Transforming growth factor (TGF) β1 regulates cell migration of non-neural cells. Hence, two hypotheses were tested: (i) that TGFβ1 affects cell migration and the expression of associated adhesion proteins in developing cortex; and (ii) that these effects are antagonized by ethanol. The effects of TGFβ1 (2.5–40 ng/ml) and ethanol (400 mg/dl) on cell migration were examined in organotypic cultures from fetal rat brains. Migration was determined by tracing the movement of cells pulse-labeled with bromodeoxyuridine. Cell migration was altered by TGFβ1 in a concentration-dependent manner: at low concentrations, cell migration was promoted whereas at high concentrations TGFβ1 impeded migration. Ethanol treatment alone reduced the rate of migration. Interestingly, the rate of cell migration in slices treated with both TGFβ1 and ethanol was the same as that in untreated cultures. The expression of cell adhesion proteins (nCAM, integrin α3, αv and β1) was differentially effected by TGFβ1 and/or ethanol. TGFβ1 increased the expression of these adhesion proteins in a progressive, concentration-dependent manner. Likewise, ethanol also increased adhesion protein expression, however, combined TGFβ1 and ethanol treatment reduced expression. Collectively, the data show that TGFβ1 alters cell migration in the developing cortex and that the TGFβ1 system is a target of ethanol toxicity.
Development of the cerebral cortex from a simple, embryonic neuroepithelium to a well organized laminar structure requires proper migration of post-mitotic neurons. Migration is a highly regulated process following an orderly inside-to-outside sequence (Angevine and Sidman, 1961; Berry and Rogers, 1965; Rakic, 1972, 1974; Miller and Nowakowski, 1988; Rakic et al., 1994). The order of this process depends upon the coordination of migrating neurons, radial glia, and proteinaceous mediators in the microenvironment.
Cellular interactions important for migration are mediated through multiple groups of proteins, including cell adhesion proteins (CAPs) and growth factors. CAPs, such as neural cell adhesion molecule (nCAM), L1, astrotactin, and integrins, mediate appropriate attachment in the extracellular environment and transduce signals to cytoskeletal elements important for cell motility (Miura et al., 1992; Zheng et al., 1996; Lambert de Rouvroit and Goffinet, 2001; Hatten, 2002; Nadarajah and Parnavelas, 2002; Sobeih and Corfas, 2002; Schmid and Anton, 2003). One growth factor that affects cell migration is transforming growth factor (TGF) β1. For example, TGFβ1 can promote the migration of a variety non-neural cells including osteoclasts (Pilkington et al., 2001), immune (Wahl et al., 1993; Olsson et al., 2000; Olsson et al., 2001), and endothelial cells (Enenstein et al., 1992). TGFβ1 is also involved in tumorgenic cell migration, particularly that of neural-derived gliomas and neuroblastomas (Paulus et al., 1995; Tsuzuki et al., 1998; Platten et al., 2000; Platten et al., 2001).
The distributions of ligands and receptors in the developing cortex in vivo are consistent with the notion that TGFβ is involved in neuronal migration. TGFβ ligands (TGFβ1 and TGFβ2) are expressed by neurons and radial glia, respectively (Flanders et al., 1991; Miller, 2003). TGFβ receptor type 1 (TGFβIr) mRNA is found in the intermediate zone (IZ) of the developing mouse cortex (Tomoda et al., 1996). Prenatally, both TGFβIr and TGFβ receptor type 2 (TGFβIIr) protein are expressed by cells in the proliferative areas. Perinatally, TGFβIr is expressed along radial glial fibers (Miller, 2003). Collectively, these studies provide compelling evidence that the TGFβ system is involved in neuronal migration in the developing brain.
TGFβ1 can influence cell migration by modulating the expression of the CAPs that mediate attachment and promote motility. Two families of adhesion proteins, integrins and cell adhesion molecules (CAMs), are upregulated by TGFβ in numerous cell types (Ignotz and Massagué, 1987; Roberts et al., 1988; Heino et al., 1989; Ignotz et al., 1989; Roubin et al., 1990; Perides et al., 1994; Stewart et al., 1997; Luo and Miller, 1999a; Miller and Luo, 2002). Integrins are heterodimeric proteins that act as essential links between proteins in the extracellular environment and the cytoskeleton (van der Flier and Sonnenberg, 2001). CAMs, e.g. nCAM and L1, mediate cellular attachment through homophilic interactions or binding with CAPs (Ronn et al., 1998; Crossin and Krushel, 2000). Studies of immature neurons and neuroblastoma cells show that TGFβ1 up-regulates expression of nCAM (Luo and Miller, 1999a; Miller and Luo, 2002), a protein intimately involved in neuronal migration (Chuong et al., 1987; Miura et al., 1992).
Migrating neurons are targets of ethanol toxicity. Brains of human fetuses exposed to alcohol in utero exhibit defects such as neuroglial heterotopias and ectopic cell clusters (Clarren et al., 1978). Studies of animal models provide insight into possible mechanisms underlying such defects. In the neocortex, late-generated neurons intended for superficial cortical layers can be found ectopically in deeper layers of cortex (Miller, 1993). A potential cause of this migrational error is the disruption of the radial glial scaffold upon which some neurons migrate because the radial glia prematurely differentiate into astrocytes (Shetty and Phillips, 1992; Miller and Robertson, 1993). Alternatively, neurons destined for the cortical plate (CP) can form heterotopic clusters beyond the limits of the cerebral wall (Clarren et al., 1978; Kotkoskie and Norton, 1988; Komatsu et al., 2001). Ethanol can disrupt signals that prompt post-mitotic neurons to migrate from the proliferative zones and reduces the rate of neuronal migration. Ethanol exposure, however, does not reduce the total number of migrating cells within the cerebral wall (Miller, 1993). Taken together, these data suggest that ethanol affects the initiation, maintenance, and completion of neuronal migration.
Ethanol interferes with the normal expression and activity of TGFβ1. In vivo, prenatal ethanol exposure alters the amount of both TGFβ ligand and receptor in the developing cortex (Miller, 2003). Notably, ethanol severely reduces the expression TGFβIr in radial glia. In vitro, ethanol blocks TGFβ1-dependent increases in nCAM expression by B104 neuroblastoma cells (Luo and Miller, 1999a). Furthermore, treatment of primary cultured cortical neurons with either TGFβ1 or ethanol increases nCAM expression (Miller and Luo, 2002). In contrast, combined treatment with TGFβ1 and ethanol attenuates nCAM expression. Thus, it appears that TGFβ1 and ethanol affect cellular changes via similar, yet unidentified, mechanisms. Functional studies of neuroblastoma cells support this conclusion. TGFβ1 and ethanol are anti-proliferative agents that initiate the same signal transduction mechanisms (Luo and Miller, 1999a,b). It is noteworthy that ethanol also affects the function of L1, by reducing L1-mediated cell–cell adhesion (Charness et al., 1994; Ramanathan et al., 1996; Wilkemeyer et al., 1999, 2002).
The present study tests the hypotheses that TGFβ1 modulates cell migration in the developing cerebral cortex and that ethanol disrupts the action of TGFβ1. Two primary goals of this study are (i) to determine whether TGFβ1 influences cell migration and CAP expression in the developing cerebral cortex; and (ii) to assess whether ethanol affects the action of exogenously applied TGFβ1.
Materials and Methods
Organotypic Slice Cultures
Pregnant Sprague–Dawley rats (Taconic, Germantown, NY) were anesthetized on gestational (G) day17 with a cocktail of ketamine (10 mg/kg) and xylazine (1.0 mg/kg). Embryos were collected and their brains were removed. Forebrains were dissected and cut coronally into 300 µm slices using a MacIlwain Tissue Chopper (Mickle Lab Engineering, Gomshell, UK). All slices were taken from the same rostrocaudal level in the brain that contained presumptive parietal cortex. Approximately six slice hemispheres were obtained per fetal brain. Slices from a single litter were pooled and carefully arranged on Millicell filter inserts with 0.40 µm pores; five or six slices per insert (Millipore, Bedford, MA).
Inserts were placed in 35 mm culture dishes (Falcon, Lincoln Park, NY) containing 20% fetal bovine serum (FBS), Minimal Essential Media with Hanks salts, 200 mM glutamine (Gibco, Carlsbad, CA), 25 mM HEPES, 100 mM dextrose, 25 mM KCl, 100 µM penicillin/streptomycin (Gibco), and 12.5 µM fungizone (Gibco). HEPES was added to further stabilize the pH of the medium during regular movements in and out of the incubator necessitated by the ethanol and 5-bromo-2-deoxyuridine (BrdU; Sigma, St Louis, MO) treatments. Cultures were incubated at 37°C with 95% O2/5.0% CO2. Slices remained in the FBS-laced medium for 2–4 h, after which the slices were incubated in a FBS-free medium containing TGFβ1 (0, 2.5, 5, 10, 20 or 40 ng/ml), ethanol (100, 200, 400 mg/dl), or TGFβ1 (10 ng/ml) and ethanol (400 mg/dl).
An enclosed chamber method was used to preserve a stable concentration of ethanol in the medium (Adickes et al., 1988; Luo and Miller, 1997). Cultures were suspended over a water bath containing 100, 200 or 400 mg/dl ethanol, depending on treatment. The chambers were sealed and CO2 was added to the chambers to maintain a concentration of 5.0% CO2.
Bromodeoxyuridine Labeling and Immunohistochemistry
Following 10 h treatment with TGFβ1 and/or ethanol (designated as time 0 hr; t0), slices were exposed to 0.0040% BrdU for 1 h (Luo and Miller, 1997; Haydar et al., 1999; Jacobs and Miller, 2000). After the 1 h exposure (t1), the BrdU-containing medium was removed and replaced with fresh, serum-free medium containing no BrdU. At t1, t6 or t12, slices were fixed in 4.0% paraformaldehyde for 30 min at room temperature. Slices were processed for cryosectioning and cut in 15 µm sections.
Cells that incorporated BrdU were visualized immunohistochemically (Miller and Nowakowski, 1988; Luo and Miller, 1997; Jacobs and Miller, 2000). Following a rinse with 0.10 M phosphate buffered saline (PBS) rinse, endogenous peroxidase activity in the sections was quenched with a wash in 3.0% H2O2 for 5 min. Sections were incubated in 2.0 N HCl for 30 min followed by a 1 min rinse in dH2O. Sections were incubated overnight at 4°C with an anti-BrdU antibody (Becton-Dickinsen, San Diego, CA) diluted 1:30 in PBS with 0.75% Triton X-100 and 10% goat serum. After a several rinses with PBS, sections were incubated in biotintylated goat anti-mouse antibody (Vector Laboratories, Burlingame, CA) diluted 1:400 in PBS for 1 h. Elite Vectastain ABC and DAB kits (Vector) were used to detect BrdU-labeled cells.
Some sections were processed immunohistochemically for the expression of nCAM. A procedure similar to that described above was followed with the following differences: (i) the sections were not treated with HCl; and (ii) the primary antibody was a mouse anti-nCAM antibody (Sigma) diluted 1:1000 in PBS.
Various controls for the immunolabeling were performed. For nCAM and BrdU labeling, the immunohistochemical procedure was performed without incubating the sections with a primary antibody. A second control for the BrdU studies included immunohistochemical processing of sections from slices not exposed to BrdU. The results of these controls were consistently negative.
Cerebral Wall Collection and Immunoblotting
Changes in the expression of CAPs in the cerebral wall were detected using a quantitative immunoblotting method. Slices were prepared and treated as described (see ‘Organotypic slice cultures’) and collected at t6 and t12. The section of cerebral wall used in the immunoblot analyses was limited laterally by the lateral extent of the ventricle and medially by the midpoint of the ventricle (Fig. 1A; dashed lines and arrows). The cerebral walls of eight slices from each treatment condition were collected and pooled.
The tissue was placed in a lysis buffer (1.0% Nonidet P-40, 0.50% deoxycholic acid, 0.010% sodium dodecylsulfate (SDS), Complete Mini protease inhibitor cocktail tablets (1 tablet per 10 ml buffer; Roche, Indianapolis, IN) in 0.010 M PBS), homogenized, and spun at 10,000 r.p.m. for 10 min. Only the supernatant was used for protein analysis. Pilot studies performed on pellet fractions consistently showed that no signal was detectable. The protein content of the supernatant was determined (Bradford, 1976). An aliquot of the supernatant containing 40 µg/ml protein was combined with electrophoresis sample buffer (300 mM Tris–HCl, 50% glycerol, 5.0% SDS, 0.025% bromophenol blue, w/wo 250 mM β-mercaptoethanol).
Samples were loaded on a 10% SDS–polyacrylamide gel, separated by electrophoresis, and transferred to nitrocellulose membranes. Non-specific immunoreactivity on the membranes was blocked for 2 h in 5.0% non-fat dehydrated milk (NFDM). Blots were then probed overnight at 4°C with one of the following: mouse monoclonal anti-nCAM antibody (1:1000 in PBS; Sigma) diluted in 2.5% NFDM, rabbit-polyclonal anti-integrin α3 antibody (1:200 in PBS; Chemicon, Temecula, CA), mouse monoclonal anti-integrin αv antibody (1:500 in PBS; Pharmigen, San Diego, CA), and rabbit polyclonal anti-integrin β1 antibody (1:200 in PBS; Santa Cruz, Santa Cruz, CA). Subsequently, the membranes were incubated for 30 min with horseradish peroxidase-linked anti-mouse or anti-rabbit secondary antibody. Tagged protein bands were detected using a chemiluminescent detection reagent (Amersham, Piscataway, NJ). In most cases, the membrane was stripped of immunolabel and re-probed with an anti-actin antibody as a loading control (1:1500; Sigma).
Two standards were loaded onto each gel. One was an internal standard, which was lysate of pooled cortices from a litter of 19-day-old rat fetuses. Two samples of this internal standard were loaded onto each gel. Rainbow protein standards (Amersham) were also loaded onto each gel to determine the molecular weight of the immunolabeled protein.
The rate of migration was calculated by determining the mean change in the mean position of the ten outermost BrdU-labeled cells between t6 and t12. The analysis focused on a 200 µm wide strip of the lateral cerebral wall (Fig. 1A; boxed region). The mean distance of the ten BrdU-labeled cells farthest from the ventricular surface was determined with the Bioquant Image Analysis System (R&M Biometrics, Nashville, TN). Measurements were taken from 3–5 slices from a single litter for each timepoint and treatment condition. A minimum of four litters (n ≥ 4) was used to represent each datum.
Densitometric analysis of the immunoblots was performed using an Image Station (Kodak, Rochester, NY). Data were corrected for loading variations within each blot using the amount of actin expression as a standard. Inter-blot variation was addressed by normalizing the data against the signal detected in the internal standard (Mooney and Miller, 2001).
Differences among treatment groups were assessed with an analysis of variance (ANOVA). In cases in which statistically significant (P < 0.05) differences were detected, specific differences (e.g. in time or treatment) were examined using individual t-tests.
Cell Migration in Organotypic Slice Cultures
In order to identify a discrete population of cells within the cerebral wall, slices were exposed to a one h pulse of BrdU (t0 to t1). At t1, BrdU-positive cells in the untreated (control) slices were confined to the ventricular zone (VZ) and subventricular zone (SZ) (Fig. 1B). This is consistent with the understandings (i) that BrdU is incorporated into cells passing through the S-phase and (ii) that cycling neural cells are distributed in the VZ and SZ (Angevine and Sidman, 1961; Hinds and Ruffett, 1971; Rakic, 1974; Miller and Nowakowski, 1988; Nowakowski et al., 1989). Six hours after introducing the BrdU (t6), most BrdU-labeled cells were in the proliferative zones, but some were distributed in the deep IZ (Fig. 1C). Presumably, this change resulted from the migration of BrdU-positive cells from the proliferative zones into the IZ. At t12, additional BrdU-labeled cells were in the IZ and they were located farther into the IZ (Fig. 1G).
To quantify the change in the distribution of BrdU-positive cells in the untreated slices, the distances of the 10 outermost labeled cells from the ventricular surface at t1, t6 and t12 were measured. At t6, the outermost 10 cells were significantly (P < 0.05) farther from the ventricular surface as compared with t1 (Figs 2 and 3A). These findings indicated that the cells had commenced their migrations by t6. By t12, the 10 outermost cells were distributed farther from the ventricular surface (Fig. 3A). The mean change in the position of the outermost cells between t6 and t12 was indicative of the rate of migration. Accordingly, the mean rate of migration of cells in untreated slices was 6.1 µm/h (Fig. 3B).
TGFβ1-promoted Migration of Post-mitotic Cells
At t1 (data not shown) and t6 (Fig. 1C,D), the distribution of the outermost BrdU-positive cells in the slices treated with TGFβ1 (10 ng/ml) was comparable to that in the untreated preparations. By t12, more BrdU-positive cells in the TGFβ1-treated slices were distributed in the IZ than at t6 and they were located more superficially (i.e. farther from the ventricular surface) as compared with untreated conditions (Fig. 1G,H). Thus, by t12, cells in the TGFβ1-treated slices had migrated farther than cells in untreated slices.
The effect of TGFβ1 (10 ng/ml) on cell migration was quantified. At t1 and t6, the outermost cells in TGFβ1-treated slices were at a similar mean distance from the ventricular surface as in untreated slices (Figs 2 and 3A). The mean position of the ten outermost cells significantly (P < 0.05) differed in TGFβ1-treated slices at t1 and t6 (Figs 2 and 3A). At t12, the outermost population in TGFβ1-treated slices was significantly (P < 0.05) farther from the ventricular surface than cohorts in untreated slices (Fig 3C). The mean rate of migration was significantly (P < 0.05) increased (74%) by treatment with TGFβ1 (Fig. 3B).
TGFβ1 affected cell migration in a concentration-dependent manner [Fig. 3B; F(7,31) = 13.038; P < 0.001]. At low concentrations (i.e. 2.5 and 5.0 ng/ml), TGFβ1 treatment induced a modest increase in the rate of migration (32 and 36%, respectively), however, these differences were not statistically significant (Fig. 4B). The rate of migration nearly doubled in slices treated with 10 ng/ml TGFβ1 (as compared with the rate of migration in untreated slices); this increase was significant (P < 0.05). At concentrations above 10 ng/ml, however, no such increase was evident. In slices treated with high amounts of TGFβ1 (40 ng/ml), the rate of migration was significantly (P < 0.05) slower than it was in either untreated slices or slices treated with 10 ng/ml of TGFβ1 (Fig. 4B). Thus, TGFβ1 did not always facilitate cell migration; indeed, at high concentrations, TGFβ1 inhibited migration.
Ethanol Impeded Cell Migration
The effect of ethanol (400 mg/dl) on the spatiotemporal change in the location of BrdU-labeled cells was determined. At t1, BrdU-positive cells in the ethanol-treated slices were in the proliferative zones, a pattern that was indistinguishable from that in the untreated and TGFβ1-treated preparations (data not shown). Unlike the untreated and TGFβ1-treated slices at t6, in the ethanol-treated slices few BrdU-labeled cells were outside the proliferative zones and were distributed closer to the ventricular surface (Fig. 1E). At t12, more cells were evident in the IZ, however, they were located deeper in the IZ than were cohorts in untreated slices (Fig. 1G,I). This implies that ethanol indeed affects cell migration.
Quantitative analysis of the period from t1 to t12 provided key information about the effects of ethanol on cell migration. At t1 the outermost BrdU-positive cells in the ethanol-treated slices were positioned similarly to those in the untreated and TGFβ1-treated preparations (Fig. 2). Moreover, there was a significant (P < 0.05) change in the mean distance of the ten outermost cells in the ethanol-treated slices between t1 and t6 (Figs 2 and 3A). That is, as in the untreated and TGFβ1-treated preparations, cell migration had commenced by t6. Further analysis showed that the outermost BrdU positive cells at t6 and t12 in ethanol-treated slices were significantly (P < 0.05) closer to the ventricular surface than were cohorts in the untreated slices (Fig. 3A). In fact, the rate of migration of the outermost ten BrdU-positive cells was significantly (P < 0.05) reduced (43%) by ethanol (400 mg/dl; Fig. 3B). Treatment with lower amounts of ethanol (100 or 200 mg/dl) also attenuated the rate of migration (data not shown). Both concentrations reduced the rate of migration 21–26% relative to controls.
The outermost BrdU-positive cells in slices co-treated with TGFβ1 (10ng/ml) and ethanol (400 mg/dl) were distributed at similar positions in the cerebral wall as those in untreated slices (Fig. 1C,F,G,J). The mean position of the outermost 10 cells in co-treated slices was not significantly different than in untreated preparations at both t6 and t12 (Fig. 3A). Furthermore, the mean rate of migration in TGFβ1-ethanol-treated slices was not significantly different than untreated cultures (Fig. 3B). Interestingly, the mean rate of migration in co-treated slices was significantly (P < 0.05) different than either conditions alone, being less than TGFβ1 but greater than ethanol (Fig. 3B). This suggests that ethanol disrupts TGFβ1-dependent changes in cell migration.
TGFβ1 Increased CAP Expression
CAPs, such as nCAM and integrin subunits α3, αv and β1, are important for cell migration in the developing brain (Chuong et al., 1987; Georges-Labouesse et al., 1998; Anton et al., 1999; Graus-Porta et al., 2001). Thus, cortical slices were examined for TGFβ1-dependent changes in CAP expression. Immunoblotting methods were used to determine whether TGFβ1 affected nCAM expression. At t6, concentrations of TGFβ1 (2.5, 10 and 40 ng/ml) significantly [F(5,14) = 27.934; P < 0.001] increased expression of nCAM (Fig. 5). The greatest increase in nCAM expression was detected in cultures treated with the highest concentration of TGFβ1. Some change in nCAM expression occurred between t6 and t12, though it was not statistically significant. On the other hand, by t12, the expression of nCAM in TGFβ1-treated slices was no longer significantly higher than untreated at t12. This was because at t12, nCAM expression in untreated cultures had risen significantly (P < 0.05) as compared with the amount of expression at t6.
The distribution of nCAM expression in the cerebral wall was also examined immunohistochemically. In untreated slices, nCAM expression was highest in the IZ and CP, though some labeling was evident in the proliferative areas (Fig. 6A). nCAM immunolabeling was particularly notable on the cell membranes where dark puncta outlined cell bodies (Fig. 6G). Changes in nCAM labeling were evident in slices treated with TGFβ1. These changes appeared to be concentration-dependent. Low concentrations of TGFβ1 (2.5 and 10 ng/ml) induced subtle changes in nCAM expression, most notable around the outside of the cell body (Fig. 6B,C,H,I). Treatment with high concentrations of TGFβ1 (40 ng/ml), however, induced substantial changes (Fig. 6D). Dark, often large patches of immunostained membrane were frequently evident (Fig. 6J).
TGFβ1 also altered expression of the integrin subunits α3, αv and β1. The pattern of integrin α3 expression was similar to that for nCAM. At t6, integrin α3 expression was increased significantly [F(5,19) = 16.347; P < 0.001] by TGFβ1 with treatments of 40 ng/ml having the largest effect (Fig. 7). Whereas integrin α3 expression in the TGFβ1-treated slices was not significantly different between t6 and t12, expression in the untreated conditions increased significantly (P < 0.05) by t12. Thus, α3 expression in the untreated slices did not vary significantly from the TGFβ1-treated slices at t12. Unlike nCAM expression, integrin αv expression only increased significantly [F(5,16) = 3.811; P < 0.05; F(5,17) = 5.771; P < 0.05; t6 and t12, respectively] following treatment with a high concentration of TGFβ1 (Fig. 8). TGFβ1 treatment also significantly [F(5,16) = 28.945; P < 0.001] increased the expression of integrin β1 at t6 with all concentrations tested (Fig. 9). On the other hand, at t12 only treatment with a high concentration of TGFβ1 significantly [F(5,14) = 7.175; P < 0.01] increased integrin β1 expression.
Ethanol Affected CAP Expression
To determine whether ethanol affected CAP expression, immunoblot and immunohistochemical analysis were used. Ethanol (400 mg/dl) significantly (P < 0.05) increased expression of nCAM at t6 (Fig. 5). Lower concentrations of ethanol (100 and 200mg/dl) likewise increased nCAM expression (data not shown). The increased expression of nCAM expression was maintained between t6 and t12. Interestingly, the ethanol-induced increase in nCAM expression was similar to that detected in slices treated with a high concentration of TGFβ1 (Fig. 5). The changes in nCAM expression were paralleled by immunohistochemical changes. nCAM immunolabeling was prominent in slices treated with ethanol, particularly in the IZ and CP (Fig. 6E). Intense regions of staining were commonly found around the outside of the cell body (Fig. 6K).
Like nCAM, expression of the integrin proteins was affected by ethanol. Ethanol significantly (P < 0.05) increased the expression of integrins α3, αv and β1 at t6 (Figs 7–9). In addition, the expression of integrin αv and integrin β1 was significantly (P < 0.05) greater in ethanol-treated slices than in untreated preparations at t12 (Figs 8 and 9).
Slices treated with both TGFβ1 (10 ng/ml) and ethanol (400 mg/dl) were examined for changes in nCAM expression. These preparations were subjected to three comparisons: with untreated slices and with cultures treated with TGFβ1 or ethanol. Co-treatment significantly (P < 0.05) increased the amount of nCAM at t6 relative to the amount of expression detected in untreated slices (Fig. 5). nCAM expression in slices co-treated with TGFβ1 and ethanol was not significantly different from that of slices treated with TGFβ1 alone. In contrast, TGFβ1-ethanol co-treatment significantly (P < 0.05) reduced nCAM expression at t6 and t12 re slices treated with ethanol alone.
In the co-treated slices, nCAM was expressed by cells in the IZ and CP. That is, the pattern of nCAM immunolabeling within the cerebral wall of co-treated slices was the same as that in slices treated with TGFβ1, ethanol, or neither substance. On the other hand, the intensity of the immunolabel appeared to be affected by co-treatment. nCAM expression in the IZ of co-treated appeared to be heterogeneous as there were cells with the intense staining characteristic of ethanol or 40 ng/ml conditions as well cells that looked similar to cells in 10 ng/ml TGFβ1 or untreated (Fig. 6F). This was best exemplified at a higher magnification (Fig. 6L).
Integrin expression in slices treated with both TGFβ1 and ethanol was similar as that for nCAM. Co-treatment significantly (P < 0.05) increased the expression of integrin αv and integrin β1 at t6 as compared with untreated conditions (Figs 8 and 9). In contrast, the expression of integrin α3 and integrin β1 was significantly (P < 0.05) lower in the co-treated slices than in slices treated with ethanol at t6 and t12 and for integrin αv at t12.
Cell Migration in Controls
Desynchrony of Cortical Development
A challenge to studying cell migration in a neuroepithelium in situ is the desynchrony in the developing brain. This is evident among proliferating cells that are distributed through all phases of the cell cycle. Thus, an acute (i.e. 1 h) pulse of BrdU indiscriminately labels a subpopulation of cells that is in or recently exited from the S-phase. Further desynchrony results from post-mitotic activity. After BrdU-labeled cells mandatorily pass through the G2- and M-phases, some cells permanently exit the cell cycle and others re-enter it (Miller and Nowakowski, 1988; Nowakowski et al., 1989; Takahashi et al., 1995; Miller, 1999). This phenomenon cannot be addressed with the present model.
The movement of post-mitotic ‘pioneer’ cells labeled by a BrdU pulse can be traced by focusing on the first cohort of cells to leave the proliferative zones. The present study measured the change in the position of such cells between t6 and t12. By t6, migration has commenced. This is noted by the change in the distribution of BrdU-positive cells between t1 and t6, the increased presence of labeled cells in the IZ at t6, and the significant change in the mean distance of the outermost cells from the ventricular surface between t6 and t12. Therefore, by examining the change in the position of leading cells between t6 and t12, insight into cell migration can be gleaned.
Pathways of Cell Migration
An assumption in the present study is that cortical migration occurs via a radial pathway. This is supported by many studies showing that the majority of neurons in the adult cortex are generated in the VZ and SZ and migrate along radial glia (Rakic, 1971, 1972). Recent data show that many GABAergic local circuit neurons (LCNs) are generated in the ganglionic eminence and migrate via distinct routes into the immature cortex, notably by tangential paths through the IZ, SZ, and MZ (Anderson et al., 1997; Wichterle et al., 2001; Jimenez et al., 2002; Ang et al., 2003). The pathway through the IZ raises the possibility that some of the BrdU positive cells measured in this study are migrating tangentially rather than radially. BrdU labeling studies, however, show that most LCNs migrating through the IZ (where the outermost 10 BrdU-positive cells are located at t6 and t12) are generated on G14 (Tamamaki et al., 1997). This is four days before the slices in this study were exposed to BrdU. Thus, it is unlikely that LCNs migrating tangentially from the GE through the IZ are included in the present analysis.
LCNs migrating through the SZ (Anderson et al., 2001) or MZ (Lavdas et al., 1999; Zecevic and Rakic, 2001) are generated within the timeframe of BrdU exposure used in the present study (Ang et al., 2003). Hence, the contribution from these cells must be taken into account. LCNs comprise ∼10% of the neurons in mature rat cortex (Peters et al., 1985). Thus, at most, LCNs account for only one of the outermost 10 cells used in the calculation of the rate of migration. This estimate must be further reduced because (i) many LCN’s are generated in the VZ and SZ and (ii) LCNs derived from the GE migrate via multiple pathways, two of which have no relevance to the present analysis, i.e. via the IZ (see above) and the MZ. Though we cannot completely discount alternative origins and migratory routes of the BrdU-positive cells analyzed in this study, the timing of the BrdU exposure and the relative contribution of future LCNs to the total migrating population reduces the impact of this confounding factor.
Though migration occurs in a discontinuous manner (Edmondson and Hatten, 1987), the net movement can be determined by examining the change in position over an extended period. In the present investigation, samples were garnered on G17 and the BrdU pulse and chase studies occurred on the equivalent of G18. In such untreated cultures, the mean net movement over six h was at a rate of 144 ± 4 µm/day. This is remarkably similar to the rate of migration in vivo for cells born on in control rats on G17. Such migrating cells move from the proliferative zones in vivo at a mean rate of 137 ± 15 µm/day (Miller, 1993, 1999).
Cell Adhesion Proteins
As with non-neural systems, cell migration during cortical development partly relies upon appropriate CAP expression (Anton et al., 1996; Zheng et al., 1996; Georges-Labouesse et al., 1998; Anton et al., 1999; Graus-Porta et al., 2001). CAPs mediate attachments formed at the leading edges of motile cells and lost at the trailing process (Lauffenburger and Horwitz, 1996). Not surprisingly, the present study shows that the IZ of the developing cerebral wall is heavily populated by nCAM-positive neurons. Specifically, nCAM is expressed at the surface of these IZ cells.
One unexpected, yet interesting, change in CAP expression in the untreated cultures is an increase in CAP expression, particularly nCAM and integrin α3, between t6 and t12. The mechanism underlying this change is unknown. Possibly, it results from the continued generation of cells that can express CAPs. Alternatively, the increased CAP expression may result from growth factors released by the slices into the medium. Serum-free medium conditioned by slices from neonates exhibits increased nerve growth factor (NGF) and basic fibroblast growth factor content (unpublished observations). It is also possible that the change in adhesion protein expression is a reflection of alterations in non-migrating cells (e.g. post-migratory neurons in the CP) or indicates changes in cellular CAP expression that do not translate to changes in cell-surface expression.
Effect of TGFβ1 on Cell Migration
TGFβ1 alters cell migration and these effects are concentration-dependent in a biphasic manner. At low and moderate concentrations, TGFβ1 promotes the rate of cell movement, whereas at high concentrations, TGFβ1 retards cell migration. Indeed, a similar concentration-dependent effect of TGFβ1 has been described for T-cell chemotaxis; exposure to low concentrations of TGFβ1 promotes chemotaxis and supraoptimal concentrations of TGFβ1 are inhibitory (Franitza et al., 2002).
Not only does exogenous TGFβ1 increase the rate of migration, it increases nCAM and integrin expression. Such an effect parallels the robust TGFβ1-mediated up-regulation of nCAM expression in simpler neural-based systems, e.g. primary cultures of developing cortical neurons (Miller and Luo, 2002) and dissociated cultures of B104 neuroblastoma cells (Luo and Miller, 1999a). Furthermore, these changes echo the effects of another member of the TGFβ family, bone morphogenic proteins, on nCAM and L1 (Perides et al., 1994).
As with cell migration, TGFβ1 effects nCAM and integrin expression in a concentration-dependent manner. Interestingly, the effects of concentration are different for cell migration and CAP expression. Whereas the effect of TGFβ1 on cell migration is biphasic, the effect on CAP expression is monophasic. That is, treatment with low or moderate concentrations of TGFβ1 induces a graded increase in both the amount of CAP expression and the rate of cell migration. At a high concentration of TGFβ1, however, the effects are divergent; the rate of cell migration falls while CAP expression continues to increase. One explanation is that there is an optimal ‘working’ concentration of TGFβ1 (∼10 ng/ml) to promote migration. At supra-optimal concentrations (∼40 ng/ml), migrating cells express such large amounts of adhesion proteins that adhesions are too strong to be conducive to cell motility. This conclusion is supported by a study examining the relationship between the rate of cell migration in Chinese hamster ovary (CHO) cells and the expression of integrins (Palecek et al., 1997). Increases in integrin expression only promote the migration of CHO cells to a certain amount after which higher integrin expression hinders cell migration.
Effect of Ethanol on Cell Migration
Cells in ethanol-treated slices migrate at half the rate (84 ± 14 µm/day) of cells in untreated cultures. In vivo, cells generated on G17 in an ethanol-treated rat migrate at a mean rate of 45 ± 5 µm/day (Miller, 1993). This difference between the two rates is statistically significant (P < 0.05). Interestingly, this means that the concentration of ethanol used in vitro (400 mg/dl) is less effective in slowing cell migration than is a lower exposure (∼150 mg/dl) in vivo. Thus, the present data support the use of the slice cultures as a conservative model of cell migration and fetal ethanol-induced changes. In previous studies of developing cells, it was shown that higher concentrations were needed in vitro to cause similar changes in vivo. Treatment of cells with 400 mg/dl of ethanol in vitro (Luo and Miller, 1997; Jacobs and Miller, 2001) has a similar effect on cell cycle kinetics as 150 mg/dl ethanol in vivo (Miller and Nowakowski, 1991).
One potential problem with examining the effects of ethanol on cell migration is that ethanol can suppress a contiguous developmental process, cell proliferation (Miller, 1989; Miller and Nowakowski, 1991; Miller and Kuhn, 1995). Conceivably, the change in cell migration results from an ethanol-induced decrease (i) in cell proliferation; (ii) in the number of cells that exit from the cycling population; and (iii) and in the timing of migration out of the proliferative zones (Miller, 1986; Miller and Nowakowski, 1991; Miller, 1993; Jacobs and Miller, 2001). Cell migration, however, begins in slices exposed to ethanol by t6. This is indicated by a significant change in the position of the outermost cells between t1 and t6 in the ethanol-treated slices. Thus, the ethanol-induced deficit in cell migration most likely results from ethanol-dependent changes in cell movement. Interestingly, there is a difference in the position of the outermost cells in the untreated and ethanol-treated conditions at t6. The outermost cells in the ethanol-treated condition are closer to the ventricular surface than controls. This suggests a delay in the onset of migration. Such a delay has been described as an effect of ethanol exposure in vivo (Miller, 1993).
Like TGFβ1, ethanol increased the expression of nCAM and integrin proteins in the immature cerebral wall. Note that ethanol reduces cell migration and increases CAP expression in a pattern reminiscent of that caused by high concentration (40 ng/ml) of TGFβ1. In the context of the migration model described above, one way ethanol may affect cell migration in the slice cultures is by increasing CAP expression beyond optimal amounts for cell migration.
CAPs are a target of ethanol toxicity. Ethanol affects adhesion protein expression and function in various neural systems. The most thoroughly studied CAP is nCAM (Luo and Miller, 1999a; Miñana et al., 2000; Miller and Luo, 2002). Gestational ethanol exposure alters normal expression of nCAM in the postnatal rat brain, including a decrease in expression overall as well as a decrease in nCAM cell surface expression in cultured astrocytes (Miñana et al., 2000). The functional, adhesive properties of CAMs, like L1, are also vulnerable to ethanol exposure. Ethanol disrupts L1-dependent cell adhesion without effecting the expression of the receptor, suggesting that ethanol physically blocks the homophilic interaction between L1 molecules (Charness et al., 1994; Ramanathan et al., 1996; Wilkemeyer et al., 1999).
The disruption of normal integrin expression within the developing cerebral cortex is a novel mechanism through which ethanol may interfere with cell migration. Ethanol can potentiate integrin expression by hepatocytes (Schaffert et al., 2001). The present data showing the effects of ethanol on integrin expression and cell migration are fascinating in light of recent studies of knockout mice deficient in integrin subunits. Mice lacking integrin α3 exhibit abnormal neuronal migration and premature differentiation of radial glia into astrocytes during cortical development (Anton et al., 1999). This phenotype is similar to that described in rats prenatally exposed to ethanol (Shetty and Phillips, 1992; Miller and Robertson, 1993). Likewise, integrin β1 knockout mice also feature inappropriate neuronal migration (Graus-Porta et al., 2001). Ectopic cells are common in the MZ and the pial surface is abrogated. Such migratory defects are not uncommon in human and rodent brains exposed to alcohol (Clarren et al., 1978; Kotkoskie and Norton, 1988; Komatsu et al., 2001; Miller and Mooney, 2004; Mooney et al., 2004).
Interaction of TGFβ1 and Ethanol on Cell Migration and CAP Expression
Combined treatment with TGFβ1 and ethanol results in a rate of migration similar to untreated conditions. This implies either that ethanol interferes with TGFβ1-dependent changes in cell migration or that TGFβ1 counters the depressive effects of ethanol. CAP expression is also effected by TGFβ1-ethanol co-treatment. Ethanol and TGFβ1 treatment alone increase CAP expression. Instead of an additive effect, however, co-treatment reduces CAP expression. This suggests that the restoration of a normal rate of migration in the TGFβ1-ethanol co-treated slices may be a consequence of a level of CAP expression that facilitates normal cell motility. Furthermore, the data indicated that TGFβ1 and ethanol might modulate CAP expression, and possibly cell migration, through similar mechanisms. Studies of neuroblastoma cells (Luo and Miller, 1999a,b) support this conclusion. TGFβ1 and ethanol are both anti-mitogenic and they initiate the ras/raf/mitogen-activated protein kinase (MAPK) pathway in a similar manner. Both substances induce a similar response in MAPK activity, i.e. a sustained increase.
The Endogenous TGFβ System Regulates Migration and Is the Target of Ethanol Toxicity
The developing cortex comprises TGFβ ligands and receptors (Heine et al., 1987; Flanders et al., 1991; Pelton et al., 1991; Tomoda et al., 1996; Miller, 2003). TGFβ1 is expressed by cells in the proliferative zones and TGFβIr and its sister receptor, TGFβIIr, are expressed by migrating neurons and radial glia. Thus, the TGFβ system is endogenous to the developing cerebral cortex, and it is strategically expressed to play a role in neuronal migration. Moreover, the TGFβ system is affected by prenatal exposure to ethanol (Miller, 2003). Exogenous TGFβ1 presumably interacts with the endogenous system. Furthermore, if migrating cells in the cerebral wall respond to exogenous TGFβ1 with a change in the rate of migration and CAP expression, it is probable that endogenous TGFβ1 has a similar function in vivo.
There are striking similarities between the phenotypes of the TGFβ knockouts and animals and humans exposed to ethanol in utero. Cardiac and skeletal dysgenesis as well as craniofacial abnormalities are hallmark phenotypes in the TGFβ knockout mice (Proetzel et al., 1995; Sanford et al., 1997). These defects echo malformations described in animals prenatally exposed to ethanol (Sulik et al., 1981; Sulik, 1984; Daft et al., 1986; Edwards and Dow-Edwards, 1991; Kotch et al., 1992; Astley et al., 1999) and children with FAS and ARND (Jones and Smith, 1973; Jones et al., 1973; Jones et al., 1974; Abel, 1981). Hence, the endogenous TGFβ system appears to be a specific target of ethanol.
We thank Renee Mezza for assistance with immunohistochemical techniques, Dr. Sandra Mooney for help with densitometry analysis, and Brad Pawlikowski for constructive comments on the manuscript. This research was supported by National Institute of Alcohol Abuse and Alcoholism (AA06916 and AA07568) and the Department of Veterans Affairs.