Organelle and Cellular Abnormalities Associated with Hippocampal Heterotopia in Neonatal Doublecortin Knockout Mice

Heterotopic or aberrantly positioned cortical neurons are associated with epilepsy and intellectual disability. Various mouse models exist with forms of heterotopia, but the composition and state of cells developing in heterotopic bands has been little studied. Dcx knockout (KO) mice show hippocampal CA3 pyramidal cell lamination abnormalities, appearing from the age of E17.5, and mice suffer from spontaneous epilepsy. The Dcx KO CA3 region is organized in two distinct pyramidal cell layers, resembling a heterotopic situation, and exhibits hyperexcitability. Here, we characterized the abnormally organized cells in postnatal mouse brains. Electron microscopy confirmed that the Dcx KO CA3 layers at postnatal day (P) 0 are distinct and separated by an intermediate layer devoid of neuronal somata. We found that organization and cytoplasm content of pyramidal neurons in each layer were altered compared to wild type (WT) cells. Less regular nuclei and differences in mitochondria and Golgi apparatuses were identified. Each Dcx KO CA3 layer at P0 contained pyramidal neurons but also other closely apposed cells, displaying different morphologies. Quantitative PCR and immunodetections revealed increased numbers of oligodendrocyte precursor cells (OPCs) and interneurons in close proximity to Dcx KO pyramidal cells. Immunohistochemistry experiments also showed that caspase-3 dependent cell death was increased in the CA1 and CA3 regions of Dcx KO hippocampi at P2. Thus, unsuspected ultrastructural abnormalities and cellular heterogeneity may lead to abnormal neuronal function and survival in this model, which together may contribute to the development of hyperexcitability.

Heterotopic neurons arise during development by a variety of mechanisms [1]. Neurons born close to the ventricles must migrate long distances to reach their final position in the cortical plate [14]. Slowed or arrested migration can therefore lead to abnormal final positioning of neurons in the migratory path [15]. The physiopathological consequences of heterotopia and especially their link with the emergence of epileptiform activities are not well understood. Rare histological and immunohistochemical studies of human heterotopia have shown that they contain both pyramidal cells and interneurons, and DiI tracing studies have revealed connections between heterotopic regions and subcortical/cortical regions [16]. More recent data in rodent models of SBH suggest that not only the heterotopia, but also the overlying cortex function abnormally [17]. However, few studies have been devoted to characterizing the morphological and ultrastructural features of neurons developing in the heterotopic and overlying cortex. This could provide clues to their later abnormal function in the adult.
Mutant mouse lines generated for genes involved in SBH and type 1 lissencephaly in human are consistently associated with heterotopic pyramidal cells in the hippocampus. Reeler mice are the most severely affected, showing a grossly disorganized hippocampus and isocortex [15,18]. Lis1 mutant mice have more subtle defects in the isocortex but also exhibit severe hippocampal lamination defects, consisting of fragmented CA1 and CA3 pyramidal cell layers, and this is associated with epilepsy [19][20][21]. Tuba1a mutant mice show a similar hippocampal phenotype [11], whilst Dcx KO mice present a pyramidal cell disorganization largely restricted to the CA3 region [6,22]. Interneuron migration abnormalities have been shown to accompany the hippocampal lamination defects in Lis1 and Dcx mutants [23,24].
During embryonic development of the WT hippocampus, neurons migrate from the ventricular zone (VZ) of the medial wall across an intermediate zone (IZ, the future stratum oriens) to reach the hippocampal plate [25]. In the Dcx KO, as well as a correctly forming pyramidal cell layer, an abnormal high density of cells is observed in the IZ during this developmental period [6]. In the adult, Dcx KO CA3 pyramidal cells are arranged in two distinct layers, compared to a single layer in WT. Furthermore, mice suffer from spontaneous epilepsy and the CA3 region shows enhanced excitability in vitro [26,27]. Since hippocampal organization is well characterized in WT, Dcx KO mice provide an excellent model to further study specific features of developing heterotopic cells, and the generation of hyperexcitability.
[TUGHTER]In WT, interneurons and oligodendrocyte precursor cells (OPCs) originate in the ventral telencephalon during embryogenesis, and migrate long distances to reach medial parts of the cortex, with interneurons reaching the CA3 region by E16 [28][29][30][31]. In late embryonic stages and postnatally, interneurons and OPCs move within the hippocampus to their final positions [28,32]. Dentate gyrus granule cell production within the hippocampus temporally matches the other cell types [33], with numerous cells produced from E16 onwards [34], migrating in a tangential subpial stream, to reach the dentate gyrus region [35], where production continues postnatally [28]. During development, cell death is also a physiological phenomenon with peaks of apoptosis observed in the rodent hippocampus in early postnatal stages [36][37][38].
In this study, we set out to characterize the Dcx KO CA3 region postnatally, performing an ultrastructural and cellular study to query the nature of the lamination defect. Few ultrastructural analyses examining the content and aspect of heterotopic cells have been reported previously [39,40]. Organelle modifications were identified in Dcx KO neuronal cells including evidence of damaged mitochondria, and more dilated, circular and vesicular Golgi apparatus forms compared to WT. In addition, the postnatal Dcx KO pyramidal layers showed a heterogeneous cell composition, exhibiting some interspersed OPCs and other cell types. Caspase-3 dependent cell death was also increased approximately two-fold. These combined data reveal abnormalities in the postnatal Dcx KO hippocampus which have not previously been described and which may provide clues to the abnormal functioning of Dcx KO pyramidal neurons in the adult.

Materials and Methods
Ethics statement, mouse breeding, genotyping and tissue preparation For each experiment, Dcx KO and WT mice maintained on the Sv129Pas background were analyzed from several litters. Mice were maintained after more than ten generations of backcrosses and genotyped to verify that the Dcx gene was inactivated [24]. Mice were generated by crossing heterozygote females with pure Sv129Pas males (Charles River, France). For all analyses male hemizygote KO mice were compared with littermate male WT mice. Newborn P0-P4 animals were anesthetized by hypothermia for 3-4 min. Prior to sacrifice, adult mice were deeply anesthetized by intraperitoneal injections of pentobarbital. For in situ hybridization and caspase-3 immunostaining, mice (n = 3 WT and KO) were perfused with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB), and postfixed in the same solution overnight. Following postfixation, brains were cryoprotected in 30% sucrose, frozen in embedding medium and cut using a cryostat at 16-20 mm. To characterize cell heterogeneity, different neural precursor marker antibodies were used. For these experiments, neonatal P0 mice were anesthetized and perfused with 2% PFA in PB, and postfixed in the same solution for 2-4 hours. Following postfixation, brains were cryoprotected in 30% sucrose, frozen in embedding medium and cut using a cryostat at 12 mm. For electron microscopy (EM) experiments, P0 mice (n = 4 WT and KO) were anesthetized and perfused with 4% PFA and 2.5% glutaraldehyde in PB. The brains were removed and placed in fresh fixative overnight at 4uC, rinsed in PB, postfixed in 2% OsO 4 (in PB), dehydrated in an ascending series of ethanol, and embedded in epoxy resin. For laser capture microdissection, 4-5 KO and WT P0 mice were anesthetized by hypothermia, and brains flash frozen by dropping in isopentane maintained at 235uC for 2 minutes.
Caspase-3 positive cells were counted in CA1 and CA3 regions of the dorsal hippocampus across a surface of approximately 5610 6 (CA1) and 2610 6 (CA3) mm 2 . A 406objective mounted on a LEICA DM6000 microscope was used for manual counting. Surface area was calculated on corresponding Hoechst images using ImageJ software. 7 sections (each 192 mm apart) per animal were analyzed in a blind manner. The density of caspase-3 positive cells was calculated by dividing their number by the corresponding area. Average densities per animal were estimated and compared between groups with a Student t-test using the GraphPrism software.

Laser capture microdissection (LCM) and quantitative PCR
To prepare RNA from the CA3 pyramidal cell region, laser microdissection was performed. Coronal brain sections (12 mm) containing the rostro-caudal hippocampus from Dcx KO and WT mice were prepared using a cryostat at 220uC and mounted on PENmembrane slides (1440-1000, PALM, Bernried, Germany). Sections were stained with cresyl violet 1% (Sigma Aldrich, France). In WT animals, the single CA3 pyramidal cell layer, incorporating CA3a (closest to CA1) and b (central CA3) regions, was microdissected (Zeiss LCM system), excising at the border with the strata oriens and radiatum. For Dcx KO animals the same region was dissected, however the internal pyramidal cell layer (SPI, closer to the stratum radiatum) was excised separately from the external layer (SPE, closer to the stratum oriens). For each animal, the entire dorsal hippocampal CA3 structure (bilateral) was excised from 60 sections and pooled.
Total RNA was isolated using a picopure RNA isolation kit (Arcturus) according to manufacturer's instructions. RNA quantity and quality were assessed using a Nanodrop ND-1000 spectrophotometer (ThermoScientific) and Agilent 2100 Bioanalyzer (Agilent Technologies, USA) respectively. To generate cDNA for qPCR experiments, RNA reverse transcription and amplification was performed using a Nugen Ovation Pico WTA system following manufacturer's instructions. Real time qPCR assays using the SYBRgreen method followed MIQE guidelines [44]. Gene-specific primers for Olig1 (forward CGTCTTGGCTTGTGACTAGCG and reverse GCCAGT-TAAATTCGGCTACTGTC), for Npy (forward GCTCTGCGA-CACTACATCAATCTC, reverse AGGGCTG-GATCTCTTGCCATA) and for Sst (forward GGAAGACATTCACATCCTGTTAGCTT reverse AGAGGTCTGGCTAGGACAACAATATTA) were designed using Primer Express Software (PE Applied Biosystems). Values were normalized to the geometric mean of 3 Normalization Factors found to be stable in all samples using the geNorm approach (http://medgen.ugent.be/jvdesomp/genorm/). These were ATP synthase, H+ transporting mitochondrial F1 complex, beta subunit (Atp5b), eukaryotic translation initiation factor 4A2 (Eif4a2) and prosaposin (Psap). For test genes, average values 6 standard deviations were calculated for at least 4 animals of each genotype. Ratios of Olig1 and Npy gene expression fold changes for SPI versus WT, and SPE versus WT were calculated, and P values obtained using the Student t-test.

Electron microscopy (EM) and morphometry
Semi-thin sections (0.5 mm) were stained with toluidine blue and images were acquired with a Coolsnap CCD camera mounted on a Provis Olympus Microscope. Ultra-thin sections (40 nm) were mounted on either 200 mesh or one slot grids, cut and double stained with uranyl acetate and lead citrate prior to observation with a Philips (CM-100) electron microscope. Digital images from the CA3 region were obtained with a CCD camera (Gatan Orius). One exhibits a clear cytoplasm and large nuclei (black asterisks) with round, elongated or lobulated shapes, enclosing light chromatin with peripherally located nucleoli, similar in aspect to the neuronal chromatin observed in WT. The second type of cell (white asterisks) has a darker cytoplasm, occasional tapered processes and encloses nuclei with dense chromatin. They are intermingled and often closely juxtaposed with neuronal-like cells. (D) In the SPI, two types of cells are shown, pyramidal neuronal-like cells with a clear cytoplasm and peripheral nucleoli (black asterisk) and a second cell type with a darker cytoplasm whose nucleoli are rarely peripheral (white asterisk). (E) In the SPE similar cell types with light (black asterisk) and dark cytoplasm (white asterisks) are also identified. Arrows in D and E indicate the irregular shape of the nuclear membrane. Scale bars: A, C: 5 mm; B, D, E: 2.5 mm. doi:10.1371/journal.pone.0072622.g002 Morphometric analysis was performed with the software Digital Micrograph. For each nucleus, its widest diameter was measured, and the mean calculated from 50 neuronal nuclei from two mice per genotype. Only neurons displaying a clearly visible nucleolus were taken into account. Golgi apparatuses (GA) were analyzed in 50 cells from two WT animals and two KO animals. Altered versus normal mitochondria were counted in a minimum of 20 WT, 20 SPI and 20 SPE cells from 4 KO mice.

EM characterization of the two Dcx KO layers in the P0 CA3 region
Histological studies to characterize the Dcx KO hippocampus were previously performed between E17 and P6 [6]. From postnatal stages two distinct layers of pyramidal cells were clearly visible in the CA3 region, continuing in the adult ( Fig. S1 and  [6,27]). Despite their abnormal positioning along the radial axis, pyramidal cells appeared correctly specified, expressing appropriate field markers (Fig. S1). We defined the two KO CA3 layers (Fig. 1) as stratum pyramidale internal (SPI, inner layer, closest to the stratum radiatum) and stratum pyramidale external (SPE, outermost layer, closest to the stratum oriens).
Well-organized and aligned CA3 pyramidal cell nuclei were identified in coronal semi-thin WT P0 sections (Fig. 1A, C). In the Dcx KO sections, the two cell layers (marked by black dotted lines in Fig 1D) were present over a larger, less cell dense region (Fig. 1B,  D), separated by a plexiform-like intermediary layer (IN). Cells in the Dcx KO SPE layer were distinguished from other cells in the stratum oriens by their neuronal nuclear aspect [45] and their contiguous arrangement, without large free spaces that characterized the neuroepithelium (Fig. 1D). Within the WT layer, cells were closely packed and their nuclei displayed round or oval shapes enclosing one or two nucleoli (Fig. 1C). SPE and SPI KO nuclei were in general more widely spaced than those in the WT layer. Nuclei in SPE possessed less regular shapes and were less well aligned than SPI nuclei. The average nuclear diameter was significantly smaller in SPE than in SPI and WT layers (expressed as mean 6 standard error WT, 8 Fig. 1E). Thus the SPE can be distinguished from the SPI at this stage.
At higher magnification ( Fig. 2A), the broad single WT pyramidal cell layer showed superposed layers of neuronal-like cells (Fig. 2B, black circles), characterized by a clear abundant cytoplasm (Fig. 2B). Dcx KO CA3 layers contained some morphologically different cell types from those observed in WT. A first cell type showed clear cytoplasm and round or elongated nuclei (Fig. 2C, black asterisks) enclosing light chromatin and peripherally located nucleoli, similar in aspect to neuronal-like cells in WT. The Dcx KO neuronal-like nuclei sometimes showed an irregular or lobulated form compared to WT (Fig. 2 compare D, E to B), and clumped chromatin (data not shown). A second cell type with darker cytoplasm (white asterisks in Fig. 2C) was observed interspersed between neuronal-like cells in both SPE and SPI. Such cells in some cases displayed processes, occasionally surrounding neuronal-like cells (Fig. S2). Their nuclei were often smaller than those of neurons, and showed irregular contours and dark chromatin, with one or several nucleoli. Thus, more heterogeneous cellular forms were observed in Dcx KO CA3 layers compared to WT.

Organelle abnormalities identified in Dcx KO neuronal cells
In WT cytoplasm, mitochondria were found to have a normal aspect, with clearly defined cristae (arrow in Fig. 3A), whereas Dcx KO neuronal-like mitochondria were often altered and swollen (arrow in Fig. 3B). Both normal and altered mitochondria were found in the same cell, including in neuritic processes. On the other hand, in darker cells closely juxtaposed to Dcx KO neuronallike cells, mitochondria possessed well defined cristae (arrowheads, Fig. 3B, C). Numbers of damaged mitochondria were further assessed in a number of randomly selected neuronal somata. No abnormal mitochondria were observed in WT cells (n = 20 cells). However, neuronal-like KO cells (n = 21 SPE, n = 28 SPI) showed only 45% normal mitochondria, similar in SPI and SPE (Fig. 3D). Thus, mitochondrial abnormalities are present in the Dcx KO CA3 region and they appear to be cell-type specific, restricted to neuronal-like cells.

Characterization of the intermediary (IN) layer separating the SPI from the SPE
The Dcx KO IN layer was found to contain numerous heterogeneous, neuritic profiles ( Fig. 4A-E). Nuclei were largely absent from this region. Neuritic profiles were not oriented uniformly and displayed heterogeneous forms, ranging from round to elongated (Fig. 4B). Some displayed a light cytoplasm with damaged mitochondria (Fig. 4C), and contained tubulo-vesicular structures (Fig. 4D) and regions devoid of cytoskeletal elements (arrowhead, Fig. 4D). Microtubules in longitudinally sectioned profiles, when obvious in Dcx KO neurites, appeared less numerous than those in WT cells (compare Fig. 4F, G).
Some neuritic contacts were observed, for example, between Dcx KO apical neurites (black asterisk, Fig. 4C) emerging from a neuronal cell located in the SPE and extending to a neuronal soma in SPI (as shown in Fig. 4A, B). This profile could represent a growing apical dendrite. Synaptic contacts were also occasionally observed at the border of the IN, in some cases likely to be immature symmetric synapses of GABAergic cells, contacting perisomatic regions of pyramidal cells (contact with an SPI soma shown in Fig. 4B, E). They contained a few synaptic vesicles (arrow) opposite the postsynaptic zone. These data indicate that, despite abnormal cellular profiles, synaptogenesis can take place at P0 in Dcx KO neuronal-like cells.

Heterogeneous cell types in close proximity to Dcx KO pyramidal cells
We set out to identify the heterogeneous cell types revealed by EM, interspersed within, or in close proximity to, the Dcx KO pyramidal cell layers. Anti-Pax6 and anti-Sox2, two markers of apical radial glial cell progenitors [46], normally present in the VZ during hippocampal development, were first used. Pax6-and Sox2-positive nuclei were only seen in their correct position in the VZ of both WT and KO brains and were not present in the pyramidal layers (Fig. S3A-F). Similarly, we used anti-nestin, which labels intermediate filaments in radial glial cell progenitors, including their processes. There was no evidence of nestin-positive somata within the pyramidal cell layers in either WT or KO (Fig. S3C-F). GFAP, a marker of radial glial cells and mature astrocytes, also showed no labeling at this stage in the pyramidal cell layer in either WT or KO hippocampi (Fig. S3G, H). Therefore the cellular heterogeneity revealed by EM is not due to abnormal positioning of radial glial progenitors or astrocytes.
Platelet-derived growth factor receptor a (Pdgfra), a marker of OPCs, was next tested. In the WT CA3 region, cell somata were present in the strata oriens and radiatum but rarely in the pyramidal cell layer (Fig. 5A). However, a number of OPCs were identified within and adjacent to the Dcx KO pyramidal cell layers (Fig. 5B-E and Fig. S3). OPC nuclei were either horizontally, or vertically oriented (Fig. 5C1-3, E1-2 arrows) and were well arborized. In order to confirm the higher abundance of OPCs in the Dcx KO stratum pyramidale compared to WT, quantitative PCR (qPCR) was performed for the OPC marker, Olig1 (Table 1 and Table S1). We found a two-fold up-regulation of this transcript in Dcx KO SPE compared to WT (Student t-test, p #0.001) and a five-fold upregulation in Dcx KO SPI compared to WT (Student t-test, p,0.01). These combined results suggest a significant excess of OPCs interspersed within the P0 Dcx KO pyramidal cell layers.
Further unlabeled nuclei were also observed in immunohistochemistry experiments, some potentially with an aspect of migrating cells based on their vertically oriented, oval shaped nuclei (Fig. 5D, E1, red arrowheads). These cells could correspond to migrating pyramidal cells or interneurons, or other cell types. Concerning interneurons, we have previously assessed the number of parvalbumin positive cells in the adult Dcx KO hippocampus, showing no overall differences in number compared to WT [27]. Somatostatin (Sst)-positive cells represent a second major subpopulation [47], which can co-express either reelin, neuropeptide Y (Npy) or calretinin markers [48]. This subpopulation is present in the CA3 region perinatally, but not normally associated with the stratum pyramidale and we hence decided to test for these cells in Dcx KO pyramidal layers. At early postnatal stages in WT, Sstinterneurons were found mainly in the stratum oriens, with some somata also found in the stratum radiatum (Fig. 6A, C, E, G) and in the hilus (data not shown). In the Dcx KO, as well as being observed in these regions, some Sst-positive cells were also found associated with the SPI, SPE or the IN region (Fig. 6B, D, F, H). Such cells were vertically, obliquely or horizontally oriented and some showed a potentially migratory morphology (Fig. 6H,  arrows). Similarly, qPCRs for Npy showed an upregulation especially in the SPI layer compared to WT (3.6 fold upregulation, Student t-test, p#0.01, Table 1 and Table S1), although the Sst transcript tested was not found significantly different in this assay. These data together suggest that, as well as OPCs, certain interneuron populations or their migrating precursors may also contribute to cell heterogeneity observed at P0 in the Dcx KO pyramidal layers.

Cell death is increased in the Dcx KO hippocampus
The abnormalities identified in the Dcx KO hippocampus also prompted us to assess cell death. We examined apoptosis at P2, a peak stage of cell death during rodent postnatal hippocampal development [37,38]. Cells expressing activated caspase-3 were identified across the rostro-caudal extent of the Dcx KO dorsal hippocampus compared to WT littermate hippocampi. At this postnatal stage in both WT and KO, activated caspase-3 labeled cells were present at highest densities in the CA3 region as well as to a lesser extent, in the CA1 region ( Fig. 7A-D, white  arrowheads). In both CA3 and CA1 regions, caspase-3 positive cells were predominantly present in the stratum oriens, although some cells could also be seen in the stratum pyramidale (Fig. 7E-F). Analyzing CA1 and CA3 regions separately, an approximately two-fold increase of caspase 3-positive cells was present in each hippocampal region in the KO compared to WT (Fig. 7G-H

Discussion
We characterized here the state of the Dcx KO hippocampus using EM, immunohistochemistry and qPCRs. We found the developing Dcx KO pyramidal layers to be quite different from WT, being less organized and more heterogeneous. Organelle abnormalities were identified in Dcx KO neuronal-like cells, which has not previously been shown to our knowledge in other heterotopia mouse models. Indeed, few previous studies have addressed ultrastructural aspects of abnormally positioned neurons by EM, although analyzes of the highly disorganized embryonic reeler cortex revealed cell mis-orientation with grossly normal neuronal differentiation [49]. Our postnatal data, potentially illustrating early consequences of abnormal neuronal position, provide a framework on which to study how functional abnormalities arise in the Dcx KO hippocampus, especially the development of cell hyperexcitability [26,27].
In WT embryos, the peak of neurogenesis for the CA3 region occurs at E14 [50], although some CA3 pyramidal cells are still produced at E16 [35]. Migration occurs across a 4 to 5 day period before cells reach the hippocampal plate, after pausing extensively in the IZ [25], perhaps waiting for coordinating signals from interneurons and dentate gyrus granule cells [31,33]. We looked at the state of cells in early postnatal development, thus at a time when migration is expected to be largely completed [25,28]. The presence and position of two Dcx KO layers, even from E17.5 [6], suggests slowed or arrested migration, with SPE cells potentially remaining displaced even in the adult [6,22]. We hence searched for differences between the SPE and SPI layers. Indeed, we found that SPE neurons showed a relatively smaller nuclear diameter than SPI neurons at P0. This may be related to the fact that they are less mature [51,52], since a previous study of immature hippocampal neurons in foetal monkey brain showed that as migrating pyramidal cells reach their final position they become progressively more complex, with larger somata and less electrondense nuclei [51]. SPE cells at P0 may hence be less mature than SPI and WT neurons, and perhaps retarded or arrested in their migration. The additional observation of increased cell death at P2 in the Dcx KO CA3 region, in particular in the stratum oriens, may also be linked to the elimination of some aberrant cells arrested in their migration.
We identified some oval-shaped nuclei within both Dcx KO layers, which may correspond to some still-migrating pyramidal cells, interneurons, OPCs, or other cell types. Indeed, probable migratory elongated somata, associated with lobulated nuclei  containing relatively dense chromatin, and tapered leading processes, were previously characterized by EM in the developing mouse brain by Pinto-Lord and colleagues [53]. Based on the identification of similar cells in our EM study, this reinforces the idea that a certain number of cells continue to migrate in the postnatal Dcx KO hippocampus.
Despite the possible immaturity of the SPE layer, some synaptic contacts were clearly being formed. Indeed, the IN layer enclosing different types of cellular profiles resembles a plexiform layer, likely to contain growing axons, and apical and basal dendrites of SPE and SPI pyramidal cell neurons respectively, as well as incoming mossy fibers. We previously showed by biocytin labeling that adult SPE apical dendrites can cross the SPI [27]. Neurites from the SPE, observed by EM in this study to extend up to and to contact SPI cells, may be, at least in part, such dendrites. Interestingly, mature apical dendrites of SPE cells and basal dendrites of SPI cells in the adult KO hippocampus showed reduced length compared to WT [27]. Stunted dendrite growth in the IN might hence be a consequence of the bi-layered organization. On the other hand, a bi-laminar organization of CA1 apparently naturally exists in the wallaby, possum and mole rat, although in this case, only the inner, superficial layer is compact, whereas the 'outer layer' represents a continuum of sparser cells [54]. Differing from this situation, already at P0, Dcx KO SPI and SPE cell layers seem relatively compact and can be distinguished from other cells in the stratum oriens, likely to be migrating granule cell precursors [28]. Thus the divided pyramidal cell layer in the Dcx KO does not appear to resemble other naturally existing and well-functioning hippocampal 'bi-layers'.
Although Dcx KO cells are relatively condensed in layers, their extracellular environment may still be permissive or attractive for colonization by OPCs and other cell types, which insert themselves into the intercellular spaces. Darker cells identified by EM may thus in some cases represent OPCs. These cells, which displayed normal organelles, may not normally express Dcx, which is largely restricted to the neuronal lineage [55,56]. Developing OPCs are proliferative cells that originate in the ventral telencephalon and then migrate throughout the CNS [57][58][59]. They rapidly increase in number in the developing central nervous system between P0 and P7 and migrate extensively to populate the developing hippocampus before converting into differentiated oligodendrocytes [32]. The final fate of the Dcx KO OPCs associated with the pyramidal cell layers at P0, is not yet known. Signaling between Dcx KO CA3 pyramidal neurons and OPCs [60], or aberrantly positioned interneurons, could perturb neurotransmission and the development of hippocampal circuits.
Organelle abnormalities were identified in immature Dcx KO neuronal-like cells, and to our knowledge, abnormal mitochondria have not previously been reported in classical cortical malformation models, although they have been associated with Zellweger syndrome [61], and more commonly in models of neurodegenerative disorders [62]. Mitochondrial and Golgi abnormalities may stem from aberrant functioning of the microtubule network leading to organelle transport problems [63]. Dcx is a microtubule-associated protein [55,64], and microtubule abnormalities were previously characterized in Dcx KO neurons [65]. As well as direct microtubule functions [66,67], Dcx may also link organelles to microtubules, indeed it has been previously shown to interact with adaptor complexes, involved in vesicle and cargo trafficking [68], to affect the parameters of molecular motor (kinesin)-directed movement [69,70], and neurons mutant for two Dcx family members also showed abnormalities in axonal transport and synaptic vesicle deficits [71]. Thus, mitochondrial and Golgi apparatus abnormalities identified here could be a reflection of defects in various microtubule-related functions of Dcx.
Alternatively, organelle differences may arise for other reasons, perhaps related to more generalized cell stress, calcium signaling abnormalities [72], perturbed neurotrophic factors [73], or due to increased oxidative stress [61]. It will be important in the future to identify the causes of these defects and to assess the situation in adult Dcx KO pyramidal cells. Apoptosis may eliminate severely deficient cells during development, and autophagy may also play a role in their survival [62], allowing them to function in the adult. Further interrogation of these defects may lead to the identification of novel neuroprotective treatment strategies for this type of lamination disorder. Figure S1 Correctly specified CA fields in the adult Dcx KO hippocampus compared to WT. In order to first verify that field identity of CA KO cells was appropriately specified, we used adult hippocampal field specific markers in in situ hybridization experiments. In situ hybridization results show Wfs1 (A, B), Necab2 (C, D) and KA1 (E, F) markers comparing adult WT (A, C, E) and Dcx KO sections (B, D, F). Wolfram syndrome gene 1 (Wfs1) is a CA1 field specific marker [74][75][76]. In the Dcx KO hippocampus, we identified mainly normally positioned CA1 pyramidal cells, although some heterotopic cells were identified close to the subiculum, expressing this marker (Fig. S1B, arrows). Disorganized cells in the KO CA3 region did not express Wfs1. N terminal EF calcium binding protein 2 (Necab2), labels the CA2/CA3a region (the part of CA3 closest to CA2) and cells in the outer border of CA3b (the middle portion of CA3, Fig. S1C). This marker also showed a grossly normal pattern in the Dcx KO and revealed the region of splitting of the pyramidal cell layer (Fig. S1D, large arrow). Some Necab2-labelled cells were also found in the Dcx KO SPI (Fig. S1D, short arrow), with fewer present in the outer border of SPE CA3b. KA1 (glutamate receptor kainate type 1) is a wellknown CA3 cell marker, whose expression is specified early on during hippocampal development [77]. The KA1 marker most intensely labels the CA3 region, and the Dcx KO double cell layer is labeled with this marker (Fig. S1E, F showing that only OPCs, but not astrocytes intercalate between the CA3 neurons in Dcx mutant hippocampus. Note astrocytes close to the ventricular surface (G, arrowheads). H is a higher magnification of the area in the inset of G. Scale bars: 50 mm. (TIF)