|Home | About | Journals | Submit | Contact Us | Français|
Cortical γ-aminobutyric acid (GABA)ergic interneurons in rodents originate mainly in ventrally positioned ganglionic eminences (GEs), but their origin in primates is still debated. We studied human fetal forebrains during the first half of gestation (5–23 gestational weeks, gw) for the expression of ventral transcription factors, Nkx2.1, Dlx1,2, Lhx6, and Mash1, important for development of neocortical interneurons. In embryonic (5–8 gw) human forebrain, these factors were expressed in the GE but also dorsally in the neocortical ventricular/subventricular zones (VZ/SVZ). Furthermore, their expression was retained in cells of all fetal cortical layers up to midgestation (20 gw). Nkx2.1 continued to be expressed not only in the GE but also in a subpopulation of neocortical interneurons. Moreover, proliferation marker Ki67 revealed that calretinin+, Mash1+, and Nkx2.1+ cells proliferate in the neocortical VZ/SVZ at midgestation. At least some of the Mash1+ progenitors in the neocortical SVZ could be colabeled with GABA, whereas others were oligodendrocyte progenitors, indicating a link between the 2 lineages. Taken together, these results suggest the existence of several categories of dorsal interneuronal progenitors in the human neocortical VZ/SVZ, in addition to ventrally derived cortical interneurons described in rodents. These human-specific developmental events may underlie human brain's higher complexity and capacity to process information.
Cortical interneurons provide inhibitory input to principal (pyramidal) cells and thus are necessary for building and fine-tuning of cortical circuitries. Whereas cortical principal neurons are derived from dorsal forebrain and migrate radially into cortical plate, majority of cortical interneurons are, at least in rodents, derived from ventral (subcortical) forebrain and migrate tangentially into developing cortex (Anderson et al. 1997; Parnavelas et al. 2000; Marin and Rubenstein 2001). This tangential migration was identified in explants and slice cultures from multiple mammalian species including mice (Anderson et al. 1997; Wichterle et al. 1999), rats (Lavdas et al. 1999), ferrets (Anderson et al. 2002), and humans (Letinic et al. 2002).
In rodents, majority of cortical interneurons originate in the medial ganglionic eminence (MGE) from progenitors that express Dlx2 and Nkx2.1—ventral transcription factors. Although in the human brain a significant part of the neocortex is actually positioned ventrally to basal ganglia, we will use the nomenclature “ventral” for subcortical structures and “dorsal” for cortex since this is commonly used in the literature. Dlx is also abundant in the human fetal brain, both in the GE and neocortical ventricular/subventricular zones (VZ/SVZ) (Letinic et al. 2002; Rakic and Zecevic 2003b). Nkx2.1, another ventral homeobox gene, has a role in specification of a subpopulation of neocortical interneurons originating in the MGE. Its genetic ablation in mice results in 50% loss of γ-aminobutyricacid (GABA)ergic cortical interneurons (Anderson et al. 1997; Sussel et al. 1999; Xu et al. 2005). Nkx2.1 is downregulated in cells that migrate to the neocortex and maintained in cells that migrate to the striatum (Nobrega-Pereira et al. 2008). Interneuronal progenitors migrating from the MGE to the neocortex still express Dlx, and another transcription factor, Lhx6, which is an LIM homeodomain transcription factor essential for their tangential migration (Grigoriou et al. 1998; Lavdas et al. 1999; Anderson et al. 2001; Du et al. 2008). Mash1 (mammalian achaete-schute homolog 1) is a proneural gene of particular interest for development of neocortical interneurons, expressed both in rodents (Horton et al. 1999; Yun et al. 2002; Long et al. 2007) and primates (Letinic et al. 2002; Petanjek et al. 2009).
Increasing body of evidence suggests that in humans and in other primates, cortical interneurons have dual origin, from the ventral forebrain, as well as from dorsal cortical VZ/SVZ (Letinic et al. 2002; Rakic and Zecevic 2003b; Petanjek et al. 2009; Rakic 2009; Fertuzinhos et al. 2009; Zecevic et al. 2010). Human brain, and particularly the cerebral cortex, has a longer developmental period, larger size, and evolutionary new areas (e.g., language areas), with enlarged upper cortical layers (layers II and III) (Hill and Walsh 2005), prominent subpial granular layer, and expanded diversity of layer I (Zecevic and Rakic 2001; Rakic and Zecevic 2003b). Moreover, in the enlarged human outer SVZ cell proliferation continues well into midterm period (Zecevic et al. 2005; Hansen et al. 2010). In order to further describe the distribution of the potential interneuronal progenitors, we studied the expression of relevant transcription factors on frozen sections of human forebrains during the first half of gestation using indirect immunohistochemistry.
We present the evidence that the expression of ventral transcription factors, including Nkx2.1, spreads over dorsal areas of human neocortex from the earliest stages of development. We also describe, for the first time, the expression of Lhx6 in the human developing cortex from early fetal stage (8 gw) to midterm (20 gw). In addition, Mash1-expressing progenitors in the neocortical VZ/SVZ could present a link between interneuronal and oligodendrocyte lineages in humans since we could label them with markers of both oligodendrocyte and interneuronal lineages. Our conclusion is that human neocortical interneurons arise from several progenitor sources, located both dorsally and ventrally. This may explain greater diversity of interneuronal subtypes in humans and may account for the expansion in information processing power that provided humans with their key evolutionary advantage over other mammalian species (DeFelipe 1999; Jones 2009; Rakic 2009).
Human fetuses from 5 to 23 gw(Gestational weeks correspond to postconceptional weeks. Full term is 40 gw.) (n = 17; see Table 1) were obtained from the Human Fetal Tissue Repository at the Albert Einstein College of Medicine (Bronx), after legal abortions, with proper consent from parents. Handling of human material was done with special care following all necessary requirements and regulations set by the Ethics Committee of the University of Connecticut and the Helsinki Declaration. In all studied cases, ultrasound and gross neuropathological examination excluded brain pathology. Tissue was dissected to smaller blocks (for details, see figure 4B in Jakovcevski and Zecevic 2005a), fixed in 4% formaldehyde solution in 0.1 M phosphate buffer, cryoprotected by immersion in 30% sucrose, embedded in Tissue Tek (Sakura), and frozen in isopentane cooled to −70 °C. Tissue was preserved at −70 °C until cutting into 15-μm thick sections used for immunohistochemistry.
Various primary antibodies used in this study are listed in Table 2. Since antibody sensitivity and specificity are of critical importance for our work, we will here briefly describe the most important antibodies that we used. For detection of calretinin, we used both rabbit polyclonal and mouse monoclonal antibodies. Double labeling with the 2 antibodies was performed in order to ascertain that both antibodies label the same cells within human cerebral cortex (data not shown). In addition, polyclonal anti-calretinin antibody from Swant was previously used for several publications from our laboratory (e.g., Zecevic et al. 1999; Rakic and Zecevic 2003b) and tested by immunoblot on the human midgestational tissue (Supplementary Fig. 1A). Polyclonal anti-calbindin antibody was also previously used by us and others on human tissue (Letinic and Kostovic 1998; Zecevic et al. 1999), and the immunoblot showed a clear single band of proper molecular weight (Supplementary Fig. 1B). For the human Nkx2.1 (TTF-1), a monoclonal antibody produced in rabbit was used. This antibody gave a clear nuclear staining of cells in both the GE and in the cortex, and immunoblot gave a band of a proper molecular weight (Supplementary Fig. 1C). Since previously we demonstrated similar immunostaining pattern for Nkx2.1 and confirmed this result with in situ hybridization (Rakic and Zecevic 2003b), we consider the staining specific both in the GE and the neocortical VZ/SVZ. Nkx2.1 antibody stained only MGE in the mouse embryo (data not shown). However, some cytoplasmic and background staining was observed in the neocortical plate and medial telencephalic wall; we considered only nuclear staining to be specific for this transcription factor. Polyclonal rabbit Lhx6 antibody, used in this study for the first time on human tissue was also tested by immunoblot (Supplementary Fig. 1D). Pan-Dlx (Dll) antibody which labels 4 Dlx isoforms expressed in the brain (Dlx1-2, 5-6) was a gift from Dr Y. Morozov (Yale University, New Haven, CT) and was also previously used successfully on human brain tissue (Letinic et al. 2002; Rakic and Zecevic 2003b). Both monoclonal antibody against Mash1 (also known as ASCL1) from Pharmingen and polyclonal anti-GABA antibody were previously used on human and monkey fetal brains (Letinic et al. 2002; Rakic and Zecevic 2003b; Fertuzinhos et al. 2009; Petanjek et al. 2009). Polyclonal anti-PDGRFRα antibody was a gift from Dr W. Stallcup (Stanford-Burnham Medical Research Institute, La Jolla, CA) and was successfully used on human tissue in several publications (Jakovcevski and Zecevic 2005a, 2005b; Jakovcevski, Filipovic, et al. 2009). To assure for staining specificity, stringent controls for immunohistochemistry were performed as described below.
Frozen brain blocks were serially sectioned in the coronal or sagittal plane at 15 μm thickness. For various transcription factors and other antigens, with the exception of PDGFRα, antigen unmasking was done in 0.01 M sodium citrate solution (pH 9.0) for 30 min at 80 °C, as previously described (Jakovcevski, Siering, et al. 2009). Sections were then incubated in blocking solution (1% bovine serum albumin [Sigma], 5% normal goat serum [Vector Laboratories], and 0.5% Tween 20 in phosphate-buffered saline) for 30 min at room temperature. Primary antibodies were applied overnight at 4 °C, whereas corresponding secondary antibodies were subsequently applied for 2 h at room temperature. For double staining, the primary antibodies from different species were mixed at optimal dilutions and detected using mixtures of appropriate secondary antibodies (multiple-labeling grade antibodies; Jackson ImmunoResearch). Nuclei were counterstained with 5-min incubation in 1% bisbenzamide (Polysciences). For each antigen, specificity of staining was tested by omitting the primary antibody or replacing it with different concentrations of appropriate normal serum. In addition, imaging and quantitative analysis were performed only on those sections on which the morphology of the labeled cells appeared to be characteristic of the cell type expected to be labeled, and the staining pattern was consistent with the subcellular distribution of the antigen (e.g., transcription factors within the nucleus etc.).
The quantifications were performed on an Axioskop microscope (Zeiss) equipped with a motorized stage and Neurolucida software–controlled computer system (MicroBrightField). Sections were observed and delineated under low-power magnification (×10 objective) with a 365/420 nm excitation/emission filter set (01, Zeiss, blue fluorescence). The nuclear staining allowed delineation of areas of interest (e.g., GE, cortical plate etc.). For cells expressing various transcription factors, GABA, calretinin, and Ki67, 2D counts of the labeled cells/nuclear profiles were performed under ×40 objective. In each delineated field and profile number was normalized to the surface area. In addition, the number of immunolabeled cell profiles normalized to overall cell number (number of nuclear profiles) was calculated and will be further referred to as a “relative density.” In large midgestational brains a grid (60 × 60 μm) was placed over area of interest and random samples 120 μm apart were counted and averaged for overall area. At least 3 sections per case and marker were counted. In some cases when we doubted the specificity of immunostaining or when parts of sections were missing and/or appeared damaged during tissue processing, we did not include results in our analysis. Due to a very limited sample size, we did not consider applying stereological methods, and we offer our quantifications just as an estimate and look forward for further studies on more cases to confirm our results.
Brain slices from 19 to 22 gw fetuses were dissected through the parietal lobe, and tissue samples were homogenized in a tissue grinder in lysis buffer with protease inhibitors. Homogenates were run on a 4–15% sodium dodecyl sulphate–polyacrlyamide gel electrophoresis and transferred to polyvinylidene fluoride transfer membranes. Membranes were blocked with 3% nonfat milk in Tris buffer saline Tween (Tris 10 mM, NaCl 150 mM, Tween-20 0.05%, pH 7.4) and incubated overnight at 4 °C with primary antibodies (calretinin, 1:1000; calbindin, 1:1000; Nkx2.1 1:1000; Lhx6 1:2000), followed by incubation for 1 h at room temperature with the horseradish peroxidase–conjugated secondary antibodies. The signals were detected by enhanced chemiluminescence (Amersham Biosciences).
Image analysis and photographic documentation were done using a confocal laser-scanning microscope (Carl Zeiss, LSM 510) and Axioplan fluorescent microscope (Carl Zeiss). Image processing was done by Adobe Photoshop CS2 software (Adobe Systems Inc.) and was limited to brightness/contrast adjustments.
We previously reported that Nkx2.1 expression in the human embryonic and fetal brain is, unlike in rodents, spread to the cortical VZ and cortical plate from early developmental stages (Rakic and Zecevic 2003b). Using a new monoclonal antibody against Nkx2.1 (TTF1), we now confirm our results for fetal stages at midgestation (20 gw, Fig. 1A–G). Strong expression of Nkx2.1 is demonstrated in the MGE (Fig. 1B !supportNestedAnchors]>]>), fading somewhat in the striato-cortical border (Fig. 1C !supportNestedAnchors]>]>). More dorsally, in the cortical VZ, however, the expression becomes stronger, and it is especially intensive in the medial cortex and the septal area (Fig. 1D–F). Notably, a number of cells in the deeper part of the cortical plate were also Nkx2.1+ (Fig. 1G). Double labeling with interneuronal markers revealed that Nkx2.1+ cells in the cortical plate are frequently calbindin+ and calretinin+, especially in the deeper cortical layers where 50% and 55% Nkx2.1 cells coexpress calbindin or calretinin versus around 20% in superficial layers (Fig. 1H–J). Since the finding of Nkx2.1+ cells in the neocortical plate was important and unexpected, we have confirmed it also by the immunoblot analysis (Fig. 2). Neocortical Nkx2.1+ cells could not have been predicted from the studies in rodents, where migrating interneurons need to downregulate Nkx2.1 in order to tangentionally migrate toward the neocortex (Nobrega-Pereira et al. 2008). This suggests that a different mechanism regulates tangentional migration in humans. Alternative interpretation is that human neocortical VZ/SVZ is an additional site of origin for Nkx2.1+ cells.
Rabbit polyclonal Dll antibody (pan-Dlx isoforms 1-2, 5-6) labeled numerous cells in the neocortical VZ/SVZ but also in the cortical plate at midterm (Fig. 3). Double labeling with Dll and Nkx2.1 antibodies could not be performed since both antibodies were produced in rabbit, but we quantified Dll+ cells and Nkx2.1+ cells on adjacent coronal sections from the posterior frontal lobe (see Figs 1 and 3A) and from the occipital lobe (data not shown). In some areas, including the striatum and the fetal neocortical layer 1, around 80% of all cells express Dll. In all studied regions of a 20-week-old brain, with the exception of the MGE, Dll+ cells are more numerous than Nkx2.1+ cells (cf. Figs 1 and Supplementary Fig. 2). It is thus probable that Nkx2.1+ cells represent a subpopulation of Dll+ cells, as described in rodents (Sussel et al. 1999), but we cannot exclude that both populations separately exist in the same brain regions or that there is only a partial overlap between the 2 populations.
The majority of the cortical interneurons in rodents originate from Nkx2.1+ progenitors in the MGE. They subsequently downregulate Nkx2.1 and upregulate Lhx6 while migrating tangentionally toward neocortex (Xu et al. 2004; Butt et al. 2005; Liodis et al. 2007; Nobrega-Pereira et al. 2008). Thus, we examined the expression pattern of Lhx6, as a marker of major MGE—derived interneuronal populations. At 8 gw, immediately after the emergence of the cortical plate, Lhx6 is strongly expressed in the GE mantle zone but also in the cortical VZ (Fig. 4A,B). Most prominent expression of Lhx6 was in the striato-cortical border, a sulcus between the lateral ganglionic eminence and the neocortical VZ (Fig. 4C !supportNestedAnchors]>]>), suggesting this as possible migratory route. Similarly to the Nkx2.1+ cells, Lhx6+ cells were also present in the cortical VZ and cortical plate at 8 gw (Fig. 4D). This expression pattern remained until midgestation, where Lhx6 is again strongly expressed in the cortical VZ (Fig. 4E). With the thickening of the cortical plate and formation of the subplate zone, Lhx6+ cells appeared very dense in the subplate, suggesting their accumulation or/and migration through this layer (Fig. 4E). This distribution closely mirrors the Lhx6 expression pattern in rodents (Lavdas et al. 1999; Fogarty et al. 2007; Liodis et al. 2007). Quantification of Lhx6+ cells in the VZ at 8, 15, and 20 gw shows peak density in the striato-cortical border at all ages, whereas relative densities of Lhx6+ cells in the cortical VZ/SVZ are highest at 15 gw, and later declines, congruent with the general decline in neurogenesis toward the end of midgestation (Fig. 4F). The distribution of Lhx6+ cells is consistent with the idea that Lhx6+ cells are migratory GE—derived interneurons. The reports in rodents show that the majority of Lhx6+ cells will express parvalbumin, or somatostatin, but not calretinin (Fogarty et al. 2007; Liodis et al. 2007). Since parvalbumin and somatostatin are very sparsely expressed at midgestation in human brain (Zecevic et al. 2010), we could not directly prove the lineage of Lhx6+ cells in tissue sections.
Cells in the neocortex labeled with transcription factors, including Dlx2, Mash1, and Lhx6 originate in the GE in both rodents and primates, some of these cells, and particularly Mash1+ cells were suggested to be the progenitors of dorsally derived interneurons (Letinic et al. 2002; Petanjek et al. 2009). We thus examined this cell population closer through a series of double-labeling experiments in midgestational human brain.
We demonstrate that in the GE all Mash1+ cells express Dll (Fig. 5A–C), in accord with a previous report in human fetal brain (Letinic et al. 2002). In the cortical VZ/SVZ, however, there was little if any colabeling of these 2 transcription factors (Fig. 5D–F). One notable difference between medial and dorsolateral cortex is relative abundance of Mash1+ cells accompanied with relative paucity of Dll+ cells in the medial cortical VZ/SVZ, whereas the opposite is true in the dorsolateral cortex (Fig. 5D–F). In the MGE, however, numerous cells were double labeled with Nkx2.1 and Mash1, but single-labeled Nkx2.1+ and Mash1+ cells were also present (Fig. 5G–I). In contrast to MGE, in the cortical VZ/SVZ, Nkx2.1+ cells were not colabeled with Mash1 (Fig. 5J–L). Overall our data suggest that Mash1+ cells in the cortical VZ/SVZ are a distinct population, not colabeled with either Dll or Nkx2.1. This Mash1+ subpopulation is especially frequent in the medial cortex at this stage of development, and may have local origin, as suggested by other studies on primates (Letinic et al. 2002; Petanjek et al. 2009).
In order to analyze which of the cells present in the midgestational VZ/SVZ are still mitotic, and thus progenitor cells, we performed a series of double-labeling experiments with Ki67, a proliferating cell marker and various interneuronal markers. Double labeling with calretinin confirmed that indeed a subpopulation of calretinin+/Ki67+ cells exist in the VZ/SVZ of the human neocortex (Fig. 6A–C). Next, we tested the proliferation of Mash1+ cells in the cortical VZ/SVZ (Fig. 6D–F). Again, in the medial cortical SVZ, we discovered that a significant proportion of Mash1+ cells undergo mitosis, as judged by their colabeling with Ki67. We next tried to colabel calretinin and Mash1 in the same area of the medial cortical SVZ, expecting that at least some of the proliferating calretinin+ cells were Mash1+. Notably, there were very few colabeled cells (Fig. 6G–I). Finally, many Nkx2.1+ cells in the cortical VZ/SVZ were colabeled with Ki67, suggesting that a subpopulation of Nkx2.1+ cells in the neocortical VZ/SVZ is still mitotic (Fig. 6J–L). At least some of the proliferating Nkx2.1+ cells in the cortical VZ/SVZ could also be oligodendrocyte progenitors, but this population is very sparse during midgestation, as we reported earlier (Rakic and Zecevic 2003a). Thus, high density and high percentage of mitotic Nkx2.1+ cells indicate that many of them are involved in neurogenesis.
Next we tried to address whether Mash1+ cells could be colabeled with GABA and found that this was indeed the case both in the neocortical VZ/SVZ and in the cortical plate (Fig. 7A–D). This lead to the conclusion that Mash1+ progenitors in the cortical SVZ could give rise to a subpopulation of calretinin− interneurons (Fig. 6G,H). Mash1+ cells in the forebrain could also be labeled with platelet-derived growth factor receptor alpha (PDGFRα), a marker of early oligodendrocyte progenitor cells (Jakovcevski and Zecevic 2005b; Jakovcevski, Filipovic, et al. 2009). Whereas in the cortical plate, and especially in the subplate, a high percentage of Mash1+ cells coexpressed PDGFRα, this percentage was considerably lower in the VZ/SVZ (Fig. 7D–G). One notable exception is a part of the SVZ between GE and cortex where we previously described a “stream” of migratory stem-like cells that express markers of early oligodendrocyte progenitor cells (Rakic and Zecevic 2003a). Most of the cells in this “stream” are Mash1+ (Fig. 7H). Our data suggest that, at midgestation, both Mash1+ interneurons and Mash1+ oligodendrocyte progenitors are present in the cortex, with higher percentage of former than latter cells. There is, however, in all cortical areas, a significant proportion of Mash1+ cells that express neither of the 2 markers that we used (Fig. 7D). These cells are most probably still noncommitted, or committed to either interneuronal or oligodendrocyte lineages, but at earlier or later stages than defined by GABA and PDGFRα expressions, respectively. Mash1+ cells could also represent a link between these 2 lineages, as discussed below. Taken together, our data indicate the existence of several distinct subpopulations of interneuronal progenitors in the cortical VZ/SVZ at midgestation.
The aim of our study was to clarify the origin of cortical interneurons in humans and to specify the subpopulations of neurons that arise from dorsal and ventral progenitors. Although we do not provide a direct answer to either of these questions, we can draw several significant conclusions. First, we discovered several possible interneuronal progenitors in the cortical VZ/SVZ, including calretinin+ cells, Mash1+ cells, and Nkx2.1+ cells. Second, all these progenitors have specific discrete distributions within the cortical VZ/SVZ, with especially high densities in the medial neocortex at midterm. In contrast, Dll+ cells, probably representing ventrally derived population, are more frequent in dorsolateral cortical areas. Third, we generated a novel set of data regarding the distribution of Lhx6+ cells in the human fetal brain. We consider Lhx6+ cells to be ventrally derived interneurons, as indicated by previous research on rodents (Lavdas et al. 1999; Fogarty et al. 2007).
Several transcription factors that have been characterized as “ventral,” that is, indicative of lineages of the ventral telencephalic origin, in human have a strong expression in the dorsal areas of neocortex, including VZ/SVZ and the cortical plate.
In the current study, we clarify Nkx2.1 protein expression at midterm in areas not previously reported in studies on mice, including neocortical proliferative zones, VZ/SVZ, and the cortical plate. Our previous study has also demonstrated the expression of Nkx2.1 messenger RNA and protein in the cortical VZ/SVZ from 5-week-old embryos until midgestation (Rakic and Zecevic 2003b). Another study on human fetal tissue, however, reported Nkx2.1 expression only in the MGE and the striatum, similar to what is seen in rodents (Fertuzinhos et al. 2009). In the study of Fertuzinhos et al. (2009), different Nkx2.1 antibody was used, and immunostaining of the dorsal cortical VZ was also observed, but considered unspecific. This is often the case with immunostainings on human tissue and careful analysis is needed to confirm the specificity of staining. The monoclonal rabbit Nkx2.1 antibody that we used produced specific nuclear staining, which was also confirmed by western blot analysis and our previous in situ hybridization findings (Rakic and Zecevic 2003b). Additionally, we used the same antibody to stain mouse embryos and got a staining pattern confined to ventral forebrain, characteristic for rodent Nkx2.1 expression (Yu X and Zecevic N, unpublished data). Thus, we are confident that our results of Nkx2.1 expression are specific. The finding of Nkx2.1+ progenitors in the human cortical VZ/SVZ raises further questions into the origin of these cells.
In rodents, Mash1 is expressed in the GE, where it is required for controlling the balance between early and late progenitors of GABAergic neurons (Horton et al. 1999; Yun et al. 2002; Long et al. 2007). Mash1 is expressed, although at lower levels, at E15.5 in cortical progenitors (Britz et al. 2006), where it may induce the expression of Dlx1 (Fode et al. 2000). Describing the cells that express 2 ventral transcription factors, Dlx2 and Mash1, Letinic et al. (2002) have postulated predominantly dorsal origin of cortical interneurons in humans. These authors identified dorsal interneuronal progenitors as Mash1+/Dll- cells and estimated that they comprise about 65% of all cortical interneurons, whereas 35% of cortical interneurons have ventral origin (Letinic et al. 2002). In addition, in the monkey, Mash1+/GAD65+ cells were reported to be late interneuron progenitors in the cortical VZ/SVZ (Petanjek et al. 2009). Cortical Mash1+ cells in our study were Dll− and Nkx2.1− but also calretinin−. However, a significant proportion of them could be colabeled with GABA (Letinic et al. 2002; this study). Thus, we confirmed that Mash1+ progenitors are present in the human cortical VZ/SVZ at midgestation and that at least some of them are cortical interneurons. Although it has been suggested that Mash1+ cells in mice could be the earliest interneuronal progenitors (Poitras et al. 2007), other groups proposed that at late embryonic stages Mash1+ cells are oligodendrocyte progenitors (Parras et al. 2007; Kim et al. 2008). Our data indicate that at midgestation in humans both views are correct.
We describe the expression pattern of Lhx6 in the human forebrain during the first half of gestation. The first Lhx6+ cells appear in the human forebrain simultaneously with the emergence of the cortical plate (around 7–8 gw). At this time point, their distribution very closely resembles that in rodent brain at the beginning of cortical development, with one notable difference—accumulation of Lhx6+ cells in the striato-cortical border. It is unclear if this specific region of the VZ is a migration route for human ventrally derived interneurons, but in the same region, we have previously reported the high expression of other transcription factors, namely Dll, Olig2, and Pax6 (Rakic and Zecevic 2003b; Jakovcevski and Zecevic 2005a; Mo et al. 2007). Others have also reported Nkx2.1 to be strongly expressed at striato-cortical border (Fertuzinhos et al. 2009).
Our experiments suggest the existence of several classes of interneuronal progenitors in the human cortical VZ/SVZ. This is in agreement with the previous work published on primate cortical interneurons (Letinic et al. 2002; Fertuzinhos et al. 2009; Petanjek et al. 2009; Rakic 2009). All these studies, together with our previous and current results (Rakic and Zecevic 2003b; Zecevic et al. 2010) have several common topics. It appears that dorsally derived interneurons are relatively late-born in comparison with ventrally derived ones. Many of those late-generated interneurons are calretinin+ cells. These cells include double-bouquet cells especially abundant in the upper cortical layers of human neocortex (DeFelipe 1997). Moreover, pyramidal cells from upper cortical layers form corticocortical connections, important for development of human-specific higher brain functions (Hill and Walsh 2005; Jones 2009; Rakic 2009).
In rodents, dorsally derived interneurons were reported to be a small population of mostly calretinin+ interneurons that populate olfactory bulbs (Kohwi et al. 2007). However, a population of neonatal GABAergic cells born in the cortical SVZ in the mouse has been reported to migrate not only to the olfactory bulbs but also to various cortical areas (Inta et al. 2008). The timing of this population (first 3 postnatal weeks) corresponds well with late neurogenesis around midterm in human. It seems that neurons which represent significant populations in humans are also present in rodents but are considerably sparser. Additionally, large primate brain with long developmental window has an advantage for studying developmental events that may take very short time in smaller experimental animals.
Our experiments suggest that development of cortical interneurons in humans might be even more complex than previously conceived. In the human cortical SVZ, there are several candidate interneuronal progenitors, including Nkx2.1+, Mash1+, and calretinin+ cells. We present some of these progenitors on a schematic drawing and suggest their possible migratory routes during early and late development (Fig. 8). It would be important to further elucidate how this diversity of progenitors during development translates into interneuronal diversity in the adult cerebral cortex.
Another very interesting observation of this study concerns the presence of both Mash1+ cells that coexpress interneuronal markers (GABA), as well as those that express PDGFRα, a marker of oligodendrocyte progenitor cells, in the dorsal cerebral cortex. We have previously shown that another transcription factor, Olig2, expressed abundantly in the human cortical SVZ by oligodendrocyte precursor cells, was not expressed in GABAergic neurons (Jakovcevski and Zecevic 2005a). Since both interneurons and oligodendrocytes have ventral telencephalic origins, it has been hypothesized that in rodents, they may have a common progenitor in the GEs (He et al. 2001). Our study is the first to suggest that in human brain this lineage relationship may be present also in the dorsal neocortical areas. The conspicuous similarities between interneurons and oligodendrocytes in function (modulation of the principal neuron's output) and origins (ventral origin in early development, later appearance of dorsally derived populations) were previously noticed and discussed (He et al. 2001; Kim et al. 2007).
Our conclusion is, thus, that great complexity of cortical interneurons in humans emerges from great complexity of progenitors during cortical development. Careful morphological analysis of human fetal tissue could provide important data essential in our efforts to understand normal functioning and pathological processes in human brain. Further research is warranted to clarify many unanswered questions, including the exact timing and precise anatomical localization of ontogenic events that lead to development of great diversity of human cortical interneurons.
National Institutes of Health and Connecticut Innovation Stem Cell Grant 2008-013. (NS041489-09 to N.Z.).
Authors would like to acknowledge Dr Stewart Anderson from Weill Cornell Medical College for critical reading of an earlier version of this manuscript, and Dr Jens Grosche from Paul-Flechsig Institute for Brain Research, Leipzig, Germany, for graphic design of figure 8. We thank Dr Brad Poulos, Tissue Repository at Albert Einstein College of Medicine for providing human fetal brain tissue for our research. Conflict of Interest: None declared.