Effects of epidermal growth factor and dibutyryl cyclic adenosine monophosphate on the migration pattern of astrocytes grafted into adult rat brain.

OBJECTIVE: Neonatal rat astrocytes transplanted into host brains migrate in specific patterns, which are determined by the developmental stage of the host brain and the region of implantation. We hypothesized that the differentiation state of the implanted astrocytes could also affect astrocyte migration. METHODS: Astrocytes derived from neonatal rats (1-4 d) were placed in culture and exposed to growth- or differentiation-promoting agents (e.g., epidermal growth factor or dibutyryl cyclic adenosine monophosphate). Treated cells were then injected into different regions of the adult rat brain. At 3, 6, and 9 days after implantation, the extent and pattern of astrocyte migration after injection into the cortex, hippocampus, and corpus callosum were assessed. RESULTS: Astrocytes pretreated with either factor did not migrate during the first 3 days after implantation into the host cortex and hippocampus, whereas untreated cells migrated extensively by Day 3. After 9 days, implanted cells that had been pretreated with dibutyryl cyclic adenosine monophosphate began to demonstrate migratory activity, whereas those exposed to epidermal growth factor remained at the site of implantation. These findings corresponded to the effects of these agents in culture. On the other hand, cells implanted into the corpus callosum migrated in spite of pretreatment. CONCLUSION: Epidermal growth factor and dibutyryl cyclic adenosine monophosphate each altered the cells in culture such that they were inhibited from migrating after transplantation into the host cortex and hippocampus. This finding suggests that the activation of either growth or differentiation cascades partially inhibits the migratory ability in these cells either through effects on their internal migratory potentials or their responsiveness to external migratory signals. In contrast, cells implanted into the corpus callosum migrated in spite of pretreatment, suggesting that this structure may present migratory cues sufficient to override the effects of treatment.

Role of glial fibrillary acidic protein expression in the biology of human glioblastoma U-373MG cells.

OBJECT: The relationship between glial fibrillary acidic protein (GFAP) expression and glial tumor cell behavior has not been well defined. The goal of this study was to examine this relationship further. METHODS: To investigate the relationship between GFAP expression and glial tumor cell behavior, the authors isolated clones from the human glioblastoma cell line, U-373MG, according to their level of GFAP expression. Immunochemical analysis demonstrated that one clone had consistently low GFAP expression (approximately 93% of cells were GFAP negative), whereas a second clone had consistently high GFAP expression (approximately 80% of the cells were GFAP positive). The structure, population doubling time, saturation density, anchorage-independent growth, migratory rate, and invasive potential of these two clones were determined in relation to their level of GFAP expression. Morphologically, both clones were composed of ameboid as well as stellate components. Although the population doubling times of the two clones were equally rapid, the clone with low GFAP expression demonstrated a slightly higher saturation density compared with the clone with high GFAP expression. In an anchorage-independent environment (soft agar), a greater difference in growth characteristics was noted between the two clones: the high-expression clone formed more colonies and these colonies were compact, well defined, and spherical, whereas the low-expression clone formed predominantly smaller, two-dimensional colonies with vague boundaries and isolated cells or groups of cells at the periphery. In contrast to these minor differences between the clones, the low-expression clone showed a markedly increased migratory rate and invasive potential compared with the high-expression clone. Therefore, the clone with reduced GFAP expression appeared more aggressive, demonstrating decreased contact inhibition, increased migratory rate, and increased invasive potential. CONCLUSIONS: These results suggest a direct correlation between GFAP expression and some measures of aggressive tumor growth and transformation properties.

Cell type-specific development of rodent central nervous system progenitor cells in culture.

OBJECT: The aim of the study was to assess the pluripotential central nervous system (CNS) progenitor cells that give rise to the many differentiated neuronal and glial cell types of the adult mammalian brain and the role of peptide growth factors such as the epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF). The action of these factors is crucial to the survival and ultimate differentiation of these CNS progenitor cells. However, the precise role of EGF and bFGF in the time course of cellular development, the acquisition of cell type specificity, and possible differential mitogenic behavior has not been clearly defined. METHODS: The authors defined the time course of CNS progenitor cell development in cultured embryonic rodent cells by using immunocytochemical analysis to identify the expression of pluripotential (nestin)-, neuron (microtubule-associated protein-2 [MAP-2])-, and glia (glial fibrillary acidic protein [GFAP])-specific proteins in response to treatment with EGF and bFGF alone or in combination. The bromodeoxyuridine (BUdR) labeling index for each treatment group was used to define the mitogenic effects of each growth factor. In this investigation, the authors observed that progenitor cells develop in a stereotypical fashion when exposed to bFGF or EGF. Marked staining for nestin was evident soon after plating. This declined over time as staining for MAP-2 and GFAP increased. When treated with EGF alone, cells maintained their nestin immunoreactivity longer than those treated with bFGF alone or in combination with EGF. Treatment with bFGF alone promoted a significant increase in MAP-2 and, to a much lesser extent, GFAP reactivity. This was observed concomitant with the decline in nestin staining. The BUdR labeling index was similar among the different treatment groups and declined similarly over time in all treatment groups. CONCLUSIONS: The effects of EGF and/or bFGF on the expression of development- and lineage-specific markers likely reflect the specific effects of these factors on developmental processes. These data indicate that bFGF exerts a preferential effect on neuronal development and, to a lesser extent, glial development, which is not explained by selective mitogenicity. The persistence of nestin staining seen in the cells treated with EGF alone indicates that EGF may function as a stem cell survival factor. This study provides evidence that CNS cell type-specific development can be altered by the manipulation of peptide growth factors that act as differentiation agents.

The role of the epidermal growth factor receptor in human gliomas: I. The control of cell growth.

The epidermal growth factor receptor (EGFR) gene is amplified in over 40% of primary human glioblastomas and overexpressed in the majority. The authors' investigations demonstrate that the function of the EGFR in glioblastomas is distinct from that in other human cancers because it does not appear to mediate the primary growth-promoting effect of EGF. Findings show that the level of EGFR expression does not directly predict the growth response to EGF, with growth stimulated in some cells but inhibited in others when cells were cultured in plastic dishes. On the other hand, when human glioblastoma cells were placed in soft agar cultures, the cell line expressing the highest levels of the EGFR demonstrated considerable colony formation in response to EGF treatment. In addition, cell lines with the highest EGFR levels were also more resistant to the growth-suppressive effects of retinoic acid when maintained in soft agar. These observations suggest that even though the overexpression of the EGFR did not confer a distinct growth advantage to glioma cells cultured on flat culture dishes, the ability of these cells to maintain anchorage-independent growth in soft agar especially in response to EGF and retinoic acid is facilitated. Because anchorage-independent growth is the best in vitro correlate to tumorigenicity, amplification and overexpression of the EGFR in human glioblastoma cells may be in part responsible for the tumorigenic potential of these cells.

The role of the epidermal growth factor receptor in human gliomas: II. The control of glial process extension and the expression of glial fibrillary acidic protein.

Our earlier investigations of the biology of the epidermal growth factor receptor (EGFR) in human gliomas demonstrated that the level of EGFR expression did not directly predict the glioma growth response to EGF, suggesting that the function of the EGFR in glioblastomas might not be limited to mediating the growth effects of EGF. We conducted the current studies to investigate the function(s) of the EGFR not related to growth control in human gliomas. These investigations show that the EGFR mediates the stimulative effects of EGF on glial process extension and glial fibrillary acidic protein (GFAP) expression. In addition, the level of EGFR expression correlates inversely with glioma cell responsiveness to differentiation promoting agents (for example, nerve growth factor and transforming growth factor-beta) that act through transmembrane tyrosine kinase receptors. Thus, glioma lines with a high level of EGFR expression (for example, T-98G cells) responded to fewer differentiation promoting factors than lines with a low level of EGFR expression (such as U-373MG cells). Our results suggest that the EGFR in gliomas may participate in mediating the process extension and GFAP stimulative effects of both EGF and other differentiation promoting agents. These properties represent components of the differentiated state in glia because their expression is stimulated by dibutyryl cyclic adenosine monophosphate in normal astrocytes. The involvement of the EGFR in the expression of these glial specific properties suggests that the EGFR may play an important role in glial differentiation.

Differential migration of astrocytes grafted into the developing rat brain.

Fetal and neonatal astrocytes migrate in specific patterns when transplanted into the adult rat host brain. However, it is unclear whether these astrocytes demonstrate the same degree of mobility during early brain development. In the present study, neonatal cortical, hippocampal, and hypothalamic astrocytes were collected from the brains of 1- to 3-day-old rats and placed in tissue culture. After 14 to 21 days, cultures enriched in astrocytes were harvested and labelled with either the fluorescent dye Fast Blue or fluorescein-labelled latex beads. They were then transplanted into the right frontal cerebrum of neonatal rats at 2, 5, 8, and 11 days postpartum. Seven days after transplantation, animals were sacrificed and their brains were fixed by immersion in aldehydes, sectioned on a cryostat, and examined with fluorescence microscopy. Transplanted astrocytes migrated along the corpus callosum, internal capsule, glial limitans, ventricular linings, and hippocampal structure. Labelled cells were also found in the contralateral hemisphere in day 2 brains. Migration in a radial fashion from the injection site toward the periphery was a particularly obvious pattern, and was most pronounced in these younger hosts. In days 5 and 8 rat brains, astrocyte migration became more restricted to the hemisphere of implantation. In 11-day-old host brains, hemispheric restriction and other region-specific influences became manifest and specifically modulated migration. Radial migration was absent in the 11-day-old host group except for cells of cortical origin. The observed results demonstrate that neonatal cortical, hippocampal, and hypothalamic astrocytes transplanted into the neonatal cerebrum migrate in patterns that are more extensive than in the adult brain.(ABSTRACT TRUNCATED AT 250 WORDS)

In vitro differentiation inhibits the migration of cultured neonatal rat cortical astrocytes transplanted to the neonatal rat cerebrum.

Neonatal rat astrocytes transplanted into the rat cerebrum migrate extensively. However, few of the molecular signals determining this migration have been defined. In the present study, in vitro modifications were designed to examine whether differentiation prior to transplantation would affect the magnitude or pattern of astrocyte migration in the neonatal host brain. Here, cortical astrocytes were collected from the brains of rats 1-3 days postpartum and purified by culturing them in DME medium supplemented with 10% calf serum. After 14-21 days, astrocytes were labelled with fluorescein-tagged latex microspheres for 16 hr; the label was then removed and replaced with either fresh medium or fresh serum-free medium plus 1 mM dbcAMP. After 48 hr, cells were harvested and then transplanted into the right frontal cerebrum of neonatal rats at 3 days postpartum by injection with a hand-held Hamilton syringe. Animals were sacrificed at 3, 6, 9, 15, 21 and 28 days after inoculation and their brains examined with fluorescence microscopy. Astrocytes not exposed to dbcAMP prior to implantation migrated along the corpus callosum, internal capsule, glial limitans, ventricular linings and the hippocampal structure. They also appeared to migrate in a radial fashion toward the periphery from the ventricular lining. Astrocytes treated with dbcAMP prior to transplantation did not appear to migrate into the neonatal parenchyma, remaining confined to the injection site for at least 6 days. Migration then appeared to commence at a normal rate after 9 days. Thus, neonatal cortical migrate outward in a pattern similar to that defined by the radial glia. Astrocytes differentiated by dbcAMP treatment, however, do not appear to migrate to any large degree in the neonatal brain until the treatment effect diminishes, suggesting that differentiation may represent an end-point to glial migration in the neonatal host brain.

Native astrocytes do not migrate de novo or after local trauma.

While transplanted astrocytes migrate in specific patterns in the recipient brains, it is not known whether native astrocytes behave similarly. The ability of normal astrocytes to migrate under non-transplant conditions was therefore explored. Native astrocytes were labelled in situ with fluorescent latex beads. These latex spheres were actively endocytosed by astrocytes in vitro, and it was therefore anticipated that these spheres would also be endocytosed by native astrocytes exposed to them. Labelling was accomplished by dissecting the pia mater away from a small region of the cerebral cortex and overlaying the area with Gelfoam containing fluorescent beads. After 2-4 h, the Gelfoam was removed and the wound was closed. At the end of 2-4 weeks, manipulated brains were harvested for fluorescence microscopy. In this analysis, fluorescent polyspheres had been taken up by both pial fibroblasts and astrocytes at the pial-glial margin. Labelled astrocytes [identified by glial fibrillary acidic protein (GFAP) staining] were neither hyperplastic nor hypertrophic. They were confined to the area of the original labelling site, and did not migrate either laterally across the pial margin or ventrally into the cortical layers. Knife wounding at the time of label application, either in the region of the label or distant from it, produced reactive astrocytes that were hypertrophic. These cells also did not migrate from the label site. These results suggest that astrocytes labelled by this method do not migrate in the absence of some transplant-derived stimulus even when stimulated by local wounding.

Demonstration of specific neuronal cell groups in rat brain by beta-galactosidase enzyme histochemistry.

beta-Galactosidase activity as illuminated by the indigogenic X-gal staining method has been used to demonstrate the presence of genetically modified cells carrying the reporter gene lacZ, coding for the E. coli enzyme. Endogenous activity has been assumed to be minimal since the pH optimum for the mammalian enzyme is 3.5-5.5, while the pH optimum for the E. coli enzyme (and thus of the staining procedure usually employed) is 7.3. Background staining has been reported to be limited to pericytes and a few specific neuronal cell groups. In contrast, our investigations of normal rat brain anatomy demonstrate that many specific neuronal cell groups possess endogenous beta-galactosidase activity when staining is performed at physiological pH. This suggests that background staining of endogenous beta-galactosidase activity in the rat brain has been underestimated. In addition, such specific activity would afford an additional means of identification and illustration of these cells.

Formation of PC12 tumors after transplantation into rat brains: dependence of time course on host age.

Rat pheochromocytoma PC12 cells form tumors when placed into the brains of Sprague-Dawley rats under specific conditions. We now show that tumorigenic potential is regulated by the microenvironment of the developing cerebrum. PC12 cell aggregates were identified in the periventricular or intraventricular spaces within 24 h after injection of cell suspensions into rat brains. In fetal or young neonatal (1-4-day-old) recipient rat brains, these cell aggregates formed large masses within 21 days. The tumor incidence declined in recipient neonates between the ages of 5 and 8 days. In both cases, tumors spread throughout the ventricular system and subarachnoid and Virchow-Robin spaces as they grew. In contrast, tumors were not generated by injections into adult rat brains or by placement of PC12 cell pellets into preformed cavities. Despite the loss of tumorigenicity, surviving cells were present at the injection site. The presence of surviving cells and the ability of another rat cell line (the C6 rat glioma line) to form tumors in adult rat brains suggest that an immune response is not solely responsible for the lack of PC12 tumorigenicity in adult rat brains. We propose that developmentally increasing local concentrations of specific factors (e.g., nerve growth factor of fibroblast growth factor) may also contribute to the suppression of tumor formation in this system.

Migratory behavior of PC12 cells transplanted into neonatal rat brain.

PC12 rat pheochromocytoma cells form tumors when transplanted into the forebrains of 1-4-day-old neonatal rats; thereafter, the incidence of tumor formation declines rapidly with increasing recipient age. The fate of PC12 cells transplanted into the forebrains of older neonates is thus not well defined. To examine the interactions of PC12 cells with this older neural environment, we transplanted [3H]thymidine-labeled PC12 cells into the brains of 5-day-old rats. In the brains of animals sacrificed 5 days after transplantation, clusters of labeled cells were found in and around the lateral and third ventricles. By 11 days after transplantation, single labeled cells were found to migrate into the hippocampus and the nearby cerebral cortex. Occasional invasion of the ventral hypothalamus from the third ventricle was also observed. Cells were rarely found to cross the midline or to invade the thalamus or the midbrain. The same pattern of labeling was found in the brains of animals sacrificed at 16 days after inoculation, suggesting that migration was completed by that time. No tumors were detectable, despite the implantation of cells in and around the ventricles. Control injections of [3H]thymidine alone or of [3H]thymidine-labeled astrocytes showed no labeling above background. These results suggest that PC12 cells migrate after inoculation into the brains of older neonatal rats. Additionally, this migration may be regionally constrained and dictated by the specific local trophic environment.

Migration of grafted rat astrocytes: dependence on source/target organ.

Neonatal rat cortical astrocytes migrate extensively after transplantation into the brains of adult hosts. However, the effects of cues provided by different sources of donor astrocytes and by different target sites of implantation on this migration is unknown. In order to investigate the significance of regional influences on glial migration, we established primary cultures of astrocytes derived from 1-3 day old rat cerebral cortex, hippocampus, and hypothalamus. After in vitro labelling with either Fast Blue or fluorescein-labelled latex beads, these astrocytes were inoculated into different target regions of the adult rat brain by stereotaxic injection with a Hamilton syringe. Astrocytes implanted into the cerebral cortex migrated extensively throughout the adult brain regardless of their donor source. These implanted cells were intimately associated with the ventricular wall, glial limitans, vasculature, and fiber bundles. Astrocytes homografted into the hippocampus and the hypothalamus migrated primarily within and around the respective homotopic target organs. Migration of astrocytes derived from cerebral cortex was also limited when injections were made into these two regions. In these latter cases, migration appeared to be less guided by other cellular or regional cues than was migration after implantation into the cerebral cortex. These results suggest that the migration patterns of astrocytes grafted after tissue culture are more dependent on the target implantation site than on the donor organ.

Orthogonal arrays are absent from the membranes of human glioblastomatous tissues.

Human astrocytic gliomas were studied with the freeze fracture technique. Orthogonal arrays of particles were noted in the plasma membranes of low-grade astrocytoma tissues. However, no such arrays were found in the plasma membranes of anaplastic glioma or glioblastoma tissues. Gap junctions were rarely seen in the membranes of these higher-grade gliomas; when seen, they consisted of relatively few particles in poorly organized plaques. These plasma membranes were dominated by randomly distributed single particles. These findings constitute aspects of the loss of differentiation in these malignant tumors.

Proto-oncogene abnormalities and their relationship to tumorigenicity in some human glioblastomas.

Human glioblastomas are highly malignant intracranial tumors, some of which demonstrate amplification of the epidermal growth factor-receptor (EGF-R) gene. Overexpression of this gene is seen in the majority of primary tumors; however, the role of the EGF-R gene in glial tumorigenesis is unknown. The authors explored the relationship between EGF-R gene expression and glioblastoma cell growth in vitro and in vivo and found that this level of EGF-R gene expression did not correlate with tumor cell growth either in soft agar or in the nude mouse. This suggests that the EGF-R gene is not involved in effecting direct growth stimulation in glial oncogenesis. Tumorigenesis involves differentiation arrest; therefore, the expression of several proto-oncogenes in neuroectodermal tumors was investigated to evaluate the potential involvement of the EGF-R gene in glial differentiation. A nonoverlapping expression of the N-myc and EGF-R genes was found in neuronal-derived and glial-derived tumors, respectively. This suggests that the EGF-R gene may be involved in differentiation or its arrest in glia.

Purification and characterization of a second form of acid lipase in human liver.

We describe here the purification and characterization of a form of acid lipase from human liver (designated ALII), which differed from the more abundant Mr-29000 form (ALI). ALII was solubilized from frozen human liver with Triton X-100 and purified 8500-fold by chromatography over concanavalin A-sepharose, CM-cellulose and finally h.p.l.c. over a Mono S column. ALII migrated as a single band on polyacrylamide-gel electrophoresis in both the presence and the absence of SDS. The Mr of ALII was estimated to be 58,500 by SDS/polyacrylamide-gel electrophoresis. Gel filtration on Sephacryl S-200 gave an apparent Mr of 69,000. 4-Methylumbelliferyl (4MU) palmitate, cholesterol oleate and triolein were substrates for ALII, with apparent Vmax values of 5000, 1100 and 2500 nmol/min per mg respectively and Km values of 1.0, 1.5 and 1.8 mM respectively. Cholesterol oleate and triolein were hydrolysed optimally by ALII at pH 4.5, whereas 4MU palmitate was hydrolysed optimally at pH 5.3. Antisera were raised against ALI and ALII and, on immunoblot analysis, no antigenic similarity was observed between ALI and ALII. Cellulose acetate electrophoresis followed by reaction with 4MU palmitate revealed two forms of lipase, corresponding to ALI and ALII. The two enzymes were also separated by hydrophobic chromatography. The activity of ALII was stimulated by several proteins and was partially inhibited by millimolar concentrations of NaCl, CaCl2 and MgSO4.

Identification of polypeptides associated with sarcolemmal vesicles enriched in orthogonal arrays.

Electron microscopy of freeze-fracture replicas from the sarcolemmas of fast-twitch muscle fibers reveals orthogonal arrays of particles. The biochemical nature of macromolecules associated with the sarcolemmal orthogonal array was investigated using muscle fragments and isolated sarcolemmal vesicles. Muscle fragments incubated in vitro with the lectin concanavalin A exhibited a clustering of orthogonal arrays into local patches. Treatment with other lectins did not result in the clustering of arrays. Clustering was inhibited by the addition of alpha-methyl-D-mannoside, a ligand which also binds concanavalin A. These results suggest that the orthogonal arrays (or associated components) specifically bind concanavalin A. Sarcolemmal vesicles from rabbit sacrospinalis (SAC) and rat extensor digitorum longus (EDL) (both primarily fast-twitch) and rat soleus (SOL) (primarily slow-twitch) were obtained by a combination of low-salt fractionation and sucrose density gradient centrifugation. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) of proteins and glycoproteins solubilized from these vesicles revealed several bands. Four of these bands were present in gels from both the rabbit and rat fast-twitch muscle sarcolemmal preparations (that contained arrays), yet were absent in gels from rat slow-twitch muscle sarcolemmal preparations (not bearing arrays). An enrichment in vesicles containing arrays was achieved by binding SAC sarcolemmal vesicles to Con A-Sepharose 4B beads. SDS-PAGE analysis of array-enriched vesicles from the concanavalin A beads revealed enrichment of three major bands at Mr 93,000, 54,000 and 49,000. These enriched bands correlate with three of the four bands common to fast-twitch EDL and SAC, yet absent in slow-twitch SOL sarcolemmal preparations. We conclude that at least one macromolecular component associated with the sarcolemmal orthogonal array is a concanavalin A binding glycoprotein. We further conclude that three candidates for this component co-purify with the morphological array, and have approximate molecular weights of 93,000, 54,000 and 49,000.

Orthogonal arrays are redistributed in the membranes of astroglia from alumina-induced epileptic foci.

Freeze-fracture techniques were used to examine the distribution of orthogonal arrays in the membranes of astroglia from specimens of cerebral cortex of epileptic and control rats. Astroglial membranes from experimental chronic epileptic cortex displayed normal densities of arrays in pial-glial regions surrounding the alumina granuloma. Astroglial membranes from deep parenchymal regions of epileptic cortex displayed increased numbers of arrays, especially in regions of membrane not normally displaying arrays. The apparent random distribution of these membrane specializations in reactive astroglia in alumina-induced epileptic foci is discussed in reference to one postulation of orthogonal array function.

The distribution of orthogonal arrays in the freeze-fractured rat median eminence.

The distribution of orthogonal arrays of particles and their relationships to gap and tight junctions have been studied in the glia of the freeze-fractured rat median eminence (ME). These rectilinear clusters of intramembrane particles are thought to represent trans-membrane channels for ions or metabolites, and were found to be densely packed on the membranous laminations of the pial-glial limitans. Additionally, arrays were found to be present on all of the perivascular glial end-feet examined. Two classes of end-feet were distinguished by their relative densities of orthogonal arrays. End-feet displaying low densities of arrays occurred more frequently in the internal zone, while end-feet displaying high densities occurred more often in the external zone. Similar distinctions based on array density could be made in membranes from other regions of the cell as well. Cross-fractures revealing the cytoplasm underlying these membranes often exposed lipid inclusion bodies, suggesting that membranes containing few arrays belong to tanycytes (or to 'astrocyte-like tanycytes'). The distribution of arrays appeared to be unrelated to the distribution of gap junctions in the membranes of astrocytes and tanycytes (and 'astrocyte-like tanycytes') of the ME, appearing near to and far from gap junctions with approximately equal frequency. Orthogonal arrays were absent from glial membranes near synaptic profiles in the ME. Arrays were also absent from the microvillous membranes of the apical surfaces of ependymal cells, from the cytoplasmic protrusions into the CSF of tanycytes, and from the vicinity of the tight and complex junctions linking the tanycyte and ependymal cell lateral membranes near their apical poles. These results suggest that there is a gradient of array density for most glia of the ME, increasing from the ventricular to the pial surface.

The distribution of orthogonal arrays and their relationship to intercellular junctions in neuroglia of the freeze-fractured hypothalamo-neurohypophysial system.

Using freeze-fracture techniques, we have investigated membrane specializations of the glia associated with the hypothalamo-neurohypophysial system of the rat. In the paraventricular (PVN) and supraoptic (SON) nuclei, astrocytes in areas of high neuronal density (i.e., magnocellular regions) display orthogonal arrays of 6--7 nm particles solely near gap junctions, while astrocytes in areas of lower neuronal density (i.e., parvocellular regions), contain additional arrays in membranes not displaying gap junctions. Arrays are especially numerous on astrocytic perivascular end-feet in both nuclei and in the laminations of the pial-glial limitans ventral to the SON. Ependymal cells near the PVN show arrays both on their lateral surfaces (displaying gap junctions) and on their apical surfaces (facing the CSF). Tight junctions are not noted on astrocytes or ependymal cells, but are noted on both the somas and myelin lamellae of oligodendroglia. Both of these latter membranes occasionally contain gap junctions as well; however, orthogonal arrays are never noted on oligodendroglia. The plasma membranes of pituicytes in the neurohypophysis display gap junctions, complex junctions, and tight junctions. Orthogonal arrays are noted near the first two of these, but not near the last. Arrays in the neutral lobe appear most dense on membranes adjacent to subpial or perivascular spaces. Pituicyte membranes containing orthogonal arrays appear infrequently near the neural stalk, increasing towards the distal end of the neural lobe. The distribution of orthogonal arrays in this system, as well as in other systems in which they have been noted, suggests a polarization of membrane activity.