In Vivo Transcriptional Profile Analysis Reveals RNA Splicing and Chromatin Remodeling As Prominent Processes for Adult Neurogenesis

www.elsevier.com/locate/ymcne Mol. Cell. Neurosci. 31 (2006) 131 – 148

In vivo transcriptional profile analysis reveals RNA splicing and chromatin remodeling as prominent processes for adult neurogenesis

Daniel A. Lim,a,* Mayte Sua´rez-Farin˜as,b Felix Naef,b Coleen R. Hacker,c Benedicte Menn,a Hirohide Takebayashi,a Marcelo Magnasco,b Nila Patil,c and Arturo Alvarez-Buyllaa,* aDepartment of Neurological Surgery and Developmental and Stem Cell Biology Program, University of California, San Francisco, CA 94143, USA bThe Rockefeller University, 1230 York Ave., New York, NY 10021, USA cPerlegen Sciences Inc., 3380 Central Expressway, Santa Clara, CA 95051, USA

Received 10 May 2005; revised 21 August 2005; accepted 4 October 2005 Available online 5 December 2005

Neural stem cells and neurogenesis persist in the adult mammalian SVZ neurogenesis and reveal RNA splicing and chromatin remodel- brain subventricular zone (SVZ). Cells born in the rodent SVZ ing as prominent biological processes for these germinal cells. migrate to the olfactory bulb (Ob) where they differentiate into D 2005 Elsevier Inc. All rights reserved. interneurons. To determine the gene expression and functional profile of SVZ neurogenesis, we performed three complementary Keywords: Subventricular zone (SVZ); Neurogenesis; Stem cell; Adult sets of transcriptional analysis experiments using Affymetrix brain; Microarray; Transcription; Transcriptional profile; Chromatin GeneChips: (1) comparison of adult mouse SVZ and Ob gene remodeling; RNA splicing expression profiles with those of the striatum, cerebral cortex, and hippocampus; (2) profiling of SVZ stem cells and ependyma isolated by fluorescent-activated cell sorting (FACS); and (3) analysis of gene expression changes during in vivo SVZ regeneration after anti- Introduction mitotic treatment. Gene Ontology (GO) analysis of data from these three separate approaches showed that in adult SVZ neurogenesis, The adult brain harbors neurogenic stem cells within the RNA splicing and chromatin remodeling are biological processes as statistically significant as cell proliferation, transcription, and subventricular zone (SVZ) of the lateral ventricle wall (Garcia- neurogenesis. In non-neurogenic brain regions, RNA splicing and Verdugo et al., 1998; Gage, 2000). In neonatal (Luskin, 1993) and chromatin remodeling were not prominent processes. Fourteen adult mice (Lois and Alvarez-Buylla, 1994; Doetsch and Alvarez- mRNA splicing factors including Sf3b1, Sfrs2, Lsm4, and Khdrbs1/ Buylla, 1996; Jankovski and Sotelo, 1996; Thomas et al., 1996), Sam68 were detected along with 9 chromatin remodeling genes cells born in the SVZ migrate a long distance to the olfactory bulb including Mll, Bmi1, Smarcad1, Baf53a, and Hat1. We validated the (Ob) where they differentiate into interneurons. SVZ astrocytes transcriptional profile data with Northern blot analysis and in situ (type B cells) are neural stem cells (Doetsch et al., 1999a,b; Laywell hybridization. The data greatly expand the catalogue of cell cycle et al., 2000) and give rise to rapidly dividing, immature-appearing components, transcription factors, and migration genes for adult cells (type C cells) that generate migratory neuroblasts (type A cells) (Lois and Alvarez-Buylla, 1994; Peretto et al., 1997; Luskin, 1998; Doetsch et al., 1999a,b). See Figs. 1B, C. SVZ ependymal cells are themselves not neurogenic (Chiasson et al., 1999; Laywell et al., 2000; Capela and Temple, 2002) but may be important for generating the SVZ neurogenic niche (Lim et al., 2000; Goldman, Abbreviations: SVZ, subventricular zone; Ob, olfactory bulb; ObC, 2003; Peretto et al., 2004). Although the SVZ cellular architecture olfactory bulb core; Ctx, cortex; St, striatum; Hp, hippocampus; ds cDNA, (Gates et al., 1995; Jankovski and Sotelo, 1996; Doetsch et al., double-stranded complementary DNA; cRNA, complementary (antisense) 1997; Peretto et al., 1997), stem cell identity (Chiasson et al., 1999; RNA; FACS, fluorescent-activated cell sorting; ECM, extracellular matrix; Doetsch et al., 1999a,b; Laywell et al., 2000; Rietze et al., 2001; RMS, rostral migratory stream; EGF, epidermal growth factor; FGF, fibroblast growth factor. Capela and Temple, 2002; Imura et al., 2003), and neurogenic * Corresponding authors. lineage (Doetsch et al., 1999a,b) have been defined, the genetic E-mail addresses: [email protected] (D.A. Lim), program for adult SVZ neurogenesis is poorly understood. [email protected] (A. Alvarez-Buylla). The transcriptional changes of differentiating neocortical (East- Available online on ScienceDirect (www.sciencedirect.com). erday et al., 2003; Karsten et al., 2003) and postnatal SVZ-derived

to have a similar migration capacity. Therefore, transcriptional analysis of in vivo SVZ neurogenesis is required to identify genes and biological processes involved in this continual generation of neurons for the Ob. Using high-density oligonucleotide (GeneChip) arrays, we undertook three complementary approaches to determine the transcriptional profile of in vivo SVZ neurogenesis. We first compared gene expression differences of the SVZ-Ob system with that of three other brain regions. We then utilized FACS methods to compare the transcriptional profiles of type B cells – the neurogenic stem cell – and the non-neurogenic ependyma. Finally, we analyzed the transcriptional changes of the SVZ as it regenerated type C and A cells from a population of type B cells. Data integrated from these three approaches identified genes, signaling pathways, and biological processes related to SVZ neurogenesis. In addition to expanding the catalogue of cell cycle components, transcription factors, and genes for migration, we identified RNA splicing and chromatin remodeling as prominent processes for adult neurogenesis. The importance of RNA splicing and chromatin remodeling has not been described for SVZ neurogenesis, and we here provide evidence that these processes are as upregulated as the expected processes of cell cycle, transcription, and neurogenesis. We focused our Results and Discussion below only on a subset of genes with special attention to RNA splicing and chromatin remodeling, however, both the raw chip image data and other data analyses are available (Supplementary data and at http:// asterion.rockefeller.edu/mayte/Neurogenesis) for future compara- tive expression profile analyses with other developmental, adult, or tumor cell populations.

Results

Fig. 1. (A) Brain regions dissected for the brain region transcriptional analysis. Dissected areas are shown in yellow. The SVZ contains three Brain region transcriptional profile analysis identified genes with populations of neurogenic precursors—type B, C, and A cells. (B) The increased expression in the SVZ-ObC neurogenic system lineage of SVZ neurogenesis. Type B cells (blue) are SVZ astrocytes that self- renew and give rise to a rapidly dividing population of immature-appearing We analyzed the transcriptional profiles of the SVZ, Ob core cells—type C cells (green). The transit-amplifying type C cells then become (ObC), and three other brain regions indicated in Fig. 1A. The ObC type A cells (red), the neuroblasts that migrate into the ObC. (C) Architecture dissection excluded the mitral and periglomerular layers, providing a of the SVZ. The ventricle is to the left. Ciliated ependymal cells (gray) line RNA sample primarily representing migratory type A cells, the ventricle wall. Some type B cells (blue) make contact with the ventricle maturing neuroblasts, and mature granule cells. The hippocampus lumen (arrow). Both type C (green) and A cells (red) are in direct contact with (Hp) dissection included the non-neurogenic CA1–CA3 regions as the type B cells. In this panel, type A cells are migrating toward the ObC in a well the dentate gyrus. The striatum (St) was the region directly direction perpendicular to the page. (D) Sagittal view of the mouse brain. Within the SVZ, there is an extensive network of type A cells migrating underlying the SVZ dissection. The cortex (Ctx) did not include the tangentially toward the ObC. This network of pathways coalesces at the corpus callosum. Biotin-labeled complementary RNAs (cRNAs) anterior of the SVZ to form the rostral migratory stream (curved arrow). The derived from each brain region were analyzed on GeneChip Mu11k rostral migratory stream enters the ObC where type A cells then migrate expression arrays, which contain more than 13,000 probe sets radially and disperse (red dots) throughout the Ob. The major biological analyzing the expression of over 11,000 unique genes. Each brain processes that occur in the SVZ alone (SVZ profile), SVZ and ObC (SO region was analyzed independently twice; the data among the profile), and ObC alone (ObC profile) are listed to the left, middle, and right, duplicates were consistent (Supplementary data S1). respectively; the cell types of the SVZ and ObC profiles are in bold. To focus our analysis on those genes more likely to be involved in SVZ neurogenesis, we filtered the data (see Experimental (Gurok et al., 2004) neurospheres have also been studied in vitro. methods) for those genes that are (1) increased in the SVZ—the Many neurospheres are derived from transit-amplifying cells (type SVZ profile, (2) increased in the ObC—the ObC profile, and (3) C cells) (Doetsch et al., 2002), and their exposure to growth factors increased in both the SVZ and ObC—the SO profile (Fig. 1D). The (EGF or FGF) in vitro deregulates the normal genetic control of SVZ profile (Supplementary data S2) contained 65 unique genes cell differentiation that occurs in vivo (Gabay et al., 2003; Santa- (71 probe sets) with increased expression in the SVZ as compared Olalla et al., 2003; Hack et al., 2004); it is likely that the expression to all other regions (ObC, Hp, Ctx, St). The ObC profile profiles of neurospheres and endogenous SVZ precursors differ. (Supplementary data S3) included 168 genes (209 probe sets), Furthermore, in vivo, SVZ neurogenesis involves a long-distance, and the SO profile (Supplementary data S4) contained 60 genes (80 directional migration to the Ob while neurospheres do not appear probe sets). Genes in the SVZ, SO, ObC profiles are shown D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 133

Table 1 Onto-Express (see Experimental methods). The functional profiles Correlation between published expression data and GeneChip brain region of SVZ, SO, and ObC gene expression are shown in the pie charts of profile analysis Figs. 3B–D. Gene Profile Published expression The SVZ is the primary site where type B and C cells are SVZ ObC Reference maintained and proliferate. Compared to the other brain regions in our analysis, the SVZ is the most proliferative. As expected, the Dlx1 SO + + Lois, 1996 biological processes of proliferation and cell cycle were prominent Dlx2 SO + + Doetsch et al., 2002 Sox2 SO + + Ferri et al., 2004 in the SVZ profile (Fig. 3B). From the SVZ, type A cells tangentially Pbx1 ObC + ++ Redmond et al., 1996 migrate into the ObC where they then turn to migrate radially into the Mash1 –++Parras et al., 2004 granule cell layer. Within the granule cell layer, the type A cells Er81/Etv1 –++Stenman et al., 2003 undergo terminal differentiation and integrate into local circuits (Fig. Vim SVZ + À Doetsch et al., 1997 1D). There is also a continual turnover of young neurons in the Ob Mki67 SVZ + À Zhu et al., 2003 involving apoptosis (Najbauer and Leon, 1995; Fiske and Brunjes, Rrm1 SVZ + À Zhu et al., 2003 2001; Petreanu and Alvarez-Buylla, 2002). The ObC profile Notch1 –+?Stump et al., 2002 therefore should reflect these later stages of SVZ-Ob neurogenesis Wnt5a ObC À + Shimogori et al., 2004 as well as granule cell turnover. Indeed, significant biological Thra SVZ + À Lemkine et al., 2005 processes were development, neurogenesis, and cell differentiation; Nog ObC + ++ Peretto et al., 2004 Nestin –+À Doetsch et al., 1997 other highly significant GO terms included CNS and brain Cd24 SO + + Calaora et al., 1996 development, negative regulation of cell proliferation, axonogenesis, and apoptosis/programmed cell death (Fig. 3D). Therefore, Genes expressed in SVZ and/or ObC but not on Mu11K chip Olig2, Emx2, Slit, Ng2, Dcx, Gli1. GO analysis described many of the major known and expected biological processes that occur in the SVZ and ObC regions. The SO profile represents gene expression common to both the clustered in a color matrix in Fig. 3A. Genes that had decreased SVZ and ObC. As expected, SO profile terms related to cell growth, expression in the SVZ, SO, and ObC can be found in transcription, protein metabolism, and development (Fig. 3C). The Supplementary data S12. most prominent biological process in the SO profile, however, was To assess the sensitivity of the SVZ, ObC, and SO profiles, we RNA splicing (Fig. 3C); terms related to RNA splicing appeared 9 surveyed the literature to identify those genes that would be times in this analysis (in Supplementary data S5), all with very high expected to be detected in our analysis. Of the genes represented significance ( P values are in the figure). GO terms related to on the Mu11K GeneChip set, we identified 15 that are highly chromatin regulation terms appeared 7 times, including terms from expressed in the SVZ and/or ObC relative to the other brain regions all three GO categories (Fig. 3C, and in Supplementary data S5). The (Hp, Ctx, St). Of these 15 genes, our analysis detected 11 (73%) SVZ profile also was significant for Festablishment and/or mainte- with a profile matching the published in situ hybridization or nance of chromatin architecture_ as well as components of chromatin immunohistochemical data (Table 1). Six other genes previously described to be expressed highly in the SVZ and/or ObC were not represented on the Mu11K GeneChip set. To validate the array data with another measure of transcript levels, we analyzed 8 genes by Northern blot: Ccnd2, Hmgb2, Mia, Pdyn, Dlx1, 2310021G01Rik, Sox11, and Col6a1. For all of the genes tested, the Northern blot data paralleled the pattern of expression observed on GeneChip analysis (Fig. 2).

Gene Ontology analysis identifies RNA splicing and chromatin regulation as prominent biological events in the SVZ and ObC brain regions

To translate the gene expression data into functional profiles, we used Gene Ontology (GO) analysis. GO provides an organized vocabulary of terms that can be used to describe a gene product’s attributes (www.geneontology.org). GO terms are organized into three categories (biological process, cellular component, and molecular function) in structures called directed acyclic graphs; these structures differ from hierarchies in that a Fchild_ (more specialized term) can have several Fparents_ (less specialized term). To analyze the GO terms of the SVZ, SO, and ObC profiles, we Fig. 2. Northern hybridizations substantiate array data. Northern blot used Onto-Express (Khatri et al., 2002, 2004). For each GO term, analysis was performed on the cRNA samples (left). Corresponding Onto-Express computes its significance ( P value), allowing one to GeneChip array data for the brain region analysis (duplicate data shown, distinguish prominent biological processes from non-significant indicated as 1 or 2 under the brackets) is shown as a color matrix (right, events. A complete list of GO terms for the SVZ, SO, and ObC red—increased expression, green—decreased expression). Eight out of profiles with associated P values is in Supplementary data S5, and 8 genes tested by Northern blot had good agreement with the array data. the parent–child relationship of the GO terms can be browsed with Fold change scale (log2) for the color matrix is shown at the bottom right. 134 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Fig. 3. The brain region transcriptional profiles. (A) Color matrix of the SVZ, SO, and ObC profiles. Genes are ordered along the vertical axis using hierarchical clustering. Duplicate profiles of the brain regions are presented on the horizontal axis. The color and color intensity of each cell in the matrix relate to the expression ratio of each gene. Red indicates a positive ratio (expression greater than the mean of the other brain regions), green indicates a negative ratio, and black indicates a ratio of 1. A color scale (log2) indicating the magnitude of the expression ratios is shown in the bottom. (B–D) GO analysis pie charts for the brain region profiles. The entire pie represents all GO terms in the analysis. Pie slices are proportional to the number of genes (in parentheses) related to a particular GO Fparent_ term (legend for color code is in the inset to the right of each panel). GO terms that are Fchildren_ of a parent term are listed next to the pie chart with an indicating line. Further parent–child relationship of the GO tree structure is indicated by indentation with hyphen. All listed GO terms are statistically significant, and color of the type indicates the GO category (see legend at the lower right of the figure). D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 135 and nucleosomes (Fig. 3B). Thus, the data suggest that both RNA because they are important biological processes but not well splicing and chromatin regulation are important biological processes described for the adult SVZ and Ob. for SVZ neurogenesis. Using the GO analysis and a review of the literature, we To determine the relative prominence of RNA splicing and identified genes related to RNA splicing and chromatin remodel- chromatin remodeling for SVZ neurogenesis in comparison to non- ing in the SVZ, SO, and ObC expression profiles. The SO profile neurogenic brain regions, we performed GO analysis on the sets of contained RNA splicing factors Sf3b1, Sfrs2, Lsm4, Snrpg, genes that were increased in the Ctx (Ctx profile), St (St profile), and Snrpd2, Hnrpa2b1, Hnrpd, Hnrpm, Hnrpdl, Hnrph1,and Hp (Hp profile) (probe set lists in Supplementary data S11, GO term Khdrbs1/Sam68, and the ObC profile contained Snrpb (Table lists in S5). No terms related to RNA splicing were statistically 2). Chromatin-remodeling genes Mll, Hat1, Hmgb3, and Baf53a significant in the Ctx, St, or Hp profiles. In the Ctx profile, the term were detected in the SO profile, Hmgb2 and H2afx were in the Fchromatin remodeling_ was associated with 2 genes and a P value SVZ profile, and the ObC profile contained Bmi1 and Smarcad1 of 0.02; however, the parent term of Festablishment and/or (Table 2). maintenance of chromatin architecture_ was not statistically significant ( P = 0.37). No GO terms related to chromatin Gene expression comparison of the type B SVZ stem cell and the remodeling were significant in the St or Hp profiles. Thus, RNA non-neurogenic ependyma reveals chromatin regulation as a splicing and chromatin remodeling were much more prominent in prominent process in type B cells the SVZ and SO profiles than in the Ctx, St, and Hp. In the Supplementary text, we identify and discuss the genes Neurogenic SVZ cells are closely associated with the non- detected in our SVZ-Ob analysis related to cell cycle, transcription, proliferative ependymal cells that line the walls of the lateral migration, and apoptosis. The majority of those genes has not been ventricle (see Fig. 1C). The SVZ and SO profiles therefore previously described for adult SVZ-Ob neurogenesis, and thus, the contained the gene expression of non-neurogenic ependyma. We data present a wealth of gene candidates for future study. In this used fluorescent-activated cell sorting (FACS) to separate the type manuscript, we focus on RNA splicing and chromatin remodeling B cells and ependyma and compared their gene expression profiles.

Table 2 Chromatin-remodeling and RNA splicing genes in the brain region profiles

Highlighted cells indicate the profile to which each probe set/gene belongs (e.g., Baf53a has cells in both the SVZ and ObC columns highlighted, indicating the SO profile). 136 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

To isolate type B cells, we used antibodies to GFAP (Doetsch et al., for Gfap did not show differential expression, however, the Gfap 1999a,b). Immunocytochemistry for this intracellular antigen mRNA was differentially represented in the representative cDNA requires permeabilization of the cell membrane. We developed libraries as shown by RT-PCR (Fig. 4J). A small fraction of the methods to isolate RNA from cells permeabilized by a non-ionic probe sets on the Mu11K arrays assess transcript levels poorly (N. detergent (Tween-20) and confirmed that the RNAs are stable Patil, personal communication), and it is possible that the probe set through the immunostaining protocol (Figs. 4H, I, K). GFAP+ cells for Gfap is problematic. NOG (Noggin) protein has been were generally round or elliptical and not ciliated (Figs. 4C, D). previously shown to be highly expressed in ependymal cells We used CD24 antibodies to purify ependymal cells (Capela and (Lim et al., 2000; Peretto et al., 2004) and SVZ astrocytes (Peretto Temple, 2002). CD24 staining was also performed with Tween-20 et al., 2004), however, we did not find elevated expression of Nog so that any changes in the gene expression profile associated with in either the SVZ profile or CD24+ cells. There may be a mismatch this agent would be comparable to those observed in the GFAP+ between transcription and translation for the Nog gene, resulting in population. To a lesser degree, CD24 antibodies also stain SVZ Type a pattern of low mRNA transcript levels but high Noggin protein A cells (Calaora et al., 1996); however, our dissociation protocol and concentrations in the SVZ and ependymal cells. It is also possible Tween-20 treatment eliminated the CD24 epitope from the surface that differential expression for any gene is not detected due to a of type A cells. CD24 antibody staining strongly labeled multi- loss of transcript during FACS or cDNA amplification. ciliated ependymal cells (Figs. 4A, B); CD24+ non-ciliated cells GO analysis showed that type B cells are significant for cell were not observed. proliferation and cell cycle, while ependymal cells are significant for SVZ cells immunostained for CD24 and GFAP were sorted by cell cycle arrest (Table 3). These data are consistent with the finding FACS (Figs. 4E, F). Total RNA from type B and ependymal cell that ependymal cells do not divide in vivo (Doetsch et al., 1999a,b; populations was isolated, and mRNAs were amplified as schema- Capela and Temple, 2002; Spassky et al., 2005). The process of tized in Fig. 4G and described in Experimental methods. The neurogenesis was also significant in type B cells and not in amplification procedure preserved the appropriate mRNA size ependyma, supporting the data that ependyma are non-neurogenic distribution as well as differential expression of GFAP and CD24 (Chiasson et al., 1999; Capela and Temple, 2002). Like the SVZ and (Figs. 4I, J). The cRNAs produced for GeneChip analysis were SO profiles, establishment and/or maintenance of chromatin also of an appropriate size distribution, and GAPDH Northern blot architecture was prominent in type B cells along with histone analysis shows a single band of expected size, indicating that the acetyltransferase activity. FmRNA metabolism,_FmRNA proc- amplification procedure did not produce degraded transcripts (Fig. essing,_ and Fnuclear mRNA splicing via spliceosome_ were not 4K). Scatter plots comparing expression profiles of duplicate significant GO terms in either cell population. Ependymal cells have samples show good reproducibility (see Supplementary data S6). a basal–apical orientation, and the GO term for Fapical plasma Differential expression of 1324 probe sets (1282 unique genes) membrane_ was significant in these cells along with peroxidase was detected between GFAP+ and CD24+ cells. 54% of the genes activity. A complete listing of GO terms for the FACS data is in had increased expression in GFAP+ cells, and 46% were increased Supplementary data S7. in the CD24+ cells. To confirm the FACS cell separation and There were 82 probe sets (78 unique genes) at the intersection of cDNA amplification, we examined the data for expected differen- the FACS data and the brain region profile data. Fold-change values tial gene expression. Cd24 itself was strongly increased (146-fold) for genes at this intersection are indicated in the tables of in the CD24+ population, paralleling the RT-PCR result of Fig. 4J. Supplementary data S2–4. Cell cycle related genes Ccnd2, Cdca7, In the SVZ, Sox2 is expressed highest in the ependyma (Ferri et al., Mki67, Rrm2, and Mcm7 were increased in type B cells; no cell 2004), and the FACS data reported Sox2 expression as 3.8-fold cycle genes were statistically significantly elevated in the CD24+ higher in the ependymal cells relative to the type B cells. Spa17 is population. Of the 10 RNA splicing genes in the SO profiles, only a component of cilia (Grizzi et al., 2004), and it was expressed 11- Snrpg was differentially expressed (1.7-fold increased in CD24+ fold higher in the ciliated CD24+ ependymal cells. The probe set cells). Of the chromatin-remodeling genes, Mll, H2afx, and Hmgb3

Fig. 4. FACS analysis of SVZ cells. (A–D) Immunostaining of dissociated SVZ cells. (A) DIC image and (B) immunofluorescent image of a multiciliated CD24- positive ependymal cell. Arrow indicates ependymal cilia. Panels C and D show respective DIC and immunofluorescence images of a GFAP-positive SVZ cell. (E, F) FACS of immunostained SVZ cells. (E) SVZ cells stained only with secondary antibodies. Cross-bars shown isolate >99% of the non-specific signal in the lower left quadrant. (F) SVZ cells stained for CD24 and GFAP. Rectangle R1 indicates the collection gate for the GFAP+, CD24À population. R2 indicates the collection gate for the GFAPÀ, CD24+ cells. (G) Schematic of cDNA amplification procedure. Briefly, mRNA is reverse transcribed from an oligo-dT primer containing a T7 RNA polymerase promoter sequence. A specific oligonucleotide (SMARTIII oligo) containing a stretch of dG nucleotides is included in the reaction, and the ‘‘strand-switching’’ activity of the reverse transcriptase copies the SMARTIII sequence to the end of the cDNA. With primers to the SMARTIII and oligo-dT T7 promoter sequences, two rounds of long-distance PCR (LD-PCR) are used to amplify the cDNA. For hybridization, cRNAs are produced from the 3V T7 promoter. See Experimental methods for details. (H) Cellular RNAs are stable through the immunostaining protocol. 1 Â 106 SVZ cells were double immunostained for GFAP and CD24. Omission of 0.1% Tween-20 results in no GFAP staining. 1% Tween-20, RNasin, and DTT were added to the staining solutions where indicated (+). After staining, cells were incubated at 4-C for an additional 1.5 h. Total cellular RNAwas then extracted and analyzed on agarose gel. No RNA degradation was detected in any staining protocol. Note that if SVZ cells are freeze thawed and incubated at 37-C, all of the 28S and 18S RNAs are degraded (right lane). (I) Analysis of ds cDNA libraries from FACS SVZ cells. A portion of the ds cDNAs after the first round of LD-PCR was used as the template in a second round of control LD-PCR reactions in which aliquots were taken after 6, 8, 10, and 12 cycles. The cDNA aliquots were analyzed on agarose gels (left panel). The size distribution of the amplified cDNAs was not biased toward smaller products by the LD-PCR. Southern blot signal for GAPDH was a single band, indicating that the initial mRNA was not heavily degraded. The linear range of amplification was determined by both the GAPDH signal intensity and visual inspection of the ethidium bromide stained cDNA population. (J) Semi-quantitative RT-PCR confirms the separation of SVZ cells by FACS. The GFAP message was more than 10-fold enriched in the cDNAs prepared from the GFAP+, CD24À cell population (R1) as compared to the GFAPÀ, CD24+ population (R2). Conversely, the CD24 message was more than 20-fold enriched in the cDNAs from the R2 population in comparison with that of the R1 population. (K) Agarose gel and Northern analysis of cRNAs from FACS-derived ds cDNAs. Size distributions were as expected for brain tissue and GAPDH messages did not show signs of mRNA degradation. D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 137 were increased in type B cells by 3.7, 9.4, and 1.6-fold, respectively days after pump removal, type C cells emerge, and after that, type A (Table 2). Therefore, some chromatin-remodeling genes may begin cells form. Within 10 days, the entire network of migrating expression in the stem cell population of the SVZ and continue into neuroblasts with clusters of B and C cells is reconstituted (Doetsch the ObC. Discussion of some of the other notable gene expression et al., 1999a,b). See Fig. 5A for illustration of SVZ regeneration. differences between type B cells and ependyma is in the We profiled gene expression at 1, 3, and 10 days (A1, A3, A10) Supplementary text. after AraC pump removal. To control for the effects of surgery, we analyzed gene expression of saline infusion at 1 day (S1) and 10 Analysis of SVZ gene expression changes during SVZ regeneration days (S10) after pump removal. We also in parallel analyzed SVZ also identifies RNA splicing and chromosome organization as from unmanipulated animals. prominent biological processes First, we identified genes whose expression was significantly regulated ( P < 0.05) in at least one comparison to untreated SVZ We next analyzed gene expression changes during in vivo (total of 1758 probe sets). SVZ dissections include a small amount of regeneration of the SVZ germinal zone. Osmotic pump infusion of underlying striatal tissue; to focus our analysis on genes expressed the anti-mitotic cytosine arabinoside (AraC) onto the surface of the strongly in the SVZ, we filtered the AraC data with the list of genes brain eliminates type A and C cells, leaving behind only type B cells (985 probe sets) that were determined to be increased in the SVZ as and ependyma. After AraC pump removal, the SVZ regenerates with compared to the underlying striatum ( P < 0.05) in the brain region remarkable fidelity. First, type B cells begin dividing. Between 2 to 4 experiment. The 229 probe sets at the intersection of these two lists 138 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Table 3 profiles are derived from filters based on expression levels relative GO term differences between type B cells (GFAP+) and ependyma to an artificial mean (see Experimental methods), they are not (CD24+) intended to indicate the absolute presence or absence of gene expression in the brain regions analyzed. For instance, a gene in the ObC profile should be expressed at a level statistically higher than the calculated average of all brain regions; however, an ObC profile gene may not necessarily be expressed exclusively in the ObC. To better understand how the expression profile data predicts in vivo expression patterns, we performed ISH for some of the genes. Dlx5 and Mrg1/Meis2 were found in the ObC profile, and ISH demonstrated that both Dlx5 and Mrg1/Meis2 are expressed in both the ObC and the SVZ (Figs. 6A, B, E, F). To provide a comparison to an SO profile gene, we performed ISH for Dlx2 in parallel (Figs. 6C, D). As assessed by ISH, ObC profile genes Dlx5 and Mrg1/ Meis2 both were more intensely expressed in the ObC as compared to the SVZ; in comparison, the SO profile gene Dlx2 was expressed higher in the SVZ than in the ObC. Therefore, ObC profile genes may be expressed in SVZ, but the ObC/SVZ expression ratio is higher than that of SO profile genes. The GeneChip data also predict that Mrg/Meis2 expression levels in the SVZ and St should be similar, and the ISH data are consistent with this prediction. Thus, the GeneChip data provide a reasonable Highlighting indicates statistical significance of the listed GO term (e.g., estimation of relative gene expression levels as assessed by ISH. F _ cell cycle arrest is significant in the CD24+ cells and not the GFAP+ cells. We next used ISH to examine the gene expression of the RNA splicing genes Sfrs2, Sf3b1, Lsm4, and Khdrbs1/Sam68 and chromatin remodeling genes Mll and Smarcad1 (Fig. 6). Sfrs2 is were then analyzed with Principle Component Analysis (PCA) to clearly expressed in the SVZ and ObC. A low level of Lsm4 allow us to separate the gene expression changes of SVZ expression was detected in the ObC, however, ISH was not evident regeneration from that of surgery and saline infusion (see outside of that region; it is likely that the ISH detection threshold Experimental methods for details of the filters and PCA). The gene for this gene was low, and we confirmed Lsm4 expression in both expression pattern of the 59 probe sets (57 unique genes) most the SVZ and ObC with RT-PCR (data not shown). Sf3b1 and related to SVZ regeneration is shown clustered in a color matrix (Fig. Khdrbs1/Sam68 were both clearly expressed in the SVZ and ObC 5B), and a list of these genes is in Supplementary data S8. at levels higher than the other brain regions. The chromatin- The 59 probe sets shown share a similar expression pattern remodeling gene Mll was expressed at moderate levels in all brain representing the initial destruction and later regeneration of the regions; however, it was detected in the SVZ and at relatively SVZ. At A1, gene expression is decreased relative to S1 (A1 < S1). higher levels in the ObC. Similarly, SWI/SNF family member Between A1 and A10, gene expression returns to near normal Smarcad1 was expressed moderately in all brain regions; however, levels (A10 ; S10) or even Fsupranormal_ levels (A10 > S10); its expression was very prominent in the SVZ and ObC. these Fsupranormal_ levels may be due to the robust surge of neurogenesis after AraC treatment, producing chains of type A cells more dense than in saline controls (Doetsch et al., 1999a,b; Discussion Doetsch and Alvarez-Buylla, 1996). We applied GO analysis to the genes regulated during SVZ We used Affymetrix GeneChips in three different approaches to regeneration. Similar to the SO profile, terms related to mRNA identify gene sets associated with in vivo SVZ neurogenesis. We splicing were the most significant (Fig. 5C). GO terms related to first obtained the gene expression profiles of five adult mouse brain regulation of cell cycle, proliferation, enzyme regulation, and regions and filtered for genes that had increased expression in the chromosome organization, and chromatin/nucleosome structure germinal SVZ and/or Ob target of neuronal differentiation. GO were also significant (Fig. 5C, and Supplementary data S9 contains analysis identified RNA splicing and chromatin remodeling as a list of all GO terms for SVZ regeneration). Of the 59 probe sets in prominent biological processes in the neurogenic SVZ and Ob this analysis, 16 (29%) were also found in the SVZ or SO profiles brain regions. Using FACS and cDNA amplification, we then (Table 4). The probability of having such an intersection at random compared the expression profiles of two SVZ cell populations is approximately 10À 50, with the expected number of genes in the important for neurogenesis: the SVZ astrocytes which function as random intersection being 0.7. Of these 16 genes, 4 had increased the stem cells (Doetsch et al., 1999a,b), and the ependymal cells expression in the FACS GFAP+ population (Table 4); the which contribute to the creation of a neurogenic niche (reviewed in probability of this intersection by chance is smaller than 10À 10. Goldman, 2003; Alvarez-Buylla and Lim, 2004); SVZ astrocytes were significant for the processes of cell proliferation, neuro- In situ hybridization (ISH) validates gene expression data genesis, and chromatin remodeling. For a more dynamic portrait of SVZ neurogenesis, we analyzed the transcriptional profiles during The SVZ, SO, and ObC expression profiles suggested genes SVZ regeneration which proceeds sequentially from B to C to A that may be important for SVZ-Ob neurogenesis. Because these cells (Doetsch et al., 1999a,b); GO analysis of the SVZ D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 139

Fig. 5. Transcriptional profile of SVZ regeneration after AraC treatment. (A) Schematic of AraC infusion and associated changes in SVZ cellular composition after AraC pump removal. At 1 day, only ependyma (gray) and type B cells (blue) remain. At 3 days, type C (green) cells return. At 10 days, all SVZ cell types including type A cells (red) have been regenerated. (B) Transcriptional profile of SVZ regeneration. The columns labeled A1, A3, and A10 represent the timepoints after AraC infusion. Columns S1 and S10 are the timepoints after control saline infusion. The SVZ column is the gene expression of unmanipulated controls. Genes are ordered along the vertical axis using hierarchical clustering. The color and color intensity of each cell in the matrix relate to the expression ratio of each gene. Red indicates a positive ratio (expression greater than the mean of the other brain regions), green indicates a negative ratio, and black indicates a ratio of 1. A color scale (log2) indicating the magnitude of the expression ratios is at the bottom of the panel. (C) GO analysis pie chart for SVZ regeneration. The entire pie represents all GO terms in the analysis. Pie slices are proportional to the number of genes (in parentheses) related to a particular GO Fparent_ term (legend for color code is in the inset to the right of each panel). GO terms that are Fchildren_ of a parent term are listed next to the pie chart with an indicating line. Further parent–child relationship of the GO tree structure is indicated by indentation with hyphen. All listed GO terms are statistically significant, and color of the type indicates the GO category (see legend at the lower right of the figure). regeneration data also found RNA splicing and chromosome and the transcriptional profiles from all three approaches were organization as prominent biological processes. unified by GO analysis, which gave us an overview of the These three approaches have distinct advantages and dis- biological processes involved. advantages. The brain region comparison yields the cleanest Supporting our experimental approaches, we found that some of expression data, but it represents the average expression profile of our expression data matched previously known regional and cell- entire regions and may reveal components beyond those related specific expression patterns, and Northern blot analysis and ISH to neurogenesis. The cell-type comparison is a more direct validated other data. A large number of genes identified in this study analysis of the neurogenic transcriptional profile, but the extra have not been previously described to be present in the SVZ or Ob amplification required for chip hybridization results in noisier and are available in the Supplementary data. In the Results section, data. The regeneration analysis is a fairly direct test for genes that we presented data mostly for the RNA splicing and chromatin are dynamically regulated during neurogenesis, yet the invasive- remodeling genes, however, taken together, the data appeared to fit ness of the procedure complicates analysis. Because the into a biological ‘‘story’’ of SVZ neurogenesis, progressing through expression data derived from these three approaches differ in cell cycle, transcriptional regulation, RNA processing, migration, quality and nature, we analyzed the GeneChip array data of the and apoptosis (see Fig. 7 and Supplementary text). three experiments separately. For the brain region and cell- Recent progress in the description of stem cell gene expression specific transcriptional profile analyses, we used the t test to has been made by comparing gene profiles of embryonic, determine differential gene expression; for the SVZ regeneration hematopoietic, and neural stem cells grown as neurospheres experiment, we used PCA to separate the gene expression due to (Ivanova et al., 2002; Ramalho-Santos et al., 2002). These analyses SVZ regeneration from that of surgery and saline infusion (see identified sets of genes that may be important for basic stem cell Experimental methods, Data analysis for details of these properties such as self-renewal; however, the process of neuro- methods). Each experimental approach provided us with a genesis was not specifically addressed. Prior gene expression studies different view of the transcriptional profile for SVZ neurogenesis, of neurogenesis have been performed with neurospheres in vitro. 140 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Table 4 Intersection with SVZ regeneration data

Highlighted cells indicate the profile to which each probe set/gene belongs (e.g., Ccnd2 has its cell in SVZ column highlighted, indicating the SVZ profile).

Neurospheres are spherical clusters of cells propagated in vitro from non-neurogenic ependyma. It is possible that the GFAP+ cells in the single cells by addition of EGF and/or FGF. Neurospheres can SVZ are intrinsically different from GFAP+ astrocytes in non- generate neurons, astrocytes, and oligodendrocytes (Reynolds and germinal regions. It will be interesting to compare the SVZ GFAP+ Weiss, 1992; Morshead et al., 1994; Gritti et al., 1996; Kukekov et transcriptional profile to those of astrocytes without stem cell al., 1999; Caldwell et al., 2001). For the transcriptional profile properties; the differences revealed by such an analysis may reveal studies, neurospheres were obtained from embryonic and early the molecular basis of the stem cell properties unique to SVZ postnatal cortex (not SVZ) (Geschwind et al., 2001; Easterday et al., astrocytes. There is very little information about the gene expression 2003; Karsten et al., 2003), embryonic striatum (contains SVZ) of ependymal cells. These important epithelial cells are born in the (Zhou et al., 2001; Wen et al., 2002), or postnatal SVZ (Gurok et al., embryo (Spassky et al., 2005) and play essential roles in brain 2004); the adult SVZ differs in gene expression and cellular cerebrospinal fluid circulation and homeostasis. Ependymal cell also composition from that of embryonic and postnatal SVZ as well as contribute to the neurogenic niche (Lim et al., 2000; Goldman, 2003; developing cortex (Tramontin et al., 2003). Also, the high levels of Peretto et al., 2004). Our transcriptional profile of the CD24+ cells exogenous growth factors (EGF or FGF) used to propagate provides a gene expression database for ependymal cells and should neurospheres deregulates normal gene expression (Gabay et al., serve as an important resource for further molecular analysis of these 2003; Hack et al., 2004), likely leading to significant alterations in cells (see Supplementary text). The gene expression profile of their transcriptional profiles. Notwithstanding these differences, isolated type A cells has also been studied (Pennartz et al., 2004); there were genes and biological processes overlapping between our therefore, to date, transcriptional profiles of type B, ependymal, and in vivo analysis and the in vitro neurosphere studies: certain cell type A cells are available, and together they should assist cycle genes (Ccnd2, Mcm3, Mcm7, S100a6, Mdk, Pcna, Gadd45b), investigators in the formation of hypotheses about gene function cytoskeletal/migration genes (Tubb3, Tagln, Racgap1), Hmgb2, in the SVZ. Fyn, and Rbp1 were common to our analysis and one or more of the neurosphere gene expression studies (Geschwind et al., 2001; RNA splicing in SVZ neurogenesis Easterday et al., 2003; Karsten et al., 2003; Gurok et al., 2004). In addition to identifying these genes, our study provided spatial (brain It has been proposed that RNA splicing is vital for region and SVZ cell type) and/or temporal (during regeneration) generating the complexity of the nervous system (Grabowski expression information. The raw data sets and complete gene lists and Black, 2001; Black and Grabowski, 2003). Alternative are available in the Supplementary data, allowing further analysis of splicing of the same gene can induce dramatic changes in neural the similarities and differences between mouse in vitro neurospheres developmental; for instance, distinct splice isoforms of Numb and in vivo SVZ neurogenesis. Such analyses along with compar- direct either proliferation or differentiation (Verdi et al., 1999). isons to human neurosphere transcriptional profiles (Wright et al., RNA splicing can regulate cell fate, transcription factor activity, 2003) may allow us to narrow down the list of genes that may be axon guidance, neurotransmitter receptor and ion channel important for neural stem cell function. function, and apoptosis; because all of these processes occur The GFAP+ and CD24+ transcriptional profiles allowed us to in the SVZ throughout adult life, the SVZ may be an ideal assign a subset of genes to either the neurogenic type B cells or the system in which to study RNA splicing function in neural D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 141

Fig. 6. In situ hybridization (ISH) validates transcriptional profile expression data. ISH was performed for Dlx2 (A, B), Dlx5 (C, D), Meis2 (E, F), Sfrs2 (G, H), Sf3b1 (I, J), Lsm4 (K, L), Khrdbs1/Sam68 (M, N), Mll (O, P), and Smarcad1 (R, S) on coronal adult brain sections. The dotted line in panel A shows the boundary between the corpus callosum (CC) and the Ctx, and the SVZ is indicated by arrows. The ventricle is to the left. Scale bars = 100 Am (A, C, E, G, I, K, M, O, R), 500 Am (B, D, F, H, J, L, N, P, S). development. In this study, we identified 11 genes for RNA identified in the regeneration experiment: Brunol4, Prpf8, and splicing that may be important for adult SVZ neurogenesis. The Hnrpab (Supplementary data S8). SO profile contained Sf3b1 (splicing factor 3b, subunit 1), Sfrs2 Sf3b1, Sfrs2, Prpf8, Lsm4, Snrpg, Hnrpa2b1, Hnrpm, Hnrph1, (splicing factor, arginine/serine-rich 2, SC35), Lsm4 (LSM4 Hnrpd, and Hnrpab are all components of the spliceosome complex homologue, U6 small nuclear RNA associated), Snrpg (small (reviewed in Jurica and Moore, 2003). The activity and specificity of nuclear ribonucleoprotein, polypeptide G), Khdrbs1/Sam68 (KH the spliceosome are regulated; for instance, changes in levels of domain containing, RNA binding, signal transduction associated Hnrpab mediate mRNA splice site selection in developing 1), and four members of the heterogeneous nuclear ribonucleo- erythroblasts (Hou et al., 2002). The heterogeneous nuclear protein family—Hnrpa2b1, Hnrpm, Hnrph1, and Hnrpd. The ribonucleoprotein (Hnrp) family members (e.g., Hnrpab) them- analysis of SVZ regeneration also recognized Sf3b1, Hnrpd, and selves are regulated by methylation at arginine (reviewed in Lsm4; additionally, three other genes for RNA splicing were McBride and Silver, 2001), and the arginine methyltransferase 142 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Hmrt1l2 (Scott et al., 1998) was in the SO profile, suggesting its profile. BMI1 physically interacts with and is antagonized by interaction with the Hnrps. Brunol4 belongs to the bruno/elav MLL (Hanson et al., 1999; Xia et al., 2003). family of RNA binding proteins that regulate mRNA processing Mll establishes and maintains specific gene expression patterns (Good et al., 2000); the human homologue of Brunol4 promotes through serial mitotic cell cycles (Yu et al., 1998; Milne et al., specific exon exclusion in developing muscle (Ladd et al., 2001). 2002). The increased expression of Mll in the B cell population and Perhaps most intriguingly, Khdrbs1/Sam68 is a prototype splice presence in the SO profile (Table 2) suggests that Mll expression site regulator whose activity is modified by extracellular signal- begins in B cells and continues through the lineage to type A cells. regulated kinase (ERK) transduction (Matter et al., 2002); as such, Mll therefore potentially regulates global developmental transcrip- Khdrbs1/Sam68 may link the SVZ precursor RNA splicing tional patterns throughout the entire SVZ neurogenic lineage. Mll machinery to changes in the extracellular environment. Khdrbs1/ regulates Dlx1, Dlx2, and Dlx5 (Ferrari et al., 2003), transcription Sam68, like the Hnrp family members, is also regulated by arginine factors in the SO profile, and MLL fusion proteins regulate Pbx3 methylation (Bedford et al., 2000). Fyn is a kinase found in the ObC and Meis1 (ObC profile) (Zeisig et al., 2004). Additionally, using profile, and FYN phosphorylation of KHDRBS1/SAM68 changes transcriptional profile analysis, Schraets et al. identified potential its subcellular localization, interaction with the spliceosome gene targets of Mll regulation (Schraets et al., 2003), and among components, and splice site selection (Hartmann et al., 1999); the the top candidates are Col6a (SO profile), Fhl1 (Four-and-a-half increased expression of Fyn in the ObC could induce Khdrbs1/ LIM domains 1, ObC profile), Nestin (neural precursor cell marker Sam68 to change mRNA splicing regulation in type A cells leading expressed in SVZ (Gates et al., 1995; Doetsch et al., 1997)), and to their cell cycle exit, change to radial migration, and integration Tenascin-C (SVZ stem cell niche ECM component (Garcion et al., into local circuits. 2004)). Hence, we have not only identified Mll in the SVZ but also Neuroblasts born in the SVZ have different destinations in the 9 genes that Mll may regulate. Ob. Some end up in the granule cell layer, while others migrate H2afx (SVZ profile, regulated during regeneration) is a histone farther into the periglomerular layer. Granule cell and periglomer- H2A variant that is critical for chromatin remodeling and ular interneurons have different synaptic organization as well as inactivation of sex chromosomes in meiosis (Fernandez-Capetillo neurotransmitter phenotypes. If these two types of Ob interneurons et al., 2003). Methylation of histone arginine residues modifies are derived from the same SVZ neural stem cell (this is currently chromatin function (reviewed in Trievel, 2004), and the arginine unclear), it is possible that alternative splicing may be critical for methyltransferase Hmrt1l2 (Scott et al., 1998) was found in the SO determining the migratory path of the neuroblasts as well as the cell profile. One of the best characterized histone modifications is fate choice. Recently, a genome-wide analysis of alternative lysine acetylation (reviewed in Sterner and Berger, 2000), and splicing determined by the Nova splicing factor has indicated that Hat1 (histone acetyltransferase 1) was in the SVZ profile. In RNA splicing may play important roles in synapse formation, addition to modifying histones, Hat1 can acetylate high mobility axonogenesis, neurite morphogenesis, and neurogenesis (Ule et al., group proteins (HMGs), which were also present in our analysis. 2005). Eph/ephrin signaling plays a role in SVZ migration and Hmgb2 (SVZ profile) and Hmgb3 (SO profile, increased in type B proliferation (Conover et al., 2000), and alternative splice forms of cells) are members of the high-mobility group B (HMGB) family, certain Eph receptors can regulate cellular repulsion or adhesion which can activate or repress transcription by modifying DNA– (Holmberg et al., 2000). Hence, alternative splicing of the same histone complexes (Ge and Roeder, 1994; Shykind et al., 1995; sets of transcripts could account for the generation of different Thomas, 2001). Hmgb2 wasalsoidentifiedinneurospheres destinations and phenotypes of SVZ-born neuroblasts. (Karsten et al., 2003; Gurok et al., 2004). In primitive blood cell precursors, enforced expression of Hmgb3 inhibits B cell and Chromatin remodeling in SVZ neurogenesis myeloid lineages (Nemeth et al., 2003), and Hmgb3-deficient mice have dysregulated lymphoid and myeloid cell development Chromatin remodeling can engage or maintain particular (Nemeth et al., 2004). genetic ‘‘programs’’ and therefore likely plays a critical role in SWI/SNF chromatin modifiers also regulate transcription. both stem cell maintenance as well as daughter cell differenti- Smarcad1 (ObC profile) is a SWI/SNF component, and ation (reviewed in Rasmussen, 2003; Cerny and Quesenberry, Smarcad1-deficient mice have impaired fertility, skeletal dyspla- 2004; Ehrenhofer-Murray, 2004). There also is increasing sias, and growth retardation (Schoor et al., 1999). Arp (actin- evidence that chromatin remodeling is important for neural related protein) family members regulate SWI/SNF complexes development (reviewed in Hsieh and Gage, 2004). Bmi1,a (reviewed in Olave et al., 2002), and Baf53a (ArpNa) was member of the Polycomb group of chromatin modifiers, is identified in the SO profile. Intriguingly, Baf53a is brain specific important for self-renewal of embryonic and postnatal SVZ stem and expressed in developing neurons in vitro (Kuroda et al., cell regulation (Molofsky et al., 2003); in the adult SVZ, we 2002). Among the 216 ‘‘stemness’’ genes common to brain, identified Bmi1 in the ObC profile. Polycomb group members blood, and embryonic stem cells are two members of the SWI/ such as Bmi1 work in concert with trithorax group proteins to SNF family of chromatin modifiers (Ramalho-Santos et al., regulate chromatin structure (Orlando, 2003); appropriately, Mll, 2002), further suggesting the importance of chromatin modifica- a member of the trithorax family, was expressed in the SO tion for stem cell regulation.

Fig. 7. Schematic of genes, biological processes, and gene interactions for SVZ neurogenesis. Data from the SVZ, SO, ObC profiles, the FACS data, and the SVZ regeneration analysis are integrated. This figure highlights 89 genes selected from the data; these genes are discussed in the Results section and Supplementary text. Genes in the SVZ, SO, and ObC profiles are arranged over a yellow background in vertical columns. Genes increased in GFAP+ and CD24+ cells are boldfaced in blue and black, respectively. Genes regulated during SVZ regeneration are circled. Known physical and genetic interactions are indicated by dotted lines and red arrows, respectively. See the legend at the lower right. D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 143 144 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Concluding remarks transcriptase protocols. Using the cDNA as template, 20 cycles of LD-PCR (BD Biosciences, Clontech) were performed. ds Any attempt to understand adult neurogenesis at the molecular T7cDNA from the LD-PCR reactions were phenol/CHCl3 level needs to take into consideration large sets of genes acting in extracted and spun through a Chromospin400 column (BD parallel. This study provides data on genes that contribute to adult Biosciences, Clontech). An aliquot of the ds T7cDNA was used neurogenesis. The data hint to the groups of genes involved in as template for a second LD-PCR, and aliquots were removed at proliferation, migration, and differentiation and reveal chromatin 6, 8, 10, and 12 cycles; these were analyzed on agarose gels by remodeling and RNA splicing as important components of these ethidium bromide staining and GAPDH Southern blot to processes. This in vivo molecular description of SVZ neurogenesis determine the linear range of amplification. provides the launching point of future studies into the regulation of this adult germinal zone. The challenge now is to understand the Analysis of regenerating SVZ contribution of individual genes in the context of the complexity revealed by this study. 2% AraC in vehicle (saline 0.9%) or vehicle alone were infused onto the surface of 2- to 3-month-old CD-1 mice for 6 days by mini-osmotic pump (Alzet, Palo Alto, CA, Model 1007D) as Experimental methods described (Doetsch et al., 1999a,b). At the end of infusion, osmotic pumps were surgically removed from their suprascapular place- Production of ds T7 cDNA from adult brain regions ment; cannulas were left in place until after animals were sacrificed. Only the SVZ from the side of cannula placement Adult (2–3 months) CD-1 (Charles River Laboratories) mouse (right side) was dissected. A total of 18 mice were used for this brains were used for RNA isolation. SVZ was dissected as experiment: 4 for the no-surgery control, 4 for A1, 3 for A3, 3 for previously described (Lim and Alvarez-Buylla, 1999), and Ctx A10, 2 for S1, and 2 for S10. Total RNA from dissected SVZ tissue and St were obtained from the same coronal slice. ObC was was isolated as described above. 3–12 Ag of total RNA from dissected from serial coronal slices of the Ob. Hp was isolated by pooled SVZ tissue for each time point/condition was converted to cutting the fimbria and blunt dissection. 10 mice were used for ds T7 cDNA using the above protocols, and equal amounts of each of the 2 experimental replicates. Dissected tissues were snap biotin-labeled cRNA were used for GeneChip hybridizations. frozen in 1.5-ml tubes with liquid N2. Tissues were disrupted in RNeasy (Qiagen) lysis buffer with needle trituration and Qiash- GeneChip probe production and hybridizations redder columns (Qiagen). DNase treated total RNA was isolated with RNeasy mini-columns (Qiagen). PolyA RNA was then Biotin-labeled cRNAs were produced from the ds T7cDNA purified with magnetic oligo-dT beads (Dynal). For each brain libraries and hybridized to Mu11K chips according to standard region, 1 Ag of polyA RNA was converted to ds T7cDNA with the protocols (Affymetrix, Santa Clara, CA). Chips were scanned on T7LD3V primer using standard Superscript II reverse transcriptase a GeneArray scanner (Affymetrix). For each brain region, cRNAs and DNA polymerase protocols (Invitrogen). were prepared from independent ds cDNA libraries from different dissection sessions. Likewise, for each FACS population, cRNAs FACS isolation of type B and ependymal cells and ds cDNA were generated from independent ds cDNA libraries prepared production from different dissection sessions and FACS runs.

Adult SVZ cells were dissociated, cleared of dead cells and Northern and Southern blots and PCR analysis debris by 22% Percoll (Sigma) step gradient as previously described (Lim et al., 2000), and passed through a 40-Am nylon Northern and Southern blots were performed according to cell strainer (BD Biosciences). All immunostaining incubations standard protocols using ExpressHyb (BD Biosciences, Clontech) and washes were performed at 0–4-C with pre-chilled buffers. or ULTRAhyb (Ambion). Probes for hybridizations were produced Biotinylated mCD24 antibody (BD Biosciences, Pharmingen) by PCR cloning. All probes were sequenced to verify their identity. was used at 1:10 and rabbit GFAP antibody (DakoCytomation) Semi-quantitative PCR analysis for CD24 and GFAP was was used at 1:100. About 1 Â 106 SVZ cells were resuspended performed as previously described (Lim et al., 2000). in 100 Al PBS containing both primary antibodies, 0.1% Tween- 20 (Sigma), and 100–200 units of RNasin (Promega) and In situ hybridization incubated for 15 min on ice. Cells were pelleted by gentle centrifugation and washed in PBS three times. Cells were then For in situ hybridization (ISH) on cryosections, we used a resuspended in 100 Al of PBS containing streptavidin-Cy2 at modification of methods previously described (Wilkinson, 1999). 1:100 and anti-rabbit F(ab)2 at 1:25 (Jackson Immunoresearch), After perfusion–fixation of 2- to 3-month old mice, brain tissues 0.1% Tween-20, and 100–200 units of RNasin and incubated were fixed with 4% PFA, cryoprotected by 10% and 20% sucrose for 10 min on ice. Cells were again washed 3 times with PBS. PBS, embedded in OCT compound (Sakura Finetechnical Co. Omission of primary antibodies resulted in no staining. Ltd., Tokyo, Japan), frozen, and sectioned at 18 Am thickness. Immunostained cells were isolated with the FACS Vantage After ISH staining, the sections were counterstained by nuclear (BD Biosciences). For each of the 2 experimental replicates, fast red. The following mouse cDNA was used for making 10,000 cells (from the SVZ of 25 mice) were collected directly digoxigenin labeled probes: Meis2 (GenBank accession num- into RNeasy lysis buffer, and DNAse-treated RNA was isolated ber:BF472214), Dlx2 (GenBank:NM_010054,nt746-1355), Dlx5 with RNeasy columns. RNA was then converted to cDNA with (GenBank:AW046057), Mll1 (GenBank:BC044818), Smarcad1 the T7LD3V primer using standard Superscript II reverse (GenBank: BC042442), Sf3b1 (GenBank:BC037098), Sfrs2 (Gen- D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 145

Bank:BC005493), Sam68 (GenBank:BC002051) and Lsm4 (Gen- PCA, a widely used data mining technique (see, e.g., Jolliffe, Bank:BC026747). Dlx2 cDNA was cloned by RT-PCR, and others 2003), creates new independent variables (the principal compo- were obtained as EST clones. nents) as those linear combinations of the original variables that capture as much of the variability of the original system as possible. Data analysis In other words, PCA models a cloud of points in high dimensional space by finding the direction along which the cloud has the largest Data analysis was performed with the R packages available at spread (the first component), the perpendicular direction with the the Bioconductor project site (www.bioconductor.org). We used second largest spread (the second component), and so on. We found the GCRMA algorithm to obtain expression measures from the that the first three principal components were enough to explain fluorescent intensities of the individual probes. This algorithm almost 90% of the variability among chips, thereby reducing our 6- employs a statistical model that uses probe sequence information dimension space to a 3-dimension space. In this new space, the first for background adjustment (Naef and Magnasco, 2003; Wu and component was basically the overall expression of the genes. The Irizarry, in press) which proved to be more sensitive than other second component described the ‘‘recovery’’from surgery and saline preprocessing methods (see http://affycomp.biostat.jhsph.edu) in- infusion, while the third component captured the gene expression cluding the GeneChip software (MAS 5). The normalization step due to the regeneration of the SVZ cellular population (see utilizes a quantile normalization algorithm (Bolstad et al., 2003; Supplementary data S10). We emphasize that these 3 new variables Bolstad, 2004) and probe sets were summarized using medianpol- are independent in the population considered, and so the recovery ish (Bolstad et al., 2003; Irizarry et al., 2003). from saline infusion and the SVZ regeneration are now independent To identify genes of the SVZ, SO, and ObC profile, we variables. The 229 probe sets were then listed by magnitude of the first filtered the data to exclude genes with low variability third component so that those genes at the top represent those most across all brain region samples (standard deviation smaller than related to SVZ regeneration and not the effect of saline or surgery. 0.15). Then the t test was used to determine the set of genes The expression array data for the top 25% of this list (59 probe sets, differentially expressed in the region under question as 56 unique genes, Supplementary data S8) was then clustered and is compared to the other brain regions. P values were adjusted shown in Fig. 4. for multiple hypothesis test as suggested in Benjamini and Clustering analysis was done using Gene Cluster 3.0 software Hochberg (2001) and Dudoit and Shaffer (2003), using the and Tree View 1.6 (Eisen et al., 1998) available at http://rana.lbl.gov/ Benjamin and Hochbert procedure; the permutations procedure EisenSoftware.htm. We used hierarchical clustering with Complete was not used. We then filtered for genes with statistical Average Linkage method and Euclidean distance as similarity significance ( P < 0.05) and with difference greater than 0.5 matrix for the SVZ, SO, and ObC profile data, and with the Pearson (i.e., more than 1.42-fold change) to obtain 71, 80, and 209 Correlation coefficient for the AraC data. filtered probe sets in the SVZ, SO and ObC profiles Analysis of GO annotations was done using the Onto-Express respectively (65, 60, 168 UniGene identifiers). Similar proce- (Khatri et al., 2002; Draghici et al., 2003; Khatri et al., 2004)a dures were carried out in other comparisons: t tests were web-based tool available at http://vortex.cs.wayne.edu/Proj- applied to FACS data (to determine those genes that are ects.html. To find those GO terms that were over-represented in differentially expressed between CD24+ and GFAP+ cells, P < the transcriptional profile in question (e.g., the SVZ, SO, or ObC 0.05) and to determine those genes expressed higher in the profiles), we compared the list of genes in the profile with the SVZ as compared to the St (to provide the list of genes that entire set of genes in Mu11K A and B chips. Significance was was used to as a filter for the AraC data, see below). assessed by using the hypergeometric distribution, and P values The gene expression analysis of SVZ regeneration is confound- were corrected for multiple hypothesis, controlling fdr (false ed by the changes induced by the surgery. Some of these effects discovery rate). Only nodes (in the ontology tree) with fdr <0.1 may be adequately controlled by comparison with the saline and >1 gene were considered. Supplementary data S11 contains control groups; however, the response to surgical lesions is variable the probe set identifiers for the SVZ, SO, ObC, Ctx, St, Hp, from animal to animal and may differ between saline and AraC GFAP+, CD24+, and SVZ regeneration profiles as well as the treated animals. For this reason, we pooled the RNA from the SVZ background Mu11kA and B chips; these probe set lists can be for each of the different time points (see Analysis of Regenerating used with the Onto-Express tool, allowing one to browse through SVZ, above). Since this pooled RNA was analyzed on a single chip the GO terms organized in the tree structure. set, we used Principal Component Analysis (PCA) after filtering the data. To filter the data, we considered the differences between untreated SVZ and all other samples (three AraC and two saline Primer sequences time points) within this experiment; we selected genes that show V differences in at least one comparison; the threshold of the t test T7LD3 :ATTCTAGAGGCCGAGGCGGCCGACATG- was based on the distribution of the differences for all genes, rather TAATACGACTCACTATAGGGCGTTTTTTTTTTTTTTTTT- than on a gene-by-gene basis. This set of 1764 probe sets was TTTTTTTTTTTTTVN (V = A,G,C; N = A,G,C,T) filtered with the list of genes that are increased in the SVZ as SMART III: AAGCAGTGGTATCAACGCAGAGTGGCCAT- compared with St ( P < 0.05, in the brain region analysis) to TATGGCCGGG eliminate from analysis those genes that normally are expressed at 5VPCR: AAGCAGTGGTATCAACGCAGAGTGGCCAT- high levels in the striatum. In order to separate the gene expression TATGG changes of SVZ regeneration from that of surgery and infusion of 3VPCR: ATTCTAGAGGCCGAGGCGGCCGACATGTAA- saline vehicle, the 229 probe sets at the intersection of these two TACGACTCACTATAGGGCG lists were analyzed by applying PCA to the expression matrix of Gapdh: CCCACTAACATCAAATGGGG, CTCACTTGT- the 229 probe sets and the 6 chips. GGCCCAGGTAT 146 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Dlx1: TCCTGAATGGTCTTCTTCCG, CTGGGGTGGTAC- Capela, A., Temple, S., 2002. LeX/ssea-1 is expressed by adult mouse CNS GAAGATGG stem cells, identifying them as nonependymal. Neuron 35, 865–875. Cerny, J., Quesenberry, P.J., 2004. Chromatin remodeling and stem cell 2310021601Rik: AGATGATAGCTGAGCAGCGG, CTGG- theory of relativity. J. Cell. Physiol. 201, 1–16. CAGAGAGGTTCAAAGC Chiasson, B.J., Tropepe, V., Morshead, C.M., Van der Kooy, D., 1999. Sox11: CAGGCACTTCTTCCCTTTTG, CAGCTCT- Adult mammalian forebrain ependymal and subependymal cells GAGGTCTATGTCACC demonstrate proliferative potential, but only subependymal cells have Col6a1: CCCCATTGGACCTAAAGGAT, CAGCACGAA- neural stem cell characteristics. J. Neurosci. 19, 4462–4471. GAGGATGTCAA Conover, J.C., Doetsch, F., Garcia-Verdugo, J.M., Gale, N.W., Yancopou- Ccnd2: CCTCACGACTTCATTGAGCA, ATGCTGCTCTT- los, G.D., Alvarez-Buylla, A., 2000. Disruption of Eph/ephrin signaling affects migration and proliferation in the adult subventricular zone. Nat. GACGGAACT Neurosci. 3, 1091–1097. Hmgb2: AGCTTGGGGAAGGAAGTCTC, AGCAAAACAG- Doetsch, F., Alvarez-Buylla, A., 1996. Network of tangential pathways for GAAGAAGGCA neuronal migration in adult mammalian brain. Proc. Natl. Acad. Sci. U. Mia: AGCCCAGAGACCTCGTTCTT, ATCAATTTTGC- S. A. 93, 14895–14900. CAGGTTTCG Doetsch, F., Garcia-Verdugo, J.M., Alvarez-Buylla, A., 1997. Cellular Pdyn: GATCAGGTAGGGCATGAGGA, TTCTCT- composition and three-dimensional organization of the subventric- GGATTCTGGGATGG ular germinal zone in the adult mammalian brain. J. Neurosci. 17, 5046–5061. Gfap: CTCAATGCTGGCTTCAAGGAGA, GACG- Doetsch, F., Caille, I., Lim, D.A., Garcia-Verdugo, J.M., Alvarez-Buylla, CAGCGTCTGTGAGGTC A., 1999a. Subventricular zone astrocytes are neural stem cells in the Cd24: ATGCAAAGGAGCCAAAACTG, GTGACCATGC- adult mammalian brain. Cell 97, 703–716. GAACAAAAGA Doetsch, F., Garcia-Verdugo, J.M., Alvarez-Buylla, A., 1999b. Regenera- tion of a germinal layer in the adult mammalian brain. Proc. Natl. Acad. Sci. U. S. A. 96, 11619–11624. Acknowledgments Doetsch, F., Petreanu, L., Caille, I., Garcia-Verdugo, J.M., Alvarez-Buylla, A., 2002. EGF converts transit-amplifying neurogenic precursors in the adult brain into multipotent stem cells. Neuron 36, 1021–1034. We thank Miguel Ramalho-Santos for the many helpful dis- Draghici, S., Khatri, P., Bhavsar, P., Shah, A., Krawetz, S.A., Tainsky, cussions and editorial comments. H.T. was supported by the M.A., 2003. Onto-Tools, the toolkit of the modern biologist: Onto- Mochida Memorial Foundation for Medical and Pharmaceutical Express, Onto-Compare, Onto-Design and Onto-Translate. Nucleic Research. This work was supported by NIH grant NS28478-12 to Acids Res. 31, 3775–3781. AAB. Dudoit, S., Shaffer, J., 2003. Multiple hypothesis testing in microarray experiments. Stat. Sci. 18, 71–103. Easterday, M.C., Dougherty, J.D., Jackson, R.L., Ou, J., Nakano, I., Paucar, Appendix A. Supplementary data A.A., Roobini, B., Dianati, M., Irvin, D.K., Weissman, I.L., Terskikh, A.V., Geschwind, D.H., Kornblum, H.I., 2003. Neural progenitor genes. Supplementary data associated with this article can be found in Germinal zone expression and analysis of genetic overlap in stem cell populations. Dev. Biol. 264, 309–322. the online version at doi:10.1016/j.mcn.2005.10.005. Ehrenhofer-Murray, A.E., 2004. Chromatin dynamics at DNA replication, transcription and repair. Eur. J. Biochem. 271, 2335–2349. Eisen, M.B., Spellman, P.T., Brown, P.O., Botstein, D., 1998. Cluster References analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. U. S. A. 95, 14863–14868. Alvarez-Buylla, A., Lim, D.A., 2004. For the long run: maintaining Fernandez-Capetillo, O., Mahadevaiah, S.K., Celeste, A., Romanienko, P.J., germinal niches in the adult brain. Neuron 41, 683–686. Camerini-Otero, R.D., Bonner, W.M., Manova, K., Burgoyne, P., Bedford, M.T., Frankel, A., Yaffe, M.B., Clarke, S., Leder, P., Richard, Nussenzweig, A., 2003. H2AX is required for chromatin remodeling S., 2000. Arginine methylation inhibits the binding of proline-rich and inactivation of sex chromosomes in male mouse meiosis. Dev. Cell ligands to Src homology 3, but not WW, domains. J. Biol. Chem. 4, 497–508. 275, 16030–16036. Ferrari, N., Palmisano, G.L., Paleari, L., Basso, G., Mangioni, M., Fidanza, Benjamini, Y., Hochberg, Y.C., 2001. Controlling the false discovery rate: a V., Albini, A., Croce, C.M., Levi, G., Brigati, C., 2003. DLX genes as practical and powerful approach to multiple testing. J. R. Stat. Soc. 57, targets of ALL-1: DLX 2,3,4 down-regulation in t(4;11) acute 289–300. lymphoblastic leukemias. J. Leukocyte Biol. 74, 302–305. Black, D.L., Grabowski, P.J., 2003. Alternative pre-mRNA splicing and Ferri, A.L., Cavallaro, M., Braida, D., Di Cristofano, A., Canta, A., neuronal function. Prog. Mol. Subcell. Biol. 31, 187–216. Vezzani, A., Ottolenghi, S., Pandolfi, P.P., Sala, M., DeBiasi, S., Bolstad, B.M., 2004. Low level analysis of high-density oligonucleotide Nicolis, S.K., 2004. Sox2 deficiency causes neurodegeneration and array data: background, normalization and summarization. Biostatistics. impaired neurogenesis in the adult mouse brain. Development 131, University of California, Berkeley, pp. 156. 3805–3819. Bolstad, B.M., Irizarry, R.A., Astrand, M., Speed, T.P., 2003. A comparison Fiske, B.K., Brunjes, P.C., 2001. Cell death in the developing and sensory- of normalization methods for high density oligonucleotide array data deprived rat olfactory bulb. J. Comp. Neurol. 431, 311–319. based on variance and bias. Bioinformatics 19, 185–193. Gabay, L., Lowell, S., Rubin, L.L., Anderson, D.J., 2003. Deregulation of Calaora, V., Chazal, G., Nielsen, P.J., Rougon, G., Moreau, H., 1996. dorsoventral patterning by FGF confers trilineage differentiation mCD24 expression in the developing mouse brain and in zones of capacity on CNS stem cells in vitro. Neuron 40, 485–499. secondary neurogenesis in the adult. Neuroscience 73, 581–594. Gage, F.H., 2000. Mammalian neural stem cells. Science 287, 1433–1438. Caldwell, M.A., He, X., Wilkie, N., Pollack, S., Marshall, G., Wafford, K.A., Garcia-Verdugo, J.M., Doetsch, F., Wichterle, H., Lim, D.A., Alvarez- Svendsen, C.N., 2001. Growth factors regulate the survival and fate of Buylla, A., 1998. Architecture and cell types of the adult subventricular cells derived from human neurospheres. Nat. Biotechnol. 19, 475–479. zone: in search of the stem cells. J. Neurobiol. 36, 234–248. D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148 147

Garcion, E., Halilagic, A., Faissner, A., ffrench-Constant, C., 2004. Jankovski, A., Sotelo, C., 1996. Subventricular zone-olfactory bulb Generation of an environmental niche for neural stem cell development migratory pathway in the adult mouse: Cellular composition and by the extracellular matrix molecule tenascin C. Development 131, specificity as determined by heterochronic and heterotopic transplanta- 3423–3432. tion. J. Comp. Neurol. 371, 376–396. Gates, M.A., Thomas, L.B., Howard, E.M., Laywell, E.D., Sajin, B., Jolliffe, I., 2003. Principle Component Analysis. Springler-Verlag, New Faissner, A., Gotz, B., Silver, J., Steindler, D.A., 1995. Cell and York. molecular analysis of the developing and adult mouse subventricular Jurica, M.S., Moore, M.J., 2003. Pre-mRNA splicing: awash in a sea of zone of the cerebral hemispheres. J. Comp. Neurol. 361, 249–266. proteins. Mol. Cell 12, 5–14. Ge, H., Roeder, R.G., 1994. The high mobility group protein HMG1 can Karsten, S.L., Kudo, L.C., Jackson, R., Sabatti, C., Kornblum, H.I., reversibly inhibit class II gene transcription by interaction with the Geschwind, D.H., 2003. Global analysis of gene expression in neural TATA-binding protein. J. Biol. Chem. 269, 17136–17140. progenitors reveals specific cell-cycle, signaling, and metabolic net- Geschwind, D.H., Ou, J., Easterday, M.C., Dougherty, J.D., Jackson, R.L., works. Dev. Biol. 261, 165–182. Chen, Z., Antoine, H., Terskikh, A., Weissman, I.L., Nelson, S.F., Khatri, P., Draghici, S., Ostermeier, G.C., Krawetz, S.A., 2002. Profiling Kornblum, H.I., 2001. A genetic analysis of neural progenitor gene expression using onto-express. Genomics 79, 266–270. differentiation. Neuron 29, 325–339. Khatri, P., Bhavsar, P., Bawa, G., Draghici, S., 2004. Onto-Tools: an Goldman, S., 2003. Glia as neural progenitor cells. Trends Neurosci. 26, ensemble of web-accessible, ontology-based tools for the functional 590–596. design and interpretation of high-throughput gene expression experi- Good, P.J., Chen, Q., Warner, S.J., Herring, D.C., 2000. A family of human ments. Nucleic. Acids Res. 32, W449–W456. RNA-binding proteins related to the Drosophila Bruno translational Kukekov, V.G., Laywell, E.D., Suslov, O., Davies, K., Scheffler, B., regulator. J. Biol. Chem. 275, 28583–28592. Thomas, L.B., O’Brien, T.F., Kusakabe, M., Steindler, D.A., 1999. Grabowski, P.J., Black, D.L., 2001. Alternative RNA splicing in the Multipotent stem/progenitor cells with similar properties arise from nervous system. Prog. Neurobiol. 65, 289–308. two neurogenic regions of adult human brain. Exp. Neurol. 156, Gritti, A., Parati, E.A., Cova, L., Frolichsthal, P., Galli, R., Wanke, E., 333–344. Faravelli, L., Morassutti, D.J., Roisen, F., Nickel, D.D., Vescovi, A.L., Kuroda, Y., Oma, Y., Nishimori, K., Ohta, T., Harata, M., 2002. Brain- 1996. Multipotential stem cells from the adult mouse brain proliferate specific expression of the nuclear actin-related protein ArpNalpha and and self-renew in response to basic fibroblast growth factor. J. Neurosci. its involvement in mammalian SWI/SNF chromatin remodeling 16, 1091–1100. complex. Biochem. Biophys. Res. Commun. 299, 328–334. Grizzi, F., Chiriva-Internati, M., Franceschini, B., Bumm, K., Colombo, P., Ladd, A.N., Charlet, N., Cooper, T.A., 2001. The CELF family of RNA Ciccarelli, M., Donetti, E., Gagliano, N., Hermonat, P.L., Bright, R.K., binding proteins is implicated in cell-specific and developmentally Gioia, M., Dioguardi, N., Kast, W.M., 2004. Sperm protein 17 is regulated alternative splicing. Mol. Cell. Biol. 21, 1285–1296. expressed in human somatic ciliated epithelia. J. Histochem. Cytochem. Laywell, E.D., Rakic, P., Kukekov, V.G., Holland, E.C., Steindler, D.A., 52, 549–554. 2000. Identification of a multipotent astrocytic stem cell in the Gurok, U., Steinhoff, C., Lipkowitz, B., Ropers, H.H., Scharff, C., Nuber, immature and adult mouse brain. Proc. Natl. Acad. Sci. U. S. A. 97, U.A., 2004. Gene expression changes in the course of neural progenitor 13883–13888. cell differentiation. J. Neurosci. 24, 5982–6002. Lemkine, G.F., Raji, A., Alfama, G., Turque, N., Hassani, Z., Alegria- Hack, M.A., Sugimori, M., Lundberg, C., Nakafuku, M., Gotz, M., 2004. Prevot, O., Samarut, J., Levi, G., Demeneix, B.A., 2005. Adult neural Regionalization and fate specification in neurospheres: the role of Olig2 stem cell cycling in vivo requires thyroid hormone and its alpha and Pax6. Mol. Cell. Neurosci. 25, 664–678. receptor. FASEB J. 19, 863–865. Hanson, R.D., Hess, J.L., Yu, B.D., Ernst, P., van Lohuizen, M., Berns, A., Lim, D.A., Alvarez-Buylla, A., 1999. Interaction between astrocytes and van der Lugt, N.M., Shashikant, C.S., Ruddle, F.H., Seto, M., adult subventricular zone precursors stimulates neurogenesis. Proc. Korsmeyer, S.J., 1999. Mammalian Trithorax and polycomb-group Natl. Acad. Sci. U. S. A. 96, 7526–7531. homologues are antagonistic regulators of homeotic development. Proc. Lim, D.A., Tramontin, A.D., Trevejo, J.M., Herrera, D.G., Garcia-Verdugo, Natl. Acad. Sci. U. S. A. 96, 14372–14377. J.M., Alvarez-Buylla, A., 2000. Noggin antagonizes BMP signaling to Hartmann, A.M., Nayler, O., Schwaiger, F.W., Obermeier, A., Stamm, S., create a niche for adult neurogenesis. Neuron 28, 713–726. 1999. The interaction and colocalization of Sam68 with the splicing- Lois, C., 1996. Long Distance Neuronal Migration in the Adult Mammalian associated factor YT521-B in nuclear dots is regulated by the Src family Brain. Rockefeller, New York, pp. 160. kinase p59(fyn). Mol. Biol. Cell 10, 3909–3926. Lois, C., Alvarez-Buylla, A., 1994. Long-distance neuronal migration in the Holmberg, J., Clarke, D.L., Frisen, J., 2000. Regulation of repulsion adult mammalian brain. Science 264, 1145–1148. versus adhesion by different splice forms of an Eph receptor. Nature Luskin, M.B., 1993. Restricted proliferation and migration of postnatally 408, 203–206. generated neurons derived from the forebrain subventricular zone. Hou, V.C., Lersch, R., Gee, S.L., Ponthier, J.L., Lo, A.J., Wu, M., Turck, Neuron 11, 173–189. C.W., Koury, M., Krainer, A.R., Mayeda, A., Conboy, J.G., 2002. Luskin, M.B., 1998. Neuroblasts of the postnatal mammalian forebrain: Decrease in hnRNP A/B expression during erythropoiesis mediates a their phenotype and fate. J. Neurobiol. 36, 221–233. pre-mRNA splicing switch. EMBO J. 21, 6195–6204. Matter, N., Herrlich, P., Konig, H., 2002. Signal-dependent regulation of Hsieh, J., Gage, F.H., 2004. Epigenetic control of neural stem cell fate. splicing via phosphorylation of Sam68. Nature 420, 691–695. Curr. Opin. Genet. Dev. 14, 461–469. McBride, A.E., Silver, P.A., 2001. State of the arg: protein methylation at Imura, T., Kornblum, H.I., Sofroniew, M.V., 2003. The predominant arginine comes of age. Cell 106, 5–8. neural stem cell isolated from postnatal and adult forebrain but Milne, T.A., Briggs, S.D., Brock, H.W., Martin, M.E., Gibbs, D., Allis, not early embryonic forebrain expresses GFAP. J. Neurosci. 23, C.D., Hess, J.L., 2002. MLL targets SET domain methyltransferase 2824–2832. activity to Hox gene promoters. Mol. Cell 10, 1107–1117. Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Molofsky, A.V., Pardal, R., Iwashita, T., Park, I.K., Clarke, M.F., Morrison, Scherf, U., Speed, T.P., 2003. Exploration, normalization, and summa- S.J., 2003. Bmi-1 dependence distinguishes neural stem cell self- ries of high density oligonucleotide array probe level data. Biostatistics renewal from progenitor proliferation. Nature 425, 962–967. 4, 249–264. Morshead, C.M., Reynolds, B.A., Craig, C.G., McBurney, M.W., Staines, Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A., W.A., Morassutti, D., Weiss, S., Van der Kooy, D., 1994. Neural stem Lemischka, I.R., 2002. A stem cell molecular signature. Science 298, cells in the adult mammalian forebrain: a relatively quiescent 601–604. subpopulation of subependymal cells. Neuron 13, 1071–1082. 148 D.A. Lim et al. / Mol. Cell. Neurosci. 31 (2006) 131–148

Naef, F., Magnasco, M.O., 2003. Solving the riddle of the bright Wnt, Fz, and Frp gene families expressed in postnatal mouse cerebral mismatches: labeling and effective binding in oligonucleotide arrays. cortex. J. Comp. Neurol. 473, 496–510. Phys. Rev., E Stat. Nonlinear Soft Matter Phys. 68, 011906. Shykind, B.M., Kim, J., Sharp, P.A., 1995. Activation of the TFIID-TFIIA Najbauer, J., Leon, M., 1995. Olfactory experience modulated apoptosis in complex with HMG-2. Genes Dev. 9, 1354–1365. the developing olfactory bulb. Brain Res. 674, 245–251. Spassky, N., Merkle, F.T., Flames, N., Tramontin, A.D., Garcia-Verdugo, Nemeth, M.J., Curtis, D.J., Kirby, M.R., Garrett-Beal, L.J., Seidel, N.E., J.M., Alvarez-Buylla, A., 2005. Adult ependymal cells are postmitotic Cline, A.P., Bodine, D.M., 2003. Hmgb3: an HMG-box family member and are derived from radial glial cells during embryogenesis. J. Neurosci. expressed in primitive hematopoietic cells that inhibits myeloid and B- 25, 10–18. cell differentiation. Blood 102, 1298–1306. Stenman, J., Toresson, H., Campbell, K., 2003. Identification of two distinct Nemeth, M.J., Cline, A.P., Anderson, S.M., Garrett-Beal, L.J., Bodine, progenitor populations in the lateral ganglionic eminence: implications D.M., 2005. Hmgb3 deficiency deregulates proliferation and differen- for striatal and olfactory bulb neurogenesis. J. Neurosci. 23, 167–174. tiation of common lymphoid and myeloid progenitors. Blood 105, Sterner, D.E., Berger, S.L., 2000. Acetylation of histones and transcription- 627–634. related factors. Microbiol. Mol. Biol. Rev. 64, 435–459. Olave, I.A., Reck-Peterson, S.L., Crabtree, G.R., 2002. Nuclear actin and Stump, G., Durrer, A., Klein, A.L., Lutolf, S., Suter, U., Taylor, V., 2002. actin-related proteins in chromatin remodeling. Annu. Rev. Biochem. Notch1 and its ligands Delta-like and Jagged are expressed and active in 71, 755–781. distinct cell populations in the postnatal mouse brain. Mech. Dev. 114, Orlando, V., 2003. Polycomb, epigenomes, and control of cell identity. Cell 153–159. 112, 599–606. Thomas, J.O., 2001. HMG1 and 2: architectural DNA-binding proteins. Parras, C.M., Galli, R., Britz, O., Soares, S., Galichet, C., Battiste, J., Biochem. Soc. Trans. 29, 395–401. Johnson, J.E., Nakafuku, M., Vescovi, A., Guillemot, F., 2004. Mash1 Thomas, L.B., Gates, M.A., Steindler, D.A., 1996. Young neurons from the specifies neurons and oligodendrocytes in the postnatal brain. EMBO J. adult subependymal zone proliferate and migrate along an astrocyte, 23, 4495–4505. extracellular matrix-rich pathway. Glia 17, 1–14. Pennartz, S., Belvindrah, R., Tomiuk, S., Zimmer, C., Hofmann, K., Tramontin, A.D., Garcia-Verdugo, J.M., Lim, D.A., Alvarez-Buylla, A., Conradt, M., Bosio, A., Cremer, H., 2004. Purification of neuronal 2003. Postnatal development of radial glia and the ventricular zone precursors from the adult mouse brain: comprehensive gene expression (VZ): a continuum of the neural stem cell compartment. Cereb. Cortex analysis provides new insights into the control of cell migration, 13, 580–587. differentiation, and homeostasis. Mol. Cell. Neurosci. 25, 692–706. Trievel, R.C., 2004. Structure and function of histone methyltransferases. Peretto, P., Merighi, A., Fasolo, A., Bonfanti, L., 1997. Glial tubes in the Crit. Rev. Eukaryotic Gene Expression 14, 147–169. rostral migratory stream of the adult rat. Brain Res. Bull. 42, 9–21. Ule, J., Ule, A., Spencer, J., Williams, A., Hu, J.S., Cline, M., Wang, H., Peretto, P., Dati, C., De Marchis, S., Kim, H.H., Ukhanova, M., Fasolo, A., Clark, T., Fraser, C., Ruggiu, M., Zeeberg, B.R., Kane, D., Weinstein, Margolis, F.L., 2004. Expression of the secreted factors noggin and J.N., Blume, J., Darnell, R.B., 2005. Nova regulates brain-specific bone morphogenetic proteins in the subependymal layer and olfactory splicing to shape the synapse. Nat. Genet. 37, 844–852. bulb of the adult mouse brain. Neuroscience 128, 685–696. Verdi, J.M., Bashirullah, A., Goldhawk, D.E., Kubu, C.J., Jamali, M., Petreanu, L., Alvarez-Buylla, A., 2002. Maturation and death of adult-born Meakin, S.O., Lipshitz, H.D., 1999. Distinct human NUMB isoforms olfactory bulb granule neurons: role of olfaction. J. Neurosci. 22, regulate differentiation vs. proliferation in the neuronal lineage. Proc. 6106–6113. Natl. Acad. Sci. U. S. A. 96, 10472–10476. Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C., Melton, Wen, T., Gu, P., Minning, T.A., Wu, Q., Liu, M., Chen, F., Liu, H., Huang, D.A., 2002. ‘‘Stemness’’: transcriptional profiling of embryonic and H., 2002. Microarray analysis of neural stem cell differentiation in the adult stem cells. Science 298, 597–600. striatum of the fetal rat. Cell. Mol. Neurobiol. 22, 407–416. Rasmussen, T.P., 2003. Embryonic stem cell differentiation: a chromatin Wilkinson, D.G. (Ed.), 1999. In Situ Hybridization: A Practical Approach. perspective. Reprod. Biol. Endocrinol. 1, 100. Oxford Univ. Press, Oxford. Redmond, L., Hockfield, S., Morabito, M.A., 1996. The divergent Wright, L.S., Li, J., Caldwell, M.A., Wallace, K., Johnson, J.A., Svendsen, homeobox gene PBX1 is expressed in the postnatal subven- C.N., 2003. Gene expression in human neural stem cells: effects of tricular zone and interneurons of the olfactory bulb. J. Neurosci. leukemia inhibitory factor. J. Neurochem. 86, 179–195. 16, 2972–2982. Wu, Z., Irizarry, R.A., Gentleman, R., Martinez-Murillo, F., Spencer, F., Reynolds, B., Weiss, S., 1992. Generation of neurons and astrocytes from 2004. A model based background adjustment for oligonucleotide isolated cells of the adult mammalian central nervous system. Science expression arrays. J. Am. Stat. Assoc. 99, 909–917. 255, 1707–1710. Xia, Z.B., Anderson, M., Diaz, M.O., Zeleznik-Le, N.J., 2003. MLL Rietze, R.L., Valcanis, H., Brooker, G.F., Thomas, T., Voss, A.K., Bartlett, repression domain interacts with histone deacetylases, the polycomb P.F., 2001. Purification of a pluripotent neural stem cell from the adult group proteins HPC2 and BMI-1, and the corepressor C-terminal- mouse brain. Nature 412, 736–739. binding protein. Proc. Natl. Acad. Sci. U. S. A. 100, 8342–8347. Santa-Olalla, J., Baizabal, J.M., Fregoso, M., del Carmen Cardenas, M., Yu, B.D., Hanson, R.D., Hess, J.L., Horning, S.E., Korsmeyer, S.J., 1998. Covarrubias, L., 2003. The in vivo positional identity gene expression MLL, a mammalian trithorax-group gene, functions as a transcriptional code is not preserved in neural stem cells grown in culture. Eur. J. maintenance factor in morphogenesis. Proc. Natl. Acad. Sci. U. S. A. Neurosci. 18, 1073–1084. 95, 10632–10636. Schoor, M., Schuster-Gossler, K., Roopenian, D., Gossler, A., 1999. Zeisig, B.B., Milne, T., Garcia-Cuellar, M.P., Schreiner, S., Martin, Skeletal dysplasias, growth retardation, reduced postnatal survival, M.E., Fuchs, U., Borkhardt, A., Chanda, S.K., Walker, J., Soden, and impaired fertility in mice lacking the SNF2/SWI2 family member R., Hess, J.L., Slany, R.K., 2004. Hoxa9 and Meis1 are key targets ETL1. Mech. Dev. 85, 73–83. for MLL-ENL-mediated cellular immortalization. Mol. Cell. Biol. Schraets, D., Lehmann, T., Dingermann, T., Marschalek, R., 2003. MLL- 24, 617–628. mediated transcriptional gene regulation investigated by gene expres- Zhou, F.C., Duguid, J.R., Edenberg, H.J., McClintick, J., Young, P., Nelson, sion profiling. Oncogene 22, 3655–3668. P., 2001. DNA microarray analysis of differential gene expression of 6- Scott, H.S., Antonarakis, S.E., Lalioti, M.D., Rossier, C., Silver, P.A., year-old rat neural striatal progenitor cells during early differentiation. Henry, M.F., 1998. Identification and characterization of two putative Restor. Neurol. Neurosci. 18, 95–104. human arginine methyltransferases (HRMT1L1 and HRMT1L2). Zhu, H., Wang, Z.Y., Hansson, H.A., 2003. Visualization of proliferating Genomics 48, 330–340. cells in the adult mammalian brain with the aid of ribonucleotide Shimogori, T., VanSant, J., Paik, E., Grove, E.A., 2004. Members of the reductase. Brain Res. 977, 180–189.