Running Rescues Defective Adult Neurogenesis by Shortening the Length of the Cell Cycle of Neural Stem and Progenitor Cells

Abstract

Physical exercise increases the generation of new neurons in adult neurogenesis. However, only few studies have investigated the beneficial effects of physical exercise in paradigms of impaired neurogenesis. Here, we demonstrate that running fully reverses the deficient adult neurogenesis within the hippocampus and subventricular zone of the lateral ventricle, observed in mice lacking the antiproliferative gene Btg1. We also evaluated for the first time how running influences the cell cycle kinetics of stem and precursor subpopulations of wild-type and Btg1-null mice, using a new method to determine the cell cycle length. Our data show that in wild-type mice running leads to a cell cycle shortening only of NeuroD1-positive progenitor cells. In contrast, in Btg1-null mice, physical exercise fully reactivates the defective hippocampal neurogenesis, by shortening the S-phase length and the overall cell cycle duration of both neural stem (glial fibrillary acidic protein⁺ and Sox2⁺) and progenitor (NeuroD1⁺) cells. These events are sufficient and necessary to reactivate the hyperproliferation observed in Btg1-null early-postnatal mice and to expand the pool of adult neural stem and progenitor cells. Such a sustained increase of cell proliferation in Btg1-null mice after running provides a long-lasting increment of proliferation, differentiation, and production of newborn neurons, which rescues the impaired pattern separation previously identified in Btg1-null mice. This study shows that running positively affects the cell cycle kinetics of specific subpopulations of newly generated neurons and suggests that the plasticity of neural stem cells without cell cycle inhibitory control is reactivated by running, with implications for the long-term modulation of neurogenesis. Stem Cells 2014;32:1968–1982

Introduction

Human studies have shown that exercise has profound benefits for cognitive function, promoting plasticity and health of the central nervous system 1, 2. In animal studies, aerobic running, either voluntary or forced, is associated with changes in the expression of neuronal plasticity-related genes 3, 4 and enhanced hippocampal adult neurogenesis 5-7. Moreover behavioral tests have confirmed the positive effect of exercise on hippocampal neurogenesis, showing an improvement of running mice in a series of specific hippocampus-dependent tasks such as pattern separation 8, contextual fear conditioning, and novel object recognition 9, 10. Many studies suggested that both precursor cell proliferation and the relative number of neurons among the progeny of neural precursors decrease in older age 11, 12. However, it has been shown that exercise is capable of overcoming the age-dependent depletion of hippocampal neurogenesis 13-15. Adult neurogenesis takes place in two main neurogenic areas: the subventricular zone (SVZ) adjacent to the lateral ventricles, which generates olfactory bulb interneurons, and the subgranular zone, which gives rise to granular neurons in the hippocampal dentate gyrus (DG) 16. The process of adult hippocampal neurogenesis has been divided into six developmental stages 17, 18, in which putative neural stem cells (NSCs, type-1 cells) develop into postmitotic neurons through three consecutive stages of progenitor cells (type-2ab and type-3 cells; 19-21). Afterward, immature neurons migrate within the granule cell layer and transiently express the calcium binding protein calretinin, followed by more mature neuronal markers such as neuronal nuclei (NeuN) and calbindin 22, 23. Newborn neurons become functionally integrated into existing DG circuitry within 3 weeks, extending their axons to CA3, as indicated by morphological and electrophysiological studies 24, 25. Interestingly, recent observations have demonstrated that new neurons of the DG become functionally active in learning circuits at late stages of their maturation (∼4–6 postnatal weeks, 26). In a previous study, we have described a mouse model in which the inactivation of the antiproliferative gene Btg1 results in a transient increase of proliferation and production of newly generated neurons in the SVZ and DG during the early postnatal stages. This overproduction of newborn neurons gives rise during adulthood to a severe depletion of stem and progenitor cells, caused by a concomitant decrease of their proliferative capability, accompanied by premature exit from cell cycle and massive apoptosis. The increased exit from cell cycle is probably dependent on an increment of p21 expression to compensate for the loss of the cell cycle regulation exerted by Btg1 27, 28. These data strongly suggest that Btg1 plays an essential role in the physiological maintenance of the neurogenic pool and in the proper control of adult neurogenesis, as also implied by its expression in the adult and developing SVZ and DG (27; Supporting Information Fig. S1). In this study, we show that running reverses the severe phenotype of Btg1 knockout (KO) mouse, by promoting proliferation and neurogenesis in the adult neurogenic niches. Moreover, for the first time, we evaluated how running influences the cell cycle kinetics of distinct hippocampal precursor subpopulations of Btg1 wild-type and KO mice. Our data illustrate that a short period of running fully restores the proliferation of stem/progenitor cells in the DG of Btg1 null-mice, by shortening the S-phase length and consequently the overall duration of the cell cycle. Notably, we show that in Btg1 KO mice, but not in the wild type, running is able to specifically recruit to the cell cycle the quiescent NSCs, accelerating their cell cycle progression and consequently allowing their expansion. These events are sufficient to reactivate the hyperproliferation observed in the Btg1 postnatal mice and to completely replenish the loss of adult neurogenesis. Such a sustained increase in cell proliferation in Btg1 null mice after running provides a large pool of activated NSCs that results in a long-term increment of neurogenesis and production of newborn neurons, which in turn contribute to improve the defective process of pattern separation detected in Btg1-null mice.

Materials and Methods

Animals

C57BL/6 Btg1 KO mice had been generated previously 27. Genotyping was routinely performed by PCR (for details see 27). Mice were reared in standard cages until P60. At 2 months of age, they were randomly assigned to running wheel or standard cages for 12 days. Distances run were recorded daily with an automatic counter. After 12 days, mice were perfused or returned to standard cages to be sacrificed at different times from the end of the run. Animal were treated following the Italian Ministry of Health and directive 2010/63/EU guidelines.

Immunohistochemistry

Brains were collected after transcardiac perfusion with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and kept overnight in PFA. Afterward, brains were equilibrated in sucrose 30% and cryopreserved at −80°C.

Hippocampi from brains embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA; http://www.sakura-americas.com/about/contact.lasso) were cut by cryostat at −25°C in 40 µm serial free floating sections. Sections were then stained for multiple labeling using fluorescent methods. After permeabilization with 0.3% Triton X-100 in PBS, sections were incubated with primary antibodies with 3% normal donkey serum in PBS for 16–18 hours. Primary and secondary antibodies used are reported in Supporting Information.

Thymidine Analogs Detection

See Supporting Information.

Stereological Analysis

See Supporting Information.

In Situ Hybridization

See Supporting Information.

Cell Cycle Analysis

S-phase length (Ts) and cell cycle length (Tc) were obtained using the equation developed by others 29, 30. To measure Ts, we administrated intraperitoneally IdU (57.5 mg/kg in 0.02 N NaOH, 0.9% NaCl saline solution), and 3 hours later CldU (42.5 mg/kg in 0.9% NaCl saline solution). For the determination of Tc, the interval between injections was 20 hours. Forty-five minutes after CldU, injected animals were sacrificed for immunohistochemistry. Tc and Ts of the glial fibrillary acidic protein (GFAP⁺), Sox2⁺, and NeuroD1⁺ progenitors were obtained through triple immunofluorescence GFAP⁺IdU⁺Cldu⁺, Sox2⁺IdU⁺Cldu⁺, and NeuroD1⁺IdU⁺CldU⁺.

G2/M length was estimated with the percentage of labeled mitosis 31. Animals were injected with BrdU (95 mg/kg in PBS) and sacrificed after 2 or 2.5 hours for immunohistochemistry against BrdU and phosphohistone-H3 (phH3). The labeling index (LI) was calculated as the percentage of phH3⁺ cells colabeled with BrdU. G1 length was obtained as Tc − (Ts + T_G2/M).

Contextual Fear-Discrimination Learning

See Supporting Information.

Statistical Analysis

Statistical analyses were performed using two-way ANOVA (genotype × running) for comparing the effects of running, respectively, in WT and KO mice. Whenever appropriate, simple effects were analyzed by Fisher's protected least significant difference (PLSD) test. All experimental groups were composed of at least four animals. Statistical analysis of mouse behavior was performed using three-way repeated measure ANOVA (genotype × running × day), followed by Fisher's PLSD test. Differences were considered statistically significant at p < .05. All data were expressed as mean values ± SEM.

Results

Impact of Btg1 Deficiency on Wheel-Running Activity

We first determined the impact of Btg1 deficiency on wheel-running activity and we found that both Btg1 wild-type and KO mice made use of the running wheel to a large extent and without significant differences in distance of run (Fig. 1A, 1B). Over a 12 days running period, the cumulative distances run were on average 79.6 km in the Btg1 wild-type and 80 km in the Btg1 KO mice (Fig. 1B).

Figure 1

Voluntary exercise promotes cellular proliferation in Btg1 WT and KO mice. (A): Schematic diagram of the experimental paradigm and BrdU injection protocol. Exercising mice were allowed access to running wheels for 12 days. BrdU (95 mg/kg) was injected 2 hours before the sacrifice of mice. (B): Average daily running distance in km/day (±SEM) over 12 days of free access to wheels. (C): Representative pictures showing the increase of BrdU-positive cells in the DG of Btg1 WT and KO RUN mice (scale bar = 100 µm). (D): Quantification of the absolute number of DG cells entering in S phase (total Brdu⁺ cells after a 2-hour pulse) showed a significant difference between Btg1 WT and KO RUN mice in comparison with their respective controls (running effect, F_1,161 = 34.3, p < .001). (E): Running induces a significant increase of Ki67⁺ cells in both the genotypes analyzed (genotype effect, F_1,217 = 28.92, p < .001; running effect F_1,217 = 98.8, p < .001). Cell number in the DG is means ± SEM of the analysis of at least three animals per group. **, p < .01; ***, p < .001. Abbreviations: DG, dentate gyrus; KO, knockout; WT, wild type.

Effects of Running on Adult Hippocampal Proliferation and Cell Cycle

In this article, our first question was whether the proneurogenic effect of voluntary wheel running was able to overcome the reduced adult hippocampal proliferation observed in the Btg1 null mice 27. To this aim, the Btg1 wild-type and Btg1 KO mice were housed in cages with free access to a running wheel for 12 days. In order to identify the number of proliferating progenitor cells entering in S phase, control and running mice of both genotypes were injected with a short pulse of BrdU 2 hours before the sacrifice (Fig. 1A). Immunohistochemical analysis showed that running stimulated the proliferation rate, as revealed by the 23% increase in the number of BrdU-positive cells in the Btg1 wild-type running mice (thereafter referred as WT RUN mice) compared to the wild-type control mice (WT CTL mice) (p < .01; Fig. 1C, 1D). A higher increase (40%) of proliferation was observed in the running Btg1 KO mice (KO RUN mice) compared to the Btg1 null sedentary mice (KO CTL mice) (p < .001; Fig. 1C, 1D). Surprisingly, we observed that the number of BrdU-labeled cells in the KO RUN mice was significantly higher also with respect to the WT CTL mice (p < .001, with an increase of 30%; Fig. 1C, 1D), reaching the proliferation rate detected in the WT RUN mice (Fig. 1C, 1D).

The cell proliferation in response to running measured by Ki67, which labels cycling cells 32, was similar to BrdU data. Indeed, both WT RUN and KO RUN mice displayed a significant increase in the number of Ki67-positive cells when compared with their respective controls (p < .001 with an increment of 30% and p < .001 with an increment of 36%, respectively; Fig. 1E). These data indicate that the basal proliferation of the Btg1 KO mice was highly stimulated by running and that a short period of running was able to revert the decreased proliferation rate of the KO CTL mice. As proliferation in the adult brain is precursor-type-specific 33, we sought to determine whether the increased proliferation detected in the KO RUN mice was restricted to a specific subpopulation within all cycling precursors (Fig. 2A). With this purpose, we used the transcription factor Sox2 to identify quiescent and proliferating NSC (type-1a and type-1b, respectively, Fig. 2A) as well as immature progenitor cells (type-2a, 18, 34). Neuronally committed progenitor cells (type-2b-3) were identified by the transcription factor NeuroD1, whose expression persists until the stage of early postmitotic neuroblasts (stage 5; 35). Our quantifications revealed that the number of dividing NSCs expressing Sox2 (i.e., Sox2⁺/Ki67⁺) was severely lowered in the KO CTL mice when compared with their WT littermates (p < .001; Fig. 2B, 2C). A similar reduction was also observed for the pool expressing Sox2⁺ (p < .001; Fig. 2B, 2C). In the KO RUN mice, we observed a reversion of this phenotype. Indeed in the Btg1-null mice, running was able to strongly increase the number of proliferating (Sox2⁺/Ki67⁺, p < .001; Fig. 2B, 2C) and total Sox2⁺ cells (p < .001; Fig. 2B, 2C) to a level comparable with that observed in the WT CTL mice (Fig. 2C). Moreover, the percentage of neural stem and progenitor cells recruited in the cell cycle (expressed by the ratio of Sox2⁺/Ki67⁺ cells to total Sox2⁺ cells) significantly increased in the KO RUN mice, relative to KO CTL (p < .05; Fig. 2C) as well as to WT CTL and WT RUN mice (p < .05 for both conditions; Fig. 2C). Finally, when we analyzed the WT RUN mice, we did not detect any significant difference with the WT CTL mice either for the proliferating (Sox2⁺/Ki67⁺, p = .8; Fig. 2C) or for the total Sox2⁺ cells (p = .9; Fig. 2C), indicating that in normal condition the beneficial impact of running did not involve an increment of the Sox2⁺ subpopulation of NSCs.

Figure 2

Subtype-specific increase of proliferating cells after physical exercise. (A): Milestones of neuronal development in the adult DG. (B): Representative fluorescence images of confocal optical sections of DG stained with Ki67 and Sox2. On the right panels, higher magnifications, corresponding to the boxed regions, show the large increase of proliferating Sox2⁺ cells in the Btg1 RUN mice. Scale bar = 100 µm, enlargements 50 µm. (C): The overall number of proliferating neural stem cells identified by expression of Ki67 and Sox2 was significantly increased in the KO RUN mice compared to the KO CTL (genotype × exercise interaction F_1,50 = 9.69, p < .05, followed by analysis of simple effects p < .001) as well as the number of neural stem cells Sox2 positive (genotype × exercise interaction F_1,50 = 9.64, p < .05, followed by analysis of simple effects p < .001). Note that the percentage of Ki67⁺/Sox2⁺ cells (ratio between Ki67⁺/Sox2⁺ cells and total Sox2⁺ cells) was greatly increased in KO RUN mice in comparison to the other experimental conditions (genotype × exercise interaction F_1,50 = 4.91, p < .05, followed by analysis of simple effects, p < .05 vs. KO CTL, WT CTL, and WT RUN mice). (D): Qualitative representation of radial glia-like neural stem cells (rNSC). The soma of these cells dwells in the subgranular zone, while a single apical process crosses the GCL and then arborizes profusely in the molecular layer. The rNSCs were characterized by the expression of both GFAP and Nestin (scale bar = 15 µm). (E): Representative confocal images showing the increased number in the KO RUN mice of dividing rNSCs, identified by means Ki67⁺/Nestin⁺/GFAP⁺, green, red, and blue, respectively; scale bar = 50 µm. (F): The quantification of the total number of NSCs showed a significant increment in the KO RUN mice if compared to KO CTL mice (genotype × exercise interaction F_1,55 = 4.23, p < .05, followed by analysis of simple effects p < .01). (G): Similarly, the number of total number of cycling cells increased in the KO RUN mice, when compared with the other experimental conditions (genotype × exercise interaction F_1,55 = 21.99, p < .001, followed by analysis of simple effects, p < .001 KO RUN vs. WT CTL, WT RUN, and KO CTL groups). (H): The percentage of rNSC recruited in the cell cycle (ratio between Ki67⁺/Nestin⁺/GFAP⁺ and total Nestin⁺/GFAP⁺ cells) was significantly increased in KO RUN mice (genotype × exercise interaction F_1,55 = 14.48, p < .001, followed by analysis of simple effects, p < .001 in comparison with the other experimental conditions). Cell numbers in the DG are means ± SEM of the analysis of at least three animals per group. *, p < .05; **, p < .01; ***, p < .001. Abbreviations: DG, dentate gyrus; gcl, granule cell layers; GFAP, glial fibrillary acidic protein; KO, knockout; sgz, subgranular zone; WT, wild type.

To further confirm these results, we counted the number of type-1a and type-1b NSCs, which represent the first step in the neurogenic cascade (Fig. 2D). Our data revealed that running was able to reverse the reduction of the NSC population in the Btg1 KO mice (p < .05; Fig. 2E, 2F). The increment of NSCs in the KO RUN mice was strictly dependent on the rise of proliferation rate as a consequence of exercise. Interestingly, in the KO RUN mice, we observed a fourfold increase of type-1b cells (Ki67⁺/Nestin⁺/GFAP⁺ cells) relative to the KO CTL mice (p < .001; Fig. 2E, 2G) and a threefold increase with respect also to WT CTL and to WT RUN mice (p < .001; Fig. 2G). Consistent with this, the proportion of type-1b cells recruited in the cell cycle (expressed as the ratio between Ki67⁺/Nestin⁺/GFAP⁺ to the total Nestin⁺/GFAP⁺) was significantly higher in the KO RUN mice compared to the other experimental conditions (p < .001; Fig. 2E, 2H). As expected, running did not exert any proproliferative effect in the NSCs of WT RUN mice (Fig. 2F–2H). These results suggested that NSCs of the Btg1 KO mice expressed a proliferative and neurogenic potential higher than NSCs of wild-type mice, and activable by running.

When we analyzed the proliferation of NeuroD1-positive cells in the DG, we detected a nearly twofold reduction of Ki67⁺/NeuroD1⁺ cells in the KO CTL mice compared to the WT CTL mice (p < .05; Supporting Information Fig. S2C). A similar decrement was also observed for the number of the NeuroD1 total cells (p < .001; Supporting Information Fig. S2D). In the Btg1-deficient mice, running greatly induced both the proliferation (p < .001; Supporting Information Fig. S2C) and the total number (p < .001; Supporting Information Fig. S2D) of NeuroD1-positive cells relative to the KO CTL mice. Altogether, these results clearly demonstrated that in the Btg1-null mice, running reactivated the burst of proliferation observed in early postnatal age.

It is also worth of note that in the WT RUN mice the number of proliferating NeuroD1 cells (Ki67⁺/NeuroD1⁺) as well as the whole subpopulation of NeuroD1⁺ progenitors was highly increased by running in comparison to WT CTL mice (p < .05 and p < .001; Supporting Information Fig. S2C, S2D), indicating that running specifically acted by promoting the proliferation of committed newly generated progenitor cells. These results suggested that in WT CTL mice the differences of proliferation rate between the GFAP/Sox2 and NeuroD1 subpopulations, revealed by running, might reflect a heterogeneity of cell cycle kinetics within the neurogenic pool.

Finally, we defined whether physical exercise would also rescue the defective neurogenesis in the SVZ of Btg1-null mice (Supporting Information Fig. S3). Using different markers to specifically detect type B NSCs (GFAP) and type A (DCX) progenitor cells, we demonstrated that running totally reactivated the postnatal hyperproliferation of subventricular newly generated neurons (Supporting Information Fig. S3C), while it did not provide any beneficial effects in the WT RUN mice, as previously shown 36.

Running Induces a Shortening of S-Phase and Cell Cycle Length in the Btg1 Null Mice

In order to determine if the increase in cell proliferation exerted by running was due to a shortening of progenitor cell-cycle length, we used a new method recently developed by others 29, 30. This protocol easily allows the precise calculation of the S phase (Ts; Fig. 3A, 3B) and cell cycle length (Tc; Fig. 3D, 3E) of distinct stem/progenitor cells within the adult DG (for detailed information see Materials and Methods). By adopting this procedure, we determined in the WT CTL mice an average Ts = 12.9 ± 0.77 hours (Fig. 3H) with 26.8% of cells out of S phase (Fig. 3C), and an average Tc = 24.95 ± 0.27 hours (Fig. 3H). In the KO CTL mice, we observed a significant lengthening of S phase (p < .01; Fig. 3H) with a strong decrease of cells that had left the S phase (p < .001; Fig. 3C), and a concomitant significant increase of Tc length (p < .001; Fig. 3H). Running resulted in a strong shortening of Ts and consequently of Tc lengths in both the genotypes analyzed. Indeed in the WT RUN group we calculated an average Ts of 10.25 ± 0.61 hours with 33.6% of cell exited from the S phase, both the values being significantly different compared with those obtained in the WT CTL group (p < .05 and p < .001, respectively, Fig. 3C, 3H). In Btg1-deficient mice, physical activity produced a significant reduction of S phase duration compared to the KO CTL (p < .001; Fig. 3H) and also to WT CTL mice (p < .01; Fig. 3H), reaching the value obtained in the WT RUN mice (Fig. 3H). We also detected in the KO RUN mice a strong increase of cells out of S phase and a significant faster Tc with respect to KO and WT CTL mice (p < .001; Fig. 3C, 3H) approaching the Tc measured in the WT RUN mice (Fig. 3H). These data indicated that the shortening of S-phase is a key regulator of running-dependent enhancement of proliferation, by promoting a faster progression of cycling cells and a shorter duration of the overall cell cycle length. However, a decrease of the cell cycle duration after running could result from changes in the length of one or more particular cell cycle phases. Thus, to investigate the combined length of the G2 and M phases, we used the percentage of labeled mitoses method, labeling the mitotic cells with antiphosphohistone H3 antibodies (see Materials and Methods). The maximal BrdU mitotic LI was reached 2.5 hours after the initial BrdU injection, regardless of the genotype and condition, indicating that the combined length of G2+M phases was comparable among the experimental groups (Fig. 3F, 3G). Consequently, we did not observe any significant difference in the G1 length, calculated as described in Materials and Methods.

Figure 3

Analysis of the cell-cycle length using the double pulse labeling method. (A): A 3-hour interval between an IdU and CldU injections produced three differentially labeled cell populations: IdU⁺/CldU⁻ cells (green, arrowhead) that exited S-phase of the cell cycle between the two injections, IdU⁻/CldU⁺ cells (red, arrow) that entered S-phase during the two injections interval, and IdU⁺/CldU⁺ double labeled cells (yellow) that were in S-phase during both injections (scale bar = 25 µm). (B): Experimental timeline of thymidine analog pulses for S-phase calculation. Exercising mice were allowed access to running wheels for 12 days. The last day of running, mice received a single i.p. injection of IdU (57.5 mg/kg), followed by a single equimolar i.p. injection of CldU (42.5 mg/kg) 2 hours later, and perfused 45 minutes thereafter. (B′): Equation for the length of S-phase. The ratio of cells that have left S phase during the 3 hours of interinjection interval (IdU⁺/CldU⁻ cells) to the total number of IdU⁺ cells is equal to 3 hours/Ts. (C): Histograms showing the percentage of cells out of S-phase after the double pulse method (calculated as the ratio of the IdU⁺/CldU⁻ to the Idu⁺ total cells). Running induced an increment of cell cycle exit in both the genotypes when compared with the respective sedentary mice (genotype × exercise interaction F_1,177 = 5.87, p < .05, followed by analysis of simple effects p < .001 KO RUN vs. KO CTL and WT RUN vs. WT CTL). Cell number in the dentate gyrus is mean ± SEM of the analysis of at least three animals per group. **, p < .01; ***, p < .001. (D): An interval of 20 hours between IdU and Cldu injections generates cycling cells that were in S-phase during the Idu injection only (IdU⁺/CldU⁻ cells; green, arrowheads), or during the CldU pulse (IdU⁻/CldU⁺ cells; red, arrow), or cells that re-entered in the cell cycle and were in S-phase during both the pulses (CldU⁺/IdU⁺ cells, yellow). Scale bar = 25 µm. (E): Experimental timeline of thymidine analog pulses for cell cycle calculation. Exercising mice were allowed access to running wheels for 12 days. The eleventh day, mice received a single i.p. injection of IdU (57.5 mg/kg), after 20 hours a single equimolar i.p. injection of CldU (42.5 mg/kg), and they were perfused 45 minutes thereafter. (E′): Equation for the length of the whole cell cycle. (F): Mitotic BrdU labeling index. The G2/M duration was estimated by the number of phH3+ cells that were also BrdU⁺ after 2 hours and 2.5 hours from BrdU exposure. Mitotic labeling index 1.00 = all mitotic cells (i.e., Ki67⁺/phH3⁺) stained with BrdU. (G): Orthogonal projection of triple-positive cells stained with Ki67 (blue), BrdU (green), and phH3 (red) after 2 (left) and 2.5 (right) hours of BrdU exposure. In the left panel, arrow and arrowhead indicated Ki67⁺/BrdU⁻/phH3⁺, and Ki67⁺/BrdU⁺/phH3⁺, respectively. After 2.5 hours (right panel), all cells were Ki67⁺/BrdU⁺/phH3⁺ (arrowhead). Scale bar = 25 µm. (H): Table showing the total cell cycle time, S-phase, G2/M phase, and G1 phase duration among the different experimental groups. All values are presented as means ± SEM except TG1. Statistical analysis for Tc and Ts reports genotype × exercise interactions (G × E) with p < .01, followed by analysis of simple effects. Abbreviations: gcl, granule cell layers; KO, knockout; sgz, subgranular zone; WT, wild type.

Furthermore, we wanted to analyze in depth the cell cycle length in distinct subpopulations of the DG, in order to distinguish the cell types whose proliferative kinetics is specifically affected by running. By combining triple immunofluorescence labeling (IdU/CldU/GFAP and IdU/CldU/NeuroD1; Fig. 4A, 4F) with the method described above, we analyzed the cell cycle length of type1-b cells, expressing the marker GFAP, and of the precursor cells positive for NeuroD1. Our data demonstrated that the Btg1-null mice showed a significant increase of S-phase length both in the GFAP⁺ and in the NeuroD1⁺ subpopulation relative to the WT CTL (p < .05 for GFAP and NeuroD1; Fig. 4B, 4D and Fig. 4G, 4I), which resulted in a significant lengthening of the cell cycle duration (Tc GFAP⁺ p < .05 Fig. 4B, 4D; Tc NeuroD1⁺ p < .05; Fig. 4G, 4I).

Figure 4

Effect of running on cell cycle length of GFAP⁺ and NeuroD1⁺ cell populations. (A): Orthogonal projections of triple labeling with IdU (green), CldU (red), separated by 3 hours, and GFAP (blue). A GFAP⁺ cell (blue) is shown that was in S-phase during both injections (Idu⁺ green and CldU⁺ red, arrowhead). Scale bar = 25 µm. (B–E): Panels representing the measured S-phase and cell cycle length for the stem GFAP⁺ cell population in the different experimental conditions, showing a shortening of Tc only in the KO RUN mice with respect to the KO CTL and WT CTL mice (genotype × exercise interaction F_1,73 = 4.07, p < .05, followed by analysis of simple effects p < .001 vs. KO CTL mice and p < .05 vs. WT CTL mice). (F): Orthogonal projection of triple labeling with IdU (green), CldU (red), separated by 3 hours, and NeuroD1 (blue). In this example, the arrowhead indicates a NeuroD1⁺ cell that was in S-phase during both injections (Idu⁺ green and CldU⁺ red), while the arrow indicates a NeuroD1⁻ cells in S-phase during both injections (Idu⁺ green and CldU⁺ red). Scale bar = 25 µm. (G–J): Panels representing the measured S-phase and cell cycle length for the NeuroD1⁺ cell population in the different experimental conditions. In this case running exerts a Tc shortening in both the genotypes if compared with their respective controls (genotype effect, F_1,64 = 6.5, p < .01; running effect F_1,64 = 18.71, p < .001). Abbreviations: gcl, granule cell layers; GFAP, glial fibrillary acidic protein; KO, knockout; sgz, subgranular zone; WT, wild type.

In the wild-type mice, the physical activity did not have any effect on the cell cycle kinetics of the GFAP⁺ subpopulation that displayed Ts and Tc comparable to those calculated in the WT CTL mice (p = .97 and p = .57, respectively; Fig. 4B, 4C). In contrast, we observed that running strongly accelerated the S phase of the NeuroD1⁺ cells when compared with the WT CTL mice (p < .05; Fig. 4G, 4H), resulting in a shortening of the overall duration of the cell cycle (p < .05; Fig. 4G, 4H).

When we analyzed the Btg1 null mice, we found that the cell cycle kinetics of both the two subpopulations was deeply affected by running. Indeed the S phase progression of GFAP⁺ and NeuroD1⁺ cells was significantly accelerated when compared with the KO CTL (p < .05 for GFAP and p < .001 for NeuroD1; GFAP Fig. 4D, 4E; NeuroD1 Fig. 4I, 4J) and also to the WT CTL mice (p < .05 for GFAP and p < .001 for NeuroD1; GFAP Fig. 4B, 4E; NeuroD1 Fig. 4G, 4J). Consequently, a specific effect of running in the KO RUN mice was a Tc significantly shorter both in the GFAP⁺ and NeuroD1⁺ cells with respect to KO CTL mice (GFAP p < .001; Fig. 4D, 4E; NeuroD1 p < .001; Fig. 4I, 4J) as well as to WT CTL mice (GFAP p < .05; Fig. 4B, 4E; NeuroD1 p < .001; Fig. 4G, 4J). We then analyzed the cell cycle kinetics with the NSCs marker Sox2 (IdU/CldU/Sox2), and we obtained similar results to those observed with the GFAP immunostaining (Supporting Information Fig. S4B).

These results suggested that in physiological conditions running enhanced proliferation with a specific action, by shortening the S phase and consequently the cell cycle duration of committed NeuroD1⁺ progenitors, thereby increasing the numbers of new neurons being generated but not the number of stem cells. In Btg1-deficient mice, the replenishment of the pool as well as the burst of proliferation observed after physical activity mainly resulted in S-phase and overall cell cycle faster progression of both NSCs (GFAP⁺ and Sox2⁺) and early differentiating (NeuroD1⁺) progenitors.

In order to directly relate changes in cell cycle kinetics to the increase of cell division number and consequently of type1-b population, we tracked proliferation of NSCs by pulse-labeling experiment. Mice received a single injection of BrdU to label one cohort of dividing cells, and the number of type1-b cells (identified by BrdU/Sox2/GFAP) was analyzed over the course of 72 hours (Supporting Information Fig. S5A). After 2 hours, we did not detect any significant decrease of type1-b cells in the KO CTL mice with respect to the WT CTL mice (as previously observed in Fig. 1D for the BrdU⁺ total cells), suggesting that in the Btg1-null mice the lengthening of S-phase results in an accumulation of the fraction of BrdU⁺ cells labeled after a short pulse (Supporting Information Fig. S5C). However, in the KO CTL mice, the number of type1-b cells declines over time, whereas running induced a large expansion of NSCs specifically in the Btg1-null mice (Supporting Information Fig. S5B, S5C), clearly providing a direct link between cell cycle shortening and type-1 stem cells expansion in the KO RUN mice.

Effect of Wheel Running on Neurogenesis and Early Maturation

Afterward, we asked whether this strong increment of progenitor cells proliferation in the KO RUN mice could directly produce an increase of neurogenesis of young (1–5-day old) newborn neurons. To this aim, mice had free access to a running-wheel for 12 days, and during the last 5 days both RUN and CTL groups were treated with five daily injections of BrdU, in order to detect new 1–5-day-old progenitors and neurons (Fig. 5A). Our data showed that KO RUN mice displayed a high increase compared to the KO CTL mice of BrdU total cells (p < .001; Fig. 5B, 5I), as well as of the total NSCs identified by nestin (p < .001; Fig. 5C), and of Dcx differentiating progenitor cells (p < .001; Fig. 5D). In these experimental paradigm, the running wheel leads the KO RUN mice to generate a number of BrdU-positive as well as nestin- and Dcx-positive cells equivalent to that in WT CTL mice. In order to display which differentiation stage was mainly associated with the effects of running in the Btg1-null mice, we analyzed the different neurogenic markers in the proliferating BrdU⁺ stem and progenitor cells. Our results showed that exclusively in the KO RUN mice, the physical exercise strongly increased both the type-1a (Nestin⁺GFAP⁺; Fig. 5E) and the type-1b (BrdU⁺Nestin⁺GFAP⁺; Fig. 5F) NSCs. Moreover in the KO RUN mice we found a large increment of the type-2a cells (BrdU⁺Nestin⁺GFAP⁻, Fig. 5G) and of the proliferating type-3 cells (BrdU⁺Nestin⁻Dcx⁺, p < .001; Fig. 5I), whose number was significantly higher than the other three experimental conditions (4.8-, 3.5-, and 1.5-fold increase with respect to KO CTL, WT CTL, and WT RUN mice, respectively). Finally, we observed that in WT RUN mice voluntary wheel running preferentially promotes the proliferation of progenitor cells of the neuronal lineage type-2 (BrdU⁺Nestin⁺Dcx⁺) and type-3 (BrdU⁺Nestin⁻Dcx⁺) as previously suggested (20, 37, respectively).

Figure 5

Increased number and differentiation rate of 1–5-day-old newborn neurons after run. (A): Schematic diagram of the experimental paradigm and BrdU injection protocol. Exercising mice were allowed access to running wheels for 12 days. BrdU (95 mg/kg) was injected the last 5 days of running. (B–D): Quantification of the number of total BrdU-positive, total nestin-positive, and total Dcx-positive cells indicated a significant increase in WT RUN as well as KO RUN mice compared to their respective controls (Brdu, genotype effect, F_1,93 = 9.1, p < .001; running effect F_1,93 = 23.5, p < .001; Nestin, genotype × exercise interaction F_1,67 = 10.67, p < .01, followed by analysis of simple effects p < .001 for both genotypes; Dcx, genotype × exercise interaction F_1,110 = 4.73, p < .05, followed by analysis of simple effects p < .05 WT RUN vs. WT CTL and p < .001 KO RUN vs. KO CTL mice). (E–I): Histograms showing the increase in the KO RUN mice with respect to the KO CTL mice of new 1–5-day-old type-1a (Nestin⁺/GFAP⁺, genotype × exercise interaction F_1,50 = 24.89, p < .001, followed by analysis of simple effects p < .001), type-1b (Brdu⁺/Nestin⁺/GFAP⁺, genotype × exercise interaction F_1,50 = 44.28, p < .001, followed by analysis of simple effects p < .001), type-2a (Brdu⁺/Nestin⁺/GFAP⁻, genotype × exercise interaction F_1,50 = 14.93, p < .001, followed by analysis of simple effects p < .001), type 2-b (BrdU⁺/Nestin⁺/Dcx⁺, genotype × exercise interaction F_1,83 = 28.57, p < .001, followed by analysis of simple effects p < .001), and type-3 (BrdU⁺/Nestin⁻/Dcx⁺, genotype × exercise interaction F_1,83 = 5.9, p < .01, followed by analysis of simple effects p < .001) stem and progenitor cells in Btg1 null mice. (J): Quantification of the total number of postmitotic 1–5-day-old neurons (stage 5; BrdU⁺/Dcx⁺/NeuN⁺) indicated a large increase in WT RUN, but not in the KO RUN if compared with their respective controls (WT RUN mice, genotype × exercise interaction F_1,111 = 12.64, p < .001, followed by analysis of simple effects p < .001 vs. WT CTL; KO RUN mice, genotype × exercise interaction F_1,50 = 44.28 followed by analysis of simple effects p > .05 vs. KO CTL). (B–J): Cell number in the DG is mean ± SEM of the analysis of at least three animals per group. *, p < .05; **, p < .01; ***, p < .001. (K): Representative images showing an increase in the KO RUN mice DG of neurogenesis, as detected by incorporation of BrdU (green) and by Dcx (red) and NeuN (blue) multiple-labeling confocal microscopy. In the left panel, higher magnifications correspond to the boxed regions and show the large increase of proliferating Dcx⁺ 1–5-day-old cells in the Btg1 RUN mice. Scale bar = 100 µm, enlargements 50 µm. Abbreviations: DG, dentate gyrus; GFAP, glial fibrillary acidic protein; KO, knockout; WT, wild type.

Type-3 progenitor cells give rise to early postmitotic neurons identified by expression of neuronal-specific differentiation marker NeuN (stages 5 and 6). We found that in KO RUN mice no significant increase of postmitotic neurons expressing Dcx and NeuN occurred, relative to neurons counted in KO CTL mice (p = .1; Fig. 5J). In contrast, in WT RUN mice, the early differentiated newborn neurons (1–5-day old) increased greatly relative to WT CTL mice (63% increase; p < .05; Fig. 5J), and also to KO RUN mice (twofold increase; p < .05; Fig. 5J).

These results showed that the decrease in adult neurogenesis observed in the Btg1-null mice model is completely reversed by voluntary physical activity, and that the neuronal lineage type-3 progenitor cells and to a lesser extent type 1–2a are the cell subpopulation targets of running in the KO RUN mice. However, we observed that the rate of differentiation and early maturation displayed by the new-born neurons of KO RUN mice is slower with respect to its proliferation rate, suggesting that in this mouse model running recruited former quiescent cells for cell division more than increasing their postmitotic differentiative fate.

Finally, we found that physical exercise in KO RUN mice can prevent the massive apoptotic-dependent NSCs loss, previously observed in Btg1-null mice 27, while it has no antiapoptotic effect in physiological condition (Supporting Information Fig. S6C, S6D), as recently reported 38, 39. These data suggest that the reduction of apoptosis triggered by running may in part explain the increased neurogenesis observed in the KO RUN mice. In line with these data, we observed that running led to a significant improvement of pattern separation in the KO RUN mice, even though not sufficient to restore the functioning of memory processes observed in the WT CTL mice (Supporting Information Fig. S7 and Supporting Information text for more details).

Running Induces in Btg1 Null Mice a Long-Lasting Benefit on Adult Neurogenesis

We next investigated the duration of the activity-induced augmentation of neurogenesis and whether running was able to maintain hippocampal plasticity into older age. To this aim Btg1 wild-type and KO mice were housed in running conditions for 12 days, and then sacrificed 14, 30, and 90 days after. We performed five daily injections of BrdU before perfusion (Fig. 6A) and used BrdU-incorporation to measure proliferative activity, DCX-immunoreactivity to assess the number of neuronally determined precursor cells, and BrdU/Dcx/NeuN labeling to estimate the amount of newborn neurons produced. Our data showed that 15 days after the end of run, the number of proliferating and differentiating cells (Dcx⁺) in the WT RUN and KO RUN mice was significantly higher in comparison with their respective control mice (Fig. 6B–6D). After 30 days, the trend observed previously was maintained even though the production of new neurons (BrdU⁺/Dcx⁺/NeuN⁺ cells) observed in the KO RUN mice significantly exceeded that measured in the WT CTL and WT RUN mice (Fig. 6B–6D). Finally, 90 days after the end of the run period, the values of neurogenesis detected in the WT CTL mice reached those observed in the KO CTL mice, and running ceased to provide any beneficial effects on neurogenesis in the WT RUN mice. Surprisingly, in the Btg1 null mice, the physical exercise exerted 90 days earlier, still produced a strong proneurogenetic effect; indeed all the three parameters analyzed (proliferation, differentiation, and production of new neurons) were significantly increased when compared with the other experimental conditions (Fig. 6B–6D). Altogether, these results suggested that in normal condition running had only a transient beneficial effect on neurogenesis, probably due to the exhaustion of the proliferative wave strictly dependent on the committed neural progenitors. In contrast in the Btg1 KO mice, the large increase of NSCs after running would translate into an expansion of the existent pool of NSC that might end into a long-term effect on adult neurogenesis (Figs. 6B–6D, 7).

Figure 6

Proneurogenic long-term effect of running in the Btg1 null mice. (A): Schematic diagram of the experimental paradigm and BrdU injection protocols. Exercising mice were allowed access to running wheels for 12 days and sacrificed after 15, 30, and 90 days. BrdU (95 mg/kg) was injected the last 5 days before the sacrifice. (B–D): Quantification of Brdu⁺, Dcx⁺, and BrdU⁺/Dcx⁺/NeuN⁺ total cells at the different time point analyzed. Statistical analysis reports genotype effect (G), running effect (R), and genotype × exercise effect (G × R) followed by analysis of simple effects. The statistical analysis of the first point of the graphs (72 days) is shown in Figure 5. NS = not significant. Cell number in the DG is means ± SEM of the analysis of at least three animals per group. (E): Representative images showing the increase of the BrdU⁺ cells (red) and Dcx⁺ cells (green) in the KO RUN mice when compared with KO CTL mice, 90 days since the end of run. Scale bar = 100 µm. Abbreviations: DG, dentate gyrus; KO, knockout; WT, wild type.

Figure 7

A model for the effect of running in the Btg1 wild-type and knockout mice. (A): In the wild-type mice running induced cell cycle shortening and consequently an increased proliferation of NeuroD1 progenitor cells. These events resulted in a transient proneurogenic effect, since 90 days after the end of the run the neurogenesis rate declined to the control level. (B): In the Btg1 knockout mice, running leads to a shortening of the cell cycle length of GFAP⁺/Sox2⁺ neural stem and Neurod1⁺ progenitor cells, inducing an expansion of these subpopulations and more specifically allowing the reactivation of the neural stem cell pool. This latter event is critical for the long-term proneurogenic effect of running, as demonstrated by the increased neurogenesis observed the KO RUN mice 90 days after the end of run. In Btg1 knockout mice, running is able to re-establish a baseline of higher proliferation in neural stem and progenitor cells, reinstating the hyperproliferation observed in early postnatal age; this also suggests that the adult pool of Btg1-null mice does not undergo full depletion, due to the slower cell cycle kinetics of the adult neurogenic proliferation. Abbreviations: KO, knockout; WT, wild type.

Discussion

In this study, we show that voluntary physical activity is able to fully reverse the deficits in adult neurogenesis observed in the Btg1-null mouse model. More interestingly, we demonstrate that the running-induced rescue of proliferation was correlated with the cell cycle kinetics, providing for the first time clear evidence that the S-phase shortening represents an intrinsic major regulator of the proneurogenic effect exerted by running. It is well known that small changes in the cell cycle length could result in significant changes in the number of neurons produced, even though there is no direct evidence about the effect of running on cell cycle progression 40.

Our study provides clear proofs of a large heterogeneity of cell cycle kinetics within the hippocampal pool of adult stem and progenitor cells in response to physical exercise. In this regard, we adopted a novel method to calculate, in running mice, the S-phase and cell cycle length 29 of the Sox2⁺ and NeuroD1⁺ subpopulations of the DG, whose expression do not overlap (35, 41, 42 and our unpublished observations).

Our results demonstrate that in the wild-type mice running leads to a cell cycle shortening and to an expansion of the NeuroD1⁺ progenitors (type-2b and −3), and that this increase would translate into a transient proneurogenic effect, very likely due to the progressive differentiation of the neuronally committed NeuroD1⁺ progenitors after the withdrawal of the stimulus. This is in accordance with previous results showing that voluntary wheel running primarily promotes the proliferation of transient amplifying progenitor cells (type-2 and type-3) 20, 43 and also their survival and maturation into functional hippocampal neurons 44. However, other studies propose that running enables an expansion of NSCs 34, 45. Our data, also supported by the detailed analysis of the cell cycle kinetics, are thus more consistent with the classic view of the transient amplifying progenitors expansion, where NSCs are not recruited into the active NSCs pool 46.

Moreover, this work shows that a short period of voluntary physical activity induces in the Btg1-null mice a striking reduction of the S-phase length, and consequently of the cell cycle, in the GFAP⁺ (type-1), Sox2⁺ (type1–2ab), and NeuroD1⁺ DG subpopulations. More interestingly, we find that in the Btg1 KO mice, running can activate quiescent NSCs (type-1a) and induce them to proliferate and expand, indicating that they can be targets to reinstate neurogenesis. This finding is intriguing, in particular because it suggests that cell-cycle kinetics might directly control the NSCs pool size. The regulation of NSCs quiescence by running in the Btg1-null mice has important implication for long-term modulation of adult neurogenesis, as demonstrated by the large increase of neurogenesis and production of new neurons observed in the Btg1 KO mice 90 days after the end of run. This increase of new neurons is functionally effective, as it rescues the defect of Btg1 KO mice to discriminate among overlapping memories (pattern separation). Conceptually, there are two possible explanations for the increase of NSC pool in the Btg1-deficient mice after running. First, a fraction of quiescent NSCs may constitute a reserve pool or backup that can be promptly activated in response to running when the antiproliferative control of Btg1 is missing. Another plausible hypothesis is that in this mouse model running can induce, in the absence of the proliferative restraint exerted by Btg1, a higher proliferation rate, re-establishing a “baseline” of the neurogenic pool, at the level of hyperproliferation observed in postnatal age 27. These results reveal an intrinsic property of the adult neurogenic pool, that is, an additional reserve of plasticity in response to external stimuli, in term of recruitment, cell cycle kinetics, and proliferation rate. Recently, it has been reported that a significant increase in the number of NSCs undergoing cell division occurs after systemic injection of kainic acid, a glutamate agonist inducing acute seizure 47, and after administration of electroconvulsive shock 48. Although in both experimental procedures an elevated rate of NSCs proliferation was detected, these studies did not analyze if the increase was mirroring an expansion of the existent NSC pool, nor if it had a long-term effect on neurogenesis.

It is also worth noting that in Btg1 KO mice running increases also the population of neuronally committed progenitor cells (type-2b-3, NeuroD1+); such an increase is comparable to that induced by running in wild-type mice, suggesting that the running-dependent expansion of NSC in Btg1 KO mice is not matched by an increase in the differentiation of the neuronally committed progenitor cells. The increased activation of NSC and of neuronally committed progenitors as well as the long lasting enhanced generation of new neurons occurring in Btg1 KO mice after exercise is summarized in the model of Figure 7.

As we have previously shown, the loss of proliferative capacity occurring in adult Btg1 KO DG NSCs is an age-dependent process of increased exit from the cell cycle that also involves an increased expression of the cyclin-dependent kinase inhibitor p21 27. Thus, the striking recovery of proliferative capacity in Btg1-null adult NSCs by a stimulus triggering neurogenesis may imply the action of p21; in fact, p21 or the cyclin-dependent kinase inhibitor p57 play a role similar to that of Btg1 in maintaining the quiescence of NSC 49, 50. Furthermore, the observation that running expands type-1 stem cells only in Btg1 KO mice, while type-2–3 progenitor cells are expanded also in Btg1 wild-type mice, suggests a different responsivity of the two populations to the inhibitory action by Btg1 on cell cycle. It has been shown that Btg1 controls the expression of D cyclins 51; these, in turn, through the phosphorylation of pRb rule the cell cycle restriction point, that is, the decisional point controlling the entrance into the cell cycle 52. Given that type-1 stem cells are a prevalently quiescent population, we can speculate that their responsivity to running is specifically inhibited by Btg1 as they remain mainly in a quiescent state directly under control of the restriction point, unlike the proliferating type-2–3 progenitor cells already committed to cycle.

A further possibility to be considered is whether the decrease of apoptosis occurring in nestin⁺caspase⁺ cells of KO RUN mice, relative to KO CTL mice, is somewhat responsible for the increased neurogenesis in KO RUN mice. At least in part, we cannot exclude this possibility; however, the number of apoptotic nestin⁺ cells in KO RUN mice is the same as in WT CTL and WT RUN mice, while the number of nestin⁺ cells in KO RUN mice is far greater than in WT CTL and WT RUN mice, suggesting that a decrease of apoptosis in itself is not sufficient to account for the large increase of new neurons.

Further work will be useful to understand the molecular mechanisms of Btg1 in maintaining NSC quiescence and the interactions with other pathways. These may also include another Btg family-related gene, Tis21/Btg2, which appears to be required for terminal differentiation of the DG progenitor cells 28, 53, and thus could play a role in sustaining the expansion of NSC throughout the exercise-induced neurogenesis.

Conclusions

This work is, to our knowledge, the first to provide clear evidence that S-phase length is one key regulator of the proneurogenic effect of running, specifically involving the NeuroD1+ subpopulation of the DG neurogenic pool. Moreover, we describe that physical activity induces a complete recovery of the impaired neurogenesis in the Btg1-deficient mice, mainly through a dramatic expansion of type-1b cells. At the origin of this action is the shortening of the cell cycle of Btg1 KO stem and also progenitor cells exerted by running. In turn, these events result in a long-term increase of neurogenesis and production of newborn neurons. These findings also highlight the key role of Btg1 in maintaining the quiescence of adult NSC.

Acknowledgments

This work was supported by the Italian Ministry of Economy and Finance to CNR (Project FaReBio; to Felice Tirone) and by the Associazione Italiana Ricerca sul Cancro project #9251 to Felice Tirone. Laura Micheli was supported by FILAS (Finanziaria Laziale di Sviluppo). We thank Diego Centonze for kindly providing the running wheel cages.

Author Contributions

A.M.: conception and design, collection and/or assembly of data, data analysis and interpretation; L.M., M.C., and L.L.: collection and/or assembly of data; J.-P.R.: provision of study material; D.S., M.C., and V.C.: neurobehavioral experiments and text, and ANOVA statistics; S.F.-V.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; F.T.: conception and design, data analysis and interpretation, drafting and critical revision of manuscript, final approval of manuscript, and financial support. S.F.V. and A.M. contributed equally to this article.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.