Author contributions: M.D.B.: conception and design, collection and assembly of data, data analysis and interpretation, drafting and critical revision of manuscript, and fund raising; M.H.: collection and assembly of data and critical revision of manuscript; A.S.: conception and design, data analysis and interpretation, drafting and critical revision of manuscript, and fund raising.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS September 17, 2012.
Cell cycle analyses of adult hippocampal neural stem and precursor cells in vivo are challenging, as there is no temporal or local discrimination of different precursor cell populations. All commonly used techniques to determine the cell cycle length of proliferating cells in the adult hippocampus do not allow discrimination between different cell types. Here, we introduce a novel procedure to precisely calculate cell cycle phase lengths of distinct precursor cell populations in vivo and thereby demonstrate a large heterogeneity of cell cycle kinetics within the pool of adult hippocampal precursor cells. Proliferating NeuroD1+ cells exhibited a significantly faster S-phase progression (Ts = 10.1 ± 0.6 hours) and shorter total cell cycle length (Tc = 22.6 ± 0.1 hours) than NeuroD1− cells (Ts = 13.5 ± 0.8 hours, Tc = 27.0 ± 0.5 hours; p < .05). Dividing glial fibrillary acidic protein (GFAP+) cells also showed significantly shorter mean Ts of 9.7 ± 0.6 hours and Tc of 22.8 ± 0.5 hours compared to the rest of uncommitted NeuroD1− precursors (p < .01). Together, NeuroD1+ neuronal progenitors and mitotic GFAP+ radial glia-like cells divide significantly faster than amplifying neural progenitor cells by accelerating their S-phase. S-phase duration seems to determine cell cycle length in the adult hippocampus. STEM CELLS 2012;30:2843–2847
The dentate gyrus of the hippocampus maintains the capability to generate new neurons throughout adulthood. This cellular plasticity is mainly controlled by mitotic activity of adult neural stem and precursor cells. Labeling cells that are in S-phase of their cell cycle with thymidine analogs 5-bromo-2-desoxyuridine (BrdU) has become the most common method for quantitative analyses of proliferating cells and their progenies, adult-born new neurons. The short half-life of BrdU in the brain after i.p. injection of approximately 0.5 hours  allows a pulse labeling of all cells that are currently in S-phase without labeling cells in G2/M or G1 phase. Thus, the relative length of S-phase compared to whole cell cycle length defines the number of proliferating cells that are labeled by BrdU. As a consequence, it is essential to know the length of the S-phase (Ts) in relation to the total cell cycle length (Tc) to estimate the size of the proliferating cell pool after BrdU labeling. So far, there is however no reliable approach to calculate the length of cell cycle phases in distinct cell types within the adult hippocampus, where no temporal or local discrimination of different precursor cells exist. A major open question thus is whether a high BrdU-labeling index of a distinct precursor population (e.g., type 2 cells in the hippocampus) is equivalent to a fast dividing highly proliferative cell pool or rather to a longer S-phase, which increases the chance to be labeled. We thus introduce a novel approach to precisely calculate Tc of distinct precursor cell populations with an easy to accomplish dual pulse labeling approach.
MATERIALS AND METHODS
Animals, Tissue Preparation, and Immunohistochemistry
Eight-week-old female C57/Bl6 mice received a single i.p. injection of 57.5 mg/kg b.wt. IdU (iododeoxyuridine) in 0.2 N NaOH 0.9% NaCl solution, followed by a single i.p. injection of 42.5 mg/kg b.wt. chlorodeoxyuridine (CldU) in sterile 0.9% NaCl solution according to the experimental protocol outline in Figure 1. Staining of IdU and CldU in brain slices was performed using standard procedures for fluorescence or diaminobenzidine detection similar to the method introduced by Dimitrova et al.  with modifications according to Vega and Peterson . Detection specificity was tested in mice that were either injected with IdU or CldU only (Supporting Information Fig. S1).
For total cell counts, IdU+ or CldU+ cells stained using the diaminobenzidine method (Supporting Information Fig. S2) were counted throughout the rostro-caudal extent of the granule cell layer and subgranular zone of the hippocampus. The optical dissector method was adopted as described previously . For proportional analyses of double-labeled cells and cell phenotype, IdU+ and CldU+ cells within the granule cell layer of each animal were analyzed using confocal microscopy in sequential scanning mode for detecting double labeling and coexpression of the various cell type markers. Double labeling was confirmed by z-series of the entire nucleus or cell. Total number of cells with a specific marker expression (e.g., total number of IdU+/NeuroD1+ double-labeled cells) was estimated by the total cell count multiplied with the ratio as the result of the proportional analyses. See Supporting Information, Materials and Methods Section, for complementary information on immunohistochemistry, quantification, and statistics.
Calculating Tc and Cell Cycle Phase Durations of Proliferating Cells in the Subgranular Zone (SGZ)
To initially determine Ts, adult mice received an IdU injection 3 hours prior to a CldU-injection and were scarified 45 minutes thereafter (T3 animals, Fig. 1A, 1B; Supporting Information Text for methodological aspects and bioavailability of halogenated thymidine analogs). The same S-phase was labeled by both compounds in some cells, while others did only incorporate IdU or CldU due to either leaving of S-phase within the 3 hours interinjection interval or entering S-phase after IdU but before CldU-injection (Fig. 1A). As described earlier , the ratio of cells that left S-phase during the 3 hours interval (IdU+/CldU− cells) to the total number of IdU+ cells is equal to 3 hours/Ts (equation 1 in Fig. 1B). To subsequently determine Tc, we chose an interinjection interval that is longer than Ts and Tc − Ts but shorter than the expected duration of 2× Tc − Ts to make sure that we label neither the same S-phase twice nor the next but one S-phase. For that reason, we chose interinjection intervals of 20 hours and as an internal control 16 hours (T20 and T16 animals). Tc can then be calculated using equation 2 in Figure 1C, 1D.
In T3 animals, we found that 24.8% ± 0.7% of total IdU+ cells was not colabeled with CldU and therefore had left S-phase within the last 3 hours (for total cell numbers, Fig. 2A). Equation 1 revealed mean Ts of 12.1 ± 0.3 hours. In T20 animals, 44.7% ± 1.4% of cells in S-phase (CldU+) was in another cell cycle phase 20 hours before (CldU+/IdU− cells). Inserted into equation 2, mean Tc was 25.3 ± 0.2 hours. Values from T16 animals revealed a very similar result of 25.8 ± 0.5 hours confirming the reliability of the method.
Since the IdU is passed to two daughter cells after leaving G2 and M-phase, the number of IdU+ cells increases over time. We indeed detected a doubling in the number of IdU+ cells compared to the number of CldU+ cells in T20 animals with no significant change in the number of IdU+ cells between T16 and T20 animals showing that every initially labeled cell had already passed through G2/M in T16 animals (Fig. 2A–C). G2/M was thus shorter than 16 hours 45 minutes (time after IdU injection) minus Ts (=4.65 hours). Conversely, there was no difference in the total numbers between CldU+ and IdU+ cells in T3 animals, thus no IdU+ cell had passed G2/M-phase 3 hours 45 minutes after IdU injection. We obtained approximate values for the duration of G2/M between 3.75 hours and 4.65 hours (∼4.2 hours) and 9.0 hours for G1-phase (Fig. 2C).
Precursor Cell Populations Differ in Mitotic Tempo
It is broadly assumed that the pool of proliferating cells in the adult hippocampus is heterogeneous with quiescent and proliferating radial glia-like cells (type 1 cells), so called highly proliferative amplifying neural progenitor cells (type 2a cells), and dividing intermediate progenitor cells and neuroblasts (type 2b and type 3 cells) [6, 7]. The major advantage of the dual pulse labeling method is that it affords a distinct cell cycle analysis for all precursor cell populations (Fig. 2D–F). Type 2 (neuronal committed) progenitors cells start to express the transcription factor NeuroD1, whereas neural precursor cells (type 1 and the majority of type 2a cells) are NeuroD1-negative . Consistently, we found no overlay between neural stem cell marker Sox2 and neuronal precursor marker NeuroD1 (Supporting Information Fig. S3). Using Nestin-GFP mice (expressing green fluorescence protein [GFP] under the Nestin gene promoter) to investigate the expression pattern of the transcription factor NeuroD1 within the Nestin-GFP+ type 1, type 2a, and type 2b cell populations, we detected 71.6% ± 2.5% of NeuroD1+ cells expressing Nestin-GFP and 51% ± 2% of Nestin-GFP+ cells being positive for NeuroD1 (Supporting Information Fig. S3). One third of Nestin-GFP+ cells (32% ± 2%) expressed Sox2, a transcription factor known to be expressed in radial glia-like stem cells (type 1 cells) and their progeny (early type 2 cells) (Supporting Information [15, 16]. Since we found a significant amount of Nestin-GFP+ cells that neither expressed Sox2 nor NeuroD1 (18% ± 4%), transition from Sox2+ to NeuroD1+ precursor cells most likely takes place at the developmental stage of type 2a cells (Supporting Information Fig. S3c). NeuroD1 seems thus to be an appropriate marker to distinguish between early neural stem/precursor cells (NeuroD1−) and late neuronal progenitor cells (NeuroD1+) in the adult mouse hippocampus (see Supporting Information Text for more details).
The majority (74% ± 6%) of cells currently being in S-phase (CldU+) were NeuroD1−, whereas 26% ± 6% expressed NeuroD1 (Fig. 2F). These proliferating NeuroD1+ cells exhibited a significantly faster S-phase progression (Ts = 10.1 ± 0.6 hours) and shorter Tc (22.6 ± 0.1 hours) than NeuroD1− cells (Ts = 13.5 ± 0.8 hours, p = .015; Tc = 27.0 ± 0.5 hours, p = .0024, Fig. 2F). We found 16% ± 3% of CldU-labeled cells being positive for glia-marker glial fibrillary acidic protein (GFAP) (Fig. 2F). It has been assumed that radial glia-like cell divide faster than their progenies once they left the quiescent state . Our results confirmed this hypothesis for dividing GFAP+ type 1 cells within the dentate gyrus with significantly shorter mean Ts of 9.7 ± 0.6 hours and Tc of 22.8 ± 0.5 hours compared to the rest of uncommitted NeuroD1− precursors (p = .017 for Ts and 0.01 for Tc; see Supporting Information for discussion on quiescent type 1 cells).
Since there was no difference of IdU-labeled GFAP+ cells over time (Fig. 2G), it is likely that the majority of proliferating GFAP+ cells divided asymmetrically, giving rise to GFAP− amplifying neural progenitor type 2 cells. In contrast, the number of new born NeuroD1+ cells (colabeled with IdU) increased approximately threefold after 16 hours, whereas there is no significant difference in the number of new born NeuroD1− cells (Fig. 2G). Together, we suggest that type 2 and type 3 cells divide symmetrically, and that type 2 cells start to express NeuroD1 soon after division. The main increase of proliferating cells during the first 20 hours is thus due to the production of NeuroD1+ cells by symmetric cell division as well as progression of neural precursor cells to neuronal progenitors.
The introduced method allows the precise calculation of cell cycle phase lengths of distinct (stem) cell populations within tissues, in which temporal or local discrimination of the different cell populations is impossible. Our study thereby demonstrates for the first time a large heterogeneity of cell cycle kinetics within the pool of adult hippocampal precursor cells. NeuroD1+ neuronal progenitor cells and proliferating GFAP+ radial glia-like cells divide significantly faster than amplifying neural progenitor cells by accelerating their S-phase. Another major finding is that the sum length of G1, G2, and M-phase was similar in all cell types (ranging from 12.5 hours to 13.5 hours) indicating that Tc is mainly determined by S-phase duration. Differences in S-phase duration are also reported in embryonic neural stem cells depending on whether the cells are expanding or committed to neuron production . Here, the authors suggest that the pool of expanding neural stem and precursor cells invest more time in S-phase for quality control of DNA replication than those committed to neuron production . However, our results indicate that in the adult hippocampus, it is only the expanding population of uncommitted precursor cells (NeuroD1−, GFAP−) that invest more time in S-phase and therefore displays a longer total cell cycle. In contrast, proliferating GFAP+ (type 1) cells divide faster than their progenies once they are activated, which is congruent with the data of Encinas et al. who had introduced the “disposable stem cell” model proposing that a stem cell is quiescent for the entire postnatal life, is activated, undergoes rapid asymmetric divisions producing progeny, and quits the pool of stem cells by differentiation . Quality control could be one reason for a longer S-phase, but in the adult brain there rather might be another advantage: A longer S-phase with a consecutive longer cell cycle and replication rate of uncommitted type 2a precursor cells could be an important regulatory mechanism to control the number of new born cells. A restricted cell cycle rate could be the precondition to regulate and control adult hippocampal neurogenesis without running the risk of depleting the pool of distinct precursors after increasing their proliferation and accelerate their differentiation. Further studies will show whether, for example, physical exercise, a stimulus of precursor cell proliferation and differentiation [6, 11], affects the cell cycle kinetics of distinct (former slowly dividing) precursor populations.
Moreover, our results show that the so far called highly proliferative type 2a cells divide relatively slowly but remain in S-phase for a relatively long time, which partly explains the high amount of labeled cells shortly after CldU injection. Another consequence of the presented data is that cumulative labeling (e.g., commonly used protocols with daily injections for several days) with an interinjection interval of 24 hours likely labels the same cells (at least during the first 2 or 3 days). We thus recommend an interinjection interval of 13 hours (Tc − Ts) to label the highest amount of mitotic cells in the dentate gyrus.
We thank Felix Brandt for fruitful discussions, Johannes Schwarz for comments on the manuscript, and Sylvia Kanzler, Andrea Kempe, and Cornelia May for technical assistance. The study was supported in part by the Deutsche Forschungsgemeinschaft (DFG) through the SFB655 “From cells to tissues” (project A23 to A.S.) and the Research Centre/Cluster of Excellence “Center for Regenerative Therapies Dresden (CRTD)” (to M.D.B. and A.S.), and the intramural research program MeDDrive of the Medical Faculty Carl Gustav Carus at Dresden University of Technology (to M.D.B.).
DISCLOSURE OF POTENTIAL C ONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.