Nkx2.1 regulates the generation of telencephalic astrocytes during embryonic development

The homeodomain transcription factor Nkx2.1 (NK2 homeobox 1) controls cell differentiation of telencephalic GABAergic interneurons and oligodendrocytes. Here we show that Nkx2.1 also regulates astrogliogenesis of the telencephalon from embryonic day (E) 14.5 to E16.5. Moreover we identify the different mechanisms by which Nkx2.1 controls the telencephalic astrogliogenesis. In Nkx2.1 knockout (Nkx2.1−/−) mice a drastic loss of astrocytes is observed that is not related to cell death. Further, in vivo analysis using BrdU incorporation reveals that Nkx2.1 affects the proliferation of the ventral neural stem cells that generate early astrocytes. Also, in vitro neurosphere assays showed reduced generation of astroglia upon loss of Nkx2.1, which could be due to decreased precursor proliferation and possibly defects in glial specification/differentiation. Chromatin immunoprecipitation analysis and in vitro co-transfection studies with an Nkx2.1-expressing plasmid indicate that Nkx2.1 binds to the promoter of glial fibrillary acidic protein (GFAP), primarily expressed in astrocytes, to regulate its expression. Hence, Nkx2.1 controls astroglial production spatiotemporally in embryos by regulating proliferation of the contributing Nkx2.1-positive precursors.

Scientific RepoRts | 7:43093 | DOI: 10.1038/srep43093 from progenitor cells in the dorsolateral subventricular zone (SVZ) 32,33 . Recent evidence, however, shows that a population of locally differentiated glia in the postnatal cortex rather constitute the primary source of astrocytes 34 .
Nkx2.1, a homeodomain transcription factor, was initially found to regulate the transcription of many thyroid [35][36][37] and lung-specific genes 38,39 . Interestingly, Nkx2.1 regulates several cell cycle related genes such as Notch1, E2f3, Cyclin B1, Cyclin B2 and c-Met in developing embryonic lungs 40 . In the embryonic brain, Nkx2.1 controls the specification of the GABAergic interneurons and oligodendrocytes that populate the ventral and dorsal telencephalic region 3,16,37,[41][42][43] . Loss of Nkx2.1 leads to ventral-to-dorsal re-specification of the pallium causing loss of GABAergic interneurons and oligodendrocytes in the dorsal telencephalic region 16,17,37 . Recently, we showed that Nkx2.1 regulates the generation of astrocytes that populate the ventral telencephalon during embryonic development and participates in axonal guidance in the anterior commissure 18,44 . We found that this Nkx2.1-derived astrocyte population is generated from three ventral telencephalic precursor regions, namely the medial ganglionic eminence (MGE), the anterior entopeduncular area (AEP)/preoptic area (POA) and the triangular septal nucleus (TS) 18,44 . Several works have revealed information on the origin of embryonic NG2 glia, also known as polydendrocytes or oligodendrocyte precursor cells 45,46 . Our work contributed by showing that embryonic NG2 glia originate from the Nkx2.1 + progenitors of the ventral telencephalon and promote precise blood vessel network development 44 . These cells have a highly complex branched morphology and are different from neurons, mature oligodendrocytes, astrocytes and microglia [45][46][47][48][49][50][51][52] . For a clear distinction, these cells will be referred to as "NG2 glia" in this study.
Here, we describe that Nkx2.1-derived astrocytes populate the corpus callosum (CC) and its surrounding regions in the embryonic dorsal telencephalon. These Nkx2.1-derived astrocytes are generated from E12.5 onwards with maximal production between E14.5 to E16.5. Interestingly, in Nkx2.1 −/− mice, we showed that the mutated Nkx2.1 (mut-Nkx2.1) leads to a drastic loss of astrocytes and NG2 glia in the dorsal telencephalic region at the midline. Since this mut-Nkx2.1 derived cell loss is not accompanied by increased cell death, we analyzed if cell proliferation of Nkx2.1 + progenitors in the three ventral precursor regions (MGE, AEP/POA and TS) was affected. In vivo BrdU incorporation showed that interestingly Nkx2.1 regulates astroglial generation by controlling the proliferation of Nkx2.1 + precursors. Similarly, in vitro neurosphere differentiation assays showed decreased generation of astroglia from Nkx2.1 −/− precursors, which may be a result of decreased proliferation of astroglial precursors upon loss of Nkx2.1 or perhaps defects in glial specification and/or differentiation. In addition, chromatin immunoprecipitation analysis showed that Nkx2.1 binds to the promoter of the glial fibrillary acidic protein (GFAP), primarily expressed in radial glia and astroglia. Co-transfection studies in HEK293 cells using tagged over-expressed Nkx2.1 and a mouse GFAP promoter construct confirmed that Nkx2.1 binds to the GFAP promoter and regulates expression of the GFAP gene. Thus, Nkx2.1 exhibits multilevel control over the generation of the dorsal telencephalic astroglia by spatially coordinating astroglial generation from three ventral precursor regions and temporally restricting maximal generation to E14.5 to E16.5. Further analysis into the complete repertoire of genes regulated by Nkx2.1 will shed light on the exact mode of action and help to delineate the different mechanisms involved in astrogliogenesis.
To identify the GABAergic interneurons, we used Gad1 (Glutamate decarboxylase 1; also called as GAD67)-EGFP knock-in mice that express the enhanced green fluorescent protein (EGFP) in GAD67 + GABAergic interneurons 53 . In combination with Nkx2.1, as with previous observations, we show down-regulation of Nkx2.1 expression in dorsal telencephalic GABAergic interneuron population 54 at E16.5, and also observe that none of the Gad1-GFP + interneurons of the CC and dorsal surrounding areas express Nkx2.1 (n = 4; Fig. 1a,b; solid arrowheads in 1b).
Further analyses using tamoxifen-inducible GLAST-Cre ERT TM /Rosa26-EYFP mice displayed the presence of many EYFP + early astrocytes outside of the germinal zones and also within the CC and surrounding area from E16.5 to E18.5 (n = 5; Fig. 1e-j). Many of these early astrocytes visualized by EYFP and GLAST co-staining showed Nkx2.1 expression also (n = 5; CC in Fig. 1f,g and MGE in 1i-j, solid arrowheads). (a-d) Double immunohistochemistry on coronal CC sections from Gad1-EGFP + mice at E16.5 for GFP and Nkx2.1 (a,b) (n = 4), and on coronal CC sections from wild-type mice at E16.5 for GLAST and Nkx2.1 (c,d) (n = 2). (e-j) Triple immunohistochemistry for EYFP, Nkx2.1 and GLAST on coronal CC (e-g) and MGE (h-j) sections from GLAST-Cre:ERT2 + /Rosa-EYFP mice at E18.5 (n = 5). Colocalization in the green and the red channels is yellow (b,c,d,f,i and j). (b), (d), (f), (g), (i), and (j) are higher magnifications of the CC and MGE region indicated by an arrow in (a), (c), (e) and (h), respectively. Cell nuclei were counterstained in blue with Hoechst (a,e, and h). (a-d) At E16.5, several Nkx2.1 + (red) nuclei were observed in the medial part of the CC (open arrowheads in b). Nkx2.1 did not label the Gad1-EGFP + interneurons (green) populating this region (solid arrowhead in b). At this age, however, most of the Nkx2.1-expressing nuclei co-expressed astroglial markers like GLAST (solid arrowheads in c and d).
The results indicate that Nkx2.1 expression is primarily maintained in the astrocyte population of the CC region from E14.5 to E18.5.
To investigate if cell death is the reason for the marked decrease in the number of glial cells observed, we analyzed the control (n = 4 for CC region; n = 5 for POA region) and Nkx2.1 −/− (n = 6 for CC region; n = 10 for POA region) brains at E16.5 for presence of cleaved-caspase 3, a key biomarker of apoptosis (Supp. Fig. S4a-d and S4i) as well as terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) assay to detect DNA fragmentation that results from different cell death processes (n = 16 for CC in controls, n = 22 for CC in knockouts; n = 6 for POA in controls, n = 5 for POA in knockouts; n = 10 for MGE in controls, n = 11 for MGE in knockouts; n = 7 for SEP in controls, n = 14 for SEP in knockouts; Supp. Fig. S4e-h and S4j). Hoechst was also used to visualize pyknotic nuclei. Quantification of dying cells labeled by cleaved-caspase 3 (Supp. Fig. S4i), and by TUNEL staining (Supp. Fig. S4j), revealed no significant differences between the Nkx2.1 −/− brains and the control brains in any of the telencephalic regions tested namely the CC, MGE, POA or septum (p-value = 0.1225 for CC and 0.4618 for POA with cleaved caspase 3 staining; p-value = 0.7934 for CC, 0.8193 for POA, 0.4032 for MGE, and 0.4879 for SEP with TUNEL assay). The size and morphology of cell nuclei were comparable in both control and mutant brains.
In conclusion, glial cells occupying the CC are under the regulation of Nkx2.1. Moreover, the significant loss of astrocytes and NG2 glia in the Nkx2.1 −/− telencephalon is not due to glial cell death.
In the VZ, SVZ and mantle zone of the control MGE, many GLAST + precursors and differentiated astroglia co-express Nkx2.1 (Fig. 6a-c and h, solid arrowheads) whereas in the germinal and mantle zones of the mutant MGE* in Nkx2.1 −/− mice very few GLAST + precursors and astroglia showed co-localization with the mut-Nkx2.1 ( Fig. 6d-f, solid arrowheads). This difference may be attributed as shown previously, to the fact that although a MGE-like structure forms in the mutant (MGE*), it has been re-specified to a more dorsal LGE (lateral ganglionic eminence)-like fate 38 . A similar reduction in cells containing GLAST and mut-Nkx2.1 were seen in the mutant POA* of Nkx2.1 −/− mice (Fig. 6i,j). Quantitative analysis revealed a very large and significant decrease of the total number of precursors (50 to 85%) and specifically GLAST + precursors (45 to 86%) expressing mut-Nkx2.1 in the VZ, SVZ of the MGE*, POA* and TS* (p-value < 0.0001 in the VZ of MGE, POA and TS in Fig. 6k,l and p-value = 0.0139 for the SVZ of MGE in Fig. 6k). Consequently, the number of GLAST + differentiated astrocytes co-expressing mut-Nkx2.1 in the parenchyma (striatum (ST); lateral preoptic area (LPOA)/Lateral hypothalamus (LH); septum (SEP)) in Nkx2.1 −/− mice (n = 4) was severely decreased (60 to 80%) compared to control mice (n = 4) (p-value < 0.0001 for striatum, LPOA and septum in Fig. 6l). Therefore, mutation of Nkx2.1 results in the severe loss of precursors and differentiated astrocytes in the brains.
Nkx2.1 directly regulates the expression of GFAP. From our results, astroglial cell populations of the embryonic telencephalon are derived from Nkx2.1 + progenitors and Nkx2.1 regulates astrocyte precursor cell proliferation. We looked if the transcription factor Nkx2.1 binds to the promoter of the GFAP gene expressed in some brain astroglia and regulates its expression. Chromatin immunoprecipitation assays were performed on lysates of E16.5 embryonic brains. First we looked for DNA elements in the GFAP gene promoter and of a negative control gene, Neurog2 that regulates dorsal precursors matching the consensus NK2 family binding sequence [GNNCACT(T/C)AAGT(A/G)(G/C)TT] 35,57 . Since a complete consensus binding sequence was not found, the core binding sequence T(C/T)AAG was chosen for analysis. As positive control, we used Lhx6 (LIM homeodomain gene) for which the binding site of Nkx2.1 is already known 58 . Then, after shortlisting the position of putative Nkx2.1 binding sites, primers for the flanking sequences of all shortlisted binding sites (up to three) were made (see methods and Supp. Fig. S5A). We performed PCR on cross-linked and sonicated DNA pulled down using an anti-Nkx2.1 monoclonal antibody. The amplification of a putative Nkx2.1 binding sequence located within the GFAP gene (Fig. 9a) was found in the chromatin immunoprecipitated with the Nkx2.1 antibody, however, no positive interaction was detected for Neurog2 (Fig. 9c). Amplification of the already known Nkx2.1 binding site within the Lhx6 promoter was positive after immunoprecipitation with the anti-Nkx2.1 antibody (Fig. 9b). Furthermore, the PCR fragment(s) amplified in the GFAP promoter contained the core sequence CTCAAGT of the Nkx2.1 consensus binding sequence. This suggests that in vivo, Nkx2.1 binds to the promoter region of the astroglial GFAP regulatory gene at the highly conserved core-binding sequence of the consensus-binding site.
To investigate the influence of the binding of Nkx2.1 to this promoter sequence on the transcription of GFAP gene, we performed co-transfection studies in HEK293 cells. We used the expression plasmid pDRIVE-mGFAP, which contains the LacZ reporter under the control of the upstream − 1679 bp mouse GFAP promoter sequence (that includes the putative Nkx2.1 binding site; Supp. Fig. S5B). This was co-transfected with the pCAG-Nkx2.1-IRES-Tomato plasmid that constitutively over-expresses Nkx2.1 and Tomato proteins under the control of the pCAG promoter. Co-transfection in HEK293 cells resulted in robust expression of the LacZ reporter (98% of Tomato + cells were LacZ + , n = 50; Fig. 9g-i). Conversely, almost no expression was seen with co-transfection of the pDRIVE-mGFAP plasmid with a control pCAG-IRES-Tomato plasmid lacking the Nkx2.1 cDNA (only 4.26% of Tomato + cells were LacZ + , n = 94; Fig. 9d-f). This suggests that activation of the GFAP promoter fragment requires the presence of Nkx2.1, which probably acts at the binding sequence we identified.
Overall, the results show that the Nkx2.1 homeobox gene regulates proliferation of astroglia that occupy the CC region during late embryonic stages.

Discussion
Nkx2.1 is implicated in the specification of GABAergic interneurons and oligodendrocytes that occupy the embryonic telecephalon 3,16,37,41-43 . Recently, our group has shown that Nkx2.1 controls the production of the GABAergic interneurons, astrocytes and NG2 glia that populate the embryonic ventral telencephalon 18,44 . In this study, we now show that Nkx2.1 regulates the generation of the dorsal astroglia that populate the corpus callosum and its surrounding regions at late embryonic stages. Nkx2.1 controls astroglia by regulation of the proliferative capacity of Nkx2.1 + precursors present in the ventral progenitor regions, namely the MGE, the AEP/POA and the TS. By controlling the production of neurons and glia that populate the entire telencephalon, Nkx2.1 is a key factor for brain patterning in embryonic development.
Origins and timing of gliogenesis in mouse brain. The exact timing of the generation and origin of embryonic glia is a topic of active investigation. Several mechanisms describing spatial and temporal differentiation of progenitors required for generation of the correct number of different types of glial cells have been proposed. Radial glia (RG) stem cells of the ventricular zone are proposed as the principal progenitor type during late embryonic brain development. During embryonic development, RG cells of the dorsal telencephalon not only generate most neurons of the cerebral cortex but also give rise later to two main glial sub-types, astrocytes and oligodendrocytes 19,22,23,26,29 . This hypothesis is supported by the generation of astrocytes and oligodendrocytes by spinal cord radial precursors 59 . Embryonic astroglia of the CC and the IG are also believed to originate from the radial glia of the dorsomedial pallium 30,31 . Contrary to what has been so far proposed, our study here shows for the first time that Nkx2.1 regulates the generation of transient midline dorsal embryonic astroglia of the CC and IG from the ventral progenitor regions. This is in accordance with our previous work showing that transient embryonic astroglia of the ventral telencephalon and transient embryonic NG2 glia occupying ventral and dorsal telencephalic regions are both derived from Nkx2.1 + progenitors from the subpallium 18,44 . Numerous studies also indicate that embryonic oligodendrocytes are produced in waves from the ventral telencephalic progenitors 16,17 . Additionally, subpopulations of telencephalic NG2 glia are documented to originate from ventral radial glia and pallial progenitors at E14 60,61 . Also, previous reports have shown that some postnatal astrocytes occupying the cerebral cortex, white matter and striatum originate from progenitor cells in the subventricular zone (SVZ) of the mammalian forebrain 33 . Recent evidence, however, shows that local populations of differentiated astrocytes constitute the primary source of postnatal glia and the SVZ progenitors contribute to only 3% of postnatal glia 34 .
The temporal competence of glial progenitors is not yet fully known. In the dorsal telencephalon, astrocyte gliogenesis was reported to occur only after neurogenesis (after E17 in mice) when the bipotential radial glial cells of the dorsal pallium differentiate into astrocytes [20][21][22]24,25,[27][28][29] . The time of generation of some of the astrocytes occupying the CC midline region however is noted to be between E13 and postnatal day 2 (P2) with a peak at Scientific RepoRts | 7:43093 | DOI: 10.1038/srep43093  PCR (a-c). A very faint signal was detected in some samples immunoprecipitated with non-specific control IgG. These results suggest that Nkx2.1 binds in vivo to the promoter region of GFAP at putative Nkx2.1 binding sequences. The relative intensities of all the bands were calculated by assigning an arbitrary value of 1 to the input band. The quantifications are indicated below the gels. The Figure represents one of three independently performed assays. As a control, human embryonic kidney 293 (HEK293) cells were co-transfected with two reporter-constructs, namely, the pDRIVE-mGFAP-LacZ and the pCAG-IRES-Tomato plasmids (d-f). Cell nuclei were counterstained in blue with Hoechst (d). In the control, only 4.26% of Tomato + cells were LacZ + , n = 94. To test the binding of Nkx2.1 to the GFAP promoter, HEK293 cells were co-transfected with two reporter constructs, namely, the pDRIVE-mGFAP-LacZ and the pCAG- Nkx2.1-IRES-Tomato plasmids (g-i). Cell nuclei were counterstained in blue with Hoechst (g). Activation of the LacZ reporter was seen upon addition of the Nkx2.1 expression vector, thus confirming that Nkx2.1 activates the GFAP promoter, most probably by binding the sequence identified as Nkx2.1 consensus. Here, 98% of Tomato + cells were LacZ + , n = 50. Bar = 50 μ m.
Scientific RepoRts | 7:43093 | DOI: 10.1038/srep43093 E14, much earlier than previously proposed 30 . Our previous work demonstrated that transient Nkx2.1-derived midline glia in the anterior commissure region are already generated by E14.5 18 . Here, we confirmed that transient Nkx2.1-derived astroglia in the CC and surrounding region are mainly generated between E14.5 to E16.5 too. Contrary to previous reports, we show that not only are these glia generated much earlier than previously proposed but also that they are generated from the ventral Nkx2.1 + progenitor regions. Hence, our study here sheds light upon both the spatial and temporal aspects of origin of embryonic telencephalic glia.
Multilevel regulation of embryonic astrogliogenesis by Nkx2.1. The Nkx2.1-derived embryonic cell population is broadly divided into GABAergic neurons, astrocytes and NG2 glia based on their expression profiles. Only Nkx2.1 + astrocyte-like cells that are GLAST + and/or GFAP + maintain Nkx2.1 expression while other Nkx2.1-derived cells that are Olig2 + NG2 glia no longer express Nkx2.1 after differentiation. Loss of Nkx2.1 function leads to a drastic reduction in the numbers of astroglia and NG2 glia in the midline dorsal (CC, IG, MZG) and ventral telencephalic (mutant MGE* and POA*, and septum) regions. The loss of astroglia and NG2 glia is not accompanied by an increase in apoptotic cells indicating that loss is likely due to incapacity of the precursors to generate astroglia and NG2 glia. Indeed, further analyses revealed that the loss of glia is associated with a decrease in Nkx2.1-derived precursor division capacity. Accordingly, we observed a drastic decrease in GLAST + precursors expressing the mut-Nkx2.1 in the VZ, SVZ of the mutant MGE*, mutant POA* and TS region, and in GLAST + differentiated astrocytes expressing the mut-Nkx2.1 in the parenchyma (striatum, LPOA/LH, septum) in Nkx2.1 −/− compared to mice. The decreased presence of the precursors and differentiated astroglial population was accompanied by reduced proliferative status of the BrdU + dividing cells containing mut-Nkx2.1 in the VZ and SVZ of mutant MGE*, mutant POA* and the septal nucleus region in Nkx2.1 −/− mice compared to control precursors. Additionally, in vitro differentiation of E14.5 Nkx2.1 −/− mutant MGE*, POA*-derived neurospheres revealed that progenitors were unable to generate GFAP + astroglia expressing mut-Nkx2.1 though they still retained the capacity to generate post-mitotic neurons. Hence, the reduction in number of astroglia and NG2 glia can be attributed to inadequate proliferation of precursors in the three subpallial domains in the absence of Nkx2.1. Nonetheless, reduced generation of astroglia in the neurosphere assay could be due to decreased specification or differentiation capacity of precursors not expressing Nkx2.1. Evidence for this comes from previously published studies where loss of Nkx2.1 results in a ventral to dorsal re-specification of the MGE* to LGE* 37 . Likewise, we showed that in Nkx2.1-Cre + ; Rosa-DTA mice, there is significant loss of differentiated astroglia at the anterior commissure midline without reduced precursor proliferation, thus, suggesting a role for Nkx2.1 after the proliferation step 18 . Nkx2.1-Cre + ; Rosa-DTA mice cause the selective ablation of Nkx2.1-derived post-mitotic cells but not Nkx2.1 + precursors due to a delay in expression and subsequent action of diphtheria toxin 18 . Also, the GLAST + astroglia at the CC midline at E16.5-E18.5 still retain Nkx2.1 expression (Fig. 1c,d), pointing towards additional Nkx2.1 functions in mature astroglia. Performing a clonal analysis with the neurosphere assay where both the number and size of the astrocytic clones are estimated could provide further insights into Nkx2.1's mode of action.
Chromatin immunopreciptation analysis performed in this study suggests that the Nkx2.1 might act by directly activating astroglial specific gene, such as GFAP as shown here. Future experiments where the putative Nkx2.1-binding site found in this study is mutated would further strengthen this result.
Nkx2.1 is important for several aspects of brain development in embryogenesis. In Nkx2.1 knockout mice several developmental abnormalities occur in the ventral forebrain 43 . Previous reports from our and other groups have shown that Nkx2.1 is not only important for regional specification of the ventral telencephalic regions, MGE and POA, but is also essential for the generation of a wide spectrum of Nkx2.1-derived lineages, including GABAergic interneurons, NG2 glia (or oligodendrocytes) and astrocytes that populate both the dorsal and ventral telencephalon from E12.5 3,14,[16][17][18]37,[42][43][44]54,58,[62][63][64] . We showed previously that the absence of NG2 glia drastically affects the vascular development in all the telencephalon leading to severe blood vessels reduction of ramifications and connections 44 . We also demonstrated that a drastic reduction in Nkx2.1-derived cell populations leads to AC commissure agenesis and commissural axon misrouting 18 . Interestingly, the timing of generation of the Nkx2.1-derived cells precedes the arrival of the commissural axons at the AC midline, and some of the precursors in the TS and astrocytes surrounding the AC use Slit2 to tunnel the axons through the anterior commissure 18 .
Also, the reduced GABAergic neuronal population and possibly the loss of astrocytes in the Nkx2.1 −/− mice lead to slight callosal axon branching and outgrowth defects in the CC tract 65 . Though Nkx2.1 −/− mice display a drastic loss of astroglia within CC and its surrounding regions including IG and MZG, the mice do not present CC agenesis and severe callosal axon bundles defects. Hence, these results show that these astroglia do not play a major role in CC axon guidance. These results contradict previously published reports where IG astroglia have been proposed to act as guidepost cells for callosal axons 31,66 . However, we cannot exclude the proposed contribution of GW radial glia in callosal axon guidance 66-68 since they are not Nkx2.1-derived and hence, are not ablated in Nkx2.1 −/− mice. Thus, radial glia precursors occupying the GW and secreting Slit2 may be key player cells for callosal axon guidance during embryonic development.
Altogether, Nkx2.1 is able to perform its vast range of roles through regulation of proliferation of Nkx2.1 + precursors as shown here, and also possibly through an effect on glial specification/differentiation and survival. It appears that Nkx2.1 mediates part of these effects through transcriptional regulation, such as that shown for astroglial gene GFAP here. Interestingly, Nkx2.1 has previously been proposed to regulate several transcriptional programs that are important in vertebrate lung morphogenesis 69 . Notably, transcriptional regulation by Nkx2.1 in early (E11.5) and late (E19.5) mouse lung development 40 . Interestingly, in mouse lungs, Nkx2.1 also directly regulates cell cycle effectors and its loss alters cell cycle progression 40 . Further investigation into the complete repertoire of transcriptional regulation exerted by Nkx2.1 could provide interesting insights into Nkx2.1 mode of action and astroglial generation in the brain.
Along the same line, previous reports have shown that Nkx2.1 regulates the transcription of many thyroid-specific genes [35][36][37] and activates pulmonary-surfactant 38 , as well as pituitary gland genes 39 . Moreover, it has been shown in vitro, that Nestin might be a target of Nkx2.1 70 . Hence, it is probable that Nkx2.1 displays functional conservation in brain, thyroid, pituitary, and lung.
Complex cellular and molecular interactions between glia, neurons and guidance cues produced by them, govern the formation of the midline structures such as the corpus callosum and anterior commissure. Several Nkx2.1-derived glial and neuronal populations populate these structures, and further understanding of the mode of regulation mediated by Nkx2.1 can help to understand the formation of dorsal and ventral telencephalic regions.

Methods
Animals. All studies on mice of either sex were performed in compliance with the national and international guidelines, and with the approval of the Federation of Swiss cantonal Veterinary Officers (authorization number 2164). Mice were housed under conditions of controlled temperature and illumination (12-hour light-dark cycle, with lights on at 07:00 am and off at 07:00 pm). Animals had ad libitum access to food and water and were monitored regularly. For staging of embryos, midday of the day of vaginal plug formation was considered as embryonic day 0.5 (E0.5). Wild-type mice maintained in a CD-1/SWISS genetic background were used for developmental analysis of the CC. We used wild-type (+ /+ ) and homozygous mutant Nkx2.1 mice 37,43,71 , which are referred as Nkx2.1 +/+ and Nkx2.1 −/− respectively, in this work. We used heterozygous GAD67-GFP knock-in mice, described as Gad1-EGFP knock-in mice 53 . Gad1-EGFP knock-in embryos can be recognized by their GFP fluorescence. PCR genotyping of these lines was performed as described previously 72  Fluorescence immunostaining was performed as follows: non-specific binding was blocked with 2% normal horse serum in PBS 1X solution with 0.3% Triton X-100 for preincubation and incubations. The primary antibodies were detected with Cy3-conjugated (Jackson ImmunoResearch laboratories, West Grove, PA) and Alexa488-, Alexa594-or Alexa647-conjugated antibodies (Molecular Probes, Eugene, OR) used at 1:300 dilution. Sections were counterstained with Hoechst 33258 (Molecular Probes), mounted on glass slides and covered in Mowiol 4-88 (Calbiochem, Bad Soden, Germany).
DAB immunostaining was performed as follows: Endogenous peroxidase reaction was quenched with 0.5% hydrogen peroxide in methanol and non-specific binding was blocked by adding 2% normal horse serum in Tris-buffered solutions containing 0.3% Triton X-100 for preincubation and incubations. The primary antibodies were detected with biotinylated secondary antibodies (Jackson ImmunoResearch, West Grove, PA) and the Vector-Elite ABC kit (Vector Laboratories, Burlingame, CA). The slices were mounted on glass slides, dried, dehydrated, and covered with Eukitt. for GFP and Alexa488), with a HeNe laser 543 nm (green excitation for Alexa 594 and CY3), with a HeNe laser 633 nm (excitation for Alexa 647 and CY5) and a Diode laser 405 nm (for Hoechst-stained sections). Z-stacks of 10-15 planes were acquired for each CC coronal section in a multitrack mode avoiding crosstalk.
All 3D Z stack reconstructions and image processing were performed with Imaris 7.2.1 software. To create real 3D data sets, we used the mode "Surpass". The colocalization between two fluorochromes was calculated and visualized by creating a yellow channel. Figures were processed in Adobe Photoshop© CS4 and CS5 and schematic illustrations in Fig. 2 were produced using Adobe Illustrator© CS4.
In vivo Quantifications. Glial cell population analysis. In 50 μ m thick brain sections of Nkx2.1 +/+ and Nkx2.1 −/− embryos at E18.5, the astroglial cells were labeled with GFAP and polydendroglial cells were labeled with NG2. Cells were counted in the CC, IG, MGE, MZG and POA regions from at least 4 brains per condition. Cell densities were reported per surface unit area (number of cells/mm 2 ). The quantification was done using Neurolucida 9.0 and Neurolucida 9.0 Explorer© software.
In 50 μ m thick brain sections of Nkx2.1 +/+ and Nkx2.1 −/− embryos at E18.5, the astroglial cells that were labeled for Olig2 or both Olig2 and GLAST were counted in the CC mid from at least 2 brains per condition. Olig2 staining labeled the glial cell bodies while GLAST labeled both the cell bodies and processes. The cell densities were determined in the medial and lateral part of the CC. The cell densities were reported per volume unit (number of cells/mm 3 ). The quantification was done using Imaris ® 7.2.1 software.
Nkx2.1 + and GLAST + and BrdU + proliferation analyses. Pregnant female mice were injected intraperitoneally with a solution of 8 mg/ml of 5-bromo-2′ -deoxyuridine in PBS to a final concentration of 50 mg/kg body weight. To trace the division rate of the subpallial precursors, pregnant females were sacrificed 2 hours post-injection. Embryos were collected after caesarean section and quickly killed by decapitation. Their brains were dissected out and fixed by immersion overnight in a solution of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) at 4 °C. In 50 μ m thick brain sections of Nkx2.1 +/+ and Nkx2.1 −/− embryos at E16.5, Nkx2.1 + cells, BrdU + dividing cells and GLAST + precursors or post-mitotic astroglial cells of the MGE, POA and TS were counted in the VZ, SVZ and in the parenchyma of each region, from at least 4 brains per condition. Nkx2.1 and BrdU staining labeled the cell bodies while GLAST labeled both the cell bodies and processes. Although the expression level of truncated Nkx2.1 protein (mut-Nkx2.1) in Nkx2.1 −/− embryos was low, it was easily identifiable using immunostaining and quantifications were performed on higher magnification images where the staining could be easily visualized. The percentage of Nkx2.1-derived dividing precursors or post-mitotic glial cells were determined as follows: In each sub region and for each condition, a sample of at least four different Z-stacks was acquired at 100x magnification using a Leica SP5 confocal microscope. The Z-stacks comprised 10 planes that were acquired in a multitrack mode avoiding any crosstalk. Thereafter, in order to exclude the possibility of quantifying the same cells more than once, snapshots of only 3 planes (from the acquired 10 planes), were analyzed. The quantification of Nkx2.1, BrdU, GLAST and Hoechst staining was done on each snapshot separately Imaris ® 7.2.1 software.
Cell death analysis. In brain sections of Nkx2.1 +/+ and Nkx2.1 −/− embryos at E16.5, apoptotic cells labeled with either cleaved-caspase 3 or TUNEL were counted in the CC, MGE, SEP, and POA from at least 2 brains per condition. 50 μ m thick brain sections were used for cleaved-caspase 3 staining whereas 10 μ m thick brain sections were utilized for TUNEL staining. Cell nuclei were counterstained with Hoechst. For each condition, at least 5 different Z-stacks were obtained at 100x magnification using a Leica SP5 microscope. The number of apoptotic nuclei were counted and reported as an absolute number per section (the surface area of one section was 24119.332 μ m 2 ). The quantification was done using Neurolucida 9.0 and Neurolucida 9.0 Explorer© software.
Neurosphere generation and microscopic analysis. The protocol has been adapted from Arsenijevic Primary culture and sphere passaging. The brains of embryos at developmental stage E14.5 were collected as described above. They were carefully removed from the skull into ice-cold sterile dissecting medium (MEM 1X) complemented with Glucose 1 M (5 ml/100 ml). Thereafter, the brains were embedded in low melting point Agarose 3% (LMP-Agar, Gibco) at 37 °C, and cut into 250 μ m thick slices using a vibratome (Leica© VT 1000 S). The sections were collected in the ice-cold dissecting medium. The areas of interest (MGE, POA and SEP) were dissected out using two tungsten needles under a stereomicroscope (Leica© MZ16F). The dissected pieces of tissue were then collected into 1 ml ice-cold sterile Hormone Mix Medium (MHM 1X) supplemented with Penicillin (50 U/ml) and Streptomycin (50 U/ml) (GIBCO). The Hormone Mix Medium is a growing medium containing DMEM and F-12 nutrient (1:1), glucose (0.6%), glutamine (2 mM), sodium bicarbonate (3 mM), HEPES buffer (5 mM), transferrin (100 mg/ml), insulin (25 μ g/ml), progesterone (20 nM), putrescine (60 μ M), selenium chloride (30 nM) 74 . Brain tissue pieces were mechanically dissociated in sterile conditions with a fire-polished pipette in the Hormone Mix Medium. The pipette was rinsed before the dissociation of each new region.
The dissociated cells were then grown in Hormone Mix Medium complemented with Pen/Strep and EGF in 6-well dishes (Nunclon Surface, NUNC Brand Products, Nalge Nunc International) at a concentration of around 10 4 -10 5 cells per 1 ml and 4 ml per dish. After 6-7 days in vitro (DIV) at 37 °C in a 5% CO 2 atmosphere, the sphere cultures were expanded. Primary spheres were dissociated mechanically and cells were plated at the density of Scientific RepoRts | 7:43093 | DOI: 10.1038/srep43093 2 × 10 6 cells for 40 ml in a flask (Nunclon Surface, NUNC Brand Products, Nalge Nunc International). Sphere passages were done every 7 DIV, by sphere dissociation and transfer of 2 × 10 6 cells to a new 40 ml flask.
Differentiation of spheres. After 7 DIV, the neurospheres of optimum size were chosen under a steremicroscope (Nikon©) and transferred individually and plated onto poly-L-ornithine coated coverslips in 24-well plates (Nunclon Surface, NUNC Brand Products, Nalge Nunc International). Each coverslip contained about ten spheres and 1 ml of Hormone Mix Medium supplemented with Pen/Strep and 2% fetal bovine serum (FBS).
Immunofluorescence on differrentiated Neurospheres. After 7 DIV, the neurospheres were fixed in 4% PFA for 20 minutes and permeabilized with 0.3% triton/PBS1X for 3 minutes. Coverslips were incubated with primary antibodies diluted in PBS containing 10% NHS for 2 hours at room temperature, followed by secondary fluorescent antibodies for 45 minutes at 37° and Hoechst staining for 5 minutes.
Neurosphere production analysis. MGE-and POA-derived neurospheres were obtained from Nkx2.1 +/+ and Nkx2.1 −/− E14.5 embryos. After 7 DIV, the neurospheres were differentiated and immunostained as mentioned above. Two different brains were used for each condition and were labeled for Nkx2.1, GFAP, GLAST, and β III tubulin. Cell nuclei were counterstained with Hoechst. For each condition, a total of at least 5 different Z-stacks in 5 different neurospheres were acquired at 100x magnification using a Leica SP5 microscope. The percentage of Nkx2.1 + /GFAP + differentiated astrocytes and Nkx2.1 + /β III tubulin + differentiated neurons were counted directly on the Z-stacks by using Imaris ® 7.2.1 software.
Chromatin Immunoprecipitation. Chromatin immunoprecipitation was conducted on E16.5 brain samples according to the instructions provided by the manufacturer (Upstate, 17-295), using 2 μ g of mouse anti-Nkx2.1 monoclonal antibody (MS699-P, Lab Vision). For crosslinking, 1% PFA was used. For sonication, six bursts of 45 seconds ON (30% power) and 30 second OFF were given and samples were kept on ice throughout. Mouse Genome Assembly data mm9 was used to map sites.
A 391 bp PCR fragment of the Lhx6 promoter that includes a Nkx2.1 binding sequence at position − 240 bp relative to the putative transcriptional start site was identified using primers 5′ -tttgtaccgagagtaggagaagg and 5′gtcctaactttgtagtgggcattt.
A 206 bp PCR fragment of the GFAP promoter that includes a putative Nkx2.1 binding sequence (ctcaagt) at position − 838 bp relative to the putative transcriptional start site was found to be a binding target and was identified using primers 5′ -tggataagaggccacagagg and 5′ -cctctcccctgaatctctcc.
Primers against two fragments of the Neurogenin2 promoter region, comprising the core Nkx2.1 binding consensus sequence (tcaag), were made. 1) Primers 5′ -cgggattctgactctcactaattc and 5′ -aatggttctaaagctcctgttgg were designed to amplify a 410 bp PCR fragment with the core consensus Nkx2.1 binding sequence at position − 668 bp relative to the putative transcriptional start site. 2) Primers 5′ -cgggattctgactctcactaattc and 5′aatggttctaaagctcctgttgg were designed to amplify another 352 bp PCR fragment with the core consensus Nkx2.1 binding sequence at position − 4073 bp relative to the putative transcriptional start site.
Following controls were used: (1) Null -beads only without any antibody, (2) IgG -beads with an isotype matched control immunoglobulin (Ig) to know the background of the assay, (3) Input -starting material taken before immunoprecipitation with antibody, and (4) Negative -non-template control used for the PCR reaction to spot any contamination.
For quantification, ImageJ was used to measure the band densities (https://imagej.nih.gov/ij/docs/menus/ analyze.html#gels). The relative intensities of all the bands (null control, IgG control, immunoprecipitated with antibody, and negative control) were calculated by assigning an arbitrary value of 1 to the input band.
Transfection of HEK293 cells. A suspension of HEK293 cells adapted to serum-free growth medium was plated at 1 × 10 6 cells in 4 ml media in a 60mm plate. For formation of the transfection complexes, a 3:1 ratio of FuGENE ® HD Transfection Reagent (μ l): plasmid DNA (μ g) was prepared and used for transfection.
The study was performed by co-transfecting an expression plasmid for constitutive over-expression of Nkx2.1 (pCAG-Nkx2.1-IRES-Tomato) or a control plasmid (pCAG-IRES-Tomato) with the pDRIVE-mGFAP plasmid containing the GFAP promoter region in front of the LacZ reporter gene. The pCAG promoter is constructed of following sequences: (C) cytomegalovirus early enhancer element, (A) promoter, the first exon and intron of the chicken beta-actin gene and (G) the splice acceptor of the rabbit beta-globin gene. The mouse GFAP promoter sequence from the pDRIVE-mGFAP plasmid (#pdrive-mgfap, InvivoGen) used for transfection experiments is shown in Supp. Fig. S5B. Transfection complexes were formed by mixing 2 μ g of each of the two plasmids with 12 μ l of Fugene transfection reagent and 188 μ l of Optimem reduced serum media. The mix was incubated at room temperature for 20 minutes and then added to the cell plates. The cell plates were kept in the 37 °C incubator and gene expression analysis was done after 24-48 hours of transfection. Fluorescence immunostaining was done to visualize the presence and level of LacZ expression. Tomato signal was visible by direct fluorescence, however, for a clearer visualization of Tomato signal anti-RFP immunostaining was done (described above).

Experimental design and statistical analysis.
For all analyses at least three independent experiments were performed. Mutant embryos were always compared with controls originating from the same litter. Qualitative, morphological and cellular change phenotypes of the control and mutant brains were ascertained blindly. Quantitative analyses were verified blindly by a second experimenter. The results from all quantifications were analyzed with the aid of Statview software (SAS Institute Inc.). For all analyses, values were first tested for Scientific RepoRts | 7:43093 | DOI: 10.1038/srep43093 normality and the variance of independent populations were tested for equality. Values that followed a normal distribution were compared using Student's t-test. To show the degree of significance for quantitation included in the study, we added the number of asterisks based on the following standard p-value criteria: ***p < 0.001; **p < 0.01; *p < 0.05.
Atlas and nomenclature. The neuroanatomical nomenclature is based on the "Atlas of the prenatal mouse brain" 75 .