Role of insulin-dependent cortical fodrin/spectrin remodeling in glucose transporter 4 translocation in rat adipocytes

Fodrin or nonerythroid spectrin is an abundant component of the cortical cytoskeletal network in rat adipocytes. Fodrin has a highly punctate distribution in resting cells, and insulin causes a dramatic remodeling of fodrin to a more diffuse pattern. Insulin-mediated remodeling of actin occurs to a lesser extent than does that of fodrin. We show that fodrin interacts with the t-soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) syntaxin 4, and this interaction is increased by insulin stimulation and decreased by prior latrunculin A treatment. Latrunculin A disrupts all actin filaments, inhibits glucose transporter 4 (GLUT4) translocation, and causes fodrin to partially redistribute from the plasma membrane to the cytosol. In contrast, cytochalasin D disrupts only the short actin filament signal, and cytochalasin D neither inhibits GLUT4 translocation nor fodrin redistribution in adipocytes. Together, our data suggest that insulin induces remodeling of the fodrin-actin network, which is required for the fusion of GLUT4 storage vesicles with the plasma membrane by permitting their access to the t-SNARE syntaxin 4.     

Axonal transport of actin in rabbit retinal ganglion cells

We labeled proteins in the cell bodies of rabbit retinal ganglion cells with [35S]methionine and subsequently observed the appearance of radioactive actin in tissues containing the axons and synaptic terminals of these neurons, i.e., the optic nerve (ON), optic tract (OT), lateral geniculate nucleus (LGN) and the superior colliculus (SC). The temporal sequence of appearance of labeled actin (which was identified by its specific binding to DNase I, its electrophoretic mobility, and its peptide map) in these tissues indicated that actin is an axonally transported protein with a maximum transport velocity of 3.4--4.3 mm/d. The kinetics of labeling actin were similar to the kinetics of labeling two proteins (M1 and M2) which resemble myosin; these myosin-like proteins were previously found to be included in the groups of proteins (groups III and IV) transported with the third and fourth most rapid maximum velocities. The similarity in transport between actin and myosin-like proteins supports the idea that a number of proteins in the third and fourth transport groups may be functionally related by virtue of their involvement in a force-generating mechanism and suggests the possibility that these proteins may be axonally transported as a preformed force-generating unit.     

Ca2+- and calmodulin-dependent stimulation of smooth muscle actomyosin Mg2+-ATPase by fodrin

Fodrin, a spectrin-like actin and calmodulin binding protein, was purified to electrophoretic homogeneity from a membrane fraction of bovine brain. The effect of fodrin on smooth muscle actomyosin Mg2+-ATPase activity was examined by using a system reconstituted from skeletal muscle actin and smooth muscle myosin and regulatory proteins. The simulation of actomyosin Mg2+-ATPase by fodrin showed a biphasic dependence on fodrin concentration and on the time of actin and myosin preincubation at 30 degrees C. Maximal stimulation (50-70%) was obtained at 3 nM fodrin following 10 min of preincubation of actin and myosin. This stimulation was also dependent on the presence of tropomyosin. In the absence of myosin light chain kinase, the fodrin stimulation of Mg2+-ATPase could not be demonstrated with normal actomyosin but could be demonstrated with acto-thiophosphorylated myosin, suggesting that fodrin stimulation depends on the phosphorylation of myosin. Fodrin stimulation was shown to require the presence of both Ca2+ and calmodulin when acto-thiophosphorylated myosin was used. These observations suggest a possible functional role of fodrin in the regulation of smooth muscle contraction and demonstrate an effect on Ca2+ and calmodulin on fodrin function.     

Cellular Origin and Biosynthesis of Rat Optic Nerve Proteins: A Two-Dimensional Gel Analysis

High resolution 2DGE (two-dimensional gel electrophoresis) was used to characterize neuronal and glial proteins of the rat optic nerve, to examine the phases of intraaxonal transport with which the neuronal proteins are associated, and to identify the ribosomal populations on which these proteins are synthesized. Neuronal proteins synthesized in the retinal ganglion cells were identified by injecting the eye with L-[35S]methionine, followed by 2DGE analysis of fast and slow axonally transported proteins in particulate and soluble fractions. Proteins synthesized by the glial cells were labeled by incubating isolated optic nerves in the presence of L-[35S]methionine and then analyzed by 2DGE. A number of differences were seen between filamentous proteins of neurons and glia. Most strikingly, proteins in the alpha- and beta-tubulin region of the 2D gels of glial proteins were distinctly different than was observed for axonal proteins. As expected, neurons but not glia expressed neurofilament proteins, which appeared among the slow axonally transported proteins in the particulate fraction; significant amounts of the glial filamentous protein, GFA, were also labeled under these conditions, which may have been due to transfer of amino acids from the axon to the glial compartment. The fast axonally transported proteins contained relatively large amounts of high-molecular-weight acidic proteins, two of which were shown to comigrate (on 2DGE) with proteins synthesized by rat CNS rough microsomes; this finding suggests that rough endoplasmic reticulum may be a major site of synthesis for fast transported proteins. In contrast, the free polysome population was shown to synthesize the principal components of slow axonal transport, including tubulin subunits, actin, and neurofilament proteins.     

Cleavage of cytoskeletal proteins by caspases during ovarian cell death: evidence that cell-free systems do not always mimic apoptotic events in intact cells

Several lines of evidence support a role for protease activation during apoptosis. Herein, we investigated the involvement of several members of the CASP (cysteine aspartic acid-specific protease; CED-3- or ICE-like protease) gene family in fodrin and actin cleavage using mouse ovarian cells and HeLa cells combined with immunoblot analysis. Hormone deprivation-induced apo-ptosis in granulosa cells of mouse antral follicles incubated for 24 h was attenuated by two specific peptide inhibitors of caspases, zVAD-FMK and zDEVD-FMK (50-500 microM), confirming that these enzymes are involved in this paradigm of cell death. Proteolysis of actin was not observed in follicles incubated in vitro while fodrin was cleaved to the 120 kDa fragment that accompanies apoptosis. Fodrin, but not actin, cleavage was also detected in HeLa cells treated with various apoptotic stimuli. These findings suggest that, in contrast to recent data, proteolysis of cytoplasmic actin may not be a component of the cell death cascade. To confirm and extend these data, total cell proteins collected from mouse ovaries or non-apoptotic HeLa cells were incubated without and with recombinant caspase-1 (ICE), caspase-2 (ICH-1) or caspase-3 (CPP32). Immunoblot analysis revealed that caspase-3, but not caspase-1 nor caspase-2, cleaved fodrin to a 120 kDa fragment, wheres both caspases-1 and -3 (but not caspase-2) cleaved actin. We conclude that CASP gene family members participate in granulosa cell apoptosis during ovarian follicular atresia, and that collapse of the granulosa cell cytoskeleton may result from caspase-3-catalyzed fodrin proteolysis. However, the discrepancy in the data obtained using intact cells (actin not cleaved) versus the cell-free extract assays (actin cleaved) raises concern over previous conclusions drawn related to the role of actin cleavage in apoptosis.     

Fodrin is part of a filamentous cortical sheath of the detergent resistant cytoskeleton of cultured cells before and after cytochalasin treatment

Cytoskeletons of cultured cells prepared under mild conditions in the presence of "stabilization' buffer contain most of the fodrin present in the cells. The fodrin in these cytoskeletons was localized by immunofluorescence microscopy and found to be present in a cortical sheath of fine filaments. In general, the filamentous distribution showed no correspondence with actin bundles as revealed by double-label fluorescence microscopy. However, in cells with large and abundant stress fibers, some colocalization of fodrin with actin bundles was seen. Treatment of cells with either cytochalasin A or D caused disorganization of the actin bundles whereas fodrin still showed a filamentous distribution in cytoskeletons of the cytochalasin-treated cells. Implications of these results for the organization of the fodrin-containing sheath of cultured cells is discussed.     

Fodrin in Centrosomes: Implication of a Role of Fodrin in the Transport of Gamma-Tubulin Complex in Brain

Gamma-tubulin is the major protein involved in the nucleation of microtubules from centrosomes in eukaryotic cells. It is present in both cytoplasm and centrosome. However, before centrosome maturation prior to mitosis, gamma-tubulin concentration increases dramatically in the centrosome, the mechanism of which is not known. Earlier it was reported that cytoplasmic gamma-tubulin complex isolated from goat brain contains non-erythroid spectrin/fodrin. The major role of erythroid spectrin is to help in the membrane organisation and integrity. However, fodrin or non-erythroid spectrin has a distinct pattern of localisation in brain cells and evidently some special functions over its erythroid counterpart. In this study, we show that fodrin and γ-tubulin are present together in both the cytoplasm and centrosomes in all brain cells except differentiated neurons and astrocytes. Immunoprecipitation studies in purified centrosomes from brain tissue and brain cell lines confirm that fodrin and γ-tubulin interact with each other in centrosomes. Fodrin dissociates from centrosome just after the onset of mitosis, when the concentration of γ-tubulin attains a maximum at centrosomes. Further it is observed that the interaction between fodrin and γ-tubulin in the centrosome is dependent on actin as depolymerisation of microfilaments stops fodrin localization. Image analysis revealed that γ-tubulin concentration also decreased drastically in the centrosome under this condition. This indicates towards a role of fodrin as a regulatory transporter of γ-tubulin to the centrosomes for normal progression of mitosis.     

Low intracellular pH induces redistribution of fodrin and instabilization of lateral walls in MDCK cells.

We have studied the effect of intracellular pH on the establishment and maintenance of the cellular polarity in MDCK cells by utilizing nigericin which causes lowering of the cytoplasmic pH. At pH below 6.5, MDCK cells lost their polarized morphology and became roundish, with an increased apical area and shortened and unstable lateral walls. The lateral wall marker proteins uvomorulin and Na,K-ATPase remained segregated to the lateral walls in the acidified cells, as shown by immunofluorescence microscopy. Fodrin, on the other hand, was released from its normal basolateral residence and was found in the cytoplasm. Actin, which normally co-localizes with fodrin along the basolateral walls, showed a dotty distribution in the cytoplasm of acidified cells, while stress fibers remained intact. Microtubular network appeared flattened, but the Golgi complex retained its apical position. The pH change-induced alterations were readily reversible, as the normal basal-apical polarity (columnar shape, distinct apical and lateral domains with apposing and stiff lateral membranes) was reformed within 10 minutes after restoring the normal pH gradient across the cell membrane. This coincided with the translocation of fodrin from the cytoplasm to the lateral walls. The results show that lowering of intracellular pH leads to temporary segregation of fodrin from the other components of the membrane skeleton assembly, and that association of fodrin with the lateral walls seems to be a prerequisite for their close apposition and for the maintenance of normal basal-axial polarity.     

Changes in axonally transported proteins during axon regeneration in toad retinal ganglion cells

In an effort to understand the regulation of the transition of a mature neuron to the growth, or regenerating, state we have analyzed the composition of the axonally transported proteins in the retinal ganglion cells of the toad Bufo marinus after inducing axon regeneration by crushing the optic nerve. At increasing intervals after axotomy, we labeled the retinal ganglion cells with [35S]methionine and subsequently analyzed the labeled transported polypeptides in the crushed optic nerve by means of one- and two-dimensional electrophoretic techniques. The most significant conclusion from these experiments is that, while the transition from the mature to the regenerating state does not require a gross qualitative alteration in the composition of axonally transported proteins, the relative labeling of a small subset of rapidly transported proteins is altered dramatically (changes of more than 20-fold) and reproducibly (more than 30 animals) by axotomy. One of these growth-associated proteins (GAPs) was soluble in an aqueous buffer, while three were associated with a crude membrane fraction. The labeling of all three of the membrane-associated GAPs increased during the first 8 d after axotomy, and they continued to be labeled for at least 4 wk. The modulation of these proteins after axotomy is consistent with the possibility that they are involve in growth-specific functions and that the altered expression of a small number of genes is a crucial regulatory event in the transition of a mature neuron to a growth state. In addition to these selective changes in rapidly transported proteins, we observed the following more general metabolic correlates of the regeneration process: The total radioactive label associated with the most rapidly transported proteins (groups I and II) increased three to fourfold during the first 8 d after the nerve was crushed, while the total label associated with more slowly moving proteins (group IV) increased about 10-fold during this same period. Among these more slowly transported polypeptides, five were observed whose labeling increased much more than the average. Three of these five polypeptides resemble actin and alpha- and beta-tubulin in their electrophoretic properties.     

Localization of fodrin during fertilization and early development of sea urchins and mice

Fodrin, a spectrin-like protein, is localized in gametes, zygotes, and embryos from sea urchins and mice. Mammalian fodrin comprises two polypeptides with molecular weights of approximately 240 kDa (alpha) and 235 kDa (beta). An antibody specific for mammalian alpha-fodrin cross-reacted with a 240-kDa polypeptide from sea urchin egg extracts. This indicates that sea urchins contain a protein of similar electrophoretic mobility and immunological properties to mammalian alpha-fodrin. When this antibody was used to stain the sea urchin gametes with indirect immunofluorescence, fodrin-specific fluorescence was localized to the acrosome of the sperm and was distributed over the entire egg near the surface in a punctate pattern similar to the distribution of polymeric actin. During sperm incorporation, the fodrin-specific fluorescence is found at the site of sperm incorporation, in the fertilization cone. After fertilization, the intensity of fodrin fluorescence increases. During mitosis and cytokinesis in sea urchins, the entire surface of the egg remains stained; the cleavage furrow also was stained but no more intensely than was the rest of the egg surface. Antibody labeling with colloidal gold followed by electron microscopy showed that fodrin was loated in the cytoplasm immediately beneath the plasma membrane. In unfertilized mouse oocytes, both actin and fodrin were stained most intensely beneath the membrane adjacent to the meiotic spindle. After insemination, the cell surfaces of the pronucleate egg and the second polar body were stained; however, the actin matrix surrounding the apposed pronuclei did not bind the fodrin antibody. During cytokinesis in the mouse, the cleavage furrow stained more intensely than did the rest of the egg cortex, and in embryos the cell borders were delineated. These results indicate that organisms as unrelated to mammals as sea urchins have fodrin-like proteins; the rearrangements of such proteins suggest that they participate in the actin-mediated events at the cell surface during fertilization and early development in both mice and sea urchins.     

Characterization of fodrin phosphorylation by spleen protein tyrosine kinase

Fodrin, an actin and calmodulin binding and spectrin-like protein present in many nonerythrocyte tissues, could be phosphorylated up to more than 1.5 mol of phosphate/mol of protein by a highly purified non-receptor-associated protein tyrosine kinase from bovine spleen. The protein phosphorylation was not affected by Ca2+/calmodulin or by F-actin. Km and Vmax values of the reaction were 91 nM and 0.35 nmol of P2 min-1 (mg of kinase)-1, respectively. Both subunits A and B of fodrin were phosphorylated, with the rate of subunit A phosphorylation much greater than that of subunit B phosphorylation. Tryptic phosphopeptide mapping of the phosphorylated subunits suggested that there were three major phosphorylation sites in subunit A and one in subunit B. Phosphotyrosylfodrin could be dephosphorylated by the calmodulin-stimulated phosphatase (calcineurin) in the presence of activating metal ions; Ni2+ was a much more effective activator than Mn2+ for this reaction. Fodrin phosphorylation by the spleen protein tyrosine kinase did not appear to alter the actin and calmodulin binding properties of the protein. On the other hand, the calmodulin-dependent stimulation of smooth muscle actomyosin Mg2+-ATPase by fodrin was enhanced by 101% +/- 3% (n = 3) upon fodrin phosphorylation. Ni2+-calcineurin, which was shown to effectively dephosphorylate the phosphotyrosyl residues on fodrin, could reverse the phosphorylation-enhanced Mg2+-ATPase stimulatory activity of fodrin.     

The spectrin-related molecule, TW-260/240, cross-links the actin bundles of the microvillus rootlets in the brush borders of intestinal epithelial cells

Previous studies have shown that molecules related to erythrocyte spectrin are present in the cortical cytoplasm of nonerythroid cells. We report here the localization by immunoelectron microscopy of one such molecule, TW-260/240, in the brush border of intestinal epithelial cells. Using highly specific antibodies against TW-260 and TW-240 as well as antibodies against fodrin, another spectrinlike molecule, we have found that the TW-260/240 molecules are displayed between rootlets at all levels of the terminal web. Occasionally, extended structures appear labeled suggestive of the fine filaments known to cross-link actin bundles. These results are in line with previous in vitro studies showing that TW-260/240 binds to, and cross-links, actin filaments. The results are discussed in terms of a model in which rootlets are immobilized in the terminal web in a matrix of TW-260/240.     

Immunolocalization of a spectrin-like protein (fodrin) in pancreatic acinar cells

A spectrin-like protein (fodrin) was localized in porcine pancreas using an immunoperoxidase procedure with antibodies raised against erythrocyte spectrin. Fodrin was primarily associated with the cell plasma membrane although some was also detectable in the cytoplasm of the acinar cells. The membrane labelling of the acinar cells was uneven such that the lateral and basal membranes were strongly labelled by anti-spectrin antibodies whereas the apical membranes were poorly labelled. The implications of the results to secretion and to the occurrence of specific membrane domains are discussed.     

Dynamic organization of cortical actin filaments during the ooplasmic segregation of ascidian Ciona eggs

Spatial reorganization of cytoplasm in zygotic cells is critically important for establishing the body plans of many animal species. In ascidian zygotes, maternal determinants (mRNAs) are first transported to the vegetal pole a few minutes after fertilization and then to the future posterior side of the zygotes in a later phase of cytoplasmic reorganization, before the first cell division. Here, by using a novel fluorescence polarization microscope that reports the position and the orientation of fluorescently labeled proteins in living cells, we mapped the local alignments and the time-dependent changes of cortical actin networks in Ciona eggs. The initial cytoplasmic reorganization started with the contraction of vegetal hemisphere approximately 20 s after the fertilization-induced [Ca²⁺] increase. Timing of the vegetal contraction was consistent with the emergence of highly aligned actin filaments at the cell cortex of the vegetal hemisphere, which ran perpendicular to the animal–vegetal axis. We propose that the cytoplasmic reorganization is initiated by the local contraction of laterally aligned cortical actomyosin in the vegetal hemisphere, which in turn generates the directional movement of cytoplasm within the whole egg.     

Calcium-dependent peripheral localization of 4.1-like proteins and fodrin in cultured human keratinocytes

Summary— Recently, several proteins immunologically related to erythrocyte membrane skeletal proteins, such as protein 4.1 and fodrin (non-erythroid spectrin), have been found in keratinocytes. In the present study, in order to investigate the roles of these proteins in cell-cell contact, we analyzed the distribution of non-erythroid protein 4.1, β-fodrin and actin in cultured human keratinocytes at low (0.15 mM) and standard (1.85 mM) Ca²⁺ concentrations. Immunofluorescence microscopy revealed that immunoreactive forms of protein 4.1, β-fodrin and actin filaments were present in the cytoplasm of cells cultured in low Ca²⁺ medium, while in cells in the standard Ca²⁺ medium, these proteins were localized at the cell boundary and partially in the cytoplasm. When cells in the low-Ca²⁺ medium were treated with 100 nM 12-O-tetradecanoylphorbol-13-acetate (TPA) for 1 h, these proteins were also present at the cell boundary. Increasing extracellular Ca²⁺ concentration from low to standard in the medium induces cell-cell contact among the cultured human keratinocytes, accompanied by the translocation of protein 4.1 and β-fodrin from the cytoplasm to the membrane. On the basis of the present study, movement of membrane skeletal proteins from the cytosol to the membrane suggests that either these proteins or the membrane skeletal lattice plays an important role in the regulation of cell-cell intergigitations in response to changes in the Ca²⁺ concentrations in culture medium, and that phosphorylation of these skeletal proteins might be involved in the regulation of the membrane skeletal proteins of keratinocytes in response to Ca²⁺.     

Relation between cell activity and the distribution of cytoplasmic actin and myosin

We documented the activity of cultured cells on time-lapse videotapes and then stained these identified cells with antibodies to actin and myosin. This experimental approach enabled us to directly correlate cellular activity with the distribution of cytoplasmic actin and myosin. When trypsinized HeLa cells spread onto a glass surface, the cortical cytoplasm was the most actively motile and random, bleb-like extensions (0.5-4.0 micrometer wide, 2-5 micrometer long) occurred over the entire surface until the cells started to spread. During spreading, ruffling membranes were found at the cell perimeter. The actin staining was found alone in the surface blebs and ruffles and together with myosin staining in the cortical cytoplasm at the bases of the blebs and ruffles. In well-spread, stationary HeLa cells most of the actin and myosin was found in stress fibers but there was also diffuse antiactin fluorescence in areas of motile cytoplasm such as leading lamellae and ruffling membranes. Similarly, all 22 of the rapidly translocating embryonic chick cells had only diffuse actin staining. Between these extremes were slow-moving HeLa cells, which had combinations of diffuse and fibrous antiactin and antimyosin staining. These results suggest that large actomyosin filament bundles are associated with nonmotile cytoplasm and that actively motile cytoplasm has a more diffuse distribution of these proteins.     

Effect of phosphorylation by cyclic AMP-dependent protein kinase on the smooth muscle actomyosin Mg2+-ATPase stimulatory activity of fodrin

Fodrin, a non-erythrocyte spectrin-like protein, has been purified from bovine brain and found to be phosphorylated by the cyclic AMP-dependent protein kinase with a maximal stoichiometry of 1.02 +/- 0.06 mol of phosphate/mol of fodrin dimer (n = 4). This phosphorylation was not affected by the presence of actin and calmodulin. The phosphorylation of fodrin was found to occur exclusively at serine residues on the beta subunit. Two-dimensional thin layer electrophoresis and chromatography of a tryptic digest of phosphorylated fodrin showed one major phosphopeptide and a few minor ones. We have previously reported that nonphosphorylated fodrin is capable of stimulating the smooth muscle actomyosin Mg2+-ATPase by 50-70% under a well-defined set of conditions such as a critical fodrin concentration and an optimal preincubation time (Wang, C., Ngai, P.K., Walsh, M.P., and Wang, J.H. (1987) Biochemistry 24, 1110-1117). We now report that phosphorylation of fodrin completely eliminates this stimulatory effect. However, phosphorylation of fodrin was able to compete with nonphosphorylated fodrin to result in the abolition of the stimulatory effect. Similarly, nonphosphorylated fodrin could overcome the inhibitory effect created by phosphorylated fodrin. The present results support the suggestion that the stimulation of the smooth muscle actomyosin Mg2+-ATPase by fodrin may be a physiological phenomenon and cyclic AMP may serve as a regulator for this effect.     

Peculiar Distribution of Fodrin in Fat-Storing Cells

Fat-storing cells (FSCs) show unique morphology containing many lipid droplets in the cytoplasm. In this study, we found that a membrane skeletal protein, fodrin, shows peculiar distribution in FSCs of rat liver. By immunofluorescence microscopy of FSCs in culture, intense labeling for fodrin was seen as coarse filaments in the cytoplasm. Especially in FSCs isolated from vitamin A-treated rats, the labeling was often seen as many small rings in the cytoplasm. In contrast, labeling for fodrin in human fibroblasts or rat adipocytes in culture was seen diffusely in the cell cortex. Distribution of actin, tubulin, vimentin, and desmin in FSCs was also examined, but none of them appeared correlated with fodrin. By immunoelectron microscopy using nanogold labeling with silver enhancement, positive labeling for fodrin was seen around some lipid droplets in FSCsin vivo.We assume that the peculiar distribution of fodrin may be related to the morphological characteristics of FSCs.     

Lack of physiological stimulation induces decreased proteolytic activity in nerve terminals

The effect of optic nerve transsection on proteolytic degradation of axonally transported proteins in the superior colliculus of the rabbit was studied. Proteolysis of labeled proteins was determined in vitro in small pieces of the superior colliculus. Within 2 hours after sectioning the optic nerve there was a decreased degradation of slowly transported labeled proteins in the nerve terminals in the superior colliculus.Special Issue dedicated to Prof. Holger Hydén.     

Segregation of two spectrin forms in the chicken optic system: a mechanism for establishing restricted membrane-cytoskeletal domains in neurons

The chicken optic system contains a brain-specific form of spectrin (alpha gamma-spectrin or fodrin) as a major membrane-associated, axonally transported cytoskeletal protein. We show here that the chicken optic system also contains an erythrocyte-specific form of spectrin (alpha beta' beta-spectrin), which has a more restricted distribution; it is confined to the plasma membrane of dendrites and cell bodies of retinal ganglion cells, is absent from the optic nerve fibers, and is not axonally transported from the retina into the optic nerve. During development of the optic system, the expression of alpha gamma-spectrin is constitutive in all cell types. On the other hand, the accumulation of alpha beta' beta-spectrin is detected in only the ganglion cells, and at a time in development which coincides with the phase of synaptogenesis. These results indicate the existence of a developmentally regulated mechanism that topologically segregates the erythroid and brain forms of spectrin from each other, and the former from axonal transport, and suggest that erythroid spectrin may be involved in establishing restricted membrane-cytoskeletal domains in neurons during synaptogenesis, and maintaining them in the adult cell.