Research Article

Oscillatory Control of Factors Determining Multipotency and Fate in Mouse Neural Progenitors

See allHide authors and affiliations

Science  06 Dec 2013:
Vol. 342, Issue 6163, pp. 1203-1208
DOI: 10.1126/science.1242366

Oscillation Stabilizes the Progenitor State

Transcription factors regulate fate choice between different neural lineages, but the same transcription factors are also expressed in neural progenitor cells. Imayoshi et al. (p. 1203, published online 31 October) analyzed the details of expression of several transcription factors in mouse neural cells. In neural progenitor cells, several different transcription factors were expressed in an oscillatory manner, whereas differentiated neurons stably expressed a single lineage-specific factor.


The basic helix-loop-helix transcription factors Ascl1/Mash1, Hes1, and Olig2 regulate fate choice of neurons, astrocytes, and oligodendrocytes, respectively. These same factors are coexpressed by neural progenitor cells. Here, we found by time-lapse imaging that these factors are expressed in an oscillatory manner by mouse neural progenitor cells. In each differentiation lineage, one of the factors becomes dominant. We used optogenetics to control expression of Ascl1 and found that, although sustained Ascl1 expression promotes neuronal fate determination, oscillatory Ascl1 expression maintains proliferating neural progenitor cells. Thus, the multipotent state correlates with oscillatory expression of several fate-determination factors, whereas the differentiated state correlates with sustained expression of a single factor.

Analyses of populations of mouse hematopoietic progenitors suggest that general fluctuations in the transcriptome affect lineage choice (1). On the other hand, analyses of individual cells do not show an effect of transcriptome fluctuation on lineage choice (2). Fate choice could be a gradual process based on transcriptome-wide fluctuations or discrete processes through different subpopulation states. To address these issues, we studied the expression patterns of fate determination factors in individual mouse cells in the multipotent state and during fate choice processes.

In the developing murine nervous system, neural progenitor cells (NPCs) proliferate (reproducing themselves) and also differentiate into three cell types, neurons, oligodendrocytes, and astrocytes (thus demonstrating multipotency). Transcription factors that regulate the proliferation of NPCs and the differentiation of each cell type include basic helix-loop-helix (bHLH) transcription factors (3, 4). For example, proneural bHLH genes such as Ascl1 (also known as Mash1) and Neurogenin2 (Ngn2) promote neuronal fate determination and suppress astrocytic gene expression (57). The bHLH gene Olig2 regulates oligodendrocyte specification, whereas the bHLH genes Hes1 and Hes5 maintain NPCs by repressing proneural gene expression (812). In addition, Ascl1 and Olig2 regulate oligodendrocyte and motor neuron development, respectively (810, 1318). One model suggests that Hes genes, expressed by NPCs, repress expression of other bHLH genes; down-regulation of Hes genes allows up-regulation of proneural genes or Olig2 and subsequent differentiation into neurons or oligodendrocytes (4). However, Ascl1 is expressed by dividing NPCs and postmitotic neurons, and Ascl1 up-regulates the expression of genes involved in cell cycle progression of NPCs as well as in cell cycle exit and neuronal differentiation (19, 20). Furthermore, Hes1 promotes the maintenance of NPCs and astrocyte differentiation (11, 12, 21, 22). Olig2 is also involved in NPC proliferation (23). Thus, these bHLH genes exert contradictory functions.

Hes1 expression oscillates with a period of about 2 to 3 hours in many cell types, including NPCs (24, 25). Ngn2 expression also oscillates in NPCs, probably because it is periodically repressed by Hes1 oscillation. In differentiating neurons, which lack Hes1 expression, Ngn2 expression is steady (25). Thus, the pattern of bHLH gene expression differs between NPCs and neurons, although our previous time-lapse imaging study only monitored messenger RNA (mRNA) production (25). Because transcription and translation can be dissociated (26), we studied protein expression at the single-cell level. We found that Hes1, Ascl1, and Olig2 protein expression oscillates in NPCs.

Variable Levels of bHLH Transcription Factor Expression in NPCs

Ascl1, Hes1, and Olig2 are expressed by NPCs in the ventral telencephalon, which generate neurons, astrocytes, and oligodendrocytes during perinatal stages of mouse development (1418, 27, 28). The expression levels of these transcription factors were variable from cell to cell (Fig. 1, A to H). Many cells were positive for all three bHLH factors (Hes1, Ascl1, and Olig2), whereas others expressed only two (Fig. 1, E to H). By contrast, differentiating neurons, oligodendrocytes, and astrocytes expressed only one of them at later stages (i.e., Ascl1, Olig2, and Hes1, respectively) (18, 22, 29). To examine the expression of these bHLH factors, we generated transgenic mice carrying reporters in which fluorescent (Venus or mCherry) or firefly luciferase (Luc2) complementary DNA (cDNA) was inserted in frame into the 5′ region of each bHLH gene in bacterial artificial chromosome (BAC) clones so that a bHLH factor fused with either Venus, mCherry, or Luc2 was expressed (table S1 and fig. S1, A to F). We also generated knock-in mice in which Venus was inserted in frame into the 5′ region of the Hes1 gene for Hes1 imaging (table S1 and fig. S2A), those in which Venus or luciferase (Eluc) was inserted in frame into the 5′ region of the Hes5 gene for Hes5 imaging (table S1 and fig. S2, B and C), and Sox2 reporter mice expressing a Luc2-Sox2 fusion protein (table S1 and fig. S1G). The reporter expression in these mice was similar to endogenous expression (fig. S3). Reporter expression also correlated well with endogenous protein expression in NPCs (fig. S4). The brain structures and the NPC competency of these reporter mice, including homozygous Venus-Hes1 fusion knock-in mice, were apparently normal (fig. S5).

Fig. 1 Variable expression levels of bHLH factors in NPCs of the ventral telencephalon.

(A to H) The expression of Hes1, Ascl1, and Olig2 in the ventral telencephalon at perinatal stages was examined by immunohistochemistry. The boxed region in (D) is enlarged in (E) to (H). Many cells were positive for all three bHLH factors (Hes1, Ascl1, and Olig2; arrowheads), whereas others were mostly positive for two of them. (I to K) Bioluminescence imaging and quantification of Luc2-Hes1 [(I) and (J)] and Luc2-Ascl1 (K) expression in slice cultures of the ventral telencephalon of reporter mice. Scale bars indicate 50 μm for (A) to (D) and (I). A.U., arbitrary units.

We used time-lapse imaging of brain slices from the ventral telencephalon of reporter mice and found that Hes1 and Ascl1 expression oscillate in NPCs (Fig. 1, I to K, and movie S1).

Oscillatory Expression of bHLH Factors in NPCs

We next prepared NPCs from the ventral telencephalon of perinatal reporter mice. We used acutely dissociated NPCs and those (designated NS cells) maintained in vitro for at least 10 generations in the presence of epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) (30, 31). Both cell types exhibited similar expression oscillations in the progenitor state and during cell fate choice. These NPCs expressed Ascl1, Olig2, and Hes1 at variable levels (fig. S6) and generated γ-aminobutyric acid–releasing (GABAergic) neurons, oligodendrocytes, and astrocytes (fig. S7) (14). Their competency to generate neurons and oligodendrocytes was significantly reduced in the absence of Ascl1 and Olig2, respectively (fig. S7, G, H″, I, and K). These NPCs also showed a reduced ability to generate astrocytes in the absence of Hes1 (fig. S7, F′ and J). Compared with the bHLH factors, the NPC-specific factor Sox2 (32) was expressed at a relatively constant level (fig. S6, A, E, and J).

We used time-lapse imaging to analyze expression patterns of bHLH factors. Hes1 protein expression oscillated in NPCs [average period of 149.9 ± 12.3 min (SD)] (Fig. 2, A, D, G, and H′; fig. S8; and movie S2). Hes5 expression in NPCs oscillated in phase with Hes1 expression (figs. S9 and S10). Ascl1 expression oscillated in NPCs (average period of 175.4 ± 29.5 min) (Fig. 2, B, E, G, and I; fig. S11; and movie S3). The Ascl1 protein level was up-regulated during S and G2 in some cells (~30%; fig. S11, C and D); after cell division, both daughter cells showed equal amounts of Ascl1, and expression oscillations resumed in cells that remained undifferentiated (fig. S11D). Olig2 protein expression also oscillated but more slowly (average period of 375.5 ± 105.8 min) than Hes1 and Ascl1 protein oscillations (Fig. 2, C, F, and G; fig. S12; and movie S4). The average expression levels of these factors are different between G1 and S, G2, and M (fig. S13), although their expression oscillated throughout the cell cycle. By contrast, two other factors characteristic of NPCs were expressed steadily: Sox2 and Nestin (figs. S13C and S14 and movie S5).

Hes1 and Ascl1 expression are inversely correlated (fig. S15, A and D) (33); Hes1 represses Ascl1 expression by binding to the Ascl1 promoter (34). Oscillating Hes1 would periodically repress Ascl1 expression, thereby driving oscillations in Ascl1 expression. Indeed, oscillatory Ascl1 expression was lost in the absence of Hes1 (fig. S11, E and F). Expression of Olig2, on the other hand, changes independently of Ascl1 and Hes1 (fig. S15, B, C, E, and F).

Fig. 2 Oscillatory expression of bHLH factors in self-renewing NPCs derived from the ventral telencephalon.

(A to F) Bioluminescence images and quantification of Luc2-Hes1 [(A) and (D)], Luc2-Ascl1 [(B) and (E)], and Luc2-Olig2 [(C) and (F)] reporter expression. (G to I) The distribution of the oscillation periods of Hes1, Ascl1, and Olig2 derived from bioluminescence measurements (n > 25 NPCs for each factor). Error bars in (G) indicate SD.

We next examined whether the oscillations in bHLH factor expression levels create a bias in differentiation competency. We separated NPCs of the fluorescent reporter mice into Hes1-high, Hes1-low, Ascl1-high, Ascl1-low, Olig2-high, and Olig2-low fractions (fig. S16, A to C). All cell fractions kept in NS media returned to original diverse levels of Hes1, Ascl1, and Olig2 expression within 2 days, suggesting that diverse levels represent different phases of expression oscillation (fig. S16, D to L). All cells generated neurospheres at similar efficiencies (fig. S16, M to R). After sorting, each cell population was cultured in a differentiation medium. Hes1-high and Hes1-low cells preferentially differentiated into astrocytes and neurons, respectively (fig. S17, A and B). Ascl1-high (and Olig2-low), and Olig2-high (and Ascl1-low) cells preferentially differentiated into neurons and oligodendrocytes, respectively (fig. S17, A and C to G). These results suggest that the different expression levels of the bHLH factors bias the fate choice of NPCs. However, such transient high expression is neither required nor sufficient for cell fate determination: NPCs with high expression of any of Ascl1, Olig2, and Hes1 were all able to differentiate into any of the three cell types (fig. S17).

Sustained Expression of bHLH Factors During Cell Fate Choice

We next examined how the expression of these bHLH factors changes during cell fate choice. During neuronal differentiation, Ascl1 protein accumulated after cell division (Fig. 3, A to D), in contrast to oscillatory expression in NPCs, and 6 to 8 hours later the early neuronal marker doublecortin (DCX) was expressed (Fig. 3, A and C to D′). Ascl1 expression continued to be up-regulated in many cells (76.7%, Fig. 3, A to C) but not in others (23.3%, Fig. 3D) after DCX expression was initiated, raising the possibility that the minimum requirement for neuronal differentiation is accumulation of Ascl1 over 6 to 8 hours during G1.

Fig. 3 Sustained Ascl1 expression in differentiating neurons.

Bioluminescence images of Luc2-Ascl1 (A to E) and Luc2-Hes1 (E′) expression were quantified. (A, C, C′, D, D′) Ascl1 expression accumulated in differentiating neurons [(A), green line; (C) and (D), orange double asterisks]. The early neuronal marker DCX was monitored by DCX-DsRed. Transient up-regulation of Ascl1 expression occurred before cell division [(C) and (D), magenta asterisks]. (B) Temporal trajectories of Luc2-Ascl1 in NS cells just after NPC or neurogenic division at time = 0 (mean in solid line, and standard errors in colored, n > 19 for each division). Division-mode effect, P = 0.0022; interaction between division mode and time, P < 0.0001, repeated measures analysis of variance (ANOVA). (E) This acutely dissociated NPC underwent asymmetric cell division. Ascl1 was equally distributed in both daughter cells after the first cell division. Ascl1 expression accumulated in a daughter neuron (green) but resumed oscillating in a daughter NPC (red), which underwent the second division. (E′) Hes1 expression was repressed before asymmetric cell division (asterisks). The suppression of Hes1 expression was maintained in a daughter neuron (blue line), whereas Hes1 oscillation resumed in a daughter NPC (red line).

In acute dissociation culture from the ventral telencephalon, many NPCs underwent asymmetric cell division, in which one daughter cell remained undifferentiated while the other differentiated into a neuron. In these NPCs, Ascl1 expression was up-regulated (at least twofold compared with the average) before cell division and seemed to be equally distributed in both daughter cells (Fig. 3E). In the daughter neuron, Ascl1 expression accumulated after cell division, whereas it resumed oscillating in the daughter NPC (Fig. 3E). Before neurogenic cell division (one or both daughter cells underwent neuronal differentiation), Ascl1 expression was transiently up-regulated (at least twofold compared with the average) in many cases: ~90% of their mother cells exhibited such transient up-regulation of Ascl1 before cell division (fig. S18B). However, ~30% of mother cells did so when they produced two daughter NPCs (figs. S11, C and D, and S18B). Thus, the transient up-regulation of Ascl1 before cell division is not decisive but merely lends a bias toward neuronal fate choice.

In those NPCs whose daughter cells underwent neuronal differentiation, Hes1 expression was repressed before cell division (Fig. 3E′ and fig. S18, A, E, and F) but not when both cells remained NPCs (fig. S18, A, C, and D). The suppression of Hes1 expression was maintained in daughter neurons (Fig. 3E′ and fig. S18, E and F), whereas Hes1 oscillation resumed in daughter NPCs (Fig. 3E′). Thus, it is likely that transient down-regulation of Hes1 expression and the concomitant up-regulation of Ascl1 before cell division directs NPCs toward neuronal fate choice and that sustained expression of Ascl1 after cell division irreversibly determines neuronal fate. Down-regulation of Hes1 could be caused by fluctuations of the expression levels of Notch intracellular domain (NICD), an active form of Notch signaling (35) (fig. S19). Indeed, when stable NICD expression was induced from the Rosa26 locus in NPCs, both Hes1 and Hes5 expression oscillated sustainably, and there was no down-regulation of these factors (figs. S20 and S21). By contrast, in the presence of a γ-secretase inhibitor, which inhibits Notch signaling activity, Ascl1 and Olig2 expression was up-regulated to stable expression (fig. S22).

During astrocyte differentiation, Hes1 protein expression still oscillated but at high average and trough levels (Fig. 4, A and B, and fig. S23, B and C). Twelve to 24 hours later, expression of the astrocyte marker glial fibrillary acidic protein (GFAP) began (fig. S23A). Ascl1 and Olig2 expression became undetectable within 10 hours during astrocyte differentiation (fig. S23, D to G). During oligodendrocyte differentiation, Olig2 protein expression oscillated but at high trough levels (Fig. 4, C and D, and fig. S24, A, C, and E). A few days after induction of oligodendrocyte differentiation, Olig2 expression was down-regulated, and expression of the mature oligodendrocyte marker 2',3′-cyclic-nucleotide-3′-phosphodiesterase (CNPase) (fig. S25) was up-regulated (Fig. 4E and fig. S24, B, D, and G). During this period, Ascl1 and Hes1 expression were down-regulated (fig. S24, H to J). Thus, bHLH fate determination factors are coexpressed in an oscillatory manner in NPCs, but, as the cell fate choice becomes established, one factor accumulates and the other two are lost. Although oscillatory expression of multiple fate determination factors underlies the multipotent state of NPCs, this oscillatory pattern gives way to stable and dominant expression of one factor during cellular differentiation.

Fig. 4 Expression dynamics of bHLH factors during gliogenesis.

(A and B) Bioluminescence imaging (A) and quantification (B) of Luc2-Hes1 expression during astrocyte differentiation. Astrocyte specification was induced at time = 0 by leukemia inhibitory factor and bone morphogenetic protein 4. (C to E) Bioluminescence imaging (C) and quantification (D) of Luc2-Olig2 expression and quantification of pCNP-Venus (E) in a single cell during oligodendrocyte differentiation, which was induced by T3 at time = 0.

Light-Induced Control of Expression Pattern

To demonstrate the functional importance of oscillatory or sustained expression patterns, we adopted the optogenetic gene expression system by using the Neurospora crassa photoreceptor Vivid that was fused with Gal4 DNA binding domain and p65 activation domain (GAVPO) (36). The codon usage was optimized for mammalian cells to increase expression efficiency, and the target mRNA was destabilized by introducing the 3′ untranslated region of mouse Ascl1 mRNA (fig. S26A). With this system, we can induce gene expression comparable to endogenous levels (fig. S26, D to G). Repeated exposure of blue light with 3-hour intervals generated oscillatory expression with a 3-hour period, whereas repeated exposure with 30-min intervals generated sustained expression at both cell-population (Fig. 5A) and single-cell levels (Fig. 5, B and C, and movie S6).

Fig. 5 Light-induced oscillatory or sustained expression of Ascl1 in NPCs.

(A to C) Time-lapse imaging (C) and quantification [(A) and (B)] of gene expression at cell-population (A) and single-cell levels [(B) and (C)] induced by blue light. (D to K) According to the schedule of light exposure (fig. S26B; light intensity of 1.11 μmol/m2 per s), oscillatory [(D) to (G)] and sustained [(H) to (K)] Ascl1 expression was induced in Ascl1-null NPCs, which were cultured in the presence of bFGF and EGF, a condition that inhibits neurogenesis. Oscillatory Ascl1 expression induced nearly no βIII-tubulin+ neuron formation (D) but significantly increased the proportion of dividing cells (PH3+) compared with the control [ubiquitin (Ub)–luc2] [(E) to (G)]. Sustained Ascl1 expression significantly increased βIII-tubulin+ neuron formation compared with the control (Ub-luc2) (H, J, K). DAPI, 4′,6-diamidino-2-phenylindole. *P < 0.05, **P < 0.01; two-tailed Student’s t test. Error bars indicate SE. (L to P) According to the schedule of light exposure (fig. S26C), sustained Ascl1 expression was induced in NPCs for indicated time lengths, and neuronal formation was examined. *P < 0.05, **P < 0.01, ***P < 0.001, one-way ANOVA followed by Tukey post hoc test. Scale bars, 12.5 μm for (F), (G), (J), and (K) and 12.5 μm for (M) to (P).

Because Ascl1 is known to promote the cell cycle progression of NPCs and their cell cycle exit and neuronal differentiation (20), we asked whether these contradictory functions of Ascl1 are regulated by different expression patterns. We introduced the Ascl1-inducible system into Ascl1-null NS cells, which produce no neurons (fig. S7) and proliferate more slowly than wild-type NS cells (20). Ascl1-null NS cells have a low proportion [4 to 5%; phosphohistone H3+ (PH3+)] of dividing cells (Fig. 5E), whereas wild-type NS cells have 14 to 15% PH3+ dividing cells (fig. S29C). However, light-induced oscillatory expression of Ascl1 increased the proportion of dividing cells in the Ascl1-null NS cell population to ~10% PH3+ (Fig. 5, E to G), suggesting that Ascl1 oscillation enhanced NPC proliferation. These NS cells did not differentiate into neurons (βIII-tubulin+) even after 3 days (Fig. 5D). The period of this oscillation is important, because a 6-hour period did not affect NPC proliferation (fig. S27). By contrast, light-induced sustained expression of Ascl1 enhanced neuronal differentiation (βIII-tubulin+) of Ascl1-null NS cells (Fig. 5, H to K). A higher level of sustained Ascl1 expression increased the efficiency of neuronal differentiation (compare Fig. 5H versus fig. S28E). Oscillatory Ascl1 expression did not induce neuronal differentiation even at a higher amplitude (fig. S28, A to C) but increased the number of proliferating NPCs (fig. S28D). By contrast, sustained Ascl1 expression at similar levels increased neuronal differentiation (fig. S28E). These results indicate that distinct expression patterns, but not the levels, of Ascl1 are important for a choice between proliferation and differentiation.

We used the Ascl1-inducible system in wild-type NS cells, where endogenous Ascl1 expression oscillates (fig. S29). Oscillatory light stimulation with 3-hour intervals did not affect the neurogenesis or proliferation of these cells (figs. S29, A to I, and S30, A to D). By contrast, sustained Ascl1 expression for 72 hours increased neuronal formation even in the presence of bFGF and EGF, a condition that inhibits neurogenesis (figs. S29, J to Q, and S30E). At least 6 to 8 hours of sustained Ascl1 expression was required for generation of neurons (Fig. 5, L to P), agreeing with the above notion about the minimal requirement for neuronal differentiation. This requirement for elapsed time suggests that only NS cells that are caught at G1 could be redirected by light-induced expression of Ascl1 into neuronal development. The Ascl1-inducible system was also introduced into the dorsal telencephalon, which normally expresses very low levels of Ascl1. The oscillatory expression of Ascl1 did not induce neuronal differentiation but did maintain Nestin+ NPCs in the ventricular zone, whereas the sustained expression of Ascl1 increased the number of βIII-tubulin+ neurons that migrated out of the ventricular zone (fig. S31). Thus, manipulation of Ascl1 gene expression can impose a choice favoring proliferation or differentiation according to whether the Ascl1 expression is oscillatory or sustained.


Our data suggest that multipotency is a state of multiple oscillating neurogenic and gliogenic determination factors and that cell fate choice is a process of sustained expression of a single factor. This switching may be induced by the fluctuations of Notch signaling (supplementary text). The detailed mechanism by which the oscillatory and sustained Ascl1 expression differentially regulates downstream gene expression remains to be determined. A recent report indicates that the proneural factor Ngn2 is differentially phosphorylated between NPCs and neurons and controls the expression of its target genes differently depending on its phosphorylation status (37). We speculate that oscillatory and sustained expression of proneural factors could be involved in the different posttranscriptional modulation that is responsible for target gene selectivity. We also demonstrated that the light-switchable gene expression system offers an efficient way to control the proliferation and differentiation of stem cells by changing the light-exposure pattern rather than using different growth factors or chemicals, showing its applicability to the regeneration technology.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S31

Table S1

References (3855)

Movies S1 to S6

References and Notes

  1. Acknowledgments: We are grateful to F. Guillemot, J. Johnson, H. Takebayashi, K. Ikenaka, D. Melton, A. Miyawaki, A. Sakaue-Sawano, Q. Lu, and Y. Yang for reagents and S. Kitano, M. Sakamoto, H. Shimojo, and Center for Meso-Bio Single-Molecule Imaging (CeMI), WPI-iCeMS, Kyoto University for technical help. This work was supported by Core Research for Evolutional Science and Technology (R.K. and H.K.), Grant-in-Aid for Scientific Research on Innovative Areas (Ministry of Education, Culture, Sports, Science, and Technology 22123002) (R.K.), Scientific Research (A) [Japan Society for Promotion of Science (JSPS) 24240049] (R.K.) and Young Scientists (A) (JSPS 24680035) (I.I.), and Takeda Foundation (R.K.). I.I. and R.K. designed the project and wrote the manuscript. A.I. developed the light-induced gene expression system. I.I., A.I., Y.H., and H.M. performed experiments, and T.F. and F.I. conducted fluorescent imaging analyses. K.K. and H.K. performed computer simulation. A national (Japanese) patent application, “Optogenetic control of proliferation and differentiation of stem cells” (2013-193582), has been filed by Kyoto University.
View Abstract

Stay Connected to Science

Navigate This Article