The laminar and area patterning of the mammalian neocortex are two organizing principles that define its functional architecture. Members of the immunoglobulin (Ig) superfamily of cell adhesion molecules influence neural development by regulating cell adhesion, migration, and process growth. Here we describe the dynamic expression of the unique Ig-containing cell adhesion molecule, MAM domain–containing glycosylphosphatidylinositol anchor 1 (MDGA1), during forebrain development in mice and compare it with other markers. We show that MDGA1 is a layer-specific marker and an area-specific marker, being expressed in layers 2/3 throughout the neocortex, but within the primary somatosensory area (S1), MDGA1 is also uniquely expressed in layers 4 and 6a. Comparisons with other markers, including cadherins, serotonin, cytochrome oxidase, RORβ, and COUP-TF1, reveal unique features of patterned expression of MDGA1 within cortex and S1 barrels. Further, our findings indicate that at earlier stages of development, MDGA1 is expressed by Reelin- and Tbr1-positive Cajal–Retzius neurons that originate from multiple sources outside of neocortex and emigrate into it. At even earlier stages, MDGA1 is expressed by the earliest diencephalic and mesencephalic neurons, which appear to migrate from a MDGA1-positive domain of progenitors in the diencephalon and form a “preplate.” These findings show that MDGA1 is a unique marker for studies of cortical lamination and area patterning and together with recent reports suggest that MDGA1 has critical functions in forebrain/midbrain development.
The neocortex, a dorsal telencephalic (dTel) structure, is the largest region of the mammalian cerebral cortex. The adult neocortex is organized in its radial dimension into 6 major layers, distinguished by differences in the morphology and density of their neurons, connectivity, and gene expression. Differences in these same sets of properties along the tangential dimension of the neocortex divide it into anatomically and functionally distinct areas (Brodmann 1909; O'Leary and Nakagawa 2002).
During development, the earliest cortical neurons accumulate superficially within the wall of dTel, just beneath the pial surface and immediately above the ventricular zone (VZ), forming a transient layer termed the preplate (PP) (Bayer and Altman 1991). Later generated neurons accumulate within the PP and form the cortical plate (CP), splitting the PP into a subplate (SP) layer deep to the CP and a marginal zone (MZ; future layer 1) predominantly populated by Cajal–Retzius (C-R) neurons that serve important functions in cortical development (McConnell 1995). The CP is progressively populated mainly by neurons of two general types, glutamatergic neurons, including all projection neurons, which are generated in the VZ and subventricular zone of dTel, and γ-aminobutyric acidergic interneurons that are generated in the ganglionic eminences in ventral telencephalon (vTel) and tangentially migrate to the cortex (Parnavelas and others 2000; Marin and Rubenstein 2003; Kriegstein and Noctor 2004).
C-R neurons are generated in multiple sites outside of the neocortex, including the cortical hem, the septum, and vTel, from where they migrate tangentially into the cortex just beneath the pial surface (Hevner and others 2001; Meyer and others 2002; Bielle and others 2005). C-R neurons secrete Reelin, a protein implicated in CP lamination (D'Arcangelo and others 1995). Accumulating evidence indicates that laminar identities of cortical neurons are genetically determined (McConnell and Kaznowski 1991; Chenn and others 1997), suggesting that specific sets of genes expressed in progenitors and/or their neuronal progeny determine laminar fate (Hevner and others 2003; Zhong and others 2004).
The mammalian neocortex has 3 basic types of areas: the primary sensory areas that receive input relayed from peripheral sensory organs; motor areas that project to subcortical regions to control voluntary movement; and higher order areas that are often multimodal, integrate sensory and motor information, and interconnect the primary areas (Northcutt and Kaas 1995). Cortical area identities are specified by regulatory genes expressed in cortical progenitor cells, and extrinsic influences, such as thalamocortical axon (TCA) input, proposed to act later within the constraints of a genetic framework established in cortex (Rakic 1988; O'Leary 1989; Chenn and others 1997; O'Leary and Nakagawa 2002). Because mice deficient for many of the genes believed to regulate cortical patterning die at birth or earlier, well before cortical areas differentiate, the analysis of their functions requires the use of genes that are differentially expressed across the cortex as markers of positional identity (Bishop and others 2000, 2002; Mallamaci and others 2000; Hamasaki and others 2004).
The identification of genes expressed in layer-specific and area-specific patterns is important to understand mechanisms that regulate the fate decisions that influence cortical patterning. We and others have identified potential candidate genes by performing screens, including differential display polymerase chain reaction (PCR) (Liu and others 2000), representational difference analysis (Li and others 2006), and microarrays (Sansom and others 2005). However, we fortuitously identified a promising candidate gene, MDGA1, that could be involved in laminar and area patterning from a differential display PCR screen designed to identify genes involved in the development of hindbrain nuclei and their connections (Gesemann and others 2001; Litwack and others 2004).
MDGA1 (MAM domain–containing glycosylphosphatidylinositol anchor 1) is a glycoprotein anchored to the external surface of the neuronal membrane by a glycosylphosphatidylinositol anchor that shares features with immunoglobulin-containing cell adhesion molecules (IgCAMs) but has a unique domain structure that distinguishes it (Litwack and others 2004). Here we show that MDGA1 is expressed in both layer-specific and area-specific patterns, marking layers 2/3 throughout the neocortex and selectively marking the primary somatosensory area (S1) by unique expression of MDGA1 in layers 4 and 6a. Further, we show that MDGA1 is expressed transiently by Reelin-expressing cortical C-R neurons that originate external to the neocortex and by the earliest diencephalic/mesencephalic neurons. These findings show that MDGA1 is a unique marker for studies of cortical lamination and arealization, and together with evidence that MDGA1 influences the radial migration of CP neurons cell autonomously (Takeuchi and O'Leary 2006), suggest that MDGA1 functions in multiple aspects of cortical development.
Materials and Methods
C57BL/6 mice were used. Morning of the vaginal plug is embryonic day (E) 0.5; the first 24 h after birth is postnatal day (P) 0. Experiments were done following institutional approved protocols. Brain sections were prepared as described previously (Hamasaki and others 2004; Litwack and others 2004).
In Situ Hybridization
The following digoxigenin- or S35-labeled riboprobes were used: cadherin (cad) 6 (mouse full-length clone; a gift from C. Kintner, Salk Institute), cad8 (241-1481 of mouse cad8 cDNA; GenBank accession number X95600; Nakagawa and others 1999), COUP-TF1 (mouse clone; gift from M.J. Tsai, Baylor), MDGA1 (rat clone: Litwack and others 2004; mouse clone: Takeuchi and O'Leary 2006, GenBank accession number DQ788983), RORβ (mouse clone; gift from M. Becker-Andre, Serono Pharmaceutical Research Institute, Switzerland), Tbr1 (163-2301 of mouse Tbr1 cDNA; GenBank accession number BC052737). Reelin probe was from IMAGE cDNA clones (ID: 734262; Invitrogen Corporation, Carlsbad, CA). In situ hybridizations using digoxigenin- and radioactive-labeled riboprobes were done as described (Tuttle and others 1999; Litwack and others 2004).
BrdU Labeling, Immunostaining, Histochemistry, etc.
Bromodeoxyuridine (BrdU) labeling was done as previously described (Bishop and others 2003). Briefly, pregnant mice were injected with BrdU (40 mg/kg intraperitoneally). One hour later, embryos were collected, sectioned, and immunostained as described (Tuttle and others 1999). Antibodies used were rabbit polyclonal anti-BrdU (1:100, Accurate Chemical & Scientific Corporation, NY), mouse monoclonal anti-βIII tubulin, Tuj1 (1:500, Covance, CA), rabbit polyclonal anticalretinin (1:1000, Chemicon, CA), and rabbit polyclonal antibody against serotonin (5 hydroxytryptamine [5HT]; 1:50 000; Immunostar, Inc., Hudson, WI), followed by goat anti-mouse Alexa 488 or goat anti-rabbit Alexa 488 or 568 antibodies (Molecular Probes, Palo Alto, CA; Invitrogen Corporation). After incubation with biotinylated anti-rabbit immunoglobulin G (Jackson Immunoresearch Laboratories, Inc., West Grove, PA), 5HT immunoreactivity was detected using an Elite ABC Kit (Vector Laboratories, Burlingame, CA). Cytochrome oxidase (CO) histochemistry was performed on floating 20- to 40-μm sections as described (Wong-Riley and Welt 1980). Sections were photographed on a microscope equipped with a digital camera (Retiga EX, Q Imaging, or SV Micro, Carl Zeiss MicroImaging, Inc., Thornwood, NY).
MDGA1 Is Expressed in Layer-Specific and Area-Specific Patterns
We first characterize MDGA1 expression at P7, a stage when cortical layers and areas can be defined, and compare it with other markers. We processed adjacent sagittal and coronal sections through P7 mouse cortex for serotonin immunostaining, CO histochemistry, and for in situ hybridization using digoxigenin-labeled riboprobes for MDGA1 and RORβ (Fig. 1). Serotonin immunostaining marks the primary sensory areas, S1 (Fig. 1A,I), V1 (Fig. 1E,I), and A1 (Fig. 1E), due to staining of dense terminations of TCAs from the thalamic principal sensory nuclei, the ventroposterior, dorsal lateral geniculate, and ventral medial geniculate, respectively (Fujimiya and others 1986; Persico and others 2001). Secondary sensory areas, including S2 (Fig. 1A) and V2 (Fig. 1E), are also marked by serotonin staining but are distinguished from the primary sensory areas by their staining pattern and intensity. At P7, CO histochemistry predominantly reveals S1 (Fig. 1C,G,K) by staining both terminals of TCAs and postsynaptic dendrites of cortical neurons enriched in the mitochondrial CO enzyme (Woolsey and Van der Loos 1970; Wong-Riley and Welt 1980). The patterned expression of the orphan nuclear receptor, RORβ, not only predominantly marks the primary sensory areas, S1, V1, and A1 (Fig. 1D,H,L) but also has distinct expression marking secondary sensory areas, S2 and V2 (Fig. 1D,H) (Nakagawa and O'Leary 2003).
In situ hybridization for MDGA1 reveals 2 tangential bands of expression in the cortex: one in superficial layers present throughout the tangential extent of the cortex and another in deep layers that has a much more limited tangential extent (Fig. 1B,F,J). The superficial expression pattern thickens in intermediate parts of the neocortex (Fig. 1B,J), coincident with the tangential extent of the deeper expression pattern (Fig. 1B,J). Comparison of MDGA1 expression (Fig. 1B,F,J) with serotonin immunostaining (Fig. 1A,E,I) and CO histochemistry (Fig. 1C,G,K) on adjacent sections demonstrates that the expanded laminar pattern of MDGA1 expression in “intermediate” cortex matches well tangentially with the serotonin (Fig. 1B) and CO marking of S1 (Fig. 1B,C,F,G,J,K). Similarly, the expanded laminar pattern of MDGA1 expression tangentially coincides with RORβ expression characteristic of S1 (Fig. 1B,D,J,L). These findings indicate that MDGA1 is expressed in superficial layers throughout the neocortex, but a thickened superficial expression and a deep layer expression is predominantly limited to S1.
A more detailed analysis of the laminar expression patterns of MDGA1 in the primary sensory areas at P7 is illustrated in Figure 2. Adjacent sections were processed for Nissl (Fig. 2A,E,I) and serotonin staining (Fig. 2B,F,J) and in situ hybridization using digoxigenin-labeled riboprobes for MDGA1 (Fig. 2C,G,K) and RORβ (Fig. 2D,H,L). The laminar expression pattern of MDGA1 can be readily delineated by comparing its pattern with those of these other markers, particularly RORβ (Fig. 2D,H,L), which is limited to layers 4 and 5 (Nakagawa and O'Leary 2003). It is clear that in S1, MDGA1 is robustly expressed not only in layers 2/3 but also in layer 4, and exhibits in layer 4 a reiterative pattern similar to serotonin staining (Fig. 2B,C). In contrast, in the remainder of the neocortex (Fig. 1), including V1 and A1 (Fig. 2), the superficial band of MDGA1 expression is specific to layers 2/3 (Fig. 2F,G,J,K). By comparing within S1 the expression of MDGA1 (Fig. 2C) with that of RORβ (Fig. 2D,H,L), it is clear that the deep laminar expression of MDGA1 is deep to RORβ expression in layer 5, being localized predominantly to the superficial part of layer 6 (layer 6a).
In summary, at P7, within S1, MDGA1 is robustly expressed in layers 2/3 and 4, as well as by a proportion of deep layer neurons, predominantly in layer 6a. Throughout the remainder of the neocortex, MDGA1 is robustly expressed in layers 2/3, with very low or no expression detected in other layers. These data indicate that MDGA1 is expressed in a layer-specific pattern that differs in an area-specific manner.
Comparison of Area-Specific Patterned Expression of MDGA1 with Other Areal Markers and within the S1 Barrelfield
Studies of area patterning of the neocortex often make use of tangential sections through a flattened cortical hemisphere to examine the expression of areal markers within the full context of the cortex. Therefore, to confirm the area specificity of MDGA1 expression and assess its utility as an area-specific marker in the tangential dimension, we examined its expression in tangential sections relative to the expression of commonly used area markers, CO and serotonin staining, and the orphan nuclear receptors, RORβ and COUP-TF1. For this analysis, we performed CO histochemistry, serotonin and Nissl staining, and in situ hybridization using digoxigenin-labeled riboprobes for MDGA1, RORβ, and COUP-TF1 on tangential sections through layer 4 of flattened P7 cortical hemispheres (Fig. 3).
It is clear that MDGA1 expression is largely restricted to S1 and delineates the body plan in S1, including the posteromedial barrel subfield (PMBSF), the cortical representation of the large facial whiskers (Fig. 3A). For example, the pattern of MDGA1 expression is very similar to that revealed by CO histochemistry, a common marker of rodent S1 and PMBSF (Fig. 3B), which complements well the cellular pattern revealed by Nissl staining (Fig. 3C). In contrast, serotonin immunostaining (Fig. 3D), which delineates the 3 primary sensory areas, S1, V1, and A1, as well as some secondary sensory areas, such as S2, is closely mirrored by the expression patterns of RORβ and COUP-TF1 (Fig. 3E,F). These marker analyses show that MDGA1 expression is a unique gene marker that predominantly marks S1 and therefore can be of significant utility in studies of cortical area patterning. In addition, these findings are the first to show the patterned expression of the area gene markers RORβ and COUP-TF1 in the tangential cortical plane, underscoring their utility as gene markers in the broader context of area patterning.
The patterned expression of MDGA1 within PMBSF (Figs 2 and 3) prompted us to examine at P7 its expression relative to barrel components marked by Nissl, CO, and serotonin stains and expression of other gene markers, including the cell adhesion molecules (CAMs), cad6 and cad8, as well as RORβ and COUP-TF1, revealed by in situ hybridization using digoxigenin-labeled riboprobes (Fig. 4). The pattern of MDGA1 expression in PMBSF indicates that it is preferentially expressed by barrel neurons (Fig. 4A). However, comparison of the pattern of MDGA1 expression with the overall cellular pattern revealed by Nissl staining (Fig. 4B) indicates that the MDGA1 pattern is not simply due to the differences of cell density in PMBSF, where cell density is highest in the barrel walls and lower in the center of the barrel and in the septal region, that is, the space between individual barrels (Fig. 4B). Thus, these findings indicate that MDGA1 is differentially expressed at P7 by layer 4 neurons in PMBSF.
This conclusion is extended by our analysis of the patterns of expression of the other gene markers used, cad6, cad8, RORβ, COUP-TF1 (Fig. 4C,F). For example, cad6 expression (Fig. 4C) resembles that of MDGA1, whereas in sharp contrast cad8 is preferentially expressed in the septal region between the barrels (Fig. 4D) (also see Gil and others 2002). RORβ and COUP-TF1 are expressed in patterns (Figs. 4E,F) that resemble those of MDGA1 and cad6. As previously described, staining patterns for serotonin and CO staining are similar to each other, with both being strongest within the barrel walls, and at low levels in the septal regions (Fig. 4G,H). In summary, MDGA1 is differentially expressed by layer 4 neurons in S1, suggesting that it could play a role in the development of barrel patterning and their functional circuits.
Development of Layer- and Area-Specific Patterns of MDGA1 Expression
To investigate the onset of the layer- and area-specific expression of MDGA1 in the developing cortex, we used S35-labeled riboprobes for MDGA1 and initially focused on ages E15.5–P21. We selected these ages because superficial layer neurons are predominantly born around E15 and later (Takahashi and others 1999; Tarabykin and others 2001), migrate radially into the overlying CP, become a distinct layer by P7 (Figs 5 and 6), and development is essentially complete before P21. Our first goal was to determine when the S1-specific expression pattern of MDGA1 could first be readily discerned. We determined that prior to birth, this pattern was difficult to accurately identify (data not shown), therefore, we focus on postnatal ages for this description.
At P0/P1, the expanded laminar expression pattern of MDGA1 that selectively marks S1 is already evident (Fig. 1B). This S1-specific expression pattern includes a significant thickening of expression in the superficial layers that can be readily defined and is due to the expression by layer 2/3 neurons found throughout the cortex, combined with MDGA1 expression by layer 4 neurons in S1 (Fig. 1A,B,E,F). Although the S1-specific expression pattern of MDGA1 in layer 6 is evident at P0/P1, it is not as crisply defined as at P7 (Fig. 5E–H), likely due in part to the continued migration of MDGA1-positive neurons through the deep layers enroute the superficial layers. Nonetheless, these findings show that MDGA1 can be used as a unique area-specific marker for S1 for analyses done as early as P0 when mice with a mutation in one of the many genes potentially involved in arealization often die.
To address the onset of superficial layer expression, we selected for analysis the posterior neocortex, the location where the visual areas will develop, because robust expression is limited to layers 2/3 (Figs 1 and 2). At E15.5, only weak MDGA1 expression is detected in developing cortex (Fig. 6A–E). At E16.5, modest expression is evident, predominantly in the intermediate zone (IZ) (Fig. 6B–F). By E18.5, stronger expression is observed in the IZ and CP, with a band of robust expression in the most superficial portion of the CP (Fig. 6C–G). By P0, the band of robust expression in the most superficial portion of CP has thickened, and what appear to be MDGA1-expressing cells are scattered deep in the CP and IZ and are likely to be mainly migrating superficial layer neurons (Fig. 6D,H). MDGA1 expression was not detected in the MZ at these ages. These findings suggest that MDGA1 is expressed by layer 2/3 neurons during their migration and settling in the CP and layer formation (also see Takeuchi and O'Leary 2006).
Although MDGA1 is robustly expressed at P7 in layers 2/3 throughout the cortex, and in addition also in layers 4 and 6a in S1 (Figs 1–52345), by P21, expression has diminished to low or nondetectable levels in the cortex (data not shown). Thus, the cortical functions of MDGA1 appear to be limited to stages associated with cortical development.
MDGA1 Is Transiently Expressed in PP and MZ and by C-R Neurons Arising from Distinct Sources Extrinsic to Neocortex
To determine expression of MDGA1 at early stages in cortical development, we performed in situ hybridization for MDGA1 from E9.5 to E13.5 (Fig. 7). To assess the potential relationship of MDGA1 expression to that of other markers indicative of specific cell types, we also performed in situ hybridization or immunostaining for relevant markers on sections adjacent to those processed for MDGA1 expression.
At E9.5 and E10.5 (Fig. 7A,B), we do not detect MDGA1 expression in the telencephalon. However, MDGA1 is expressed in the diencephalon and mesencephalon in a layer of neurons, identified by the neuron-specific marker, Tuj1, positioned immediately beneath the pial surface and just superficial to the BrdU-positive VZ in a manner resembling a PP (Fig. 7A′,B′). These MDGA1-expressing neurons are continuous with a MDGA1-positive proliferative domain in diencephalon (compare Fig. 7A and 7A′; shown at higher magnification in Fig. 7G,G′). These findings suggest that the earliest diencephalic and mesencephalic neurons are generated by MDGA1-expressing progenitors in the diencephalon and migrate beneath the pial surface to form a nascent PP in the diencephalon and mesencephalon. These neurons, though, do not express either Reelin or Tbr1 (data not shown), in contrast to dTel C-R neurons (described below).
MDGA1 expression is first detected in the cortical PP at E12.5, but is sparse. However, at the same age (E12.5), robust MDGA1 expression is detected in several structures identified as sources of C-R neurons (see Discussion), vTel (Fig. 7C,D,J), septum (Fig. 7C,I), and the cortical hem (Fig. 7D,H). These findings suggest that MDGA1 is expressed by these origins of C-R neurons. This suggestion is supported by our finding that expression of MDGA1 in septum (Fig. 7I,I″) and vTel (Fig. 7J,J″) is largely coincident with expression of Tbr1, one of the essential transcription factors for the development of C-R neurons (Hevner and others 2001), as well as with Reelin, a classic marker for C-R neurons (Meyer and Goffinet 1998). In addition to MDGA1 being expressed in these 3 sources of C-R neurons, we find that MDGA1-expressing cells are distributed within the MZ/PP in a pattern contiguous with these sources. A low density of MDGA1 expression is observed in the cortical MZ/PP at E12.5 (Fig. 7C,D), but MDGA1 expression increases substantially by E13.5 (Fig. 7E,F) consistent with the continued emigration of MDGA1-expressing C-R neurons into the neocortex from their MDGA1-expressing extrinsic origins. These relationships of MDGA1 to sources of C-R neurons and their migratory routes are very similar to those described for other markers of C-R neurons (Meyer and Goffinet 1998; Meyer and Wahle 1999), suggesting that MDGA1 marks at least a proportion of Reelin-positive cortical C-R neurons that originate from these sources extrinsic to the neocortex.
At E13.5, in the more mature, most lateral part of the developing cortex, the PP is just beginning to be split into a superficial MZ and a deep SP by the earliest generated CP neurons (which will form deep layer 6) that aggregate within it. At this stage, MDGA1 expression already appears downregulated in this most lateral, mature portion of the cortical wall (Fig. 7F,K). Over the next day or 2, MDGA1 expression diminishes to very low or nondetectable levels in the remnants of the PP, that is, the MZ and SP, such that by E15.5, MDGA1 expression is not detected in either layer. Taken together, these findings suggest that MDGA1 marks the earliest neurons in the nascent diencephalon, that MDGA1 marks C-R neurons that arise from multiple extracortical sources and emigrate into the cortical MZ/PP, and that the early expression of MDGA1 in the cortical PP/MZ, presumably by C-R neurons, is transient.
MDGA1 is a unique cell surface glycoprotein that is expressed predominantly in the developing nervous system, both central and peripheral (Litwack and others 2004; Takeuchi and O'Leary 2006; present study). In the present study, we examine the expression pattern of MDGA1 during forebrain development in mice, comparing its patterned expression with other markers, and find that it has very unique and dynamic patterns of expression. Our findings show that 1) MDGA1 is a layer-specific marker, marking all or a high proportion of layer 2/3 neurons and 2) MDGA1 is a marker for S1, an area in which it selectively marks layer 4 neurons in addition to layer 2/3 neurons, as well as a proportion of layer 6a neurons. In addition, our findings show that 3) MDGA1 is a marker for C-R neurons that arise from multiple sources outside of the neocortex and migrate into it and 4) MDGA1 marks the earliest population of diencephalic/mesencephalic neurons that appear to arise from a discrete domain of MDGA1-positive progenitors in the diencephalic VZ. Based on these expression patterns, functions of IgCAMs, including MDGA1 (Takeuchi and O'Leary 2006), we suggest that MDGA1 has important functions in multiple aspects of forebrain development. Finally, we present the patterned expression of several additional markers (e.g., RORβ, COUP-TF1) in the novel context of the entire cortex in tangential sections and their relationship to MDGA1 expression and classic markers including serotonin and CO, as well as details of the expression patterns of each of these markers and cadherins that mark distinct compartments of S1 barrels.
MDGA1 as a Layer-Specific Marker
We show that MDGA1 is expressed by layer 2/3 neurons throughout the neocortex of postnatal mice. During development, MDGA1 is expressed in patterns consistent with its expression by migrating layer 2/3 neurons, suggesting a cell-autonomous role for it in controlling the migration and settling of these neurons, as recently shown by using RNA interference (RNAi), targeting different sequences of mouse MDGA1 (Takeuchi and O'Leary 2006). Together, these findings indicate that MDGA1 is a unique marker for studies of cortical lamination and arealization and has critical functions in the development of these fundamental cortical organizations.
MDGA1 Is an Area-Specific Marker Selectively Identifying the Somatosensory Area (S1)
We show that MDGA1 is a marker for S1, an area in which it selectively marks layer 4 neurons in addition to layer 2/3 neurons, as well as a proportion of layer 6a neurons. Thus MDGA1 is useful as a marker for studies of area patterning and may be involved in this process. Our finding that the S1 area-specific pattern of MDGA1 expression can be identified as early as birth indicates that MDGA1 can be even more broadly useful as a marker for studies of area patterning, especially because mice deficient for genes potentially involved in regulating area patterning often die at birth, before areas emerge and are definable by classic stains such as CO or serotonin.
A few previous studies have reported area-specific gene expression in the neocortex. Occ1, which encodes a primate homologue of TSC-36/follistatin-related protein, is preferentially expressed in layers II, III, IVA, and IVC of V1 of macaque monkeys but is also weakly expressed in S1 and A1 (Tochitani and others 2001). Latexin, an inhibitor for carboxypeptidase A, is expressed in glutamatergic neurons of the secondary or nonprimary sensory areas with substantially weaker expression in the primary sensory areas (Arimatsu and others 1999). Zhong and others (2004) report several genes specific to cortical layer 4 or layers 2/3, including unc5h4 and stem cell factor whose expression patterns appear specific to some areas. Patterned expression of the orphan nuclear receptor, RORβ, matches well with S1, V1, and A1 (Nakagawa and O'Leary 2003; present study). However, none of these genes exhibit an “area-specific” pattern limited to one cortical area. In contrast, although MDGA1 is expressed robustly in layers 2/3 throughout the entire neocortex, the area-specific expression of MDGA1 is due to its unique expression in layers 4 and 6a of S1. Interestingly, the H-2Z1 transgenic mouse line, characterized by a LacZ transgene driven under control of regulatory elements from a major histocompatibility complex class I gene, also selectively marks S1 (Cohen-Tannoudji and others 1994).
Within layer 4 of S1, the majority of layer 4 neurons expresses MDGA1, including most that form barrels, an anatomical and functional unit characteristic of layer 4 of rodent S1. Given that IgCAMs regulate dendrite and axon growth, synaptogenesis, and other developmental processes (Hirano and others 2003) and that MDGA1 influences the radial migration of cortical neurons (Takeuchi and O'Leary 2006), we speculate that MDGA1 is involved in barrel development. This possibility is particularly intriguing because barrel development requires a directed development of primary dendrites and their arborizations, and our preliminary findings using RNAi techniques indicate that MDGA1 influences dendritic development of cortical neurons (A Takeuchi and DDM O'Leary, unpublished data). Our findings that MDGA1 and other CAMs, such as cad6 and cad8, mark specific barrel components suggests that they could cooperate to develop barrels.
Our findings also show that MDGA1 selectively marks a subset of layer 6a neurons in S1. It is of interest to determine the type of layer 6a neuron that MDGA1 marks, but that issue will be easier to address once we successfully generate MDGA1-specific antibodies that work not only on blots but also on tissue sections (Litwack and others 2004; and A Takeuchi and DDM O'Leary, unpublished data).
MDGA1 Marks Cortical C-R Neurons Generated External to the Cortex
We present evidence that at early stages of cortical development, prior to the generation of CP neurons, MDGA1 is expressed by Reelin-positive C-R neurons that originate from multiple discrete sources outside of the neocortex and emigrate into it. At E12.5, expression of MDGA1 in the cortical PP is low but robust expression is observed in the septum, the vTel and the cortical hem. All 3 structures generate C-R neurons that emigrate tangentially into the cortex and become distributed in the cortical MZ (Meyer and others 2002; Takiguchi-Hayashi and others 2004; Yamazaki and others 2004; Bielle and others 2005). As early as E9.5, a neural tube stage days before the generation of cortical C-R neurons, MDGA1 marks a population of neurons in the nascent diencephalon and mesencephalon that are generated by MDGA1-expressing progenitors in the diencephalic VZ and appear to migrate beneath the pial surface to form a layer that resembles a nascent PP in the diencephalon and mesencephalon. However, in contrast to their apparent dTel counterparts, that is, C-R neurons, they do not express either Reelin or Tbr1.
C-R neurons are the predominant cortical source of the secreted extracellular protein, Reelin (D'Arcangelo and others 1995; Ogawa and others 1995; Marin and Rubenstein 2003). Each of the 3 extraneocortical sources of C-R neurons also express Tbr1. Tbr1 knockout mice have cortical lamination defects correlated with a downregulation of Reelin (Hevner and others 2001), and the defects appear similar to those in mice deficient for functional Reelin protein (D'Arcangelo and others 1995). In addition, both the septal and vTel sources of C-R neurons express Dbx1. Genetic ablation of Dbx1-positive neurons results in a parallel loss of C-R neurons (Bielle and others 2005). Although C-R neurons are heterogeneous, they appear to share MDGA1 as a common marker.
Interestingly, MDGA1 expression in C-R neurons, and in the cortical MZ/PP in general (presumably reflecting expression in C-R neurons), is transient and diminishes to nondetectable levels by E14.5–E15.5. We suggest that MDGA1 controls the tangential migration of C-R neurons from their origins in the septum, cortical hem, and vTel to the cortex, as well as the tangential migration of the earliest population of MDGA1-expressing diencephalic and mesencephalic neurons that arise from a MDGA1 VZ domain in diencephalon. We base this suggestion on the relationship of MDGA1 expression to C-R neurons and their origins, the transient nature of MDGA1 expression, being downregulated shortly after the neurons complete migration, and recent functional studies that show a cell-autonomous role for MDGA1 in radial cortical neuronal migration (Takeuchi and O'Leary 2006).
Our findings show that MDGA1 is a unique marker to study cortical development, including laminar and area patterning, and is likely involved in the development of these features. However, much needs to be learned to further understand MDGA1 and its functions, including 1) does MDGA1 act and/or signal via heterophilic or homophilic interactions, 2) is MDGA1 a receptor, part of a receptor complex, is it a ligand, or some combination of these possibilities, 3) what class of layer 6a neurons does MDGA1 mark in S1, 4) does MDGA1 control the tangential migration of specific forebrain neurons, for example, Reelin-expressing C-R neurons that in turn control other important aspects of cortical development, and 5) what other functions does MDGA1 have in development?
This work was supported by National Institutes of Health grants R37 NS31558 and P01 NS31249 (DDMO'L) and a fellowship (AT) from Uehara Memorial Foundation. We thank Carlos Perez-Garcia and Shen-ju Chou for discussion and comments on the manuscript and are grateful to John Rubenstein for advice on early brain structures. Conflict of Interest: None of the authors have a financial interest related to this work.