Maintenance of high proteolipid protein level in adult central nervous system myelin is required to preserve the integrity of myelin and axons.

Proteolipid protein (PLP) is the most abundant integral membrane protein in central nervous system (CNS) myelin. Expression of the Plp-gene in oligodendrocytes is not essential for the biosynthesis of myelin membranes but required to prevent axonal pathology. This raises the question whether the exceptionally high level of PLP in myelin is required later in life, or whether high-level PLP expression becomes dispensable once myelin has been assembled. Both models require a better understanding of the turnover of PLP in myelin in vivo. Thus, we generated and characterized a novel line of tamoxifen-inducible Plp-mutant mice that allowed us to determine the rate of PLP turnover after developmental myelination has been completed, and to assess the possible impact of gradually decreasing amounts of PLP for myelin and axonal integrity. We found that 6 months after targeting the Plp-gene the abundance of PLP in CNS myelin was about halved, probably reflecting that myelin is slowly turned over in the adult brain. Importantly, this reduction by 50% was sufficient to cause the entire spectrum of neuropathological changes previously associated with the developmental lack of PLP, including myelin outfoldings, lamellae splittings, and axonal spheroids. In comparison to axonopathy and gliosis, the infiltration of cytotoxic T-cells was temporally delayed, suggesting a corresponding chronology also in the genetic disorders of PLP-deficiency. High-level abundance of PLP in myelin throughout adult life emerges as a requirement for the preservation of white matter integrity.

The K2P -channel TASK1 affects Oligodendroglial differentiation but not myelin restoration.

In multiple sclerosis, demyelination occurs as a consequence of chronic autoimmunity in the central nervous system causing progressive neurological impairment in patients. After a demyelinating event, new myelin sheaths are formed by adult oligodendroglial progenitor cells; a process called remyelination. However, remyelination often fails in multiple sclerosis due to insufficient recruitment and differentiation of oligodendroglial precursor cells. A pivotal role for the two-pore-domain potassium (K2P ) channel, TASK1, has already been proven for an animal model of multiple sclerosis. However, the mechanisms underlying the TASK1-mediated effects are still elusive. Here, we tested the role of TASK1 channels in oligodendroglial differentiation and remyelination after cuprizone-induced demyelination in male mice. We found TASK1 channels to be functionally expressed on primary murine and human, pluripotent stem cell-derived oligodendrocytes. Lack of TASK1 channels resulted in an increase of mature oligodendrocytes in vitro as well as a higher number of mature oligodendrocytes and accelerated developmental myelination in vivo. Mechanistically, Task1-deficient cells revealed a higher amount of phosphorylated WNK1, a kinase known to be involved in the downstream signaling of the myelination regulator LINGO-1. Furthermore, we analyzed the effect of genetic TASK1 ablation or pharmacological TASK1 inhibition on disease-related remyelination. Neither channel inhibition nor lack of TASK1 channels promoted remyelination after pathological demyelination. In summary, we conclude that functional TASK1 channels participate in the modulation of differentiating oligodendroglial cells in a previously unknown manner. However, while being involved in developmental myelination our data suggest that TASK1 channels have no major effect on remyelination.

Changes in ultrastructure and expression of steroidogenic factor-1 in ovaries of zebrafish Danio rerio exposed to glyphosate.

Glyphosate is a broad-spectrum organophosphate (OP) herbicide, highly soluble in water, and when applied in terrestrial systems it penetrates into soil, eventually reaching the aquatic community and affecting nontarget organisms. The aim of this study was to evaluate the toxicity of glyphosate on ovaries of zebrafish (Danio rerio). Ovaries (n = 18 per triplicate) were exposed to 65 µg/L of glyphosate [N-(phosphonomethyl) glycine] for 15 d. This concentration was determined according to Resolution 357/2005/CONAMA/Brazil, which establishes the permissible concentration of glyphosate in Brazilian inland waters. Nonexposed ovaries (n = 18 per triplicate) were used as control. Subsequently, morphology and expression of steroidogenic factor-1 (SF-1) of exposed and nonexposed ovaries was determined. No apparent changes were noted in general morphology of exposed and nonexposed ovaries. However, a significant increase in diameter of oocytes was observed after exposure to glyphosate. When ovarian ultrastructure was examined the presence of concentric membranes, appearing as myelin-like structures, associated with the external membranes of mitochondria and with yolk granules was found. After glyphosate exposure, immunohistochemistry and immunoblotting revealed greater expression of SF-1 in the oocytes, which suggests a relationship between oocyte growth and SF-1 expression. These subtle adverse effects of glyphosate on oocytes raised a potential concern for fish reproduction. These results contribute to understanding glyphosate-induced toxicity to nontarget organisms, showing subcellular and molecular impairments that may affect reproduction in +female fish.

Myelin structure is unaltered in chemotherapy-induced peripheral neuropathy.

PURPOSE: Alterations in mRNA for myelin proteins are reported in animal models of chemotherapy-induced peripheral neuropathies (CIPN); however, ultrastructural changes in aldehyde-fixed and plastic-embedded myelin are not evident by electron microscopy. Therefore, we used X-ray diffraction (XRD) to investigate more subtle changes in myelin sheath structure from unfixed nerves. EXPERIMENTAL DESIGN: We used in vivo chronic animal models of CIPN in female Wistar rats, administering cisplatin (CDDP 2mg/kg, i.p. twice/week), paclitaxel (PT 10mg/kg, i.v. once/week) or bortezomib (0.20mg/kg, i.v. three times/week) over a total period of 4weeks. Animal weights were monitored, and tail nerve conduction velocity (NCV) was determined at the end of the treatments to assess the occurrence of peripheral neuropathy. Sciatic nerves were collected and the myelin structure was analyzed using electron microscopy (EM) and XRD. RESULTS: All the rats treated with the chemotherapy agents developed peripheral neuropathy, as indicated by a decrease in NCV values; however, light and electron microscopy indicated no severe pathological alterations of the myelin morphology. XRD also did not demonstrate significant differences between sciatic nerves in treated vs. control rats with respect to myelin period, relative amount of myelin, membrane structure, and regularity of membrane packing. CONCLUSIONS: These results indicate that experimental peripheral neuropathy caused by CDDP, PT, and bortezomib-which are among the most widely used chemotherapy agents-does not significantly affect the structure of internodal myelin in peripheral nerve.

Ultrastructural identification of oligodendrocyte/myelin proteins in corpus callosum of hypothyroid animals.

Thyroid hormone (T3) deficiency impairs the development of the CNS, particularly myelination. We have previously described an increase in the frequency of morphological abnormalities in the central myelin sheath in a hypothyroidism model, which reinforced the hypothesis of a role for T3 in myelin compaction. However, there are no data concerning the cellular distribution of myelin proteins in hypothyroid animals. In the present work, we describe the distribution of 2',3'-cyclic nucleotide 3'-phosphodiesterase (CNPase), myelin basic protein (MBP) and proteolipid protein (PLP) throughout the central myelin sheath of a hypothyroidism model. We used euthyroid and hypothyroid adult rats at 90 days of age. In order to induce hypothyroid status, animals received 0.02% methimazol from the 19th gestation day onwards. After perfusion with a fixative mixture, small pieces of corpus callosum were obtained, dehydrated and embedded in LR White resin. Ultrathin sections were immunoreacted, using specific antibodies revealed by a secondary antibody coupled to colloidal gold particles of 10nm. Gold particle density per region of myelin sheath for each one of these proteins was obtained. In normal animals, CNPase, PLP and MBP were identified in sites that had already been described in previous studies. In hypothyroid animals, CNPase was identified in the region corresponding to compact lamellae, which normally does not contain this protein, while, in this same region, PLP and MBP immunolabeling were decreased. These results suggest that thyroid hormone deficiency impairs the distribution of the major oligodendrocyte/myelin markers. This effect may justify the reduction in myelin sheath compaction previously demonstrated in a similar model of hypothyroidism.

Structure and stability of internodal myelin in mouse models of hereditary neuropathy.

Peripheral neuropathies often result in abnormalities in the structure of internodal myelin, including changes in period and membrane packing, as observed by electron microscopy (EM). Mutations in the gene that encodes the major adhesive structural protein of internodal myelin in the peripheral nervous system of humans and mice--P0 glycoprotein--correlate with these defects. The mechanisms by which P0 mutations interfere with myelin packing and stability are not well understood and cannot be provided by EM studies that give static and qualitative information on fixed material. To gain insights into the pathogenesis of mutant P0, we used x-ray diffraction, which can detect more subtle and dynamic changes in native myelin, to investigate myelin structure in sciatic nerves from murine models of hereditary neuropathies. We used mice with disruption of one or both copies of the P0 gene (models of Charcot-Marie-Tooth-like neuropathy [CMT1B] or Dejerine-Sottas-like neuropathy) and mice with a CMT1B resulting from a transgene encoding P0 with an amino terminal myc-tag. To directly test the structural role of P0, we also examined a mouse that expresses P0 instead of proteolipid protein in central nervous system myelin. To link our findings on unfixed nerves with EM results, we analyzed x-ray patterns from unembedded, aldehyde-fixed nerves and from plastic-embedded nerves. From the x-ray patterns recorded from whole nerves, we assessed the amount of myelin and its quality (i.e. relative thickness and regularity). Among sciatic nerves having different levels of P0, we found that unfixed nerves and, to a lesser extent, fixed but unembedded nerves gave diffraction patterns of sufficient quality to distinguish periods, sometimes differing by a few Angstroms. Certain packing abnormalities were preserved qualitatively by aldehyde fixation, and the relative amount and structural integrity of myelin among nerves could be distinguished. Measurements from the same nerve over time showed that the amount of P0 affected myelin's stability against swelling, thus directly supporting the hypothesis that packing defects underlie instability in "live" or intact myelin. Our findings demonstrate that diffraction can provide a quantitative basis for understanding, at a molecular level, the membrane packing defects that occur in internodal myelin in demyelinating peripheral neuropathies.

Ultrastructural identification of peripheral myelin proteins by a pre-embedding immunogold labeling method.

Ultrastructural immunolabeling of peripheral nervous system components is an important tool to study the relation between structure and function. Owing to the scarcity of certain antigens and the dense structure of the peripheral nerve, a pre-embedding technique is likely appropriate. After several investigations on procedures for pre-embedding immunolabeling, we propose a method that offers a good compromise between detection of antigenic sites and preservation of morphology at the ultrastructural level, and that is easy to use and suitable for investigations on peripheral nerve biopsies from humans. Pre-fixation by immersion in paraformaldehyde/glutaraldehyde is necessary to stabilize the ultrastructure. Then, ultrasmall gold particles with silver enhancement are advised. Antibodies against myelin protein zero and myelin basic protein were chosen for demonstration. The same technique was applied to localize a 35 kDa myelin protein.

Nogo-A expression in mature oligodendrocytes of rat spinal cord in association with specific molecules.

Nogo-A is known as an oligodendrocyte/myelin-associated molecule having an inhibitory effect on neurite outgrowth in the central nervous system. During development, starting from P21 Nogo-A was detected in the cytoplasm of mature oligodendrocytes with compact myelin sheaths in the rat spinal cord. COS7 cells transfected with recNogo-A displayed strong Nogo-A immunoreactivity in their cytoplasm as well as on the mitotic spindle. Nogo-A was not detected in membrane protein fractions from transfected plus biotinylated COS7 cells. Nogo-A was co-immunoprecipitated with alpha-tubulin and myelin basic protein (MBP) from rat brain tissue. These results show that Nogo-A is expressed in association with tubulin and MBP in the mature oligodendrocytes.

Mechanical ventilation of rat lung: effect on surfactant forms.

Mechanical ventilation of the lung could affect surfactant turnover by alteration of its secretion, recycling, and degradation. In vitro studies of surfactant subfractions recoverable from lavage fluid have led to predictions about surfactant physiology in vivo that include morphological transformations. We used electron microscopy to study in situ lipid forms in alveoli of rat lungs after two ventilation strategies [15 min at pressures (cmH(2)O) of 20/0 or 20/10]. In control animals, 4% of the lipid profile area in the surface lining layer was myelin figures (MF), 14% was tubular myelin, 37% was vesicular forms (VF), and the remainder (45%) was hypophase. Compared with controls, the length-normalized sum of the lipid forms and the hypophase was two times as great in the lungs of the 20/0 group. MF were threefold higher in the 20/0 group and fivefold higher in the 20/10 group. VF doubled after ventilation at 20/0, but VF were the same as control after ventilation at 20/10. The results showed that a ventilation pattern of 20/0 compared with that of 20/10 group was associated with a significantly larger VF, suggesting an increased net production of these surfactant forms during a large tidal volume breathing pattern. These morphological results are consistent with published results using physical methods of fractionating lung lavage.

Ins and outs of peripheral myelin protein-22: mapping transmembrane topology and intracellular sorting.

Molecular genetic studies have shown that the peripheral myelin protein 22 (PMP22) is a key gene in hereditary peripheral neuropathies and appears to be essential for the formation and maintenance of myelin in the PNS. Based on the amino acid sequence the predicted structure of PMP22 protein contains two major distinct hydrophilic regions and four transmembrane domains. To analyze the cellular localization and membrane topology of PMP22 we inserted an octapeptide tag-sequence at the amino or at the carboxyl terminus of the PMP22 open reading frame and generated different chimeric constructs which were expressed in HeLa cells. The expression of the tagged PMP22 protein and its orientation with respect to the plasma membrane were analyzed using antibodies raised against specific PMP22 epitopes and the tag sequence. Combined indirect, double-immunofluorescence labeling and confocal microscopy showed that PMP22 is synthesized in the rough endoplasmic reticulum of transfected cells and passes through the Golgi apparatus to the cell surface. We determined the transmembrane organization of PMP22 providing the first experimental evidence that confirms the cytoplasmic disposition of its N and C termini and the extracellular localization of the two hydrophilic domains containing amino acids 28-40 and 118-131. This study provides the basis for further analysis aimed to identify functional domains of wild-type PMP22 and the cellular sorting of mutant forms of PMP22.

Clinical and pathological phenotype of the original family with Charcot-Marie-Tooth type 1B: a 20-year study.

Charcot-Marie-Tooth type 1B is an uncommon form of hereditary motor and sensory neuropathy caused by mutations in the P(0) myelin protein gene on chromosome 1. We report here a 20-year observation of 13 members of the first family with Charcot-Marie-Tooth disease to demonstrate linkage to chromosome 1 and now known to have a C270A mutation in the P(0) gene altering the extracellular domain of the protein. Affected individuals generally show an early age at onset, often indicated by delayed ability to walk. Proximal muscle weakness of the lower extremities is common and often marked, but the individuals remain ambulatory and there is no decrease in life span. Motor nerve conduction velocities of the fastest fibers are severely slowed (mean, 9-11 m/sec), even when compared with 3 families having Charcot-Marie Tooth type 1A (mean, 19-21 m/sec). Variability of disability between family members suggests that genetic and environmental factors in addition to the P(0) mutation play a role in the final phenotype. Nerve biopsy specimens demonstrate hypertrophy, onion bulb formation, loss of myelinated fibers, and occasional myelin thickening similar to that described in P(0) myelin knockout mice. Autopsy of the 92-year-old great-grandmother in this family demonstrated diffuse involvement of sensory and motor nerves, with loss of myelin in the posterior columns of the spinal cord and loss of anterior horn neurons but without other involvement of the central nervous system. This family demonstrates the long-term phenotypic consequences on the peripheral nervous system of a specific point mutation in the P(0) myelin gene.

Myelin-containing corpora amylacea in vestibular root entry zone.

The vestibular nerve of patients with Meniere's disease and vascular cross-compression syndrome of the root entry zone due to the antero-inferior cerebellar artery was studied. All patients underwent vestibular neurectomy using the retrosigmoid approach, which permits the removal of a long nerve segment. CA were found in the cytoplasm of astrocytes that had not shown signs of degeneration at the central portion of the vestibular root entry zone. No membrane intervened between CA and the surrounding cytoplasm, which was rich in filaments, in particular near the CA, and poorly equipped with other organelles. CA were round or oval inclusions measuring 10-12 microns in diameter. The matrix of the CA was composed of low-density amorphous material, with irregular masses displaying a medium density. A network of randomly oriented filaments and bilaminar, osmiophilic lipid fragments with the same structure and thickness of myelin layers were embedded in the matrix. The CA rich in bilaminar fragments were recognizable also at low magnification for their high electron density. In the astrocytic cytoplasm, near the CA, round or ovalshaped, electron-dense bodies with a multilamellar structure were often visible. These results confirm the hypothesis that CA may contain degenerating myelin embedded in a microenvironment rich in glucose polymers and that CA could be an indicator of neurodegeneration.

Ultrastructural distribution of PMP22 in Charcot-Marie-Tooth disease type 1A.

Peripheral Myelin Protein-22 (PMP22) is a membrane glycoprotein which represents up to 5% of total protein in myelin of peripheral nerves. Mutations affecting the PMP22 gene have been linked to the inherited peripheral neuropathies Charcot-Marie-Tooth disease type 1A (CMT1A; duplications and point mutations), Dejerine-Sottas syndrome (DSS; point mutations), and hereditary neuropathy with liability to pressure palsies (HNPP; deletions). In this study, we determined the ultrastructural distribution of PMP22 and other myelin proteins in normal human peripheral nervous system (PNS) nerves and in CMT1 patients with or without the CMT1A duplication on chromosome 17. Our results demonstrate that PMP22, P0 protein, and myelin basic protein are present in compact myelin of all patients examined. PMP22 was also present in the plasma membrane of Schwann cells of unmyelinated fibers and onion bulbs. Although the precise biological role of PMP22 remains to be discovered, our results support the hypothesis that this protein serves multiple functions in Schwann cells.

Myelin vesicles: what we know and what we do not know.

In the following review, we address difficulties that have arisen when attempting to convert the myelin multilayers into vesicles. The emphasis is on CNS myelin of adult mammals although both central nervous system (CNS) and peripheral nervous system (PNS) myelin are considered. The ability to prepare vesicle of myelin membrane has yet not been feasible. We hope to clarify some aspect of this problem and offer some possible approaches. Special attention is paid to myelin swelling phenomena because these indicate ways in which the myelin multilayer can break down. Images of isolated myelin are reviewed with special attention to the ways in which the multilayer actually breaks down. Attempts at reproducing a procedure for vesiculating myelin are summarized, and a critique is given to account for the inability to reproduce the published results. Finally, novel approaches for vesiculating myelin are proposed, which are based on well-characterized swelling phenomena.

The human entorhinal cortex: a cytoarchitectonic analysis.

The entorhinal cortex of man is in the medial aspect of the temporal lobe. As in other mammalian species, it constitutes an essential component of the hippocampal formation and the route through which the neocortex interacts with the hippocampus. The importance of knowing its architecture in detail arises from the possibility of extrapolating it to experimental findings, notably in the nonhuman primate. We have investigated the cytoarchitectonic features of the human entorhinal cortex by using as a base our previous study (D.G. Amaral, R. Insausti, and W.M. Cowan [1987] J. Comp. Neurol. 264:326-355) of the nonhuman primate entorhinal cortex. We prepared serial sections of the temporal lobe from 35 normal brains. Thionin- and myelin-stained series were made of all cases. Sections spaced 500 microns apart through the full rostrocaudal extent of the entorhinal cortex were analyzed. The human entorhinal cortex is made up of six layers, of which layer IV does not appear throughout all subfields of the entorhinal cortex. The overall appearance resembles that of the adjacent neocortex in lateral and caudal portions. In harmony with general structural principles in the nonhuman primate entorhinal cortex, our analysis supports the partitioning of the human entorhinal cortex into eight different subfields. (1) The olfactory subfield (EO), the rostralmost field, is little laminated. (2) The lateral rostral subfield (ELr), laterally located, merges with the laterally adjacent perirhinal cortex. (3) The rostral subfield (ER) is between EO and ELr, with better differentiation of layers II and III than EO. (4) The medial intermediate subfield (EMI) is located at the medial border. (5) The intermediate field (EI) is a lateral continuation of EMI; lamina dissecans (layer IV) can be best appreciated in this field. (6) The lateral caudal subfield (ELc) laterally borders on EI as a continuation of ELr. (7) The caudal subfield (EC) lies caudal to the beginning of the hippocampal fissure, with a distinctive, clear space (Vc) between layers V and VI. (8) The caudal limiting field (ECL) forms the caudal termination of the entorhinal cortex. Thus our parcellation of the entorhinal cortex in man is largely parallel to that arrived at in the monkey. This close homology provides a rational basis for the application to clinical problems of anatomical and functional information obtained in experimental work in nonhuman primates.

Effects of carbonic anhydrase II (CAII) deficiency on CNS structure and function in the myelin-deficient CAII-deficient double mutant mouse.

In the choroid plexus carbonic anhydrase II (CAII) supports the transport of bicarbonate ions, sodium ions, and water from blood to the CSF, and in the myelin sheath CAII supports compaction of myelin by stimulating cotransport of ions and water out from between the myelin membranes. In view of the latter, it is surprising that mutant mice deficient in CAII (Car-2n) have compact myelin. Since myelin basic protein also takes part in myelin compaction, we bred double CAII-deficient, myelin-deficient (Mld) mutant mice, in which the adults would have some compact myelin sheaths and a partial deficiency in myelin basic protein, with a view to examining oligodendrocytes and myelin sheaths in the double mutant. Like the parent Mld strain, the double mutants displayed tremors and seizures; however, the onset of seizures was delayed significantly in the double mutants, and the lifespan increased by several months. Like the brains of Car-2n mutants, those of double mutants (MldCar-2n) were deficient in mRNA and protein for CAII and showed upregulation of a different isozyme, CAIV. In the double mutants, oligodendrocytes were reduced in number, and the myelin sheaths and oligodendrocytes were swollen. The partial protection against seizures, which CAII deficiency conferred, suggests that acidosis in the central nervous system (CNS) of the Car-2n and MldCar-2n mice, due to absence of CAII from the choroid plexus, may downregulate the activity of NMDA receptors.(ABSTRACT TRUNCATED AT 250 WORDS)

The anterolateral funiculus in the spinal cord in amyotrophic lateral sclerosis.

We carried out a morphometric study on the myelinated fibers in the anterolateral funiculus (ALF) and lateral corticospinal tract (LCS) in the cervical segment of the spinal cord of 13 patients with classic amyotrophic lateral sclerosis (ALS), 6 of whom had been on a respirator; 5 age-matched subjects were used as controls. The results obtained revealed that: (1) the fiber-size distributions of the myelinated fibers in the ALF and LCS of the control subjects had peaks at 2 microns; (2) there were marked and significant losses of large myelinated fibers in the ALF and LCS of ALS patients; (3) the patients who required respirator support showed more severe degeneration in the ALF than those who required none; and (4) the degree of myelinated fiber loss in the LCS did not correlate with either the illness duration or the history of respirator use.

Cryo-electron microscopy of myelin treated with detergents.

After detergent extraction, myelin lamellae disaggregate into diverse globular fragments that correspond to the globular structures observed in normal cryofixed myelin. The interaction of these structural units in the bilayer is probably the morphological basis that provides the stability of the myelin lamellae and their lamination.

Reconstitution of proteolipid protein: some properties and its role in interlamellar attachment.

Proteolipid apoprotein (PLP) isolated from human brain was reconstituted in dioleoylphosphatidylcholine vesicles by dialysis from 2-chloroethanol, using a dialysis buffer of pH 5.0. Under these conditions, and in contrast with dialysis carried out at pH 7.4, well-defined unilamellar vesicles containing the protein were formed. As judged by electron microscopy and quasi-elastic light scattering, the size of the vesicles was determined by the initial protein/lipid ratio used for reconstitution. When the vesicles were incubated in a buffer at neutral pH, aggregation of the vesicles was observed, but their structure remained intact. Asymmetric aggregation occurred when the reconstituted vesicles were incubated with large unilamellar vesicles (LUVs) devoid of protein. This aggregation was accompanied by loss of membrane integrity, as revealed by extensive leakage of the LUVs, and by membrane lipid dilution, indicative of the occurrence of membrane fusion. Destabilization of the vesicles depended on the presence of negatively charged phosphatidylserine in the membrane of the LUVs. Similar effects, but to a lesser extent, were seen when the LUVs contained sulphatide, a negatively charged lipid prominently present in myelin. DM 20, a natural mutant of PLP, appeared to be far less potent in causing membrane lipid dilution than PLP. This could suggest that a distinct protein sequence of PLP, which is absent from DM 20, may be involved in triggering the observed membrane destabilization. Temperature-dependent experiments indicate that this sequence in PLP displays dynamic properties, its exposure being affected by conformational criteria. Exposure of this particular domain, in conjunction with its affinity for negatively charged lipid, could be related to a perturbation of the integrity of the myelin sheath, as will be discussed.

Functional topography of the myelin-associated glycoprotein. I. Mapping of domains by electron microscopy.

The functional topography of the myelin-associated glycoprotein (MAG) was investigated by electron microscopic analysis of rotary-shadowed molecules of a MAG fragment (MAG 90) comprising the five immunoglobulin-like domains of the extracellular part of the molecule. MAG 90 molecules appeared as rod-like structures (18.5 +/- 1.2 nm long and 4.0 +/- 0.8 nm wide) with a globular domain at one end. Antibodies directed against the amino- and carboxy-terminus of MAG 90 interacted with the non-globular terminal region, indicating that the molecule is bent in the globular region with the amino- and carboxy-terminal arms in close apposition to each other. An antibody which interferes with the binding of MAG to neurons interacted predominantly with the globular domain of MAG 90. The fibril-forming collagen types I, III and V bound mainly to the non-globular terminal region of MAG 90, whereas the majority of heparin molecules interacted with the globular region of the molecule. The L2/HNK-1 carbohydrate structure was localized at the non-globular region in the protein fragment comprising the fourth and fifth immunoglobulin-like domains.