The extracellular matrix glycoprotein tenascin-R regulates neurogenesis during development and in the adult dentate gyrus of mice Free

Proportionately small numbers of GABAergic interneurons in the neocortex and hippocampus provide inhibitory input and regulate the activity of large populations of principal excitatory cells. Deficits in the balance of excitation and inhibition lead to brain dysfunction and appear to be pivotal in the pathogenesis of major neuropsychiatric disorders, such as schizophrenia and autism (Chao et al., 2010; Kehrer et al., 2008; Uhlhaas and Singer, 2012). Abnormal inhibition of neuronal excitability in neuropsychiatric disorders is often related to abnormalities in the generation of GABAergic neurons (Lee et al., 2011; Levitt et al., 2004; Volk et al., 2000). Identifying the molecular basis underlying the generation of interneurons is therefore expected to augment our understanding of psychiatric disorders.

The ability of neural stem or progenitor cells (NSCs) to produce different types of neurons is determined by intrinsic genetic programs (Martynoga et al., 2012; Miyata et al., 2010; Shen et al., 2006). Cell fate can also be specified by signals from the local environment in the nervous system (Borello and Pierani, 2010; Guillemot and Zimmer, 2011). For example, extracellular matrix molecules can interact with cell surface receptors and regulate the generation of neurons by promoting or impeding the self-renewal and fate determination of NSCs (Barros et al., 2011; Ma et al., 2008; Tanentzapf et al., 2007).

Among the extracellular matrix molecules involved in neurogenesis is tenascin-R (TNR), a large glycoprotein expressed in the central nervous system by subpopulations of interneurons, motoneurons and oligodendrocytes (Nörenberg et al., 1996; Wintergerst et al., 1993). TNR, with its epidermal-growth-factor- and fibronectin-type-III-homologous repeats (EGFL and FNIII, respectively) and fibrinogen knob, is a multifunctional molecule implicated in cell adhesion and repulsion, and the promotion and inhibition of neuritogenesis, as well as synaptic and ion channel activity (Chiquet-Ehrismann and Tucker, 2011; Dityatev et al., 2010; Pesheva and Probstmeier, 2000). TNR also interacts with the GABA(B) receptor through its associated carbohydrate, HNK-1, and can thus directly influence signaling in cells expressing this receptor (Saghatelyan et al., 2004b). Furthermore, TNR is important for activity-dependent recruitment of neuroblasts in the adult mouse forebrain (Saghatelyan et al., 2004a). In vitro, the EGFL and FNIII domains 6–8 of TNR have been shown to modulate NSC proliferation and differentiation through interactions with the cell-surface receptor β1 integrin (Liao et al., 2008). Deficiency in TNR leads to increased anxiety and severe cognitive deficits in mice (Freitag et al., 2003; Montag-Sallaz and Montag, 2003).

The present study was designed to analyze the role of TNR in differentiation of NSCs in the murine dentate gyrus. We provide evidence that TNR deficiency leads to increased numbers of GABAergic neurons in the dentate gyrus, when compared with wild-type (TNR^+/+) littermates early during development, and to increased numbers of GABAergic neurons and granule cells in the adult. In addition, we propose signaling pathways that might underlie these phenotypes during development and in the adult.

We first analyzed the morphology of the dentate gyrus in adult mice. TNR^−/− mice showed an increase in the volume of the granule cell layer of the dentate gyrus at 3 months of age and throughout adulthood when compared with TNR^+/+ littermate mice (Fig. 1A,B). Although the volume of the granule cell layer of the dentate gyrus remained stable in TNR^+/+ mice, it increased between 3 and 18 months of age in TNR^−/− mice. This abnormality of the granule cell layer was not due to changes in cell density (143,000±3000 cells/mm³ in TNR^+/+ mice versus 149,000±6000 cells/mm³ in TNR^−/− mice in 3-month-old animals), but was a result of an increase in the total number of granule cells (Fig. 1A–C). Previous work has shown that adult (4- to 6-month-old) TNR^−/− mice have more parvalbumin immunoreactive (PV⁺) inhibitory neurons in all subfields of the hippocampus compared with TNR^+/+ littermates (Morellini et al., 2010). A likely explanation of this abnormality is that TNR is involved in regulating the generation of inhibitory interneurons during embryonic development. To test this notion, we analyzed numbers of GAD65/67⁺ (GABAergic) neurons (Fig. 1D) in the dentate gyrus at embryonic day (E) 18, an age at which the generation of inhibitory interneurons is in its final stage (Danglot et al., 2006). The number of GAD65/67⁺ interneurons was twice as high in TNR^−/− mice when compared with TNR^+/+ littermates (Fig. 1E). A similar difference was found at postnatal days (P) 7 and 90 (Fig. 1E). These findings suggest that TNR is involved in the regulation of the generation of GABAergic inhibitory interneurons during embryogenesis.

GABAergic neurons in the dentate gyrus are mainly generated between E13 and E14 (Danglot et al., 2006). Given the increase in the number of GABAergic inhibitory interneurons at E18 in TNR^−/− mice, it appeared most likely that TNR deficiency influences the cell fate determination of NSCs at early embryonic stages. To address this issue, we first analyzed whether TNR is expressed in the ganglionic eminence, the origin of hippocampal neurons during embryonic development (Fig. 2A). Indeed, TNR was expressed in close proximity to proliferating (Ki67⁺) cells (Fig. 2A). The expression of TNR in mouse embryos was confirmed by western blot analysis of total brain homogenate (Fig. 2B). However, in vitro TNR was not detectable in cultured nestin⁺ NSCs. Two days after the induction of differentiation by withdrawal of FGF-2 and EGF, immunoreactivity for TNR was detected in a subset of differentiating cells (Fig. 2C). Taken together, our results demonstrate that TNR is not expressed by nestin⁺ NSCs, but is expressed by differentiating or differentiated cells.

To determine whether TNR is expressed in the adult hippocampus, immunohistochemical analysis of TNR expression was performed and showed prominent expression of TNR in the sub-granular layer of the dentate gyrus. In particular, TNR was detectable in close proximity to Ki67⁺ progenitor cells (Fig. 2D). Western blot analysis of the hippocampus of TNR^+/+ mice showed sustained expression of TNR in the hippocampus at both embryonic and adult stages (Fig. 2E).

Molecules of the extracellular matrix can interact with integrin receptors (Tanentzapf et al., 2007). By immunohistochemical analysis, we determined the distribution of TNR and β1 integrin in the ganglionic eminence of TNR^+/+ mice at E14, the time when generation of interneurons is most prominent. In line with previous observations (Campos et al., 2004), integrin expression was detected in the germinal neuroepithelium throughout the ganglionic eminence (Fig. 3A). In TNR^+/+ mice, TNR and β1 integrin were co-expressed in the ganglionic eminence (Fig. 3A). In addition, TNR could be co-immunoprecipitated with β1 integrin from brain homogenate at embryonic day E14 (Fig. 3B) indicating that TNR and β1 integrin are associated during mouse brain development.

We then examined whether the expression of β1 integrin is altered in TNR^−/− mice using neurospheres as a model system because these three-dimensional mixtures of stem cells, committed precursors and differentiated cells are similar to embryonic brain tissue in cell and extracellular matrix composition (Campos et al., 2004). We showed that β1 integrin is expressed and partially colocalizes with TNR (Fig. 3C) and that cells expressing TNR are in the vicinity of Ki67⁺ proliferating cells (Fig. 3E). We also probed for the expression of β1 integrin in neurospheres by western blot analysis and observed a 30% decrease of β1 integrin expression in TNR^−/− neurospheres when compared with neurospheres derived from TNR^+/+ mice (Fig. 3D).

Because β1 integrin is an important part of the signaling mechanisms that modulate NSC maintenance and differentiation through the MAPK pathways (Campos et al., 2004), we analyzed whether the reduction of β1 integrin expression would affect signal transduction through the MAPK pathway. Western blot analysis revealed that phosphorylation of p38 MAPK was significantly increased in TNR^−/− versus TNR^+/+ neurospheres, whereas no changes were observed in the overall p38 MAPK protein levels. No difference was detected between TNR^−/− and TNR^+/+ neurospheres in the levels of p44/42 MAPK and in the phosphorylation of p44/42 MAPK (Fig. 3F).

Proneural basic helix-loop-helix factors play important roles in the specification of neural cell types (Guillemot, 2007). Ablation of β1 integrin in embryonic stem cells leads to an increased expression of the lineage-specific gene Ascl1 (also known as Mash1), which encodes a basic helix-loop-helix transcription factor that plays an important role in the specification of GABAergic interneurons and favors increased neuronal differentiation (Rohwedel et al., 1998). Given that TNR^−/− mice have increased numbers of GABAergic interneurons and reduced expression of β1 integrin, we hypothesized that the effects of TNR deficiency on neuronal differentiation are mediated by ASCL1.

Thus, we first studied the localization of ASCL1 and β1 integrin in E14 mouse brains. ASCL1-expressing cells in the wall of the lateral ventricles at the levels of the ganglionic eminence co-expressed β1 integrin (Fig. 4A). Many of the ASCL1-immunoreactive cells co-expressed Ki67 in the proliferative zones of the ganglionic eminence (Fig. 4B), where TNR is also expressed (Fig. 2A). Colocalization of ASCL1 with β1 integrin was also confirmed in neurospheres derived from E14 mouse brains (Fig. 4C). Western blot analysis of cell lysates of neurospheres derived from the E14 brains showed a 70% increase of ASCL1 expression in TNR^−/− versus TNR^+/+ neurospheres (Fig. 4D). Immunohistochemical analysis revealed that this increase in ASCL1 expression in TNR^−/− neurospheres was accompanied by enhanced nuclear localization of ASCL1 (supplementary material Fig. S1).

Given the conspicuous increase in the number of inhibitory interneurons in the dentate gyrus of TNR^−/− mice, we examined how the lack of TNR affected neurogenesis in the ganglionic eminence. TNR^−/− and TNR^+/+ animals were analyzed at E14–E16 – at the peak of the generation of inhibitory interneurons.

Increased neurogenesis in the absence of TNR could be caused by decreased cell death, increased NSC proliferation, increased neuronal differentiation at the expense of glial differentiation or a combination of these factors. Immunohistochemical analysis for activated caspase-3 at E15.5 and E18 did not show differences in apoptosis between TNR^−/− and TNR^+/+ mice, indicating that a lack of TNR did not affect cell death during embryonic development in vivo (supplementary material Fig. S2).

By contrast, proliferation in the ganglionic eminence as determined by counting of Ki67⁺ cells was reduced by 30% in TNR^−/− mice when compared with TNR^+/+ mice (Ki67⁺ cells: TNR^+/+, 113,900±11,100; TNR^−/−, 79,200±5200) (Fig. 5A). This decreased proliferation was confirmed in NSCs derived from E14 TNR^−/− brains versus NSCs derived from TNR^+/+ brains in vitro (Fig. 5A). Thus, deficiency in TNR leads to decreased proliferation of NSCs during development both in vitro and in vivo.

To examine the cellular mechanism that leads to decreased proliferation of NSCs in TNR^−/− mice, we assessed cell-cycle exit and re-entry by BrdU-incorporation experiments. At E14, mice of both genotypes were labeled with a pulse of BrdU, which is incorporated into the DNA during the S-phase of the cell cycle. At E15, 24 hours after the BrdU pulse labeling, immunohistochemical analysis with antibodies against Ki67 and BrdU was performed. TNR^−/− mice had a 27% higher ratio of cells that had exited the cell cycle (BrdU⁺Ki67⁻ cells / BrdU⁺ cells) when compared with TNR^+/+ controls (Fig. 5B). This finding suggests that TNR favors cell cycle re-entry and impedes differentiation of NSCs.

The influence of TNR on neuronal differentiation was further assessed by counts of newly generated neurons in the ganglionic eminence. At E14, TNR^−/− mice and TNR^+/+ littermates were pulse-labeled with BrdU. Two days after BrdU administration (E16), the number of newly generated neurons, identified as Tuj1 and BrdU double-labeled cells, was increased by 25% in TNR^−/− mice when compared with TNR^+/+ mice (BrdU⁺/Tuj1⁺ cells: TNR^+/+, 552±22; TNR^−/−, 669±15) (Fig. 5C). These findings were confirmed in NSCs cultured in vitro (Fig. 5C), indicating that TNR deficiency increases neuronal differentiation of NSCs both in vitro and in vivo.

To specifically assess whether the number of newly generated cells committed to the GABAergic cell fate was increased by TNR deficiency, we labeled newly generated cells with antibodies against GAD65/67 and BrdU 2 days after BrdU injection. Quantification showed that there was an about 20% increase in the number of GAD65/67⁺/BrdU⁺ double-labeled cells in the dentate gyrus of TNR^−/− mice compared with TNR^+/+ littermate mice (GAD65/67⁺/BrdU⁺ cells: TNR^+/+, 299±16; TNR^−/−, 373±28) (Fig. 5D), indicating that TNR deficiency increases GABAergic differentiation in the developing brain.

It has been suggested that adult NSCs are influenced by GABAergic interneurons that are immunoreactive for the calcium-binding protein PV (Danglot et al., 2006). To gain more information on the features of GABAergic interneurons that potentially interact with hippocampal NSCs, we performed double immunofluorescence labeling with antibodies against PV and Ki67, and found that PV⁺ axon terminals were closely associated with Ki67⁺ NSCs (Fig. 6A,B). Stereological analysis showed a 67% increase in the total number of PV⁺ interneurons in adult TNR^−/− mice compared with TNR^+/+ littermates (Fig. 6C). To test whether an increase in PV⁺ interneurons leads to alteration of the GABAergic perisomatic coverage of dividing hippocampal NSCs immunoreactive for PSA-NCAM, immunohistochemical analysis was performed to distinguish GABAergic terminals estimated by the expression of the vesicular GABA transporter VGAT on PSA-NCAM⁺ NSCs. Analysis of GABAergic perisomatic inputs to PSA-NCAM⁺ NSCs showed a 20% increase in TNR^−/− mice compared with TNR^+/+ mice (Fig. 6D). Thus, TNR deficiency is associated with increased numbers of PV⁺ interneurons in the dentate gyrus, leading to increased GABAergic input to hippocampal NSCs.

GABAergic signaling increases neuronal differentiation of NSCs in the adult dentate gyrus and decreases proliferation of NSCs in the adult subventricular zone (Liu et al., 2005; Tozuka et al., 2005). Thus, we first evaluated proliferation by counting the number of Ki67⁺ cells in the dentate gyrus. TNR^−/− mice showed a 70% reduction in the overall number of proliferating cells compared with TNR^+/+ mice (Ki67⁺ cells: TNR^+/+, 3600±800; TNR^−/−, 1120±300) (Fig. 7A). Apoptotic cells were rarely detected in the adult dentate gyrus and, thus, there was no difference in the number of cells undergoing apoptosis in the adult dentate gyrus of TNR^−/− mice and TNR^+/+ littermates at postnatal days 120 and 540 (supplementary material Fig. S2).

To evaluate whether increased numbers of PV⁺ interneurons and enhanced GABAergic innervation led to an alteration in neuronal differentiation of adult hippocampal progenitor cells in TNR^−/− mice, proliferating NSCs were labeled with BrdU. Mice were sacrificed 28 days after BrdU injection. The neuronal progeny of hippocampal NSCs immunostained with antibodies against BrdU and NeuN was determined. Compared with TNR^+/+ mice, TNR^−/− mice showed increased numbers of BrdU⁺/NeuN⁺ neurons (BrdU⁺/NeuN⁺ cells: TNR^+/+, 1134±126; TNR^−/−, 1692±189) (Fig. 7B). Thus, TNR deficiency decreased proliferation and increased neuronal differentiation of NSCs in the adult dentate gyrus.

Intrinsic programs and extrinsic factors imposed by the local extracellular environment tightly control the fate determination of NSCs (Barnabé-Heider et al., 2005; Ma et al., 2008; Seuntjens et al., 2009; Shen et al., 2006; Tanentzapf et al., 2007). In the present study we identified TNR, an extracellular matrix glycoprotein that is expressed in the developing and adult nervous system, as an important factor in regulating the generation of GABAergic interneurons and granule neurons in the developing and adult murine dentate gyrus.

The majority of GABAergic neurons in rodents arise from the medial and caudal eminence (Danglot et al., 2006). We show that during embryonic development TNR is not only strongly expressed in the developing murine cortex, but also in the medial and caudal ganglionic eminence in close proximity to proliferating cells. Notably, TNR is also expressed in the developing human cerebral cortex in a spatio-temporally controlled pattern (El Ayachi et al., 2011). Ablation of TNR leads to increased numbers of GABAergic and PV⁺ interneurons in the dentate gyrus of embryonic mice. It is interesting in this context that TNR stimulates the detachment of migrating neuroblasts in the rostral migratory stream in an activity-dependent manner, and promotes their migration and differentiation into inhibitory neurons in the murine olfactory bulb (Saghatelyan et al., 2004a). These findings are compatible with our present results indicating that TNR signals to NSCs in the subgranular zone of the hippocampus to regulate the generation of inhibitory interneurons. Deficiency in TNR has been shown to lead to severe cognitive deficits in mice (Montag-Sallaz and Montag, 2003) and intellectual impairment in a human patient (Dufresne et al., 2012). These deficits could be caused by increased GABAergic neurons, and especially PV⁺ interneurons, leading to an imbalance between excitatory and inhibitory neurotransmission.

Other extracellular matrix molecules have been shown to participate in the regulation of neurogenesis in the neurogenic niche. Ablation of tenascin-C (TNC) in the neurogenic niche promotes the generation of neurons and influenced proliferation of oligodendrocyte precursors, radial glial cells, and retinal stem/progenitor cell by regulating the cell-cycle exit (Besser et al., 2012; Czopka et al., 2009; Garcion et al., 2001; Garcion et al., 2004). Interestingly, TNC^−/− mice show a reduction in expression of ASCL1 (Mash1) by retinal stem/progenitor cells (Besser et al., 2012), in contrast to our study where TNR deficiency resulted in increased ASCL1 expression in NSCs in the ganglionic eminence. Enzymatic degradation of chondroitin sulfate expressed by proteoglycans in the stem cell niche promotes differentiation and migration of neural progenitor cells (Gu et al., 2009), indicating an influence of these extracellular matrix molecules on neurogenesis. The importance of the extracellular matrix in regulating neurogenesis has been supported by studies on matrix metalloproteinases, which can regulate neurogenesis by proteolytic modulation of extracellular matrix molecules (Bovetti et al., 2007; Kang et al., 2008; Tonti et al., 2009; Zhang et al., 2007). Deficiency in tenascin-R affects developmental neurogenesis in the mouse olfactory bulb, but not neurogenesis in the adult, thus highlighting that regulation of neurogenesis can vary during an animal's lifetime (David et al., 2013).

In the dentate gyrus of the hippocampus, granule neurons are generated during development later than inhibitory interneurons (Danglot et al., 2006). Moreover, in the adult forebrain, the dentate gyrus is one of the few areas where a continuous supply of newborn neurons occurs (Tozuka et al., 2005). Although the signaling mechanisms that regulate neurogenesis in the adult dentate gyrus have remained largely unknown, increasing evidence indicates that the proliferation and differentiation of NSCs in the dentate gyrus are influenced by neurotransmitters. Importantly, NSCs in the dentate gyrus receive GABAergic input (Bhattacharyya et al., 2008) that promotes their differentiation into neurons (Tozuka et al., 2005). Furthermore, GABAergic signaling decreases the proliferation of NSCs in the subventricular zone of the adult brain (Liu et al., 2005). Potential candidates among the diverse populations of interneurons that may be responsible for the GABAergic influence on NSCs are PV⁺ interneurons (Bhattacharyya et al., 2008; Tozuka et al., 2005). Accumulating evidence indicates that TNR, a component of perineuronal nets (Dityatev et al., 2010) influences synaptic plasticity and GABAergic signaling in the hippocampus (Saghatelyan et al., 2001). Given that the number of GABAergic neurons, and in particular PV⁺ interneurons, is increased in the dentate gyrus of adult TNR^−/− mice, leading to enhanced GABAergic contacts with dividing NSCs, it seems reasonable to suggest that TNR deficiency promotes neuronal differentiation and inhibits proliferation of hippocampal NSCs in the adult dentate gyrus by increased GABAergic signaling, thus explaining the increase in the number of excitatory granule neurons in the dentate gyrus of TNR^−/− mice. In view of the close association of TNR and proliferating cells in the adult hippocampus of TNR^+/+ mice, it is likely that TNR directly contributes to neurogenesis in the adult, similar to what we observed in embryonic mice. It is noteworthy that no effect of TNR on adult neurogenesis was observed in the olfactory system (David et al., 2013).

The mechanism by which decreased proliferation of NSCs contributes to increased numbers of excitatory granule neurons is presently not known. It is interesting in this context that decreased proliferation of hippocampal NSCs is associated with increased neuronal differentiation (Chen et al., 2012; Jung et al., 2008; Li et al., 1998; Spella et al., 2011; Theriault et al., 2005; Yabut et al., 2010; Zhao et al., 2010) and that 80% of newly generated neurons die within the first month after generation in the adult hippocampus (Kempermann et al., 2003). Thus, decreased apoptosis of newly generated neurons might have contributed to the increase in numbers of granule neurons in the hippocampus of adult TNR^−/−mice. In a recent study, PV⁺ neurons were shown to regulate fate decision in the adult murine hippocampus in that optogenetic activation of PV⁺ interneuron activity affects fate decisions of NSCs by decreasing their proliferation (Song et al., 2012). Furthermore, alterations in long-term potentiation in TNR^−/− mice could be rescued by pharmacological inhibition with GABA_A receptor antagonists (Morellini et al., 2010). Thus, we suggest that increased numbers of GABAergic neurons and in particular PV⁺ interneurons, being generated in the developing hippocampus of TNR^−/− mice, contribute to the maintenance of increased neurogenesis in the adult TNR^−/− hippocampus, which also leads to the generation of principal neurons.

Extracellular matrix molecules exert their regulatory effect on NSCs through intracellular signaling pathways mediated by interaction with cell surface receptors (Brown, 2011; Fietz et al., 2012; Hynes, 2009; Tanentzapf et al., 2007). TNR has been shown to associate with sodium channel β-subunits (Xiao et al., 1999), the cell surface receptor F3/contactin-1 (Pesheva et al., 1993), acetylated gangliosides (Probstmeier et al., 1999) and receptor protein tyrosine phosphatase (RPTP)-zeta/beta (Milev et al., 1998; Xiao et al., 1997). In addition, a cognate receptor interacting with TNR in NSCs is β1 integrin, which is a prominent sensor of signaling from extracellular matrix components expressed by neural cells (Liao et al., 2008). β1 integrins are highly expressed in murine neurospheres (Gu et al., 2009), and more than 90% of human embryonic-stem-cell-derived NSCs express β1 integrins (Ma et al., 2008). Blocking the functions of β1 integrin inhibits proliferation of NSCs (Ma et al., 2008). In the present study, we show that TNR colocalizes with β1 integrin in the developing medial and caudal ganglionic eminence. Furthermore, we show that TNR and β1 integrin co-immunoprecipitate, most likely through their EGFL or FN6-8 domains (Liao et al., 2008) during development, suggesting that they cooperate in determining neurogenesis in the developing brain. It will be important to analyze which domains determine the cell fates of hippocampal NSCs and whether there are differences in β1 integrin signaling in development and in adulthood. Integrins mediate cellular responses via intracellular signaling cascades and regulation of gene expression (Campos et al., 2004; Tanentzapf et al., 2007). The influence of β1 integrins on NSC maintenance is regulated, at least to some extent, through the MAPK signaling pathway (Campos et al., 2004). First steps towards analysis of the molecular mechanisms underlying TNR signaling pathways in TNR^−/− mice revealed reduced β1 integrin expression that was accompanied by an upregulation of p38 MAPK and an enhanced expression of the transcription factor ASCL1. This observation is in agreement with a study showing that loss of β1 integrin function leads to increased ASCL1 expression (Rohwedel et al., 1998), which in turn modulates GABAergic differentiation (Berninger et al., 2007; Guillemot, 2007). ASCL1 is expressed by NSCs in the stem cell niches of the medial and caudal ganglionic eminence, and leads to cell cycle exit and neuronal differentiation of NSCs (Nieto et al., 2001). In the present study, upregulation of endogenous ASCL1 suggests an activation of this pathway in TNR^−/− mice, thereby leading to decreased proliferation through increased levels of ASCL1, increased cell cycle exit and increased GABAergic neuronal differentiation in the absence of TNR. It thus seems plausible to suggest that the observed effects of TNR deficiency during development of the hippocampus are caused by the reduction of β1 integrin expression, which appears to be an important component in lineage-specific signaling pathways.

Taken together, we propose that TNR functions as a negative-feedback signaling molecule that regulates the generation of appropriate numbers of GABAergic neurons in the dentate gyrus during development. TNR secreted by inhibitory interneurons might influence the fate of NSCs by binding to β1 integrin, regulating intracellular p38 MAPK signaling and expression of ASCL1, thereby controlling GABAergic neuronal differentiation during development. In the adult, direct input to NSCs from GABAergic neurons, and in particular PV⁺ neurons that are increased in TNR^−/− mice, promotes the differentiation of NSCs into excitatory granule neurons (Tozuka et al., 2005), an observation that might explain the increased numbers of granule neurons in TNR^−/− mice. Furthermore, direct effects of TNR as shown during development could contribute to the regulation of adult neurogenesis in the hippocampus. A schematic view of our proposed model is shown in Fig. 8. On the basis of this model, further studies should identify whether TNR is a potential target for mimetic small molecules that might beneficially influence the abnormal development of interneurons in the mammalian dentate gyrus in diverse neurological disorders.

The generation of TNR-deficient (TNR^−/−) mice and wild-type (TNR^+/+) littermates used in this study has been described (Weber et al., 1999). Numbers and age of the animals studied in different experiments are given in the figure legends. Experiments were conducted in accordance with the German and European Community laws on protection of experimental animals and approved by the local authorities of the City of Hamburg.

Neural stem cells (NSCs), defined in this article as neural stem or progenitor cells, were isolated from the brains of C57BL/6J mice at embryonic day 14 (E14) as described (Dihné et al., 2003). Briefly, ganglionic eminences were removed, mechanically dissociated using a pipette, and maintained in a defined medium composed of a 1∶1 mixture of Dulbecco's Modified Eagle's Medium and F-12 supplemented with glucose (0.6%, Sigma-Aldrich, Deisenhofen, Germany), sodium bicarbonate (3 mM, Invitrogen, Karlsruhe, Germany), B27 (2%, Invitrogen), glutamine (2 mM; Invitrogen) and HEPES buffer (5 mM, Sigma-Aldrich). The medium was supplemented with epidermal growth factor (EGF; 20 ng/ml; PreproTech, Rocky Hill, NY), and basic fibroblast growth factor (FGF-2; 20 ng/ml; PreproTech). The cells were maintained as neurospheres which were then passaged by mechanical dissociation approximately every fifth day and reseeded at a density of 50,000 cells/ml. Experiments were performed with neurospheres taken from passages 3 and 6.

For measurement of NSC proliferation, dissociated NSCs were plated onto 15 mm glass coverslips coated with 0.01% poly-L-lysine (PLL, Sigma-Aldrich). After 5 days in culture, 5-bromo-2-deoxyuridine (BrdU) (10 µM; Sigma-Aldrich) was administered in EGF and FGF-2 containing culture medium 8 hours before the cells were fixed with 4% formaldehyde. The percentage of BrdU-positive cells was determined by immunocytochemical analysis with monoclonal rat antibody against BrdU (1∶100; Abcam, Cambridge, MA). For analysis of differentiation, neurospheres were mechanically dissociated and plated at a density of 50,000 cells/ml onto PLL-coated coverslips. Precursor cells were first maintained in an undifferentiated state for 2 days after plating in EGF and FGF2 containing serum-free culture medium. Growth factors were then removed, and precursor cells were allowed to differentiate for an additional 7 days. Then, coverslips were washed in phosphate-buffered saline, pH 7.3 (PBS), and cells were fixed for 30 minutes with 4% formaldehyde in PBS.

For immunocytochemical analysis of fixed cells, cultured cells were washed in PBS, and fixed in 4% formaldehyde. For immunohistochemical analysis of hippocampus, postnatal and adult mice were anesthetized and transcardially perfused with fixative consisting of 4% formaldehyde and 0.1% CaCl₂ in 0.1 M cacodylate buffer (Sigma-Aldrich), pH 7.3, for 15 minutes at room temperature. Brains were removed, incubated in 4% formaldehyde for 24 hours followed by incubation in 15% sucrose in PBS for 48 hours at 4°C. 25-µm-thick serial coronal sections were cut in a cryostat (CM3050; Leica, Nussloch, Germany) and collected on SuperFrost Plus glass slides (Roth, Karlsruhe, Germany). Mouse embryos were fixed in ice-cold 4% formaldehyde for 2 hours. After fixation, the tissue was washed extensively in ice-cold PBS and transferred to 15% sucrose solution overnight at 4°C.

Antigen retrieval was performed by incubation of the tissue sections in 0.01 M sodium citrate solution, pH 9.0, for 30 minutes at 80°C followed by blocking with PBS containing 0.2% v/v Triton X-100, 0.02% w/v sodium azide (Sigma-Aldrich) and 5% v/v appropriate non-immune serum (Jackson ImmunoResearch via Dianova, Hamburg, Germany) for 30 minutes. Primary antibodies diluted in PBS containing 0.5% w/v λ-carrageenan (Sigma-Aldrich) were applied for 48 hours at 4°C. After washing in PBS, secondary antibodies were applied for 1 hour at room temperature. Finally, coverslips were washed twice with PBS. To determine total cell numbers in vitro, cells were counterstained with DAPI (Sigma-Aldrich) for 2 minutes and the ratio of cell type-specific marker-positive cells of all DAPI-positive cells was calculated. For BrdU staining, DNA was denatured with 2 M HCl for 30 minutes at 37°C. Monoclonal rat antibody against BrdU (1∶100; Abcam) was applied overnight at 4°C. For negative controls, primary antibody was omitted.

For cryosectioning of neurospheres, they were fixed in 4% formaldehyde in PBS for 30 minutes at room temperature. After fixation, the neurospheres were incubated in 30% sucrose in PBS for 4 hours at 4°C. Finally, 14-µm-thick cryosections were prepared.

Primary antibodies used were mouse monoclonal antibodies against ASCL1 (1∶200, BD Pharmingen, CA), NeuN (1∶1,000, Millipore, MA), nestin (1∶50; Developmental Studies Hybridoma Bank, Iowa City, IA), parvalbumin (PV) (clone PARV-19, 1∶1000, Sigma-Aldrich), polysialylated neural cell adhesion molecule [PSA-NCAM, 1∶1000, Millipore), TNR (clone 619 (Morganti et al., 1990)], polyclonal rabbit antibodies against β1 integrin (1∶200, Chemicon, Temecula, CA), caspase-3 (1∶2000, R&D Systems, Minneapolis, MN), glutamate decarboxylase (GAD) 65 and 67 (1∶500, Sigma-Aldrich), Ki67 (1∶500, Abcam), neuronal class III β-tubulin (Tuj1) (1∶2000; Covance, Berkeley, CA) and VGAT (anti-vesicular GABA transporter, 1∶1000, VGAT, Synaptic Systems, Göttingen, Germany). Cyanine 2 (Cy2)-, Cy3- and Cy5-conjugated secondary antibodies (Dianova) were used to detect primary antibodies.

Numerical densities were estimated using the optical dissector method as described (Nikonenko et al., 2006). Counting was performed on an Axioskop microscope (Carl Zeiss Microimaging) equipped with a motorized stage and Neurolucida software-controlled computer system (MicroBrightField Europe, Magdeburg, Germany). The volume of the granule cell layer was estimated using spaced serial sections (250 µm interval) and the Cavalieri principle.

Analysis of cell cycle exit was performed as described previously (Lien et al., 2006). Briefly, mice received an intraperitoneal injection of 100 mg/kg BrdU in PBS. 24 hours later, mice were perfused and cryosections were cut as described above and immunostained with antibodies against Ki67 and BrdU. BrdU⁺ cells that had left the cell cycle after the BrdU injection did not express Ki67 and could thus be identified as BrdU⁺/Ki67⁻, whereas the cells that remained in an active cell cycle could be identified as BrdU⁺/Ki67⁺.

Estimation of perisomatic puncta in the dentate gyrus was performed as described (Nikonenko et al., 2006) with a slight modification. Stacks of images of 1 µm thickness were obtained using sections double-stained for PSA-NCAM and VGAT using a TCS SP2 confocal microscope (Leica). One merged image (red and green channel) per cell at the level of the largest cross-sectional area of the cell body was used to measure soma area and area of perisomatic puncta. Area of VGAT-positive puncta around each PSA-NCAM positive cell was measured. All measurements were performed using the Image Tool 2.0 software (University of Texas Health Science Center, San Antonio, TX).

Brain protein extracts were generated by lysing brain with RIPA buffer for 1 hour at 4°C. Protein extracts were cleared with protein A/G agarose beads (Santa Cruz Biotechnology) for 3 hours at 4°C and then incubated with anti-β1-integrin antibody (1∶1000, Chemicon) or non-immune IgG (1∶1000, Chemicon) overnight at 4°C. Protein-A/G agarose was added to capture the immunocomplexes for 6 hours at 4°C under constant agitation. Immunoprecipitated proteins were eluted from agarose beads by 2× SDS sample buffer (125 mM Tris-HCl, 4% SDS, 30% glycerol, 10% β-mercaptoethanol and 0.00625% Bromophenol Blue, pH 6.8).

SDS-PAGE and transfer were performed according to standard laboratory protocols. Membranes were probed with antibodies against ASCL1 (1∶500, BD Pharmingen), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1∶1000; Millipore), TNR [clone 619 (Morganti et al., 1990)], actin (1∶1000, Sigma-Aldrich), β1 integrin (1∶1000, Chemicon), phospho-p38 (Thr180/Tyr182), phospho-p44/42 (Thr202/Tyr204) or total p38 and total p44/42 antibodies (1∶1000; Cell Signaling Technology, Danvers, MA, USA). Immunoreactivity was visualized using the enhanced chemiluminescence detection system (ECL, Thermo Fisher Scientific, Bonn, Germany).

All numerical data are presented as means ± s.e.m. Statistical analyses were performed as indicated in the text and figures using the SPSS 11.5 software package (SPSS, Chicago, IL).

The authors are grateful to Eva Kronberg for animal care, Dr Fabio Morellini and Achim Dahlmann for animal care and genotyping. Melitta Schachner is New Jersey Professor of Spinal Cord Research and is supported by the Li Kashing Foundation at Shantou University Medical College, China.