The neuronal endoplasmic reticulum: its cytochemistry and contribution to the endomembrane system. I. Cell bodies and dendrites

Ultrastructural localization of delta-opioid receptor and Met5-enkephalin immunoreactivity in rat insular cortex

The insular cortex has been implicated in the reinforcing properties of opiates as well as in the integration of responses to sensory-motor stimulation. Moreover, the delta-opioid receptor (DOR) and the endogenous opioid ligand, Met5-enkephalin (ENK) are known to be prominently distributed in insular limbic cortex. To examine the anatomical sites for opioid activation of DOR in rat insular cortex, we used immunoperoxidase for detection of an antiserum raised against a peptide sequence unique to the DOR alone, and in combination with immunogold-silver labeling for ENK. Light microscopy showed intense DOR-like immunoreactivity (DOR-LI) in pyramidal cells and interneurons in deep laminae, and in varicose processes in both superficial and deep layers of the insular cortex. Ultrastructural analysis of layers V and VI in insular cortex showed that the most prominent immunoperoxidase labeling for DOR was in dendrites. This labeling was associated with asymmetric excitatory-type junctions postsynaptic to unlabeled terminals. Dendritic DOR-LI was also distributed along selective portions of non-synaptic plasma membranes and subsurface organelles. In dually labeled sections, dendrites containing DOR-LI sometimes received synaptic input from ENK-labeled terminals or more infrequently colocalized with ENK. Other axon terminals were exclusively immunolabeled for DOR or more rarely contained both DOR and ENK immunoreactivity. Within labeled axon terminals, distinct segments of the plasma membrane and membranes of immediately adjacent synaptic vesicles showed the largest accumulation of the peroxidase reaction product for DOR. These results indicate that in rat insular cortex DOR is primarily heteroreceptive, but also serves an autoreceptive function on certain ENK-containing neurons. Our results also provide the first ultrastructural evidence that in rat insular cortex endogenous opioids interact through the DOR (1) to modulate the postsynaptic responses to other excitatory afferents and (2) to presynaptically regulate the release of other neurotransmitters. The modulatory actions on both ENK-containing and non-ENK-containing neurons may contribute significantly to the reinforcing properties of exogenous opiates acting on the DOR in limbic cortex.

A neuroendocrine-specific protein localized to the endoplasmic reticulum by distal degradation

Regulated endocrine-specific protein, 18-kDa (RESP18), was previously cloned from rat neurointermediate pituitary based on its coordinate regulation with proopiomelanocortin and neuroendocrine specificity. RESP18 has no homology to any known protein. Although RESP18 is translocated across microsomal membranes after in vitro translation, AtT-20 pituitary tumor cells, which endogenously synthesize RESP18, do not release it into the culture medium. In this work, immunostaining and subcellular fractionation have identified RESP18 as an endoplasmic reticulum (ER) protein. Biosynthetic labeling and temperature block studies of AtT-20 cells demonstrated the localization of RESP18 to the ER lumen by a unique mechanism, degradation by proteolysis in a post-ER pre-Golgi compartment. Proteases in this compartment were saturated by exogenous RESP18 overexpression in AtT-20 cells. Furthermore, a calpain protease inhibitor enhanced secretion of RESP18 from AtT-20 cells overexpressing RESP18. Saturation and inhibition of the RESP18 degrading proteases allowed RESP18 to enter secretory granules and acquire a post-translational modification, likely O-glycosylation; this modified 21-kDa RESP18 isoform was the only RESP18 secreted. Rat anterior pituitary extracts contain 18-kDa and O-glycosylated RESP18 with similar properties. Exogenous RESP18 expression in hEK-293 cells demonstrated ER localization and RESP18 metabolism similar to AtT-20 cells, indicating that the cellular machinery involved in localizing RESP18 is not specific to neuroendocrine cells. The data implicate a novel ER localization mechanism for this neuroendocrine-specific luminal ER resident.

The organization of the endoplasmic reticulum and the intermediate compartment in cultured rat hippocampal neurons

The boundaries of the organelles of the biosynthetic endomembrane system are still controversial. In this paper we take advantage of the unique architectural organization of neurons to investigate the localization of a spectrum of compartment-specific markers with the goal of defining the location of the rough endoplasmic reticulum (ER), smooth ER, intermediate compartment, and the Golgi complex. Markers of the rough ER (signal sequence receptor), Golgi complex (mannosidase II), and the trans Golgi network (TGN38) were essentially restricted to the cell body and the initial segment of one of the cell's dendrites. In contrast the cytochemical reaction product for glucose 6 phosphate, a classical ER marker, in addition to staining ER structures in the cell body also reacted with smooth ER elements that extended into both axons and dendrites. These peripheral smooth ER elements also reacted at the immunofluorescence level for ER marker 3-hydroxy-3-methylglutaryl-coenzyme A reductase, as well as for calnexin and protein disulfide isomerase. We also analyzed the location of rab1, rab2, p58, the KDEL receptor, and beta-subunit of coatomer. These intermediate compartment markers were found predominantly in the cell body but also extended to the proximal parts of the dendrites. Collectively, our data argue that the ER of hippocampal neurons consists of functionally and spatially distinct and separated domains, and they stress the power of the hippocampal neuron system for investigations of the organization of the ER by light microscopy.

NSP-encoded reticulons, neuroendocrine proteins of a novel gene family associated with membranes of the endoplasmic reticulum

The novel NSP gene was previously shown to encode, among a variety of neuroendocrine cell types, two 3′-overlapping transcripts, a 3.4 kb one for NSP-A (776 amino acids) and a 1.8 kb one for NSP-C (208 amino acids). The deduced proteins, which were predicted to possess distinct amino-terminal regions, appeared to exhibit some architectural resemblance to known neuroendocrine proteins. In this paper the biochemical characterization and subcellular localization of the two proteins is addressed. In vitro translation of NSP-A and -C RNA produced proteins of about 135 and 23 kDa, respectively. Proteins of similar molecular mass were also detected in immunoprecipitation and western blot analyses of neural and endocrine cells using specific anti-NSP-A or -C antisera; some heterogeneity of NSP-A was observed. NSP-A, but not NSP-C, appeared to be highly phosphorylated and preferentially on serine residues. In immunocytochemical studies, we demonstrated that NSP-A and -C are associated with the endoplasmic reticulum; NSP-A was found to co-localize with SERCA2b, a membrane-associated Ca(2+)-ATPase of the endoplasmic reticulum. In Purkinje cells, we found NSP-immunostaining in the perikaryon, the extensive dendritic tree and the axon, also suggesting association with the smooth endoplasmic reticulum. Biochemical studies of NSP-A provided evidence that NSP-A is strongly associated with microsomal membranes and analysis of deletion mutants of NSP-A revealed that the hydrophobic carboxy-terminal portion of the protein, which is also present in NSP-C, is critical for membrane binding. Through database searches, finally, we found two different NSP-related sequences, one in a sequenced region of human chromosome 19, and the second in a human, pancreatic islet-derived partial cDNA, suggesting that the NSP gene is the prototype of a larger gene family. The results of our studies seem to indicate that the NSP-encoded proteins are novel, membrane-anchored components of the endoplasmic reticulum for which we propose the name reticulons.

Molecular analysis of expression in rat brain of NSP-A, a novel neuroendocrine-specific protein of the endoplasmic reticulum

Previous studies have established that the novel neuroendocrine-specific NSP gene encodes three carboxy-terminally overlapping proteins, NSP-A, NSP-B and NSP-C which are anchored to membranes of the endoplasmic reticulum. Here, we report results of studies in which expression of NSP-A in rat brain was investigated. Immunization of mice with a bacterial hybrid protein containing almost all NSP-A sequences led to the isolation of five monoclonal anti-NSP-A antibodies. The corresponding epitopes were found to be mapping to two regions unique to NSP-A. In Western blot analysis of rat cerebrum and cerebellum using these antibodies, proteins of about 145 kDa were detected. An immunohistochemical study of rat brain revealed the presence of NSP-A in many brain regions, particularly in cerebellar Purkinje cells, in neurons of the superior colliculus and of the pyriform and enthorhinal cortex, in fibers of the basal ganglia and several hippocampal regions including CA3 (stratum lucidum) and the dentate gyrus, in the induseum griseum and in the subcommissural organ, suggesting a role of NSP-A in many areas of the brain.

Spatial heterogeneity of caffeine- and inositol 1,4,5-trisphosphate-induced Ca2+ transients in isolated snail neurons

Inositol 1,4,5-trisphosphate- and caffeine-induced Ca2+ release was examined in neurons isolated from the mollusc Helix pomatia using Ca2+ indicator fura-2 and fluorescent digital-imaging microscopy technique. Extracellular application of caffeine caused a fast and pronounced augmentation of [Ca2+]i whose amplitude and kinetics differ in the centre of the cell and near its membrane. Mean values of caffeine-induced increase of [Ca2+]i were 0.97 +/- 0.11 microM at the periphery and 0.53 +/- 0.13 microM in the centre. The rates of rise and relaxation of caffeine-evoked [Ca2+]i transients were faster near the membrane. Pressure injection of inositol, 1,4,5-trisphosphate into the same neurons produced an abrupt and significant increase of [Ca2+]i in the centre (mean value of inositol 1,4,5-trisphosphate-induced elevation = 0.55 +/- 0.11 microM) while the response was smaller or even absent near the cellular membrane. Inositol 1,4,5-trisphosphate- and caffeine-induced Ca2+ transients did not affect each other. The data obtained indicate that in snail neurons these two calcium pools are not overlapping and at least some part of the caffeine-sensitive store is located close to the cellular membrane and that the inositol 1,4,5-trisphosphate-sensitive one is located in the centre of the cell.

The endoplasmic reticulum of Purkinje neuron body and dendrites: molecular identity and specializations for Ca2+ transport

Immunofluorescence and immunogold labeling, together with sucrose gradient separation and Western blot analysis of microsomal subfractions, were employed in parallel to probe the endoplasmic reticulum in the cell body and dendrites of rat cerebellar Purkinje neurons. Two markers, previously investigated in non-nerve cells, the membrane protein p91 (calnexin) and the lumenal protein BiP, were found to be highly expressed and widely distributed to the various endoplasmic reticulum sections of Purkinje neurons, from the cell body to dendrites and dendritic spines. An antibody (denominated anti-rough-surfaced endoplasmic reticulum), which recognized two membrane proteins, p14 and p40, revealed a similar immunogold labeling pattern. However, centrifugation results consistent with a widespread distribution were obtained for p14 only, while p40 was concentrated in the rough microsome-enriched subfractions. The areas enriched in the inositol 1,4,5-triphosphate receptor and thus presumably specialized in Ca2+ transport (stacks of multiple smooth-surfaced cisternae; the dendritic spine apparatus) also exhibited labeling for BiP and p91, and were positive for the anti-rough-surfaced endoplasmic reticulum antibody (presumably via the p14 antigen). Additional antibodies, that yielded inadequate immunocytochemical signals, were employed only by Western blotting of the microsomal subfractions, while the ryanodine receptor was studied by specific binding. The latter receptor and the Ca2+ ATPase, known in other species to be concentrated in Purkinje neurons, exhibited bimodal distributions with a peak in the light and another in the heavy subfractions. A similar distribution was also observed with another lumenal protein, protein disulfide isomerase. Taken as a whole, the results that we have obtained suggest the existence in the endoplasmic reticulum of Purkinje neurons of two levels of organization; the first identified by widespread, probably general markers (BiP, p91, possibly p14 and others), the second by specialization markers, such as the inositol 1,4,5-triphosphate receptor and, possibly, p40, which appear restricted to areas where specific functions appear to be localized.

Characteristics and function of Ca(2+)- and inositol 1,4,5-trisphosphate-releasable stores of Ca2+ in neurons

Molecular, biochemical and physiological evidence for the existence of releasable Ca2+ stores in neurons is strong. There are two separate molecules that function as release channels from those Ca2+ stores, the RyanR and InsP3R, and both have multiple regulatory sites for positive and negative control. Perhaps most intriguing is the biphasic, concentration-dependent action of cytosolic Ca2+ on both channels, first to stimulate release then, at higher concentration, to depress release. Whether the InsP3R and RyanR channels regulate Ca2+ release from different or identical functional compartments will need to be defined for each neuron type and perhaps even for each intracellular region within neurons since the evidence for functional separation of stores is mixed. The identification of Ca2+ storage and releasing capacity throughout all subcellular regions of neurons and the increasing evidence for a role for Ca2+ stores in neuronal plasticity suggests that the further characterization of the functional properties of Ca2+ stores will be an increasingly important and expanding area of interest in neurobiology.

Freeze-fracture organization of chromatin and cytoplasm in neurons and astroglia of rat cerebellar cortex

The cytology of the cell nucleus and cytoplasm of neurons and astroglia of the rat cerebellar cortex has been investigated by freeze-fracture electron microscopy. The main differential characteristics in the cytoplasm of the several cell types of the cerebellar cortex were: (1) the organization of endoplasmic reticulum elements, including special configurations of lamellar bodies and hypolemmal complexes, (2) the polarity, extension and arrangement of Golgi cisterns and associated tubulovesicular elements; (3) the connection pattern among different membrane-bounded cellular compartments; and (4) the architecture of endomembranes (i.e. presence of pits and fenestrations). In the nucleus, the main differential features were the the three-dimensional view of the nuclear envelope, the distribution of nuclear pores and the aggregation pattern of chromatin, visualized as clusters of nuclear particles in cross-fractures. The quantitative analysis of chromatin revealed four peaks of nuclear particle sizes (8, 12, 17 and 21 nm) that may correspond to variable degrees of coiling of the polynucleosomal chain in the chromatin fibre. Significant differences were observed in the proportion, numerical density and size distribution of aggregated nuclear particles in heterochromatin domains among the different cell types of the cerebellar cortex. The percentage of nuclear particles in aggregates varied from 10% in Purkinje cells to 64% in granule cells. Astrocytes and Bergmann glia showed intermediate values (about 40%). The percentage of nuclear particles in aggregates showed a significant (P less than 0.05) negative linear correlation with the nuclear volume, the number of pores per unit nuclear volume and the total number of pores per nucleus. In granule cells and astroglia, heterochromatin domains had a greater percentage of large nuclear particles (greater than 10 nm) than did euchromatin domains, whereas in interneurons, Purkinje and Golgi cells heterochromatin and euchromatin showed a similar proportion of large particles. Nuclear particles in euchromatin exhibited a similar pattern of distribution in all cerebellar cells.

Ultrastructural study of cholinergic neurons in the medial septal nucleus and vertical limb of the diagonal band of broca in the basal forebrain of the rat

The morphology, ultrastructure and synaptic relationships of the cholinergic and non-cholinergic neurons in the medial septal nucleus (MS) and vertical limb of the diagonal band of Broca (VDB) in the basal forebrain of the rat were studied at the light and electron microscopic levels. The cholinergic neurons were localized immunocytochemically using a monoclonal antibody against choline acetyltransferase (ChAT). Morphometric and statistical analyses showed that ChAT-labelled cells presented a predominantly oval morphology in both nuclei. The sizes of the neurons were significantly larger in the VDB nucleus. Within the two nuclei, two populations of cholinergic neurons were differentiated. One of the large immunolabelled neurons presented deep indentations and prominent nucleoli in their non-immunoreactive nuclei. Their cytoplasm contained a well-organized endomembrane system composed of short cisternae of rough endoplasmic reticulum (RER). One or two lamellar bodies with a peculiar ultrastructure were frequently found intercalated in this system. The Golgi areas presented numerous coated vesicles, sequestration and multivesicular bodies, which was indicative of an intense metabolic activity in these cells. The second population of small immunolabelled neurons exhibited reduced cytoplasm with a poorly developed endomembrane system and apparent absence of lamellar bodies. The neighbouring non-immunolabelled neurons presented a different type of organization of the endomembrane system which was composed of scattered and loosely arranged elongated cisternae of RER and infrequent lamellar bodies, with a structure different from that seen in the large cholinergic neurons. We propose that the structural differences in composition of the endomembrane system and lamellar bodies observed in the three types of neurons in this study indicate different metabolic activities. Symmetrical and asymmetrical synaptic contacts were observed on somata and dendrites of labelled neurons, the latter being more frequent. ChAT-labelled axon boutons were never seen. The absence of immunolabelled axon terminals and the presence of immunolabelled myelinated axons leads us to suggest that the majority of neurons in these areas are of the long projecting type.

Optimization of differential immunogold-silver and peroxidase labeling with maintenance of ultrastructure in brain sections before plastic embedding

The limited success of immunogold labeling for pre-embedding immunocytochemistry of neuronal antigens is largely attributed to poor penetration of large (5-20 nm) colloidal gold particles. We examined the applicability of using silver intensification of 1 nm colloidal gold particles non-covalently bound to goat anti-rabbit immunoglobulin (1) for single labeling of a rabbit antiserum against the catecholamine synthesizing enzyme, tyrosine hydroxylase (TH), and (2) for immunogold localization of rabbit anti-TH simultaneously with immunoperoxidase labeling of a mouse monoclonal antibody against the opiate peptide, leucine-enkephalin (LE). Vibratome sections were collected from acrolein fixed brains of adult rats. These sections were immunolabeled without use of freeze-thawing or other methods that enhance penetration, but damage ultrastructure. By light microscopy, incubations in the silver intensifier (Intense M, Janssen) for less than 10 min at room temperature resulted in a brownish-red reaction product for TH. This product was virtually indistinguishable from that seen using diaminobenzidine reaction for detection of peroxidase immunoreactivity. Longer incubations produced intense black silver deposits that were more clearly distinguishable from the brown immunoperoxidase labeling. However, by light microscopy, the gold particles seen by electron microscopy were most readily distinguished from peroxidase reaction product with shorter silver intensification periods. The smaller size of gold particles with shorter periods of silver intensification also facilitated evaluation of labeling with respect to subcellular organelles. Detection of the silver product did not appear to be appreciably changed by duration of post-fixation in osmium tetroxide. In dual-labeled sections, perikarya and terminals exhibiting immunogold-silver labeling for TH were distinct from those containing immunoperoxidase labeling for LE. These results (1) define the conditions needed for optimal immunogold-silver labeling of antigens while maintaining the ultrastructural morphology in brain, and (2) establish the necessity for controlled silver intensification for light or electron microscopic differentiation of immunogold-silver and peroxidase reaction products and for optimal subcellular resolution.

Morphometric evidence from C-synapses for phased Nissl body response in alpha-motoneurones retrogradely intoxicated with diphtheria toxin

Diphtheria toxin (DTX) kills cells by inactivating ribosomal translocation and when used to retrogradely intoxicate cat intercostal motoneurones produces marked morphological alterations in Nissl bodies, including those specifically sited postsynaptic to C-type axon terminals. Here, qualitative examinations of 'intoxicated' postsynaptic Nissl bodies reveal a progressive structural alteration marked by rER dilatation, rER lamellae fragmentation but retention of both the highly ordered multilamellate organization and ribosomal attachment until final stages of Nissl body dissolution. Morphometric results identified 3 broad phases to the postintoxication response which differed in the degree of rER cisternal dilation, and the numerical and spatial relationships between rER-lamellae, rER-bound ribosomes and rER-associated polyribosomes. These phases reflect the known molecular basis of diphtheritic toxicity and contrast with the fast developing Nissl body reaction associated with the neurotoxin ricin which also invokes ribosomal dysfunction and has been used to mimic certain features of motor neurone disease. The cytopathology of DTX and ricin are compared in the Discussion.

Transcytosis of macromolecules through the blood-brain barrier: a cell biological perspective and critical appraisal

A critical appraisal is presented of nearly two decades of research publications and review articles advocating the bidirectional transcytosis of fluid-phase molecules, most notably native horseradish peroxidase (HRP), through the normal and experimentally modified blood-brain barrier (BBB). Extracellular routes circumventing the BBB in normal and pathological states and artifact introduced in histological preparation of CNS tissue exposed to blood-borne peroxidase are emphasized. The potential for transcytosis of macromolecules entering the nonfenestrated cerebral endothelium by the processes of non-specific fluid phase endocytosis (e.g., HRP), adsorptive endocytosis (e.g., lectins) and receptor-mediated endocytosis (e.g., ligands) is analyzed in the context of the cellular secretory process and the complimentary events of endocytosis and exocytosis at the luminal and abluminal plasma membranes. Available data suggest that the cerebral endothelium is polarized with regard to endocytosis and the internalization of cell surface membrane; hence, the transcytosis of specific macromolecules through the BBB may be vectorial. If these data are correct, the blood-brain barrier is not absolute, whereas its counterpart, the brain-blood barrier, may be.

Neurons containing retrogradely transported Fluoro-Gold exhibit a variety of lysosomal profiles: a combined brightfield, fluorescence, and electron microscopic study

The advantages of axonally transported Fluoro-Gold as a retrograde fluorescent marker are numerous. The objective of the present study was to determine whether transported Fluoro-Gold is visible in either semi-thin sections for light microscopy or thin sections for electron microscopy. Rats received injections of Fluoro-Gold into either the striatum or thoracic spinal cord. After appropriate survival times, labelled neurons were observed with the fluorescence microscope in brain regions that are known to project to the injected areas. Sections that contained labelled cells were embedded in plastic and examined with a fluorescence microscope. Semi-thin sections of unosmicated tissue displayed high-resolution fluorescent labelling of somata and dendrites. In contrast, osmicated tissue did not fluoresce, but numerous dark granules were observed in the dendritic and perikaryal cytoplasm of labelled neurons in toluidine blue stained sections that were examined with brightfield optics. The unosmicated tissue did not display these granules, and this finding suggested that the granules are composed of membranes. Neurons in other brain regions that are known not to project to the injection sites did not contain these dark granules. Adjacent thin sections examined with the electron microscope displayed numerous electron-dense, lysosome-like organelles in the cytoplasm of labelled neurons. The electron density of these organelles was greater than that of lysosomes in unlabelled neurons. Three types of distinctive organelles were observed in these preparations: (1) relatively dense concentric lamellar bodies of various sizes; (2) heterogeneous or lipofuscin-like lysosomes; and (3) coarse grained lysosomes. Control sections and unlabelled neurons did not display these organelles. Therefore, these organelles appear to correlate with Fluoro-Gold localized within the somata and dendrites of retrogradely labelled neurons. It is not known if they are the Fluoro-Gold itself, or represent a physiological effect on membranes. The results of this study indicate that Fluoro-Gold may be useful for tract tracing at the electron microscopic level.

Lamellar bodies are markers of cholinergic neurons in ferret nucleus basalis

Lamellar bodies are composed of stacks of closely-packed, ribosome-free cisterns which are in continuity with the rough endoplasmic reticulum. In the ferret nucleus basalis stained for choline acetyltransferase it was shown, by correlating light with electron microscopy, that only the cholinergic cells there possess lamellar bodies. The significance of lamellar bodies in the cholinergic neurons of the nucleus basalis is not known, but these structures may reflect a peculiar aspect of the functioning of the cholinergic cells which will need to be investigated further.

Concanavalin A binding sites of rough endoplasmic reticulum containing intracisternal microtubules in canine neurones

Lectin histochemistry was used to identify sugar residues of IM-containing RER in elderly canine sympathetic ganglionic neurones. IM-inclusions stained with ConA-peroxidase conjugate, but not with soybean agglutinin (SBA), wheat germ agglutinin (WGA), peanut agglutinin (PNA), Dolichos biflorus agglutinin (DBA), Ricinus communis agglutinin (RCA I) and Ulex europaeus agglutinin (UEA I). ConA-binding sites were visualized within cisternae of RER containing IM; reaction product was localized in IM-containing RER cisternae and IM. Inhibition with specific sugars (0.1 M alpha-methyl-D-mannoside or 0.5 M D-glucose) blocked the binding of ConA to IM-inclusions and normal Nissl substance. When a low sugar concentration (5 x 10(-3) M alpha-methyl-D-mannoside or 0.2 M D-glucose) was employed, IM-inclusions were still strongly ConA-positive, but normal Nissl substance was not. These results demonstrate that IM-containing RER have an excessive amount of carbohydrates (mannose or glucose-rich sugars) which are essentially detected in flattened RER under normal conditions and further indicate that glycoproteins in IM differ from those in cytoplasmic microtubules.

A histochemical study of the regional distribution in the rat brain of enzymatic activity hydrolyzing glucose- and 2-deoxyglucose-6-phosphate

A modified Wachstein-Meisel lead salt method using glucose-6-phosphate or 2-deoxyglucose-6-phosphate as substrates was employed at the light microscopic level to map the rat brain for glucose-6-phosphatase (G-6-Pase). As has been described, most of the activity of the enzyme resided in neuronal cell bodies and dendritic stems. No differences were found between the results obtained with the two substrates. Two categories of brain structures with heavy and with moderate staining could be distinguished while the majority of brain regions contained only barely discernible neurons. Structures displaying very high enzyme activity included nuclei of cranial nerves, nuclei of the reticular formation, Purkinje cells, and some parts of the limbic system, e.g., CA 3 and CA 4 pyramidal fields of the hippocampus. It is pointed out that accurate biochemical determinations of G-6-Pase activity will critically depend on painstaking microdissection of nuclei and cell layers. The histochemical results may be pertinent to the interpretation of the 2-deoxyglucose method for assessment of regional glucose utilization rates in brain. The present observations make it unlikely that regional variations in G-6-Pase activity account for differences in uptake and retention of radioactivity from (1-14C)glucose and (14C)2-deoxyglucose reported previously by our group.

Feline 'C'-type terminals possess synaptic sites associated with a hypolemmal cistern and Nissl body

Previous findings suggest that axosomatic 'C'-type terminals on rat spinal motoneurones possess 'active' synaptic sites associated with the characteristic subsynaptic cistern and postsynaptic Nissl body, but 'C'-terminals in cat do not. A re-examination of feline 'C'-terminals undertook to compare the synaptic ultrastructure in aldehyde-fixed material stained by either osmium or ethanolic phosphotungstic acid (E-PTA). In agreement with previous findings osmicated cat 'C'-terminals failed to reveal synaptic complexes in regions possessing the subsynaptic cistern ('cisternal regions'). In contrast, cisternal regions of E-PTA stained 'C'-terminals exhibited a linear array of presynaptic dense projection all opposed by a single extended postsynaptic density which abutted directly onto both postsynaptic membrane and the cistern. This close topographical relationship is suggested to be functionally significant in the trans-synaptic trophism previously demonstrated for feline 'C'-terminals.

Tubular profiles do not form transendothelial channels through the blood-brain barrier

The contribution of tubular profiles within the mammalian cerebral endothelium to the formation of transcellular channels was analysed following exposure of the endothelium to native horseradish peroxidase (HRP) dissolved in saline or dimethyl sulphoxide (DMSO) administered intravenously in mice. Within 5-15 min, but not at 30 min to 2 h postinjection, peroxidase-positive extravasations were evident within the parenchyma of the forebrain and brainstem of mice exposed and not exposed to DMSO. The extravasations may be associated with the rupture of interendothelial tight junctions at the level of arterioles as a consequence of the perfusion-fixation process. Ultrastructural inspection of endothelia within and away from areas of peroxidase extravasation revealed the following intraendothelial, peroxidase-positive organelles: presumptive endocytic vesicles, endosomes (a prelysosomal compartment), multivesicular and dense bodies, and tubular profiles. Statistical analysis of the concentration of HRP-labelled presumptive endocytic vesicles, which may coalesce to form tubules, within endothelia from mice injected intravenously with HRP-DMSO compared to mice receiving HRP-saline revealed no significant difference. HRP-positive tubular profiles were blunt-ended, variable in length and width, and appeared free in the cytoplasm or in continuity with dense bodies. Labelled tubules free in the cytoplasm were positioned parallel to the luminal and abluminal plasma membranes and were less frequently oblique or perpendicular to these membranes. Tubular profiles analysed in serial thin sections or with a goniometer tilt stage did not establish membrane continuities with the luminal and abluminal plasma membranes. Peroxidase-positive tubular profiles were similar morphologically to those exhibiting acid hydrolase activity but did not share morphological and enzyme cytochemical similarities with the endoplasmic reticulum that stained for glucose-6-phosphatase (G6Pase) activity. G6Pase-positive profiles of endoplasmic reticulum were not observed to contribute to a transendothelial canalicular network. Our results suggest that: (i) peroxidase-labelled tubules, acid hydrolase-positive tubules, and G6Pase-positive endoplasmic reticulum do not form transcellular channels through the cerebral endothelium; (ii) tubular profiles labelled with blood-borne HRP in the cerebral endothelium are associated with the endosome apparatus and/or the lysosomal system of organelles; and (iii) DMSO does not appear to alter the permeability of the blood-brain barrier to blood-borne protein.

Alzheimer's disease: paired helical filaments and cytomembranes

Observations were made with the electron microscope on biopsies from the frontal cerebral cortex of four patients showing the clinical symptoms of advanced Alzheimer's disease (AD). Sections from all four biopsies showed neuronal dendrites, and to a lesser extent, perikarya, packed with paired helical filaments (PHFs) and a correlated complete loss of the normal content of microtubules. It was usually impossible to visualize the beginning and end of the PHF since it passed out of the plane of section. However, not infrequently, PHFs could be seen apparently arising at or from the surfaces of cytomembranes. Such membranes (in perikarya or dendrite) often formed irregular stacks or lamellated bodies. These membranes were invariably agranular (i.e. not studded with ribosomes) and so might be considered as pathological forms of smooth endoplasmic reticulum (SER). A correlation is established between the known biochemical nature of PHFs and their postulated membrane origin.

Ultracytochemical evidence for the presence of GERL in pinealocytes of the Mongolian gerbil (Meriones unguiculatus)

Ultracytochemical reactions for the demonstration of acid phosphatase, glucose-6-phosphatase and thiamine pyrophosphatase, as well as zinc iodide-osmium tetroxide impregnation, revealed the existence of GERL (Golgi apparatus-Endoplasmic Reticulum-Lysosomes) in pinealocytes of the Mongolian gerbil (Meriones unguiculatus). The spatial arrangement of this structure was studied on thick sections using a goniometric stage. Although it was not possible to determine whether GERL in pinealocytes belongs to the Golgi apparatus or to endoplasmic reticulum, it can be concluded that its presence in studied cells signifies that they are considerably more active synthetically than has been believed to date.

Ultrastructural localization of concanavalin A-binding sites in the Golgi apparatus of various types of neurons in rat dorsal root ganglia: functional implications

The localization of concanavalin A (con A) binding sites has been determined at the electron-microscopic level in the six types of neurons (A1, A2, A3, B1, B2, C) of rat dorsal root ganglia. In all ganglion cells, con A stained the plasma membrane, the nuclear envelope, the cisternae of the rough endoplasmic reticulum, and the matrix of some multivesicular bodies. In contrast, the con A reactivity of the Golgi apparatus varied according to cell type. In type B1 and B2 cells and possibly in type A3 cells, the lectin was exclusively located in three or four saccules on the cis side of the Golgi stacks, whereas the TPPase-positive saccules and the trans sacculotubular elements were unstained with con A. In type A1, A2, and C neurons, all Golgi saccules as well as the trans sacculotubular elements were stained with the lectin. These results suggest that different types of glycoproteins were produced in these two groups of neurons. In the type A1, A2, and C cells, the persistence of the lectin reactivity in the TTPase-positive saccules or sacculotubular elements on the trans side of the Golgi stacks suggests the presence of glycoproteins with oligosaccharide side chains rich in alpha-D-mannosyl residues in terminal positions. In contrast, the disappearance of the con A reactivity in equivalent elements of the Golgi stacks in type B1, B2, and A3 cells suggests the addition at this level of other sugar residues characteristic of complex oligosaccharide side chains. The majority of the vesicular elements associated with the Golgi apparatus, as well as lysosomes, were unstained with con A.

Cytochemical identification of cerebral glycogen and glucose-6-phosphatase activity under normal and experimental conditions. II. Choroid plexus and ependymal epithelia, endothelia and pericytes

Intracellular glycogen and glucose-6-phosphatase (G6Pase) activity were identified cytochemically within epithelia of the choroid plexus and ependyma of the cerebral ventricles including the median eminence and area postrema, the cerebral endothelium and pericytes from control, salt-stressed and fasted adult mice. Identification of glycogen was obtained by employing osmium tetroxide-potassium ferrocyanide and the periodic acid-thiocarbohydrazide-silver protein technique as ultrastructural contrast stains. A lead-capture method was used to localize G6Pase activity with glucose-6-phosphate or mannose-6-phosphate as substrate. Cerebral G6Pase functions predominantly as a phosphohydrolase to convert glucose-6-phosphate to glucose. Some glucose-6-phosphate in vivo may be derived from the breakdown of glycogen stores. Within the sampled cell types, presumptive glycogen appeared as electron-dense, isodiametric particles scattered throughout the cytoplasm. Reaction product for G6Pase activity was localized consistently within the lumen of the nuclear envelope and endoplasmic reticulum and frequently within an outer saccule of the Golgi complex under normal conditions. Choroid plexus epithelia from stressed mice exhibited a qualitative increase in cytoplasmic glycogen and a decrease in G6Pase activity; the other cell types did not express demonstrable alterations in glycogen concentration and G6Pase activity. The results indicate that glycogen and G6Pase activity are prevalent within non-neural cells of the adult mammalian CNS. Glucose utilization in the choroid plexus epithelium may be altered by stressful conditions that influence the functional activity of this cell.

Histochemical localization of acetylcholinesterase in relation to motor neurons and internuclear neurons of the cat abducens nucleus

Three distinct patterns of AChE localization have been observed in relation to cat abducens motor neurons and internuclear neurons labelled by retrograde transport of horseradish peroxidase. First, AChE was localized predominantly within cisternae of granular endoplasmic reticulum and agranular reticulum of motor neuron somata, dendrites and axons, but was absent from internuclear neurons. AChE was also associated with saccules of the Golgi apparatus in the motor neurons, but was was absent from all other cytoplasmic organelles. Second, AChE was observed on the soma-dendritic and axonal surface membrane of the motor neurons, particularly at sites of apposition of synaptic endings of all morphological types, but was usually absent from the surface membranes of internuclear neurons. Third, AChE was associated both extracellularly and intracellularly with certain synaptic endings that contained spheroidal synaptic vesicles and that contacted both motor neurons and internuclear neurons. A similar pattern of staining of synaptic endings was observed at the neuromuscular junctions in the lateral rectus muscle. Axotomy of the VIth nerve resulted in loss of intracellular AChE associated with the Golgi apparatus and extracellular AChE on the somatic surface membrane of the motor neurons. The patterned localization of AChE contrasted with the localization of butyrylcholinesterase, which was associated predominantly with astrocytes. The findings suggest different roles of AChE as a function of the different patterns of localization.

Apparent absence of a translocase in the cerebral glucose-6-phosphatase system

In the hepatocyte endoplasmic reticulum, a substrate transporter could provide a means of regulating hydrolysis of glucose-6-phosphate by specifically modulating access of the substrate to the hydrolase. Several characteristics of the cerebral microsomal enzyme suggest that such an hypothesis is untenable in the brain. These are: (a) the inability of the enzyme in either untreated or detergent-disrupted brain microsomes to distinguish between glucose-6-phosphate and mannose-6-phosphate; (b) the close agreement of the apparent Km values for either substrate in intact or disrupted microsomal preparations; (c) the constancy of the latency toward both substrates over a wide concentration range; (d) the inability of nonpenetrating, covalently-linking reagents [e.g., 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid (DIDS)] to affect the accessibility of the hydrolase to its substrate; (e) the absence of a putative transporter polypeptide, such as that of the liver, in experiments where tritiated H2DIDS, polyacrylamide gel electrophoresis, and radioautography are applied to brain microsomes.

High resolution autoradiographic determination of the topographic distribution of radioactivity in the hippocampal formation after injection of [1-14C]glucose or 2-deoxy[14C]glucose

Using high resolution autoradiography, the accumulation of radioactivity after intravenous injection of [1-14C]glucose was measured in the corpus callosum, hippocampus, dorsal hippocampal commissure, somatosensory cortex, inferior colliculus and pontine periaqueductal grey. Autoradiograms were prepared by thaw-mounting 4 micron frozen sections on nuclear emulsion-coated slides, and were evaluated quantitatively with a computer-assisted video system for automated counting of silver grains. In all brain regions examined, silver grain densities were greater in rats killed 30 min after injection of [1-14C]glucose compared to rats killed 10 min after injection. After intravenous injection of [1-14C]glucose or 2-deoxy[14C]glucose, the relative uptake and retention of radioactivity in different hippocampal subregions was compared. Striking differences were found in the hippocampus between 2-deoxy[14C]glucose and [1-14C]glucose autoradiograms. After injection of 2-deoxy[14C]glucose, there were large variations in the uptake and retention of radioactivity among different pyramidal cell fields. The CA 3 pyramidal cell field retained considerably more radioactivity than other pyramidal cell fields after injection of 2-deoxy[14C]glucose, while after injection of [1-14C]glucose, the retention of radioactivity was similar in all pyramidal cell fields. After [1-14C]glucose injection, the dentate gyrus contained relatively high levels of radioactivity and more 14C accumulated in the granular layer, compared to the molecular layer. In contrast, after 2-deoxy[14C]glucose injection, there was uniformly less radioactivity throughout the dentate gyrus when compared to rats injected with [1-14C]glucose and there was no preferential accumulation of 2-deoxy[14C]glucose in the granular layer compared to the molecular layer.(ABSTRACT TRUNCATED AT 250 WORDS)

Acetylcholinesterase activity and type C synapses in the hypoglossal, facial and spinal-cord motor nuclei of rats. An electron-microscope study

Using the electron-microscope technique of Lewis and Shute, we studied the localization of the acetylcholinesterase (AChE) activity in the hypoglossal, facial and spinal-cord motor nuclei of rats. The technique used selectively detects synapses with subsynaptic cisterns (type C synapses) as well as heavy deposits of reaction products in the rough endoplasmic reticulum, in fragments of the nuclear envelope, in some Golgi zones and on parts of the pericaryal plasma membrane, the axolemma and the dendritic membrane. In C synapses, AChE activity was located in the synaptic cleft and on the membrane of presynaptic boutons. Some C synapses exhibited distinct synaptic specialization in the form of multiple 'active zones'. These zones were characterized by dense presynaptic projections, short dilations of the synaptic cleft, and postsynaptic densities localized between the postsynaptic membrane and the outer membrane of the subsynaptic cistern. Within the postsynaptic densities, rows of rod- or channel-like structures were observed. The subsynaptic cisterns were continuous with the positive rough endoplasmic reticulum. The results are discussed in terms of the possible role of C synapses in the regulation of AChE synthesis in postsynaptic cholinergic neurons and/or in the regulation of AChE release into the extracellular space as well as in the establishment of new synaptic contacts.

Endocytic and exocytic pathways of the neuronal secretory process and trans-synaptic transfer of wheat germ agglutinin-horseradish peroxidase in vivo

The lectin wheat germ agglutinin (WGA) conjugated to horseradish peroxidase (HRP) was employed to study the endocytic and exocytic pathways of the secretory process in neurons and the potential for trans-synaptic transfer of molecules within the CNS. WGA-HRP binds to surface membrane oligosaccharides and enters cells by adsorptive endocytosis. The lectin conjugate was administered intranasally or into the cerebral ventricles of mice; postinjection survival times ranged from 5 minutes to 6 days. Due to binding of the lectin to ependymal cells subsequent to an intraventricular injection, only select populations of neurons (i.e., hippocampal formation; paraventricular nuclei; midbrain raphe; VI, X, XII motor nuclei; among others) were exposed extracellularly to WGA-HRP and became labeled by retrograde axoplasmic transport from axon terminals or by direct cell body/dendritic uptake. WGA-HRP delivered intranasally was endocytosed by first-order olfactory neurons and transported by anterograde axoplasmic flow to the terminal field within the glomerular layer of the main olfactory bulb; eventually perikarya of the mitral cell layer were labeled, presumably by anterograde trans-synaptic transfer of the lectin conjugate. In the variety of neurons analyzed ultrastructurally following exposure to WGA-HRP, the proposed sequence of intracellular pathways through which peroxidase reaction product was traced over time was: cell surface membrane----endocytic structures----endosomes (presecondary lysosomes)----transfer vesicles----transmost Golgi saccule----vesicles, vacuoles, and/or dense core granules. WGA-HRP also labeled vesicles and tubules that were channeled to and/or derived from spherical endosomes, dense bodies, and multivesicular bodies. The peroxidase-positive, membrane-delimited products of the trans Golgi saccule contributed to anterograde axonal transport vectors and accumulated within axon terminals. A second contribution to these vectors was provided by peroxidase-labeled tubules and dense bodies believed to represent components of the lysosomal compartment. Profiles of the axonal reticulum comparable to those that stained cytochemically for glucose-6-phosphatase activity, a marker for the endoplasmic reticulum, were not associated with the transport of WGA-HRP. Trans-synaptic transfer of WGA-HRP from primary olfactory neurons to postsynaptic cells in the olfactory bulb was reflected in peroxidase-positive endocytic vesicles, endosomes, dense bodies, and the trans Golgi saccule.(ABSTRACT TRUNCATED AT 400 WORDS)

The neuronal endoplasmic reticulum: its cytochemistry and contribution to the endomembrane system. II. Axons and terminals

The morphology and cytochemistry of the endoplasmic reticulum (ER) in axons and terminals of a number of different types of neurons in brains from mice were investigated ultrastructurally. The neurohypophysis received particular attention because the morphology and enzyme cytochemical activities of many of the preterminal swellings of hypothalamo-neurohypophysial axons are altered by chronic salt-stress. Membrane contrast and enzyme cytochemical staining techniques were employed to characterize the axonal reticulum and to determine if organelles representing the lysosomal system in the axon and the tubular profiles participating in the anterograde axonal transport of native horseradish peroxidase (HRP) are associated with the ER. Potential enzyme cytochemical markers for the axonal ER included glucose-6-phosphatase (G6Pase), thiamine pyrophosphatase, nucleoside diphosphatase, and acid hydroxylase activities. The anterograde transport of HRP was analyzed in undamaged hypothalamo-neurohypophysial neurons and in facial and hypoglossal motoneurons of mice receiving the protein in the lateral cerebral ventricle. The ER pervaded the axon and appeared as parallel, 20-40-nm-wide tubules interconnected by oblique anastomoses. Membrane thickness of the axonal reticulum measured 60-100 A, which is similar to that of the perikaryal ER. Enzyme cytochemical activities associated with the ER or lysosomes were not conspicuous in axons and terminals under normal conditions but became prominent in some axons and preterminal swellings manifesting an autophagic appearance within neurohypophyses from salt-stressed mice. Only G6Pase activity was a marker for the ER in these axons and preterminals. Many ER profiles in non-incubated sections and in G6Pase cytochemical preparations of salt-stressed neurohypophyses were wrapped around or interspersed among secretory granules, multilamellar bodies, and vacuoles that may represent forms of lysosomes involved in autophagy and crinophagy. Acid hydrolase activities were localized within the vacuoles as well as within 80-130-nm-wide, blunt-ended tubules in pituitary stalk axons; similar reactive tubules were confluent with large secondary lysosomes in neurosecretory cell bodies and may be derived from these lysosomes. Morphologically identical tubules transporting HRP in the anterograde direction were observed only in the salt-stressed hypothalamo-neurohypophysial neuron. The HRP-positive tubules very likely are affiliated with the lysosomal system.

Ultrastructural localization of acetylcholinesterase in substantia nigra: a comparison between rat and guinea pig

The distribution and ultrastructural localization of acetylcholinesterase (AChE) was examined in the substantia nigra of rat and guinea pig. Although the pars compacta, in both species, is clearly defined when stained with thionin, there is an apparent discrepancy in the distribution of AChE at the light microscope level. In the rat substantia nigra the enzyme appears to be concentrated mainly in the pars compacta, whereas in the guinea pig the AChE seems homogeneous throughout the entire substantia nigra. Ultrastructural studies, however, reveal a close correspondence in the morphology of rat and guinea pig substantia nigra. The subcellular localization of AChE is also similar. The between-species discrepancy seen with the light microscope was attributed to relative differences in level of intensity of staining for AChE. In both rat and guinea pig, the enzyme is localized in two types of neuron and in the extracellular space. The ultrastructural distribution of AChE is discussed with reference to neurochemical studies on its release from nigral neurons.

Further studies of the secretory process in hypothalamo-neurohypophysial neurons: an analysis using immunocytochemistry, wheat germ agglutinin-peroxidase, and native peroxidase

The axonal endoplasmic reticulum (ER) and synaptic-like (micro)vesicles within axon terminals of the neurohypophysis and their contribution to the secretory process in hypothalamo-neurohypophysial neurons have been investigated cytochemically in normal mice and in mice given 2% salt water to drink for stimulation of hormone synthesis in and release from these neurons. Cytochemical techniques included the peroxidase-antiperoxidase (PAP) immunocytochemical method for localization of neurophysin, wheat germ agglutinin-horseradish peroxidase (WGA-HRP) as a tracer for the anterograde axonal transport of membrane from within the perikaryon, and blood-borne native horseradish peroxidase (HRP) as a tracer for internalized axon terminal membrane. The primary antiserum employed was directed against neurophysins I and II, the carrier proteins for the peptide hormones oxytocin and vasopressin, respectively. PAP reaction product was observed over neurosecretory granules but never over the endoplasmic reticulum, microvesicles or other organelles in axons and terminals of the neurohypophysis. WGA-HRP was delivered extracellularly to cell bodies of paraventricular neurons by cerebral ventriculocisternal perfusion. Internalized perikaryal surface membrane tagged with WGA-HRP was recycled through the innermost Golgi saccule (GERL) from which neurosecretory granules were formed. The anterograde axonal transport of membrane-bound WGA-HRP was manifested within the neurosecretory granules; WGA-HRP did not label the axonal reticulum or terminal microvesicles in the neurohypophysis. Blood-borne native HRP endocytosed into neurohypophysial terminals was associated with a plethora of microvesicles measuring 40-70 nm in diameter and vacuoles similar in size to the 100-300-nm-wide neurosecretory granules. The microvesicles contributed to the formation of numerous vacuoles. The internalization of axon terminal membrane as microvesicles incorporating HRP was quantitatively greater than vacuoles in both salt-stressed and control mice. The results suggest that in the hypothalamo-neurohypophysial system of the mouse the axonal ER and terminal microvesicles are not involved in the transport, storage, and exocytosis of neurosecretory material and perhaps other molecules processed through the innermost Golgi saccule. Nevertheless, a prominent population of the microvesicles within axon terminals of the neurohypophysis does participate in the secretory process. These vesicles are involved directly in the internalization of the terminal surface membrane subsequent to release of secretory granule content.(ABSTRACT TRUNCATED AT 400 WORDS)

Isolation and partial characterisation of neuronal growth cones from neonatal rat forebrain

We have devised a method for the isolation of viable neuronal growth cones from neonatal rat forebrain. The method involves differential and density gradient centrifugation and exploits the relatively low buoyant density (approximately 1.018 g/cm3) of growth cones. There are no known biochemical markers for growth cones and it was necessary therefore to monitor for their presence during the isolation using transmission electron microscopy. Several criteria were used to identify isolated growth cones including the presence of filopodia, an extensive system of branching, tubular smooth endoplasmic reticulum and a region rich in microfilaments subjacent to the plasma membrane. These morphological features are similar to those of growth cones identified unequivocally in intact developing brain and in tissue culture. Electron microscopical analysis showed that greater than 90% of membrane-bound, identifiable objects in one fraction were growth cones by these criteria. The major contaminant consisted of membrane sacs and vesicles of unidentified origin. There were only small amounts of isolated rough endoplasmic reticulum and mitochondria. Isolated growth cones were roughly spherical in shape with a diameter of 1.9 +/- 0.5 micron (mean +/- 1 SD). They usually contained mitochondria, large granular vesicles and small vesicles, and occasionally contained coated vesicles, lysosomes, lamellar bodies and multivesicular bodies, and only very rarely, intermediate filaments. Occasionally, growth cones had rudimentary synapses on them. The viability of isolated growth cones was investigated by observing their behaviour in short-term culture. After a few hours in culture on poly-D-lysine-coated coverslips, growth cones flattened down and extended filopodia-like processes. This behaviour was inhibited by cytochalasin B and reversibly by cold (4 degrees C). We conclude that physiologically active growth cones can be isolated rapidly and in large numbers by the method described here.

Effect of microtubule assembly status on the intracellular processing and surface expression of an integral protein of the plasma membrane

We studied the effects of changes in microtubule assembly status upon the intracellular transport of an integral membrane protein from the rough endoplasmic reticulum to the plasma membrane. The protein was the G glycoprotein of vesicular stomatitis virus in cells infected with the Orsay-45 temperature-sensitive mutant of the virus; the synchronous intracellular transport of the G protein could be initiated by a temperature shift-down protocol. The intracellular and surface-expressed G protein were separately detected and localized in the same cells at different times after the temperature shift, by double-immunofluorescence microscopic measurements, and the extent of sialylation of the G protein at different times was quantitated by immunoprecipitation and SDS PAGE of [35S]methionine-labeled cell extracts. Neither complete disassembly of the cytoplasmic microtubules by nocodazole treatment, nor the radical reorganization of microtubules upon taxol treatment, led to any perceptible changes in the rate or extent of G protein sialylation, nor to any marked changes in the rate or extent of surface appearance of the G protein. However, whereas in control cells the surface expression of G was polarized, at membrane regions in juxtaposition to the perinuclear compact Golgi apparatus, in cells with disassembled microtubules the surface expression of the G protein was uniform, corresponding to the intracellular dispersal of the elements of the Golgi apparatus. The mechanisms of transfer of integral proteins from the rough endoplasmic reticulum to the Golgi apparatus, and from the Golgi apparatus to the plasma membrane, are discussed in the light of these observations, and compared with earlier studies of the intracellular transport of secretory proteins.