The brains of individuals with developmental dyslexia have neocortical neuronal migration abnormalities including molecular layer heterotopias, laminar dysplasias, and periventricular nodular heterotopias (PNH). RNA interference (RNAi) of Dyx1c1, a candidate dyslexia susceptibility gene, disrupts neuronal migration in developing embryonic neocortex. Using in utero electroporation, we cotransfected cells in the rat neocortical ventricular zone (VZ) at E14/15 with short hairpin RNA vectors targeting Dyx1c1 along with either plasmids encoding enhanced green fluorescent protein or plasmids encoding monomeric red fluorescent protein only. RNAi of Dyx1c1 resulted in pockets of unmigrated neurons resembling PNH. The pattern of migration of transfected neurons was bimodal, with approximately 20% of the neurons migrating a short distance from the VZ and another 40% that migrated past their expected lamina. Approximately 25% of the transfected brains had hippocampal pyramidal cell migration anomalies. Molecular layer ectopias, which were not related to injection site artifacts, were also seen in 25% of the animals. These results support the hypothesis that targeted disruption of the candidate dyslexia susceptibility gene, Dyx1c1, results in neuronal migration disorders similar to those seen in the brains of dyslexics.
Developmental dyslexia is a language-based learning disability affecting 4–10% of the population, which is characterized by a difficulty with learning to read despite adequate motivation, intelligence, and educational opportunity. The complex nature of this disorder is reflected in the wide-ranging differences reported between the brains of dyslexics and nondyslexics, including symmetry of language-related regions (Galaburda et al. 1985; Shapleske et al. 1999), reduced gray matter volume (Silani et al. 2005; Vinckenbosch et al. 2005), and altered patterns of white matter organization (Klingberg et al. 2000; Schwartzman et al. 2005). We have reported small neocortical malformations in the brains of dyslexics (Galaburda and Kemper 1979; Galaburda et al. 1985; Humphreys et al. 1990). These neuronal migration anomalies, consisting of nests of neurons and glia in the molecular layer (ectopias), intracortical laminar dysplasias, and occasional instances of focal microgyria, are located primarily in the lateral surface of the left hemisphere, including the perisylvian, temporo-occipital, temporoparietal, and frontal regions. Similarly, recent reports have demonstrated an increased incidence of developmental dyslexia in patients with periventricular nodular heterotopias (PNH) (Chang et al. 2005; Sokol et al. 2006).
The known genetic etiology of dyslexia similarly reflects the complexity of this disorder. Thus, linkage analysis points to dyslexic susceptibility loci on chromosomes 1, 2, 3, 6, 7, 11, 15, 18, and X (see Fisher and Francks 2006). Recently, candidate dyslexia susceptibility genes have been proposed at some of these intervals. ROBO1, an axon guidance and neuronal migration gene located on Chr 3, has been reported to be a candidate dyslexia susceptibility gene (Hannula-Jouppi et al. 2005), and 2 genes, DCDC2 and KIAA0319, have been identified on Chr 6 (Francks et al. 2004; Cope et al. 2005; Meng, Smith, et al. 2005; Paracchini et al. 2006; Schumacher et al. 2006). DYX1C1 (also known as EKN1), a candidate dyslexia susceptibility gene located on Chr 15, has been proposed, although there is controversy regarding its generalizability outside of Finnish populations (Taipale et al. 2003; Chapman et al. 2004; Scerri et al. 2004; Wigg et al. 2004; Bellini et al. 2005; Marino et al. 2005).
These genetic results have informed research on the anatomic substrates of developmental dyslexia. We have demonstrated that DCDC2, KIAA0319, and DYX1C1 are genes involved in neuronal migration (Meng, Smith, et al. 2005; Paracchini et al. 2006; Wang et al. 2006). Specifically, we used in utero electroporation to transfect premigratory neurons in the ventricular zone (VZ) with short hairpin RNA (shRNA) targeted against each of these candidate dyslexia susceptibility genes. Examination of the brains 4 days post transfection revealed in each case that transfected neurons were arrested in migration. What is not yet known is the adult neocortical phenotype of animals transfected with these shRNAs, which would be of interest for comparison with the adult dyslexic brain. Here we report the results of a study in which we examined the brains of adult rats that had in utero electroporation of shRNA targeted against Dyx1c1. We hypothesized that the neuronal migration abnormalities induced by this treatment would result in neocortical malformations similar to those seen in postmortem dyslexic brains.
Materials and Methods
In Utero Electroporation
In utero electroporation was performed at the University of Connecticut, and all procedures were approved by the Institutional Animal Care and Use Committee of that institution. In all Dyx1c1 shRNA treatments, plasmids encoding shRNA (pU6DyxHPB) and plasmids encoding enhanced green fluorescent protein (eGFP) (pCAGGS-eGFP) were cotransfected into the VZ. The remaining animals received transfection only with plasmids encoding monomeric red fluorescent protein (mRFP) (pCAGGS-RFP). Ten Wistar dams (Charles River Laboratory, Wilmington, MA) were anesthetized at E14/15, the uterine horns were exposed, and approximately half of the pups were randomly assigned to receive Dyx1c1 shRNA plasmids (1.5 μg/μL) plus eGFP (0.5 μg/μL), whereas the remaining pups were injected with mRFP (0.4 μg/μL).
The plasmids were microinjected by pressure (General Valve picospritzer) through the uterine wall into one randomly chosen lateral ventricle of each embryo, using a pulled glass capillary (Drummond Scientific, Broomall, PA). Equal numbers of mRFP and Dyx1c1 shRNA + eGFP injections were made. Electroporation was achieved by discharge of a 500 μF 250 V capacitor charged to 50–100 V (Bai et al. 2003). A pair of copper alloy plates (1 × 0.5 cm) pinching the head of each embryo through the uterus were the conduit for the voltage pulse. For each embryo, plasmids were injected in one hemisphere, and diffusion of the plasmids into the opposite hemisphere was encouraged by gently tapping the embryo skull until dye was seen to migrate into the opposite hemisphere. A voltage pulse was discharged across the embryo skull with the positive electrode adjacent to the transfected hemisphere, the electrodes were repositioned to the opposite hemisphere, and a second pulse was delivered. This resulted in the transfection of cortex in both hemispheres. Five embryos were harvested 2 days post transfection for comparison of the size of the initial transfection in both the mRFP and Dyx1c1 shRNA + eGFP conditions. Thirty-six males were weaned on P21 and were housed into pairs. Animals were reared under a 12:12 light/dark cycle with food and water available ad lib and were behaviorally tested (Threlkeld et al. forthcoming). In order to assess the efficiency of cotransfection (i.e., what proportion of cells are cotransfected when unequal molar ratios similar to those used in the Dyx1c1 shRNA + eGFP condition), 4 animals were cotransfected with eGFP (3.2 μg/μL) and mRFP (0.5 μg/μL). As shown in the results (and similar to previous reports, e.g., Bai et al. 2003), essentially all cells are cotransfected with both plasmids even when one plasmid is used at a much lower concentration.
At P90–100, animals were deeply anesthetized (Xylanzine/Ketamine 100 mg/mL) and sacrificed by transcardial perfusion with 0.9% saline followed by 4% paraformaldehyde. The brains were removed from the skull and visualized whole under fluorescence for the presence of eGFP (N = 21) or mRFP (N = 15). Twelve brains (6 Dyx1c1 shRNA + eGFP and 6 mRFP) were coronally sectioned at 40 μm on a freezing microtome after cryoprotection in 30% sucrose buffer. The remaining 24 brains (15 Dyx1c1 shRNA + eGFP and 9 mRFP) were dehydrated in graded ethanol and embedded in 12% celloidin before being sectioned in the coronal plane at 30 μm. One series of every 10th section was stained for Nissl substance (Thionin for the frozen sections, cresyl violet for the celloidin). An adjacent series of free-floating sections was processed for immunohistochemical detection of either eGFP (Chemicon, 1:200) or mRFP (Chemicon, Temecula, CA, 1:1000) using ABC protocols. The celloidin was first removed from the celloidin-embedded tissue by incubation in a 1:1 solution of ethyl ether and 100% ethanol for 30 min.
For assessment of the number of cells initially transfected in the Dyx1c1 shRNA + eGFP and mRFP conditions, embryos were harvested 2 days after transfection, a time when previous experiments have indicated that proliferation in the transfected population has ceased (Wang et al. 2006). For the investigation of cotransfection efficiency at molar ratios similar to those used in the current experiment, embryos were harvested 4 days post transfection. Pregnant dams were anesthetized as above and embryos harvested. Brains were quickly removed, fixed by 4% paraformaldyhyde, and stored in sodium phosphate buffer. They were sectioned on a vibratome (Leica VT1000S, Leica Microsystems Inc., Bannockburn, IL) at 80 μm, and the sections mounted on glass slides and coverslipped.
In Situ Hybridization
In order to better interpret the knockdown findings, we determined the expression of Dyx1c1 in the prenatal brain by in situ hybridization during various stages of embryonic development. We obtained time-mated pregnant females (Charles River) and sacrificed the litters on E13, E15, E17, or E19. Embryos from these litters were immediately frozen, their brains were cut in the sagittal or coronal plane on a cryostat at 18 μm thickness, and the slides were processed for in situ hybridization of Dyx1c1 as described below.
The cDNA prepared from frontal, parietal, and occipital lobes of human embryonic brain (20 weeks, Biochain Institute, Hayward, CA) was amplified with respective forward and reverse primers (GGG AGA AAT TCA GAA AAT ATA TTT AC and TTA AGA TTT TAG TTC TGT TCC TTG AATT) for 35 cycles. All fragments were then cloned into t vector (Invitrogen, Carlsbad, CA) and sequenced to verify Dyx1c1 amplification. Rat embryonic and postnatal brain cDNAs were synthesized from total RNA and amplified with the primers ATG CCG GTG CGA GTG AGC GAG and CAT CAT CTC GCC TAG GGC GTA TC to rat Dyx1c1. The amplified DNA was gel-purified, cloned, and sequence verified to be Dyx1c1. Nonradioactive in situ hybridizations were done by UB-In Situ (Natick, MA), as previously described (Berger and Hediger 2001), using a digoxigenin-labeled cRNA probe. The antisense and sense probes were obtained from the polymerase chain reaction (PCR) products, amplified from rat E14 brain cDNA, and cloned in pGEMT-Easy flanking T7 and SP6 promoters. Two probes, one from the first 400 bp generated from PCR primer pairs and the second full-length cDNA yielded similar results.
All analyses of adult brains were performed blind with respect to condition. Nissl-stained sections were surveyed for the presence of malformations and the location noted. The location of immunohistochemically labeled cells was charted in an adjacent series from 5 Dyx1c1 shRNA + eGFP and 4 mRFP brains using Neurolucida (MBF Biosciences, Williston, VT), and the average number of labeled cells per section was computed. The distance of labeled cell migration from the white matter to the pial surface was assessed on 4 Dyx1c1 + eGFP and 4 mRFP brains. Section drawings by Neurolucida were imported into Canvas X (ACD Systems, Miami, FL), and a counting rectangle subdivided into 10 equal-size bins was sized to extend from the white matter to the pial surface. The width of the counting box was held constant at approximately 500 μm. The number of labeled cells within each decile was manually counted and recorded. For each brain, we measured one rectangle in each hemisphere from 8 sections. A total of 2929 cells were counted in animals cotransfected with Dyx1c1 shRNA + eGFP and 5678 in those transfected with mRFP alone. The percentage of labeled cells in each decile was determined for each animal, and the mean value across all animals within each condition was determined. Differences in frequency distribution between the 2 conditions were assessed using chi-square analysis.
In order to determine the size of the initial transfection in transfected (mRFP) and cotransfected (Dyxt1c1 shRNA + eGFP) conditions, images of embryonic brains harvested 2 days post transfection were obtained by confocal microscopy (Leica TCS SP2, Leica Microsystems) with a 40× objective. Image J (http://rsb.info.nih.gov/ij/) was used to count the number of labeled neurons in each of 3 sections in each brain automatically. Three brains were assessed in the mRFP condition and 2 brains in the Dyx1c1 shRNA + eGFP condition.
To determine the efficiency of cotransfection at molar ratios similar to those used in shRNA transfections, we examined 4 brains that were harvested 4 days following transfection of eGFP and mRFP plasmids at approximately 6:1 molar ratio. Confocal microscopy and sequential scanning for the 2 flours was used to assess single labeling and double labeling (indicating cotransfection with both plasmids).
Photomicrographs were adjusted for exposure and sharpened (unsharp mask filter) using Adobe Photoshop (Adobe Inc., San Jose, CA). Some images were acquired using the Virtual Slice Module of Neurolucida. Image montages were created in Canvas X (ACD Systems).
Dyx1c1 Is Ubiquitously Expressed at a Modest Level in the Forebrain
In situ hybridization of Dyx1c1 in embryonic rat embryos revealed that this gene is expressed relatively ubiquitously in the forebrain during forebrain development (Fig. 1A). The level of expression is relatively modest, with higher concentrations in the neocortex, hippocampus, and choroid plexus. There is no literature regarding Dyx1c1 expression in rat brain, but examination of in situ hybridizations of adult mouse brains at the Allen Brain Atlas (http://www.brainatlas.org/aba/) also demonstrated a relatively modest level of ubiquitous expression of Dyx1c1 in the forebrain, with higher levels in the neocortex, hippocampus, and choroid plexus. We also queried The GeneNetwork (http://genenetwork.org/search.html), an online, publicly accessible database of transcript expression (from Affymetrix 430 v2 microarrays) from dissected hippocampus, cerebellum, eye, striatum, and forebrain for a variety of mouse strains, to assess Dyx1c1 transcript expression in adulthood. These data confirmed relatively modest expression of the transcript in all 5 regions assayed (Fig. 1B).
Cotransfection Is Highly Efficient
Examination of embryos cotransfected with eGFP and mRFP at an approximately 6:1 molar ratio 4 days post transfection revealed that nearly all transfected neurons were transfected with both plasmids (Fig. 2). Examination of thousands of neurons in all 4 animals revealed only an occasional neuron that was transfected with only one of the plasmids. This indicates a near perfect efficiency of cotransfection using molar ratios even greater than those employed in the Dyx1c1 shRNA + eGFP condition (approximately 5:1). This supports the contention that nearly every eGFP-labeled neuron in this condition was also transfected with Dyx1c1 shRNA.
There Are Fewer Dyx1c1 RNA interference–Transfected Neurons in the Neocortex
We counted the number of transfected neurons 2 days post transfection in 3 mRFP and 2 Dyx1c1 shRNA + eGFP animals and found similar numbers of labeled neurons in these conditions (±SEM=492±13.5 vs. 478±1.0, respectively; F1,4 < 1, not significant; Fig. 3A). These results support the contention that the initial number of transfected neurons did not differ between the 2 conditions. Examination of the adult brains, however, revealed fewer labeled cells in Dyx1c1 shRNA + eGFP–transfected animals as compared with those transfected with mRFP alone (see Fig. 3B,C,D). This was confirmed quantitatively: the average number of labeled cells in each section was significantly higher in the mRFP than in the Dyx1c1 shRNA + eGFP condition (±SEM=954.6±118.0 vs. 347.0±90.7, respectively; t = 4.2, degrees of freedom [df] = 7, P < 0.01). Taken together, these results suggest a loss of neurons in Dyx1c1 shRNA + eGFP–transfected animals in adulthood.
There Are Pockets of Unmigrated Cells in the White Matter
There were periventricular clusters of unmigrated cells in all Dyx1c1 shRNA + eGFP (see Fig. 4) but in none of the mRFP animals. These PNH were located directly either at the ependymal layer deep to the white matter, within the white matter itself, or at the cortical white matter border and were easily visible in Nissl-stained sections. Examination of these cells in immunohistochemically stained sections revealed that only a subset was immunopositive for eGFP. This suggested that these neurons did not migrate because of a secondary rather than a direct effect of Dyx1c1 RNA interference (RNAi) transfection. The morphology of the unmigrated immunopositive cells was clearly neuronal, although their normal radial orientation was disturbed (Fig. 4D,D′).
Laminar Displacement of shRNA-Transfected Neurons
In all the cases examined, mRFP+ neurons migrated to supragranular locations, predominantly to neocortical layer 3. Relatively few mRFP-labeled neurons were located in infragranular layers. In contrast, eGFP+ neurons in the Dyx1c1 shRNA condition were commonly found in infragranular layers, and those that migrated supragranularly were found superficial to the layer 3 location seen in mRFP-transfected animals (see Fig. 5A,B). This was confirmed quantitatively by migration distance analysis (Fig. 5C). The distribution of labeled cells through the thickness of the neocortex in Dyx1c1 shRNA + eGFP–transfected animals differed significantly from that of neurons transfected with mRFP (χ2 = 3920.0, df = 9, P < 0.0001). Thus, whereas the distribution in mRFP animals was Gaussian, with a peak at 70–80% of the distance to the pial surface, in the Dyx1c1 shRNA + eGFP cases, it was essentially bimodal, with peaks at the white matter border and at a location closer to the pial surface. A chi-square analysis of just the neurons in the upper 50% of the cortex was also significant (χ2 = 2554.9, df = 4, P < 0.0001), indicating that the neurons in the Dyx1c1 shRNA + eGFP condition migrated to more superficial locations than neurons transfected with mRFP only.
Malformations of the Hippocampus
Of the 21 animals cotransfected with eGFP and Dyx1c1 shRNA, 5 had malformations of the hippocampus, one of which was bilateral. There were no hippocampal malformations seen in any of the brains of animals transfected with mRFP only. The hippocampal malformations consisted predominantly of displaced cells from the pyramidal layer into the stratum radiatum and stratum oriens (see Fig. 6A,B,D,E). There were also ectopic collections of cells in the stratum radiatum that appeared to be pyramidal in morphology. Interestingly, the morphology of the eGFP+ cells was more typical of neocortical pyramidal cells than of hippocampal pyramidal cells (Fig. 6C,F,G). Specifically, these neurons had extended apical dendrites and less elaborate dendritic arborization than typically seen in hippocampal pyramidal cells. Examination of immunohistochemically stained sections revealed that only a small percentage of the displaced neurons were eGFP+. Thus, some of the displaced neurons were not transfected with shRNA targeted against Dyx1c1, thereby suggesting a secondary effect of the shRNA transfection in the hippocampus.
Molecular Layer Ectopias in the Cerebral Cortex
Collections of ectopic neurons in the molecular layer were seen in both mRFP (13/15) and Dyx1c1 shRNA + eGFP (13/21) animals. These ectopic neurons were often accompanied by laminar dysplasias in the subjacent cortex, which appear to be related to the disturbances associated with the injection of plasmids into the ventricle at E14 (Rosen et al. 1992, 1995). This is confirmed by examination of immunohistochemically stained sections where the physical displacement of labeled cells can be seen (see Fig. 7).
Five of the 21 brains cotransfected with eGFP and Dyx1c1 shRNA, and none of the mRFP-transfected cases, had 2 separate collections of ectopic neurons in layer I of the cortex: one collection resembled the disturbance at the injection site described above, whereas the other showed a different morphology. In each case where a second collection of ectopic neurons was found, it was in an area distant from the site of injection. In addition, these malformations were distinguishable from the injection-related malformations by the general lack in the latter of disturbed cortex in the subjacent layers (see Fig. 8A,B,C). The second type of molecular layer ectopias was remarkably similar in appearance to ectopias seen in human dyslexic brains (Fig. 8D). Examination of the immunohistochemically stained sections in this second set of ectopias revealed that, in contrast to injection site ectopias, there was no displacement of labeled neurons into more superficial lamina in the area of the malformation. Moreover, transfected neurons surrounded the ectopia, but there were few, if any, labeled neurons within the cluster of cells (see Fig. 8E). This suggested once again that at least some component of the malformations caused by Dyx1c1 shRNA might arise from changes in cell interactions rather than from direct RNAi effects on cells.
The results of the current experiment support previous work suggesting that RNAi of the rodent homolog of the candidate dyslexia susceptibility gene DYX1C1 disrupts neuronal migration in developing cerebral cortex (Wang et al. 2006). Four days following transfection, there was a nearly complete block of migration into the cortical plate. Moreover, this block was rescued by concurrent overexpression of Dyx1c1 indicating that the RNAi effect is specific to knockdown of Dyx1c1 expression. When examined in adulthood, the pattern of the disturbance in the cerebral cortex has evolved. There is an approximately 60% reduction of transfected cells in Dyx1c1 shRNA–transfected brains as compared with those transfected with mRFP alone. Neurons transfected with shRNA targeted against Dyx1c1 exhibit an essentially bimodal pattern of neuronal migration, with approximately 20% of the surviving neurons remaining in the white matter and in layer VI and 60% migrating to supragranular layers. Interestingly, two-thirds of these supragranular neurons appear to have migrated to laminae beyond those seen in animals transfected with fluorescent protein alone at the same age. In addition to these migrational disturbances, molecular layer ectopias not related to the injection site are seen in approximately 25% of the Dyx1c1 shRNA–transfected animals, which appear to be similar to those seen in the brains of postmortem dyslexics. We unexpectedly found that targeted knockdown of Dyx1c1 also disrupts the anatomic organization of the hippocampus.
Neuronal Migration and Developmental Dyslexia
Of the 4 candidate dyslexia susceptibility genes currently identified, all 4 have been shown to play a role in neuronal migration. Using in utero electroporation of shRNA targeted against the genes DCDC2, KIAA0319, and DYX1C1, we have demonstrated that each of these genes disrupts neuronal migration to the neocortex (Meng, Smith, et al. 2005; Paracchini et al. 2006; Wang et al. 2006). The fourth dyslexia candidate gene, ROBO1 (Hannula-Jouppi et al. 2005), has also been well characterized as being important for axon guidance and neuronal migration (Nguyen Ba-Charvet et al. 1999; Zhu et al. 1999; Hivert et al. 2002). Combined roles in neuronal migration and the establishment of neural circuits make these genes nicely placed to be behind the abnormally functioning cortex seen in developmental dyslexia.
In previous experiments involving in utero electroporation of shRNA targeted against dyslexia susceptibility genes, brains were examined within 4 days of transfection. As a result, the effect on the eventual organization of the forebrain was not assessed. In this experiment, we found forebrain malformations in the brains of animals sacrificed in adulthood. Intriguingly, we found molecular layer ectopias that are remarkably similar to those seen in postmortem dyslexic brains (Galaburda and Kemper 1979; Galaburda et al. 1985; Humphreys et al. 1990). In addition, the clusters of unmigrated neurons seen in the Dyx1c1 shRNA + eGFP condition closely resemble PNH, the presence of which has been associated with an increased incidence of developmental dyslexia (Chang et al. 2005; Sokol et al. 2006). This lends support the purported link between neuronal migration disorders and developmental dyslexia.
As a cautionary note, it must be mentioned that the Dyx1c1 locus described in the initial report in a Finnish population (Taipale et al. 2003) was also found in a Canadian sample, but with different alleles and haplotypes (Wigg et al. 2004). Also, Chapman et al. (2004) found linkage to the same region of Chr 15 in a US population, but again did not directly replicate the role of Dyx1c1. Other studies have failed to provide support for the Dyx1c1 locus in Italian (Bellini et al. 2005; Marino et al. 2005), English (Scerri et al. 2004), and US (Meng, Hager, et al. 2005) populations. Thus, although our data support a role of Dyx1c1 in neuronal migration, definitive support for its role as a candidate dyslexia susceptibility gene outside the Finnish population awaits further studies.
Dyx1c1 Disrupts Neuronal Migration
There was a significant diminution of labeled neurons in the cortex of the Dyx1c1 shRNA + eGFP condition when compared with the mRFP condition, with the latter having nearly 3-fold as many labeled neurons. This could be the result of changes in the proliferation of neurons or in subsequent cell death. Wang et al. (2006) demonstrated that there was no change in the proliferation of cells after treatment with RNAi against Dyx1c1, and our analysis did not demonstrate any difference in size of the initial transfection between animals cotransfected with shRNA and eGFP and those transfected with fluorescent protein alone (Fig. 2A,B). Because relatively equal numbers of neurons are transfected in both conditions, the diminution of labeled cells in the Dyx1c1 shRNA + eGFP condition in this study most likely represents an increase in cell death in this population of transfected cells.
The periventricular collections of unmigrated neurons were not surprising given the results from the previous experiments, where few neurons from Dyx1c1 shRNA + eGFP cases had migrated away from the VZ when examined 4 days after transfection. It was, however, unexpected that the unmigrated neurons in this study comprised only 20% of the labeled neurons, with the overwhelming majority of labeled neurons continuing on to migrate into the cortical plate. But it is also clear that the migration of the labeled cells in Dyx1c1 shRNA + eGFP animals differs from those transfected only with mRFP. In mRFP animals, the distribution of labeled neurons throughout the depth of the neocortex was Gaussian, with approximately 75% of the cells located in layer 3. In contrast, labeled cells in the Dyx1c1 shRNA + eGFP condition were distributed essentially in a bimodal manner, with one collection of cells remaining in the white matter and another group migrating up to neocortical layer 2.
The reason for the migration of labeled cells to layer 2 of the cerebral cortex in the Dyx1c1 shRNA + eGFP condition as compared with layer 3 in the mRFP condition is not known. It could be that knockdown of Dyx1c1 function serves mostly to delay neuronal migration. Thus, transfected neurons remain in the VZ in the immediate aftermath of the transfection. They then begin their migration at the same time that later generated neurons begin theirs. The position in the upper layers of the cerebral cortex would be consistent with the possibility that these cells have been respecified to upper layer fates similar to the results from transplantation experiments of early generated neurons into latter VZ environments (e.g., McConnell 1985, 1990). Alternatively, it could be that knockdown of Dyx1c1 function in these neurons fundamentally changes their ability to migrate. In this case, neurons migrate throughout the cortical layers and collect in layer 2 because they cannot migrate past the external glial limiting membrane. At this point, we cannot distinguish between these possibilities.
Malformations in the Forebrain
We have previously shown that puncture wounds to the cortical plate lead to ectopic collections of neurons in the molecular layer (Rosen et al. 1992, 1995). In the current experiment, ectopic collections of neurons were seen in the majority of cases and were related to the puncture of the cortical plate by the injection of plasmids. In addition to these obvious traumatic malformations, 25% of the Dyx1c1 shRNA + eGFP animals had other ectopic collections of neurons in the molecular layer, which are similar to those seen in humans with developmental dyslexia (see above).
The malformations of the hippocampus were an unexpected finding. These malformations are remarkably similar to those seen in the progeny of mothers injected with methylazoxymethanol (MAM) at E15 (Chevassus-Au-Louis et al. 1998). In the case of these induced malformations, the neurons in the hippocampus exhibit neocortical-like neuronal activity and are hypothesized to have migrated from the neocortex (Castro et al. 2002). In our case, it could be that the in utero electroporation-labeled neurons from the VZ destined to be hippocampal. The paucity of labeled neurons in the hippocampus of mRFP-transfected animals, however, argues against this interpretation. In addition, the long apical dendrites and modest dendritic arborization of these ectopic immunopositive neurons more closely resemble neocortical pyramidal cells. It is more likely, therefore, that the labeled neurons in the hippocampal malformations are neocortical neurons that have mismigrated from the VZ.
Secondary Effects of Dyx1c1 Knockdown
There are a number of indications that there were secondary effects of Dyx1c1 knockdown in the developing forebrain. Few of the neurons that comprise the nontraumatic molecular layer ectopias, the periventricular collections of unmigrated neurons, and the hippocampal malformations were eGFP+. This result could theoretically be due to the possibility that some eGFP-negative cells are transfected only with Dyx1c1 shRNA but not eGFP plasmids. This is unlikely given the high degree of cotransfection efficiency at the molar ratios used in this experiment. It is more likely that there are secondary effects of the knockdown of Dyx1c1 function that act to disrupt neuronal migration in the forebrain beyond the direct effects on transfected cells.
It is tempting to speculate that disturbances to radial glial cells may underlie these secondary effects. Radial glial cells function not only as the scaffold by which neocortical neurons migrate from the VZ to the cortical plate (Rakic 1988) but also serve as neuronal precursors in the neocortex (Noctor et al. 2002; Fishell and Kriegstein 2003). In addition, radial glial cells are involved in the maintenance of the early external glial limiting membrane (Marin-Padilla 1995). We have previously shown that disruptions of the external glial limiting membrane can cause molecular layer ectopias to occur (Rosen et al. 1992; Sherman et al. 1992). Prenatal exposure to MAM not only induces hippocampal malformations but also results in periventricular nodules and other neocortical heterotopias (Zhang et al. 1995). MAM exposure directly affects the radial glia by disrupting their attachment to the cortical surface (Zhang et al. 1995) and hastening their transition to astrocytes (Noctor et al. 1999; Gierdalski and Juliano 2003), and it has been hypothesized that this disruption of the radial glial cells is the main cause of the malformations in MAM-exposed animals (Paredes et al. 2006). Thus, disturbances of radial glial cells during the period of neuronal migration have been associated with all 3 types of malformations exhibited by our model. It is therefore tempting to hypothesize that Dyx1c1 acts not only to disrupt the machinery important for neuronal migration in neurons (Wang et al. 2006) but also may, perhaps indirectly through the injury to the migrating neurons, disrupt the radial glia. Future research will directly test this hypothesis.
This work was supported, in part, by National Institutes of Health grant HD20806. The authors wish to acknowledge the expert technical assistance of Shira Anconina, Cullen Owens, and Zachary Snow. Conflict of Interest: None declared.