Abstract
In the developing mammalian basal telencephalon, neural progenitors from the subpallium generate the majority of inhibitory medium spiny neurons (MSNs) in the striatum, while both pallial- and subpallial-derived progenitors contribute to excitatory and inhibitory neuronal diversity in the amygdala. Using a combination of approaches, including genetic fate mapping, cell birth dating, cell migration assays, and electrophysiology, we find that cells derived from the Emx1 lineage contribute to two distinct neuronal populations in the mature basal forebrain: inhibitory MSNs in the striatum and functionally distinct subclasses of excitatory neurons in the amygdala. Our cell birth-dating studies reveal that these two populations are born at different times during early neurogenesis, with the amygdala population born before the MSNs. In the striatum, Emx1-lineage neurons represent a unique subpopulation of MSNs: they are disproportionately localized to the dorsal striatum, are found in dopamine receiving, reelin-positive patches, and are born throughout striatal neurogenesis. In addition, our data suggest that a subpopulation of these Emx1-lineage cells originate in the pallium and subsequently migrate to the developing striatum and amygdala. Our intersectional fate-mapping analysis further reveals that Emx1-lineage cells that coexpress Dlx exclusively generate MSNs but do not contribute to the excitatory neurons in the amygdala. Thus, both the timing of neurogenesis and differential combinatorial gene expression appear to be key determinants of striatal versus amygdala fate decisions of Emx1-lineage cells.
Introduction
Neuronal diversity in the mature telencephalon is generated during embryogenesis, in which progenitor pools are formed from the regional gene expression of key transcription factors. Multiple ventral telencephalic (subpallial) progenitor pools contribute to the developing striatum, the inhibitory center of the telencephalon vital for movement and impulse control (Wichterle et al., 1999, 2001; Marin et al., 2000; Nery et al., 2002). The majority (90–95%) of the striatum is composed of GABAergic medium spiny neurons (MSNs) (Gerfen, 1992), which are derived primarily from the lateral ganglionic eminence (LGE) (Wichterle et al., 1999, 2001). These cells migrate radially and separate into two primary striatal compartments: the patches, which are formed from MSN precursors, the majority of which are born early [in the mouse between embryonic day 12.5 (E12.5) and E13.5]; and the surrounding matrix, the majority of which are born later (E13.5–E15.5) (Fishell and van der Kooy, 1987, 1991; Gerfen, 1992; Krushel et al., 1993; Mason et al., 2005).
Another major basal telencephalic structure is the amygdala, a key component of the limbic system. The amygdala plays an important role in modulating fear, aggression, and emotionality (Swanson and Petrovich, 1998; Sah et al., 2003; Maren and Quirk, 2004). Similar to the striatum, multiple progenitor pools contribute to the amygdala, and several lines of evidence indicate that this structure is derived from dorsal (pallial) and ventral (subpallial) progenitors, at least some of which are born early in embryonic neurogenesis (Nery et al., 2002; Carney et al., 2006; Remedios et al., 2007; Bai et al., 2008; Hirata et al., 2009; Soma et al., 2009).
Previous work has revealed that progenitor populations marked by the pallial-expressed transcription factor, Emx1, appear to contribute to numerous telencephalic structures, including the striatum (Willaime-Morawek et al., 2006; Willaime-Morawek and van der Kooy, 2008), olfactory bulb (Fogarty et al., 2007; Kohwi et al., 2007; Young et al., 2007), and amygdala (Puelles et al., 2000; Gorski et al., 2002; Medina et al., 2004; Tole et al., 2005). However, the relationship between the origin and timing of the generation of Emx1+ progenitors and their ultimate fate in the mature striatum and amygdala remains unknown.
In this study, we used a multidisciplinary approach to investigate key developmental aspects of the striatum and amygdala, including the genetic heterogeneity and origin of the Emx1 lineage, the timing of their genesis, and their ultimate fate in these two structures. Our data reveal a relationship between the timing of the generation and the postnatal fate of Emx1-lineage progenitors in the striatum and amygdala. By use of an intersectional genetic fate-mapping strategy, our data also reveal that, in addition to timing, differential combinatorial gene expression within Emx1-lineage cells may be an important determinant of their distinct fates in the striatum versus the amygdala. These findings provide novel insight into the plasticity of a subset of telencephalic neural progenitors from a broadly common lineage, wherein their ultimate neuronal fate in the postnatal brain appears to be related to both the timing of their birth and combinatorial gene expression during embryogenesis.
Materials and Methods
Animal use.
Swiss Webster (Taconic Farms), Z/EG, ROSA-YFP, ROSA-LacZ (Jackson Laboratory) (Novak et al., 2000; Srinivas et al., 2001), Emx1-Cre (K. Jones, University of Colorado, Boulder, CO) (Gorski et al., 2002), and Small eye (Sey) (Hill et al., 1991) mice used in these studies were maintained according to the protocols approved by Children's National Medical Center and Georgetown University Medical Center. Emx1-Cre, ROSA-YFP, and Sey+/−; Emx1-Cre mice were maintained on a mixed C57BL/6 × SW background; Z/EG mice were maintained on an SW background. For the intersectional fate-mapping studies (see Fig. 7) Emx1-Cre, Dlx5/6-Flpe, and RCE: dual mice were maintained according to the protocols approved by the New York University School of Medicine, NY. For these experiments, the neo cassette from Emx1-Cre driver mice (T. Iwasato, BSI RIKEN, Japan) was removed before crossing. Dlx5/6-Flpe driver mice (G. Miyoshi and G. Fishell, unpublished observations) expressed the Flpe site-specific recombinase under control of the intergenic enhancer residing between Dlx5 and Dlx6 (Stenman et al., 2003). To generate RCE: dual reporter mice, a CAG promoter with floxed and flrted stop cassettes followed by EGFP reporter was targeted to the R26R locus (Sousa et al., 2009). For staging of the embryos, midday of vaginal plug detection was considered as E0.5. For postnatal animals, the day of birth was considered as postnatal day 0 (P0). The genotyping of animals was performed as described previously (Gorski et al., 2002; Carney et al., 2009).
Tissue preparation and histology.
For immunofluorescence and in situ hybridization at embryonic ages, brains were fixed in 4% paraformaldehyde (PFA) for 2 h and overnight, respectively. Brains were cryoprotected by sucrose immersion, embedded in Histoprep (Fisher Scientific), and frozen. Serial coronal sections of embedded tissue were cut at 20–30 μm thickness using a cryostat and mounted on glass slides.
Immunohistochemistry.
Cryostat mounted sections were air dried and rinsed three times in PBS before blocking for 1 h in 10% normal donkey serum diluted in PBS with 0.2% Triton to prevent nonspecific binding. Primary antibodies were diluted in 1% serum diluted in PBS with 0.2% Triton; sections were incubated in primary antibody overnight at 4°C. The primary antibodies used were as follows: goat anti-Pax6 (1:200), mouse anti-NeuN (1:500; Covance), rabbit anti-Ng2 (1:400, Covance), mouse anti-GFAP (1:200, Sigma), rabbit anti-DARPP-32 (1:1000, Covance), rat anti-BrdU (1:250, Serotec), mouse BrdU (1:50, Sigma), rat anti-GFP (Nacala, 1:2000), rabbit anti-Gsx2 (1:1500, gift from K. Campbell, Cincinnati Children's Hospital, Cincinnati, OH), mouse anti-CC1 (1:50, Calbiochem), rabbit anti-Tbr1 (1:1000; gift from R. Hevner, University of Washington, Seattle, WA), guinea pig anti-Dlx2 (1:1500; gift from K. Yoshikawa, Osaka University, Osaka, Japan), rat anti-CTIP2 (1.300; Abcam), rabbit anti-tyrosine hydroxylase (anti-TH) (1:1000; Millipore Bioscience Research Reagents), mouse anti-reelin (1:1000, Millipore Bioscience Research Reagents), rabbit anti-μ-opioid receptor-β (anti-MOR1-β) (1:10,000; Millipore), and rat anti-somatostatin (ant-SST) (1:1000; Millipore). To detect primary antibodies, secondary antibodies raised in mouse, rat, goat, guinea pig, and rabbit were used (Cy3 and Cy5 at 1:200, FITC at 1:50; all from Jackson Immunoresearch). Sections were incubated for 2 h in 1% serum in PBS with 0.2% Triton and were washed and coverslipped with gel mount (Sigma) or Vectashield with DAPI (Vector Labs). Sections coverslipped with gel mount were incubated first with To-Pro-3 iodide (1:100, Invitrogen) for 10 min and then washed and coverslipped.
BrdU labeling.
To label cells in S phase, E9.5, E11.5, E13.5, E14.5, and E15.5 Emx1-Cre; ROSA-YFP pregnant dams were administered BrdU (Sigma), which was dissolved in PBS, by intraperitoneal injection at a dose of 100 mg/kg. To achieve short-term labeling, the dams were killed after 2 h and embryos were prepared for processing as described above. Cryostat sections were processed for immunohistochemical labeling of YFP using rat anti-GFP, as described above, and postfixed with 4% PFA for 15 min. A sodium citrate pretreatment was performed before processing sections for BrdU immunohistochemistry, as described previously (Tang et al., 2007). Briefly, a food steamer (Oster 6.1 quart model) was preheated to 100°C, and slides were placed horizontally in the incubator, covered with 500 μl of 10 mm sodium citrate, and incubated for 15 min. Slides were allowed to cool for 2 min before standard immunohistochemistry procedures were continued.
In situ hybridization.
Tissue was prepared as described above. Nonradioactive-dioxygenin-labeled RNA in situ hybridization was performed as described previously (Carney et al., 2009). The probes used in this study were Emx1 (Yoshida et al., 1997) and Cre (Hirata et al., 2009).
DiI cell migration assay.
E11.5 and E13.5 brains were dissected in ice-cold HBSS (Invitrogen), the skull and meninges were removed, and brains were placed in 3% low-melting-point agarose (Fisher Scientific). Slices (250–300 μm) were cut using a vibratome (VT1000S; Leica), and sections were allowed to recover at 37°C in MEM supplemented with l-glutamine (1:100), penicillin/streptomycin (1:100), and 10% fetal bovine serum (all from Invitrogen). DiI crystals were placed in the lateral pallium and the slices were cultured in neurobasal medium supplemented with l-glutamine (1:100), penicillin/streptomycin (1:100), and B-27 (1:50; Invitrogen) for 2 d in vitro (DIV). After culture, sections were washed in PBS, postfixed in 4% PFA for 15 min, and washed again in PBS. Sections were then coverslipped using concave slides (Fisher Scientific), and slices in which DiI was targeted to the ventral or lateral pallium and DiI+ cells were observed migrating ventrally were immediately photographed.
In utero electroporation.
Electroporations were performed as previously described (Gal et al., 2006). CAG-RFP was used at a concentration of 4 μg/μl, diluted in PBS and Fast-Green. RFP (1 μl) was injected into the lateral ventricles of E12.5–E14.5 embryo brains, and micro tweezertrodes were oriented with the positive electrode over the cortical primordium; four pulses of 33 mV of current were then applied, with 50 ms intervals. Embryos were then allowed to survive for 1 d, for 3 d, or to full term, and they were killed postnatally at P15 or P30. Brains were processed as in preparation for immunohistochemistry. The location of electroporation was determined by examining the brain in serial coronal sections. Only animals with RFP+ electroporated cells in the pallium or cortex were analyzed.
Electrophysiology.
Electrophysiological recordings were performed from fluorescent Emx1-lineage cells located in the striatum and basolateral nucleus of the amygdala. Briefly, animals were deeply anesthetized (with CO2) until nonresponsive and then decapitated. Brains were removed and immediately immersed for 2–3 min in an ice-cold oxygenated (95% O2/5% CO2) sucrose slicing solution containing the following (in mm): 234 sucrose, 11 glucose, 24 NaHCO3, 2.5 KCl, 1.25 NaH2PO4 · H2O, 10 MgSO4, and 0.5 CaCl2. Coronal slices containing the striatum or amygdala were cut on a vibratome (Leica) at 250 μm. Slices were collected and placed in an oxygenated incubation chamber containing preheated (32°C) oxygen-equilibrated artificial cerebral spinal fluid (ACSF) containing the following (in mm): 126 NaCl, 26 NaHCO3, 10 glucose, 2.5 KCl, 1.25 NaH2PO4 · H2O, 2 MgCl2 · 6H2O, and 2 CaCl2 · 2H2O; pH 7.4. The slices were incubated for 1 h at 32°C and then allowed to cool to room temperature until being transferred to the recording chamber. Neurons were located and visualized with a fixed-stage upright microscope (E600 FN Nikon) equipped with a 4× objective and a 60× insulated objective, infrared (IR) illumination, Nomarski optics, an IR-sensitive video camera (Cohu), and fluorescent lamp (D-FI universal epi-fluor illuminator, Nikon) equipped with a 450–490 λ filter. Glass pipettes (nonfilament borosilicate glass, Garner Glass) were pulled with a Flaming/Brown micropipette puller (model P-97, Sutter Instruments) to a resistance of 3–5 MΩ. For all recordings the intracellular pipette solution consisted of the following (in mm): 130 K-gluconate, 10 KCl, 10 HEPES, 10 EGTA, and 2 MgCl2. Recordings were performed at room temperature with continuous perfusion (2 ml/min) of ACSF. Cells were recorded in current-clamp mode using a multiclamp 700A amplifier (Axon) and digitizer (DigiDATA, Axon). For all cells, membrane potential and input resistance values were recorded. For all current-clamp recordings, current was injected into the cell to keep the cell at −60 mV. Cells were then characterized on the response to depolarizing and hyperpolarizing current pulses for a duration of 600 ms in 12 consecutive sweeps. For striatal cells, threshold responses were measured and rheobase currents were recorded. Cells were analyzed off-line using pClamp software (Axon) and specialized graphing software (Origin).
In vitro differentiation assay.
E13.5 Emx1-Cre; ROSA-YFP brains were dissected out in ice-cold HBSS and divided into medial pallium, dorsal pallium, ventral/lateral pallium, and LGE regions. Cells were disassociated and plated in MEM with supplements on PDL-coated coverslips, as described in detail previously (Costa et al., 2008). After 1 DIV, 100 μl of B27 was added to the medium. Cells were cultured for 10 d, fixed, immunostained with antibodies to GAD65/67 and CTIP2, and subsequently counterstained with DAPI.
Microscopy.
In situ hybridization photographs were taken using an Olympus BX51 microscope. Fluorescent photographs were taken using a Zeiss LSM 510 confocal microscope. For confocal image analysis, each fluorophore was scanned sequentially and z-stacks of the images obtained were collapsed into a single projection image or presented as individual optical sections. Figures were prepared using ImageJ (NIH) and Adobe Illustrator; brightness and contrast adjustments were applied equally across all images.
Data analysis.
Postnatal sections from Emx1-cre; ROSA-YFP brains were photographed as described above. For each immunohistochemical marker, three coronal sections of dorsal striatum at bregma levels 0.14–0.98 were examined from n = 3 animals, except where noted. The following criteria were applied to determine colocalization of cell subtype markers with YFP fluorescence. (1) Cells were counted from individual optical sections, not collapsed projection images. (2) Cells were counted as double positive if an immunopositive YFP cell body was clearly colocalized with the fluorophore of interest and contained a nucleus that was also DAPI positive. (3) For BrdU quantification, cells were counted as BrdU+ if they contained medium to strong labeling with BrdU, defined as labeling at least 50% of the DAPI nuclear stain. Percentages of Emx1-lineage and Emx1-negative MSNs (see Fig. 3) were calculated by dividing the total number of BrdU+ YFP+ (or YFP−) DARPP-32+ cells by the number of YFP+ (or YFP−) DARPP-32+ cells for each age. For the amygdala, percentages of Emx1-lineage excitatory cells (see Fig. 3) were calculated by dividing the total number of BrdU+ YFP+ Tbr1+ cells by the number of YFP+ Tbr1+ cells for each age.
Results
Fate of Emx1-lineage cells in the postnatal striatum and amygdala
To determine the fate of Emx1-lineage cells in the postnatal striatum and amygdala, we crossed previously generated Emx1-Cre mice (Gorski et al., 2002) to ROSA-YFP reporter mice. We observed that YFP+ Emx1-lineage cells were distributed preferentially in the dorsal striatum (Fig. 1A), a region that receives layer V cortical inputs, primarily from neocortical areas. To determine the cellular fate of Emx1-lineage cells in the striatum, we performed immunohistochemical analyses with antibodies to YFP and all mature striatal cell types, including inhibitory neuronal populations and glia. Additionally, we labeled with markers for the patch and matrix domains of the striatum. When labeling was done with neuronal markers, we found that 44% (89/198, n = 3) of the Emx1-lineage cells in the striatum colocalized with NeuN, and that most (96%, 67/70 n = 2) YFP+/NeuN+ cells also coexpressed DARPP-32 (Fig. 1B). YFP+/DARPP-32+ cells in the dorsal striatum revealed represented ∼4% (105/2713, n = 3) of the total DARPP-32+ MSN population. These Emx1-lineage cells also colabeled with CTIP2 (Fig. 1C), which, along with DARPP-32, labels MSNs (Gerfen, 1992; Arlotta et al., 2008). In contrast, immunolabeling with several markers for interneuronal subtypes [parvalbumin, calretinin (CR), SST, neuropeptide Y, and neuronal nitric oxide synthase] revealed that approximately only one YFP+ cell per section also expressed CR, with no YFP+ cells colocalizing with other markers of inhibitory neurons (data not shown). In addition, we found that many Emx1-lineage cells expressed CC1 (26%, 49/188, n = 3), GFAP (9%, 17/179, n = 3), or NG2 (21%, 43/240, n = 3), markers for mature oligodendrocytes, astrocytes, or glial progenitors, respectively (supplemental Fig. 1, available at www.jneurosci.org as supplemental material; Table 1).
Emx1-lineage cells give rise to patch MSNs in the mature striatum. A, Low-power image of P30 Emx1-Cre; ROSA-YFP striatum shows YFP+ Emx1-lineage cells present at high levels in the dorsal striatum. In this panel, lateral is to the right and medial is to the left. B, C, Higher magnification of a single confocal optical section of the mediodorsal striatum shows triple labeled YFP+/DARPP-32+/NeuN+ neurons (B, arrowheads). YFP+ cells are also labeled by the MSN marker CTIP2 (C, arrowheads). D, Few Emx1-lineage neurons are labeled by the matrix projection neuron marker calbindin (arrowheads point to a positive cell, double arrowheads point to a negative cell). E, YFP+ cells are found outside of the domain of SST+ fibers, which demarcates the matrix (arrowheads). F–H, Immunolabeling with patch markers μ-opioid receptor-1β (F, arrowheads) and reelin (G, arrowheads) reveals colocalization with YFP+ Emx1-lineage neurons. YFP+ cells are also surrounded by TH+ fibers (H, arrowheads). I, J, Patch-clamp recordings of YFP+ cells in the striatum reveal that Emx1-lineage neurons are both D1- and D2-type MSNs. I, Trace from a single D1-like YFP+ cell; current held at −60 mV. Average resting membrane potential was −87.2 mV (n = 2). J, Trace from a single D2-like YFP+ cell; current held at −60 mV. Average resting membrane potential was −85.4 mV (n = 6). CC, Corpus callosum, Str, striatum. Scale bars: (in A) A, 500 μm; B–H, 100 μm.
As the largest population of striatal neurons derived from the Emx1 lineage were DARPP-32+/CTIP2+ MSNs, we wanted to determine whether these neurons contributed to both major subdivisions of the striatum, patch (also called striosomes) and matrix. We labeled the patch domains with reelin and MOR1-β, and the matrix domains with SST and CB, as SST+ fibers from the cortex allow demarcation of the matrix and CB labels projection neurons in the matrix (Gerfen, 1985, 1992; Alcántara et al., 1998; Arlotta et al., 2008). Few Emx1-lineage MSNs were found in the SST+ fiber-rich matrix, and a small number of Emx1-lineage colocalized with CB (Fig. 1D,E). In contrast, we found that the majority (91%, 98/108, n = 2) of Emx1-lineage MSNs were in reelin-positive patches, while only 44% (1153/2630, n = 2) of YFP-negative MSNs were present in reelin-positive patches (Fig. 1F). This result was further supported by the colocalization of YFP+ cells with MOR1-β expression (Fig. 1G). In addition, we found that 100% (28/28, n = 2) of Emx1-lineage MSNs were in TH+ domains (Fig. 1H). TH labeling marks dopaminergic projections from the substantia nigra, also referred to as dopaminergic islands. Thus, Emx1-lineage progenitors gave rise to patch MSNs innervated by dopaminergic inputs from the substantia nigra.
In addition to patch/matrix subdivisions, striatal MSNs can also be classified based on their output projections. MSNs receive excitatory projections primarily from layer V of the cortex, and send inhibitory projections out of the striatum through two primary pathways (Gerfen, 1985). Direct pathway MSNs project directly to the substantia nigra, express D1 dopamine receptors (D1 nigral MSNs of the direct pathway), and facilitate movement (Gerfen et al., 1990). Indirect pathway MSNs project to the substantia nigra via the globus pallidus, and express D2 dopamine receptors (D2 pallidal MSNs of the indirect pathway), and inhibit movement. To determine whether Emx1-lineage MSNs were part of one or both of these circuits, we performed whole-cell current-clamp recordings using previously established physiological criteria to differentiate between D1 and D2 MSN subtypes (Ade et al., 2008; Gerfen et al., 2008). Based on the spiking threshold of YFP+ cells, we found that Emx1-lineage neurons were both high threshold D1-like MSNs (Fig. 1I) (2/8 cells, rheobase = 90–100 pA) and low-threshold D2-like MSNs (Fig. 1J) (Table 1) (6/8 cells; rheobase = 15–30 pA). As D1 MSNs make up 35% of striatal neurons, and D2 MSNs account for 40% of striatal neurons (Gerfen, 1992), our findings suggest that Emx1-lineage neurons disproportionately contribute to the D2 MSN subtype, as 75% of Emx1-lineage cells were D2-like MSNs. Thus, collectively these data reveal that Emx1-lineage progenitors gave rise to MSNs that preferentially localized to the patch domains, which receive cortical input from deep layer V neurons and subcortical input from ventral tier dopaminergic neurons from the substantia nigra. Furthermore, we found that Emx1-lineage-derived MSNs provide output to the substantia nigra via both direct and indirect pathways, although more Emx1-lineage MSNs contributed to the D2 MSN subpopulation that is responsible for inhibition of movement.
Emx1-lineage cells contribute to multiple neural subtypes in the basal forebrain
In addition to contributing to the postnatal striatum, consistent with previous studies (Gorski et al., 2002; Medina et al., 2004), we observed that Emx1-lineage cells also contributed to the postnatal amygdala. However, in previous studies the cellular fate of Emx1-lineage cells was not determined. We therefore examined whether Emx1-lineage cells in the mature amygdala gave rise to excitatory neurons similar to their pallial counterparts in the cerebral cortex, or to inhibitory neurons similar to their subpallial counterparts in the striatum. We found that, in contrast to the striatum, Emx1-lineage YFP+ cells in the mature basolateral complex (BLC) of the amygdala colocalized exclusively with Tbr1 in the both the basolateral and lateral nuclei (Fig. 2A–C), with no YFP+ cells expressing inhibitory neuronal markers. Previous studies (for review, see Sah et al., 2003) have revealed that, based on measures of firing properties, excitatory neurons in the lateral and basolateral amygdala comprise two distinct electrophysiological subtypes, pyramidal-like and burst-firing neurons. To examine the physiological fate of amygdala Emx1-lineage cells, we performed patch-clamp slice recordings in the lateral and basolateral amygdala nuclei. This analysis revealed that Emx1-lineage cells generated these two cell types, which were interestingly segregated to separate nuclei. Emx1-lineage cells in the lateral nucleus exclusively gave rise to pyramidal-like excitatory neurons (n = 7), whereas Emx1-lineage cells in the basolateral nucleus only gave rise to burst-firing excitatory neurons (n = 7) (Fig. 2I–K, Table 1).
Emx1-lineage cells give rise to excitatory neurons in the basolateral complex of the amygdala. A, As shown at P30, YFP+ Emx1-lineage cells are present throughout the basolateral and lateral nuclei of the amygdala. B, C, Higher-magnification optical sections of boxed regions of the lateral nucleus and basolateral nucleus shown in A reveal that YFP+ cells in both of these nuclei also express Tbr1, a marker of excitatory neurons. (Arrowheads in B and C show YFP+/Tbr1+ cells; DAPI is shown in blue). D–K, Cells of the amygdala in Emx1-Cre; Z/EG animals were also analyzed by immunohistochemistry (D, E) and by patch-clamp slice recordings at P21 (F–K). Emx1-lineage cells in Z/EG reporter animals (D) also express Tbr1 in the BLC (E). Low magnification of a bright-field view of the amygdala (F) shows location of recording. Higher magnification of bright field (G) and fluorescent images (H) illustrate a GFP+ cell before patching. I, Coronal section from an Emx1-Cre; ROSA-LacZ brain at the level of the basolateral complex shows many recombined cells in the lateral (J) and basolateral (K) amygdala. Colored circles illustrate the differential distribution of Emx1-lineage excitatory neurons identified by recordings in J and K. J, Traces from a representative GFP+ cell in the lateral nucleus of the amygdala; all GFP+ cells recorded in this nucleus exhibited a pyramidal-like firing pattern of action potentials (n = 7). K, Traces from a representative GFP+ cell recorded in the basolateral nucleus; this population generated action potentials in a burst-firing pattern (n = 7). BL, Basolateral nucleus of the amygdala; L, lateral nucleus of the amygdala; DEn, dorsal endopiriform nucleus. Scale bars: (in A) A–C, F, H, J, 300 μm; E, G, I, 50 μm.
Differences in the timing of the generation of striatal and amygdala Emx1-lineage neurons
As our above data revealed that the Emx1 lineage generates different cell types in the striatum and the amygdala, we sought to determine whether developmental differences in timing of the birth of Emx1+ progenitors is related to the generation of these two disparate neuronal populations. To accomplish this, we used BrdU birth dating to label cells in S phase at E9.5, E11.5, E13.5, E14.5, and E15.5. In the striatum, a minority of Emx1-lineage MSNs were labeled with BrdU at early ages of BrdU administration [4% (4/103, n = 3) at E9.5; 15% (9/59, n = 2) at E11.5] (Fig. 3A,D). BrdU administration at later ages resulted in more labeling of Emx1-lineage MSNs [23% (15/63, n = 2) at E13.5, 29% (21/78, n = 3) at E14.5, and 27% (14/50, n = 2) at E15.5] (Fig. 3B–D). These birth dates were also coincident with the timing of neurogenesis of Emx1-negative MSNs, in which fewer Emx1-negative MSNs are born early [3% (86/3339, n = 3) at E9.5, and 11% (355/3221, n = 2) at E11.5]. Birth dates of later born Emx1-negative MSNs were also similarly distributed [33% at E13.5 (670/2032, n = 2), 31% at E14.5 (1272/4093, n = 3), and 27% at E15.5 (703/2261, n = 2)] (Fig. 3D). Emx1-lineage MSNs contributed between 2 and 3% of the total dividing MSN progenitor pool at all ages examined (3% at E9.5, n = 3, 2% at E11.5, n = 2, 3% at E13.5, n = 2, 2% at E14.5, n = 3, and 2% at E15.5, n = 2).
Differential birth dates of Emx1-lineage cells in the striatum and amygdala. A, BrdU administration at E11.5 labels Emx1-lineage neurons in the dorsal striatum at P15 (arrowheads point to a BrdU+/YFP+/DARPP-32+ cell; double arrowheads point to a BrdU−/YFP+/NeuN+ cell). B, Administration of BrdU at E13.5 results in more YFP+ cells double-labeled. (Arrowheads point to a BrdU+/YFP+/NeuN+ cell; double arrowheads point to a BrdU−/YFP+/NeuN+ cell). C, BrdU administration at E14.5 results in many colabeled Emx1-lineage cells at P15. (Arrowheads point to BrdU+/YFP+/DARPP-32+ cells; double arrowheads point to BrdU−/YFP+/DARPP-32+ cells). D, Quantification of striatal birth dates. For Emx1-lineage MSNs, percentages were calculated by dividing the total number of YFP+ (or YFP−)/DARPP-32+/BrdU+ cells by the number of YFP+ (or YFP−)/DARPP-32+ cells for each age; for Emx1-negative MSNs percentages, YFP− cells in the same sections were quantified. Emx1-lineage MSNs: 4% (4/103, n = 3) at E9.5, 15% (9/59, n = 2) at E11.5, 23% (15/63, n = 2) at E13.5, 29% (21/78, n = 3) at E14.5, and 27% (14/50, n = 2) at E15.5. Emx1-negative MSNs: 3% (86/3339, n = 3) at E9.5, 11% (355/3221, n = 2) at E11.5, at E13.5, 33% (670/2032, n = 2), at E14.5, 31% (1272/4093, n = 3), and 27% (703/2261, n = 2) at E15.5. E, F, In the amygdala, BrdU labels 7% of Tbr1+/YFP+ cells at E9.5. E, BrdU alone; few cells are labeled in the basolateral nucleus (BL) at E9.5. E′, Low-power image of the basolateral nucleus. F, Higher-magnification single optical sections of boxed section in E′; single arrowheads point to triple-labeled BrdU+/Tbr1+ YFP+ cells; double arrowheads point to BrdU−/YFP+/Tbr1+ cells. G, H, BrdU administration at E11.5; BrdU+ cells are present throughout the BLA. BrdU-labeled Emx1-lineage cells are present throughout both the BL and lateral nucleus (L) (26.4% of YFP+/Tbr1+ cells were also heavily labeled with BrdU); arrowheads in H point to BrdU+/YFP+/Tbr1+ cells. I, J, BrdU at E13.5; in the BLA very few cells are BrdU labeled. However, intercalated cell masses both medially and laterally are heavily labeled by BrdU (I). Low-power (I′) and higher-power (J) optical sections show that few Emx1-lineage cells are labeled with BrdU at E13.5 in the BL (1.5% of YFP+/Tbr1+ cells were heavily labeled) (arrowheads point to the only BrdU+/YFP+/Tbr1+ cell in the section). K, Graph of the percentages of striatal- and amygdala-fated Emx1-lineage BrdU+ cells born at different times during embryogenesis. In the amygdala, 7% were born at E9.5 (48/731, n = 6), 14% at E11.5 (89/741, n = 6), 2% at E13.5 (9/555, n = 3), 1% at E14.5 (7/279, n = 3), at 0.1% at E15.5 (1/608, n = 3). LITC, Lateral intercalated cells; MITC, medial intercalated cells; PCx, piriform cortex. Scale bars: (in A), A–C, F, H, J, 100 μm; E, G, I, 150 μm.
We next examined whether Emx1-lineage neurons in the amygdala are generated at different times than Emx1-lineage neurons in the striatum. Similar to the striatum, BrdU administration at E9.5 resulted in a small (7%, 48/731, n = 6) population of heavily labeled BrdU+/Tbr1+ Emx1-lineage excitatory neurons in the lateral and basolateral amygdala (Fig. 3E,F). In contrast, administration of BrdU 2 d later at E11.5 resulted in a larger population (14%, 89/741, n = 6) of strongly BrdU-labeled excitatory neurons of the Emx1 lineage in both the postnatal lateral and basolateral amygdala, compared with the striatum. These cells were distributed throughout the lateral and basolateral nuclei (Fig. 3G′,H). Additionally, also in contrast to the striatum, only very few (2%, 9/555, n = 3) amygdala neurons were heavily BrdU labeled when BrdU was administered at E13.5 (Fig. 3I,J). Interestingly, BrdU administration at this age selectively labeled intercalated cells, which are located at the boundary of the lateral and basolateral nuclei (Fig. 3I). These specialized inhibitory neurons regulate feedforward cortical inputs and fear circuitry outputs (Marowsky et al., 2005; Hefner et al., 2008; Likhtik et al., 2008). Very few neurons were born in the amygdala at later ages, with 1% labeled at E14.5 (7/279, n = 3), and at E15.5, with 0.1% labeled (1/608, n = 3) (data not shown). Therefore, Emx1-lineage cells that contribute to the amygdala precede the generation of Emx1-lineage cells that contribute to the striatum; these differences in timing are shown graphically in Figure 3K.
Origins of basal telencephalic Emx1-lineage neurons
Previous expression studies (Puelles et al., 2000; Gorski et al., 2002) revealed that Emx1 is expressed throughout the lateral and dorsal pallial domains, beginning during early neurogenesis. This restricted expression pattern of Emx1 raised the question as to whether striatal and amygdala Emx1-lineage cells were pallial derived. To test this we took a number of approaches. We first examined expression of Emx1 during early neurogenesis, beginning at E9.5, when Emx1 is first present, and compared this expression pattern to the pattern of Cre expression and the location of Emx1-lineage cells as revealed from Emx1-Cre; ROSA-LacZ crosses. We reasoned that recombined Emx1-lineage cells observed outside the domain of Emx1 and Cre expression most likely migrated from an Emx1+ source. Indeed, at E11.5 and E13.5, β-gal staining in Emx1-Cre; ROSA-LacZ brains revealed that Emx1-lineage cells were present in the ventral pallium (VP) and dorsal lateral ganglionic eminence (dLGE), beyond the domain of Emx1 and Cre mRNA expression (supplemental Fig. 2C,D,G,H,K,L, available at www.jneurosci.org as supplemental material). This result suggested that Emx1+ cells had migrated from the pallium to the subpallium during early neurogenesis. In fact, late in embryogenesis (older than E15.5) and early in postnatal development, cortically derived Emx1-lineage cells have been shown to migrate ventrally into the developing striatum (Willaime-Morawek et al., 2006; Willaime-Morawek and van der Kooy, 2008).
To directly examine whether pallial-derived Emx1-lineage cells indeed migrate from the pallium to the subpallium early in development, we placed DiI in the lateral and dorsal–lateral pallium of E11.5 (Fig. 4A–F) and E13.5 (Fig. 4G–L) Emx1-Cre; ROSA-YFP brains and cultured for 36–48 h in vitro. DiI+ cells were observed migrating into the dLGE, thus confirming that pallial progenitors had the capacity to migrate ventrally. We also observed DiI+/YFP+ Emx1-lineage cells migrating ventral–laterally from the dorsal–lateral pallium into the ventral pallium and LGE (Fig. 4D–F,J–L).
A subpopulation of Emx1-lineage progenitors migrates ventrally from the developing cerebral cortex into the LGE during early neurogenesis. A–C, Low-power images of a coronal Emx1-Cre; ROSA-YFP slice show DiI placement at E11.5 in the lateral pallium (arrowheads mark cells that have migrated). C′, DiI crystal location. D–F, Higher magnification of the boxed region in C reveals that a subpopulation of DiI+ cells has migrated from the lateral pallium to the ventral pallium and LGE. Arrowheads point to several cells that have migrated ventrally that are also YFP+. Eighty-three percent (5/6) of the slices had ventrally migrating DiI-labeled cells that also colabeled with YFP. G–I, I′, DiI placed in the lateral pallium at E13.5 also reveals cells migrating ventrally into the VP and LGE. J–L, Higher magnification of the boxed region in I; arrowheads point to DiI+/YFP+ cells in the LGE. Sixty-seven percent (4/6) of the slices had ventrally migrating DiI-labeled cells that also colabeled with YFP. M–Q, Embryos were electroporated in utero with 4 μg/μl CAG-RFP plasmid. The lateral pallium was specifically targeted at embryonic ages as indicated, and embryos were killed 3 d later (E15.5), at P15, or at P30. M, N, Short-term analysis of electroporated brains shows that the electroporation was targeted specifically in the lateral pallium. N, Higher magnification of the boxed region in M shows cells with ventrally oriented leading processes (arrowheads) (n = 5 electroporations). O, Coronal sections of a P15 animal electroporated at E13.5; RFP+ cells are present in the dorsal striatum near the SVZ (n = 3). P, Coronal section of a P30 animal electroporated at E14.5; cells are present in the dorsal striatum. Q, Image of the adjacent cortex; RFP+ cells are present in the upper layers of the cortex, indicating correct embryonic dorsal pallium targeting (n = 2). Scale bars: (in C), A–C, 250 μm; D–F, 50 μm; (in L), G–I, 250 μm; J–L, 125 μm; M, 160 μm; N, 80 μm; O–Q, 100 μm.
To further examine this migratory path using an in vivo assay, we electroporated the dorsolateral pallium in utero at E12.5 with a construct that constitutively expresses RFP in electroporated cells. Three days later, at E15.5, we observed RFP+ cells that had migrated away from the site of electroporation in the dorsolateral pallium and toward the ventral telencephalon (Fig. 4M,N), consistent with previous work showing pallial-to-subpallial migration using in utero electroporations (Bai et al., 2008; Soma et al., 2009). This observation was also consistent with our in vitro findings of an early pallial-to-subpallial migratory wave. From our long-term analysis of cortically electroporated brains, we also found a small number of RFP+ cells in the dorsal striatum (Fig. 4O–Q).
Although the above evidence revealed that Emx1-lineage cells migrate from the pallium to the subpallium, we wanted to examine whether some of these cells might also be locally derived. To accomplish this, we labeled Emx1-Cre; ROSA-YFP embryos with a short pulse of BrdU at either E11.5 or E13.5, and killed after 2 h (supplemental Fig. 3, available at www.jneurosci.org as supplemental material). We reasoned that any BrdU+ cells in the LGE would most likely be derived locally as 2 h would not be enough time for a pallial-derived cell population to migrate ventrally into the LGE. Interestingly, a few Emx1-lineage cells in the LGE were also heavily labeled with BrdU at E13.5, although not at E11.5, suggesting that at E13.5 there may also be a locally derived Emx1+ population. Thus, in addition to arising from the pallium, it is likely that a small subpopulation of Emx1+ cells present in the LGE at E13.5 and later were locally derived. This finding is also consistent with a recent study revealing that Emx1-Cre mice have the capacity to recombine in subpallial Gsx2-positive cells (Waclaw et al., 2009).
Generation of MSNs from Emx1+ pallial progenitors in vitro
Given the above evidence that dorsal Emx1-lineage cells were able to migrate ventrally, and that pallial RFP-labeled cells were found in the postnatal striatum, we next sought to determine whether pallial-derived Emx1-lineage cells had the capacity to generate striatal MSNs. To accomplish this, we dissected distinct domains of the E13.5 Emx1-Cre; ROSA-YFP pallium and LGE (schema shown in supplemental Fig. 4E, available at www.jneurosci.org as supplemental material) and allowed them to differentiate into neurons in vitro. Differentiated MSNs were identified by triple immunostaining with YFP, CTIP2 and GAD65/67 (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). We found that Emx1-lineage cells from the medial pallium were inefficient at producing MSNs, while cells taken from either the dorsal pallium or the region of the ventral/lateral pallium were able to produce MSNs with a greater efficiency (supplemental Fig. 4A–C,F, available at www.jneurosci.org as supplemental material). In contrast, Emx1-lineage cells in the LGE were not as efficient at making MSNs as cells taken from the dorsal or ventral/lateral pallium (supplemental Fig. 4D,F, available at www.jneurosci.org as supplemental material). Thus, this in vitro data revealed that embryonic Emx1-lineage cells derived from the pallium have the capacity to generate MSNs.
Differential combinatorial gene expression in striatal and amygdala Emx1 lineages
Once we established that Emx1-lineage cells were able to migrate ventrally from the pallium and had the potential to become inhibitory MSNs, we sought to determine whether subpallial Emx1-lineage cells were expressing genes characteristic of their putative pallial origin or of their new subpallial location. To accomplish this, sections from Emx1-Cre; ROSA-YFP embryos were coimmunostained for YFP and individually for the subpallial markers Gsx2 and Dlx2, or the pallial markers Pax6 and Tbr1. At E11.5 Emx1-lineage cells in the LGE were predominately found in the dLGE, and were also present near the pallial–subpallial border (PSB) (Fig. 5A). These cells coexpressed the subpallial ventricular zone (VZ) homeobox transcription factor Gsx2 (Fig. 5B). By E13.5, many more cells were present throughout the LGE, and these cells had leading processes extending in multiple directions, suggesting that they were moving between the LGE mantle, subventricular zone (SVZ), and VZ (Fig. 5C,D,D′). These Emx1-lineage cells also expressed Gsx2 in the VZ, and in addition, coexpressed Dlx2, a homeobox transcription factor expressed in the VZ and SVZ of the subpallium (Fig. 5E,F). These data indicate that, in contrast to the Emx1-lineage cells that remain in the pallium, the subpopulation of Emx1-lineage cells that are present in the LGE express subpallial genes characteristic of their new environment.
Expression of subpallial and pallial markers in Emx1-lineage cells during embryogenesis correspond to their basal telencephalic location. A, Confocal images from sections from Emx1-Cre; ROSA-YFP brains illustrate that several Emx1-lineage cells are present at E11.5 in the dLGE in both the VZ and SVZ (arrowheads). B, YFP+ cells in the dLGE express Gsx2, which marks the VZ of the subpallium. Arrowheads in B show YFP+/Gsx2+cells. C, D, By E13.5, many Emx1-lineage cells are present in the LGE. D, Higher magnification of the boxed region in C. E, F, Confocal optical images reveal that a subpopulation of YFP+ cells in the LGE expresses the subpallial transcription factors Gsx2 (E) and Dlx2 (F) (arrowheads show double-labeled cells and double arrowheads show YFP+ cells only). G, Optical sections reveal that Emx1-lineage cells in the dLGE at E11.5 do not express the pallial marker Pax6 (double arrowheads). However, Emx1-lineage cells in the VP are also Pax6+ (arrowheads). H, I, Low-power optical section (H) and higher-power optical section of boxed region in H (I) showing that recombined cells are also present more caudally in the region of the developing piriform cortex and amygdala. Many of these cells are Tbr1+. J, K, By E13.5, many YFP+ cells are present beyond the Pax6+ ventral pallium and are located in the LGE in both the VZ and the SVZ. K, Higher-magnification optical section of boxed region in J reveals that in the LGE, YFP+ cells do not colabel with Pax6 (arrows). L, M, In the more caudal regions, including the presumptive amygdala, many Emx1-lineage cells are present, and most express Tbr1. Higher-magnification optical section of boxed region in L is shown in M. Scale bars: (in B), A, C,H,L, 300 μm; B,G,K, 200 μm; (in E), E, 100 μm; (in F), F, 100 μm; (in I), I, 100 μm; (in M), M, 100 μm. AMY, Amygdala, CGE, caudal ganglionic eminence; PCx, presumptive piriform cortex; Str, striatal primordium.
We further examined whether Emx1-lineage cells located in the subpallium continued to express pallial genes consistent with their probable pallial origin. We immunolabeled E11.5 and E13.5 sections with Pax6, and found that some Emx1-lineage cells at the ventral pallial aspect of the PSB continued to express Pax6 protein. However, consistent with their expression of subpallial markers as shown above, cells in the LGE VZ and SVZ did not express Pax6 (Fig. 5G,J,K). We also examined sections more caudally at the level of the caudal ganglionic eminence and the developing amygdala. While Pax6 was not expressed at this plane of section in the amygdala primordium at detectable levels (data not shown), Tbr1, a marker of the excitatory neuronal lineage, was highly expressed, and most Emx1-lineage cells at this age expressed Tbr1 in the amygdala primordium and piriform cortex, but not in the dLGE or ventral palliumVP (Fig. 5H,I,L,M) (data not shown).
As our above evidence collectively indicates that pallial-derived Emx1+ progenitors contribute to both excitatory and inhibitory neuronal populations in the basal telencephalon, we next sought to determine whether interaction with the PSB, a region through which many of these cells presumptively migrate, is important in the regulation of this differential fate. To accomplish this, we examined the status of Emx1-lineage cells in loss of Pax6 function (Sey/Sey; Small eye) mice in which the PSB is severely disrupted (Fig. 6). In these mutants, proliferation is increased in the cortex, differentiation is impaired, migration out of the pallial VZ is reduced, and migration at the PSB is abnormal (Schmahl et al., 1993; Carić et al., 1997; Chapouton et al., 1999; Warren et al., 1999; Quinn et al., 2007; Carney et al., 2009). We found an increased number of Emx1-lineage progenitors in the LGE of Emx1-Cre; Sey/Sey mice at E13.5, although this difference was not apparent at earlier ages (Fig. 6A–D). Analysis at E17.5, when many LGE progenitors have already differentiated into MSNs, revealed that there were increased numbers of CTIP2+/Emx1-lineage cells in the striatal primordium in Emx1-Cre; Sey/Sey mice compared with control mice (Fig. 6E,F). In contrast, in the developing amygdala, the number of Tbr1+/YFP+ neurons was not different between controls and Sey/Sey mice at either E13.5 or E17.5 (Fig. 6G–J). These data indicated that there was not an obvious defect in proliferation, migration, or fate of Emx1-lineage cells in the amygdala. Therefore, disruption of the PSB results in an increase in the numbers of Emx1-derived MSNs in the striatum, but not the amygdala, indicating a role for Pax6 in one or more of several developmental processes: (1) determination of the dorsal to ventral migratory route of Emx1-lineage cells to the LGE, (2) regulation of the fate of these cells in the striatum, or (3) regulation of the size of the Emx1-derived MSN progenitor pool.
Disruption of the PSB in Sey/Sey mice alters the number of Emx1-lineage cells in the striatal primordium but not in the amygdala primordium. A, Wild-type (WT) E11.5 Emx1-Cre; ROSA-YFP section labeled with Gsx2; a few YFP+/Gsx2+ cells are present in the LGE. B, In the Sey mutant, Gsx2 protein expands dorsally, and few Emx1-lineage cells are present in the LGE. C, In the WT E13.5 mouse; YFP+ cells colabel with Gsx2 in the VZ of the LGE. D, In the E 13.5 Sey mutant, the number of YFP+ cells is ectopically expanded past the PSB (arrowheads), and Gsx2 expression is expanded dorsally, but there is no change in the number of YFP+ cells or their position in the LGE (110 CTIP2+/YFP+ cells in n = 3 WT, 108 CTIP2+/YFP+ cells in n = 3 mutants). E, At E17.5 in the WT mouse, scattered YFP+/CTIP2+ cells are present in the dorsal–lateral striatum and many YFP+ cortical fibers are visible in the dorsal–medial striatum (see inset); few CTIP2+ cells are YFP+ (arrowheads). F, In the Sey mutant, significantly more CTIP2+/YFP+ cells (arrowheads) are present (see inset), and the YFP+ cortical fibers are markedly diminished; CTIP2+ YFP+ cells in mutants are 182% of CTIP2/YFP+ cell numbers in controls, p < 0.05 (95 CTIP2+/YFP+ cells in n = 3 WT, 173 CTIP2+/YFP+ cells in n = 3 mutants). G, At E13.5, the WT amygdala primordium contains many YFP+/Tbr1+ cells. H, In the Sey mutant, numerous YFP+ cells are still present. I, J, In addition, YFP+/Tbr1+ cells are present in both WT and mutants at E17.5. Quantification of E13.5 and E17.5 revealed no significant differences in the numbers of YFP+ cells in the amygdala primordium [Sey mutants had 102% of control Tbr1+/YFP+ cell numbers at E13.5 (415 Tbr1+/YFP cells in Sey/Sey mice, n = 3, 405 Tbr1+/YFP cells in WT mice, n = 3) and 115% of controls at E17.5 (233 Tbr1+/YFP cells in Sey/Sey mice, n = 3, 202 Tbr1+/YFP cells in WT mice, n = 3)]; these differences were not statistically significant. Scale bars: (in B), A–D, 100 μm; (in F), E–J, 100 μm.
Our above analyses of gene expression profiles of Emx1-lineage cells during embryogenesis (Fig. 5) revealed what appeared to be two distinct pools of basal telencephalic Emx1-lineage cells: ones that express subpallial markers and presumably contribute to the striatum to generate MSNs, and ones that maintain expression of pallial markers and contribute to excitatory neurons in the amygdala. To directly examine whether Emx1-lineage cells that also expressed subpallial transcription factors had a different fate in the basal telencephalon than Emx1-lineage cells that did not express subpallial transcription factors, we used an intersectional genetic fate-mapping approach (Kim and Dymecki, 2009) in which Emx1-Cre mice were crossed to Dlx5/6-Flpe mice (Miyoshi and Fishell, unpublished observations). Their progeny were crossed to RCE: dual reporter mice, which express EGFP only when both Cre and Flpe have been expressed in the same cell (Sousa et al., 2009). During embryogenesis, we found that Emx1; Dlx-lineage cells were remarkably restricted to the LGE and not present in other telencephalic regions (Fig. 7A,B). Many of these cells were labeled with the neuronal progenitor marker TUJ1 (Fig. 7C,D). Consistent with the restricted pattern of cell distribution in the embryonic brain, we observed that in the postnatal brain Emx1; Dlx-lineage cells were found almost exclusively in the postnatal striatum (Fig. 7E) and were notably absent from the amygdala (data not shown). Furthermore, Emx1; Dlx-lineage cells expressed DARPP-32 and NeuN, but did not express markers for glial cells or interneurons, indicating that they were also restricted in their cellular fate (Fig. 7F–I) (data not shown). Thus, Emx1-lineage neurons that express Dlx genes during development exclusively give rise to an MSN subpopulation in the striatum.
Cells derived from Emx1 and Dlx combinatorial lineages are present in the embryonic LGE and contribute exclusively to the postnatal striatum. A, B, At E13.5, GFP+ Emx1; Dlx-lineage cells are present throughout the SVZ and mantle of the LGE. Higher magnification of boxed region in A is shown in B. C, D, Many GFP+ Emx1; Dlx-lineage cells express TUJ1 (arrowheads), an early neuronal marker, as shown at lower power in C and high power in D. E, F, Low-power (E) and high-power (F) images of a P30 Emx1-Cre; Dlx5/6-Flpe; RCE: dual brain shows the distribution of intersectional lineage cells (arrowheads) in the striatum. G–I, Optical sections of GFP+ cells immunolabeled with NeuN (G) or DARPP-32 (I) reveal these Emx1; Dlx-lineage cells are NeuN+ and DARPP-32+ (G, I) but are not positive for the glial lineage marker Ng2 (H). Arrowheads show double-labeled cells for each marker, whereas double arrowheads show non-double-labeled cells. Scale bars: (in A) A, 100 μm; B, C, 50 μm; (in E) E, 500 μm; (in F) F, 100 μm.
Discussion
Emx1 lineage contribution to the striatum and amygdala
In this study, using a combination of approaches, we investigated the timing, fate and origins of Emx1-lineage cells in the striatum and amygdala, two major structures of the basal telencephalon. In relation to fate, most prominently, our immunohistochemical and electrophysiological characterization in the striatum reveals that Emx1-lineage neurons in the striatum comprise a subset of striatal MSNs. Emx1-lineage striatal neurons display a number of specific characteristics distinct from non-Emx1 lineage MSNs. First, striatal Emx1-lineage cells, while present throughout the dorsal–ventral and rostral–caudal axes, are disproportionately localized in the dorsal portion of the striatum, which receive connections from neurons in deep layer V of the cortex. Emx1-lineage MSNs are also preferentially localized to the patch (striosome) compartment of the striatum. Patch neurons receive inputs primarily from the limbic cortex, in contrast to matrix neurons, which receive inputs primarily from neocortical areas (Gerfen, 1984; Donoghue and Herkenham, 1986). Previous work has also shown that one of the targets of patch neurons is the basolateral nucleus of the amygdala, indicating that this Emx1-lineage subpopulation may be an important subcomponent of a limbic circuit that modulates information between the cortex and amygdala (Ragsdale and Graybiel, 1988). It is interesting to note that while Emx1-lineage neurons contribute mainly to patches in the striatum our BrdU analysis indicates that this cell population is born throughout striatal development. As previous work has shown that patch neurons are born earlier than matrix neurons (Fishell and van der Kooy, 1987; Fishell et al., 1990; Krushel et al., 1993; Song and Harlan, 1994; Krushel et al., 1995; Olsson et al., 1997; Mason et al., 2005) our data interestingly suggest that Emx1-lineage neurons may produce patch neurons over a longer period of time than the Emx1-negative MSN progenitor pool.
Intriguingly, our results also indicate that, in addition to the LGE, at least a subpopulation of the Emx1-lineage MSNs of the striatum appear to be derived from the embryonic pallium, a region not previously recognized as a source of basal telencephalic inhibitory projection neurons. The LGE origin is supported by our short-term BrdU birth-dating experiments which reveal some Emx1-lineage dividing cells in the VZ of the LGE at E13.5 (but not E11.5), as well as recent work has shown that Emx1-Cre mice can be used to recombine cells in the LGE (Waclaw et al., 2009). However, we suggest that an early-generated Emx1+ progenitor pool from the developing cerebral cortex is also an important source of striatal neurons. Our evidence for this is as follows: (1) at embryonic stages we find that Emx1-lineage cells are present in regions that extend beyond the pallial Emx1 mRNA-expressing domain, suggestive of an active migration ventrally; (2) our in vitro and in vivo migration assays reveal that Emx1-lineage cells from the pallium migrate ventrally into the LGE and striatum, respectively; (3) our in vitro differentiation assay indicates that the E13.5 pallium has the capacity to generate MSNs; and (4) our embryonic Sey/Sey mutant analysis shows increased numbers of Emx1-lineage MSNs in the striatal primordium. This putative novel source of striatal inhibitory neurons indicates a greater flexibility in telencephalic dorsal–ventral patterning than previously recognized.
In addition to the striatum, the other major target of Emx1-lineage cells is the amygdala. Similar to other telencephalic structures, our work and the work of others have revealed that numerous spatially separate progenitor populations contribute to the developing amygdala (Puelles et al., 2000; Medina et al., 2004; Tole et al., 2005; Hirata et al., 2009). Gene expression and fate-mapping studies have suggested one such population is the cortically derived Emx1 lineage, which is expressed in the developing amygdala; analysis of Emx1-Cre mice has revealed recombined cells in the mature amygdala (Fig. 2) (Puelles et al., 2000; Gorski et al., 2002; Medina et al., 2004). Our fate-mapping analysis in this study demonstrates definitively that Emx1-lineage progenitors give rise to excitatory neurons in the amygdala. Our electrophysiological analyses extend the above findings, revealing that these Emx1-lineage cells generate two subclasses of excitatory neurons: pyramidal-like neurons in the lateral nucleus of the amygdala, and burst-firing neurons in the basolateral nucleus of the amygdala.
Differential cell birth date and genetic heterogeneity of Emx1-derived striatal and amygdala neurons
Our fate-mapping data reveal that Emx1-lineage progenitors generate strikingly disparate neuronal subtypes depending on which structure they will eventually populate. The factors that regulate the decision of Emx1-lineage cells to contribute to one of these specific pools is not known, however, it is likely contingent upon one of three non-mutually-exclusive mechanisms: (1) differential timing of progenitor cell birth, (2) location along the anterior–posterior and/or dorsal–ventral axis of the developing cerebral cortex, or (3) local heterogeneity within the cortical VZ. Our analyses here indicate that both timing and genetic heterogeneity may play a causal role in cell fate decisions within the Emx1 lineage in determining their differentiation into striatal inhibitory neurons or amygdala excitatory neurons. The differences in the timing of the generation of Emx1-lineage amygdala neurons versus striatal neurons is quite striking, with Emx1-lineage amygdala excitatory neurons generated between E9.5 and E11.5 and Emx1-lineage MSNs generated between E11.5 and E15.5. This early generation of amygdala neurons is consistent with other studies that have indicated that the amygdala is a relatively early-born structure (McConnell and Angevine, 1983; Carney et al., 2006; Hirata et al., 2009; Soma et al., 2009). This differential timing for the generation of neuronal diversity is also highly reminiscent of the mechanisms for the generation of neuronal diversity in numerous regions of the neuraxis including the developing layers of the cerebral cortex, the patch matrix of the striatum, and dorsal–ventral diversity in the spinal cord (Miller and Nowakowski, 1988; Song and Harlan, 1994; Brittis et al., 1995; McConnell, 1995; Lai et al., 2008).
In addition, our gene expression and intersectional fate-mapping analyses indicate that genetic heterogeneity of Emx1-lineage cells also likely informs cell fate decisions. In support of this, Emx1-lineage cells in the embryonic LGE express Gsx2 and Dlx2, transcription factors required for inhibitory neuronal fate, while Emx1-lineage cells in the developing amygdala primordium exclusively express Pax6, a transcription factor required for pallial fate (Stoykova et al., 1996; Toresson and Campbell, 2001; Yun et al., 2001; Corbin et al., 2003; Stenman et al., 2003; Cobos et al., 2005; Carney et al., 2009). Our Emx1; Dlx intersectional fate mapping provides direct insight into the putative combinatorial code for the differential specification of striatal and amygdala neurons, as the subpopulation of Emx1-lineage neurons in the striatum appear to require Dlx expression to become inhibitory neurons. This is consistent with previous work using Emx1 and Dlx genetic fate mapping to analyze the origins of TH+ and CR+ interneurons olfactory bulb neurons, which in part appear to be generated from the ventral pallium and the LGE (Kohwi et al., 2007; Batista-Brito et al., 2008), regions in which overlap between Emx1 and Dlx lineages occurs (Kohwi et al., 2007). Consistent with our findings, previous work has indicated that Dlx genes are necessary for the initiation of GABA production, and so it may be that Emx1-lineage cells must also express Dlx family members to become specified to an inhibitory neuronal phenotype (Stühmer et al., 2002; Cobos et al., 2005).
Although our data are strongly suggestive that both timing and genetic heterogeneity are key determinants in the differential fate of striatal versus amygdala Emx1-lineage neurons, the exact mechanisms of this determination are unclear. It is possible that Emx1-lineage neurons in the developing cerebral cortex (or LGE) upregulate Dlx genes (and other genes that specify the ventral program) before they become postmitotic and migrate to the striatum. In support of this, previous analyses have revealed that a subpopulation of cortical progenitors express Dlx2 in the ventricular zone (Nery et al., 2003), and our in vitro differentiation assays shown here reveal that Emx1+ cortical progenitors have the capacity to generate MSNs. Alternatively, the differential fates of ventrally migrating Emx1-lineage cells whose ultimate destination is the striatum or amygdala may also be influenced by external cues at the boundary between the developing cortex and striatum (the PSB). This area expresses a number of molecules that are secreted locally, including Sfrp2 and Tgfα, and the more caudally expressed Fgf7 (Stoykova et al., 1996; Toresson and Campbell, 2001; Carney et al., 2009). As Emx1-lineage cells from the pallium migrate ventrally through the region at the PSB, and more caudally, at the level of the developing amygdala, it is possible that these migrating Emx1+ cells upregulate different transcription factors that endow them with the capacity to migrate to different locations and differentiate into different cell subtypes when compared with Emx1+ cells that remain in the cerebral cortex. Indeed, our observed increase in the numbers of Emx1-derived neurons in the LGE in Sey/Sey mice, which display a disrupted PSB, is consistent with such a role for the PSB in fate and/or migratory route determination.
Footnotes
This work was supported by the following grants: National Institute on Drug Abuse Grant R01 NIDA020140 (J.G.C.), National Research Service Award Fellowship NS564662 (L.A.C.), and Cellular Imaging Core of Mental Retardation and Developmental Disabilities Research Center Grant P30HD40677. We thank the following people for sharing of reagents: K. Campbell for the Gsx2 antibody, R. Hevner for the Tbr1 antibody, K. Yoshikawa for the Dlx2 antibody, M. Baydyuk for the TH antibody, Elizabeth Powell for the MOR1-β antibody, and T. Haydar for the CAG-RFP construct. We thank J.-M. Mangin for assistance with confocal microscopy and electrophysiological data analysis and preparation and M. R. Costa for advice regarding the in vitro differentiation assay. We also thank the following people for their critical comments and reading of this manuscript: V. Gallo, T. Haydar, and J. L. Olmos-Serrano. L.A.C. and J.G.C. designed and planned the experiments and wrote this manuscript. L.A.C. performed all of the experiments with the exception of the electrophysiology, which was done with M.M.H. L.A.C. also performed all of the data analysis and figure preparation. G.M. generated the Dlx-Flpe mouse and provided relevant crosses in the laboratory of G.F. R.S.E.C. performed a subset of BrdU injections and perfusions and provided technical assistance and intellectual input. V.H.S. generated the RCE: dual reporter mice. T.H. conducted preliminary analysis of early Emx1 expression and recombination. K.R.J. provided Emx1-Cre mice.
- Correspondence should be addressed to Joshua G. Corbin, Center for Neuroscience Research, Children's Research Institute, Children's National Medical Center, 111 Michigan Avenue NW, Washington, DC 20010. jcorbin{at}cnmcresearch.org
