|Home | About | Journals | Submit | Contact Us | Français|
Neural stem cells (NSCs) continually produce new neurons in postnatal brains. However, the majority of these cells stay in a non-dividing, inactive state. The molecular mechanism that is required for these cells to enter proliferation still remains largely unknown. Here, we show that nuclear receptor TLX (NR2E1) controls the activation status of postnatal NSCs in mice. Lineage tracing indicates that TLX-expressing cells give rise to both activated and inactive postnatal NSCs. Surprisingly, loss of TLX function does not result in spontaneous glial differentiation, but rather leads to a precipitous age-dependent increase of inactive cells with marker expression and radial morphology for NSCs. These inactive cells are mis-positioned throughout the granular cell layer of the dentate gyrus during development and can proliferate again after reintroducing ectopic TLX. RNA-seq analysis of sorted NSCs revealed a TLX-dependent global expression signature, which includes the p53 signaling pathway. TLX regulates p21 expression in a p53-dependent manner and acute removal of p53 can rescue the proliferation defect of TLX-null NSCs in culture. Together, these findings suggest that TLX acts as an essential regulator that ensures the proliferative ability of postnatal NSCs by controlling their activation through genetic interaction with p53 and other signaling pathways.
Postnatal neural stem cells (NSCs) exist normally in the subgranular zone (SGZ) of the dentate gyrus (DG) and the subventricular zone (SVZ) of the lateral ventricles (LV) (Lois and Alvarez-Buylla, 1993; Kuhn et al., 1996; Lie et al., 2004; Zhao et al., 2008). These cells may play a critical role in certain forms of learning and memory and may significantly contribute to the maintenance of brain homeostasis (Imayoshi et al., 2008; Zhao et al., 2008). In the SGZ, type-1 NSCs have long radial glia-like processes spanning the entire granule cell layer. They express Nestin, GFAP, Sox2, and basic lipid binding protein (BLBP). While the majority of them remain in an inactive state, some of these NSCs slowly divide and give rise to transiently amplifying type-2 cells with short processes and tangential orientation. These Type-2 cells rapidly proliferate and generate Type-3 cells, which resemble immature neuroblasts and express doublecortin (DCX). They eventually mature into granule neurons, which functionally integrate into the existing neural networks. In the LV, glia-like, GFAP+Nestin+ NSCs (type B cells) are located adjacent to the ependyma, a thin layer of cells lining the ventricle. These slow-dividing type B cells give rise to transiently amplifying Dlx2+ type C cells, which produce type A (DCX+PSA-NCAM+) neuroblasts. Newly generated neuroblasts migrate into the olfactory bulbs and become granule or periglomerular interneurons. Neurogenesis in both the SVZ and the SGZ continues throughout the adult life but decreases dramatically with age (Seki and Arai, 1995; Kuhn et al., 1996; Tropepe et al., 1997).
Despite advances in understanding adult NSCs, it still remains unclear how their activity is molecularly regulated and what signals are responsible for the age-dependent decline in replication. Previously, we and others have identified that TLX (NR2E1) is expressed in the neurogenic niche and is required for adult neurogenesis in the SGZ and the SVZ (Shi et al., 2004; Liu et al., 2008; Zhang et al., 2008). Furthermore, TLX-dependent NSCs and neurogenesis play a role in spatial learning and memory (Zhang et al., 2008). TLX is a member of the nuclear hormone receptor superfamily and functions as a transcriptional repressor by recruiting corepressors (Yu et al., 1994; Monaghan et al., 1995; Wang et al., 2006; Zhang et al., 2006; Sun et al., 2007; Yokoyama et al., 2008). The function of TLX is largely thought to prevent precocious differentiation of NSCs into mature neurons or glial cells during development (Roy et al., 2004; Shi et al., 2004; Li et al., 2008). This notwithstanding, the role of TLX in NSCs is far from clear, since our detailed analysis using genetic tracers revealed that cells expressing markers for NSCs still exist in postnatal TIx-null brains. These NSCs are mis-positioned and rapidly lose their ability to proliferate in an age-dependent manner. Interestingly, these inert cells can be reactivated by reintroducing TLX. Furthermore, our RNA-seq analysis and in vitro cell culture studies showed a genetic interaction of TLX with the p53 signaling pathway. Together, these findings suggest that TLX function is essential in postnatal NSCs by controlling the switch from quiescence to activation.
pTlx-CreERT2 mice were generated through Recombineering Technology (http://web.ncifcrf.gov/research/brb/recombineeringInformation.aspx). Briefly, a tamoxifen-inducible CreERT2 gene (Feil et al., 1997) was inserted through homologous recombination into the first exon of the Tlx locus in a BAC clone (RP24-344A4). The correctly recombined BAC clones were confirmed by restriction digestion and sequencing. The genomic DNA was released by sequential digestion with BsiWI and AscI and separated from the vector backbone through a CL-4B sepharose column. Transgenic animals were then produced by pronuclear injection of fertilized mouse eggs by The Transgenic Core Facility at UT Southwestern. Twenty-three founders were identified after genotyping and were further screened for inducible expression of EYFP after crossing into Rosa-EYFP reporter mice and treatment with Tamoxifen. The pTlx-CreERT2 mice were kept in a mixed background of FVB, C57BL/6J and 129S1/SvImJ.
The strategies and methods for generating TlxLacZ/LacZ mice or mice with conditional alleles of Tlx have been described (Yu et al., 2000; Zhang et al., 2008). In short, TlxLacZ/LacZ mice were generated by replacing exons 3–5 with LacZ and Neo genes. Thus, expression of LacZ is under the direct regulation of the endogenous Tlx promoter and enhancers. To create a conditional allele of Tlx, exon 2 was flanked by two loxP sites through homologous recombination. Cre-mediated recombination resulted in a null allele of Tlx by creating stop codons after reading frame shifts. Detailed information has also been provided for the generation and characterization of transgenic Nes-GFP mice (Yamaguchi et al., 2000) and conditional allele of Trp53 (Marino et al., 2000). All mice were housed under a 12-h light/dark cycle and had free access to food and water in a controlled animal facility. No significant phenotypic differences were observed between male and female mice; thus, both genders were included in the analysis. Experimental protocols were approved by the Institutional Animal Care and Use Committee at UT Southwestern.
Tamoxifen (Sigma) was dissolved through bursts of sonication in sesame oil with a final stock concentration of 20 mg/ml. Mice were injected intraperitoneally once daily with 4 mg tamoxifen per 20 g body weight or sterile sesame oil (vehicle) for 5 consecutive days. The mice were sacrificed 24 hr after the last injection or at the indicated time points. Dividing cells in vivo were labeled by intraperitoneal injection of BrdU, CldU or IdU (in PBS) at the indicated dose for the specified duration.
Tlx cDNA was subcloned into either CMV-ires-GFP or hGfap-GFP-T2A lentiviral vector in which the expression of the transgene was driven by the CMV or hGfap promoter (Lee et al., 2008). HEK293T cells were cotransfected with lentiviral and packaging plasmids (pMDL, VSV-G and pREV) by CaPO4 method. Virus-containing supernatants were collected at 24, 48 and 72 hr post transfection, pooled and filtered through a 0.22 μm filter to remove cellular debris. Viral particles were then concentrated by centrifugation at 25,000 rpm for 2 hr at 4°C. Lentiviral titers were measured in either HEK293 cells (CMV promoter) or primary astrocytes (hGfap promoter). One μl of viruses (1 × 108 cfu/ml) were stereotactically injected into the DG. We used the following coordinates from bregma for Tlx-null mice: anterior/posterior (AP), −1.3 mm; medial/lateral (ML), ±2 mm; and dorsal/ventral from skull (DV), −1.5 mm.
The adult mice were sacrificed via CO2 overdose and perfused with 1xPBS followed by ice-cold 4% paraformaldehyde (PFA) in PBS. Brains were dissected and post-fixed overnight with 4% PFA at 4°C, followed by cryoprotection with 30% sucrose solution in PBS for another 24 hr. Frozen brains were sectioned at 40 μm with a sliding microtome (Leica) and free-floating sections were collected and stored in antifreeze solution at −20°C. For immunostaining, sections were washed 3 times with PBS and blocked for 1 hr at RT with blocking solution (3% BSA/0.2% Triton X-100 in PBS). The sections were then incubated overnight at 4°C with primary antibodies diluted in blocking solution. After 3 rinses with PBST buffer (0.2% Triton X-100 in PBS), the sections were further incubated with the corresponding secondary antibodies in blocking solution for 2 hr at RT. When necessary, nuclei were stained with Hoechst 33342 (Hst, Sigma) (1 μg/ml in PBS). Sections were washed and mounted onto Superfrost glass slides (Fisher Scientific) with mounting medium containing diazabicyclo-octane (DABCO, Sigma). For BrdU, CldU and IdU detection, prior to incubation with primary antibodies, sections were treated with 50% formamide in 2xSSC buffer for 2 hr at 65°C, followed by further treatment with 2 M HCl for 30 min at 37°C and equilibration with 0.1 M boric acid (pH8.5). Sequential detection of CldU and IdU were conducted essentially as described (Tuttle et al., 2011). Briefly, HCl-treated brain sections were first incubated overnight with antibody against IdU/BrdU (mouse clone 3D4) at 4°C, followed by high stringency wash with fresh TBST buffer (36mM Tris, 50mM NaCl, 0.5% tween-20; pH 8.0) at 37°C. The sections were then incubated overnight with antibody against CldU/BrdU (rat BU1/75) at 4°C, washed with PBST and detected with corresponding secondary antibodies.
The following primary antibodies were used: GFP (rabbit, 1:500, Molecular Probes; chick, 1:1000, Aves Labs); GFAP (mouse, 1:500, Sigma; guinea pig, 1:1000, Advanced Immunochemicals); BrdU/CldU (rat BU1/75, 1:500, Accurate); BrdU/IdU (mouse clone 3D4, 1:1000, BD Pharmingen); NeuN (rabbit, 1:500, Millipore); Sox2 (rabbit, 1:500, Millipore); BLBP (rabbit, 1:500, Millipore); Nestin (mouse, 1:200, Pharmingen); Ki67 (rabbit, 1:500, Novocastra); MCM2 (rabbit, 1:500, Cell Signaling); DCX (goat, 1:150, Santa Cruz); Olig2 (rabbit, 1:500, Millipore), pSMAD1/5/8 (rabbit, 1:500, Cell Signaling); GSTπ (mouse, 1:200, BD Sciences); S100β (rabbit, 1:1000, Swant); RIP (mouse, 1:250, Hybridoma Bank, Iowa); and Histone H3 (rabbit, 1:500, Cell Signaling). Alexa Fluor 488-, 594-, or 647-conjugated secondary antibodies produced in goat or donkey (Invitrogen) were used for indirect fluorescence. Images were taken using a Zeiss LSM510 confocal microscope. A Cell Counter software plugin in the ImageJ program was used to count cells. Data were obtained from 12 random sections from 3–5 mice in each group.
NSCs from 6-to-8-week-old Tlx+/LacZ;CreERTM (CZ) or Tlxflox/LacZ;CreERTM (FCZ) mice were isolated and cultured in growth medium (DMEM/F12 medium supplemented with N2 (Invitrogen), heparin (5ug/ml, Sigma), EGF (20ng/ml, Peprotech) and bFGF (20ng/ml, Peprotech)), as previously described (Zhang et al., 2008). Tlx+/LacZ NSCs were sorted based on LacZ expression using FluoReporter lacZ Flow Cytometry Kits, according to the user’s manual (Invitrogen). To acutely delete Tlx, FCZ cells were treated with 10 nM 4-hydroxytamoxifen (Sigma) for the indicated duration. Similarly, Tlx-null NSCs were isolated from E18.5 cortices and cultured in growth medium. These cells were labeled the following day with 1 μM 1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI, Molecular Probes) for 15 min at RT, followed by lentiviral transduction. Four hr later, the cells were washed with growth medium and continuously cultured for another 14 days with medium change in every other day. DiI and GFP were used for gating in flow cytometry. For immunocytochemistry, cultured cells on chamber slides (BD Sciences) were fixed with 4% PFA, washed with PBS, blocked for 30 min at RT, followed by overnight incubation with primary antibodies in blocking solution at 4°C. For detection of BrdU-labeled cells, the fixed cells on slides were treated with 2 M HCl at 37°C for 30 min, followed by washing with PBS and incubation with primary antibodies.
Three-week-old mice (n = 5–9 for each genotype) were used. Brains were cut into 1 mm coronal sections with a brain matrix, and fresh tissues surrounding the lateral ventricles were microdissected on ice. Tissues were enzymatically digested for 30 min at 37°C with PPD solution, which consists of papain (2.5 u/ml), DNase I (250 u/ml) and dispase II (1.0 u/ml) in DMEM culture medium supplemented with 4.5 g/L glucose. After trituration with a 5-ml pipette, cells were sequentially washed 3 times with DMEM/10% FBS and once with PBS/4% FBS. They were then filtered through a 40 μm cell strainer for live cell sorting based on GFP expression. GFP+ cells were directly sorted into TRIzol® LS reagent (Invitrogen) and the total RNAs were isolated using RNeasy Mini Kit (Qiagen). RNA quality was determined by Bioanalyzer (Agilent). cDNAs were synthesized from 50 ng of total RNA and were further amplified with an RNA-seq Ovation system (NuGen). A cDNA library was prepared and subjected to parallel sequencing using Genome Analyzer II (Illumina). Sequence tags were analyzed as described (Masui et al.). A total of 20.8 million unique tags were obtained for wild-type cells and 19.7 million with RNAs from Tlx−/− cells. Based on the rpkm (reads per kilobase of exon model per million mapped reads) value for Mcm2, which is an essential gene for licensing DNA replication, a cut-off value of 2.0 was used for either wild-type or Tlx−/−. Gene expression changes were calculated as the ratio of the rpkm for wild-type to that of Tlx−/− and further analyzed with DAVID program for KEGG signaling pathways.
Total RNAs from cultured or FAC-sorted cells were isolated using TRIzol reagent (Invitrogen) and RNeasy Mini Kit (Qiagen). The Superscript III system (Invitrogen) and random primers were used to synthesize cDNA from 1.0 μg total RNA from cultured cells. For sorted cells, 20 ng of total RNA was used to make the first strand cDNAs, which were further amplified using the RNA-seq Ovation system, according to manufacturer’s protocol (NuGEN). Gene expression was analyzed using the SYBR Greener system (Invitrogen) on a 384-well ABI 7900HT thermocycler (Applied Biosystems). Primer sequences for PCR reactions are available upon request.
Differences between groups were determined for significance using the two-tailed Student’s t test with equal variance or ANOVA. A p value of < 0.05 was considered significant.
We employed an inducible approach to specifically examine the identity and lineage of Tlx-expressing cells in the adult mouse brain. Such a strategy has been used to show that TLX-expressing cells are type B NSCs in the SVZ (Liu et al., 2008). However, the activation status of those cells and the identity of the cells in the SGZ are not known. Through homologous recombination, the synthetic gene CreERT2 was inserted into the first exon of the Tlx locus in a 148-kb BAC genomic clone, which covers the entire Tlx gene and the flanking genomic sequences (Fig. 1A). Such a recombination strategy presumably preserved all of the regulatory elements required for Tlx expression. The resulting mouse strain was termed pTlx-CreERT2. After breeding to the reporter Rosa-GFP mouse line, tamoxifen (TM)-induced expression of GFP was identified in the SVZ and the SGZ in the adult mouse brains (Fig. 1B). Some of these GFP+ cells exhibit radial glia-like morphology and coexpress the marker GFAP, two hallmarks of adult NSCs (Fig. 1C, arrows). These cells resemble type-1 or type B stem cells within the SGZ or SVZ, respectively. In addition, GFP expression can also be identified in GFAP-negative cells, with a tangential orientation in the SGZ. These cells morphologically resemble those of type-2 transient amplifying cells (Fig. 1C, asterisks).
NSCs exist in either an inactive or activated state. We used marker MCM2 to distinguish these two states. MCM2, together with MCM4, 6 and 7, forms the pre-replication complex, which is essential for the initiation of eukaryotic genome replication. Its expression is present throughout all phases of the proliferative cell cycle but down-regulated when cells exit from cycle into quiescent, fully differentiated or replicative senescent states (Stoeber et al., 2001; Wharton et al., 2001; Blow and Hodgson, 2002; Maslov et al., 2004; Williams and Stoeber, 2007). Down-regulation of MCM2 in intestinal stem cells marks their quiescent state (Kingsbury et al., 2005). Careful analysis through confocal microscopy revealed that MCM2 is expressed in ~20% (20 ± 2.5%) of GFP+GFAP+ cells that have a radial morphology, showing that Tlx-expressing cells generate both dormant and activated stem cells (Fig. 1D, indicated by arrows or arrow heads, respectively). Furthermore, MCM2 was also detected in all of the GFP+GFAP− cells with short processes and tangential orientation in the SGZ. These cells resemble type-2 transient amplifying cells in this neurogenic niche (Fig. 1D, denoted by asterisks). Interestingly, some of the MCM2+GFP+GFAP− type-2 cells can be detected in close contact with MCM2+GFP+GFAP+ type-1 cells (Fig. 1D, denoted by asterisk), suggesting that TLX-expressing activated type-1 cells give rise to rapidly dividing type-2 cells. Together, these results suggest that TLX lies at the top level of a hierarchy that is required for postnatal neurogenesis.
TLX is essential for NSC proliferation and neurogenesis in the postnatal brain (Shi et al., 2004; Liu et al., 2008; Zhang et al., 2008). The fate of stem cells lacking Tlx is not clear, although there is an indication that they undergo spontaneous differentiation into mature astrocytes and thus deplete NSCs (Shi et al., 2004). Taking advantage of the LacZ marker that was knocked into the Tlx locus, we examined the fate of Tlx-expressing cells. Surprisingly, β-gal staining for LacZ gene expression can be robustly detected in the neurogenic niche even in the adult Tlx-null brain, suggesting that deletion of Tlx does not cause cell death or elimination of those Tlx-expressing cells (Fig. 2A). Importantly, Nestin+ or Sox2+GFAP+ cells can still be detected in the neurogenic niche in the adult Tlx-null brains (Fig. 2B).
To examine the early fate of stem cells lacking Tlx, we isolated TLX-positive cells based on LacZ expression and cultured them in monolayer. These cells also contained a floxed allele of Tlx and a CreERTM transgene (Fig. 2C). TM treatment can induce robust deletion of the conditional allele of Tlx in a time-dependent manner. However, qRT-PCR analysis showed that the markers for stem cells, such as Sox2 and Nestin, are still expressed. In fact, the expression of Nestin is rather enhanced after inducible deletion of Tlx in these cultured cells. In contrast, deletion of Tlx results in a significant down-regulation of Ezh2, a member of the polycomb group (PcG) that regulates methylation status of core histones and embryonic stem cell development.
To further investigate the identity of those Tlx-null stem cells, we crossed the Tlx-null mice into the pNestin-GFP (Nes-GFP) transgenic background. Expression of gfp is under a neural-specific Nestin enhancer, which marks NSCs and transient amplifying cells in the neurogenic niche (Yamaguchi et al., 2000). In agreement with our immunostaining for endogenous Nestin expression, Nes-GFP+ cells with radial morphology can be observed in both of the neurogenic niches (the SVZ and the SGZ) in adult Tlx-null brains (Fig. 2D and data not shown). Interestingly, many GFP+ cells in Tlx-null brains are mis-localized in the outer layer of the dentate blades (see below), although they still maintain their normal outward orientation. Through confocal microscopy, we examined in detail the identity of these Nes-GFP+ cells and found that they co-express GFAP, BLBP, and Sox2 and that they exhibit radial glia morphology in the DG (Fig. 2E). These are characteristics of postnatal NSCs (Ihrie and Alvarez-Buylla, 2008; Zhao et al., 2008; Kriegstein and Alvarez-Buylla, 2009). Furthermore, these GFP+ cells lack marker staining for astrocytes (S100β and glutamine synthetase (GS); Fig. 3A–B), oligodendrocytes (Olig2, GSTπ, and RIP; Fig. 3C–D), or immature (DCX) or mature (NeuN) neurons (Fig. 3E–F). In contrast to a previous report (Shi et al., 2004), these data clearly indicate that deletion of Tlx during embryogenesis does not lead to a depletion of cells that have characteristics of NSCs or result in spontaneous differentiation of NSCs into mature astrocytes at the time points examined.
The above data demonstrates that cells with morphology and marker expression resembling that of NSCs persist in the adult Tlx-null brains, raising the question about the status of these cells. By 4 weeks of age, Nes-GFP+ cells in Tlx-null brains are no longer able to generate DCX+ immature neurons (Fig. 3E). In addition, they do not proliferate. This is indicated by a lack of BrdU incorporation (Fig. 2B) or staining for Ki67, an endogenous marker for cell proliferation. Importantly, these cells also lack MCM2 expression, suggesting that they are in a non-licensed state for replication (see below).
Next, we examined the time course during which TLX is required for stem cell activation. Ki67, PCNA and MCM2 are endogenous markers that were used to label the status of Nes-GFP+ NSCs. Expression of Ki67 is present throughout the cell cycle but not in the G0 or early G1 phase, thus tightly associated with active cell proliferation (Gerdes et al., 1984; Scholzen and Gerdes, 2000). At 7 days of age (postnatal day 7, P7), ~20% of Nes-GFP+ cells in the DG were labeled by Ki67-staining in both the wild-type and the mutant brains (Fig. 4A). However, 7 days later, the number of Ki67+GFP+ cells dropped more than 8 fold to 2.4% in the mutant brains, while Ki67+GFP+ cells from their wild-type littermates continued to proliferate at a rate of 14.2%. By 3 weeks of age (P21), only 0.7% of GFP+ cells in Tlx-null mice expressed Ki67, which was in sharp contrast to 10.8% in the control brains (Fig. 4B). It is known that the proliferation rate of stem cells drops with age, which was manifested in the control brains where we observed a 40% decrease in proliferating stem cells between 7 and 21 days of age. However, this rate of age-dependent decline was dramatically increased to 96% in Tlx-null NSCs.
Similar to Ki67, PCNA expression identifies cells in late G1, S, and G2-M phase, thus is associated with active cell proliferation (Giordano et al., 1991). The percentage of PCNA-labeled cells was much higher than those labeled by Ki67 (comparing Fig. 4D to Fig. 4B). This may be due to differential sensitivity of the antibodies used and/or an additional role of PCNA in DNA repair (Stoimenov and Helleday, 2009). This notwithstanding, deletion of Tlx resulted in a 72% and 90% reduction of PCNA+ cells among the total Nes-GFP+ cells at P14 and P21, respectively (Fig. 4D), which is consistent with the results obtained with Ki67-staining (Fig. 4B).
Formation of the pre-initiation complex for DNA replication and its chromosomal loading are the prerequisite steps for cell to proliferate (Geng et al., 2003; Lea et al., 2003; Chuang et al., 2009). One of the key components of this complex is the MCM2 protein, which regulates the helicase activity of the Mcm4/6/7 hexamer (Ishimi et al., 2001). Unlike Ki67 or PCNA, expression of MCM2 marks all activated cells including those leaving the G0 to the G1 phase of the cell cycle, thus also identifies noncycling cells with proliferative potential (Kayes et al., 2009; Torres-Rendon et al., 2009). At 7 days of age, 83% of Nes-GFP+ cells express MCM2 in both the Tlx-null mice and their littermate controls. This number dropped to 30% seven days later, suggesting a sharp increase with age in the number of stem cells that are in an inactive stage. Interestingly, we did not observe any difference between Tlx mutants and their wild-type controls at this early postnatal stage (Fig. 4B–C). However, by 3 weeks, MCM2+GFP+ cells decreased by 80% in Tlx-null compared to their littermate controls, resulting in a mere 3.5% of stem cells in proliferation-competent state. Because the decrease of Ki67+ or PCNA+ cells (at 2 weeks) precedes that of MCM2+ cells (at 3 weeks), these data indicate that a loss of TLX function first results in an age-dependent decrease of active proliferation, followed by an exit of cell cycle indicated by a non-licensed state.
The granule cell layer of the DG is established in an outside-in layering pattern: the early generated cells form the outer layer, whereas later born cells progressively populate the deeper inner layer. The SGZ within the inner layer constitutes a neurogenic niche for postnatal NSCs (Zhao et al., 2008). The cell bodies of these cells are localized in this microenvironment while their long, radial processes pass through the dentate granular cell layer and protrude outward. We found that many of the Nes-GFP+ cells in the adult Tlx-null brains are localized in the outer layer of the dentate blade, yet still maintain the correct orientation of their radial processes (Fig. 2D–E). Such mis-positioning of stem cells in the DG of Tlx-null mice can be detected as early as postnatal day 7 and becomes most obvious by the adult stage, when nearly 50% of these Nes-GFP+ cells are localized in the outer half of the granular cell layer in the Tlx-null mice (Fig. 5A–B).
To trace the origin of these mis-positioned Nes-GFP+ cells, we birth-dated cells born in embryonic or paranatal stages with BrdU. After 3 to 5 weeks, the cellular identity of label-retaining cells and their positions in the DG were analyzed. Although the majority of BrdU-retaining cells are mature NeuN+ neurons, a small subset of these cells displays radial morphology and coexpresses Nes-GFP and GFAP, which are characteristics of NSCs (Fig. 5C). Close to 100% and 62% of these cells are mis-positioned in the outer layer of the Tlx-null DG when cells are labeled at E15.5 and P0, respectively. In contrast, none of the label-retaining cells can be detected in the outer layer when cells are pulse-labeled at P7 (Fig. 5D–E). Such mis-positioned BrdU-retaining cells are rarely observed in wild-type mice. These data suggest that the cells that are mis-positioned in the outer layer of the Tlx-null DG originate from the embryonic or paranatal stages. Although dysregulated migration of label-retaining cells from inner layer to outer layer cannot be fully excluded, such results are consistent with an outside-in layering pattern of the DG and further indicate that deletion of TLX leads to inactive NSCs that are temporally “frozen” in their original birth-place.
Our results demonstrate that deletion of TLX leads to rapid age-dependent inactivation of NSCs. Since they do not yet spontaneously differentiate into either glial or neuronal cells, these inactive cells may become senescent. We performed staining for senescence-associated β-galactosidase at pH6.0. Unexpectedly, both wild-type and mutant brains showed staining in the DG and other regions of the hippocampus, although the staining in the mutant brains were stronger. We also conducted the same staining using in vitro cultured cells and observed similar robust staining in both wild-type or mutant cells in which Tlx is conditionally deleted. Thus, such staining cannot clearly differentiate whether the cells in the Tlx-null brains are senescent. Reasoning that quiescent but not senescent stem cells can be reactivated, we then performed rescue experiments by reintroducing Tlx. We initially tried an in vitro culture system. For this, Tlx-null cells from E18.5 cortices were isolated and labeled with DiI, followed by transduction with lentivirus expressing either GFP or GFP-T2A-TLX (Fig. 6A). Fourteen days later, flow cytometry showed that TLX-reintroduction caused a dramatic reduction of label-retaining cells when compared to that of GFP-transduced cells (GFP, 19.7 ± 1.7%; GFP-T2A-TLX, 1.9 ± 0.93%; n=3, p < 0.0001). This data suggests that re-expression of TLX rescued and promoted proliferation of the cultured Tlx-null cells.
To examine the in vivo role of TLX in inactive NSCs, we first checked the basal level of proliferating type-1 NSCs in both 9-week-old Tlx-null mice and their littermate controls after BrdU-labeling (100 mg/kg, twice a day for 5 days). Whereas 9 out of 215 Nes-GFP+;GFAP+ cells were also labeled by BrdU in control mice, only 1 cell was BrdU+ after counting 255 Nes-GFP+;GFAP+ cells in Tlx-null mice, indicating a very low level of basal proliferation of type-1 cells after deletion of Tlx (Fig. 6B). Through lentiviral delivery, Tlx was then reintroduced into Tlx-null;Nes-GFP brains at 6 weeks of age. Three weeks later, proliferating cells were labeled by BrdU incorporation (100 mg/kg, twice a day for 5 days). We observed a significant increase in the number of BrdU+;Nes-GFP+;GFAP+ cells in the Tlx-null brains after re-expressing Tlx (6.25 vs. 0.25 cell/injection site when comparing Tlx-virus-transduced mutant mice to those transduced with control viruses). Such increase is most likely due to reactivation of originally non-dividing stem cells rather than a mere enhancement of the rate/frequency of those already proliferating cells, since the latter case would result in clusters of BrdU-labeled cells (Encinas et al., 2011), which were not detected in this study (Fig. 6C).
To further directly examine the role of TLX in label-retaining NSCs, we performed a pulse-chase-pulse experiment to investigate whether exogenous TLX could enable Tlx-null NSCs to proliferate in adult mice. For this, we first labeled proliferating NSCs with one pulse of CldU (50 mg/kg) each at E18.5 and E19.5. After 6 weeks of chase, the DG of Tlx-null mice was stereotactically injected with lentiviruses expressing GFP or GFP-T2A-TLX under the control of hGfap promoter (Lee et al., 2008). Proliferating cells were then pulsed again with IdU (100 mg/kg, twice a day for 7 or 10 days). By immunohistochemistry and confocal microscopy, Tlx-induced GFP+CldU+IdU+ cells could be observed as early as 7 days post injection (dpi) (Fig. 6D). Analysis at 10 dpi showed more triple labeled cells in mice injected with viruses expressing GFP-T2A-TLX versus control GFP (18.3 vs. 3.75 cells/section). Together, these data demonstrate that re-expression of TLX enables label-retaining cells to proliferate.
How does TLX regulate stem cell activation? Previously, we have used cultured Tlx+ NSCs for global gene expression analysis after acute deletion of Tlx and found that many genes involved in cell proliferation showed significant changes (Zhang et al., 2008). Considering the potential side effects of long-term culture and the exogenous growth factors on stem cells, we wished to determine the function of TLX in endogenous NSCs. For this purpose, we took advantage of the recently developed RNA-seq technology. Lateral ventricles were dissected from 3-week-old mice (5–9 mice for each genotype). After enzymatic dissociation, Nes-GFP+ cells were isolated by fluorescent activated cell sorting (FACS) based on GFP expression, which resulted in over 97% purity. Total RNA was purified from these sorted cells and converted into cDNA, which was then amplified using the Ovation RNA-seq system from NuGEN.
After sequencing of the prepared cDNA libraries, around 20 million sequence tags for each genotype were uniquely mapped onto the mouse genome. The expression for MCM2, which is 2 rpkm (Reads Per Kilobase of exon model per Million mapped reads), was chosen as the cutoff. Such analysis identified 1721 genes with 2-fold changes. Importantly, we did not detect any significant changes of many house-keeping genes (such as Hprt, Rpl23 and Rpp30). Additionally, the expression of several stem cell markers, such as Nestin, Sox2, or Blbp/Fabp7, was not altered, suggesting that we indeed isolated and used the same population of cells for RNA-seq analysis (Fig. 7A). Together, these results demonstrated the robustness of our RNA-seq data.
KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis of RNA-seq data showed that several signaling pathways were significantly altered upon deletion of Tlx in stem cells (Fig. 7B). Among these, changes of genes involved in DNA replication, cell cycle and p53 signaling may contribute to the inactivation of NSCs upon deletion of Tlx. Using independently FACS-isolated Nes-GFP+ cells and qRT-PCR, we analyzed and confirmed the genes involved in DNA replication. Significantly, these genes were also identified in our previous microarray analysis using cultured cells (Zhang et al., 2008) (Fig. 7C–D). To further analyze the role of TLX in DNA replication, we isolated and cultured Tlx+ NSCs and found that the DNA content reached a plateau by 60 h after TM-induced deletion of Tlx, whereas the control cells continued their growth (Fig. 7E).
Of the genes involved in cell cycle control and p53 signaling, p21/Cdkn1a is significantly up-regulated in Tlx-null stem cells (Fig. 7A). This up-regulation of p21 is confirmed by use of an inducible deletion system in cultured NSCs by qRT-PCR (Fig. 8A), indicating a direct link between p21 expression and TLX function. This is further supported by data showing that TLX directly binds to the promoter region of p21 (Sun et al., 2007). Because of its demonstrated role in maintaining stem cell quiescence (Cheng et al., 2000; Perucca et al., 2009), these data strongly raised the possibility that p21 may be a critical component in mediating TLX-deletion-induced NSC inactivation. We tested this possibility by ectopic expression of p21 in cultured NSCs and indeed observed a significant reduction of Ki67+ cells (Fig. 8B).
It is well established that the expression of p21 is under the direct control of the p53 signaling pathway in most cellular contexts. We examined whether p53 is also involved in Tlx deletion-induced upregulation of p21 in postnatal NSCs. We isolated these cells from 2-month-old mice harboring floxed alleles of Tlx and/or p53. Consistent with our previous observation (Zhang et al., 2008), Cre-mediated deletion of Tlx results in upregulation of p21. Such upregulation is completely abolished upon acute removal of p53, suggesting that TLX controls p21 expression in a p53-dependent manner (Fig. 8C). Besides p21, our RNA-seq also identified altered expression of several other genes that are involved in the p53 signaling pathway. These changes can be further confirmed by qRT-PCR analysis of independently isolated RNA samples from sorted Nes-GFP+ cells (Fig. 8D). These results suggest that p53 may play a role in controlling the activation status of Tlx-null NSCs. To test this possibility, we acutely deleted Tlx with or without concomitant removal of p53 expression in cultured adult NSCs. As demonstrated before (Zhang et al., 2008), Cre-mediated deletion of Tlx led to a significant reduction of BrdU+ cells. Interestingly, this proliferation defect of Tlx-null NSCs was completely rescued by concomitant deletion of p53 (Fig. 8E–F). It should also be noted that deletion of p53 alone resulted in a small but significant increase of proliferating cells, which is consistent with a demonstrated role of p53 in adult NSCs (Meletis et al., 2006). Together, these data show that TLX genetically interacts with the p53 signaling pathway to tightly regulate the activity of postnatal NSCs.
Postnatal neurogenesis depends on the continued presence of infrequently dividing NSCs in the SGZ and the SVZ. However, the underlying driving force for NSCs to become activated remains largely unknown. Using inducible lineage tracing and genetic markers for adult NSCs, we show here for the first time that Tlx -expressing cells generate both activated and non-proliferative postnatal NSCs and that it is required for NSC activation and positioning in the neurogenic niche. Our whole-genome RNA-seq data reveals that TLX acts as a key coordinator for multiple signaling pathways in the regulation of NSC behavior. Specifically, TLX genetically interacts with the p53 pathway to control NSC activation (Fig. 8G). Together, these data uncover a key role of TLX in controlling the proliferative capacity of postnatal neural stem cells, but not their radial glia morphology and the expression of multiple NSC markers.
TLX was recently shown to be expressed in type B cells in the LV (Liu et al., 2010). In agreement with this finding, our inducible lineage tracing further demonstrated that Tlx-expressing cells give rise to both type-1 cells in the SGZ and type B cells in the SVZ. This implies a much broader role of TLX in the regulation of neurogenic stem cells in the adult brain. Consequently, deletion of Tlx leads to completely abolished neurogenesis in both of the neurogenic regions (Shi et al., 2004; Liu et al., 2008; Zhang et al., 2008). Type-1/B cells are slowly dividing stem cells that give rise to transiently but rapidly amplifying Type-2/C cells (Kriegstein and Alvarez-Buylla, 2009). Interestingly, both active and inactive NSCs are derived from Tlx-expressing cells. We cannot unambiguously differentiate which of the two types of NSCs express Tlx due to the lack of a sensitive antibody that recognizes endogenous TLX in the adult neurogenic regions. Nevertheless, three pieces of data indicate that Tlx is expressed in inactive NSCs. First, the LacZ reporter, which was knocked into the endogenous Tlx locus, can still be robustly detected in both the SVZ and the SGZ in Tlx-null mice. Second, whole-genome sequencing of total RNAs clearly shows the expression of undeleted exons in sorted non-proliferative Nes-GFP+ cells from 3-week-old Tlx-null brains. Third, it is known that aging is accompanied by a significant decrease of proliferating stem cells; however, our qRT-PCR analysis cannot detect a significant change in Tlx expression between young (1 month old) and aged mice (26 months old) (data not shown). Expression of Tlx in non-proliferative NSCs raised the possibility that the activity of TLX itself in NSCs may be under regulation through either post-translational modifications or ligand-binding. The ligand for TLX (a member of the nuclear hormone receptor superfamily) remains unknown. Future studies are clearly needed to identify this ligand, which may provide a unique opportunity to control TLX-dependent postnatal neurogenesis.
The function of TLX during development is to prevent precocious cell cycle exit and premature differentiation (Roy et al., 2004; Zhang et al., 2006; Li et al., 2008). By contrast, the role of TLX in postnatal NSCs is not clear, although it has been proposed to inhibit spontaneous differentiation of NSCs into astrocytes (Shi et al., 2004). We were able to detect significantly more astrocytes in postnatal Tlx-null mice. However, these cells are largely localized to the gliogenic regions with typical stellate morphology and the expression of S100β and glutamine synthetase, which are characteristics of differentiated astrocytes. Since our RNA-seq showed that genes involved in bone morphogenetic protein (BMP) and leukemia inhibitory factor (LIF) signaling, such as Bmpr1b (2.1 fold increase), Bmper (2.08 fold decrease), Bmp4 (8.3 fold increase), and Lifr (2.25 fold increase), are significantly altered upon deletion of Tlx, we suspect that enhanced response of neural progenitor cells to BMP and/or LIF signaling may be the underlying mechanism for the observed increase of astrocytes in gliogenic regions. On the other hand, cells within the neurogenic niche of Tlx-null brains exhibit radial glia morphology, a characteristic of adult NSCs. Although they are GFAP+, they do not express markers for mature glia (Fig. 3); rather, they still express markers for stem cells, such as Nestin, Sox2, and BLBP (Fig. 2E). Our whole-genome RNA-seq data further supports this conclusion by showing unaltered expression of these stem cell markers (Fig. 7A). These data clearly demonstrate that loss of TLX function does not lead to spontaneous differentiation of NSCs directly into mature glia, but rather results in non-proliferative cells that genetically and morphologically resemble NSCs.
Neurogenesis continues into adulthood but reduces considerably with the aging process (Seki and Arai, 1995; Kuhn et al., 1996; Tropepe et al., 1997). Such age-dependent reduction may be attributed to both decreased proliferation and/or division-coupled astrocytic differentiation (Hattiangady and Shetty, 2008; Dranovsky et al., 2011; Encinas et al., 2011). By using MCM2 as a marker for all activated stem cells and Ki67 and PCNA to label actively proliferating cells, our detailed analysis of postnatal NSCs revealed a previously unappreciated role of TLX in stem cell biology. Loss of TLX function results in an age-dependent dramatic reduction of proliferating cells with a compensatory increase in the number of MCM− inactive NSCs. Remarkably, ectopic expression of TLX can reactivate these inactive NSCs in vivo. Such a result is consistent with a recent report showing that overexpression of Tlx is sufficient to enhance neurogenesis even in aged animals (Liu et al., 2010). Increased expression of tumor suppressors INK4A/ARF has been linked to NSC aging (Molofsky et al., 2006; Levi and Morrison, 2008). Yet, we did not detect any expression change on these two genes through our extensive genome-wide analysis using either cultured adult NSCs or sorted live NSCs from wild-type or Tlx-null animals. This suggests that TLX controls stem cell aging in an INK4A/ARF-independent manner. Our unbiased global expression analysis revealed that many genes in p53 signaling (such as p21, Gadd45, Btg2, Ccnd1 and Ccnd2) are significantly altered. Interestingly, our data shows that ectopic expression of p21 is sufficient to inhibit proliferation of cultured NSCs. Conversely, concomitant loss of p53 function rescues the Tlx-deletion-induced proliferation defect of cultured NSCs. Such a role of p21 and p53 in postnatal NSCs is consistent with their demonstrated function in controlling stem cell quiescence (Cheng et al., 2000; Kippin et al., 2005; Gil-Perotin et al., 2006; Meletis et al., 2006; Liu et al., 2009; Perucca et al., 2009). In addition to p53 signaling, however, it should be noted that TLX also controls the expression of a plethora of other genes that may also play important roles in the regulation of NSC activation (Fig. 8G). Future studies are required to tease out the in vivo function of these genes in NSCs and during adult neurogenesis.
Emerging evidence indicates that slowly dividing, self-renewable NSC-like cells are the cellular origin for brain tumors (Stiles and Rowitch, 2008). Therefore, our finding that TLX plays a key role in NSC activation may have therapeutic implications in brain tumorigenesis. Recent data demonstrating that increased activity of TLX may initiate brain tumor formation in fly or mice (Kurusu et al., 2009; Liu et al., 2010) supports our findings. Overexpression of Tlx also has been associated with certain human brain tumors, including astrocytomas, ependymomas and glioblastomas (Taylor et al., 2005; Phillips et al., 2006; Sim et al., 2006). Mechanistically, our systematic genome-wide analysis using microarray (Zhang et al., 2006; Zhang et al., 2008) and RNA-seq (current study) revealed that TLX controls a plethora of genes involved in DNA replication, cell cycle, p53 signaling and pathways in cancers. These data suggest that TLX lies at a nodal point to control NSC activation and replication by coordinating a complex genetic network. Because of its essential role in controlling postnatal neurogenesis, it will be interesting to examine whether TLX is also required for the development of certain type of brain tumors and to determine the relationship of adult neurogenesis and tumorigenesis. Our finding that the proliferation defect of Tlx-deleted NSCs can be rescued by concomitant removal of p53 suggests that targeting TLX alone may not be sufficient to control tumorigenesis. However, because of its restricted expression in the nervous system and the potential regulation by a ligand, TLX could be an excellent drug target in combination with other therapeutic strategies to treat certain human brain tumors.
The authors thank Ronald Evans for sharing Tlx mutant mice; Amelia Eisch for providing Nes-GFP mice; Michael Brenner for providing hGfap enhancer, Pierre Chambon for providing the CreERT2 plasmid; Richard Lu for sharing the Olig2 antibody; Ray MacDonald and Galvin Swift for assistance with RNA-seq; the Transgenic Core Facility at UT Southwestern for generating BAC transgenic mice; and the Flow Cytometry Core for FACS. The authors also thank Eric Olson, Jane Johnson, Steven Kliewer and Jenny Hsieh for comments, Ronald Evans and Fred Gage for intellectual support, and Pamela Jackson for administrative assistance. C-L.Z. is a W. W. Caruth, Jr. Scholar in Biomedical Research. This work was supported by the Whitehall Foundation (2009-12-05), the Welch Foundation (I-1724), the American Heart Association (09SDG2260602), the NIH (1DP2OD006484 and R01NS070981) and a Startup Fund from UT Southwestern to C-L.Z. No conflict of interest is claimed.
Conflict of Interest: None