New Striatal Neurons in a Mouse Model of Progressive Striatal Degeneration Are Generated in both the Subventricular Zone and the Striatal Parenchyma

Acute striatal lesions increase proliferation in the subventricular zone (SVZ) and induce migration of SVZ neuroblasts to the striatum. However, the potential of these cells to replace acutely degenerated neurons is controversial. The possible contribution of parenchymal progenitors to striatal lesion-induced neurogenesis has been poorly explored. Here, we present a detailed investigation of neurogenesis in the striatum of a mouse model showing slow progressive neurodegeneration of striatal neurons, the Creb1Camkcre4Crem−/− mutant mice (CBCM). By using BrdU time course analyses, intraventricular injections of a cell tracker and 3D reconstructions we showed that neurodegeneration in CBCM mice stimulates the migration of SVZ neuroblasts to the striatum without altering SVZ proliferation. SVZ-neuroblasts migrate as chains through the callosal striatal border and then enter within the striatal parenchyma as individual cells. In addition, a population of clustered neuroblasts showing high turnover rates were observed in the mutant striatum that had not migrated from the SVZ. Clustered neuroblasts might originate within the striatum itself because they are specifically associated with parenchymal proliferating cells showing features of intermediate neuronal progenitors such as clustering, expression of EGF receptor and multiple glial (SOX2, SOX9, BLBP) and neuronal (Dlx, Sp8, and to some extent DCX) markers. Newborn striatal neurons had a short lifespan and did not replace projection neurons nor expressed sets of transcription factors involved in their specification. The differentiation failure of endogenous neuroblasts likely occurred cell autonomously because transplanted wild type embryonic precursors correctly differentiated into striatal projection neurons. Thus, we propose that under progressive degeneration, neither SVZ derived nor intra-striatal generated neurons have the potential to differentiate into striatal projection neurons.


Introduction
In rodents, the subventricular zone (SVZ)-olfactory bulb (OB) system and the dentate gyrus (DG) of the hippocampus are the major sites of adult constitutive neurogenesis under physiological conditions [1]. Different brain lesion paradigms can further induce neurogenic events in normally non-neurogenic sites, such as the CA1 of the hippocampus, the striatum or the neocortex [2][3][4][5][6][7]. It is known that after a lesion, neural progenitors in both the SVZ and DG increase their proliferation, and that SVZ neuroblasts migrate toward injured areas [7]. The possible contribution of parenchymal progenitors to lesion-induced neurogenesis has been poorly explored. However, recent data indicate that mild ischemia can induce genesis of cortical interneurons from progenitors located in the rat neocortical layer 1 [8]. Interestingly enough, parenchymal neuronal progenitors normally occur in the adult rabbit striatum [9]. Contrary to what is generally thought these studies maintain that the mature brain parenchyma can be permissive for the generation of neurons.
Newborn cells in injured regions generally show low survival rate, and although replacement of a few lost neuronal cells has been reported, the fate/functional meaning of lesion-induced neuroblasts remains unclear [2,3,5,6]. In the striatum, where reactive neurogenesis has been extensively analyzed [2,3,6], the potential of newborn cells to replace medium spiny neurons, the main striatal cell type, is controversial [2,9]. Indeed, lesion induced striatal neuroblasts are generated in the SVZ [9,11] and express Sp8, a transcription factor associated to SVZ neuroblasts committed to becoming OB and striatal calretinin interneurons [9,12]. This support the fact that acute injuries of the striatum do not alter the intrinsic differentiation potential of SVZ-derived neuroblasts [9].
It is to be noted that most data regarding lesion-induced neurogenesis in the striatum have been obtained in acute models of neurodegeneration, in which neuronal death mostly occurs synchronously [2,3,6,[10][11][12]. By contrast, only a few data are available in models of slow progressive neurodegeneration [13]. In such a state, factors modulating newborn neuron production, migration and survival, such as inflammation and astrogliosis, are expected to differ in respect to acute injuries [7,14,15].
Here, we present the first detailed investigation of induced neurogenesis in the striatum of a mouse model of slow progressive neurodegeneration, the Creb1 Camkcre4 Crem 2/2 double mutant mice, characterized by genetic ablation of the cAMP-dependent transcription factors CREB and CREM (CBCM; [16]]. In these animals, the loss of Creb1 in a Crem null background, is conditionally restricted to CamKIIa expressing mature neurons. This leads to a slow progressive degeneration of the DG, CA1 and dorso-lateral striatum [16]. Interestingly, impairment of CREB mediated transcription has been proposed as a possible cause of cell death in Huntington disease (HD). [17,18]. In these mice we investigated both the origin and fate of newborn striatal neurons.
Our data indicate that similarly to acute lesions, neurogenesis occurs within the striatum of CBCM mice, although in absence of proliferation induction in the SVZ. Notably, in parallel to migration of SVZ-neuroblasts to the mutant striatum, we found a population of clustered neuroblasts generated from intrastriatal proliferating cells showing features of intermediate neuronal progenitors such as clustering and co-expression of peculiar glial and neuronal markers [1]. However, as in acute models of neurodegeneration, newborn neuronal cells in CBCM striatum showed transient existence and did not express any specific markers of either mature or immature striatal projection neurons.

Animals, 5-bromo-2-deoxyuridine injections, and tissue preparation
All experimental procedures were in accordance with the European Communities Council Directive of 24 November 1986 (86 609 EEC), the Italian law for the care and use of experimental animals (DL116 92) and approved by the Italian Ministry of Health and the Bioethical Committee of the University of Turin. All experiments were designed to minimize the numbers of animals used and their discomfort.
Animal experiments were performed on male and female mice with targeted deletion of the Creb1 gene achieved by the Cre-loxP system. Creb1 loxP/loxP mice in which the exon carrying CREB1 DNA binding domain is flanked by loxP sites were crossed with the transgenic line carrying the Cre recombinase expressed under the control of the CamkIIa promoter and with mutants containing a germline deletion of the Crem gene [16]. Since Crem 2/2 male mice are sterile, we used the following mating schemes to obtain experimental mutants: (i) Camkcre4 Creb1 loxP/loxP Crem +/2 males 6Creb1 loxP/loxP Crem 2/2 females (25% probability of obtaining a double mutant; CBCM).
Animals not expressing the Cre were used as controls. Twenty eight mice aged six to nine months, 14 CBCM mutants and 14 controls, were intraperitoneally injected with 5-bromo-2-deoxyuridine (BrdU; Sigma, Steinheim, Germany; 40 mg/kg body weight in 0.1 M Tris) and than killed 2 h after a single BrdU injection (n = 4), or 5 (n = 3), 15 (n = 4) and 28 days (n = 3) after the first of 5 days of one daily BrdU injection. Animals were deeply anesthetized with a ketamine/xylazine solution (100 and 33 mg/ kg body weight, respectively) and transcardially perfused with icecold saline solution (0.9% NaCl), followed by a freshly prepared solution of 4% paraformaldehyde (PFA) plus 2% picric acid in 0.1 M sodium phosphate buffer, pH7.4 Brains were then postfixed overnight, cryoprotected, frozen at 280uC, and sectioned in series on a cryostat. Most brains were cut into 40 mm thick sections, with the exception of six hemispheres that were used for the phenotypic analysis of BrdU+ cells at 15 d and 28 d, which were cut into 25 mm thick sections, and the Cell tracker green injected brains that were cut into 60 mm thick sections.

Stereotaxic Surgery and Cell Tracker Green (CTG) Injections
Seven month-old adult CBCM mice were anesthetized by intraperitoneal injection of ketamine (100 mg/kg; Ketavet; Bayer, Leverkusen, Germany) supplemented by xylazine (5 mg/kg; Rompun; Bayer) and positioned in a stereotaxic apparatus (Stoelting, Wood Dale, IL). The skull was exposed by a skin incision and small holes drilled through the skull and 1 ml of Cell Tracker Green CMFDA (10 mM in dimethylsulfoxide; Molecular Probes, Eugene, OR, Cat #2925) was bilaterally injected at stereotaxic coordinates of 0.3 mm AP, +/2 1 mm ML, and 21.5 mm DV, using a glass micropipette and a pneumatic pressure injection apparatus (Picospritzer II, General Valve Corp, Fairfield, IL). Infusions were performed in 10-15 minutes and the micropipette was left in place for a further 10 minutes. After the micropipette was removed, the skin was sutured with 8.0 mm silk thread. Following surgery, the animals were sacrificed at 24 hr (n = 3) or seven days (n = 3) following surgery.).

Preparation of cell suspension and transplantation
For embryonic precursor transplantation, donor cells were dissected from embryo [embryonic day 15 (E15) EGFP mice. The preparation of donor Lateral Ganglionic Eeminence (LGE) progenitor cells was performed as described previously [19,20]. Briefly, the LGE of donor embryos was isolated and mechanically dissociated to a single-cell suspension in ice-cold Leibovitz L15 medium. The obtained suspension was centrifuged and resuspended in the same solution. The donor cell suspensions were resuspended at a final concentration of 5610 4 cells/ml.
Cell transplantation was performed on CBCM (n = 3) and control mice (n = 3), following the same surgical procedure described previously for neuronal tracer injection. In all experiments, 1.5 ml of the cell suspension were bilaterally injected at +1. mm AP, +/2 2. mm ML, and 2.5 mm DV to reach the dorso-lateral left striatum. Infusions were performed in 10-15 minutes and the micropipette was left in place for a further 10 minutes. Animals were left to survive for 21 days after surgery.

Light and confocal 3D reconstructions
For 3D reconstructions, one hemisphere of a CBCM mouse and one of a control mice were serially coronally sectioned (40 mm thick) and immunolabeled with anti-DCX. For the reconstruction of the SVZ, dPSB and striatal chains along the entire brain, 140 and 120 consecutive sections, from control and CBCM mice respectively, encompassing the entire SVZ and lateral ventricle were digitally captured (voxel size 0.760.7640 mm) and reciprocally aligned using Reconstruct software version 1.1 (Copyright 2004, John C. Fiala, Boston University, Boston, MA; http://tech. groups.yahoo.com/group/reconstruct_users). The dPSB chains, CC DCX+ cells and striatal DCX+ clusters were 3D rendered as traces slabs, in which traces were filled with colour and extruded along the z-axis, while the SVZ was rendered as Boissonnant surface, in which subsequent traces were interpolated. For higher resolution 3D reconstructions subsequent sections were acquired by the confocal microscope at 206(N.A. 0.7; size 0.760.762 mm), or 1006 (N.A. 1.3; 0.1360.1361 mm). Multiple stacks from the same section were mounted with Vias (CNIC, University of New South Wales, Australia) and the sequences of optical planes from subsequent sections were imported in Reconstruct for alignment. Tracings of 206 reconstructions were obtained in reconstruct using the wildfire region growing tool, and were represented as traces slabs and rendered in Blender (Stichting Blender Foundation, Amsterdam, Netherlands; www.blender.org). Tracings of the blue and cyan groups shown in supplementary Video S1 were obtained through an automatic algorithm in Neuronstudio ( [21]; CNIC, University of New South Wales, Australia) and were rendered in V3D neuron ( [22]; http://penglab.janelia.org/proj/v3d). Single DCX+ cells were traced and analyzed with a semi-automatic procedure in Neuromantic 1.6 (Darren Myatt, University of Reading, UK).

Quantifications and statistical analyses
Volumes were evaluated with Neurolucida 7.0 in a 1-in-6 series of sections (40 mm thick) labeled with DAPI from CBCM (n = 3) and control mice (n = 3). Total number of cells was evaluated stereologically with Neurolucida in a 1-in-6 series of sections (40 mm thick) in both control (n = 3) and CBCM (n = 4) mice. In the present study the optical fractionator principle was modified, in that, only the labelled cells in the uppermost focal plane were excluded to avoid oversampling, as described by others [23,24].
The percentage of Ki67+/DCX+ cells (K/D+) and Ki67+ cells associated with DCX+ cells (K/aD cells) labeled for SOX2, SOX9, BLBP, EGFr, pan-Dll and BrdU was evaluated in at least 80 cells per animal for each cell type.
Analysis of the maturation of newly generated neuronal precursors in the striatum of CBCM mice. This analysis was performed on sections labelled simultaneously for BrdU, DCX, NeuN and CaMKII-Cre in CBCM mice at 15days (n = 3) and 28 days (n = 3) from BrdU administration. Total number of BrdU+ cells in the striatum was counted stereologically in Neurolucida. The proportion of cells colabeled for BrdU and combinations of DCX, NeuN and CaMKII-Cre compared to the total BrdU+ population was calculated by sampling 100 BrdU positive cells per animal. This proportion was then multiplied for the total number of BrdU+ cells to obtain estimates of the total number of each specific subpopulation of BrdU+ cells.
Statistical analyses were performed by using Statistical Package for the Social Science 14.0 (SPSS, Chicago, IL). Analysis of variance (ANOVA) was followed by Tukey's post hoc test when appropriate.

The dorsal SVZ expands laterally in CBCM mice
In CBCM mice the loss of Creb1 under the control of CamKIIa promoter, in absence of Crem expression, leads to progressive neurodegeneration of the dorsolateral striatum [16]. According to previous studies [16], the volume of the caudate/putamen of six month old CBCM mutants was reduced about 70% compared to control mice (Volume of the Striatum: Controls = 6.561.1 mm 3 ; Mutants = 2.260.3 mm 3 ; F 2,1 = 43.83 p = 0.003; n = 3; Fig. 1A,B) while the volume of the lateral ventricle was about 250% expanded (Volume of the lateral ventricle: Controls = 0.9960.30 mm 3 ; Mutants = 2.4760.62 mm 3 ; F 2,1 = 13.69 p = 0.021; n = 3). In acute models of striatal neurodegeneration the SVZ increases its proliferative activity and acts as a source of new neurons for the striatum [2,6,11,25]. In CBCM mice, despite a severe striatal neurodegeneration the SVZ volume and the number of proliferating cells labeled by a two hour BrdU pulse in this region were comparable to those of controls (Volume of the SVZ: CBCM = 0.060+/20.004 mm 3 ; Controls = 0.044+/20.011 mm 3 , F 2,1 = 5.8, p = 0.07; BrdU+ cells: CBCM 5344+/2894; controls 5656+/22092 cells; p = 0.677; n = 3 animal/group). Nevertheless, doublecortin (DCX) immunostaining in mutants revealed an expansion of the dorsal portion of the SVZ (dSVZ, Fig. 1A,B, Fig. 2B; Volume of the dSVZ: CBCM = 0.0049+/20.0009 mm 3 ; Controls = 0.0251 +/20.0084 mm 3 , F 2,1 = 17.3, p = 0.014). A net increase in DCX+ cells was also found along the boundary between the corpus callosum and the striatum, here referred to as the dorsal pallial-subpallial boundary (dPSB; arrow in Fig. 1A,B; dark green in Fig. 2A,B, Fig. S1). As shown by a 3D reconstruction study, neuroblasts in the dPSB formed laterally oriented chains organized in a network directly linked to the dSVZ (Fig. 3B). In addition, many individual DCX+ cells were observed along the striatal side of the dPSB (Fig. 3A, arrows), suggesting the occurrence of SVZ neuroblast migration through the dPSB to the striatum. Thus, in CBCM mice the lateral extension of the SVZ along the dPSB might represent a specific migratory pathway leading neuroblasts towards the dorso-lateral striatum.
In order to investigate whether striatal DCX+ cells were newly generated, BrdU was given for 5 consecutive days (1 injection/day; 50 mg/Kg) and the animals were left to survive for 5 ( Fig. 1D,F). These data indicate that newborn striatal DCX+ cells are initially organized mostly in clusters.

Morphology and spatial distribution of striatal DCX+ neuroblasts
Clustered DCX+ cells. DCX+ clusters in mutant striatum showed a stereotyped distribution, typically occupying a dorsal position in the anterior striatum, a central and ventro-lateral position in the middle striatum and a medial position in the caudal striatum ( Fig. 2A; Fig. S1). Clusters were restricted to the grey matter (Fig. 1C) and generally showed irregular profiles, lacking a preferential orientation (Fig. 2C). A 3D reconstruction showed that unlike SVZ and dPSB chains, striatal DCX+ clusters were scattered and not organized as a network ( Fig. 2A,C; Fig. S1). None of the clusters directly come in contact with the SVZ or the dPSB.
Individual DCX+ cells. The spatial distribution of striatal individual DCX+ cells overlapped that of clustered cells. In addition, individual DCX+ cells were also found in the lateral subcallosal region (Fig. 3C,D). In contrast to the clustered cells, individual cells were found both in the grey matter and within the fiber tracts of the internal capsule where they were often organized in fascicles (Fig. 1C dotted line).
To better understand the spatial organization of the individual DCX+ cells and their relationships with the SVZ and the DCX+ clusters, we performed a 3D reconstruction of the middle striatum, spanning 640 mm along the rostro-caudal axis (Fig. 3). In the reconstructed volume we identified nine spatially segregated groups of individual DCX+ cells (Fig. 3C,D blue and violet tones). None of these groups directly contacted the SVZ, while three of them did, indirectly through the dPSB (dPSB connected group, blue tones Fig. 3D). The dPSB connected groups were restricted to the dorsal caudate-putamen and extended caudoventrally for several hundred microns with a slope similar to that of the bundles of the internal capsule ( Fig. 3D; Fig. S2). Some of the DCX+ cells in the dPSB connected groups were bipolar (blue in Fig. 4A, F), but most of them showed a more complex morphology with long neurites often oriented parallel to the internal capsule fibers (cyan in Fig. 4A, B, F). Interestingly, the dPSB connected groups were only rarely associated with the clustered DCX+ cells (red in Fig. 3D; Supplementary Video S1).
The dPSB unconnected groups were located in the medial and ventral caudate-putamen (Fig. 3C-D). Cells in these groups were only rarely oriented along the internal capsule fiber bundles ( Fig. 4C-E, G; data not shown). In contrast to the dPSB connected groups, those unconnected were closely associated with the striatal DCX+ clusters (Fig. 3D).
Overall, these data indicate that a subset of individual DCX+ cells in CBCM striatum is indirectly linked with the SVZ through the dPSB. By contrast, clustered DCX+ cells and individual cells surrounding them are mostly separated from the SVZ, suggesting they are independent from this germinative region.

SVZ-derived neuroblasts enter the striatum of CBCM mice as individual cells and do not contribute to the population of clustered DCX+ cells
In order to verify whether SVZ neuroblasts migrate into the CBCM striatum, we performed intraventricular injections of Cell Tracker Green (CTG). This cell tracker had previously been shown to efficiently label migrating SVZ neuroblasts over several days [26,27,9]. At 24 hours from CTG injection virtually all periventricular DCX+ cells in the SVZ of CBCM mice were labeled by CTG (Fig. 5 A-C9). By contrast, very few DCX+ cells incorporated CTG in the dSVZ, and no CTG/DCX co-labeled cells were identified in the RMS, OB, dPSB and striatal parenchyma (Fig. 5 A-C9 and data not shown). At 7 d post injection CTG+/DCX+ cells migrated out of the periventricular SVZ (Fig. 5 F-F9 and data not shown) and reached the RMS and OB (data not shown), but also the dPSB and the underlying dorso-lateral caudate-putamen (Fig. 5D,E). In the striatum CTG+/DCX+ cells were always identified as individual cells, mostly oriented parallel to the internal capsule fiber bundles (Fig. 5 D, E). No clustered DCX+ cells were observed to be CTG+ at both 24 hours and 7 d survival time (Fig. 5 D, E  arrows).
These data confirm that at least a subset of individual DCX+ cells in the striatum of the CBCM mice derive from the SVZ. In parallel, they support the hypothesis that striatal DCX+ clusters do not derive from this germinative region. The latter cells are in physical contact with neuroblasts and show specific features such as clustering, high proliferative activity and co-expression of glial and neuronal markers [28,29,30,1]. We previously demonstrated the occurrence of intermediate neuronal progenitor-like cells in the striatum of the adult rabbit [9]. To unravel whether DCX+ clusters in the CBCM striatum could be related to the presence of local neuronal progenitors, we firstly investigated the occurrence of cells expressing the endogenous marker of proliferation, Ki67 [31], and their association with DCX+ clusters. In the CBCM striatum the number of Ki67+ cells was about ten times higher than in controls (Fig. 6 C; F 1,2 = 107.1; p = 0.008, n = 3) and 70.663.8% of these cells were organized into clusters of at least four cells (Ki67+ clusters; Fig. 6A,B,D). As expected, Ki67+ clusters in control brains were exclusively found in the DG and SVZ (data not shown).
In contrast to the striatum of control mice, where only rare Ki67+ cells were labelled for DCX, in mutants 42,869,8% of Ki67+ cells co-expressed DCX (K/D+, Fig. 6B0,E). Moreover, 32,8611% of Ki67+ cells were negative for DCX but were either in contact with DCX+ cells or were part of Ki67+ clusters contacting DCX+ cells (we will refer to these cells as Ki67+ cells associated with DCX or K/aD; Fig. 6B0,E). Thus, in the striatum of CBCM mice, Ki67+ cells are mostly clustered and closely associated with DCX+ cells. The analysis of the association of K/ D+ and K/aD cells with individual or clustered Ki67+ and DCX+ cells (Fig. 6G,H), revealed that 80,1610,0% of the K/D+ cells were simultaneously part of Ki67+ and DCX+ clusters (Fig. 6G), while 65,268,8% of K/aD cells belong to Ki67+ clusters contacting DCX+ clusters (Fig. 6H). These data indicate that the Ki67+ clusters are mainly associated, and partly overlapped with the DCX+ clusters. In particular, the K/D+ cells accounted for 53,966,1% of the cells of the Ki67+ clusters (Fig. 6F) and for 25,063,1% of the cells of the DCX+ clusters.
A high proportion of both K/D+ and K/aD cells were labeled by a two hours BrdU pulse (Fig. 7A), indicating these cells were actually proliferating. Notably, K/aD cells displayed a higher proliferation rate compared to K/D+ cells (K/aD 69,567,6%; K/ D+ 43,464,1%; p = 0,012).
Thus, about one fourth of the clustered DCX+ cells do proliferate, and as in the SVZ and DG [32,30] these dividing neuroblasts are clustered together with cells that do not express DCX and shows a higher proliferation rate.
Phenotypic analysis of K/aD and K/D+ cells. To characterize the putative progenitor nature of K/aD and K/D+ cells we performed multiple immunofluorescence labeling for DCX, Ki67 and for: 1) molecules normally expressed by SVZ and DG intermediate neuronal progenitors, such as the glial markers BLBP, SOX2 and SOX9, [30,29]; 2) the EGF receptor (EGFr), a specific marker of SVZ intermediate progenitors, or type C cells, and 3) the marker of subpallial neurons pan-Dlx [32].
These data indicate that most of the K/aD cells and a part of the K/D+ cells in the CBCM striatum co-express a set of markers associated with intermediate neuronal progenitors in the adult brain. In particular, the organization and antigenic profile of the K/aD cells closely resemble those of SVZ type C cells. These striatal intermediate progenitor-like cells are specifically associated with clustered DCX+ cells and might account for their local genesis.

DCX+ neuroblasts in CBCM striatum do not express sets of transcription factors associated with the specification of striatal projection neurons
To investigate the lineage and commitment of striatal neuroblasts in CBCM mice, we analyzed the expression of transcription factors (TFs) associated with the specification of either SVZ neuronal precursors (Sp8 and Pax6; [33,34,35]), or striatal interneurons (Nkx2.1 and Lhx6; [36]) and projection neurons (Isl1 and CTIP2; [37]).
Striatal DCX+ or Ki67+ cells were never found to be immunopositive for PAX6, Nkx2.1, Lhx6, nor yet Isl-1 (data not shown). On the other hand, all DCX+ cells co-expressed Sp8 and CTIP2 ( Fig. 7F and Fig. S2B,C). Interestingly, K/aD cells expressed Sp8 and some of them were also immunopositive for CTIP2 ( Fig. 7F; Fig. S2B). Thus, striatal DCX+ neuroblasts and part of the K/aD cells share the expression of Sp8 and CTIP2 suggesting they belong to the same lineage. Notably, this combination of TFs is not associated with the specification of any known striatal neurons during development.

Survival and differentiation of DCX+ neuroblasts are severely impaired in the CBCM striatum
To investigate the fate of newly generated cells in the CBCM striatum we injected animals with BrdU for five consecutive days (1 injection/day; 50 mg/Kg) and analyzed the phenotype of BrdU+ cells 15 d (n = 3) and 28 d (n = 3) after the first injection. Simultaneous labeling for BrdU, the immature neuronal marker DCX, the mature neuronal markers NeuN and CaMKII-Cre ( Fig. 8A-C) showed that between 15 d and 28 d after BrdU administration, the number of the BrdU+ cells expressing at least one of the tested neuronal markers had greatly decreased (64586457 cells at 15 d vs 4576220 cells at 35 d; F 1,2 = 77,77; p = 0,001). At 15 d all BrdU+ cells in the neuronal lineage expressed DCX and 29.568% of them were co-labeled for NeuN, whereas only rare cells expressed CaMKII-Cre (Fig. 8B). At 28 d we estimated that 250646.1 BrdU+ cells/hemisphere expressed according to the legend in A. C) Reconstructions at higher resolution (voxel size 0,760,762 mm) of dorsal SVZ chains included in three consecutive sections (#124-127; green) and striatal DCX+ clusters taken from the reconstruction shown in Fig. 3B (red). The ependymal surface is shown in yellow. SVZ chains are elongated and coalesce in the dSVZ. By contrast, DCX+ clusters (red) have irregular shapes, and lack any preferred orientation. Scale Bars 100 mm in B. doi:10.1371/journal.pone.0025088.g002 CaMKII-Cre, NeuN but not DCX, indicating these cells had reached a relatively mature stage of differentiation. These few mature neurons represented only about 3.8% of the neuroblast population at 15 d.
By coupling BrdU with markers of striatal projection neurons (DARP-32 and FOXP-1), interneurons (Parvalbumin, Somato-statin, Calretinin) and dopaminergic cells (TH), at both 15 d and 28 d survival, only rare double labeled BrdU/Calretinin cells were found (Fig. 8 E,F). BrdU/Calretinin double positive cells were also observed in the dSVZ (Fig. 8D), a region described as a source of such cells in stroke lesioned striatum [10,12].  Fig. 2. In the reconstructed volume DCX+ clusters (red), the dPSB (dark green) and the dSVZ (pale green) are shown together with nine independent groups of individual DCX+ cells. Three of them contact the dPSB (dPSBconnected groups; blue tones) while the other six do not (dPSB-unconnected groups;orange, and pink/violet tones). The contact between the dark cyan group and the dPSB occurs anteriorly to the reconstructed volume. Scale bars 100 mm in A, 23 mm in A9, 10 mm in C. doi:10.1371/journal.pone.0025088.g003 Overall, these data indicate that a great majority of newly generated neurons in the striatum of CBCM mice die either before or immediately after expressing the CaMKII-Cre failing to replace the medium spiny neurons.

Differentiation of transplanted embryonic striatal precursors is not impaired in the CBCM striatum
Survival and differentiation failure of newborn neurons are common features of both pathologic and physiologic models of striatal neurogenesis [2,9]. This could be due to a combination of cell autonomous mechanisms and environmental cues. In order to investigate whether the CBCM striatum is permissive for the survival and differentiation of correctly specified cells, we transplanted striatal precursors derived from the lateral ganglionic eminence (LGE) of E15 [38] EGFP+ mice into the striatum of adult CBCM and control mice. At 21 days after transplantation, numerous EGFP+ cells were found in the striatum of both CBCM and control mice (Fig. 9 A-G). Among these cells, many were projection neurons, identifiable for the high spine density and DARPP-32 expression (Fig. 9 C,D). Accordingly, several EGFP+ axons travelled within the internal capsule and reached the globus pallidus ( Fig. 9 H,I). In this latter region, axonal fibers were observed to arborize more extensively in CBCM mice than in controls (Fig. 9 H,I).
It is to be noted that in both control and CBCM mice numerous transplanted cells expressing NeuN or DCX dispersed widely from the injection site into the striatum and neocortex (Fig. 9 F,G and data not shown). By contrast, EGFP+/DARPP-32+ projection neurons were mostly restricted to the transplantation site ( Fig. 9A-E). These results indicate that the striatal environment, both in CBCM and control mice, is permissive for differentiation and survival but not migration of striatal projection neuron precursors. Moreover, these data also suggest the differentiation failure of endogenous neuroblasts in CBCM striatum is mostly due to cell autonomous factors.

Discussion
Several studies have analyzed the induction of neurogenesis in non-neurogenic regions of the adult brain following acute injuries [7]. On the other hand, the neurogenic response to a slow progressive neurodegeneration has been poorly investigated. In this study we examined induced neurogenesis in the striatum of a mouse model of slow progressive neuronal degeneration: the CREB1 Camkcre4 CREM 2/2 double transgenic mice (CBCM; [16]). The analyses were performed at advanced stages of neurodegeneration (6-8 month old animals), in which the volume of the striatum was reduced by about 70%.   Origin of the striatal neuroblasts SVZ contribution. In CBCM mice, we demonstrated that SVZ neuroblasts migrate towards the degenerated striatum. These cells reach the CBCM striatum through a specific pathway that runs along the boundary between the corpus callosum and the striatum, or dPSB. This region appears as a lateral extension of the dSVZ. Similarly to the RMS-OB migratory route [39], SVZderived neuroblasts initially migrate organized in chains within the dPSB and then detach from the chains to enter the mature striatal parenchyma as individual cells. In contrast to our results in CBCM mice, in vitro slice cultures of the striatum shortly after acute ischemic injury, showed that SVZ neuroblasts enter the striatum from the whole periventricular SVZ, often organized as chains [40,41]. Moreover, differently from acute striatal lesions [2,6,3] the SVZ showed normal proliferative levels in CBCM mice. These differences further support the fact that acute and chronic degenerative conditions are associated with tissue responses that differentially modulate distinct aspects of adult neurogenesis [14,15].
Local neuronal genesis contribution. Although our analyses in CBCM mice indicated that SVZ neuroblasts migrate individually within the striatum, about half of the striatal DCX+ cells were organized in clusters. Since more than 60% of the clustered DCX+ cells were generated within five days, our data suggest these cells might not have migrated from the SVZ. In line with this hypothesis, we found strong evidence that clustered Parenchymal generation of neuroblasts had previously been demonstrated in the normal rabbit striatum, where similar clustered proliferating neuroblasts were observed [9]. This finding was further supported by in vitro experiments in which BrdU labeled neuroblasts migrated out of striatal explants obtained from BrdU pulse labeled animals [9]. Unfortunately, in CBCM mice we could not obtain in vitro migration of neuroblasts neither from SVZ nor striatal explants, which is likely due to the poor viability of such explants in aged mice (Luzzati et al. unpublished observation).
Since SVZ migrating neuroblasts are able to divide after their genesis [42,43], and that cell division would likely dilute CTG labeling, we cannot exclude an SVZ origin of K/D+ cells. Nonetheless, considerable evidence supports the hypothesis that these cells were produced in the context of a more specialized neurogenic system. Indeed, as in the SVZ, DG and rabbit striatum [24,9,29], proliferating neuroblasts of CBCM striatum were specifically associated with Ki67+/DCX negative cells (K/aD cells). In contrast to the SVZ neuroblasts migrating to the CBCM striatum, the K/D+ and K/aD cells showed characteristic features of intermediate progenitors including: mitotic activity, clustering, co-expression of markers of both glial (BLBP, SOX2, SOX9) and neuronal lineages (pan-Dlx; Sp8 and in part DCX), expression of the EGFr, and close contact with post-mitotic neuroblasts (Fig. 10) [32,30,28,29]. In particular, the organization and molecular profile of the K/aD cells closely resemble those of SVZ type C cells. The origin of these putative striatal parenchymal neuronal progenitors remains to be determined, however, two main possibilities can be considered: i) they could derive from the SVZ through displacement of type B/C cells, or alternatively ii) they could represent local cells that became neurogenic in response to neurodegeneration. The latter possibility would be consistent with previous in vitro studies indicating the presence of neurogenic progenitors in different brain regions including the striatum [44,45], and with recent in vivo evidence showing the activation of local parenchymal progenitors in the neocortex after mild ischemia [8]. Both the migratory capacity of SVZ progenitors and the neurogenic potential of local cells would be of great interest for neural stem cell research, and deserve further investigation.

Commitment and fate of striatal neuroblasts
In agreement with other pathologic and physiologic models of striatal neurogenesis [2,6,9], phenotypic analyses showed that induced neuroblasts in the CBCM striatum did not survive or differentiate into striatal neuronal types, except for a few calretinin expressing cells [9,46,10]. This could be due to cell-autonomous rather than environmental factors, because LGE derived embryonic striatal precursors engrafted into the CBCM striatum correctly differentiated into projection neurons.
One possible cell-autonomous factor reducing the survival of endogenous striatal neuroblasts in CBCM mice is CREB deletion, which occurs under the control of CaMKIIa promoter [16]. However, in these animals CaMKIIa drives Cre expression at late neuronal differentiation stages, leading to normal development  [16], and only 25% reduction in adult-generated OB interneuron survival [27]. The dramatic loss of striatal newborn neurons occurring in CBCM mice suggests that other cell-autonomous factors can affect the survival of these cells. As in the post-stroke rat striatum [10], striatal neuroblasts in CBCM mice do not express a set of TFs specifically involved in the differentiation of medium spiny neurons, but express Sp8, which is also found in SVZ cells fated to become OB interneurons and striatal calretinin interneurons [35,12]. Interestingly, in the normal post-natal guinea pig striatum we recently observed a population of neuroblasts having a transient existence and showing Sp8+ expression [47]; Luzzati F unpublished results]. These cells were in contact with the ventral PSB, and similar to the neuroblasts of the dPSB connected groups of CBCM mice, they were closely associated with internal capsule fiber bundles [47], (Luzzati F unpublished results). These observations are consistent with the possibility, so far not considered, that striatal neurodegeneration may stimulate the production of a transient neuronal cell type. that might be involved in some forms of reconfiguration of the incoming cortical and/or thalamic inputs.

Concluding remarks
The mature brain parenchyma has been classically considered non permissive for neuronal progenitor activity. This idea comes mostly from the observations that transplantation of stem/ progenitor cells into the mature brain parenchyma mainly gives rise to glial cells [48,49,50]. Transplantation of SVZ neurospheres in the striatum of CBCM mice yielded similar results (Luzzati et al. unpublished results). Nonetheless, our data in CBCM mice and in normal rabbits, together with recent studies in the neocortex [9,8] indicate that parenchymal neuronal production can occur in specific circumstances. A deeper understanding of the biology of the parenchymal neuronal progenitors could feed the development of cell replacement/restorative therapies based on local neuronal production. Future studies should also address whether, as for SVZ neuronal progenitors [51,52], activation of ectopic neurogenic progenitors, either derived from migrating SVZ progenitors or local cells, can give rise to brain tumors. Video S1 LSCM serial section reconstruction of two dPSB connected groups. The first part of this video shows a 360u rotation, beginning with a posterior view, of the maximum intensity projection (MIP) volume rendering of an LSCM serial section reconstruction (sections #122-127; voxel size 0,1360,1361 mm) encompassing a cuboid volume of the dorsolateral striatum (base: 2406240 mm). In this volume the lateral part of the dPSB and part of two dPSB connected groups (blue and cyan of Fig. 3D) are shown. In the second part of the video tracings of the dPSB chains and of the two dPSB connected groups, obtained with neuronstudio, are shown. Note the complete absence of clustered DCX+ cells in these groups.  . Proposed developmental progression of local parenchymal progenitors in CBCM striatum. In this model highly proliferating Ki67+ clusters give rise to DCX+ neuroblasts initially organized in clusters and then as individual cells. These cells eventually differentiate into DCX2, NeuN+, CamKIIa-Cre+ mature neurons. The Ki67+ cluster are made by the K/aD cells that express EGFr but not DCX and K/D+ cells that express DCX but not the EGFr, both cell types express SOX2, BLBP, SOX9, Dll and Sp8. Postmitotic neuroblasts downregulate BLBP but maintain SOX2 and SOX9 expression until they are part of the DCX+ clusters. The entire lineage is characterized by Sp8 and Dll expression. CTIP2 is also expressed by all DCX+ cells but only a few K/aD cells. A question mark indicates that the expression of Sp8, CTIP2 and Dll has not been determined in DCX negative newly generated striatal neurons. doi:10.1371/journal.pone.0025088.g010