Neurogenesis and widespread forebrain migration of distinct GABAergic neurons from the postnatal subventricular zone

Abstract

Most forebrain GABAergic interneurons in rodents are born during embryonic development in the ganglionic eminences (GE) and migrate tangentially into the cortical plate. A subset, however, continues to be generated postnatally in the subventricular zone (SVZ). These interneurons populate the olfactory bulb (OB) reached via migration in the rostral migratory stream (RMS). Employing transgenic mice expressing EGFP in 5-HT₃-positive neurons, we identified additional migratory pathways in the early postnatal brain. Time-lapse imaging experiments revealed massive migration of EGFP-positive cells from the SVZ into numerous forebrain regions, including cortex, striatum, and nucleus accumbens. The neuronal fate of the migratory EGFP-labeled cells was indicated by their doublecortin (DCX) expression. Birthdating experiments, by using 5-bromo-2′-deoxyuridine (BrdU) and retrovirus-based experiments, provided evidence that migrating neuroblasts were born in the SVZ postnatally and developed a distinct GABAergic phenotype. Our results demonstrate that the SVZ is a reservoir of GABAergic interneurons not only for the OB, but also for other cortical and subcortical areas.

Classical neurodevelopmental studies led to the view that the generation of neocortical neurons occurs during embryonic development in the regions of primary neurogenesis (1). In contrast to neocortical glutamatergic neurons that originate in the pallial ventricular zone (VZ), various subpopulations of neocortical GABAergic interneurons are generated in different areas of the subpallial ganglionic eminences (GE), from where they migrate tangentially into the cortical plate (2, 3). Toward the end of embryonic development and continuing postnatally, an additional proliferative process of secondary neurogenesis takes place in the subventricular zone (SVZ), the dentate gyrus (DG), and external cerebellar layer (ECL) (4, 5). The 3 sites of secondary neurogenesis that are displaced from the regions of primary neurogenesis represent germinal zones that proliferate intensively at early postnatal stages and even throughout life (SVZ and DG). However, the impact of secondary neurogenesis is spatially restricted, providing significant numbers of granule neurons only for the olfactory bulb (OB), the hippocampus, and the cerebellar cortex. Because neocortical layers are thought to be devoid of late-generated neurons, neurogenesis of both excitatory and inhibitory neocortical neurons is considered completed before birth. The forebrain SVZ is the main site of secondary neurogenesis (6), harboring stem cells that have been displaced early in development from various parts of the embryonic GE and cortex (7). The SVZ develops as the VZ disappears during the latter third of prenatal development and reaches maximal size during the first postnatal week (8), when it generates the majority of OB granule cells (9, 10). In the early postnatal brain, the SVZ is also the major site for glial cell generation (11).

In this study, we have generated transgenic mice in which EGFP is specifically expressed both in immature and differentiated 5-HT₃-positive GABAergic interneurons. The 5-HT₃ receptors are the only ionotropic receptors in the large serotonin receptor family, and of the 2 subunits 5-HT_3A and 5-HT_3B, only the former is expressed in the brain in subpopulations of GABAergic interneurons (12). By using these mice for in vivo visualization of neuroblasts, we detected massive migration of neuronal precursors from the SVZ not only toward the OB, but also to cortical and subcortical structures during the first 3 postnatal weeks. Additional birthdating and fate mapping experiments demonstrated that these postnatally born cells differentiate and develop a distinct GABAergic phenotype.

Results

Faithful Transgene Expression in 5-HT₃-EGFP Transgenic Mice.

To aid in the identification of 5-HT_3A expressing interneurons, we generated transgenic mice by using the bacterial artificial chromosome (BAC) technology (13) to express the in vivo marker, EGFP, under the control of the 5-HT_3A promoter (Fig. S1A). EGFP expression was similar in offspring from 3 founders and 1 line was chosen for further detailed analysis. The specificity of EGFP expression in transgenic mice was demonstrated by in situ hybridization, electrophysiological, and pharmacological methods (Fig. S1 B–I). Similar to other subtypes of GABAergic interneurons (e.g., parvalbumin-, somatostatin-, and calretinin-positive cells), that were shown to be generated during embryonic development in the GE (1, 14), massive tangential migration of 5-HT₃/EGFP-positive neuroblasts from the caudal ganglionic eminence (CGE) into the cortical plate could be visualized in coronal slices from embryonic animals (E14.5) (Fig. S1 J and K).

At a cellular level, EGFP labeling in the adult occurs in NeuN-positive neurons and does not colocalize with the astrocyte and oligodendrocyte markers GFAP and CNP, respectively (Fig. S2 A–C″). Double labeling experiments demonstrated that EGFP expression is restricted to GABAergic interneurons (Fig. S2 D–D″). EGFP-positive cells comprise a heterogeneous population of GABAergic interneurons that colocalize mainly with calretinin (CR) or cholecystokinin (CCK), but never with parvalbumin or somatostatin (Fig. S2 E–G and Table S1). These results are in agreement with the cortical expression pattern of 5-HT₃ receptors (12).

Transgene Expression Shows Widespread Migratory Patterns of Neuroblasts in the Juvenile Brain.

Much to our surprise, careful expression analysis revealed that, in addition to the expected expression of EGFP in defined GABAergic subpopulations during development and in the adult, fluorescent cells were also present in brain regions associated with postnatal neurogenesis, suchas the rostral migratory stream (RMS) and the OB (Fig. 1A). Thus, in situ hybridization studies indicate the presence of high mRNA levels both for 5-HT_3A and EGFP in the RMS in adult mice (Fig. S1E). Moreover, in sections from young animals (1–4 weeks old), we detected several other pathways comprising numerous small EGFP-positive cells with migratory appearance. Based on their location, we termed these pathways dorsal migratory pathway (DMP), ventral migratory pathway (VMP), and external migratory pathway (EMP). The DMP, best seen on sagittal sections, extended above the hippocampus and was directed toward the occipital cortex (Fig. 1B). The VMP, also detectable on sagittal sections, contained many individual EGFP-positive cells dispersing from the SVZ to the striatum and nucleus accumbens (Fig. 1B). Finally, the EMP could be best visualized in horizontal sections, emerging from anterior parts of the SVZ, and extending along the external capsule toward latero-dorsal brain regions (Fig. 1C). Interestingly, many EGFP-positive cells in the vicinity of the RMS and the other migratory pathways are oriented radially toward the adjacent brain regions and thus appear to have exited the migratory streams (Fig. 1B, Inset).

Fig. 1.

Multiple migratory pathways for neuroblasts are revealed in the postnatal brain of EGFP/5-HT₃ transgenic mice. (A) Overview of EGFP expression in a sagittal section of a P20 transgenic animal. (B) EGFP immunocytochemistry in sagittal sections from P10 transgenic animals showing the RMS, 2 novel pathways termed VMP and DMP, and cells detached from the RMS (Bottom Right). (C) EGFP immunohistochemistry showing the EMP on horizontal sections from P10 transgenic mice. (D–E″) Colocalization of EGFP (green) and DCX (red) in cells in the DMP (D–D″) and EMP (E–E″) in sagittal and horizontal sections, respectively, from P10 transgenic animals. (F and G) Colocalization of EGFP (green) and DCX (red) in the RMS, in cells detached from the RMS in P9 sagittal sections (F), and in migratory cells that have exited the EMP in P16 horizontal sections (indicated by arrows in G). (B, C, F, and G) Plane of sections is indicated in the Inset. [Scale bars: (A) 200 μm, (B) 50 μm, (C) 200 μm, (D–E″) 20 μm, (F) 50 μm, (G) 75 μm.]

To identify the cell type of EGFP-positive cells with migratory appearance in all migratory pathways, we performed double-labeling studies with neuronal and glial markers. In the SVZ and in the newly identified pathways there was no colocalization of EGFP with glial markers (data not shown). We found, however, strong colocalization of EGFP with the markers for immature neurons doublecortin (DCX) (Fig. 1 D–G), Tuj1, and PSA-NCAM (data not shown). Virtually all EGFP-positive cells with immature appearance in all migratory pathways were also DCX-positive (e.g., 98.7 ± 0.4% of all EGFP-positive cells were DCX-positive in the EMP; n = 616 cells, 3 animals). EGFP/DCX expression could also be detected in migratory cells that appear to have exited the different migratory pathways (Fig. 1 F and G). In the cortex, EGFP/DCX-positive cells were located predominantly in lower layers. However, especially in the cingulate cortex, the medial prefrontal cortex (mPFC), and the infralimbic cortex, EGFP/DCX-positive cells could be found all of the way up to cortical layers I–II (data not shown). Interestingly, small EGFP/DCX-positive cells with immature appearance were never detected in the DG. Thus, 5-HT₃ expression in neuronal precursors in the postnatal brain is specific for GABAergic SVZ-generated cells and not granule cell precursors that derive from the DG.

Time-Lapse Imaging Studies Reveal Migration of Neuroblasts into Cortical Structures in the Postnatal Brain.

To directly demonstrate that EGFP-labeled neuroblasts exit the streams and migrate into adjacent cortical regions, time-lapse imaging experiments were performed in acute slices fromtransgenic mice (P7–P19) for several hours (3–20 h) (Fig. 2, Fig. S3, and Movies S1–S6). EGFP-labeled cells in RMS and all pathways moved bidirectionally (Movie S1 showing migration in the VMP, RMS, and striatum at P16). Numerous cells exiting the streams and showing directed movement toward the cortex could be visualized (Movies S2 and S3 for the RMS and Movie S4 for the DMP). The quantitative evaluation of the movement of EGFP-positive cells, visualized in Movies S2 (P7) and S4 (P10), revealed that 75.7% of the moving cells migrated toward the pial surface. Interestingly, the direction of migration for 24.3% of the cells was toward the RMS or DMP (n = 144) (Fig. 2 A and B). During experiments that lasted several hours, neuroblasts that exited the stream could be followed all of the way up to layer IV (Movie S5 represents a higher magnification of the indicated cells in Fig. S3B). Both in the immediate vicinity of the streams and in the cortex, migration of individual neuroblasts and chain-like migration of groups of cells could be visualized, the latter being more preponderant close to the streams (Movies S2 and S6, and Fig. S4).

Fig. 2.

EGFP-positive neuroblasts migrate from the RMS and DMP into the cortex. (A) Cell tracking of EGFP-positive cells from a 20-h movie (Movie S2) in a P7 sagittal slice, by using imageJ software. Colored dots represent the starting point of the cell trajectory. (B) Quantification of the migratory EGFP-positive cells from the Movies S2 and S4. The number of moving cells at P7 from the RMS and at P10 from DMP into the cortex is indicated on the y-axis and the direction of movement (expressed as angle between their trajectory and the stream) on the x-axis. Data were pooled because there was no difference between the 2 ages. (C) Higher magnification of 11 frames taken from Movie S2 in a P7 sagittal slice. Time elapsed between frames is 110 min. (←) Speed of migration was 9 μm/h for the indicated cell. (*) Another migrating neuroblast. The size of each frame is 60 × 240 μm.

Postnatal Generation and Fate Mapping of 5-HT₃-EGFP-Positive Neurons by BrdU Experiments.

5-bromo-2′-deoxyuridine (BrdU) injections (20 mg/kg) at early postnatal stages (P4) were carried out to determine time of birth for postnatally migrating EGFP-positive neuroblasts. At 72 h post-BrdU injection, BrdU/EGFP double-positive cells were detected exclusively in the SVZ, RMS, DMP, and their immediate vicinity, indicating that EGFP-positive neuroblasts were derived from mitotically active progenitors located in the SVZ or the streams (Fig. 3 A and B). After 10 days, numerous BrdU/EGFP/DCX triple-positive cells could already be identified in brain structures more distant from the RMS and other pathways, such as in deep cortical layers of the frontal cortex (Fig. 3 C–D′), in more dorsal areas of the cortex (Fig. 3 E and E′), and in many subcortical forebrain areas such as the striatum and nucleus accumbens (Fig. 3 F and F″). A quantitative evaluation in the cortical region above the RMS in P4-injected mice revealed that 21.7 ± 0.6% (n = 590 cells, 3 animals) of all BrdU-positive cells were EGFP-labeled cells.

Fig. 3.

BrdU birthdating and differentiation of postnatally generated EGFP-positive interneurons. (A and B) Location of EGFP/BrdU double-positive cells 72 h after BrdU injection at P4 in the DMP (A) and at the posterior edge of the alveus of the hippocampus (B). (C and D′) Numerous EGFP/BrdU-positive cells with migratory appearance in layer VI of the frontal cortex, 10 days after BrdU injections at P4. (D and D′) EGFP/BrdU-positive cell with morphology indicative for somal translocation. Leading and lagging process of the cell are indicated by arrows in D′. (E–F″) Triple stainings showing colocalization of EGFP (green), DCX (blue), and BrdU (red) in the secondary visual cortex (E and E′), striatum (F′), and around the anterior commissure (F″) 10 days after BrdU injections at P4. Within the striatum (F), both single EGFP-positive cells and an impressive number of chains of migratory cells (arrows) can be detected. (C′, D′, E′, F′, and F″) Enlargements of the boxed areas in C, D, E, and F, respectively. Colocalization is indicated by arrows (B, C′, E′, F′, and F″). (G) Triple stainings illustrating the colocalization of EGFP (green), NeuN (blue), and BrdU (red) in layer VI of the primary somatosensory cortex, 28 days after BrdU injections at P4. (H–I′) Examples of triple-labeled EGFP (green), CR (blue), and BrdU (red) cells in striatum (H) and in layer IV of the frontal cortex (I and I′), 28 days after BrdU injections at P4 (H) or at P7 (I and I′). (I′) Enlargement of Inset in I. Cx, cortex; Hi, hippocampus; CPu, caudate putamen. [Scale bars: (A, C, D, E, F, I) 200 μm; (C′, I′) 50 μm; (B, D′, E′, F′, F″, G, H) 10 μm.]

To determine the fate and distribution of postnatally generated EGFP-positive cells, transgenic mice were analyzed 28 days after BrdU injections at P4, P7, P11, P15, and P30. BrdU/EGFP-positive cells were found in many cortical and subcortical areas and had by this stage reached maturity. BrdU/EGFP-positive cells expressed NeuN, a marker for mature neurons (Fig. 3G). Similar to the well-studied OB, where the generation of new neurons declines rapidly in the first 3 weeks after birth (15), we found a gradual decrease of newly generated 5-HT₃-positive cells located in the cortex (Table 1). However, in the vicinity of the RMS, individual neuroblasts directed toward the cortex could still be identified in both transgenic and wild-type adult animals (Fig. S5 A and B). Results obtained with BrdU injections at P90 are comparable to those at P30: Few newborn 5-HT₃-positive neurons with similar characteristics as in young animals could be detected in lower cortical layers (Fig. S5 C–G ‴).

Table 1.

Semiquantitative evaluation and final cortical location of EGFP-positive interneurons generated before and after birth

Age at treatment, days	Treatment stage	Upper layers*	Lower layers*	Total	No. of cells, n†
12.5	Embryonic‡	70%	46%	60.6%	n = 546
14.5	Embryonic	72.8%	47.2%	60%	n = 590
16.5	Embryonic	11.1%	26.1%	18.1%	n = 772
19.5	Embryonic	4.5%	20%	14.3%	n = 362
4	Postnatal§	2.7%	19.6%	14.2%	n = 1540
7	Postnatal	1.2%	13.7%	10.5%	n = 648
11	Postnatal	1.5%	2.4%	2%	n = 902
15	Postnatal	0.6%	2.4%	1.6%	n = 1817
30	Postnatal	0%	0.7%	0.4%	n = 472

BrdU/EGFP-double positive cells were counted 1 month after BrdU injections at the indicated ages.

*Numbers represent the percentage of EGFP-positive cells that also express BrdU.

^†“n” is the number of EGFP-positive cells counted.

^‡Embryonic treatment: 50 mg/kg, 5 times/day, administered i.p. to pregnant mice to establish birthdating during embryonic development based on previous studies (Xu Q, Cobos I, De La Cruz E, Rubenstein JL, Anderson SA (2004) Origins of cortical interneuron subtypes. J Neurosci 24:2612–2622) investigating birthdating of GABAergic interneurons.

^§Postnatal treatment: 20 mg/kg, 5 times/day, injected i.p., and the protocol had to be slightly modified to overcome toxicity.

In agreement with other studies (14), a large fraction of cortical EGFP-positive interneurons were born during embryogenesis in the GE (Table 1 and Fig. S1K). Also, as previously suggested for other types of GABAergic interneurons (3), the final distribution of 5-HT₃ cells seems to depend on the time and place of origin. Whereas most of E12–14 born 5-HT₃ cells populated the upper layers, postnatally born cells were mainly found in lower layers (Table 1). Furthermore, the cortical distribution of postnatally born 5-HT₃ cells was not uniform along the medio-lateral axis: In medio-frontal areas, such as the mPFC, the cingulated and infralimbic cortex BrdU/EGFP-positive cells were distributed in all cortical layers ≤ layer I. In more lateral regions of the cortex, BrdU/EGFP-positive cells were found exclusively in lower cortical layers adjacent to the callosal system, and were practically absent on very lateral sections (Fig. S6).

Postnatally generated 5-HT₃/EGFP-positive cells do not comprise a morphologically homogenous population. The majority coexpressed the interneuronal marker CR (Fig. 3 H–I′). Thus, quantitative evaluations of the data deriving from triple-labeling experiments after BrdU injections at P4 indicated that 83.1 ± 2.2% (n = 346 cells, 4 animals) of BrdU/EGFP-positive cells in the frontal cortex and 85.6 ± 2.8% (n = 219 cells, 4 animals) in the striatum were CR-positive. BrdU/EGFP/CR triple-positive cells could be identified in many brain regions including the caudate-putamen (Fig. 3H), the mPFC, the infralimbic cortex, the cingulate cortex, the olfactory tubercle, and the nucleus accumbens (data not shown). These cells exhibit monopolar morphology, a small-sized cell body, and intense CR-immunoreactivity. There is, however, yet another subpopulation of postnatally born 5-HT₃/EGFP-positive neurons that were CR-negative, showed multipolar morphology, and were often localized in cortical layers II–IV. Of note, bipolar 5-HT₃/EGFP/CR/BrdU-positive cells in upper cortical layers were found exclusively when injected during embryonic development (between E12.5–E14.5) and corresponded to the cell population described previously by Yozu et al. (16). Representative examples of embryonically and postnatally generated 5-HT₃ /EGFP-positive cells, illustrating their different morphology, are shown in Fig. S6.

Migration Pattern and Fate Mapping of EGFP-Positive Cells Using Viral Injections.

To follow newborn migratory cells in vivo, we carried out viral-mediated cell labeling by using a replication-incompetent retroviral vector expressing red fluorescent protein (RFP) (17). We examined the final destination of labeled cells 2 weeks after viral delivery into the SVZ of P3–4-old animals (11 animals analyzed). In agreement with previous studies, numerous viral-infected neurons were found in the OB, and glial cells (astrocytes and oligodendrocytes) were identified in the cortex and striatum (18) (data not shown). In addition, however, we also detected NeuN/EGFP/RFP- and CR/EGFP/RFP-positive cells in different cortical layers (Fig. 4 D–I), indicating that postnatally generated neuroblasts from the SVZ migrate and develop into mature 5-HT₃-positive neurons in the cortex.

Fig. 4.

EGFP-positive neurons infected with retrovirus in the SVZ reach different cortical areas. (A and B) Reconstructions of a sagittal section from a P18 mouse infected with retrovirus-RFP in the SVZ at P4, showing EGFP (A) and RFP (B) labeling. Red arrow indicates the injection site. (C and D) Enlargements of Insets in A and B showing examples of triple-labeled EGFP (green), RFP (red), and NeuN (blue) immunostained neurons in the cortex. (E) Example of EGFP (green), RFP (red), and CR (blue) triple-positive neuron detected in cortical layer V of a transgenic mouse 2 weeks after retroviral injection. [Scale bars: (B) 200 μm; (C–E) 10 μm.]

The unlikely scenario that neuroblast migration occurred as a consequence of the genetic manipulation in the transgenic animals was ruled out by several control experiments. Firstly, neuroblasts in wild-type animals exhibit similar migration patterns as those observed in transgenic mice as revealed by DCX stainings. In P10 animals, the EMP and DMP, as well as DCX-positive cells detaching from the streams, could be identified (Fig. S7A). Secondly, retroviral experiments revealed the existence of newborn DCX-positive neuroblasts, migrating out of the RMS 6 days after infection in wild-type animals (Fig. S7 B–B ″). Finally, we analyzed retroviral-infected wild-type animals 2 weeks after virus delivery into the SVZ and, similar to the experiments in transgenic animals, we found NeuN/RFP double-positive neurons in the cortex, remote from the injection site (Fig. S7 C–C″).

Discussion

Here we describe the identification and characterization of migratory patterns for 5-HT₃ interneurons and their precursor cells during postnatal development. These experiments rely primarily on the faithful expression of EGFP, transgenically driven by the 5-HT₃ promoter in the mouse brain, and were validated in wild-type mice. In addition to the expected expression in distinct GABAergic interneurons, EGFP positivity could also be detected in the postnatal SVZ and RMS. Unexpectedly, we found massive migration of neuroblasts from the SVZ that was not restricted to the RMS and OB, but included numerous cortical and subcortical regions. We show that in addition to the well-known and much-studied RMS, neuroblasts from the postnatal SVZ migrate extensively along all structures and elongations of the callosal system before they disperse as individual cells into adjacent cortical and subcortical areas. According to their location, corresponding to parts of the dorsal corpus callosum and external capsule, we named these new pathways DMP and EMP, respectively. The former most likely corresponds to the structure that the authors of a previous study described as the “subcallosal zone” (19). Although it was initially described as a germinal zone that produces mainly oligodendrocytes, subsequent studies suggested that this dorsal area is also able to produce interneurons after birth (20, 21). The migratory pathway directed from the SVZ to the nucleus accumbens we termed VMP, and the migratory cells in this pathway are probably identical to those described by De Marchis and colleagues (22).

The neuronal fate of the EGFP-positive cells in the RMS and in the migratory pathways was demonstrated by expression of markers for immature neurons, such as DCX. At first glance it appears puzzling that the extensive migration of neuroblasts from the early postnatal SVZ described here has so far remained unknown. However, this is not surprising considering that most detailed DCX expression studies have been carried out in adult rodent brain (23–24). There has been a paucity of similar investigations for the juvenile brain. Furthermore, DCX expression is transient and does not allow the investigation of subsequent neuroblast maturation. The extent and complexity of postnatal migration of this distinct GABAergic subpopulation can, however, be easily visualized and followed in the time-lapse experiments by using brain slices from young 5-HT₃/EGFP transgenic animals.

Birthdating experiments demonstrated that migrating 5-HT₃/EGFP-positive neuroblasts that populated other forebrain regions than the OB were born postnatally. BrdU labeling has been routinely used over the last decades to demonstrate neurogenesis during embryonic development and in the adult. However, it is associated with some limitations because of toxicity and rapid clearance from the brain (25). It is possible that pronounced BrdU toxicity in the juvenile brain or insufficient labeling by just one pulse of BrdU may be the reason why this population was not detected before. The results obtained by BrdU labeling were extended by fate-mapping experiments using virus mediated cell-labeling (26). The fact that more than 20% of all P4-labeled newborn cells found in the cortex were EGFP-positive, together with the visualization of the movement of these cells by time-lapse imaging, demonstrates that the migration of neuroblasts from the RMS into the frontal cortex is quantitatively a robust phenomenon. Although it is of note that there is a marked reduction of SVZ neurogenesis during postnatal development that affects all postnatal interneuron generation including those destined for the OB (9, 10). However, the addition of newborn 5-HT₃-positive interneurons in the cortex does not cease completely in adult mice, as was previously suggested also for other species (27–28).

Like many other GABAergic interneuron subclasses (29–31), 5-HT₃-positive interneurons do not constitute a homogenous subclass, but rather several subpopulations of GABAergic interneurons whose fate is determined, at least in part, by the time and the place of their birth. Thus, a large number of 5-HT₃-positive interneurons were generated during prenatal development with a peak around E14. A large proportion of these cells coexpress CR. These data are in agreement with previous birth-dating experiments demonstrating that CR-positive cells, particularly bipolar interneurons in upper cortical layers, are generated around E14 in the CGE (2, 32, 33). Also, for CR-expressing neurons a unique outside-in neurogenesis gradient has been demonstrated (16, 34). In this study we provided evidence that another subpopulation of 5-HT₃-positive interneurons continues to be generated postnatally and their site of origin is the SVZ. This cell population is heterogenous when coexpression of CR or cell morphology is taken into account. Whilst a small proportion of postnatally born 5-HT₃-positive cells were CR-negative and had a multipolar cell body, the majority of postnatally born 5-HT₃-positive cells were CR-positive, were preferentially located in lower cortical layers, had a small cell body and 1 main process.

Elegant in vivo studies demonstrated that subclasses of GABAergic interneurons differentially contribute to defined network activity that in turn is associated with specific behavior (35). The functional role of the cell population described here at the circuit level and for behavior remains to be established. The findings in this study might be relevant in the context of diseases resulting from abnormal development of the cell population described here. Interestingly, it has been reported that the human SVZ is particularly well developed in comparison to other species, and the majority of GABAergic interneurons in humans arise from the VZ and SVZ (36, 37). Moreover, regions with the highest number of postnatally generated 5-HT₃-positive cells, such as the prefrontal cortex, are believed to be critically involved in the pathogenesis of several neuropsychiatric disorders with presumed neurodevelopmental background. Particularly schizophrenia is thought to be associated with disturbances in the prefrontal cortex during postnatal/juvenile development (38, 39). Future studies aiming at interrupting the migration and/or differentiation of this cell type should help to establish the link between malfunction of the brain as a result of abnormal development.

Materials and Methods

Generation of 5-HT₃-EGFP Transgenic Mice.

A transgene was constructed by homologous recombination of a bacterial artificial chromosome (BAC) containing the 5-HT_3A gene and used for pronucleus injection into mouse zygotes. Technical details about constructs, cloning, and analysis, demonstrating the correct expression of the transgene, are provided in SI Methods.

Immunohistochemistry.

Immunostaining was carried out on 50-μm or 100-μm free-floating sections. The following primary antibodies were used: Polyclonal rabbit anti-EGFP antibody, 1:10,000 (Molecular Probes); the polyclonal chicken anti-EGFP antibody (Abcam, 1:2,000; or the monoclonal anti-EGFP antibody, 1:300 (Chemicon); goat anti-doublecortin, 1:500 (Santa Cruz); mouse or rabbit anti-calretinin antibody, each at 1:5,000 (Swant); mouse anti-NeuN, 1:1,000 (Chemicon); and rat anti-BrdU antibody, 1:400 (Accurate). For visualization of primary antibodies, slices were incubated with FITC- or Alexa 488- (Molecular Probes) conjugated anti-rabbit, anti-chicken, anti-mouse, or anti-goat IgG; anti-mouse, anti-goat, or anti-rat Cy3-coupled secondary antibody; and anti-mouse or anti-rabbit Cy5-coupled secondary antibody (Jackson Immuno Research Laboratories). Sections were analyzed by using an upright fluorescent microscope (Zeiss Axioplan 2), a LEICA TCS-NT confocal microscope (Leica Microsystems), or a Zeiss Axiovert 200M confocal microscope.

Time-Lapse Imaging.

Sagittal brain sections at 250-μm thickness were generated from transgenic mice (P7–P19) by using a vibratome (Leica VT1000S; Leica Microsystems). A stack of 21 frames (spanning 100 μm) was taken every minute with an Olympus microscope and a 20×:0.5NA and a 40× objective. A focus projection of the stack was generated by using ImageJ for visualization of cells throughout the stack.

Birth-Dating Analysis.

Pregnant mice were injected i.p. five times with 50-mg/kg BrdU (Sigma) (body weight) every 2 h at E12.5, E14.5, E16.5, and E19.5. For postnatal neurogenesis, the same regime at 20 mg/kg was used at P4, P7, P11, P15, and P30. Adult (P90) transgenic animals were injected i.p. with 50-mg/kg BrdU twice per day, during 3 days. Animals were killed after 3, 10, or 28 days. The experiments were done in accordance with institutional guidelines and were approved by the local Committee on Animal Care and Use (Karlsruhe, Germany). To determine the number of EGFP/BrdU-positive cells, every sixth section of 50 μm (300-μm intervals) of 1 cerebral hemisphere from each animal was processed for immunohistochemistry as previously described. EGFP/BrdU-positive cells were counted by using a LEICA TCS-NT confocal microscope with a 40× oil-immersion objective. Colocalizing cells were identified and counted by confocal Z sectioning within two 40× magnification optical fields for each region in the deep (layers IV–VI) and superficial (layers I–III) frontal cortex above the RMS.

Retroviral Production.

We used a replication-deficient Moloney murine leukemia retrovirus expressing RFP under the control of CAG promoter, kindly provided by Dr. F. H. Gage (Salk Institute, La Jolla, CA). Viral preparation was performed as previously described (17). Briefly, retroviral particles were produced in the packaging-cell line HEK 293 (Human Embryonic Kidney 293). Three different plasmids (CMV-gag/pol, CMV-vsvg, and CAG-RFP) were transfected into the cells by using calcium phosphate precipitation. Virus containing media were collected after 48 h, concentrated by ultracentrifugation, and purified through a sucrose cushion.

Viral Injections.

Postnatal day 3–4 EGFP-positive or wild-type mice from the same litters were anesthetized by hypothermia. The virus solution (1.5 μl) was delivered to the right lateral ventricle of the pups with a glass capillary by stereotaxic injections by using the following coordinates (relative to bregma in mm): Anterior, 0.4; lateral, 1; ventricular, 1.8. Animals were killed after 6 or 14–15 days after surgery.