GFAP (50%) or for S-100 protein (100%); NF epitopes, class III beta-tubulin, and calbindin 28-kD were expressed in 2 (33%), 5 (83%) and 4 (67%) cases, respectively. Cytoplasmic staining for somatostatin (50%), met-enkephalin (50%), 5-hydroxytryptamine (33%), beta-endorphin (33%) and neuropeptide Y (17%) was noted in SEGA, but not in tubers. Ultrastructurally, the giant cells of tubers and the cells of SEGA contained numerous intermediate filaments, frequent lysosomes and occasional rectangular or rhomboid membrane-bound crystalloids exhibiting lamellar periodicity and structural transition to lysosomes. Some SEGA cells showed features suggestive of neuronal differentiation, including stacks of rough endoplasmic reticulum, occasional microtubules and a few dense-core granules. Furthermore, in one case of tuber, a process of a single large cell was seen to be engaged in synapse formation. Intermediate filaments within a few cells of both lesions were decorated by gold particle-labeled GFAP antiserum. Within the tumor cells of SEGA, irregular, non-membrane-bound, electron-lucent areas often contained somatostatin-immunoreactive particles, whereas the latter could not be detected in tuber. The present study provides further evidence of divergent glioneuronal differentiation, both in the giant cells of tubers and the cells of SEGA. The findings of similar cells at different sites, including the subependymal zone, white matter ("heterotopias"), and cortex indirectly supports the idea that these lesions of TS result from a migration abnormality.

Ultrastructural study of the optic nerve in blind mole-rats (Spalacidae, Spalax)

The optic nerve in two species of subterranean mole-rats (Spalacidae) has been examined at the ultrastructural level. The axial length of the eye and the diameter of the optic nerve are 1.9 mm and 52.5 microns in Spalax leucodon, and 0.7 mm and 80.8 microns in Spalax ehrenbergi, respectively. An anti-glial fibrillary acidic protein postembedding procedure was used to distinguish glial cell processes from axons. In both species, the optic nerve is composed exclusively of unmyelinated axons and a spatial distribution gradient according to the size or the density of fibers is lacking. The optic nerve of S. leucodon contains 1790 fibers ranging in diameter from 0.07-2.30 microns (mean = 0.57 microns), whereas in S. ehrenbergi, only 928 fibers, with diameters of 0.04-1.77 microns (mean = 0.53 microns) are observed. In S. ehrenbergi, a higher proportion of glial tissue is present and the fascicular organization of optic fibers is less obvious. Distribution gradients according to size frequency or density of fibers in the optic nerve are absent in both species. Comparison with other mammals suggests that although ocular regression in microphthalmic species is correlated with a significant decrease in the total number of optic fibers and the relative proportion of myelinated fibers, no difference in the absolute size range of unmyelinated axons is observed. The total absence of myelinated fibers in Spalax may be related to the subcutaneous location of the eyes.(ABSTRACT TRUNCATED AT 250 WORDS)

Ultrastructural characteristics of glial fibrillary acidic protein expression in epoxy resin-embedded human brain tumors

Thirteen surgically removed, epoxy resin (Durcupan ACM or Epon 812)-embedded human brain tumors were examined for glial fibrillary acidic protein (GFAP) content in semithin and ultrathin sections with the immunogold-silver staining method. Mild aldehyde fixation and the hydrophobic resin embedding did not interfere with the antigenicity, since silver intensification of the immunogold marker provided excellent visualization of the reaction on both light microscopic and ultrastructural levels. The GFAP reaction was usually localized on the glial intermediate filament bundles, usually correlating well with the amount of filaments. The unstained filamental regions of two ependymomas might correspond to the vimentin expression revealed by double labeling in semithin sections. Occasional GFAP immunopositivity without filamental appearance was observed in one of the oligodendrogliomas, as patchy electron-dense cytoplasmic corpuscules, in Rosenthal fibers and in some mainly necrobiotic tumor cells, reflecting a possible connection with glial filaments.

2,5-Hexanedione-induced intermediate filament aggregates contain ubiquitin-protein conjugate immunoreactivity and resemble Rosenthal fibres

A number of chronic degenerative disorders including cerebellar astrocytomas and Parkinson's disease are characterized by the presence of cytosolic inclusions which contain intermediate filament (IF) aggregates and ubiquitin-protein conjugate immunoreactivity. In cerebellar astrocytomas these inclusions are known as Rosenthal fibres. 2,5-hexanedione (HD) treatment is known to induce IF aggregates in cells in culture. HD-induced aggregates have therefore been studied as a potential model for the clinical inclusions. Exposure of astrocyte cultures to 2 mM HD for 2 or 4 weeks led to the formation of aggregates of the IFs (glial fibrillary acidic protein and vimentin). The aggregates contained ubiquitin-protein conjugates, which, on electron microscopy appeared to be localized in a peripheral shell. In addition, ubiquitin mRNA levels were found to be elevated approximately threefold by HD treatment. HD-induced inclusions and Rosenthal fibres were found to share a number of features. HD administration, therefore, appears to be a suitable model for the production of pathological inclusions.

LR Gold embedding of nervous tissue for immunoelectron microscopy studies

A major drawback of all acrylic resins commonly used for post-embedding immunocytochemical studies of the central nervous system is the disruption of the ultrastructural morphology, due to the high lipid content of neural tissue. We have investigated the suitability of the acrylic resin LR Gold, which has been employed recently for immunogold labeling studies in several non-neural tissues. Optimal preservation of both antigenicity and ultrastructure of nervous tissue was obtained after en bloc staining with uranyl acetate, followed by total dehydration in acetone and curing at low temperature. Cell membranes and myelin sheaths, which are usually lost with other acrylic resins, were well maintained. The degree of antigenicity of LR Gold-embedded tissues was comparable to that of LR White-embedded one, but the morphologic detail was much better preserved. The use of LR Gold is particularly advantageous for studying neurodegenerative disorders such as Alzheimer disease.

Preferential expression of alpha B-crystallin in astrocytic elements of neuroectodermal tumors

Recently the authors have identified a major component of Rosenthal fibers as alpha B-crystallin, a major lens protein. In the current study the authors investigated the expression of alpha B-crystallin in four cultured glioma cell lines and in 115 human neuroectodermal tumors. alpha B-crystallin was expressed differentially by those glioma cell lines, but not by neuroblastoma cell lines. Northern blot analysis revealed two distinct messages for alpha B-crystallin in C-6, whereas only a single message in U-373MG and G26-24. In human surgical specimens positive immunostaining was frequently observed in the following brain tumors: pilocytic astrocytoma of the juvenile type, anaplastic astrocytoma, glioblastoma multiforme, and subependymal giant cell astrocytoma. The astrocytic elements of mixed oligoastrocytomas, glioblastomas with sarcomatous components, and gangliogliomas were likewise strongly stained. In contrast, little immunoreactivity was observed in ependymal and choroid plexus tumors. Thus, alpha B-crystallin is mainly expressed by astrocytic tumors among neuroectodermal neoplasms, without regard to the presence of Rosenthal fibers.

Immunoelectron microscopic localization of elastic tissue components in archival tissue samples

Tissue samples that have been stored for many years, in different media and under a variety of conditions, have been examined by modern techniques of immunoelectron microscopy, using antibodies against elastic tissue components. A range of postembedding restorative procedures has been identified, which will allow reliable immunolocalization of antibodies against the elastic tissue component of such specimens. These methods have been applied successfully to autopsy-derived material, fixed in buffered formaldehyde, to archival material stored frozen at -70 or -20 degrees C, to specimens fixed for electron microscopy and stored for many years in buffer, and even to archival material from formaldehyde-fixed, paraffin-embedded blocks, reprocessed for electron microscopic examination. The successful restorative methods included pre-treatment of the sections with 6 M guanidine hydrochloride, or 1 M Tris/saline, each containing 100 mM dithiothreitol (a reducing agent) followed by alkylation with 220 mM iodoacetamide. The application of these techniques allowed reliable study of elastic tissue antibody distributions in archival tissues that could not be obtained again, as well as comparative studies with tissues processed many years previously.

Immunoelectron microscopy of Rosenthal fibers

Seventeen intracerebral gliomas containing Rosenthal fibers (RF) were studied by an immunoperoxidase method for localization of ubiquitin (UB), glial fibrillary acidic protein (GFAP), desmin and vimentin (VIM). The majority of RF showed an immunohistochemically negative core surrounded by a ring of overlapping reactions for UB, GFAP and VIM. Many RF were entirely negative for UB and intermediate filaments (IF). Immunoelectron microscopic localization of UB and GFAP was performed on seven selected tumors. UB was found in all RF and on IF in the proximity of RF. GFAP reaction was localized on astrocytic IF, including those trapped within RF, and within the granular component of some RF. In contrast to the light microscopic studies, neither GFAP- nor UB-negative RF were found on immunoelectron microscopy. VIM reaction on IF and a few RF was demonstrated in one tumor processed at low temperature into Lowicryl; it was much weaker than that for GFAP. Many cells with RF contained lysosome-like inclusions with material displaying electron density similar to adjacent RF; few of these inclusions were reactive for UB. It is concluded that RF formation is associated with ubiquitination of astrocytic IF. GFAP- and VIM-immunoreactive IF and products of their disintegration contribute to RF material. It is also suggested that the lysosomal system of astrocytes partially degrades RF.