Adult neurogenesis, the continuous process of generating new neurons from neural progenitors throughout life, is predominately restricted to two adult brain regions in mice: the ventricular-subventricular zone (V-SVZ) of the lateral ventricles and the dentate gyrus (DG) of the hippocampus, in which adult-born neurons contribute to olfactory bulb and hippocampal functions, respectively (Bond et al., 2015; Lim and Alvarez-Buylla, 2016; Deng et al., 2010). Adult hippocampal neurogenesis relies on a population of multipotent radial glial-like neural stem cells (NSCs) that have both neurogenic and gliogenic potential (Steiner et al., 2004; Encinas et al., 2011; Bonaguidi et al., 2011; Pilz et al., 2018). The majority of NSCs remain out of the cell cycle in a reversible quiescent state (Doetsch et al., 1999; Andersen et al., 2014; Codega et al., 2014; Urbán et al., 2016). An important regulatory step of adult neurogenesis is the transition of NSCs from quiescence to activation (reviewed by Kempermann et al., 2015). Too little stem cell activation results in an insufficient number of new neurons being generated (Andersen et al., 2014). By contrast, excessive activation generates a transient burst in neurogenesis followed by a sharp decline resulting from the depletion of NSCs, which have limited self-renewal capacity (Paik et al., 2009; Renault et al., 2009). Therefore, tight regulation of the transition between quiescent and active NSC states is crucial to ensure the long-term maintenance of the NSC pool and neurogenesis. Niche-derived signals play an important role in regulating this transition (Fuentealba et al., 2012; Choe et al., 2016; Imayoshi et al., 2010; Lie et al., 2005; Mira et al., 2010; Petrova et al., 2013; Qu et al., 2010). For example, Notch and BMP signalling have an integral role in maintaining NSC quiescence (Lavado and Oliver, 2014; Ables et al., 2010; Bonaguidi et al., 2008; Mira et al., 2010). However, less is known about the niche-derived signals that promote the activation of quiescent NSCs.

Wnt ligands and antagonists are expressed by multiple cell types within the DG niche, including NSCs (Lie et al., 2005; Qu et al., 2013; Jang et al., 2013; Seib et al., 2013). In the absence of Wnt, β-catenin is phosphorylated for degradation by GSK3β (Daugherty and Gottardi, 2007). Upon Wnt ligand binding, β-catenin is stabilised and translocates to the nucleus, where it forms a complex with TCF/LEF transcription factors to activate Wnt/β-catenin target genes, such as Axin2 (Lustig et al., 2002; Nusse and Clevers, 2017). Wnt/β-catenin signalling is differentially active in cells along the neurogenic lineage and regulates both progenitor proliferation and newborn neuron maturation (Lie et al., 2005; Qu et al., 2013; Wexler et al., 2009; Kuwabara et al., 2009; Seib et al., 2013; Jang et al., 2013; Rosenbloom et al., 2020; Heppt et al., 2020).

Stimulating Wnt/β-catenin signalling in the DG promotes proliferation and neurogenesis (Lie et al., 2005; Seib et al., 2013; Jang et al., 2013). The absence of the Wnt inhibitor Dickkopf (DKK1) results in a specific increase in the generation of neuronally committed intermediate progenitor cells (Seib et al., 2013). However, the absence of another Wnt inhibitor, SFRP3, stimulates proliferation but does not favour a neurogenic lineage choice of NSCs (Jang et al., 2013). Inhibiting Wnt/β-catenin signalling has also yielded contrasting results, because it was found to impair the generation of newborn neurons in vivo but induce neuronal differentiation in vitro (Qu et al., 2013, 2010; Wexler et al., 2009; Kuwabara et al., 2009; Lie et al., 2005). These discrepancies could result from the use of different experimental approaches to modulate Wnt/β-catenin signalling, such as overexpressing a Wnt ligand versus deleting a Wnt inhibitor, or from the use of in vitro versus in vivo approaches (Lie et al., 2005; Seib et al., 2013; Jang et al., 2013; Kuwabara et al., 2009; Wexler et al., 2009). The majority of these studies have used systemic or hippocampal-wide modulation of Wnt/β-catenin signalling, which do not discriminate between cell-autonomous and non-cell-autonomous effects; neither do they allow identification of the step(s) in the NSC lineage in which Wnt/β-catenin signalling acts (Lie et al., 2005; Seib et al., 2013; Jang et al., 2013; Qu et al., 2013). In addition, the use of constitutive knockout mice makes it difficult to distinguish between a developmental and an adult neurogenesis phenotype (Seib et al., 2013; Jang et al., 2013; Qu et al., 2013, 2010). Furthermore, earlier studies did not robustly distinguish between NSCs with radial morphology (NSCs) and intermediate progenitor cells (IPCs); therefore the role of Wnt/β-catenin signalling in NSCs, particularly in their transition between active and quiescent states, remains unclear (Kuwabara et al., 2009; Seib et al., 2013).

Here, we used genetic and pharmacological tools to measure and manipulate Wnt/β-catenin signalling specifically in active and quiescent adult NSCs both in vivo and in vitro. We show that both quiescent and active NSCs respond to Wnt/β-catenin signalling, that the response of NSCs to Wnt/β-catenin stimulation is dose and cell state specific, but that Wnt/β-catenin signalling is not essential for cell-autonomous NSC homeostasis or for the ability of NSCs to proliferate and generate neuronal progeny. Together, these findings reconcile some of the current contradictions relating to the role of Wnt/β-catenin signalling in adult hippocampal NSCs.

To study the role of Wnt/β-catenin in the transition of hippocampal stem cells from quiescence to activation, we first investigated whether quiescent and active NSCs express components of the Wnt/β-catenin signalling pathway. Although expression of Wnt/β-catenin signalling components in adult hippocampal NSCs has been reported in previously published single-cell sequencing datasets, too few NSCs were identified in most of these datasets to allow a robust comparison between quiescent and active states (Hochgerner et al., 2018). We re-analysed a previously generated single-cell RNA-sequencing dataset containing 2947 NSCs (Fig. 1A) (Harris et al., 2021). We found that both quiescent and active NSCs heterogeneously expressed components of the Wnt/β-catenin signalling pathway, as well as Wnt ligands (Wnt7a and Wnt7b) and Wnt inhibitors (Dkk3 and Sfrp1), suggesting that NSCs respond to, and directly regulate, Wnt activity levels in the DG (Fig. 1A).

Fig. 1.

NSCs in vivo respond to Wnt/β-catenin signalling independently of their activation state. (A) Heatmap of publicly available (GSE159768) single-cell RNA-sequencing data showing the expression of Wnt/β-catenin pathway components from in vivo hippocampal quiescent and active NSCs (Harris et al., 2021). (B) BATGAL and GFAP immunolabelling in the DG of a 2-month-old BATGAL Wnt/β-catenin reporter mouse (Maretto et al., 2003). Arrows indicate BATGAL+ NSCs (SGZ cell body and GFAP+ radial process). The white-boxed area is enlarged in the adjacent panel to show three BATGAL+ NSCs. White-dashed lines denote the SGZ. (C) Quantification of the proportion of BATGAL+ cells in the DG of 2-month-old BATGAL mice, showing NSCs (GFAP+ NSCs, 29±5.44%; n=4), IPCs (GFAP− SOX2+, 1.5±0.6%; n=4), neuroblasts (NBs, DCX+, 11.5±0.5; n=2), neurons (NeuN+, 55.67±4.44%, n=3) and astrocytes (GFAP+ SOX2−, 16±4.44%, n=3). (D) BATGAL immunolabelling in Ki67+ NSCs in 2-month-old BATGAL mice. White-dashed lines denote Ki67+ BATGAL+ NSCs and Ki67+ BATGAL− NSCs. (E) Quantification of the proportion of Ki67+ immunolabelling in BATGAL+ (2±1%) and BATGAL− NSCs (1.33±0.99%) shown in D. n=3. (F) BATGAL, GFAP/SOX2 and Id4 immunolabelling in the DG of 2-month-old BATGAL mice. Single arrowheads indicate Id4-positive BATGAL-positive NSCs. Double arrowheads indicate Id4-positive, BATGAL-negative NSCs. (G) Quantification of the data shown in F. n=3. Data were analysed as follows: Student's t-test (A), ordinary one-way ANOVA with Tukey's multiple comparisons test (C) and unpaired two-tailed Student's t-test (E); ns, P>0.05, **P<0.01, ***P<0.001. Data are mean±s.e.m. A.U., arbitrary units. Scale bars: 20 µm in D; 50 µm in B,F.

Fig. 1.

NSCs in vivo respond to Wnt/β-catenin signalling independently of their activation state. (A) Heatmap of publicly available (GSE159768) single-cell RNA-sequencing data showing the expression of Wnt/β-catenin pathway components from in vivo hippocampal quiescent and active NSCs (Harris et al., 2021). (B) BATGAL and GFAP immunolabelling in the DG of a 2-month-old BATGAL Wnt/β-catenin reporter mouse (Maretto et al., 2003). Arrows indicate BATGAL+ NSCs (SGZ cell body and GFAP+ radial process). The white-boxed area is enlarged in the adjacent panel to show three BATGAL+ NSCs. White-dashed lines denote the SGZ. (C) Quantification of the proportion of BATGAL+ cells in the DG of 2-month-old BATGAL mice, showing NSCs (GFAP+ NSCs, 29±5.44%; n=4), IPCs (GFAP− SOX2+, 1.5±0.6%; n=4), neuroblasts (NBs, DCX+, 11.5±0.5; n=2), neurons (NeuN+, 55.67±4.44%, n=3) and astrocytes (GFAP+ SOX2−, 16±4.44%, n=3). (D) BATGAL immunolabelling in Ki67+ NSCs in 2-month-old BATGAL mice. White-dashed lines denote Ki67+ BATGAL+ NSCs and Ki67+ BATGAL− NSCs. (E) Quantification of the proportion of Ki67+ immunolabelling in BATGAL+ (2±1%) and BATGAL− NSCs (1.33±0.99%) shown in D. n=3. (F) BATGAL, GFAP/SOX2 and Id4 immunolabelling in the DG of 2-month-old BATGAL mice. Single arrowheads indicate Id4-positive BATGAL-positive NSCs. Double arrowheads indicate Id4-positive, BATGAL-negative NSCs. (G) Quantification of the data shown in F. n=3. Data were analysed as follows: Student's t-test (A), ordinary one-way ANOVA with Tukey's multiple comparisons test (C) and unpaired two-tailed Student's t-test (E); ns, P>0.05, **P<0.01, ***P<0.001. Data are mean±s.e.m. A.U., arbitrary units. Scale bars: 20 µm in D; 50 µm in B,F.

Quiescent NSCs expressed Wnt receptors (Fzd1 and Fzd2) at a higher level compared with active NSCs, which corroborates previous reports showing that quiescent NSCs are enriched for cell surface receptors (Fig. 1A) (Shin et al., 2015; Hochgerner et al., 2018; Artegiani et al., 2017; Cheung and Rando, 2013). The Wnt transducer molecule Tcf7l2 is more highly expressed in quiescent than in active NSCs, whereas β-catenin (Ctnnb1) and GSK3β (Gsk3b) are upregulated in active NSCs (Fig. 1A). However, these differences in expression do not translate into a differential expression of Wnt targets between quiescent and active NSCs. Axin2 was expressed at low levels by very few quiescent (5%) and active (7%) NSCs (Fig. 1A), which might suggest that few NSCs respond to Wnt/β-catenin signalling, but could also reflect the limitation of single cell RNA-sequencing in detecting lowly expressed genes (Haque et al., 2017).

The similar Axin2 levels between quiescent and active NSCs suggest that cells in these two states respond similarly to Wnt/β-catenin signalling (Fig. 1A). To further characterise the response of NSCs and other DG cells to Wnt/β-catenin signalling in vivo, we used β-galactosidase (BATGAL) Wnt/β-catenin reporter mice, which express β-galactosidase (Glb1) under the control of 7xTCF/LEF-binding sites (Maretto et al., 2003). We identified NSCs based on their glial fibrillary acidic protein (GFAP) expression, their subgranular zone (SGZ) cell body localisation and their radial process extending through the granule cell layer (GCL) (Fig. 1B). We found that a higher proportion of NSCs (29±5.4%) and neurons (55.7±4.4%) responded to Wnt/β-catenin signalling compared with IPCs (1.5±0.6%), neuroblasts (11.5±0.5%) and astrocytes (16±4.4%) (Fig. 1C), corroborating previous reports (Garbe and Ring, 2012; Heppt et al., 2020). To investigate whether the Wnt/β-catenin response correlates with NSC activation, we quantified proliferation in BATGAL-positive and -negative NSCs (Fig. 1D). The proportion of proliferating NSCs was similar between these two cell states (Fig. 1E). Two different populations can be distinguished within the quiescent NSC pool, with resting NSCs having a higher proliferative potential compared with dormant NSCs (Harris et al., 2021; Urbán et al., 2016). We quantified Id4 immunostaining levels, which are lower in resting compared with dormant NSCs (Harris et al., 2021) alongside BATGAL levels in Id4-positive NSCs (Fig. 1F). Our results showed no correlation between Id4 and BATGAL levels, indicating no difference in Wnt signalling response between resting and dormant NSCs (Fig. 1G). Overall, these data show that, in vivo, NSCs respond similarly to Wnt/β-catenin signalling, independently of their activation state.

To investigate the effects of Wnt/β-catenin inhibition in NSCs in vivo, we generated GlastCreERT2; β-catfl/fl ex3-6; RYFP mice to delete β-catenin conditionally in Glast-expressing NSCs by tamoxifen-inducible, Cre-mediated recombination (Huelsken et al., 2001; Mori et al., 2006; Srinivas et al., 2001). However, the β-catenin allele failed to recombine in NSCs (Fig. S1); therefore, we generated a second β-catenin floxed mouse line: GlastCreERT2; β-catfl/fl ex2-6; RYFP mice (hereafter referred to as β-catdel ex2-6 mice, Fig. S2A) (Brault et al., 2001). The expression of the recombined Ctnnb1 transcript was significantly decreased in YFP-positive fluorescence-activated cell (FAC)-sorted cells from the DG of β-catdel ex2-6 mice compared with controls (Fig. S2B,C), indicating the successful recombination of the β-catfl/fl ex2-6 allele. β-catenin protein levels were low in the DG, especially when compared with the SVZ (Fig. S2D). Nevertheless, we were able to detect β-catenin staining in the radial processes of NSCs in control mice, which was eliminated in recombined (YFP-positive) NSCs (Fig. S2E,F), confirming the deletion of β-catenin protein. However, Axin2 transcript levels were very low and unchanged between YFP-positive FAC-sorted cells from the DG of β-catdel ex2-6 mice compared with controls (Fig. S2G). Given that beta-galactosidase accumulates in the cell, it can be a more-sensitive readout of Wnt activity. Therefore, we crossed β-catdel ex2-6 mice with BATGAL mice and found that the proportion of BATGAL-positive NSCs was reduced in β-catdel ex2-6 mice compared with controls, confirming that the loss of β-catenin impairs Wnt/β-catenin activity in NSCs (Fig. S2H-J). To examine the effect of β-catenin deletion on NSCs, we administered tamoxifen to 2-month-old β-catdel ex2-6 and control mice and performed immunofluorescence analysis 30 days later, focusing on recombined YFP+ cells (Fig. 2A,B). The proportion of Ki67-positive NSCs was not significantly different between β-catdel ex2-6 and control mice, indicating that β-catenin deletion did not perturb NSC proliferation (Fig. 2C). We also quantified the total number of NSCs 30 days (Fig. 2D) and 90 days (Fig. S3A,B) after tamoxifen administration and found no significant difference between genotypes, indicating that NSC maintenance was unaffected by β-catenin deletion. Overall, these data suggest that the proliferation and maintenance of NSCs are unaffected by loss of Wnt/β-catenin signalling.

Fig. 2.

NSCs and adult hippocampal neurogenesis are unaffected by the NSC-specific deletion of β-catenin and inhibition of Wnt/β-catenin signalling. (A) Two-month-old control and β-catdel ex2-6 mice were administered tamoxifen for 5 consecutive days and sacrificed 30 days after the first tamoxifen injection. (B) YFP, GFAP, SOX2 and Ki67 immunolabelling in the DG of control and β-catdel ex2-6 mice 30 days after tamoxifen administration. White-dashed lines denote the SGZ. (C-F) Quantification of the data shown in B. C: proportion of Ki67+ NSCs (control versus β-catdel ex2-6: 3.33±0.61% versus 3.333±0.76%). D: total number of NSCs (YFP+ GFAP+ SOX2+ radial cells in the SGZ) normalised to the length of the SGZ (mm) (control versus β-catdel ex2-6: 21.2±1.73 versus 23.56±1.52). E: proportion of proliferating (Ki67+) IPCs (SOX2+ YFP+ GFAP− cells in the SGZ; control versus β-catdel ex2-6: 56±4.56% versus 47.67±4.03%). F: total number of IPCs normalised to the length of the SGZ (mm) (SOX2+ YFP+ GFAP− cells in the SGZ; control versus β-catdel ex2-6: 20.18±2.31 versus 22.43±2.62). n=6. (G) YFP, TBR2 and DCX immunolabelling in the DG of control and β-catdel ex2-6 mice 30 days after tamoxifen administration. (H,I) Quantification of the data shown in G. H: total number of TBR2+ IPCs normalised to the SGZ length (mm) (TBR2+ YFP+ cells in the SGZ; control versus β-catdel ex2-6: 14.29±1.51 versus 17.16±3.09). I: total number of neuroblasts (NBs) normalised to the SGZ length (mm) (DCX+ YFP+ cells; control versus β-catdel ex2-6: 50.72±4.49 versus 50.53±5.69). n=6 and n=3 for control and β-catdel ex2-6 mice, respectively. Data analysed using unpaired two-tailed Student's t-test; ns, P>0.05. Scale bars: 50 µm in B,G. Data are mean±s.e.m.

Fig. 2.

NSCs and adult hippocampal neurogenesis are unaffected by the NSC-specific deletion of β-catenin and inhibition of Wnt/β-catenin signalling. (A) Two-month-old control and β-catdel ex2-6 mice were administered tamoxifen for 5 consecutive days and sacrificed 30 days after the first tamoxifen injection. (B) YFP, GFAP, SOX2 and Ki67 immunolabelling in the DG of control and β-catdel ex2-6 mice 30 days after tamoxifen administration. White-dashed lines denote the SGZ. (C-F) Quantification of the data shown in B. C: proportion of Ki67+ NSCs (control versus β-catdel ex2-6: 3.33±0.61% versus 3.333±0.76%). D: total number of NSCs (YFP+ GFAP+ SOX2+ radial cells in the SGZ) normalised to the length of the SGZ (mm) (control versus β-catdel ex2-6: 21.2±1.73 versus 23.56±1.52). E: proportion of proliferating (Ki67+) IPCs (SOX2+ YFP+ GFAP− cells in the SGZ; control versus β-catdel ex2-6: 56±4.56% versus 47.67±4.03%). F: total number of IPCs normalised to the length of the SGZ (mm) (SOX2+ YFP+ GFAP− cells in the SGZ; control versus β-catdel ex2-6: 20.18±2.31 versus 22.43±2.62). n=6. (G) YFP, TBR2 and DCX immunolabelling in the DG of control and β-catdel ex2-6 mice 30 days after tamoxifen administration. (H,I) Quantification of the data shown in G. H: total number of TBR2+ IPCs normalised to the SGZ length (mm) (TBR2+ YFP+ cells in the SGZ; control versus β-catdel ex2-6: 14.29±1.51 versus 17.16±3.09). I: total number of neuroblasts (NBs) normalised to the SGZ length (mm) (DCX+ YFP+ cells; control versus β-catdel ex2-6: 50.72±4.49 versus 50.53±5.69). n=6 and n=3 for control and β-catdel ex2-6 mice, respectively. Data analysed using unpaired two-tailed Student's t-test; ns, P>0.05. Scale bars: 50 µm in B,G. Data are mean±s.e.m.

We then investigated how the loss of β-catenin affects later steps in adult hippocampal neurogenesis. We quantified the percentage of proliferating IPCs (Fig. 2E), the total number of IPCs (Fig. 2F-H) and the total number of neuroblasts (Fig. 2G,I), and did not find any significant difference between β-catdel ex2-6 and control mice 30 days after β-catenin deletion. We also quantified the total number of neuroblasts and newly generated neurons 90 days after tamoxifen administration and again observed no difference between genotypes (Fig. S3C-G). Overall, these data show that loss of Wnt/β-catenin signalling in NSCs in vivo does not impair the behaviour or maintenance of NSCs; neither does it impair the generation and survival of new neurons in the adult hippocampus.

After having assessed the effects of disrupting Wnt/β-catenin signalling in NSCs in vivo, we investigated how stimulating Wnt/β-catenin signalling would affect NSCs. To do so, we generated GlastCreERT2; Catnblox(ex3)/wt RYFP mice [hereafter referred to as Catnbdel(ex3) mice] to stabilise β-catenin conditionally in Glast-expressing NSCs (Harada et al., 1999). Catnblox(ex3) is a conditional constitutively active allele of β-catenin (Catnb) in which exon 3, which encodes the GSK3β phosphorylation sites that mark β-catenin for degradation, is flanked by LoxP sites (Harada et al., 1999). Cre-mediated recombination of the Catnblox(ex3) allele resulted in β-catenin stabilisation and ligand-independent activation of downstream Wnt/β-catenin signalling in targeted cells. Upon recombination, we observed an increase in both β-catenin levels and in the intensity of the BATGAL reporter in Catnblox(ex3)/BATGAL mice compared with BATGAL controls (Fig. S4A-E).

We injected 2-month-old Catnbdel(ex3) and control mice with tamoxifen for 5 consecutive days and analysed their NSCs 10 and 30 days later (Fig. S4A). We first assessed the effect of stabilising β-catenin in NSCs by quantifying the total number of NSCs in Catnbdel(ex3) and control mice (Fig. S4F-K). We found a significant decrease in the number of NSCs in Catnbdel(ex3) mice compared with controls at 30 days (Fig. S4G) but not at 10 days after tamoxifen administration (Fig. S4J), suggesting that stabilising β-catenin causes NSC loss between 10 and 30 days later. This could be the result of increased NSC proliferation and subsequent depletion. We observed an increase in the proportion of Ki67-positive NSCs 30 days but not 10 days after tamoxifen administration (Fig. S4H,K). Therefore, the loss of NSCs at 30 days could be because of increased proliferation between the 10 and 30 day time points. However, we noticed that the cellular organisation of the DG was already disrupted in Catnbdel(ex3) mice 10 days after β-catenin stabilisation. Many YFP-positive recombined cells were displaced throughout the GCL and molecular layer (ML) (Fig. S4I). Some of these cells retained NSC characteristics (GFAP- and SOX2-positive cells with radial morphology) but their cell bodies were not correctly located in the SGZ (Fig. S4L). Such cells were also present in control mice, but the proportion of displaced NSCs was increased three-fold in Catnbdel(ex3) compared with control mice (Fig. S4M,N). This suggests that stabilising β-catenin in NSCs promotes their displacement from their correct SGZ location, which might cause their subsequent loss from the DG. Given that β-catenin regulates cell adhesion at adherens junctions (Bienz, 2005), this displacement phenotype could result from disrupted cell adhesion, which precludes using this mouse model to investigate the effects of stimulating Wnt/β-catenin signalling in NSCs in vivo. As an alternative approach, we used an established in vitro model of hippocampal NSCs that allows manipulation of their quiescent and active states in a niche-independent setting (Blomfield et al., 2019).

Dissociated hippocampal NSCs in adherent cultures can be maintained in a proliferative state by the presence of Fgf2 and can be induced into a reversible state of quiescence through the addition of Bmp4 (Blomfield et al., 2019; Martynoga et al., 2013; Mira et al., 2010). Using a published bulk RNA-sequencing dataset (Blomfield et al., 2019), we found that NSCs in this in vitro model system largely recapitulated the expression of Wnt pathway components observed in quiescent and active NSCs in vivo (Fig. 3A), including the upregulation of Wnt receptors in quiescent NSCs compared with active NSCs, corroborating published reports (Lie et al., 2005; Wexler et al., 2009). Moreover, quiescent and active NSCs expressed Wnt ligands at both the RNA and protein levels (Fig. 3A,B), suggesting that they can self-regulate their behaviour by autocrine/paracrine Wnt/β-catenin signalling, as previously proposed (Qu et al., 2013, 2010; Wexler et al., 2009). As with NSCs in vivo, we found that quiescent and active NSCs in vitro expressed Axin2 at similarly low levels, suggesting that their response to Wnt/β-catenin signalling is independent of their activation state (