Untitled

Article
Coordinated changes in cellular behavior ensure the

lifelong maintenance of the hippocampal stem cell
population
Graphical Abstract Authors
Lachlan Harris, Piero Rigo,
Thomas Stiehl, ..., Noelia Urbán,
Anna Marciniak-Czochra,
François Guillemot
Correspondence
[email protected]
In Brief
Harris et al. show that multiple cellular
changes work in concert during early life
to preserve the hippocampal stem cell
population throughout adulthood in mice.
In particular, more proliferating stem cells
return to quiescence instead of
differentiating. The changes are
coordinated by increasing degradation of
the pro-activation factor ASCL1.
Highlights
d More proliferating hippocampal stem cells return to shallow
quiescence with age
d Dormant stem cells enter deeper quiescence with age
d These changes drive the transition from developmental to

adult neurogenesis
d Increasing degradation of ASCL1 protein by HUWE1

coordinates these changes
Harris et al., 2021, Cell Stem Cell 28, 863–876

May 6, 2021 ª 2021 The Author(s). Published by Elsevier Inc.
https://doi.org/10.1016/j.stem.2021.01.003 ll
ll
OPEN ACCESS
Article
Coordinated changes in cellular behavior
ensure the lifelong maintenance of
the hippocampal stem cell population
Lachlan Harris,1 Piero Rigo,1 Thomas Stiehl,2,3,4 Zachary B. Gaber,1,7 Sophie H.L. Austin,1 Maria del Mar Masdeu,1,6
Amelia Edwards,5 Noelia Urbán,1,8 Anna Marciniak-Czochra,2,3,4 and François Guillemot1,9,*
1Neural Stem Cell Biology Laboratory, The Francis Crick Institute, London NW1 1AT, UK
2Instituteof Applied Mathematics, Heidelberg University, 69120 Heidelberg, Germany
3Interdisciplinary Center for Scientific Computing (IWR), Heidelberg University, 69120 Heidelberg, Germany
4Bioquant Center, Heidelberg University, 69120 Heidelberg, Germany
5Advanced Sequencing Facility, The Francis Crick Institute, London NW1 1AT, UK
6Present address: Orchard Therapeutics Limited, 108 Cannon St., London EC4N 6EU, UK
7Present address: Nuclera Nucleics, Cambridge Science Park, Cambridge CB4 0GD, UK
8Present address: Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA), Vienna Biocenter Campus (VBC), Dr.
Bohr Gasse 3, 1030 Vienna, Austria

9Lead contact
*Correspondence: [email protected]
https://doi.org/10.1016/j.stem.2021.01.003
SUMMARY
Neural stem cell numbers fall rapidly in the hippocampus of juvenile mice but stabilize during adulthood,
ensuring lifelong hippocampal neurogenesis. We show that this stabilization of stem cell numbers in young
adults is the result of coordinated changes in stem cell behavior. Although proliferating neural stem cells in
juveniles differentiate rapidly, they increasingly return to a resting state of shallow quiescence and progress
through additional self-renewing divisions in adulthood. Single-cell transcriptomics, modeling, and label
retention analyses indicate that resting cells have a higher activation rate and greater contribution to neuro-
genesis than dormant cells, which have not left quiescence. These changes in stem cell behavior result from a
progressive reduction in expression of the pro-activation protein ASCL1 because of increased post-transla-
tional degradation. These cellular mechanisms help reconcile current contradictory models of hippocampal
neural stem cell (NSC) dynamics and may contribute to the different rates of decline of hippocampal neuro-
genesis in mammalian species, including humans.
INTRODUCTION typic pattern has been observed in most mammals; the numbers
of new neurons generated in the DG are high in juveniles but
Neural stem cells (NSCs) persist in restricted brain areas during decline rapidly in young adults and are maintained at low levels
adulthood, mostly in the dentate gyrus (DG) of the hippocampus throughout adulthood (Amrein, 2015; Snyder, 2019). This transi-
and the ventricular-subventricular zone (V-SVZ) of the lateral tion from high levels of hippocampal neurogenesis in juveniles to
ventricles. In mice, proliferative DG precursor cells enter quies- lower and sustained levels in adults differs from an aging process
cence during the second week after birth. By the end of the sec- because it occurs early in life, suggesting that it is adaptive. Inter-
ond postnatal week, quiescent DG precursors have acquired the estingly, whether the same transition occurs in humans to extend
elongated and branched morphology and arrangement in a sin- neurogenesis beyond childhood has been disputed recently,
gle cell layer that typify adult NSCs (Berg et al., 2019; Hochgerner highlighting the lack of understanding of the processes that pre-
et al., 2018; Li et al., 2013; Nicola et al., 2015; Noguchi et al., serve neurogenesis between infancy and old age in model spe-
2019). There is therefore a clear histological transition between cies such as mice (Moreno-Jiménez et al., 2019; Sorrells
developmental neurogenesis, which extends from mid-embryo- et al., 2018).
genesis to the second postnatal week, and adult hippocampal Hippocampal NSCs sit atop the lineage hierarchy of neuro-
neurogenesis, which starts around post-natal day 14 (P14) and genesis, and their numbers also decline over time. Changes in
continues throughout life. NSC behavior are therefore likely to be involved in the transition
However, hippocampal neurogenesis continues to change to lower and sustainable levels of neurogenesis during adult-
considerably after P14 in mice as the DG transitions from devel- hood. However, the behavior of adult hippocampal NSCs is still
opmental to adult rates of neuronal production. Indeed, a stereo- not well understood, and current models diverge on key points
Cell Stem Cell 28, 863–876, May 6, 2021 ª 2021 The Author(s). Published by Elsevier Inc. 863
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
ll
OPEN ACCESS Article
Figure 1. Proliferating NSCs increasingly return to quiescence with time

(A) Images demonstrating that loss of hippocampal NSCs is rapid in young mice (0.5–2 months) but more gradual in adults (>2 months of age).
(B) Quantification of NSC numbers from 0.5–18 months of age.
(C) The rate of NSC depletion is lower than predicted by the disposable stem cell model after 2 months of age.
(D) Quantification of the fraction of proliferating NSCs from 0.5–18 months of age.
(E) Schematic of the different hippocampal NSC states. Dormant NSCs have never proliferated, whereas resting NSCs have returned to quiescence from a
proliferating state.
(F) Proliferating NSCs were labeled via EdU injections followed by a 48-h chase.
(G) In young mice, EdU+ NSCs rarely returned to quiescence; most were Ki67+.
(H) Return to quiescence increased in frequency with age (Ki67–).
(I) Quantification of return to quiescence at different ages.
(J) The ability of EdU-incorporating NSCs in 1- and 6-month-old mice to persist in the hippocampal niche was determined by measuring the fraction of EdU+
NSCs that remained as NSCs 10 and 30 days after a 5-day EdU labeling protocol.
(K) Representative image of an EdU+ NSC that persisted for 30 days after labeling in a 6-month-old mouse.
(legend continued on next page)
864 Cell Stem Cell 28, 863–876, May 6, 2021

ll
Article OPEN ACCESS
(Lazutkin et al., 2019). In particular, Encinas et al. (2011) tween 2 and 6 months of age) and became decoupled from pro-
described a ‘‘disposable stem cell model’’ where NSCs that liferation so that it became substantially lower than predicted by
have left quiescence progress rapidly through a series of neuro- the disposable stem cell model (Figure 1C). We therefore
genic divisions without returning to quiescence and eventually reasoned that proliferating NSCs may acquire distinct properties
differentiate into astrocytes, leading to a decrease in the NSC in early adulthood that function to slow depletion and preserve
pool. Bonaguidi et al. (2011) proposed a contrasting ‘‘long- the NSC population throughout adult neurogenesis.
term self-renewal’’ model where many NSCs return to quies-
cence after dividing so that the NSC pool declines only slightly Proliferating NSCs progressively acquire the capacity to
with age. Finally, we have proposed a model supporting hetero- return to quiescence
geneity in stem cell behavior where degradation of the pro-acti- We investigated whether proliferating hippocampal NSCs might
vation factor ASCL1 in proliferating stem cells allowed a subset avoid depletion in young adults by returning to quiescence. We
of these cells to return to quiescence (Urbán et al., 2016). refer to proliferating NSCs that have returned to quiescence as
Here we show that these different models are valid but resting NSCs to distinguish them from dormant NSCs that
describe NSC behavior at different stages throughout adult life. have remained quiescent since establishment of the DG niche
At the onset of adult neurogenesis, all proliferating NSCs are (Figure 1E; Urbán et al., 2016). To label proliferating hippocampal
lost rapidly through differentiation, whereas by 6 months of NSCs, mice of different ages (0.5–18 months) were all given the
age, more than half of the dividing NSCs return to quiescence. thymidine analog 5-Ethynyl-20 -deoxyuridine (EdU) via injections
We determine that cells that have returned to quiescence and were culled after a 48-h chase (Figure 1F; Chehrehasa
(resting NSCs) have molecular and cellular properties distinct et al., 2009). EdU+ NSCs were identified as having remained
from quiescent cells that have never proliferated (dormant cycling or having returned to a quiescent state depending on
NSCs) and that they play an increasingly important part in main- whether they co-expressed Ki67 protein. In 2-week-old mice,
taining NSC proliferation over time. Finally, we find that these the hippocampal NSC niche is histologically mature and com-
coordinated changes are caused by an increase in post-transla- prises a large population of mostly quiescent NSCs (~90%; Fig-
tional degradation of the ASCL1 protein. Our results demon- ure 1D; Berg et al., 2019; Noguchi et al., 2019). However, all
strate that progressive and coordinated changes in NSC proper- EdU+ NSCs in 2-week-old mice remained cycling at the end of
ties enable transition from high levels of neurogenesis coupled the chase period (EdU+Ki67+, 36 of 36), indicating that, at onset
with stem cell depletion in juveniles to lower but sustainable of adult neurogenesis, proliferating NSCs do not have the capac-
levels of neurogenesis associated with stem cell self-renewal ity to return to quiescence (Figures 1G and 1I). In 1-month-old
throughout adult life. mice, in contrast, we found that a small fraction (6.68% ±
1.17%) of EdU+ NSCs returned to quiescence (EdU+Ki67–).
RESULTS The fraction of proliferating NSCs that exited the cell cycle
then increased in a gradual and substantial manner so that the
Rapid NSC depletion in juvenile mice and reduced majority of EdU+ NSCs returned to a quiescent state in mice
depletion in adults by 6 months of age (Figures 1H and 1I).
We set out to examine the cellular properties of hippocampal We confirmed these results with independent approaches.
NSCs from juvenile stages to old age. At the onset of adult neuro- First we labeled proliferating NSCs by genetic means, utilizing
genesis in 0.5-month-old mice (Berg et al., 2019; Noguchi et al., mice containing a Ki67-creERT2 allele (Basak et al., 2018) and
2019), there were approximately 46,000 NSCs per DG that un- a tdTomato cre-reporter (Madisen et al., 2010). In these Ki67TD
derwent an immediate and rapid decline so that their number mice, transcription of creERT2 occurs in cells progressing
had more than halved in 2-month-old mice (Figures 1A and through the cell cycle. We administered tamoxifen to 1- and 6-
1B). After this rapid loss of NSCs in juveniles, the depletion month-old Ki67TD mice to label proliferating NSCs and culled
rate, defined as the percentage of the hippocampal NSC pool the mice at the end of the injection period. Consistent with the
that was lost each day, slowed in young adults and reached EdU labeling experiment, proliferating NSCs (tdTomato+) re-
much lower levels in 6-, 12-, and 18-month-old mice (Figure 1C). turned to a quiescent state (Ki67–) at a greater rate in 6-month-
The disposable stem cell model proposed for hippocampal old mice than in 1-month-old mice (Figures S1A–S1D).
NSCs (Encinas et al., 2011) states that NSC activation leads to Further, we confirmed the increased rate of return to quies-
a series of rapid asymmetric divisions and eventual loss of the cence with age by analyzing EdU-labeled NSCs with MCM2 la-
NSC through astrocytic or neuronal differentiation (Pilz et al., beling. Mcm2 is highly transcribed throughout the cell cycle,
2018). Because this model links NSC depletion to their activity, including G1 phase (Braun and Breeden, 2007; Tsuruga et al.,
a lower rate of depletion might reflect a reduction in NSC activity. 1997; Young and Tye, 1997), unlike Ki67 (Miller et al., 2018).
Indeed, the depletion rate decreased in proportion to the size of We determined that the MCM2 protein is extremely stable and
the proliferating (Ki67+) NSC pool in juveniles (between 1– that it is detected not only throughout the entire cell cycle but
2 months; Figures 1C and 1D). However, the depletion rate also for at least 72 h after a cell has exited the cell cycle (by re-
slowed considerably more than NSC activity in young adults (be- turning to quiescence or differentiating), which is approximately
(L) Quantification of NSC persistence.

Graphs show mean ± SEM. In (B)–(D), 3 mice were analyzed per time point, except at 12 months, where 8 mice were analyzed. Dots represent individual mice in (I)
and (L). Statistics: t test in (C) and (L), and one-way ANOVA in (I). *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar (located in K’): 19 mm in (A), 10 mm in (G), (G’), (H), and
(H’), and 11.25 mm in (K) and (K’).
Cell Stem Cell 28, 863–876, May 6, 2021 865

ll
OPEN ACCESS Article
Figure 2. Resting NSCs increasingly
contribute to the proliferative NSC pool with
time
(A) Mice received EdU via drinking water for 2 weeks
to label proliferating and resting NSCs, followed by
a 20-h chase.
(B) Dormant NSCs that proliferated during the 20-h
chase period were identified as EdU–Ki67+ NSCs.
(C) The activation rate of dormant NSCs, normalized
to the total size of the NSC pool, decreased with
age, indicating deepening quiescence of this pop-
ulation.
(D) The absolute numbers of EdU–Ki67+ cells also
decreased with age.
(E) The contribution of dormant NSCs to the prolif-
erative NSC pool decreased with age, whereas the
contribution of resting NSCs increased despite the
resting NSC pool remaining relatively small even at
advanced ages.
Graphs show mean ± SEM. Dots represent indi-
vidual mice in (C) and (D); in (E), 3 mice were
analyzed per time point. Statistics: one-way ANOVA
in (C) and (D). Scale bar in (B), 10 mm.
and 1L). These findings demonstrate

that proliferating hippocampal NSCs
progressively acquire the capacity to re-
turn to quiescence, resulting in long-term
twice as long as Ki67 (Figures S1E–S1H). Thus, although an self-renewal that supports maintenance of the NSC pool
EdU+ NSC that is Ki67– can have returned to G0 phase or be pro- with age.
gressing through a G1 phase that is longer than 40 h, an EdU+
NSC that is MCM2– must have left the cell cycle and returned Resting NSCs increasingly contribute to the
to quiescence for at least 72 h. We labeled proliferating cells in proliferative NSC pool
1- and 6-month-old mice with EdU injections and quantified Next we wanted to find out whether dormant NSCs, which have
the number of EdU+ NSCs that were MCM2– 1 week later to not proliferated previously, may also alter their properties with
allow sufficient time for MCM2 to degrade (Figure S1I). We found time. To address this, we gave mice of different ages (1–
that, in 6-month-old mice, 20.9% of EdU+ NSCs retained their 18 months) EdU in the drinking water for 2 weeks and culled
stem cell identity and were MCM2– and, thus, had returned to them after a 20-h chase. Resting NSCs do not remain quiescent
quiescence for at least 72 h (Figure S1J). This represented an over the long term, particularly in young mice (Figure 1L), and
~200-fold increase over 1-month-old mice (Figure S1J). More- therefore most resting and proliferating NSCs are labeled by pro-
over, the absolute number of EdU+MCM2 NSCs was 23-fold longed EdU exposure in these paradigms, meaning that the
higher in 6-month-old mice than in 1-month-old mice (Figures EdU– NSC population essentially corresponds to the dormant
S1K and S1L), despite the much lower levels of proliferation NSC pool. We found that the fraction of dormant NSCs that
that occur with age. These independent lines of evidence became active and proliferated during the 20-h chase period
demonstrate that proliferating hippocampal NSCs increasingly (i.e., EdU–Ki67+ NSCs; Figure 2B) declined progressively and
return to quiescence with time. substantially with age, indicating that dormant NSCs were pro-
We reasoned that because proliferating NSCs increasingly gressing into a deeper state of quiescence over time (Figures
return to quiescence rather than differentiating, they might 2C and 2D).
persist as NSCs for longer in older mice, helping to preserve Even in 6- to 18-month-old mice, the numbers of resting
the NSC pool. To examine the long-term persistence of prolif- NSCs remained small compared with the size of the dormant
erating NSCs in the hippocampal niche, we labeled prolifer- pool (Figure 2E). We therefore quantified the functional signifi-
ating NSCs with EdU for 5 days in drinking water and exam- cance of resting NSCs by determining their contribution to
ined their fate after 10 and 30 days (Figure 1J). In 1-month-old the proliferative NSC pool at different ages. As expected, in ju-
mice, only 2.46% ± 1.25% of EdU+ NSCs retained their NSC venile mice where the resting pool is very small and the rate of
identity 10 days after labeling, and none retained it after activation of dormant cells is relatively high, the vast majority of
30 days (Figure 1L). Strikingly, in 6-month-old mice, there proliferating NSCs arose from the dormant NSC pool. However,
was an approximately 15-fold increase in the fraction of prolif- in 6- to 18-month-old mice, when the resting pool has grown
erating NSCs that persisted as NSCs after the 10-day chase and dormant cells have become more deeply quiescent, resting
(29.37% ± 8.5%), and a third of those (9.3% ± 1.89%) re- NSCs were the origin of 55%–63.8% of proliferating NSCs
tained their NSC identity after the 30-day chase (Figures 1K despite comprising only 1.5%–3% of the NSC population

ll
Article OPEN ACCESS
(legend on next page)

ll
OPEN ACCESS Article
(Figure 2E). These findings imply that resting hippocampal provided a good fit for the numbers of total NSCs and active
NSCs are in a shallower state of quiescence and have a higher NSCs, as already found in our previous model of hippocampal
rate of activation than dormant NSCs. They also support a NSC dynamics (Ziebell et al., 2018). However, the model was a
novel model where, with time, hippocampal NSC proliferation poor fit for resting NSCs (the quiescent label-retaining cell pop-
is generated increasingly from resting NSCs that have the ca- ulation) because it did not take into account that dormant and
pacity to shuttle between active and quiescent states, delaying resting cells could have different activation rates (Figure S2B).
NSC depletion and ensuring long-term maintenance of the We therefore retained the time-dependent activation rate from
NSC pool. the initial model but allowed the absolute value of the activation
rate to differ between resting NSCs and dormant NSCs. This
Increasing numbers of NSC self-renewing divisions change resulted in dramatic improvement of the fit of the model
over time (Akaike information criterion [AIC] = 53 compared with 120 for the
We next wondered whether returning to the resting state and model where both populations have the same activation rate;
escaping differentiation might provide NSCs with additional op- Figures S2B and S2C). In the top-ranked model (AIC = 36; Fig-
portunities for self-renewing divisions. To address this possibil- ure 2D), resting NSCs across the first 6 months of life had a me-
ity, we developed a mouse model to count the number of self- dian activation rate 29.2-fold higher than that of dormant NSCs.
renewing divisions each hippocampal NSC undergoes prior to We tested a variety of other scenarios (e.g., time-dependent cell
depleting from the niche. We used a labeling system based cycle duration), and in isolation these models were poor (Table
on the histone 2B (H2B)-green fluorescent protein (GFP) fusion S1; Data S1). Likewise, when we added time-dependent self-
protein, which accumulates in nuclei and is then repressed renewal and time-dependent cell cycle duration to the model
through doxycycline (Dox) administration, resulting in dilution of best fit, they did not make improvements (AIC > 36), suggest-
of the GFP label following cell division (Kanda et al., 1998; Tum- ing that these features contribute less to time-dependent
bar et al., 2004). We compared juvenile (1.5 month) and adult changes than activation rates. Furthermore, refitting of the model
(6 month) cohorts in which the H2B-GFP label was targeted to the data under the assumption that activation rates, prolifera-
to NSCs using the Glast gene promoter (Figures 3A–3E; Mori tion rate, and self-renewal are age dependent predicts that age-
et al., 2006). related changes of proliferation rate and self-renewal are small
We found that, at both ages, most NSCs underwent between 1 compared with the changes in activation rates (Data S1). Accu-
and 3 self-renewing divisions before losing their stem cell iden- mulation of resting NSCs with a high activation rate and the
tity. However, the patterns of H2B-GFP dilution were different decreasing activation rate of dormant NSCs are the main fea-
between ages, indicating different division patterns (Figures tures that explain the overall increased contribution of resting
3F–3I). For example, in a label dilution experiment where mice NSCs to proliferation (Table S1) and the decreased depletion
received Dox for identical lengths of time, 64% of NSCs self-re- of NSCs over time.
newed only once in juvenile mice (versus 50% in adults), whereas
12% of NSCs underwent three or more self-renewing divisions Single-cell RNA sequencing and pseudotemporal
(versus 28% in adults) (Figures 3G and 3H). Thus, the increased ordering demonstrate state- and age-dependent
self-renewal with time is an additional mechanism contributing to changes to hippocampal NSC quiescence
preservation of the NSC pool in adults. The age-dependent changes in hippocampal NSC properties
and in particular the different activation rates of resting and
Mathematical modeling of time-dependent changes dormant NSCs warranted transcriptomic characterization to
To extend these observations, we built a mathematical model of identify the molecular mechanisms underlying these properties.
hippocampal NSC population dynamics at different ages. For To achieve this, we crossed the Nestin-GFP mouse line that la-
model fitting, we used the numbers of total NSCs (Figure 1B), bels NSCs (Mignone et al., 2004) with Ki67TD mice to label prolif-
proliferating NSCs (Ki67+, referred to as active cells throughout erating cells and their progeny (Basak et al., 2018). The gener-
modeling section; Figure 1D), and resting NSCs (EdU+Ki67–; ated mice (hereafter called Ki67TD-NES) were injected with
Figure 2E) from mice 0.5–12 months of age. In our model (Fig- tamoxifen at 1-month, 2-months, and 6–8 months of age to irre-
ure S2A; Data S1), we found that a decreasing activation rate versibly label the progeny of proliferating cells with red
Figure 3. NSCs perform more self-renewing divisions with time

(A) A juvenile cohort (P14) of H2B-GFP mice was injected with tamoxifen to induce GFP expression in NSCs. The mice received Dox for 10 days to stop label
incorporation and EdU to mark dividing cells.
(B) An adult cohort (5 months old) received the same treatment.
(C) A second cohort of adult mice (4.5 months old) received Dox for 30 days to control for the increased time a proliferating NSC persists before depleting in
adult mice.
(D) The H2B-GFP label becomes diluted in EdU+ NSCs.
(E) In quiescent EdU– NSCs, the label remained undiluted.
(F) The GFP signal from EdU+ NSCs was categorized into discrete bins through automated analysis. These bins corresponded to the numbers of self-renewing
divisions.
(G) Label dilution profile of EdU+ NSCs from juvenile mice (10-day chase).
(H) The label dilution profile of EdU+ NSCs from adult mice (10-day chase) showed increased self-renewal compared with juveniles.
(I) Label dilution profile of EdU+ NSCs from adult mice (30-day chase) also showed increased self-renewal compared with juveniles.
Statistics: Fisher’s exact test in (G)–(I). Scale bar (located in D): 15 mm in (D) and (E) and 10 mm in the subset panels in (D) and (E). div, divisions.

ll
Article OPEN ACCESS
Figure 4. The quiescence depth of hippocampal NSCs is dependent on the proliferation history of the stem cell and the age of the mouse
(A) Cohorts of 1-, 2-, and 6- to 8-month-old Ki67TD-NES mice were given tamoxifen and culled after 8 days, and then the DG was dissociated.
(B) GFP+tdTomato+ and GFP+tdTomato cells were collected by flow cytometry and sequenced using a 10x Genomics platform.
(C) Uniform Manifold Approximation and Projection (UMAP) plot showing the 24,203 cells sequenced from 8 experiments.
(D) After two iterations of subsetting and re-clustering, a dataset of 2,947 NSCs was ordered using Slingshot, revealing a pseudotime trajectory from the most
quiescent NSCs (blue) to proliferating NSCs (red).
(E) Pseudotime progression is correlated negatively with Apoe expression and positively with Ccnd2 expression.
(F) Strong concordance between genes associated with this pseudotime trajectory and those from Shin et al. (2015).
(G) UMAP plot showing the locations of dormant (quiescent and GFP+tdTomato–), resting (quiescent and GFP+tdTomato+), and proliferating NSCs (cell cycle
gene expression and GFP+tdTomato+/–).
(H) Plotting pseudotime positions reveals that resting NSCs are in a shallower state of quiescence than dormant NSCs.
(I) UMAP plot showing the location of dormant cells according to mouse age.
(J) Plotting pseudotime positions of dormant NSCs grouped by age reveals a progressive increase in quiescence depth.
Dots in UMAP and violin plots represent individual cells. Statistics: Mann-Whitney U test in (H) and (J). ***p < 0.001.
fluorescence, their DG was disassociated, and GFP+tdTomato– tory tool Slingshot to infer progression along a pseudotime curve
and GFP+tdTomato+ cells were sorted by flow cytometry and from quiescence to activation (Figures 4D and 4E; Street et al.,
sequenced with the 10x Genomics platform (Figures 4A and 2018), which we validated against an existing pseudotime anal-
4B; Figure S3). The dataset of 24,203 cells was subsetted into ysis of 123 hippocampal progenitors (Shin et al., 2015;
NSCs and intermediate progenitor cells (IPCs) (Figure 4C; Fig- Figure 4F).
ures S4A and S4B) and then re-clustered to exclude IPCs and We then analyzed the positions along the pseudotime axis of
isolate only quiescent and proliferating NSCs (Figures S4C and NSCs in different states (proliferating, resting, and dormant)
S4D). From this final dataset of 2,947 NSCs, we used the trajec- and of different ages. We identified proliferating NSCs in

ll
OPEN ACCESS Article
Figure 5. Resting NSCs display a distinct
transcriptional profile indicative of a
shallow quiescent state
(A) Resting NSCs exhibit lower expression of
quiescence marker genes than dormant NSCs
(scRNA-seq).
(B) Resting NSCs have increased expression of
genes involved in biosynthesis compared with
dormant NSCs (scRNA-seq).
(C) To validate the differential gene expression
analysis, 12-month-old mice received EdU for
2 weeks and were culled after a 20-h chase.
(D) Resting NSCs had reduced staining intensity
for ID4 than dormant NSCs.
(E) Image of resting NSC (EdU+Ki67–, arrow-
heads) and dormant NSCs (EdU–Ki67–, arrows)
with ID4 co-staining.
Graphs are violin plots. Dots in violin plots repre-
sent individual cells. Statistics: t test performed on
Pearson’s residuals with false discovery rate
(FDR)-corrected p value in (A) and (B) and t test in
(D). *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar in
(E), 15 mm. a.u., arbitrary units.
The effect of age on the depth of

dormant NSC quiescence was also strik-
ing (Figures 4I and 4J). The distribution of
NSC pseudotime positions from 1-
month-old mice was clearly overrepre-
sented in shallower states of quiescence,
whereas NSCs at 2 months of age were
distributed evenly across the quiescence
spectrum, and NSCs from 6- to 8-month-
old mice were over-represented in
deeper states of quiescence. Thus,
dormant hippocampal NSCs demon-
strate an early and continuing progres-
sion into deeper quiescence. The single-
cell RNA sequencing (scRNA-seq) data
support the EdU labeling data and math-
ematical modeling reporting the exis-
G2/S/M phase based on cell cycle scoring algorithms (Butler tence of a small population of resting NSCs that exist in a
et al., 2018). However, to distinguish proliferating NSCs in G1 shallow quiescent state and the deepening quiescence of
phase from quiescent NSCs in G0 phase among recombined dormant NSCs.
(tdTomato+) cells, we used the expression of Mcm and other
G1/S-phase genes (Figure S5; STAR methods). Supporting Division history affects the molecular properties of
our thresholding approach, resting NSCs expressed an inde- quiescent hippocampal NSCs
pendent set of G1/S-phase genes at similarly low levels as The previous findings suggest that resting NSCs might diverge
dormant cells (Figure S5D) and had total mRNA levels that molecularly from dormant NSCs. To directly address this, we
were half of those of proliferating NSCs and equivalent to those used the scRNA-seq data to compare gene expression in the
of dormant NSCs, indicative of a quiescent state (Figure S5E). different NSC populations. Gene Ontology analysis of differential
As predicted from the 2-week EdU labeling experiments (Fig- gene expression between resting and dormant NSCs (880
ure 2E), resting NSCs were rare (Figure 4G). They segregated genes; Table S3), showed that resting NSCs had reduced
along the pseudotime axis at positions intermediate between expression of genes associated with lipid metabolism, suggest-
proliferating NSCs and dormant NSCs (tdTomato– and quies- ing shallower quiescence because NSCs require de novo lipo-
cent), with only a minority of resting cells occupying deeply genesis and fatty acid oxidation to maintain quiescence (Kno-
quiescent positions (Figures 4G and 4H). Our single-cell tran- bloch et al., 2013, 2014; Figure 5A). Resting NSCs also
scriptomics analysis therefore shows that proliferating NSCs showed increased expression of genes associated with ribo-
that return to a state of quiescence are among the shallowest somal biogenesis, mRNA processing, and translation elonga-
quiescent NSCs. tion, demonstrating upregulation of protein biosynthesis

ll
Article OPEN ACCESS
pathways, which has been associated with exit from quiescence labeled proliferating, resting, and dormant NSCs in 1-month-
(Figure 5B; Cabezas-Wallscheid et al., 2017). Furthermore, mul- old Ascl1neo/neo mice by EdU retention and Ki67 labeling as
tiple genes/pathways that have been associated with mainte- before (Figure 6E). Compared with age-matched controls,
nance of quiescence, such as Clu (Basak et al., 2018), Hopx EdU+ NSCs in hypomorphic mice were 2.5 times more likely to
(Berg et al., 2019), Notch2 (Engler et al., 2018), and Id4 (p < return to quiescence (Figure 6F), and dormant NSCs were 24
0.01) (Blomfield et al., 2019), were downregulated in resting times less likely to exit quiescence (Figure 6G). As a
NSCs compared with dormant NSCs (Figure 5A; Table S3). We result, resting NSCs contributed more to the proliferative pool
validated the dataset by confirming that expression of ID4 was in Ascl1neo/neo mice than in controls (Figure 6H) and the total
downregulated at a protein level in resting versus dormant number of NSCs was increased, presumably because of a lower
NSCs (Figures 5C–5E). We extended this analysis to genes ex- depletion rate (Figure 6I). Therefore, hippocampal NSCs in
pressed differentially between resting and proliferating NSCs. Ascl1neo/neo mice display the behavior of NSCs from older
Excluding cell cycle genes that were upregulated uniquely in wild-type mice, which provides direct evidence that the declining
proliferating NSCs, many of the aforementioned quiescence- levels of ASCL1 drive the time-dependent changes in NSC
specific genes were expressed in resting NSCs at an intermedi- behavior and the resulting reduction in NSC depletion rate.
ate level between dormant and proliferating states (Figures 5A
and 5B; Table S4). Thus, although resting NSCs are quiescent, Declining ASCL1 levels are due to increased post-
they show increased expression of metabolic and biosynthetic translational degradation
pathways and are primed toward re-activation. In our scRNA-seq analysis of hippocampal NSCs in 1-month-old
and 6-month-old mice, the levels of Ascl1 transcripts did not
Declining ASCL1 levels explain time-dependent change significantly (Figure 7A). We therefore reasoned that
changes in quiescence the steep time-dependent decline in ASCL1 protein levels must
The fact that the properties of dormant, resting and proliferating be primarily due to altered post-translational regulation, such
NSCs change coordinately over time suggested that a common as through increased expression or activity of ID proteins or
molecular mechanism might underlie these changes. The tran- HUWE1, which promote targeting of ASCL1 for proteasomal
scription factor ASCL1 was a strong candidate to contribute to destruction (Blomfield et al., 2019; Urbán et al., 2016). Indeed,
the changing properties of NSCs; induction of ASCL1 expression expression of Huwe1 increased in 6-month-old NSCs relative
is required for activation of quiescent NSCs (Andersen et al., to its expression in 1-month-old NSCs (Figure 7B), whereas,
2014), and a reduction of ASCL1 protein levels allows prolifer- among Id genes, Id4 was upregulated but Id1–Id3 were downre-
ating hippocampal NSCs to return to a resting state (Urbán gulated with age (Table S5). We therefore focused on the role of
et al., 2016) and increases the likelihood that embryonic NSCs HUWE1 in driving the time-dependent change in ASCL1 protein
proliferate rather than differentiate (Imayoshi et al., 2013). We expression. HUWE1-dependent degradation of ASCL1 has been
reasoned that the changing properties of hippocampal NSCs shown to allow hippocampal NSCs to stop dividing and return to
might reflect progressively lower expression levels of ASCL1 a resting state (Urbán et al., 2016), and it also suppresses activa-
over time. tion of dormant NSCs (Figures S7A and S7B). Thus, increased
Ascl1 is transcribed by most NSCs (Blomfield et al., 2019). expression of Huwe1 and/or an increase in efficiency of
However, because of low protein levels as a result of degradation HUWE1-mediated ASCL1 degradation with age might contribute
induced by the E3 ubiquitin ligase HUWE1 (Urbán et al., 2016) to the progressive reduction in ASCL1 levels.
and by Inhibitor of differentiation/DNA binding (ID) factors (Blom- To test whether HUWE1 activity increases with age, we
field et al., 2019), ASCL1 protein cannot be detected by immuno- deleted Huwe1 from hippocampal NSCs by administering
labeling in quiescent NSCs and can only be seen when ASCL1 tamoxifen to Huwe1fl/y; Glast-creERT2 mice at 1- and 6-months
protein levels are at their highest; i.e., in proliferating NSCs. To of age and analyzed the mice 1 week later (Figure 7C). The ex-
increase ASCL1 detection, we examined its expression in a pected increase in the number of NSCs expressing ASCL1
mouse line that expresses an ASCL1-VENUS fusion protein (Im- was substantially larger in 6-month-old mice (12.1-fold increase)
ayoshi et al., 2013). Labeling of the fusion protein with anti-GFP than in 1-month-old mice (3.9-fold increase) when normalized to
antibodies greatly enhanced detection of ASCL1 protein so that age-matched controls (Figure 7D; Figure S7D). Similarly, the
low levels were readily detectable even in quiescent NSCs (Fig- number of Ki67+ NSCs increased by a larger amount in 6-
ure S6). We found that there was a progressive but dramatic month-old mice (3.8-fold) than in 1-month-old mice (1.4-fold)
reduction in the proportion of NSCs (including quiescent NSCs) (Figure 7E; Figure S7E), indicating that HUWE1 activity, which
that express detectable levels of ASCL1-VENUS throughout eliminates ASCL1 expression and suppresses NSC proliferation,
adulthood. 54% of NSCs expressed ASCL1-VENUS in 0.5- is greater in adult than in juvenile mice. Finally, we asked whether
month-old mice, 19% in 2-month-old mice, and 3% in 12- Huwe1 has a role in long-term maintenance of the NSC pool and
month-old mice (Figures 6A–6C). There was a parallel reduction whether this, too, is age dependent (Figure S7F). We found that
in the intensity of ASCL1-VENUS fluorescence in NSCs from 6- the decline in NSC number was small 2 months after Huwe1
month-old mice relative to 1-month-old mice (Figure 6D). Thus, deletion from 1-month-old mice but significantly larger 2 months
the observed changes in hippocampal NSC properties with after deletion at 5 months of age (Figures S7G and S7H). These
age correlate with a reduction in ASCL1 protein. results demonstrate that the expression and activity of Huwe1 in-
To establish causality, we utilized an Ascl1 hypomorphic crease with time to reduce ASCL1 protein levels and mediate the
mouse line (Ascl1neo/neo) that expresses reduced levels of associated changes in quiescence that serve to preserve the
Ascl1 in adult hippocampal NSCs (Andersen et al., 2014). We NSC pool during adult neurogenesis.

ll
OPEN ACCESS Article
Figure 6. Ascl1 protein levels decrease with time and cause progressive changes in NSC behavior
(A) ASCL1-VENUS staining in hippocampal NSCs from 1-month-old mice (arrowheads indicate positive cells) using an anti-GFP antibody.
(B) ASCL1-VENUS staining in NSCs from 6-month-old mice.
(C) Fewer NSCs are positive for ASCL1-VENUS with age.
(D) The expression intensity of the ASCL1-VENUS protein also decreases with age.
(E) Ascl1neo/neo and control mice were given EdU to label proliferating and resting NSCs and were culled following a 20-h chase.
(F) Hippocampal NSCs were returning more to quiescence after proliferating (EdU+Ki67) in Ascl1neo/neo mice than in controls.
(G) Dormant NSCs activated less frequently in Ascl1neo/neo mice.
(H) Resting NSCs contributed more to the proliferative NSC pool in Ascl1neo/neo mice than in controls.
(I) Ascl1neo/neo mice had more NSCs than controls.
Graphs represent the mean ± SEM. Dots represent individual mice, except in (D), where dots represent individual cells (minimum of 20 cells analyzed per mouse, 2
mice per age). Statistics: one-way ANOVA in (C) and t test in (D) and (F)–(I). *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar (located in A): 21.9 mm in (A) and (B). a.u.,
arbitrary units.
DISCUSSION et al., 2010). Previous experiments have demonstrated that

NSCs in the adult niches of the V-SVZ and DG lie on a trajectory
The changes in NSC behaviors we identify serve to preserve the from quiescence to activation, including a population of primed
NSC pool and long-term adult neurogenesis and are therefore NSCs in the V-SVZ (Llorens-Bobadilla et al., 2015; Shin et al.,
distinct from stem cell exhaustion, which occurs during aging. 2015). However, whether all NSCs operate along the same tra-
Most of the changes we describe occur in the first 6 months of jectory, whether the trajectory can be reversed, and whether
life when mice are of breeding age, suggesting an adaptive func- subpopulations of NSCs along this trajectory can be linked to
tion. Aging is defined by functional decline and accumulation of specific cellular behaviors (i.e., returning to quiescence) has re-
cellular damage, resulting in loss of fitness (López-Otı́n et al., mained unclear. We find that hippocampal NSCs that have
2013). In contrast, we found that the reduction of NSC numbers recently divided and returned to a resting state are in a much
during this period is not associated with loss of NSC function. shallower quiescent state than NSCs that have never left quies-
On the contrary, the number of self-renewing divisions increases. cence. Although resting NSCs present lower expression levels of
Moreover, our transcriptional analysis of NSCs in 1-month-old quiescence-associated transcription factors and higher levels of
and 6-month-old mice shows little change over time in inflamma- metabolic genes than dormant NSCs, they nevertheless are able
tory, mitochondrial, and lysosomal pathways that typify stem cell to remain quiescent because of active degradation of the pro-
aging (Leeman et al., 2018). Instead, we propose that hippocam- activation factor ASCL1, which prevents NSC activation and
pal NSCs progress, during juvenile and early adult stages, cell cycle re-entry (Urbán et al., 2016). However, the reduced
through a transition between developmental neurogenesis, char- expression of quiescence-associated genes maintains resting
acterized by large NSC numbers and high rates of neuronal pro- NSCs in shallow quiescence, which allows them to reactivate
duction, and proper adult neurogenesis, characterized by smaller much more readily than dormant cells and sustain NSC prolifer-
numbers of NSCs and lower rates of neuronal production. ation and neurogenesis over the long term.
Adult tissue stem cell compartments often have several stem Our data suggest that current conflicting models of hippocam-
cell populations that differ in their level of quiescence (Barker pal NSC behavior can be reconciled by considering changes in

ll
Article OPEN ACCESS
Figure 7. Expression and activity of Huwe1 increases with time

(A) Ascl1 transcript levels in NSCs are largely unchanged between young and adult mice (scRNA-seq).
(B) The expression of the ubiquitin-ligase Huwe1 increases in older mice (scRNA-seq).
(C) 1- and 6-month-old Huwe1fl/y and control mice were culled 1 week after receiving tamoxifen.
(D) The fold change increase in ASCL1+ NSCs in Huwe1fl/y mice (relative to controls) was larger in the older cohort.
(E) The fold change increase in Ki67+ NSCs in Huwe1fl/y mice (relative to controls) was also larger in the older cohort.
(F) Images of control and Huwe1fl/y mice at 1 month of age; arrows indicate ASCL1+ NSCs.
(G) Images of control and Huwe1fl/y mice at 6 months of age; arrows indicate ASCL1+ NSCs.
Graphs represent mean ± SEM or are violin plots in (A) and (B). Dots represent single cells in (A) and (B) and individual mice in (D) and (E). Statistics: t test
performed on Pearson’s residuals with FDR-corrected p value in (A) and (B) and t test in (D) and (E). *p < 0.05, **p < 0.01. Scale bar (located in G): 28 mm in (F)
and (G).
behavior over time. We find that NSCs uniformly adhere to the are expressed differentially in NSCs between adults and juve-
disposable stem cell model in juvenile mice (Encinas et al., niles (Table S5). Identifying niche signals that control the phos-
2011). Thereafter, heterogeneity emerges (Urbán et al., 2016), phorylation status of ASCL1 and its interaction with HUWE1
so that some hippocampal NSCs acquire the capacity to return might provide insights into timing mechanisms that drive
to quiescence and long-term self-renewal increases, as reported changes in hippocampal NSC behavior during transition from
by Bonaguidi et al. (2011). Importantly, the age of the mice used developmental to adult neurogenesis. Our results demonstrate
in the original studies might have contributed to their divergent how a series of early, progressive, and coordinated changes in
conclusions. Two-month-old mice were used in the bromodeox- hippocampal NSC properties function to preserve proliferation
yuridine (BrdU) labeling experiments that originated the dispos- beyond the juvenile period in mice. Whether similar mechanisms
able stem cell model (Encinas et al., 2011), and 10-week-old are present in humans to extend neurogenesis beyond child-
mice were used when live imaging led to similar conclusions hood warrants investigation.
(Pilz et al., 2018). Three- to four-month-old mice were used in la-
beling experiments where a subpopulation of hippocampal Limitations of study
NSCs were found to return to quiescence (Urbán et al., 2016), There are several limitations in the present study. First, our anal-
whereas a substantial part of the clonal experiments that posited ysis of the emergence of resting NSCs and the deepening quies-
long-term self-renewal was analyzed in 1-year-old mice (Bona- cence of dormant NSCs with age was restricted to scRNA-seq
guidi et al., 2011). These observations suggest that use of mice and nucleotide retention analysis. Further investigations should
of different ages contributed to development of contradictory determine whether these age-dependent states involve lasting
models of hippocampal NSC behavior. The effect of time we reorganization of chromatin, especially considering that ASCL1
describe here renders these models compatible. is a pioneer transcription factor (Raposo et al., 2015). Second,
We attribute the changes in hippocampal NSC dynamics the resolution of the H2B-GFP dilution method could not distin-
throughout adulthood to the declining levels of ASCL1 protein. guish between 3 and more self-renewing divisions. Therefore, in
We found that the reduction in ASCL1 levels with age was due future studies, the magnitude of the age-dependent increase in
to increased post-translational degradation by HUWE1. The self-renewal would be best quantified by intravital live-imaging
mechanisms leading to transcriptional upregulation of Huwe1 approaches.
and/or its increased activity are unclear. It has been shown
that the phosphorylation status of substrates can affect STAR+METHODS
HUWE1-mediated ubiquitination (Forget et al., 2014). Interest-
ingly, ASCL1 stability is phospho dependent (Ali et al., 2014), Detailed methods are provided in the online version of this paper
and we identified a number of phosphatases and kinases that and include the following:

ll
OPEN ACCESS Article
d KEY RESOURCES TABLE REFERENCES
d RESOURCE AVAILABILITY
B Lead contact Ali, F.R., Cheng, K., Kirwan, P., Metcalfe, S., Livesey, F.J., Barker, R.A., and
Philpott, A. (2014). The phosphorylation status of Ascl1 is a key determinant
B Materials availability
of neuronal differentiation and maturation in vivo and in vitro. Development
B Data and code availability
141, 2216–2224.
d EXPERIMENTAL MODEL AND SUBJECT DETAILS
Ameijeiras-Alonso, J., Crujeiras, R.M., and Rodrı́guez-Casal, A. (2018).
d METHOD DETAILS Multimode: An R Package for Mode Assessment. arXiv, arXiv:1803.00472
B Mathematical modeling https://arxiv.org/abs/1803.00472.
B Cell culture Amrein, I. (2015). Adult hippocampal neurogenesis in natural populations of
B Preparation and sectioning of mouse brain tissue mammals. Cold Spring Harb. Perspect. Biol. 7, a021295.
B Antibodies, immunofluorescence, cell counts Andersen, J., Urbán, N., Achimastou, A., Ito, A., Simic, M., Ullom, K.,
B Tamoxifen treatment and administration of EdU Martynoga, B., Lebel, M., Göritz, C., Frisén, J., et al. (2014). A transcriptional
B Calculation of observed/expected depletion rate mechanism integrating inputs from extracellular signals to activate hippocam-
B Calculation of contribution to NSC proliferation pal stem cells. Neuron 83, 1085–1097.
B H2B-GFP label dilution experiments Barker, N., Bartfeld, S., and Clevers, H. (2010). Tissue-resident adult stem cell
TD-NES populations of rapidly self-renewing organs. Cell Stem Cell 7, 656–670.
B Single-cell analysis of Ki67 mice
d QUANTIFICATION AND STATISTICAL ANALYSIS Basak, O., Krieger, T.G., Muraro, M.J., Wiebrands, K., Stange, D.E., Frias-
Aldeguer, J., Rivron, N.C., van de Wetering, M., van Es, J.H., van
B Single-cell data
Oudenaarden, A., et al. (2018). Troy+ brain stem cells cycle through quies-
B Gene Ontology
cence and regulate their number by sensing niche occupancy. Proc. Natl.
B Statistical analysis of cell counts Acad. Sci. USA 115, E610–E619.
Becht, E., McInnes, L., Healy, J., Dutertre, C.A., Kwok, I.W.H., Ng, L.G.,
SUPPLEMENTAL INFORMATION Ginhoux, F., and Newell, E.W. (2018). Dimensionality reduction for visualizing
single-cell data using UMAP. Nat. Biotechnol. Published online December 3,
Supplemental Information can be found online at https://doi.org/10.1016/j. 2018. https://doi.org/10.1038/nbt.4314.
stem.2021.01.003. Berg, D.A., Su, Y., Jimenez-Cyrus, D., Patel, A., Huang, N., Morizet, D., Lee, S.,
Shah, R., Ringeling, F.R., Jain, R., et al. (2019). A Common Embryonic Origin of
Stem Cells Drives Developmental and Adult Neurogenesis. Cell 177, 654–
ACKNOWLEDGMENTS 668.e15.
Blomfield, I.M., Rocamonde, B., Masdeu, M.D.M., Mulugeta, E., Vaga, S., van
We gratefully acknowledge Lan Chen; the Biological Resources, Advanced
den Berg, D.L., Huillard, E., Guillemot, F., and Urbán, N. (2019). Id4 promotes
Sequencing, Flow Cytometry, and Light Microscopy Facilities; and the
the elimination of the pro-activation factor Ascl1 to maintain quiescence of
Research Illustration team of the Francis Crick Institute for technical support
adult hippocampal stem cells. eLife 8, e48561.
and Ana Martin-Villalba for feedback on the manuscript. This work was sup-
ported by the Francis Crick Institute, which receives its funding from Cancer Bolte, S., and Cordelières, F.P. (2006). A guided tour into subcellular colocal-
Research UK (FC0010089), the UK Medical Research Council (FC0010089), ization analysis in light microscopy. J. Microsc. 224, 213–232.
and the Wellcome Trust (FC0010089). This work was also supported by the Bonaguidi, M.A., Wheeler, M.A., Shapiro, J.S., Stadel, R.P., Sun, G.J., Ming,
UK Medical Research Council (project grant U117570528 to F.G.), the Well- G.L., and Song, H. (2011). In vivo clonal analysis reveals self-renewing and
come Trust (Investigator Award 106187/Z/14/Z to F.G.), and the German multipotent adult neural stem cell characteristics. Cell 145, 1142–1155.
Research Foundation (DFG; SFB 873, subproject B08 to A.M.C.). Brandt, M.D., Hu €bner, M., and Storch, A. (2012). Brief report: Adult hippocam-
pal precursor cells shorten S-phase and total cell cycle length during neuronal
AUTHOR CONTRIBUTIONS differentiation. Stem Cells 30, 2843–2847.
Braun, K.A., and Breeden, L.L. (2007). Nascent transcription of MCM2-7 is
Conceptualization, L.H., T.S., N.U., A.M.-C., and F.G.; Data Curation, L.H.; important for nuclear localization of the minichromosome maintenance com-
Formal Analysis, L.H., P.R., Z.B.G., and T.S.; Validation, L.H., P.R., and T.S.; plex in G1. Mol. Biol. Cell 18, 1447–1456.
Investigation, L.H., P.R., S.H.L.A., M.d.M.M., and T.S.; Visualization, L.H.,
Briggs, J.A., Weinreb, C., Wagner, D.E., Megason, S., Peshkin, L., Kirschner,
P.R., and T.S.; Methodology, T.S.; Writing – Original Draft, L.H. and F.G.;
M.W., and Klein, A.M. (2018). The dynamics of gene expression in vertebrate
Writing – Review & Editing, all authors; Supervision, F.G.; Funding Acquisition,
embryogenesis at single-cell resolution. Science 360, eaar5780.
F.G.; Project Administration, F.G.
Burnham, K.P., and Anderson, D.R. (2002). Model selection and multimodel
inference: A practical information-theoretic approach (Springer).
DECLARATION OF INTERESTS Butler, A., Hoffman, P., Smibert, P., Papalexi, E., and Satija, R. (2018).
The authors declare no competing interests. Integrating single-cell transcriptomic data across different conditions, technol-
ogies, and species. Nat. Biotechnol. 36, 411–420.
Received: March 2, 2020 Cabezas-Wallscheid, N., Buettner, F., Sommerkamp, P., Klimmeck, D., Ladel,
Revised: October 9, 2020 L., Thalheimer, F.B., Pastor-Flores, D., Roma, L.P., Renders, S., Zeisberger,
Accepted: January 7, 2021 P., et al. (2017). Vitamin A-Retinoic Acid Signaling Regulates Hematopoietic
Published: February 12, 2021 Stem Cell Dormancy. Cell 169, 807–823.e19.
Chehrehasa, F., Meedeniya, A.C., Dwyer, P., Abrahamsen, G., and Mackay-
SUPPORTING CITATIONS Sim, A. (2009). EdU, a new thymidine analogue for labelling proliferating cells
in the nervous system. J. Neurosci. Methods 177, 122–130.
The following references appear in the Supplemental Information: Brandt et al., Encinas, J.M., Michurina, T.V., Peunova, N., Park, J.H., Tordo, J., Peterson,
2012, Burnham and Anderson, 2002, Marciniak-Czochra et al., 2009, Stiehl D.A., Fishell, G., Koulakov, A., and Enikolopov, G. (2011). Division-coupled as-
and Marciniak-Czochra, 2011, Stiehl and Marciniak-Czochra, 2017, Ziebell trocytic differentiation and age-related depletion of neural stem cells in the
et al., 2014. adult hippocampus. Cell Stem Cell 8, 566–579.

ll
Article OPEN ACCESS
Engler, A., Rolando, C., Giachino, C., Saotome, I., Erni, A., Brien, C., Zhang, R., Moreno-Jiménez, E.P., Flor-Garcı́a, M., Terreros-Roncal, J., Rábano, A.,
Zimber-Strobl, U., Radtke, F., Artavanis-Tsakonas, S., et al. (2018). Notch2 Cafini, F., Pallas-Bazarra, N., Ávila, J., and Llorens-Martı́n, M. (2019). Adult hip-
Signaling Maintains NSC Quiescence in the Murine Ventricular- pocampal neurogenesis is abundant in neurologically healthy subjects and
Subventricular Zone. Cell Rep. 22, 992–1002. drops sharply in patients with Alzheimer’s disease. Nat. Med. 25, 554–560.
Forget, A., Bihannic, L., Cigna, S.M., Lefevre, C., Remke, M., Barnat, M., Mori, T., Tanaka, K., Buffo, A., Wurst, W., Ku €hn, R., and Götz, M. (2006).
Dodier, S., Shirvani, H., Mercier, A., Mensah, A., et al. (2014). Shh signaling Inducible gene deletion in astroglia and radial glia–a valuable tool for functional
protects Atoh1 from degradation mediated by the E3 ubiquitin ligase Huwe1 and lineage analysis. Glia 54, 21–34.
in neural precursors. Dev. Cell 29, 649–661. Nakamura, K.C., Kameda, H., Koshimizu, Y., Yanagawa, Y., and Kaneko, T.
Hafemeister, C., and Satija, R. (2019). Normalization and variance stabilization (2008). Production and histological application of affinity-purified antibodies
of single-cell RNA-seq data using regularized negative binomial regression. to heat-denatured green fluorescent protein. J. Histochem. Cytochem. 56,
Genome Biol. 20, 296. 647–657.
Hagihara, H., Toyama, K., Yamasaki, N., and Miyakawa, T. (2009). Dissection Nicola, Z., Fabel, K., and Kempermann, G. (2015). Development of the adult
of hippocampal dentate gyrus from adult mouse. J. Vis. Exp. (33), 1543. neurogenic niche in the hippocampus of mice. Front. Neuroanat. 9, 53.
Hochgerner, H., Zeisel, A., Lönnerberg, P., and Linnarsson, S. (2018). Noguchi, H., Castillo, J.G., Nakashima, K., and Pleasure, S.J. (2019).
Conserved properties of dentate gyrus neurogenesis across postnatal devel- Suppressor of fused controls perinatal expansion and quiescence of future
opment revealed by single-cell RNA sequencing. Nat. Neurosci. 21, 290–299. dentate adult neural stem cells. eLife 8, e42918.
Imayoshi, I., Isomura, A., Harima, Y., Kawaguchi, K., Kori, H., Miyachi, H., Pilz, G.A., Bottes, S., Betizeau, M., Jörg, D.J., Carta, S., Simons, B.D.,
Fujiwara, T., Ishidate, F., and Kageyama, R. (2013). Oscillatory control of fac- Helmchen, F., and Jessberger, S. (2018). Live imaging of neurogenesis in
tors determining multipotency and fate in mouse neural progenitors. Science the adult mouse hippocampus. Science 359, 658–662.
342, 1203–1208. Raposo, A.A.S.F., Vasconcelos, F.F., Drechsel, D., Marie, C., Johnston, C.,
Kanda, T., Sullivan, K.F., and Wahl, G.M. (1998). Histone-GFP fusion protein Dolle, D., Bithell, A., Gillotin, S., van den Berg, D.L.C., Ettwiller, L., et al.
enables sensitive analysis of chromosome dynamics in living mammalian cells. (2015). Ascl1 Coordinately Regulates Gene Expression and the Chromatin
Curr. Biol. 8, 377–385. Landscape during Neurogenesis. Cell Rep. 10, 1544–1556.
Knobloch, M., Braun, S.M., Zurkirchen, L., von Schoultz, C., Zamboni, N., Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch,
Araúzo-Bravo, M.J., Kovacs, W.J., Karalay, O., Suter, U., Machado, R.A., T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an
et al. (2013). Metabolic control of adult neural stem cell activity by Fasn-depen- open-source platform for biological-image analysis. Nat. Methods 9, 676–682.
dent lipogenesis. Nature 493, 226–230. Shin, J., Berg, D.A., Zhu, Y., Shin, J.Y., Song, J., Bonaguidi, M.A., Enikolopov,
Knobloch, M., von Schoultz, C., Zurkirchen, L., Braun, S.M., Vidmar, M., and G., Nauen, D.W., Christian, K.M., Ming, G.L., and Song, H. (2015). Single-Cell
Jessberger, S. (2014). SPOT14-positive neural stem/progenitor cells in the hip- RNA-Seq with Waterfall Reveals Molecular Cascades underlying Adult
pocampus respond dynamically to neurogenic regulators. Stem Cell Reports Neurogenesis. Cell Stem Cell 17, 360–372.
3, 735–742. Snyder, J.S. (2019). Recalibrating the Relevance of Adult Neurogenesis.
Lazutkin, A., Podgorny, O., and Enikolopov, G. (2019). Modes of division and Trends Neurosci. 42, 164–178.
differentiation of neural stem cells. Behav. Brain Res. 374, 112118. Sorrells, S.F., Paredes, M.F., Cebrian-Silla, A., Sandoval, K., Qi, D., Kelley,
Leeman, D.S., Hebestreit, K., Ruetz, T., Webb, A.E., McKay, A., Pollina, E.A., K.W., James, D., Mayer, S., Chang, J., Auguste, K.I., et al. (2018). Human hip-
Dulken, B.W., Zhao, X., Yeo, R.W., Ho, T.T., et al. (2018). Lysosome activation pocampal neurogenesis drops sharply in children to undetectable levels in
clears aggregates and enhances quiescent neural stem cell activation during adults. Nature 555, 377–381.
aging. Science 359, 1277–1283. Srinivas, S., Watanabe, T., Lin, C.S., William, C.M., Tanabe, Y., Jessell, T.M.,
Li, G., Fang, L., Fernández, G., and Pleasure, S.J. (2013). The ventral hippo- and Costantini, F. (2001). Cre reporter strains produced by targeted insertion
campus is the embryonic origin for adult neural stem cells in the dentate gyrus. of EYFP and ECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4.
Neuron 78, 658–672. Stiehl, T., and Marciniak-Czochra, A. (2011). Characterization of stem cells us-
Llorens-Bobadilla, E., Zhao, S., Baser, A., Saiz-Castro, G., Zwadlo, K., and ing mathematical models of multistage cell lineages. Math. Comput. Model.
Martin-Villalba, A. (2015). Single-Cell Transcriptomics Reveals a Population 53, 1505–1517.
of Dormant Neural Stem Cells that Become Activated upon Brain Injury. Cell Stiehl, T., and Marciniak-Czochra, A. (2017). Stem cell self-renewal in regener-
Stem Cell 17, 329–340. ation and cancer: Insights from mathematical modeling. Curr. Opin. Syst. Biol.
López-Otı́n, C., Blasco, M.A., Partridge, L., Serrano, M., and Kroemer, G. 5, 112–120.
(2013). The hallmarks of aging. Cell 153, 1194–1217. Street, K., Risso, D., Fletcher, R.B., Das, D., Ngai, J., Yosef, N., Purdom, E.,
Madisen, L., Zwingman, T.A., Sunkin, S.M., Oh, S.W., Zariwala, H.A., Gu, H., and Dudoit, S. (2018). Slingshot: cell lineage and pseudotime inference for sin-
Ng, L.L., Palmiter, R.D., Hawrylycz, M.J., Jones, A.R., et al. (2010). A robust gle-cell transcriptomics. BMC Genomics 19, 477.
and high-throughput Cre reporting and characterization system for the whole Stuart, T., Butler, A., Hoffman, P., Hafemeister, C., Papalexi, E., Mauck, W.M.,
mouse brain. Nat. Neurosci. 13, 133–140. 3rd, Hao, Y., Stoeckius, M., Smibert, P., and Satija, R. (2019). Comprehensive
€ger, W., and Wagner, W. (2009).
Marciniak-Czochra, A., Stiehl, T., Ho, A.D., Ja Integration of Single-Cell Data. Cell 177, 1888–1902.e21.
Modeling of asymmetric cell division in hematopoietic stem cells–regulation of Tsuruga, H., Yabuta, N., Hashizume, K., Ikeda, M., Endo, Y., and Nojima, H.
self-renewal is essential for efficient repopulation. Stem Cells Dev. 18, (1997). Expression, nuclear localization and interactions of human MCM/P1
377–385. proteins. Biochem. Biophys. Res. Commun. 236, 118–125.
Martı́n-Suárez, S., Valero, J., Muro-Garcı́a, T., and Encinas, J.M. (2019). Tumbar, T., Guasch, G., Greco, V., Blanpain, C., Lowry, W.E., Rendl, M., and
Phenotypical and functional heterogeneity of neural stem cells in the aged hip- Fuchs, E. (2004). Defining the epithelial stem cell niche in skin. Science 303,
pocampus. Aging Cell 18, e12958. 359–363.
Mignone, J.L., Kukekov, V., Chiang, A.S., Steindler, D., and Enikolopov, G. Urbán, N., van den Berg, D.L., Forget, A., Andersen, J., Demmers, J.A., Hunt,
(2004). Neural stem and progenitor cells in nestin-GFP transgenic mice. C., Ayrault, O., and Guillemot, F. (2016). Return to quiescence of mouse neural
J. Comp. Neurol. 469, 311–324. stem cells by degradation of a proactivation protein. Science 353, 292–295.
Miller, I., Min, M., Yang, C., Tian, C., Gookin, S., Carter, D., and Spencer, S.L. Walker, T.L., and Kempermann, G. (2014). One mouse, two cultures: isolation
(2018). Ki67 is a Graded Rather than a Binary Marker of Proliferation versus and culture of adult neural stem cells from the two neurogenic zones of individ-
Quiescence. Cell Rep. 24, 1105–1112.e5. ual mice. J. Vis. Exp. e51225.

ll
OPEN ACCESS Article
Wang, L., Sharma, K., Deng, H.X., Siddique, T., Grisotti, G., Liu, E., and Roos, Zhao, X., Heng, J.I., Guardavaccaro, D., Jiang, R., Pagano, M., Guillemot, F.,
R.P. (2008). Restricted expression of mutant SOD1 in spinal motor neurons Iavarone, A., and Lasorella, A. (2008). The HECT-domain ubiquitin ligase
and interneurons induces motor neuron pathology. Neurobiol. Dis. 29, 400–408. Huwe1 controls neural differentiation and proliferation by destabilizing the
Young, M.R., and Tye, B.K. (1997). Mcm2 and Mcm3 are constitutive nuclear N-Myc oncoprotein. Nat. Cell Biol. 10, 643–653.
proteins that exhibit distinct isoforms and bind chromatin during specific cell Ziebell, F., Martin-Villalba, A., and Marciniak-Czochra, A. (2014). Mathematical
cycle stages of Saccharomyces cerevisiae. Mol. Biol. Cell 8, 1587–1601. modelling of adult hippocampal neurogenesis: effects of altered stem cell dy-
Young, K.M., Psachoulia, K., Tripathi, R.B., Dunn, S.J., Cossell, L., Attwell, D., namics on cell counts and bromodeoxyuridine-labelled cells. J. R. Soc.
Tohyama, K., and Richardson, W.D. (2013). Oligodendrocyte dynamics in the Interface 11, 20140144.
healthy adult CNS: evidence for myelin remodeling. Neuron 77, 873–885. Ziebell, F., Dehler, S., Martin-Villalba, A., and Marciniak-Czochra, A. (2018).
Yu, G., Wang, L.G., Han, Y., and He, Q.Y. (2012). clusterProfiler: an R package Revealing age-related changes of adult hippocampal neurogenesis using
for comparing biological themes among gene clusters. OMICS 16, 284–287. mathematical models. Development 145, 145.

ll
Article OPEN ACCESS
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER

Antibodies
Mouse monoclonal anti-Ascl1 BD PharMingen RRID:AB_396479
Rat monoclonal anti-GFAP Invitrogen RRID:AB_86543
Chicken polyclonal anti-GFP Abcam RRID:AB_300798
Mouse monoclonal anti-Ki67 BD Biosciences RRID:AB_393778
Rat monoclonal anti-Sox2 Invitrogen RRID:AB_11219471
Goat polyclonal anti-tdTomato Sicgen RRID:AB_2722750
Rabbit polyclonal anti-MCM2 Cell Signaling RRID:AB_2142134
Rabbit polyclonal anti-ID4 BioCheck RRID:AB_2814978
Rabbit monoclonal anti S100b-647 Abcam RRID:AB_2868562
Goat polyclonal anti-DCX Santa Cruz Biotechnology RRID:AB_2088494
Alexa Fluor 488-affiniPure F(ab’)2 Donkey Jackson ImmunoResearch RRID:AB_2340849
anti-mouse IgG
anti-rabbit IgG
anti-rat IgG
Alexa Fluor 488-affiniPure Donkey anti- Jackson ImmunoResearch RRID:AB_2340375
ch IgY
Alexa Fluor Cy3-affiniPure F(ab’)2 Donkey Jackson ImmunoResearch RRID:AB_2340817
anti-mouse IgG
anti-rat IgG
anti-rabbit IgG
anti-goat IgG
Donkey anti-Rat IgG Alexa Fluor 594 ThermoFisher Scientific RRID:AB_2535795
anti-mouse IgG
anti-rabbit IgG
anti-rat IgG
anti-goat IgG
Chemicals, peptides, and recombinant proteins
Doxycycline Sigma-Aldrich Cat# D9891
Tamoxifen Sigma-Aldrich Cat# T5648
Cornflower oil Sigma-Aldrich Cat# C8267
5-Ethynyl-20 -deoxyuridine Santa Cruz Biotechnology Cat# sc-284628
Recombinant Mouse BMP-4 R&D Systems Cat# 5020-BP-010
Recombinant Murine FGF2 Peprotech Cat# 450-33
Laminin Sigma-Aldrich Cat# L2020
Heparin Sigma-Aldrich Cat# H3393-50KU
DMEM/F12 + GLUTAMAX ThermoFisher Scientific Cat# 31331093
DMEM/F12 without phenol red ThermoFisher Scientific Cat# 21041025
(Continued on next page)
Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021 e1

ll
OPEN ACCESS Article
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
DAPI ThermoFisher Scientific Cat# D1306
Critical commercial assays
Click-iT EdU Alexa Fluor 647 Imaging Kit Invitrogen Cat# C10340
Click-iT PLUS EdU Alexa Fluor 647 Invitrogen Cat# C10640
Imaging Kit
Neural Tissue dissociation kit (P) Milteny Biotec Cat# 130-092-628
Deposited data
scRNA-sequencing data This study GEO: GSE159768
Code to analyze scRNA-sequencing data This study https://github.com/harrislachlan/
lifelong_stemcells
scRNA-sequencing data Shin et al., 2015 GEO: GSE71485
Experimental models: cell lines
Primary adult hippocampal wildtype neural Francois Guillemot Blomfield et al., 2019
stem cell line no. 5
Experimental models: organisms/strains
Slc1a3tm1(cre/ERT2)Mgoe (Glast-CreERT2) Mori et al., 2006 MGI:5466676
Gt(ROSA)26Sortm1(EYFP)Cos (RYFP) Srinivas et al., 2001 MGI:2449038
Gt(ROSA)26Sortm9(CAG-tdTomato)Hze Madisen et al., 2010 MGI: 3809523
(tdTomato)
Ascl1tg1(venus)Rik (Ascl1Venus) Imayoshi et al., 2013 MGI: 6369044
Huwe1tm1Alas (Huwe1fl) Zhao et al., 2008 MGI: 4439480
C.129P2(B6)-Gt(ROSA)26Sortm1(tTA)Roos/J Wang et al., 2008 Cat# (JAX Stock): 008603
(R26tTa)
Tg(Nes-EGFP)33Enik (Nestin::GFP) Mignone et al., 2004 MGI: 5523870
Mki67tm2.1(cre/ERT2)Cle (Ki67-creERT2) Basak et al., 2018 MGI: 5816737
Tg(tetO-HIST1H2BJ/GFP)47Efu Kanda et al., 1998 MGI: 3044190
(H2B::GFP)
Software and algorithms
Seurat Stuart et al., 2019 RRID:SCR_007322 (v3.1.2)
Slingshot Street et al., 2018 RRID:SCR_017012 (v1.2.0)
Multimode Ameijeiras-Alonso et al., 2018 https://cran.r-project.org/web/packages/
multimode/ v1.4
GraphPad Prism 8 GraphPad Software RRID:SCR_002798
FIJI v1.0 Schindelin et al., 2012 RRID:SCR_003070
RESOURCE AVAILABILITY
Lead contact
Further information and requests for reagents should be directed to and will be fulfilled by the Lead Contact, François Guillemot
([email protected]).
Materials availability
This study did not generate new reagents.
Data and code availability

The accession number for the raw RNA sequencing data reported in this paper is GEO: GSE159768. The code to reproduce the an-
alyses can be found at Github (https://github.com/harrislachlan/lifelong_stemcells).
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Wild-type animals in this study (Figures 1 and 2) were on a C57BL/6J genetic background (000664, The Jackson Laboratory). In the
scRNA-seq data (Figures 4 and 5) and in a limited number of genetic and EdU-labeling experiments (Figure 1I; Figures S1A–S1D)
e2 Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021

ll
Article OPEN ACCESS
mice were on a C57BL/6J/ CD1 mixed background heterozygous for the following transgenes, Ki67-creERT2 (Basak et al., 2018),
Nestin-GFP (Mignone et al., 2004) and tdTomato (Madisen et al., 2010). In contrast, all other transgenic mice were maintained on
a mixed genetic background with littermates serving as controls. Mice with conditional Huwe1 (Zhao et al., 2008), Ascl1neo/neo (An-
dersen et al., 2014), R26tTa (Wang et al., 2008), and YFP (Srinivas et al., 2001) alleles have been described before; as have the cre-
driver lines Glast-creERT2 (Mori et al., 2006), and the reporter lines Ascl1-venus (Imayoshi et al., 2013), and H2B-GFP (Kanda et al.,
1998). Both male and female mice were used throughout the study, an exception to this was for the x-linked Huwe1 conditional allele,
where only males were used. The effects seen throughout the study were consistent across sexes.
All experimental protocols involving mice were performed in accordance with guidelines of the Francis Crick Institute, national
guidelines and laws. This study was approved by the UK Home Office (PPL PB04755CC). Throughout the study, mice were housed
in standard cages with a 12 h light/dark cycle and ad libitum access to food and water.
METHOD DETAILS
Mathematical modeling
Details of mathematical modeling are found in Data S1.
Cell culture
Adherent cultures of primary hippocampal neural stem and progenitors cells (AHNSC line #5) were propagated in basal media con-
taining 20ng/ml FGF (Peprotech) (Blomfield et al., 2019). Once cells had reached 70% confluence, quiescence was induced with the
addition of BMP4 (R&D) at 20ng/ml, and media was replaced every 3 days.
Preparation and sectioning of mouse brain tissue

Mice were perfused transcardially with phosphate buffered saline (PBS), followed by 4% paraformaldehyde (10-20 mL) and post-
fixed for 16-24 h before long term storage in PBS with 0.02% sodium azide, at 4 C. Brains were sectioned in a coronal plane at
40 mm using a vibratome (Leica). The entire rostral-caudal extent of the hippocampus was collected in a 1 in 12 series.
Antibodies, immunofluorescence, cell counts

At least 1 series per mouse (5-6 sections), per experiment, was stained and analyzed. For experiments requiring the detection of fix-
ation-sensitive antigens (ASCL1, TBR2, Ki67, MCM2) free-floating sections were subjected to heat-mediated antigen retrieval in so-
dium citrate solution (10 mM, pH 6.0) at 95 C for 10 min. A shorter duration of heat-retrieval was used (2 min) if the detection of GFP
was required, which is heat sensitive (Nakamura et al., 2008). Following incubation at 95 C, sections were cooled for 5 min before
being rinsed with PBS and blocked in 2% normal donkey serum diluted in PBS-Triton X-100 (0.1%) for 2 hours. Sections were
then incubated overnight at 4 C with primary antibodies diluted in blocking buffer, washed 3 times in PBS for 10 min, and incubated
with fluorescent secondary antibodies at room temperature for 2 h (Jackson ImmunoResearch). Sections were stained with 40 ,6-dia-
midino-2-phenylindole (DAPI, Thermo Fisher Scientific) and mounted onto Superfrost slides (Thermo Fisher Scientific) with Aqua Pol-
yMount (Polysciences).
The stained sections were imaged using a 40X oil objective on a SP5 or SP8 Leica confocal microscope, with a step size of 1-2 mm
(X/Y pixel diameter = 0.28-0.38 mm). Cell counts were then performed on the imaged sections. To present normalized data in units of
cells/DG, the area of the sampled SGZ was measured (SGZ length*z stack depth). The calculated surface area was then scaled to an
age-appropriate reference dataset from C57BL/6J mice of different ages (0.5, 1, 2, 6, 12 and 18 months) where the total SGZ surface
area had been calculated. For measurements of fluorescence intensity the corrected total cellular fluorescence (CTCF) was calcu-
lated using the following formula: integrated density – (area of region of interest*mean fluorescence of background). The data for fluo-
rescence intensity is presented as arbitrary units in Figures 5 and 6. All cell counts were performed blind to the genotype being
analyzed.
Throughout the analysis we identified NSCs as cells containing a radial GFAP-positive process linked to a SOX2-positive nucleus in
the subgranular zone (SGZ). These cells were defined as quiescent or proliferating depending on expression of the cell-cycle marker
Ki67. In the long-term self-renewal/persistence experiment (Figure 1K) we performed further validation that NSC identity was main-
tained long-term by ensuring the NSC was negative for the astrocytic marker S100b and therefore had not transformed into an astro-
cyte as postulated by the disposable stem cell model (Encinas et al., 2011; Martı́n-Suárez et al., 2019).
Tamoxifen treatment and administration of EdU

Tamoxifen solution (10-20 mg/mL) was prepared for intraperitoneal injections by dissolving the powder (Sigma-Aldrich) in a mix of
10% ethanol and 90% cornflower oil (Sigma-Aldrich). The dose and injection regime were selected according to the relative difficulty
of inducing recombination of the floxed allele, whereas the route of administration changed from intraperitoneal injection to oral
gavage dosing for experiments that lasted longer than 5-days, based on veterinary advice.
For experiments involving the Huwe1fl lines, mice were injected with 80mg/kg of tamoxifen per day, whereas the daily dose deliv-
ered to Ki67TD mice in Figure S1 and H2B-GFP mice in Figure 3, was 100mg/kg. Finally, for the Ki67TD-NES mice used in the scRNA-
seq experiments they were delivered tamoxifen via oral gavage for 6 consecutive days at 100mg/kg. In all experiments, control mice

ll
OPEN ACCESS Article
received tamoxifen at the same doses of experimental mice. The control mice were littermates that were wild-type for the floxed al-
leles and contained the cre transgene or were homozygous for the floxed alleles but negative for cre.
The labeling of cells progressing though S-phase was performed by either intraperitoneal injections of EdU (20 mg/kg) or through
administration of EdU in the drinking water (0.2 mg/ml). For long-term EdU water administration, fresh solution was replaced at least
every 56 hours. The sensitivity of EdU detection kits allowed for lower concentration of EdU to be used than in standard BrdU drinking
water assays (typically 1mg/mL), and avoided obvious cellular toxicity (Young et al., 2013).
Calculation of observed/expected depletion rate

The observed NSC depletion rate was measured by first quantifying the reduction in NSC number between two time points (for
example, between 0.5 months and 1 month of age). This was converted to the rate of change per day, which was normalized to
the size of the NSC pool at the earlier time point (i.e., 0.5 months). This was repeated for all time points (2, 6, 12 and 18 months).
The expected rate of depletion predicted by the disposable stem cell model was calculated by reducing the observed depletion
date at the earliest instance (between 0.5-1 months of age) by the percentage reduction in the size of the proliferating NSC pool
for all later time points (2, 6, 12 and 18 months).
Calculation of contribution to NSC proliferation

The contribution of dormant and resting NSCs to the proliferative (Ki67+) NSC pool is reported in Figures 2 and 6 from experiments in
which mice received EdU in the drinking water for 2 weeks, prior to a 20 h chase period and culling. The number of proliferating (Ki67+)
NSCs originating from the resting NSC pool was determined by multiplying the fraction of those Ki67+ NSCs that were EdU+, with the
age-specific probability that the cell had at some point been in a resting state during the 2-week EdU period (determined by the frac-
tion of total EdU+ NSCs that were Ki67–). This calculation assumes at minimum: the probability of a cell returning to quiescence is
uniform throughout the 2-week EdU administration period, and that the likelihood of a cell returning to quiescence is the same after
the 1st, 2nd or nth division. Finally, the fraction of proliferating NSCs originating from the dormant NSC pool was determined by sub-
tracting the contribution of resting NSCs from the total.
H2B-GFP label dilution experiments

The dilution of the H2B-GFP nuclear label was used as a proxy to determine the number of self-renewing NSC divisions. Juvenile
(0.5 months) and adult (5 months) cohorts were injected with tamoxifen (100mg/kg), each day for 5 days, to activate the creERT2 recom-
binase, recombine the R26tTa locus and induce H2B-GFP expression. The H2B-GFP label was allowed to build up for 14 days after
tamoxifen administration, which was sufficient to produce bright and even labeling of NSC nuclei. After H2B-GFP labeling, the juvenile
and adult cohorts were given 2 mg/ml doxycycline in drinking water, with EdU (0.2mg/mL) and sucrose (1% w/v) for 10 days, switching off
H2B-GFP expression. This 10-day chase period corresponds to the approximate lifespan of an EdU+ NSC in 1-month old mice (Fig-
ure 1L). The solution was changed 3-times per week. After doxycycline treatment, juvenile mice (now 6-weeks of age) and adult mice
(now 6 months of age) were immediately perfused and processed for immunostaining. As an additional control experiment, we later
repeated the experiment in a new adult cohort using different build-up and chase periods. The length of the build-up period was increased
to 25 days to allow for a more robust H2B-GFP labeling, considering that the expression of the Rosa26 locus decreases with age. Impor-
tantly, we did not notice any difference in the sensitivity of H2B-GFP signal detection between the different build-up periods, suggesting
that the extent of labeling was similar. Furthermore, the label-dilution period was adjusted to 30-days to match the approximate lifespan
of an EdU+ NSC in adult mice, where it takes 30 days for more than 90% of the EdU+ NSC population to differentiate (Figure 1L). The
experiment started when the mice were 4.5 months old to ensure they were 6-months old at the end of the chase period.
Sections were stained using antibodies against GFAP and SOX2 and were also stained for EdU and DAPI. Cell counts and GFP
fluorescence intensity measurements were performed on the entire DG at different levels along the rostrocaudal axis. The GFP fluo-
rescence intensity of EdU+ NSCs was compared to the mean GFP fluorescence intensity of 2-3 neighboring EdU– NSCs, to calculate
a ratio. To minimize variability caused by staining or imaging artifacts, the EdU– cells that were selected for comparison had to be in
the same X-Y image tile (291*291 mm) and their center of mass within 8 z-planes (6.04 mm) of the EdU+ NSC. Segmentation of nuclei
was performed using the Fiji plugin ‘3D Object Counter’ (Bolte and Cordelières, 2006), and the raw integrated density for each nu-
cleus was used for the measurement output. Data points were excluded if the standard error of the mean of the average fluorescence
intensity of EdU– NSCs was greater than 15% of the mean, as inclusion of this data detrimentally affected normalization.
All datasets from the juvenile and adult cohorts were pooled together to estimate the most probable number of clusters (modes) in
the distribution, using the ‘‘modetest’’ function of the R package ‘‘multimode’’ (Ameijeiras-Alonso et al., 2018). Modes and anti-
modes were then located by means of the ‘‘locmodes’’ function. and used to create bins that separated the values based on the num-
ber of cell divisions (n) that single EdU+ NSCs performed. The number of values falling in each bin was counted for each dataset.
Therefore, we obtained the fraction of cells that divided n-times in each dataset.
Single-cell analysis of Ki67TD-NES mice

Sample preparation
Each scRNA-seq experiment comprised 2-5 male and female Ki67TD; Nestin-GFP mice that had received tamoxifen via oral gavage
for 6 consecutive days (100mg*kg) and culled on the 8th day. In total, 1 experiment was performed with 1-month-old mice, 3 exper-
iments with 2-month old mice, and 3 experiments with 6-8-month-old mice (Table S2).

ll
Article OPEN ACCESS
Mice were killed by cervical dislocation, their brains removed and the DG was microdissected (Hagihara et al., 2009; Walker and
Kempermann, 2014). The dissected DG was then disassociated using the Neural Tissue dissociation kit (P) (Milteny Biotec) according
to manufacturer’s instructions, with the following exceptions: a 37 C orbital shaker was used during the enzymatic digestions, and
we used manual trituration with fire-polished pipettes to aid dissociation following the incubations with enzymatic mix 1 and 2. The
dissociated cells were centrifuged and resuspended in 750 mL recovery media (0.5% PBS-BSA in DMEM/F-12 without phenol red
and 1ug/ml DAPI. The cells were sorted on the MoFlo XDP (Bechman Coulter) using a 100mm nozzle with a pressure of 30 psi,
and a sort efficiency of more than 80%. The events were first gated to remove debris (fsc-h versus ssc-h), to remove aggregates
(typically pulse-width versus area then pulse-height versus area/width) and to remove dead cells (fsc-h versus DAPI fluorescence).
Cells were then gated for tdTomato expression according to a control mouse that expressed Nestin-GFP alone; whereas the GFP
gate was set according to the expected distribution of GFP fluorescence intensities based on prior control experiments. Cells
were collected in 700 ml of recovery media in 1.5mL tubes, and spun down at 500G for 7 min at 4 C. After removing all but 50ml of
the supernatant, the cells were then gently resuspended using a wide-bore pipette. The single-cell suspension (to a maximum of
10,000 cells) was then loaded into the 10x Chromium.
Sequencing and mapping
We prepared one library for each of the 8 samples across the 7 experimental days. The libraries for 2-month-old mice were prepared
with 10x Genomics Chemistry, Single Cell 30 version 2, while the 1-month and 6-month libraries were prepared with version 3 (Table
S2). After sequencing, cellRanger count (Ver. 3.0.2) was used to map the FASTQ files to our custom mouse genome (mm10-3.0.0),
which contained 4 functional elements. The first element was the GFP coding sequence to detect transcription of Nestin-GFP that
appears as ‘‘eGFP’’ in the final data matrix. The second, was the 50 floxed sequence of the tdTomato allele that detects transcription
of the intact tdTomato locus and appears as ‘‘tdTomatoLoxP’’ in the final data matrix. Finally, we added 30 regions of the tdTomato
locus that are transcribed upon recombination, which encode the woodchuck hepatitis virus post-transcriptional regulatory element
(appears as ‘‘WPRE’’ in the final data matrix) and the bovine growth hormone polyA signal (appears as ‘‘bGHpolyA’’ in the final data
matrix). The coding sequence of the tdTomato gene was also originally included in the custom genome as a 5th functional custom
element, however very few transcripts mapped to this region, and the origins of these transcripts (intact versus recombined tdTomato
locus) were unclear as the tdTomato coding sequence is directly downstream of the 50 polyA signal but more than 600 base pairs
upstream of the 30 polyA signal. As such the data was remapped with this feature disabled. The disabling of this genomic region
is marked by the empty feature ‘‘tdTomatoCDS’’ in the final data matrix. The custom genome was made with the cellRanger mkref
command.
Seurat analysis: Quality control
The cellRanger aggr command was used to merge the count files of the 8 individual libraries without normalization. The merged ma-
trix was then read into Seurat (Ver. 3.1.2) for analysis (Stuart et al., 2019). GEM beads were kept for further analysis if they contained
more than 500 genes and fewer than 10% mitochondrial reads in order to remove GEM beads that contained only background signal
or a dead cell. Doublets/multiplets were then removed based on the expression of more than 1 marker gene per GEM bead (e.g., co-
expression of the astrocytic marker Aldoc with the oligodendrocyte marker Mog). In total, 24,203 of 26,036 (92.95%) passed these
quality control steps. Later, during the subsetting/re-clustering of the data further small doublet clusters appeared, which were iden-
tified and removed based on the lack of specific marker expression (Briggs et al., 2018).
Seurat analysis: Distinguishing G1 from G0
We then added metadata that marked each cell for tdTomato expression and cell-cycle status. Cells were identified as tdTomato+ if
they contained < 4 reads of the intact tdTomato locus, and more than > 1 read of the recombined tdTomato locus based on ground-
truth testing of our first dataset, in which we sequenced the tdTomato+ and tdTomato– populations separately (Figure S3). We also
identified whether a cell was in G0 or proliferating using ground-truth testing. We isolated bona-fide S-phase cells through the Cell-
CycleScoring function in Seurat, and bona-fide G0-phase NSCs based on UMAP position and plotted the expression in these cells of
genes known to be highly and stably expressed across both G1- and S-phases (i.e., Mcm2-7, Ccne1/2, and Pcna) (Braun and Bree-
den, 2007).We binarized each cell as positive or negative for each marker and plotted every cell according to their index score (range
of possible values 0-9). These plots revealed a clear separation of two cellular populations which we used to distinguish in the
remainder of the dataset, tdTomato+ cells that were in G1 (that we marked as proliferating NSCs) or in G0 (resting NSCs). Specifically,
cells in which we detected > 1 index gene were scored as in G1, while all other cells were considered quiescent/post-mitotic. Cells
sequenced with Version 3 of the 10x Chemistry had on average twice the number of genes detected, so the thresholds for tdTomato
and cell-cycle scoring were doubled.
Seurat analysis: Data visualization
The merged Seurat object was split according to library/experimental day and the data transformed using the SCTransform function.
The 5,000 most variable genes and anchors across the 8 datasets were then identified and the data integrated using the default pa-
rameters. The integrated dataset was then visualized using UMAP (Becht et al., 2018). Elbow plots were used to determine the num-
ber of significant principal components. Following cluster identification with known marker genes, we extracted all NSCs and IPCs
and re-clustered these cells using the same workflow. We iterated this process a further time to then extract only NSCs, this time
splitting and integrating the Seurat object according to 10x Chemistry (V2 or V3), as this best eliminated technical variation without
removing biological signal (assessed by clustering of replicates). We appended metadata classifying these cells as dormant NSCs
(quiescent and tdTomato–), resting NSCs (quiescent and tdTomato+) or proliferating NSCs (proliferating and tdTomato+/–). The final
dataset contained 2,947 cells which we processed for differential gene expression and pseudotime ordering.

ll
OPEN ACCESS Article
Pseudotime ordering
To construct a pseudotime ordering of NSCs we used the trajectory inference tool Slingshot (Street et al., 2018) (Ver. 1.2.0) and UMAP
dimensionality reduction. The curve generated by Slingshot was used to plot the pseudotime position of each cell between condi-
tions and to generate a set of genes statistically associated (p < 0.01) with pseudotime progression. For the reconstruction of the
pseudotime trajectory from Shin et al. (2015) we read in the normalized counts (TPM) from GSE71485 into Seurat. Next PCA analysis
was performed using the 5 top components and the data was then visualized using UMAP. We excluded a cluster of cells that lacked
significant expression of Hopx that largely corresponded to contaminating oligodendrocytes, as reported in the original study. We
compared the set of genes that were statistically associated with pseudotime with our dataset.
QUANTIFICATION AND STATISTICAL ANALYSIS
Single-cell data
Differential gene expression analysis was performed using the FindMarkers function in Seurat using the Pearson residuals located in
the ‘‘scale.data’’ slot of the SCT assay using Student’s t test (Hafemeister and Satija, 2019). No minimum log fold change threshold
was enforced; however, to be included in the final analysis, all genes had to be expressed by a minimum of 20% of cells in at least one
of the two conditions being compared and were considered as statistically significant only if they had an FDR adjusted P-value of <
0.05. For visualisation of gene expression differences between groups the normalised data from the "RNA" assay are presented as
violin plots throughout the manuscript.
Gene Ontology
The Gene Ontology analysis was performed on statistically significant genes using the clusterProfiler package (3.12.0) (Yu et al.,
2012) with the onotology term ‘‘BP’’ for biological process. This analysis was performed separately on 1) all statistically significant
genes together, 2) upregulated genes only and 3) downregulated genes only. The results of these analyses are reported in Tables
S3, S4, and S5.
Statistical analysis of cell counts

The statistical testing approach were specified prior to data collection and implemented using Graphpad Prism (version 8.0) or in R
(3.6.2). Two-tailed unpaired Student’s t tests or Mann-Whitney U tests were performed when comparing two groups. For experiments
involving one or two independent variables, one or two-way ANOVA was performed, respectively. Any significant main effect de-
tected by ANOVA was followed by multiple t tests when required using a pooled estimate of variance where appropriate, and signif-
icance was corrected for using the Holm-Sidak method in Prism 8.0 (Graphpad). Fisher’s exact test was used in Figure 3. Details of
statistical tests for specific experiments are found in figure legends.

Untitled

Uploaded by

Copyright:

Available Formats

Untitled

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Untitled

Uploaded by

Copyright:

Available Formats

Article

Coordinated changes in cellular behavior ensure the

d Dormant stem cells enter deeper quiescence with age

d These changes drive the transition from developmental to

d Increasing degradation of ASCL1 protein by HUWE1

Harris et al., 2021, Cell Stem Cell 28, 863–876

Bohr Gasse 3, 1030 Vienna, Austria

Figure 1. Proliferating NSCs increasingly return to quiescence with time

(legend continued on next page)

864 Cell Stem Cell 28, 863–876, May 6, 2021

(L) Quantification of NSC persistence.

Cell Stem Cell 28, 863–876, May 6, 2021 865

and 1L). These findings demonstrate

866 Cell Stem Cell 28, 863–876, May 6, 2021

(legend on next page)

Cell Stem Cell 28, 863–876, May 6, 2021 867

Figure 3. NSCs perform more self-renewing divisions with time

868 Cell Stem Cell 28, 863–876, May 6, 2021

Cell Stem Cell 28, 863–876, May 6, 2021 869

The effect of age on the depth of

870 Cell Stem Cell 28, 863–876, May 6, 2021

Cell Stem Cell 28, 863–876, May 6, 2021 871

DISCUSSION et al., 2010). Previous experiments have demonstrated that

872 Cell Stem Cell 28, 863–876, May 6, 2021

Figure 7. Expression and activity of Huwe1 increases with time

Cell Stem Cell 28, 863–876, May 6, 2021 873

874 Cell Stem Cell 28, 863–876, May 6, 2021

Cell Stem Cell 28, 863–876, May 6, 2021 875

876 Cell Stem Cell 28, 863–876, May 6, 2021

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021 e1

Data and code availability

EXPERIMENTAL MODEL AND SUBJECT DETAILS

e2 Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021

Preparation and sectioning of mouse brain tissue

Antibodies, immunofluorescence, cell counts

Tamoxifen treatment and administration of EdU

Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021 e3

Calculation of observed/expected depletion rate

Calculation of contribution to NSC proliferation

H2B-GFP label dilution experiments

Single-cell analysis of Ki67TD-NES mice

e4 Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021

Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021 e5

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis of cell counts

e6 Cell Stem Cell 28, 863–876.e1–e6, May 6, 2021

You might also like