Endosomes and Golgi vesicles in adsorptive and fluid phase endocytosis.

We studied with morphometric methods the endocytosis by pheochromocytoma cells of a conjugate of wheat germ agglutinin with ferritin (WGA-Ft) and of horseradish peroxidase (HRP). Quantitative studies indicated that WGA-Ft was cleared slowly from cell surfaces and that it was not recycled to the surface. Cells labeled with WGA-Ft for 15 min at room temperature were washed and incubated in medium containing HRP for 15 or 30 min at 37 degrees C. The greatest proportion of labeled vesicles and tubules contained only WGA-Ft (83.4% at 15 min and 85.3% at 30 min). A very small fraction of labeled vesicles and tubules contained only HRP (0.2% at 15 min and 0.9% at 30 min). Vesicles and tubules at the Golgi apparatus were labeled almost exclusively with WGA-Ft (97% at 15 min and 30 min); the rest had both labels. Most labeled lysosomes contained both labels (80.1% at 15 min and 80.8% at 30 min). Of the remainder more contained WGA-Ft alone (20% at 15 min and 10.9% at 30 min), then HRP alone (none at 15 min and 8.2% at 30 min). In contrast to the various and varying patterns of labeling with WGA-Ft and HRP of the other organelles studied, the vast majority of endosomes contained both markers (94.1% at 15 min and 100% at 30 min); the rest contained WGA-Ft only. These results demonstrate that endosomes are recipients of both fluid phase and adsorptive endocytosis markers; these findings are consistent with the hypothesis that endosomes mediate the sorting out and subsequent intracellular traffic of membrane bound and fluid phase markers. Cisterns of the Golgi apparatus did not contain WGA-Ft; in sharp contrast, when WGA-HRP was used, the cisterns of the Golgi apparatus consistently contained HRP.

Plasma membrane and internalized immunoglobulins of lymph node cells studied with conjugates of antibody or its Fab fragments with horseradish peroxidase.

Normal rat and mouse lymphoid cells were incubated at 0 degrees -4 degrees C for 1 h with purified rabbit or sheep antirat (mouse) immunoglobulin (Ig)-horseradish peroxidase (PO) conjugates or with Fab fragments of antibody coupled with peroxidase. Cells were subsequently washed and incubated in fresh medium, without labeled antibody or Fab fragments for 5-30 min at 20 degrees or 37 degrees C. With the use of the diaminobenzidine (DAB) method, distribution of peroxidase was studied in the light and electron microscopes. Fab fragments of antirat Ig antibody were iodinated with (125)I and subsequently coupled with horseradish PO. Plasma membrane and internalized immunoglobulins were detected by electron microscope autoradiography and peroxidase cytochemistry. Single- (Fab-PO), and double- ([(125)I]Fab-PO) labeled lymphoid cells showed identical patterns of surface or internal distribution of immunoglobulins. In the electron microscope, Fab-PO conjugates at 0 degrees -4 degrees C resulted in a diffuse specific staining of the plasmalemma of lymphocytes and plasma cells. Most of the small dark lymphocytes (T cells?) did not show plasma membrane Ig. Macrophages did not show plasmalemma staining, but displayed nonspecific cytoplasmic staining after incubation at 20 degrees or 37 degrees C with antibody or Fab-PO conjugates. Lymphocytes and plasma cells, after incubation with antibody-PO conjugates at 0 degrees -4 degrees C, had patchy deposits of oxidized DAB on their plasma membranes. Macrophages, similarly treated, had no plasmalemmal staining. Patch and cap formation on the plasma membrane of lymphocytes and plasma cells was seen regularly after antibody-PO incubation at 37 degrees C. Internalization patterns were different in lymphocytes and plasma cells. In lymphocytes, peroxidase staining was observed in small round or oval vesicles clustered at one pole of the cell (30 min at 37 degrees C). In plasma cells, peroxidase staining was seen in clusters of tubules resembling the Golgi apparatus. Internalization of plasma membrane IgG was less pronounced after antibody-PO labeling as compared to Fab-PO labeling.

Quantitative ultrastructural autoradiographic studies of iodinated plasma membranes of lymphocytes during segregation and internalization of surface immunoglobulins.

Rat spleen lymphocytes were iodinated (125 I) with lactoperoxidase. Quantitative autoradiographic studies on cells fixed immediately after iodination showed 19-24% of intracytoplasmic grains at 3HD and over from the plasma membrane. Normalization of grain density distribution and comparison of resulting curves with the universal curve of grain scatter of 125 I showed that a significant percentage of intracytoplasmic grains (36%) originates from intracytoplasmic labeled sources rather than from scattering from the heavily labeled plasma membrane. Damaged cells had a threefold grain density than intact cells. Radioactivity counts in sliced polyacrylamide gels of iodinated cells revealed 65-72% of total radioactivity in five peaks of apparent mol wt of 44, 50, 57, 90 and 195 thousand daltons. Segregation and internalization of anti-immunoglobulin-Ig-horseradish peroxidase (HRP) complexes from the iodinated plasma membrane proteins of lymphocytes was studied with quantitative autoradiography (125 I) and peroxidase cytochemistry; 64% of grains at 1.5HD (1,500 A) from the plasma membrane were within the cap zone, and 36% of grains remained outside the capped immunoglobulins; 45-57% of grains internalized together with Fab-anti-Ig-Ig-HRP, and 68% of grains internalized together with anti-Ig-Ig-HRP. These studies indicate that (a) iodination of rat spleen lymphocytes results in a significant internal labeling and that (b) immunoglobulins segregate into caps and internalize together with other iodinated plasma membrane proteins while a significant percentage of iodinated proteins (36%) are excluded from the immunoglobulin caps or internalization sites (32-55%).

Superior sensitivity of conjugates of horseradish peroxidase with wheat germ agglutinin for studies of retrograde axonal transport.

We have compared the retrograde axonal transport of horseradish peroxidase (HRP), to the retrograde transport of HRP conjugated with wheat germ agglutinin (WGA). Morphometric studies have shown that WGA-HRP conjugates were 40 times more sensitive than free HRP, in the tracing of retrograde connections from the rat submandibular gland to the superior cervical ganglion. Also, WGA-HRP was more sensitive than free HRP in the tracing of retrograde connections from the rat tongue to the hypoglossal nucleus. Our findings with WGA-HRP are consistent with the observations by Schwab et al. who reported (-125I) WGA is a highly sensitive retrograde tracer (Brain Research 152:145, 1978 (22)).

Detection of plasma cell immunoglobulins in tissue sections optimally fixed for ultrastructural immunocytochemistry.

We describe the ultrastructural localization of plasma cell immunoglobulins in vibratome sections of popliteal lymph nodes. Fixation with glutaraldehyde-paraformaldehyde gave better tissue and antigen preservation than paraformaldehyde or periodic acid lysine-paraformaldehyde; biotinylated Fab fragments of sheep anti-mouse IgG-streptavidin-biotinylated horseradish peroxidase (HRP) or Fab-HRP conjugates gave similar results. With both immunoreagents, excellent tissue preservation and antigen detection was observed in the first layer of cells sectioned with the vibratome. Conjugates of anti-mouse IgG with HRP did not show any staining. Peroxidase stain was observed in the nuclear envelope, cisternae of the rough endoplasmic reticulum, and the Golgi apparatus complex. In the Golgi apparatus, staining was seen consistently in cisternae of the cis face and in adjacent vesicles; the trans cisternae showed weak or no stain, and adjacent vesicles, "coated" vesicles, and granules were not stained. This study shows that high quality of tissue preservation and antigen detection, by both light and ultrastructural immunocytochemistry, is feasible in tissue fixed with glutaraldehyde-paraformaldehyde followed by vibratome sectioning and immunostaining with Fab-biotin-streptavidin-biotin-HRP, or Fab-HRP.

Identification of a novel protein (G210) specific to the Golgi apparatus.

A monoclonal antibody, 3C9, has enabled the detection of a novel Golgi-specific protein in bovine tissues. Immunohistochemical studies at the light microscopic level have detected the 3C9 antigen only in certain cells: exocrine pancreas, gut epithelium, and thymus epithelium. Examination of gut and pancreas by immunoelectron microscopy showed a localization exclusive to the Golgi apparatus. The relative molecular weight of the antigen detected by immunoblotting is 210,000 daltons. The antigen is not extracted from microsomal membranes of bovine gut epithelium by sodium carbonate solutions. Furthermore, the 3C9 antigen enters into the detergent phase when Triton X-114 partitioning methods are used. These data strongly suggest that this novel antigen is an intrinsic membrane protein, resident in the Golgi apparatus of certain cells. Moreover, they enhance the hypothesis that the distribution of enzymes and polypeptides in the Golgi apparatus is cell specific.

Immunocytochemical visualization of the Golgi apparatus in several species, including human, and tissues with an antiserum against MG-160, a sialoglycoprotein of rat Golgi apparatus.

We used a monoclonal antibody (10A8), derived from mice immunized with fractions enriched in Golgi apparatus of rat brain neurons, to isolate an intrinsic membrane sialoglycoprotein of 160 KD from rat brain. By immunoelectron microscopy the sialoglycoprotein, named MG-160, was localized in medical cisternae of the Golgi apparatus of neurons, glia, adenohypophysis, and cultured rat pheochromocytoma (PC 12). The monoclonal antibody (MAb) reacted only with rat tissues. Because the epitope(s) recognized by a monoclonal antibody may be restricted, localization of an antigen by a single MAb may not reflect the extent of the distribution of antigen in various species and tissues. Therefore, to further investigate the presence and localization of MG-160 or of an antigenically related protein in several species and tissues, we used a polyclonal antiserum raised against MG-160 purified by antibody (10A8) affinity chromatography. Immunoblots of crude microsomal fractions from rat brain probed with the antiserum against MG-160 showed two to three prominent bands of approximately 160, 150, and 68 KD. Immunoblots of crude microsomal fractions from human, chicken, and frog brains showed prominent bands of 130-140 and 68 KD. Immunoblots of crude membrane fractions from Saccharomyces cerevisiae showed prominent bands of approximately 110-120 and 80 KD. Light microscopic immunocytochemical studies with frog, chicken, mouse, rat, rabbit, bovine, and human brains and with several other rat and human tissues showed a staining pattern consistent with the Golgi apparatus. Immunoelectron microscopy with rat and human brain and with rat myocardium and pituitary showed prominent and exclusive staining of cis, medial, and occasionally trans cisternae of the Golgi apparatus. The cisternae of the trans Golgi network were not stained. These findings are consistent with the hypothesis that a polypeptide related to MG-160 is present in the Golgi apparatus of several tissues in human, rodents, chicken, and frog and possibly in Saccharomyces cerevisiae. The antiserum to MG-160 represents a reliable reagent for immunohistochemical visualization of the Golgi apparatus in brain and several other human tissues obtained at autopsy, fixed with Bouin's, and embedded in paraffin.

Monoclonal antibody 2H1 detects a 60-65 KD membrane polypeptide of the rough endoplasmic reticulum of neurons and selectively stains cells of several rat tissues.

Monoclonal antibody (MAb) 2H1, raised in mice immunized with membrane fractions from cultured rat pheochromocytoma cells (clonal line PC-12), detects a polypeptide from rat brain and PC-12 cell membranes of 60-65 KD apparent molecular mass. The polypeptide has been localized by immunoelectron microscopy in the rough endoplasmic reticulum (RER) of neurons. By light microscopic immunocytochemistry, several rat tissues and two rat-derived cultured cell types show selective patterns of staining with 2H1. In the central nervous system, the antibody stains neuronal cytoplasm; in the spleen, staining is seen only in certain cells of the marginal zone of the white pulp, and in lymph nodes, in plasma cells, and in areas populated by monocytes and macrophages. Whereas astrocytes and adrenal medullary cells in situ are virtually unstained with 2H1, primary cultures of astrocytes and PC-12 cells, which are derived from adrenal medullary cells, stain intensely with 2H1. The strong staining of cultured astrocytes and PC-12 cells with 2H1 suggests that the levels of the 60-65 KD polypeptide are up-regulated during cell proliferation and growth. Only a few hepatocytes stain with 2H1; intestinal epithelial and pancreatic cells are not stained with 2H1. The organelle-specific antibody 2H1 may prove a useful probe in structural and functional studies of membranes of the rough endoplasmic reticulum in neurons, and in certain cells of the immune system.

Cloning and sequence analysis of the human MG160, a fibroblast growth factor and E-selectin binding membrane sialoglycoprotein of the Golgi apparatus.

The amino acid sequence of MG160, a membrane sialoglycoprotein of the medial cisternae of the rat Golgi apparatus, is more than 90% identical with CFR, a fibroblast growth factor (FGF) binding protein of chicken membranes, and with ESL-1, a ligand for E-selectin of plasma membranes of myeloid cells; furthermore, MG160, isolated by immunoaffinity chromatography from rat brain membranes, binds to basic FGF. The gene for MG160 has been assigned to human chromosome 16q22-23. To characterize this protein further in the human, its cDNA was cloned and sequenced. The protein has a large luminal domain composed of an initial proline-glutamine-rich segment, encoded by an uninterrupted exonic sequence of several CAG-CAA repeats. Expansion of CAG repeats has been implicated in the etiology of several neurodegenerative diseases. The proline-glutamine-rich segment is followed by 16 cysteine-rich repeats that contain five potential asparagine-linked glycosylation sites, which are conserved in the human, rat, mouse, and chicken. The large intralumenal domain of the protein is followed by a single transmembrane domain and a 13-amino-acid cytoplasmic carboxy-terminal tail, which is identical to that in the chicken, rat, and mouse. The overall amino acid identifies between MG160, CFR, and ESL-1 range from 88% to 95%. In several human fetal and adult tissues, three mRNA transcripts for MG160 of 10 kb, 5 kb, and 3.8 kb were identified by Northern blot analysis of poly(A)-selected mRNAs. These transcripts may represent alternatively spliced mRNAs of the protein or mRNAs encoding closely related proteins of the Golgi apparatus and/or plasma membranes.

Conjugates of horseradish peroxidase (HRP) with cholera toxin and wheat germ agglutinin are superior to free HRP as orthogradely transported markers.

The relative sensitivity of horseradish peroxidase (HRP) conjugates of cholera toxin (CTHRP) and wheat germ agglutinin (WGHRP) as orthogradely transported markers was compared with that of free HRP (FHRP) following unilateral, intravitreal injections of each probe in the rat. Both CTHRP and WGHRP labeled all of the recognized retinal efferents of the rat and were 30-50 times more sensitive than FHRP. In additional experiments the 'receptors' specificity of CTHRP and WGHRP was also demonstrated.

Horseradish peroxidase (HRP) conjugates of cholera toxin and lectins are more sensitive retrogradely transported markers than free HRP.

Horseradish peroxidase (HRP) conjugates of 8 different lectins (wheat germ agglutinin, ricinus communis I and II, peanut agglutinin, lens culinaris, soybean agglutinin, limulus polyhemus, ulex europaeus I) and cholera toxin (CT) or free HRP (FHRP) were individually injected into the submandibular gland (SMG) or anterior chamber (AC) of the eye and the retrogradely labeled neurons in the superior cervical ganglion (SCG) were quantitated. The effect of using 3 different cross-linking reagents (glutaraldehyde, p-benzoquinone and periodic acid) on the results obtained with HRP conjugates of wheat germ agglutinin (WG) was also examined. The results in 100 rats demonstrated the superior sensitivity of ligand-HRP conjugates over FHRP as retrogradely transported markers. After SMG injections, HRP conjugates of CT, WG and soybean agglutinin were 20-50 times more sensitive than FHRP. After AC injections, HRP conjugates of CT and WG consistently yielded labeled SCG neurons while FHRP failed to do so even when the amount of FHRP injected was increased 10-fold. The sensitivity of HRP conjugates of WG was similar after SMG injections using each of the 3 cross-linking reagents, but AC injections of conjugates produced with p-benzoquinone yielded twice as many labeled SCG neurons as the other WG conjugates. Mixtures of conjugates with and without FHRP were no more sensitive than the most sensitive individual ligand-HRP conjugates, except after SMG injections of a conjugate mixture with FHRP. Additional experiments demonstrated the specificity of the ligand-"receptor" interaction of WG and CT and that the superior sensitivity of these ligand-HRP conjugates does not depend on the tissue destruction produced by the injection procedure.

A quantitative comparison of the efficiency of orthograde axonal transport and transsynaptic transport of iodinated (125I) wheat germ agglutinin (I-WGA) and horseradish peroxidase labeled I-WGA (I-WGA-HRP) in the rat visual system.

A quantitative study of the orthograde axonal transport of iodinated wheat germ agglutinin (I-WGA) and I-WGA conjugated to horseradish peroxidase (I-WGA-HRP) demonstrated labeling of both the lateral geniculate nucleus and the visual cortex following intraocular injections. I-WGA was 13 times more efficient than I-WGA-HRP as a marker of the primary projection from the retina to the geniculate and 3 times more efficient than I-WGA-HRP as a marker of the second order projection to the cortex.

The Golgi apparatus-complex of neurons and astrocytes studied with an anti-organelle antibody.

An antiserum reacting with a 135-kDa antigen of rat liver Golgi apparatus-complex was used to stain, by light microscopic and ultrastructural immunocytochemistry, sections of rat cerebellum and by immunoblot homogenates of whole brain, isolated neurons and a fraction of enriched neuronal Golgi apparatus. In sections of rat cerebellum fixed with periodate-lysine-paraformaldehyde and immunostained with the direct peroxidase or peroxidase-antiperoxidase methods, the Golgi apparatus-complex in perikarya of neurons and glia was stained. Occasionally, nuclear envelopes and cisternae of the rough endoplasmic reticulum of neurons and glia were stained. Immunostain was not observed in peripheral dendrites, axons and presynaptic terminals. In striking contrast, peripheral smooth cisternae of astrocytic perikarya and processes were stained. Immunoblots of whole-brain membrane fractions, homogenates of isolated neurons and an enriched neuronal fraction of the Golgi apparatus-complex showed a principal single band of 64-kDa apparent mol. wt. We have concluded that the putative 64-kDa antigen(s) is distributed in cisternae of the Golgi apparatus-complex and occasionally in the nuclear envelope and rough reticulum, within the perikarya of neurons and glia. A second important distribution of the 64-kDa antigen(s), involving peripheral cisternae in perikarya and processes of astrocytes, is consistent with the hypothesis that the Golgi apparatus-complex of these cells extends to the periphery of these cells. The functional implications of the peripheral localization of the 64-kDa antigen(s) in astrocytes are discussed.

The golgi apparatus is present in perisynaptic, subependymal and perivascular processes of astrocytes and in processes of retinal Müller glia.

The Golgi apparatus (GA) of innervated rat and chicken skeletal muscle is present in a typical perinuclear location, and in subsynaptic areas where it disperses after denervation. It was suggested that the subsynaptic segments of the GA are linked with functions involved in the maturation and targeting of synaptic proteins. Similarly, the GA of rat myocardium is found in a perinuclear location and between myofibrils, adjacent to the T system of tubules. These findings raise the question whether the GA of polarized cells is present in a typical perinuclear location, for the performance of general "housekeeping" functions, and in distal areas, for the mediation of specialized functions. Astrocytes may contain GA within their long cytoplasmic processes which are difficult to identify in thin sections. To ensure the astrocytic origin of GA in otherwise unidentifiable small processes, we used transgenic mice expressing the rat MG160 medial Golgi sialoglycoprotein only in the GA of astrocytes, and visualized the GA with monoclonal antibody 10A8 (mAb10A8) which reacts only with rat MG160. Thus, we identified cisternae of the GA in distal perisynaptic and subependymal processes, in perivascular foot plates of cerebral astrocytes, and in processes of the Müller glia in the retina. A similar strategy may be adopted in future investigations aiming at the detection of elements of the GA in distal processes of neurons and oligodendrocytes. The functional implications of GA in perisynaptic astrocytic processes and other processes are unknown. However, the isolation and molecular characterization of the perisynaptic subset of astrocytic Golgi may be feasible, since others have purified the astrocytic glutamate transporter 1 (GLT1) from crude synaptosomal fractions in which astrocytic processes are probably unavoidable contaminants.

Cholera toxin and wheat germ agglutinin conjugates as neuroanatomical probes: their uptake and clearance, transganglionic and retrograde transport and sensitivity.

Horseradish peroxidase (HRP) conjugates of 6 different lectins and cholera toxin (CTHRP) were quantitatively compared with respect to: (a) their behavior at the injection site and (b) their ability to label, by means of transganglionic and retrograde axonal transport, axon terminals and neurons in the medulla of the rat subsequent to injections of each probe into the anterior two thirds of the tongue. HRP conjugates of wheat germ agglutinin (WGHRP) and CTHRP were more sensitive than any of the other lectin-HRP conjugates. Both were far superior to free-HRP (FHRP) in demonstrating these projections and CTHRP was the most sensitive transganglionic and retrograde probe. Additional experiments demonstrated that this superiority was not an artifact of the volume of material injected into the tongue nor of the injection site area or survival time selection. These experiments demonstrated further that CTHRP and WGHRP remain at the injection site approximately twice as long as FHRP and that their removal from or degradation in retrogradely labeled neurons requires approximately twice as much time as that required for FHRP. These observations, together with earlier studies from this laboratory, suggest the following conclusions: (1) CTHRP and WGHRP are superior in sensitivity to FHRP for studies of neuronal connectivity; and (2) HRP conjugates of ligands such as CTHRP and WGHRP are internalized, transported and/or degraded by mammalian neurons in a manner which differs from that of FHRP, a macromolecule for which neuronal plasma membrane 'receptors' are lacking.

In vivo uptake of wheat germ agglutinin-horseradish peroxidase conjugates into neuronal GERL and lysosomes.

In published studies we have shown that lectins and cholera toxin bind at 4 degrees C on surfaces in cultured neurons and neuroblastoma cells and, after incubations of labeled cells at 37 degrees C, undergo endocytosis (adsorptive endocytosis) into GERL (Golgi-endoplasmic reticulum-lysosome). In this study conjugates of wheat germ agglutinin (WGA) with horseradish peroxidase (HRP), or free HRP, were injected in rat submandibular glands or over superior cervical ganglia, and their uptake was examined by electron microscopic cytochemistry for HRP. Lysosomes and vesicles and cisternae of GERL were involved in the uptake of WGA-HRP injected into the submandibular gland or into the superior cerival ganglion; HRP, administered in a similar fashion, but in amounts 10-15-fold higher than WGA-HRP, underwent endocytosis into lysosomes. The present in vivo studies, as well as previous in vitro studies, indicate that neuronal GERL is part of the endocytic pathway of various ligands which bind on corresponding surface (plasma membrane) "receptors".

Aggregates of mutant protein appear progressively in dendrites, in periaxonal processes of oligodendrocytes, and in neuronal and astrocytic perikarya of mice expressing the SOD1(G93A) mutation of familial amyotrophic lateral sclerosis.

Mice expressing the G93A and other mutations of Cu,Zn superoxide dismutase (SOD1(G93A)) are valid models for the familial form of amyotrophic lateral sclerosis (FALS) with SOD1 mutations and, probably, for sporadic ALS. Adult mice become progressively paralyzed and show most of the histopathological lesions reported in sporadic ALS, i.e. neuronal loss, astrogliosis, ubiquitin and Lewy body-like inclusions, dystrophic axons and fragmentation of the Golgi apparatus (GA) of motor neurons. In transgenic mice, the mutant protein and ubiquitin aggregate within pathological 13 nm thick filaments [Stieber A, Gonatas JO, Gonatas NK. J Neurol Sci 2000;173:53-62]. This immunocytochemical and quantitative study of mice expressing SOD1(G93A) establishes the chronological order and cellular localization of aggregates of SOD1 and their correlation with fragmentation of the GA. Young asymptomatic mice expressing SOD1(G93A) showed aggregates of mutant SOD1 within neurites, prior to the detection of SOD1 in the perikarya of spinal cord motor neurons and astrocytes. Both dendrites and the periaxonal oligodendroglial cytoplasm, surrounding atrophic axons, contained SOD1 as revealed by immunoelectron microscopy The perikarya of a small percentage of spinal cord motor neurons contained both fragmented GA and aggregates of SOD1; however, about 50% of motor neurons with fragmented GA did not contain SOD1 in the perikaryon, suggesting that aggregates of mutant protein may not directly cause fragmentation of the GA. The mechanism of the putative toxic effect by the mutant protein remains to be clarified. The isolation and biochemical characterization of the filamentous aggregates of mutant protein and ubiquitin from spinal cords of transgenic mice expressing mutations of the SOD1 gene may offer clues on pathogenetic mechanisms.

Aggregation of ubiquitin and a mutant ALS-linked SOD1 protein correlate with disease progression and fragmentation of the Golgi apparatus.

Transgenic mice that express the G93A mutation of human Cu,Zn superoxide dismutase (SOD1(G93A)), found in familial amyotrophic lateral sclerosis (FALS), showed clinical symptoms and histopathological changes of sporadic ALS, including fragmentation of the neuronal Golgi apparatus (GA). The finding of fragmented neuronal GA in asymptomatic mice, months before the onset of paralysis, suggests that the GA is an early target of the pathological processes causing neuronal degeneration. Transgenic mice expressing human SOD1(G93A) have aggregates of mutant protein and ubiquitin in neuronal and glial cytoplasm; they appeared first in the neuropil and later in the perikarya of motor neurons, where they were adjacent to fragmented GA. The aggregates of SOD1(G93A) appeared in neuronal perikarya of asymptomatic mice containing fragmented GA. The numbers of neurons with deposits of SOD1(G93A) and fragmented GA progressively increased with age. Immuno-electron microscopy using colloidal gold showed labeling of ubiquitin and SOD1 over 13 nm thick cytoplasmic filaments. Spinal cord extracts showed a 20-fold increase of SOD1(G93A) in transgenic mice compared to the wild-type protein in controls. The results suggest a causal relationship between the aggregation of mutant SOD1 and ubiquitin, fragmentation of the Golgi apparatus of motor neurons and neurodegeneration.

The neuronal Golgi apparatus is fragmented in transgenic mice expressing a mutant human SOD1, but not in mice expressing the human NF-H gene.

Fragmentation of the Golgi apparatus (GA) of motor neurons was first described in sporadic amyotrophic lateral sclerosis (ALS) and later confirmed in transgenic mice expressing the G93A mutation of the gene encoding the enzyme Cu,Zn superoxide dismutase (SOD1(G93A)) found in some cases of familial ALS. In these transgenic mice, however, the fragmentation of the neuronal GA was associated with cytoplasmic and mitochondrial vacuoles not seen in ALS. The present new series of transgenic mice expressing 14-17 trans gene copies of SOD1(G93A), compared to 25 copies in the mice we studied previously, showed consistent fragmentation of the GA of spinal cord motor neurons, axonal swellings, Lewy-like body inclusions in neurons and glia, but none of the cytoplasmic or mitochondrial vacuoles originally reported. Thus, this animal model recapitulates the clinical and most neuropathological findings of sporadic ALS. Neurofilaments (NF) accumulate in axons and, less often, in neuronal perikarya in most cases of sporadic ALS and they have been implicated in its pathogenesis. In order to investigate whether fragmentation of the neuronal GA also occurs in association with accumulation of perikaryal NFs, we studied the organelle in transgenic mice expressing the heavy subunit of human neurofilaments (NF-H) which developed a motor neuronopathy resembling ALS. The neuronal GA of mice expressing NF-H, however, was intact despite massive accumulation of NFs in both perikarya and axons of motor neurons. In contrast, in transgenic mice expressing SOD1(G93A), the GA was fragmented despite the absence of accumulation of perikaryal NFs. These findings suggest that, in transgenic mice with neuronopathies caused by the expression of mutant SOD1(G93A) or the human NF-H, the GA and the perikaryal NFs are independently involved in the pathogenesis. The evidence suggests that the GA plays a central role in the pathogenesis of the vast majority of sporadic ALS and in FALS with SOD1 mutations.

Disruption of the structure of the Golgi apparatus and the function of the secretory pathway by mutants G93A and G85R of Cu, Zn superoxide dismutase (SOD1) of familial amyotrophic lateral sclerosis.

The Golgi apparatus of motor neurons (GA) is fragmented in sporadic amyotrophic lateral sclerosis (ALS), in familial ALS with SOD1 mutations, and in mice that express SOD1G93A of familial ALS, in which it was detected months before paralysis. In paralyzed transgenic mice expressing SOD1G93A or SOD1G85R, mutant proteins aggregated not only in the cytoplasm of motor neurons, but also in astrocytes and oligodendrocytes. Furthermore, aggregation of the G85R protein damaged astrocytes and was associated with rapidly progressing disease. In order to gain insight into the functional state of the fragmented GA, we examined the effects of S0D1 mutants G93A and G85R in Chinese Hamster Ovary Cells (CHO). In contrast to cells expressing the wt and G93A, the G85R expressers had no SOD1 activity. However, cells expressing both mutants, and to a lesser degree the wt, showed decreased survival, fragmentation of the GA, and dysfunction of the secretory pathway, which was assessed by measuring the amount of cell surface co-expressed CD4, a glycoprotein processed through the GA. The G93A and wt proteins were partially recovered in detergent insoluble fractions; while the recovery of G85R was minimal. Both mutants showed equal reductions of cell survival and function of the secretory pathway, in comparison to the wt and cells expressing mutant alsin, a protein found in rare cases of fALS. These results are consistent with the conclusion that the two SOD1 mutants, by an unknown mechanism, promote the dispersion of the GA and the dysfunction of the secretory pathway. This and other in vitro models of mutant SOD1 toxicity may prove useful in the elucidation of pathogenetic mechanisms.