BMP Signaling Regulates the Tempo of Adult Hippocampal Progenitor Maturation at Multiple Stages of the Lineage
Allison M. Bond,
Department of Neurology, Northwestern University's Feinberg School of Medicine, Chicago, Illinois, USA
Correspondence: Allison M. Bond, BA, Northwestern University's Feinberg School of Medicine, 303 E. Chicago Avenue, Ward 10-233, Chicago, Illinois 60611, USA. Telephone: 312-503-2795; Fax: 312-503-0872; e-mail: firstname.lastname@example.org
Novel environmental stimuli, such as running and learning, increase proliferation of adult hippocampal neural stem cells (NSCs) and enlarge the population of new neurons. However, it remains unclear how increased numbers of new neurons can be generated in a time frame far shorter than the time required for proliferating stem cells to generate these neurons. Here, we show that bone morphogenetic protein (BMP) signaling in the subgranular zone regulates the tempo of neural progenitor cell (NPC) maturation by directing their transition between states of quiescence and activation at multiple stages along the lineage. Virally mediated overexpression of BMP4 caused NPC cell cycle exit and slowed the normal maturation of NPCs, resulting in a long-term reduction in neurogenesis. Conversely, overexpression of the BMP inhibitor noggin promoted NPC cell cycle entry and accelerated NPC maturation. Similarly, BMP receptor type 2 (BMPRII) ablation in Ascl1+ intermediate NPCs accelerated their maturation into neurons. Importantly, ablation of BMPRII in GFAP+ stem cells accelerated maturation without depleting the NSC pool, indicating that an increased rate of neurogenesis does not necessarily diminish the stem cell population. Thus, inhibition of BMP signaling is a mechanism for rapidly expanding the pool of new neurons in the adult hippocampus by tipping the balance between quiescence/activation of NPCs and accelerating the rate at which they mature into neurons. Stem Cells2014;32:2201–2214
In the mammalian hippocampus, new granule cell neurons are generated in the subgranular zone (SGZ) of the dentate gyrus (DG) throughout life. Within the SGZ, neural stem cells (NSCs) generate a series of intermediate neural progenitor cells (NPCs) that, in turn, give rise to newborn granule neurons [1-3]. Novel experiences, such as running and learning, cause a rapid increase in neurogenesis which is associated with improved cognition [4, 5]. This behavioral enhancement can occur after as few as 7 days of exercise  or 7 days of learning , much less time than it takes for an NSC to mature into a neuron under normal conditions [8, 9]. This suggests that mechanisms other than increased proliferation of stem cells are necessary to generate such a rapid expansion of the new neuron population. Accelerated NPC maturation has been shown to contribute to the increase in neurogenesis that results from running [10, 11], and is potentially one of the fastest ways to expand the pool of new neurons. However, the mechanisms underlying this process remain unknown.
Bone morphogenetic proteins (BMPs) are secreted signaling molecules that belong to the larger transforming growth factor beta superfamily of signaling ligands, and regulate NSC fate and maturation throughout development [12-14]. BMPs and the BMP inhibitor, noggin, are endogenously expressed in the adult DG [15-19], and running induced neurogenesis is correlated with an overall decrease of BMP signaling in the DG . BMP signaling promotes NSC quiescence, whereas its inhibition conversely stimulates NSC division and subsequent neurogenesis [20-27]. However, a potential role for BMP signaling at later stages of the lineage remains unexplored.
Here, we define the effects of BMP signaling on NPC maturation by specifically modulating BMP signaling in vivo using virally mediated overexpression of BMP4 and noggin or conditional ablation of the BMP receptor type 2 (BMPRII) in distinct NPC subtypes. We show that inhibition of BMP signaling enhances neurogenesis by activating NPCs at multiple stages of the lineage and by accelerating their maturation. We conclude that inhibition of BMP signaling provides a mechanism for rapidly expanding the pool of new neurons to respond to the demands of the environment.
Materials and Methods
Generation of Transgenic Mice and Tamoxifen Administration
Generation of BMPRII floxed mutant mice (BMPRIIflx/flx) was described previously . BMPRIIflx/flx mice were crossed with fluorescent reporter mice harboring Rosa-CAG-LSL-ZsGreen1 transgene (RosazsG/zsG, , Jackson Laboratory; http://www.jax.org/, Bar Harbor, ME.) to homozygosity to generate mice with a genetically traceable marker. Double mutants (BMPRIIflx/flx; RosazsG/zsG) were then crossed with transgenic mice harboring Cre recombinase under the control of either human glial fibrillary acidic protein (hGFAP) or mouse achaete-scute complex homolog 1 (Ascl1) promoters [30, 31]. The triple transgenic (hGFAP-CreERT2; BMPRIIflx; RosazsG or Ascl1-CreERTM; BMPRIIflx; RosazsG) mice were maintained with homozygous or heterozygous floxed-BMPRII allele, and were mated to generate mice used for tamoxifen induced BMPRII ablation at 8–10 weeks of age. See Supporting Information Methods for description of tamoxifen preparation and injection.
Viral Vector Production and In Vivo Virus Injections
To generate a traceable viral vector, an internal ribosome entry site (IRES2) and the fluorescent protein mCherry were cloned into the pBOB lentiviral vector (from the lab of Inder M. Verma, , Addgene #12337; https://www.addgene.org/, Cambridge, MA), resulting in the control vector, pBOB-IRES2-mCherry (LV-Control). The pSecTag2 mammalian expression vector (Life Technologies; http://www.lifetechnologies.com/us/en/home.html, Grand Island, NY) was used to replace the initiation site of murine BMP4 or noggin cDNA with an N-terminal secretion tag, and each was cloned upstream of the IRES2-mCherry to generate the BMP4 overexpression vector, pBOB-secBMP4-IRES2-mCherry (LV-BMP4), and the noggin overexpression vector, pBOB-secNoggin-IRES2-mCherry (LV-Noggin), respectively. CAG-GFP retrovirus from Dr. Fred Gage's lab was used to label dividing progenitors (, Addgene #16664; https://www.addgene.org/, Cambridge, MA). See Supporting Information Methods for additional details.
Immunostaining was performed using standard techniques ( and Supporting Information Methods). See Supporting Information Methods for primary antibodies used and dilutions. Secondary antibodies used were fluorophore-conjugated Alexa Fluor goat or donkey secondary antibodies (1:250, Life Technologies; http://www.lifetechnologies.com/us/en/home.html, Grand Island, NY).
Confocal Imaging and Quantification
All imaging and quantification were done blinded to genotype and condition. Sections of similar rostral-caudal position in the dorsal DG were imaged using a Leica SP5 Confocal Microscope (http://www.leica-microsystems.com/, Wetzlar, Germany). See Supporting Information Methods for details.
Postnatal NPCs were cultured from C57Bl/6 mice as previously described .
RNA and Protein Extraction, Real-Time Reverse Transcriptase PCR, and Western Blotting
RNA and protein extraction, real-time reverse transcriptase PCR, and Western blotting were performed using standard techniques. See Supporting Information Methods for details.
Statistical analysis was performed with ANOVA (with Tukey post hoc test) or with two-tailed, unpaired Student's t test, as indicated in the figures and text.
BMP Signaling Regulates Cell Cycle Entry of Early and Late Progenitors
To define the role of BMP signaling in regulating the hippocampal stem/progenitor cell lineage, we first modulated BMP signaling in the DG using stereotaxic lentivirus injections. mCherry fluorescent protein tagged lentivirus overexpressing mouse BMP4 (LV-BMP4), a BMP highly expressed endogenously in the DG , was constructed to increase BMP signaling (Supporting Information Fig. S1A). Conversely, lentivirus overexpressing the BMP inhibitor, noggin (LV-Noggin), was constructed to reduce BMP signaling (Supporting Information Fig. S1A). An empty viral vector with the mCherry reporter served as a control (LV-Control; Supporting Information Fig. S1A). Viral vectors were first validated in vitro in postnatal NPC cultures. Infection of NPCs with LV-BMP4 increased BMP4 mRNA over 350-fold (Supporting Information Fig. S1B) with markedly increased secretion of BMP4 protein (Supporting Information Fig. S1C). Infection of NPCs with LV-Noggin increased noggin mRNA 125-fold (Supporting Information Fig. S1B) with significantly increased secretion of noggin protein (Supporting Information Fig. S1C). SMAD1/5/8 is a downstream phosphorylation target of BMPs, and levels of phosphorylated SMAD1/5/8 protein (pSMAD1/5/8) were used to assess overall BMP signaling. LV-BMP4 increased pSMAD1/5/8 levels, while LV-Noggin reduced pSMAD1/5/8 levels as predicted (Supporting Information Fig. S1D). When injected into DG, the lentivirus remained activated over the experimental timeline as shown by persistent mCherry expression at 28 and 56 days post-injection (dpi; Supporting Information Fig. S1E). In addition, Noggin and BMP transcripts detected by qRT-PCR remained elevated at 28dpi in DG injected with the respective lentivirus (Supporting Information Fig. S1F, S1G). Thus, LV-BMP4 and LV-Noggin persistently modulate BMP signaling in vivo in the DG.
LV-BMP4, LV-Noggin, or LV-Control were stereotaxically injected into the DG of adult mice and brains were examined 7dpi (Fig. 1A). First, we determined whether altered BMP signaling affected the cell cycle status of early SGZ progenitors as marked by SRY-box 2 (SOX2) and Ki67 expression. BMP4 overexpression markedly reduced the number of cycling SOX2+ progenitors (SOX2+Ki67+), whereas noggin overexpression more than tripled the number of SOX2+ Ki67+ cells in the SGZ (Fig. 1B, 1E; LV-Control 3,356 ± 1,628, LV-BMP4 418 ± 59, LV-Noggin 11,038 ± 1,961, Tukey post hocp < .05). However, neither BMP4 nor noggin changed the total number of SOX2+ progenitor cells (Fig. 1C, 1E; LV-Control 56,802 ± 6,178, LV-BMP4 58,291 ± 4,604, LV-Noggin 67,940 ± 5,083, ANOVA p = .31). Thus, BMP4 reduced the proportion of cycling SOX2+ cells by more than 80%, whereas noggin increased this proportion more than threefold compared to control (Fig. 1D; LV-Control 5.15% ± 2.24%, LV-BMP4 0.73% ± 0.13%, LV-Noggin 15.70% ± 1.67%, Tukey post hocp < .01). Thus, endogenous BMP signaling promotes SOX2+ progenitor cell cycle exit, and inhibiting BMP signaling with noggin recruits SOX2+ progenitors into the cell cycle.
We next evaluated the effects of BMP signaling on the doublecortin positive (DCX+) neuroblast population. BMP4 overexpression had no effect on the size of the DCX+ population, while overexpression of noggin more than doubled the DCX+ population (Fig. 1F, 1G; LV-Control 10,635 ± 2,689, LV-BMP4 18,628 ± 3,177, LV-Noggin 33,881 ± 5,384, Tukey post hocp < .01). However, no DCX+ cells colabeled with Ki67 after overexpression of BMP4, while noggin overexpression increased the number of colabeled cells by almost 10-fold (Fig. 1H, Supporting Information Fig. S1H; LV-Control 269 ± 139, LV-BMP4 0 ± 0, LV-Noggin 2,276 ± 535, Tukey post hocp < .05). Thus, BMP4 prevented DCX+ cells from cycling whereas noggin profoundly increased the proportion of DCX+ cells in cell cycle (Fig. 1I; LV-Control 1.95% ± 0.91%, LV-BMP4 0% ± 0%, LV-Noggin 6.89% ± 1.44%, Tukey post hocp < .05). Together these results indicate that BMP signaling promotes quiescence in early and late progenitors, and that inhibition of BMP signaling allows recruitment of these cells into the cell cycle.
BMP Signaling Regulates the Transition Between Quiescence and Activation in Intermediate Progenitor Populations
SOX2 is expressed in a wide range of early progenitors (type 1–2) and DCX expression is found both in late progenitors (type 2b and type 3) and in immature neurons [1, 2]. Because BMP signaling appeared to regulate cell division, we next focused specifically on the intermediate progenitor population, which constitutes most of the dividing cell population in the SGZ. We increased the resolution within the intermediate progenitor population using Ascl1 (Mash1) and NeuroD1 staining to distinguish between type 2a and type 2b intermediate progenitors, respectively. We injected LV-BMP4, LV-Noggin, or LV-Control into the DG of adult mice and on day 7 administered the thymidine analog ethynyl deoxyuridine (EdU) prior to examination (Fig. 2A). The EdU+ population of cells was reduced when BMP4 was overexpressed and was increased when noggin was overexpressed, confirming that BMP signaling regulates the overall level of proliferation (Fig. 2B, 2E; LV-Control 12,766 ± 2,928, LV-BMP4 3,485 ± 798 Tukey post hocp < .05, LV-Noggin 30,920 ± 3,787 Tukey post hocp < .001).
BMP4 overexpression severely reduced the size of the Ascl1+ intermediate progenitor population (Fig. 2C, 2E; LV-Control 63,276 ± 1,339, LV-BMP4 7,843 ± 613, Tukey post hocp < .001) and prevented Ascl1+ progenitor proliferation (Fig. 2D, 2E; LV-Control 7,107 ± 2,081, LV-BMP4 0 ± 0, Tukey post hocp < .01). Noggin overexpression did not change the overall size of the Ascl1+ population (Fig. 2C, 2E; LV-Noggin 63,094 ± 5,694, Tukey post hocp < .001), but greatly increased the number of dividing Ascl1+ progenitors (Fig. 2D, 2E; LV-Noggin 19,917 ± 2,275, Tukey post hocp < .001), resulting in a threefold increase in the dividing proportion of the Ascl1+ population (Fig. 2F; LV-Control 11.38% ± 2.90%, LV-Noggin 31.39% ± 1.27%, Tukey post hocp < .001). Thus, endogenous BMP signaling prevents Ascl1+ progenitors from dividing, and noggin inhibition mobilizes quiescent Ascl1+ progenitors to re-enter cell cycle.
Next we examined the effects of altered BMP signaling on NeuroD1+ intermediate progenitors, which are derived from Ascl1+ progenitors. BMP4 overexpression reduced the NeuroD1+ population (Fig. 2G, 2J; LV-Control 65,529 ± 3,235, LV-BMP4 17,281 ± 6,815, Tukey post hocp < .01) and prevented NeuroD1+ progenitors from proliferating (Fig. 2H; LV-Control 672 ± 672, LV-BMP4 0 ± 0, ANOVA p < .05). In contrast, noggin overexpression increased the size of the NeuroD1+ population (Fig. 2G, 2J; LV-Noggin 139,116 ± 11,336, Tukey post hocp < .001) and increased the number of dividing NeuroD1+ progenitors (Fig. 2H; LV-Noggin 2,451 ± 961, ANOVA p < .05). Modulating BMP signaling did not affect the proportion of the NeuroD1+ population that was actively dividing (Fig. 2I; LV-Control 0.93% ± 0.93%, LV-BMP4 0% ± 0%, LV-Noggin 2.24% ± 0.9%, ANOVA p = .19), suggesting that changes in the size of the NeuroD1+ population likely reflect changes earlier in the lineage.
The enhanced proliferation observed with BMP signaling inhibition could be due to a shortened cell cycle. Cell cycle analysis using a dual thymidine labeling protocol where iododeoxyuridine (IdU) and chlorodeoxyuridine (CldU) were administered with either a 3-hour or 16-hour separation time allowed us to determine the length of S-phase or total cell cycle length, respectively . Modulating BMP signaling did not change the S-phase length (Supporting Information Fig. S1I; LV-Control 9.48 ± 1.52, LV-Noggin 11.38 ± 0.58, p = .24) or total cell cycle length of NPCs (Supporting Information Fig. S1I; LV-Control 23.69 ± 0.17, LV-Noggin 24.62 ± 0.34, p = .06). This indicates that the enhanced cell division observed after BMP signaling inhibition is due to progenitor activation and entry into cell cycle, rather than a shortening of cell cycle.
To define the longer term effects of modulating BMP signaling, we analyzed virally injected brains at 28dpi (Fig. 3A). BMP4 overexpression decreased the number of Ki67+ cells by more than 70% (Fig. 3B; LV-Control 10,310 ± 718, LV-BMP4 2,881 ± 565, Tukey post hocp < .001). By contrast, the number of Ki67+ cells after noggin overexpression did not differ significantly from control (Fig. 3B; LV-Noggin 8,642 ± 718). BMP4 overexpression maintained a marked reduction in the number of SOX2+Ki67+ cells, whereas sustained noggin overexpression returned SOX2+Ki67+ cell numbers to control levels (Fig. 3C; LV-Control 5,898 ± 848, LV-BMP4 2,095 ± 482 Tukey post hocp < .01, LV-Noggin 6,340 ± 1,108). Despite the effect of BMP4 on NPC division, neither BMP4 nor noggin changed the size of the SOX2+ population (Fig. 3D; LV-Control 54,967 ± 6,468, LV-BMP4 59,087 ± 1,637, LV-Noggin 54,323 ± 2,298, ANOVA p = .66). Sustained BMP4 overexpression decreased the number of DCX+ neuroblasts by 84%, while noggin overexpression led to a return to control numbers (Fig. 3E, 3F; LV-Control 31,339 ± 3,916, LV-BMP4 4,899 ± 971, LV-Noggin 30,489 ± 4,422, Tukey post hocp < .001). To validate our observations with Ki67 immunostaining, bromodeoxyuridine (BrdU) was administered 1 day prior to analysis (27dpi) to measure proliferation (Fig. 3A). BMP4 overexpression decreased the number of BrdU-labeled cells by 68%, whereas sustained noggin overexpression did not alter proliferation (Fig. 3G, 3H; LV-Control 9,769 ± 1,241, LV-BMP4 3,089 ± 1,003, LV-Noggin 11,362 ± 2,926, ANOVA p < .05). Together these data suggest that a chronic increase in BMP signaling markedly diminishes neurogenesis in the DG by promoting NPC cell cycle exit. However, the effect of BMP signaling inhibition on NPC cell cycle entry observed at 7dpi is not sustained.
BMP4 Slows NPC Maturation and Noggin Accelerates NPC Maturation
The effects of BMP signaling on NPC division are only relevant to the functional hippocampal circuit if they translate into changes in neuron generation. Therefore, we asked whether modulating BMP signaling would also affect NPC maturation into neurons. We used a green fluorescent protein (GFP)-labeled retrovirus to birthdate a cohort of dividing newborn NPCs in DG infected with LV-BMP4 or LV-Noggin, and analyzed the maturation of this cohort at 28 and 56dpi (Fig. 4A). Under normal conditions, most NPCs mature into neurons by 28dpi, and by 56dpi these newborn neurons have integrated into the existing circuitry [9, 36, 37]. As expected, most GFP+ cells in the LV-Control condition had long dendritic processes characteristic of mature dentate granule cells (DGCs) at 28dpi (Fig. 4D). Similarly, with LV-Noggin most GFP+ cells were mature DGCs. However, with LV-BMP4 very few GFP+ cells had a neuronal morphology, and instead most were small cells located in the SGZ with few processes (Fig. 4D). Immunohistological markers of development were used to determine the maturation state of the GFP+ cells. Few of the control GFP+ cells were SOX2+, presumably because they had matured past this developmental stage, and noggin overexpression resulted in even fewer GFP+ SOX2+ cells (Fig. 4B; LV-Control 20.03% ± 5.08%, LV-Noggin 7.83% ± 4.28%). In contrast, after BMP4 overexpression more than 60% of the GFP+ cells were SOX2+, suggesting an arrest or delay in development (Fig. 4B; 64.47% ± 3.05%, Tukey post hocp < .001). BMP signaling promotes an astrocytic fate in cultured NPCs [38, 39], and astrocytes can express SOX2 . To determine whether BMP4 overexpression caused a change in cell fate or a delay in development, we examined expression of S100β, a marker of mature astrocytes. The proportion of GFP+ S100β+ cells did not significantly change with BMP4 overexpression (Fig. 4C; LV-Control 4.03% ± 2.46%, LV-BMP4 11.78% ± 2.38%, p = .11), suggesting that SOX2 colocalization indicates a developmental delay at an early progenitor stage. When we examined neuronal maturation of the GFP+ cohort, most control GFP+ cells were NeuN+ and even more were NeuN+ with noggin overexpression (Fig. 4E; LV-Control 61.00% ± 11.36%, LV-Noggin 78.02% ± 3.43%). In contrast, few GFP+ cells were NeuN+ with BMP4 overexpression (Fig. 4E; 14.50% ± 3.70%, Tukey post hocp < .01). Although NeuN expression signifies neuronal maturation, the later expression of calbindin demarcates a shift to adult-like connectivity [9, 41, 42]. The proportion of GFP+ cells that colocalized with calbindin increased by almost 60% when noggin was overexpressed, signifying a more advanced stage of maturation (Fig. 4F; LV-Control 39.28% ± 6.31%, LV-Noggin 62.01% ± 6.77%, p < .05).
These data support the idea that inhibition of BMP signaling accelerates NPC maturation; however, it remains unclear whether BMP signaling blocks or delays normal neuronal maturation. To distinguish between these two possibilities we examined GFP+ cells 56dpi. Similar to the earlier time point, BMP4 overexpression at 56dpi decreased the proportion of GFP+ NeuN+ cells (Fig. 4G; LV-Control 90.05% ± 7.94%, LV-BMP4 46.76% ± 12.91%, p < .05). Interestingly, within the LV-BMP4 condition the proportion of GFP+ NeuN+ cells almost doubled between 28 and 56dpi (Fig. 4E, 4G), suggesting that BMP signaling delayed, rather than blocked, maturation. As newborn neurons mature they migrate radially out of the SGZ into the granule cell layer (GCL); thus migration is a measure of maturation. BMP4 overexpression impaired neuronal migration out of the SGZ at both 28 and 56dpi (Fig. 4H; 28dpi: LV-Control 14.7 µm ± 2.1, LV-BMP4 8.1 µm ± 1.1, p = .18; 56dpi: LV-Control 18.8 µm ± 1.8, LV-BMP4 13.1 µm ± 1.0, p < .05). Together these data indicate that BMP signaling slows normal neuronal maturation, while inhibition of BMP signaling accelerates maturation.
Inhibition of BMP Signaling Does Not Deplete the Stem Cell Pool
Viral overexpression of BMP ligand and inhibitor provided an effective method to modulate BMP signaling throughout the hippocampal niche. However, it did not allow us to distinguish between the effects of BMP signaling on NSCs and NPCs. To address this question, we used inducible Cre recombinase-mediated deletion of the BMP type 2 receptor (BMPRII) to cell autonomously ablate BMP signaling at different points along the lineage. Activated BMP receptors are composed of BMPRII and a BMP type 1 receptor, BMPRIa or BMPRIb. Because we wanted to eliminate all BMP signaling, we ablated the obligatory BMPRII, rather than one of the type 1 receptors. Cre-mediated ablation of BMPRII in neural stem/progenitors cells in vitro reduced BMPRII RNA levels and pSMAD1/5/8 protein levels (Supporting Information Fig. S2A, S2B). Additionally, conditional ablation of BMPRII in vivo reduced BMPRII immunostaining in recombined cells (Supporting Information Fig. S2C). To examine BMPRII function at different points along the lineage, we used hGFAP-CreERT2; BMPRIIflx/flx (GFAP-RII cKO) and Ascl1-CreERTM; BMPRIIflx/flx (Ascl1-RII cKO) transgenic mouse lines to conditionally ablate BMPRII in GFAP+ type 1 NSCs or Ascl1+ type 2 NPCs, respectively. These transgenic mice also carried the fluorescent reporter ZsGreen1 (ZG) within the ROSA locus, allowing us to perform fate mapping studies of recombined cells.
Recent studies suggest that exogenously activating NSCs diminishes this population long-term [25, 43]. We wanted to know whether activating neural progenitors by inhibiting BMP signaling would deplete the NSC population. First, we determined whether sustained noggin overexpression throughout the niche would deplete the NSC population. Although short-term overexpression of noggin enhanced activation of NPCs, sustained noggin overexpression did not expand or deplete the radial GFAP+SOX2+ stem cell population (Fig. 5A–5C; LV-Control 12,832 ± 784, LV-BMP4 12,856 ± 1,014, LV-Noggin 11,820 ± 1,137, ANOVA p = .75). Next we examined whether BMPRII ablation in GFAP+ NSCs affects the size of the stem cell population 28 days after tamoxifen induction (D28; Fig. 5D). Deletion of BMPRII in GFAP+ NSCs did not alter the total number of ZG+ cells located in the GCL (Fig. 5E; BMPRII+/flx 84,621 ± 8,645, BMPRIIflx/flx 79,862 ± 12,038, p = .74). However, BMPRII ablation increased the size of the radial GFAP+ SOX2+ NSC pool by 44% (Fig. 5F, 5H; BMPRII+/flx 10.97% of ZG+ population ± 1.12%, BMPRIIflx/flx 15.82% ± 1.17%, p < .01), suggesting that inhibition of BMP signaling in GFAP+ NSCs causes expansion of the stem cell population. BMPRII ablation did not change the total number of SOX2+ cells (Fig. 5G; BMPRII+/flx 46.21% ± 3.21%, BMPRIIflx/flx 51.40% ± 2.03%, p = .23). A recent study showed that low levels of Ascl1 are expressed in a subset of type 1 stem cells in the SGZ , so next we asked whether BMPRII ablation in Ascl1+ cells might affect the size of the NSC population. BMPRII ablation did not significantly change the number of radial GFAP+ SOX2+ NSCs at D28 (Supporting Information Fig. S2E; BMPRII+/flx 3,903 ± 1,099, BMPRIIflx/flx 2,843 ± 858, p = .48). We conclude that blocking BMP signaling in NSCs does not deplete the stem cell pool, and in fact blocking BMP signaling in GFAP+ NSCs expands the radial stem cell population.
BMPRII Ablation Accelerates NPC Maturation by Activating Intermediate Progenitor Populations
Next we wanted to examine whether BMPRII ablation in type 1 GFAP+ NSCs might affect neuronal maturation (Fig. 6A). BMPRII ablation increased the proportion of ZG+ cells that matured into NeuN+ neurons by 30% at D28 (Fig. 6B, 6C; BMPRII+/flx 33.85% ± 2.64%, BMPRIIflx/flx 44.26% ± 3.26%, p < .05). We then asked whether the increased number of newborn neurons resulted from enhanced survival. Immunostaining with the apoptotic cell marker cleaved caspase 3 showed similar numbers in both groups, suggesting that increased neuronal survival was not a significant contributing factor (Supporting Information Fig. S2D). These results suggest that BMPRII ablation either increases the number of neurons generated or accelerates the maturation of progenitors into neurons.
To address these two possibilities, we examined Ascl1-RII cKO mice on D7, 14, and 28 to determine how BMPRII ablation in more committed type 2 Ascl1+ NPCs affects maturation (Fig. 6D). While the total number of ZG+ cells increased over time in both control and mutant DG, due to ongoing NPC division, the rate at which the ZG+ population expanded increased with BMPRII ablation (Fig. 6E). At both D7 and D14, there were more ZG+ cells in the BMPRIIflx/flx GCL (D7: BMPRII+/flx 32,724 ± 2,774, BMPRIIflx/flx 49,968 ± 4,785, p < .01; D14: BMPRII+/flx 37,601 ± 2,973, BMPRIIflx/flx 63,710 ± 9,771, p < .05). However, this difference was not maintained, and by D28 the two genotypes had an equivalent number of ZG+ cells (Fig. 6E; BMPRII+/flx 73,278 ± 10,314, BMPRIIflx/flx 72,477 ± 11,743, p = .96). Because noggin inhibition of BMP signaling expanded the intermediate progenitor population, we examined whether BMPRII ablation had a similar effect. At D7 BMPRII ablation expanded the ZG+ Ascl1+ population (Fig. 6F, 6I; BMPRII+/flx 18,024 ± 2,633, BMPRIIflx/flx 396,801 ± 5,662; p < .05). BMPRII ablation also expanded the ZG+ NeuroD1+ population (Fig. 6G, 6J; BMPRII+/flx 14,585 ± 2,112, BMPRIIflx/flx 22,981 ± 1,652; p < .05). These populations likely expanded due to increased proliferation of intermediate progenitors as shown by a doubling of EdU incorporation within the recombined cells (Fig. 6H, 6K; BMPRII+/flx 7,666 ± 1,766, BMPRIIflx/flx 23,689 ± 6,143; p = .1).
Finally, we examined whether BMPRII ablation affected neuronal maturation. BMPRII ablation increased the rate at which the ZG+ population matured into NeuN+ neurons (Fig. 6L, 6M). At both D7 and 14 BMPRII ablation increased the number of ZG+ neurons in the GCL (D7: BMPRII+/flx 7,055 ± 869, BMPRIIflx/flx 11,182 ± 1,043, p < .01; D14: BMPRII+/flx 22,246 ± 3,265, BMPRIIflx/flx 46,133 ± 7,902, p < .05). However, by D28, the two genotypes had an equivalent number of ZG+ neurons (Fig. 6L, 6M; BMPRII+/flx 50,201 ± 9,922, BMPRIIflx/flx 51,216 ± 9,238, p = .94). Together these data suggest that at the intermediate progenitor stage, endogenous BMP signaling restricts the rate of neurogenesis by slowing neuronal maturation. In contrast, blocking BMP signaling accelerates NPC maturation without altering the number of new neurons a progenitor can generate.
Enhanced cognition due to novel environmental experiences is thought to result from increased proliferation of NSCs, which leads to the generation of new neurons [2, 4, 45, 46]. However, cognition is enhanced in response to the environment much faster than the normal time course of development from NSC to neuron [6, 7]. Therefore, it is likely that an accelerated rate of NPC maturation is responsible for the rapid generation of new neurons. We show that inhibition of BMP signaling accelerates neurogenesis by mobilizing quiescent progenitors at multiple stages along the lineage. Thus, we propose that inhibition of BMP signaling presents a mechanism for rapidly enlarging the pool of new neurons by increasing the tempo of NPC maturation.
Previous studies have focused on the role of BMP signaling in regulating the stem cell population in the adult hippocampus [23, 25]. Here, we were interested in whether BMP signaling affects later stages of the lineage. Because BMP signaling is known to regulate NSC quiescence and activation, we first focused on whether BMP signaling might regulate this transition in NPCs. Our Ki67 and EdU colabeling data in the viral overexpression experiments indicate that BMP signaling regulates the transition between quiescent and activated states at many stages along the lineage. Increased BMP signaling pushed SOX2+ progenitors, Ascl1+ and NeuroD1+ intermediate progenitors, and even DCX+ neuroblasts into quiescence, resulting in diminished numbers of late progenitors and ultimately reducing the generation of new neurons. In contrast, inhibition of BMP signaling recruited progenitors at multiple stages into cell cycle, causing them to divide and expand the intermediate progenitor and neuroblast populations. BMP signaling clearly acts directly on early NPC populations (SOX2+ and Ascl1+) to dictate what proportion of these populations is actively dividing. However, it is less clear whether BMP signaling acts directly on later NPC populations (NeuroD1+ and DCX+), or if changes in the size of these populations reflects effects on earlier stages. Regardless, the balance between BMP signaling and its inhibition regulates the cell cycle status of multiple progenitor populations along the lineage.
We propose that the effects of BMP signaling on NPC quiescence/activation control the tempo of hippocampal neurogenesis. BMP signaling biases progenitors to enter a quiescent state, thus increasing the amount of time it takes for them to mature into neurons (Fig. 7). In contrast, inhibition of BMP signaling pushes progenitors into an activated state, which aids their progression down the lineage (Fig. 7). Under normal conditions, the balance of these opposing forces influences the rate of neurogenesis. It was alternatively possible that modulating BMP signaling changed the length of the cell cycle and hence the rate of maturation, but this was not the case (Supporting Information Fig. S1I). Recruitment of existing pools of quiescent progenitors back in the cell cycle provides a mechanism for very rapid changes in neurogenesis in response to the environment. In fact, we have previously shown that voluntary wheel running, a proneurogenic stimulus, increases noggin expression and decreases overall BMP signaling levels in the hippocampus . Our observations suggest that activity-dependent enhancement of neurogenesis in the SGZ reflects, at least in part, inhibition of BMP signaling.
Although increased levels of BMP signaling had long-term effects on neurogenesis, the neurogenic effect of BMP signaling inhibition was not sustained. Both overexpression of noggin and BMPRII ablation increased the generation of neuroblasts and neurons in the short-term, but at later time points, the rate of neurogenesis returned to control levels. This suggests that BMP signaling regulates the tempo of maturation but does not alter the number of neurons that a progenitor has the potential to generate. This interpretation is supported by our retroviral birthdating experiment that demonstrated that BMP4 overexpression did not block development or change the fate of progenitors, but rather slowed progenitor maturation. This idea is also supported by our finding that BMPRII ablation accelerates the rate of neuron generation, but does not change the number of neurons generated in the long-term.
It is possible that blocking BMP signaling increases the number of immature neurons that are generated, but that survival of these new neurons is limited by the availability of another factor. This possibility is supported by observations of the additive effects of running and environmental enrichment . Running is thought to enhance neurogenesis primarily by increasing progenitor proliferation, while environmental enrichment is thought to enhance neurogenesis by promoting survival of newborn cells [4, 48]. Both conditions enhance neurogenesis on their own, but when done sequentially—running, then enrichment—there is an additive effect in increasing neurogenesis . Thus, it is possible that the increase in neurogenesis observed with inhibition of BMP signaling could be sustained with environmental enrichment or additional survival-promoting factors.
Our observation that the stem cell pool expands at 28 days after BMPRII ablation in GFAP+ NSCs indicates that BMP signaling regulates the size of the stem cell pool. However, the size of the stem cell pool did not expand when BMP signaling was inhibited throughout the niche by noggin overexpression. This presumably reflects the fact that BMPRII ablation initially only affected GFAP+ NSCs whereas noggin overexpression affected all stages of progenitors. We show that in addition to activating NPCs, BMP signaling inhibition accelerates their maturation, and this effect seems to be most robust in Ascl1+ intermediate progenitors. Thus, when noggin is overexpressed throughout the niche, accelerated progenitor maturation would likely require a constant resupply of progenitors from the NSCs, which may not allow for expansion of the stem cell pool. These results differ from previously reported work which suggested that noggin administration enhances cell division in the short-term but reduces cell division and depletes the stem cell pool long-term . This inconsistency may be due to a difference in experimental design; in the previous study noggin administration lasted only 7 days, followed by a 21-day lag period, whereas we examined sustained noggin administration. Although this previous work hypothesized that inhibition of BMP signaling depleted the NSC pool, our longer term experimental design allowed us to demonstrate that the NSC pool is not only maintained but actually increases in response to long-term inhibition of BMP signaling. Furthermore, in prior studies using transgenic overexpression of noggin, we found that long-term inhibition of BMP signaling (>4 months) not only does not deplete hippocampal NSCs but actually expands the NSC pool .
We conclude that BMP signaling is an important regulator of hippocampal progenitor quiescence and activation at multiple stages of the lineage. Through its regulation of cell cycle status, BMP signaling influences the maturation rate of hippocampal progenitors. These findings provide insight into a potential mechanism for rapid changes in neurogenesis in response to environmental experiences.
We would like to thank Drs. Jane Johnson and Flora Vaccarino for their kind contribution of the Ascl1-CreERTM and hGFAP-CreERT2 mouse lines. We would also like to thank John G. Cooper for his contribution to supplementary data. This work was funded by the National Center for Research Resources (NCRR) and the National Center for Advancing Translational Sciences (NCATS), NIH through Grant Number TL1R000108 as well as the NIH NS 20013 and NS 20778.
A.M.B. and C.-Y.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; E.A.M., T.M., and O.E.: collection and/or assembly of data; J.A.K.: conception and design, manuscript writing, and final approval of manuscript. A.M.B. and C.-Y.P. contributed equally to this article.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.