Cortical and Clonal Contribution of Tbr2 Expressing Progenitors in the Developing Mouse Brain

1Intermediate progenitor cells (IPCs) are transient transit-amplifying progenitors present during neocortical neurogenesis that arise by asymmetric divisions of radial glial progenitors. These progenitors reside in the subventricular zone (SVZ) where the majority terminally divide to produce a pair of neurons. Additionally, they differ from radial glial progenitors in lacking apico-basal polarity and largely undergoing terminal divisions.

2Intermediate progenitors are defined by expression of the long noncoding RNA Svet1 and by transcription factors such as T-box brain-2 (Tbr2) and Cux2. Both of these transcription factors are known to play significant role in IPC biology and mice deficient for Tbr2 and Cux2 display defects in IPC specification and subsequent generation of neurons.

3Evolutionarily, IPCs are thought to be instrumental in the expansion of the neocortical thickness by amplifying the neuronal output from radial glia. It has been shown that nonventricular surface divisions increase at late embryonic stages, during the formation of upper-layers, therefore relating IPC-derived neurons to supragranular layers. Previous fate-mapping using a Neurod6 (NEX) driven Cre mouse line has shown IPC contribution towards the formation of upper layers. This is further supported by the large decrease in the upper cortical layer thickness and Cux1+ cells in Tbr2 conditionally deleted mice. In macaque fetal cortices, during periods of infragranular layer generation, nearly 30% of tritiated thymidine is incorporated into nonventricular zone progenitors (inner/outer SVZ). With recent studies showing large numbers of IPCs in both these regions, it suggests that IPCs could also be involved in formation of infragranular layers, at least in larger brains.

4In mice, using a Tbr2::GFP (BAC) transgenic line, it has been shown that a significant Tbr2+ population exists at E10–E11 suggesting that IPCs might also contribute towards genesis of infragranual layers. Since impermanent labeling in Tbr2::GFP brains precludes studying the final laminar positions of the cells derived early from Tbr2+ IPCs, the ultimate fate, distribution and proportions of all Tbr2+ progenitors currently remains elusive.

5A quantitative understanding of the neuronal output from IPCs towards cortical neurogenesis is thus at present absent and impeding our understanding of neuronal cell number regulation during development and evolution of the forebrain. Moreover, no previous study has examined the clonal generation of neurons from individual Tbr2+ IPCs. Further, while neural progenitors across species share some characteristics, their diversity in primate cortices has only recently been described. A quantitative comparison of rodent and primate progenitors will hence lead to insights into the evolution of the neocortex.

6We thus investigate IPC dependent lineages using the Tbr2Cre and Ai9-tdTomato mouse lines. In addition, we use a novel clonal analysis method: CLoNe that is based on stable integration of Cre-inducible fluorophores by transposases using in utero electroporation to clonally follow IPC contribution towards the cortical layers.

7All animal experiments were conducted in accordance with the UK Animals (Scientific Procedures) Act (1986). C57BL6/J background mice were used throughout the study. The date of vaginal plug was observed at E0.5 and embryos accordingly timed.

8Tbr2Cre mice were bred with Ai9-tdTomato reporter mice to label all neurons derived from the Tbr2+ IPCs. Pups born from this mating were perfused at P7 (N = 3) and P21 (N = 4) and mice were used for cell counts and immunostaining.

9For layer-wise comparison in the Tbr2flox/flox, Sox1Cre mice were crossed with Tbr2flox/flox and animals with deletions in both alleles were selected. Pups (N = 3) were perfused at P8.

10Images were taken at the level of the S1 cortex.

11Probes were prepared by ligating into pGEMT (Promega), the PCR product from the following primers on a known Cre expressing plasmid.

12Forward (5′-3′): GCCGCCACCATGGCCAATTT

13Reverse (5′-3′): GCGGCCGCTATCACAGATCT

14Standard ISH protocols were used as described in ). Briefly, DIG-labeled riboprobes were synthesized using T7 and SP6 RNA polymerase according to manufacturer’s protocols (Roche). Tissue was fixed with 4% paraformaldehyde (PFA) (Sigma Aldrich) for 4 h, cryoprotected with 30% sucrose overnight and frozen in OCT (TissueTek). Sections of 16 μm thickness were cut on a cryostat (Jung CM3000; Leica) and stored at −80° until use.

15Sections were postfixed with 4% PFA for 5 min, deproteinized with 0.1 N HCl for 5 min and acetylated with acetic anhydride (0.25% in 0.1 M triethanolamine hydrochloride). Prehybridization was carried out at room temperature (RT) for at least 1 h in a solution of 50% formamide, 200 μg/mL E. coli transfer RNA, 1× Denhardt's solution, 10% dextran sulfate, 10 mM Tris (pH 7.6), 600 mM NaCl, 0.25% sodium dodecyl sulfate, and 1 mM ethylenediaminetetra-acetic acid (EDTA). Hybridizations were carried out overnight at 65°C with riboprobes diluted in the prehybridization solution. Sections were washed in 30 mM NaCl/3 mM sodium citrate first at 65°C for an hour and then at RT for 5 min. Slides were blocked with 1% blocking reagent (Roche) in 100 mM Tris–Cl (pH 7.5) and 150 mM NaCl. This was followed with overnight incubation in anti-DIG antibody (1:500, Horseradish peroxidase conjugated, Roche) at 4°C. Sections were developed using the TSA-plus Cyanine 3 Kit (PerkinElmer) according to the manufacturer’s instructions, counterstained with 4′,6-diamidino-2- phenylindole (DAPI) and mounted with phosphate-buffered saline (PBS).

16Plasmids used for clonal analysis have been described previously The transposase expressing vector (or mPB), was kindly provided by Wellcome Trust Sanger Institute (Yusa et al. 2009) and is required for insertion of the fluorophore plasmids into the genome of progenitor cells.

17In-utero electroporation was performed as reported previously with modifications. Time-mated pregnant dames were administered Vetergesic intraperitoneally (0.08 mg/kg) for analgesia prior to anesthesia with Isoflurane. A midline laparotomy was undertaken and uterine horns exposed gently. DNA (mixed with Fast Green, Sigma) was injected into the lateral ventricles using pulled glass capillaries (Harvard Apparatus). Square pulses (at 30–35 V, 5 × 50 ms with 950 ms resting periods) were delivered using tweezer electrodes (CUY650P3, Nepagene) connected to a square pulse generator (Electro Square Porator, Harvard Apparatus). Warm PBS was used to keep the embryo sacs at near body temperature along with a warm blanket.

18Following electroporation, the uterine horns were gently placed back in the abdominal cavity. Surgical incisions were sutured (Vicryl 4/0, Ethicon) and the skin held together using surgical clips. Animals were allowed to recover after removal of anesthesia and kept on heating pads until they regained movement.

19Drop-fixation of embryonic brains was carried out at 4°C overnight in 4% paraformaldehyde (PFA; Sigma-Aldrich) and subsequently stored in PBS with 0.05% sodium azide (Sigma). For postnatal ages, animals were transcardially perfused first with cold PBS and then with cold 4% PFA before collecting the brains. Brains were sectioned on the Leica VT1000S vibrating microtome at 40–50 μm and collected in PBS. Sections were washed in PBS thrice for 5 min each and permeabilised in PBS–Triton-X (0.1%; PBST) for 30 min. Following blocking with 5% serum (in PBS-T) for 2 h, primary antibody diluted in PBS-T with 1% serum was added and the sections incubated overnight at 4°C. The following day, the sections were again washed before incubation with secondary fluorescent-labeled antibodies (Molecular Probes, Invitrogen) for 2 h at room temperature. Sections were counter-stained with DAPI (Invitrogen) and mounted on slides with PBS and covered with a coverslip. The edges were sealed with clear nail-varnish and allowed to dry.

20Images of Nissl stained sections (40 μm thickness) were taken using a Leica light microscope (Leica DMR 500). Cells were counted in 100 μm widths using NeuroLucida (MBF Bioscience) and Adobe Photoshop CS6 (Adobe Systems Inc). Three sections each from N = 3 brains were analyzed for the cell counts and heterozygotes and homozygotes for the deletion were compared.

21Images were captured using a Zeiss LSM 710 confocal microscope (Carl Zeiss Microimaging). Similar image parameters (gain, pinhole, and wavelengths) were maintained for images from all brains. Z-stacks were taken individually for each channel and then collapsed to get maximum intensity projections. Images were analyzed using ImageJ (Image Analysis in Java, NIH) or Adobe Photoshop CS6 (Adobe Systems Inc.).

22Confocal images of multi-channel fluorescently labeled cells were analyzed using a histogram based multi-level thresholding algorithm on MATLAB software (R2013a; Mathworks).

23Briefly, this algorithm segments an image based on discreet threshold values, computed iteratively using the histogram. The optimum number of thresholds (n) required to segment the image (7–12 usually) is determined based on comparison of the PSNR (peak signal-to-noise ratio) values associated with different thresholding levels.

24Finally, thresholded values were compared with the figure to verify that they showed similarly labeled cells.

25We also used Imaris software (Bitplane) to analyze cell clones. For this, we used multichannel. lsm files to extract mean cell intensity in all 4 fluorescent channels. We then thresholded the intensity into 5 levels of 51 intensity points each. Based on this thresholding, each cell was awarded a cell code and reiteration of each code was checked in the dataset. Finally clonal sizes were rechecked for frequencies to understand distribution.

26In all 3 brains labeled at E12.5 and perfused at P7 was used for this analysis, In each brain, 3 consecutive 150 µm slice was subjected to SeeDB as per published protocol and imaged using confocal microscopy.

27The expression of Tbr2 has been previously described in the developing brain, and specifically in progenitors of the SVZ. In contrast to other SVZ markers, such as the Cux2 gene, Tbr2 is solely expressed in the IPCs while Cux2 expression can be seen in the upper cortical layers of the cerebral cortex and in some fate-restricted populations of radial glial cells, thereby complicating its use as an IPC marker. Thus we chose to use a Tbr2Cre mouse line to fate map the entire progeny of IPCs in the mouse brain.

28We first characterized the expression of the Cre-recombinase in the Tbr2Cre knock-in mouse by fluorescent FISH at E12.5. We found that the Cre-recombinase is selectively expressed in the cortical proliferative zones but not in the thin cortical plate (CP) at this stage (Fig. 1A,A′). We compared Cre mRNA expression with that of Tbr2 protein on an adjacent section and found it to be largely correlative in the SVZ (Fig. 1B,B′). Cells in the Tbr2 negative region of the VZ were also positive for Cre mRNA as shown previously indicating that transcription already occurs during the transition to IPC fate as with the Tbr2 gene.

29In order to label the progeny of IPCs, we employed 2 separate strategies. Firstly, we crossed the Tbr2Cre with the Ai9-tdTomato reporter mice (henceforth referred to as Ai9) to label all cells born from the Tbr2+ lineage across the brain during development. Secondly, we performed in-utero electroporation in Tbr2Cre pregnant mice at 12.5dpc (E12.5) in order to label lineages in a temporal fashion (N = 4 litters, n = 6 brains). To achieve this, we used a cocktail of transposase expressing plasmid and STOP-floxed fluoroprotein expressing plasmids. This cocktail labels clonal progenies of focally electroporated progenitors that pass through a Tbr2+ lineage with the same combination of fluorophores thus enabling their detection (Fig. 1C;).

30We also performed short-term labeling (E12.5–E14.5, N = 2 litters, 4 brains) with only CAG-STOP-EYFP that marks cells exiting the SVZ (Fig. 1D). Importantly, no label was found in the ventricular zone that houses radial glial progenitors displaying the specificity of the labeling. Immunostaining for Tbr2 showed only a small proportion of EYFP+ cells in the SVZ (Fig. 1D) with the majority in the intermediate zone (IZ) and CP. No colocalisation of EYFP+ cells with Tbr2 was observed indicating that expression of the fluorophores occurs during exit from SVZ.

31In order to test for combinatorial and clonal fluorescent labeling of progenies, we electroporated a cocktail containing mT-Sapphire (membrane anchored), GFP (cytoplasmically targeted) and mCherry (nuclear targeted) at E12.5 and collected the brains 2 days later (N = 2 animals, 3 brains). As with the single fluorophore electroporation, most of the cells were positioned above the SVZ with some even sending axons through the IZ towards subcortical targets. We also observed 2 independent pairs of clones exiting the SVZ. The first pair (Fig. 1E; thick arrowheads) was negative for Sapphire, weakly positive for GFP and strongly expressing mCherry while the latter was positive for all 3 (thin arrowheads). Thus, clonal labeling of progenies of IPCs is possible using this method. In addition, the descent of axons from these cells (Fig. 1E,F) showed that a proportion of these cells are subcortically projecting and hence infragranular in nature. To confirm this, we performed immunostaining for Ctip2 and Tbr1 separately in E14.5 Ai9 brains (Fig. 1F–J). Both Ctip2 and Tbr1 colocalised with Ai9+ cells (Fig. 1F–J) above the IZ showing that infragranular layer neurons do arise from Tbr2+ IPCs during early neurogenic periods.

32As neurons continue to migrate well into postnatal stages, we addressed the final positions of neurons generated from all Tbr2+ IPCs in Ai9 brains at postnatal Day 21 (P21, N = 4 animals) after the completion of neuronal migration (Fig. 2A). We found Ai9+ cells across all layers of the cortex though this varied across different regions. Cingulate areas and the lateral cortex had Ai9+ neurons spread across the length of the cortex while in more dorsal regions fewer Ai9+ cells were present in infragranular layers (Fig. 2A,A″). We stained for known layer-specific markers such as Cux1, Otx1, and Nurr1 to confirm cell identities (Fig. 2B–D). Immunostaining for these markers revealed that Ai9+ neurons displayed fate identities in keeping with their positions. Majority of Ai9+ cells in supragranular layers also labeled for Cux1 while a smaller proportion of infragranular layers was positive for Otx1 (Fig. 2B′,C′, arrowheads). Surprisingly, a subset of Nurr1+ subplate (SP) cells was found to be Ai9+ (Fig. 2D′, arrowheads). SP neurons are among the earliest born cells in the cortex and are involved in mediating corticothalamic and thalamocortical innervation.

33Additionally, we performed a quantitative analysis to ascertain the laminar contribution from Tbr2+ IPCs, in Ai9 brains at P21. We observed that Tbr2+ IPCs gave rise to neurons in all layers of the cortex (Fig. 2E) with varying proportions. Layers 2/3 and Layer 4 had the highest proportions of Ai9+ cells with 40.2% ± 6.4 (mean ± SD) and 31.3% ± 5.4, respectively, in the S1 somatosensory cortex. In comparison, Layer 6 and SP had fewer cells with 29% ± 4.9 and 27.2% ± 6. Layer 5 contained the fewest cells born through an IPC with 19.8% ± 4.2. Since Ai9+ cells were counted as percentage of all cells in the cortex (thereby including glial cells and interneurons in the denominator) the percentages of pyramidal glutamatergic neurons derived from Tbr2+ IPCs are under-represented in our observations. It is widely accepted that IPCs contribute towards upper-layer genesis. However, our cell counting showed that while there was a significant difference between the contribution towards supragranular (35.55% ± 3.24) and infragranular layers (25.44% ± 2.84), the latter received a substantial contribution through IPCs. Looking at the cortical thickness as a whole, the contribution from IPCs was found to be 30.03% ± 1.95 in S1 showing that contribution towards cortical layers is more homogeneous than previously thought.

34It was reported that Tbr2 conditional knockout cortices showed a large decrease in the Cux1+ population of neurons with no significant decrease observed for Er81+ and FoxP2+ neurons. However as these markers do not reveal all pyramidal cells in the respective layers, we undertook an unbiased layer-wise cell count (neurons and glia on Nissl stained sections) in the somatosensory cortex of Sox1Cre; Tbr2fl/fl conditional knockouts (Fig. 2F) at P8. Decrease in neuronal cell numbers was observed only for Layer 4 and Layer 6 of P8 cortex (Fig. 2G) with layer 4 showing a large decrease (∼43% of wild-type, P = 0.009, Multiple t-test with Sidak correction) and Layer 6 a relatively lesser reduction (14.5%, P = 0.08). None of the other layers showed an appreciable decrease except for a minor increase in the Layer 5 neuronal cell count (∼5%, P = 0.068). Hence the decrease in supragranular layers seems to stem from a decrease in Layer 4 only.

35We stained for glial fibrillary acidic protein (GFAP) and gamma-amino butyric acid (GABA) in the P7 cortex to rule out glial and interneuron fate of Ai9+ cells (Fig. 2H–L). Expression of GFAP was largely limited to the white matter with some expression in Layers 1 and 5 (Fig. 2H). Some overlap of Ai9+ cells with GFAP+ processes (Fig. 2I–K) in the corpus callosum and rostral migratory stream could be seen which is in keeping with previous observations of Tbr2+ neuroblasts differentiating into glutamatergic interneurons in the olfactory bulb. On the other hand, GABA immunoreactivity did not overlap with Ai9+ cells in the cerebral cortex (Fig. 2L, a total of 12 sections were studied from 3 brains) maintaining that cells generated through the Tbr2+ lineage were glutamatergic as previously asserted.

36To further validate our observations, we also performed fate mapping by in utero electroporation of a Cre inducible fluorophore in Tbr2Cre cortex. For this, a mix of CAG-STOP-EYFP and mPB (transposase) plasmids was electroporated into the dorsal cortex at E12.5 and E15.5 followed by analysis at P8. When electroporation was performed at E12.5 (N = 3 brains each at P8), the resultant neurons generated were situated throughout the thickness of the cortex (Fig. 2H). Majority of cells were present in Layers 2–6, correlating with their birthdates being at the time of electroporation or afterwards. This also shows that the fluorophore cassettes were integrated in the radial glial progenitor genome. Cells were also observed in Layer 6 as expected (Fig. 2M) but fewer in number. In contrast pups collected from electroporations at E15.5 had labeled cells only in Layers 2–4 (Fig. 2N) and a strong band of fiber labeling could be observed in Layer 5; a known output target for Layer 2–3 neurons.

37As shown in Figure 2A, we observed consistent areal variations in Ai9+ cells in cortical layers throughout the cortex. This variation seemed to stem from differences in number of labeled cells in infragranular layers as dense labeling could be observed in the supragranular layers (Fig. 5A).

38We decided to investigate this further at the level of the presumptive visual cortex (Fig. 3A) by counting the number of Ai9+ cells in each layer between the dorsal-medial (Fig. 3A′) and ventrolateral regions (Fig. 3A″).

39In doing so, we found that while approximately equal Ai9+ cells could be seen in Layers 2/3 (58.33 ± 4.6 versus 46.67 ± 4.5, n = 2 brains, P = 0.12 Multiple t-test with correction), there was an appreciable difference in Layer 4 (66.67 ± 5.2 versus 46.51 ± 3.9, n = 2, P = 0.048) and in Layer 5 (12.50 ± 3.5 versus 38.24 ± 2.7, n = 2 brains, P = 0.014). No difference was noticed when all layers were grouped together suggesting against differences in cell densities (35.1 versus 35.62) (Fig. 3B).

40These results suggest a difference in cell-cycle kinetics between these closely located regions also manifesting as reciprocal differences between Layers 4 and 5.

41We were also interested in the clonal output from individual IPCs to the different cortical layers. For this we performed in utero electroporation as described above with the inclusion of plasmids carrying different fluorophores (mCherry, EYFP, EGFP and mSapphire; ) targeted to different cellular locations (cytoplasmic, membrane, and nuclear) in combination with transposase at E12.5 and E15.5 and collected at P8 for analysis.

42Clones were analyzed by histogram based thresholding algorithm (Fig. 4A). Images generated above (Fig. 4B,C) were analyzed and distances between pixels holding similar values (representing clonally related cells) measured. We found that related cells in the same layer were separated by a mean distance of 36.8 ± 20.5 μm (n = 23 cell pairs across 4 brains) while those in different layers had a separation of 142.1 ± 76.8 μm (n = 19 cell pairs across 3 brains) (Fig. 4D). Additionally, clonally related cells separated by cortical layers were clustered closer than unrelated ones (142.1 ± 76.8 μm vs. 294.9 ± 105.4 μm; P < 0.0001) (Fig. 4D). However, this did not apply to clonal clusters within the same layer or to the sum of the 2. As clonally born neurons are known to disperse long distances, we decided to use tissue clearing methods such as SeeDB to image and analyze labeled neurons in serial section each of 100–150 μm thickness. Our analysis on brains electroporated at E12.5 revealed that depending on the number of cells labeled (between 200 and 400 per brain), the distinct labeling patterns generated varied between 20 and 30 (data not shown). We observed that the larger clone sizes had cells distributed between more than one radial column suggesting that IPCs can give rise to neurons in adjacent radial columns (Fig. 4E).

43We also saw that clone sizes of >8 were generated more often that those between 1 and 8 (Fig. 4F) by almost a factor of 2 (19.7 vs. 10.8, data from 3 different brains at P7 and 200–400 cells per brain, t-test with Sidak correction). Further, frequencies of clonal size of >16 showed the highest representation compared with other sizes. As these clonal sizes can be achieved only when both early and late born IPCs are labeled, this further demonstrates the fidelity of our method. Additionally, it also shows that more IPCs are born during early cortical development (thus giving larger clonal sizes). We also found that not all combinations of an arithmetic progression were represented in the clonal distribution and most values between 9 and 16 (except for 11, 12, and 16) were absent in our analysis.

44SP cells have been described as a transient neuronal population that undergoes programmed cell death during late embryonic and early postnatal development. However, a significant proportion of these cells do survive and are labeled by a variety of immunohistochemical markers. Some studies have postulated that the acetylcholine esterase expressing interstitial white matter cells might be the surviving vestiges of the SP neurons.

45To examine if Tbr2 derived SP cells remain as interstitial cells, we chose to look for AChE+ tdTomato+ cells in the SP and white matter at P21 and P7 by using an antibody against ChAT. However, at P21, we could find no ChAT+ cells in the SP/WM region (n = 3 brains) despite ChAT+ cells being present in the globus pallidus (Fig. 5A,B). In order to see whether ChAT+ cells are present during early postnatal ages, we also looked at P7 cortices. Similar to P21, no ChAT+ cells were seen at this age either despite their presence in the ventrobasal areas of the forebrain (Fig. 5C,D).

46As the SP has been described to be a heterogeneous population, we performed immunostaining for other SP markers. We found that Tbr2 derived SP cells were not positive for the interneuron marker parvalbumin (Fig. 5E) but expressed other SP markers such as CTGF and Nurr1 (Fig. 5F,G).

47In addition to our fate mapping in the cortex, we also observed Tbr2 expression in previously undescribed regions. We observed Tbr2 immunoreactive cells in the embryonic septum and lateral olfactory tract (lot) (Fig. 6A) as well as in a small population of migratory-shaped cells along the margins of the rostral medial telencephalic wall proceeding ventrally towards the septum. A similar migration has been previously described though the nature of the progenitors was not commented on. We also observed a large number of cells in the ventro-medial cortex and septum at E14.5 in coronal sections of Ai9 brains (Fig. 6B) a large proportion of which colocalised with Tbr1 (Fig. 6B′,B″; arrowhead) indicating that they are glutamatergic in nature. Ai9+ cells in the lot were also positive for calbindin (Fig. 6D inset) as reported previously.

48Ai9 labeled cells were also observed at thalamic eminence and the prethalamus at E14.5 and appeared to cross the diencephalon–telencephalon boundary (DTB) seemingly emanating from the ventricular zone of the third ventricle (Fig. 6C,D). It has been reported that cells from this region migrate and contribute towards the posterior accessory olfactory bulb (pAOB) and the medial amygdaloid nucleus. Thus, it seems that Tbr2+ cells contribute towards the development of the vomeronasal system both by generating neurons that set up the lot and also by contributing towards the stream of neurons migrating from the prethalamus towards the pAOB.

49We also examined sections from postnatal Ai9 brains (P7). As expected, we found labeled cells in the lot and the olfactory tubercle (Fig. 6E, inset).

50Our analysis of labeled cells also revealed Tbr2+ derived cells in other regions. We observed expression of Ai9 in the hypothalamus and prethalamus (Fig. 6E′,E″). Ai9+ cells in the hypothalamus however failed to colocalise with either tyrosine hydroxylase (TH) or choline acetyltransferase (ChAT; data not shown) indicating that they were generated from a different progenitor pool and were not neurons of the basal forebrain. In addition, Ai9 staining could also be seen in the cells of indusium griseum, bed nucleus of anterior commissure (BAC) and the anterior commissure itself (Fig. 6F–H).

51Our results show that in contrast to accepted paradigms, Tbr2+ IPCs contribute towards all layers of the cortex including the earliest born SP layer. As IPCs are an acquired progenitor group during cortical evolution, it suggests that the expansion of the cortex has been more homogeneous among all layers than previously thought. These observations are further supported from our current quantitative analysis of the somatosensory cortex of Sox1Cre; Tbr2fl/fl conditional knockouts where we describe reduction in cell numbers in all layers. Our study provides the first analysis on the spatial relation between clonally derived neurons from IPCs and clonal size. Accordingly, clonally related cells positioned in different layers cluster closer than unrelated cells. Finally, our study identified extracortical contributions of Tbr2+ expressing progenitors that are involved in the generation of glutamatergic neurons of the lateral olfactory tract, the olfactory tubercle, and ventromedial hypothalamic nucleus.

52The contribution from Tbr2+ progenitors towards the cerebral cortical layers has not been studied directly, quantitatively and systematically. While its origin has been well described, the fate of its progenies has not been followed into later postnatal stages. Ambiguity persists regarding the contribution towards the upper and lower layers. A study utilizing Tbr2 conditional knockout describes a decrease of immunoreactivity for an upper layer marker while another report describes a loss in both upper and lower layers. In the latter case however, haploinsufficiency at the Foxg1 locus is known to decrease the specification of IPCs by ∼38% and is associated with a decrease in supragranular layer neurons. Thus use of this Cre line would lead to erroneous interpretation. Using Tbr2::GFP, it has been reported that early born IPCs contribute towards infragranular layers. However, this is an indicative assumption as the GFP label is impermanent and laminar fates of neurons cannot be determined using this transgenic.

53In contrast, transgenics with a single copy of Tbr2 develop normally and do not show reduced expression. Thus the Tbr2Cre mouse line is better suited for fate-mapping of IPCs as shown by us using immunohistochemistry and reporter mouse lines. Using this line, we show that Tbr2+ IPCs give rise to both upper and lower layer neurons but in different proportions. Importantly, we describe that IPCs give rise to a subpopulation of SP neurons proving important during early corticogenesis and circuit formation.

54Our study also demonstrated that Tbr2+ IPCs do not produce GABAergic or astrocytic cells. It is reported that the rodent somatosensory cortex contains ∼36% glia and ∼64% neurons. Further, among neurons, ∼25% is constituted by GABAergic interneurons with excitatory cells making the majority of cells. In simpler terms, for every 100 cells counted in the somatosensory cortex of rodents, 64 cells are neurons of which 16 cells would be interneurons. Considering from results shown above that Tbr2+ IPC derived neurons account for 30% of all cortical cells, it implies that ∼62.5% of excitatory neurons (30/48) are produced via Tbr2+ IPCs with the number being even higher in upper layers and remaining constant for lower layers (exact calculations are difficult since astrocytes and oligodendrocytes show layer specific clustering). This number differs from the predictive studies that use pH3+ cell numbers in the SVZ to calculate the contribution from IPCs perhaps due to difference in the region used for analysis. Our study hence does not support the idea for an exclusive path for the generation of glutamergic neurons but instead suggests most glutamatergic neurons to be born via a RGC to IPC to neuron pathway.

55Our layer-wise cell counts from the conditional knockout as well as from the Ai9 brains indicates that while neurons born from IPCs come to lie across all cortical layers, the decrease in the conditional knockout is only seen in Layers 4 and 6. This suggests redundancy in Tbr2 function observed in relation to other layers (than L4) during cortical neurogenesis. Single cell profiling has described gene clusters that are expressed in the SVZ and might be involved in specifying neuronal fate. In addition, Tbx1, another T-box gene that has been described to affect proliferation of IPCs is correlated with DiGeorge Syndrome leading to a decrease in supragranular layer neurons. Such clusters and paralogs might rescue some of the deleterious effects of Tbr2 gene knockout. This problem might be magnified in animals with larger brains having a larger Tbr2+ population.

56Analysis of clonally generated cells from Tbr2+ IPCs shows that they do not disperse long distances but are in close proximity with one another. Also, we show that clonal clusters are located within limits that suggest their arrangements in columns as previously described. We also show for the first time that IPCs can generate neurons towards adjacent radial columns despite coming from the same RGC pool. It will be interesting to further study whether the clonally related cells in our study are also functionally related.

57The regulation of cortical patterning and arealisation is attributed to multiple factors. Studies have shown an inherent difference in cell-cycle kinetics between different regions at embryonic ages that lead to differences in cortical numbers. This has further been supported by artificial modification of cell-cycle parameters leading to changes in cortical numbers and also by genetic and transcriptomic studies. In addition, modulation of thalamic input has also been shown to influence cortical parcellation. However, the contribution of specific progenitor cells in areal cortical number differences has not been looked into in detail.

58Using the fate mapping approach, we show a difference in cortical numbers in specific laminae between dorsomedial and adjoining ventrolateral cortex at the level of both somatosensory and visual cortices indicating that differential IPC division between adjoining areas might lead to areal differences in the mature cortex.

59This study also describes expression of Tbr2 in regions outside of the cerebral cortex namely in the prethalamus, embryonic septum, and hypothalamus. The septal expression is particularly interesting as it aides in the development of the lot, which has been shown to be affected in the conditional Tbr2 knockout thus arguing for a role of these cells in extracortical regions. We also show expression in the prethalamus during embryonic development. A case can hence be made for the involvement of Tbr2+ IPCs in specification and development of these regions and ultimately to the development of the brain. However, it remains to be seen as to whether the transcriptional regulation in these regions is similar to that studied in the SVZ.

60Finally, we show expression in the bed nucleus of anterior commissure and the indusium griseum, regions that are closely located with major fiber tracts of the brain, i.e., the anterior commissure and corpus callosum both of which are absent in the conditional Tbr2 knockout mouse and in human with mutations in the TBR2 gene.

61Intermediate progenitors have been hypothesized to be an important addition in the evolutionary expansion of the cerebral cortex. It is also known that a mutation spanning the TBR2 locus in humans causes microcephaly with polymicrogyria. However, from mouse models of loss of function of Tbr2 it is not clear as to which populations of cortical neurons are affected. In this study, we have shown the pan-cortical contribution from IPCs and to several other regions crucial for the development of the rodent brain. The evolutionary impact of IPCs will thus have to be revisited to study if the expansion is limited to the neocortex or if it applies to these other regions of expression as well. At a clonal level, it would again be interesting to compare differences among mammals with an expanded germinal zone as well as with avian and reptilian counterparts. However, this might need designing of new experimental strategies.

62The study was designed by N.A.V., F.G.-M., and Z.M. N.A.V. performed the fate-mapping and clonal analysis experiments. F.G.-M. performed some of the clonal analysis experiments and established the method used for Figure 3. S.A. wrote the algorithm for image segmentation. A.C. performed and analyzed the Tbr2cKO cell counting. Tbr2Cre mouse strains and Tbr2-deficient tissues were generated and provided by S.J.A. and E.J.R. N.A.V., F.G.-M., and Z.M. wrote the paper.

63This study was funded by Medical Research Council, UK and Biotechnology and Biological Sciences Research Council, UK grants to Z.M. S.J.A. was supported by the German Research Foundation (DFG; Emmy Noether Programme AR 732/1-1). F.G.-M. is funded by Human Frontiers Science Program and N.A.V. is funded by Goodger and Felix Scholarships. Funding to pay the Open Access publication charges for this article was provided by University of Oxford RCUK Access Block Grant.