|Home | About | Journals | Submit | Contact Us | Français|
Here we analyze the role of the Lhx6 lim-homeobox transcription factor in regulating the development of subsets of neocortical, hippocampal and striatal interneurons. An Lhx6 loss-of-function allele, that expresses placental alkaline phosphatase (PLAP), allowed analysis of the development and fate of Lhx6-expressing interneurons in mice lacking this homeobox transcription factor. There are Lhx6+;Dlx+ and Lhx6−;Dlx+ subtypes of tangentially migrating interneurons. Most interneurons in Lhx6PLAP/PLAP mutants migrate to the cortex, although less efficiently, and exhibit defects in populating the marginal zone and superficial parts of the neocortical plate. By contrast, migration to superficial parts of the hippocampus is not seriously affected. Furthermore, whereas parvalbumin+ and somatostatin+ interneurons do not differentiate, NPY+ interneurons are present; we suggest that these NPY+ interneurons are derived from the Lhx6−;Dlx+ subtype. Striatal interneurons show deficits distinct from pallial interneurons, including a reduction in the NPY+ subtype. We provide evidence that Lhx6 mediates these effects through promoting expression of receptors that regulate interneuron migration (ErbB4, CXCR4, CXCR7), and through promoting the expression of transcription factors either known (Arx) or implicated (bMaf, Cux2, and NPAS1) in controlling interneuron development.
Current evidence shows that most rodent cortical interneurons are derived from progenitor domains in the prenatal subcortical telencephalon (subpallium) (reviewed in: Flames and Marin, 2005; Marin and Rubenstein, 2003; Wonders and Anderson, 2006). Ongoing studies are aimed at elucidating the precise regions within the subpallium that generate these cells, and the mechanisms that regulate their specification, migration and differentiation into distinct interneuron subtypes in pallial structures (i.e. cortex, hippocampus, parts of the amygdala, and olfactory bulb). Here, we have examined the role of the mouse Lhx6 LIM-Homeodomain transcription factor in regulating these processes.
The subpallium consists of four major subdivisions that have distinct molecular and morphological features: the lateral ganglionic eminence (LGE), medial ganglionic eminence (MGE), septum (SE) and preoptic area (POA) (Flames et al., 2007; Puelles and Rubenstein, 2003). Sulci approximate the transitions between most of these domains at early embryonic stages; however, the caudal part of the LGE and MGE lack a clear morphological demarcation, leading this region to be called the caudal ganglionic eminence (CGE), even though it has distinct LGE and MGE components. Efforts are underway to identify subdivisions within the progenitor domains of the LGE, MGE, SE and POA; the most advanced effort has used transcription factor expression patterns (Flames et al., 2007).
The MGE is thought to be the source of a large share of pallial interneurons –including the majority of interneurons that express parvalbumin and somatostatin. On the other hand, there are at least two types of calretinin+ and neuropeptide Y (NPY)+ interneurons: those expressing somatostatin are thought to derive from the MGE, and those that don’t express somatostatin are thought to derive from the CGE (Butt et al., 2005; Pleasure et al., 2000; Wonders and Anderson, 2006; Xu et al., 2004). Furthermore, there are temporal constraints that correlate with the production of specific types of interneurons and the layers that they populate (Miyoshi et al., 2007; Rymar and Sadikot, 2007).
Flames et al. (2007), using transplantation of domains within the MGE, found that its dorsal and ventral subdivisions produce different ratios of cortical interneuron subtypes. They correlate this with molecular features of the MGE to provide evidence for a transcription factor code that participates in generating distinct interneuron subtypes. Fogarty et al. (2007), using several lines of mice that express Cre in different patterns in the embryonic basal ganglia, provided evidence for the origins of different types of cortical interneurons. Ghanem et al. (2007) established that two different enhancers of the Dlx1&2locus (i12b and URE2) are expressed in distinct subsets of cortical interneurons. Differences in the expression of i12b and URE2 in the subpallium suggest why they are expressed in different subtypes of interneurons.
Here we have explored pallial interneuron development by studying the expression and function of Lhx6. Lhx6 expression begins in scattered cells in the ventricular zone of the MGE and POA, becomes broadly expressed in the subventricular zone, and is maintained in many projection neurons and interneurons (Alifragis et al., 2004; Choi et al., 2005; Flames et al., 2007; Garcia-Lopez et al., 2008; Grigoriou et al., 1998; Lavdas et al., 1999; Liodis et al., 2007; Marin et al., 2000; Sussel et al., 1999). Thus, subsets of projection neurons are Lhx6+ in the globus pallidus, ventral pallidum, bed nucleus stria terminals, medial amygdala and septum. The vast majority of MGE-derived interneurons that migrate to the pallium, and perhaps striatum, are also Lhx6+. Mature cortical interneurons (principally of the parvalbumin+ and somatostatin+ subtypes) express Lhx6 (Cobos et al., 2005; Cobos et al., 2006; Fogarty et al., 2007; Liodis et al., 2007).
Molecular regulation of interneuron specification, differentiation and migration has begun to be elucidated (reviewed in: Flames and Marin, 2005; Marin and Rubenstein, 2003; Wonders and Anderson, 2006). Lhx6 expression in the MGE is dependent on the function of the Nkx2.1 homeobox gene (Sussel et al., 1999). Nkx2.1 neonatal mutants lack somatostatin+ and NPY+ interneurons (Pleasure et al., 2000). Previous studies implicated Lhx6 function in tangential migration of immature interneurons (Alifragis et al., 2004; Liodis et al., 2007) and in the development of parvalbumin+, somatostatin+ and calretinin+ interneurons (Liodis et al., 2007).
Lhx6 expression in the subpallium is maintained in mice lacking either the Dlx1&2 or the Arx homeobox genes (Colombo et al., 2007; Marin et al., 2000; Petryniak et al., 2007). In both mutants interneuron migration to the cortex is greatly diminished, demonstrating that Lhx6 is not sufficient for migration (Anderson et al., 1997; Colombo et al., 2007; Kitamura et al., 2002). Interneuron migration is disrupted in the Dlx1&2 mutants for several reasons including: 1) immature interneurons over-express the Pak3 kinase, which contributes to premature elaboration of neurites and reduced migration (Cobos et al., 2007); 2) reduced expression of cytokine receptors, CXCR4 and RDC1 (CXCR7) (Long et al., submitted) CXCR4 function is required for the correct laminar positioning of developing interneurons (Li et al., 2008; López-Bendito et al., 2008; Stumm et al., 2007; Stumm et al., 2003) 3) Molecular changes in the MGE result in reduced Arx and Lhx6 expression (Cobos et al., 2005; Petryniak et al., 2007); and increased Olig2 expression – the latter leads to increased numbers of oligodendrocytes (Petryniak et al., 2007).
Immature interneurons are repelled from ventral migration by an unknown factor in the basal telencephalon (Marin et al., 2003). Conversely, they are attracted to the cortex (Marin et al., 2003; Wichterle et al., 2003), in part by neuregulin1/ErbB4 signaling (Flames et al., 2004). En route, semaphorin/neuropilins repulsion is implicated in targeting the interneurons to their appropriate targets (Marin et al., 2001; Tamamaki et al., 2003). Interneuron movement is stimulated by receptor tyrosine kinase signaling through trkB/BDNF and cMet/HGF (Polleux et al., 2002; Powell et al., 2001). BDNF regulates the distribution of marginal zone interneurons (Alcantara et al., 2006). Mice lacking either GDNF or GFR-alpha-l have reduced numbers of cortical interneurons (Pozas and Ibanez, 2005). Finally, there is evidence that Robo1 contributes to interneuron migration (Andrews et al., 2006), while analysis of Slit1&2 single and double mutants failed to identify a clear defect in tangential migration (Marin et al., 2003).
Herein, we explore the roles of Lhx6 in regulating cortical and hippocampal interneuron development. Previous studies have implicated Lhx6 in promoting interneuron tangential migration (Alifragis et al., 2004; Liodis et al., 2007), and in the development of interneurons expressing parvalbumin, somatostatin and calretinin (Liodis et al., 2007). Here, we have taken advantage of an Lhx6 loss-of-function allele that expresses placental alkaline phosphatase (PLAP) (Choi et al., 2005) to follow the development and fate of Lhx6-expressing interneurons in mice lacking this transcription factor. This analysis demonstrates that Lhx6−/− interneurons do migrate to the cortex, although less efficiently. However, they exhibit defects in populating the marginal zone and superficial parts of the neocortical plate. In addition, they lack expression of several markers normally expressed by cortical interneurons. Further analysis demonstrates that there are two subtypes of tangentially migrating interneurons, one Lhx6+;Dlx+ and another which is only Dlx+, suggesting that some interneurons are not dependent on Lhx6 function. By contrast, migration to superficial parts of the prenatal hippocampus does not appear to be seriously affected. While previous studies suggest that development of hippocampal interneurons follow the same general rules as for neocortical interneurons (Pleasure et al., 2000), there have been hints for differences in their development. For instance, although neonatal Dlx1/2−/− mutants lack hippocampal interneurons, their neocortex still has ~20% normal numbers (Anderson et al., 1997; Pleasure et al., 2000). As described herein, compared to Dlx1/2−/− mutants, loss of Lhx6 function has a complementary change in the relative distribution of interneurons between the neocortex and hippocampus.
Furthermore, whereas parvalbumin and somatostatin subtypes do not differentiate in the Lhx6 mutant, NPY+ interneurons are present. We suggest that late-born NPY+ interneurons are derived from the Lhx6−;Dlx+ subtype and provide evidence that Lhx6 mediates these effects through promoting expression of receptors that regulate interneuron migration (ErbB4, CXCR4, CXCR7), and through promoting the expression transcription factors either known (Arx) or implicated (bMaf, Cux2 and NPAS1) in controlling interneuron development.
Lhx6PLAP were provided by Regeneron and genotyped as per Choi et al., 2005. Prenatal studies were performed on a C57BL/6J background (> 10 generations); postnatal studies on a CF1 background (1 generation), which may promote greater postnatal survival. Animals were treated in accordance with the protocols approved by the NICHD and UCSF animal use committee.
Embryos were immersion fixed in 4% paraformaldehyde (PFA) in phosphate buffered saline pH 7.4 (PBS) for 4–12 hours; postnatal animals were fixed by cardiac PFA perfusion; brains were post-fixed overnight, cryoprotected in a gradient of sucrose to 30%, frozen in embedding medium (OCT, Tissue-Tek, Torrance, CA) and cut using a cryostat.
In situ hybridization experiments (Cobos et al., 2007) were performed using digoxigenin riboprobes on 20 μm frozen sections. We thank the following people for cDNAs: Drs. Brian Condie (GAD67), Cary Lai (ErbB4), Sabine Cordes (bMAF), Thomas Jessell (ER81), Harold Cremer (Cux2), Vassilis Pachnis (Lhx6 and Lhx7), Kunio Kitamura (Arx), Thomas Lufkin (SS), Andy McMahon (Shh), Steve McKnight (NPAS1), Sam Pleasure (CXCR4), TH Rabbits (LMO3/RBTN3), ATCC (NPY and RDC1) and Image (Thrombospondin). The Dlx1 and Nkx2.1 plasmids were generated in the Rubenstein lab.
Immunofluorescence and immunoperoxidase stainings were performed per Zhao et al. (2003); cryostat sections were 20 μm on slides (for embryonic ages), or 40 μm free-floating (for postnatal ages). Antibodies: rat anti-somatostatin (Chemicon, MAB354, 1/150); goat anti-choline acetyltransferase (Chemicon, AB144P, 1/250), rabbit anti-calretinin (Chemicon, AB5054, 1:2500) or (Swant, 1/2000); rabbit anti-NPY (Sigma, N9528, 1/5000); mouse anti-parvalbumin (Sigma, P3088, 1/2000); rat anti-CTIP2 (abcam, 1/1000); rabbit anti pan-DLX (kindly provided by Grace Boekhoff-Falk; Panganiban et al., 1995), 1/250), guinea pig anti-DLX2 (kindly provided by Kazuaki Yoshikawa; Kuwajima et al., 2006), 1/2000); chicken anti-GFP (1/2000, Aves labs), rabbit anti-NPY (Immunostar, 1/2000); rabbit anti-PLAP (Serotec, 1/50); rabbit anti-TBR1 (kindly provided by Robert Hevner, 1/1000); rat anti-BrdU abcam (ab6326; 1/5000); rabbit anti-NPAS1 (kindly provided by Steve McKnight, 1/1000; Erbel-Sieler et al., 2004). Secondary antibodies were: Alexa 488 goat anti-rabbit, Alexa 488 goat anti-chicken, Alexa 594 goat anti-rat, Alexa 594 goat anti-guinea pig. Immunoperoxidase staining was performed using the ABC elite or M.O.M. kit (Vector Laboratory).
Paraffin sections (4 μm thick) were treated with an antigen retrieval solution (buffer A, Electron Microscopy Sciences) using a PickCell 2100-Retriever (Electron Microscopy Sciences), for phospho-histone H3 (rabbit; Upstate, #06570, 1/500) and caspase-3 (rabbit; Cell Signaling, #9661, 1/200 to 1/400).
The calbindin D28k antibody was characterized by Western blot. The antibody reacts specifically to a single band at 28 kDa on Western blot of rat hippocampal cells (Kim et al., 2006). Immuno-staining of mouse embryo brain sections produces a specific staining pattern of calbindin positive neurons identical to that published previously (Jacobowitz & Abbott, 1997, Chemoarchitectonic Atlas of the Developing Mouse Brain, CRC press).
On immunoblot prepared from nuclear extract from Jurkat cells immunoprecipitated with anti-Sir2 antibody, the antibody detects 2 bands representing CTIP2 at about 120kD. Immunohistochemistry show expression patterns that are indistinguishable from the RNA expression patterns.
As described recently (Xu et al., 2008), the calretinin antibody reacts to a 32 kDa band on Western blot of rat brain specific for calretinin. Immuno-staining shows a pattern of cellular morphology and distribution in the cortex consistent with published reports (Fonseca et al., 1995).
This antibody reacts specifically with calretinin in tissue originating from human, monkey, rat and mouse. This antiserum does not cross-react with calbindin D-28k or other known calcium binding proteins, as determined by its distribution in the brain, as well as by immunoblots using brain sections. (Schwaller et al., 1994).
The antibody to cleaved caspase-3 was characterized by the manufacturer on Western blot. It detects the large fragment (17/19 kDa) of activated caspase-3 resulting from cleavage adjacent to Asp 17 in apoptotic C6, NIH/3T3, and Jurkat cells. It does not recognize full-length caspase-3 or other cleaved caspases. Immuno-staining detects apoptotic cells both in culture and in tissue sections. The specificity of the staining has been confirmed by the use of blocking peptide.
As described recently (Xu et al., 2008), the antibody to ChAT reacts to a single band of 68–70 kDa on Western blot of rat brain. Immuno-staining shows a pattern of cellular morphology and distribution in the basal telencephalon consistent with previous reports (Corcoran et al., 2004; Wozniak et al., 1989).
DLX2 Immunoreactivity disappears in Dlx1/2 mutant brain section (Long et al., 2007). In Dlx1/2 mutants, a deletion removes most of the coding exons for both Dlx1 and Dlx2 – but does not remove exon 1 – thus an N-terminal truncated protein could be produced – however, since the guinea pig anti-DLX2 antibody is directed to the N-terminal amino acids (1–154), we conclude that there is very little of this protein is present in the Dlx1/2 mutant brain.
pan-DLX immunohistochemistry to invertebrate tissues show expression patterns that are indistinguishable from the RNA expression patterns (Panganiban et al., 1995). Immunohistochemistry to vertebrate tissues show expression patterns that are indistinguishable from the sum of the Dlx1,2,5,6 RNA expression patterns (Stuhmer et al., 2002; Stuhmer et al., 2002). In this paper, Supplemental figure 2 shows a co-labeling experiment using the DLX2 and pan-DLX antibodies on mouse brain section. This data strongly suggests that all the DLX2+ cells are also pan-DLX+ and vice versa.
According to the manufacturer, this anti-GFP antibody is suited for detection of native GFP, GFP variants, and most GFP fusion proteins by western blot and immunohistochemistry analysis using transgenic mice expressing the GFP gene product. No staining was detected on brain sections from wild type mice.
The antibody was analyzed by western blot analysis and immunohistochemistry using transgenic mice expressing the GFP gene product (according to the manufacturer). No staining was detected on brain sections from wild type mice.
Lack of staining on NPAS1 knockout mice brain sections (The NPAS1 mutant mouse was generated by inserting the E. coli β-galactosidase gene into NPAS1 gene which may produce a peptide composed of NPAS1’s N-terminal 41 amino acids fused to β-galactosidase) (Erbel-Sieler et al., 2004).
The NPY antibody was analyzed by the manufacturer using dot blot. It shows that the antibody reacts with NPY, with no cross-reactivity with substance P, neurokinin A, neurokinin B, vasoactive intestinal peptide (VIP), calcitonin gene related peptide, calcitonin, and somatostatin. Immunostaining labels NPY-containing neurons that can be blocked by pre-incubation of the antibody with NPY.
This antibody was analyzed by the manufacturerusing the Biotin-Streptavidin/HRP detection method on rat brain tissue. All staining was blocked by pre-absorption of the diluted antiserum with excess NPY (10-6 M). Cross reactivity experiments in which diluted NPY antiserum was absorbed with excess (10-6M) peptide YY, avian pancreatic polypeptide, B-endorphin, vasoactive intestinal peptide, cholecystokinin or somatostatin showed no affect in blocking the intensity of staining.
As previously described (Pantazopoulos et al., 2006), the parvalbumin antibody detects a single 12 kDa band on Western blot of rabbit skeletal muscle corresponding to parvalbumin. It does not cross-react with other members of the EF-hand family of calcium binding proteins. Immunohistochemistry shows a pattern of staining in the cortex consistent with that reported previously (Liodis et al., 2007; Xu et al., 2008).
The antibody to phospho-histone H3 was characterized by the manufacturer with Western blot on acid extracts of colcemid treated Hela cells. It shows that the antibody detects a single band at 17 kDa corresponding to phospho-histone H3. Immunohistochemistry shows that the antibody labels mitotic cells.
No staining was detected on brain sections from wild type mice.
As described recently (Xu et al., 2006), the specificity of the antibody to somatostatin has been confirmed by blocking of its immunoreactivity by synthetic somatostatin peptide.
Lack of staining on Tbr1 knockout brain section and by western blot using brain protein extract and TBR1 antibody immunostaining on brain sections match in situ hybridization data (Englund et al., 2005).
Brightfield images (PLAP staining and in situ hybridization) were taken using SZX7 microscope (Olympus, Japan) and DP70 camera (Olympus, Japan). 25x magnification confocal images were taken using a Zeiss LSM 510 confocal microscope (pinhole: 2 μm). Images were manipulated with Photoshop CS2 software (Adobe). Contrast and brightness of these images were adjusted for a better visualization.
We used the counting approach described in Guillery, 2002. Tissue sections were 20 μm in thickness. We only counted cells with a clear nucleus (for Table 3); when it was not possible to clearly discern the nucleus, we counted cell somata (for Table 4: counting hippocampal neurons and striatal interneurons). We measured the size of the nucleus or somata to determine whether or not there was difference between genotypes (see sizes in Supplemental Method). As no size difference was noted, we did not apply the Abercrombie correction factor. Within the neocortex and hippocampus, we counted cells within a grid (see below). For the hippocampus and striatum we determined that the Lhx6 mutation did not result in a discernable change in the size of these structures. (See supplemental method).
To count cortical cells positive for DLX2 and or PLAP at E14.5 and E18.5, we used the confocal images shown in Figure 1; boxes were drawn to delimit the approximate positions of the MZ, CP, IZ/SVZ (for E14.5 and MZ), and layer II–IV, layer V, layer VI, IZ and SVZ (for E18.5; aided by the expression of CR for MZ, CTIP2 for layer V and TBR1 for layer VI; see Fig. 3). The area for each box was as follow:
E14.5: MZ: 6115 μm2; CP: 16163 μm2; IZ/SVZ: 19657 μm2. E18.5: MZ: 4238 μm2; layer II/IV: 8150 μm2; layer V: 19071 μm2; layer VI: 20864 μm2; IZ: 15322 μm2; SVZ: 14833 μm2.
To count E18.5 PLAP+ cortical cells, we used the images in Figure 2; boxes were drawn to delimit the MZ, layer II/IV, layer V, layer VI, IZ and SVZ, aided by the expression of CR for MZ, CTIP2 for layer V and TBR1 for layer VI; see Fig. 3. We counted three different dorso-ventral positions (see boxes in Fig. 2C and 2D). The area for each box was: MZ: 6953 μm2; layer II/IV: 11821 μm2; layer V: 15993 μm2; layer VI: 26772 μm2; IZ: 21904 μm2; SVZ: 15298 μm2. The same boxes were used for the same layer in the three different regions.
To count the E18.5 cortical cells positive for pan-DLX and or GFP (Lhx6-BAC-GFP), we used the confocal images in Supplemental Fig. 1; boxes were drawn to delimit the approximate positions of the MZ, CP, IZ and SVZ. The area for each box was: MZ: 16463 μm2; CP: 105557 μm2; IZ: 60952 μm2; SVZ: 47773 μm2. Numbers shown in Table 3, Table 4 and supplemental Table 1 are the areal density (number of positive cells/surface area of the box μm2)). The values were multiplied by 1000 to make it easier to read.
To count cortical cells labeled by in situ hybridization, we drew a box on images of the cortex of P17 animals; the box extended from the ventricular zone to the pial surface. Three horizontal lines divided the box into three strata; these were drawn at approximately 9%, 53% and 85% from the pial surface. We counted cells in these boxes, and estimate that they correspond to different cortical layers: 0–9%: ~layer I; 9%–53% ~layers II–III; 53%–85% ~layers IV–VI.
To count striatal interneurons positive for SOM, PV, NPY, or ChAT, three successive coronal sections, 40 μm thick and 240 μm apart, were selected from a region anterior to the crossing of the anterior commissure. The entire striatum on both sides of these sections was photographed. Positively labeled neurons were manually counted. The numbers for the striatum on each side of these sections were then averaged. The counting was performed exactly the same for both control and Lhx6 homozygous mutant animals. The differences between these two were statistically analyzed using the student’s t-test.
Lhx6 expression in the adult neocortex primarily marks parvalbumin+ and somatostatin+ interneurons (Cobos et al., 2005; Cobos et al., 2006; Liodis et al., 2007). We compared Lhx6 and Dlx expression in the prenatal mouse cortex; we chose Dlx expression, because most (perhaps all) immature cortical interneurons express these genes (Anderson et al., 1999; Anderson et al., 1997; Cobos et al., 2007; Cobos et al., 2005; Cobos et al., 2006; Stuhmer et al., 2002b). We studied the co-expression of DLX2 and all DLX proteins (pan-distaless antibody) with marker genes expressed from the Lhx6 locus at E14.5 and E18.5 (Fig. 1; Supplemental Fig. 1). We identified Lhx6+ cells using either PLAP expressed by the endogenous Lhx6 locus (the heterozygote mouse has only one functional Lhx6 allele; (Choi et al., 2005), or green fluorescent protein (GFP) expressed from a BAC transgenic mouse (Lhx6 function is believed to be inactivated in this allele; Cobos et al., 2006; Gong et al., 2003).
We compared the numbers of Dlx+ and Lhx6+ cells in different layers of the prenatal neocortex at E14.5 and E18.5 (Table 3; Supplemental Table 1). Double-immunofluorescence labeling identified two types of cells: Dlx+;Lhx6− and Dlx+;Lhx6+ using both the Lhx6PLAP and Lhx6-GFP mice; we did not unequivocally identify Dlx−;Lhx6+ cells. Between E14.5–E18.5, there is an increase in Dlx+;Lhx6−, particularly in the subventricular zone (Fig. 1A″, C″). To examine whether Dlx+;Lhx6− cells contribute to distinct interneuron subtypes, we studied the phenotype of mice lacking Lhx6..
In the Lhx6PLAP allele, PLAP replaces part of exon 2 and all of exons 3, 4, 5, and 6; this deleted region encodes amino acid residues 25 to 264, which includes a large part of the N-terminus, the entire LIM1 and LIM2 domains, and 49 out of the 62 amino acid residues of the homeodomain (Choi et al., 2005). The increased intensity of PLAP expression in the Lhx6PLAP/PLAP mutants, relative to the Lhx6PLAP/+, could primarily reflect PLAP dosage.
PLAP labels the MGE progenitors and radially migrating cells that become pallidal projection neurons and tangentially migrating cells, which contribute to the population of cortical interneurons (Fig. 2; Supplemental Figs. 3, 4). Lhx6PLAP/PLAP mice express PLAP in the MGE progenitor domains, its mantle (pallidum) and in cells tangentially migrating to the cortex at E12.5, E14.5, E16.5 and E18.5 (Fig. 2; Supplemental Figs. 3, 4). This demonstrates that cells lacking Lhx6 expression can tangentially migrate to the cortex. However, both the superficial (marginal zone) and deep (intermediate/subventricular zone) E14.5 cortical tangential migrations in the Lhx6PLAP/PLAP mutants do not advance as far dorsally (particularly in rostral regions) (Fig. 2B, B′; Supplemental Fig. 3). The PLAP+ mutant cells continue to express DLX2 (Fig. 1B–B″).
At E16.5 and E18.5, the deficit in the mutant’s deep tangential migration has recovered, whereas the marginal zone and neocortical plate defect persists (Fig. 1C′, D′; Fig. 2C′, D′; Supplemental Fig. 4). Those interneurons that remain in the superficial layers are DLX2+;PLAP− (Fig. 1D–D). There are very few PLAP+ cells in dorsomedial parts of the neocortex. Surprisingly, the hippocampal primordium does have PLAP+ cells in the marginal zone suggesting that neocortical and the hippocampal interneurons have different dependence on Lhx6 function (Fig. 2; Supplemental Fig. 4).
We compared the laminar distribution of PLAP+ cells in the neocortex of Lhx6+/PLAP and Lhx6PLAP/PLAP mice at E18.5 (Table 4), using the following markers: Tbr1 (subplate and layer VI), CTIP2 (layer V), and calretinin (marginal zone, subplate and intermediate zone) (Fig 3). The phenotype varied as a function of dorsoventral position, although in all regions the mutant had ~10-fold fewer PLAP+ cells in the marginal zone (MZ) and presumptive layers II–IV (Table 4). This phenotype was most severe in the dorsal neocortex. Layer V also had fewer PLAP+ cells, but the reduction was ~2–3-fold. On the other hand, presumptive layer VI, subplate, intermediate zone and subventricular zone had increased PLAP+ cells.
Lhx6 is expressed in the progenitors in the MGE, and neurons in the globus pallidus, bed nucleus stria terminals and the ventral pallium (Flames et al., 2007; Garcia-Lopez et al., 2008; Grigoriou et al., 1998; Lavdas et al., 1999; Liodis et al., 2007; Marin et al., 2000; Sussel et al., 1999). Based on the PLAP expression, MGE development appears to be preserved in the Lhx6PLAP/PLAP mutants (Figs. 2, 4A–B; Supplemental Figs. 3, 4). Proliferation in the MGE appears normal based on the number of phospho-histone 3+ (M-phase) cells at E12.5, E14.5, E16.5 and E18.5 (Supplemental Fig. 18). In addition, we did not detect increased levels of prenatal apoptosis (caspase-3+ cells) (Supplemental Fig. 16).
To obtain molecular information on whether loss of Lhx6 function alters the development of pallidal neurons, we compared gene expression from Lhx6PLAP/PLAP and Lhx6PLAP/+ E12.5, E14.5, E16.5 and E18.5 brains. We studied MGE marker expression using in situ RNA hybridization for: Arx, bMaf, Cux2, CXCR4, Dlx1, ER81, ErbB4, GAD1 (GAD67), Lhx6, Lhx7, Lmo3, Nkx2.1, NPAS1, NPY, RDC1, somatostatin and thrombospondin (Figs. 4, ,5;5; Supplemental Figs. 5, 6, 7, 8).
Expression of Lhx6 RNA and PLAP activity are maintained in the SVZ domain of the MGE in the Lhx6PLAP/PLAP mutants at E12.5, E14.5, E16.5 and E18.5 (Fig. 4A–D′, 5A–D′; Supplemental Figs. 5, 7, 8). However, by E14.5, in the MZ of the Lhx6PLAP/PLAP MGE, PLAP activity is much easier to detect than Lhx6 RNA, by contrast to the Lhx6PLAP/+, where Lhx6 RNA and PLAP expression are both strong (Fig. 4B, B′, D, D′). This suggests that Lhx6 RNA expression fails to be maintained in most of the postmitotic cells that migrate from the progenitor domains into the MZ, where the neurons mature. The residual expression of Lhx6 RNA overlaps with expression of Lhx7 and Nkx2.1 (Supplemental Fig. 5), suggesting that these TFs preserve Lhx6 expression in progenitor cells.
Lack of Lhx6 function leads to reduced MGE and pallidal expression of three transcription factors (TFs) Cux2, bMaf and NPAS1 at E12.5 and E14.5 (arrows; Fig. 4Q–R′, U–V′; Supplemental Fig. 5). Cux2 and bMaf expression become restricted to a small domain in the dorsal MGE; bMaf expression is nearly lost from a thin “stream” of SVZ cells that extends from the VZ of the ventral MGE (arrow; Fig. 4Q–R′). NPAS1 expression is reduced in the dorsal MGE (arrow), but is maintained in the ventral MGE and CGE (arrowhead) (Supplemental Fig. 5).
Despite these molecular defects, expression of other TFs in MGE progenitor and postmitotic zones (including the presumptive globus pallidus), appear preserved (Arx, Dlx1, Er81, Lhx7, Lmo3 and Nkx2.1; Fig. 4E–H′; Supplemental Fig. 5).
The differential effect on TF expression suggests that Lhx6 has a selective role in controlling the differentiation of specific sets of effector genes and/or subsets of MGE-derived cells. Consistent with this idea is the maintained MGE/pallidal expression of the ErbB4 receptor tyrosine kinase and the GABA synthesizing enzyme (GAD1) (Fig. 4M–N′; I–J′). However, RNA expression of NPY is virtually eliminated and somatostatin expression is greatly reduced (arrow; Fig. 4S–T′; O–P′). Furthermore, expression of the cytokine receptors CXCR4 and RDC1 (CXCR7, CMKOR1) are greatly reduced (arrow; Fig. 4K–L′; Supplemental Figs. 5, 6). These CXCL12 (SDF1) receptors are implicated in regulating cortical interneuron migration (Li et al., 2008; López-Bendito et al., 2008; Stumm et al., 2007; Stumm et al., 2003; Y. Wang, J. Long, D. Crawford, T. Behrens, and J. Rubenstein, unpublished).
Analysis of PLAP, Arx, Dlx1 and Lhx7 expression in the E16.5 and E18.5 basal ganglia of Lhx6PLAP/PLAP mutants did not identify major anatomical or molecular defects (Fig. 5; Supplemental Figs. 4, 6, 7), except for the periventricular accumulation of PLAP+ cells. This ectopia is located in the dorsal MGE and expresses DLX2 (Supplemental Fig. 9). BrdU birthdating shows that many of these cells had their last mitosis around E13.5 (Supplemental Fig. 9).
To determine the molecular profile of tangentially migrating cells in the Lhx6PLAP/PLAP mutants from E12.5–E18.5, we examined expression of genes known to mark, or regulate, their development (Fig. 4; Supplemental Figs. 5, 6, 7). Cortical expression of Dlx1 and GAD67 in Lhx6PLAP/PLAP mutants at E14.5 was moderately decreased, similar to the reduction in PLAP expression (Fig. 4F, F′, J, J′; see higher magnification in Supplemental Fig. 3, 6). Furthermore, interneuron expression of Arx was greatly reduced and interneuron expression of Cux2, CXCR4, ErbB4, bMaf, NPY, RDC1, somatostatin and thrombospondin were not detectable in the Lhx6PLAP/PLAP mutants (Fig. 4; see higher magnification in Supplemental Fig. 6). Arx encodes a homeobox gene whose function is required for development of cortical interneurons and is implicated in promoting migration (Colombo et al., 2007; Kitamura et al., 2002). In the Lhx6PLAP/PLAP mutants, Arx expression is maintained in the MGE and LGE, thus the Lhx6PLAP/PLAP mutants should not fully phenocopy the Arx mutant.
ErbB4 function is linked to the attraction of interneurons into the cortex (Flames et al., 2004). While ErbB4 expression is maintained in the MGE (arrows; Fig. 4M–N′), ErbB4+ cells fail to spread into the LGE and the cortex (arrowhead; Fig. 4M′, N′; Supplemental Fig. 6). This suggests that Lhx6 is required to maintain ErbB4 expression in the immature interneurons migrating from the MGE en route to the cortex. CXCR4 and RDC1 encode cytokine receptors that transduce CXCL12 (SDF1) signaling (Li et al., 2008; López-Bendito et al., 2008; Stumm et al., 2007; Stumm et al., 2003). CXCL12 (SDF1) is an attractant for Lhx6+ interneurons (Li et al., 2008). CXCR4 regulates the laminar organization of migrating cortical interneurons. RDC1 is expressed in developing cortical interneurons (J. Long and J. Rubenstein, unpublished results). While RDC1 function in migrating cortical interneurons is not known, it regulates migration of lateral line cells in zebrafish (Dambly-Chaudiere et al., 2007; Valentin et al., 2007). CXCR4 and RDC1 expression in the Lhx6PLAP/PLAP mutant MGE are greatly reduced (arrows; Fig. 4K–L′; Supplemental Fig. 6), and expression in tangential streams to the cortex is not detectable (arrowheads; Fig. 4K′, L′; Supplemental Fig. 6).
In sum, loss of Lhx6 function results in important molecular defects in migrating interneurons that can account, at least in part, for reduced tangential migration and abnormal laminar location of those interneurons that reach the cortex. These results suggest that Lhx6 may not have a major function in regulating expression of some markers like Dlx1 and GAD67, but rather it regulates the number and types of interneurons reaching the cortex, and their position within cortical layers and regions.
With advancing age (E16.5 and E18.5), the cortical SVZ of Lhx6PLAP/PLAP mutants showed strong expression of markers of tangentially migrating interneurons (Arx, Dlx1, ErbB4, GAD67) (Fig. 5E–J′, M–N′; Supplemental Figs. 7, 8). This result is consistent with PLAP expression (Figs. 1C′, D′, 2C–D′; Supplemental Fig. 4) and the observation that there are increasing numbers of Dlx+;Lhx6− tangentially migrating interneurons in the cortical SVZ/IZ late in gestation (Fig. 1C–C; Supplemental Fig. 1). To establish the effect of the mutation on interneuron subtypes and their distributions within the cortex and hippocampus, we studied late gestation and early postnatal Lhx6PLAP/PLAP mutants.
At E18.5, PLAP expression shows that cortical interneurons have failed to adequately populate the marginal zone and cortical plate of Lhx6PLAP/PLAP mutants (Fig. 2D–D″; Supplemental Fig. 4). Furthermore, these layers have very low expression of Arx, ErbB4 bMaf, RDC1, and somatostatin (Fig. 5G–H′, K–R′; Supplemental Fig. 8). However, marginal zone and cortical plate expression of Dlx1, GAD67 and NPAS1 are better preserved (Fig. 5E–F′, I–J′, U, U′, V, V′; Fig. 5A, A′, B, B′, C, C′, D, D′, K, K′); The NPAS1+ cells are DLX+ (Supplemental Fig. 19).
Unlike the neocortex, many molecular features of hippocampal interneurons are preserved in the E18.5 Lhx6PLAP/PLAP mutants. Interneurons, in the superficial layers of the CA fields, maintain robust expression of PLAP, Arx, Dlx1, GAD67 and NPAS1 (Figs. 5, ,7;7; Supplemental Figs. 4, 19), with lower levels of ErbB4, and NPY (Fig. 5 and not shown). On the other hand, hippocampal expression of bMaf, RDC1 and somatostatin was not detected (Fig. 5K, K′, O, O′, Q, Q′; Supplemental Fig. 8).
Neonatal Lhx6PLAP/PLAP mutants were indistinguishable from the heterozygote or wild type littermates for at least one week (based on simple examination of their size, movements and behaviors). By P10 the mutants appeared smaller than the controls, and by P14–P18, they were ~half the size and less active than controls. No Lhx6PLAP/PLAP mutants were detected by weaning at P21.
Analysis of the postnatal neocortex and hippocampus at P11, P15 and P18 showed profound interneuron defects (Figs. 6, ,7;7; Supplemental Figs. 12, 13; and data not shown). As reported by Liodis et al., (2007), parvalbumin immunoreactivity (at P15 and P18) was reduced to just a few cells in deep neocortical layers. Strongly positive somatostatin immunoreactive cells (at P15 and P18) were almost eliminated; weakly positive cells were detectable. Calbindin and calretinin immunoreactivity were greatly reduced (at P15 and P18); dysmorphic calbindin+ cells remained in three layers (Supplemental Figs. 12, 13).
On the other hand, cortical and hippocampal expression of Dlx1 (P18), Dlx2 (P11), GAD67 (P18) and NPY (at P11, P15 and P18) remained strong (Figs. 6 and and7;7; Supplemental Figs. 11, 12, 13); these aspects of the Lhx6 mutant were previously not reported. The number of Dlx1, Dlx2, GAD67 and NPY expressing neocortical cells were increased; this was easiest to quantify using in situ RNA hybridization (Fig. 6; see numbers in the legend to Fig. 6). Some of the NPY+ cells are probably not interneurons because their laminar distribution differs from Dlx1 and GAD67 expression (arrow Fig. 6F′). Double immunofluorescence labeling showed that many NPY+ cells co-express DLX2 in the neocortex and hippocampus, providing evidence that these are interneurons (Fig. 7 and Supplemental Fig. 19). Unfortunately, because postnatal expression of PLAP was too low in both the Lhx6PLAP/+ and Lhx6PLAP/PLAP mice, we could not determine whether the NPY+ interneurons were PLAP+. Thus, while early (E12.5 and E14.5) NPY expression in the MGE and migrating cells is dependent on Lhx6 (Fig. 4), later expression (P11 and P18) in cortical and hippocampal interneurons is Lhx6-independent.
Finally, like Dlx1, NPAS1 expression was maintained in the mutant hippocampus (Fig. 7; Supplemental Fig. 19). In the neocortex, NPAS1 expression was maintained in superficial layers, but was reduced in deep layers (Fig. 5 U, U′, V, V′; Supplemental Fig. 19).
Postnatal analysis of Lhx6PLAP/PLAP mutant basal ganglia showed no overt morphological defects (Fig. 8; Supplemental Figs. 10, 15). The Lhx6PLAP/PLAP mutant continues to express parvalbumin in the globus pallidus (Supplemental Fig. 10). However, there was an abnormal distribution of calretinin, NPY, parvalbumin and somatostatin immunoreactivity in the septum, nucleus accumbens, striatum, BSTN and amygdala (Fig. 8; Supplemental Fig. 10). Some of these defects are due to reduced interneuron numbers (legend to Fig. 8) – for instance, the striatum has a ~2-fold reduction of somatostatin+ and NPY+ interneurons.
Here we provide evidence that Lhx6 is expressed in a large subset of the DLX+ tangentially migrating immature interneurons. At the earliest stages of tangential migration, the vast majority of interneurons are Dlx+;Lhx6+ (Fig. 1; Supplemental Fig. 1). With progressive time, there are increasing numbers of Dlx+;Lhx6− tangentially migrating immature interneurons; this cell type migrates in both the deep and superficial streams, but is predominantly found in the IZ and SVZ. While our analyses do not discern the anatomical origins of Dlx+;Lhx6+ and Dlx+;Lhx6− cells, a parsimonious hypothesis is that the Dlx+;Lhx6+ are derived from the MGE, whereas the Dlx+;Lhx6− cells arise from the caudal LGE – the dorsal CGE (dCGE) (Flames et al., 2007). The CGE is currently implicated in producing late-born interneurons that express calretinin and NPY (Butt et al., 2005; Fogarty et al., 2007; Xu et al., 2004).
Consistent with previous studies, we find that Lhx6 function participates in tangential migration to the cortex. Unlike previous studies, our analyses had the advantage of a cell autonomous marker of the Lhx6 homozygous mutant cells, as PLAP was inserted into the deleted Lhx6 allele (Choi et al., 2005). Unlike Alifragis et al., 2004, who used siRNA to reduce Lhx6 expression, we find that loss of Lhx6 does not block tangential migration to the cortex, except perhaps in a subset of cells that remain as a periventricular ectopia (Supplemental Fig. 9). Our results are similar to Liodis et al., 2007, who studied a different Lhx6 mutant. The reduced number of cortical GAD67+ cells in their Lhx6−/−mutant resembles the Lhx6PLAP/PLAP mutant, particularly in the E14.5 dorsal neocortex. As development proceeds the superficial zone is more affected (Figs. 1, ,2,2, ,3,3, ,4;4; Supplemental Figs. 3, 5, 6, 7, 8). We suggest that the reduced migration is caused in part by reduced ErbB4 expression, a receptor mediating neuregulin1 attraction of tangentially migrating cortical interneurons (Flames et al., 2004).
Lhx6PLAP/PLAP mutant PLAP+ cells are greatly reduced in the marginal zone and superficial layers of the E16.5 and E18.5 cortical plate. However, they are found in increased numbers in layer VI, the intermediate and subventricular zones (Table 4; Figs. 1, ,2,2, ,3).3). These data support a model that Lhx6 regulates the laminar position of the Lhx6PLAP+ interneurons.
On the other hand, DLX2+;Lhx6PLAP− interneurons do not show the deficit in populating superficial layers of the cortex (Fig. 1D; Table 3), providing evidence that the laminar phenotypes of Lhx6PLAP+ cells are due to autonomous molecular deficits in responding to environmental cues. The postnatal cortex (P11, P15 and P18) has interneurons (Dlx1+, Dlx2+ or GAD67+ cells) in all cortical layers (Fig. 6), but with skewing of their laminar distribution (~2-fold increase in deep layers of Dlx1+ or GAD67+ cells; see Fig. 6 legend).
Little is known about how interneurons radially migrate into the cortical plate, although there is evidence that cytokine signaling participates in this process. Mice lacking either the CXCL12 (SDF1) cytokine, or its receptor (CXCR4), show defects in interneuron laminar position (Li et al., 2008; López-Bendito et al., 2008; Stumm et al., 2007; Stumm et al., 2003). In particular, CXCR4 mutants show deficits in interneuron localization to the marginal zone, presumably because CXCR4 mediates their attraction to SDF1 secreted by the meninges.
CXCR4 expression is not detectable in immature cortical interneurons in the Lhx6PLAP/PLAP mutants (Supplemental Figs. 5, 6). Thus, this could account for the decreased number of interneurons in the marginal zone. However the laminar phenotype is more severe than in the CXCR4 mutant. We found that the Lhx6PLAP/PLAP mutants also lack detectable expression of a new SDF1 receptor named RDC1 (CXCR7; CMKOR1) (Figs. 4, ,55)(Balabanian et al., 2005). RDC1 expression is down-regulated in the basal ganglia and cortical interneurons in Dlx1&2−/− mutants (Long et al., submitted). RDC1 (CXCR7) regulates the migration of lateral line cells in zebrafish (Dambly-Chaudiere et al., 2007; Valentin et al., 2007). While the function of this receptor has not been established in the development of cortical interneurons, perhaps the combined reduction in CXCR4 and RDC1 contributes to the lamination defects of the immature interneurons.
While the overall number of cortical PLAP+ cells is reduced in Lhx6PLAP/PLAP mutants, there are increased numbers in the SVZ at E16.5 and E18.5, where Arx and Dlx1 expression is also robust (Table 4; Figs. 1, ,2,2, ,3,3, ,5;5; Supplemental Fig. 3, 6, 7, 8). Furthermore, ErbB4 expression, which is not detectable in the mutant migratory cells at E12.5 and E14.5 (Fig. 4; Supplemental Fig. 5), is present in the SVZ migratory pathway at E16.5 and E18.5 (Fig. 5, ,6,6, ,7).7). Thus, Lhx6 function appears to be less important for interneurons that migrate within the SVZ at E16.5 and E18.5. Note that the wild-type SVZ is enriched in migratory cells that express DLX proteins but not Lhx6 (Fig. 1; Table 3; Supplemental Fig. 1). Thus, we propose that there is a subset of SVZ cells that are independent of Lhx6 function, but require Arx, Dlx and ErbB4 function (Fig. 9). While the fate of these cells are not certain, it is known that late born calretinin+ interneurons are produced by the caudal ganglionic eminence and at least this population is not dependent on Nkx2.1 function (Butt et al., 2005; Fogarty et al., 2007; Xu et al., 2004). Thus, we examined the types of cortical and hippocampal interneurons that are present in the postnatal Lhx6PLAP/PLAP mutant.
Lhx6PLAP/PLAP mutants survive until ~P20; the cause of death is not known. The mutant cortex shows increased apoptosis at P11 and P16 (Supplemental Fig. 17 and not shown), which may contribute to the low numbers of PLAP+ cells. We were not able to determine whether the apoptosis in the cortex was limited to the interneurons that lacked Lhx6 expression because of the low PLAP expression postnatally. Prenatally, we did not find increased apoptosis in the mutant telencephalon (Supplemental Fig. 16).
The Lhx6PLAP/PLAP cortex and hippocampus have severe deficits in parvalbumin+ and somatostatin+ cells (Fig. 6; Supplemental Figs 11, 12, 13), consistent with Liodis et al. (2007), and the expression of Lhx6-GFP in these interneurons (Cobos et al., 2006). There is also a reduction in calretinin+ interneurons in the Lhx6PLAP/PLAP mutants (as did Liodis et al., 2007). This finding was unexpected as previously we observed little co-expression of Lhx6-GFP in these cells (Cobos et al., 2006). Perhaps, Lhx6 is required in progenitors for some calretinin+ interneurons; this is likely as a subset of calretinin+ interneurons also express somatostatin (Xu et al., 2006; Gonchar et al., 2008). Alternatively, the postnatal cortical milieu does not support calretinin expression. Furthermore, given that Lhx6 is expressed in the hypothalamus, it is possible that additional physiological defects could contribute to the mutant’s phenotype.
Unlike the other interneuron subtypes, NPY+ cells were not reduced in the neocortex or hippocampus at later stages (Figs. 6, ,7;7; Supplemental Figs. 11, 12, 13). In fact, the number of cells expressing NPY increases (~4-fold at P18; Fig. 6). However, quantifying the increase has been difficult, because we believe that some of the increase is due to expression in cells other than interneurons. None-the-less, we confirmed that in the hippocampus and superficial layers of the neocortex that most of the NPY+ cells are interneurons, based on co-expression with DLX2 (Fig. 6 and not shown).
The preservation of NPY-expressing interneurons is consistent with the observation that few Lhx6-GFP+ interneurons express NPY (Cobos et al., 2006). Of note, however, NPY expression in the MGE at E12.5 and E14.5 is not detectable in the Lhx6PLAP/PLAP mutants (Fig. 4). This suggests that early NPY expression is dependent on Lhx6, whereas late expression is Lhx6-independent. Cavanagh and Parnavelas (1990) previously presented evidence for two types of NPY+ interneurons; an early born population that decreases postnatally, and a later born population that persists. Our results suggest that the early born population depends on Lhx6, whereas the later-born NPY+ interneurons are largely Lhx6-independent. This conclusion is consistent with fate-mapping suggesting the NPY+ interneurons come from both the MGE and LGE/CGE (Butt et al., 2005; Fogarty et al., 2007).
Striatal NPY+ interneurons show the opposite result as in the cortex; the Lhx6PLAP/PLAP mutants exhibited a decrease in NPY+ cells (Fig. 8; Supplemental Fig. 10). Perhaps this reflects the loss of NPY expression in the MGE at E12.5 and E14.5 (Fig. 4). Furthermore, while parvalbumin and somatostatin expression are essentially eliminated from the pallium, there are only modest decreases in the numbers of striatal interneurons expressing these proteins. These results reflect the specificity of the genetic pathways that control the development of pallial and striatal interneurons (Fig. 9). Furthermore, these results highlight differences between Lhx6 and Lhx7(8) in the genesis of striatal interneurons. While Lhx7(8) mutants have large reduction in subcortical cholinergic neurons (Zhao et al., 2003; Mori et al., 2004; Fragkouli et al., 2005), this cell type appears normal in the Lhx6 mutants (Supplemental Fig. 15). These findings add to earlier studies on the roles of the Dlx1&2, Mash1 and Nkx2.1 transcription factors on striatal interneuron development (Marin et al., 2000).
Lhx6 RNA and PLAP expression is maintained in MGE progenitors of the Lhx6PLAP/PLAP mutants (Figs. 4, ,5;5; Supplemental Figs. 4, 6, 7). Thus, Lhx6 function is not required in these cells to maintain Lhx6 RNA expression. On the other hand, while Lhx6 RNA can be detected in pallidal neurons at E12.5, its expression becomes progressively weak in pallidal neurons at E14.5, E16.5 and E18.5. Furthermore, it is not detected as MGE-derived cells tangentially migrate into the LGE/CGE and cortex at any age, whereas PLAP activity and immunoreactivity is preserved in pallidal neurons and in tangentially migrating interneurons. (Figs. 4, ,5;5; Supplemental Figs. 4, 6, 7). Thus, perhaps Lhx6 is required to maintain its transcription in these cells, and the PLAP activity represents perdurance of the protein.
Lack of Lhx6 function leads to reduced MGE and pallidal expression of three transcription factors, bMaf, Cux2 and NPAS1 (Fig. 4; Supplemental Fig. 5). While the function of these genes has not been established in the MGE/pallidum, they may contribute to the Lhx6PLAP/PLAP phenotype. On the other hand, MGE/pallidal expression of Arx, Dlx1, ER81, Lmo3 and Nkx2.1 appear normal (Fig. 4; Supplemental Fig. 5), and may explain why pallidal nuclei show no obvious phenotype.
As Lhx6PLAP/PLAP interneurons tangentially migrate into the LGE and cortex, they show reduced Arx, bMaf and Cux2 expression (Fig. 4; Supplemental Fig. 6). Constitutive loss of Arx function results in severe reductions in the numbers of cortical interneurons (Colombo et al., 2007; Kitamura et al., 2002). Arx expression appears normal in the MGE/pallidum of the Lhx6PLAP/PLAP mutants; therefore, Arx expression may be reduced as interneurons enter the LGE. It is unknown how reduced Arx levels in tangentially migration interneurons affect their migration.
Dlx1 expression appears less affected in Lhx6PLAP/PLAP mutants. Dlx1&2function are required for tangential migration (Anderson et al., 1997), suggesting why this process is largely normal in Lhx6PLAP/PLAP mutants. Furthermore, Dlx1 regulates the number of calretinin, somatostatin and NPY neurons (Cobos et al., 2005). Perhaps the maintenance of Dlx1 expression is related to the persistence of NPY interneurons.
NPAS1 expression is reduced in deep cortical layers, but is maintained in many cortical and hippocampal interneurons at E18.5 and P11 (Figs. 5, ,7,7, and not shown). In fact, there are ~2-fold more NPAS1+ cells in the mutant P11 hippocampus (Fig. 7). Expression of this transcription factor is reduced in Dlx1−/− mutants (Cobos et al., 2005). Thus, we propose that NPAS1 is a marker of interneurons that are in part derived from the Lhx6-independent lineage that depend on Dlx1 function, and perhaps come primarily from the LGE/dorsal CGE. We also hypothesize that NPAS1 is expressed in NPY+ interneurons; proof of this awaits double-labeling experiments.
In sum, appear to be two major subtypes of immature pallial interneurons that differ in their origins, expression of transcription factors and fate (Fig. 9; Butt et al., 2005; Flames et al., 2007; Fogarty et al., 2007; Liodis et al., 2007; Wonders and Anderson, 2006; Xu et al., 2004). In the absence of Lhx6, some cells fail to migrate out of the MGE; many of these are born around E13.5, and remain as a periventricular ectopia, showing that Lhx6 is essential for migration of a subset of MGE-derived cells (Supplemental Fig. 9). However, most of the PLAP+ MGE-derived cells, which are fated to become pallial interneurons, migrate into the cortex and hippocampus. These cells continue to express PLAP and DLX2 in the cortex and hippocampus at E18.5 (Fig. 1 and not shown); therefore they have the potential to develop along the pathway of Dlx+;Lhx6−interneurons. Postnatally, the vast majority of mutant PLAP+ cells either lose PLAP expression and/or die (Supplemental Figs. 11, 14); thus we cannot follow their fate. We postulate that the increase in NPAS1 expression in the postnatal hippocampus may reflect a fate change of some MGE-derived cells.
Quantification of numbers of cells that are positive for GFP (Lhx6-GFP), pan-DLX and both GFP & pan-DLX immunoreactivity in the E14.5 and E18.5 cortex of Lhx6PLAP/+ and Lhx6PLAP/PLAP, using data from Supplemental Figure 1. The numbers represent the areal density (positive cells per 1000 μm2 of the box surface area) in cortex subregions. Abbreviations: see Table 3.
This work was supported by the research grants to JLRR from: Nina Ireland, Larry L. Hillblom Foundation, NIMH RO1 MH49428-01 and K05 MH065670; PF: Swiss National Science Foundation (PA00A-117463 and PBGEA-112882); HW from Intramural Research Program of the National Institute of Child Health and Human Development. We thank Regeneron for providing the Lhx6 mutant mice.