Recent studies have revealed that the adult mammalian brain has the capacity to regenerate some neurons after various insults. However, the precise mechanism of insult-induced neurogenesis has not been demonstrated. In the normal brain, GFAP-expressing cells in the subventricular zone (SVZ) of the lateral ventricles include a neurogenic cell population that gives rise to olfactory bulb neurons only. Herein, we report evidence that, after a stroke, these cells are capable of producing new neurons outside the olfactory bulbs. SVZ GFAP-expressing cells labeled by a cell-type-specific viral infection method were found to generate neuroblasts that migrated toward the injured striatum after middle cerebral artery occlusion. These neuroblasts in the striatum formed elongated chain-like cell aggregates similar to those in the normal SVZ, and these chains were observed to be closely associated with thin astrocytic processes and blood vessels. Finally, long-term tracing of the green fluorescent-labeled cells with a Cre-loxP system revealed that the SVZ-derived neuroblasts differentiated into mature neurons in the striatum, in which they expressed neuronal-specific nuclear protein and formed synapses with neighboring striatal cells. These results highlight the role of the SVZ in neuronal regeneration after a stroke and its potential as an important therapeutic target for various neurological disorders.
New neurons are continuously generated within two restricted regions of the adult mammalian brain: the subventricular zone (SVZ) of the lateral ventricle (Alvarez-Buylla and Garcia-Verdugo, 2002) and the subgranular zone (SGZ) of the hippocampal dentate gyrus (Gage, 2000). In the SVZ, GFAP-expressing cells include a neurogenic cell population that gives rise to neuroblasts (Doetsch et al., 1999; Garcia et al., 2004), which migrate anteriorly and form an extensive network of chains (Doetsch and Alvarez-Buylla, 1996; Lois et al., 1996). Most of the neuroblasts migrate through the rostral migratory stream (RMS) into the olfactory bulb (OB), in which they differentiate into interneurons (Altman, 1969; Lois and Alvarez-Buylla, 1994; Kornack and Rakic, 2001; Pencea et al., 2001a). These interneurons in the OB integrate with the existing circuitry and functionally contribute to olfaction (Gheusi et al., 2000; Carleton et al., 2003). The results of in vitro studies suggest that adult human SVZ cells also have neurogenic potential (Pincus et al., 1998; Roy et al., 2000; Sanai et al., 2004). In the rodent brain, cerebral ischemia enhances neurogenesis in the SVZ and SGZ (Liu et al., 1998; Arvidsson et al., 2001; Jin et al., 2001), and ectopic neurogenesis has been observed in the ipsilateral striatum in animal models of middle cerebral artery occlusion (MCAO) (Arvidsson et al., 2002; Zhang et al., 2002) and in degenerated hippocampal CA1 (Nakatomi et al., 2002; Bendel et al., 2005) in animal models with global cerebral ischemia. These newly born neurons may compensate for the loss of neuronal function caused by strokes. A precise understanding of the mechanism of neuronal regeneration should contribute to devising novel strategies to treat patients with cerebral ischemia.
There are two possible sources of the newly generated neurons in the ischemic striatum after MCAO. The first possible source is the SVZ, because neuroblasts born in this region have been proposed to migrate toward the striatum after ischemic injury (Arvidsson et al., 2002; Zhang et al., 2002, 2004; Jin et al., 2003). The second possible source is the striatal parenchyma, because it may also contain latent progenitor cells that can be activated and become neurogenic when stimulated by neurotrophic factors (Palmer et al., 1995; Pencea et al., 2001b). Because more specific cell labeling and tracing techniques were needed to identify the source of this ectopic neurogenesis, in this study, we performed region- and cell-type-specific cell labeling and long-term tracing experiments in combination with light and electron microscopic analyses. The results revealed the SVZ to be the principal source of the neuroblasts that form chain-like structures migrate laterally toward the injured striatal regions and differentiate into mature neurons.
Materials and Methods
Adult 9- to 16- week-old mice (20–32 g) were used in this study. Wild-type ICR mice were purchased from SLC (Shizuoka, Japan). CAG-CAT-enhanced green fluorescent protein (EGFP) transgenic mice (Kawamoto et al., 2000) were provided by Dr. Jun-ichi Miyazaki (Osaka University Medical School, Suita, Osaka, Japan). GFAP tv-a (Gtv-a) transgenic mice (Holland and Varmus, 1998) were purchased from The Jackson Laboratory (Bar Harbor, ME). All animal-related procedures were approved by the Laboratory Animal Care and Use Committee of Keio University and conducted in accordance with the guidelines of the National Institutes of Health.
Focal cerebral ischemia.
During surgery, the mice were anesthetized with a nitrous oxide/oxygen/isoflurane mixture (69/30/1%) administered through an inhalation mask. Rectal temperature was maintained at 37.0°C by placing the animals on a heating bed (model BMT-100; Bio Research Center, Nagoya, Japan). A laser Doppler flowmeter probe (model ALF21; Advance, Tokyo, Japan) was attached to the surface of the ipsilateral cortex to monitor regional cerebral blood flow. MCAO was induced by the intraluminal filament technique reported previously (Shibata et al., 2000; Hayashi et al., 2003; Yamashita et al., 2005). In brief, the right carotid bifurcation was exposed, and the external carotid artery was coagulated distal to the bifurcation. A silicone-coated 8-0 filament was then inserted through the stump of the external cerebral artery and gently advanced (9.0–10.0 mm) to occlude the middle cerebral artery. After 30 min occlusion, the filament was gently withdrawn, and the incision was closed.
Adenovirus preparation and injection into the striatum.
The Cre-encoding recombinant adenovirus AxCANCre (Kanegae et al., 1995) was kindly provided by Dr. Izumu Saito (University of Tokyo, Tokyo, Japan). Concentrated and purified virus stocks (1.7 × 107 viral particles/ml) were prepared by the standard procedure as described previously (Kanegae et al., 1994). The viral suspension (1 μl) was diluted in 1000 μl of PBS before injection, and a 300 nl volume of diluted viral suspension was stereotaxically injected into the striatal parenchyma [anterior, lateral, depth (in mm): 0.5, 2.0, 3.0–4.0] of the CAG-CAT-EGFP transgenic mice, as shown in Figure 2A.
In vivo plasmid transfection of subventricular zone cells.
Plasmid DNA-polyethylenimine (PEI) complexes were prepared according to the instructions of the manufacturer. In short, the pxCANCre plasmid DNA (10 μg) was diluted in a sterile solution of 5% glucose to a final volume of 18.8 μl and complexed with 1.2 μl of linear PEI (in vivo jet PEI; PolyPlus Transfection, Illkirch, France). A 2 μl volume of the PEI–plasmid complexes was stereotaxically injected bilaterally into the lateral ventricle [anterior, lateral, depth (in mm): −0.7, 1.0, 2.0] of the CAG-CAT-EGFP transgenic mice as shown in Figure 3A.
Preparation and transplantation of RCAS-EGFP-virus-producing chicken fibroblasts.
The plasmid for generating the replication-competent avian retrovirus RCAS-EGFP was a gift from Hiroyuki Yaginuma (Fukushima Medical University School of Medicine, Fukushima, Japan). Chicken fibroblast cell line DF-1 (American Type Culture Collection, Manassas, VA) was transfected with the RCAS-EGFP plasmid using the FuGENE6 transfection reagent (Roche, Mannheim, Germany) as described previously (Sato et al., 2002). The virus-producing cells were washed three times with PBS to remove virus particles and pelleted by centrifugation. The pellets were resuspended in 50 μl of PBS and placed on ice. A 100 nl volume of the cell suspension (3–5 × 105 cells) was stereotaxically injected into the ipsilateral striatum [anterior, lateral, depth (in mm): 0.5, 2.7, 3.2–4.0; −0.3, 3.0, 3.2–4.0] or the SVZ (0.5, 1.2, 2.0–2.5; −0.3, 1.5, 2.0–2.5) (see Fig. 4F,I). GFAP-promoter-driven TVA expression in striatal astrocytes was significantly increased after MCAO (data not shown). Therefore, to label target cells at maximal efficiency, DF-1 cells were grafted at 5 d after MCAO (see Fig. 4E). No immunosuppressive agents were given to the animals.
For light microscopy, brains were fixed by perfusion with 4% paraformaldehyde, and, after postfixation overnight, 50 μm sections were cut with a vibrating blade microtome (VT1000S; Leica, Heidelberg, Germany). The following primary antibodies were used: goat anti-doublecortin (Dcx) antibody, 1:100 (Santa Cruz Biotechnology, Santa Cruz, CA); mouse anti-βIII-tubulin antibody (Tuj-1), 1:200 (Sigma, St. Louis, MO); rabbit anti-GFP antibody, 1:200 (MBL, Woburn, MA) or mouse anti-GFP antibody, 1:1000 (gift from Dr. Shohei Mitani, Tokyo Women’s Medical University School of Medicine, Tokyo, Japan); mouse anti-GFAP antibody, 1:200 (Sigma); mouse anti-neuronal-specific nuclear protein (NeuN) antibody, 1:200 (Chemicon, Temecula, CA); rabbit anti-glutathione S-transferase (GST)-π antibody, 1:500 (MBL); rabbit anti-TVA antibody, 1:100 (gift from Dr. Andrew D. Leavitt, University of California, San Francisco, CA); and rat anti-platelet/endothelial cell adhesion molecule-1 (PECAM-1) antibody conjugated with biotin, 1:100 (BD Biosciences PharMingen, San Diego, CA). The antibodies against Dcx, βIII-tubulin, GFP, GFAP, NeuN, GST-π, and TVA were detected with secondary antibodies conjugated with Alexa Fluor (Invitrogen, Carlsbad, CA). For detection of PECAM-1, sections were incubated with ABC Elite complex (Vector Laboratories, Burlingame, CA) and treated with a tyramide signal amplification kit (TSA tetramethyl rhodamine system; PerkinElmer, Boston, MA). To count Dcx-positive neuroblasts in the ischemic striatum, sections were incubated with goat anti-Dcx antibody (1:500) and then with a biotin-labeled secondary donkey anti-goat IgG antibody. After incubation with ABC Elite complex, the signal was visualized with diaminobenzidine tetrahydrochloride. Sections stained without primary antibodies showed no signals (data not shown).
For preembedding immunostaining for electron microscopy, brains were perfused with 4% paraformaldehyde and 0.1% glutaraldehyde, and, after postfixation overnight, 50 μm coronal sections were cut with a vibrating blade microtome. The sections were stained with goat anti-Dcx antibody (1:100) or rabbit anti-GFP antibody (1:100) as described above.
Electron microscopic analysis.
Electron microscopic analyses were performed as described previously (Doetsch et al., 1997). Briefly, the sections labeled by GFP or Dcx immunohistochemistry were postfixed in 2% osmium for 1.5 h, incubated in 2% uranyl acetate for 2 h, dehydrated, and embedded in Araldite (Durcupan ACM; Fluka BioChemika, Ronkonkoma, NY). To study the overall organization of the SVZ and the striatum, we cut serial 1.5 μm semithin sections with a diamond knife and stained them with 1% toluidine blue. Sections containing GFP-positive cells with neuronal morphology were selected under light microscopy and used for preparation of ultrathin sections. Ultrathin (0.05 μm) sections were cut with a diamond knife, stained with lead citrate, and examined with a Jeol (Tokyo, Japan) 1010 electron microscope.
The Dcx-positive cells in the peri-infarct region, located in the medial half of the striatum, were counted under a microscope (Axioplan 2; Zeiss, Tokyo, Japan) in six sections from three levels of the caudate–putamen (1.0, 0.5, and 0 mm rostral to the bregma) of each animal, because these sections consistently contained the infarct area in the striatum ipsilateral to the insult. To count the number of GFAP-positive cells in the GFAP/TVA double-labeled sections (see Fig. 4A,B), three section levels were selected as described above, and four areas in the ipsilateral SVZ of each section were chosen randomly and captured at 100× magnification with a confocal laser microscope (LSM510; Zeiss). For the cell-type-specific marker/GFP (see Fig. 2–6) or βIII-tubulin/Dcx (see Fig. 1E) double-labeling immunohistochemistry, confocal images were taken from four areas in the peri-infarct region of the striatum. LSM510 software was used to measure the distance between GFP/Dcx double-positive neuroblasts and the ventricular wall.
The spatial relationships between the neuroblasts and blood vessels in the striatum were studied by capturing a series of 0.8 μm optical sections with a confocal laser microscope from PECAM-1/Dcx double-immunostained sections and reconstructing them into stacked Z-dimension images with Volocity software (Improvision, Lexington, MA). Based on their morphology, Dcx-positive neuroblast aggregates in the striatum were classified as either “spherical clusters” or “chains” as reported previously (Zhang et al., 2004).
Values are expressed as means ± SEM. The differences in numbers of Dcx-positive cells in the striatum were evaluated for statistical significance by ANOVA with Bonferroni’s correction. The orientation of chain-forming neuroblasts was classified as mediolateral or dorsoventral. The frequencies of chains in each orientation were compared using the χ2 test. Statistical significance was assumed at a p value of 0.05.
Focal cerebral ischemia induces a transient increase in neuroblasts in the adult mouse striatum
In this study, we used a mouse model of middle cerebral artery occlusion (Yamashita et al., 2005), in which the infarcted region is mainly located in the lateral portion of the corpus striatum (Fig. 1A). As the first step in identifying the source of the neuroblasts in the ischemic striatum, we counted the number of cells expressing Dcx, a marker of neuroblasts (Brown et al., 2003; Yang et al., 2004), in the ischemic brains fixed preoperatively and at 14, 18, 21, and 35 d after ischemia induction (Fig. 1B,C). The numbers of Dcx-positive cells 18 and 21 d after ischemia induction were significantly greater than the preoperative number (Fig. 1C). Dcx-positive cells were distributed widely in the striatum at 18 d after MCAO (Fig. 1D), similar to results reported previously (Thored et al., 2006). Ninety-eight percent of the Dcx-positive cells (n = 1108 cells) found in the striatum 18 d after MCAO were positive for the neuronal marker βIII-tubulin (Lee et al., 1990) (Fig. 1E), indicating that they were of the neuronal and not the glial lineage. Based on this finding, we designed a series of experiments in which we performed region- and cell-type-specific labeling and tracing to identify the source of the Dcx-positive neuroblasts in the striatum (see below).
Striatal cells do not generate neuroblasts
Because of the regionally restricted emergence of neuroblasts in the striatum (Fig. 1), we hypothesized that striatal cells generate neuroblasts when subjected to ischemia. To test this hypothesis, we specifically labeled striatal cells and traced their fates with a Cre-loxP recombination system (Zinyk et al., 1998). We stereotaxically injected the Cre-encoding adenoviral-vector AxCANCre (Kanegae et al., 1995; Araki et al., 2000) into the striata of the CAG-CAT-EGFP transgenic mice, in which GFP expression is induced in Cre-introduced cells and their progeny under control of the ubiquitous CAG promoter (Kawamoto et al., 2000). Because adenoviral vectors efficiently infect most, if not all, cell types, regardless of their mitotic activity (Benihoud et al., 1999), if the Dcx-positive neuroblasts were born in the striatum, some should be labeled with GFP. The brain was examined for the presence of GFP-labeled cells after inducing MCAO (Fig. 2A,B). At 5 d after AxCANCre injection, GFP-positive cells were present exclusively in the striatum (Fig. 2C), although none were detected around the SVZ, the other possible source of striatal neuroblasts (see below). Close examination of the sections revealed that virtually all of the cells labeled with Hoechst 33258 expressed GFP in the vicinity of the injection sites (Fig. 2D). These GFP-positive cells expressed the astrocytic marker GFAP (Fig. 2E) (22 cells per 56 GFP-positive cells), the mature neuronal marker NeuN (Fig. 2F) (13 cells per 50 GFP-positive cells), the oligodendrocyte marker GST-π (Hsieh et al., 2004) (Fig. 2G) (8 cells per 55 GFP-positive cells), the endothelial marker PECAM-1 (Sheibani et al., 1997) (Fig. 2H) (6 cells per 55 GFP-positive cells), or NG2, which is reportedly expressed in neuronal progenitor cells in the cortex (Dayer et al., 2005) (Fig. 2I) (8 cells per 44 GFP-positive cells), suggesting that this method labels various types of striatal cells. We then investigated whether the GFP-positive cells in the striatum were labeled with Dcx at 18 d after MCAO. None of the striatal GFP-positive cells observed (n = 695) expressed Dcx (Fig. 2J,K), suggesting the Dcx-positive neuroblasts to have been derived from brain regions outside of the striatum.
SVZ cells generate neuroblasts that migrate toward the ischemic striatum
The results of previous studies have suggested the SVZ as another possible source of striatal neuroblasts (Arvidsson et al., 2002; Zhang et al., 2002, 2004; Jin et al., 2003). A specific and efficient method of labeling cells in the SVZ, a very thin cell layer lining the ventricular wall that is difficult to target, is needed to investigate whether these cells generate striatal neuroblasts. In this study, we chose intraventricular injection of plasmid DNA mixed with PEI, because it has been demonstrated to introduce foreign genes specifically into the SVZ (Lemkine et al., 2002). To verify the specificity and efficiency of this method, the Cre-encoding plasmid pxCANCre mixed with PEI was injected into the lateral ventricles of CAG-CAT-EGFP transgenic mice (Fig. 3A), and GFP-positive cells were observed in the SVZ alone at 5 d after the injection (Fig. 3C). We thus concluded that the Cre expression was specifically targeted to the regions close to the ventricular wall. Cell counts revealed that 50% of the GFP-expressing cells (n = 42) were GFAP-positive astrocytes and 6.3% (n = 111) were Dcx-positive neuroblasts. Analysis of the fate of the GFP-labeled SVZ cells in the non-ischemic control mice (n = 4) and MCAO mice (n = 6) on day 18 (Fig. 3B) revealed large numbers of GFP-positive cells in the OB and RMS of all brains in both the control group and the MCAO group (data not shown), but examination of the striatum revealed a significant difference in the distribution of GFP-positive cells between these two groups. GFP-positive cells were restricted to the SVZ in the control group, with none being observed in the striatum (Fig. 3D). In contrast, in the MCAO group, GFP-positive cells (82.0 ± 12.1 cells per hemisphere) were found in the ipsilateral striatum (Fig. 3E), and 37% (n = 45 cells) were positive for Dcx and had a long leading process (Wichterle et al., 1997), similar to that of the neuroblasts migrating in the RMS (Fig. 3F). The distribution pattern of these SVZ-derived GFP/DCX double-immunopositive cells (Fig. 3G) was very similar to that of DCX-positive cells (Fig. 1D). These results suggest that SVZ-derived neuroblasts migrate toward the striatal parenchyma after a stroke.
SVZ GFAP-expressing cells generate striatal neuroblasts after cerebral ischemia
Previous studies have shown that SVZ GFAP-expressing cells include a neurogenic cell population, which gives rise to neuroblasts that migrate anteriorly in the RMS and differentiate into neurons in the OB (Doetsch et al., 1999; Garcia et al., 2004; Imura et al., 2006). To determine whether SVZ GFAP-expressing cells are the source of neuroblasts in the damaged striatum, we administered the RCAS-EGFP avian retrovirus (Sato et al., 2002) into the brains of Gtv-a mice, in which only GFAP-positive cells express TVA, a receptor required for infection by the avian retrovirus (Holland and Varmus, 1998; Doetsch et al., 1999; Seri et al., 2001). To confirm that TVA is specifically expressed in GFAP-positive cells in the ischemic brain, we performed double immunohistochemistry for GFAP and TVA. Confocal microscopic analyses of the SVZ tissue sections from Gtv-a transgenic mice, 8 d after ischemia induction, revealed that virtually all of the TVA-expressing cells were positive for GFAP (Fig. 4A,B), although the different subcellular localizations of each of the proteins did not allow quantitative analysis. In addition, all of the TVA-expressing cells (n = 327) were negative for Dcx (Fig. 4C,D). Injection of the RCAS-EGFP virus into wild-type ischemic brains yielded no labeled cells (data not shown), indicating that infection with this virus is dependent on TVA. Thus, only GFAP-positive cells should be infected with RCAS-EGFP retrovirus even in the ischemic brain.
In this study, we transplanted the chicken fibroblast cell-line DF-1 cells producing the RCAS-EGFP virus, because it was shown to yield high virus titers at injection sites for gene transfer (Holland and Varmus, 1998). We transplanted DF-1 producing the RCAS-EGFP virus into the SVZ or the striatum of the Gtv-a transgenic mice 5 d after MCAO and characterized the fate of GFP-positive cells 18 d after MCAO (Fig. 4E). Most of the GFP-positive cells in the brains, of which the striatum had been injected with RCAS-EGFP-producing cells (Fig. 4F), were still positive for GFAP (Fig. 4G), and no GFP/Dcx double-immunopositive cells were observed (none of 498 GFP-positive cells) (Fig. 4H), consistent with the results of the adenoviral infection experiment (Fig. 2). Conversely, when cells producing the RCAS-EGFP virus were transplanted into the SVZ (Fig. 4I), all of the brains studied (n = 3) contained GFP/Dcx double-positive cells (5 cells per 13 GFP-positive cells in total) in the peri-infarct region of the striatum at 18 d (Fig. 4J). We therefore concluded that SVZ GFAP-expressing cells produce the neuroblasts observed in the ischemic striatum.
Migrating neuroblasts form chain-like structures along blood vessels in the striatum
The results described above (Figs. 2⇑–4) indicate that most of the Dcx-positive neuroblasts migrating in the ischemic striatum originated in the SVZ. We then examined the Dcx-positive cells migrating in the striatum by immunofluorescence and electron microscopic techniques to study the morphology of the SVZ-derived neuroblasts in greater detail. Recent studies have found blood vessels to be involved in the regulation of neuronal progenitors in vitro or in vivo (Palmer et al., 2000; Louissaint et al., 2002; Shen et al., 2004), and insult-induced neurogenesis in the striatum is also reportedly influenced by blood vessels (Sun et al., 2003; Chen et al., 2005). Possible interactions between the migrating neuroblasts and blood vessels in the striatum were examined 18 d after MCAO in sections of the brain that had been stained for Dcx and PECAM-1, a marker of vascular endothelial cells (Sheibani et al., 1997). Some of the neuroblasts formed aggregates and others remained isolated in the striatum. The neuroblast aggregates were classified as spherical clusters or chains based on their morphology, as described previously (Zhang et al., 2004). All of the spherical clusters (n = 6) were identified at some distance from PECAM-1-positive vascular endothelial cells (Fig. 5A), whereas all of the chains (n = 12) were wound around vascular endothelial cells (Fig. 5B,C) (supplemental movie 1, available at www.jneurosci.org as supplemental material). Most of the neuroblasts in the chain were elongated parallel to the blood vessels, and, interestingly, most of the chains (83%, n = 10) were oriented in the mediolateral direction (Fig. 5D–F). These findings suggest blood vessels to be involved in the chain formation and lateral migration of SVZ-derived neuroblasts toward the injured tissue. To further investigate the fine morphology of the chain structure, we examined anti-Dcx-stained sections electron microscopically. The morphology of the neuroblasts in the striatum resembled that in the intact SVZ (Doetsch et al., 1997). Their nuclei were spindle-shaped and dark, and the cytoplasm was smooth and scant (Fig. 5H). Zonula-adherens-like small junctional complexes were observed between these neuroblasts (Fig. 5I,K). The chains of neuroblasts were frequently associated with thin astrocytic processes aligned close to blood vessels (Fig. 5I,J). Thus, the chain-like structures that developed in the striatum exhibited cellular organization and morphology similar to those of the normal SVZ.
SVZ-derived neuroblasts differentiated into mature neurons and formed synapses in the striatum
Previous studies have shown that a small population of striatal neuroblasts in the ischemic brain can survive and mature into neurons (Arvidsson et al., 2002; Teramoto et al., 2003). We tested whether SVZ-derived neuroblasts were capable of differentiating into mature neurons after prolonged survival. For this purpose, we analyzed GFP-positive cells labeled by injection of the PEI–pxCANCre plasmid complex 90 d after MCAO (Fig. 6A,B). The results showed that a subpopulation of the GFP-positive cells (29%; n = 5 cells) in the striatum expressed NeuN, a marker of mature neurons (Fig. 6C), and the GFP-labeled cells frequently showed a morphology typical of mature neurons (Fig. 6D). Moreover, immunoelectron microscopic examination revealed that the GFP-positive axons contained abundant presynaptic vesicles and formed synapses with neighboring cells (Fig. 6E,F). These observations indicated that SVZ-derived neuroblasts are capable of differentiating into mature neurons and being integrated into the preexisting neuronal circuit.
SVZ GFAP-expressing cells produce neuroblasts in the peri-infarcted striatum
Migration of SVZ neuroblasts toward the injured striatum in MCAO models similar to ours has been suggested previously by the results of experiments using bromodeoxyuridine (BrdU) labeling (Arvidsson et al., 2002; Parent et al., 2002; Sundholm-Peters et al., 2005; Thored et al., 2006) or DiI labeling (Jin et al., 2003). Intraperitoneally administered BrdU can be incorporated into dividing cells and dying cells throughout the body (Kuan et al., 2004), such that it is not useful for determining where the cells were labeled. Because DiI is lipid soluble and transported to neighboring cells, when DiI is injected into the lateral ventricles, it labels various types of cells directly adjacent to or projecting into the ventricular wall independently of cell migration (Frielingsdorf et al., 2004). Both BrdU and DiI are diluted by cell divisions, which makes it difficult to study the phenotypes of the matured progeny of labeled cells after a long period. Moreover, the SVZ is a very thin cell layer lining the ventricular wall and is difficult to specifically target by stereotaxic viral injections. To overcome these technical limitations, we injected Cre-encoding plasmid into the lateral ventricles of transgenic mice carrying a floxed GFP gene, which made it possible to specifically label SVZ cells and trace their progeny until 90 d after MCAO. The results of our experiments provide direct evidence that the SVZ is the principal source of the neuroblasts that migrate into the ischemic striatum and differentiate into mature neurons. We also performed cell-type-specific labeling by the RCAS-TVA system (Bell and Brickell, 1997), which has been used to study the fate of GFAP-expressing cells in vivo (Holland and Varmus, 1998; Doetsch et al., 1999; Seri et al., 2001). The results of previous studies indicate that SVZ GFAP-expressing cells generate only olfactory bulb neurons in the normal brain (Doetsch et al., 1999), whereas the results of our study strongly suggest that SVZ GFAP-expressing cells give rise to neuroblasts that migrate into the injured striatum. However, because we could not directly show that the cells originally infected with the RCAS-EGFP virus were actually expressing GFAP, we cannot rule out the possibility that other cell types in the SVZ generated the EGFP-positive neuroblasts. Interestingly, SVZ cells have been implicated in the generation of glial cells in the mechanically injured cerebral cortex (Goings et al., 2004) and the corpus callosum damaged by experimental autoimmune encephalomyelitis (Picard-Riera et al., 2002). Thus, it is possible that SVZ GFAP-expressing cells generate various types of progenitors, which have the ability to migrate to diverse damaged brain regions and replace the various types of cells lost by brain injury or brain disease.
Striatal cells do not generate neuroblasts in the ischemic brain
The results of this study suggest neuroblasts appearing in the peri-infarcted striatum to be derived from brain regions outside of the striatum (Fig. 2), and this finding is consistent with the idea that the distribution of the neurogenic cell population is restricted to a niche in which a special microenvironment supports their survival and function in the adult brain (Herrera et al., 1999; Shihabuddin et al., 2000). However, we were unable to rule out the possibility that the striatum contains a population of rare neurogenic cells undetectable under our experimental conditions. Previous in vitro studies have revealed that cells isolated from the striatum can generate new neurons when stimulated by intrinsic or extrinsic factors (Palmer et al., 1995; Wang et al., 2004), suggesting that latent neural progenitor cells reside in the striatal parenchyma. These latent neural progenitor cells may be activated by manipulating the signaling pathways that determine their cell fate. Noggin, an antagonist for bone morphogenetic protein, has been reported to promote neuronal differentiation of stem cells (Lim et al., 2000). Moreover, forced expression of Pax6 (paired box gene 6) (Heins et al., 2002) and a dominant-negative Olig2 (oligodendrocyte lineage transcription factor 2) mutant (Buffo et al., 2005) induce neurogenesis in the cortex. These factors may be useful for activating the neurogenic potential of progenitor cells in the striatal parenchyma.
Neuroblasts emigrate from the SVZ into the injured region
The present results show that neuroblasts that migrate into the ischemic striatum form elongated aggregates that are morphologically similar to those in the SVZ, as reported previously in rats (Zhang et al., 2004). The chain-like structures were observed to be in contact with thin processes of astrocytes in the striatum, similar to those in the RMS (Lois et al., 1996) and SVZ (Doetsch et al., 1997). Astrocyte-derived migration-inducing activity has been reported to stimulate chain migration by neuroblasts in vitro (Mason et al., 2001) and may be involved in the lateral chain migration in the ischemic striatum as well.
Our SVZ-specific labeling experiments using a Cre-loxP system provided direct evidence that a subpopulation of SVZ neuroblasts, which migrate anteriorly in the normal brain, are laterally redirected to the injured striatum. There are two possible mechanisms to explain the ischemia-induced change in the direction of neuroblast migration. First, ischemia may disrupt mechanisms that restrict neuroblasts within the SVZ. Second, ischemia may promote lateral migration by neuroblasts through altered expression of guidance molecules, such as Slit proteins, which determine the direction of neuroblast migration in the SVZ as repellents (Wu et al., 1999; Sawamoto et al., 2006). Chemotactic cytokines, such as stromal cell-derived factor 1a, expressed in the injured striatum may also be involved in this process (Robin et al., 2006; Thored et al., 2006). Our results demonstrate that SVZ-derived neuroblasts directionally migrate in chain-like structures parallel to blood vessels toward the infarcted region. Blood vessels in the damaged striatum may play a crucial role in the migration and/or survival of these neuroblasts, perhaps by releasing diffusible factors, such as BDNF (Louissaint et al., 2002).
SVZ cells generate mature neurons in the striatum
Previous reports have described BrdU-labeled newborn cells as expressing markers for mature neurons within the damaged striatum as early as 30 d after the induction of ischemia (Arvidsson et al., 2002; Parent et al., 2002). In the present study, we examined the phenotype of SVZ-derived GFP-labeled cells by light and electron microscopy after an extended survival period (90 d). The labeled cells were found to possess long processes, express NeuN, and form synaptic structures in the damaged striatum 90 d after ischemia induction. These results strongly suggest SVZ cells to have the ability to generate functional mature neurons that survive in the damaged striatum for considerable periods.
Based on all of the results of this study, we conclude that SVZ is the main source of the neuroblasts that migrate toward the brain region infarcted by cerebral ischemia, in which they differentiate into mature neurons. Our findings indicate that, as an important endogenous cell source, SVZ is a promising therapeutic target for various neurological disorders. However, it remains unclear what types of neurons are generated, although some of the SVZ-derived neurons showed a dopaminergic phenotype (our unpublished observation). In addition, the number of newborn neurons is too small for recovery of neurological functions (Arvidsson et al., 2002). Thus, it will be necessary to add appropriate interventions to enhance the proliferation, survival, and/or neuronal maturation of SVZ cells and their progeny. The mechanisms of insult-induced neurogenesis described herein are anticipated to be of fundamental importance to studying molecular mechanisms that control SVZ cells and their progeny and for developing novel neuronal self-repair strategies.
This work was supported by grants from Bridgestone Corporation; The Ministry of Education, Culture, Sports, Science, and Technology; The Ministry of Health, Labour, and Welfare; Mitsui Life Social Welfare Foundation; and Japan Science and Technology Agency (Core Research for Evolutional Science and Technology). We are grateful to Arturo Alvarez-Buylla, Magdalena Götz, and Tatsuhiro Hisatsune for valuable discussions; Tetsu Yoshida, Masako Katsumaru, and Mario Soriano Navarro for expert technical assistance; Noboru Sato and Hiroyuki Yaginuma for the RCAS-EGFP retrovirus vector; Izumu Saito for the AxCANCre adenovirus; Andrew D. Leavitt for the anti-TVA antibody; Jun-ichi Miyazaki for the CAG-CAT-EGFP mice; Shohei Mitani for the anti-GFP antibody; and members of our laboratories for useful advice and encouragement.
↵*T.Y. and M.N. contributed equally to this work.
- Correspondence should be addressed to Kazunobu Sawamoto, Bridgestone Laboratory of Developmental and Regenerative Neurobiology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku, Tokyo 160-8582, Japan. Email: