In the developing nervous system migrating neurons and growing axons are guided by diffusible and/or substrate-bound cues, such as extracellular matrix-associated laminin. In a previous work we demonstrated that laminin molecules could self-assemble in two different manners, giving rise to matrices that could favor either neuritogenesis or proliferation of cortical precursor cells. We investigated whether the ability of astrocytes to promote neuritogenesis of co-cultivated neurons was modulated by the assembling mode of the laminin matrix secreted by them. We compared the morphologies and neuritogenic potentials of laminin deposited by in vitro-differentiated astrocytes obtained from embryonic or neonatal rat brain cortices. We showed that, while permissive astrocytes derived from embryonic brain produced a flat laminin matrix that remained associated to the cell surface, astrocytes derived from newborn brain secreted a laminin matrix resembling a fibrillar web that protruded from the cell plane. The average neurite lengths obtained for E16 neurons cultured on each astrocyte layer were 198±22 and 123±13 μm, respectively. Analyses of surface-associated electrostatic potentials revealed that embryonic astrocytes presented a pI of -2.8, while in newborn cells this value was -3.8. Removal of the sialic acid groups on the embryonic monolayer by neuraminidase treatment led to the immediate release of matrix-associated laminin. Interestingly, laminin reassembled 1 hour after neuraminidase removal converted to the features of the newborn matrix. Alternatively, treatment of astrocytes with the cholesterol-solubilizing detergent methyl-β-cyclodextrin also resulted in release of the extracellular laminin. To test the hypothesis that sialic-acid-containing lipids localized at cholesterol-rich membrane domains could affect the process of laminin assembly, we devised a cell-free assay where laminin polymerization was carried out over artificial lipid films. Films of either a mixture of gangliosides or pure ganglioside GT1b induced formation of matrices of morpho-functional features similar to the matrices deposited by embryonic astrocytes. Conversely, films of phosphatidylcholine or ganglioside GM1 led to the formation of bulky laminin aggregates that lacked a defined structure. We propose that the expression of negative lipids on astrocytes can control the extracellular polymerization of laminin and, consequently, the permissivity to neuritogenesis of astrocytes during development.

During development of the nervous system, neuron precursor cells must migrate, differentiate and establish proper connections with defined targets. Regulation of these three stages requires multiple extracellular signals, which are provided upon interaction with other cells, instructive substrates and/or gradients of soluble molecules (Bixby and Harris, 1991; Tessier-Lavigne and Goodman, 1996; Höpker et al., 1999; Pires Neto et al., 1999). The large extracellular glycoprotein, laminin (Colognato and Yurchenco, 2000), has been implicated in the morphogenesis of the nervous system due to: (1) its increased expression along CNS developing pathways (Morissette and Carbonetto, 1995; Luckenbill-Edds, 1997); (2) CNS developmental impairments provoked by either spontaneous or targeted mutations of laminin genes (Arahata et al., 1995; Ryan et al., 1996); and (3) abundant evidence for its ability to promote cell migration, differentiation and axonal guidance in vitro (Liesi, 1990; Hunter and Brunken, 1997). Conversely, mice knocked out for one of the most ubiquitous laminin subunits, the γ1 chain, do not survive beyond embryonic day 5.5, a fact that has prevented direct observation of presumptive failures in neural development associated to laminin depletion (Smyth et al., 1999).

In adult basement membranes, laminin is organized as a continuous mesh-like network, spatially separated from the bulk extracellular matrix. During embryonic development, however, laminin appears in the CNS either as cell-surface-associated deposits or arranged in large polymeric aggregates scattered within the extracellular space (Luckenbill-Edds, 1997). A detailed study of laminin aggregates found in the developing brain described the occurrence in vivo of four different structural patterns (Zhou, 1990). Accordingly, part of laminin accumulated in the embryonic nervous system was reported to be extractable with physiological buffers, whereas the protein found in mature basement membranes is not (Edgar, 1991; Kuecherer-Ehret et al., 1990). More recently, using whole mounts of newborn retina, we have reported that the laminin deposited on the inner limiting membrane at peripheral retina corresponded to a homogeneous matrix, while the laminin found in central retina lacked any distinguishable structural organization (Freire et al., 2002). Because the retina matures from center to periphery, we proposed that the transition between the two types of matrices could be related to pacing of retinal development. Corroborating this hypothesis, we and others have shown that the supramolecular structure of laminin, i.e. the organization of laminin trimers into a polymeric array, modulates neuritogenesis, as well as migration and proliferation (Garcia-Abreu et al., 1995a; Garcia-Abreu et al., 1995b; Freire et al., 2002). It is thus tempting to hypothesize that the differences in the supramolecular organization of laminin, constitutes an additional developmental cue in the CNS. Here we approached the question of how the diversity of laminin deposits found in vivo can be generated by astrocytes.

Astrocytes are thought to be a major source of extracellular matrix (ECM) molecules in the CNS (Pindzola et al., 1993; Powell et al., 1997). These cells, even when differentiated in vitro, have been shown to preserve their capacity to support neuritogenesis, as well as their ability to deposit extracellular matrix (Grierson et al., 1990; Meiners et al., 1995; Garcia-Abreu et al., 1995a; Garcia-Abreu et al., 1995b; Martinez and Gomes, 2002). We derived astrocytes from rat cerebral cortices at distinct ages and characterized their associated laminin matrices for both morphology and neuritogenic potential. We showed that cells isolated during embryonic life supported neuritogenesis with higher efficiency than cells isolated from newborn animals. Such difference in the neuritogenic potential of distinct astrocyte monolayers correlated with the organization of the laminin matrices deposited by the glial cells. In addition, we demonstrated that laminin secreted by embryonic astrocytes acquired the properties of laminin secreted by newborn-derived cells after digestion of sialic acid groups or, alternatively, upon detergent-induced disruption of membrane-associated lipid domains.

Astrocyte primary cultures

Astrocyte primary cultures were prepared from cerebral cortices of 16- or 18-day-old embryos (E16; E18), newborn (P0) or 6-day old rats (P6) as previously described (Gomes et al., 1999). Briefly, after decapitation, brain structures were removed and the meninges carefully stripped off. Tissues were washed in PBS-0.6% glucose (Glucose, Sigma, St Louis, MO) and dissociated into single cells in a medium consisting of Dulbecco's minimum essential medium (DMEM) and nutrient mixture F12 (DMEM/F12, Gibco-Invitrogen, Grand Island, NY), enriched with glucose (33 mM), glutamine (2 mM) and sodium bicarbonate (3 mM). Cells were seeded at a density of 2.0×105 cells/cm2 either on glass coverslips or on 25 cm2 culture flasks (Costar) previously coated with polyornithine (1.5 μg/ml, Mr 41,000, Sigma), in DMEM/F12 supplemented with 10% fetal calf serum (Gibco-Invitrogen). Cultures were maintained at 37°C in a humidified 5% CO2 chamber. Culture medium was changed for the first time at 24 hours after plating, and then at 3 day intervals. Typically, cultures reached confluence after 10 days.

Treatment with neuraminidase and methyl-β-cyclodextrin

Confluent astrocyte monolayers were treated for 30 minutes at 37°C either with 0.1 mU/ml neuraminidase from Vibrio cholerae (Sigma) in sodium acetate buffer (50 mM, pH 5.5, containing 9 mM CaCl2 and 154 mM NaCl) or with 0.12% (w/v) methyl-β-cyclodextrin (CD) (Sigma) in Tris-HCl buffer (50 mM, pH 7.4, containing 25 mM KCl, 5 mM MgCl2 and 1 mM EDTA). After washing with PBS, monolayers were used as substrates for culture of E16 neurons or directly prepared for laminin immunostaining.

Co-cultures

After reaching confluence, glial monolayers were washed 3 times and incubated for 2 additional days in serum-free medium. After this period, cells freshly dissociated from 16-day-old embryonic rat cerebral cortices were plated at 5×104 cells/cm2 onto the monolayers to generate neurons. After 24 hours in the absence of serum, co-cultures were fixed and prepared for evaluation of neurite outgrowth by immunostaining with an antibody against the neuronal marker, Tau.

Immunocytochemistry

To visualize cytoskeleton markers, both glial cultures and co-cultures were fixed with 4% paraformaldehyde for 20 minutes, washed 3 times with PBS, and permeabilized for 5 minutes at room temperature with 0.2% Triton X-100, containing 1% bovine serum albumin (BSA). After permeabilization, cells were again washed 3 times with PBS and blocked for 30 minutes with BSA 5%. Immunocytochemistry was performed as previously described (Gomes et al., 1999; Freire et al., 2002). Neuronal morphology was observed using a rabbit anti-Tau antibody (Sigma; 1:200 dilution) and astrocytes, using a rabbit anti-glial fibrillary acidic protein (GFAP) (Dako, Carpinteria, CA, 1:50 dilution). To visualize extracellular laminin, astrocyte monolayers were quickly fixed with paraformaldehyde (4 minutes), washed 3 times, blocked with BSA 5% and incubated with rabbit anti-laminin (Sigma; 1:30 dilution). It should be noted that analyses of laminin were performed on non-permeabilized confluent astrocyte monolayers, which means that only the protein deposited at the upper face of the culture was available for recognition by the antibody. The secondary antibody used in all cases was a Cy3-labeled goat anti-rabbit (Sigma; 1:5000 dilution), incubated for 2 hours at room temperature. Negative controls, performed by omitting the primary antibody, showed no reactivity. Coverslips were mounted in glycerolpropylgallate and observed in a Zeiss Axioplan microscope.

Morphometry and statistical analyses

Cells stained for the neuronal marker Tau were photographed in a Nikon TE 300 fluorescence microscope. Photos were scanned and total neurite lengths analyzed using the Sigma Scan Pro Software (Jandel Scientific). A total of 100 neurons in six or seven fields chosen randomly were considered independently of their sizes. Each experiment was repeated at least three times. Statistical significance was evaluated using the Student's t-test. Error bars represent standard errors.

Preparation of laminin matrices

Phosphatidylcholine (bovine brain), GT1b (bovine brain), and Gmix (type IV: bovine brain) were purchased from Sigma and ganglioside GM1, from Calbiochem (La Jolla, CA). Chloroform/metanol solutions of gangliosides [either pure or in mixtures of varying molar ratios of ganglioside to phosphatidylcholine (PC)] were evaporated under nitrogen, resuspended in 20 mM Tris-HCl, pH 7 (final concentration of 1 mM) and vigorously vortexed. Small unilamellar vesicles were formed by sonicating the turbid suspensions using a Branson sonifier (Sonic Power Company, Danbury, CT) equipped with a titanium microtip probe. Sonication was carried out in an ice bath, alternating cycles of 20 seconds at 50% full power with 60 seconds resting intervals until a transparent solution was obtained (usually 8 cycles). After sonication, vesicles were centrifuged for 4 minutes in an Eppendorf centrifuge to remove titanium released from the probe and filtered in disposable Millipore filter (pore 0.22 μm). Drops of lipid suspensions were placed onto glass coverslips and incubated for 20 minutes. After sedimentation, 1 mM CaCl2 was added directly to the drops to initiate fusion of the vesicles, which proceeded for 1 additional hour at 37°C. Laminin (natural mouse laminin isolated from Engelbreth-Holm-Swarm tumour, Invitrogen), was diluted to a final concentration of 50 μg/ml in either 20 mM sodium acetate, pH 4 or Tris-HCl, pH 7, each containing 1mM CaCl2. Aliquots of 200 μl were placed immediately after dilution onto coverslips, pre-coated or not with lipids, and incubated at 37°C for 12 hours. Laminin-coated coverslips were then washed 3 times with PBS and used either directly for immunostaining or as substrates for neuronal cultures.

Measurement of zeta-potential (ξ)

Astrocytes derived from embryonic or newborn rats were mechanically detached from monolayers, centrifuged, washed and suspended in buffers (20 mM) of different pHs, varying between 3.0 and 11.0: Glycin pH 3, sodium acetate pH 5, Tris pH 7 or 9 or CAPES pH 11. In some experiments, cells were treated with 0.1 U/ml of neuraminidase from Vibrio cholerae (Sigma) for 30 minutes at room temperature before ξ measurements. Cells suspended in each of the solutions were submitted to ξ measurements at 25°C, as previously described (Silva Filho et al., 1987), using a Zeta-Meter (PenKem) whose electrical current was kept at 100 Volts. Typically, no more than 2 seconds were spent for each individual measurement. Values of ξ reported here represent means of three individual populations of 10-12 cells.

Characterization of laminin matrices deposited by astrocyte monolayers.

Astrocyte cultures were prepared from rat cerebral cortices at four developmental ages (E16, E18, P0 and P6). Analyses of the laminin matrices secreted by each monolayer at their medium-exposed surfaces (see Materials and Methods for details) showed that either embryonic or postnatal cultures produced distinct matrices (Fig. 1). Laminin deposited by either E16 or E18 astrocytes remained associated to cell surfaces, displaying an approximately periodical arrangement, characterized by geometrical figures (Fig. 1A,E). Conversely, cells isolated either at P0 or P6 produced fibrous laminin deposits that protruded from the cell surface, projecting out to the extracellular space (Fig. 1B,F). Panels D and G, in Fig. 1, compare the two matrices at higher magnification. Such comparison also reveals that the P0 matrix contains compact struts of aggregated laminin (G), while the embryonic matrix forms a delicate mesh (D).

Neuritogenesis on astrocyte monolayers

To investigate whether the differences in secreted laminin correlated with differences in permissivity for neuritogenesis, neuronal cells isolated from E16 brains were plated over confluent astrocytic monolayers (similar to those whose associated laminin matrices were visualized in Fig. 1). After 24 hours in the absence of serum, neurons were observed using immunolabeling for the neuronal marker Tau (Fig. 2). Neuritogenesis was significantly more pronounced on the E16 than on the P0 monolayer. Quantitative analyses showed that the average sizes of neurites were 198±22 and 123±13 μm on embryonic and newborn monolayers, respectively (Fig. 2C). Concomitantly, 86% of the neuronal population exhibited 2 or more neurites per cell body when plated on E16 astrocytes, while less than 50% presented this feature on P0 carpets (Fig. 2D). Interestingly, neurons plated on newborn astrocytes had a tendency to form small clusters as indicated by the arrows in panel 2B. Such clusters had previously been observed by us on artificial laminin matrices polymerized at neutral pH, where they were found to correspond to proliferating cells (Freire et al., 2002). Together, these results suggest a correspondence between the appearance of laminin aggregates deposited by astrocytes and their ability to induce neuritogenesis.

Electrochemical potentials of cultivated astrocytes

We have previously reported that morphologies and neuritogenic potentials of artificial laminin polymers were controlled by the pH of the dilution buffer (Freire et al., 2002). Here we investigated whether the distinguished morphofunctional properties of the extracellular matrices produced by astrocyte cultures could be determined by local pH (a pH restricted to the microenvironment of the external face of the plasma membrane). To assess the actual pH at the surface of astrocytes, we designed experiments for the direct measurement of zeta potentials (ξ) of cells in solution. The resulting values were used to calculate the pH at the surface of astrocytes (pHsurface), using the equation pHsurface=pHbulk+ Fξ/3.303RT (adapted from van der Goot et al., 1991). In neutral buffer P0 astrocytes presented ξ=-11.0 mV, which corresponds to a negligible difference of 0.019 pH unit between the cell surface and the bulk solution. Conversely, E16 astrocytes presented a significantly lower ξ of -14.0 mV. Using this value, we found a difference between surface and bulk pHs of 0.236 unit. It is important to note that the latter value represents an average of potentials sampled by different regions of a non-homogeneous cell membrane. Considering that some negative molecules in plasma membranes, such as glycolipids and proteoglycans, tend to segregate into cholesterol-rich membrane rafts (Carey et al., 1996; Hakomori et al., 1998), one should expect local deviations from bulk pH at these regions to be actually larger than calculated here.

Data concerning ξ measurements at increasing buffer pHs are summarized in Fig. 3. Both the shape and the slope of the curves obtained for each cell type were similar. Embryonic and newborn astrocytes presented negatively charged surfaces at physiological bulk pH. As the pH of the bulk solution was reduced, ξ progressively increased, reaching values as high as +4 and +6 mV. Zeta potentials were zero at pH 2.8 and at pH 3.8 in embryonic and newborn cells, respectively. Such values corresponded to isoelectrophoretic points (IEP) at each situation. Previous treatment of E16 and P0 cells with neuraminidase for 30 minutes in neutral buffer led to increases in mean ξ values. Such effect was more pronounced in E16 cells, where ξ increased from -14.0 to -7.5 mV. Neuraminidase treatment of P0 astrocytes led to a smaller increase of ξ from -11 to -8.2 mV. It is important to note that neuraminidase treated cells reincubated in fresh culture medium for 30 or 60 minutes did not restore the mean ξ values of the controls (Fig. 3, inset).

Effect of neuraminidase on astrocyte monolayers

The finding that the pH at the surface of embryonic astrocytes is lower than the pH at the surface of newborn cells was in agreement with the hypothesis that the morphological differences in the laminin matrices secreted by each cell layer could be determined by their respective surface pHs. We then tested whether the presence of sialic acid contributes to the permissivity of the monolayers and their associated laminin matrices. We treated E16 astrocyte cultures with neuraminidase and analyzed their laminin matrices. Neuraminidase led to immediate loss of immunoreactivity for laminin (Fig. 4A,B). After 1 hour, a time interval not sufficient to restore sialic acid groups (see inset in Fig. 3), cells had already replaced their extracellular laminin (Fig. 4C,D). Interestingly, the appearance of the laminin matrix deposited by neuraminidase-treated E16 astrocytes was very similar to the appearance of the matrices originally deposited by P0 astrocytes (see Fig. 1B).

To evaluate the permissivity of the reassembled matrix, neurons were plated onto E16 astrocyte monolayers 1 hour after neuraminidase removal and neurite outgrowth was analyzed 24 hours later. On neuraminidase-treated embryonic matrices, neurons developed the short neurites and presented the tendency to cell clumping characteristic of the monolayers derived at P0 (Fig. 5; compare with Fig. 2B). These results support the view that the mode of laminin organization on the cellular surface depends on the presence of exposed sialic acid groups.

Effect of lipid films on the formation of artificial laminin matrices

Results presented here suggest that sialic acid-containing molecules can play a role in orienting formation of laminin matrices by cultivated astrocytes. To investigate this possibility directly, we devised a cell-free assay, where we tested whether substrates made of pure sialic-acid-containing lipids would themselves orient the assembly of laminin. Panel A in Fig. 6 shows the appearance of a control laminin matrix assembled directly on a glass coverslip at neutral pH and in the absence of lipids. Large protein aggregates protrude from the optical plane and no organized pattern can be distinguished. Pre-coating of surfaces with phosphatidylcholine (PC) (the neutral phospholipid predominating in biological membranes) led to a subtle decrease in the size of laminin aggregates as compared with the control (Fig. 6B). It is interesting to note that the laminin matrices deposited on PC were still structured, because complete blockade of polymerization, as promoted by EDTA or pH alkalinization (Yurchenco et al., 1985), led to total disappearance of protein aggregates (Fig. 6D,H). Laminin matrices formed on GM1, a ganglioside containing only one sialic acid group, presented a morphology similar to the one assembled on PC (Fig. 6C). Conversely, when ganglioside GT1b containing 3 sialic acid groups, or when a heterogeneous mixture of gangliosides (Gmix) were used, highly structured polymers were obtained (Fig. 6E,F). The patterns seen on these two matrices were comparable to the organization of polymers obtained in the absence of lipids but upon acidification of the aggregation buffer, a condition previously reported to increase laminin neuritogenic potential greatly (Fig. 6G) (Freire et al., 2002).

Morphometric analyses of neurons plated on each matrix revealed that neurites on GT1b and on the ganglioside mixture (Fig. 7B,C) were markedly longer than neurites on PC or GM1 (Fig. 7A). It is interesting to note that the average neurite size on laminin assembled on GT1b or Gmix reached a maximum of 45-50 μm, which is similar to the average size of approximately 50 μm found for neurites extending on laminin assembled directly onto glass at neutral pH (Freire et al., 2002). These values are all below the limit of about 150 μm reported for laminin assembled on glass at acidic pH (Freire et al., 2002). Nevertheless, the observation that the average neurite size obtained in the presence of PC and GM1 (15-25 μm) is much lower than obtained in the absence of lipids suggests that lipids themselves, besides an effect on the orientation of laminin assembly, might also directly inhibit the growth of neurites (Hynds et al., 1997), thus contributing to an overall decrease in the sizes of neurites observed in the assay. Results presented in this section show that the presence of an artificial negative surface can modulate the morphology and the neuritogenic potential of laminin aggregates, thus mimicking the effect of negative groups on the membrane of astrocytes.

Effect of detergents on astrocyte monolayers

Methyl-β-cyclodextrin is a detergent capable of disrupting cholesterol-stabilized lipid domains on the plasma membrane. Conversely, Triton X-100 solubilizes phospholipids, hence leading to cell permeabilization, whereas not affecting cholesterol-stabilized domains. As gangliosides are known to segregate into lipid microdomains on the plasma membrane (Masserini et al., 1988; Hakomori et al., 1998), we have evaluated the effect of methyl-β-cyclodextrin (CD) and Triton X-100 on the laminin matrix associated to the embryonic astrocyte monolayer. Treatment of E16 astrocytes with 1% Triton X-100 for 1 hour permeabilized cells, as revealed by clear intracellular labeling with anti-laminin antibody (Fig. 8B). Conversely, when astrocytes were treated with CD a complete loss of immunoreactivity for laminin was observed (Fig. 8C). This result shows that, although the cell membrane was preserved upon CD treatment (no intracellular labeling), membrane-associated laminin was probably released due to destabilization of cholesterol-rich lipid domains. Considering that laminin receptors, such as β1-subunit-containing integrins and dystroglycan, localize to lipid rafts (Keshet et al., 1999; Keshet et al., 2000; Thorn et al., 2000; Claas et al., 2001; Baron et al., 2003), our interpretation is that anchorage of laminin polymers depends on the integrity of lipid domains on the plasma membrane.

The main contribution of the present work has been to establish a mechanism by which astrocytes could control the morphofunctional properties of the laminin matrices deposited by them. Up until now, it has been known that: (1) the structural organization of laminin matrices could control neuritogenesis (Garcia-Abreu et al., 1995a; Garcia-Abreu et al., 1995b; Freire et al., 2002); (2) distinct laminin polymers coexisted in the developing nervous system (Edgar, 1991; Kuecherer-Ehret et al., 1990; Zhou, 1990); and (3) astrocytes secreted laminin during CNS development (Powell et al., 1997). However, the question about how glial cells could guide the formation of different laminin polymers potentially capable of modulating axonal growth was still open. Previous attempts to address this issue proposed, for instance, that polymers found in vivo at the extracellular matrix (ECM) corresponded to laminin aggregates that resulted from an unbalance in the production of the three chains (Edgar, 1991). Alternatively, differences in the organization of laminin bound to Schwann cells were proposed to reflect the distribution of receptors at the cell surface (Tsiper and Yurchenco, 2002). In this paper, we show that the assembly of laminin into distinct polymers can be guided by the amount of sialic-acid-containing molecules segregated in cholesterol-stabilized microdomains of the plasma membrane. Because laminin is recognized by membrane receptors, which can themselves be localized in lipid rafts (Thorne et al., 2000; Claas et al., 2001), it is possible that laminin patterns actually reflect the organization of their receptors, the distribution of which can be affected by the destabilization of lipid rafts. Although this interpretation cannot be ruled out, the observation that artificial lipid films can modulate the assembly of laminin in the absence of plasma membrane proteins indicates that lipid domains can directly orient laminin self-polymerization. In this regard, a recent paper has shown an example of how an artificial laminin matrix induces the reorganization of integrin receptors in oligodendrocytes (Baron et al., 2003).

Astrocytes are thought to play an essential role in guiding axonal extension in the developing CNS by releasing soluble factors, by exposing membrane-associated adhesion molecules and by depositing ECM molecules (Smith et al., 1986; Smith et al., 1990; Sajin and Steindler, 1994; Powell et al., 1997). Previous studies have associated the morphologies of laminin deposits found on monolayers of cultivated astrocytes with permissivity for neurite outgrowth. Two papers (Garcia-Abreu et al., 1995a,b) showed that astrocytes isolated from either more or less permissive regions of the developing brain retained their original permissivities in vitro and were characterized by distinct ECM deposits. In two other papers (Grierson et al., 1990; Meiners et al., 1995) described and characterized heterogeneity within a single astrocyte monolayer. The two identified subpopulations were named by the authors as `flat' and `rocky' to designate, respectively, the absence or presence of bumps on the surface of monolayers. Our results were qualitatively similar to those reported by Meiners and co-workers, i.e. in both cases uneven surfaces were less permissive for neurite outgrowth than flat ones. In their study, roughness of the surfaces was attributed to distinct expression of the protein tenascin and proteoglycans, whereas no difference in laminin expression between the two identified populations was observed. While their immunofluorescence for laminin seemed not to have detected extracellular labeling, other studies, including ours, clearly demonstrated that invitro-derived astrocytes can present associated laminin matrices (Ard and Bunge, 1988; Garcia-Abreu et al., 1995a; Garcia-Abreu et al., 1995b; Martinez and Gomes, 2002). Therefore, we believe that discrepancies between our results and those reported by Meiners and co-workers have probably resulted from differences in the preservation of associated ECM.

Several reasons have led us to search for a possible role of surface glycolipids in guiding polymerization of extracellular laminin. First, our previous work had suggested that negative lipids could modulate laminin self-assembly (Freire and Coelho-Sampaio, 2000; Freire et al., 2002). Second, we knew that surface charges had been proposed to fine-tune the cellular response to ECM molecules, such as chondroitin and heparan sulfate proteoglycans (Powell et al., 1997; Wettreich et al., 1999). Third, demonstrations that formation of fibronectin fibrils was totally dependent on the presence of gangliosides in the membrane (Spiegel et al., 1985) prompted us to search for a parallel effect of gangliosides on the assembly of laminin matrices. Two results presented in this paper constitute direct evidence that negative lipids segregated in microdomains of the plasma membrane are important in directing laminin assemblies. The first showed that the disruption of microdomains led to the disruption of laminin assemblies; and the second showed that artificial lipid surfaces can guide laminin organization in the absence of other plasma membrane components. Moreover, we presented complementary data corroborating our proposal: (1) that the more permissive embryonic astrocytes possessed a significant more negative surface potential than astrocytes derived from newborn animals; (2) that neuraminidase treatment led to an increase in the surface charge concomitant with a release of bound laminin; and (3) that re-assembly of laminin on an embryonic astrocyte layer pre-treated with neuraminidase followed the pattern of newborn cells.

Astrocytes not only constitute the pavement for neuronal migration and axonal growth during morphogenesis, but also appear as a background substrate in injured adult CNS. In the latter case, however, astrocytes are a key component of reactive gliosis, a major impediment to axonal regeneration (Hatten et al., 1991; Ridet et al., 1997; Reier et al., 1988).

Considerable effort has been made over the last decade to understand the molecular mechanisms underlying the switch from a permissive to a non-permissive phenotype of astrocytes in these two scenarios. It has been shown that normal astroglial maturation is accompanied by a decrease in molecules, such as laminin, NCAM, L1 and heparan sulfate proteoglycan, which are known to promote axonal outgrowth (Ard and Bunge, 1988; McKeon et al., 1995). In parallel, there occurs an increase in the synthesis of molecules known to inhibit neurite outgrowth, such as chondroitin sulfate proteoglycan and tenascin (McKeon et al., 1991; Faissner and Steindler, 1995). Our work now provides additional data in demonstrating that, besides regulation of laminin expression, astrocyte aging can modulate the organization pattern of the secreted protein. While the embryonic laminin matrix remained associated to the plane of cell surfaces, displaying a geometrical arrangement, the matrices produced by newborn astrocytes were mainly composed of fibrous laminin deposits that protruded to the extracellular space. The fact that we did not observe differences in the intensity of laminin staining in embryonic and newborn astrocytes suggests that the pattern of laminin deposition is more critical for axonal growth than the amount of laminin secreted.

Laminin has been detected immunohistochemically in astrocytes of the embryonic, but not the injured, mammalian CNS (Jucker et al., 1996). In the injured adult CNS, laminin is produced only in some reactive astrocytes located near the site of CNS lesions (McKeon et al., 1991; Giftochristos and David, 1988; Liesi et al., 1984). Although the potent axon growth and guidance capacities of laminin in the context of the CNS have been extensively demonstrated in vitro (Chamak and Prochiantz, 1989; Garcia-Abreu et al., 1995a; Freire et al., 2002; Martinez and Gomes, 2002; Mendes et al., 2003), its role in regeneration still lacks confirmation. In vivo, there is a great deal of evidence in support of the critical role of laminin in PNS regeneration but relatively little is known about a possible role of laminin in the process of axon regeneration in the mammalian CNS (Bonner and O'Connor, 2001; Grimpe et al., 2002). The fact that glial scars are enriched in laminin (Grimpe and Silver, 2002) suggests that high content of this protein is not per se permissive to neurite outgrowth. This observation leads to two, non-exclusive, possibilities: (1) there might be a balance in favor of neuritic inhibitory molecules in injured brain tissues, despite the presence of abundant laminin; (2) laminin assembled within the injured brain might be structurally distinct from the matrix formed under physiological conditions. Evidence reported here supports the last hypothesis.

In the nervous system, sialylation of glycoproteins and glicolipids plays an important role at times of extensive neuronal plasticity, such as during development and regeneration. Gangliosides are a large group of sialized glycosphingolipids widely expressed in the mammalian nervous system (Lloyd and Furukawa, 1998). Because the expression of gangliosides in CNS accompanies neural maturation, it is believed that these molecules are the main factors responsible for pacing brain development (Rosenberg and Stern, 1996). In particular, a striking correlation between ganglioside expression and neuritogenesis has previously been established (Hogan et al., 1988; Rosenberg et al., 1992). The regulatory role of such lipids is usually attributed to their ability to form microdomains in cell membranes (Masserini et al., 1988), allowing for selective sorting of polarized neuronal structures (Hogan et al., 1988; van Echten and Sandhoff, 1989) or for dimerization-induced activation of growth factor receptors (Farooqui et al., 1997). It has been proposed that biochemical engineering of the side chain sialic acid might activate β1 integrins thus increasing binding to ECM molecules and favoring neuritogenesis (Buttner et al., 2002). In addition, the recent identification of gangliosides specifically associated with integrin and vinculin in point contacts of growth cones of dorsal root ganglia strongly implicates these molecules in integrin-ECM interactions (Negreiros et al., 2003).

Despite the abundant information on neuronal expression of gangliosides, little is known about a parallel increase in glial gangliosides throughout development. It has been reported, however, that gangliosides GD1, GD2 and GD3, are present in developing glia, and have their expressions decreased after birth (Kotani et al., 1995). These data fit well with our view that laminin matrices formed in the presence of low levels of gangliosides, such as those derived from newborn astrocytes or after neuraminidase treatment, are poor substrate for neuritogenesis.

Here we have reported that digestion of sialic acid groups with neuraminidase reduces by 50% the surface potential of embryonic astrocytes, a finding that supports a contribution of sialylation in the establishment of the putative negative field guiding laminin assembly. Our proposal is that the electrostatic potential created by negative glycolipids or glycoproteins on astrocyte surfaces favors a laminin-assembling mode characterized by the formation of flat surfaces, which, in turn, will favor astrocyte-guided neurite outgrowth. It is important to note that we have focused on the role of sialic acid associated with glycolipids because experimental assessment in this case was more direct (use of artificial lipid films). However, it is possible that glycoproteins also contribute with the formation of an electrostatic potential at the astrocyte surface. In this regard, the polysialylated neural cell adhesion molecule could play an important role, as demonstrated in the development of the inner ear, where polysialylation was found to modulate axonal navigation (Hrynkow et al., 1998a; Hrynkow et al., 1998b). In the end, our work points to the need to establish a new concept underlying the mechanism by which gangliosides modulate neuritogenesis. We propose that the effect of gangliosides on axonal growth during development is mainly exerted by organizing the astrocyte-associated ECM. The fact that astrocyte maturation and activation share several common features opens the possibility that laminin organization can be manipulated to reverse CNS pathological conditions.

We thank Adiel Batista do Nascimento for his excellent technical assistance. The authors are in debt to R. Linden and M. P. Gomes for their critical reading of this manuscript. This work was partially supported by grants from the Fundação Universitária José Bonifácio and Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ).

Arahata, K., Ishii, H. and Hayashi, Y. K. (
1995
). Congenital muscular dystrophies.
Curr. Opin. Neurol.
8
,
385
-390.
Ard, M. D. and Bunge, R. P. (
1988
). Heparan sulfate proteoglycan and laminin immunoreactivity on cultured astrocytes: relationship to differentiation and neurite growth.
J. Neurosci.
8
,
2844
-2858.
Baron, W., Decker, L., Colognato, H. and ffrench-Constant, C. (
2003
). Regulation of integrin growth factor interactions in oligodendrocytes by lipid raft microdomains.
Curr. Biol.
13
,
151
-155.
Bixby, J. L. and Harris, W. A. (
1991
). Molecular mechanisms of axon growth and guidance.
Annu. Rev. Cell. Biol.
7
,
117
-159.
Bonner, J. and O'Connor, T. P. (
2001
). The permissive cue laminin is essential for growth cone turning in vivo.
J. Neurosci.
21
,
9782
-9791.
Büttner, B., Kannicht, C., Schmidt, C., Löster, K., Reutter, W., Lee, H. Y., Nöhring, S. and Horstkorte, R. (
2002
). Biochemical engineering of cell surface sialic acids stimulates axonal growth.
J. Neurosci.
22
,
8869
-8875.
Carey, D. J., Bendt, K. M. and Stahl, R. C. (
1996
). The cytoplasmic domain of syndecan-1 is required for cytoskeleton association but not detergent insolubility. Identification of essential cytoplasmic domain residues.
J. Biol. Chem.
271
,
15253
-15260.
Chamak, B. and Prochiantz, A. (
1989
). Influence of extracellular matrix proteins on the expression of neuronal polarity.
Development
106
,
483
-491.
Claas, C., Stipp, C. S. and Hemler, M. E. (
2001
). Evaluation of prototype transmembrane 4 superfamily protein complexes and their relation to lipid rafts.
J. Biol. Chem.
276
,
7974
-7984.
Colognato, H. and Yurchenco, P. (
2000
). Form and function: the laminin family of heterotrimers.
Dev. Dyn.
218
,
213
-234.
Edgar, D. (
1991
). The expression and distribution of laminin in the developing nervous system.
J. Cell Sci.
15
,
9
-12.
Farooqui, T., Franklin, T., Pearl, D. K. and Yates, A. J. (
1997
). Ganglioside GM1 enhances induction by nerve growth factor of a putative dimmer of TrkA.
J. Neurochem.
68
,
2348
-2355.
Faissner, A. and Steindler, D. (
1995
). Boundaries and inhibitory molecules in developing neural tissues.
Glia
13
,
233
-254.
Freire, E. and Coelho-Sampaio, T. (
2000
). Self-assembly of laminin induced by acidic pH.
J. Biol. Chem.
275
,
817
-822.
Freire, E., Gomes, F. C. A., Linden, R., Moura Neto, V. and Coelho- Sampaio, T. (
2002
). Structure of laminin substrate modulates cellular signaling for neuritogenesis.
J. Cell Sci.
115
,
4867
-4876.
Garcia-Abreu, J., Moura Neto, V., Carvalho, S. L. and Cavalcante, L. A. (
1995a
). Regionally specific properties of midbrain glia: interactions with midbrain neurons.
J. Neurosci. Res.
40
,
471
-477.
Garcia-Abreu, J., Cavalcante, L. A. and Moura Neto, V. (
1995b
). Differential patterns of laminin expression in lateral and medial midbrain glia.
Neuroreport
6
,
761
-764.
Giftochristos, N. and David, S. (
1988
). Laminin and heparan sulphate proteoglycan in the lesioned adult mammalian central nervous system and their possible relationship to axonal sprouting.
J. Neurocytol.
17
,
385
-397.
Gomes, F. C. A., Garcia-Abreu, J., Galou, M., Paulin, D. and Moura Neto, V. (
1999
). Neurons induced GFAP gene promoter of cultured astrocytes from transgenic mice.
Glia
26
,
97
-108.
Grierson, J. P., Petroski, R. E., Ling, D. S. and Geller, H. M. (
1990
). Astrocyte topography and tenascin cytotactin expression: correlation with the ability to support neuritic outgrowth.
Brain Res. Dev. Brain Res.
55
,
11
-19.
Grimpe, B. and Silver, J. (
2002
). The extracellular matrix in axon regeneration.
Prog. Brain Res.
137
,
333
-349.
Grimpe, B., Dong, S., Doller, C., Temple, K., Malouf, A. T. and Silver, J. (
2002
). The critical role of basement membrane-independent laminin γ1 chain during axon regeneration in the CNS.
J. Neurosci.
22
,
3144
-3160.
Hakomori, S., Yamamura, S. and Handa, A. K. (
1998
). Signal transduction through glyco(sphingo)lipids. Introduction and recent studies on glyco(sphingo)lipid-enriched microdomains.
Ann. New York Acad. Sci.
845
,
1
-10.
Hatten, M. E., Liem, R. K., Shelanski, M. L. and Mason, C. A. (
1991
). Astroglia in CNS injury.
Glia
4
,
233
-243.
Hogan, M. V., Saito, M. and Rosenberg, A. (
1988
). Influence of monensin on ganglioside anabolism and neurite stability in culture chick nervous.
J. Neurosci. Res.
20
,
390
-394.
Höpker, V. H., Shewan, D., Tessier-Lavingne, M., Poo, M. and Holt, C. (
1999
). Growth-cone attraction to netrin-1 is converted to repulsion by laminina-1.
Nature
401
,
69
-73.
Hrynkow, S. H., Morest, D. K., Brumwell, C. and Rutishauser, U. (
1998a
). Spatio-temporal diversity in the microenvironments for neural cell adhesion molecule, neural cell adhesion molecule-polysialic acid, and L1-adhesion molecule expression by sensory neurons and their targets during cochleovestibular innervation.
Neuroscience
87
,
401
-422.
Hrynkow, S. H., Morest, D. K., Bilak, M. and Rutishauser, U. (
1998b
). Multiple roles of neural cell adhesion molecule, neural cell adhesion molecule-polysialic acid, and L1 adhesion molecules during sensory innervation of the otic epithelium in vitro.
Neuroscience
87
,
423
-437.
Hunter, D. D. and Brunken, W. J. (
1997
). B2 laminins modulate neuronal phenotype in rat retina.
Mol. Cell Neurosci.
10
,
7
-15.
Hynds, D. L., Burry, R. W. and Yates, A. J. (
1997
). Gangliosides inhibit growth factor-stimulated neurite outgrowth in SH-SY5Y human neuroblastoma cells.
J. Neurosci. Res.
47
,
617
-625.
Jucker, M., Tian, M. and Ingram, D. K. (
1996
). Laminins in the adult and aged brain.
Mol. Chem. Neuropathol.
28
,
209
-218.
Keshet, G. I., Ovadia, H., Taraboulos, A. and Gabizon, R. (
1999
). Scrapie-infected mice and PrPc knockout mice share abnormal localization and activity of neuronal nitric oxide synthase.
J. Neurochem.
72
,
1224
-1231.
Keshet, G. I., Bar-Peled, O., Yaffe, D., Nudel, U. and Gabizon, P. (
2000
). The cellular prion protein colocalizes with the dystroglycan complex in the brain.
J. Neurochem.
75
,
1889
-1897.
Kotani, M., Terashima, T. and Tai, T. (
1995
). Developmental changes of ganglioside expressions in postnatal rat cerebellar cortex.
Brain Res.
700
,
40
-58.
Kuecherer-Ehret, A., Graeber, M. B., Edgar, D., Thoenen, H. and Kreutzberg, G. W. (
1990
). Immunoelectron microscopic localization of laminina in normal and regenerating mouse sciatic nerve.
J. Neurocytol.
19
,
101
-109.
Liesi, P. (
1990
). Extracellular matrix and neuronal movement.
Experientia
46
,
900
-907.
Liesi, P., Kaakkola, S., Dahl, D. and Vaheri, A. (
1984
). Laminin is induced in astrocytes of adult brain by injury.
EMBO J.
3
,
683
-686.
Lloyd, K. O. and Furukawa, K. (
1998
). Biosynthesis and functions of gangliosides: recent advances.
Glycoconj. J.
15
,
627
-636.
Luckenbill-Edds, L. (
1997
). Laminin and the mechanism of neuronal outgrowth.
Brain Res. Rev.
23
,
1
-27.
Martinez, R. and Gomes, F. C. A. (
2002
). Neuritogenesis induced by thyreoid hormone-treated astrocytes is mediated by epidermal growth factor/mitogen-activated protein kinase-phosphatidylinositol 3 kinase pathways and involves modulation of extracellular matrix proteins.
J. Biol. Chem.
277
,
49311
-49318.
Masserini, M., Palestini, P., Venerando, B., Fiorilli, A., Acquotti, D. and Tettamanti, G. (
1988
). Interactions of proteins with ganglioside-enriched microdomains on the membrane: the lateral phase separation of molecular species of GD1a ganglioside, having homogeneous long-chain basic composition, is recognized by Vibrio cholerae sialidase.
Biochemistry
27
,
7973
-7978.
McKeon, R. J., Schreiber, R. C., Rudge, J. S. and Silver, J. (
1991
). Reduction of neurite outgrowth in a model of glial scarring following CNS injury is correlated with the expression of inhibitory molecules on reactive astrocytes.
J. Neurosci.
11
,
3398
-3411.
McKeon, R. J., Hoke, A. and Silver, J. (
1995
). Injury-induced proteoglycans inhibit the potential for laminin-mediated axon growth on astrocytic scars.
Exp. Neurol.
136
,
32
-43.
Meiners, S., Powell, E. M. and Geller, H. M. (
1995
). A distinct subset of tenascin/CSPG-rich astrocytes restricts neuronal growth in vitro.
J. Neurosci.
15
,
8096
-8108.
Mendes, F. A., Onofre, G. R., Silva, L. C. F., Cavalcante, L. A. and Garcia- Abreu, J. (
2003
). Concentration-depnedent actions of glial chondroitin sulfate on the neuritic growth of midbrain neurons.
Dev. Brain. Res.
142
,
111
-119.
Morissette, N. and Carbonetto, S. (
1995
). Laminin alpha 2 chain (M chain) is found within the pathway of avian and murine retinal projections.
J. Neurosci.
15
,
8067
-8082.
Negreiros, E. M., Leao, A. C., Santiago, M. F. and Mendez-Otero, R. (
2003
). Localization of ganglioside 9-O-acetyl GD3 in point contacts of neuronal growth cones.
J. Neurobiol.
57
,
31
-37.
Pindzola, R. R., Doller, C. and Silver, J. (
1993
). Putative inhibitory extracellular matrix molecules at the dorsal root entry zone of the spinal cord during development and after root and sciatic nerve lesions.
Dev. Biol.
156
,
34
-48.
Pires Neto, M. A., Braga-de-Souza, S. and Lent, R. (
1999
). Extracellular matrix molecules play diverse roles in the growth and guidance of central nervous system axons.
Braz. J. Med. Biol. Res.
32
,
633
-638.
Powell, E. M., Meiners, S., DiProspero, N. A. and Geller, H. M. (
1997
). Mechanisms of astrocyte-directed neurite guidance.
Cell Tissue Res.
290
,
385
-393.
Reier, P. J. and Houle, J. D. (
1988
). The glial scar: its bearing on axonal elongation and transplantation approaches to CNS repair.
Adv. Neurol.
47
,
87
-138.
Ridet, J. L., Malhotra, S. K., Privat, A. and Gage, F. H. (
1997
). Reactive astrocytes: cellular and molecular cues to biological function.
Trends Neurosci.
20
,
570
-577.
Rosenberg, A. and Stern, N. (
1996
). Changes in sphingosine and fatty acid components of the gangliosides in developing rat and human brain.
J. Lipid Res.
7
,
122
-127.
Rosenberg, A., Sauer, A., Noble, E. P., Gross, H. J., Chang, R. and Brossmer, R. (
1992
). Developmental patterns of ganglioside sialosylation coincident with neuritogenesis in cultured embryonic chick brain neurons.
J. Biol. Chem.
267
,
10607
-10612.
Ryan, M. C., Christiano, A. M., Engvall, E., Wewer, U. M., Miner, J. H., Sanes, J. R. and Burgeson, R. E. (
1996
). The functions of laminins: lessons from in vivo studies.
Matrix Biol.
15
,
369
-381.
Sajin, B. and Steindler, D. A. (
1994
). Cells on the edge: boundary astrocytes and neurons.
Perspect. Dev. Neurobiol.
2
,
275
-289.
Silva Filho, F. C., Santos, A. B., de Carvalho, T. M. and de Souza, W. (
1987
). Surface charge of resident, elicited, and activated mouse peritoneal macrophages.
J. Leukoc. Biol.
41
,
143
-149.
Smith, G. M., Miller, R. H. and Silver, J. (
1986
). Changing role of forebrain astrocytes during development, regenerative failure, and induced regeneration upon transplantation.
J. Comp. Neurol.
251
,
23
-43.
Smith, G. M., Rutshauser, U., Silver, J. and Miller, R. H. (
1990
). Maturation of astrocytes in vitro alters the extent and molecular basis of neurite outgrowth.
Dev. Biol.
138
,
377
-390.
Smyth, N., Vatansever, H. S., Murray, P., Meyer, M., Frie, C., Paulsson, M. and Edgar, D. (
1999
). Absence of basement membranes after targeting the LAMC1 gene results in embryonic lethality due to failure of endoderm differentiation.
J. Cell Biol.
144
,
151
-160.
Spiegel, S., Yamada, K. M., Hom, B. E., Moss, J. and Fishman, P. H. (
1985
). Fluorescent gangliosides as probes for the retention and organization of fibronectin by ganglioside-deficient mouse cells.
J. Cell Biol.
100
,
721
-726.
Tessier-Lavigne, M. and Goodman, C. S. (
1996
). The molecular biology of axon guidance.
Science
274
,
1123
-1133.
Thorne, R. F., Marshall, J. F., Shafren, D. R., Gibson, P. G., Hart, I. R. and Burns, G. F. (
2000
). The integrins alpha3β1 and alpha6β1 physically and functionally associate with CD36 in human melanoma cells. Requirement for the extracellular domain OF CD36.
J. Biol. Chem.
275
,
35264
-35275.
Tsiper, M. V. and Yurchenco, P. D. (
2002
). Laminin assembles into separate basement membrane and fibrillar matrices in Schwann cells.
J. Cell Sci.
115
,
1005
-1015.
van der Goot, F. G., Gonzalez-Manas, J. M., Lakey, J. H. and Pattus, F. (
1991
). A `molten-globule' membrane-insertion intermediate of the pore-forming domain of colicin A.
Nature
354
,
408
-410.
van Echten, G. and Sandhoff, K. (
1989
). Modulation of ganglioside biosynthesis in primary cultured neurons.
J. Neurochem.
52
,
207
-214.
Wettreich, A., Sebollela, A., Carvalho, M. A., Borojevic, R., Ferreira, S. T. and Coelho-Sampaio, T. (
1999
). Acid pH modulates the interaction between human granulocyte-macrophage colony stimulating factor and glycosaminoglycans.
J. Biol. Chem.
274
,
31468
-31475.
Yurchenco, P. D., Tsilibary, E. C., Charonis, A. S. and Furthmayr, H. (
1985
). Laminin polymerization in vitro. Evidence for a two-step assembly with domain specificity.
J. Biol. Chem.
260
,
7636
-7644.
Zhou, F. C. (
1990
). Four patterns of laminin-immunoreactive structure in developing rat brain.
Brain Res. Dev. Brain Res.
55
,
191
-201.