Arcuate Nucleus Expression of NKX2.1 and DLX, and Lineages Expressing These Transcription Factors in NPY⁺, POMC⁺ and TH⁺ Neurons in Neonatal and Adult Mice

Animals

Brains were obtained from 26 adult and 17 postnatal day 0 (PO) mice. The University of California San Francisco's Institutional Animal Care and Use Committee approved the handling and use of all animals for these studies. Adult mice were anesthetized with avertin and perfused transcardially with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS; per 1 liter: 0.8g of NaCl, 0.02g of KCl, 0.144g of Na2HPO4, 0.024g of KH2PO4, pH to 7.4). PO mice were anesthetized on ice; brains were removed and placed in 4% PFA.

Immunocytochemistry and histochemistry

Following postfixation (2–4 hours) and cryoprotection in 10% sucrose for at least four hours and 20% sucrose in PBS overnight, PO and adult brains were sectioned coronally (20μm) on a cryostat and collected onto slides (FisherBrand Superfrost Slides, Fisher Scientific). Slides were washed in PBS, 3 × 10 minutes, and incubated for one hour at room temperature in blocking solution (1% bovine serum albumin, 5% normal goat serum, 0.3% Triton X-100 in PBS). After pretreatment, sections were placed overnight at 4°C in primary antisera diluted in blocking solution (For a list of antibodies, see Table 1). The following day, after three ten-minute rinses in PBS, sections were incubated for 1–2 hours at room temperature in the appropriate fluorescently labeled secondary antibodies (Alexa- 594 or Alexa-488). After three ten-minute rinses in PBS, sections were cover-slipped using Fluoromount G (Fisher Scientific).

Free-floating Immunohistochemistry: Following intracardial perfusion with 4% paraformaldehyde in PBS the brains were removed and postfixed overnight in the same fixative and cryoprotected by immersion in 30% sucrose. Free-floating cryostat sections (40μm) were processed using standard procedures. Briefly, sections were treated with 1% hydrogen peroxide in PBS for 30 minutes, and then washed for 30 minutes in PBS. Sections were blocked in 10% goat serum in PBST (0.1% TritonX-100 in PBS) for 2 hours, and then incubated with Tyrosine hydroxylase primary antibody (1:1000) overnight at 4°C. After primary incubation, sections were washed for 1 hour in PBS, and then incubated in biotinylated rabbit secondary antibody (Vector Labs, Burlingame, CA; 1:500) for 2 hours at room temperature. Sections were then washed for 1 hour in PBS, incubated with ABC solution (Elite kit, Vector) for 2 hours, washed for 1 hour in PBS, and developed in 0.5mg/ml diaminobenzamine (Sigma, St. Louis, MO) and 0.005% peroxide. Sections were mounted, dehydrated with alcohol and xylene washes, and cover slipped with Mount-Quick (Ted Pella, Redding, CA).

Immunocytochemical Controls: Omission of primary antibodies eliminated all immunoreactivity reported below. Elimination of one of the primary antibodies with application of both secondary antibodies confirmed secondary antibody specificity, i.e. the secondary antisera reacted only with IgG of the appropriate species. In cases where null mutant mice are available, antibodies were used on mutant tissue and staining was not detected (See Table 1).

Imaging

Images were collected using a Zeiss Meta 510 scanning microscope. For each double label immunohistochemical image, the two channels were collected separately with single wavelength excitation and then merged to produce the composite image. All images were composed in Adobe Photoshop 7.0 (Adobe Systems, Mountain View, CA) adjusting only brightness and contrast.

Nuclear profile counts and statistics

Cells on one side of the ARN were counted. A square area that began at the third ventricle was defined and used for profile counts. For TH immunoreactive profiles and double labeled profiles, the area of the ARN above the third ventricle on both the left and right side were chosen for analysis. Nuclear profiles were counted in a single section plane (n=3, where n= number of sections). The number of antibody⁺ profiles, GFP⁺ or β–galactosidase⁺ profiles and antibody⁺/GFP⁺ or β–galactosidase⁺ profiles were counted in PO and adult mice. The percent of antibody double labeled and GFP double labeled profiles was determined (n=3) and the over-counting of profiles was corrected for by applying Abercrombie's formula: T/T+h, where T= section thickness and h= mean diameter of the objects along the z axis (20/ 20+ 6.785). The numbers of cell profiles labeled by immunohistochemistry were counted; the average number of labeled cells, and the percentage of double labeled cells were determined. The Abercrombie correction factor was applied and the standard deviation (variance) was determined for each value (Tables 2–4).

Gad67 (Gad1)-GFP is co-expressed with DLX, NKX2.1, NPY and TH in the neonatal and adult ARN

We studied Gad67 (Gad1) expression in the ARN and adjacent regions of the hypothalamus using GFP expression from the GAD67 locus (Tamamaki et al., 2003) on the day of birth (P0) (Fig. 1) and in adulthood (Fig. 2). We compared expression of Gad67 (GFP) to the expression of the DLX and NKX2.1 transcription factors, the NPY and POMC neuropeptides and the tyrosine hydroxylase (TH) enzyme. In the neonatal brain (P0) Gad67-GFP was detected in cell bodies and nerve fibers, whereas the VMH was largely devoid of GFP⁺ cells but contained some GFP⁺ processes (Figure 1 B, F, J, N and R). The pan distaless antibody, which binds DLX proteins from invertebrates and vertebrates (Panganiban et al., 1995; Stuhmer et al., 2002a), detected DLX expression in a subset of Gad67 (GFP)⁺ P0 ARN cells (Figure 1, A–D); DLX expression was not detected in the VMH. 92.4% DLX⁺ (pan distaless) cells are Gad67 (GFP)⁺ (Figure 1 C and D, arrowheads, Table 2). NKX2.1 was expressed in a subset of Gad67-GFP⁺ cells in the P0 ARN (Figure 1 E–H) and a subset of Gad67-GFP⁻ VMH cells (Lee et al., 2001; Marin et al., 2002; Kim et al., 2006). NKX2.1 was also expressed in cells lining the ventricle (which become ependymal cells, or tanycytes, in the mature hypothalamus) (Lee et al., 2001).

POMC is expressed in anorexigenic neurons (activated in response to leptin; Cowley et al., 2001). POMC⁺ neurons in the P0 ARN did not exhibit detectable Gad67 (GFP)⁺ immunoreactivity (Figure 1 I–L) (POMC⁺ cells were not detected in the VMH). NPY, which is expressed in orexigenic cells (inhibited by leptin; Inui et al., 1998; Serguei et al., 2004), are largely Gad67 (GFP)⁺ (Figure 1 M–P). There were a few NPY⁺ cells in the VMH that are Gad67 (GFP)⁺ (Figure 1 O, asterisk). Dopaminergic cells in the P0 ARN express tyrosine hydroxylase (TH). TH⁺ cells were present in similar distribution to DLX⁺ cells (Figure 1Q). TH⁺ cells that are large in diameter clearly expressed Gad67 (GFP) (Figure 1 S and T, arrowheads), whereas smaller diameter TH⁺ cells did not appear to be GAD67⁺ (GFP) (Figure 1S and T, arrows).

In the adult ARN, Gad67 (GFP) showed a similar pattern of co-expression with DLX, NKX2.1, NPY and TH as in the P0 ARN (Fig. 2). There were a few salient differences. First, there were a subset of POMC⁺ cells that were Gad67 (GFP)⁺ (Figure 2 I–L, arrowhead), however the majority of POMC expressing cells (~95%) in the adult ARN lacked GAD67 (GFP) immunoreactivity. Thus, like Hentges et al., 2004, we found evidence that a subset of POMC⁺ cells are GABAergic, although we observed roughly 6-fold fewer double-positive cells, perhaps because of differences in our methods (they used double in situ hybridization). Second, NPY immunoreactivity was difficult for us to detect in cell bodies in the adult mice, making it difficult to determine whether the NPY-expressing neurons also expressed Gad67 (Figure 2 M–P).

Fate mapping of Nkx2.1-expressing cells in the P0 ARN using the Nkx2.1-Cre BAC transgene

We crossed mice with the Nkx2.1-Cre BAC transgene (Xu et al., 2008) with the Z/EG “recombination reporter” mice (Novak et al., 1999) to identify cells that had expressed Nkx2.1 at some point in their development (Figure 3 B–D, F–H, J–L, N–P, R–T and V–X). Cre mediated recombination activates GFP expression in the Z/EG mice. First, we compared GFP and NKX2.1 protein expression at P0 (Figure 3 A–D). The regional pattern of expression in the rostroventral hypothalamus was very similar (particularly in the ARN; Figure 3 C and D), although there were some differences. There were fewer GFP⁺ ependymal cells (arrow in Figure 3 B), and there was less GFP labeling in the VMH (Figure 3 B). In the ARN, there is a small population of NKX2.1⁺ cells (14.7%) that did not display GFP immunoreactivity (Figure 3 C and D, arrows), and 25.3% of GFP⁺ cells lack NKX2.1 immunoreactivity (Figure 3 C and D, asterisk, Table 3). Presumably some of the GFP⁺ only cells expressed Nkx2.1 at an earlier point in their lifetime. Lack of GFP-expression in some NKX2.1⁺ cells suggests that the Nkx2.1-Cre BAC may lack regulatory elements that drive expression in these regions, or could be due to under-reporting because of incomplete recombination.

We next assayed whether Nkx2.1-lineage cells co-expressed DLX, POMC, NPY and TH pan-DLX antibody at PO. Cells co-expressing DLX and GFP (Nkx2.1 lineage) were detected in lateral areas of the ARN (Figure 3G, boxed region), although 55.2% of DLX⁺ cells were GFP-negative (Figure 3G and H, arrows). Approximately 76.6% of POMC⁺ cells were GFP⁺ (Figure 3 K and L, arrowheads). Approximately two-thirds of NPY⁺ cells arise from the Nkx2.1 lineage (Figure 3 O and P); NPY⁺;GFP⁻ cells tend to be closer to the ventricle (Figure 3 O,P, arrow). 54% of TH⁺ cells were GFP⁺ (Figure 1 S and T, arrowheads), many cells that display only TH immunoreactivity were present in lateral areas of the ARN (Figure 3 S and T, arrows).

Fate mapping of Nkx2.1-expressing cells in the adult ARN using the Nkx2.1-Cre BAC transgene

Analysis in the adult involved crossing the Nkx2.1-Cre BAC transgene (Xu et al., 2008) with the Rosa-LacZ reporter mouse. All NKX2.1⁺ cells were also β-gal⁺ (i.e. had expressed Cre), while some NKX2.1⁻ cells were also β-gal⁺ indicating they had expressed Nkx2.1 at an earlier point in their development (Figure 4 C and D). Approximately 56.5% of ARN β-gal⁺ cells were pan-DLX⁺; 29% of ARN pan-DLX⁺ cells were β-gal⁻ (Figure 4 E–H). Nkx2.1-Cre descendents contributed to 85.8% of the POMC⁺ cells (Figure 4 I–L) and 56.6% of the TH⁺ cells (Figure 4 M–P).

Dlx 5/6i lineage in the P0 ARN

Dlx5 and Dlx6 are expressed during development of the hypothalamus (Eisenstat et al., 1999; Liu et al, 1999; Ghanem et al., 2007; Puelles and Rubenstein, unpublished); their expression is driven by at least two intergenic enhancers (I5/6i and I5/6ii) (Zerucha et al., 2000; Ghanem et al., 2007) and by an uncharacterized extragenic element(s) (Dye and Rubenstein, unpublished). We have generated a transgenic mouse that expresses Cre under the control of Dlx5/6i (Monory et al., 2006; Kohwi et al., 2007), which largely recapitulates the expression of transgenic mice that express LacZ under the control of I5/6i (Zerucha et al., 2000; Stuhmer et al., 2002).

We crossed the Dlx5/6i-Cre transgene mouse with the Z/EG “recombination reporter” mice to identify cells that had expressed the Dlx5/6 intergenic enhancer (i) at some point in their development (“Dlx5/6i lineage”). GFP⁺ cells of the Dlx5/6i lineage were present throughout the ARN, but were absent from the VMH (Figure 5 B, F, J, N and R). GFP⁺ axons were also apparent in the ARN and median eminence (Figure 5B, light green staining, open arrow). Pan-DLX and GFP double immunostaining revealed that there were cells that displayed only DLX immunoreactivity (Figure 5 C and D, asterisks), suggesting that Dlx5/6i was not expressed in all DLX⁺ ARN neurons. There were also some GFP⁺;DLX⁻ cells suggesting that DLX expression was down-regulated in some cells that had expressed Dlx5/6i at an earlier point in their lifetime.

Double immunohistochemical labeling of NKX2.1⁺ and GFP⁺ (I5/6i lineage) in the ARN (Figure 5 E–H) revealed that 66.5% of NKX2.1⁺ cells were derived from the I5/6i lineage (Figure 5 G and H, arrowheads, Table 4). Cells that displayed either POMC or NPY immunoreactivity were GFP⁻ (Figure 5 I–K; M–P, arrows). On the other hand, TH was expressed in 20.8% of the Dlx5/6i lineage population in the ARN (Figure 5 Q–T, arrowheads in S and T, Table 4).

Dlx5/6i lineage in the adult ARN

Similar to the Dlx5/6i lineage in the ARN at PO, NKX2.1⁺ cells were present in 46.3% of Dlx5/6i-Cre⁺ descendents in the adult ARN (Figure 6 A–D, arrowheads pointing to white cells in C and D, Table 4). NPY immunoreactivity was mainly present in nerve fibers and was largely absent from cell bodies (Figure 6 E, G and H). NPY immunoreactive nerve fibers were in close contact with Dlx5/6i-Cre descendents in the ARN (Figure 6 G and H, arrowheads). POMC was not present in Dlx5/6i-Cre descendents (Figure 6 I–L, arrows in K and L). 57.6% of TH⁺ cells were present in Dlx5/6i-Cre descendents. (Figure 6 M–P, arrowheads in O and P, Table 4).

Dlx1 Regulates the number of tyrosine hydroxylase⁺ ARN Neurons

Given DLX expression of TH⁺ cells of the ARN, we studied these cells in adult Dlx1^−/− mutants (Dlx2^−/− and Dlx5^−/− die the day of birth precluding this analysis). We found a ~2-fold reduction in the number of TH⁺ ARN neurons in postnatal day 60 Dlx1^−/− mutants (Dlx1^+/+: 54.67±6.74; Dlx1^−/−: 25.67±2.44) (Figure 7; Supplemental Figure 1). There was also an obvious reduction in the TH⁺ neuropile in the ARN, but not in the median eminence. TH was also reduced in dorsomedial hypothalamus (arrow, Figure 7), but not in the ventral midbrain (not shown), or its projection through the cerebral peduncle (Figure 7).

Dlx and Nkx2.1 expression in the hypothalamus

Previous studies in the mouse have focused on the expression of Nkx2.1 and the Dlx genes (Dlx1, 2, 5 and 6) in progenitors and immature neurons of the prenatal telencephalon and diencephalon (Eisenstat et al., 1999; Marin et al., 2002; Sussel et al., 1999). However, these genes are also expressed postnatally in the forebrain (Cobos et al., 2005; Lee et al., 2001; Marin et al., 2000; Mastronardi et al., 2006; Saino-Saito et al., 2003). Within the diencephalon, Nkx2.1 is broadly expressed in the basal hypothalamus, including the mammillary, ventromedial and ARN nuclei (Puelles and Rubenstein, 2003). In the diencephalon, the Dlx genes are expressed in a transverse alar domain that extends from the prethalamus (ventral) thalamus (reticular nucleus, zona incerta) through the posterior entopeduncular area (PEP) to the anterior hypothalamus and the suprachiasmatic area (SCH) (Puelles and Rubenstein, 2003). Dlx expression is also present ventral to the PEP-SCH domain, where the dorsomedial hypothalamic and tuberomammillary and ARN areas are Dlx⁺.

In the telencephalon, there is evidence that Dlx gene expression marks virtually all differentiating GABAergic and dopaminergic neurons (Kohwi et al., 2007; Saino-Saito et al., 2003; Stuhmer et al., 2002a,b); Nkx2.1 expression marks a subset of differentiating GABAergic and cholinergic neurons derived from the basal telencephalon (Marin et al., 2000).

Here we focused on Nkx2.1 and Dlx expression in the neonatal and adult ARN and ventromedial hypothalamic nuclei, exploring their expression in GAD67-, NPY-, POMC- and TH-expressing neurons. Nkx2.1 conditional mutants show differentiation defects in the region of the ventromedial hypothalamus and ARN nucleus, with alterations in reproductive functions (Mastronardi et al., 2006). Ongoing studies are investigating the effects of mutations in the Dlx genes on ARN development.

Validation of the Dlx5/6i-Cre and Nkx2.1-Cre transgenic lines

Throughout our work we found a high concordance of the recombination of patterns of both the Dlx5/6i-Cre and Nkx2.1-Cre transgenic lines and the endogenous expression of DLX and NKX2.1 in the ARN. However, we did detect some cells that underwent recombination and that did not express the transcription factors (Figure 3 D, Figure 4 D, Figure 5 D, Figure 6D). This could be due to earlier expression of the transgene, in a correct pattern, and subsequent of the down-regulation of the endogenous gene; however, we can't rule out whether there is ectopic expression of the transgene. Likewise, we also observed some DLX⁺ cells that did not show Dlx5/6i-Cre activity. This could reflect the fact that there are multiple Dlx5/6 and Dlx1/2 enhancers, which could drive DLX expression more broadly in the ARN (Ghanem et al., 2007). None-the-less, the recombination patterns driven by the Dlx5/6i-Cre and Nkx2.1-Cre transgenic lines provide useful information for future studies aiming to use these Cre lines to alter gene expression in the developing ARN.

Dlx lineage marks GABAergic and dopaminergic ARN neurons, and Dlx1 regulates the number of TH⁺ ARN neurons

Over 90% of DLX⁺ (pan-Distal-less antibody-reactive) cells in the neonatal and adult ARN nucleus express Gad67 (Gad1) (Figure 1; Table 2) showing that DLX-expressing cells are GABAergic (as in the telencephalon). However, without demonstrating that Gad65 (Gad2) and Gad67 (Gad1) expression are the same, we can't be certain that we have labeled all GABAergic neurons. We did not observe POMC co-expression in neurons of the Dlx56i-Cre lineage (Figures 5 and 6; Table 4). While roughly 50% of TH⁺ cells were in the Dlx56i-Cre lineage (Figures 5 and 6; Table 4). Thus, we conclude that Dlx expression in the ARN is restricted to GABAergic and dopaminergic (TH⁺) neurons. Surprisingly, NPY⁺ neurons, which are GABAergic, do not appear to be in the Dlx5/6i-Cre lineage. Furthermore, the distribution of pan-Distal-less⁺ and NPY⁺ cells appears different (Figure 1 and 2), suggesting that most ARN NPY⁺ cells do not express Dlx genes, although further analyses are needed to be certain. The observation that POMC⁺ ARN neurons are not in the Dlx5/6i lineage is expected given that these neurons are not GABAergic. In sum, Dlx expression in the ARN is largely limited to dopaminergic (TH⁺) neurons, and GABAergic neurons that are NPY negative; future analysis is needed to define the identity and properties of these GABAergic neurons, including establishing whether they are projection or local circuit neurons. Consistent with these analyses, we found that adult (P60) Dlx1^−/− mutants have ~2-fold fewer TH⁺ ARN neurons (Fig. 7). This adds to previous studies demonstrating the requirement of the Dlx genes for olfactory bulb and diencephalic TH⁺ neurons (Andrews et al., 2003; Long et al., 2007; Qiu et al., 1995). Future studies will define the mechanisms through which the Dlx genes control TH neuron numbers.

Nkx2.1 lineage marks subsets of POMC, GABAergic and dopaminergic ARN neurons

In the ARN, 60–70% of NKX2.1⁺ cells are GAD67⁺ (Figure 1; Table 2). Analysis of the Nkx2.1-Cre lineage shows that many of these cells are DLX⁺ and NPY⁺ (Figures 3 and 4; Table 3), supporting a model that NKX2.1 is expressed during the development of several types of GABAergic neurons. In addition, Nkx2.1-Cre induces recombination in TH⁺ and POMC⁺ neurons (Figures 3 and 4; Table 3); thus, NKX2.1 may be expressed in most subtypes of ARN neurons. However many GAD67⁺, TH⁺ and POMC⁺ cells do not arise from NKX2.1⁺ lineage, as assessed by Cre recombination. This is either because Nkx2.1 is not expressed in all cells, or due to incomplete recombination induced by the Nkx2.1-Cre transgene.

Furthermore, Nkx2.1 is broadly expressed in the ventromedial hypothalamus (a glutamatergic structure), which lacks Dlx expression (Figures 3 and 4). Thus, expression of Nkx2.1, and cells in this lineage, are not restricted to GABAergic and dopaminergic neurons of the ARN and contribute to many cell populations within the different hypothalamic regions.

Utility of Dlx5/6i-Cre and Nkx2.1-Cre mice for modifying gene expression in the ARN nucleus

Our analysis provides insights into the origins, transcriptional regulation and molecular properties of ARN neurons. Furthermore, this study demonstrates the utility of the Dlx5/6i-Cre and Nkx2.1-Cre mouse lines for modifying gene expression in ARN neurons.

The Nkx2.1-Cre allele will induce recombination in all known ARN neuronal cell types (POMC, NPY, GABA and TH). As Nkx2.1 expression begins in the neuroepithelium, it is likely that it can induce loxp recombination in progenitor cells; as a result, the ependyma also shows recombination (Figures 3 and 4). Thus, gene expression in ependymal tanycytes of the ARN and median eminence can be modified; tanycytes are implicated in regulating many hypothalamic functions (Rodríguez et al., 2005). Based on the recombination pattern at P0 and in the adult, Nkx2.1-Cre may not induce recombination in all ARN neurons; this may pose a problem for phenotypic analyses. Furthermore, phenotypic analysis of mutants generated using Nkx2.1-Cre need to consider the fact that Nkx2.1 is also expressed in the telencephalon (Xu et al., 2008), other regions of the hypothalamus, the thyroid and the lung.

The Dlx5/6i-Cre allele will induce recombination in only a subset of ARN neuronal cell types (GABA and TH). As Dlx5/6 expression begins in the subventricular zone, it is likely that it can induce loxp recombination in these secondary progenitor cells; as a result, cells can have genetic changes during midembryogenesis (~E11) when Dlx5/6 expression begins. Phenotypic analysis of mutants generated using Dlx5/6i-Cre need to consider the fact that this enhancer is also expressed in the telencephalon and other regions of the hypothalamus (Kohwi et al., 2007; Monory et al., 2006).