During cerebral cortical development, γ-aminobutyric acidergic (GABAergic) interneurons arise from a different site than projection neurons. GABAergic cells are generated in the subpallial ganglionic eminence (GE), while excitatory projection neurons arise from the neocortical ventricular zone. Our laboratory studies a model of cortical dysplasia that displays specific disruption of GABAergic mechanisms and an alteration in the overall balance of excitation in the neocortex. To produce this model, the birth of neurons on a specific gestational day in ferrets (embryonic day 33 [E33]) is interrupted by injection of the antimitotic methylazoxymethanol (MAM). We hypothesized that migration of interneurons might be disrupted in this cortical dysplasia paradigm. We observed that although interneurons migrate into the neocortex in both normal and dysplastic cortex, the migrating cells become disoriented over time after E33 MAM treatment. Coculture experiments using normal GE and MAM-treated cortex (and vice versa) demonstrate that cues dictating proper orientation of migrating interneurons arise from the cortex and are not intrinsic to the migrating cells. As a consequence, interneurons in mature brains of MAM-treated animals are abnormally distributed. We report that GABAA receptor activation is crucial to the proper positioning of interneurons migrating into the cortex from the GE in normal and MAM-treated animals.
A large proportion of neurons that ultimately reside in the cerebral cortex originate from subcortical sources. The medial and lateral ganglionic eminences (GEs), as well as the caudal GE, are important sources for neurons that migrate into the cerebral cortex (Wonders and Anderson 2005). Neurons originating from these sites use several routes to travel tangentially into the neocortex and are reported to comprise the majority of γ-aminobutyric acidergic (GABAergic) neurons (Anderson et al. 1997; Wichterle et al. 2001; Polleux et al. 2002). The mechanisms that mediate the tangential migratory route are beginning to be described and appear to consist of both repulsive and attractive cues (Marin and Rubenstein 2003).
We study a model of cortical dysplasia in which layer 4 cells in the somatosensory cortex of the ferret fail to form and migrate into the neocortex (Noctor et al. 1999; Palmer et al. 2001). This is accomplished by injection of an antimitotic methylazoxy methanol (MAM), into a pregnant ferret on the gestational day when the bulk of layer 4 cells are generated (embryonic day 33 [E33]). This treatment results in several consistent features that include a marked reduction in the number of cells in layer 4, whereas the cells in other layers remain similar to normal cortex in size and density (Noctor et al. 2001). In addition, the thalamic afferent fibers do not focus on layer 4 in the MAM-treated animals but disperse throughout the cortical layers. GABAA receptors also distribute over multiple layers compared with their usual focus in layer 4 (Palmer et al. 2001; Jablonska et al. 2004). The topography of the body map is normal in MAM-treated ferret cortex, but the ability to process complex stimuli through the cortical layers is severely impaired after E33 MAM treatment (Noctor et al. 2001; McLaughlin and Juliano 2005). It should be noted that the timing of MAM injection is crucial in ferrets, and treatment on earlier gestational days results in dramatically different and more severe effects when injected, for example, on E24 (Noctor et al. 1999; Hasling et al. 2003).
Our findings suggest that the balance of excitation and inhibition and aspects of GABAergic elements are disturbed after injection of MAM on E33, which led us to question the ability of neurons leaving the GE to migrate properly into the neocortex. We conducted a series of experiments to evaluate the capacity of neurons to migrate effectively from the GE of normal and MAM-treated ferrets. Ferrets are excellent animals to use for a study of migration; because they are altricial, many neurons are born and migrate into the neocortex for 7–10 days postnatal, depending on the rostrocaudal region (Jackson et al. 1989; Noctor et al. 1997). We report here that neurons leaving the GE of animals with a relative absence of layer 4 migrate less effectively than in normal animals. This impairment involves specific changes in activity levels due to abnormal GABAA signaling affecting the ability of cells leaving the GE to successfully reach their target.
Materials and Methods
Timed pregnant ferrets were purchased from Marshall Farms (New Rose, NY). Normal fetuses at E27–E28, E33–E34, or E38 (the gestation period of ferret is 41–42 days) were obtained by caesarean section and their brains removed. The caesarean section was performed under sterile conditions using isofluorane anesthesia under the supervision of a veterinarian. We also used newborn (P0) and P28 ferrets: normal or treated previously with an antimitotic MAM acetate (MAM, Midwest Research Institute, Kansas City, MO) on the 33rd day of gestation (i.e., E33 MAM-treated animals). Pregnant ferrets, anesthetized with isofluorane (1%), were injected intraperitoneally (IP) with MAM (16 mg/kg) diluted in a sterile saline buffer. P0 and P28 ferrets (normal or E33 MAM-treated) were anesthetized with an IP injection of pentobarbital sodium (50 mg/kg) prior to brain removal.
After brain removal and under sterile conditions in a laminar airflow hood, the brains were cut into 500-μm-thick coronal slices using a tissue chopper (Stoelting, Wood Dale, IL). During the dissection, brains and slices were perfused with cold and oxygenated artificial cerebrospinal fluid (containing in mM: CaCl2 2.4, KCl 3.2, MgSO4 1.2, NaCl 124, NaHCO3 26, NaH2PO4 1.2, glucose 10). Slices containing the GE and the cortical plate (CP) were cultured on inserts (Millipore, Bedford, MA) and placed into 6-well plates in Minimum Essential Medium Eagle (Gibco, Carlsbad, CA) containing 10% decomplemented fetal bovine serum (Gibco) and 4% G1,2 solution (0.5 mg/mL gentamycin, 15% glucose, 50 mM L-glutamine). In some cases, the following drugs were added to the medium: bicuculline (BIC, 10 μM; Sigma, St Louis, MO), muscimol (10 μM, Sigma), tetrodotoxin (TTX, 1 μM; Calbiochem, La Jolla, CA). To prepare cocultured slices (at E38 or P0), explants of normal GE were paired with E33 MAM-treated cortical explants and vice versa. To label neurons migrating tangentially, crystals of DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) (Molecular Probes, Eugene, OR) were placed in the GE, or CellTracker™ Orange CMTMR (5-(and-6)-(((4-chloromethyl)benzoyl)amino) (Molecular Probes) was injected in the GE using a micropipette and a microinjector (World Precision Instruments, Sarasota, FL). The organotypic slices were incubated between 2 and 5 days in culture (DIC) in an incubator (95% O2; 37 °C), and then the slices were fixed overnight by immersion in a solution containing 4% paraformaldehyde in 0.1 M phosphate buffer.
Explants of GE in Matrigel
Coronal slices were cut with a tissue chopper in normal animals (E27, E33, P0). Explants from the ventricular zone (VZ)/subventricular zone of the GE were microdissected and embedded in Matrigel (BD Biosciences, Bedford, MA). After 5 DIC (same medium described previously), the explants were fixed overnight in 4% paraformaldehyde in 0.1 M phosphate buffer.
Peroxidase immunocytochemistry was performed on 75-μm-thick vibratome cut coronal sections from brains obtained at E27, E33, P0, or P28. The brains obtained at E27 and E33 were fixed by immersion in 4% paraformaldehyde for 1 day prior to cutting. At P0 and P28, the animals were perfused transcardially after injection of 50 mg/kg of pentobarbital Na and their brains removed and placed in a solution of 4% paraformaldehyde. After cutting, the sections were incubated overnight at 4 °C with rabbit IgG polyclonal antibodies against GABA (1/200, Sigma) or mouse IgG monoclonal antibodies against calcium-binding proteins (calbindin, 1/200, Swant, Bellinzona, Switzerland; calretinin, 1/1000, Swant; parvalbumin, 1/1000, Swant) diluted in a buffer containing phosphate-buffered saline (PBS), bovine serum albumin (BSA, 2%), and Triton X-100 (0.1% or 0.05% for double labeling with DiI). The sections were washed in Tris buffer (0.05 M, pH 7.6) and incubated with the appropriate secondary antibodies (Vectastain ABC kit, Vector Laboratories, Burlingame, CA). Peroxidase was revealed with 3,3′-diaminobenzidine (0.1%) diluted in Tris buffer with H2O2 (0.2%). The sections were washed and mounted in Mowiol (13% Mowiol, 10% glycerol in 0.2 M Tris buffer, pH 8.5). For fluorescence immunocytochemistry, the explants or slices were incubated overnight at 4 °C with rabbit IgG polyclonal antibodies against GABA, mouse IgG monoclonal antibodies against all α subunits of the GABAA receptor (1/50, Chemicon, Temecula, CA), rabbit IgG polyclonal antibodies against Tbr1 (1/1000, kindly provided by R. Hevner), or mouse IgG monoclonal antibodies against glutamate decarboxylase 65/67 (GAD65/67) (1/1000, Chemicon). After washes in PBS, the corresponding secondary antibodies were used (anti-rabbit or anti-mouse Alexa-488 or Alexa-546, 1/200, Molecular Probes).
For birthdating studies, normal and E33 MAM kits (P0, n = 1 for each or P2, n = 2 for each) were injected IP 3 times at 45 min intervals, with bromodeoxyuridine (BrdU; 50 mg/kg, Sigma) diluted in saline. The ferrets, normal and E33 MAM-treated, were perfused at P0 (BrdU injection at P0) or at P28 (BrdU injection at P2) with 4% paraformaldehyde in phosphate buffer and 4% sucrose. Slices were cut on a vibratome at 100 μm thickness. For BrdU immunoreactivity combined with GAD immunoreactivity, the slices were incubated 3 days at 4 °C in PBS containing BSA (2%), Triton X-100, and the polyclonal rabbit anti-GAD65/67 (1/1000, Chemicon). After washes in PBS, the slices were incubated 1 h at 37 °C in HCl 2N, then rinsed in 0.1 M borate buffer, pH 8.5, for 10 min. After again washing in PBS, the slices were incubated for 3 days at 4 °C in PBS containing BSA (2%), Triton X-100, and polyclonal rat anti-BrdU (1/50, Accurate Chemical and Scientific Corp., Westbury, NY). The slices were washed and incubated with the secondary antibodies: Alexa Fluor 546–conjugated goat anti-rabbit IgG and Alexa Fluor 488–conjugated goat anti-rat IgG.
To detect apoptotic cells containing fragmented DNA, 50 μm vibratome sections were processed for TUNEL staining (Boehringer Mannheim, Indianapolis, IN). Sections from normal or MAM-treated ferrets (P0) were incubated for 60 min at 37 °C in a mixture containing terminal deoxynucleotidyl transferase plus nucleotides labeled with fluorescein. Sections were mounted in Mowiol. The number of TUNEL+ cells was counted in the GE VZ in a field of 250 × 275 μm. Two fields per slice were counted. Three slices for each animal were analyzed. Data were pooled from 3 normal or MAM-treated ferrets (P0). Statistical significance was determined using a Mann–Whitney Test.
Quantitative Analysis of Distribution
To quantify the laminar distribution of the cells, images of neocortical sections were divided into 10 equal bins from the pial surface to the VZ (P0) or to the white matter (P28). The number of cells in each bin is plotted as a percentage of the total number of cells. Statistical analysis was conducted using a 2-way analysis of variance (ANOVA) followed by pairwise multiple comparison procedures (Holm-Sidak method). Values represent mean ± standard error of the mean (SEM).
Quantitative Analysis of Directionality
To assess the directionality of migrating neurons, the orientation of the leading process of a labeled and migrating cell was measured in relation to the pia. A 500-μm wide band of cortex close to the boundary between the cortex and the GE was used for measurement in each case (Supplementary Fig. 1A). Migrating cells were analyzed through the full thickness of the band, excluding the VZ. An angle was calculated for each process and placed into a bin that contained the 90° (out of 360°) either oriented radially (either toward the pia or toward the ventricle) or oriented tangentially (medially toward the midline or laterally) (Supplementary Fig. 1B). Orientations measured either radially toward the pia or tangentially in the medial direction were considered the “proper” directions. A chi-square contingency test was used to determine differences in the directionality of migration between various ages and treatments.
Calculation of the Crossing Index
DiI crystals were injected into GE, and the organotypic cultures were maintained 5 DIC. After fixation of the slices, the crossing index was determined as described by Cuzon et al. (2006). This evaluated whether tangentially migrating cells successfully invade the neocortex in MAM-treated compared with normal slices or after pharmacological treatment compared with nontreated slices. A 500-μm wide section of cortex medial to the corticostriate junction (CSJ) was divided into 5 bins of 100 μm in width (Supplementary Fig. 1C). Adjacent and lateral to these bins is a single bin 200 μm in width (shaded in Supplementary Fig. 1C). The crossing index was calculated by dividing the number of cells in the first wider bin, located at the CSJ into the number of cells in each of the more medially positioned 5 bins. Statistical analysis used a 2-way ANOVA followed by pairwise multiple comparison procedures (Holm-Sidak method). Values represent mean ± SEM.
Normal Pattern of Migration in Embryonic and Postnatal Ferrets
In our initial experiments, we compared migration patterns from the medial and lateral GE of the ferret. We primarily studied ages after MAM treatment (E33) because the ferret embryos were otherwise normal prior to this date. At E33 or older, we found it difficult to distinguish a medial eminence. At earlier time points (i.e., prior to E33), a clear medial GE can be seen, which fuses with the lateral eminence by E33 or earlier. Supplementary Figure 2 shows examples of rostrocaudal sections of ferret brains at different embryonic and postnatal ages indicating that although a medial eminence is present at early stages of corticogenesis in the ferret (E27), it fuses with the lateral ganglionic eminence (LGE) by E33 and a single eminence is present until postnatal ages. In addition, when we placed DiI crystals at more medial sites at any age, very few, if any, neurons migrated away from the injection site, while large numbers of cells migrated toward the neocortex when injected into lateral portions of the GE. These observations suggest that in the ferret, an important site of origin for neurons that migrate tangentially from the subpallium into the neocortex is the single GE. In addition, the bulk of our experiments were conducted after E33 because we compared the migration patterns between normal and E33 MAM-treated animals. The Dil injections into ferret organotypic cultures described in this study were made into lateral parts of the single GE either after E33 or in experiments that included dates before E33.
At all ages studied in the normal animal (E27, E33, and P0), numerous cells leave the GE and migrate into the neocortex after injections of DiI into organotypic cultures; the pattern of migration was similar to that already described by others (Anderson et al. 2001; Wichterle et al. 2001). After 2 DIC, the cells leave the GE in 2 rough streams, best seen in Figure 1C, one stream in the intermediate zone (IZ), but close and running parallel to the neocortical VZ, and one that is just deep to the developing CP (Fig. 1C). Cells run in a plane with these sites or turn and move vertically into the CP. After 5 DIC, the 2 streams are still visible after DiI injection at E33/34 and less so when the dye is injected at P0 (Fig. 1A,B,D–G). We observed that neurons leave the GE in large numbers up to P2 (data not shown); we did not assess dates later than this.
Normal Distribution of GABAergic Neurons
Because a large percentage of neurons leaving the GE are reported to be GABAergic, we evaluated the distribution of GABA immunoreactive cells at ages comparable to our study of tangentially migrating neurons in normal ferret brain. GABA+ cells are morphologically similar to the migrating neurons in the organotypic cultures and also occur in 2 rough streams at E27 and less so at P0. Figure 2A–D demonstrates that GABA+ cells appearing to migrate are present in the IZ and more superficially in the CP. Distinct ventral and dorsal streams of GABAergic cells are more easily visible on E27 than on P0, although a collection of cells in the more ventral position can still be seen at P0 (Fig. 2D).
To determine if the cells leaving the GE of normal ferret are primarily GABAergic, we placed GE explants obtained at E27, E33, and P0 in Matrigel and allowed cells to migrate out for 5 days. The explants were stained with bisbenzimide to identify cells and counterstained using antibodies directed against GABA. Figure 2E–H shows that the majority of cells leaving the GE are GABA+ for all developmental ages studied, and the relative percentage of GABAergic neurons remains stable at different ages.
Distribution of Cells Leaving the GE in E33 MAM-Treated Brains
Because we noticed disrupted GABAergic elements in E33 MAM-treated brains, we assessed the distribution of cells originating in the GE in organotypic cultures of E33 MAM-treated cortex. Since MAM is injected on E33, we would expect no differences between groups before or on this date, we therefore only assessed the P0 time point. After injection of DiI into the GE of P0 organotypic cultures, the pattern and the number of migrating cells after 2 DIC look similar to the normal animal. There are 2 streams situated roughly ventral, paralleling the VZ, and roughly dorsal, underlying the CP (Fig. 3A). To determine if migrating neurons leaving the GE in MAM-treated animals reached the CP in similar proportion to the normal animal, we quantified the distribution of neurons leaving the GE. The total number of these neurons was assigned a bin according to their position en route to the cortex. This distribution indicated that equal amounts of migrating neurons were found between the VZ and into the CP in normal and MAM-treated organotypic cultures at P0 (Fig. 3B, graph on the left). We also determined the crossing index to assess possible differences in the number of cells entering the neocortex in normal versus MAM-treated cultures (as described in Supplementary Fig. 1C). The crossing index shows no significant difference between the 2 groups (Fig. 3B, graph on the right). Altogether, these data strongly suggest that when comparing normal and MAM-treated brains, cells leaving the GE show the same migratory pattern and invade the neocortex in similar numbers.
In the distribution of migrating cells after 5 DIC, however, labeled cells in the MAM-treated organotypic cultures appeared oriented differently than those in the normal animal (Fig. 3C,D). To verify this observation, we quantified the orientation of migrating, DiI-labeled cells by measuring the angle of each leading process as described in the Materials and Methods. This analysis demonstrates that after 2 DIC, the cells leaving the GE in both normal and MAM-treated cortex are similarly oriented. After 5 DIC, however, fewer cells orient their leading processes toward the pia or the medial edge of the hemisphere (i.e., the “correct” orientations) in MAM-treated cortex (Fig. 3E). Therefore, as the cells leave the GE from both normal and MAM-treated cultures during the first 2 days, they initially travel tangentially with similar orientations. As the migration continues, the cells move radially into the CP and those entering the MAM-treated cortex become more disoriented en route. A chi-square contingency analysis indicates that the directionality of cells migrating in normal and E33 MAM-treated organotypic cultures after 2 DIC is not different from one another, whereas after 5 DIC significantly more cells travel in “incorrect” directions in the E33 MAM-treated cultures than in the normal cultures (P < 0.001). This suggests that migrating cells in the MAM-treated brains are oriented properly as long as they migrate strictly tangentially (i.e., at 2 DIC). Once they turn toward the CP, however, they lack one or several factors necessary to orient properly toward the CP. We also emphasize that equal numbers of the migrating cells in normal and MAM-treated organotypic cultures reach the CP (Fig. 3B) but those in MAM-treated slices are disoriented.
To assess in greater detail the migratory dynamic of interneurons in normal and MAM-treated organotypic cultures, the orientation of the leading process was evaluated depending on the position of a migrating cell within the cortical wall, including the marginal zone (MZ), CP, or IZ (Fig. 4). The percentage of cells in all orientations in each region is shown in Table 1. These data are presented in a semi quantitative manner in Figure 4. The primary direction of migration is indicated with a large arrow in each zone; if no specific direction emerged, smaller arrows were used to demonstrate multiple directions. The first distinction between the 2 groups emerges after 2 DIC, when cells leaving the GE in MAM-treated cortex are not primarily oriented radially toward the pia in the CP but migrate in a tangential direction (Fig. 4A,B). After 5 DIC, the pattern of migration diverges even more strongly between normal and MAM-treated cultures. In normal organotypic cultures, cells continue to move tangentially in the MZ or toward the CP in all regions (Fig. 4C). In MAM-treated cultures, however, the cells have no dominant direction of movement in the CP or IZ, whereas cells in the MZ orient predominantly opposite to the tangential direction observed in the normal cultures (Fig. 4D). This diversion from the normal pattern suggests that cues available to the normal slice are missing from the MAM-treated cortex.
|Orientation (% of cells)||2 DIC||5 DIC|
|B. E33 MAM|
|Orientation (% of cells)||2 DIC||5 DIC|
|B. E33 MAM|
Apoptosis and Proliferation in the GE of Normal and MAM-Treated Ferrets
To assess if a MAM injection performed at E33 can have a prolonged negative effect on cell proliferation in the GE at P0 (1 week after the MAM injection), we compared the number of apoptotic cells in the VZ of the GE in normal and MAM-treated animals (3 animals of each analyzed) as revealed by the TUNEL method (Supplementary Fig. 3A). We found no significant differences in the number of TUNEL-positive cells in the GE of normal and MAM-treated animals (Supplementary Fig. 3B). To assess the state of proliferation in both normal and MAM-treated animals, BrdU was injected at P0 and the animals were then fixed and processed for BrdU immunoreactivity later. Supplementary Figure 3C shows that roughly equal numbers of cells are produced in the GE of normal and MAM-treated animals. These results suggest that MAM injections in pregnant ferrets at E33 do not affect the survival or the proliferation of progenitors in the GE at P0.
Distribution of GABAergic Cells in Juvenile Ferret Cortex
Because the tangentially migrating cells were disoriented after MAM treatment, we assessed their distribution in more mature cortex, using specific markers for interneurons. We examined the parietal cortex in somatosensory area 3b, the cortical region where we conducted all previous quantitative studies (Noctor et al. 1997, 1999, 2001; Palmer et al. 2001; McLaughlin and Juliano 2005). Cells immunoreactive for GAD65/67 were plotted across the depth of the cortex in normal (n = 7) and MAM-treated brains (n = 8) at P28. The thickness of the cortex was divided into 10 equal bins, and the number of immunoreactive cells found in each bin plotted as a percentage of the total number of labeled cells. The distribution of the GAD+ cells in the MAM-treated cortex differed from the normal distribution (Fig. 5A,B). In both groups, immunoreactive cells are present in all layers but are more highly represented in the upper layers in normal cortex. In MAM-treated cortex, however, GAD+ cells fail to congregate in the upper layers and overall more cells were found in the lower and middle parts of the cortex. A statistical comparison between the 2 distributions indicates that they differ significantly from one another (2-way ANOVA, P < 0.05). This difference between the pattern of normal and MAM-treated GAD+ cells suggests that cells migrating in the MAM-treated brain are less effective in reaching their target in the upper layers and that more cells remain “stuck” in the lower layers. We note that the difference in the pattern of distribution does not reflect the total number of neurons, which are not significantly different (student t-test) in the normal (57 ± 7.8) and MAM-treated cortex (49.5 ± 8).
We also evaluated the distribution of GABAergic cells according to birthdate. BrdU was injected into normal and MAM-treated kits on P2 (n = 2 for each); at this age, cells populating layer 2 of somatosensory cortex are being generated (Noctor et al. 1997). We evaluated the position of BrdU+ cells in normal and MAM-treated cortex at P28. Histograms generated in a fashion similar to those described above indicated that the vast majority of BrdU+ cells in the normal brain reside in layer 2 (bins 1–3), with a very small percentage distributed through the deeper layers (11.59%). In the MAM-treated brains, although the largest fraction of cells are in the appropriate layer, 44.70% of the BrdU+ cells distribute throughout the cortical thickness outside of layer 2, suggesting they have not migrated properly to their target layer (Fig. 5C,D). Note that the number of BrdU+ cells is not different in normal versus MAM-treated animals (141 ± 8 and 158 ± 29, respectively). This result indicates that at P2, proliferation is not affected in the GE of MAM-treated animals (see also Supplementary Fig. 3C).
To determine if a portion of the misplaced cells were GABAergic, the sections with BrdU+ cells were also immunoreacted with antibodies against GAD. When the position of these double-labeled cells were plotted, although the majority of the BrdU/GAD+ cells were in the appropriate site (i.e., layer 2) in both normal and MAM-treated cortex, 14.54% of the double- labeled cells were outside of the target layer 2 in the normal animal, compared with 31.34% of similar cells outside of layer 2 in MAM-treated cortex (Fig. 5E,F). This change in laminar distribution of cells generated on P2 helps to explain the diminished number of GAD+ cells in the upper layers of the overall pattern shown in Figure 5A,B.
To further evaluate the differentiation and distribution of interneurons in normal versus MAM-treated animals, we studied the distribution and morphology of cells expressing selected calcium-binding proteins: calretinin, calbindin, and parvalbumin. Overall, the distributions of calcium-binding proteins in treated cortex suggest a reallocation of GABAergic cells after MAM delivery, especially for the laminar profile of calbindin and parvalbumin (Fig. 6A,B). For these markers, more immunoreactive cells cluster in the central regions of cortex, and fewer cells reach the upper cortical layers compared with the normal pattern. We tested the correlation coefficient and covariance of the distributions of labeled cells and found that the distributions of calbindin and parvalbumin were significantly different between normal and MAM-treated brains. The distribution of calretinin cells in MAM-treated cortex, however, does not differ significantly from the normal distribution. This is interesting, as several studies suggest that interneurons expressing calretinin originate from the caudal eminence and are not dependent on the transcription factor Nkx2.1 for their specification, as do parvalbumin containing cells (Xu et al. 2004, 2005).
The size and the morphology of the interneurons expressing calcium-binding proteins do not appear different in normal versus MAM-treated animals. This suggests that if subpopulations of interneurons (calbindin or parvalbumin) fail to reach their proper target, their differentiation is not affected because they acquire the proper morphology as observed in normal animals (Fig. 6C).
Are Large Axonal Pathways Involved in Mediating the Orientation of Migrating Cells?
We demonstrated in a previous study that thalamocortical axons were disturbed in E33 MAM-treated somatosensory cortex. These axons, however, are relatively normal until they grow into the CP (i.e., around P7); at P0–1, they appear to be distributed as in the normal animal (Noctor et al. 2001; Palmer et al. 2001). This makes it unlikely that they are the source of disorientation. Another possibility is that corticothalamic axons are disturbed in this model because they have been implicated in mediation of tangential migration (Denaxa et al. 2001). To determine if corticofugal fibers appeared disrupted after E33 MAM treatment, we placed DiI crystals in the cortex of normal and E33 MAM-treated brains at P0 according to the method described by Molnar and Cordery (1999). The pattern of descending projections in MAM-treated cortex appears normal; the axons descend, turn sharply in a lateral direction in a thick band, and descend toward the internal capsule (Supplementary Fig. 4A). This is also not surprising because the bulk of corticofugal axons are generated from cells in layers 5 and 6, which are established prior to the MAM injection on E33. In another evaluation of the quality of layers 5 and 6 after MAM treatment, we assessed the distribution of Tbr1 cells at P0 in normal and E33 MAM-treated animals; Tbr1 is a transcription factor gene of the T-box family expressed in early-born neurons of the preplate and layer 6 (Bulfone et al. 1995). The distribution of Tbr1+ cells was similar in normal and MAM-treated cortex at P0, suggesting that lower layer cells are normally distributed in MAM-treated cortex (Supplementary Fig. 4B).
Are the Alterations Observed in MAM-Treated Cortex Intrinsic to the Interneurons?
Because the problem with migration orientation did not appear to occur from axonal pathways, we wanted to determine whether the source of disorientation in the MAM-treated brains originated in the cells leaving the GE or resulted from cues originating in the cortex. To this end, explants of GE and cortex were obtained from normal and MAM-treated brains at either E38 or P0 and paired isochronically so that a MAM-treated GE was situated with a piece of normal cortex and vice versa (Fig. 7A–C). Crystals of DiI were placed in the GEs, and the cocultures remained for 5 DIC. Because approximately equal numbers of cells migrate into the CP in both normal and MAM-treated cortex, the goal of this experiment was to determine whether the influence on the orientation of the migrating cells initiated from cortical cues or was intrinsic to the cells themselves. The angle of the leading process of the migrating cells was determined for all combinations of cocultures as described earlier. This analysis indicates that the factor determining if the GE cells migrate in the proper orientation is whether the cortical explant is normal or MAM-treated. More cells that migrate into normal cortex are oriented properly than those migrating into MAM-treated cortex, whether the cells originate in normal or MAM-treated GE (Fig. 7A–C). A chi-square contingency analysis demonstrated a significant difference between the directionality of migrating cells in cocultures of normal cortex and MAM-treated GE compared with those in MAM-treated cortex and normal GE (P < 0.001).
Do Overall Activity Levels Influence the Orientation of Migrating Cells?
Many of our findings suggest that overall activity levels are altered in MAM-treated cortex. This notion is strengthened by the altered distribution of interneurons in juvenile ferrets (Figs 5 and 6) and previous findings of changes in GABAA receptor distribution and cortical response profiles (Jablonska et al. 2004; McLaughlin and Juliano 2005). In the embryonic mouse, cells migrating tangentially express GABAA receptors (Soria and Valdeomillos 2002). To assess if tangentially migrating cells also express GABAA receptors in ferrets, we labeled migrating cells by placing crystals of DiI into the GE of normal slices. After 5 DIC, the slices were fixed and immunoreacted against the α chain of the GABAA receptor. DiI-positive cells leaving the GE clearly express GABAAα receptors (Supplementary Fig. 5).
To determine if altering overall activity levels in normal cortex mimicked the impairment in orientation seen in MAM-treated cortex, we exposed normal cortical organotypic slices to BIC (10 μM), muscimol (10 μM), or TTX (1 μM). Each of these drugs has the capacity to alter global activity using different mechanisms. After 5 DIC, the orientation of the DiI-positive cells was assessed as described previously. Disruption of the orientation of migrating cells in normal cultures only occurred in the presence of muscimol (Fig. 8). This information indicates that increased activation of GABAA signaling in normal cortex alters the course of tangentially migrating neurons, while changing global activity patterns using other mechanisms do not modify the neuronal trajectory. BIC had no effect on the orientation of the migrating cells, but using this GABAA receptor antagonist affects their distribution. When organotypic cultures are treated with BIC, fewer DiI-labeled cells reach the upper layers of cortex after 5 DIC (Fig. 9). The number of migrating cells is statistically increased in the VZ but does not modify the crossing index (Fig. 9D). This suggests that although the migrating interneurons are able to enter the CP, to reach their proper position within the neocortex, these cells rely on appropriate activation of GABAA receptors.
We developed a model of cortical dysplasia in ferret where layer 4 is severely diminished after exposure to MAM on E33. Although many human dysplasias have genetic causes, the majority of children born with cortical dysplasia cannot be associated with a specific gene and are presumed due to environmental or combinations of environmental and genetic causes (Montenegro et al. 2002; Sisodiya 2004). Additional evidence points to trauma and infectious diseases during pregnancy contributing to human cortical dysplasia (Marin Padilla 1999; Deukmedjian et al. 2004). Typical findings in human malformation disorders include altered excitability and disturbances in GABAergic mechanisms (Di Luca et al. 1994; Ross 2002; Calcagnotto and Baraban 2003; Gardoni et al. 2003; Bast et al. 2004; Deukmedjian et al. 2004; Levitt et al. 2004 ). Our study shows that in a ferret model of cortical dysplasia, cortical cues are essential to proper migration and positioning of interneurons arising from the GE. We also report that appropriate GABAA receptor activation is critical to the ultimate orientation and laminar position of these cells.
Tangential Migration in Normal Ferret
Large numbers of GABAergic neurons migrate from the GE into the cerebral cortex during corticogenesis in the ferret (cf., Anderson et al. 1999, 2002). In the rodent, the medial GE appears to be a strong source of tangentially migrating cells, although the LGE has been shown to contribute cells to the CP, especially during later stages of corticogenesis (Anderson et al. 2001). In mice, the medial ganglionic eminence (MGE) starts to diminish at E15 (E15–E16 corresponding to the generation of layer 4; Bayer and Altman 1987) and the fusion of MGE with the LGE is complete by E17 (Jimenez et al. 2002). In ferrets, the fusion of the MGE and the LGE is complete by E33 (layer 4 is generated at E33–E34 in the somatosensory cortex, Noctor et al. 1997). In comparison with rodent gestational days, the ferret fusion is somewhat earlier. In addition, ferret corticogenesis is protracted compared with cortical development in the mouse allowing for an extended time for the production of tangentially migrating neurons. We were predominantly interested in studying tangential migration after E33 because MAM treatment occurs on that day. Our studies show that in the ferret, large numbers of cells leave lateral portions of the GE and migrate into the neocortex at dates from E33–P2, which correspond to generation of layers 2–4 in parietal cortex (Noctor et al. 1997).
Tangential Migration in the Absence of Layer 4
In E33 MAM-treated animals, the ability to migrate tangentially from the GE to the neocortex is maintained and equal numbers of cells reach the CP as in normal animals. After E33 MAM treatment, cells leave the GE in similar numbers, travel in bulk close to the neocortical VZ, and only become disoriented after more than 2 DIC. Marin and Rubenstein (2003) suggest that distinct factors are involved with phases of tangential migration from the GE into the neocortex. Our data suggest that a factor en route to the cortex or in the cortex itself has an effect on the pattern of migration. Evidence supporting this assumption includes our observation that the initial phase of cells leaving the GE after MAM treatment does not show disruption of orientation, which only occurs after 2 DIC. When we analyzed the orientation of tangentially migrating cells according to their position within the cortical wall, we found subtle distinctions after 2 DIC, but only after 5 DIC did dramatic differences occur. In normal cortex, the cells originating in the GE continue to migrate in “correct” orientations, whereas in MAM-treated cortex, the GE-derived cells move in all orientations and appear to lose clues directing them to their proper target.
In addition, normal GE cells migrating into MAM-treated cortex are disoriented in our coculture experiments, whereas GE cells migrating into normal cortex are oriented properly, regardless of their origin. When assessing the detailed pattern of migration across the cortical wall in normal organotypic cultures, we also observed that a substantial proportion of cells migrating in the normal MZ entered the CP as did cells in the IZ; a similar pattern was reported by Tanaka et al. (2003).
Structural Influences on Interneurons Migration
Another reported influence on the ability of interneurons to reach the neocortex is the structural support of specific axonal systems. One specific suggestion for such guidance is the support of corticothalamic axons (Denaxa et al. 2001). Corticofugal axons, however, appear relatively normal in MAM-treated cortex. In addition, the distribution of cells containing Tbr1 is similar in the normal and MAM-treated cortex. This adds to the probability that corticofugal projections are relatively normal in the MAM-treated animals because Tbr1 contributes to the development of these fibers (Hevner et al. 2002). This appears to rule out corticofugal axons as impacting negatively on the orientation of migration.
An additional possibility that may alter the ultimate position of migrating interneurons is the influence of migrating projection neurons. Hevner et al. (2004) suggest that radially migrating pyramidal neurons could provide cues to migrating interneurons that signal appropriate laminar position. In MAM-treated cortex, cues like these may be aberrant. Although we previously demonstrated the basic distribution of neurons outside of layer 4 in MAM-treated animals is relatively normal in terms of cell size and position (Noctor et al. 2001), as observed in this study and by others in our group, slight features of cortical architecture show alterations in the MAM model (Palmer et al. 2001; Jablonska et al. 2004). The current study demonstrates that disturbances in the migration and final disposition of interneurons are subtle (i.e., orientation) and occur at the later stages of their path to the CP. After entering the CP, they do not locate in the proper layers, as we observed that GAD+ or calcium-binding protein immunoreactive cells appear to congregate in more central cortical regions in juvenile ferrets. Because the main alteration in the distribution and orientation of the migrating neurons after E33 MAM treatment occurs during the later stage of migration, it is possible that the path of later born pyramidal neurons migrating into the CP might also be subtly disturbed after MAM administration and influence the ultimate position of cells emanating from the GE.
Involvement of GABA Receptors during Tangential Migration
Another possibility is that an alteration in the balance of excitation and inhibition, initiated by the lack of layer 4 cells, translates into changes in activity levels. Activity cues in turn may be important in signaling positional information to migrating cells (Ming et al. 2001). As we noted, changes in markers related to GABA or GABA receptors and changes in activity patterns are characteristic of MAM treatment on E33 (e.g., Jablonska et al. 2004; McLaughlin and Juliano 2005). GABA is an attractant to migrating neurons of the neocortex and may affect more subtle aspects of migration (Behar et al. 1996, 1998). GABA is also implicated in the mechanism of tangential migration. Studies in mouse show that blockade of GABA signaling through GABAB receptors, which are expressed by the migrating cells, alters the ultimate position of cells leaving the GE (Lopez-Bendito et al. 2003). In addition, migrating cells express functional GABAA receptors (Soria and Valdeomillos 2002). We show here that in ferrets, migrating cells originating from the GE also express GABAA receptors. Cuzon et al. (2006) demonstrated that blocking GABAA receptors with BIC in mice decreases the number of migrating cells entering the neocortex. With a similar concentration of BIC, we found that blocking GABAA receptors alters the laminar positioning of the migrating interneurons in normal ferrets but does not diminish the number entering the cortex, as measured by the crossing index. Both reports indicate, however, that migration of interneurons is impeded with blockade of GABAA receptors. We also found that abnormal activation of GABAA receptors using muscimol disrupts the orientation of neurons arising from the GE in normal ferret cortex, which mimics the disorientation observed in MAM-treated animals. Neither blocking overall activity with TTX nor blocking GABAA receptors with BIC altered orientation, suggesting that this feature of migrating cells is intimately linked with GABAA signaling.
Overall, our results in ferrets suggest that activation of GABAA receptors is necessary for cells leaving the GE to navigate appropriately toward and within the superficial cortical layers. Abnormal activation or blockade of GABAA receptors does not affect the ability of interneurons to invade the neocortex but will alter the ultimate laminar distribution of the interneurons (Fig. 10). We suggest that the relative absence of layer 4 in the somatosensory cortex after MAM treatment leads to a cascade of effects including improper termination of thalamocortical afferents, abnormal distribution of GABAA receptors, disorientation of neurons migrating tangentially, abnormal distribution of the interneurons, and an inability to transfer information through the cortical layers (Noctor et al. 2001; Jablonska et al. 2004; McLaughlin and Juliano 2005).
Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.
This work was funded by PHS RO1 NINDS NS 24014 (SLJ). We thank Stewart Anderson for helpful comments and critically reading the manuscript and also Sarah Dhandu for technical assistance and excellent care of the ferrets. Conflict of Interest: None declared.