The mammalian subventricular zone (SVZ) contains progenitors derived from cerebral cortex radial glia cells, which give rise to glutamatergic pyramidal neurons during embryogenesis. However, during postnatal life, SVZ generates neurons that migrate and differentiate into olfactory bulb γ-aminobutyric acid (GABA)ergic interneurons. In this work, we tested if SVZ cells are able to produce glutamatergic neurons if confronted with the embryonic cortical ventricular zone environment. Different from typical SVZ chain migration, cells from P9–P11 SVZ explants migrate into embryonic cortical slices individually, many of which radially oriented. An average of 82.5% of green fluorescent protein–positive cells were immunolabeled for neuronal marker class III β-tubulin. Invading cells differentiate into multiple morphologies, including a pyramidal-like morphotype. A subset of these cells are GABAergic; however, about 28% of SVZ-derived cells are immunoreactive for glutamate. Adult SVZ explants also give rise to glutamatergic neurons in these conditions. Taken together, our results indicate that SVZ can be a source of glutamatergic cortical neurons when submitted to proper environmental cues.
The subventricular zone (SVZ) is a continuous source of new neurons in postnatal and adult mammals (for review, see Gage 2000). The newly generated neurons migrate to the ipsilateral olfactory bulb (OB; Altman 1969; Luskin 1993) where they differentiate into granular and periglomerular neurons (Luskin 1993). A known feature of these cells is that they differentiate into γ-aminobutyric acid (GABA)ergic neurons (Betarbet et al. 1996). It was recently described that a subpopulation of periglomerular neurons could be devoid of GABA (Gutièrrez-Mecinas et al. 2005). However, recent evidence shows that all OB periglomerular neurons activate the promoter of the gene for 1 of the 2 forms of glutamic acid decarboxilase, 67KD form of glutamic acid decarboxilase (GAD67), as detected by green fluorescent protein (GFP) expression in transgenic mice (Panzanelli et al. 2007). In addition, a subpopulation of periglomerular neurons (Kosaka et al. 1985; Betarbet et al. 1996; Hack et al. 2005) and superficial granular neurons (Kohwi et al. 2005) are double labeled for GABA and tyrosine hydroxylase.
It is suggested that the telencephalic embryonic germinal layer, the ventricular zone (VZ), gives rise to different neurochemical phenotypes in a spatial-dependent fashion (Marin and Rubenstein 2001). In the telencephalon, glutamatergic neurons are generated by the dorsal VZ (Schuurmans et al. 2004) and GABAergic neurons from its ventral domain (Anderson et al. 1997). At the end of embryonic neurogenesis, radial glia (RG), the main progenitors of the VZ (Noctor et al. 2001), start to transform into other cell types, including SVZ astrocytes (Alves et al. 2002; Tramontin et al. 2003) and ependymal cells (Spassky et al. 2005). Postnatal progenitors, derived from RG, are also known to give rise to inhibitory interneurons of the OB, irrespective of dorsoventral position (Merkle et al. 2007). These observations lead to the question of whether dorsal progenitors, which initially make glutamatergic neurons, respecify their program for neurotransmitter choice after embryonic neurogenesis ends. An alternative possibility is that dorsal telencephalic progenitors are still capable of giving rise to glutamatergic neurons, and the external signals present in the postnatal brain do not allow or stimulate this phenotype. SVZ neuroblasts already express GABA in route to the OB (Bolteus and Bordey 2004), suggesting an early commitment by these cells with a GABAergic destiny.
Recent data suggest that most GABA-immunoreactive neuroblasts of the SVZ are not endowed with the GABA synthetic machinery typical of differentiated neurons (De Marchis et al. 2004; Sequerra et al. 2007). Instead, an alternative synthetic pathway has been identified, which converts putrescine into GABA (Sequerra et al. 2007). Therefore, neuroblasts in the SVZ may not be irreversibly committed to a GABAergic destiny. The question herein addressed is whether the dorsal embryonic VZ environment could drive postnatal and adult SVZ progenitors to a glutamatergic phenotype.
Materials and Methods
In this work, we used wild-type postnatal and embryonic mice (Mus musculus, Swiss), transgenic mice with the expression of enhanced green fluorescent protein (EGFP) driven by the chicken β-actin promoter and cytomegalovirus enhancer (TgEGFP+ BCF1 background; Okabe et al. 1997). Procedures for the use of animals were in accordance with the Committee for Ethics in the Use of Animals for Research of the Instituto de Biofísica Carlos Chagas Filho of the Universidade Federal do Rio de Janeiro and follow National Institutes of Health’s guidelines for animal research.
To obtain embryonic slices, pregnant mothers received an ip injection of chloropentobarbital (60 mg/mL sodic pentobarbital and 0.042 mg/mL chloral hydrate) of 3.5 mL/kg of body weight, with 30% more volume added after 5 min. Embryos were removed by C-section and decapitated. For postnatal explants, animals were anesthetized with ether inhalation before decapitation. For adult explants, animals were anesthetized with ether inhalation and euthanized by cervical dislocation.
Briefly, embryonic and postnatal brains were removed from crania in cold (4 °C) Gey’s basal salt solution in sterile conditions. The brains were coronally sliced using a tissue chopper (McIlwain, O’Fallon, MO) at 350 μm thickness. Slices and explants were plated on membranes permeable to gases (Petriperm; Sigma, Saint Louis, MO) coated with Poly-L-lysine (Sigma; 10 mg/mL) and covered with culture medium containing 60% Dulbecco’s Modified Eagles Medium (GIBCO, Grand Island, NY), 30% Hank’s solution (GIBCO), 10% fetal calf serum (GIBCO), 1% solution of penicillin (10 000 U/L) + streptomycin (10 mg/mL) (GIBCO), 1% Fungizone (GIBCO), and 1% glucose. Slices were cultivated at 37 °C for up to 5 days in a 5% CO2/95% air incubator (Heraeus Instruments, Hanau, Germany), and culture medium was entirely changed after 12 and 72 h.
We performed 2 coculture groups (Supplementary Fig. 1): 1) Heterochronic and heterotopic cocultures: Postnatal or adult SVZ explants were cocultured with telencephalic embryonic coronal slices. In this culture model, SVZ explants obtained from P9–P11 (birthday—P0) or 3-month-old TgEGFP+ mice were plated inside the ventricular lumen of E15 telencephalic coronal slices touching its dorsal ventricular surface. Slices used represent the rostral 2/3 of the embryonic telencephalon. SVZ explants were dissected from the ventricular wall adjacent to the striatum and from the SVZ rostral projection present in OB coronal slices (see also Sequerra et al. 2007). 2) Homotopic and homochronic cocultures: Postnatal SVZ explants were cocultured with postnatal OB slices; P9 OB coronal slices were plated after removal of the SVZ. Then, TgEGFP+ SVZ explants were dissected from OB coronal slices and used as replacement for the homotopic SVZ that was removed. The GFP+ SVZ explant was plated touching the internal face of the granular layer exposed by the excision of the original SVZ.
In Vitro and In Situ Immunohistochemistry
For immunohistochemistry, cultures were prefixed in formaldehyde vapor by placing the Petriperm culture plates without their lids into a chamber containing paper towel embedded with formaldehyde for 10 min. Subsequently, medium was removed and cultures were immersed in buffered 4% paraformaldehyde for 15 min. Cultures processed for glutamate immunohistochemistry were immersed in buffered 4% paraformaldehyde/1% glutaraldehyde. After several washes in phosphate-buffered saline (PBS; pH 7.4), cocultures were incubated for 72 h at 4 °C with primary antibodies diluted in PBS with 0.3% Triton-X 100 (Reagen, Curitiba, PR, Brazil) and 5% normal goat serum (Invitrogen, Carlsbad, CA). Primary antibodies were the following: anti-GABA (rabbit 1:2000; Sigma), anti-glutamate (rabbit 1:250; Chemicon, Temecula, CA; antibody made against vesicular glutamate), anti–class III β-tubulin (Tuj1, mouse 1:500; Covance, Emeryville, CA), anti-glial fibrillary acidic protein (GFAP; rabbit 1:400; Dako, Carpinteria, CA), anti-S100β (rabbit 1:500; Sigma), anti-Tbr1 (1:200; Chemicon), anti-Ctip2 (rat 1:500; Abcam, Cambridge, MA), anti-Satb2 (mouse 1:200; Abcam), and anti-PSD-95 (rabbit 1:100; Santa Cruz, Santa Cruz, CA). Secondary antibodies were incubated for 2 h at room temperature. Secondary antibodies were the following: goat anti-rabbit IgG conjugated with Cy3 (1:800; Jackson ImmunoResearch, West Grove, PA), goat anti-mouse IgG conjugated with Cy3 (1:800; Jackson ImmunoResearch), and goat anti-rabbit conjugated with Cy5 (1:500; Jackson ImmunoResearch). Pieces of Petriperm membrane were then mounted on glass slides and coverslipped using gel-mounting medium (Biomeda, Foster City, CA).
For in situ immunohistochemistry, P11 Swiss mice (2 animals) were ether anaesthetized and intracardially perfused with 4% paraformaldehyde and 1% glutaraldehyde in phosphate buffer. Brains were removed and sectioned with a vibratome (Vibratome 3000; Pelco, Redding, CA) at 50 μm. Free-floating sections were processed as described above for cocultures. Sections were mounted on glass slides and coverslipped using gel-mounting medium (Biomeda).
Microscopy and Quantification
Conventional fluorescence microscopy was performed in an Eclipse TE200 inverted microscope (Nikon, Tokyo, Japan), equipped with a CoolSNAP-Pro cf CCD camera (Media Cybernetics, Silver Spring, MD; monochrome). Images were acquired with the aid of Image-Pro Express software (version 220.127.116.11) and edited with Photoshop CS2 (Adobe, San Jose, CA). In one particular figure (Supplementary Fig. 3A,B), a Zeiss Inverted Fluorescent Microscope (Axiovert 200M) equipped with an ApoTome structural microscopy module (Zeiss, Gottingen, Germany) was used. Confocal microscopy was performed in a Zeiss Axiovert 200 microscope equipped with LSM 510 Meta NLO confocal system. Images were collected and processed with the LSM Image Browser software (Zeiss). Neurons were classified into 5 morphotypes (pyramidal, horizontal bipolar, multipolar, periglomerular like, and nondescript) based on dendritic patterns and the existence of a thin process identifiable as an axon. These morphotypes were counted as a percentage of total EGFP-positive differentiated cells (n = 11 slices in 3 independent experiments, 30 animals, 76 cells, and field visualization at ×63 objective—oil). Typical migratory morphologies were excluded from the quantification. We also counted the percentage of EGFP cells that were positive for the neuronal marker class III β-tubulin under ×40 objectives (n = 11 slices in 2 independent experiments, 19 animals used, and 273 cells) and the percentage of EGFP cells that were labeled with anti-glutamate antibody (n = 8 slices in 2 independent experiments, 18 animals used, 312 cells, and field visualization at ×63 objective—oil). In all cases, all labeled cells were counted within focal depths that included immunolabeling. Numbers are expressed as mean ± SEM.
SVZ Cells Invade Embryonic Telencephalic Slices and Differentiate into Multiple Morphotypes
To test possible effects of the dorsal VZ environment on the differentiation of SVZ progenitors, we cocultured telencephalic slices with SVZ explants derived from constitutively expressing GFP transgenic mice at P9–P11 (Okabe et al. 1997; Supplementary Fig. 1). After 2 days in vitro (DIV), SVZ-derived cells had entered the slice and showed bipolar morphologies typical of migrating cells (Fig. 1A). These cells were oriented both radially (arrows in Fig. 1A) and tangentially (arrowhead in Fig. 1A) and were found dispersed throughout the cortical depth (Fig. 1B,C, arrows) without any particular laminar pattern. An average of 82.5 ± 3.8% of GFP-positive cells (11 slices in 2 independent experiments) were identified as immature neurons since they were immunolabeled for class III β-tubulin (Fig. 2E, Tuj1 antibody; Menezes and Luskin 1994).
After 3 DIV, SVZ cells derived from EGFP mice started to differentiate within the wild-type cortical plate (CP) into multiple morphotypes, some of them resembling cortical (Fig. 1B–F) rather than OB neurons. Analysis of differentiated cell types at 5 DIV revealed pyramidal-like cells (Fig. 1C; 15.5 ± 4.7% of EGFP cells, 11 slices, and 2 independent experiments), characterized by a pyramidal-shaped cell body, a prominent apical dendrite oriented to the pia, basal dendrites, and a thin axon running in the opposite direction, toward the white matter (Fig. 1C; Peters and Kara 1985). Another characteristic of these cells is the presence of small processes sprouting from the dendrites, some of which are filopodia arising from the dendritic shafts and some shorter possibly emerging dendritic spines (Fig. 1D). Other cell types were found, such as horizontal bipolar neurons (Fig. 1E; 5.68 ± 1.96%), multipolar cells with highly branched neurites radiating in all directions (Fig. 1F; 55.03 ± 7.81%), morphologically nondescript (Fig. 1G; 1.30 ± 1.30%), and cells resembling OB periglomerular neurons (Fig. 1H; 23.03 ± 8.51%). The latter subtype is characterized by absence of an axon and presence of a single thick process that branches into multiple thinner putative dendrites, arising from the soma (Schneider and Macrides 1978).
Glial SVZ Cells Do Not Migrate into the Embryonic Telencephalic Tissue
We observed SVZ-derived cells differentiated into multipolar morphological types that were compatible with neuronal and astrocytic phenotypes. To test if SVZ-derived cells were also present as astrocytes in the embryonic dorsal telencephalic slice, we performed immunohistochemistry for GFAP and S100β, astrocytic lineage markers. SVZ explants contained many GFAP+ cells, and these were able to migrate centrifugally out of the explant into the culture plate (Fig. 2A). However, GFP+ cells in the dorsal telencephalic tissue did not express GFAP (Fig. 2A, arrowheads) or S100β (Fig. 2D). The few GFP+/GFAP+ cells found within the slice (Fig. 2B,C, 4 GFP+ cells out of 118 in 3 slices from independent experiments) were restricted to the presumptive VZ. Absence of labeling by poor antibody penetration has been ruled out since GFAP+ cells from the host were found within the slice (GFAP+/GFP−, data not shown).
SVZ-Derived Pyramidal-Like Neurons Are Glutamatergic and Express PSD-95 in Short Filopodia Arising from Dendritic Shafts
We have analyzed GABA and glutamate expression by SVZ-derived neurons within the embryonic telencephalic tissue. We found multiple SVZ-derived cells expressing GABA in the CP (Fig. 3A), some presenting a typical multipolar morphology (Fig. 3A). Double labeling was confirmed with confocal reconstruction (Fig. 3B,B′).
To search for a possible glutamatergic phenotype, we immunolabeled cultured slices with an antibody that recognizes synaptic vesicle glutamate fixed with glutaraldehyde. Glutamate immunolabeling was restricted to the CP (Fig. 3C) as expected for normal cortical distribution of glutamatergic neurons. SVZ-derived cells that disperse tangentially underneath the CP in the intermediate zone and SVZ are not labeled with glutamate (Fig. 3C, arrow). After 5 DIV, GFP+ pyramidal-like neurons were found labeled with glutamate within the CP (Fig. 3D,E). A mean of 27.6 ± 11.6% (n = 5) of the GFP-positive SVZ-derived cells were double labeled for glutamate. These double-labeled cells include cells with immature morphology that could not be classified. All the pyramidal-like cells in these slices were glutamate positive, and no other morphological type was found labeled. Therefore, the postnatal SVZ contains progenitors capable of differentiating into glutamatergic neurons of the cerebral cortex. We have also tested Tbr1 immunoreactivity, a marker for a subpopulation of cortical glutamatergic neurons (Hevner et al. 2001) in pyramidal-shaped GFP+ cells, with no observable immunostaining (Fig. 3F, 19 slices in 2 independent experiments). Additionally, we tested if these cells are immunoreactive for Ctip2 or Satb2 since these markers are expressed in deep and superficial cortical layer neurons, respectively (Arlotta et al. 2005; Britanova et al. 2005). We found no double labeling (Supplementary Fig. 2), although both antibodies labeled neighboring cells in the cortical tissue.
An important feature of our morphological characterization of the pyramidal-like neurons is the presence of small protrusions on the dendrites. These resemble short dendritic filopodia as most of them lack the morphological features of more mature dendritic spines, such as bulbous heads and necks (Yuste and Bonhoeffer 2004). However, their position on the dendritic shafts and the spacing of these filopodia were compatible with that of nascent spines rather than dendritic branches appearing in more or less 10-μm intervals. To further characterize these structures, 5-DIV slices were submitted to immunohistochemistry for PSD-95, a protein found in the postsynaptic density of glutamatergic synapses (Cho et al. 1992). The expression of PSD-95 was also detected in GFP+-derived SVZ cells. Although some cells present a diffuse somatodendritic immunostaining pattern, frequent clusters of PSD-95 were present along dendrites (Fig. 3G,H, arrows). Through confocal microscopy optical sectioning, it is possible to demonstrate that some of these clusters are located within the protrusions that occur along the dendrites (inset in Fig. 3H).
Rare SVZ-Derived Cells Express Glutamate in the OB Granular Layer
Since postnatal SVZ generates glutamatergic neurons when confronted with the embryonic telencephalon environment, we investigated if SVZ-derived cells are also capable of expressing this neurotransmitter when cocultivated with their normal target tissue, the OB, of the same age. This approach tests for the alternative of an SVZ progenitor capable of generating glutamatergic neurons destined to the OB, which could have been overlooked in previous studies. SVZ explants from coronal OB slices were dissected from P9–P11 mice and cocultured with age-matched SVZ explants derived from GFP transgenic mice, touching the internal face of the granule cell layer. After 5 DIV, glutamate immunolabeling is present in the glomerular layer (Fig. 4A), probably reflecting the presence of neuropil and short axon cell bodies (Aungst et al. 2003). GFP+ cells in this region were devoid of glutamate immunolabeling (Fig. 4A).
We found glutamate+ cells in the granular layer (Fig. 4B); however, rare GFP-positive cells were colabeled (Fig. 4C). When that was the case, labeled cells presented a smooth unbranched radial bipolar morphology (Fig. 4D). Although GFP-positive cells invaded the whole extension of the granular layer of the OB slice, glutamate-labeled cells (both from the explants and from the slice) were restricted to a small extension of this structure (data not shown). Since it is not expected to observe glutamate-positive neurons in the OB granular layer (Shepherd et al. 2007), we performed double-labeling immunohistochemistry in OB sections of P11 mice. Rare glutamate-positive cells were found in the granular layer, but these were not labeled with the neuronal specific marker, class III β-tubulin (Fig. 4E, Tuj1 antibody). This indicates a non-neuronal or immature nature of glutamate-expressing cells within the granule layer.
Adult SVZ Generates Glutamatergic Progenitors in Embryonic Telencephalic Slices
To observe if SVZ capacity to generate glutamatergic pyramidal-like neurons postnatally is a transitory characteristic, we cocultured SVZ explants from 3-month-old mice with embryonic telencephalic slices (as shown in Supplementary Fig. 1) for 5–7 DIV. We observed GFP-positive cells within cortical slices, most of which presented a polarized morphology (Fig. 5A–D). These immature pyramidal morphologies were also found coexpressing class III β-tubulin and glutamate within the CP (Fig. 5E–K). Confocal orthogonal sectioning showed these putative pyramidal neurons to be colabeled with glutamate immunolabeling. These data show that the SVZ potential for generating glutamatergic neurons is a long-lasting phenomenon.
Our present results show that postnatal SVZ explants give rise to glutamatergic neurons and multiple cortical neuronal morphotypes when cocultured with embryonic telencephalon slices. These include glutamatergic neurons with morphologies coherent with those of immature pyramidal neurons. In fact, around 28% of the SVZ-derived invading cells were identified as glutamatergic. Some of these glutamate immunoreactive cells displayed immature profiles, with dendrites beginning to extend apical processes. Among more differentiated morphotypes, an average of 16% pyramidal-like neurons was identified. PSD-95 immunohistochemistry revealed labeled clusters in dendritic protrusions supporting the existence of dendritic filopodia with putative dendritic spine characteristics in pyramidal-like neurons. Coherently, the recruitment of PSD-95 to filopodia was previously described as a step toward their stabilization into mature spines (Prange and Murphy 2001). We recognize that the time constraints in our culture model obligate us to characterize very immature phases in pyramidal cell differentiation. However, we have no reason to believe that these refer to another cell type since all cells were bipolar with an apical dendrite directed toward the pia and coherently we never observed multipolar cells (also with their morphologies quite clear) expressing glutamate for example. In addition, we show that the adult SVZ also generates glutamatergic neurons, indicating that the progenitors capable of giving rise to glutamatergic cortical neurons remain in the SVZ during the whole life span of the animal.
Since neuroblasts, RG, astrocytes, and possibly intermediate progenitors are present in postnatal SVZ (Alves et al. 2002; Peretto et al. 2005), all 4 could be candidates for the generation of neurons in the embryonic cortex. We observed that the great majority of the SVZ cells migrating into the slices were neuroblasts, as revealed by class III β-tubulin immunolabeling. Why the cells labeled for GFAP do not migrate into E15 slices remains unknown. It appears that embryonic cortex may be nonpermissive or even repulsive to astrocytes at this age. In a previous attempt of heterochronic transplantation of postnatal SVZ cells to the embryonic brain, Lim et al. (1997) did not find any SVZ cells invading the cortical wall. However, when cells did cross the ventricular surface in other regions and incorporated into the host brain parenchyma, they were class III β-tubulin–positive neuroblasts. Coherently, the embryonic CP is largely devoid of astrocytes, and astrocyte invasion into the cortical parenchyma only starts at later embryonic stages (Choi 1989; Levers et al. 2001).
In the SVZ, neuroblasts migrate in chains using each other as substrates (Lois et al. 1996). In contrast, postnatal SVZ-derived cells that enter the slice migrate individually. Their orientations are compatible with radial and tangential migration modes, typically displayed by glutamatergic and GABAergic cells, respectively (Marin and Rubenstein 2001), adding to previously mentioned features of neurons generated in the embryonic telencephalon. It has been shown that migrating neuroblasts of the SVZ retain proliferative capacity (Menezes et al. 1995); therefore, one possibility is that invading cells reenter the cycle within the VZ before migrating to the CP or, even later, within the CP. Nevertheless, no evidence of GFP+ mitotic figures was found in the embryonic slices.
Recent demonstration that cell fusion can occur spontaneously between blood-borne cells and neurons (Alvarez-Dolado et al. 2003; Weimann et al. 2003) makes it possible that our results could arise from cell fusion between cells of donor explants and host brain slices. Although we have not thoroughly excluded this possibility, we have reasons to believe that cell fusion does not play a significant role on the appearance of glutamatergic pyramidal-like cells derived from our explants. First, we have not observed the occurrence of double-nucleated cells any time during the culture period (Supplementary Fig. 3). Second, our observation that explant-derived cells go through several steps of maturation, such as migratory, immature, and mature profiles as time of culture increases, excludes the possibility that non-neuronal GFP-positive cells like microglia would fuse with more mature cells within the embryonic slice. Third, we have not at any time seen satellite cells closely apposed to GFP-positive pyramidal cells, such as the microglia/pyramidal cell pairs described in Ackman et al. (2006). We could not exclude that cell fusion may occur at the VZ between progenitor cells; however, this is also highly unlikely. First, several groups that used modified neural stem cells for transplantation have excluded the occurrence of cell fusion (i.e., Muotri et al. 2005) between these cells. Second, given that heterokarya may be unstable (Nern et al. 2009), the loss of donor nuclei would give rise to decreased GFP expression that we have not detected. The lack of evidence for cell fusion in our model is in agreement with the low incidence of this phenomenon reported in other brain regions outside the cerebellum (Alvarez-Dolado et al. 2003).
Upon coculturing SVZ explants with postnatal OB, glutamate-positive cells were unexpectedly found in the granular layer derived from both host OB and donor SVZ. Several lines of evidence suggest that these cells are non-neuronal: 1) Endogenous glutamate-positive cells of the OB granular layer are not labeled for Tuj1 antibody, 2) extensive characterization of OB granular neurons has shown that they are all GABAergic (Shepherd et al. 2007), and 3) it has been shown that SVZ astrocytes can be labeled by the glutamate immunohistochemistry (Platel et al. 2007). This further supports the hypothesis that the neuronal glutamatergic phenotype triggered in SVZ progenitors requires signals specific to the embryonic telencephalon. We could not, however, rule out the possibility of existence of cells with a yet undefined morphology.
Data shown here indicate that the postnatal SVZ is capable of generating multiple neuronal types with a broader neurotransmitter repertoire than previously thought. Why would this potential be hidden in vivo? Two hypotheses are postulated: The first one is that progenitors with a broader potential would be capable of generating glutamatergic neurons, besides GABAergic, when exposed to proper signals. The second possibility is that restricted glutamatergic progenitors remain quiescent until recruited. Recent data favor the hypothesis that neurotransmitter bipotent progenitors may be present in the SVZ since dorsal RG, which are known to give rise to cortical pyramidal glutamatergic neurons early on, originate OB GABAergic neurons postnatally (Merkle et al. 2007; Ventura and Goldman 2007). This suggests that signals present in the embryonic environment may reinstruct SVZ progenitors to a differentiation pathway that is normally inactive in postnatal ages. In our experiments, explants from different regions of the SVZ, striatal, and OB SVZ generated glutamatergic neurons. For this reason, we believe that glutamatergic progenitors in our explants are in route migratory neuroblasts. We cannot, however, rule out the possibility that quiescent restricted progenitors are present throughout the rostral migratory stream. In support of the environmental influence hypothesis, it has been shown that neurotransmitter specification can be changed by environmental factors, such as glutamate and GABA signaling in Xenopus laevis (Root et al. 2008). This signaling changes calcium-spiking profiles in progenitors and consequently neurotransmitter specification (Borodinsky et al. 2004). The choice of SVZ neuroblasts to synthesize GABA from putrescine instead of using GAD (Sequerra et al. 2007) favors the idea that these progenitors maintain neurochemical plasticity during their migration. Our present results add up to the latter ones in showing neurotransmitter phenotype plasticity regulated by environmental factors. Whether bipotent or plastic, SVZ progenitors give rise to glutamatergic neurons triggered by environmental factors.
In contrast, Merkle et al. (2007) have suggested by means of heterotopic transplantation of SVZ cells in vivo that specification of neuronal type would be cell autonomous and environment independent. In contrast, our data suggest that the heterochronic transplantation to the embryonic dorsal VZ environment provides SVZ cells with signals that could release a suppressed program or directly stimulate the pathway that leads to the pyramidal neuron phenotype. In light of our data, a possible interpretation of the results in Merkle et al. (2007) is that the positional identity endows progenitors with a restricted spectrum of possible phenotypic destinies. We suggest that heterotopic transplantation within the SVZ would not be sufficient to uncover other phenotypes because putative cues present along the rostral migratory stream must be the same for all neuroblasts since it is a common pathway for all migratory cells destined to the OB.
Macklis and collaborators have shown that endogenous telencephalic progenitors are able to partly replace cortical projection neurons lost by targeted apoptotic cell death (Magavi et al. 2000; Chen et al. 2004). The source of these endogenous progenitors is still unclear. The local parenchyma may contain putative progenitor cells capable of cell replacement (Palmer et al. 1999). In addition, the marginal zone/layer I was recently shown to give rise to both neurons and glia in the embryo, and it is still unclear if it can become neurogenic in response to neural injury in postnatal or mature brains (Costa et al. 2007). Macklis and collaborators suggest that a major candidate for this cell replacement is the underlying postnatal SVZ since they found BrdU-labeled doublecortin-positive cells in the white matter, midways between SVZ, and the cerebral cortex, shortly after the photolytic lesion (Magavi et al. 2000; Chen et al. 2004). In accordance to this idea, Fagel et al. (2009) have shown that, after perinatal chronic hypoxia, Tbr1+ cells are generated in the postnatal cerebral cortex. This phenomenon is correlated with an increase in SVZ proliferation, and both events are FGFr1 dependent. Our work directly supports the hypothesis that the postnatal SVZ is a potential source for replacement of pyramidal neurons in postnatal and adult cerebral cortex.
We analyzed GFP-positive cells for Tbr1, Ctip2, and Satb2 expression since these are considered pyramidal cell markers of different cortical layers (Hevner et al. 2001; Arlotta et al. 2005; Britanova et al. 2005). However, GFP-positive pyramidal neurons do not express any of these markers in our embryonic slice preparation. During development, cortical neurons from different layers are generated at different times (Angevine and Sidman 1961) and are sequentially specified (McConnell 1988). However, after embryogenesis, progenitors may acquire phenotypic determination via an alternative specification pathway.
During the review process of this paper, another article appeared showing that dorsal SVZ progenitors that expressed Tbr1 and Tbr2 gave rise to glutamatergic interneurons of the OB glomerular layer (Brill et al. 2009). We have not identified glutamatergic neurons in the glomerular layer with our coculture study. This could be due to the short survival period of our cultures allied to the low frequency of occurrence of these cells described by Brill et al. (2009). The transitory expression of Tbr1 shown by those authors in glutamatergic neuronal progenitors of the SVZ opens the possibility that the glutamatergic neurons we detected could also have expressed Tbr1 transiently. An important point to be discussed here is whether the glutamatergic phenotype is intrinsic or could be influenced by extracellular signals. Tbr-positive progenitors were found to be restricted to the dorsal portion of the SVZ (Brill et al. 2009); however, these authors do not show if the generation of these progenitors is intrinsic to this region or if it is being induced by signals present in the dorsal adult telencephalon. Therefore, as we are observing the generation of glutamatergic neurons derived by ventral SVZ (striatal), it is possible that the extracellular dorsal environment is inducing the expression of this phenotype. Since the proportion of glutamatergic cells in the model herein presented seems to be higher than the one found by Brill et al. (2009) in vitro studies, we suggest that this phenotype is at least partially controlled by environmental cues. Another possibility is that the embryonic CP is selecting glutamatergic neurons, thereby increasing its proportions in relation to what would normally be expected. Again, our data establish direct evidence that the postnatal SVZ and adult SVZ are sources for glutamatergic neurons capable to differentiate into cortical morphotypes.
Taken together, our results indicate that progenitors capable of generating pyramidal neurons are continuously present in the proximity of the lateral ventricles postnatally and until adulthood. This potential is revealed by stimulation with signals present in embryonic telencephalon. The knowledge of the external factors that direct SVZ cells to a glutamatergic phenotype will be an important next step for achieving successful cell replacement therapies in cerebral cortex and other central nervous system regions.
Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro, Conselho Nacional de Pesquisa e Desenvolvimento, CNPq/PRONEX, FINEP research grant “Rede Instituto Brasileiro de Neurociência (IBN-Net)” 01.06.0842-00.
The authors thank Adiel Batista do Nascimento for animal care, Carla Moreira Furtado, MSc, and Elizabeth Cunha Penna de Moraes, MSc, for technical assistance, and Marcos Romualdo Costa for helpful comments on the text. E.B.S. and L.M.M. were recipients of DSc and Post Doctoral fellowships from CNPq and PROCAD/CAPES respectively. Conflict of Interest: None declared.