Immunostaining

For immunohistochemistry, anaesthetized P2 and P18 mice were perfused transcardially with 4% paraformaldehyde (PF) in PB and the brains post-fixed in PF overnight. Vibratome sections (30 µm) were obtained in sagittal or coronal planes and transferred to 0.2% Triton X-100 (PBS-T). Culture samples were fixed with PF for 15 min and then washed in PBS-T. Sections and cultures were blocked in 2% bovine serum albumin (Sigma) for 45 min, and then incubated overnight in the primary antibody. After rinsing samples were incubated with secondary antibodies for 90 min and then counterstained with Hoechst (10 µg mL⁻¹, Sigma).

Sections and cultures were immunostained with the following: anti-Doublecortin (DCX, 1:2,000; Chemicon); anti-glial-fibrillary-acid-protein (GFAP, 1:500; Dako); and Anti-β-Tubulin (TuJ1, 1:1,000; gift from Dr. Frankfurter). To visualize the antibodies different secondary antibodies were used: biotinylated-Goat-anti-Guinea-Pig (Atom Vector, 1:200); Alexa 568 and 488 (Molecular Probes, 1:2,000). Biotinylated antibodies were revealed using Alexa 488 Streptavidin (Molecular Probes, 1:2,000). For all antibodies, control samples were stained omitting the primary antibody and no staining was seen.

Astrocyte Monolayer Cultures

Purified astrocyte monolayers were generated following the modified protocol of Banker and Goslin (1991). Postnatal mice (P1–3) were decapitated and brains were removed and placed in cold HBSS (GIBCO). Brains were embedded in 3% low-melting-point agarose, (Pronadisa) diluted in dissecting medium (L15 supplemented with 10% FBS, GIBCO) at 43°C. After cooling, embedded brain was vibratome sectioned (200 µm), and dissected (Fig. 1A). We selected both OB (Fig. 1A, excluding the subependymal-, glomerular-, and nerve-layers to avoid contamination with ensheathing cells or RMS elements) and RMS (Fig. 1A, RMS, horizontal arm excluding the OB area) as permissive regions. As nonpermissive regions, the anterior frontal cortex (Fig. 1A, Cx) was chosen, as well as other areas adjacent to the RMS (Fig. 1A, AA). Selected tissues were collected separately in tubes and mechanically dissociated. Cells were resuspended in DMEM/F12 medium (glutamine-free, GIBCO) supplemented with 10% FBS, penicillin/streptomycin (GIBCO), and Glutamax (GIBCO), and then plated in flasks poly-l-lysine-coated (10 µg mL⁻¹, Sigma). Cell cultures were maintained in a 5% CO₂ atmosphere at 37°C, replacing the culture medium twice a week.

Open in a separate window

Fig. 1

RMS-explants cultured on astrocyte monolayers derived from permissive (RMS, OB) and nonpermissive areas (frontal Cx and other adjacent RMS regions). (A) Shaded areas indicate the regions removed to generate the different astrocyte monolayers. (B–D) Immunohistochemistry for GFAP in sagittal sections at P2 (cell nuclei counterstained with bisbenzimide are shown in grey) from the OB (B), RMS (C) and CX (D). (E–G) Neuroblasts migrating from a RMS-explant (TuJ1, green) on astrocyte monolayers (GFAP, red) from OB (E), RMS (F) and Cx (G). (H) Neuroblast M.I. is statistically higher in those cells cultured on OB and RMS-derived-astrocytes than in those cultured on astrocytes from adjacent areas (where neuroblast migration does not occur in vivo). (I) Graph illustrates the percentage of GFAP-positive cells in the astrocyte monolayers, showing a minimum of 91% astrocytes in the different monolayers. (J) Linear correlation analysis between cell density and M.I., showing no co-relation between both parameters. Horizontal lines between groups indicate significant differences. Error bars show SEM. Scale bar = 400 µm in A; 50 µm for B–D; 200 µm for E–G.

After 12–15 days in vitro (DIV), purified astrocytes were obtained by shaking flasks at 280 rpm o/n at 37°C. Attached cells were trypsinized (trypsin-EDTA; GIBCO) and seeded onto poly-l-lysine-coated-plates (10 µg mL⁻¹), at the same cell density depending on the experiment (see below).

Contact Co-Cultures of RMS-Explants on Astrocyte Monolayers

The same number of astrocytes from each source (4–6 × 10⁴ cells/well) was plated in 48-well plates and grown to confluence (2–7 days). Before explant addition, plates were shaken again to remove cells other than astrocytes. RMS-explants (P1–P3) were then placed in each well and co-cultured for 1 DIV.

Complementary experiments were performed to examine the response of RMS-explants in the presence of two different astrocyte monolayers in the same dish. Astrocytes from different origins were separately plated in parallel in the same well, as described below. Plates were prepared by dividing the dish with a strip of parafilm^® (~200-µm wide), applied by pressure, after which the plates were sterilized and then poly-l-lysine coated (Rydholm et al., 2005). Two drops (15 µL) containing 15,000 astrocytes from two different origins were plated on each side of the parafilm strip that was removed after reaching confluence. Two RMS-explants were placed on each monolayer, while in other experiments only one explant was plated between both astrocyte monolayers. Co-cultures were maintained for 1 DIV as described above.

Contact-Free Co-Culture of RMS-Explants and Astrocyte Monolayers

To test for diffusible factors, RMS-explants were plated on coverslips covered with collagen and suspended above the astrocyte monolayer by three paraffin pillars (modified from Banker and Goslin, 1991). RMS-explants obtained as above were embedded in three-dimensional collagen matrices prepared from 10 µL of Type-I-collagen (BD Biosciences). After gelling, coverslips were inverted and fixed to the astrocyte monolayer, which prevented direct physical contact but allowed these cell types to share the culture medium (400 µL of culture medium). The cultures were incubated for 3 days as described above.

Dynamic Analysis of Migration by Video Time-Lapse Recording

The dynamic nature of neuroblast migration was analyzed in contact co-cultures on the different astrocyte monolayers. Video recording of cell migration started either 6 h before fixation, or immediately after plating. Images were acquired with a 10× objective at 3-min intervals for 6 h in at least four different explant positions. In some cases, explants were recorded during the first 18 h at 5-min intervals. Time-lapse video images were acquired with an inverted microscope (Leica, DMI6000B) equipped with a CO₂ incubation system (Leica BL).

Quantitative RT-PCR Analysis of Differential Gene Expression

Total RNA was extracted from confluent astrocyte cultures and real-time RT-PCR was performed to assay the expression of genes encoding ECM and cell adhesion molecules (PAMM-013; Superarray). Astrocyte cultures were shaken and replated, and the total RNA was extracted using the RNEasy kit (Qiagen). RNA from different astrocyte sources (500 ng) was used to generate cDNAs, and signals from genomic DNA and impurities that might affect the RT-PCR assay were ruled out by quality control real-time RT-PCR-array (PAMM-999; Superarray). Fluorescent emission was recorded in real-time (Chromo-4-Four-Color Real-Time-PCR-Detector; MJ Research). All the above procedures were carried out according to the manufacturer instructions.

For tissue samples, RNA from OB-, RMS-, and Cx-areas were extracted from P2 mice as above and real time RT-PCR was used to test for the expression of following genes (Power SYBR Green RNA-to-CT 1-Step Kit, Applied Biosystems): Thbs-1 (PrimerBank ID:7110725a1), Sparc (6678077a1), Ecm1 (4103926a1), Cttna1 (6753294a1), Lama3 (1922889a1), Cdh2 (6680 902a1). All of these primer sequences were obtained from PrimerBank (Wang and Seed, 2003). Data were normalized to 18S-ribosomal-RNA (sense-CGGCTACCA CATCCAAGG, antisense-GCTGCTGGCACCAGACTT).

Heterogeneity in Astrocyte Morphology

GFAP-positive astrocytes isolated from different regions displayed distinct morphologies. Astrocytes from both OB (Fig. 2A,B) and RMS (Fig. 2C,D) were elongated with a fusiform morphology (Fig. 2B,D). This bipolar morphology was more common and evident in RMS-derived-astrocyte monolayers, whereas in OB-derived-astrocytes the elongated morphology coexisted with other more flattened cells. However, both OB- and RMS-derived-monolayers were clearly distinct from Cx-derived-monolayers which rarely included elongated astrocytes (Fig. 2E,F). Rather, monolayers from Cx-derived-astrocytes exhibited a flattened star-like and bushy morphology, with many fine terminal processes protruding from the main cellular processes (Fig. 2F). This heterogeneity reflects intrinsic differences in the morphological response with respect to similar in vitro conditions.

Open in a separate window

Fig. 2

Morphologies of astrocytes from different origins immunostained with anti-GFAP. (A,B) OB. (C,D) RMS. (E,F) Cx. OB- and RMS-astrocytes show bipolar morphology while Cx-astrocytes have a stellate and flat morphology. Scale bar = 50 for A, C, E; 20 µm for B, D, F.

Nonsoluble Factors from Astrocytes Mediate the Migration Index Differences

To determine whether surface bound or soluble molecules mediated the significant M.I. differences, we performed two additional assays. First, the M.I. was measured when two RMS-explants were plated on two different monolayers within the same culture dish (Fig. 3A) separated by a band without cells (Fig. 3B). In this assay, the M.I. differed on the alternative substrates, despite their presence in the same culture well. The relationship between the M.I. from both explants in the same well was defined as the side ratio (Fig. 3C). When OB-astrocytes were opposite cortical-astrocytes, the side ratio evidenced a higher M.I. for OB, than Cx-derived-astrocytes (side ratio mean ± SEM; 4.63 ± 0.55). Similarly, when OB-derived-astrocytes were opposite those from the RMS the M.I. was higher for OB (2.31 ± 0.21). Finally, RMS-derived-astrocytes promoted a higher migration than Cx-derived-astrocytes (1.71 ± 0.09). In all cases the side ratios were greater than 1.0, indicating that any soluble factors released into the culture medium did not contribute significantly to migration. This relationship was maintained when only one RMS-explant was placed with a choice of two alternative astrocyte monolayers in the same culture well (Fig. 3D,E). The striking asymmetric migration onto the RMS-substrate rather than the Cx-substrate (Fig. 3D) was not evident with RMS-explants on paired Cx-monolayers (Fig. 3E).

Open in a separate window

Fig. 3

Differences in neuroblast migration on different astrocyte monolayers are not mediated by soluble factors. (A) Two-explants cultured on different astrocytes sources stained with bisbezimide. Dotted lines delimited the migration limits for each explant. (B) Dotted square in (A) shows the free cell line between the different astrocytes sources. (C) Graph corresponding to the neuroblast migration ratio between two RMS-explants plated on two different astrocyte monolayers in the same well. (D) RMS-explant cultured at the boundary between RMS and Cx-astrocytes stained for Tuj1 (green) and GFAP (red). Note the difference in migration is maintained even when the RMS-explant is positioned in between different astrocyte monolayers. (E) RMS-explant plated in between two different Cx astrocyte monolayers. Note there are no significant differences in the migratory area. (F) Cartoon representing co-cultured RMS-explants on collagen over astrocyte monolayers. (G) RMS-explant co-cultured in collagen over astrocytes stained with anti-TuJ1. (H) Comparison of the migration index in the collagen co-culture on different astrocyte monolayers. W/O indicates assay without astrocytes. Error bars show SEM. Scale bar = 1.5 mm in A; 300 µm in B; 200 µm for D-E; 100 µm in G.

To further test whether soluble factors affect migration, the RMS-explants were isolated from the underlying astrocyte monolayers. The explants were plated on a plastic coverslip embedded in a three-dimensional collagen matrix and suspended above the astrocyte monolayers by three pillars that prevented the collagen matrix, containing the explant, from making direct contact with the monolayer (Fig. 3F). After 3 DIV, the explants were immunostained for TuJ1 (Fig. 3G). Under these conditions, there were no significant differences in the migration between OB-, RMS-, and Cx-derived astrocyte monolayers (Fig. 3H), even when compared with cultures without monolayers (P = 0.309). Thus, contact mediated factors exert a more decisive influence on the migration differences than soluble cues produced by the astrocyte monolayers.

Dynamics of Neuroblast Migration: Time-Lapse Video Microscopy Study

To explore the dynamic temporal and cytological aspects of the differences in M.I., we analyzed the behavior of RMS neurons using time-lapse video-microscopy. We compared the migratory behavior of cells from RMS-explants cultured on OB-, RMS- or Cx-derived-astrocyte monolayers, capturing frames at 3-min intervals and tracking individual cells for the last 6 h in the culture. Under these conditions, differences in the migration of neuroblasts on the three substrates were immediately apparent, even a short time after plating the RMS-explant on the different monolayers (Fig. 4A–C; Supp. Info. Movies 1–3). After tracking 30 cells, their trajectories were plotted from the origin point (Fig. 4D–F; Supp. Info. Movies 4–6); these trajectories occupied areas consistent with our previous results (Fig. 4D–F).

Open in a separate window

Fig. 4

Time-lapse videomicroscopy of RMS-explants cultured on different astrocyte monolayers. (A–C) Neuroblast migration on astrocyte monolayers from OB (A), RMS (B) and Cx (C). The extension of cell migration is greater in OB- and RMS-derived astrocytes than on Cx-astrocytes. (D–F) Individual cell trajectories plotted over the last 6 h of the culture and adjusted to a common origin point. (G) Average speed of cell migration during the 6-h period. (H) Average of the maximum distance traveled away from the explant during the 6-h period. (I) Average of the cell stationary periods during the 6 h. (J) Average of the ratio between the total distance covered and the maximum distance reached from the migratory origin. (K) Histogram of the step length shown in percentages. Each step is defined as 3-min period. (L) Graph showing the temporal average cell distance from the migration origin. (M) Graph represents the relationship between the number of paused cells and time. A linear regression analysis was used to evaluate the relationship between the number and the time of stopped cells. The slope was statistically non-zero for RMS and Cx-derived-astrocytes (P < 0.0001 in both cases), but not for OB-derived-astrocytes (P = 0.424). Horizontal lines between groups indicate significant differences. Error bars show SEM. Scale bar = 50 µm for A–C; 75 µm for D–F.

We compared the average speed of RMS cells in the final 6 h of a 24-h culture period and RMS neuroblasts migrated faster on OB- than on Cx-derived-astrocytes (mean ± SEM; OB = 72.4 ± 2.93 µm h⁻¹; Cx = 57.47 ± 3.09 µm h⁻¹; P < 0.05). No significant differences were observed in migration speed on the OB- and RMS-derived-astrocytes (64.40 ± 3.7 µm h⁻¹) or RMS-and Cx-derived substrates. We then quantified the maximum net displacement during the 6-h period (Fig. 4H), although no significant differences were observed for this value on the different astrocyte substrates (P = 0.06). Since cell migration is interrupted by periods of no movement (Bovetti et al., 2007a; Wichterle et al., 1997), we analyzed the pauses in cell migration relative to the migratory substrate (Fig. 4I). RMS neuroblasts paused for significantly longer periods of time on Cx- than on OB-derived-astrocytes (P < 0.01).

To compare the trajectories from RMS-explants on the different monolayers, we analyzed the ratio between the trajectory and the net displacement, defined as the “winding index” (Fig. 4J). Cells with a lower “winding index” migrated in a straight path from the start to the end point, while those with a higher index followed a more tortuous route. However, there were no significant differences in this index between the RMS cells cultured on the different monolayers (P = 0.821).

To explore the mechanisms underlying the migratory behavior of these cells along the time course, we next analyzed the dynamics of different parameters during 6-h sequences. We determined the distance traveled by neuroblasts in each 3-min interval (referred to as a step) and represented in a histogram as percentage step distance (Fig. 4K). The most repetitive distance interval was 2 µm for all astrocyte sources. However, in relation to the staccato migratory behavior described for RMS neuroblasts, the number of cell pauses (defined as distances less than 1 µm) occupied an important step percentage. We found that RMS cells co-cultured on Cx-derived-astrocytes had the highest percentage of pauses, represented as a 0 value in the step length axis (Fig. 4K). These results were consistent with the preceding analyses in which such pauses were also more frequent on Cx-derived-astrocytes.

We then analyzed the average distance traveled by cells from the point of origin in relation to time. Cell dynamic curves on the different monolayers show that, while the average distance increased linearly on OB-, RMS- and Cx-derived-astrocytes, the distance traveled by RMS cells saturated ~200 min from the start point (Fig. 4L). This correlates with the increase in the number of paused cells in relation to the culture time on both RMS- and Cx-derived-astrocytes (P < 0.0001), but not for explants on OB-derived-astrocytes (P = 0.424; Fig. 4M).

In summary, time-lapse experiments revealed many dynamic parameters related to the cell migratory behavior that help to explain the differences in migration on OB- and Cx-derived astroglia seen in the statically analyzed cultures.

Astroglia Heterogeneity Differentially Affects Neuroblast Migration

Our results demonstrate that astrocytes isolated from permissive areas support more extensive migration than those from nonpermissive areas. Throughout life SVZ neuroblasts migrate toward the OB in complex chains ensheathed by astrocytes (Lois et al., 1996; Peretto et al., 1997). Three main lines of evidence suggested that astroglia were not essential for migration: (1) mature astroglia are not present during the early period of neurogenesis (Law et al., 1999); (2) SVZ explants exhibit chain migration in Matrigel in the absence of astroglia (Wichterle et al., 1997); and (3) specialized contacts exist between neuroblasts but not between astrocytes and neuroblasts (Lois and Alvarez-Buylla, 1994). By contrast we show here, together with previous results (Alves et al., 2002; Pencea and Luskin, 2003), that GFAP is expressed in the SVZ/RMS by the first postnatal week, suggesting that astrocyte progenitors could substitute for the absence of mature astrocytes (Alves et al., 2002). In addition, although chain migration can occur in the absence of astroglia, Matrigel is a mixture of ECM components secreted by EHS tumors (Kleinman and Martin, 2005) that may substitute for glial-derived factors. Thus, although there is no evidence for specialized contacts within the RMS, astrocytes may regulate cell migration directly as a cellular scaffold, or, by providing a molecular scaffold similar to artificial substrates in vitro. Consistent with this hypothesis (see Fig. 5G), ECM proteins are abundant in RMS (Thomas et al., 1996) as well as in OB, where an ECM layer appears to be interposed between astrocytes and neuroblasts (Bovetti et al., 2007b).

Consistent with the new data we present here, additional lines of evidence support the notion that astrocytes may be a pivotal determinant of RMS neuroblast migration. For example, neuroblast migration is 20% faster in chains within glial tubes than in glial-free environments (Bovetti et al., 2007a). Similarly, it is slower in nonglial ensheathed parenchymal chains (Luzzati et al., 2003). Moreover, astrocytes enhance migration by regulating extracellular GABA (Bolteus and Bordey, 2004) and releasing a soluble protein (Mason et al., 2001). Finally, astroglial tunnels may constitute a physical barrier that defines boundaries preventing the exit of migrating cells (Alves et al., 2002; Jankovski and Sotelo, 1996).

The role of glia in patterning and boundary establishment is known elsewhere in the CNS. Thus, radial glia delimits migratory neuronal fate with regard to process orientation (Noctor et al., 2001). Similarly, differences in permissiveness of medial- and lateral-glia promote a barrier for axonal growth across the mid-brain midline (Garcia-Abreu et al., 1995, 2000). Furthermore, astrocytes prevent neurite extension beyond the “walls” of the barrels in somatosensory cortex (Steindler et al., 1995). In addition, astrocytes play a prominent role in creating an environment that is nonpermissive to axonal growth in lesions (Stichel and Müller, 1998), while in a complementary fashion cortical or striatal astrocytes attract more axons in a neural co-graft than ventral mesencephalic astrocytes (Petit et al., 2001).

The heterogeneity among astroglia reported here suggests that specific combinations of diverse populations establish migration limits, not only as a physical impediment but also by determining the environment permissiveness. For example, less permissive astroglia outside of the RMS may prevent cells from escaping the pathway rich in permissive astroglia. Moreover, OB-astroglia strongly enhances migration which may help to explain cell-detachment and radial migration from the RMS into the OB. Although several molecular factors influence this process (Hack et al., 2002; Ng et al., 2005; Saghatelyan et al., 2004), their cellular source is not clear. Interestingly, radial migration into the OB in vivo is faster than tangential migration along the RMS (Bovetti et al., 2007a). Thus, a more permissive astroglia population may facilitate migration, especially in the OB where neuroblasts now migrate in isolation rather than as chains which requires establishing contacts with other cellular substrates. Indeed, neuroblasts in the OB are closely associated with astrocyte processes that are apposed to blood vessels that may also contribute to the migratory scaffold (Bovetti et al., 2007b). Cell detachment from the RMS chains could be an integrating phenomenon that involves both molecular chemoattractants and a permissive astroglial population that would enhance migration.

Neurons Display Different Migratory Behavior Depending on the Astrocyte Origin

Time-lapse analyses showed neuroblasts migrate faster and have shorter stationary periods on OB-astrocytes relative to those on Cx-astrocytes. Our estimates of velocity are consistent with in vivo data (Nam et al., 2007). Furthermore, neuroblast stationary periods were augmented on RMS- and Cx-astrocytes, producing a reduction in the average distance from the origin. This could be related to the cell speed decrease in the RMS (Fig. 1C in Bovetti et al., 2007a), or the numerous stops in the more rostral RMS (Nam et al., 2007), suggesting that migration in RMS is limited by astrocyte-derived-factors. In addition, both in vivo (Nam et al., 2007) and in vitro data (see results, Fig. 4D–F) show that RMS is composed of different motile cell classes defined by their trajectories and speed; intragroup variability could mask significant differences for other parameters in time-lapse analyses.

Nonsoluble Factors Associated with Astrocytes are Sufficient to Account for the Differences in RMS Cell Migration

RMS migration is influenced by different soluble and cell-contact molecules (for review Lledo et al., 2006; Menezes et al., 2002). However, our collagen assays failed to identify soluble glial-derived factors that enhance cell migration, as reported in Matrigel or poly-d-lysine rat cultures (Mason et al., 2001). Additionally, our data showed that nonsoluble astrocyte-bound-factors are directly responsible for differences in migration. Accordingly, our data lead to the conclusion that astrocytes can enhance migration through cell–cell contact (Lim et al., 1999). Thus, we examined the expression of ECM and cell-adhesion related-genes. By analyzing different astroglial populations, this novel approach revealed new and significant differences among these populations not described in earlier studies (Cahoy et al., 2008). Of importance, most genes that differed significantly in RMS- and OB-astrocytes when compared with Cx-astrocytes were upregulated suggesting a proactive effect.

Diverse factors upregulated in the permissive monolayers were specifically related to RMS migration in vivo. Thus, laminins (Belvindrah et al., 2007; Murase and Horwitz, 2002), integrins (Belvindrah et al., 2007), and metalloproteases (Bovetti et al., 2007a) are decisive in this migration. Collagens in the RMS (Eagleson et al., 1996), are also implicated as scaffolding and regulatory elements. In addition, other proteins including Sparc, expressed in RMS (Vincent et al., 2008), or cadherin/catenin system proteins, implicated in chain migration (Taniguchi et al., 2006), may influence RMS migration. Lack of Thbs1 leads to a widening of the RMS and consequently allows fewer neuronal precursors to reach the OB (Blake et al., 2008). This is even more evident in light of our data, which shows the importance of Thbs1, Ecm1, Lama3, and Ctnna1 as potential mediators in postnatal migratory behavior. Thus, our data suggest that permissive astrocytes could generate a more favorable environment for migration through the upregulation of some specific genes. However, lower migration rates of neuroblast from the RMS explants on nonpermisive astrocyte monolayers, could also be related to inhibitory properties of astrocytes for cell migration. In fact, the presence of inhibitory factors such as chondroitinsulfate proteoglycans surrounding the SVZ/RMS, i.e. NG2-expressing cells, might be preventing neuroblasts from dispersing into the parenchyma (Komitova et al., 2009).

The authors thank Alfonso Araque, Albert Blanchart, and Fernando García-Moreno for helpful comments on the manuscript. They also thank Dolores Montoya and Wolfang Pita for their technical assistance.

Grant sponsor: MICINN; Grant number: BFU2007-60351/BFI; Grant sponsor: OLFACTOSENSE Consortium (CAM); Grant number: P-SEM-0255-2006; Grant sponsor: Castilla-La Mancha Health Research Foundation; Grant number: FISCAM PI2007-66.